Since 2000, there has been an explosion of activity in the field of plasmon-enhanced Raman spectroscopy (PERS), including surface-enhanced Raman spectroscopy (SERS), tip-enhanced Raman spectroscopy (TERS) and shell-isolated nanoparticle-enhanced Raman spectroscopy (SHINERS). In this Review, we explore the mechanism of PERS and discuss PERS hotspots — nanoscale regions with a strongly enhanced local electromagnetic field — that allow trace-molecule detection, biomolecule analysis and surface characterization of various materials. In particular, we discuss a new generation of hotspots that are generated from hybrid structures combining PERS-active nanostructures and probe materials, which feature a strong local electromagnetic field on the surface of the probe material. Enhancement of surface Raman signals up to five orders of magnitude can be obtained from materials that are weakly SERS active or SERS inactive. We provide a detailed overview of future research directions in the field of PERS, focusing on new PERS-active nanomaterials and nanostructures and the broad application prospect for materials science and technology.
The properties of a surface have crucial roles in determining its adhesion, friction, wear and wetting behaviour, as well as surface chemical processes such as catalysis, corrosion, sintering, composite formation and electrochemistry1,2. With the development of surface engineering techniques and nanotechnology, research has shifted from homogeneous to composite materials, from almost perfect single-crystal surfaces to surfaces with functionalized ‘active sites’, and from thin-film materials to promising 2D materials with one to several atomic layers in thickness. Meanwhile, characterization techniques have been developed to reveal the structure–property relationships of these emerging materials. However, for most techniques, the signals obtained from the surfaces are either too weak to detect or are difficult to resolve into distinct components because of the low spectral and spatial resolution. Therefore, it is necessary to develop in situ techniques with ultrahigh sensitivity, surface specificity, high spectral resolution, and high spatial and temporal resolution.
One such technique is surface-enhanced Raman spectroscopy (SERS)3,
PERS enhancement is strongly dependent on the optical properties, shape and
aggregation of nanomaterials24. In
the 1970s and 1980s, SERS-active substrates, such as roughened Au and Ag electrodes,
colloidal aggregates and microlithographic particle surfaces were widely
explored3,10,25 (Fig. 1). However, at that time, techniques to
fabricate metallic structures at the nanometre scale were limited. PERS benefited
considerably from developments in nanoscience in the 1990s26: in particular, from new techniques to synthesize
and characterize SERS-active nanoparticles or nanostructures. Some examples of
SERS-active nanostructures are Au and Ag nanoparticles with nanometre-sized gaps
(nanogaps) and nanometre-sized tips (nanotips), and structured surfaces with
nanometre-sized holes (nanoholes), voids, bumps, grooves or ridges15,26,
In the early stage, SERS was predominantly used to probe (sub-)monolayer molecules
adsorbed on roughened coinage-metal surfaces or on metal colloids with an average
SERS enhancement factor of 105–107 (Fig. 1). A great effort has been made to extend
SERS for the surface analysis of various adlayer structures on other materials that
are important to electrochemistry, corrosion inhibition and heterogeneous catalysis,
such as Group VIIIB transition metals (Pt, Pd, Ru, Rh, Fe, Co and Ni)38,39. During the 1990s and 2000s, these transition metals, if
prepared as appropriate nanomaterials, exhibited38,41,
To improve the SERS enhancement factor on transition-metal surfaces, a strategy (which we refer to as ‘borrowing SERS’) was devised to coat transition metals as an ultrathin shell or overlayer on the surface of Au and Ag nanostructures26. Boosted by the long-range effect of the enhanced electromagnetic field generated from the highly SERS-active Au or Ag cores, the chemisorption information of probe molecules on the transition-metal overcoating can be extracted by SERS measurements with an enhancement factor of up to 104–105. In principle, this strategy could be extended to probe other overcoating materials beyond metals without SERS activity on Au and Ag nanostructures (Fig. 1).
There are, however, several shortcomings of the borrowing SERS strategy. The SERS enhancement factor of the overcoating or shell material strongly depends on the thickness of the coating. In general, to obtain a strong SERS enhancement, the shell thickness should be less than 2 nm (that is, about seven atomic layers) for transition metals and 5 nm for dielectric materials. It is challenging to coat target materials, such as insulators or semiconductors, as an ultrathin shell on the surfaces of Au or Ag nanostructures. Furthermore, the shell should be free of pinholes in the coating materials to avoid the adsorption of analytes directly on the surface of Au or Ag nanostructures, which leads to misinterpretation of the Raman signal15. Moreover, in many cases, it is necessary to characterize the pristine surface structures of materials in their native environment for a given application. For example, the surface structures of silicon wafers and of blades in motors cannot be represented or modelled by an ultrathin structure or even a structure with nanoscale curvature. Therefore, new working modes of SERS were developed.
Two such working modes are those of TERS11,
More than 500 review articles on PERS have been published24,45,55,
In the following section, we describe the first and second generations of SERS hotspots. We then discuss new, third-generation PERS hotspots generated by electromagnetic coupling in hybrid structures that contain both PERS-active nanostructures and probe materials. We focus our discussion on three key PERS working modes — contact-mode (SERS), shell-isolated mode (SHINERS) and non-contact mode (TERS) — and highlight recent advances in PERS applications in a range of material systems. Finally, we offer our perspective on the promising interplay between plasmon-enhanced spectroscopy and materials science, with an emphasis on new plasmonic materials and new nanostructures for PERS, and other plasmon-enhanced spectroscopic techniques.
First- and second-generation hotspots
The SERS enhancement factor is approximately proportional to the fourth power of the
local electric field (Box 2). More
importantly, the electromagnetic field around plasmonic materials is not uniformly
distributed but highly localized in spatially narrow regions (SERS hotspots), such
as nanotips, interparticle nanogaps or particle–substrate
The first-generation hotspots were made from single nanostructures, such as nanospheres and nanocubes (Fig. 2a,b), or nanorods freely suspended in a homogeneous medium. These hotspots exhibit moderate SERS activity; however, some rationally designed single nanoparticles with sharp corners and/or with intraparticle gaps, such as Au and Ag nanostars, nanoflowers and mesocages, exhibit much higher SERS activity72.
The second generation of SERS hotspots are generated from coupled nanostructures with controllable interparticle nanogaps (for example, the nanoparticle dimers shown in Fig. 2c,d, or oligomers and nanoparticle arrays shown in Fig. 2e,f) or interunit nanogaps in nanopatterned surfaces (Fig. 2g,h). Such hotspots exhibit excellent SERS activity. Because the average SERS intensities from coupled plasmonic nanostructures are typically two to four orders of magnitude greater than those from single nanostructures66,73, they are more commonly used for trace-molecule detection.
The size of SERS hotspots from coupled nanostructures is extremely small (1–5 nm), but the Raman signals of probe molecules at the hotspots contribute significantly to the total Raman signal74,75. For example, the hotspot of a Ag nanosphere dimer (2-nm gap size) occupies less than 1% of the total area but can contribute more than 50% of the total SERS signal, provided that the molecules are uniformly distributed on the surface64,76. The extremely strong electromagnetic field in the tiny volume of the second-generation hotspots from coupled nanostructures is crucial for detecting and analysing trace amounts of molecules, including single molecules, located at the hotspots. An implication of this is that the probe molecule should be located at the hotspots.
Two main approaches (bottom-up and top-down) have been developed to prepare coupled
nanostructures that can support second-generation hotspots for SERS. Bottom-up
methods were first used to prepare highly monodispersed Au and Ag nanoparticle
aggregates with dense SERS hotspots for application in the detection of trace
molecules in the solution or gas phase10,77. However,
nanoparticle aggregates are often not structurally well defined. Thus, high-yield
and stable Au and Ag nanoparticle dimers and oligomers with well-defined nanogaps
were prepared, to improve the reproducibility of single-molecule detection30,33,78,
Top-down techniques, such as electron-beam lithography, focused-ion beam and photon lithography, are also used to fabricate coupled nanostructures75,92. The advantage of top-down nanotechnology is its ability to fabricate SERS-active nanostructures with a wide range of geometries, ultrafine features, good reproducibility and large-scale uniformity69; examples include ordered clusters or arrays of pillars, holes, grooves and ridges with sharp corners and small gap size. Hierarchical structures, such as particle-in-cavity93, particle-in-groove and particle-on-film94,95 structures, have been developed for SERS analysis, because they can strengthen the plasmonic response and boost the overall SERS enhancement factor. However, challenges of top-down nanotechnology lie in the difficulties in controlling the nanogap below 5 nm and the atomic-scale surface roughness.
At present, second-generation SERS hotspots are dominant in both fundamental and
applied research. For example, hotspots are commonly designed for the verification
of SERS mechanisms and the evaluation of the performance of newly developed
nanostructures in SERS applications. Trace-molecule detection (for example,
single-molecule detection in molecular electronics devices96) and biomolecule analysis (for example, pesticide
residues, drugs, tumour and DNA sequences) have been successfully achieved40,97,
In general, first- and second-generation hotspots are not well suited for the surface analysis of many materials. For example, widely used materials, such as silicon wafer or ceramics, cannot be squeezed into the extremely tiny and narrow regions of the hotspots formed by coupled nanostructures, such as those shown in Fig. 2c–h. Therefore, it is highly desirable to design plasmonic nanostructures that can have hotspots right on the surface of the materials to be probed. This can be achieved by taking into account the effect of the materials to be probed when designing the plasmonic nanostructures, because the localized SPR response and local electromagnetic field distribution not only depend on the shape and size of these plasmonic nanostructures, but also depend sensitively on the dielectric properties of probe materials close to such nanostructures. Hotspots on probe material surfaces, which are generated from hybrid structures consisting of plasmonic nanostructures and the probe materials, can be considered third-generation hotspots. Representative examples are shown in Fig. 3.
The simplest example is a single Au nanoparticle on a flat Si or Pt surface62,94,101,
The enhancement can also be tuned by using different plasmonic nanostructures. For example, the average SERS enhancement factor over the projected area of a single Au nanocube on the Si surface (Fig. 3b, left) is about two orders of magnitude larger than that of a single Au nanosphere (Fig. 3a, left). When a bare Au nanosphere dimer is placed on a Si or Pt surface, hotspots can be simultaneously produced at the particle–substrate and interparticle nanogaps70 (Fig. 3c) for a certain range of incident wavelengths. For dielectric materials (for example, Si), the average SERS enhancement factor at the surface in the presence of a Au nanosphere dimer (Fig. 3c, left) is about three orders of magnitude larger than that of a single Au nanosphere (Fig. 3a, left) at the surface. The average SERS enhancement factor increases with the refractive index of the dielectric material. For metal surfaces (for example, Pt), the average SERS enhancement factor at the surface in the presence of a Au nanosphere dimer (Fig. 3c, right) is about three orders of magnitude larger than that of a single Au nanosphere (Fig. 3a, right) at the surface, owing to the additional plasmonic coupling between the Au nanosphere dimer and the metal surface70. When a shell-isolated Au@SiO2 dimer is placed on a Si or Pt surface, hotspots can be simultaneously produced at the particle–substrate and antiparticle nanogaps (Fig. 3d). Additionally, the chemically inert shell can avoid unwanted or unexpected interfering signals from the interfaces. To obtain higher sensitivity, nanoparticle clusters can be used instead of single nanoparticles on the surface of the probe material. The SERS enhancement also significantly depends on the particle–substrate nanogap, and on the wavelength and polarization states of the incident laser.
Contact-mode SERS for surface characterization. Third-generation hotspots can be realized in three different working modes (Fig. 4). The earliest and simplest is contact-mode SERS (Fig. 4b), in which SERS-active bare Au or Ag nanoparticles are coated on the surface of the probe material to enhance the surface Raman signals104. The bare nanoparticles either directly contact the probe material surface or contact self-assembled probe molecule layers on a metal substrate (the latter is referred to as gap-mode SERS in the literature62,67,94,105).
Contact-mode SERS exhibits excellent enhanced Raman signals in various applications; however, it has several significant limitations, which may be why it has not been more widely adopted by the materials science community in the past three decades. For example, solute molecules or contaminants in solution or gas can easily be adsorbed on the surfaces of the bare Au nanoparticles most commonly used in SERS and can then diffuse into the particle–substrate hotspots (Fig. 4b). As a consequence, the SERS spectra might be dominated by contaminated Raman signals, leading to the misinterpretation of SERS spectra of adsorption and surface processes.
In addition, interfacial charge transfer may occur between some hard materials (such as Pt, Ni or Co) and bare Au or Ag nanoparticles because of their differences in work functions (Fig. 4b). Unwanted interference may be stronger for soft materials (such as liquid crystals, polymer films and molecular self-assemblies), because the contact of bare Au nanoparticles may result in the formation of new chemical bonds or induce a photochemical reaction under strong laser illumination15 (Fig. 4b). Therefore, one must be cautious about interference from the direct contact between bare Au or Ag nanoparticles and the probe material to avoid misinterpretation of the surface Raman spectra47. Moreover, for hybrid structures including single or clustered Au or Ag nanoparticles on a metal substrate, the nanoparticles and the metal substrate may fuse under strong laser illumination. The fusion of Au or Ag nanoparticles with the metal substrates damages the hybrid structure, thereby adversely affecting the local electromagnetic field distribution and the SERS performance of hybrid structures. Accordingly, probe adsorbates on fused metal surfaces also affect the original surface Raman spectra of probe molecules on the metal substrates.
To overcome these drawbacks of contact-mode SERS, two working modes with more versatile and reliable surface analysis capability have been developed: TERS, which operates in non-contact mode, and SHINERS (Fig. 4c,d). These two modes can support third-generation hotspots and the details are discussed in the following two sections.
TERS. TERS was invented11,
Although other methods can characterize the surface morphology of materials with atomic resolution, it remains challenging to investigate the structure, chemical composition and morphology of material surfaces simultaneously. TERS combines the advantages of scanning probe microscopy (that is, high spatial resolution) and SERS (that is, highly sensitive molecular information). In TERS, a plasmonic Au or Ag tip with a radius of about 20 nm is used to create a single hotspot between the tip and the probe material. When the tip approaches a sample surface, the electromagnetic field and the Raman signal of the sample are significantly enhanced to provide high detection sensitivity46 (Fig. 4c). Because the enhanced field is highly localized to the tip apex, it can selectively probe the local chemical and electronic structure at a spatial resolution of about 10 nm.
In contrast to SERS and SHINERS (described in the following section), TERS offers only one hotspot for Raman measurements. Thus, its sensitivity is typically much lower than that of SERS and SHINERS. However, TERS provides nanometre spatial resolution, which is not achievable with either SHINERS or SERS. If a metallic or dielectric material with a high refractive index is to be investigated, gap-mode TERS can be used to improve the field enhancement, sensitivity and spatial resolution14.
SHINERS. SHINERS was invented15 in 2010 and uses shell-isolated nanoparticles composed of plasmonic Au or Ag cores with ultrathin (1–5 nm) and chemically and electrically inert shells (for example, of SiO2 or Al2O3). The advantages of SHINERS are fourfold. First, the ultrathin yet pinhole-free shells separate the cores from the material surface (and the environment), thus ensuring that there is almost no interference from Au and Ag cores. Second, the chemically inert shell effectively avoids interparticle and particle–metal substrate fusion, which significantly improves the stability of the nanoparticles and the probe structures. Third, the shell thickness can be used to control the nanogap between the Au or Ag core particle and the substrate, and consequently determines the particle–substrate electromagnetic coupling. Last, the Au and Ag cores can boost the local electromagnetic field to enhance the Raman signals from the probe substrate without distorting its structure15,108 (Fig. 3d).
The shell-isolated mode in SHINERS is different from contact-mode SERS in that the probe materials cannot directly interact with the Au and Ag cores because of the inert, isolating ultrathin shells (Fig. 4d). Because it is implicit in the name SERS that the surface of the SPR-active nanostructure directly touches the probe target (either molecules or materials), SHINERS cannot be classified as a SERS technique15. This is also true for TERS. Because SERS, TERS and SHINERS involve local field enhancement due mainly to the enhancement of SPR, each can be considered as a unique PERS technique.
SHINERS overcomes several disadvantages of contact-mode SERS for surface analysis but at the cost of reduced Raman enhancement. Both interparticle plasmonic coupling and nanoparticle–substrate coupling in shell-isolated nanoparticle-on-substrate systems are moderate when compared with bare nanoparticle-on-substrate systems. The average SHINERS enhancement factor is about six and five times weaker than that from a bare Au nanoparticle dimer on Si and Pt surfaces, respectively (Fig. 3c,d).
In practice, shell-isolated nanoparticles are simply spread as ‘smart dust’ over the surfaces of materials with diverse compositions and morphologies15. However, SHINERS requires high-quality, pinhole-free shell-isolated nanoparticles. Therefore, if the aforementioned limitations due to bare nanoparticles are acceptable for a given application, it is appropriate to use SERS with bare nanoparticles to attain a higher sensitivity than SHINERS. However, when interference effects due to bare nanoparticles are severe, SHINERS is the technique of choice.
Applications of SERS, TERS and SHINERS
Applications based on third-generation hotspots using SERS, TERS and SHINERS can be divided into two categories. The first is the surface characterization of materials such as liquid crystals, silicon or polymers, in which the structure of the surface may be different from the bulk one, as well as nanomaterials, such as graphene, carbon nanotubes and magnetic nanocrystals. The second category is the determination of the molecular composition and surface bonding of molecular adlayers or surface species interacting with the substrate, such as oxygen species on Pt, hydrogen on Si, corrosion inhibitors on stainless steel or self-assembled layers on structured materials. PERS can be used to investigate several interface-related processes such as growth kinetics of films, mechanically induced conformational changes, ion-intercalation processes, corrosion and electrocatalysis. Some selective examples are shown in Fig. 5 and detailed in the following sections. Such investigations would greatly deepen our fundamental understanding of the structure–property relationships of emerging materials.
Applications of contact-mode SERS. Polyimide alignment layers (PALs) with high pretilt angles have a crucial role in controlling the orientation of liquid crystals. By mechanical rubbing of the PALs, as shown in Fig. 5a, the bulk orientation of liquid crystals can be controlled in a preferential direction by interface-induced alignment between the liquid crystals and PALs. However, this phenomenon, as well as the surface structures of the PALs before and after rubbing, is not well understood. SERS with bare Ag colloids has been used to reveal that both the pendent cyanobiphenyls in the side chains and backbones in the first PALs possess nearly planar chain conformations before surface rubbing109. After rubbing, some of the backbone moieties (for example, imide-phenylene) tilted, and the conformation of the pendent side chains at the surface was significantly different (Fig. 5a). This conformational rearrangement of the side chains resulted in the formation of fold-like bent structures on the surface, which directly leads to a large negative pretilt angle of the long axis of cyanobiphenyls with respect to the rubbing direction.
Liquid-phase chemical bath deposition is an important technique for depositing metals, semiconductors and dielectric thin films over large areas because of its low cost and scalability. However, in situ characterization of the growth kinetics during the early stages is challenging. SERS was recently used110 to study the early-stage growth kinetics during the liquid-phase chemical bath deposition of CdS. No Raman signal was observed for the CdS film after 5 minutes of deposition on a pure quartz surface; by contrast, the Raman signal (with peaks at 305 and 600 cm−1) of CdS on Ag nanoparticles was pronounced and increased with deposition time. Interestingly, the Raman intensity of CdS saturated after 10 minutes of deposition (∼20-nm thickness), which can be explained in terms of preferential deposition of CdS on itself (Fig. 5b).
Vanadium oxide is an abundant, low-cost metal oxide that is well suited for ion intercalation. In a recent study111, SERS was used to probe the ion-intercalating processes in electrodes by placing a bare single Au nanoparticle on a vanadium oxide surface to form a Au nanoparticle–dithiol–vanadium oxide nanojunction. When cycling the electrochemical potentials, real-time chemical and structural changes (obtained through variations of the Raman frequencies and intensities) were found to correlate with ion intercalation–extraction processes at the vanadium oxide/electrolyte interface. Thus, this study provided a unique SERS-based strategy to probe ion intercalation processes.
Applications of TERS. A typical STM-based TERS set-up is illustrated in Fig. 6a. The tip is brought to about 1 nm from the material surface, which is then illuminated by a laser of suitable wavelength. As the tip is scanned over the surface, a TERS image of the Raman characteristics of the material can be obtained with high spatial resolution.
Barium titanate, BaTiO3, is an important ferroelectric perovskite material used in piezoelectric actuators and non-volatile memory. The nanoscale characterization of this material may lead to a fundamental understanding of its nanoscale ferroelectric order and domain behaviour. TERS is a particularly powerful technique for characterizing the phases in crystal domains. For example, the Raman-active longitudinal optical mode of BaTiO3 is characteristic of the tetragonal phase and is a marker of the ferroelectric state112. Ferroelectric domains have been successfully identified in a BaTiO3 nanorod using TERS at a spatial resolution that is not obtainable using a conventional Raman microscope (Fig. 6b). The lattice constants differ by only 0.044 Å between the cubic paraelectric phase and the tetragonal ferroelectric phase, making their distinction difficult to observe even with high-resolution transmission electron microscopy in combination with electron diffraction techniques. However, TERS can be used to sensitively identify the difference between phases, and it has great potential in characterizing ferroelectric materials.
Chemical analysis of defects, strain and surface chemistry at nanometre resolution is highly desirable in carbon materials, including carbon nanotubes and graphene. Recently, TERS has been used for chemical analysis of single-layer graphene on a Au substrate52. Using the 2D-band (2,662 cm−1) signature of single-layer graphene and the D-band (1,350 cm−1) signature of the defects, a spatial resolution of better than 12 nm was obtained, with full spectral information in every pixel. Two defects within the surrounding graphene layer were clearly identified (Fig. 6c). In a separate study53, 1.7-nm spatial resolution was achieved under ambient conditions, simultaneously providing chemical and topological information of individual carbon nanotubes. In this study, different types of carbon nanotubes, local defects and bundling effects were visualized in real space, as shown in Fig. 6d. Most importantly, this method does not rely on an ultrahigh-vacuum and cryogenic environment and is suitable for analysing soft materials, which is not possible with general electron microscopy techniques. More recently, TERS has been successfully used to plasmonically generate and detect spin waves (a collective oscillation of spin moments in a crystal, which may have an important role in spintronics) in α-Fe2O3 nanocrystals113. The two-magnon mode (spin wave) of α-Fe2O3 nanoparticles, which is not observable in bulk materials, becomes prominent at 1,587 cm−1 in the TERS spectrum as a result of phonon-mediated coupling of the spin of d-orbital electrons with the strongly localized electromagnetic field in the vicinity of the tip apex.
In gap-mode TERS, if the localized SPR of a Ag tip and Ag substrate is tuned to match
the electronic resonance of the molecule, the local field may be further confined to
produce ultrahigh spatial resolution. Such tuning can be realized by vacuum
bombardment of the Ag tip. An ultrastable, low-temperature and ultrahigh-vacuum STM
in the TERS system can efficiently prevent thermal drift of the STM system and
suppress lateral movement of the molecule. This approach allows spatial resolution
of TERS on the sub-nanometre scale and has been applied to resolve the structure and
configuration of a single porphyrin molecule on a surface114, as shown in Fig.
6e. Different vibrational modes show slightly different patterns.
Although the detailed mechanism underlying this unprecedented spatial resolution is
still under debate, this work has already launched a surge of theoretical
TERS can be used not only to characterize materials under static conditions as described above, but also to probe dynamic processes. In molecular electronics, the molecular structure and interface between the molecule and conducting lead are two of the most important criteria in searching for suitable materials. It remains a great challenge to characterize and gain detailed structural information of the molecule–metal–molecule junction during the electron transport. Fishing-mode TERS was developed to increase the probability of forming a Au–4,4′-bipyridine–Au junction in the gap between a Au tip and Au substrate118 (Fig. 6f). This allows mutually verifiable single-molecule conductance and Raman signals with single-molecule contributions to be acquired simultaneously under ambient conditions, which is not achievable using conventional techniques (for example, ultrahigh-vacuum-based inelastic electron tunnelling spectroscopy). The conducting molecular junctions had a Raman signature that was distinct from the broken molecular junctions. Such a strong bonding interaction between the molecule and the drain may account for the nonlinear dependence of the conductance on the bias voltage.
Owing to its high spatial resolution under ambient conditions, TERS can also be applied to reveal conformational changes in biomaterials: for example, the secondary structure (that is, of the α-helix and β-sheet) on fibril surfaces with a lateral resolution better than 2 nm (Ref. 119). The commercialization of TERS instruments and the fast development of nanoscale infrared spectroscopy make it very likely that nanoscale vibrational spectroscopy will have an important role in future nanoscale chemical analysis.
To analyse complex chemical reactions at surfaces, it is necessary to use PERS methods under in situ conditions, whereby, for example, they operate under an applied voltage in the case of electrochemical systems and under light illumination in the case of photocatalytic systems. Electrochemical TERS is a particularly promising recent development in this area, because it can easily tune the electronic state of the electrode surface, with the laser focused in the gap between the tip and electrode surface. This offers a flexible way to control and characterize the surface reaction at the nanometre scale120.
Applications of SHINERS. Single-crystal surfaces with atomically flat and well-defined surface states are the model systems for many fundamental studies (for example, heterogeneous catalysis). To circumvent the previously discussed issue of interference in contact-mode SERS, SHINERS has been successfully used to study the relationship between surface structures of materials and their reactivity. By simply casting Au nanoparticles coated with SiO2 (Au@SiO2) at Au(hkl) electrodes, surface Raman signals have been obtained during the electrooxidation of a single-crystal Au surface51 (Fig. 5c). The Au–OH bending mode (δAuOH) at ∼790 cm−1 was detected at the beginning of the Au oxidation process and diminished after complete oxidation of the Au surface. This finding provided the first direct insight into the chemical nature of the intermediate species, which had long been under speculation. Moreover, the intensities of the bending mode δAuOH at three single-crystal facets increased in the order of Au(100) < Au(110) << Au(111), which revealed that the surface coverage of hydroxide species is very sensitive to the crystal orientation of the surface. Interestingly, this order is the opposite to the order of activity in the oxygen reduction reaction under the same conditions. This may be because hydroxide ions formed during Au reduction retard the oxygen reduction reaction. In situ electrochemical SHINERS offers a unique opportunity for a real-time investigation of electro- and photocatalytic reactions, as well as corrosion inhibitors121 at structurally well-defined surfaces.
The formation of interfacial species in Li-ion batteries strongly influences the battery performance. SHINERS has also been used to monitor interfacial processes in Li-ion batteries in situ49. In this study, the SHINERS signals were identified to relate to the formation of a solid electrolyte interphase on the Si electrode before reduction of ethylene carbonate during charge–discharge cycling (Fig. 5d). In contrast to the normal Raman signal from the bulk Si electrode, the SHINERS signal revealed different amorphization rates at the surface and bulk of the electrode. SHINERS has further been applied to directly observe the formation of Li2O during electrochemical cycling on a Li-rich cathode. Formation of Li2O on the cathode leads to the generation of H2O and LiOH, as well as changes in the electrolyte, which eventually results in a deterioration in performance50. Thus, SHINERS offers a unique method to deepen our understanding of surface reactions that heavily influence the overall battery performance.
Silicon is currently the most important semiconductor material in the electronic industry. SHINERS can be used to analyse the surface functional groups by simply drop coating shell-isolated nanoparticles at atomically flat Si surfaces15,122. A strong Si−H band at about 2,149 cm−1 was observed15 when the surface was treated with HF solution, which disappeared after O2 plasma treatment. This example demonstrates the great potential that SHINERS holds in unravelling interfacial phenomena that occur during industrial processing of semiconductors, as well as in various chemical or electrochemical processes. This technique has also been used to characterize the structures of living bacteria (for example, the yeast cell wall) and to detect pesticide residues on fruits and vegetables for food safety.
To further expand the applications of PERS in materials science, it is necessary to improve the enhancement and spatial resolution. Progress in materials science — especially in nanomaterials — has aided the development of PERS, which, in turn, has greatly contributed to our understanding in many branches of materials science. However, this mutually beneficial relationship is still in its infancy. Below, we detail some of the directions that this field will take in the near future, the materials that will enable this and the materials systems that will benefit from these advances.
New plasmonic materials for a broader wavelength range.
The success of PERS relies on the development of new nanostructures in the visible
and near-infrared region. Several alternative plasmonic materials beyond coinage
metals have been predicted or even developed, including metal nitrides, silicides,
borides and germanides for visible plasmonics123,
Other new plasmonic materials have been explored that might enable PERS to be carried out in different wavelength regions. Some examples (also listed in Table 1) include Al, Ga, In, Sn, Tl, Pb and Bi for ultraviolet plasmonics128; transparent conductive oxides, including indium tin oxide, Al:ZnO and Ga:ZnO; n-type semiconductors, including n-GaAs, n-InP and n-Si for near-infrared plasmonics; III–V or IV–IV semiconductors for mid-infrared plasmonics129; and Sn:In2O3 and graphene for terahertz plasmonics130,131. These alternative plasmonic materials will pave the way for diverse applications, including plasmon-enhanced spectroscopy, localized SPR sensing, superlenses and negative-index metamaterials123.
New nanostructures for third-generation PERS hotspots. Electromagnetic coupling in hybrid structures should be carefully considered when designing PERS-active nanomaterials for materials analysis. In practice, the electromagnetic coupling in hybrid structures of Au or Ag nanoparticle clusters on metallic substrates can be very strong if the particle–substrate nanogap can be well controlled, as is the case with the shell-isolated strategy. But for hybrid structures with Au or Ag nanoparticle clusters on dielectric material substrates with a low refractive index, such as SiO2 and biomembranes, only a small proportion of hotspots is located at the material surface.
One possible way to overcome this problem is to develop Au and Ag nanocrystals with sharp corners (for example, nanocubes, nanobars or concave nanocrystals) that support stronger intraparticle electromagnetic coupling101,132,133. It is also possible to create spoof surface plasmons using prism-coupling attenuated-total-reflectance based on periodic metal grooves that are fabricated on thin Au or Ag films. These nanostructures can support hotspots in air or solution.
Hierarchical micro- and nanostructures represent a new type of PERS-active substrate29,31,88,93,134,135. By means of subwavelength and/or deep-subwavelength electromagnetic coupling, hierarchical structures could effectively bridge the gap between the several-wavelength scale in the far field and the several-nanometre scale of the hotspots in the near field. In particular, it is necessary to develop hierarchical structures for plasmon-enhanced spectroscopies working in infrared regions in which the gap between the length scales is one to two orders of magnitude larger than that in the UV and visible regions. It is possible that hierarchical photonic–plasmonic devices could be designed to imbue TERS with far greater versatility. To fabricate new multiscale plasmonic materials, it is essential to understand the mechanism of the multiscale electromagnetic coupling.
To maximize the local electromagnetic field, it is possible to design nanostructures
that support plasmonic Fano-resonance or double resonance due to near-field or
far-field electromagnetic interference between nanostructures35,136. Some Fano-resonance nanostructures are well suited for
PERS applications because they significantly enhance the local electromagnetic field
at one wavelength and enhance the emission efficiency at another
The shell-isolated strategy can improve the stability of nanoparticles and target systems if the ultrathin shell materials are chemically stable and inert. However, the commonly used SiO2 dissolves in alkaline media, which limits the applications of SHINERS. Besides, the shell SiO2 is typically porous, which restricts the large-scale preparation of pinhole-free shell-isolated nanoparticles with shell thickness down to 2 nm. Therefore, it is highly desirable to develop other shell materials that can be easily prepared. It is now possible to develop ultrathin and compact Au@MnO2 and Au@Ag2S nanoparticles in alkaline systems141,142. The thinnest shell materials are 2D materials, such as graphene143 and hexagonal boron nitride (h-BN), with which the SHINERS sensitivity could be further increased by one order of magnitude; however, it is highly challenging to avoid creating pinholes in these materials. Although the PERS activity of Ag@shell nanoparticles is typically one to two orders of magnitude larger than that of Au@shell nanoparticles, it is still challenging to prepare pinhole-free, shell-isolated Ag nanoparticles with shell thickness down to 2 nm (Ref. 144). In addition, the shell-isolated strategy can also be extended to TERS by preparing a shell-isolated Au or Ag tip of a scanning probe microscope.
New materials for other spectroscopic methods. Beyond
PERS, there are other plasmon-enhanced spectroscopies, such as plasmon-enhanced
infrared absorption spectroscopy by means of mid-infrared plasmonic materials,
tip-enhanced infrared or terahertz spectroscopies145,
To develop PE-NLS methods, it is necessary to design corresponding plasmonic nanomaterials and nanostructures. For example, surface-enhanced CARS has been used to achieve single-molecule detection on well-designed nanostructures with nanoquadrumer discs139. The nanoquadrumer discs support three localized SPR peaks, which simultaneously match the three beams (two incident beams at different wavelengths and one anti-Stokes Raman-scattered beam) in CARS. In the development of plasmon-enhanced spectroscopy methods, research has focused on second-generation hotspots for trace-molecule detection. Further desirable developments would be to create third-generation PE-NLS hotspots that allow for energy and phase matching at the nanoscale, in order to improve the low enhancement effect that currently hinders these methods153.
Owing largely to the tremendous progress made in the development of new plasmonic materials and nanostructures, we anticipate a growing number of applications of PERS in the near future, ranging from in operando characterization to the on-line inspection of materials processing.
This work is financially supported by the National Natural Science Foundation of China (91427304, 21533006, 21321062 and 21522508) and Ministry of Science and Technology (2015CB932300). The authors thank X-Y. Cao for the English editing of this manuscript.