Surface-enhanced Raman spectroscopy: substrate-related issues
- First Online:
- Cite this article as:
- Lin, X., Cui, Y., Xu, Y. et al. Anal Bioanal Chem (2009) 394: 1729. doi:10.1007/s00216-009-2761-5
After over 30 years of development, surface-enhanced Raman spectroscopy (SERS) is now facing a very important stage in its history. The explosive development of nanoscience and nanotechnology has assisted the rapid development of SERS, especially during the last 5 years. Further development of surface-enhanced Raman spectroscopy is mainly limited by the reproducible preparation of clean and highly surface enhanced Raman scattering (SERS) active substrates. This review deals with some substrate-related issues. Various methods will be introduced for preparing SERS substrates of Ag and Au for analytical purposes, from SERS substrates prepared by electrochemical or vacuum methods, to well-dispersed Au or Ag nanoparticle sols, to nanoparticle thin film substrates, and finally to ordered nanostructured substrates. Emphasis is placed on the analysis of the advantages and weaknesses of different methods in preparing SERS substrates. Closely related to the application of SERS in the analysis of trace sample and unknown systems, the existing cleaning methods for SERS substrates are analyzed and a combined chemical adsorption and electrochemical oxidation method is proposed to eliminate the interference of contaminants. A defocusing method is proposed to deal with the laser-induced sample decomposition problem frequently met in SERS measurement to obtain strong signals. The existing methods to estimate the surface enhancement factor, a criterion to characterize the SERS activity of a substrate, are analyzed and some guidelines are proposed to obtain the correct enhancement factor.
KeywordsSilverGoldSubstrate preparationSubstrate cleaningPhotodecompositionSurface enhancement factor
Raman spectroscopy is capable of obtaining fingerprint information of species by detecting the vibrational bands. However, the Raman signal of most systems is very weak and is only about 10-10 times the intensity of the incident laser [1, 2]. The signal becomes even weaker when the adsorbed species are concerned, because there are only monolayer or submonolayer species on the surface. In Raman measurements other than in a pure liquid or solid, a certain kind of enhancement effect should be employed to boost the signal, such as the resonance Raman effect, the surface enhanced Raman scattering (SERS) effect , or the tip-enhanced Raman effect . The first observation of surface-enhanced Raman spectra was in 1974, when Fleischmann et al.  reported the first high-quality Raman spectra of monolayer-adsorbed pyridine on an electrochemically roughened Ag electrode surface. Three years later, the Van Duyne group  and the Creighton group  independently pointed out that the million-fold enhancement is due to a kind of surface enhancement effect. These exciting early observations have stimulated intensive experimental and theoretical works and SERS has found wide application in surface sciences, analytical sciences, biological sciences, etc. [8–10].
Meanwhile, the investigation of the SERS mechanisms has never stopped. It is now widely accepted that electromagnetic enhancement and chemical enhancement are the two major enhancement mechanisms contributing to the giant SERS, with the former being the dominant contribution [11–13]. The chemical enhancement mechanism reflects the enhancement as a result of the chemical interaction between the adsorbates and the metal surface. Among the various types of chemical enhancements, including chemical bonding enhancement, resonance enhancement of a surface complex, and photon-induced charge-transfer enhancement (PICT) [14, 15], PICT is the most important. It describes the optical resonance excitation of the charge-transfer state formed by the surface complex of the metal adatom and the adsorbate [16–18]. Chemical enhancement is a short-range effect occurring in the range of molecular scale. Electromagnetic enhancement comes from the interaction of light (both incident and scattered) with the substrate. It is a long-range effect not depending on the properties of the probe molecules. Localized surface plasmon resonance (LSPR), the lightning rod effect, and the image field effect have all been considered to contribute to SERS [19, 20]. Among them, LSPR makes the major contribution to the electromagnetic field enhancement and therefore SERS. The LSPR strength and frequency are influenced by the incident laser wavelength, the morphology of the substrate, and the surrounding medium. By controlling the composition, shape, size, and the interparticle spacing of nanoparticles and their assemblies, one can tune the LSPR to obtain the optimized SERS substrate at the desired wavelength [21, 22].
However, whether SERS can achieve a broader application really depends on the SERS activity and the reproducibility of the substrate . Ever since the first observation of SERS on an electrochemically roughened Ag electrode, the preparation methods of SERS substrates have gone through the following stages: (1) random and nonuniform substrates prepared by electrochemical oxidation and reduction cycles(s) (EC-ORC) or vacuum deposition methods; (2) nanoparticle sols with a large size distribution prepared by wet chemical synthesis or a laser ablation method; (3) nanoparticles with controlled size and shape prepared by a chemical synthesis method; and (4) large-area surface nanostructures with defined size, shape, and interparticle spacing prepared by self-assembly, template, or lithography methods. Especially, in the last 15 years, the preparation of SERS substrates has become more controllable, benefiting from the development of nanoscience and nanotechnology. As a result, SERS has found a further application in quantitative analysis [24–26], trace analysis [27–29], and the analysis of bio-related systems [30–32].
The substrate should have high SERS activity and therefore provide high sensitivity. By controlling the size (more than 50 nm) and interparticle spacing (less than 10 nm) of nanoparticles, one can tune the LSPR frequency of the substrate to match the incident laser frequency and the effective coupling between nanoparticles can be induced to maximize the enhancement.
The substrate should be uniform so that the deviation in enhancement over the whole surface can be less than 20%, which requires a relatively ordered arrangement of the nanoparticles on the substrate.
The substrate should have good stability and reproducibility. Even after a long shelf time, the enhancement effect can still be maintained. The deviation in the enhancement should be less than 20% for different batches of substrates prepared by the same method.
The substrate should be clean enough so that it can be applied to study not only strong adsorbates but also some weak adsorbates or even unknown samples.
Unfortunately, at present, it is still difficult to obtain SERS substrates that can simultaneously meet all of the above requirements. According to the specific application purpose, one has to make some trade-offs. For example, in quantitative analysis, a uniform and reproducible substrate is extremely important; however, in trace analysis, the maximized enhancement is the prerequisite. In bio-related detection, a clean and highly enhanced substrate is generally required owing to the complexity in biosystems studied to allow for a reasonable assignment of the detected spectral bands.
In the present critical review, we will firstly introduce various methods to prepare SERS substrates. We have introduced this aspect in several of our previous reviews [33, 34], and the focus of present review will be on the preparation of metal nanoparticles and highly ordered SERS substrates and the advantages and challenges when they are used as SERS substrates. We will then deal with the issue of contaminants frequently met in SERS, especially in SERS analysis of trace sample and unknown systems. A method will be proposed to handle the laser-induced decomposition problem frequently met in SERS measurement with a confocal micro Raman instrument. Finally, we will show how to correctly calculate the surface enhancement factor.
Methods for preparing SERS substrates
EC-ORC and vacuum deposition methods
The first SERS spectra were obtained on a rough Ag surface prepared by EC-ORC . By application of an oxidation potential to the metal electrode, the electrode will be oxidized to soluble compounds or will form a surface complex. A reduction potential will then reduce these dissolved species on the surface, forming surface nanostructures. To ensure reproducibility, the metal surfaces should first be mechanically polished, ultrasonically cleaned to obtain a clean surface, and then roughened in certain electrolytes (KCl or H2SO4) with a special waveform of potential or current control to allow the oxidation and reduction of the surface . Using this method, we can easily obtain SERS-active Ag and Cu surfaces. For the Au surface, special care has to be taken regarding the waveform applied and the cleanliness of the surface . The common enhancement achieved using this method can be as high as 104–106 for Ag, Au, and Cu, and 101–104 for other transition metal substrates, such as Fe, Co, Ni, Rh, Pd, and Pt . EC-ORC is a very simple method for preparing a SERS substrate with very good stability. It has less chance of being contaminated. With such a conductive substrate, potential-dependent SERS measurement can be performed to study the SERS mechanism, especially the chemical enhancement mechanism. Such a substrate has a random surface structure formed by nanoparticles with a wide size and shape distribution. Therefore, different areas of the surface may have quite different enhancements. As common practice, one has to check the surface to find the most representative spots for a series of Raman measurements to ensure the reproducibility of the SERS experiment [33, 36]. To obtain a higher enhancement on the transition metal surface for studying the surface adsorbates, a very thin layer (several atomic layers to several tens of atomic layers) of transition metals (Pt group metal) has been deposited on highly SERS active Ag or Au surfaces by electrochemical methods to utilize the long-range effect of SERS [37, 38]. The method has been used to study the adsorption and reaction of molecules on the transition metal surfaces and has provided a wealth of meaningful data and conclusions for understanding the electrochemical interfaces and the electrocatalytic processes . It should be pointed out that the influence of the substrate metal on the physical and chemical properties of overlayer metals and the stability and the reversibility of the thin layer still limit a wider application of the method. To overcome these problems, our group used core–shell nanoparticle methods to obtain SERS substrates of transition metals with a high enhancement, which will be discussed in detail later. In parallel to EC-ORC, those laboratories equipped with a vacuum deposition system tend to use vacuum deposition methods to prepare SERS substrates. The main purpose is to form a rough metal film or discontinuous metal islands in the vacuum chamber by depositing metal atoms on the substrates, which can be a Si wafer, a glass slide, graphite, or a mechanically polished rough substrate or even nanostructured substrates [40–44]. By controlling the temperature of the substrate and the deposition rate during the deposition process, one can obtain metal films with different SERS activity. For example, if the metal vapor is deposited on a cold substrate (typically lower than 120 K), a rough metal film will be obtained. With increase of the temperature, the migration of the metal atoms will be accelerated, leading to the formation of an island-like structure.
Both EC-ORC and vacuum deposition methods have difficultly to produce very uniform structures; therefore, different points may have different enhancement effects. Furthermore, both methods need specialized equipment, such as a potentiostat or a vacuum deposition system. Which of the two methods is selected really depends on the instrumentation available in the laboratory. To overcome these problems, nanoassembly or nanolithography methods have been used to fabricate ordered substrates, followed by vacuum deposition or electrochemical methods to produce comparably ordered substrates, which will be discussed further later.
Chemical synthesis for producing well-dispersed nanoparticles
Since the first application of Ag and Au sols as SERS substrates in 1979 by Creighton et al. , Au and Ag sols have been widely used as SERS substrates owing to their strong SERS activity. With the fast development of nanoscience and nanotechnology in the mid-1990s, the control of the shape, size, and composition of Ag and Au nanoparticles has become quite sophisticated. As a result, Ag and Au sols have become the most widely used SERS substrates. Theoretical calculations indicate that the Raman enhancement of a single Au nanoparticle is about 103–104, and that for a single Ag nanoparticle can reach about 106–107 [53, 54]. Experimental results obtained with 647-nm excitation indicate that Ag nanoparticles with a diameter of about 200 nm and Au nanoparticles with a diameter of about 60 nm show the highest enhancement [55, 56]. Although spherical Au nanoparticles show surface plasmon resonance absorption at about 530 nm, the efficiency of inducing localized electromagnetic field enhancement is not high owing to the extension of the interband transition of Au into this wavelength region; therefore, reasonably good enhancement can only be obtained by using red excitation, typically 632.8 or 647 nm. Theoretical results have shown that metal nanoparticles with a sharp edge and angle will give an extra enhancement owing to the “lightning rod effect” [57, 58]. Experimental results also indicate that Au nanocubes or nanotriangles have a strong enhancement effect over nanospheres, among which the enhancement effect of nanocubes can be as high as 109 [59, 60], although in this finding one should not neglect the coupling effect of a Au substrate on enhancement. Most of the above-mentioned nanoparticles show surface plasmon resonance absorption in the visible region; however, to apply SERS to study biological samples or cells, it is better to choose excitation light in the near-infrared region. Owing to the weak absorption of the near-infrared light by tissue and the lower energy of the light, it can penetrate deeper into the tissue and effectively avoid the interference of fluorescence. As a result, there has been special interest in preparing SERS substrates with LSPR absorption in the near-infrared region in recent years . Among various type of nanoparticles, Au nanorods have a very unique absorption in the near-infrared region. Different from the single absorption band of single spherical nanoparticles, nanorods show transverse and longitudinal LSPR modes due to the geometric asymmetry. The former occurs in the short-wavelength region and the latter occurs in the long-wavelength region, and usually the latter shows a stronger absorption than the former. The longitudinal band will redshift and increase in intensity with increasing aspect ratio, which dominates the optical properties of nanoparticles . For example, Au nanorods with an aspect ratio of 4 show an absorption band at about 800 nm. They are ideal substrates for 785-nm excitation and have been used for cancer cell detection . However, owing to the isotropic feature of nanorods, the enhancement obtained from them may also depend very much on the direction of polarization of the light and the long axis of the nanorods, which may lead to some complications in using them as substrates.
Compared with the SERS effect of Au and Ag nanoparticles, that of transition metal nanoparticles is relatively weak. Transition metal nanoparticles are normally used for studying the electrocatalytic and corrosion processes of transition metals. The pioneering work on using transition metal nanoparticles as SERS substrates dates back to 1984, when Parker et al.  synthesized Rh sols and used them to detect the SERS of pyridine. It was not until 1997, after the demonstration that there is a SERS effect on transition metals [33, 63], that SERS was also observed on Pd , Fe , and Pt [66, 67] nanoparticles. Owing to the difficulty in synthesizing large transition metal nanoparticles, most of the previous studies were performed using small nanoparticles (less than 20 nm in diameter), and as a result the enhancement was usually weak. More recently, with use of a special capping agent, Pt nanocubes and Pd nanotriangles have been synthesized and used as SERS substrates. Both of them show higher enhancement than the respective spherical nanoparticles, possibly owing to the special effect of the sharp edge [68, 69]. To overcome the weak SERS activity of transition metal nanoparticles, we further coated the Au core with an optimized SERS activity with a very thin layer (one to ten atomic layers) of a transition metal shell (Pt, Pd, Ni, Co, etc.) and obtained Au core–transition metal shell nanoparticles. With the assistance of the long-range effect of the SERS of the Au core, the SERS of molecules adsorbed on transition metal shells can be significantly improved. The enhancement can be as high as 104–105 [70, 71]. Such core–shell nanoparticles have been used for studying the electrochemical process and interfacial structure [70–72].
Both experimental and theoretical results indicate that a stronger enhancement effect will be produced when the single nanoparticles form aggregates of two or multiple nanoparticles owing to the coupling of the electromagnetic field. For instance, in single-molecule SERS, enhancement as high as 1014 has been claimed [73, 74]. A high concentration of probe molecules or some salts (such as NaCl) will help to form aggregates in the sol and improve the SERS activity [51, 75]. It should be pointed out that in the multiple-nanoparticle system, the SERS enhancement will not only be influenced by the size, shape, and interparticle spacing, but also by the polarization of the laser. In such a system, the interparticle spacing and arrangement is random. Therefore, the experimentally obtained signal is the averaged spectra contributed by systems of different enhancements, which will lead to a poor reproducibility of the spectral intensity in the multiple-nanoparticle coupling system. Especially, the sol is usually stabilized by the surface charge or the steric hindrance of polymeric molecules on the nanoparticle surfaces. One would expect chemical degradation or aggregation of the nanoparticles in the sol owing to the change in surface charge or the desorption or decomposition of polymeric molecules. Furthermore, after the addition of sample molecules and salt, the nanoparticles may form precipitates after the formation of aggregates, which may lead to a significant decrease of the SERS signal in a few minutes . Therefore, to obtain signals with very high reproducibility, it is better to use highly ordered substrates; these will be discussed later.
Highly ordered SERS substrates
The methods to prepare highly ordered SERS substrates include the nanoparticle assembly method, the Langmuir–Blodgett (LB) method, the template method, and nanolithography and nanoimprint methods. In the following part, the preparation method and the advantages and disadvantages will be analyzed for different methods.
Chemical assembly method
An alternative method is to functionalize the Au nanoparticle surface with CTAB to form a positively charged molecular bilayer. Then a droplet of the functionalized nanoparticles is dispersed on a precleaned ITO glass surface to form a very uniform layer of Au nanoparticles, with the interparticle spacing controlled by the length of the CTAB molecule (approximately 8 nm). Owing to the positive charge of the bilayer surrounding the nanoparticles, the nanoparticles will not form aggregates on the surface during the process of evaporation of the solvent. The small interparticle spacing enhances coupling between nanoparticles, which will lead to a shift of the LSPR of the substrate to the near-infrared region and a 108 enhancement . The substrate has a very uniform SERS signal and good reproducibility and stability. It should be pointed out that even for a strongly adsorbed molecule such as p-mercaptoaniline, the signal of CTAB can still be detected. This will limit to some extent its application for detecting molecules with a weak interaction with the substrate or a weak Raman signal. The problem can be solved by a kind of surface cleaning method, as will be discussed later. Although this method can effectively avoid the aggregation of nanoparticles, it is still difficult to obtain a defect-free surface and to obtain a substrate with a large area. The LB technique can potentially solve this problem.
Anodic aluminum oxide template method
Nanolithography and nanoimprint
The ultimate method to obtain a highly ordered array is to use the top-down nanolithography and related nanoimprint methods. Typically, a layer of polymeric photoresist (positive or negative) is cast on the solid substrate (such as Si, glass, or Au film), followed by patterning with ultraviolet light, an electron beam, or a focused ion beam directly on the photoresist surface or indirectly with the assistance of a mold. After exposure and development, the remaining photoresist can be used as a mold, on which SERS-active metals are deposited by vacuum physical vapor deposition. After the mold has been lifted off, a highly ordered nanostructured SERS-active substrate with a structure identical or complementary to that of the mold is formed [100, 101]. In the case of a Au film substrate, the resist masks obtained can also be used as stencils for ion beam etching of the exposed Au substrate. Then, the residual polymeric photoresist can be dissolved to expose the SERS structures . The nanolithography method allows the preparation of highly ordered and uniform SERS substrates with nanostructures having a wide diversity of shapes and geometries compared with nanosphere lithography. However, it is still a challenge to routinely obtain a spacing of less than 10 nm to allow the maximized electromagnetic coupling. The use of focused ion beam or electron beam lithography to make a SERS substrate of large area is still time-consuming and the cost is too high for commercialization. Only those laboratories having the facilities are able to use them to fabricate SERS substrates for mechanistic study. Owing to the above-mentioned facts, these two methods will not be widely used because of the low efficiency and high cost. However, if the two methods are used to make molds to be used in nanoimprint, batch production will be possible. Nanoimprint normally directly writes on the Si or quartz slide with an electron beam to produce the desired nanopattern as the mold. The mold is then aligned and pressed into the photoresist covering on the substrate. After curing, the mold is lifted off and the substrate is deposited with the desired metal. After removal of the photoresist, a highly ordered nanostructure with SERS activity can be obtained . Compared with nanolithography, nanoimprint has a much higher efficiency and lower cost. However, the problems associated with the mold, such as the weak coupling and low enhancement effect, still exist. Although there have been attempts to obtain the rough nanoarrays by changing the deposition conditions or posttreatment to increase the SERS activity of the fabricated nanostructures , such kinds of treatment will also lead to the problem of nonuniformity.
SERS substrate cleaning
If the target molecule has a strong interaction with the nanoparticle surface and the signal of the target molecule is strong, then the target molecule will probably replace the contaminants on the surface. In this case, the interference of the contaminants can be neglected.
If the interaction of the target molecule with the nanoparticle surface is weak or the signal is weak, then it is difficult for the target molecules to occupy the surface active site. Meanwhile, the existing signal of the contaminants will strongly interfere with the signal of the target molecules.
Especially when the substrate is used for detecting species with low concentration or biological samples, the requirement of a clean SERS substrate will be even higher because of the low surface coverage and the complexity of the biological environment.
An approach to reliable SERS measurement by the defocusing method
A common experience of a SERS experimentalist with a micro Raman system is that the SERS signal will keep changing with time with the extended illumination under a laser power on the order of milliwatts, which is considered a very low laser power in the conventional Raman measurement. The main reason is due to the use of a microscope objective with a large numerical aperture (NA), through which the size of the laser spot on the sample becomes much smaller than that in the conventional macro Raman system . With the same laser power on the sample, the laser power density of a confocal Raman microscope is typically about 3 orders of magnitude higher than that in the conventional Raman instruments. Such a power density may not be a problem for the study of nonabsorbing or very stable molecules or crystals; however, it will be a disaster for some organic molecules or biomolecules [114, 115]. Common strategies to reduce this effect are to lower the laser power or to use a movable sample [2, 18, 116].
Moving the sample requires a specially designed stable moving stage and will sometimes disturb the systems to be studied; therefore, the most frequently used method is to reduce the laser power. As we know, the Raman intensity is linearly proportional to the laser power; therefore, a decrease in the laser power will also lead to a proportional decrease of the Raman intensity . More recently, there has been increasing use of the line focusing method to expand the point focus to a line focus to match the slit. This method can significantly decrease the laser power density but retains a high signal collection efficiency . Its application is still limited by the optical design in that not all instruments can be installed with a line focusing lens.
Evaluation of enhancement effect of SERS—calculation of surface enhancement factor
If the coverage of the molecule is less than a monolayer, the ASEF will be underestimated. Nsurf can be estimated more accurately by limiting the number of molecules to be slightly less than the number in monolayer during adsorption.
The ASEF can be obtained after substituting all the known data and constants into Eq. 5. The roughness factor R can be estimated by calculating the surface area using a simple sphere model for monolayer-dispersed nanoparticles. However, for multilayer nanoparticles or massive electrode surfaces, it is better to use electrochemical methods, such as differential capacitance or cyclic-voltammetric methods to estimate R. σ is generally obtained from the literature from previous studies and can also be estimated from known atom sizes and bond distances. The latter may often lead to an underestimation of σ and the ASEF value [12, 13, 43–48]. The advantage of using this method is that the value is not directly related to the area of the laser spot, as the area can be removed by taking the ratio, eliminating possible error. If the solution can be confined to a very thin layer with known thickness of less than 10 μm, we can still use Eq. 5. But the calculation becomes simple by substituting the h value with the known thickness of the solution layer.
It is advised to make the thickness less than 10 μm for a transparent liquid sample to avoid the overestimation of the numbers. For the solid sample, the film thickness can be prepared to less than 1 μm to avoid the interference due to the absorption of the solid sample. The film thickness can be calibrated by optical methods or by scanning electron microscopy. It is recommended that a SERS laboratory prepare such a type of standard sample to be used as a reference standard for calculating the surface enhancement factor. Only when there is an accepted method to be used by different groups can the surface enhancement effects of different substrates be compared.
Different methods have been introduced to fabricate SERS substrates, including EC-ORC, vacuum deposition methods, well-dispersed nanoparticle sols, nanoparticle aggregate film surfaces, and well-ordered substrates. The advantages and disadvantages of different substrates in terms of application have been analyzed and proper fabrication methods should be selected for a special purpose of SERS. The well-dispersed nanoparticles show a weak but a homogenous enhancement. The spectral reproducibility of the nanoparticle aggregates or an electrochemically fabricated substrate is bad but very strong enhancement can be achieved by controlling the experimental conditions. Highly ordered substrates show excellent surface uniformity, but with a lower enhancement effect than nanoparticle aggregates. In short, each method has its advantages and weaknesses. Therefore, one should choose the right method to obtain a SERS-active substrate for a specific application. Well-dispersed nanoparticles and well-ordered substrates can be used for quantitative analysis. The nanoparticle aggregates can be effectively used for qualitative analysis with extremely high sensitivity.
Different methods to obtain clean SERS substrates have been analyzed and a method of iodide adsorption to repel the surface contaminants followed by the electrochemical oxidation method of iodide has been proposed. The method has been demonstrated to be useful for SERS detection of weak adsorbates and for trace analysis.
A defocusing method was proposed to effectively eliminate the laser-induced photodecomposition of surface adsorbate at the high laser power density of a micro Raman system. A 20-μm deviation of the laser focus from the sample surface can effectively eliminate the decomposition evidenced by the spectra free of carbonaceous background. Whether the substrate should be cleaned and how to clean it really depends on the systems to be measured. It is a great challenge to obtain a clean substrate with high enhancement, good uniformity, high reproducibility, and high stability.
Some existing problems in calculating the surface enhancement factor were discussed and a method to calculate the ASEF was introduced to allow the SERS activities of substrates prepared by different methods to be compared.
It is believed that the rapid development of nanoscience and nanotechnology will accelerate the process toward this goal, which will eventually allow a complete understanding of the SERS effect and a wide application of SERS in analytical science and biomedical sciences.
This work was supported by the National Basic Research Program of China (973 Program nos. 2009CB930703, 2007CB935603 and 2007DFC40440), the Natural Science Foundation of China (20673086, 20620130427, 20825313, and 20827003), and the Ministry of Education of China (NCET-05-0564).