Chapter 1: Raman Spectroscopic Sensing in Food Safety and Quality Analysis

Photons, which make up light, may interact with a molecule when the incident light is directed onto the molecule. The photons could be adsorbed by the molecule, if the photon energy matches the energy gap between the ground state and the excited state of the molecule. The molecule is promoted to a higher energy excited state after the adsorption of photon energy.^1,2 

On the other hand, if the photon energy does not match the differences between two energy levels of a molecule, the photon can also interact with the molecule and be scattered from it.³  For example, when a single frequency of light is used to irradiate samples, the photons interact with the molecules and polarize the electron clouds near the nuclei to form a virtual state for this molecule. The lifetime of this virtual state is very short, leading to the re-radiation of the photons. When the energy of a re-radiated photon is equal to the energy of an incident photon, the scattering is called “Rayleigh scattering”, which is the dominant scattering and does not involve any energy transfer. When the molecule in the ground state is promoted to the virtual state and then returns to the excited state, the scattering is called “Stokes scattering”. This process involves energy transfer from photons to the molecule.¹  Likewise, when a molecule in the excited state is promoted to the virtual state and then returns to the ground state, the scattering is called “anti-Stokes scattering”. This scattering process transfers the energy from the molecule to photons. Raman scattering is the inelastic scattering of a photon, including both Stokes and anti-Stokes scattering. However, Raman spectroscopy usually records Stokes scattering because anti-Stokes scattering is very weak compared to Stokes scattering. For example, Coherent Anti-Stokes Raman Spectroscopy (CARS) collects anti-Stokes Raman scattering. Due to the weak scattering intensity, it has not been widely used for routine analysis (Figure 1.1).

By recording the Raman scattering and calculating the energy loss of incident photons, Raman spectroscopy provides detailed vibrational information for the molecules in a sample.⁴  The Raman spectra are like “fingerprints” to the molecules, which are very useful for the identification of molecules. However, as stated above, the dominant scattering of the photon–molecule interaction is Rayleigh scattering. Raman scattering is a very weak effect. Therefore, many methods have been used to enhance the signals of Raman spectroscopy, resulting in a new term called “surface enhanced Raman spectroscopy” (SERS).

The Raman spectrometer consists of three major components: an excitation source, a sampling apparatus and a detector. Modern Raman spectrometers use a laser as the excitation source, a spectrometer for the detector and either a microscope or a fibre-optic probe for the sampling apparatus. The laser is very important in Raman spectrometry because it directly influences the Raman response from the targeted molecules.

A laser is a device that emits light through a process of light amplification due to the stimulated emission of electromagnetic radiation. The word “laser” is an acronym for Light Amplification by Stimulated Emission of Radiation. A laser device consists of a gain medium, an energy supply and an optical cavity. The gain medium is used to amplify light by stimulated emission. The energy is usually supplied by electrical current. The optical cavity is used to amplify the light. It is made by a pair of mirrors. The light could bounce between the two mirrors and pass the gain medium, where the light is amplified each time. One of the mirrors has a small transparent area, which allows the laser to spread out and form a narrow beam. This light beam could be used for Raman spectroscopy.

Laser sources for Raman spectroscopy are divided into three categories, including ultra-violet laser, visible laser and near-infrared laser. The ultra-violet laser has a wavelength of 244 nm, 257 nm, 325 nm and 364 nm. The visible laser has a wavelength of 457 nm, 473 nm, 488 nm, 514 nm, 532 nm, 633 nm and 660 nm. The near-infrared laser has a wavelength of 785 nm, 830 nm, 980 nm and 1064 nm. For Raman spectroscopy, the most widely used lasers are the visible lasers with a wavelength of 532 and 633 nm, and the near-infrared laser with a wavelength of 785 nm.

A high-performance laser is required to obtain high-quality Raman spectra. It is also important to select a suitable excitation wavelength in order to get the best Raman response. The Raman intensity is positively proportional to λ⁻⁴, where λ represents the laser wavelength. Thus, the infrared laser (785 nm) will result in a lower Raman intensity compared to the visible laser (514 nm and 633 nm). Second, the spatial resolutions are different between different lasers. The laser spot can be calculated according to the equation that diameter=1.22 λ/NA. NA is the numerical aperture of the microscope. For example, when a 0.75/50 microscope lens is used, the laser spot has a diameter of 1.28 μm and 0.73 μm for a 785-nm laser and a 532-nm laser, respectively. Thus, the 514-nm laser has a smaller laser spot. Third, the laser performance also depends upon the target chemicals. The infrared laser (785 nm) is useful in suppressing the fluorescence for organic chemicals. The 514-nm and 633-nm lasers are useful for the resonance Raman experiments because most of the excitation wavelength of fluorescence is close to this region. Ultra-violet lasers are useful for the analysis of macromolecules, such as proteins, because the ultra-violet laser can be used to achieve the resonance for those biomolecules. Finally, when a laser is used for surface enhanced Raman spectroscopy (SERS), the laser performance depends upon the SERS substrate as well. The maximum SERS enhancement is achieved when the laser wavelength is close to the localized surface plasmon resonance (LSPR) of the substrate.

Surface enhanced Raman spectroscopy is a technique that can greatly enhance the Raman signals from molecules or bacteria adsorbed onto the surface of a substrate (Figure 1.2).^5,6  The enhancement factor can be as high as 10¹⁰. Traditional analytical methods, such as High Performance Liquid Chromatography (HPLC), Gas Chromatography (GC) and Enzyme-Linked Immunosorbent Assay (ELISA), are time consuming and labour intensive, and often require complicated sample pre-treatment.⁷  The running and maintenance costs for HPLC and GC are also very high. In contrast, SERS does not require complex sample pre-treatment and expensive solvent/column for operation. Furthermore, SERS only requires a few seconds for a single measurement, which is much faster than HPLC and GC. Due to its high sensitivity, unique spectroscopic fingerprinting features and non-destructive data acquisition, the SERS technique has extensive potential to be used for the detection and identification of chemical and microbiological contaminants in foods as well as food quality analysis.⁸ 

Unlike the HPLC profile, the SERS spectrum is highly dependent upon the SERS substrates and the equipment used. The SERS spectrum often has relatively big variations compared to other analytical methods. It is highly necessary to use statistical methods to analyze the SERS spectra and interpret them accurately. The most commonly used methods are Principal Component Analysis (PCA) and Partial Least Squares (PLS).

PCA is a statistical procedure that converts a set of correlated variables into a set of linearly uncorrelated variables called “principal components”.⁹  The number of principal components is smaller than the number of variables. By using an appropriate orthogonal transformation, a series of principal components can be identified. The first principal component must have the largest possible variance and the second principal component has the second largest possible variance. This analogy is applied to other principal components. In other words, PCA is a transformation method. This orthogonal transformation could transform the data to a new coordinate system. In this new coordinate system, the largest variance is projected onto the first coordinate and the second largest variance is projected onto the second coordinate. The PCA technique is used to find those coordinate systems and evaluate the contribution of each component, which means that it can be used to find the major contributing peaks for a specific molecule from the complex Raman spectra.

For example, in order to comprehensively study an object, we have to include a variety of factors that have influence on the object. However, not all of the factors contribute equally to the behaviour of this object. Some factors contribute more than other factors. The factors may also be related to each other. We hope we can find the internal structure of the data that can best explain the variance of these data. Let's assume there are a few factors named X₁, X₂, X₃… X_p. We hope we can find one equation as follows.

The equation will lead to the following equations (n≤p)

We hope that the variance of C₁ is the largest, the variance of C₂ is the second largest, and so on. The eigenvectors (a_j1, a_j2, a_j3, … a_jp) are orthogonal. In addition, any two of the variances (C₁, C₁, … , C_n) are not correlated. The sum of the variance of C equals the variance of X as in the following equation.

Then, what we have to do is to identify the matrix of a_ij. That is the correlated eigenvector matrix. We can solve the matrix A(a_ij) using the singular value decomposition method or other methods. After solving the matrix, we could calculate the variance of each component and the contribution of each component to the total variances. In this way, we can obtain the most important components and understand the data structure more effectively. The number of principal components is usually much smaller than the number of factors. PCA is a very useful technique in analyzing Raman data from different chemicals or microorganisms. For example, Raman spectroscopy was used to distinguish three types of pesticides in apples.¹²  These three pesticides have extensive similar peaks that are derived from the matrix. In order to differentiate the pesticides, PCA could be used to eliminate the matrix effect and find the most important peaks that contribute to the separation of these pesticides.

PLS is a technique that combines the features of PCA and multiple linear regressions. It is particularly useful to predict a set of dependent variables from a large set of independent variables. For example, it can be used in SERS when we test a series of concentrations of pesticides (dependent variables). Normally, a pesticide has a lot of characteristic peaks (independent variables), which are correlated to each other (proportional). By using PLS, we can build a model to effectively predict the pesticide concentrations by correlating with the intensities of the most important characteristic peaks. In general, PLS is used to find the fundamental relations between the Y matrix (Dependent variables) and the X matrix (Predictor variables) when the number of predictor variables is higher than the number of dependent variables.

To be specific, let's assume we have I observations for J predictors. Therefore, we will be able to have I observations of dependent variables K (K<J). The observations will be stored in the matrix of I×K (Y) and the predictors will be stored in the matrix of I×J (X), such as in the following two matrices.

The purpose of PLS is to predict Y by X and to find their common structures (1<K<J). PLS can find the components from X that are also relevant to Y. PLS searches a set of components (latent vectors) that can perform a simultaneous decomposition of X and Y under the condition that these components explain the maximum covariance between X and Y. Then, we perform the regression step where the decomposition of X is used to predict Y.

In the first step, the independent variables are decomposed as X=TP^T+E with TT^T=I, where T is the score matrix, P is the loading matrix, I is the identity matrix and E is the error matrix. In a similar way, Y is estimated as Ŷ=TBC^T+F, where B is the diagonal matrix with regression weights and F is the error matrix. The column of the T matrix is the latent vector. Then, the following step is regression. We have to find two sets of weights w and c in order to create a linear combination of the columns of X and Y for the purpose of maximizing the covariance of X and Y. That is to say that we have to find the first pair of vectors as the following equation shows:

After the first latent vector is identified, it is subtracted from X and Y. All the latent vectors are identified until X becomes an empty matrix.

The dependent variables are then predicted by multivariate regression using the equation of Ŷ=TBC^T=XB_PLS where B_PLS=P^T+BC.¹⁰  Finally, through cross-validation, we can identify the latent variables for the best generalization to predict new dependent variables.

The general detection scheme of SERS is illustrated in Figure 1.3. Food contaminants will be extracted by water/organic solvent and directly placed onto the commercial or custom-made SERS substrate. The Raman laser will focus on the substrate, which can significantly enhance the Raman signals. A variety of food contaminants, including pesticides, antibiotic residues and bacterial toxins, have been measured using a Raman spectrometer coupled with highly reproducible SERS substrates. SERS can provide a fast and cost-effective method for the identification of food contaminants while providing similar accuracy compared to HPLC and GC methods. The rapid SERS methods with the emergance of high-performance substrate will make in situ food safety monitoring possible in the food processing chain.

Chemical contaminants, such as illegal chemical additives, growth hormones and natural toxins, represent a major group of contaminants in foods. There are many routes for chemical contamination to enter the food supply chain. Some chemical contaminants are intentionally added into foods to boost the nutritional values or alter the organoleptic properties of foods. Natural toxins could occur because of inappropriate food storage conditions, for example, food mycotoxins. Growth hormones are added into the feeding material of animals to increase the growth rate. Illegal pesticides are used to control pests in food crops and increase the crop yield. For example, in 2006, Greenpeace tested some vegetables sold in two grocery stores in Hong Kong and discovered that 70% of tested samples contained pesticide residues. Thirty percent of tested vegetable samples exceeded safety levels and several samples were tested to be positive for illegal pesticides. The disappearance of illegal pesticides could also impose potential risks for drinking water. Chemical contamination has been posing serious risks to human health in recent years. Constant testing is one of the most effective ways to prevent contaminated foods entering consumer markets. The most employed testing methods are chromatographic-based detection methods, such as HPLC, HPLC-MS and GC-MS. However, with the fast development of Raman equipment and SERS substrates, the SERS technique presents a new opportunity for rapid food analysis.

Chemical contaminants, such as pesticides and melamine, are often small molecules. The small molecules could be extracted by solvents and directly placed onto SERS substrates for analysis. Depending upon the contaminants and the performance of SERS substrates, the detection limit of those chemical contaminants often falls into the range of parts per billion (ppb) or parts per million (ppm).

Since the outbreak of illness from melamine in 2007, a number of papers have been published on the detection of melamine by SERS methods. Melamine, a notorious chemical contaminant, was added to dairy products to boost their protein content. The detection of melamine by SERS has been proposed by many researchers. In early studies, commercial substrates were used to study the feasibility of the SERS technique. Q-SERS, a commercial SERS substrate, was used to detect melamine by Liu and coauthors.¹¹  Q-SERS is made of aggregated gold nanoparticles on a silicon wafer. The gaps of the gold nanoparticles provide “hot spots” for SERS detection. Melamine is first extracted by an organic solvent/water solution from dairy products. It was reported that the detection limit for SERS using Q-SERS substrate was 2 ppm of melamine in liquid milk.

Another commercial substrate, Klarite, is a pattern of inverted gold pyramids developed by the Renishaw Company in the UK using a lithography method. These substrates have also been extensively used in the SERS measurement of melamine in dairy products. It was reported that SERS with Klarite substrates could rapidly detect 0.1% (w/w) melamine in wheat gluten, 0.05% in chicken feed, 0.05% in cakes and 0.07% in noodles, respectively.^12–14  PLS was used in SERS spectral analysis. Other types of substrates, such as silver dendrites, were also used in melamine analysis.¹⁵  The authors used molecularly imprinted polymers rather than an organic solvent to extract the melamine and deposited the melamine onto silver nano-dendrites for SERS analysis. It was suggested that the limit of detection and limit of quantification were 0.012 mmol L⁻¹ and 0.039 mmol L⁻¹ of melamine in whole milk, respectively. Fractal-like gold nanostructures were also proposed to detect melamine in agri-food products.¹⁶  It was reported that the enhancement factor of the fractal-like gold nanostructures could reach an impressive enhancement level of ∼4×10⁷. They could be used to detect melamine and other chemical contaminants, such as dye, and the lowest detectable concentration for the dye molecules was at the ∼0.2 ppb level.

The SERS technique was also used to detect pesticides and other chemical contaminations. Various high-performance substrates have been developed for the analysis of pesticides. Dithiol-functionalized metal nanoparticles were used to create plasmonic “hot spots” and enhance Raman spectroscopic response to the pesticides. It was reported that the dithiol aids in the nanoparticle linkage and creates inter-particle junctions where sensitive “hot spots” required for SERS enhancement are present, creating a specific environment in the nanogaps between silver and gold nanoparticles. They showed a high sensitivity of SERS for the detection of organochlorine pesticides with a limit of detection reaching 10⁻⁸ M. Zhang and others also present a new way to assemble vertically aligned gold nanorod arrays for the detection of pesticides.^17,18  According to the report, the standing nanorod arrays were closely packed onto the gold film, which generated a strong electromagnetic field and uniformly distributed SERS “hot spots” on the surface of the array. Their results demonstrated that SERS could detect as low as 0.1 ppm of carbaryl and a good prediction was made by the multiple linear regression models (R>0.97). In addition, SERS has also been used to detect atrazine, mycotoxins and antibiotics in foods using similar procedures.^19–21 

The extraction of pesticides from fresh produce is required for the aforementioned SERS methods. However, the detection would be largely simplified if the SERS measurement could be performed directly on the fresh produce. For example, shell-isolated SERS was proposed to measure pesticides on the surface of fresh produce.²²  Shell-isolated SERS signals amplification was provided by gold nanoparticles with an ultrathin silica or alumina shell. It was proposed that shell-isolated SERS could be directly used to detect the pesticides on the surface of citrus fruits without solvent extraction.

Not all the chemical contaminants can generate enough Raman spectroscopic responses for an acceptable detection limit by the SERS method. Therefore, a labelling chemical, or so-called Raman dye, was used to amplify the signals derived from the chemical contaminants. These Raman dyes, such as rhodamine B and mercaptobenzoic acid, can generate intensive SERS signals. The detection scheme is usually based upon the changes of Raman signals of the dye that is induced by the addition of the target analyte. The use of Raman dye in SERS detection can sometimes achieve greater sensitivity compared to the direct detection method. For example, a novel “turn-off” biosensing strategy for the detection of thrombin was reported. It was based upon SERS and the mediation of spacing between 4-mercaptobenzoic acid (4-MBA) labelled gold nanoparticles (AuNPs). The detection limit was determined to be 160 fM.^23,24  A method for the detection of mercury ions was also designed using droplet-based microfluidics combined with SERS.²⁵  Quantitative analysis of mercury(ii) ions was achieved by calculating the spectral peak area of rhodamine B with a detection limit between 100 and 500 ppt. A similar strategy was used to detect As ions coupled with the microfluidic sampling technique.²⁶ 

Multiple outbreaks of illness have been associated with foodborne pathogens, such as E. coli O157:H7 in ground beef in the US recently.²⁷  E. coli O157:H7 can cause severe illness such as diarrhoea and acute kidney failure. It is an urgent requirement to develop rapid and convenient methods to monitor foodborne pathogens, such as E. coli O157:H7, Salmonella and Listeria, during food harvesting and food processing. The rapid identification of foodborne pathogens will enable the food supply chain to take faster action on the contaminated products and improve overall food safety.

Raman signals are enhanced by the strong electromagnetic field between the junctions or small gaps among the gold/silver nanomaterials. A small distance, such as 1–2 nm, is necessary for the generation of a strong electromagnetic field. However, the size of bacteria means it is difficult to fit into a small nanogap. Bacteria are usually identified through Raman signals from the bacteria cell membrane and cell wall. It was reported that SERS coupled with in situ coating of bacteria with silver nanoparticles could be utilized to discriminate between three strains of E. coli and one strain of Staphylococcus epidermidis by hierarchy cluster analysis. Using SERS mapping, a detection limit of 2.5×10² cells mL⁻¹ can be achieved.²⁸  SERS, along with novel silver nanorod array substrates and PCA analysis, has also been used for the direct detection of pathogenic bacteria, such as E. coli O157:H7, E. coli DH 5α, Staphylococcus aureus, S. epidermidis and S. Typhimurium.²⁹  In addition, it has been shown that SERS was able to detect single bacterial cells adsorbed onto the silver dendrites with a limit of detection as low as 10⁴ CFU mL⁻¹.^30,31 

SERS can also be used to detect bacteria spores. Gold SERS-active substrates were successfully applied to detect and discriminate among five Bacillus spores (B. cereus ATCC 13061, B. cereus ATCC 10876, B. cereus sp., B. subtilis sp. and B. stearothermophilus sp.). PCA results indicate that the Raman shift range between 900 and 1200 cm⁻¹ contributed significantly to the total data variance. It was also pointed out that a dipicolinic acid band at 998 cm⁻¹ could serve as a biomarker for differentiation among bacterial spores.³² 

Label-free SERS methods have shown promising results in discriminating between different bacteria. However, in the scenario of foodborne pathogens, it often requires a low detection limit to ensure food microbiological safety. Such a low detection limit is very challenging to achieve by label-free SERS methods. Therefore, many researchers have developed so-called indirect detection methods using immuno-magnetic enrichment and SERS nanoprobes. The SERS nanoprobes are labelled with Raman dye for signal amplification. For example, Wang and coauthors have used both immune-enrichment and Raman nanoprobes to detect both S. Typhimurium and S. aureus (Figure 1.4). It was validated in a blind test that the limit of detection is 10³ CFU mL⁻¹ with high specificity.^33,34  Similar strategies have been used in the detection of E. coli. The authors used gold-coated magnetic spherical nanoparticles, which were prepared by immobilizing biotin-labelled anti-E. coli antibodies onto avidin-coated magnetic nanoparticles, in the separation and concentration of E. coli cells. The limit of detection was reported to be 8 and 24 CFU mL⁻¹, respectively.³⁵ 

DNA-based amplification coupled with SERS has also been used to detect foodborne pathogens. DNA-based methods have many advantages compared to antibody-based methods. The extraction and amplification of DNA fragments from the target bacteria make the detection more specific. For example, S. Enteritidis has been quantified by a loop-mediated isothermal amplification SERS assay. According to the study, the target DNA was amplified by LAMP and then labelled with Au-nanoprobes composed of gold nanoparticle-modified with specific cy5/DNA probes, which allows detection by SERS. The sensitivity of the developed assay is much higher compared to traditional PCR methods.³⁶  A novel approach of using SERS for the sequence-specific detection of DNA was proposed. The study utilized magnetic nanoparticles (MNPs) for the enrichment of the target molecules and the detection of the target DNA by labelled Raman dye.³⁷ 

In recent years, there has been increasing interest in the use of novel vibrational spectroscopic methods, such as SERS, as a rapid analytical technique to analyze food composition and determine food quality.^12,13  Raman signals are generated by the inelastic scattering of the incident light from a sample and the shift of frequency or wavelength of the scattered light results from characteristic molecular vibrations.³⁸  Characteristic information of different food components could be obtained by the fingerprinting Raman spectra without complicated sample pre-treatment, such as liquid–liquid extraction, HPLC and GC.¹⁴  The SERS technique has been widely applied for the analysis of lipids, proteins, oligosaccharides and other food components.³⁹ 

Proteins are bio-functional macromolecules consisting of several polypeptides, which typically folded into a globular or fibrous form.⁴⁰  Food protein can also interact with lipids, polysaccharides and minerals. The content of food proteins is a very important nutritional index of food products. The interaction of food protein with other components also influences food quality and food properties. However, current analytical methods for food proteins, such as HPLC, infrared spectroscopy and gel electrophoresis, are complex and often need cumbersome purification. As a new approach to analyze food proteins, SERS could rapidly reflect the fingerprinting information of various proteins in the food product without destroying the food and complex pre-treatments. Unlike the conventional analytical methods, SERS not only differentiates bands representing vibrational modes of the peptide backbone and its side chains, but can also provide information about spectral positions, intensities of protein secondary, tertiary and quaternary structures, side chain orientations and the local environments.⁴¹  In addition, SERS has a great advantage compared with fluorescence-based methods. The broad emission spectra from molecular fluorophores in the proteins make simultaneous detection of multiple proteins impossible. Furthermore, susceptibility to photo bleaching of fluorescence-based methods also makes the detection limit very high.⁴²  In comparison, these problems do not exist in the SERS technique. An increasing number of studies validated that SERS has a great potential to be applied in the analysis of food proteins and deciphering the interactions of protein–protein, protein–polysaccharides and protein–lipids.⁴³ 

Most of the lipids identified in foods are in the form of triglycerides, cholesterol and phospholipids.⁴⁴  Some dietary fats are necessary to facilitate absorption of fat-soluble vitamins (A, D, E and K) and carotenoids. Therefore, the quality of oil is important to the health of consumers. In recent years, Raman spectroscopy has been used to test the quality of oil, such as sunflower and olive oil. Muik and others investigated the chemical changes taking place during lipid oxidation in several edible oils by Raman spectroscopy.⁴⁵  The authors detected the formation of saturated and unsaturated aldehydes using Raman spectra of pure chemicals. The Raman shifts are very close among different cooking oils. However, the ratio of peak intensity is different among the oils. As a result, the oil composition could be differentiated by Raman spectroscopy.

The SERS technique has also been used to differentiate cooking oil and adulterated illegal cooking oil in China. The illegal cooking oil is made from swill and cooked oil with a high peroxide value.⁴⁶  The 1660 cm⁻¹ shift of the CC stretching mode of cis unsaturated fatty acid will vanish and the intensity of the CO vibration will increase. For conventional Raman scattering, this change could be detected only if the oil has been adulterated with a large portion of illegal oil. To detect the low concentration of illegal oil, specific noble metal nanoparticles, such as silver nanoparticles, have to be supplemented into the oil to enhance Raman signal intensity. In brief, SERS is a rapid and useful method for the detection of oil quality and safety.

The conventional methods to measure the concentration of polysaccharides and detect the structure of polysaccharides are HPLC, mass spectroscopy (MS) and nuclear magnetic resonance (NMR).⁴⁷  Both MS and NMR methods require high-purity samples of polysaccharides. For NMR methods, radioactive elements have to be used and this can pose health risks for the operators. However, these disadvantages could easily be avoided by using SERS techniques. Mrozek and colleagues successfully applied SERS in the analysis of polysaccharides in aqueous solution.⁴⁸  Different types of monosaccharides were differentiated by SERS when the concentrations were 0.01 M (Figure 1.5). They concluded that the SERS approach is a valuable method for sensitive detection and characterization of carbohydrates. The identification of oligosaccharide and quantification of a mixture could be fulfilled by SERS. They demonstrated the feasibility of using Raman spectroscopy for the analysis of small quantities of chemically similar oligosaccharides and their mixtures at a concentration of 1 mM. The conventional HPLC methods sometimes could not separate similar oligosaccharides and often lead to the overlap of peaks. In comparison, by using the SERS technique, not only could the characterization of oligosaccharides be accomplished quickly, but quantification of oligosaccharides could also be fulfilled at the same time without sample pre-treatment. They also identified that Raman spectra of millimolar concentrations of aqueous mono- and oligosaccharides can be obtained by drop coating deposition onto gold or silver foil substrates in a nano-size range. Their study also indicated that spectra of individual sugars were identified with 100% accuracy and mixtures of the two sugars were quantified with an average error of 2.7% in the relative maltotetraose/stachyose. The SERS technique provides new ways for rapid and accurate detection of polysaccharides in the food industry. As a result, SERS could be applied to monitor the production process of sugar-containing products. In addition, rapid detection of saccharide, such as glucose, has opened a new door for controlling the product quality where the glucose is used as a quality indicator.

With the development of high-performance Raman-active substrates in recent years, SERS coupled with various extraction/enrichment methods has been validated to be a rapid and promising tool for the detection of chemical and bacterial contaminations in agri-foods. SERS has also shown its potential to be used in the analysis of food composition and food quality due to its fingerprinting spectral features and fast measurement performance. SERS is expected to be used to replace chromatography-based methods in the near future for the analysis and control of standard food quality and safety.