Research Status in the Use of Surface-Enhanced Raman Scattering (SERS) to Detect Pesticide Residues in Foods and Plant-Derived Chinese Herbal Medicines

Surface-enhanced Raman scattering (SERS) technology has unique advantages in the rapid detection of pesticides in plant-derived foods, leading to reduced detection limits and increased accuracy. Plant-derived Chinese herbal medicines have similar sources to plant-derived foods; however, due to the rough surfaces and complex compositions of herbal medicines, the detection of pesticide residues in this context continues to rely heavily on traditional methods, which are time consuming and laborious and are unable to meet market demands for portability. The application of flexible nanomaterials and SERS technology in this realm would allow rapid and accurate detection in a portable format. Therefore, in this review, we summarize the underlying principles and characteristics of SERS technology, with particular focus on applications of SERS for the analysis of pesticide residues in agricultural products. This paper summarizes recent research progress in the field from three main directions: sample pretreatment, SERS substrates, and data processing. The prospects and limitations of SERS technology are also discussed, in order to provide theoretical support for rapid detection of pesticide residues in Chinese herbal medicines.


Introduction
Pesticides are chemicals or biological agents used in agriculture and forestry to prevent and control the growth of pests, including microbes, insects, and weeds, which would alter the growth of desirable plants [1].Pesticides encompass a wide range of chemicals with broad functions, and their use can greatly improve the yield of crops.However, some pesticides are chemically stable and slow to degrade fully, and they can leave residues on the plant material or in the environment.Tese residues could then enter the human body through consumption of the plant or via the water or atmosphere.Even small amounts of some residues can impact human health, for example, by inhibiting the immune system or by increasing the risk of cancer [2,3].Terefore, it is important to develop tools to sensitively detect pesticide residues in agricultural products to ensure food safety.
Chinese herbal medicines (CHMs) are collected, processed, prepared, and used based on China's traditional medical theory, and most of them are plant-based medicines that have been used for thousands of years [4].In modern times, pesticides have been widely used in order to reduce damage caused by pests and diseases during cultivation.Te presence of pesticide residues on CHMs can endanger human health, but they can also alter the efcacy of medicinal products.Pesticide residues also afect the trade in CHMs: exported CHMs are frequently returned due to contamination, causing economic losses and negatively affecting the virtuous circle of the Chinese medicinal materials' market [5].Because pesticide residues have such an important and multifaceted impact on CHMs, we have focused our analysis on this particular agricultural product.
In this review, the current situation of pesticide residues in foods and plant-derived CHMs is introduced.We summarize existing detection methods, including traditional chromatographic detection methods and emerging rapid detection methods mostly based on spectroscopy.In particular, we focus on the advantages of the use of surfaceenhanced Raman scattering (SERS) in the detection of pesticide residues.We introduce current research regarding the application of SERS technology to pesticide residue detection and summarize challenges that remain in the detection process.Finally, we present prospects of future applications of SERS technology in the detection of pesticide residues in CHMs.

Methods of Detection of Pesticide Residues in
Chinese Herbal Medicines  [6].In an analysis of 99 batches of Chuanxiong samples, the rate of detection of dichlorodiphenyltrichloroethane (DDT), chlorobenzamide, and other pesticides was 68.69% [7].Te rate of detection of pyrethroid pesticides in 40 batches of wolfberry samples from diferent sources was 42.9% [8].
Examples of other pesticide residues that have been found associated with CHMs are shown in Table 1.

Methods of Detecting Pesticide Residues in CHMs.
Te most important traditional methods of detection are chromatography and chromatography coupled to mass spectrometry (MS), while spectroscopic techniques are usually used for rapid detection.Te processes of detection can be generally divided into four steps: collection, extraction, purifcation, and detection.

Traditional Methods.
Gas chromatography-MS (GC-MS) couples the strong separation ability of GC with the accurate identifcation ability of MS [19].Tis technique is associated with high sensitivity and accuracy and is a robust method.Detection limits can be as low as 0.001 μg•kg −1 , which allows for the trace detection of several types of pesticide residues [20].However, this technique is unable to measure compounds that are strongly polar, nonvolatile, or thermally unstable.In addition, GC-MS is relatively expensive, as it relies on sophisticated instruments and highly trained professionals.Liquid chromatography-(LC-) MS can separate thermally unstable and nonvolatile samples and thus can complement the limitations of GC [21].LC-MS is also associated with highly accurate, sensitive, and reliable results.Te detection limit is as low as 0.01 μg•kg −1 , leading to appropriately qualitative and quantitative capabilities [22,23].However, bulky and cumbersome instrumentation has been difcult to adapt to the rapid detection demands of the market.
Tin-layer chromatography uses the adsorption capacity of stationary and mobile phases to achieve the separation of analytes [24].Te method is simple and intuitive, it can analyze multiple samples at the same time, and the detection limit of some pesticides is 0.05 μg•kg −1 [25].In general, this method is associated with low sensitivity, though.Its lack of sufcient separating ability often means that it can only serve as an initial means of separation.
In short, at present, traditional methods of detection are highly accurate and have low detection limits, but the procedures used to perform detection tend to be complicated, expensive, and time consuming.Tese defciencies make them difcult to adapt to the current CHM market demands for rapid application and detection.

Spectroscopic Methods.
Compared with traditional methods, spectroscopic methods tend to be associated with rapid and simple operation, allowing them to meet the current need for rapid detection technology.Ultraviolet (UV) spectroscopy is based on transitions of valence electrons in target molecules [26].Tis technique tends to be simple to operate and requires short detection times.Te detection limits of chlorpyrifos and prothioconazole are 4.0 μg•L −1 and 0.38 μg•L −1 , respectively [27,28].Nonetheless, it has several defciencies that greatly limit its application in pesticide residue detection, such as high detection limits and a lack of specifcity, and its adaptability needs to be improved.
Near-infrared (NIR) spectroscopy is used to analyze information carried by the substance according to the light transmitted or refected by an NIR light source [29].It is associated with fast detection speeds, high efciency, and low cost, and it can realize the qualitative and quantitative analysis of multiple pesticides, such as profenofos, diazinon, and chlorpyrifos [30][31][32][33].However, at present, the range of application of NIR spectroscopy technology is limited, because it has low sensitivity in the detection of some trace substances, and its ability to detect analytes in liquid samples is insufcient.
Terahertz (THz) spectroscopy is a kind of molecular spectroscopy that is associated with good spectral resolution and strong penetration.It can permit the rapid, nondestructive, and label-free detection of pesticide residues [34].When combined with chemometrics, qualitative and quantitative analyses of pesticides can be realized, and the detection limit of some pesticides is as low as 0.01 mg•L −1 [35,36].Unfortunately, THz spectroscopy instrumentation tends to be large and expensive, so portable and rapid detection is unavailable.In addition, analytical results are afected by moisture; the sensitivity and speed of detection thus need to be further improved.

2
International Journal of Analytical Chemistry Raman spectroscopy, based on the Raman scattering efect discovered by Indian scientist C. V. Raman, yields spectra associated with inelastic scattering and can be used to interrogate the internal structural information of molecules [37].Te signal arising from chemicals can be very specifc and is often referred to as a "fngerprint" [38].Raman spectroscopy is free from water interference, requires short detection times, and is simple to operate.Tese advantages have led to its adaptation to the analysis of pesticide molecular structures and the detection of pesticide residues [39].However, the conventional Raman spectrum signal is very weak, which limits its application in analysis.At present, there are two main ways to enhance the Raman signal: one is the improvement in the laser and the optical path and the other is the use of plasmon-enhanced Raman spectroscopy, which combines nanomaterials with electromagnetic radiation [40].
SERS technology, which combines Raman and nanomaterial technologies, can greatly enhance the Raman signal, leading to the trace detection of pesticide residues.SERS also has characteristics typically associated with more conventional methods, including rapid detection, simple operation, instrumentation portability, and low detection costs.It also has a wide application range and is resistant to interference by of-target molecules.
Currently, SERS is mainly used in the detection of pesticide residues in agricultural products, but there are a few reports regarding its use in detection of pesticide residues in CHMs.For example, SERS was combined with chemometric methods to establish a quantitative prediction model of deltamethrin in Corydalis yanhusuo.In this application, gold nanoparticles with a diameter of 75 nm were prepared by a chemical reduction method and then used as a SERS detection enhancement reagent.Tis technique led to efective predictions and a low detection limit of 0.186 μg•L −1 [41].Tis was the frst application of SERS technology to pesticide residues in CHMs, and it provides a new feasible direction for the rapid detection of pesticide residues.
Te advantages and disadvantages of various detection methods as applied to the detection of pesticide residues in CHMs are shown in Table 2. To better illustrate the comparison of these detection methods, we developed scores for the application scope, detection limit, cost, sampling time, usability, and portability of a series of detection methods.In this analysis, positive characteristics, including wider application scope, lower detection limit, lower cost, shorter sampling time, and increased usability and portability, are associated with higher scores.As shown in Figure 1, the key strengths of SERS technology are its ability to realize real-time, on-site, and rapid detection of CHMs and its broad application prospects.

Existing Studies on the Use of SERS in the Detection of Pesticide Residues
Recently, several studies have been performed on the use of SERS technology for detecting pesticide residues (Figure 2).Notably, this technology has been used to detect pesticide residues in CHMs, though further research into this application is warranted.

Pretreatment Methods. Sample pretreatment can reduce
or even eliminate matrix efects and improve the sensitivity and accuracy of SERS-based detection and analysis.However, it is a time-consuming aspect of the detection of pesticide residues, so developing efcient pretreatment methods is of great signifcance.Common pretreatment methods include solid-phase extraction and solid-phase microextraction, as well as a modifed form of solid-phase extraction called "quick, easy, cheap, efective, rugged, and safe" (QuEChERS) technology.Te performance of pretreatment methods is typically evaluated based on recovery rates and other indexes [42,43].
Solid-phase extraction (SPE) technology is based on solid-liquid chromatographic principles and tends to be relatively simple to implement and to yield highly accurate results [44].Solid-phase extraction technology has been applied to detect pesticide residues in tea, red bean, orange, and apple.It facilitates the removal of interfering pigments, sugars, proteins, and other impurities, and it is associated with high recovery rates [45][46][47].However, traditional solidphase extraction technology is expensive, has limited enrichment capacity, and produces variable results.
Solid-phase microextraction (SPME) can avoid some shortcomings of solid-phase extraction and is associated with short operation times and less sample consumption International Journal of Analytical Chemistry International Journal of Analytical Chemistry [42].Te use of SPME prior to detection of pesticide residues in apples, rice, and wheat has been shown to reduce matrix efects with recovery rates ranging from 79.3% to 106.8% [48][49][50].Tis method is not efective in the treatment of substances with similar polarity or those in complex matrixes.
Te QuEChERS technique is a simple and rapid pretreatment method related to solid-phase extraction.It minimizes the treatment process, reduces the amount of solvent required, and minimizes environmental pollution (Figure 3) [43].Using this technique, anhydrous MgSO 4 , primary secondary amine, C18, and graphitized carbon black (GCB) were used as reagents to purify analytes in the detection of chlorpyrifos residues in tea and soil.Te pretreatment was found to eliminate interference by impurities such as organic acids and pigments, and the recovery rate was between 93.2% and 103.8% [51,52].Interestingly, when Fe 3 O 4 was used in the pretreatment of agricultural residues in tea, the rate of recovery was better than that when GCB was used [53,54].
Another substrate that has been used in QuEChERS is multiwalled carbon nanotubes (MWNTs), which have nanoscale hollow tubular structures and large specifc surface areas [41,55].Tey have been applied to the detection of triazole residues in vegetables and deltamethrin residues in C. yanhusuo.Tese nanotubes were found to remove pigments, fatty acids, and other interfering substances, and they were associated with high recovery rates.
With its advantages of simplicity and speed, QuEChERS technology has become the preferred pretreatment method for SERS detection of agricultural residues.It showed excellent performance in eliminating the matrix efect of CHMs such as Codonopsis and Astragalus [8,56], making it possible for QuEChERS-SERS technology to realize rapid detection of pesticide residues in CHMs.Te advantages and disadvantages of several common pretreatment methods as applied to the detection of pesticide residues with SERS are shown in Table 3.It is important to note that some purifcation materials can adsorb pesticides; therefore, it is necessary to pay attention to the amount of material used during processing.

Nanoreinforced Substrates.
Tere are a wide variety of pesticides, and the residues can be present at low concentrations.Fortunately, the Raman signal can be improved by up to 10 orders of magnitude by the use of high-activity enhanced nanoparticle substrates with the optimized size and shape [57].SERS substrates include both substrates composed of a single precious metal and those composed of composite materials.TEM images of diferent nanosubstrates are shown in Figure 4.
Single precious metal substrates mainly consist of nanogold or silver.Tese particles come in diferent sizes and a variety of shapes, such as spheres, fowers, rods, bipyramids, and stars [58][59][60][61].Gold nanoparticles with an average diameter of 30 nm were used as a reinforcing substrate to detect pesticide residues in Chinese cabbage; here, the limit of detection was 0.5 mg•kg −1 [62].Silver nanofowers were used to detect the residues of methomyl and acetamiprid in green tea.In this application, voids on the surfaces of the nanoparticles were found to produce local plasmon resonance, which enhanced the Raman signals.In these experiments, the detection limits were 5.58 × 10 −4 μg•mL −1 for methomyl and 1.88 × 10 −4 μg•mL −1 for acetamiprid [44].When fexible double-cone gold nanoparticles were employed, methylthionine on the surfaces of apples was detected nondestructively with a limit of 31.58 ng•cm −2 (Figure 5) [63].
Other highly active SERS composite substrates have been prepared.For example, inert materials and nanomaterials have been used in the assembly of single metal substrates, International Journal of Analytical Chemistry and amino acids and DNA aptamers have been used as surface modifcation molecules [64,65].Te resulting shellcore structure can compensate for shortcomings of raw materials while retaining desirable physical and chemical properties.For example, nanoparticles (NPs) consisting of a silver core and a gold shell (Ag@Au) were found to be stable and to enhance the Raman signal by a factor of 10 7 ; using this method led to limits of detection of thiram and thiabendazole on apple peel of 1 mg•kg −1 [66].
Similarly, cysteamine modifcation of AuNPs enhances the afnity of acephate to the gold surface, and the resulting detection limit was found to be 0.5 mg•L −1 [67].When a target molecule was linked to a DNA aptamer and then embedded into a nano-tetrahedron based on Au@Ag NPs and a DNA skeleton, as shown in Figure 6, the Raman signal was greatly enhanced, leading to a detection limit as low as 0.0021 ng•mL −1 [68,69].
Tus, both single precious metal substrates and composite substrates have good SERS enhancement efects.Te former are simpler to prepare and have been shown to enhance the Raman signal of pesticide residues, but they tend to be less stable and are difcult to store, so they need to be produced on demand.Te latter have better adsorption and afnity and higher stability and sensitivity, whereas they are more complicated to prepare.

Spectral Data Analysis.
Prediction models can be established by combining spectral technology with chemometrics.Tese models involve a rapid spectral analysis that can be completed without sophisticated training, and they can facilitate detection.Tese processes are accurate and efcient and include spectral preprocessing, model establishment, and model performance evaluation.
An original Raman spectrum contains both chemical information and system disturbance signals.Terefore, it is necessary to preprocess the original spectrum to ensure the accuracy of SERS detection.Common spectral preprocessing methods include the standard normal variate (SNV), multiplicative scatter correction (MSC), mean center (MC), frst derivative (D1), and second derivative (D2).When comparing pretreatment performances of SNV, MSC, MC, D1, and D2 in the detection of deltamethrin in wheat, the MC prediction set was found to have the best performance, and the performance following the use of a partial least squares (PLS) model training set was signifcantly improved [81].
Both single-variable and multivariable models can realize quantitative detection of pesticide residues by SERS.A single-variable model is based on the linear equation relating concentration to peak strength.For example, when the residue of thiram on peach peel was detected with a fexible substrate, a linear equation was established with an R 2 of 0.9756, meaning that this model was able to meet the requirements of quantitative detection (Figure 8) [82].
Multivariable modeling methods include PLS, genetic algorithm-partial least squares (GA-PLS), ant colony optimization (ACO), and successive projection algorithm (SPA).Te performances of diferent prediction models are different because models have a variety of spectral intervals and variable selection procedures.At present, the prediction performance of a model is mainly evaluated by the correlation coefcient (R 2 ), relative percentage diference (RPD), and root mean square error (RMSE), including the root mean square error of calibration (RMSEC) and the root mean square error of prediction (RMSEP).For example, RMSECV and RMSEP have been used as model evaluation indexes, and MC was used as a pretreatment method to compare the performance of PLS, GA-PLS, ACO-PLS, and SPA-PLS.Te results showed that MC-SPA-PLS had the best performance and good repeatability in trace detection [81].Other research on the application of convolutional neural networks and other methods in SERS detection suggest that these methods are efcient in extracting characteristic spectra and lead to improved results [83].In the actual detection process, a multivariable model can enhance efciency and accuracy through analyzing the characteristics of diferent models and selecting an appropriate data processing algorithm according to the scale of required data.However, there is no model database for SERS detection of pesticide residues, and a large amount of preparatory work is required prior to large-scale implementation of this method.

Conclusion and Prospects
SERS technology has obvious benefts for the rapid detection of pesticide residues as compared with other detection methods.It allows fast detection with high sensitivity, its operation is relatively simple, and its instrumentation is portable.Efective sample pretreatment and fexible substrates are prerequisites for rapid detection.QuEChERS is the preferred pretreatment method for SERS detection of agricultural residues, because it is simple and fast, can greatly reduce the pretreatment time, and can remove matrix efects to the greatest extent while ensuring a high recovery rate.Flexible enhanced substrates have the advantages of simple preparation, good stability, and strong Raman signal enhancement efects, and they can permit the realization of nondestructive detection of trace contaminants.As shown in Table 4, SERS is presently able to achieve trace detection or even ultratrace detection, and the detection limits are lower than relevant provisions of the maximum residue limit of pesticides in the current national standard for food safety.It is likely, then, that SERS technology can provide conditions for rapid detection of pesticide residues in CHMs.Besides,  10 International Journal of Analytical Chemistry combining spectral technology with chemometrics can establish prediction models, and the accuracy of quantitative detection can be greatly improved.However, some defciencies remain in the application of SERS technology for detecting pesticide residues in CHMs.First, CHMs have complex chemical compositions, and some physical and chemical properties are similar to those of pesticide components, which makes extraction, analysis, and detection more difcult.Second, the complicated chemical composition of CHMs means that matrix efects are more complex; importantly, SERS-based detection is associated with an amplifed matrix efect, which afects the accuracy of detection results.Tird, there are no specifc standards available for pesticide residues in the detection of CHMs by SERS, which hinders the implementation of SERS technology.
Research into the detection of pesticide residues in CHMs by SERS should take several forms: (1) optimization of enhanced substrates by changing the size and morphology to develop materials with good reinforcement efects and high stability and to allow additional over the amount of reactants, reaction temperature, reaction time, and stirring; (2) optimization of pretreatment methods: simple and fast pretreatment methods should be chosen according to the properties of the tested substance to ensure high recovery rates and accuracy; and (3) optimization of algorithm models and the establishment of SERS pesticide residue detection databases, as well as acceleration of the establishment of SERS pesticide residue detection standards for CHMs.With this research progress, it is believed that SERS technology will begin to have a broader market appeal, and it will be possible to realize rapid and real-time detection of pesticide residues in CHMs.

Figure 1 :Figure 2 :
Figure 1: Comparison of the characteristics of diferent methods used for detecting pesticide residues.

Table 1 :
Pesticide residues associated with some Chinese herbal medicines.

Table 2 :
Comparison of advantages and disadvantages of methods used to detect pesticide residues.

Table 3 :
Comparison of advantages and disadvantages of pretreatment methods.