Food Chemistry 270 (2019) 494–501

Contents lists available at ScienceDirect

Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem

A novel metastable state nanoparticle-enhanced Raman spectroscopy T

coupled with thin layer chromatography for determination of multiple
pesticides
⁎
Yan Kang, Ting Wu, Wanchao Chen, Long Li, Yiping Du
Shanghai Key Laboratory of Functional Materials Chemistry, and Research Centre of Analysis and Test, School of Chemistry & Molecular Engineering, East China
University of Science and Technology, Shanghai 200237, PR China

A R T I C LE I N FO A B S T R A C T

Keywords: A novel and highly sensitive metastable state nanoparticle-enhanced Raman spectroscopy (MSNERS) was re-
Metastable state nanoparticle-enhanced Raman ported in this work, which employed an amphiphilic polymer polyurethane-Ag nanoparticle (AgNPs) as the
spectroscopy MSNERS substrate. Polyurethane could form micelle to incorporate nanoparticles and analytes, where targets
Pesticide could be in close contact with the metal surface, which was effective for further enhancing the detection sen-
Polyurethane
sitivity. Time-dependent visible absorption spectra and time-dependent Raman spectra indicated that poly-
Micelle
Thin layer chromatography
urethane could greatly enhance the stability of AgNPs aggregates during the volatilization process, which was a
great improvement of MSNERS substrate. Combining this MSNERS substrate with the rapid separation method of
thin layer chromatography (TLC), this TLC-MSNERS was successfully applied to analyze mixed pesticides on fruit
skin and the detection limits of thiabendazole, triazophos and phosmet were 0.02 μg/mL, 0.8 μg/mL, and 0.6 μg/
mL, respectively. It enhanced 1 order of magnitude the signals of analytes in comparison to that of traditional
TLC-SERS method.

1. Introduction appropriate hot spots and the volatilization-induced automatic con-

It is always a research hot topic in surface enhanced Raman spec-
troscopy (SERS) by controlling hot spots of Au or Ag nanoparticles to
enhance SERS signal. Recently, highly sensitive SERS signal was ob-
tained based on a state translation process of Au or Ag colloid: from wet
state to dry state, namely metastable state nanoparticle-enhanced
Raman spectroscopy (MSNERS) (Liu et al., 2014; Yang et al., 2011). It
was reported that nanoparticles tended to approach closer to generate
hot spots during the solvent volatilization process. In this case, the
target analytes could be trapped in the gap of hot spots and strong SERS
signal could be obtained at this moment (Dong, Weng, Yang, & Liu,
2015; Yang, Li, Liu, Tang, & Liu, 2015; Zhu, Li, Yu, & Yang, 2014). The
strongest SERS signal was observed when the droplet tended to become
dry. Therefore, it is critical to seize the moment before the solvent
complete drying to collect the best SERS signal. In the previously re-
ported literature, the common method to obtain the optimal signal was
by continuous recording the Raman spectra during the solvent eva-
poration process and chose the strongest signal. Although many studies
have demonstrated that the MSNERS substrate showed higher detection
sensitivity compared with the traditional SERS method due to forming
centration of the target molecules during the detection process (Li et al.,
2014; Wu, Zhou, Yang, & Liu, 2013; Yang, Liu, Ma, & Liu, 2012), the
signal intensity fluctuated a lot during the acquiring process due to the
evaporation process changing with time in real application. In other
words, the optimal signal could only be observed spontaneously. Robust
control over the accurate acquiring time to seize the strongest signal
remains a challenge because of the fluctuations of hot spots’ state and
analyte concentration. Customarily, it needs to acquire a large number
of spectra to screen the most intense signal. Therefore, although
MSNERS is a sensitive technique, there is still room for improvement of
the substrate to obtain stable signal.
In fact, keeping SERS signal stable for a longer time is quite chal-
lenging in the dynamic evaporation process due to the difficulty in
tuning the balance between persistent solvent evaporation and stability
of the resulting nanoparticles aggregates. In this work, we adopted a
novel method to improve the performance of the MSNERS substrate by
adding a polymer surfactant, polyurethane, which is a water soluble
amphiphilic polymer. Polyurethane could form micellar cluster in the
volatilization process, which could reduce the effect caused by volati-
lization-induced automatic concentration of the target molecules

⁎
Corresponding author.
E-mail address: yipingdu@ecust.edu.cn (Y. Du).

https://doi.org/10.1016/j.foodchem.2018.07.070
Received 25 January 2018; Received in revised form 4 June 2018; Accepted 11 July 2018
0308-8146/ © 2018 Published by Elsevier Ltd.
Y. Kang et al.
during the detection process. More importantly, according to the results
from the time-dependent visible diffuse spectra, it was found that the
presence of polyurethane could stabilize different forms of nanoparticle
aggregates for a great while in the dynamic evaporation process. It was
proposed that polyurethane-bridged nanoparticles in the 3D hotspot
matrix were predominant in the aggregates in our system and this
structure could be kept stable for a great while in the evaporation
process. Due to the stability of aggregates, the optimal SERS signal with
slight fluctuation was obtained in this work during the acquisition
process before droplet became dry and this strong signal could be kept
stable for several minutes. This feature was of great help and con-
venience for building up the quantitative relationship between the
Raman intensity and the corresponding concentration of analytes. The
enhancement in stability of this MSNERS substrate was a great im-
provement in comparison to that of previous reported metastable sub-
strates, which was believed to provide bright future for practical ap-
plication of MSNERS substrate.
In general, for SERS analysis, only those molecules readily adsorbed
or close to the surface of nanoparticles can be detected. However, there
are many interference components in real samples. The signal from the
low concentration of analytes or weak adsorbates is likely to be buried
in the signal from high concentration of analytes or strongly adsorbed
molecules. Therefore, it is best to separate the targets from a mixed
system prior to the SERS detection for real-world applications. Thin
layer chromatography (TLC) is a facile, rapid and cost-effective tech-
nique for separating different components from mixtures (Dawan,
Satarpai, Tuchinda, Shiowatana, & Siripinyanond, 2016; Sokolović &
Šimpraga, 2006). Often the developed TLC spots are scraped off and
extracted from the TLC plate by means of repeated rinsing and cen-
trifugation, and further examined by infrared spectroscopy (Sharma &
Kumar, 2017) or mass spectrometry (Stanger et al., 2017) for further
identification, which are time-consuming and labor-intensive. SERS in
conjunction with TLC has achieved on-site detection. It is easy to ac-
quire Raman signal of multiple targets after the components are sepa-
rated by the TLC plate. TLC-SERS technique has been used for detecting
kinds of targets in different fields, such as environmental aromatic
pollutants (Li et al., 2011), fiber of artist dyestuffs (Brosseau et al.,
2009), drugs (Fang, Qi, Lu, & Yang, 2016; Lucotti et al., 2012; Lv et al.,
2015; Xiao, Hui, Zhu, Yan, & Feng, 2016; Zhu, Cao, Cao, Chai, & Lu,
2014), pesticides (Yao et al., 2013), biochemistry (Xie et al., 2017),
polycyclic aromatic hydrocarbons (Chen, Huang, & Zhao, 2015), dyes
(Kong, Squire, Chong, & Wang, 2017; Wang et al., 2014). Although this
coupling technique possesses great potential for rapid analysis in
complex system, the sensitivity of traditional SERS substrates still needs
to be improved. In this work, coupling TLC with this improved MSNERS
technique (TLC-MSNERS), good correlations between targets con-
centration and intensity were obtained. A highly reliable, sensitive and
practical analysis method was established for rapid detection of mul-
tiple pesticides on fruit skin, which demonstrates that this method
possesses great potential to realize the quantitative detection of dif-
ferent targets in real samples.
2. Materials and methods
2.1. Materials
Silver nitrate, sodium citrate and sodium chloride were purchased
from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China) without
further purification. Isopropanol, petroleum benzin and di-
chloromethane were purchased from Shanghai Lingfeng Chemical
Reagent Co., Ltd. (Shanghai, China). Phosmet, thiabendazole, triazo-
phos and all other reagents of analytical grade were obtained from
Aladdin-Reagent Co., Ltd. (Shanghai, China). Polyurethane was pur-
chased from Shanghai Xian Ding Biotechnology Co., Ltd. (Shanghai,
China). All solutions were prepared with Milli-Q water (18 MΩ·cm,
Billerica, MA, USA). TLC plates of silica gel 60-F254 (Ø: 200 nm, LT:
495
Food Chemistry 270 (2019) 494–501
0.2 mm) with glass back plate were purchased from Shanghai Jijusheng
Chemical Co., Ltd. (Shanghai, China). The plate containing fluorescing
additive, F254, was used for easy spot visualization. Various eluents
were prepared by mixing petroleum benzin, dichloromethane and iso-
propanol in different proportions to elicit the best separation of pesti-
cides on TLC plates.
2.2. Preparation of AgNPs
AgNPs were prepared by reducing silver nitrate using sodium citrate
(Lee & Meisel, 1982). In a typical experiment, 18 mg AgNO3 was dis-
solved in 100 mL of H2O and brought to boiling. A solution of 1% so-
dium citrate (2 mL) was added drop by drop to the AgNO3 solution. The
mixed solution was kept on boiling for 45 min. The prepared colloid
was rinsed with water and ethanol for several times and stored at
−4 °C. Before the SERS measurements, the colloid was brought to dif-
ferent concentrations for use.
2.3. Chromatography and detection method
The TLC-MSNERS method was designed for the detection of mul-
tiple pesticides. All the TLC plates were directly used as received. On
each plate, a line parallel to the short side was drawn with a pencil and
the distance to the edge was about 5 mm, slightly higher than the un-
folding reagent. For the separation of different pesticides, 200 μL of
sample solution containing different concentrations of pesticides was
spotted onto the line, along with three standard solutions of 200 μL of
50 μg/mL thiabendazole, 50 μg/mL triazophos and 50 μg/mL phosmet
spotted side by side, respectively, allowing to develop at the same time.
The latter served as position marks, through which the exact spots of
different pesticides in the mixture after development were found.
Chromatography was performed in a vial with an optimized mobile
phase (proper ratio of petroleum benzin, dichloromethane and iso-
propanol). After that, the positions of high concentration standards
could be easily distinguished under 254 nm illumination with a hand-
held UV lamp (Shanghai Jihui Scientific Analyze Instrument CO., Ltd.,
China). The mixture was separated and their positions of three pesti-
cides in the sample were then marked out according to the locations of
standards. Later, 50 μL mixture of AgNPs and polyurethane as the SERS
substrate was dropped to each spot and the SERS spectra were recorded
using Raman spectrometer allowing qualitative and quantitative ana-
lysis of the components of the analyte mixture.
2.4. Sample preparation
Cherries were purchased from a local supermarket in Shanghai.
After thoroughly cleaning the cherries with ultrapure water, 1 mL
mixture of 5 μg/mL thiabendazole, 5 μg/mL phosmet and 5 μg/mL
triazophos was evenly sprayed on the surface and dried at room tem-
perature. Then, 10 mL solution of methanol and dichloromethane (1:1,
v/v) were employed as the extracting reagent. The sample was im-
mersed in the extracting solution under ultrasound (Shanghai Shengyan
Ultrasonic Equipment, China) for 20 min and extracted 2 times. Finally,
the extracts were filtered and concentrated for use. 100 μL of con-
centrated extracts were spotted onto the TLC for developing.
2.5. Instrumentation and characterization
SERS experiments were conducted using a Renishaw InVia Reflex
confocal microscope (Renishaw, UK) equipped with a 50× objective
and a high-resolution grating with 1200 grooves/mm. Acquisition of
SERS spectra employed laser of 785 nm as the excitation source and
laser power of approximately 3 mW. Time-dependent visible diffuse
reflection absorption spectra were recorded by VIS miniature optical
fiber spectrometer (Insion, Germany).
Y. Kang et al.
3. Results and discussions
3.1. The effect of the concentrations of AgNPs on SERS signal
The concentration of AgNPs is an important parameter to influence
the intensity of SERS signal, and here we employed thiabendazole as
probe molecule to evaluate the effects. 30 μL of 0.5 μg/mL thiabenda-
zole was sampled on the TLC plate. Then, 30 μL of Ag sol/polyurethane
(v/v, 3:2) with different concentrations of AgNPs were added onto the
insecticide spot. As depicted in Fig. S1 (Supplementary material), the
signal intensity of thiabendazole shows an obvious increase with col-
loidal silver concentration increased from 0.4 mM to 1.6 mM, whereas a
decrease is observed beyond 1.6 mM. The increase of colloid con-
centration led to more particle-particle interactions and encouraged
more particles to be aggregated so as to produce massive hot-spots,
which was beneficial to generate strong SERS signal. Appropriate ag-
gregation caused a boost of SERS intensity, but excessive aggregation
could reduce the SERS activity of the silver colloids (Li et al., 2011).
Therefore, when the concentration of AgNPs was higher than 1.6 mM,
excessive aggregation resulted in a decrease in SERS signal. Hence,
1.6 mM silver colloid was chosen to carry out the next experiments.
3.2. Characterization of MSNERS substrate in state translation process by
time-dependent visible diffuse reflection spectra
As mentioned in the introduction, MSNERS is based on the trans-
lation of nanoparticles from wet state to dry state during the liquid
phase evaporation process. The state of nanoparticles, especially the
state of its aggregates, could greatly affect the properties of colloid as a
SERS substrate. To monitor the evolution of the plasmonic properties of
the colloid during its evaporation, an integrating sphere in conjunction
with a spectrometer was used to collect the visible diffuse reflection
spectra of 80 μL Ag sols-polyurethane (3:2, v/v). The time-dependent
optical behavior of the colloids is shown in Fig. 1. According to the
changes in intensity, peak position and peak width of the visible diffuse
reflection spectra, we divided the whole process into three different
stages with decreased liquid-phase content. At the first stage of
Fig. 1. Time-dependent visible diffuse absorption spectra of silver sol-polyurethane during the evaporation process, (A) the first stage, (B) the second stage, (C) the
third stage, (D) the value of absorbance intensity of peak at 415 nm, 606 nm and 704 nm.
496
Food Chemistry 270 (2019) 494–501
evaporation (Fig. 1A), an absorbance maximum at 415 nm can be ob-
served during the initiative 30 s, which is typically ascribed to the
surface plasmon resonance (SPR) peak of AgNPs. With the increasing of
evaporation, the content of liquid phase decreased and a new peak at
704 nm appeared. It was reported that the driving force for nano-
particles aggregation in state translation process was short-range at-
traction coming from the solvent capillary forces, which effectively
acted as surface tension leading to a drop in surface energy upon ag-
gregation (Yang et al., 2011). At this point, capillary forces drove the
particles close together, and the average interparticle gap became
smaller and nanoparticle-nanoparticle aggregates was formed (Liu
et al., 2014). On the other hand, aggregates growth was limited by the
increasing electrostatic energy of the aggregates, which counter-
balanced the gain in surface energy (Yang et al., 2011). Therefore, a
balance between capillary forces and electrostatic repulsion resulted in
the formation of a small amount of AgNPs clusters, producing a SPR
peak at 704 nm (Fig. 1A). At the second stage of evaporation (Fig. 1B),
the SPR peak of AgNPs at 415 nm changed and broadened with the
increasing of evaporation, and this peak fused with the peak at 704 nm
to form a much stronger and broader peak after evaporation for 300 s. It
can be seen from Fig. 1D that the intensity value at 704 nm greatly
increases from 300 s to 410 s and fluctuates slightly in the following ca.
3 min in the evolution process, indicating that a large amount of ag-
gregates was formed and these aggregates kept stable for ca. 3 min
during this stage. Accordingly, as the content of the liquid phase de-
creased, the cohesion between particles increased. The capillary force
became dominant when the relevant length scales among nanoparticles
became smaller, which drove the nanoparticles as well as the previously
formed aggregates much closer, resulting in generating massive ag-
gregates during this stage (Liu et al., 2014). At the last stage of eva-
poration (Fig. 1C), the colloids tended to dry. The SPR peak of AgNPs
disappeared and a new peak at 606 nm appeared. The value of peak
intensity at 704 nm and 606 nm (Fig. 1D) increases after evaporation
for 580 s and doesn’t drop until the droplet becomes dry (after 700 s).
At this stage, as the content of the liquid phase further decreased, the
adhesive force between Ag particles was far greater than their gravities,
driving all the particles in the droplet to become aggregates. The
Y. Kang et al.
droplet was full of nanoparticle-nanoparticle aggregates, nanoparticle-
aggregate aggregations and aggregate-aggregate aggregations. The
above results reflected that the aggregating process of AgNPs was dy-
namic and time-dependent. It can be seen from Fig. 1D that the particles
and aggregates at each stage can keep stable for more than 100 s during
the evaporation process, which may be attributed to the better balance
between capillary force and electrostatic repulsion force. In general, it
is challenging for nanoparticle aggregates to keep stable in a dynamic
process. In this work, polyurethane, an amphiphilic polymer, formed
micellar cluster in the volatilization process. This feature was capable of
stabilizing the colloid-polymer and aggregation-polymer mixtures sys-
tems for a longer time (Urban et al., 2013). Therefore, the system could
maintain the metastable state of each stage for more than one minute
when forming different kinds of aggregates in the dynamic process.
3.3. Time-dependent SERS spectra of thiabendazole during the state
translation process
Time-dependent diffuse reflection spectra had displayed that AgNPs
showed different degrees of agglomeration at different stages in the
evaporation process, which could refer to fluctuations in intensity or
spectral shape of SERS signal. Therefore, we investigated the SERS
activity of this substrate during the state translation process and tried to
find the best state for the acquisition of signal. Thiabendazole was
served as a probe to track this process. The SERS spectra were recorded
with integration time of 10 s and laser power of 3 mW. When a droplet
of 80 µL of Ag sol-polyurethane (3:2, v/v) and thiabendazole was
placed, time-dependent SERS spectra were collected at intervals of 25 s
in the range of 0–850 s during the evaporation process (Fig. 2A and B).
Within the first 200 s, no Raman signals were observed. With the eva-
poration of solvent, weak SERS signals of thiabendazole gradually ap-
peared after 200 s and increased slowly in the following 100 s. Then,
the signal intensity enhanced quickly in the followed 80 s and reached
its maximal value at 500 s. Interestingly, the signal intensity fluctuated
slightly in the range of 500–675 s (Fig. 2C). Subsequently, the signal
intensity started to drop and relative weak signal could still be observed
after the droplet was dry. It can be seen from Fig. 2C that the strongest
SERS signal can keep stable for about 3 min (in the range of 500–675 s),
which is long enough for acquiring SERS spectra in practical applica-
tions. The rule of changes in SERS signal intensity was in agreement
with the one in visible spectra in the evaporation process. Different
aggregation extents resulted in forming different hot spots, making the
signal intensity of thiabendazole different. It was reported that AgNPs
could construct a 3D hotspot matrix when the liquid phase in the
droplet was evaporating. This 3D architecture was constituted by li-
quid-bridged nanoparticles, which could maintain specific gap sizes
between every two adjacent particles (Liu et al., 2014). In general, gap
sizes will become closer and closer as the evaporation of water is
continuous. In this study, polyurethane could form much micellar
cluster to incorporate AgNPs in the volatilization process and the 3D
Fig. 2. Time-dependent Raman spectra acquired on Ag sol-polyurethane, (A) in the range of 0–675 s; (B) in the range of 700–850 s the third stage; (C) the relationship
between time and intensity of the peak at 780 cm−1 and 1008 cm−1 of thiabendazole according to (A) and (B).
497
Food Chemistry 270 (2019) 494–501
hotspot matrix was constituted by polyurethane-bridged and water-
bridged nanoparticles. According to the results from time-dependent
visible spectra, the aggregates in our system could keep stable for a
while in the evaporation process, which implied that polyurethane-
bridged nanoparticles in the 3D hotspot matrix was predominant and
this structure would be more stable. In this case, the interparticle gaps
in the 3D space of this droplet shrank slowly as the water evaporated.
The nanoparticles and its aggregates could be stabilized for a longer
time. Besides its stability for hot spots, polyurethane micellar cluster
could reduce the effect induced by volatilization-induced automatic
concentration of the target molecules during the detection process.
These features allowed for sufficient settling time to collect Raman
signals in different stages. Therefore, this SERS substrate was more
practical and easier to operate in applications. According to the results,
the best SERS spectra were acquired in the range of 500–675 s in the
next experiments.
3.4. Optimization of the concentration of polyurethane
The concentration of polyurethane played an important role in the
stabilization and enhancement of SERS signal. The effect of different
concentrations of polyurethane (varied in four steps as follows: 30%,
40%, 50% and 60%) was investigated in this work. SERS spectra of
0.15 μg/mL thiabendazole was acquired at different concentrations of
polyurethane mixed with silver colloid in the evaporation process. As
depicted in Fig. S2, SERS signal intensity of the probe molecule re-
corded at every concentration shows low value at first, followed with a
sharp increase and reaching the maximal value, and then continually
drops. Among the four concentrations, the substrate with 40% poly-
urethane showed the best SERS activity. SERS intensity recorded on
substrate with 30% polyurethane increased after evaporation for 200 s
and started to drop after 400 s. SERS intensity acquired on 50% and
60% polyurethane was weaker than that of 40% polyurethane. Poly-
urethane with different concentrations could form different nano-
particles aggregates in the evaporation process. When substrate with
40% polyurethane was applied, hot spots generated from aggregates
could be the most appropriate for producing SERS signal, trapping more
analytes in hot spots. Hence, substrate with 40% polyurethane was the
best option for application.
3.5. Determination of multiple pesticides by TLC-MSNERS
In practice, for the protection from diseases and insects, different
pesticides are blended for spraying on fruits and crops, resulting in
multi-pesticide residues. To provide an example for quantitative de-
tection on fruit, we mixed three pesticides of thiabendazole, triazophos
and phosmet in different ratios, separating the mixtures by TLC and
quantifying the concentrations by their respective SERS signal intensity
by MSNERS substrate.
Y. Kang et al.
Fig. 3. Spots 1, 2, 3 are three reference chemicals, 1- triazophos, 2- phosmet, 3- thiabendazole; Spot d is a mixed sample of phosmet, triazophos and thiabendazole.
SERS spectra obtained from the spots after separation (phosmet (a), triazophos (b), thiabendazole (c)) on the TLC plate and spot d without separation.
3.5.1. Separation of thiabendazole, triazophos and phosmet by TLC
Combining this improved MSNERS substrate with TLC, this devel-
oped method was utilized to detect mixed pesticides of thiabendazole,
triazophos and phosmet. Selecting the appropriate unfolding agent is
the first step in achieving effective separation. Mobile phase of di-
chloromethane, petroleum benzin and isopropanol
(V1:V2:V3 = 80:20:3) was employed to unfold the mixed pesticides in
this work. The separated spots were then located under UV illumination
and were individually analyzed by SERS under the optimal condition.
Fig. 3 displays the schematic diagram of the results of separation and
detection, where spots 1, 2, 3 represent reference chemicals, triazophos,
phosmet and thiabendazole, respectively, and spot d is the mixed pes-
ticides before unfolding. The spots of the separated components (spots
a, b, c in the right side) were marked out according to the locations of
the reference chemicals (spots a, b, c in the left side) after unfolding
from spots 1, 2, 3. SERS spectra for the three separated spots and ori-
ginal mixed samples on the TLC plate are shown in parts a, b, c and d in
Fig. 3, respectively. SERS spectra from spots a, b, c are different from
each other, and these spectra match well with those obtained from
reference chemical spots, indicating that the separated spot a is
phosmet, spot b is triazophos and spot c is thiabendazole. These results
demonstrated that the mixed components have been successfully se-
parated and distinguished using the developed TLC-MSNERS method.
498
Food Chemistry 270 (2019) 494–501
3.5.2. Detection of thiabendazole, triazophos and phosmet by TLC-
MSNERS
Under optimized condition, the quantitative detection of multiple
pesticides by TLC-MSNERS was investigated. Several standard mixture
solutions of thiabendazole, triazophos and phosmet with different
concentrations were prepared. The mixtures were separated by TLC
with the aforementioned mobile phase. Their SERS spectra were re-
corded after dropping Ag sol-polyurethane on the separated spots under
optimal conditions. Fig. 4(A, C, E) illustrates the SERS spectra of thia-
bendazole, triazophos and phosmet with different concentrations after
separation. The intensity of representative peaks of each analyte was
measured, and the concentration dependent SERS intensities at their
respective representative peaks of the three pesticides are shown in
parts B, D, F of Fig. 4, respectively. It can be seen that the signal in-
tensity develops with the increase of the concentration of targets. The
lowest detection concentration of this TLC-MSNERS for thiabendazole
(at 780 cm−1), triazophos (at 1002 cm−1), and phosmet (at 508 cm−1)
was 0.02 μg/mL, 0.8 μg/mL, and 0.6 μg/mL, respectively. These dif-
ferences in the lowest detection concentration among different com-
pounds may be due to their different SERS activities and affinities for
AgNPs. Therefore, different the lowest detection concentrations for
different pesticides are observed. This value of lowest detection con-
centration can be further lowered to meet the demand for detection of
much lower concentration of pesticides if a preconcentration proce-
dure, such as liquid–liquid extraction or solid phase extraction, is
Y. Kang et al. Food Chemistry 270 (2019) 494–501

Fig. 4. SERS spectra of (A) thiabendazole at different concentrations (a) 0.03 μg/mL, (b) 0.17 μg/mL, (c) 0.67 μg/mL, (d) 6.7 μg/mL, (e) 33.3 μg/mL, (f) 166.7 μg/mL,
(g) 666.7 μg/mL. (C) triazophos at different concentrations (a) 1.33 μg/mL, (b) 3.33 μg/mL, (c) 16.67 μg/mL, (d) 66.67 μg/mL, (e) 200 μg/mL. (E) phosmet at
different concentrations (a) 1.67 μg/mL, (b) 3.33 μg/mL, (c) 16.67 μg/mL, (d) 33.33 μg/mL, (e) 100 μg/mL, (f) 166.67 μg/mL. Plots of intensities versus con-
centrations for (B) thiabendazole at 780 cm−1, (D) triazophos at 1002 cm−1, (F) phosmet at 508 cm−1. Three replicate samples from one representative experiment.
Error bars represent the mean value of three replicate samples and corresponding standard deviation.

applied. Thus, the presented method does match the requirement of
rapid and preliminary quantitative analysis of different pesticides, not
only for high concentrations but also for relatively low concentrations.
In order to compare the sensitivity of this TLC-MSNERS method
with traditional TLC-SERS, a series of concentrations of thiabendazole,
triazophos and phosmet solutions were measured using these two
methods. Silver sol-polyurethane (40%) was used in TLC-MSNERS
method and 1.6 mM silver sol was used in the TLC-SERS method. As
displayed in Fig. 5, the LOD of TLC-MSNERS for thiabendazole was
0.02 μg/mL, which enhanced more than 1 order of magnitude in
comparison to that of the TLC-SERS method with LOD of 0.3 μg/mL
(Fig. 5A). Similarly, TLC-MSNERS showed a better sensitivity than TLC-
SERS method for triazophos and phosmet (Fig. 5B–C). The results de-
monstrated that the TLC-MSNERS method had better sensitivity.
3.5.3. Analysis of multiple pesticides on cherry
To further test the performance of the presented method, this
method was employed to investigate an artificially produced model
sample of pesticides on cherry. 1 mL mixture of 5 μg/mL thiabendazole,
5 μg/mL phosmet and 5 μg/mL triazophos were sprayed to the surface

Fig. 5. SERS spectra of different pesticides with different concentrations using TLC-MSNERS method versus TLC-SERS method, (A) thiabendazole, (B) triazophos and
(C) phosmet.

499
Y. Kang et al. Food Chemistry 270 (2019) 494–501

Table 1 Commission of Shanghai Municipality. Ting Wu has received research

Analytical results (n = 3) for pesticides on cherry by TLC-MSNERS and HPLC. grants from Fundamental Research Funds for the Central Universities.
Samples Added The TLC-MSNERS method HPLC Yan Kang declares that she has no conflict of interest. Long Li declares
(μg/ that she has no conflict of interest. Wanchao Chen declares that he has
mL) Detection Recovery (%) Detection Recovery (%) no conflict of interest. Yiping Du declares that he has no conflict of
(μg/mL) (μg/mL) interest. Ting Wu declare that she has no conflict of interest.
Thiabendazole 0 0 0 0 0
5 4.4 ± 0.3 88 ± 6 4.6 ± 0.4 92 ± 8 Ethical approval
Triazophos 0 0 0 0 0
5 4.2 ± 0.4 84 ± 8 4.5 ± 0.3 90 ± 6 This article does not contain any studies with human participants or
animals performed by any of the authors.
Phosmet 0 0 0 0 0
5 4.1 ± 0.3 82 ± 6 4.4 ± 0.2 88 ± 4
Informed consent

of cherry, drying at room temperature. Methanol and dichloromethane Not applicable.

(1:1, v/v) were employed as the extracting reagent, respectively. 5 mL
extracting reagent was added into a vial where the cherry was placed
and sonicated for 20 min to extract the pesticides for 2 times. The ob-
tained 10 mL of sonicated extract was filtered and concentrated into
1 mL. TLC-MSNERS method was used to estimate the concentration of
these pesticides and the listed results were calculated according to the
experimental conditions (Table 1). HPLC is a traditional and mature
method to detect pesticides in mixed system (Harshit, Charmy, &
Nrupesh, 2017). For comparison, the extracts were also analyzed by
HPLC and the results were listed in Table 1. Comparing of the results
obtained from two methods, it can be seen that the detection perfor-
mance of this TLC-MSNERS method is similar to that of HPLC tech-
nique. The deviations of detection concentration are less than 10%
between two techniques. However, TLC-SERS technique is rapider and
lower-cost than HPLC, demonstrating that this developed method is an
excellent candidate for rapid detecting various pesticides simulta-
neously.
4. Conclusions
In summary, an improved MSNERS substrate was fabricated for
sensitive SERS analysis in this work. This method was achieved by
adding an appropriate amount of amphiphilic polymer polyurethane,
which could stabilize different states of nanoparticles and aggregates
for a period of time in the dynamic evaporation process. Time-depen-
dent SERS spectra indicated that the strongest signal acquired on this
MSNERS substrate could be maintained several minutes in this process,
which was a great improvement to enhance the capability of practical
application for MSNERS method. Combining the separation ability of
TLC, TLC-MSNERS has been successfully applied to analyze multiple
pesticides on fruit skin simultaneously, which greatly shortened the
analysis time. In addition, this TLC-MSNERS method could enhance 1
order of magnitude in comparison to that of traditional TLC-SERS
method. Polyurethane micelle facilitated the targets detaching from
TLC plate to enhance detection sensitivity. The obtained results de-
monstrated this coupling technique can facilitate the measurement of a
wider range of analytes in the complicated system at low concentration
and with a short analysis time, providing a valuable tool for both the
emergency monitoring of food safety accidents and routine food quality
monitoring.
Acknowledgements
This work was supported by Science and Technology Commission of
Shanghai Municipality (No. 17142202600) and Fundamental Research
Funds for the Central Universities (No. 222201714047). The authors
are grateful to Dr. L. Li and Dr. L.J. Zhao for helpful discussions.
Conflict of interest
Yan Kang has received research grants from Science and Technology
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the
online version, at https://doi.org/10.1016/j.foodchem.2018.07.070.
References
Brosseau, C. L., Gambardella, A., Casadio, F., Grzywacz, C. M., Wouters, J., & Duyne, R. P.
V. (2009). Ad-hoc surface-enhanced Raman spectroscopy methodologies for the de-
tection of artist dyestuffs: Thin layer chromatography-surface enhanced Raman
spectroscopy and in situ on the fiber analysis. Analytical Chemistry, 81(8), 3056–3062.
Chen, J., Huang, Y.-W., & Zhao, Y. (2015). Detection of polycyclic aromatic hydrocarbons
from cooking oil using ultra-thin layer chromatography and surface enhanced Raman
spectroscopy. Journal of Materials Chemistry B, 3(9), 1898–1906.
Dawan, P., Satarpai, T., Tuchinda, P., Shiowatana, J., & Siripinyanond, A. (2016). A
simple analytical platform based on thin-layer chromatography coupled with paper-
based analytical device for determination of total capsaicinoids in chilli samples.
Talanta, 162, 460–465.
Dong, R., Weng, S., Yang, L. B., & Liu, J. (2015). Rapid detection and direct readout of
drugs in human urine using dynamic surface-enhanced Raman spectroscopy and
support vector machines. Analytical Chemistry, 87(5), 2937–2944.
Fang, F., Qi, Y., Lu, F., & Yang, L. (2016). Highly sensitive on-site detection of drugs
adulterated in botanical dietary supplements using thin layer chromatography
combined with dynamic surface enhanced Raman spectroscopy. Talanta, 146,
351–357.
Harshit, D., Charmy, K., & Nrupesh, P. (2017). Organophosphorus pesticides determi-
nation by novel HPLC and spectrophotometric method. Food Chemistry, 230,
448–453.
Kong, X., Squire, K., Chong, X., & Wang, A. X. (2017). Ultra-sensitive lab-on-a-chip de-
tection of Sudan I in food using plasmonics-enhanced diatomaceous thin film. Food
Control, 79(Supplement C), 258–265.
Lee, P. C., & Meisel, D. (1982). Adsorption and surface-enhanced Raman of dyes on silver
and gold sols. Journal of Physical Chemistry, 86(17), 3391–3395.
Li, D., Qu, L., Zhai, W., Xue, J., Fossey, J. S., & Long, Y. (2011). Facile on-site detection of
substituted aromatic pollutants in water using thin layer chromatography combined
with surface-enhanced Raman spectroscopy. Environmental Science and Technology,
45(9), 4046–4052.
Li, P., Dong, R., Wu, Y., Liu, H., Kong, L., & Yang, L. (2014). Polystyrene/Ag nanoparticles
as dynamic surface-enhanced Raman spectroscopy substrates for sensitive detection
of organophosphorus pesticides. Talanta, 127, 269–275.
Liu, H., Yang, Z., Meng, L., Sun, Y., Wang, J., Yang, L., ... Tian, Z. (2014). Three-di-
mensional and time-ordered surface-enhanced Raman scattering hotspot matrix.
Journal of the American Chemical Society, 136(14), 5332–5341.
Lucotti, A., Tommasini, M., Casella, M., Morganti, A., Gramatica, F., & Zerbi, G. (2012).
TLC–surface enhanced Raman scattering of apomorphine in human plasma.
Vibrational Spectroscopy, 62(5), 286–291.
Lv, D., Cao, Y., Lou, Z., Li, S., Chen, X., Chai, Y., & Lu, F. (2015). Rapid on-site detection of
ephedrine and its analogues used as adulterants in slimming dietary supplements by
TLC-SERS. Analytical and Bioanalytical Chemistry, 407(5), 1313–1325.
Sharma, V., & Kumar, R. (2017). Fourier transform infrared spectroscopy and high per-
formance thin layer chromatography for characterization and multivariate dis-
crimination of blue ballpoint pen ink for forensic applications. Vibrational
Spectroscopy, 92, 96–104.
Sokolović, M., & Šimpraga, B. (2006). Survey of trichothecene mycotoxins in grains and
animal feed in Croatia by thin layer chromatography. Food Control, 17(9), 733–740.
Stanger, R., Tran, Q. A., Smith, N., Kennedy, E., Stockenhuber, M., Lucas, J., & Wall, T.
(2017). Separation and analysis of high range extractable molecules formed during
coal pyrolysis using coupled thin layer chromatography-imaging mass spectrometry
(TLC-LDI-IMS). Fuel, 196, 269–279.
Urban, A. S., Shen, X., Wang, Y., Large, N., Wang, H., Knight, M. W., ... Halas, N. J.
(2013). Three-dimensional plasmonic nanoclusters. Nano Letters, 13(9), 4399–4403.
Wang, C., Cheng, F., Wang, Y., Gong, Z., Fan, M., & Hu, J. (2014). Single point calibration
for semi-quantitative screening based on an internal reference in thin layer

500
Y. Kang et al.
chromatography-SERS: The case of Rhodamine B in chili oil. Analytical Methods,
6(18), 7218–7223.
Wu, Y., Zhou, F., Yang, L., & Liu, J. (2013). A shrinking strategy for creating dynamic
SERS hot spots on the surface of thermosensitive polymer nanospheres. Chemical
Communications, 49(44), 5025–5027.
Xiao, L., Hui, C., Zhu, Q., Yan, L., & Feng, L. (2016). Analysis of low active-pharmaceu-
tical-ingredient signal drugs based on thin layer chromatography and surface-en-
hanced raman spectroscopy. Journal of Pharmaceutical and Biomedical Analysis, 131,
410–419.
Xie, Z., Wang, Y., Chen, Y., Xu, X., Jin, Z., Ding, Y., ... Wu, F. (2017). Tuneable surface
enhanced Raman spectroscopy hyphenated to chemically derivatized thin-layer
chromatography plates for screening histamine in fish. Food Chemistry, 230, 547–552.
Yang, L., Li, P., Liu, H., Tang, X., & Liu, J. (2015). A dynamic surface enhanced Raman
spectroscopy method for ultra-sensitive detection: From the wet state to the dry state.
Chemical Society Reviews, 44(10), 2837–2848.
501
Food Chemistry 270 (2019) 494–501
Yang, L., Liu, H., Ma, Y., & Liu, J. (2012). Solvent-induced hot spot switch on silver
nanorod enhanced Raman spectroscopy. Analyst, 137(7), 1547–1549.
Yang, L., Liu, H., Wang, J., Zhou, F., Tian, Z., & Liu, J. (2011). Metastable state nano-
particle-enhanced Raman spectroscopy for highly sensitive detection. Chemical
Communications, 47(12), 3583–3585.
Yao, C., Cheng, F., Wang, C., Wang, Y., Guo, X., Gong, Z., ... Zhang, Z. (2013). Separation,
identification and fast determination of organophosphate pesticide methidathion in
tea leaves by thin layer chromatography–surface-enhanced Raman scattering.
Analytical Methods, 5(20), 5560–5564.
Zhu, Q., Cao, Y., Cao, Y., Chai, Y., & Lu, F. (2014). Rapid on-site TLC-SERS detection of
four antidiabetes drugs used as adulterants in botanical dietary supplements.
Analytical and Bioanalytical Chemistry, 406(7), 1877–1884.
Zhu, Y., Li, M., Yu, D., & Yang, L. (2014). A novel paper rag as 'D-SERS' substrate for
detection of pesticide residues at various peels. Talanta, 128, 117–124.