RSC Advances

View Article Online

PAPER View Journal | View Issue

Discrimination of sibutramine and its analogues

based on surface-enhanced Raman spectroscopy
Published on 12 December 2014. Downloaded by Carleton University on 12/05/2015 08:23:02.

Cite this: RSC Adv., 2015, 5, 5886

and chemometrics: toward the rapid detection of
synthetic anorexic drugs in natural slimming
products†
Hui Mao,ab Meihui Qi,a Yujie Zhou,a Xiaoyan Huang,a Liying Zhang,a Yang Jin,a
Yan Penga and Shuhu Du*a

Nowadays there is growing concern about the adulteration of synthetic anorexics in natural slimming
products (NSP) due to the increasing use of weight-reducing aids. Here, we report surface-enhanced
Raman spectroscopy (SERS) coupled with chemometrics for rapid discrimination and detection of
sibutramine (SIB) and its analogues (mono-desmethylsibutramine, MDS; di-desmethylsibutramine, DDS)
in konjac pressed candy (KPC), which is a NSP used in the treatment of obesity. It has been
demonstrated that the two characteristic peaks at 1060 cm
1 (in SERS spectrum of SIB) and 1407 cm
1
(in SERS spectrum of MDS) were used to discriminate between SIB, MDS and DDS. Moreover, the two
spectra for KPC with MDS and KPC with DDS only exhibited some slight differences. After the second
derivative transformation, the differences were much more clear, especially for the peak at 1407 cm
1 in
the SERS spectrum of KPC with MDS. Meanwhile, partial least-squares regression was used for
Received 2nd September 2014
Accepted 11th December 2014
quantitative analysis of the samples. The limit of detection for SIB, MDS and DDS was 5.00  10
8 M,
5.00  10
7 M and 1.00  10
6 M, respectively. Compared with previously reported studies, the method
DOI: 10.1039/c4ra09636c
is simple (no pretreatment), less time-consuming and inexpensive, and can be applied to rapidly screen
www.rsc.org/advances SIB and its analogues in NSP.

2010, the European Medicines Agency (EMA) and the Chinese

1 Introduction State Food and Drug Administration (SFDA) recommended the
Obesity, which directly affects the quality of people's lives, is market withdrawal of SIB owing to an association with
one of the most serious public health problems of the 21st increased cardiovascular events and strokes.5 To avoid the SIB
century. The prevalence has increased at an alarming rate, examination of local authorities related to public health control
resulting in the search for a number of weight-reducing aids and surveillance, mono-desmethylsibutramine (MDS) or di-
including sibutramine (SIB), orlistat, phentermine, rimonabant desmethylsibutramine (DDS), which has the same function as
and especially natural slimming products (NSP) due to them SIB,5 has been used as the substitute of SIB adulterated into NSP
having no signicant side effects on the human body. However, by some illegal manufacturers. However, MDS and DDS, as
the weight-reducing effect of NSP is not usually obvious in the main desmethyl active metabolites of SIB, can still cause an
short-term. As a result, some manufacturers illegally adulterate adverse reaction. As their chromatographic behaviors are not
synthetic anorexic drug in NSP in order to promote their efficacy similar to that of SIB in the routine detection using high
of weight loss. SIB, as a reuptake inhibitor of 5-hydroxytrypta- performance liquid chromatography (HPLC), the two metabo-
mine and noradrenaline,1–3 which makes people lose weight by lites are easily missed, leading to false-negative judgments.
stimulating thermogenesis,4 has been one of the most Without adequate quality control, the adulterated NSP may
commonly adulterated anorexic medicines found in NSP. In cause consumers to suffer from several unpredictable distresses
ranging from headache, vertigo and numbness to serious
cardiovascular effects.6 Therefore, the identication and
a
School of Pharmacy, Nanjing Medical University, Nanjing, 211166, China. E-mail: measurement of SIB and its analogues illegally adulterated in
shuhudu@njmu.edu.cn; Fax: +86 25 86868476; Tel: +86 25 86868476
b
numerous brands of NSP available in the market are important
Nuclear Medicine Department, Nanjing General Hospital of Nanjing Military
to protect consumers from risks.
Command, Nanjing, 210002, China
† Electronic supplementary information (ESI) available. See DOI:
Up to date, various conrmatory detection methods for the
10.1039/c4ra09636c screen of SIB and/or its analogues have been reported, such as

5886 | RSC Adv., 2015, 5, 5886–5894 This journal is © The Royal Society of Chemistry 2015
 View Article Online

Paper RSC Advances

HPLC,7 ultra-HPLC,8 liquid chromatography-mass spectrometry spectral data and develop qualitative and quantitative models
(LC-MS),9 gas chromatography-mass spectrometry (GC-MS),10 for SIB, MDS and DDS.
ion mobility spectrometer (IMS),11 IMS-MS,12
chemiluminescence-molecularly imprinted sol–gel lms
(CL-MIFs),13 ow injection electrochemiluminescence (FI- 2 Experimental
ECL),14 capillary electrophoresis,15 infrared spectroscopy,16 and 2.1 Chemicals and materials
so on. Due to the structural similarity and the trace amount (0.1
Sodium citrate (Na3C6H5O7$2H2O, 99.8%), silver nitrate
mg per capsule11 or tablet) in pharmaceutical formulations, the
(AgNO3, 99%), KI, KNO3, KBr, K2CO3, KCl, ammonium
identication of SIB and its analogues usually requires
formate (HPLC grade) and formic acid (HPLC grade) were
sophisticated sampling beforehand separation/extraction and
Published on 12 December 2014. Downloaded by Carleton University on 12/05/2015 08:23:02.

purchased from Sinopharm Chemical Reagent Co.,

pre-concentration procedures in above techniques, which are
Ltd. (Shanghai, China). Sibutramine hydrochloride mono-
highly expensive or time-consuming and also cause environ-
hydrate, mono-desmethylsibutramine hydrochloride, di-
mental problems. The method for the rapid identication and
desmethylsibutramine hydrochloride, genistein, daidzein,
detection of trace SIB or its analogues in NSP still remains a
salidroside, chlorogenic acid, caffeic acid and gallic acid were
challenge.
purchased from National Institutes for Food and Drug Control
Raman spectroscopy, as a vibrational spectroscopic
(Beijing, China). All of these reagents were used without
method, has been considered as a useful analytical technique
further purication. Konjac pressed candy (KPC), which
to evaluate food safety and quality.17,18 Although Raman
mainly contained konjac powders (tablet, 1 g), and the herbal
spectrum can provide spectroscopic ngerprint information
slimming tea (HST), which mainly contained Green tea,
about various chemical and biochemical species19–21 in
Lonicerae japonicae os, Cassiae semen, Nelumbinis folium,
complex matrix without sample pretreatment, the signal of
Gynostemmatis pentaphylli herba, Crataegi fructus, Sennae
normal Raman spectroscopy is weak, and thus only suitable
folium and Mel (pack, 5 g), were purchased from a local
for detecting target compounds present in high concentration.
drugstore. Ultrapure water (18.2 MU cm) was produced using a
Surface-enhanced Raman spectroscopy (SERS) technique
Milli-Q apparatus (Milford, MA, USA).
signicantly enhances the sensitivity of normal Raman spec-
troscopy detection. In the SERS technique, molecules are
absorbed onto the roughened surface of noble metals, which 2.2 Apparatus and analytical methods
results in signicant enhancement of the Raman signals by The morphologies and structures of NPs were examined by a
many orders of magnitude in highly localized optical elds of JEM-1010 transmission electron microscope (Tokyo, Japan).
nanostructures. In the last decades, both gold (Au) and silver The hydrodynamic sizes of NPs were determined using a Mal-
(Ag) nanoparticles (NPs) are most frequently used nano- vern Zetasizer Nano-ZS90 particle size analyzer (Malvern
materials for fabrication of SERS-active nanosubstrates.22–25 Instrument, UK). UV-vis absorption spectra were obtained by
Although Au has some advantages, i.e., chemical stability and UV-2501 spectrometer (Tokyo, Japan) in the range from 200 to
better biocompatibility, Ag is a much more efficient optical 800 nm with 2 nm resolution, using 1 cm path length quartz
enhancer, and for the same particle size and shape it yields cuvettes. Raman measurements were conducted with Thermo
enhancement factor 2 or 3 orders of magnitude larger than Fisher DXR Raman Microscope equipped with a CCD detector
that of Au.26 Nowadays, SERS technique has become one of the in backscattered conguration using a 10 objective (Madison,
most widely pursued spectroscopic tools for rapid discrimi- USA), and three types of lasers were used to produce different
nation and detection of various contaminants and adulter- excitation wavelengths (532 nm, 633 nm and 780 nm). Liquid
ants,27,28 including banned additives melamine and its chromatography-mass spectrometry (LC-MS) analysis was per-
derivatives,29 banned food dyes Sudan I and its derivatives30 formed using an Agilent 6410B triple quad LC-MS/MS (Santa
and banned seafood drugs crystal violet and malachite Clara, USA). Please see the details of LC-MS analysis condition
green.31,32 It is thus expected that the use of simple and inex- in ESI.†
pensive SERS method can overcome the drawbacks of conr-
matory techniques and provide a direct spectroscopic
identication of trace analyte in complicated matrix. 2.3 Preparation of Ag NPs
In this work, we report the rapid identication and detection All glassware was thoroughly immersed overnight with freshly
of trace SIB and its analogues in NSP using SERS technique in prepared aqua regia (3 : 1 HCl–HNO3) and washed with ultra-
combination with chemometrics. Experimental SERS condi- pure water before use. The preparation of Ag NPs was based on a
tions such as excitation wavelength, concentration of Ag NPs previous method.33 Briey, 25 mg of silver nitrate was dissolved
colloid and type of aggregating agent were rst optimized. in 63 mL of ultrapure water and then heated to boiling under
Subsequently, konjac pressed candy (KPC), as one common type magnetic stirring. To this solution, 1.25 mL of 2% trisodium
of NSP, was selected as sample, because it could be the major citrate was added and kept boiling for 1 h. Aer cooled to room
target of SIB or its analogues monitoring. Finally, multivariate temperature, the Ag NPs colloid was obtained and stored at
statistical methods including discriminant analysis (DA) and 4.0  C for further use. The concentration of Ag NPs was 0.2 nM,
partial least-squares (PLS) analysis were used to analyze SERS which was calculated using Beer's law and the extinction coef-
cient (3) (42 nm Ag NPs is 3.7  1010 M
1 cm
1).34

This journal is © The Royal Society of Chemistry 2015 RSC Adv., 2015, 5, 5886–5894 | 5887
 View Article Online

RSC Advances Paper

2.4 Theoretical calculation between three structural analogues in spiked sample, second
derivative transformation and smoothing were also applied
The molecular vibrational spectra of the investigated
when necessary to capture a more signicant variance. The
compounds (including SIB, MDS and DDS) were simulated by
spectral region used for the DA and PLS was from 1800 to 900
density functional theory (DFT) within Gaussian 09 program
cm
1.
package. On the basis of the molecular geometries optimized by
B3LYP/6-31+G** method, the frequency calculations were
carried out at the same level to validate the minima and to 2.7 SERS detections of SIB, MDS and DDS in spiked samples
obtain the vibrational information. Meanwhile, the interaction Briey, SIB with different amounts was rst spiked into one
between Ag atom and investigated compounds was simulated gram of KPC powders, and then the mixture was dissolved in 10
Published on 12 December 2014. Downloaded by Carleton University on 12/05/2015 08:23:02.

through the basis sets of 6-31+G** and LanL2DZ, which per- mL ultrapure water by sonication, statically cooled down and
formed to non-metal (C, N and O) and Ag atoms. Based on the ltered. 10 mL of ltrate was added into 90 mL of Ag NPs colloid.
optimized geometries, the Raman spectra calculations were The Raman spectra were recorded with the same condition as
carried out at the same computational level. The assignments of the detection of analytes in aqueous solution. Meanwhile,
calculated vibrational modes were made with the help of Gauss accuracy of the method was determined by standard addition
View 5.0. The calculated wavenumbers of all normal modes technique. Known amounts of reference standard SIB in a range
were scaled by a factor of 0.987 to correct systematic errors of of low, medium and high concentrations were added to the KPC
the theoretical methodology. powders and analyzed under the above measuring conditions.
The experiments were replicated six times for each concentra-
tion and the accuracy was calculated as the % of analyte
2.5 SERS measurements of SIB and its analogues
recovered. MDS and DDS was also tested using the same
In brief, 5.00 mM SIB solution was prepared and diluted with method, respectively.
ultrapure water to the predetermined concentrations. Then,
10 mL of the analyte solution was added into 90 mL of Ag NPs
colloid in a 1.5 mL centrifuge tube, followed by shaking for 1 3 Results and discussion
min. The mixture was rst sucked into a capillary glass tube, 3.1 Characterizations of Ag NPs
and then xed onto a glass slide. The Raman spectra were UV-vis spectroscopy and transmission electron microscopy
recorded using 532 nm laser with 10 mW power and 10 (TEM) were applied to characterize the size and morphology of
objective. The collecting time was 10 s with 2 rounds of accu- the synthesized Ag NPs. The maximum absorption of the
mulations and the pinhole aperture was 25 mm. The experi- measured UV-vis spectrum of Ag NPs colloid provides the
ments were repeated six times for every concentration. MDS and information on the diameter of particles, moreover, its full
DDS were also tested using the same method. width at half-maximum (FWHM) is used to estimate the NPs
dispersion.35 As shown in Fig. S1A,† the maximum UV-vis
2.6 Spectral data analysis absorption for Ag NPs colloid was at 409 nm, and the FWHM
was approximately 90 nm, which indicated that the size
Multivariate statistical tools (DA and PLS) were adopted to
dispersion of Ag NPs was narrow. Meanwhile, TEM image
analyze the spectral data with TQ Analyst soware (7.2.0.161
clearly displayed that the diameter of Ag NPs was 42  9 nm
Release, Thermo Fisher Scientic). Through computing the
(Fig. S1B†). These results revealed that the size of Ag NPs was
unknown's distance from each class center in Mahalanobis
mainly homogeneous and the shape of the most of NPs was
distance units, DA classication technique can be used to
spherical.
determine the class or classes of known materials which are
most similar to an unknown material. The closer each distance
value is to zero, the better is the match. Aer the material was 3.2 Optimization of experimental conditions for SERS
successfully classied by DA, PLS was applied to adulterated The optimum of the experimental conditions for SERS detection
samples (which only contain one synthetic anorexic per sample) is related to various parameters including excitation wave-
to obtain a calibration based on the level of adulteration. For length, aggregating agent, pH value, etc. In order to select the
PLS analysis, spectra of 40 samples with ve different concen- optimal detection condition, a series of tests were performed in
trations of constituent were used. Among these samples, thirty this study.
samples were used for calibration and ten samples were used Excitation wavelength. It is known that the basic mecha-
for validation. The calibration data were used to create the nisms of SERS mainly include charge-transfer (CT) mechanism
method calibration model, while the validation data were used and electromagnetic mechanism. For the CT mechanism, it is
to validate the model. Evaluation of the models was made related to the chemical complexation of the analyte molecules
through correlation coefficient (R) and root-mean-square error to the surface of the metallic NPs. In the case of surface
of prediction (RMSEP). The higher the R value or the lower the complexation, electronic absorption of the complexed analyte
RMSEP value is, the better predictability the model has. Before may shi with respect to the free molecule, and even overlap the
the analysis, spectral data were normalized using a standard excitation wavelength leading to resonant effect.36 Fig. 1A
normal variate (SNV) algorithm and mean centered to correct showed the SERS spectra of SIB absorbed on Ag NPs for three
for instrumental effects on the data. To further discriminate excitations (532, 633 and 780 nm). The intensity of peak at 1691

5888 | RSC Adv., 2015, 5, 5886–5894 This journal is © The Royal Society of Chemistry 2015
 View Article Online

Paper RSC Advances

cm
1 in the spectrum recorded with the 532 nm laser line was indicator of particles-aggregating degree), it has been noted that
stronger than those obtained by 633 and 780 nm laser. The extra addition of anions can enhance the SERS signal of SIB to a
surprising differences on the peak intensities should be caused certain extent. The SERS activation induced by anions can be
by the charge-transfer Ag complex with an electronic absorption ascribed to two effects: one is the aggregation of Ag NPs
in the spectral region of the 532 nm laser. The UV-vis spectrum producing “hot spots”; the other is modication of the Ag
of SIB (Fig. S2a†) showed a strong absorption at 223 nm, very far surface chemistry, resulting from the addition of strongly
from the 633 and 780 nm to have any resonant effect. However, binding anion. As clearly presented in Fig. 1B, the Br
, as the
the UV-vis spectrum of the SIB complexed with Ag NPs displayed more strongly binding anion, gave the greatest enhancement
a broad shoulder that extended up to 500 nm region among different kinds of anions, while Cl
gave the least
(Fig. S2b†). Thus, the 532 nm laser was in pre-resonance with enhancement for SIB. So, KBr was chosen as the optimal
Published on 12 December 2014. Downloaded by Carleton University on 12/05/2015 08:23:02.

the absorption of the complex, and was used as excitation aggregating agent to conduct the following optimization
source in the following tests. experiments. Additionally, the stability of Ag NPs colloid with
Aggregating agent. The aggregating agent is crucial for the addition of KBr was also examined, as shown in Fig. S5.†
application of SERS using metallic NPs. It is usually an elec- Clearly, the absorbance of Ag NPs with KBr is very stable and the
trolyte, and can induce the NPs aggregation by increasing the Raman intensity of SIB adsorbed on Ag NPs is nearly unchanged
ionic concentration inside the metallic colloid and decreasing within an hour, which ensures the measurement reliability.
the electrostatic barrier. Fig. S3† showed the UV-vis absorption Substrate concentration. In order to maximize the Raman
spectra of Ag NPs colloid blended with different aggregating signal enhancement, Ag NPs colloid at different concentrations
agents (the concentration of each aggregating agent is 10 mM). was obtained by the treatment of centrifugation or dilution.
The absorbance of Ag NPs colloid decreased aer mixed with KI, Fig. S6† revealed that the Raman intensity of peak at 1691 cm
1
KNO3, KBr, K2CO3 and KCl, respectively, and maximum continuously increased with the Ag NPs concentration in range
absorption at 409 nm shied to 415 nm, 415 nm, 414 nm, 407 of 0.007–0.2 nM. It could be ascribed to the increasing “hot
nm and 406 nm, respectively. From the previous literature, the spots”, which could maximize the localized electric eld and
evolution of the absorbance of Ag NPs colloid could be related to then greatly amplify Raman signals of SIB in the colloid. The
the aggregation of colloidal NPs.41 When aggregating agents further increase of substrate concentration led to the decrease
containing anions are mixed with Ag NPs colloid, the energy of the Raman peak intensity. It was because that Ag NPs colloid
states on particles' surface are affected by chemical contribu- with high concentration was easily precipitated. Therefore,
tions of the anions and the inter-band transitions between NPs optimal concentration of Ag NPs colloid was 0.2 nM.
through anions. The change of particles' surface energy states pH value. When the Ag NPs colloid at different pH values was
caused by additional anions is regarded as energy hybridization used, it was found that considerable changes occurred in the
effect on particles' surface.42 For this reason, the maximum Raman spectra of SIB. As shown in Fig. S7,† the intensity of
absorption of Ag NPs colloid changed in the presence of addi- band (at 1691 cm
1) obtained at pH 7.4 was stronger than that
tional anions compared with the absence of additional anions. at pH 2.5, but this band nearly vanished at pH 11.5. These
The different interactions of Ag NPs with aggregating agents phenomena may be related to variations in molecular proton-
were also conrmed by TEM and dynamic light scattering ated and deprotonated forms as pH values under ambient
(Fig. S4†). conditions37–39 or slight differences on adsorption geometries of
Fig. 1B displayed a series of SERS spectra of SIB absorbed on target molecule on nano-roughened metallic surface.40 To ach-
Ag NPs colloid to which the same amount of KI, KNO3, KBr, ieve maximum SERS intensity, we chose Ag NPs colloid with pH
K2CO3 and KCl was added, respectively. Through comparison of 7.4 as substrate for the detection of SIB. Meanwhile, for the
the six spectra (the feature peak at 1691 cm
1 was chosen as the

Fig. 1 (A) SERS spectra of 5.00  10
4 M SIB in Ag NPs colloid excited by different excitation wavelengths. (B) Effects of aggregating agents on
SERS intensity of 1.00  10
5 M SIB.

This journal is © The Royal Society of Chemistry 2015 RSC Adv., 2015, 5, 5886–5894 | 5889
 View Article Online

RSC Advances Paper

Fig. 2 Chemical structures of SIB, MDS and DDS.

other two target analytes (MDS and DDS), similar optimization ring, respectively. In comparison to MDS and DDS, the SIB
Published on 12 December 2014. Downloaded by Carleton University on 12/05/2015 08:23:02.

results were also obtained (Fig. S8–11†). showed the characteristic peak at 742 cm
1, corresponding to
the symmetry and asymmetry stretch vibrations of the N1–C2
bond. A close analysis of spectra of MDS and DDS solids
3.3 SERS discrimination of SIB, MDS and DDS
revealed no dramatic differences except the two weak feature
Fig. 2 showed the molecular structures of SIB, MDS and DDS. peaks at 1311 cm
1 (deformation vibration of the C2–N1–C7
The structural difference between three molecules is only in the bond) and 1390 cm
1 (deformation vibration of the N–H bond
amounts of substituent (methyl) of amino-group. To discrimi- and in-plane deformation vibration of C–H bond from phenyl
nate between three structural analogues, the Raman spectra of ring) for MDS.
the three analytes were determined. The observed Raman bands Fig. 3B presented the normal Raman spectra of three analyte
with their assignment accomplished with the help of theoretical solutions. Compared with the Raman spectra of solids (Fig. 3A),
calculations were listed in Table 1. it was observed that all bands of three analytes were too weak to
Fig. 3A displayed the Raman spectra of three analyte solids. discriminate between three structural analogues.
Their dominant bands appeared at 1595 cm
1, 1100 cm
1 and Fig. 3C exhibited the SERS spectra of three analytes in Ag NPs
629 cm
1, which was assigned to the stretching vibration of C–C colloid. In comparison to the normal Raman spectra (Fig. 3B),
bonds from phenyl ring, deformation vibration of the CH2 we found that the intensities of some bands were obviously
groups and out-of-plane deformation vibration of the phenyl

Table 1 Raman (experimental and calculated) wavenumbers (in cm
1) of SIB, MDS and DDS with their vibrational assignmenta

Raman

Analyte Solid (cm
1) Calc. (cm
1) SERS (cm
1) Vibrational assignment

SIB 1657 1621 1691 u(C7–N–C8) + n(C–C) (phenyl ring)

1595 1598 1593 n(C–C) (phenyl ring)
1461 1513 — d(CH3) + d(CH2)
1441 1491 1415 n(C–C) (phenyl ring) + d(CH3)
1100 1102 1095 s(CH2)
1061 1051 1060 n(C2–N1–C8)
1009 1003 1000 Phenyl ring breathing
742 741 — n(N1–C2) + d(CH2)
629 628 628 Phenyl ring out-of-plane def. + four-membered ring in-plane def.
MDS — 1685 1694 d(NH) + d(C7–N–H)
1595 1598 1595 n(C–C) (phenyl ring)
1460 1467 — d(CH2) + d(CH3)
1390 1410 1407 d(NH) + d(CH) (phenyl ring)
1311 1325 — d(C2–N1–C7) + n(C4–C5)
1095 1117 1092 s(CH2)
1012 1011 999 Phenyl ring breathing
723 749 726 s(CH2) + s(C–H) (phenyl ring)
631 645 622 Phenyl ring out-of-plane def. + four-membered ring in-plane def.
DDS — 1693 1690 d(NH2)
— 1620 1634 n(C–C) (phenyl ring) + d(C–H) (phenyl ring) + u(NH2)
1593 1598 1599 n(C–C) (phenyl ring)
1470 1470 1466 d(NH2)
1445 1389 1442 s(NH2) + n(C2–N1)
1098 1070 1092 s(CH2)
997 992 999 Phenyl ring breathing
726 735 722 s(C–H) (phenyl ring)
631 644 632 Phenyl ring out-of-plane def. + four-membered ring in-plane def.
a
r, rocking; d, scissoring; u, wagging; s, twisting; n, stretching; def., deformation.

5890 | RSC Adv., 2015, 5, 5886–5894 This journal is © The Royal Society of Chemistry 2015
 View Article Online

Paper RSC Advances

Published on 12 December 2014. Downloaded by Carleton University on 12/05/2015 08:23:02.

Fig. 4 (A) SERS spectra of KPC with SIB, KPC with MDS and KPC with
DDS (2.5 mg g
1), (B) second derivative transformations of above
Fig. 3 The Raman spectra of SIB, MDS and DDS: (A) solid, (B) water spectra and (C) Cooman plot for the classification of KPC with SIB, KPC
solution and (C) in Ag NPs colloid. with MDS and KPC with DDS.

enhanced. The enhancement effects can be explained by two bonds in phenyl ring and CH3 scissoring vibration). By further
mechanisms: (1) the electromagnetic mechanism based on the inspecting spectra of MDS and DDS, we observed that there was
coupling of the radiation eld with surface plasmons of Ag NPs; one obvious feature peak at 1407 cm
1 (deformation vibrations
(2) the chemical mechanism based on a change in the molec- of the NH bond and CH bonds in phenyl ring) in spectrum of
ular polarizability arising from the interaction with the metal MDS.
substrate. As compared to MDS and DDS, SIB displayed two In order to apply SERS spectroscopy to differentiate the three
characteristic peaks including 1060 cm
1 (stretching vibration analytes even in complicated matrix, we have to identify the
of C2–N1–C8 bond) and 1415 cm
1 (stretching vibrations of C–C feature bands that allow an easy discrimination of SIB and its

This journal is © The Royal Society of Chemistry 2015 RSC Adv., 2015, 5, 5886–5894 | 5891
 View Article Online

RSC Advances Paper

Published on 12 December 2014. Downloaded by Carleton University on 12/05/2015 08:23:02.

Fig. 5 (A) Schematic drawing for the detection of SIB and its analogues from the KPC by the SERS spectroscopy using Ag NPs as Raman sensor,
and the SERS spectra of KPC spiked with (B) SIB, (C) MDS and (D) DDS by the enhancement of Ag NPs colloid.

analogues. In general, a good discriminant peak will be rela- compared the SERS spectra of KPC (blank), KPC with SIB, KPC
tively intense and isolated from nearby peaks which can with MDS and KPC with DDS (Fig. 4A), and discriminant peaks
possibly interfere. Such peaks can serve as markers to identify of each target analyte in KPC were determined. For KPC with
the presence of each compound in mixtures. Herein, we SIB, the peaks at 1060 and 1415 cm
1 were relatively uncluttered

Table 2 Real detections and recovery tests of SIB, MDS and DDS spiked in KPC, respectively (n ¼ 6)

Detected concentration (mg g
1) Recovery (%)

Sample Spiked concentration (mg g
1) Raman method LC-MS method Raman method RSD (%) LC-MS method RSD (%)

SIB 0 nda nda — — — —

1.25 1.24 1.27 99.2 4.19 102 4.44
2.50 2.47 2.52 98.8 3.40 101 3.50
3.75 3.74 3.69 99.7 2.16 98.4 2.08
MDS 0 nda nda — — — —
1.25 1.22 1.24 97.6 4.60 99.2 4.92
2.50 2.43 2.41 97.2 3.03 96.4 4.68
3.75 3.70 3.72 98.7 2.47 99.2 2.50
DDS 0 nda nda — — — —
1.25 1.24 1.25 99.2 4.73 100 4.15
2.50 2.45 2.44 98.0 3.56 97.6 3.41
3.75 3.71 3.73 98.9 2.55 99.5 2.36
a
Not detected.

5892 | RSC Adv., 2015, 5, 5886–5894 This journal is © The Royal Society of Chemistry 2015
 View Article Online

Paper RSC Advances

by nearby peaks, and served as discriminant peaks. For KPC In order to verify the usefulness of the proposed method
with MDS and KPC with DDS, the two spectra exhibited some for the identication and detection of three anorexics in
slight differences. Aer the second derivative transformation practical applications, the selectivity of the SERS method for
(Fig. 4B), the differences looked much more clear, especially for the analytes was rst measured using the structure analogues
the peak at 1407 cm
1 which was used to discriminate between (including genistein, daidzein, salidroside, chlorogenic acid,
MDS and DDS. These ndings demonstrated that SERS was able caffeic acid and gallic acid) because these aromatic
to differentiate spectral patterns between the three analogues in compounds (from herbs) might affect the detection of target
KPC samples. To be more intuitive, DA model, as a multivariate analytes in real sample analysis. As shown in Fig. S15,† it is
statistical tool, was also introduced based on the spectra of KPC clearly seen that there is the peak at 999 cm
1 in few SERS
spiked with SIB (n ¼ 50), KPC spiked with MDS (n ¼ 50) and KPC spectra of aromatic compounds, but there is not the peak in
Published on 12 December 2014. Downloaded by Carleton University on 12/05/2015 08:23:02.

spiked with DDS (n ¼ 50) without the second derivative trans- the SERS spectra of KPC (blank) (Fig. 4A). The result indicates
formation. DA accurately classied 100% of all samples either that a highly selective SERS strategy is attained with an
as the SIB or its two analogues. Results of DA were shown in a impressive specicity for three target analytes. Subsequently,
Cooman plot (Fig. 4C), which was constructed by plotting the the analysis of three anorexics in real samples was demon-
Mahalanobis distance between each anorexic drug. Eigen strated here by the use of the SERS sensor (Fig. 5). Similar to
analysis, one of the diagnostic checks, denes how much vari- three anorexics in water solution, PLS was also applied for
ation of spectral information is explained by the use of principal quantitative analysis of three anorexics spiked in KPC
components. For this case, eigen analysis revealed that it was (Fig. S16 and S17†). The relative data of recoveries were listed
possible to obtain 99.4% of desired information with only six in Table 2, which indicated that the obtained results were
principal components. The obtained results illustrated that highly consistent with those from the LC-MS method.
SERS combined with DA could be used to differentiate the three However, the SERS method does not need to pretreat samples
analytes intuitively. before assay, and the whole test can be accomplished within
3 min. These conrm that the SERS-based Ag NPs sensor
has a high accuracy and reliability to meet the requirements
3.4 SERS detections of SIB, MDS and DDS in spiked samples of practical application.
Before the detection of trace SIB, MDS and DDS in KPC, the
detection sensitivity to analytes in water using the SERS method
has been tested by the addition of 10 mL analyte aqueous solu- 4 Conclusion
tion into 90 mL of Ag NPs colloid. Fig. S12A† showed SERS
spectra of SIB. With the increase of SIB concentration from 5.00 We have developed a rapid and effective method which can
 10
8 M to 5.00  10
6 M, the Raman intensity gradually clearly discriminate between SIB and its analogues adulterated
increases. The intensity of the strongest peak at 1691 cm
1, in KPC based on SERS data, DA and second derivative trans-
assigned to u(C7–N–C8) + n(C–C) (phenyl ring), was used for the formation. Citrate-reduced Ag NPs were used as the SERS
quantitative evaluation of SIB level and the limit of detection substrate. Optimal excitation wavelength (532 nm), substrate
(LOD) was determined to be 5.00  10
8 M (equal to 16.7 ng concentration (0.2 nM), pH value (7.4) and aggregating agent
mL
1) from three standard deviations above the background. (KBr) for high Raman enhancement were obtained. The SERS
Meanwhile, similar tests were also performed for MDS and DDS. discrimination of SIB and its analogues was achieved by DA
As shown in Fig. S12B and C,† the LOD of MDS and DDS was method, mainly relying on the differences in Raman spectral
5.00  10
7 M (equal to 151 ng mL
1) and 1.00  10
6 M (equal window between 1000 and 1600 cm
1. SIB, MDS and DDS was
to 288 ng mL
1), respectively (herein, the intensity of the successfully quantied using PLS model, respectively, and the
stronger peak at 999 cm
1, assigned to phenyl ring breathing, LOD for SIB, MDS and DDS was 5.00  10
8 M (equal to 16.7 ng
was used for the quantitative evaluation of MDS and DDS mL
1), 5.00  10
7 M (equal to 151 ng mL
1) and 1.00  10
6 M
levels). Subsequently, PLS analysis was applied on spectra of (equal to 288 ng mL
1), respectively. Moreover, the recoveries of
three target analytes. RMSEP values obtained from the PLS SIB, MDS and DDS in the spiked KPC were in the ranges of 98.8–
model of SIB with different latent variables were shown in 99.7%, 97.2–98.7% and 98.0–99.2%, respectively. Compared
Fig. S13A.† The lowest RMSEP value was achieved when 4 latent with LC-MS method, this new technique was simple, rapid and
variables were used, suggesting that the optimal number of inexpensive (each test only needed a droplet of 10 mL sample
latent variables to construct a PLS model was 4. The optimal and was accomplished within 3 min), and thus can be suitable
number for building PLS model was 3 for MDS and 4 for DDS for high throughput screening analysis, which is performed
(Fig. S13B and C†). PLS prediction result for SIB was shown in with a motorized sample holder stage together with a 96 wells
Fig. S14A† by plotting calculated SIB concentrations against plate to make the method automatic and reduce the total
actual SIB concentrations (R ¼ 0.993; RMSEP ¼ 0.102 mM). For analysis time of batch samples. Although this work is mainly
MDS and DDS, R value was 0.989 and 0.981, and RMSEP value focused on the SERS analysis of SIB, MDS and DDS in KPC, we
was 1.08 mM and 1.22 mM, respectively (Fig. S14B and C†). With have demonstrated that the SERS technique is generally appli-
R values all larger than 0.980, results of PLS analysis suggested cable to detect SIB, MDS and DDS of other NSP such as HST
that PLS could serve as a reliable method to quantify SIB, MDS (Fig. S18–20 and Table S1†). Importantly, the monitoring and
and DDS. tracking of synthetic anorexic drugs illegally adulterated in NSP

This journal is © The Royal Society of Chemistry 2015 RSC Adv., 2015, 5, 5886–5894 | 5893
 View Article Online

RSC Advances Paper

may greatly reduce the possibility of occurrence of negative 19 E. G. Anastasaki, C. D. Kanakis, C. Pappas, L. Maggi,
effects on public health. A. Zalacain, M. Carmona, G. L. Alonso and M. G. Polissiou,
J. Agric. Food Chem., 2010, 58, 6011–6017.
Acknowledgements 20 F. J. Garcı́a de Abajo, Rev. Mod. Phys., 2007, 79, 1267–1290.
21 D. Kurouski, S. Zaleski, F. Casadio, R. P. Van Duyne and
This work is supported by the National Natural Science Foun- N. C. Shah, J. Am. Chem. Soc., 2014, 136, 8677–8684.
dation of China (no. 21275075) and the Specialized Research 22 H. Zhu, M. L. Du, M. Zhang, P. Wang, S. Y. Bao, M. Zou,
Fund for the Doctoral Program of Higher Education of China Y. Q. Fu and J. M. Yao, Biosens. Bioelectron., 2014, 54, 91–101.
(no. 20113234110001). 23 C. Yuen, W. Zheng and Z. W. Huang, Biosens. Bioelectron.,
2010, 26, 580–584.
Published on 12 December 2014. Downloaded by Carleton University on 12/05/2015 08:23:02.

References 24 L. F. Sallum, F. L. Felipe Soares, J. A. Ardila and

R. L. Carneiro, Talanta, 2014, 118, 353–358.
1 G. P. Luscombe, R. H. Hopcro, P. C. Thomas and 25 X. Jiang, Y. Lai, W. Wang, W. Jiang and J. Zhan, Talanta,
W. R. Buckett, Neuropharmacology, 1989, 28, 129–134. 2013, 116, 14–17.
2 A. Balcioglu and R. J. Wurtman, Neuropharmacology, 2000, 26 C. D. Bleye, E. Dumont, E. Rozet, P. Sacré, P. Chavez,
39, 2352–2359. L. Netchacovitch, G. Piel, P. Hubert and E. Ziemons,
3 C. Gundlah, K. F. Martin, D. J. Heal and S. B. Auerbach, J. Talanta, 2013, 116, 899–905.
Pharmacol. Exp. Ther., 1997, 283, 581–591. 27 B. H. Liu, G. M. Han, Z. P. Zhang, R. Y. Liu, C. L. Jiang,
4 I. P. Connoley, Y. L. Liu, I. Frost, I. P. Reckless, D. J. Heal and S. H. Wang and M. Y. Han, Anal. Chem., 2012, 84, 255–261.
M. J. Stock, Br. J. Pharmacol., 1999, 126, 1487–1495. 28 L. He, Y. Liu, M. Lin, J. Awika, D. R. Ledoux, H. Li and
5 S. P. Chen, M. H. Tang, S. W. Ng, W. T. Poon, A. Y. Chan and A. Mustapha, Sens. Instrum. Food Qual. Saf., 2008, 2, 66–71.
T. W. Mak, Clin. Toxicol., 2010, 48, 832–838. 29 B. Liu, M. Lin and H. Li, Sens. Instrum. Food Qual. Saf., 2010,
6 K. Stypułkowska, A. Błażewicz, J. Maurin, K. Sarna and 4, 13–19.
Z. Fijałek, J. Pharm. Biomed. Anal., 2011, 56, 969–975. 30 W. Cheung, L. T. Shadi, Y. Xu and R. Goodacre, J. Phys. Chem.
7 J. L. Jung, H. C. Maren and G. W. Wolf, Forensic Sci. Int., C, 2010, 114, 7285–7290.
2006, 161, 221–222. 31 K. Lai, Y. Zhang, R. Du, F. Zhai, B. A. Rasco and Y. Huang,
8 E. Deconinck, K. Verlinde, P. Courselle and J. O. De Beer, J. Sens. Instrum. Food Qual. Saf., 2011, 5, 19–24.
Pharm. Biomed. Anal., 2012, 59, 38–43. 32 Y. Y. Zhang, Y. Q. Huang, F. L. Zhai, R. Du, Y. D. Liu and
9 J. W. Bae, C. I. Choi, C. G. Jang and S. Y. Lee, Biomed. K. Q. Lai, Food Chem., 2012, 135, 845–850.
Chromatogr., 2011, 25, 1181–1188. 33 P. C. Lee and D. P. Meisel, J. Phys. Chem. C, 1982, 86, 3391–
10 S. Strano-Rossi, C. Colamonici and F. Botrè, Rapid Commun. 3395.
Mass Spectrom., 2007, 21, 79–88. 34 R. Kanjanawarut and X. Su, Anal. Chem., 2009, 81, 6122–
11 J. D. Dunn, C. M. Gryniewicz-Ruzicka, J. F. Kauffman, 6129.
B. J. Westenberger and L. F. Buhse, J. Pharm. Biomed. 35 N. Leopold and B. Lendl, J. Phys. Chem. B, 2003, 107, 5723–
Anal., 2011, 54, 469–474. 5727.
12 M. D. Likar, G. L. Cheng, N. Mahajan and Z. L. Zhang, J. 36 R. A. Álvarez-Puebla, J. Phys. Chem. Lett., 2012, 3, 857–866.
Pharm. Biomed. Anal., 2011, 55, 569–573. 37 C. E. Talley, L. Jusinski, C. W. Hollars, S. M. Lane and
13 F. W. Wan, J. H. Yu, S. G. Ge, M. Yan, K. X. Yu and M. Zhang, T. Huser, Anal. Chem., 2004, 76, 7064–7068.
Adv. Mater. Lett., 2010, 1, 164–169. 38 S. W. Bishnoi, C. J. Rozell, C. S. Levin, M. K. Gheith,
14 S. Wang, J. Yu, F. Wan, S. Ge, M. Yan and M. Zhang, J. B. R. Johnson, D. H. Johnson and N. J. Halas, Nano Lett.,
Lumin., 2011, 131, 1515–1519. 2006, 6, 1687–1692.
15 L. M. Carvalho, M. Martini, A. P. Moreira, S. C. Garcia, 39 S. F. Zong, Z. Y. Wang, J. Yang and Y. P. Cui, Anal. Chem.,
P. C. Nascimento and D. Bohrer, Microchem. J., 2010, 96, 2011, 83, 4178–4183.
114–119. 40 J. P. Schulte, S. Grass and L. Treuel, J. Raman Spectrosc., 2013,
16 F. Lu, S. Li, J. Le, G. Chen, Y. Cao, Y. Qi, Y. Chai and Y. Wu, 44, 247–254.
Anal. Chim. Acta, 2007, 589, 200–207. 41 X. Dong, H. M. Gu, J. Kang, X. J. Yuan and J. W. Wu, Colloids
17 X. N. Lu, H. M. Al-Qadiri, M. Lin and B. A. Rasco, Food Surf., A, 2010, 368, 142–147.
Bioprocess Technol., 2011, 4, 919–935. 42 A. P. Zhang and Y. Fang, Chem. Phys., 2007, 332, 284–288.
18 R. M. El-Abassy, P. Donfack and A. Materny, Food Chem.,
2011, 126, 1443–1448.

5894 | RSC Adv., 2015, 5, 5886–5894 This journal is © The Royal Society of Chemistry 2015