Journal of Chromatography A, 1423 (2015) 39–46

Contents lists available at ScienceDirect

Journal of Chromatography A
journal homepage: www.elsevier.com/locate/chroma

Cloud-point extraction is compatible with liquid chromatography

coupled to electrospray ionization mass spectrometry for the
determination of bisoprolol in human plasma
Joanna Giebułtowicz a,∗ , Grzegorz Kojro a,b , Katarzyna Buś-Kwaśnik c , Piotr J. Rudzki c ,
Ryszard Marszałek a , Andrzej Leś c,d , Piotr Wroczyński a
a
Bioanalysis and Drugs Analysis Department, Faculty of Pharmacy, Medical University of Warsaw, 1 Banacha Street, 02-097 Warsaw, Poland
b
Valeant sp. z o.o. sp. j., 15 Przemysłowa 2 Street, 35-959 Rzeszow, Poland
c
Pharmaceutical Research Institute, 8 Rydygiera Street, 01-793 Warsaw, Poland
d
Warsaw University, Faculty of Chemistry, 1 Pasteura Street, 02-093 Warsaw, Poland

a r t i c l e i n f o a b s t r a c t

Article history: Cloud-point extraction (CPE) draws increasing interest in a number of analytical fields including bioanal-
Received 26 May 2015 ysis, but combining CPE and LC–MS with electrospray ionization (ESI) in the determination of drugs in
Received in revised form 21 October 2015 biological fluids such as plasma, serum or blood has not been reported so far. Bisoprolol was deter-
Accepted 25 October 2015
mined in human plasma by CPE using Trition X-114 as a surfactant and metoprolol as the internal
Available online 29 October 2015
standard. NaOH concentration, temperature and Trition X-114 concentration were optimized. All anal-
yses were performed using liquid chromatography–electrospray ionization-tandem mass spectrometry
Keywords:
(LC–ESI–MS/MS). All validation experiments met international acceptance criteria and no significant
Cloud-point extraction
Pharmacokinetics matrix effect was observed. The compatibility of CPE and LC–ESI–MS/MS was confirmed using clinical
Sample preparation plasma samples and appropriate statistical tests. The determination of bisoprolol concentration in human
Bioanalysis plasma in the range 1.0–70 ng mL−1 by the CPE method leads to the results which are equivalent to those
Liquid chromatography coupled to mass obtained by the widely used liquid–liquid extraction method. The results revealed that a structural ana-
spectrometry logue may be an appropriate internal standard when CPE is used as the extraction technique. CPE offers
significant practical advantages over the classical extraction methods, including a positive impact on the
environment, therefore its wider application in future pharmacokinetic studies is justifiable.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction significant advantages over traditional liquid–liquid extraction

The quantitative analysis of small molecule drugs in human
plasma is used in many fields of pharmacy and medicine, e.g. the
results of pharmacokinetic studies are crucial for the drug design
and development of new medicines. They indicate proper dosing
scheme and provide the final proof of the drug’s efficacy, safety
and quality during bioequivalence studies. The introduction of
liquid chromatography coupled to mass spectrometry (LC–MS/MS)
and ultra-high performance liquid chromatography (UHPLC) has
caused vast progress in the instrumentation of bioanalysis in
the recent years. However, the advances in the commonly used
sample preparation techniques do not follow. There is a clear need
for novel approaches to sample preparation which would have
∗ Corresponding author. Tel.: +48 22 572 09 49; fax: +48 22 572 09 76.
E-mail addresses: joanna.giebultowicz@wum.edu.pl (J. Giebułtowicz),
gkojro72@wp.pl (G. Kojro).
(LLE) or solid phase extraction (SPE) methods.
Cloud-point extraction (surfactant-mediated extraction, CPE)
draws increasing interest in the number of analytical fields includ-
ing bioanalysis [1]. This technique is based on such low surfactant
concentrations that allows the surfactant to exist as a homoge-
neous isotropic liquid phase. This phase separates into two isotropic
phases containing a surfactant in different concentrations. In the
surfactant micellar-rich phase the hydrophobic organic compo-
nents present in the sample are concentrated and separated from
more hydrophilic compounds [2]. CPE is considered a simple, safe
and environment-friendly sample preparation technique which
requires neither organic solvents nor expensive equipment.
There have been numerous reports on the CPE application for the
determination of drugs in biological fluids such as plasma, serum
or blood [1,3–16]. However, to the best of our knowledge, the com-
bination of CPE and LC–MS with electrospray ionization (ESI) in the
analysis of those biological fluids has not been reported until now.
Other methods used UV–vis [1,5,7,9–13] or fluorescence detectors

http://dx.doi.org/10.1016/j.chroma.2015.10.076
0021-9673/© 2015 Elsevier B.V. All rights reserved.
40 J. Giebułtowicz et al. / J. Chromatogr. A 1423 (2015) 39–46
[16] and the only use of CPE and LC–MS combination was reported
[3] for APCI as the ionization source. LC–ESI–MS/MS is the state-
of-the-art bioanalytical technique and is more frequently used in
the bioanalysis of drugs than LC–APCI–MS/MS. However, the sur-
factants introduced to the sample may cause matrix effects and ESI
is more sensitive to this phenomena than APCI [17]. Therefore, as
regards further applications of CPE in the bioanalysis of drugs, it is
of critical importance that the possibility of its compatibility with
LC–ESI–MS/MS is evaluated.
Bisoprolol is a selective ␤-1-adrenoceptor antagonist bereft of
any partial agonist effect [18]. Taking into account the social impor-
tance of bisoprolol, novel methods for its rapid, convenient and
environmentally safe determination in human plasma are still
needed. Previously reported methods of LC–MS/MS detection of
bisoprolol [19–28] did not use CPE as serum or plasma prepara-
tion technique. The most common one was protein precipitation
by the addition of organic solvents (e.g. acetonitrile) [20,21,24–26].
Although this technique is rapid and simple, its disadvantages
result from insufficient sample clean-up and include a shorter HPLC
column life, increased risk of significant matrix effects and con-
tamination of a mass spectrometer [29]. The SPE procedures using
Oasis polymeric mix sorbent cartridges were also applied for the
determination of bisoprolol alone or among other beta-antagonists
and beta-agonists [28]. SPE based on the hydrophilic-lipophilic bal-
anced sorbent beds is successfully used in the multi-analysis of
compounds of different polarity [30]. However, this technique may
be considered time consuming and rather expensive, especially if
there is a need for analyzing large numbers of samples in a pharma-
cokinetic study. LLE was also successfully applied in the bioanalysis
of bisoprolol [19,22,23]. This technique is often used in bioanaly-
sis as it is fast and inexpensive. Our previous experience confirms
that its application allows to avoid the contamination of an MS
ion source and extends the duration of an HPLC column [29,31,32].
However, LLE requires the use of organic solvents which are toxic
to laboratory staff and the environment, as well as expensive to dis-
pose of. CPE is an alternative technique with all the advantages of
LLE. Additionally, CPE is regarded safer and more environmentally
friendly due to the extraction being based on surfactants which are
less toxic and less expensive to dispose of [33]. An extensive search
did not return any reports on the application of CPE to the biso-
prolol determination nor the statistical comparison of CPE and the
well-established sample preparation techniques.
In this paper we introduce a novel combination of sample prepa-
ration with the separation and detection techniques. The aim of the
study was to assess the compatibility of CPE and LC–ESI–MS/MS and
the applicability of CPE to the bisoprolol determination in spiked
human plasma and clinical samples.
2. Materials and methods
2.1. Chemicals
The reference standard of bisoprolol fumarate was a kind gift
from ICN Polfa Rzeszów S.A. (Poland). Metoprolol tartrate (internal
standard, IS) and Trition X-114 was purchased from Sigma-Aldrich
(St. Louis, US). The solvents, HPLC gradient grade methanol, acetoni-
trile and formic acid 98% were purchased from Merck (Darmstadt,
Germany). Ultrapure water was obtained from the Millipore water
Table 1
Optimized MRM conditions and ion transitions.
Substance [M + H] + ion DP [V] EP [V]
Bisoprolol 326 91 10
Metoprolol (IS) 268 86 10
DP—declustering potential, EP—entrance potential, CE—collision energy, CXP—cell exit potential.
purification system (Milli-Q, Billerica, US). Blank plasma was
obtained from the Regional Center of Blood Donation and Treat-
ment (Warsaw, Poland).
2.2. Chromatographic and mass spectrometric conditions
The instrumental analysis was performed using Agilent 1260
Infinity (Agilent Technologies, Santa Clara, CA, US), equipped with
a degasser, autosampler, binary pump coupled to a hybrid triple
quadrupole/linear ion trap mass spectrometer QTRAP 4000 (AB
Sciex, Framingham, MA, US). The Turbo Ion Spray source was oper-
ated in the positive mode. The curtain gas, ion source gas 1, ion
source gas 2 and collision gas (all high purity nitrogen) were set
at 345 kPa, 207 kPa, 276 kPa and “high” instrument units, respec-
tively. The ion spray voltage and source temperature were 5000 V
and 600 ◦ C, respectively. The target compounds were analyzed in
the MRM mode (Table 1).
Chromatographic separation was achieved with a Symmetry
C18 column (150 mm × 4.6 mm, 3.5 ␮m, Waters, Milford, US). The
column was maintained at 40 ◦ C at the flow rate of 0.5 mL min−1 .
The mobile phases consisted of HPLC grade water with 0.1% formic
acid as the eluent A and methanol with 0.1% formic acid as the elu-
ent B. The gradient (%B) was as follows: 0 min. 10%; 1 min. 10%;
8 min. 90%, 16 min. 90%. The re-equilibration of the column to the
initial conditions lasted 4 min. The injection volume was 10 ␮L.
2.3. Standard solutions, calibration standards and quality control
samples
The standard stock solutions of bisoprolol and IS (1 mg mL−1 )
were prepared by dissolving an appropriate amount of the sub-
stance in methanol. The working solutions (10 ng mL−1 ) were
prepared by diluting standard stock solutions with water. The solu-
tions were stored at 4 ◦ C. The calibration standards for bisoprolol
were made at the following concentrations: 0.3, 0.9, 3.0, 10, 20, 30,
45, 60 and 70 ng mL−1 . The quality control samples were prepared
at the following levels: 0.9, 30 and 60 ng mL−1 . The calibration stan-
dards and quality control samples were stored at −80 ◦ C until use.
Sodium citrate was used as an anticoagulant.
2.4. Plasma samples from the pharmacokinetic study
Plasma samples from the pharmacokinetic study in humans
were obtained from the Pharmaceutical Research Institute [34].
All samples were previously analyzed by LLE and the LC–MS/MS
method adapted from Ding et al. [35]. Briefly, plasma sample (1 mL)
was mixed with the IS (metoprolol, 50 ␮L), 1 M NaOH (0.1 mL) and
shaken for 5 min with ethyl acetate (5 mL). After centrifugation the
organic layer was evaporated to dryness. The residue was recon-
stituted in 60% aqueous methanol (200 ␮L) and 30 ␮L aliquot was
injected onto the LC–MS/MS consisting of Alliance 2695 liquid chro-
matograph (Waters, Milford, MA, USA) coupled to a Quattro micro
API triple quadrupole mass spectrometer equipped with an electro-
spray ion source (Waters, Manchester, UK). The method was fully
validated according to the international guidelines [36] and all the
analysis were performed in accordance with the Good Laboratory
Practice (GLP).
Quantitative CE [V] CXP [V] Qualitative CE [V] CXP [V]
product ion product ion
116 27 8 222 21 12
116 27 8 133 37 10
 J. Giebułtowicz et al. / J. Chromatogr. A 1423 (2015) 39–46
2.5. Sample preparation
2.5.1. CPE
50 ␮L of the IS working solution, 300 ␮L of 7% Triton X-114
solution and 400 ␮L of 0.2 M NaOH were added to 250 ␮L plasma
sample and vortexed for 5 min. Next, the sample was incubated in
a water bath at 55 ◦ C for 20 min and centrifuged at 10 000 rpm at
25 ◦ C for 5 min. Then the aqueous phase was separated by decant-
ing. After the CPE 700 ␮L of acetonitrile was added to the micellar
phase. Alternatively to create more environmental friendly method
ethanol can be used in that step as well. The sample was incubated
at −20 ◦ C for 20 min and centrifuged at 10 000 rpm at 25 ◦ C for
5 min. The supernatant (500 ␮L) was diluted with an equal volume
of water (500 ␮L) and transferred to the vial. Ten ␮L aliquot was
injected onto the LC–MS/MS. All CPE experiments were performed
at the Bioanalysis and Drugs Analysis Department of the Medical
University of Warsaw.
2.5.2. LLE (LLE2)
The LLE2 experiments were performed at the Bioanalysis and
Drugs Analysis Department of the Medical University of War-
saw. Plasma sample (250 ␮L) was mixed with the IS (metoprolol,
50 ␮L), 1 M NaOH (25 ␮L) and shaken for 5 min with ethyl acetate
(1.25 mL). After centrifugation the organic layer was evaporated to
dryness. The residue was reconstituted in 60% aqueous methanol
(250 ␮L) and 10 ␮L aliquot was injected onto the LC–MS/MS.
2.6. Method validation
The method previously developed at the Pharmacology Depart-
ment of the Pharmaceutical Research Institute was fully validated
and the stability of bisoprolol was confirmed in various conditions
[34]. Therefore, only partial validation was performed for this study
according to the European Medicines Agency (EMA) [37] and US
Food and Drug Administration (FDA) [36] guidelines. Briefly, the lin-
earity range was selected as 0.3–70 ng mL−1 as described previously
[34]. The accuracy and precision of the method were determined
within-run and between-run using LLOQ (0.3 ng mL−1 ) and qual-
ity control (QC) samples (0.9, 30 and 60 ng mL−1 ). Carry over and
recovery was also studied.
As the aim of the study was to evaluate the compatibility of
CPE and LC–ESI–MS/MS, particular attention was paid to the matrix
effect assessment. Ion suppression or enhancement was initially
studied by the steady post-column infusion of the analyte, i.e. biso-
prolol or the IS at the concentration 60 ng mL−1 and the injection
of the extracted blank plasma sample. The second approach was
based on the method by Matuszewski et al. and allowed the quan-
titative assessment of the absolute and relative matrix effect by the
comparison of the analyte peak areas in the standard solutions and
Fig. 1. Optimization of the CPE method (n = 5). Experimental conditions: (A) 7% Triton X-114, 0.5 M NaOH; (B) extraction temperature at 55 ◦ C, 0.5 M NaOH; (C) extraction
temperature at 55 ◦ C, 7% Triton X-114.
41
post-extraction spiked samples. Various lots of matrix used for the
test included haemolyzed and hyperlipidaemic plasma samples.
2.7. Statistical analysis
The statistical evaluation of the results was performed with the
SAS software version 9.4. for Windows. The Shapiro–Wilk test was
used to evaluate normal distribution of QC results. The hypothesis
on the equality of variances was verified by the F-test (p ≥ 0.05).
The linear regression for bisoprolol concentration measured using
the CPE and LLE methods was calculated using the least squares
method, the 95% confidence limit band of the mean response at
particular LLE points and the 95% prediction interval limits.
3. Results and discussion
3.1. Optimization of the CPE procedure
Triton X-114 is more frequently chosen for the analysis of bio-
logical fluids [1,8–10,12,16,20,38–40] than other surfactants such
as Triton X-100 and Genapol X-080. All experiments were per-
formed using Triton X-114 due to its main advantages namely low
cloud-point temperature (23 ◦ C) and high density which facilitate
phase separation by centrifugation [16]. During the optimization of
the CPE conditions the following changes in the sample preparation
protocol were made: plasma samples were fortified with bisoprolol
and water instead of the IS working solution. The IS working solu-
tion was added in the last step of the extraction process instead of
water.
3.1.1. Effect of the equilibrium temperature
The CPE is based on the separation of two phases, when the
aqueous solutions of the surfactant are heated above the cloud-
point temperature. Theoretically, the optimal temperature for the
extraction is 15–20 ◦ C greater than the cloud-point of the surfac-
tant. With the increase of the equilibrium temperature, the volume
of the surfactant-rich phase decreases due to the disruption of the
hydrogen bonds and the dehydration of the phase [16]. However,
too high temperature may reduce the recovery—because of both the
decomposition of the heat labile compounds as well as the thermal
instability of the surfactant aggregates [8]. The range of 40–55 ◦ C
was selected for the optimization experiments. Due to a possi-
ble analyte instability, thermal stability problems of the surfactant
aggregates and denaturation of plasma proteins it was decided
not to perform experiments in higher temperatures. The analyti-
cal signal of bisoprolol increased with increasing the equilibrium
temperature, thus 55 ◦ C was selected as the working equilibrium
temperature (Fig. 1A). One can observe that standard deviation
for selected temperature was the highest one, still during method
validation acceptance criteria for precision were met.
42 J. Giebułtowicz et al. / J. Chromatogr. A 1423 (2015) 39–46

3.1.2. Effect of the surfactant concentration Table 2

Accuracy and precision of bisoprolol determination by CPE and LLE.
Increasing surfactant concentration improves the recovery, but
above a certain point it dilutes the compound of interest and Nominal bisoprolol concentration [ng/ml] 0.9 30.0 60.0
decreases the preconcentration rate [12]. The concentrations of Within-run accuracy [%]
Triton X-114 were optimized in the range of 5–20% which corre- CPE (n = 5) 89.9a 91.4 98.2
sponded to the concentration of 1.5–6% in the extraction mixture. LLE (n = 6) 107.9 97.6 95.2
No significant correlation between the bisoprolol peak area and the LLE2 (n = 5) 109.3 96.7 102.5
Within-run precision [%]
surfactant concentration was observed (Fig. 1B). It was observed
CPE (n = 5) 12.36a 3.55 3.98
that the surfactant concentration over 7% resulted in a lower signal LLE (n = 6) 3.62 4.08 2.70
of bisoprolol, probably due to a higher phase volume ratio (vol- LLE2 (n = 5) 8.67 6.42 2.13
ume of the surfactant-rich phase/volume of the aqueous solution) Between-run accuracy [%]
CPE (n = 15) 95.1b 93.1 98.4
resulting in the dilution of the extracted bisoprolol. Thus, based on
LLE (n = 9) 107.1 98.6 93.6
the results of the experiment, 7% of Triton X-114 was selected to LLE2 (n = 5) 104.4 93.3 100.5
achieve the best analytical signals. Between-run precision [%]
CPE (n = 15) 12.60b 6.30 3.80
3.1.3. Effect of the NaOH concentration LLE (n = 9) 3.67 5.03 6.50
LLE2 (n = 5) 10.44 8.24 4.36
The pH optimization is an important step in the CPE method
development. For organic molecules, especially ionizable ones, the CPE and LLE2 was performed in Medical University of Warsaw; LLE was performed
in Pharmaceutical Institute.
maximum extraction efficiency is achieved at those pH values a
n = 6.
where the uncharged form of the analyte prevails and therefore b
n = 16.
the analyte is favored to be partitioned into the micellar phase
[10]. Since bisoprolol is a weak base, the NaOH concentrations were concentration of bisoprolol. The values of regression parameters
tested in the range of 0.05–0.5 M that corresponded to the concen- for the curve, described by the equation: y = ax + b, were calculated
tration of 0.02–0.2 M in the extraction mixture. The best result was as: a = 0.00509 (0.00012), b = 0.00802 (0.00085) and r2 = 0.999. All
obtained when 0.2 M NaOH was added (Fig. 1C), however, the dif- regression parameters were statistically significant (p < 0.05). The
ferences between the analyte signal in 0.2 and 0.5 M NaOH were accuracy for LLOQ was 105% (RSD = 12%, n = 5) within one day and
within the measurement uncertainty. The phenomenon of a lower 107% (RSD = 9.8%, n = 15) within three runs. The accuracy for the QC
bisoprolol signal in higher NaOH concentrations might be caused samples (Table 2) within one day ranged 89–98% (RSD 3.6–13%)
by the co-extraction of the interfering compounds at a higher pH. while between-runs ranged 93–98% (RSD 3.8–13%). The carry-over
experiment in which blank human plasma samples were analyzed
3.2. Influence of the surfactants and selection of internal standard after the highest calibration standards did not show any signals
influencing quantification.
It is well known that the surfactants might seriously influence The recovery for bisoprolol in CPE was moderate and was
the accuracy and precision of bioanalytical methods. One of the rea- 46% (RSD = 12%) for high concentrations (60 ng mL−1 ) and 61%
sons might be difficulty of pipetting viscous liquids which results (RSD = 14%) for low concentrations (0.9 ng mL−1 ). The correspond-
in a relatively high uncertainty of the extraction yield of the com- ing recovery for LLE was 86% for high concentrations (60 ng mL−1 )
pound. Too low volume of the surfactant-rich phase might also and 74% for low concentrations (0.9 ng mL−1 ). In CPE the extraction
result in an unsatisfactory accuracy of the analytical method. This efficiency can be improved using salts like NaCl or Na2 SO4. These
can be overcome by the use of an internal standard. However, in salts are not compatible with LC–MS/MS technique and may cause
the case of methods coupling CPE and mass spectrometry there high matrix effect. Moreover, nonvolatile salts can deposit in source
is a problem of the surfactant’s negative effect on such ionization and plug capillaries thus requiring more cleaning and maintenance
types as MALDI or ESI [41]. Therefore, in CPE a very careful eval- operations. Alternatively violate salts like ammonium acetate or
uation of matrix effects should be an essential part of the method formate can be used, when low recoveries result in failing the vali-
validation. In order to reduce the matrix effect the samples were dation requirements for LLOQ. However, these salts has never been
diluted with acetonitrile and water after the extraction and the gra- used previously in CPE.
dient chromatographic separation was developed. No changes in During the evaluation of the matrix effect by the post-column
the quality of MS spectra nor in chromatograms after CPE exper- infusion of the analyte, the injection of the extracted matrix
iments were observed. There was no indication for additional ESI resulted in similar changes of the signal for both bisoprolol and
source cleaning. metoprolol (Fig. 2). Moreover, near the retention time of the com-
Generally, isotope-labeled compounds are recommended for pounds the signal was barely affected by the matrix and rather sta-
the LC–MS bioanalytical methods as internal standards to reduce ble. The observation was confirmed by the quantitative evaluation
matrix effects and increase the reliability of the results [37]. For this of the matrix effect studied in the bisoprolol concentrations of 1.0
study a compound of similar structure, i.e. metoprolol, was selected and 50 ng mL−1 using the calculation method by Matuszewski et al..
as the internal standard. This internal standard was proved to be an The absolute matrix effect for bisoprolol and the IS were 98% and
appropriate standard both in the SPE [27] and LLE extraction [22,34] 101% for the lower concentration and 105% and 108% for the higher
of bisoprolol from plasma. The structural analogue was selected for concentration, respectively. The relative matrix effect was assessed
this study to assess the compliance of CPE and LC–MS in more dif- as 11.5% for 1.0 ng mL−1 and 4.4% for 50 ng mL−1 , respectively.
ficult conditions than when the isotope labeled internal standard During the experiment we have also tested the influence of
is applied. NaOH and Triton concentrations on the absolute matrix effect. The
absolute matrix effect varied from 97% to 112% (RSD = 4.3%), but
3.3. Method validation no significant trends were observed. The phenomenon might be
a results of adequate chromatographic separation of analytes and
All validation experiments met the acceptance criteria. The inferences like, e.g., Triton. Thus, in CPE this aspect of analysis is
calibration curve obtained by the weighted linear regression probably more important than in other extraction techniques. In
analysis was linear in the range of 0.3–70 ng mL−1 regarding further study we aim to explore the effect of pH and chromato-
the peak area ratio of bisoprolol and the IS versus the nominal graphic separation on the absolute matrix effect in CPE.
 J. Giebułtowicz et al. / J. Chromatogr. A 1423 (2015) 39–46 43

XIC of +MRM (7 pairs): 268.133/116.100 Da ID: metoprolol 8 from Sample 3 (matryca i 1' 60ng/ml bisopr i metoprolol) of 2014-04-30bisoprolol i m... Max. 3341.7 cps.

18.98
3342

3000 OH 12.19 13.87

H 17.93 18.59
O N
2500
+ 11.90
H
IS
Intensity, cps

2000 11.65 17.66

O 15.02
1500 11.26
16.98

1000 8.68 10.72

3.33 4.07 8.22 8.95
0.92 1.17 2.14 2.33 3.19 4.81 5.10
6.59 7.11 7.30
500

0
1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 17.0 18.0 19.0
Time, min
XIC of +MRM (7 pairs): 326.259/116.000 Da ID: bisoprolol 1 from Sample 3 (matryca i 1' 60ng/ml bisopr i metoprolol) of 2014-04-30bisoprolol i met... Max. 4.8e4 cps.

OH 13.87 19.24
4.8e4 H 12.84 18.98
4.5e4 O N
18.19
4.0e4 11.77
+
3.5e4 H

3.0e4
O
Bisoprolol 11.32
17.25
Intensity, cps

15.05
2.5e4 O 11.13 16.99

2.0e4 9.13 9.49

2.33 3.33 4.26 4.42 6.97 7.26 8.24 8.91
1.5e4 0.26 1.28 2.13 3.19
1.0e4

5000.0

0.0
1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 17.0 18.0 19.0
Time, min

Fig. 2. Chromatograms recorded during the evaluation of the matrix effect by the steady post-column infusion and the injection of the extracted blank plasma sample for
metoprolol (internal standard, top) and bisoprolol (bottom).

These results indicate that CPE is a suitable technique for sample results. The statistical evaluation by the F-test indicated that the
preparation prior to the LC–ESI–MS/MS analysis as no significant hypothesis of the equality of variances was rejected for the low QC
matrix effect was observed. However, the chromatographic sep- (p = 0.003), as opposed to the medium and high QCs (p = 0.647 and
aration is probably crucial to reduce the matrix effect caused by p = 0.106, respectively). Therefore, one can conclude that for lower
the surfactant. Similar results of a negligible matrix effect in the concentrations, around 0.9 ng mL−1 , the precision of a bioanalytical
CPE–LC–MS method were observed by Liu et al. [3] using the atmo- method is significantly lower when CPE is applied instead of LLE.
spheric pressure chemical ionization (APCI) source. The results of To confirm that hypothesis we performed additional analysis
the validation also revealed that the structural analogue may be an using LLE method in the same laboratory, where CPE experiments
appropriate internal standard when CPE is used as the extraction were conducted. Due to instrumental differences we modified LLE
technique. method (LLE2) as described in Section 2.5.2. The results confirmed
our previous observation that in lower concentration (low QC) CPE
has significantly lower precision than LLE2 method (p = 0.042).
3.4. Comparison of the CPE and LLE methods

Surprisingly, there were only two reports in literature regarding

3.4.2. Application of the CPE method to clinical samples
the comparison of the CPE and LLE methods for the determination of
The final assessment of the compatibility of CPE and
drugs in biological matrices. However, the comparisons were based
LC–ESI–MS/MS was performed using 28 clinical plasma sam-
on the presentation of a chromatogram only in Zhou et al. [40] or
ples selected from different subjects and representing both high
in the brief description (including linearity, accuracy, precision and
(around maximum plasma concentration Cmax ) and low (in the
recovery) by Zhang et al. [11]. An extensive search did not find any
elimination phase) bisoprolol concentrations. Representative chro-
statistical comparisons of both methods; the lack of such analyses
matogram is presented in Fig. 3. A strong linear correlation of the
is contrary to the current trends in chemical metrology. To fill this
LLE and CPE data is evidenced by the Pearson’s correlation coef-
niche, a cross-validation of the bioanalytical methods using CPE and
ficient of 0.9802 (n = 28, SD = 0.0076) as presented in Fig. 4. The
LLE, supported by a detailed statistical evaluation, was performed
compatibility of CPE and LLE was further supported by the fact
[37,42,43]. The difference in the sample preparation techniques
that only 2 of 28 (7%) observations were located outside the 95%
was assumed to be the main source of variability. Other factors pos-
prediction limits.
sibly influencing the method’s performance, such as using different
A more detailed evaluation was based on the approach similar to
laboratories and analysts, different sample preparation techniques
the incurred sample reanalysis. The LLE results were considered as
as well as different LC–MS/MS methods and instruments, were con-
initial values, while the CPE results were considered as repeat val-
sidered less significant. The original data from the bioanalytical
ues. The acceptance criteria of 67% of the CPE results within ±20%
method full validation and pharmacokinetic study obtained pre-
of the mean were defined according to the cross validation part
viously in a GLP certified laboratory for the regulatory purposes
of the EMA guideline [37]. It was observed that 75% of 28 results
were used as reference [34].
met the acceptance criteria, however, the majority of the results
with % difference exceeding ±20% were found among the lower
3.4.1. Statistical evaluation of the validation parameters concentration (Table 3, Fig. 5). When data was limited to the LLE
Both methods met the acceptance criteria for accuracy and pre- results with the measured bisoprolol concentration not lower than
cision at all QC concentration levels (Table 2), confirming that 1.0 ng mL−1 , the statistics were far better: 91% of 22 results met the
both sample preparation techniques are capable to deliver reliable acceptance criteria, the mean %difference was lower and the SD of
44 J. Giebułtowicz et al. / J. Chromatogr. A 1423 (2015) 39–46

XIC of +MRM (7 pairs): 326.259/116.000 Da ID: bisoprolol 1 from Sample 16 (18/15) of 2014-04-17bisoprolol pacjencji.wiff (Turbo Spray) Max. 2016.7 cps.

8.36
2000

1900 Bisoprolol
1800

1700

1600

1500

1400

1300

1200
In te ns ity , c p s

1100

1000

900

800

700

600

500

400

300

200

100

0
1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 17.0 18.0 19.0
Time, min

XIC of +MRM (7 pairs): 268.133/116.100 Da ID: metoprolol 8 from Sample 16 (18/15) of 2014-04-17bisoprolol pacjencji.wiff (Turbo Spray) Max. 7.6e4 cps.

7.67
7.5e4

7.0e4 IS
6.5e4

6.0e4

5.5e4

5.0e4

4.5e4
In te ns ity , c p s

4.0e4

3.5e4

3.0e4

2.5e4

2.0e4

1.5e4

1.0e4

5000.0

0.0
1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 17.0 18.0 19.0
Time, min

Fig. 3. Chromatogram of sample obtained from healthy volunteer after 5 h of oral administration of bisoprolol fumarate. The determined concentration was 2.4 ng mL−1 . The
blue line represents the main ion transitions m/z 326 → 116 and m/z 268 → 116 for bisoprolol (top) and metoprolol (internal standard, bottom). The red line represents the
second ion transition specific m/z 326 → 222 and m/z 268 → 133 for bisoprolol and metoprolol, respectively.
 J. Giebułtowicz et al. / J. Chromatogr. A 1423 (2015) 39–46 45

60 60%
y = 0.9646x + 1.2289
R² = 0.9608 40%
50

% difference
CPE result [ng/ml]

20%
40
0%
30
-20%
0 10 20 30 40 50 60
20 LLE result [ng/ml]

Fig. 5. The %difference, i.e. (CPE result—average result)/average result × 100%, plot-
10 ted against bisoprolol concentrations measured using LLE.

0 The observations from the clinical samples analysis were in

0 10 20 30 40 50 60 line with the result of the statistical evaluation of the meth-
LLE result [ng/ml] ods’ precision. Therefore, it was concluded that the determination
of bisoprolol concentration in human plasma in the range
Fig. 4. Correlation of bisoprolol concentrations in clinical samples measured using 1.0–70 ng mL−1 by the CPE method leads to the results which are
CPE and LLE (n = 28). equivalent to those obtained by the widely used LLE method.

4. Conclusion
%difference decreased below 15%. The positive value of the mean
%difference may indicate that the application of CPE may lead to
It has been experimentally proved that cloud-point extraction
higher concentrations measured than in the case of LLE; however,
is compatible with LC–ESI–MS/MS. This novel sample prepara-
this difference seems insignificant, as in both data sets, i.e. n = 28
tion technique was successfully applied for the first time for the
and n = 22, the mean %difference was below 15%.
bisoprolol determination in human plasma. The bisoprolol plasma
concentrations in the clinical samples obtained by the CPE and LLE
methods were comparable in the range 1.0–70 ng mL−1 . The results
Table 3
Comparison of LLE and CPE results for clinical plasma samples.
indicate that further research on combining CPE and LC–MS is rec-
ommended and may lead to significant advances in the bioanalysis
Sample no. LLE result CPE result %Difference of drugs. The study provides the first report on the statistical com-
[ng/ml] [ng/ml]
parison of CPE and the conventional extraction method. CPE offers
1 1.51 2.60 27% significant practical advantages and can have a positive impact on
2 0.30 0.54 28%
the environment, therefore its wider application in future pharma-
3 0.30 0.56 30%
4 38.82 45.10 7% cokinetic studies is justifiable.
5 33.71 42.90 12%
6 33.71 36.10 3%
Conflict of interest statement
7 30.04 30.60 1%
8 3.26 4.14 12%
9 1.96 2.14 4% The authors declare no conflict of interest regarding the content
10 0.80 2.00 43% of this article.
11 24.31 23.10 −3%
12 36.44 36.30 0%
13 53.07 47.40 −6% Acknowledgments
14 43.43 47.50 4%
15 2.49 4.60 30% The authors are grateful to all investigators of the former biso-
16 6.90 7.02 1%
17 2.60 2.37 −5%
prolol bioequivalence study [34] for kindly providing their clinical
18 0.70 2.13 51% samples for the additional analysis.
19 46.44 42.60 −4%
20 42.73 47.50 5%
21 44.71 39.20 −7%
References
22 2.97 3.18 3%
23 0.89 1.30 19% [1] K. Madej, K. Persona, M. Wandas, E. Gomółka, Sequential cloud-point extraction
24 0.35 0.59 26% for toxicological screening analysis of medicaments in human plasma by high
pressure liquid chromatography with diode array detector, J. Chromatogr. A
25 13.34 15.10 6%
1312 (2013) 42–48.
26 28.11 33.00 8%
[2] K. Madej, K. Persona, Drug screening in human plasma by cloud-point extrac-
27 44.20 33.80 −13%
tion and HPLC, Cent. Eur. J. Chem. 11 (2013) 94–100.
28 37.95 36.70 −2% [3] W. Liu, K. Bi, X. Liu, J. Zhao, X. Chen, Cloud-point extraction combined with
All data n= 28 LC–MS for analysis of memantine in rat plasma, Chromatographia 69 (2009)
837–842.
% of samples meeting acceptance criteria 75%
[4] N. Gao, H. Wu, Y. Chang, X. Guo, L. Zhang, L. Du, Y. Fu, Mixed micelle cloud
mean difference 10%
point-magnetic dispersive ␮-solid phase extraction of doxazosin and alfuzosin,
SD of difference 16%
Spectrochim. Acta A: Mol. Biomol. Spectrosc. 134 (2015) 10–16.
CPE result over 1 ng/ml n= 22 [5] R. Heydari, N.S. Elyasi, Ion-pair cloud-point extraction: a new method for the
% of samples meeting acceptance criteria 91% determination of water-soluble vitamins in plasma and urine, J. Sep. Sci. 37
(2014) 2724–2731.
Mean difference 4%
[6] F. Nazari Serenjeh, P. Hashemi, M. Safdarian, Z. Kheirollahi, Semi-automated
SD of difference 10%
cloud point extraction with cold column trapping of surfactant-rich phase for
46 J. Giebułtowicz et al. / J. Chromatogr. A 1423 (2015) 39–46
phenazopyridine determination in human serum, J. Iran. Chem. Soc. 11 (2014)
733–739.
[7] Z. Pourghobadi, R. Heydari, R. Pourghobadi, M. Rashidipour, Determination of
gabapentin in human plasma using simultaneous cloud point extraction and
precolumn derivatization by HPLC, Monatsh. Chem. Chem. Mon. 144 (2013)
773–779.
[8] H. Ma, J. You, Y. Liu, Cloud-point extraction combined with HPLC for deter-
mination of larotaxel in rat plasma: a pharmacokinetic study of liposome
formulation, J. Sep. Sci. 35 (2012) 1539–1546.
[9] W.-J. Zhao, W. Liu, J.-B. Chen, Z.-M. Zhou, M.-M. Yang, Use of cloud point extrac-
tion with derivatizing reagent for the extraction and determination of isoniazid,
J. Chromatogr. Sci. 49 (2011) 154–158.
[10] F. Han, R. Yin, X.-L. Shi, Q. Jia, H.-Z. Liu, H.-M. Yao, L. Xu, S.-M. Li, Cloud point
extraction-HPLC method for determination and pharmacokinetic study of flur-
biprofen in rat plasma after oral and transdermal administration, J. Chromatogr.
B 868 (2008) 64–69.
[11] H. Zhang, H.-K. Choi, Analysis of meloxicam by high-performance liquid chro-
matography with cloud-point extraction, Anal. Bioanal. Chem. 392 (2008)
947–953.
[12] X. Liu, X.-H. Chen, Y.-Y. Zhang, W.-T. Liu, K.-S. Bi, Determination of arbidol in rat
plasma by HPLC–UV using cloud-point extraction, J. Chromatogr. B 856 (2007)
273–277.
[13] M.D. Rukhadze, S.K. Tsagareli, N.S. Sidamonidze, V.R. Meyer, Cloud-point
extraction for the determination of the free fraction of antiepileptic drugs in
blood plasma and saliva, Anal. Biochem. 287 (2000) 279–283.
[14] A.B. Tabrizi, S. Naini, K. Parnian, S. Mohammadi, F.E. zad, S.P. Anvarian, A. Abdol-
lahi, Determination of triamterene in human plasma and urine after its cloud
point extraction, Quím. Nova 37 (2014) 1182–1187.
[15] H. Wu, G.-Y. Zhao, L.-M. Du, Determination of ofloxacin and gatifloxacin
by mixed micelle-mediated cloud point extraction-fluorimetry combined
methodology, Spectrochim. Acta, A: Mol. Biomol. Spectrosc. 75 (2010)
1624–1628.
[16] X.Y. Qin, J. Meng, X.Y. Li, J. Zhou, X.L. Sun, A.D. Wen, Determination of ven-
lafaxine in human plasma by high-performance liquid chromatography using
cloud-point extraction and spectrofluorimetric detection, J. Chromatogr. B 872
(2008) 38–42.
[17] B.K. Matuszewski, M.L. Constanzer, C.M. Chavez-Eng, Strategies for the
assessment of matrix effect in quantitative bioanalytical methods based on
HPLC−MS/MS, Anal. Chem. 75 (2003) 3019–3030.
[18] G.-F. Li, K. Wang, R. Chen, H.-R. Zhao, J. Yang, Q.-S. Zheng, Simulation of the phar-
macokinetics of bisoprolol in healthy adults and patients with impaired renal
function using whole-body physiologically based pharmacokinetic modeling,
Acta Pharmacol. Sin. 33 (2012) 1359–1371.
[19] H. Chang, J. Li, J. Li, X. Guan, F. Sun, Z. Qian, K. Bi, G. Fan, Simulta-
neous determination of amlodipine and bisoprolol in rat plasma by a liquid
chromatography/tandem mass spectrometry method and its application in
pharmacokinetic study, J. Pharm. Biomed. Anal. 71 (2012) 104–110.
[20] G.-Y. Liu, W. Wang, J.-Y. Jia, C. Lu, Y.-M. Liu, M.-Q. Zhang, Y. Liu, S.-J. Li, C. Yu,
Liquid chromatography tandem mass spectrometry method for determination
of bisoprolol in human plasma using d5-bisoprolol as the internal standard,
Biomed. Chromatogr. 24 (2010) 574–580.
[21] O. Gonzalez, G. Iriarte, E. Rico, N. Ferreirós, M.I. Maguregui, R.M. Alonso, R.M.
Jiménez, LC–MS/MS method for the determination of several drugs used in
combined cardiovascular therapy in human plasma, J. Chromatogr. B 878
(2010) 2685–2692.
[22] G. Peste, N. Bibire, M. Apostu, A. Vlase, C. Oniscu, A New Liquid
Chromatography-Tandem Mass Spectrometry Method for Determination of
Bisoprolol in Human Plasma Samples, J. Biomed. Biotechnol. 2009 (2009),
http://dx.doi.org/10.1155/2009/736327, Article ID 736327, 8 pages.
[23] G. Peste, C. Oniscu, A. Vlase, Experimental research for determination of
bisoprolol fumarate in human plasma samples using liquid chromatography-
tandem mass spectrometry (LC–MS/MS) technique, Rom. Biotechnol. Lett. 15
(2010) 5141–5145.
[24] M.F. Tutunji, H.M. Ibrahim, M.H. Khabbas, L.F. Tutunji, Simultaneous determina-
tion of bisoprolol and hydrochlorothiazide in human plasma by HPLC coupled
with tandem mass spectrometry, J. Chromatogr. B 877 (2009) 1689–1697.
[25] M. Liu, D. Zhang, Y. Sun, Y. Wang, Z. Liu, J. Gu, Liquid
chromatographic–electrospray tandem mass spectrometric determination of
bisoprolol in human plasma, Biomed. Chromatogr. 21 (2007) 508–513.
[26] S. Li, G. Liu, J. Jia, Y. Liu, C. Pan, C. Yu, Y. Cai, J. Ren, Simultaneous determina-
tion of ten antiarrhythic drugs and a metabolite in human plasma by liquid
chromatography—tandem mass spectrometry, J. Chromatogr. B 847 (2007)
174–181.
[27] J. Bhatt, G. Subbaiah, S. Kambli, B. Shah, M. Patel, A. Saxena, A. Baliga,
S. Nigam, H. Parekh, G. Yadav, A high throughput and sensitive liquid
chromatography–tandem mass spectrometry (LC–MS/MS) method for the
estimation of bisoprolol in human plasma using multiplexing technique, J.
Chromatogr. B 852 (2007) 374–381.
[28] M. Josefsson, A. Sabanovic, Sample preparation on polymeric solid phase
extraction sorbents for liquid chromatographic-tandem mass spectrometric
analysis of human whole blood—a study on a number of beta-agonists and
beta-antagonists, J. Chromatogr. A 1120 (2006) 1–12.
[29] J. Musijowski, M. Filist, P.J. Rudzki, Sensitive single quadrupole LC/MS method
for determination of lapatinib in human plasma, Acta Pol. Pharm. 71 (2014)
p7.
˛
[30] J. Giebułtowicz, G. Nałecz-Jawecki, Occurrence of antidepressant residues in
the sewage-impacted Vistula and Utrata rivers and in tap water in Warsaw
(Poland), Ecotoxicol. Environ. Safe. 104 (2014) 103–109.
[31] J. Musijowski, E. Piórkowska, P.J. Rudzki, Determination of sunitinib in human
plasma using liquid chromatography coupled with mass spectrometry, J. Sep.
Sci. 37 (2014) 2652–2658.
[32] M. Filist, K. Buś-Kwaśnik, H. Ksycińska, P.J. Rudzki, Simplified LC–MS/MS
method enabling the determination of azithromycin in human plasma after
a low 100 mg dose administration, J. Pharm. Biomed. Anal. 100 (2014)
184–189.
[33] H. Filik, S.D. Çekiç, Cloud point extraction of pesticide residues, in: M.
Stoytcheva (Ed.), Pesticides in the Modern World—Trends in Pesticides Analy-
sis, InTech, Rijeka, Croatia, 2011.
[34] K. Buś-Kwaśnik, H. Ksycińska, A. Leś, K. Serafin-Byczak, P.J. Rudzki, J. Raszek, T.
Łazowski, A. Bielak, A. Wybraniec, Bioequivalence and pharmacokinetics of two
10-mg bisoprolol formulations as film-coated tablets in healthy white volun-
teers: a randomized, crossover, open-label, 2-period, single-dose, fasting study,
Int. J. Clin. Pharmacol. Ther. 50 (2012) 909–919.
[35] L. Ding, X. Zhou, X. Guo, Q. Song, J. He, G. Xu, LC–ESI–MS method for the deter-
mination of bisoprolol in human plasma, J. Pharm. Biomed. Anal. 44 (2007)
520–525.
[36] FDA Guidance for Industry, Center for Drug Evaluation and Research
(CDER).Bioanalytical Method Validation, Food and Drug Administration,
Rockville, MD,U.S, 2001.
[37] Guideline on bioanalytical method validation, EMEA/CHMP/EWP/192217/
2009, European Medicines Agency. Committee for Medicinal Products for
Human Use (CHMP), 2011.
[38] Z.-M. Zhou, D.-Y. Zhao, J. Wang, W.-J. Zhao, M.-M. Yang, Study of cloud point
extraction and high-performance liquid chromatographic determination of iso-
niazid based on the formation of isonicotinylhydrazone, J. Chromatogr. A 1216
(2009) 30–35.
[39] G. Ren, Q. Huang, J. Wu, J. Yuan, G. Yang, Z. Yan, S. Yao, Cloud point
extraction—HPLC method for the determination and pharmacokinetic study
of aristolochic acids in rat plasma after oral administration of Aristolochiae
Fructus, J. Chromatogr. B 953–954 (2014) 73–79.
[40] J. Zhou, P. Zeng, H.H. Tu, F.Q. Wang, Development and application of
high-performance liquid chromatography for the study of ampelopsin phar-
macokinetics in rat plasma using cloud-point extraction, J. Sep. Sci. 34 (2011)
160–168.
[41] A. 43 Nordström, J.V. Apon, W. Uritboonthai, E.P. Go, G. Siuzdak, Surfactant-
enhanced desorption/ionization on silicon mass spectrometry, Anal. Chem. 78
(2005) 272–278.
[42] D. Gansser, Experience with a cross-validation approach, Chromatographia 55
(2002) S71–S74.
[43] M.T. Gilbert, I. Barinov-Colligon, J.R. Miksic, Cross-validation of bioana-
lytical methods between laboratories, J. Pharm. Biomed. Anal. 13 (1995)
385–394.
 Talanta 83 (2011) 980–987

Contents lists available at ScienceDirect

Talanta
journal homepage: www.elsevier.com/locate/talanta

Tyrosinase immobilized magnetic nanobeads for the amperometric assay of

enzyme inhibitors: Application to the skin whitening agents
Veronica Harceaga Sima a , Stephanie Patris b , Zeynep Aydogmus c , Ahmad Sarakbi b ,
Robert Sandulescu a , Jean-Michel Kauffmann b,∗
a
University of Medicine and Pharmacy, Faculty of Pharmacy, 400012 Cluj-Napoca, Romania
b
Université Libre de Bruxelles (ULB), Faculty of Pharmacy, 1050 Brussels, Belgium
c
Istanbul University1 , Faculty of Pharmacy, 34116 Istanbul, Turkey

a r t i c l e i n f o a b s t r a c t

Article history: The immobilization of tyrosinase onto glutaraldehyde activated streptavidine magnetic particles and
Received 22 September 2010 subsequent retention onto a magnetized carbon paste electrode for the amperometric assay of tyrosi-
Received in revised form 26 October 2010 nase inhibitors is described. Tyrosine was used as substrate as it is the first substrate in the melanogenesis
Accepted 1 November 2010
process. The sensing mode is based on monitoring the decrease of the amperometric signal correspond-
Available online 10 November 2010
ing to the electrochemical reduction of dopaquinone enzymatically generated. This current decrease is
due to the presence of inhibitors acting directly on the enzyme or inhibitors acting on the product of the
Keywords:
enzymatic reaction, i.e. dopaquinone. The methodology is designed for the evaluation of the inhibitory
Magnetic nanoparticles
Tyrosinase
potency of the most frequently used active substances in cosmetic marketed products against hyper-
Skin whitening agent pigmentation such as kojic acid, azelaic acid and benzoic acid. These compounds bind to the tyrosinase
Biosensor active center. Ascorbic acid is also investigated as it interrupts the synthesis pathway of melanin by reduc-
ing the melanin intermediate dopaquinone back to l-dopa. By comparing the obtained IC50 , under the
same experimental conditions, the order of their inhibitory potency was: kojic acid (IC50 = 3.7 × 10−6 M,
Ki = 8.6 × 10−7 M), ascorbic acid (IC50 = 1.2 × 10−5 M), benzoic acid (IC50 = 7.2 × 10−5 M, Ki = 2.0 × 10−5 M)
and azelaic acid (IC50 = 1.3 × 10−4 M, Ki = 4.2 × 10−5 M) in close agreement with literature spectrophoto-
metric inhibition data using the soluble tyrosinase.
© 2010 Elsevier B.V. All rights reserved.

1. Introduction Hyperpigmentation is a complex process. There are many fac-

Melanin, the dark pigment produced by skin cells in the most
inner layer of the epidermis, plays an important role in protect-
ing human skin from the harmful effects of the UV sun radiations.
Melanin also determines our phenotypic appearance. However,
abnormal accumulation of melanin in the basal layer of the dermis
is responsible for hyperpigmentation including melasma, freckles
and senile lentigines [1]. These undesirable conditions of the human
skin are serious esthetic problem for human beings nowadays since
both appearance and quality of life are becoming more and more
important. This problem is particularly prevalent in middle aged
and elderly individuals [2] and this has encouraged researchers to
seek for new potent inhibitors of the hyperpigmentation process.
Recently, a global market demand has developed on this subject.
For the past few decades, tyrosinase inhibitors have been a great
concern solely due to the key role of tyrosinase in mammalian
melanogenesis [3].
∗ Corresponding author.
E-mail address: jmkauf@ulb.ac.be (J.-M. Kauffmann).
1
On leave.
tors that influence the activity of melanocites, but it is well known
and unanimously accepted that tyrosinase activity plays the main
role in the pathway of melanin biosynthesis. It catalyzes the first
two reactions of the melanogenesis process namely: the hydroxy-
lation of l-tyrosine to l-dopa as monophenolase, using molecular
oxygen as co-substrate, and the oxidation of l-dopa to dopaquinone
as diphenolase. These steps, and mostly the first oxidizing step, are
the rate-limiting steps in melanin synthesis since the subsequent
reactions can proceed spontaneously at physiological pH [3].
A number of tyrosinase and melanogenesis inhibitors, from both
natural and synthetic sources, have been identified [4]. However,
the definition of “tyrosinase inhibitor” is sometimes misleading, the
terminology being also used to refer to melanogenesis inhibitors,
whose action mainly reside in some interference in melanin forma-
tion, regardless of any direct inhibitor–enzyme interaction [3]. Only
specific tyrosinase inactivators and inhibitors should be regarded
as “true inhibitors”, which bind to the enzyme and inhibit its activ-
ity. Some skin whitening agents that act as tyrosinase inhibitors,
such as kojic acid and azelaic acid, or as melanogenesis inhibitors
such as hydroquinone, arbutin and ascorbic acid are the most
well known to most dermatologists [5]. The interference with
melanin synthesis can occur in different ways. Kojic acid is a

0039-9140/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.talanta.2010.11.005
 V.H. Sima et al. / Talanta 83 (2011) 980–987
chelator of the copper of the enzyme active site [6], arbutin is a
pro-drug of hydroquinone that is an alternative substrate, azelaic
acid blocks the tyrosine access to the active site and ascorbic acid
chemically reduces the dopaquinone back to l-dopa, thus avoid-
ing dopachrome and melanin formation [7]. Also benzoic acid, a
preservative very frequently used in cosmetic and food industry [8],
has been demonstrated to have inhibitory properties for tyrosinase
[9].
The need for the quantification and comparative analysis at high
throughput of the inhibitory activity of possible new products and
of the existing ones is becoming more and more of interest for the
evaluation and development of efficient inhibitors of the hyperpig-
mentation process.
Until now, several methods for the screening of skin-whitening
agents have been proposed. The method most frequently applied
exploited the spectrophotometric quantification of dopaquinone,
before and after the inhibition of soluble tyrosinase [6,10]. Another
method with tyrosinase in solution relied on electron spin reso-
nance spectrometry that measured the accumulation of eumelanin
radicals in melanin culture cells [11]. In vivo studies evaluated the
inhibitors potency by photometry [12]. Usually in each of the cur-
rently developed method, the inhibitory strength of the studied
compounds was compared to that of a standard inhibitor, namely
kojic acid [6]. Spectrophotometry is not a primary choice analyti-
cal method for the quantification of an on going high ratio reaction
which is the case of the enzymatic formation of dopaquinone due
to the low reproducibility of the obtained results. The use of spec-
trometry has the disadvantage of the need of culture cells and the
photometric one is applied to in vivo studies so it can be used just for
substances that have been approved for human use and the imple-
mentation of such study involves ethic problems. Thus they are not
suitable for rapid assays in scientific research of novel potential skin
whitening agents. The use of biosensors would be an interesting and
valuable alternative for the rapid screening and comparison of the
inhibitory potency of the existing compounds and the novel ones
due to its relative simplicity in use and the high reproducibility of
the analytical data.
To the best of our knowledge there have been very few tyrosi-
nase based biosensors fabricated for the evaluation of the inhibitory
potency of kojic acid. A Clark-type oxygen electrode that monitored
oxygen consumption during catechol conversion in the presence
of kojic acid inhibitor was reported [13]. Another biosensor for
kojic acid quantification used dopamine as substrate and it mon-
itored the oxygen consumption with a Clark type electrode [14].
There have been a couple of tyrosinase biosensors developed for
the evaluation of the inhibitory potency of benzoic acid [8,15]
but none of them used l-tyrosine as enzyme substrate. Tyrosinase
based biosensors usually used phenol or its derivatives as sub-
strate with the enzyme being combined with an electrochemical
transducer to sense either the oxygen consumption of the overall
enzymatic reaction [14] or the enzymatic production of the elec-
troactive quinone species [16]. There was one article reported of
a tyrosinase based biosensor for quantitative analysis of phenols.
Tyrosinase was immobilized to core–shell magnetic particles and
subsequently attached to the surface of a carbon paste electrode
with the help of a permanent magnet [17].
Tyrosinase is widely distributed in microorganisms, animals,
humans and plants. The enzyme extracted from the mushroom
Agaricus bisporus is highly homologous with the mammalian one
making it well suited as a model for studies on melanogenesis.
Practically all studies of inhibitors conducted so far have used
the mushroom tyrosinase likely because of its readily commercial
availability [3]. Yet alternative systems for testing can be obtained
like mammalian tyrosinase, melanocytic cultures, co-cultures of
keratinocytes and melanocytes, skin culture to finally in vivo appli-
cation to animal skin [5,18].
981
Different phenolic compounds can be enzymatically oxidized
by tyrosinase and several substrates have been proposed in the
literature for the evaluation of the enzyme inhibitors. Dopamine
was used as a substrate in the construction of an amperometric
biosensor for the detection of tyrosinase inhibitors such as kojic
acid, benzoic acid and SCN− ion [14]. Catechol, p-cresol, m-cresol,
phenol and p-chlorophenol were used as enzyme substrates for
the determination of benzoic acid based on its inhibition prop-
erties, using an amperometric biosensor device [19–21]. l-Dopa
was used as enzyme substrate in spectrophotometric evaluation of
skin-whitening agents [10]. Also l-tyrosine was used as a substrate
in spectrophotometric evaluation of different inhibitors [22,23].
Cellular tests with human melanocytes were used for the trans-
formation rate study of the tyrosinase substrate l-dopa using a
spectrophotometric method [24].
Recently, biosensors based on the inhibition of tyrosinase activ-
ity in the presence of a monophenol or o-diphenol substrate have
been proposed for the determination of triazine pesticides [25], car-
bamate and organophosphate pesticides [16], diuron and atrazine
[26], herbicides such as alachlor, diazinon and carbaryl [27], propyl
gallate [28], 2,4-dichlorophenol [29], fluoride [30], benzoic acid
[19], cinnamic acid and sorbic acid [31].
Besides choosing the adequate transducer and enzyme in the
construction of a biosensor, an important role is played by the
enzyme immobilization method. The use of magnetic particles
(beads) in biomedical and pharmaceutical sciences has increased
significantly in recent years [32]. The masterbeads are specially
designed for allowing simple and effective immobilization of lig-
ands such as enzyme through activation with bi-functionalized
reactants [33], like glutaraldehyde. Immobilization of biomolecules
onto the surface of magnetic particles can result in a number of
additional functionalities: on one hand the nanostructure of the
magnetic particles permits a high enzyme loading without affect-
ing its natural structure and consequently its activity and on the
other hand the beads can be attracted by a magnet and be strongly
retained in close proximity to the electrode surface and read-
ily washed away if needed [34]. The streptavidine masterbeads
(NMPS) are mainly designed for immunological analyses [35] but
they can very well be used for enzyme immobilization. Based on
the IUPAC definition [36] such a configuration is strictly speaking
not a biosensor, i.e. a self-contained integrated device, but since it
has the enzyme, during the assays, in close spatial contact with the
electrode we may consider it as a biosensor.
In the present work, a novel method for the evaluation of the
potency of the tyrosinase and melanogenesis inhibitors used as
skin whitening agents is proposed. The biosensor was fabricated
using tyrosinase immobilized via glutaraldehyde activated strep-
tavidine magnetic nanoparticles and retained onto a “magnetized”
carbon paste electrode, mCPE. The interest in using a carbon paste
electrode lies in its low background current, allowing for highly
sensitive assay to be performed in the applied potential range
close to 0.0 V versus an Ag/AgCl reference electrode. The employed
tyrosinase was the mushroom tyrosinase as it is highly homology
with the human enzyme. The substrate used was the l-tyrosine,
the enzyme’s natural substrate in the skin and the investigated
inhibitors were kojic acid, azelaic acid, ascorbic acid, the most
frequently used active substances in marketed products for hyper-
pigmentation, and benzoic acid, a well known preservative in the
cosmetic industry. The obtained configuration was used to moni-
tor the reduction current of the enzymatically generated oxidized
species of l-tyrosine, i.e. the dopaquinone, in the presence of molec-
ular oxygen (Fig. 1). The biosensor was applied to the screening
of inhibitors which induced a decrease of the reduction current
proportional to the inhibitor concentration and strength.
In practice, the steady state current for l-tyrosine was recorded
as I0 . The biosensor response in the case of the addition of an
982 V.H. Sima et al. / Talanta 83 (2011) 980–987
Fig. 1. Biosensor’s mechanism of action.
inhibitor was recorded as I1 . The concentration of inhibitors was
correlated with the percentage of inhibition (In %) which was calcu-
lated using the relationship: In (%) = [(I0 − I1 )/I0 ] × 100. With I0 and
I1 being the biosensor current signals before and after the addi-
tion of the inhibitor, respectively, I1 was expected to be smaller
comparing to I0 as the process was inhibited.
2. Materials and methods
2.1. Material
The mushroom tyrosinase (EC 1.14.18.1) 5370 Units/mg was
purchased from Sigma. l-Tyrosine and l-dopa were provided by
Janssen Chimica. The Masterbeads Streptavidine (500 nm) were
purchased from Ademtech France. Dopamine, glutaraldehyde 25%
wt and monosodium phosphate was provided by Sigma–Aldrich
and disodium phosphate was provided by Merck. The carbon paste
was from Metrohm. Kojic acid was purchased from Kreglinger
Europe, azelaic acid and ascorbic acid from Acros Organics and
benzoic acid from Fluka.
All reagents were of analytical grade and were used as received.
All solutions were prepared with distilled water and stored in the
refrigerator (4 ◦ C) when not in use. The phosphate buffer was pre-
pared by mixing two stock solutions of monosodium phosphate
0.1 M and disodium phosphate 0.1 M to the desired pH. The stock
solutions of l-tyrosine, l-dopa, kojic acid, benzoic acid, azelaic acid
and ascorbic acid were prepared in phosphate buffer.
2.2. Construction of the tyrosinase-NMPS biosensor
2.2.1. Enzyme immobilization
The Streptavidin Masterbeads were monodispersed and were
based on superparamagnetic particles composed of a magnetic core
encapsulated by a hydrophilic polymer shell coated with strepta-
vidin. The magnetic particles had a diameter of 500 nm (RSD max
20%), density of approx. 2.0 g/cm3 , magnetic susceptibility approx.
40 emu/g, specific surface area of 5 m2 /g, iron oxide content approx.
70% m/m and the solid content of 10 mg/mL [33].
An amount of 1 mL 10 mg/mL NMPSs was washed with 0.01 M
phosphate buffer pH 6.5 (5 mL) then reacted with 1 mL of 2.5%
wt glutaraldehyde solution (in 0.01 M phosphate buffer, pH 6.5)
at room temperature for 30 min. After washing with 0.01 M phos-
phate buffer pH 6.5 (5 mL), the glutaraldehyde-treated NMPSs
were reacted with 1 mL tyrosinase solution (3 mg/mL in phosphate
buffer 0.01 M, pH 6.5) during 1 h at room temperature to obtain
the tyrosinase-NMPSs. The resulting suspension was washed with
0.01 M phosphate buffer pH 6.5 (5 mL) and stored in 1 mL 0.01 M
phosphate buffer pH 6.5 at 4 ◦ C. During the washing steps, the
beads were trapped by magnetic forces by placing the reacting
Eppendorf tube close to a strong magnet and the supernatant was
carefully pipetted off. The final concentration of tyrosinase-NMPSs
was 10 mg/mL (referring to the amount of NMPS) [35].
A 10 mg/mL tyrosinase-bead suspension was used for l-
tyrosine response optimization and an approximately 1.25 mg/mL
tyrosinase-bead suspension was used in the inhibition assays.
The initial 1 mL amount of the tyrosinase-NMPSs was diluted and
it resulted finally in 8 mL suspension of 1.25 mg/mL tyrosinase-
NMPSs that allowed performing approximately 800 inhibition
assays.
Visible spectrophotometry was applied for the determination
of the efficiency of the immobilization method of tyrosinase onto
the NMPSs. The method consisted in determining the amount of
immobilized tyrosinase by measuring the absorbance of tyrosinase
in a 60 times diluted tyrosinase solution used in the immobiliza-
tion process, before and after the enzyme was immobilized onto the
activated NMPSs. The percentage of immobilization, calculated by
referring to a calibration curve (absorbance versus tyrosinase con-
centration) realized in 0.01 M phosphate buffer pH 6.5 ( = 280 nm),
was found to be around 40 ± 5%. Therefore it was inferred that
the inhibition assays were realized with approximately 8 Units of
tyrosinase immobilized at the electrode surface.
2.2.2. Biosensor fabrication
The working carbon paste electrode (mCPE) consisted of a home
made electrode: inside a plastic rod, a permanent magnet (Neody,
3 mm diameter) was inserted leaving a depression at the surface of
approximately 2 mm for housing the solid carbon paste layer, and
a copper wire was used as the electrical conductor.
Before each experiment the solid carbon paste was manually
poured in the hole and the resulting mCPE was smoothed on a clean
paper surface. The diameter of the active surface was 3 mm. A vol-
ume of 10 ␮L of the Tyrosinase-NMPSs suspension was spread over
the surface of the electrode (in position up side down) allowing
the particles to settle, being attracted within a few seconds by the
magnet. Subsequently, the tyrosinase-NMPSs immobilized mCPE
was oriented in the right position in a three-electrode cell. The
strong attraction of the Tyrosinase-NMPSs by the magnet housed
inside the working electrode allowed appropriate stirring during
the amperometric experiments.
2.3. Apparatus
The experiments were performed by using as potentiostat a
BASi EPSILON system with a C3 Cell Stand. Amperometry was per-
formed in a conventional three electrodes setup with the biosensor
as working, a Pt wire as auxiliary and an Ag/AgCl 3 M KCl as refer-
ence electrode, respectively. The cyclic voltammetry experiments
were performed in the same conventional three electrodes setup
used in amperometry. All the cyclic voltammetry experiments were
recorded at 100 mV s−1 or 10 mV s−1 in a potential window com-
prised between −0.6 V and +1.1 V versus Ag/AgCl 3 M KCl reference
electrode. The spectrophotometric assays were performed using
an Agilent Hewlett Packard Diode Array UV Vis Spectrophotometer
Model 8453. The pH of the solution was controlled with a Metrohm
827 pH Lab system. All experiments were performed at room tem-
perature (23 ◦ C).
2.4. Amperometric assay
During amperometric experiments the biosensor potential was
kept at −100 mV under continuous stirring conditions with a mag-
netic spinbar at 300 rpm. The working potential was imposed and
the background current was allowed to reach a steady state value
within approx. 10 min. Different amounts of tyrosinase standard
solution or inhibitor solution were added every 100 or 200 s and the
current was recorded as a function of time in 10 mL 0.1 M phosphate
buffer solution.
 V.H. Sima et al. / Talanta 83 (2011) 980–987
2.0E-06
I (A)
1.5E-06
b
1.0E-06
5.0E-07 a ence of l-tyrosine and l-dopa using 10 ␮L of the tyrosinase-NMPS
0.0E+00
-1.0 -0.5 0.0 0.5 1.0 1.5
-5.0E-07
E (V) vs Ag/AgCl
-1.0E-06
-1.5E-06
Fig. 2. Repetitive CV for l-tyrosine 4.76 × 10−5 M, 0.1 M phosphate buffer pH 6.5,
mCPE, scan rate, a: 10 mV s−1 , b: 100 mV s−1 .
After each amperometric inhibition assay, the tyrosinase-
NMPSs were readily removed by flushing a water burst over the
biosensor surface. The surface of the mCPE was renewed after each
inhibition assay.
3. Results and discussions
3.1. Working conditions
The working conditions for an optimal response to the sub-
strate of the tyrosinase based biosensor have been tested. Since
the biosensor was designed for testing inhibitors of the skin pig-
mentation process it was taken into account to operate as close as
possible to the natural environment of the skin tyrosinase which is
a transmembrane protein of melanosomes [37].
3.1.1. Selection of the applied potential
It is well known that l-tyrosine enzymatic oxidation pro-
duces dopaquinone and it was expected that this product could
be detected by electroreduction at the mCPE. Likewise the elec-
trochemical oxidation of l-tyrosine was supposed to generate
dopaquinone. Cyclic voltammetry (CV) performed at 10 and
100 mV/s (from 0.0 mV to 1.1 V, and back to −0.6 V) in 0.1 M phos-
phate buffer pH 6.5, showed the irreversible oxidation peak of
l-tyrosine (1 × 10−3 M and 4.76 × 10−5 M) since no reduction peak
was observed in the investigated potential domain (Fig. 2). Same
experiments were performed also for l-dopa which gave an oxida-
tion peak at 0.550 V but still no reduction peak was observed. The
absence of a reduction current and the progressive decrease of the
l-tyrosine peak during multiple scanning suggested that the gen-
erated dopaquinone escaped rapidly from the electrode solution
interface likely due its high reactivity giving rise to polymer-like
structures fouling the electrode surface [38]. Amperometry of l-
tyrosine (2.44 × 10−5 M) at the tyrosinase-NMPS mCPE, thanks to
the low background current compared to CV, permitted, however,
to detect a reduction current at potentials near 0 V versus Ag/AgCl
likely corresponding to the electroreduction of some enzymatically
generated dopaquinone remaining at the electrode solution inter-
face. Amperometric experiments were realized at 0, −50, −100,
−150, −200 and −250 mV. By decreasing the applied potential the
reduction current increased but the response steady-state became
less stable. An applied potential of −100 mV was considered opti-
mal taking into account the magnitude of the reduction current,
the ratio between signal and background current, the steady-state
of the response plateau and the possible interfering species at the
applied potential, in agreement with other tyrosinase based biosen-
sors described in the literature [21].
3.1.2. Selection of the substrate
To the best of our knowledge there is no tyrosinase based amper-
ometric biosensor for the evaluation of skin-whitening agents that
983
used l-tyrosine as a substrate, reported in the literature. Moreover
it has been demonstrated that the intensity of inhibition varied
considerably depending on the characteristics of the substrate [19].
The biosensor response was tested comparatively both in the pres-
suspension (10 mg/mL) deposited onto the surface of the mCPE. It
was observed that, in the concentration range 4.98 × 10−6 M and
2.91 × 10−5 M, the sensitivity to l-dopa (−5.56 × 10−4 A/M) was
almost 6 times higher comparing to l-tyrosine (−9.37 × 10−5 A/M)
and the response time (the time necessary for reaching 95% of the
maximum response) to l-dopa (56 s) was faster comparing to l-
tyrosine (86 s). l-Tyrosine, however, was chosen as model substrate
for subsequent experiments. This decision was based on the fact
that, in vivo, l-tyrosine is the natural substrate of tyrosinase and
also the starting point of melanin synthesis and its oxidation, to
l-dopa, is the rate-limiting step of the whole process [3]. The sensi-
tivity to l-tyrosine was considered high enough for assuring reliable
results and the lag period of its first enzymatic oxidizing step to l-
dopa was sufficiently short for further readily implementation of
the testing conditions.
3.1.3. Selection of the pH
The biosensor response was studied in the pH range comprised
between 5.5 and 7. It is well known that enzyme activity is highly pH
dependent and that the optimum pH for an enzymatic assay must
be determined empirically [39]. By decreasing the pH from 7.0 to
5.5, the amperometric signal increased but no stable steady-state
plateau was obtained at pH 6.0 and 5.5. For further experiments,
the pH chosen was 6.5. At this pH, the current and the steady-state
of the biosensor response were considered optimal, in agreement
with different literature data where the working pH of the tyrosi-
nase based assays vary between 6.0 and 7.0. [9,14,22,23]. The
selected pH corresponded to the Sigma declared tyrosinase stabil-
ity pH and is also close to the normal pH range of the skin which is
considered to vary between 4 and 6.5.
3.1.4. Amount of tyrosinase-NMPS spiked onto the mCPE
The amperometric response was determined as a function of l-
tyrosine, in the concentration range 4.98 × 10−6 and 2.91 × 10−5 M,
using different amounts (5 ␮L, 10 ␮L and 20 ␮L) of the 10 mg/mL
tyrosinase-NMPS suspension spiked onto the surface of the mCPE.
The response increased proportionally by raising the amount of
immobilized enzyme onto the surface of the electrode. The sen-
sitivities obtained were: −6.82 × 10−5 A/M, −9.37 × 10−4 A/M
and −1.68 × 10−4 A/M for 5, 10 and 20 ␮L spiked suspension,
respectively. The repeatability of 5 ␮L deposited beads was poor
comparing to 20 ␮L and 10 ␮L deposits. The latter was selected
in subsequent work since it offered good sensitivity and repeata-
bility (RSD = 2.8%, n = 3) and it consumed less beads then a 20 ␮L
deposit. For the subsequent inhibition assays it was decided to use
a diluted suspension of tyrosinase-NMPS (1.25 mg/mL) since it gave
a good response to l-tyrosine (3.33 × 10−4 M) and allowed to reduce
substantially the cost of the assays.
3.1.5. Selection of the substrate concentration for the inhibition
studies
The opinions about the substrate concentration to be consid-
ered in inhibition assays are quite contradictory. Kok et al. [40]
concluded, when measuring the inhibition potency of a competi-
tive inhibitor with an acetylcholinesterase and a choline oxidase
biosensor, that the inhibition percentage increased by raising the
substrate concentration. Therefore they worked in enzyme satu-
rated substrate conditions. Shan et al. [19] demonstrated that for
a tyrosinase biosensor for the determination of benzoic acid, the
concentration of the catechol substrate did not affect the maxi-
mum inhibition percentage but it affected the sensitivity of the
984 V.H. Sima et al. / Talanta 83 (2011) 980–987
method. It must be noted, however, that the use of a high sub-
strate concentration would not yield sensitive inhibition responses
when the quantification of a competitive inhibitor is performed by
simultaneous addition of the inhibitor and the substrate. Because
in competitive inhibition the substrate competes with the inhibitor
for the enzyme active site, and the inhibition, especially at low
inhibitor concentrations, would likely not be detected [40].
Two different l-tyrosine concentrations were studied;
3.33 × 10−4 M and 4.76 × 10−5 M, and the obtained IC50 for
kojic acid were found to be equal within the experimental error
(RSD = 2.3%). The percentage of inhibition was evaluated from
the response of the uninhibited form of the enzyme versus the
response of the enzyme partially inhibited. Taking into account that
our device was not designed for the quantification of inhibitors but
for the comparative screening and quantification of their inhibitory
potency we considered that all the immobilized enzyme molecules
must take part in the reaction and this could only be possible in
substrate concentration corresponding to the saturation portion
of the activity versus substrate curve [40]. Enzyme saturated
substrate concentration, namely 3.33 × 10−4 M l-tyrosine, was
then selected for further inhibition studies.
Since the concentration of substrate was quite high, it had to be
established if the quantity of the dissolved oxygen in the solution
was not a limiting factor for a 3.33 × 10−4 M l-tyrosine concentra-
tion. After 30 min of air bubbling into the working buffer solution
the biosensor response did not change, but after 30 min of nitrogen
bubbling the biosensor response decreased dramatically. The latter
was also a further indirect evidence of the generation of an elec-
troactive species by the immobilized tyrosinase. It was concluded
that in the selected experimental conditions the quantity of oxy-
gen was not a limiting factor and that the enzymatic process was
saturated by the substrate.
3.1.6. Michaelis–Menten curve
A typical Michaelis–Menten plot and its linearization by the
Lineweaver–Burk plot (y(A−1 ) = −7.91 × 103 × (M−1 )−8.53 × 107 ,
R2 = 1.000, RSD slope = 3.3%, RSD intercept = 7.7%, n = 3) was
obtained for l-tyrosine between 2.49 × 10−6 M and 4.41 × 10−4 M
in the above described optimal working conditions of the tyrosinase
based biosensor, namely: applied potential (Eapp ) = −100 mV, 0.1 M
phosphate buffer pH 6.5, 10 ␮L of 1.25 mg/mL tyrosinase-NMPS
suspension spiked onto the mCPE surface. The kinetic parameters
Km app and Imax were calculated as the average of 3 consecu-
tive determinations: Km app = 9.7 × 10−5 M (RSD = 3.1%, n = 3) and
Imax = −1.2 × 10−8 A (RSD = 7.5%, n = 3).
The obtained Km app values (Table 2) can be compared to
those reported in the literature for the enzyme in solution, i.e.
Km app = 6.2 × 10−4 M [23] or 3.3 × 10−4 M [41] at pH 6.5, illustrating
the fact that the applied immobilization technique maintained the
enzyme affinity for its substrate.
3.2. Inhibitors assay
The developed biosensor was designed for testing inhibitors of
the skin pigmentation process. When testing their potency, there
are many factors which contribute to the final results: type of
enzyme, quantity of immobilized enzyme, immobilization method,
type of substrate, substrate concentration, time of contact between
the enzyme, substrate and inhibitor, pH, temperature, applied
potential, rate of solution stirring. In order to obtain a correct com-
parison of the inhibition potency of the studied compounds, it was
worked under the same experimental conditions [19].
As the developed device was a novel analytical tool, in order to
test its efficiency, it was applied to the study of compounds that
have already demonstrated their activity in clinical practice or by
using other analytical methods.
Table 1
IC50 for kojic acid, benzoic acid, azelaic acid and ascorbic acid.
IC50 (␮M) Kojic acid Benzoic acid Azelaic acid Ascorbic acid
IC50 1 3.66 72.0 125 12.4
IC50 2 3.57 72.0 127 11.6
IC50 3 3.71 72.6 128 11.0
IC50 4 3.72 70.9 119 12.2
IC50 5 3.63 70.5 127 11.7
IC50average 3.66 71.6 125 11.8
The proposed biosensor was intended to test inhibitors that
have two different mechanisms of action, namely: true enzyme
inhibitors which bind to the active site of the enzyme and inhibitors
which consume the dopaquinone intermediate of the melanin syn-
thesis pathway.
It was checked that the NMPS mCPE (i.e. without tyrosinase)
gave no response under the selected experimental conditions and
in the presence of the studied compounds. It was found that l-
tyrosine, kojic acid, benzoic acid, azelaic acid and ascorbic acid gave
no amperometric signal.
Also blank experiments at the tyrosinase-NMPS-mCPE without
addition of the l-tyrosine substrate were performed. No ampero-
metric response was observed for the tested inhibitors.
3.2.1. “True” enzyme inhibitors
Kojic acid, benzoic acid and azelaic acid inhibit the melanin for-
mation by competing with tyrosinase natural substrate, l-tyrosine,
for the binding to the enzyme active site. As illustrated in Fig. 3a
typical time-dependent response was obtained at the tyrosinase-
NMPSs biosensor by l-tyrosine injection followed by stepwise
additions of kojic acid. It can be seen that the biosensor baseline
was reached within 600 s allowing the experiment to be per-
formed. Once the steady state amperometric response of l-tyrosine
(3.33 × 10−4 M) was observed, successive additions of inhibitor
produced a diminution of the reduction current. This clearly indi-
cated that kojic acid interfered with the production of dopaquinone
at the electrode surface. The inhibition percentage (In %) increased
with the inhibitor concentration. Kojic acid was tested in the con-
centration range 2.50 × 10−6 to 1.24 × 10−5 M where it inhibited
between 33% and 88% of the response to l-tyrosine. Same experi-
ments were performed for benzoic acid in the concentration range
3.98 × 10−5 to 1.96 × 10−4 M where it inhibited between 34% and
81% of the response and also for azelaic acid in the concentration
range 9.90 × 10−5 to 4.76 × 10−4 M where it inhibited between 43%
and 88% of the response.
The inhibition calibration curve of kojic acid is shown in
the insert of Fig. 3. IC50 , the concentration of inhibitor which
inhibited 50% of the l-tyrosine signal, was calculated from 5
different curves by plotting 1/In(%) versus 1/inhibitor concen-
tration (Table 1): kojic acid IC50 = 3.7 × 10−6 M, RSD = 1.6%, n = 5
(y = 4.54 × 10−8 x + 7.61 × 10−3 , R2 = 0.9999, RSD slope = 3.1%, RSD
intercept = 4.1%), benzoic acid IC50 = 7.2 × 10−5 M, RSD = 1.2%, n = 5
(y = 7.88 × 10−7 x + 8.99 × 10−3 , R2 = 0.9990, RSD slope = 5.1%, RSD
intercept = 6.3%) and azelaic acid IC50 = 1.3 × 10−4 M, RSD = 3.0%,
n = 5 (y = 1.60 × 10−6 x + 7.76 × 10−3 , R2 = 0.9996, RSD slope = 9.3%,
RSD intercept = 7.6%).
The study of the kinetics and of the mechanism of inhibition
of kojic acid, benzoic acid and azelaic acid was also carried out.
The response of the tyrosinase-NMPSs biosensor to l-tyrosine was
studied in the absence and in the presence of different concentra-
tions of inhibitor (Fig. 4). Table 2 data show that approximately
the same maximum current was obtained but different values
for the apparent Michaelis–Menten constant (Km app ), determined
following a Lineweaver–Burk plot, were calculated when various
amounts of kojic acid was added in the initial testing solution. The
same experiment was performed and the same conclusions were
 V.H. Sima et al. / Talanta 83 (2011) 980–987 985

0 500 1000 1500 2000

0.00E+00 time (s)
-1.00E-09 Average of 5 curves
-2.00E-09
-3.00E-09 0.030

I (A)
Response 1 Kojic acid
-4.00E-09 0.025
Response 2 Kojic acid

1/In (%)
0.020
-5.00E-09 Response 3 Kojic acid
0.015 Response 4 Kojic acid
-6.00E-09 Response 5 Kojic acid
0.010 y = 5E-08x + 0.0076
2 Average
-7.00E-09 0.005 R = 0.9999 Linear (Average)

-8.00E-09 0.000
0.0E+00 2.0E+05 4.0E+05 6.0E+05
-1
1/conc (mol *L)

Fig. 3. Tyrosinase based biosensor amperometric response. l-Tyrosine 3.33 × 10−4 M, kojic acid inhibitor between 2.50 × 10−6 and 1.24 × 10−5 M, 0.1 M phosphate buffer pH
6.5, Eapp = −100 mV, 10 ␮L of a 1.25 mg/mL tyrosinase-NMPS suspension spiked onto the mCPE. Insert: inhibition calibration curve.

0.0E+00 5.0E+04 1.0E+05 1.5E+05 2.0E+05 2.5E+05 3.0E+05 3.5E+05 4.0E+05 4.5E+05
0.00E+00
-2.00E+09 1/L-tyrosine (M-1)

-4.00E+09 y = -7915.9x - 8E+07 a

-6.00E+09 R2 = 1

-8.00E+09
1/I (A-1)

-1.00E+10 y = -24977x - 9E+07

R2 = 0.9995
b
-1.20E+10
-1.40E+10

-1.60E+10
y = -44490x - 1E+08
-1.80E+10 c
R2 = 0.9994
-2.00E+10

Fig. 4. Tyrosinase based biosensor amperometric data: 1/signal versus 1/[l-tyrosine]. l-Tyrosine between 2.48 × 10−6 and 4.39 × 10−4 M, a: without kojic acid, b: in the
presence of 2.09 × 10−6 M kojic acid, c: in the presence of 4.18 × 10−6 M kojic acid, 0.1 M phosphate buffer pH 6.5, Eapp = −100 mV, 10 ␮L of a 1.25 mg/mL tyrosinase-NMPS
suspension deposited onto the mCPE.

dropped out also for benzoic acid and azelaic acid, data shown
in Table 2. For the three inhibitors, a different value of Imax was
obtained (−1.1 × 10−8 A, RSD = 9.2%, −1.2 × 10−8 A, RSD = 21.1% and
−5.3 × 10−9 A, RSD = 5.0%, n = 3 for kojic acid, benzoic acid and aze-
laic acid respectively). This was not due to the nature of the studied
molecule but to the fact that the inhibitors were tested at different
periods of time after the preparation of the tyrosinase-NMPS sus-
pension. However, as it can be seen from the stability tests, this
does not influence the inhibition parameters (see below).
Since, for the same inhibitor Imax did not change and Km app
increased proportionally with the inhibitor concentration, for
kojic acid (y(A−1 ) = 8.13 × 10+1 x(M−1 ) + 9.93 × 10−5 , R2 = 0.9994),
benzoic acid (y(A−1 ) = 8.42x(M−1 ) + 6.95 × 10−5 , R2 = 0.9941) and
azelaic acid (y(A−1 ) = 2.04x(M−1 ) + 7.55 × 10−5 , R2 = 0.9946), a com-
petitive inhibition process was inferred for the three studied
molecules. This conclusion is in agreement with literature data
for kojic acid [14], azelaic acid [7] and benzoic acid [19]. From
the primary Lineweaver–Burk data, secondary plots were gener-
ated, by plotting the slopes from the primary plots versus inhibitor
concentration (Table 2) in order to determine the apparent inhi-
bition constant, Ki (Fig. 5). The obtained Ki were 8.6 × 10−7 M,
2.0 × 10−5 M and 4.2 × 10−5 M for kojic acid, benzoic acid and aze-
laic acid, respectively.
3.2.2. Inhibitors of the melanogenesis process
Another inhibitor tested was ascorbic acid. Unlike kojic acid,
benzoic acid or azelaic acid, ascorbic acid does not interact with the
enzyme. It is known to inhibit the formation of melanin by reducing
the dopaquinone back to l-dopa and thus it interrupts the melanin
synthesis pathway.
The inhibition experiments were conducted in the same manner
as in the case of true tyrosinase inhibitors. A time-dependent curve
of the tyrosinase-NMPSs biosensor response to l-tyrosine substrate
followed by successive additions of ascorbic acid was realized. The
amperometric response increased as l-tyrosine was added into the
electrochemical cell. Then, due to successive additions of ascorbic
acid into the l-tyrosine solution (3.33 × 10−4 M) the reduction cur-
rent decreased clearly indicating that ascorbic acid interfered with
the reduction of dopaquinone at the electrode surface. The inhi-
bition percentage (In %) of ascorbic acid increased by raising its

Table 2
Kojic acid, benzoic acid and azelaic acid inhibition data.

Kojic acid Without kojic acid 2.09 × 10−6 M kojic acid 4.18 × 10−6 M kojic acid
Km (M) 9.7 × 10−5 2.7 × 10−4 4.4 × 10−4
Imax (A) −1.2 × 10−8 −1.1 × 10−8 −9.8 × 10−9

Benzoic acid Without benzoic acid 3.33 × 10−5 M benzoic acid 6.67 × 10−5 benzoic acid
Km (M) 8.2 × 10−5 3.3 × 10−4 6.4 × 10−4
Imax (A) −1.1 × 10−8 −1.5 × 10−8 −1.0 × 10−8

Azelaic acid Without azelaic acid 8.32 × 10−5 M azelaic acid 1.64 × 10−4 M azelaic acid
Km (M) 8.2 × 10−5 2.3 × 10−4 4.2 × 10−4
Imax (A) −5.0 × 10−9 −5.4 × 10−9 −5.6 × 10−9
986
-1.0E-04 -5.0E-05 0.0E+00 5.0E-05 1.0E-04 1.5E-04
0.00E+00
-1.00E+04
-2.00E+04
-3.00E+04
Slope
y = -4E+08x - 8085.5
b
-4.00E+04 R2 = 0.9978
-5.00E+04 a
y = -9E+09x - 7507.1
-6.00E+04 R2 = 0.9991
-7.00E+04
-8.00E+04
Fig. 5. Tyrosinase based biosensor data: slope of the data versus [In] (Table 2).
a: kojic acid, b: benzoic acid, c: azelaic acid, 0.1 M phosphate buffer pH 6.5,
Eapp = −100 mV, 10 ␮L of a 1.25 mg/mL tyrosinase-NMPS suspension deposited onto
the mCPE.
concentration. Ascorbic acid was tested in the concentration range
6.62 × 10−6 to 3.23 × 10−5 M where it was found to inhibit between
33% and 82% of the biosensor response. The inhibitory potency,
IC50 , of ascorbic acid was calculated: IC50average = 1.2 × 10−5 M,
RSD = 4.8%, n = 5 (y = 1.60 × 10−6 x + 7.76 × 10−3 , R2 = 0.9996, RSD
slope = 6.1%, RSD intercept = 7.2%) (Table 1).
By comparing the IC50 obtained in the same experimental
conditions for all the four tested inhibitors, the order of their
inhibitory potency was: kojic acid (IC50 = 3.7 × 10−6 M), ascorbic
acid (IC50 = 1.2 × 10−5 M), benzoic acid (IC50 = 7.2 × 10−5 M) and
azelaic acid (IC50 = 1.3 × 10−4 M).
Even if the obtained values of IC50 and Ki of the studied com-
pounds cannot be directly compared with literature data due to
the various factors that may contribute to the final outcomes, the
obtained results are in agreement with our expectations, based on
their practical utilization and on the general conclusions from the
literature. Both the IC50 and the Ki show that kojic acid was the
most potent inhibitor in comparison with ascorbic, benzoic and
azelaic acid. Kojic acid is considered the standard inhibitor due to
its high inhibitory potency [6] and is known as the most potent
active ingredient in the commercialized skin-whitening cosmetic
products. Ascorbic acid is well known for its medium inhibitory
property therefore it is found as adjuvant in most of the cosmetic
creams for hyperpigmentation. Azelaic acid is mainly used as an
anti-acne agent but also as adjuvant in skin-whitening creams
due to its inhibitory properties along with other more powerful
inhibitors. Benzoic acid is above all an antibacterial agent but due to
its inhibitory characteristics it is considered a preferential preserva-
tive in cosmetic creams for hyperpigmentation treatments. Benzoic
acid is generally used as a tyrosinase inhibitor in food industry.
It should be outlined that the developed device allowed the
quantification and the characterization of inhibitors from the point
of view of their interaction with tyrosinase and the enzymatic reac-
tion in the presence of the natural substrate l-tyrosine. In the skin
pigmentation process these compounds may interference also in
other biochemical steps having as final result a decrease in melanin
formation. For example, azelaic acid inhibits melanin formation
through other mechanism besides its interaction with the tyrosi-
nase active site. It may also interfere with DNA synthesis and
mitochondria activity in hyperactive and abnormal melanocytes
[7].
3.2.3. Stability and reproducibility
The long-term stability of the immobilized tyrosinase-NMPS
was evaluated by measuring the biosensor response in the pres-
V.H. Sima et al. / Talanta 83 (2011) 980–987
2.0E-04 120.0
Remaining activity (%)
100.0
conc inhibitor (M)
80.0
60.0
40.0
20.0
0.0
0 5 10 15 20 25 30 35
Time (days)
Fig. 6. Tyrosinase-NMPS stability test. Tyrosinase based biosensor, l-tyrosine sub-
strate 3.33 × 10−4 M, 0.1 M phosphate buffer pH 6.5, Eapp = −100 mV, 10 ␮L of a
y = -4E+08x - 15201 1.25 mg/mL tyrosinase-NMPS suspension dilution deposited onto the mCPE.
c
R2 = 0.9974
ence of l-tyrosine (3.33 × 10−4 M). The biosensor retained about
66% (RSD = 1.1%, n = 3) of the original response after one week, 46%
(RSD = 2.6%, n = 3) after the second week, 23% (RSD = 2.1%, n = 3)
after the third week and 5% (RSD = 3.2%, n = 3) after four weeks
(Fig. 6). When not in use the tyrosinase-NMPSs suspension was
maintained at 4 ◦ C. IC50 of kojic acid was determined at different
periods of time following enzyme immobilization. Interestingly it
was observed that IC50 was not affected by the loss of enzyme activ-
ity: the day of suspension preparation IC50 = 3.8 × 10−6 M, after one
week IC50 = 3.6 × 10−6 M, after two weeks IC50 = 3.7 × 10−6 M and
after three weeks IC50 = 3.7 × 10−6 M, RSD = 1.5%. After four weeks
the IC50 was not anymore determined because the amperometric
signal to l-tyrosine was considered too small, around 1 nA, to give
reliable results.
The IC50 for 5 consecutive determinations gave RSD values of
1.6%, 1.2%, 3.0% and 4.8% for kojic acid, benzoic acid, azelaic acid and
ascorbic acid, respectively. The good reproducibility of the results
can be explained by the fact that the inhibition signal was normal-
ized to the initial signal of the substrate for each curve. Thus, the
final results were not influenced by the variations of substrate’s sig-
nal which can occur due to the loss of the enzyme activity or due
to random errors that can happen during biosensor’s preparation.
4. Conclusion
A tyrosinase-streptavidine magnetic beads-magnetized carbon
paste biosensor for the screening of tyrosinase “true inhibitors”
and of the melanogenesis process inhibitors has been realized. The
biosensor operation mode is based on monitoring the decrease
of the amperometric signal of the electrochemical reduction of
dopaquinone, enzymatically generated from l-tyrosine substrate,
due to the presence of tyrosinase inhibitors that bind to the enzyme
active site such as kojic acid, benzoic acid or azelaic acid or due to
the presence of compounds that reduce the dopaquinone interme-
diate back to l-dopa interrupting the synthesis process like ascorbic
acid. The enzyme immobilization is performed via glutaraldehyde
activated streptavidine magnetic particles retained onto the sur-
face of a magnetized carbon paste electrode. To the best of our
knowledge this is the first electrochemical tyrosinase based biosen-
sor designed for the screening of the inhibitory potency of different
skin-whitening agents that uses the enzyme’s skin natural sub-
strate, l-tyrosine. The use of l-tyrosine as substrate in the design
of the biosensor is the main novelty of the proposed device being
the first electrochemical biosensor that mimics the two enzymatic
steps of melanin formation in skin. The methodology is simple
in use, suitable for rapid analysis in the evaluation of inhibitors
potency.
The obtained results are in agreement with the expected rank-
ing of the inhibitory potency of the studied skin whitening agents.
The reproducibility of the results is very good and is not influenced
 V.H. Sima et al. / Talanta 83 (2011) 980–987
neither by the loss of the enzyme activity nor by other variations
that can occur during the immobilization process or which can be
due to electrode sensitivity fluctuation. In this respect the use of
magnetic particles for tyrosinase immobilization can be regarded
as an interesting approach since it allows their use “on demand”
and since the sensing electrode is not affected by an enzymatic
immobilization process.
Other substances acting through one of the above mentioned
mechanisms could be evaluated using the developed device. Tak-
ing into consideration the relative simplicity of fabrication and use
of this biosensor, it may represent an interesting and valuable ana-
lytical tool especially for the evaluation of tyrosinase – inhibitors.
Acknowledgements
Thanks are expressed to WBI and AUF for fellowship to V.S. and
TUBIKA (Turkey) for fellowship to Z.A.
References
[1] Y.J. Kim, H. Uyama, Cell. Mol. Life Sci. 62 (2005) 1707–1723.
[2] S. Parvez, M. Kang, H.S. Chung, H. Bae, Phytother. Res. 21 (2007) 805–816.
[3] T.S. Chang, Int. J. Mol. Sci. 10 (2009) 2440–2475.
[4] A. Rescigno, F. Sollai, B. Pisu, A. Rinaldi, E. Sanjust, J. Enzyme Inhib. Med. Chem.
17 (2002) 207–218.
[5] F. Solano, S. Briganti, M. Picardo, G. Ghanem, Pigment Cell Res. 19 (2006)
550–571.
[6] J.M. Noh, S.Y. Kwak, H.S. Seo, J.H. Seo, B.G. Kim, Y.S. Lee, Bioorg. Med. Chem.
Lett. 19 (2009) 5586–5589.
[7] S. Briganti, E. Camera, M. Picardo, Pigment Cell Res. 16 (2003) 101–110.
[8] S. Li, Y. Tan, P. Wang, J. Kan, Sens. Actuators B 144 (2010) 18–22.
[9] S.E. Stanca, I.C. Popescu, J. Mol. Catal. B Enzyme 27 (2004) 221–225.
[10] S.H. Jeon, K.H. Kim, J.U. Koh, K.H. Kong, Bull. Korean Chem. Soc. 26 (2005)
1135–1137.
[11] K. Maeda, M. Fukuda, J. Pharmacol. Exp. Ther. 276 (1996) 2765–2769.
[12] S.S.K. Tai, C.G. Lin, M.H. Wu, T.S. Chang, Int. J. Mol. Sci. 10 (2009) 4257–4266.
[13] K. Streffer, H. Kaatz, C.G. Bauer, A. Makower, T. Schulmeister, et al., Anal. Chim.
Acta 362 (1998) 81–90.
[14] Y. Hasebe, K. Oshima, O. Takise, S. Uchiyama, Talanta 42 (1995) 2079–2085.
987
[15] D. Shan, Q. Shi, D. Zhu, H. Xue, Talanta 72 (2007) 1767–1772.
[16] Y.D. Tanimoto de Albuquerque, L.F. Ferreira, Anal. Chim. Acta 596 (2007)
210–221.
[17] Z. Liu, Y. Liu, H. Yang, Y. Yang, G. Shen, R. Yu, Anal. Chim. Acta 533 (2005) 3–9.
[18] N. Smit, J. Vicanova, S. Pavel, Int. J. Mol. Sci. 10 (2009) 5326–5349.
[19] D. Shan, Q. Li, H. Xue, S. Cosnier, Sens. Actuators B 134 (2008) 1016–1021.
[20] S. Li, Y. Tan, P. Wang, J. Kan, Sens. Actuators B Chem. 144 (2010) 18–22.
[21] M.D. Morales, S. Morante, A. Escarpa, M.C. Gonzalez, A.J. Reviejo, J.M. Pingarron,
Talanta 57 (2002) 1189–1198.
[22] Z.P. Zheng, K.W. Cheng, J. Chao, J. Wu, M. Wang, Food Chem. 106 (2008)
529–535.
[23] O. Nerya, R. Musa, S. Khatib, S. Tamir, J. Vaya, Phytochemistry 65 (2004)
1389–1395.
[24] S. Okombi, D. Rival, S. Bonnet, A.M. Mariotte, E. Perrier, A. Boumendjel, Bioorg.
Med. Chem. Lett. 16 (2006) 2252–2255.
[25] L. Campanella, A. Bonanni, E. Martini, N. Todini, M. Tomassetti, Sens. Actuators
B 111–112 (2005) 505–514.
[26] T.M. Anh, S.V. Dzyadevych, M.C. Van, N.J. Renault, C.N. Duc, J.M. Chovelon,
Talanta 63 (2004) 365–370.
[27] X. Wang, L. Chen, S. Xia, Z. Zh, J. Zhao, J.M. Chovelon, N.J. Renault, Int. J. Elec-
trochem. Sci. 1 (2006) 55–61.
[28] M.D. Morales, M.C. Gonzalez, B. Serra, A.J. Reviejo, J.M. Pingarron, Sens. Actua-
tors B 106 (2005) 572–579.
[29] L. Kong, S. Huang, Z. Yue, B. Peng, M. Li, J. Zhang, Microchim. Acta 165 (2009)
203–209.
[30] E. Asav, E. Yorganci, E. Akyilmaz, Talanta 78 (2009) 553–556.
[31] I. Narli, S. Kiralp, L. Toppare, Anal. Chim. Acta 572 (2006) 25–31.
[32] B. Rittich, A. Spanova, D. Horak, M.J. Benes, L. Klesnilova, K. Petrova, A. Rybnikar,
Colloids Surf. B 52 (2006) 143–148.
[33] Ademtech (Bordeaux-France), Product Specification Leaflet,
http://www.ademtech.com/products.aspx?id p=23.
[34] D. Yu, O. Dominguez Renedo, B. Blankert, V. Sima, R. Sandulescu, J. Arcos, J.M.
Kauffmann, Electroanalysis 18 (2006) 1637–1642.
[35] S. Patris, C. De Vriese, F. Prohoroff, E. Burgoa Calvo, J. Arcos Martinez, J.M.
Kauffmann, Electroanalysis 22 (2010) 41–48.
[36] D.R. Thevenot, K. Toth, R.A. Durst, G.S. Wilson, Biosens. Bioelectron. 16 (2001)
121–131.
[37] A. Slominski, D.J. Tobin, S. Shibahara, J. Wortsman, Physiol. Rev. 84 (2004)
1155–1228.
[38] E. Sanjust, G. Cecchini, F. Sollai, N. Curreli, A. Rescigno, Arch. Biochem. Biophys.
412 (2003) 272–278.
[39] A. Amine, H. Mohammadi, I. Bourais, G. Palleschi, Biosens. Bioelectron. 21
(2006) 1405–1423.
[40] F.N. Kok, F. Bozoglu, V. Hasirci, Biosens. Bioelectron. 17 (2002) 531–539.
[41] B. Lejczak, P. Kafarski, E. Makowiecka, Biochem. J. 242 (1987) 81–88.
 Talanta 186 (2018) 279–285

Contents lists available at ScienceDirect

Talanta
journal homepage: www.elsevier.com/locate/talanta

Calcium-selective electrodes based on photo-cured polyurethane-acrylate T

membranes covalently attached to methacrylate functionalized poly(3,4-
ethylenedioxythiophene) as solid-contact
⁎ ⁎
Cristina Ocañaa,b, Natalia Abramovaa,c, Andrey Bratova, , Tom Lindforsb, Johan Bobackab,
a
BioMEMs Group, Instituto de Microelectrónica de Barcelona, Centro Nacional de Microelectrónica (IMB-CNM, CSIC), Campus UAB, Bellaterra 08193, Spain
b
Laboratory of Analytical Chemistry, Faculty of Science and Engineering, Johan Gadolin Process Chemistry Centre, Åbo Akademi University, Biskopsgatan 8, FIN-20500
Turku-Åbo, Finland
c
Laboratory of Artificial Sensors Systems, ITMO University, Kronverskiy pr. 49, St. Petersburg 197101, Russia

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

Keywords: We report here the fabrication of solid-contact calcium-selective electrodes (Ca2+-SCISEs) made of a poly-
Tetramethacrylate poly(3 urethane acrylate ion-selective membrane (ISM) that was covalently attached to the underlying ion-to-electron
4-ethylenedioxythiophene) transducer (solid-contact). Methacrylate-functionalized poly(3,4-ethylenedioxythiophene) (Meth-PEDOT) and
MWCNT Meth-PEDOT films containing either multiwalled carbon nanotubes (MWCNT) or carboxylated MWCNT
All-solid-state ion-selective electrode
(cMWCNT) were used as solid contacts. The solid contacts were deposited by drop-casting on screen-printed
Calcium
electrodes and characterized by cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS) and
Polyurethane membrane
potentiometry. Covalent binding between the solid contact and the ISM was obtained via photopolymerization in
order to increase the robustness of the Ca2+-SCISEs. The performance of the Ca2+-SCISEs was studied by
measuring their potentiometric response and their sensitivity to light, oxygen and carbon dioxide. Meth-PEDOT
was found to be a promising solid-contact material to develop low-cost and easy to prepare ISEs.

1. Introduction miniaturization [24], low-cost production and compatibility with mass

Over the years, the synthesis and characterization of conducting
polymers has attracted attention due to their unique electronic and
optical properties [1,2] in many applications such as electrochromic
devices [3–5], light emitting diodes [6–8], energy storage devices
[9–11], biosensors [12–14] and all-solid-state ion sensors [15–18].
Among electronically conducting polymers, there is a considerable in-
terest in using poly(3,4-ethylenedioxythiophene) (PEDOT) because of
its low oxidation potential, high stability in ambient environments
[19,20], mechanical flexibility and stable oxidized form [21]. Fur-
thermore, addition of electrically conducting nanomaterials (e.g. gra-
phene or carbon nanotubes) to PEDOT is known to improve the elec-
trical conductivity by better connecting individual conducting PEDOT
domains in the composite film [22,23].
In this work, methacrylate-functionalized PEDOT (Meth-PEDOT),
Meth-PEDOT containing MWCNT (Meth-PEDOT-MWCNT) and
cMWCNT (Meth-PEDOT-cMWCNT) were used as the ion-to-electron
transducer in potentiometric solid-contact calcium-selective electrodes
(Ca2+-SCISEs). In recent years, the classical liquid contact ISEs are
being replaced by SCISEs due to the possibility of robust
production by standard microfabrication techniques [25]. Most of the
reported SCISEs with conductive polymers use plasticized poly(vinyl
chloride) (PVC) as the ion-selective membrane (ISM) [26–28]. How-
ever, the PVC-based ISEs suffer from diffusion of water through the ISM
and the possible formation of a water layer or water pools at the ISM/
solid-contact (SC) or SC/substrate interfaces resulting in poor potential
stability and weakening of the ISM adhesion to the substrate [29,30].
Photocurable membranes are alternative materials to PVC-ISMs.
Several ISEs using photocurable ISMs have been developed since their
introduction in the middle of 1980s and 1990s [31,32]. Photo-cured
polymeric systems show many advantages over PVC-ISMs such as the
possibility to use standard photolithography processes, which may be
beneficial for mass production of ISEs, and the opportunity to obtain
ISMs with low leaching rates of the active components [33]. Urethane-
acrylate ISMs are one of the most used photo-cured membrane types
because of their fast curing rates and their compatibility with common
plasticizers and ionophores [34]. In addition, this type of ISMs present
high durability and reduced biofouling when SCISEs are used in bio-
logical samples [35] and they may possibly be chemically “anchored”
to the solid electrode substrate [32]. However, it must be noted that it is

⁎
Corresponding author.
E-mail addresses: andrei.bratov@imb-cnm.csic.es (A. Bratov), jbobacka@abo.fi (J. Bobacka).

https://doi.org/10.1016/j.talanta.2018.04.056
Received 23 February 2018; Received in revised form 17 April 2018; Accepted 19 April 2018
Available online 25 April 2018
0039-9140/ © 2018 Elsevier B.V. All rights reserved.
C. Ocaña et al.
not the first time that Meth-PEDOT and ISM matrix are copolymerized.
Rzewuska et al. reported SCISEs based on the copolymerization of
polyacrylate-based membrane and Meth-PEDOT on a glassy carbon
electrode surface forming a single phase membrane matrix [36].
However, in this approach the polymerization time was 5 min and in
our previous studies we observed that when the polymerization time
increases the selectivity of the SCISEs becomes worse due to the sodium
tetrakis[3,5-bis-(trifluoromethyl)phenyl]borate photobleaching [37].
The main aim of this work is electrochemical characterization of
Meth-PEDOT, Meth-PEDOT-MWCNT and Meth-PEDOT-cMWCNT and
their application as ion-to-electron transducers in polyurethane (PU)
based Ca2+-SCISEs. In this paper, we present a simple, low-cost and
robust method for preparing Ca2+-SCISEs, which might be beneficial
for different fields such as ranching where they require easy to use and
semiautomatic analytical systems with very low costs per analysis. The
end-capped methacrylate groups of Meth-PEDOT functioned as reactive
sites for obtaining better bonding and stronger adhesion to the PU-ac-
rylate ISM during the copolymerization process. This was expected to
enhance the durability of the Ca2+-SCISEs.
2. Experimental
2.1. Chemicals
Tetramethacrylate end-capped poly(3,4-ethylenedioxythiophene)
solution (Meth-PEDOT, 0.5 wt% dispersion in propylene carbonate
carbonate and p-toluenesulfonate as charge compensating ion), KCl,
KNO3, CaCl2, Ca(NO3)2 and carboxylated MWCNTs were obtained from
Sigma Aldrich. MWCNTs were obtained from DropSens (Oviedo,
Spain). Calcium Ionophore II (ETH 129), bis(2-ethylhexyl) sebacate
(DOS), potassium tetrakis(p-chlorophenyl)borate (KTpClPB), ETH500
and hexafluorobutyl acrylate (HFBuA) were obtained from Fluka.
Aliphatic urethane diacrylate (oligomer Ebecryl 270), cross-linker and
hexanediol diacrylate (HDDA) were from UCB Chemicals and the
photoinitiator 2,2- dimethoxy-2-phenylacetophenone (IRG 651) from
Ciba-Geigy. All other chemicals used were analytical reagent grade and
all solutions were prepared using ultrapure deionized water with the
resistivity of 18.2 MΩ cm (Milli-Q system, Millipore, Billerica, MA).
Screen-printed electrodes (SPEs) with silver ink and flash gold
(d=2 mm) prepared on PET polymer substrates were supplied by FAE
(FAE S.A., Spain).
2.2. Meth-PEDOT, Meth-PEDOT-MWCNT/cMWCNT deposition
5.0 µL (2 ×2.5 µL) of Meth-PEDOT solution containing either
0.2–0.5 wt% MWCNT or cMWCNT solution was drop-cast on the screen-
printed electrodes to form SC layers of Meth-PEDOT-MWCNT and
Meth-PEDOT-cMWCNT. After that, the electrodes were placed on a
heating plate at 50 °C to increase the evaporation rate of the propylene
carbonate solvent. All further experiments were performed at room
temperature.
2.3. Ca2+-selective ISM deposition
The photocurable ISM composition was prepared as presented pre-
viously [38]. Briefly, the main polymer composition was prepared by
mixing the aliphatic urethane diacrylate oligomer Ebecryl 270 with the
reactive diluent HDDA and the photoinitiator Irgacure 651 in the ratio
(w/w) of 81:17:2. In the next step, 0.3 g of the main polymer compo-
sition was dissolved in 0.2 ml THF and the plasticizers DOS and HFBuA,
Ca2+-ionophore II and lipophilic salts were then added to this solution.
This ISM solution was placed in an ultrasonic bath until it was homo-
geneous and then left unstirred for several hours to allow evaporation
of THF from the solution. This resulted in the following ISM composi-
tion: main polymer mixture (56.5 wt%), DOS (20.0 wt%), HFBuA
(20.0 wt%), Ca2+-ionophore II (1.0%), KTpClPB (0.5 wt%) and ETH
280
Talanta 186 (2018) 279–285
500 (2.0 wt%).
ISMs were deposited on the screen-printed electrode substrates (Ag
and Au) by applying the ISM solution with a microsyringe on the Meth-
PEDOT, Meth-PEDOT-MWCNT and Meth-PEDOT-cMWCNT solid-con-
tact layers. After the ISM deposition, the electrodes were exposed to
365 nm UV light for 20 s using a UV Curing Light Lamp 8141 (Düren,
Germany) to form covalent bonds between the ISMs and the acrylate
groups of SCs. A scheme of the copolymerization process is shown in
Fig. S1. Crosslinking between the PU-ISM and polyaniline functiona-
lized with methacrylate groups was confirmed by FTIR-ATR analysis in
previous studies [32]. Therefore, we assume that crosslinking occurs
also between the PU-ISM and the Meth-PEDOT used in this work. The
thicknesses of the ISM was ca. 200 µm, as measured with a micrometer
screw gauge (resolution: 1 µm) after the polymerization process.
2.4. Cyclic voltammetry (CV) and electrochemical impedance spectroscopy
(EIS)
Cyclic voltammograms and impedance spectra were recorded by
using a one-compartment three-electrode cell connected to the Autolab
potentiostat equipped with a frequency response analyzer (AUT20.
FRA2-AUTOLAB, Eco Chemie, B.V., The Netherlands). The screen-
printed electrodes, a glassy carbon rod and a double junction Ag/AgCl/
3 M KCl//0.1 M LiOAc served as the working (WE), auxiliary (AE) and
reference electrode (RE), respectively. CVs were recorded between
− 0.5 V and + 0.4 V in 0.1 M KNO3 at a scan rate ν = 0.01 Vs−1. The
impedance spectra were measured at the open circuit potential (Edc) in
0.1 M KNO3 within the frequency range (f = 10 kHz – 0.01 Hz) using an
excitation signal amplitude ΔEac = 10 mV. The impedance spectra were
fitted to equivalent circuits with the ZView software (Scribner
Associates Inc., USA).
2.5. Chronopotentiometric measurements
Constant-current chronopotentiograms were recorded in 0.1 M
CaCl2 with the Autolab potentiostat described above by passing a
constant current of ± 1 nA through the Ca2+-SCISEs for 60 s [39].
2.6. Potentiometric measurements
The potentiometric response of the Ca2+-SCISEs was measured in a
Faraday's cage with a 16-channel potentiometer (Lawson Labs, Inc.,
input impedance: 1015 Ω) and by using the double junction Ag/AgCl/
3 M KCl//0.1 M LiOAc as the RE. The Ca2+-SCISEs were preconditioned
under stirring in 10−3 M CaCl2 for 3 days prior to the potentiometric
measurements. Calibration plots were obtained in CaCl2 solutions from
10−7 M to 10−2 M by stirring the solutions for 2 min at each con-
centration. The activity coefficients were calculated according to the
Debye-Hückel equation.
The selectivity coefficients towards Na+, K+, Mg2+, NH4+ and Li+
ions were determined at 0.1 M concentrations of interfering ions with
the mixed solution method recommended by IUPAC [40]. The po-
tentiometric aqueous layer tests were performed by recording the Ca2+-
SCISE potentials in 0.1 M CaCl2, 0.1 M NaCl and 0.1 M CaCl2 solutions
(in this order). The CO2 and O2 sensitivities of the Ca2+-SCISEs were
obtained by measuring the potential of the Ca2+-SCISEs in a 0.1 M
CaCl2 solution that was purged with pure O2, N2 and CO2 gas in the
following sequence: N2 (1 h), O2 (1 h), N2 (1 h), CO2 (1 h), N2 (1 h). The
light sensitivity was measured in 0.1 M CaCl2 solution by monitoring
the Ca2+-SCISE potentials as the illumination of the electrode surfaces
changed in the following order: darkness (30 min), ambient room light
(30 min), cold light (30 min), ambient room light (30 min) and darkness
(30 min). The Leica CLS 150 XE cold light source (150 W, 21 V) was
directed through the electrolyte solution towards the Ca2+-SCISEs
surfaces.
C. Ocaña et al.
Fig. 1. CVs of screen-printed Meth-PEDOT electrodes containing: (1) 0 wt%
MWCNT/cMWCNT, (2) 0.2 wt% MWCNT, (3) 0.5 wt% MWCNT, (4) 0.2 wt%
cMWCNT and (5) 0.5 wt% cMWCNT; CVs show the 100th cycle in 0.1 M KNO3
with ν = 10 mV s−1.
3. Results and discussion
3.1. CV and EIS characterizations
According to Nikolskii and Materova, the SCISEs must have re-
versible transition from ionic to electronic conductivity at the ISM/SC
interface and sufficiently high exchange currents in comparison with
the current generated by the potential measurement circuitry to obtain
a stable electrode potential [40]. These requirements imply that the
solid-contact should have a high redox capacitance to become less po-
larizable and provide a more stable potential [41].
As shown in Fig. 1, the CVs of the Meth-PEDOT, Meth-PEDOT-
MWCNT and Meth-PEDOT-cMWCNT films measured after 100 potential
cycles in 0.1 M KNO3 showed the typical oxidation and reduction be-
havior of PEDOT [42]. The rather symmetrical shape of the CVs of all
Meth-PEDOT films indicates high reversibility of the doping process.
The Meth-PEDOT-MWCNT/cMWCNT films (containing CNTs) give in
general higher oxidation and reduction currents than the Meth-PEDOT
film prepared without CNTs. The highest currents are obtained with the
Meth-PEDOT film containing 0.5 wt% cMWCNT. The improved redox
behavior is due to the MWCNT and cMWCNT facilitating the formation
of a more extended electrically conductive network in the Meth-PEDOT
matrix by connecting isolated conducting PEDOT domains, which en-
hances the electron transfer at the PEDOT-SC/screen-printed electrode
interface [43]. The improved redox behavior is more significant for
cMWCNT than for MWCNT due to the presence of carboxyl groups,
which can more efficiently participate in the charge compensation
process of the Meth-PEDOT films during its oxidation and reduction in
0.1 M KNO3.
The impedance spectra of the Meth-PEDOT and Meth-PEDOT-
MWCNT/cMWCNT solid-contacts in Fig. 2a support the results obtained
by CV. The impedance spectra show an almost vertical capacitive be-
havior at low frequencies except for Meth-PEDOT, which shows a ty-
pical diffusional behavior indicating slower ion and electron transport
than in the other SCs. The impedance data for Meth-PEDOT-MWCNT/
cMWCNT was fitted to the equivalent circuit presented in Fig. 2b
composed by the solution resistance (R), the finite-length Warburg
diffusion element (ZD) and the constant phase element (CPE) in series.
The equivalent circuit is similar to the one developed earlier for elec-
tropolymerized PEDOT [40]. The ZD element is related to the
281
Talanta 186 (2018) 279–285
Fig. 2. a) Electrochemical impedance spectra of the screen-printed Meth-
PEDOT electrodes containing: (1) 0 wt% MWCNT/cMWCNT, (2) 0.2 wt%
MWCNT, (3) 0.5 wt% MWCNT, (4) 0.2 wt% cMWCNT and (5) 0.5 wt%
cMWCNT. The spectra were measured at the open circuit potential (Edc) in
0.1 M KNO3 with ΔEac = 10 mV and f = 100 kHz – 10 mHz. b) The equivalent
circuit used for fitting the impedance spectra where R, ZD and CPE are the
solution resistance, finite-length Warburg diffusion element and constant phase
element, respectively.
diffusional time constant (τD), the pseudocapacitance (CD) and the
diffusion resistance (RD = τD / CD) [44]. Fig. S2 compare the experi-
mental and fitted EIS data in the form of Nyquist plots and Table S1
summarizes the results of the fittings. The impedance spectrum for
Meth-PEDOT (0 wt% MWCNT/cMWCNT) differs from the rest of the
electrodes (Fig. 2a) and did not give a good fit to the equivalent circuit
in Fig. 2b, as shown in Fig. S2. Therefore, Meth-PEDOT (0 wt%
MWCNT/cMWCNT) was left out from the comparison shown in Table
S1. As expected, the addition of MWCNTs/cMWCNTs increases the
capacitance of Meth-PEDOT (Table S1) since the CNTs interconnect
isolated PEDOT segments, thus improving the oxidation and reduction
of PEDOT and lowering its resistance. Additionally, charging of the
MWCNT/cMWCNT surface may also contribute to the higher capaci-
tance. The CPE and n values of 1.10–3.40 mF and 0.94–0.98, respec-
tively, describe almost an ideal capacitor (n = 1) and the Meth-PEDOT-
cMWCNT films show the highest capacitance of all SCs studied here.
This assumption is supported by the results of the EIS fittings shown in
Table 1 revealing that the Meth-PEDOT-cMWCNTs have the largest
redox capacitance (CD) and the smallest diffusion resistance (RD). These
data suggest that the Meth-PEDOT-cMWCNT composite films are best
suited as solid-contacts among the Meth-PEDOT films studied here for
the PU based Ca2+-SCISEs.
3.2. Potentiometric response of the Meth-PEDOT-based solid-contacts
Initially the potentiometric response of the Meth-PEDOT and Meth-
PEDOT-MWCNT/cMWCNT films was studied in the absence of calcium
ions in KNO3 solutions with results presented in Fig. 3. The Meth-
C. Ocaña et al. Talanta 186 (2018) 279–285

Table 1
Results of the chronopotentiometry measurements with the Ca2+-SCISEs where
R, ΔE/Δt and CL are the bulk resistance of the ISM, potential drift of the Ca2+-
SCISEs and the redox capacitance of the SC. We also show for comparison the
redox capacitances of the bare Meth-PEDOT and Meth-PEDOT-MWCNT/
cMWCNT solid-contacts (prepared on screen-printed electrodes, without ISM)
obtained for from CV (CCV) and EIS (CEIS) measurements.
wt% wt% R (MΩ) ΔE/Δt CL (µF) CCV (µF) CEIS (µF)
MWCNT cMWCNT (µV/s)

– – 89.1 34.4 29.1 ± 8.2 400 -a

0.2 – 89.3 21.1 47.3 ± 7.9 521 250
0.5 – 91.1 21.9 45.7 ± 11 943 382
– 0.2 116.8 28.2 35.3 ± 9.2 980 610
– 0.5 85.7 33.3 30.0 ± 10 3070 2475

a
Value not available, because EIS data were not fitted to the equivalent
circuit.

Fig. 4. Chronopotentiograms of the Ca2+-SCISEs prepared with the Meth-

PEDOT solid-contacts containing: (1) 0 wt% MWCNT/cMWCNT, (2) 0.2 wt%
MWCNT, (3) 0.5 wt% MWCNT, (4) 0.2 wt% cMWCNT and (5) 0.5 wt%
cMWCNT. The measurements were done in 0.1 M CaCl2 by first applying
+ 1 nA for 60 s and then −1 nA for 60 s.

The cationic response in presence of mobile Ca2+ cations is illustrated

by Eq. (3):
We may also note here that PEDOT electropolymerized in the pre-
sence of the immobile and negatively charged polystyrenesulfonate
anion (PSS-) gives a cationic potentiometric response [19] due to the
oxidation-reduction reaction similar to presented in Eq. (3).

(Meth−PEDOT 0cMWCNT−)2 Ca2 + ⇌ (Meth−PEDOT+cMWCNT−) 2

+ Ca2 + + 2e− (3)

3.3. Chronopotentiometric measurements with Ca2+-SCISEs

Constant-current chronopotentiometric measurements were per-

Fig. 3. Potentiometric response of screen-printed Meth-PEDOT solid-contacts
containing either MWCNTs or cMWCNTs in 10−8-10−1 M KNO3: (1) 0 wt%
MWCNT/cMWCNT, (2) 0.2 wt% MWCNT, (3) 0.5 wt% MWCNT, (4) 0.2 wt%
cMWCNT and (5) 0.5 wt% cMWCNT.
PEDOT and the Meth-PEDOT-MWCNTs films showed an anionic re-
sponse at concentrations > 10−5 M implying that the small and mobile
NO3- (in comparison to the bulky immobilized MWCNTs) function as
the main charge compensating anion in the oxidation and reduction
reaction of Meth-PEDOT according to Eqs. (1) and (2), which includes
also electron transfer between Meth-PEDOT and the electronically
conducting substrate (electrical contact). We assume that the less bulky
p-toluenesulfonate (pTS) used originally as the charge compensating
anion in the commercial Meth-PEDOT formulation (according to Sigma-
Aldrich) is exchanged to NO3- in the PEDOT matrix during the con-
ditioning of the electrodes in 0.1 M KNO3. In Eq. (2), we assume (for
simplicity) that the MWCNTs have a surface charge close to zero:
Meth−PEDOT 0 + NO3− /pTS− ⇌ Meth−PEDOT+NO3− +e−
Meth−PEDOT 0MWCNT + NO3− ⇌ Meth−PEDOT+MWCNTNO3− +e−
On the other hand, the potentiometric response of the Meth-PEDOT-
cMWCNT films was only slightly anionic indicating that the films ex-
change both anions and cations, in addition to electron transfer, re-
sulting in a thermodynamically more well-defined ISM/SC interface.
formed in order to critically evaluate the potential stability of the dif-
ferent types of Ca2+-SCISEs [19]. The slope (ΔE/Δt) of the response
curves provides a direct measure of the potential stability of the Ca2+-
SCISEs upon passage of a small current. As can be seen in Fig. 4 and
Table 1, all Ca2+-SCISEs prepared with the Meth-PEDOT and Meth-
PEDOT-MWCNT/cMWCNT solid-contacts showed rather similar po-
tential drifts. From these measurements in Fig. 4, we can estimate the
total resistance of the PU-ISM and the SC, and the redox capacitance
(CL) of the SC covered by the PU-ISM. We should point out here that the
resistance of the SC is usually negligible compared to that of the ISM.
For the Ca2+-SCISEs studied, the bulk resistance is obtained from the
potential difference (ΔE) in the chronopotentiograms when the current
direction is reversed at t = 60 s. As expected, all PU-ISMs had relatively
high bulk resistances of 86–117 MΩ (Table 1). However, the buried SCs
had rather similar redox capacitances varying between 29 and 47 µF.
Especially the Meth-PEDOT and Meth-PEDOT-cMWCNTs had almost
identical redox capacitances of 29–35 µF showing that the cMWNCTs
(1) do not improve the redox capacitance of the PEDOT-SC in Ca2+-SCISEs.
We determined therefore also the redox capacitance of the non-coated
Meth-PEDOT and Meth-PEDOT-MWCNT/cMWCNT solid-contacts
(without ISM) from the CV and EIS measurements. Table 1 reveals that
(2)
after deposition of the PU-ISM the capacitances decreased to 3.6–10.7%
from their initial values for the uncoated SCs. For the Meth-PEDOT
prepared 0.5 wt% cMWCNTs, the redox capacitance dropped to only ca.
1% due to its considerably higher initial capacitance than for the other
SCs. It can thus be concluded that the redox process of the SCs beneath

282
C. Ocaña et al. Talanta 186 (2018) 279–285

Table 2
Reproducibility of the standard potential (E0), potentiometric slopes (S) of the
linear part of the calibration curves and the detection limit (LOD) for the Ca2+-
SCISEs (n = 6).
wt% MWCNT wt% cMWCNT E0 (mV) S (mV/pCa) LOD (M)

– – 429.2 ± 12.9 28.8 ± 0.8 (8.2 ± 2.1)·10−6

0.2 – 422.4 ± 50.2 29.3 ± 1.0 (3.4 ± 3.1)·10−6
0.5 – 461.4 ± 16.3 30.3 ± 1.9 (7.8 ± 3.9)·10−6
– 0.2 484.9 ± 13.2 30.6 ± 0.7 (6.3 ± 2.0)·10−6
– 0.5 366.6 ± 37.0 29.3 ± 1.9 (7.1 ± 2.4)·10−6

Ca2+-SCISEs with Meth-PEDOT and Meth-PEDOT (0.2 wt% cMWCNT)

as solid-contact showing the lowest standard deviation of E°
( ± 13 mV; n = 6). In contrary to other studies [43,45], our results
indicate that we did not obtain any significantly improvements in the
potentiometric response of the Ca2+-SCISEs by adding MWCNTs/
cMWCNTs to the methacrylate functionalized PEDOT-SC. This dis-
crepancy can be related to the fact that the transduction mechanism
(i.e. oxidation and reduction reaction) of the SC is limited by the PU-
ISM. This is probably due to the much higher bulk resistance of the PU-
ISM (86–117 MΩ) compared to the plasticized PVC-ISM (usually ca.
2–5 MΩ) [19,40], indicating that the diffusion coefficients of ionic
species are lower in the PU-ISM. It cannot therefore provide the SC with
a sufficient amount of charge compensating anions or/and cations. Fig.
S3 shows the potential traces of the Ca2+-SCISE with Meth-PEDOT in
order to illustrate the short-time potential stability and the response
time of the sensor.
The selectivity coefficients of the Ca2+-SCISEs towards Na+, K+,
2+
Mg and Li+ were determined by the mixed solution method in the
presence of 0.1 M background concentration of the interfering ions
(Table S2). We obtained the following selectivity coefficients log KCa2+ , Na
+

= − 3.3, log KCa, K = − 3.4, log KCa, Li = − 3.1 and log KCa, Mg =
2++ 2++ 2+ 2+

− 3.8. These values are in good agreement with previously reported

selectivity coefficients for ISFETs and Ca2+-SCISEs with the same ISM
composition as used in this work [32,38]. This indicates that the se-
lectivity of the Ca2+-SCISEs is not influenced by the type of the SC. It
may be mentioned that ISE with PVC membranes prepared with 2-ni-
trophenyl octyl plasticizer show better selectivity [46]. However, it is not
possible to use compounds containing nitro-group in formulations of
photocurable polymers as this nitro-group inhibits the radical initiation
of the polymerization.

Fig. 5. a) Calibrations plots of the Ca2+-SCISEs in 10−7 to 10−2 M CaCl2 so-

lution. Dotted line represents the ideal Nernstian plot. b) The potentiometric
aqueous layer test in 0.1 M primary ion (Ca2+) and interfering ion (Na+) so-
lutions for the Ca2+-SCISEs prepared with the Meth-PEDOT solid-contacts
containing: (1) 0 wt% MWCNT/cMWCNT, (2) 0.2 wt% MWCNT, (3) 0.5 wt%
MWCNT, (4) 0.2 wt% cMWCNT and (5) 0.5 wt% cMWCNT.
the ISM and the ion transport associated with it, which is necessary for
the oxidation and reduction reaction of Meth-PEDOT, is hampered by
the PU-ISM [39].
3.4. Calibration of the Ca2+-SCISEs
Fig. 5a shows the potentiometric response of the different Ca2+-
SCISEs in 10−7 M to 10−2 M CaCl2 solutions (calibration from low to
high concentrations). All electrodes showed almost Nernstian slopes,
which were calculated from the linear part of the calibration curves,
and rather similar detection limits (LOD) of (3.4–8.2)× 10−6 M
(Table 2). We determined the reproducibility of the standard potential
(E°) by extrapolating the linear section of the calibration curve to aCa
= 1 (i.e. log aCa = 0). Table 2 summarizes the reproducibility of the
different Ca2+-SCISEs. The best reproducibility was obtained for the
3.5. Potentiometric aqueous layer test
One of the major problems of SCISEs is penetration of water through
the polymer membrane and its condensation at the membrane/SC in-
terface. This produces instabilities in sensor response and membrane
adhesion failure. The water layer/pool formation at the interface PU-
ISM/SC was studied by the potentiometric aqueous layer test, in which
the potential drifts of the Ca2+-SCISEs were measured in primary and
interfering ion solutions for more than 20 h [47]. As can be seen in
Fig. 5b, Ca2+-SCISEs prepared with and without MWCNT showed ra-
ther stable potentials and relatively quick recovery of the sensor signal
after changing the solutions. This indicates that the cross-linking be-
tween the SCs and the ISM prevents the aqueous layer formation at the
PU-ISM/SC interface.
3.6. Oxygen, carbon dioxide and light sensitivity of the Ca2+-SCISEs
We also investigated the influence of oxygen (O2) and carbon di-
oxide (CO2) on the potential stability of the Ca2+-SCISEs which are of
fundamental importance to optimize the performance of the sensors.
Fig. S4 reveals that all the Ca2+-SCISEs showed practically no O2 sen-
sitivity and negligible CO2 sensitivity. In order to determine the light
sensitivity, the potential drift of the Ca2+-SCISEs were measured when

283
C. Ocaña et al.
the illumination conditions changed from darkness, room light to cold
light. Fig. S5 shows stable electrode potentials for the Ca2+-SCISEs
prepared with Meth-PEDOT and Meth-PEDOT containing 0.2 wt%
cMWCNT, whereas the electrodes prepared with Meth-PEDOT-MWCNT
showed a minor potential drift during the first hour after the illumi-
nation was changed from dark to room light. We conclude that the
Ca2+-SCISEs based on Meth-PEDOT and Meth-PEDOT containing
cMWCNT are not light sensitive under normal measurement conditions
in room light.
3.7. Durability of the Ca2+-SCISEs
To check the quality of the membrane/conducting polymer inter-
face formed by copolymerization in prevention of water penetration
Ca2+-SCISEs were kept in constant contact with 10−3 M CaCl2 solution
during three months. After this period of time it was found that all
electrodes except those prepared with Meth-PEDOT containing 0.2 wt%
MWCNT (no response) and 0.5 wt% cMWCNT show close to Nernstian
slopes of 26 ± 1 mV/pCa2+. The Meth-PEDOT-0.2% MWCNT lost it
response and sensors with 0.5% cMWCNT showed sensitivity less than
20 mV/decade. The average E0 drift during this time was ca. 2 mV per
day and detection limits remained as low as 5 × 10−6 M. These data
confirm that the cross-linking between the conductive Meth-PEDOT-SC
and the PU-ISM gives SCISEs with a relatively long lifetime compared
with other developed SCISES which have lifetime between 2 and 6
month [48–50]. These results also show that the adhesion between the
ISM and the screen-printed substrate is excellent.
4. Conclusions
We report here a low-cost and simple fabrication method of dis-
posable Ca-ion sensors based on ISE with solid contact formed by Meth-
PEDOT and Meth-PEDOT-MWCNT/cMWCNT conducting polymer
films. Electrochemical characterization was performed using cyclic
voltammetry, electrochemical impedance measurements and potentio-
metric measurements. The results of CV and EIS experiments reveal that
the Meth-PEDOT-cMWCNT has a higher redox capacitance than Meth-
PEDOT-MWCNT and Meth-PEDOT. However, after coating the solid
contacts with a photo-cured polyurethane-based Ca2+-selective mem-
brane (ISM) the redox capacitance of the solid contact was essentially
limited by the high bulk resistance of the ISM. Consequently, no sig-
nificant improvements in potentiometric response of the Ca2+-SCISEs
were obtained by adding MWCNT into the conducting polymer films in
terms of sensitivity, selectivity, LOD, lifetime as well as robustness to
the influence of light, O2 and CO2. No aqueous layer formation was
detected at the interfaces between the solid-contacts and the mem-
branes of the Ca2+-SCISEs by using the water layer test. Based on the
results of this work it can be concluded that Meth-PEDOT is a promising
solid-contact material to develop low-cost and easy to prepare ISEs.
Acknowledgments
This work was supported by the People Programme (Marie Curie
Actions) of the 7th Framework Programme of the European Union
(FP7/2007–2013) under REA grant agreement no. 600388
(TECNIOspring programme) and from the Agency for Business
Competitiveness of the Government of Catalonia (ACCIÓ). N.A. also
acknowledges the Government of Russian Federation (Grant 074-U01)
Appendix A. Supplementary material
Supplementary data associated with this article can be found in the
online version at http://dx.doi.org/10.1016/j.talanta.2018.04.056.
284
Talanta 186 (2018) 279–285
References
[1] J.D. Stenger-Smith, Intrinsically electrically conducting polymers, Synth., Charact.,
their Appl. Progress. Polym. Sci. 23 (1998) 57–79.
[2] A.O. Patil, A.J. Heeger, F. Wudl, Optical-properties of conducting polymers, Chem.
Rev. 88 (1988) 183–200.
[3] B.P. Jelle, G. Hagen, Electrochemical multilayer deposition of polyaniline and
Prussian blue and their application in solid state electrochromic windows, J. Appl.
Electrochem. 28 (1998) 1061–1065.
[4] Q. Pei, G. Zuccarello, M. Ahlskog, O. Inganäs, Electrochromic and highly stable poly
(3,4-ethylenedioxythiophene) switches between opaque blue-black and transparent
sky blue, Polymer 35 (1994) 1347–1351.
[5] J.L. Reddinger, G.A. Sotzing, J.R. Reynolds, Muticoloured electrochromic polymers
derived from easily oxidized bis[2-(3,4-ethylenedioxy)thienyl]carbazoles, Chem.
Commun. (1996) 1777–1778.
[6] Y. Qiu, L. Duan, L.D. Wang, Flexible organic light-emitting diodes with poly-3,4-
ethylenedioxythiophene as transparent anode, Chin. Sci. Bull. 47 (2002)
1979–1982.
[7] L.S. Yu, S.A. Chen, High efficiency polymer light emitting diodes based on poly
(phenylene vinylene)s with balanced electron and hole fluxes, Synth. Met 132
(2002) 81–86.
[8] K.L. Paik, N.S. Baek, H.K. Kim, J.H. Lee, Y. Lee, White light-emitting diodes from
novel silicon-based copolymers containing both electron-transport oxadiazole and
hole-transport carbazole moieties in the main chain, Macromolecules 35 (2002)
6782–6791.
[9] W.H. Meyer, Polymer electrolytes for lithium-ion batteries, Adv. Mater. 10 (1998)
439.
[10] K. Gurunathan, A.V. Murugan, R. Marimuthu, U.P. Mulik, D.P. Amalnerkar,
Electrochemically synthesised conducting polymeric materials for applications to-
wards technology in electronics, optoelectronics and energy storage devices, Mater.
Chem. Phys. 61 (1999) 173–191.
[11] B.E. Conway, Transition from supercapacitor to battery behavior in electrochemical
energy-storage, J. Electrochem Soc. 138 (1991) 1539–1548.
[12] A. Kros, R.J.M. Nolte, N. Sommerdijk, Conducting polymers with confined dimen-
sions: track-etch membranes for amperometric biosensor applications, Adv. Mater.
14 (2002) 1779–1782.
[13] L.H. Dall'Antonia, M.E. Vidotti, S.I.C. de Torresi, R.M. Torresi, A new sensor for
ammonia determination based on polypyrrole films doped with dodecylbenzene-
sulfonate (DBSA) ions, Electroanalysis 14 (2002) 1577–1586.
[14] A. Chaubey, M. Gerard, R. Singhal, V.S. Singh, B.D. Malhotra, Immobilization of
lactate dehydrogenase on electrochemically prepared polypyrrole-poly-
vinylsulphonate composite films for application to lactate biosensors, Electrochim.
Acta 46 (2000) 723–729.
[15] T. Lindfors, S. Ervela, A. Ivaska, Polyaniline as pH-sensitive component in plasti-
cized PVC membranes, J. Electroanal. Chem. 560 (2003) 69–78.
[16] R.E. Gyurcsanyi, A.S. Nyback, K. Toth, G. Nagy, A. Ivaska, Novel polypyrrole based
all-solid-state potassium-selective microelectrodes, Analyst 123 (1998) 1339–1344.
[17] A. Michalska, A. Hulanicki, A. Lewenstam, All-solid-state potentiometric sensors for
potassium and sodium based on poly(pyrrole) solid contact, Microchem. J. 57
(1997) 59–64.
[18] A. Cadogan, Z. Gao, A. Lewenstam, A. Ivaska, D. Diamond, All-solid-state sodium-
selective electrode based on a calixarene ionophore in a poly(vinyl chloride)
membrane with a polypyrrole solid contact, Anal. Chem. 64 (1992) 2496–2501.
[19] A. Peramo, M.G. Urbanchek, S.A. Spanninga, L.K. Povlich, P. Cederna, D.C. Martin,
In situ polymerization of a conductive polymer in acellular muscle tissue constructs,
Tissue Eng. Part A 14 (2008) 423–432.
[20] S.M. Richardson-Burns, J.L. Hendricks, D.C. Martin, Electrochemical polymeriza-
tion of conducting polymers in living neural tissue, J. Neural Eng. 4 (2007) L6–L13.
[21] A. Elschner, S. Kirchmeyer, W. Lövenich, U. Merker, K. Reuter, PEDOT, Princ. Appl.
Intrinsically Conduct. Polym. (2011).
[22] A.S. Alshammari, M. Shkunov, S.R.P. Silva, Correlation between wetting properties
and electrical performance of solution processed PEDOT:PSS/CNT nano-composite
thin films, Colloid Polym. Sci. 292 (2014) 661–668.
[23] T. Lindfors, R.-M. Latonen, Improved charging/discharging behavior of electro-
polymerized nanostructured composite films of polyaniline and electrochemically
reduced graphene oxide, Carbon 69 (2014) 122–131.
[24] F. Sundfors, R. Bereczki, J. Bobacka, K. Toth, A. Ivaska, R.E. Gyurcsanyi,
Microcavity based solid-contact ion-selective microelectrodes, Electroanalysis 18
(2006) 1372–1378.
[25] R.E. Gyurcsanyi, N. Rangisetty, S. Clifton, B.D. Pendley, E. Lindner, Microfabricated
ISEs: critical comparison of inherently conducting polymer and hydrogel based
inner contacts, Talanta 63 (2004) 89–99.
[26] J.X. Zhang, Y.X. Guo, S.J. Li, H. Xu, A solid-contact pH-selective electrode based on
tridodecylamine as hydrogen neutral ionophore, Meas. Sci. Technol. 27 (2016) 9.
[27] P. Sjoberg, A. Määttänen, U. Vanamo, M. Novell, P. Ihalainen, F.J. Andrade, et al.,
Paper-based potentiometric ion sensors constructed on ink-jet printed gold elec-
trodes, Sens Actuator B-Chem. 224 (2016) 325–332.
[28] M. Apostu, N. Bibire, G. Tantaru, M. Vieriu, A.D. Panainte, L. Agoroaei, Ion-selec-
tive membrane electrodes for the determination of heavy metals Construction
characterization and applications, Rev. Chim. 66 (2015) 657–659.
[29] N. He, T. Lindfors, Determination of water uptake of polymeric ion-selective
membranes with the coulometric karl fischer and FT-IR-attenuated total reflection
techniques, Anal. Chem. 85 (2013) 1006–1012.
[30] N. Abramova, A. Bratov, Application of photocured polymer ion selective mem-
branes for solid-state chemical sensors, Chemosensors 3 (2015) 190–199.
C. Ocaña et al.
[31] A. Beltran, J. Artigas, C. Jimenez, R. Mas, J. Bartroli, J. Alonso, Development of
durable nitrate-selective membranes for all-solid state ISE and ISFET sensors based
on photocurable compositions, Electroanalysis 14 (2002) 213–220.
[32] N. Abramova, J. Moral-Vico, J. Soley, C. Ocana, A. Bratov, Solid contact ion sensor
with conducting polymer layer copolymerized with the ion-selective membrane for
determination of calcium in blood serum, Anal. Chim. Acta 943 (2016) 50–57.
[33] D.N. Reinhoudt, J.F.J. Engbersen, Z. Brzozka, H.H. van der Vlekkert, G.W.N. Honig,
H.A.J. Holterman, et al., Development of durable K+-selective chemically modified
field effect transistors with functionalized polysiloxane membranes, Anal. Chem. 66
(1994) 3618–3623.
[34] N. Abramova, A. Bratov, Photocurable pH-sensitive membrane for ion-selective
field effect transistors, Talanta 81 (2010) 208–212.
[35] G.S. Cha, D. Liu, M.E. Meyerhoff, H.C. Cantor, A.R. Midgley, H.D. Goldberg, et al.,
Electrochemical performance, Biocompatibility, and adhesion of new polymer
matrices for solid-state ion sensors, Anal. Chem. 63 (1991) 1666–1672.
[36] A. Rzewuska, M. Wojciechowski, E. Bulska, E.A.H. Hall, K. Maksymiuk,
A. Michalska, Composite Polyacrylate−Poly(3,4- ethylenedioxythiophene) mem-
branes for improved all-solid-state ion-selective sensors, Anal. Chem. 80 (2008)
321–327.
[37] A. Bratov, N. Abramova, J. Mu¤oz, C. Dominguez, S. Alegret, J. Bartroli,
Photocurable polymer matrices for potassium-sensitive ion- selective electrode
membranes, AnalChem 67 (1995) 3589–3595.
[38] N. Abramova, A. Bratov, A. Ipatov, S. Levichev, A. Aris, E.M. Rodriguez, New flow-
through analytical system based on ion-selective field effect transistors with opti-
mised calcium selective photocurable membrane for bovine serum analysis, Talanta
113 (2013) 31–35.
[39] J. Bobacka, Potential stability of all-solid-state ion-selective electrodes using con-
ducting polymers as ion-to-electron transducers, Anal. Chem. 71 (1999)
4932–4937.
[40] G. Guilbault, R.A. Durst, M.S. Frant, H. Freiser, E.H. Hansen, T.S. Light, et al., Pure
285
Talanta 186 (2018) 279–285
Appl. Chem. 46 (1976) 127–132.
[41] J. Bobacka, Conducting polymer-based solid-state ion-selective electrodes,
Electroanalysis 18 (2006) 7–18.
[42] J. Bobacka, A. Lewenstam, A. Ivaska, Electrochemical impedance spectroscopy of
oxidized poly(3,4-ethylenedioxythiophene) film electrodes in aqueous solutions, J.
Electroanal. Chem. 489 (2000) 17–27.
[43] Z. Mousavi, J. Bobacka, A. Lewenstam, A. Ivaska, Poly(3,4-ethylenediox-
ythiophene) (PEDOT) doped with carbon nanotubes as ion-to-electron transducer in
polymer membrane-based potassium ion-selective electrodes, J. Electroanal. Chem.
633 (2009) 246–252.
[44] J.R. McDonald, Impedance Spectroscopy, John Wiley, New York, 1987.
[45] Z. Mousavi, A. Teter, A. Lewenstam, M. Maj-Zurawska, A. Ivaska, J. Bobacka,
Comparison of multi-walled carbon nanotubes and Poly(3-octylthiophene) as ion-
to-electron transducers in all-solid-state potassium ion-selective electrodes,
Electroanalysis 23 (2011) 1352–1358.
[46] U. Schefer, D. Ammann, E. Pretsch, U. Oesch, W. Simon, Neutral carrier based
calcium(2+)-selective electrode with detection limit in the sub-nanomolar range,
Anal. Chem. 58 (1986) 2282–2285.
[47] M. Fibbioli, W.E. Morf, M. Badertscher, N.F. de Rooij, E. Pretsch, Potential drifts of
solid-contacted ion-selective electrodes due to zero-current ion fluxes through the
sensor membrane, Electroanalysis 12 (2000) 1286–1292.
[48] M. Wagner, G. Lisak, A. Ivaska, J. Bobacka, Durable PEDOT:PSS films obtained from
modified water-based inks for electrochemical sensors, Sens. Actuators B: Chem.
181 (2013) 694–701.
[49] H. Xu, Y. Wang, Z.Y. Luo, Y.W. Pan, A miniature all-solid-state calcium electrode
applied to in situ seawater measurement, Meas. Sci. Technol. 24 (2013) 6.
[50] A. Matusevich, M. Pietrzak, E. Malinowska, Miniaturized F–selective all-solid-state
potentiometric sensors with conductive polymer as an intermediate layer, Sens
Actuator B-Chem. 168 (2012) 62–73.
Academic Press is an imprint of Elsevier
225 Wyman Street, Waltham, MA 02451, USA
525 B Street, Suite 1800, San Diego, CA 92101–4495, USA
125 London Wall, London, EC2Y 5AS, UK
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK

First edition 2015

© 2015 Elsevier Inc. All rights reserved.

No part of this publication may be reproduced or transmitted in any form or by any means,
electronic or mechanical, including photocopying, recording, or any information storage and
retrieval system, without permission in writing from the publisher. Details on how to seek
permission, further information about the Publisher’s permissions policies and our
arrangements with organizations such as the Copyright Clearance Center and the Copyright
Licensing Agency, can be found at our website: www.elsevier.com/permissions.

This book and the individual contributions contained in it are protected under copyright by
the Publisher (other than as may be noted herein).

Notices
Knowledge and best practice in this field are constantly changing. As new research and
experience broaden our understanding, changes in research methods, professional practices,
or medical treatment may become necessary.

Practitioners and researchers must always rely on their own experience and knowledge in
evaluating and using any information, methods, compounds, or experiments described
herein. In using such information or methods they should be mindful of their own safety and
the safety of others, including parties for whom they have a professional responsibility.

To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors,
assume any liability for any injury and/or damage to persons or property as a matter of
products liability, negligence or otherwise, or from any use or operation of any methods,
products, instructions, or ideas contained in the material herein.

ISBN: 978-0-12-803300-5
ISSN: 1871-5125 (Series)

For information on all Academic Press publications

visit our website at http://store.elsevier.com/
 Polyhedron 87 (2015) 194–201

Contents lists available at ScienceDirect

Polyhedron
journal homepage: www.elsevier.com/locate/poly

Iron(III) complexes with pentadentate Schiff-base ligands: Influence

of crystal packing change and pseudohalido coligand variations on spin
crossover
Christoph Krüger a, Peter Augustín b, L’ubor Dlháň b, Ján Pavlik b,⇑, Ján Moncol’ b, Ivan Nemec c,
Roman Boča b,d, Franz Renz a
a
Institute of Inorganic Chemistry, Leibniz University Hannover, D-30167 Hannover, Germany
b
Institute of Inorganic Chemistry, FCHPT, Slovak University of Technology, SK-812 37 Bratislava, Slovakia
c
Regional Centre of Advanced Technologies and Materials, and Department of Inorganic Chemistry, Faculty of Science, Palacký University, CZ-779 00 Olomouc, Czech Republic
d
Department of Chemistry, FPV, University of SS Cyril and Methodius, SK-917 01 Trnava, Slovakia

a r t i c l e i n f o a b s t r a c t

Article history: Two novel iron(III) complexes involving pentadentate Schiff-base ligands, [Fe(LBr)(L1)], show temperature
Received 22 August 2014 induced incomplete spin crossover of a gradual nature. While for the L1 = NCSe coligand the transition
Accepted 6 November 2014 temperature (326 K or 317 K) lies above room temperature, the N 3 coligand causes its drop down to
Available online 21 November 2014
143 K or 140 K (two values were obtained by different models). This shift is associated with a significant
decrease of the enthalpy of the transition from about DH = 6.1 kJ mol1 to ca. DH = 1.7 kJ mol1, while the
Keywords: entropy of the transition is about DS = 19 J K1 mol1 and DS = 12 J K1 mol1, respectively. Two analo-
Spin crossover
gous complexes with Cl and NCS coligands remain high-spin over the whole temperature range.
Iron(III) complexes
Pseudohalides
Ó 2014 Elsevier Ltd. All rights reserved.
Schiff-base ligand
Ising-like model

1. Introduction the monodentate ligands form a distorted octahedral coordination

Spin crossover (SCO) is one of the most fascinating phenomena
occurring in coordination compounds of transition-metal ions and
has gained more and more interest during the last decades. Already
in the early 1930s Cambi and his co-workers first reported on
dithiocarbamato complexes of iron(III) with unusual magnetic
properties [1–3]. Since that time a lot of iron(III) SCO systems have
been published and the achievements in this research were
reviewed and categorized by Van Koningsbruggen et al. in 2003
or recently, by Bruker et al. or Murray in 2013 [4]. Nowadays, these
types of molecular switches seem to be very attractive for potential
applications such as switching devices or sensors [5].
There is a plethora of mononuclear iron(III) SCO compounds of
the general composition [Fe(X-L5)L1] involving pentadentate N3O2-
donating Schiff-base ligands H2X-L5, originating in the condensa-
tion of derivatives of salicylaldehydes and aliphatic triamines; X
denotes the substitution of the aromatic rings. The sixth coordina-
tion site of iron can be occupied by various monodentate organic or
inorganic ligands [6–11]. The pentadentate H2X-L5 together with
⇑ Corresponding author.
E-mail address: jan.pavlik@stuba.sk (J. Pavlik).
environment, which is a usual condition for the occurrence of SCO
in Fe(III) compounds, but it should be noted, that SCO has been
observed also for pentacoordinate Fe(III) compounds [12]. Further-
more, the monodentate ligand can be replaced by a bridging
ligand/complex to build polynuclear complexes which often exhi-
bit SCO [6,10,13–17,7,18–26].
Previous studies have established that ligand design is crucial
for the possibility of fine tuning the SCO properties, especially, by
increasing the ligand rigidity, modification of the intermolecular
interactions between the SCO molecules [9,11] or just by the sub-
stitution at the variable sixth coordination site [27–30].
The six-coordinate iron(III) compounds exhibit SCO from the
low-spin (LS) to high-spin (HS) state (S = 1/2 ? S = 5/2), while
SCO from the intermediate state (S = 3/2 ? S = 5/2) is observed
typically for the five-coordinate complexes [12,31]. The spin tran-
sition can be induced physically (such as temperature, light or
pressure change) as well as chemically (such as a solvate, ligand
or pH change) [32–40].
Herein, we report on four novel mononuclear iron(III) com-
plexes [Fe(LBr)(L1)] involving a pentadentate Schiff-base ligand
H2LBr and Cl, N 3 , NCS

and NCSe as L1 coligands. The penta-
Br
dentate ligand H2L was prepared by the condensation
of 5-bromosalicylaldehyde with an asymmetric triamine –

http://dx.doi.org/10.1016/j.poly.2014.11.014
0277-5387/Ó 2014 Elsevier Ltd. All rights reserved.
 C. Krüger et al. / Polyhedron 87 (2015) 194–201
Scheme 1. Synthetic route for the preparation of the complexes [Fe(LBr)Cl]0.5H2O (1), [Fe(LBr)N3]CH3OH (2), [Fe(LBr)NCS] (3) and [Fe(LBr)NCSe] (4) using the Schiff-base
ligand H2LBr.
1,6-diamino-4-azahexane (Scheme 1). The complexes [Fe(LBr)(-
Cl)]0.5H2O (1), [Fe(LBr)(N3)]CH3OH (2), [Fe(LBr)(NCS)] (3), and
[Fe(LBr)(NCSe)] (4) have been characterized by single-crystal X-
ray diffraction, elemental analysis, IR spectroscopy and magnetic
measurements. In our previous work we reported on two isostruc-
tural compounds with a general formula [Fe(LCl)(NCS/Se)], which
exhibit rather gradual SCO near room temperature with TC = 280
(NCS) and 293 K (NCSe). In this study, the effect of the peripheral
substitution of Cl by the Br atom is examined in terms of the crystal
packing and intermolecular interactions, and magnetic properties
of the compounds lie in the center of interest.
2. Results and discussion
2.1. Structural data
Cell parameters of the investigated compounds are listed in
Table 1; the remaining crystallographic data are deposited in Sup-
plementary material and they can be retrieved from the Cambridge
Crystallographic Data Centre.
The molecular structures of the compounds 1–4 are very similar
to the structures of the [Fe(LX)(L1)] compounds (where L1 = Cl, N
3,
NCS, NCSe; X marks the substitutions by functional groups on
the aromatic rings of the primary ligand H2L = N,N0 -bis(1-
hydroxy-2-benzyliden)-1,6-diamino-4-azahexane) reported previ-
ously [9–11,14]. The pentadentate Schiff base ligand (LBr)2 chelates
the central iron(III) atom by three nitrogen atoms (one amine
(Nam), two imine (Nim)) and two oxygen atoms.
The nitrogen atoms are arranged in a fac manner while the oxy-
gen atoms are in the cis positions. The sixth coordination site of the
central atom is occupied by the monodentate chlorido or pseud-
ohalido ligand with the donor atom (Cl or Nterm) in the trans posi-
tion to the oxygen atom from the more flexible ‘‘propyl’’ part of the
pentadentate ligand. The metal–ligand bond lengths of the com-
pounds under study are summarized in Table 2.
The compound [Fe(LBr)(Cl)]0.5H2O (1) is isostructural with the
previously reported [Fe(LCl)(Cl)]0.5H2O and [Fe(LCl)(CN)]0.5H2O
complexes [9]. The asymmetric unit contains two complex mole-
cules and one molecule of the crystal water. The chromophore
bond lengths and angular distortion parameter, R,1 correspond to
the HS state of the complex molecule (R(1) = 62.3 and 52.3°) and
they are similar to those observed for other iron(III) chlorido com-
plexes with pentadentate Schiff base ligands with the longest Fe–
Cl bond (2.3458(6) and 2.3569(6) Å). The Fe–Nam bond lengths are
a bit shorter (2.178(2) and 2.198(3) Å) and the Fe–Nim and Fe–O
are the shortest bond lengths (Fig. 1 and Table 2). The crystal water
is involved in the hydrogen bonding pattern including four [Fe(LBr)
1
The angular distortion from the ideal octahedron R is calculated as sum of the
absolute values of the deviations of the twelve cis chromophore angles from the ideal
value 90°.
195
(Cl)] molecules non-covalently bridged (O–H  O hydrogen bonding)
by two water molecules and N–H  Br (Fig. 1). The water molecules
provide hydrogen bonding of significant strength with
d(O  O) = 2.657(4) and 2.868(6) Å, while N–H  Br contacts are
much weaker: d(N  Br) = 3.499(2) Å.
The bond lengths in the chromophore of [Fe(LBr)(N3)] CH3OH (2)
correspond to the HS state of the metal center (Fig. 2): the Fe–Nam
bonds are the longest with d(Fe–Nam) = 2.183(4) Å, while the Fe–
Nim bonds are significantly shorter (2.109(4), 2.117(4) Å) and the
Fe–Nterm bond is the shortest one (2.048(3) Å). The Fe–O bond
lengths do not differ significantly from those observed for the LS
structure of 1 and they amount to 1.916(3) and 1.968(3) Å.
Remarkably, the dihedral angle (previously correlated to the occur-
rence of SCO in Fe(III) compounds with hexadentate ligands [41])
between the least-square planes of the aromatic rings is very nar-
row (a = 35.0°) in comparison with the other compounds reported
in this article (Table 2). Also the angular distortion parameter
(R(2) = 72.3°) is very large for the HS structure of this type of com-
pounds (typical values for [Fe(X-L5)L1] HS compounds are close to
60°).
The structure of 2 is formed of the [Fe(LBr)(N3)] molecules aligned
in supramolecular chains and held by the N–H  N hydrogen bonds
between the nitrogen atoms from the amine and azido groups of the
adjacent complex molecules with d(N  N) = 3.036(6) Å. The crystal
structure contains methanol molecules coupled with the [Fe(LBr)
(N3)] molecules by a relatively short hydrogen bond between the
methanol oxygen atom and phenolato group of the complex mole-
cule (Fig. 2): d(O  O) = 2.781(6) Å.
The compound [Fe(LBr)(NCS)] (3) crystallizes in the monoclinic
space group Pn. The chromophore bond lengths and angular distor-
tion parameter (R(3) = 56.3°) refer to the HS state (Table 2).
The crystal structure is formed of the chains of the [Fe(LBr)(NCS)]
molecules interconnected by weak NH  S hydrogen bonds between
the nitrogen atoms from the amine groups and sulfur atoms of the
thiocyanato ligand from the adjacent complex molecules
(d(N  S) = 3.323(7) Å). The thiocyanato sulfur atoms are further
involved in other non-covalent interactions – the supramolecular
[Fe(LBr)(NCS)]n chains are mutually interlocked by weak Br  S
non-covalent contacts (Fig. 3) with d(BrS) = 3.481(3) and
3.564(4) Å (sum of the van der Waals radii of the Br  S pair is
3.650 Å).
The bond lengths (Fig. 4) within the chromophore of [Fe(LBr)
(NCSe)] (4) at T = 193 K are close to the typical LS values with the
longest bond between the iron atom and the amine nitrogen atom:
d(Fe–Nam) = 2.013(7) Å. The Fe–Nim, Fe–Nterm and Fe–O bond
lengths are significantly shorter: d(Fe–Nim) = 1.937(7), 1.967(7);
d(Fe–Nterm) = 1.944(6); d(Fe–O) = 1.896(6), 1.872(5) Å. The N–Fe–
N and N–Fe–O bond angles of the trans bonds in the chromophore
are close to linearity and their values range from 174.9° to 177.6°.
Also, the cis chromophore angles adopt values close to the
right angle (ideal octahedron): from 84.4° to 94.4°. This causes a
196 C. Krüger et al. / Polyhedron 87 (2015) 194–201

Table 1
Crystallographic data for compounds 1–4.

1 2 3 4
Br Br Br
[Fe(L )Cl]0.5H2O [Fe(L )(N3)]CH3OH [Fe(L )(NCS)] [Fe(LBr)(NCSe)]
M/g mol1 581.49 611.09 595.11 642.01
T/K 90(1) 293(1) 100(1) 193(1)
Crystal system triclinic monoclinic monoclinic monoclinic
Space group P1 P21/c Pn P21/c
a (Å) 12.1643(7) 12.6077(3) 12.1647(6) 8.31180(10)
b (Å) 12.9750(7) 15.7666(4) 7.9347(4) 12.99640(10)
c (Å) 14.8456(8) 12.7549(4) 12.4122(6) 20.23030(10)
a (°) 77.065(3) 90 90 90
b (°) 72.484(2) 112.623(3) 115.540(2) 97.503(3)
c (°) 66.470(2) 90 90 90
V (Å)3 2034.22(19) 2340.34(11) 1081.00(9) 2166.64(3)
Z 2 4 2 4
Rint/R1/wR2(I > 2r(I)) 0.0272/0.0259/0.0613 0.0281/0.0500/0.1102 0.0717/0.0472/0.1257 0.0629/0.0782/0.2352
CCD deposit number 998903 998904 998905 998906

Table 2
Selected structural parameters for the title compounds.

Compound Behavior Fe–Nam (Å) Fe–X (Å)a Fe–Nim (Å)b Fe–O (Å)c Rd(°) ae(°) Ref.
Br f f
[Fe(L )(Cl)]0.5H2O (1) HS 2.188 2.352 2.088 1.959 62.3/52.3 63.1/54.2 This work
[Fe(LBr)(NCS)] (3) HS 2.207 2.111 2.099 1.927 56.3 67.7 This work
[Fe(LBr)(N3)]CH3OH (2) SCO 2.183 2.048 2.113 1.942 72.3 35.0 This work
[Fe(LBr)(NCSe)] (4) SCO 2.013 1.944 1.952 1.884 27.8 85.9 This work
[Fe(LCl)(Cl)]0.25H2O HS 2.199f 2.361f 2.095 1.958 66.9/53.1 62.2/53.4 [9]
[Fe(LCl)(NCO)] HS 2.205f 1.996f 2.093 1.954 55.4/57.3 69.0/61.6 [9]
[Fe(LCl)(NCS)] SCO 2.000 1.944 1.932 1.879 27.1 84.5 [9]
[Fe(LCl)(NCSe)] SCO 2.001 1.952 1.932 1.881 26.9 85.5 [9]
[Fe(LCl)(CN)]H2O LS 2.012f 1.959f 1.934 1.903 19.3/28.8 65.9/56.0 [9]
[Fe(L)(CN)]CH3OH LS 2.000 1.962 1.922 1.880 24.6 76.0 [10]
[Fe(3tBu5Me-L)(NCS)] HS 2.191 2.100 2.096 1.916 56.0 75.6 [11]
[Fe(3MeO-L)(NCO)]CH3OH HS 2.218f 2.081f 2.078 1.946 54.0f 73.3/82.8 [11]
[Fe(3MeO-L)(N3)] HS 2.211 2.085 2.085 1.936 57.4 58.1 [11]
[Fe(3MeO-L)(CN)]CH3OH LS 2.028f 1.971f 1.935 1.904 26.0f 69.1/71.5 [11]
[Fe(3EtO-L)(CN)]H2O LS 2.010 1.975 1.944 1.904 20.8 66.4 [11]
a
The bond length between the iron atom and the donor atom of the monodentate ligand.
b
The average value calculated from the iron-imino nitrogen bond lengths.
c
The average value calculated from the Fe–O bond lengths.
d
The angular distortion from the ideal octahedron R is calculated as sum of the absolute values of the deviations of the twelve cis chromophore angles from the ideal value
90°.
e
The dihedral angle (a) between the least-square planes of the aromatic rings.
f
The average value calculated from two bond lengths/parameters. Ligand abbreviations: H2L = N,N0 -bis(1-hydroxy-2-benzyliden)-1,6-diamino-4-azahexane, H2LBr =
N,N0 -bis(1-hydroxy-4-bromo-2-benzyliden)-1,6-diamino-4-azahexane, H2LCl = N,N0 -bis(1-hydroxy-4-chloro-2-benzyliden)-1,6-diamino-4-azahexane, H23MeO-L = N,N0 -
bis(1-hydroxy-6-methoxy-2-benzyliden)-1,6-diamino-4-azahexane, H23EtO-L = N,N0 -bis(1-hydroxy-6-ethoxy-2-benzyliden)-1,6-diamino-4-azahexane, H23tBu5Me-L = N,N0 -
bis(1-hydroxy-4-methyl-6-tert-buthyl-2-benzyliden)-1,6-diamino-4-azahexane.

Fig. 1. Molecular structure of 1 (ORTEP drawing with thermal ellipsoids at the 50% probability level). Selected bond lengths (Å): Fe1–O2 = 1.8986(19), Fe1–O1 = 2.0246(18),
Fe1–N3 = 2.087(2), Fe1–N1 = 2.091(3), Fe1–N2 = 2.178(2), Fe2–O4 = 1.9097(19), Fe2–O3 = 2.0012(18), Fe2–N4 2.081(2), Fe2–N6 = 2.094(2), Fe2–N5 = 2.198(3), Fe2–
Cl2 = 2.3569(6). Right – the crystal packing in 1. Hydrogen atoms are omitted for clarity except for those involved in the hydrogen bonds (dashed lines).

low-value of the spin dependent parameter of the angular distor- consists of supramolecular dimers linked by rather weak non-
tion from the ideal octahedron: R(4) = 27.8°. The compound 4 is covalent N–H  Br interactions with d(N  Br) = 3.506(6) Å.
isostructural with [Fe(LCl)(NCR)] complexes [9] (R = S and Se) and This alignment is supported by p–p stacking interactions of the
 C. Krüger et al. / Polyhedron 87 (2015) 194–201 197

Fig. 2. Molecular structure of 2 (ORTEP drawing with thermal ellipsoids at the 50% probability level. Selected bond lengths (Å): Fe1–N1 = 2.117(4), Fe1–N3 = 2.109(4), Fe1–
N2 = 2.183(4), Fe1–N4 = 2.048(3), Fe1–O1 = 1.968(3), Fe1–O2 = 1.916(3). Right – the crystal packing in 2. Hydrogen atoms are omitted for clarity except for those involved in
the hydrogen bonds (dashed lines).

Fig. 3. The supramolecular chains [Fe(LBr)(NCS)]n in 3. Right: Interconnection of the chain substructures by Br  S non-covalent contacts (dashed lines) in 3. Hydrogen atoms
are omitted for clarity (except for those involved in the hydrogen bonds). Selected bond lengths (Å): Fe1–N1 = 2.099(5), Fe1–N2 = 2.207(9), Fe1–N3 = 2.099(5), Fe1–
N4 = 2.111(7), Fe1–O1 = 1.905(7), Fe1–O2 = 1.948(5).

Fig. 4. Molecular structure of 4 (ORTEP drawing with thermal ellipsoids at the 50% probability level). Right: The crystal packing in 4. Hydrogen atoms are omitted for clarity
except for those involved in the hydrogen bonds (dashed lines). Selected bond lengths (Å) in 4: Fe1–O1 = 1.896(6), Fe1–O2 = 1.872(5), Fe1–N1 = 1.937(7), Fe1–N4 = 1.944(6),
Fe1–N3 = 1.967(7), Fe1–N2 = 2.013(7).

neighboring complex molecules with the centroid–centroid dis-
tance of 3.700(5) Å and by short C  Br contacts with
d(C  Br) = 3.474(8) Å (sum of the van der Waals radii of the C  Br
pair is 3.650 Å, Fig. 4).
2.2. Magnetic data
The magnetic functions of complexes 1–4 are shown in Fig. 5.
The complex 1, [Fe(LBr)Cl]2 H2O, is high-spin in the whole temper-
ature interval (S = 5/2), however, the structural data show that
there are two units bridged by a rather strong hydrogen bond (vide
supra) therefore the value of the magnetic moment is higher than
in the case of the monomeric complex 2. The model of exchange
coupled dimers combined with the zero field splitting (abbr. ZFS)
was employed in the data analysis (for more details, see ESI)
[42]. The temperature dependence of the magnetic susceptibility
and the field dependence of the magnetization (ESI, Fig. S1) are fit-
ted simultaneously thus allowing more reliable determination of
the sign and value of independent parameters, namely the isotro-
pic exchange coupling constant J, the axial zero-field splitting
parameter D, the isotropic gyromagnetic factor g, the Weiss con-
stant H and the temperature-independent susceptibility aTIM
(not to be confused with the dihedral angle where no lower index
is used). Their optimum values were found as: J/hc = 0.145 cm1,
g = 1.984, aTIM = 4.76  109 m3 mol1, D/hc = 1.02 cm1,
H = 0.130 K; the corresponding discrepancy factors are
198 C. Krüger et al. / Polyhedron 87 (2015) 194–201

10 7
B = 0.1 T
B = 0.1 T
6
8

5
25 25
6
4

χmol/(10-6 m3 mol-1)
20
μeff/μB

μeff/μB
χmol/(10-6 m3 mol-1)
20

15 3 15
4
10 10
2
5 5
2
1 0
0
0 10 20 30 40
0 10 20 30 40
0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300

T/K T/K
6 6 6 6

B = 0.1 T B = 0.1 T
5 5 5 5

χmolT / cm3.mol-1.K
χmolT / cm3.mol-1.K

4 4 4 4
μeff/μB
μeff/μB

3 3 3 3

2 2 2 2

1 1 1 1

0 0 0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300 350 400

T/K T/K

Fig. 5. Effective magnetic moment for 1 (up left), 2 (bottom left), 3 (up right) and 4 (bottom right). The inset for 1 and 3 shows the low temperature fit of the magnetic
susceptibility. The empty circles for 2 and 4 show the vmT product which is directionally proportional to the high spin fraction. Circles – experimental data, solid lines – fitted.

R(v) = 0.039 and R(M) = 0.042 for the susceptibility and magnetiza-
tion, respectively. Here; J is too small in order to cause a maximum
at the susceptibility curve above 1.9 K, nevertheless, it is responsi-
6
ble for a rapid drop of the effective magnetic moment at the lowest
temperatures (Fig. 5).
Also the complex 3, [Fe(LBr)(NCS)], is high-spin in the whole
5
temperature range and the same approach can be used as for 1
except that no magnetic exchange is present. The data fitting gave
the following set of optimum parameters: g = 1.926, aTIM = 4
2.98  109 m3 mol1, D/hc = 0.53 cm1, H = 0.130 K;
R(v) = 0.011, R(M) = 0.015. In this case the drop of the effective
μeff/μB

magnetic moment at low temperature originates in the zero-field 3

splitting of the HS state and the molecular field effect (Fig. 5).
For the complexes 2, [Fe(LBr)(N3)] and 4, [Fe(LBr)(NCSe)] gradual
SCO between SL = 1/2 and SH = 5/2 is visible. For 4, the effective 2
magnetic moment stays almost constant between 5 and 200 K; [Fe(LCl)(NCSe)]
[Fe(LCl)(NCS)]
below 5 K its decrease is registered. Above T > 200 K gradual SCO Br
[Fe(L )(NCSe)]
proceeds and it is not complete until T = 390 K when leff = 5.61 lB; 1
[Fe(LCl)(CN)]
SH = 5/2 assumes leff = 5.92 lB. For both complexes the magnetiza-
[Fe(LBr)(N3)]
tion per formula unit at M1 = Mmol/NAlB = 1.5 is much higher than
M1 = 1.0 expected for the SL = 1/2 and gL = 2 system (ESI, Fig. S1). 0
0 50 100 150 200 250 300 350 400
This fact was modeled by two approaches: (i) the model A assumes
a rather large magnetogyric factor gL > 3 owing to the presence of T/K
low-lying excited states belonging to the high-spin state; (ii) the
Fig. 6. Comparison of the spin crossover behavior for related complexes (including
model B assumes a presence of a ‘‘frozen’’ high spin state, i.e. by
compounds from [9]).
 C. Krüger et al. / Polyhedron 87 (2015) 194–201 199

Table 3
Review of spin crossover parameters in Fe(III) complexes with pentadentate Schiff-base ligands.

Compound Model SL–SH Tc Magnetic parameters Ref.

(K)
[Fe(LBr)(N3)]CH3OH A 1/2–5/2 143 gL = 2.95, HL = 0.09 K, aTIM,L = 36.8  109 m3 mol1, gH = 2.02, aTIM,H = 0.0  109 m3 mol1, E/kB = 197 K, This
(2) C/kB = 76.0 K, mL = 336 cm1, mH = 292 cm1; R(v) = 0.028 and R(M) = 0.048; DH = 1.64 kJ mol1, work
DS = 11.5 J K1 mol1
B 1/2–5/2 140 gL = 2.12, HL = 0.00 K, aTIM,L = 30.0  109 m3 mol1, gH = 2.00, HH = 0.46 K, aTIM,H = 0.0  109 m3 mol1, This
L = 302 cm1, m
E/kB = 203 K, C/kB = 84.0 K, m H = 263 cm1, xfrz = 0.11; R(v) = 0.021 and R(M) = 0.031; work
DH = 1.69 kJ mol1, DS = 12.1 J K1 mol1
[Fe(LCl)(NCS)] A 1/2–5/2 280 gL = 2.38, HL = 0.91 K, aTIM,L = +5.0  109 m3 mol1, gH = 2.0, E/kB = 694 K, C/kB = 180 K, m
L = 306 cm1, [9]
mH = 250 cm1; DH = 5.77 kJ mol1, DS = 20.6 J K1 mol1
[Fe(LCl)(NCSe)] A 1/2–5/2 293 gL = 2.20, HL = 0.20 K, aTIM,L = +2.6  109 m3 mol1, gH = 2.0, E/kB = 787 K, C/kB = 197 K, m
L = 320 cm1, [9]
mH = 254 cm1; DH = 6.54 kJ mol1, DS = 22.3 J K1 mol1
[Fe(LBr)(NCSe)] (4) A 1/2–5/2 326 gL = 3.11, HL = 0.48 K, gH = 2.14, aTIM,H = 0.41  109 m3 mol1, E/kB = 725 K, C/kB = 215 K, mL = 275 cm1, This
mH = 239 cm1; R(v) = 0.020 and R(M) = 0.036; DH = 6.03 kJ mol1, DS = 18.5 J K1 mol1 work
B 1/2–5/2 317 gL = 2.20, HL = 0.00 K, aTIM,L = 0.00  109 m3 mol1, gH = 2.05, HH = 0.90 K, aTIM,H = 0.00  109 m3 mol1, This
E/kB = 743 K, C/kB = 226 K, m L = 224 cm1, m
H = 195 cm1, xfrz = 0.11; R(v) = 0.039 and R(M) = 0.040; work
DH = 6.18 kJ mol1, DS = 19.5 J K1 mol1
[Fe(LCl)(CN)] 1/2 – gL = 2.36, HL = 0.060 K, aTIM = +5.7  109 m3 mol1 [9]

cooling down, not all molecules switch their state (ESI, Fig. S2). The
obtained value of the transition enthalpy is DH = 6.03 kJ mol1 and
6.18 kJ mol1, the transition entropy DS = 18.5 J K1 mol1 and
19.5 J K1 mol1 and the characteristic temperature Tc = 326 K
and 317 K for the model A and B, respectively. In both models C/
kB < Tc so that a thermal hysteresis is absent. The optimum param-
eters resulting from the fitting procedure are collected in Table 3
and the fitted data are displayed as solid lines in Fig. 5 for the case
of model B.
Substitution of the NCSe ligand for N3 led to significant lower-
ing of the transition temperature, but the complex 2 still shows
SCO (Fig. 5). The same fitting procedure was applied with the
results DH = 1.64 kJ mol1 and 1.69 kJ mol1, DS = 11.5 J K1 mol1
and 12.1 J K1 mol1 and Tc = 143 K and 140 K, for models A and B,
respectively. The cooperativeness is again below the limiting value
for the hysteresis.
Thermal evolution of the effective magnetic moment for [Fe(LCl/Br)
X] complexes is compared in Fig. 6. The quantitative characteris-
tics, as they result from the fitting procedure are listed in Table 3.
It can be concluded that SCO centered near the room temperature
is associated with a considerable transition enthalpy of the order of
DH  5.8–6.5 kJ mol1. The transition entropy varies as DS = 18.5–
22.3 J K1 mol1, which means that a considerable contribution
from molecular vibrations assists.
3. Discussion
A detailed analysis of the crystal and molecular structures
reveals that the occurrence of SCO in the series of the [Fe(LCl/Br)
(L1)] compounds is influenced by a number of factors. First, the
expected isostructurality of the isothiocyanato and isoselenocy-
anato compounds is not preserved within these two series: only
the compounds [Fe(LCl)(NCS)], [Fe(LCl)(NCSe)] and [Fe(LBr)(NCSe)]
(4) are isostructural (P21/c, Fig. 6) and they exhibit SCO. The
[Fe(LBr)(NCS)] compound (3) possesses different crystal packing
(Pn, Fig. 4) and it stays HS over the whole temperature range
(Fig. 5).
When inspecting their molecular structures at the chromophore
level (compounds 3 and 4), it is apparent that they do not differ in
bond lengths or bond angles significantly from the typical LS/HS
forms of the SCO compounds, or from pure HS compounds
reported for this type of complexes previously [9,12]. Thus there
is no evidence that there exists any structural difference in the
[Fe(LBr)(NCS)] complex molecule, which would lead to lowering
of the overall ligand-field strength resulting in stabilization of
the HS state. Only a very slight difference between the ligand field
strengths of the LCl and LBr ligands can be deduced from the tran-
sition temperatures of the SCO compounds: Tc([Fe(LCl)
(NCSe)]) = 293 K and Tc([Fe(LBr)(NCSe)])  320 K. The alteration of
the L1 ligand from NCSe to NCS does not lead to a sizable
decrease of the increment to the crystal-field strength (D = gM fL);
fL(NCS)=1.02 while fL(NCSe) = 1.05 (fL(Cl) = 0.78).
When inspecting the topology of the complex molecule, a sig-
nificant difference can be found: the a parameter describing the
dihedral angle between the planes of the aromatic rings differs sig-
nificantly when comparing three SCO compounds (a = 84.5–85.9°)
with compound 3 (a = 67.7°, Table 2). This difference opens a ques-
tion if the molecular shape can play a dominant role in hindering
the SCO behavior in 3 as it was observed previously for a group
of [FeIII(R-L6)]+ complexes [41] (H2R-L6 = variously substituted H2
sal2trien). In these compounds the purely LS or HS complexes pre-
fer different molecular shapes (a is significantly larger for HS com-
pounds) [44]. However, this might not be the case for 3 and it can
be documented by the example of compound 2, which has a very
low value of the a parameter (a = 35.0°, Table 2) but still exhibits
SCO (Fig. 6). Furthermore, there are several examples of purely
HS [Fe(La)(NCS)] compounds[12] (H2La stands for 2-{3-[2-
(1-hydroxynaphthalen-2-yl)methylenaminoethylaminopropylimi-
no}methyl)naphthalene-1-ol) with very low a values ranging from
17° to 25°; this is in a clear contradiction with the magneto-struc-
tural correlation found for the [FeIII(R-L6)]+ complexes.
Remarkably, the crystal structure of 2 consists of the supramo-
lecular chains of the [Fe(LBr)(N3)] molecules held by N–H  N
hydrogen bonds. It must be noted that such structural motif (i.e.
assembly to chain supramolecular structure by NH  N/O hydrogen
bonding) was observed for similar SCO compounds [Fe(Ln)(NCO)]
and [Fe(Ln)(N3)] previously [11] (H2Ln stands for N,N0 -bis(2-
hydroxy-naphthylidene)-1,6-diamino-4-azahexane); however, in
the case of 2 an additional methanol molecule is found in the crys-
tal structure (Fig. 2). On the other hand, the previously reported
compound with similar composition, [Fe(3MeO-L)(N3)], does not
possess the above mentioned structural motif and it does not exhi-
bit SCO.[11] This brings discussion back to the importance of the
isostructurality, or in other words, of preservation of the SCO
molecular packing motifs and it underlines the importance of
molecular organization in the solid state for the occurrence and
character of SCO [45].
200 C. Krüger et al. / Polyhedron 87 (2015) 194–201
4. Conclusion
In this work we reported on a series of Fe(III) complexes of the
general formula [Fe(LBr)(L1)] involving the pentadentate Schiff-
base ligand (LBr)2 and (pseudo)halide coligands L1 = Cl(1), N 3 Black single crystals were obtained by slow evaporation at room
(2), NCS (3) and NCSe (4). These compounds were characterized
by single-crystal X-ray diffraction and by magnetic measurements,
which revealed that two of them, [Fe(LBr)(N3)]CH3OH and
[Fe(LBr)(NCSe)], exhibit weakly cooperative incomplete SCO with
Tc = 143 K or 140 K (2) and 326 K or 317 K (4) (modelled by two
different approaches), while compounds 1 and 3 are purely HS.
In the case of the compound 1, the non-occurrence of SCO was
explained on the basis of low ligand-field strength arising from the
weak p-donor chlorido ligand. Furthermore, this compound exhib-
its weak intermolecular magnetic exchange coupling (J/
hc = 0.145 cm1) mediated by a water molecule involved in hydro-
gen bonds interconnecting phenolato oxygen atoms from the
neighboring [Fe(LBr)(Cl)] molecules. The stabilization of the HS
state in the [Fe(LBr)(NCS)] compound was explained on the basis
of the different crystal packing motif in comparison with the
herein reported [Fe(LBr)(NCSe)] compound or previously reported
[Fe(LCl)(NCS)] and [Fe(LCl)(NCSe)], which all exhibit SCO and they
are mutually isostructural.
5. Experimental
5.1. Synthesis
5.1.1. [Fe(LBr)Cl]0.5H2O, 1
The Schiff-condensation of 5-bromo-2-hydroxybenzaldehyde
(2.0 mmol) with an asymmetric triamine, 1,6-diamino-4-azahex-
ane, H2N(CH2)3NH(CH2)2NH2 in the ratio of 2:1 in 30 cm3 of metha-
nol resulted in a pentadentate Schiff-base ligand (bright yellow
solution) according to Scheme 1. The reaction mixture was stirred
at 55 °C for 30 min. This mixture was combined with a methanol
solution of FeCl36H2O (2.0 mmol in 10 cm3) at 55 °C for 15 min.
Then, the solution of triethylamine (3.8 mmol in 5 cm3 of methanol)
was added drop wise. After 40 min of stirring at the boiling temper-
ature, the black powder was separated by filtration on a fritted fun-
nel, washed with cold methanol and diethyl ether (yield 36%). The
filtrate was left to evaporate spontaneously. A few days later, black
crystals were separated. Yield: 37%. C19H20Br2FeClN3O2.5
(M = 581.49). Anal. Calc.: C, 39.25; H, 3.47; N, 7.23. Found: C,
39.06; H, 3.57; N, 7.32%.
5.1.2. [Fe(LBr)(N3)] CH3OH, 2
A methanol solution of 1 (8.734.102 mmol, 20 cm3) was com-
bined with a methanol solution of NaN3 (9.61102 mmol in
10 cm3) and the mixture was stirred for 60 min at the boiling tem-
perature. The prepared solution was filtered and the filtrate was
left for evaporation. The next day black crystals were isolated.
Yield: 35%. C20H23Br2FeN6O3 (M = 611.09). Anal. Calc.: C, 39.31; H,
3.79; N, 13.75. Found: C, 39.66; H, 3.56; N, 14.07%.
5.1.3. [Fe(LBr)(NCS)], 3
Compound 1 (0.2 g, 0.4 mmol) was dissolved in 80 cm3 metha-
nol and combined with 0.041 g (0.42 mmol) KNCS in 20 cm3 of
methanol. Then, the solution was stirred for 60 min and filtered.
Black single crystals were obtained by slow evaporation at room
temperature. C20H19FeN4O2Br2S1 (M = 595.11). Anal. Calc.: C,
40.37; H, 3.22; N, 9.41. Found: C, 39.98; H, 3.21; N, 9.30; IR
(ATR): m~(N–H) = 3166 cm1, m~(C–H) = 2913 cm1 (m),
1 1
m~(NCS) = 2052 cm , m~(C@N) = 1616 cm .
5.1.4. [Fe(LBr)(NCSe)], 4
Compound 1 (0.2 g, 0.4 mmol) was dissolved in 80 cm3 of meth-
anol and combined with 0.061 g (0.42 mmol) KNCSe in 20 cm3 of
methanol. Then, the solution was stirred for 60 min and filtered.
temperature. C20H19FeN4O2Br2Se1 (M = 642.01). Anal. Calc.: C,
37.42; H, 2.98; N, 8.73. Found: C, 37.32; H, 2.93; N, 8.43, IR
(ATR): m~(N–H) = 3219 cm1, m~(C–H) = 2921 cm1 (m),
1 1
m~(NCSe) = 2061 cm , m~(C@N) = 1616 cm .
5.2. X-ray structure analysis
Single crystal X-ray diffraction data were collected on diffrac-
tometers Bruker X8 APEX-II or Oxford Diffraction Xcalibur with
the Sapphire CCD detector and fine-focused sealed tube (Mo Ka
radiation, k = 0.71073 Å or Cu Ka radiation, k = 1.54184 Å). All
structures were solved by direct methods using SHELXS-97 [46]
incorporated into the WINGX program package [47]. For each struc-
ture its space group was checked by the ADSYMM procedure of the
PLATON software [48]. All structures were refined using full-matrix
least-squares on Fo2-Fc2 with SHELXL-2014 or SHELXL-97 with aniso-
tropic displacement parameters for all non-hydrogen atoms [46].
The hydrogen atoms were placed into the calculated positions
and they were included into the riding model approximation, with
Uiso = 1.2/1.5 Ueq. The structure were drawn by MERCURY software
[49].
5.3. Magnetic data
The magnetic data were taken with the SQUID apparatus
(MPMS-XL7, Quantum Design) using the RSO mode of detection
with ca 20 mg of the sample encapsulated in a gelatin-made sam-
ple holder. The magnetic susceptibility taken at B = 0.1 T was cor-
rected for the underlying diamagnetism and converted to the
effective magnetic moment. The magnetization was measured at
two temperatures: T = 2.0 and T = 4.6 K.
Acknowledgments
Grant Agencies (Slovakia: VEGA-1/0233/12, APVV-0014-11,
APVV-0132-11; Germany: DAAD SK/DE; Czech Republic:
PrF_2014_09) are acknowledged for the financial support.
Appendix A. Supplementary data
CCDC 998903–998906 contain the supplementary crystallo-
graphic data for complexes 1–4, respectively. These data can be
obtained free of charge via http://www.ccdc.cam.ac.uk/conts/
retrieving.html, or from the Cambridge Crystallographic Data
Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: (+44) 1223-
336-033; or e-mail: deposit@ccdc.cam.ac.uk. Supplementary data
associated with this article can be found, in the online version, at
http://dx.doi.org/10.1016/j.poly.2014.11.014.
References
[1] L. Cambi, L. Szegö, Ber. Dtsch. Chem. Ges. (A and B Series) 64 (1931) 2591.
[2] L. Cambi, L. Szegö, Ber. Dtsch. Chem. Ges. (A and B Series) 66 (1933) 656.
[3] L. Cambi, L. Malatesta, Ber. Dtsch. Chem. Ges. (A and B Series) 70 (1937) 2067.
[4] (a) P.J. van Koningsbruggen, Y. Maeda, H. Oshio, Top. Curr. Chem. 233 (2004)
259;
(b) M. Nihei, T. Shiga, Y. Maeda, H. Oshio, Coord. Chem. Rev. 251 (2007) 2606;
(c) J. Olguin, S. Brooker, Spin-Crossover in Discrete Polynuclear Complexes, in:
Malcolm A. Halcrow (Ed.), Spin-Crossover Materials: Properties and
Applications, Wiley, 2013;
(d) K.S. Murray, The Development of Spin-Crossover Research, in: Malcolm A.
Halcrow (Ed.), Spin-Crossover Materials: Properties and Applications, Wiley,
2013;
 C. Krüger et al. / Polyhedron 87 (2015) 194–201 201

(e) P.N. Martinho, C. Rajnák, M. Ruben, Nanoparticles, Thin Films and Surface [23] F. Renz, P. Kerep, D. Hill, R. Müller-Seipel, M. Klein, Hyperfine Interact. 168
Patterns From Spin Crossover Materials and Electrical Spin State Control, in: (2006) 1051.
Malcolm A. Halcrow (Ed.), Spin-Crossover Materials: Properties and [24] M. Gembický, R. Boča, F. Renz, Inorg. Chem. Commun. 3 (2000) 662.
Applications, Wiley, 2013. [25] F. Renz, V. Martinez, M. Klein, Hyperfine Interact. 184 (2008) 251.
[5] (a) P. Gütlich, A.B. Gaspar, Y. Garcia Beilstein, J. Org. Chem. 9 (2013) 342; [26] F. Renz, V. Martinez, M. Klein, M. Schott, T. Hoffmann, M. Blumers, I. Fleischer,
(b) A. Bousseksou, G. Molnár, L. Salmon, W. Nicolazzi, Chem. Soc. Rev. 40 G. Klingelhöfer, R. Boča, M. Menzel, Hyperfine Interact. 184 (2008) 259.
(2011) 3313; [27] S. Hayami, Z.Z. Gu, H. Yoshiki, A. Fujishima, O. Sato, J. Am. Chem. Soc. 123
(c) G. Molnár, L. Salmon, W. Nicolazzi, F. Terki, A. Bousseksou, J. Mater. Chem. (2001) 11644.
C 2 (2014) 1360. [28] J.K. Tang, J.S. Costa, S. Smulders, G. Molnár, A. Bousseksou, S.J. Teat, Y.G. Li, G.A.
[6] R. Boča, Y. Fukuda, M. Gembický, R. Herchel, R. Jaroščiak, W. Linert, F. Renz, J. van Albada, P. Gamez, J. Reedijk, Inorg. Chem. 48 (2009) 2128.
Yuzurihara, Chem. Phys. Lett. 325 (2000) 411. [29] S. Dorbes, L. Valade, J.A. Real, C. Faulmann, Chem. Commun. (2005) 69.
[7] F. Renz, P. Kerep, D. Hill, M. Klein, Hyperfine Interact. 168 (2007) 981. [30] M.S. Haddad, W.D. Federer, M.W. Lynch, D.N. Hendrickson, Inorg. Chem. 20
[8] R. Boča, I. Nemec, I. Šalitroš, J. Pavlik, R. Herchel, F. Renz, Pure Appl. Chem. 81 (1981) 131.
(2009) 1357. [31] P. Gütlich, Eur. J. Inorg. Chem. (2013) 581.
[9] C. Krüger, P. Augustín, I. Nemec, Z. Trávníček, H. Oshio, R. Boča, F. Renz, Eur. J. [32] H.A. Goodwin, Coord. Chem. Rev. 18 (1976) 293.
Inorg. Chem. 5–6 (2013) 902. [33] P. Gütlich, Struct. Bonding 44 (1981) 83.
[10] I. Šalitroš, R. Boča, Ľ. Dlháň, M. Gembický, J. Kožíšek, J. Linares, J. Moncoľ, I. [34] E. König, Struct. Bonding 76 (1991) 51.
Nemec, L. Perašínová, F. Renz, I. Svoboda, H. Fuess, Eur. J. Inorg. Chem. (2009) [35] P. Gütlich, A. Hauser, H. Spiering, Angew. Chem. 106 (1994) 2109.
3141. [36] P. Gütlich, Y. Garcia, H.A. Goodwin, Chem. Soc. Rev. 29 (2000) 419.
[11] I. Nemec, R. Herchel, R. Boča, Z. Trávníček, I. Svoboda, H. Fuess, W. Linert, [37] P. Gütlich, Y. Garcia, P.J. van Koningsbruggen, F. Renz, in: F. Palacio, J.
Dalton Trans. 40 (2011) 10090. Schweizer, E. Ressouche (Eds.), Introduction to Physical Techniques in
[12] S. Mossin, B.L. Tran, D. Adhikari, M. Pink, F.W. Heinemann, J. Sutter, R.K. Molecular Magnetism, Part 1 – Structural and Magnetic Methods, University
Szilagyi, K. Meyer, D.J. Mindiola, J. Am. Chem. Soc. 134 (2012) 13651. Press, Zaragoza, 2000.
[13] I. Nemec, R. Boča, R. Herchel, Z. Trávníček, M. Gembický, W. Linert, Monatsh. [38] A. Hauser, J. Jeftić, H. Romstedt, R. Hinek, H. Spering, Coord. Chem. Rev. 190–
Chem. 140 (2009) 815. 192 (1999) 471.
[14] I. Šalitroš, R. Boča, R. Herchel, J. Moncoľ, I. Nemec, M. Ruben, F. Renz, Inorg. [39] Spin crossover in transition metal compounds I, II, IIIP. Gütlich, H.A. Goodwin
Chem. 51 (2012) 12755. (Eds.), Topics in Current Chemistry, 233, Springer, Berlin, 2004.
[15] R. Herchel, R. Boča, M. Gembický, J. Kožíšek, F. Renz, Inorg. Chem. 43 (2004) [40] P. Gütlich, P. van Koningsbruggen, F. Renz, Struct. Bonding 107 (2004) 27.
4103. [41] M.A. Halcrow, Chem. Soc. Rev. 40 (2011) 4119.
[16] F. Renz, P. Kerep, Polyhedron 24 (2005) 2849. [42] R. Boča, Handbook of Magnetochemical Formulae, Elsevier, Amsterdam, 2012.
[17] F. Renz, P. Kerep, Hyperfine Interact. 156/157 (2004) 371. [44] I. Nemec, R. Herchel, I. Šalitroš, Z. Trávníček, J. Moncoľ, H. Fuess, M. Ruben, W.
[18] F. Renz, M. Klein, S. Jung, P. Homenya, R. Saadat, D. Nariaki, Unimagazin Linert, CrystEngComm 14 (2012) 7015.
(Leibniz Universität Hannover) 1–2 (2011) 24. [45] J. Tao, R.-J. Wei, R.-B. Huang, L.-S. Zheng, Chem. Soc. Rev. 41 (2012) 703.
[19] F. Renz, D. Hill, M. Klein, J. Hefner, Polyhedron 26 (2007) 2325. [46] G.M. Sheldrick, Acta Crystallogr., Sect. A 64 (2008) 112.
[20] R. Boča, I. Šalitroš, J. Kožíšek, J. Linares, J. Moncoľ, F. Renz, Dalton Trans. 39 [47] L.J. Farrugia, J. Appl. Crystallogr. 45 (2012) 849.
(2010) 2198. [48] (a) A.L. Speck, Acta Crystallogr., Sect. D. Biol. Crystallogr. 65 (2009) 148;
[21] F. Renz, S. Jung, M. Klein, M. Menzel, A.F. Thünemann, Polyhedron 2009 (1818) (b) Y. Le Page, MISSYM1.1 – a flexible new release, J. Appl. Crystallogr. 21
28. (1988) 983.
[22] F. Renz, C. Zaba, L. Rossberg, S. Jung, M. Klein, G. Klingelhöfer, A. Wünsche, S. [49] C.F. Macrae, P.R. Edgington, P. McCabe, E. Pidcock, G.P. Shields, R. Taylor, M.
Reinhardt, M. Menzel, Polyhedron 28 (2009) 2036. Towler, J. van de Streek, J. Appl. Crystallogr. 39 (2006) 453.
 Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955

Contents lists available at ScienceDirect

Spectrochimica Acta Part A: Molecular and

Biomolecular Spectroscopy
journal homepage: www.elsevier.com/locate/saa

Continuous Wavelet Transform, a powerful alternative to Derivative

Spectrophotometry in analysis of binary and ternary mixtures: A
comparative study
Eman S. Elzanfaly, Said A. Hassan ⇑, Maissa Y. Salem, Badr A. El-Zeany
Department of Analytical Chemistry, Faculty of Pharmacy, Cairo University, Kasr El-Aini Street, 11562 Cairo, Egypt

h i g h l i g h t s g r a p h i c a l a b s t r a c t

 Numerical Differentiation and

Wavelet Transform as approaches for
derivative calculation.
 Wavelet Transform is a powerful
alternative to traditional derivative
algorithms.
 Simple, selective and precise
spectrophotometric methods for
binary and ternary mixtures.
st
 The 1 spectrophotometric method
for analysis of Amlodipine, Aliskiren
and Hydrochlorothiazide.
 Methods validated as per ICH
guidelines, parameters found to be
within the limits.

a r t i c l e i n f o a b s t r a c t

Article history: A comparative study was established between two signal processing techniques showing the theoretical
Received 21 April 2015 algorithm for each method and making a comparison between them to indicate the advantages and
Received in revised form 20 June 2015 limitations. The methods under study are Numerical Differentiation (ND) and Continuous Wavelet
Accepted 25 June 2015
Transform (CWT). These methods were studied as spectrophotometric resolution tools for simultaneous
Available online 30 June 2015
analysis of binary and ternary mixtures. To present the comparison, the two methods were applied for
the resolution of Bisoprolol (BIS) and Hydrochlorothiazide (HCT) in their binary mixture and for the
Keywords:
analysis of Amlodipine (AML), Aliskiren (ALI) and Hydrochlorothiazide (HCT) as an example for ternary
Derivative Spectrophotometry
Continuous Wavelet Transform
mixtures. By comparing the results in laboratory prepared mixtures, it was proven that CWT technique
Amlodipine is more efficient and advantageous in analysis of mixtures with severe overlapped spectra than ND.
Aliskiren The CWT was applied for quantitative determination of the drugs in their pharmaceutical formulations
Bisoprolol and validated according to the ICH guidelines where accuracy, precision, repeatability and robustness
Hydrochlorothiazide were found to be within the acceptable limit.
Numerical Differentiation Ó 2015 Elsevier B.V. All rights reserved.

1. Introduction

UV–VIS absorption spectroscopy is a well-established technique

for rapid and accurate determination of analytes in a mixture form
without prior separation, if the interferences between the spectra
⇑ Corresponding author.
can be eliminated. One of the techniques used for elimination of
E-mail addresses: said.hassan@pharma.cu.edu.eg, saidmonem_84@yahoo.com
(S.A. Hassan).
interference in spectroscopy is signal processing. The most

http://dx.doi.org/10.1016/j.saa.2015.06.100
1386-1425/Ó 2015 Elsevier B.V. All rights reserved.
946 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955
commonly used signal processing technique is Derivative
Spectrophotometry (DS).
Rutherford [1] was the first to apply derivative methods suc-
cessfully to experimental data, he suggested that the sensitivity
of mass spectroscopy would be increased on applying the first
derivative technique. Derivative methods were first applied to
UV–VIS and IR spectroscopy in 1953 by Hammond and Price [2],
followed by Morrison [3] and in 1955 Giese and French [4]
described the theory of the derivative technique. In 1959, Martin
[5] introduced the theoretical basis for second and higher orders
of derivative, and in 1968, Morrey [6] generated first to fourth
order derivatives with a digital computer. In 1964, Savitzky and
Golay [7] introduced least-squares procedures for the smoothing
and differentiation of fluctuating data.
Derivative calculation is a powerful technique used in analytical
chemistry to resolve spectra, sharpen peaks, remove background
interference, and carry out quantitative analysis. The most impor-
tant field is its use in resolution enhancement technique to facili-
tate the detection and location of poorly resolved components in
complicated experimental signals [8]. It was applied successfully
to chromatography [9], electrochemistry [10–12], flow-injection
analysis [13] and in all fields of spectroscopy [14–16]. In the region
of UV–VIS spectroscopy, DS played an important role in resolution
of overlapped spectra [17,18]. The zero-crossing derivative spectra
method was the most common procedure for the simultaneous
determination of single components as well as binary and ternary
mixtures from their overlapping spectra [19–21]. DS was also
applied to ratio spectra for analysis of binary and ternary mixtures
[22–24].
In the spectral studies, the application of the usual DS to the
original absorption spectra possesses several disadvantages such
as: peak intensity diminishes with higher order derivation, it
required the additional smooth function as well as the scaling
factor processes. These parameters may deform the obtained
derivative spectra from the original ones. As a result, the usual
derivative method produces several errors in the quantitative anal-
ysis [25]. These drawbacks suggested using other signal processing
techniques as powerful alternatives for derivative calculation such
as Wavelet Transform [26,27].
A Wavelet Transform (WT) involves the decomposition of a sig-
nal function or vector e.g., a spectrum of a chemical species into
simpler, fixed building blocks at different scales and positions
[28]. Different algorithms were proposed to achieve this purpose.
In 1977, Coifman and Weiss [29] developed the automatic decom-
position method, based on work of Calderón and Zygmund [30],
which had a very strong influence on the development of wavelet
theory. In 1984, Grossmann and Morlet [31] proposed the
Continuous Wavelet Transform (CWT). In 1989, Mallat [32] intro-
duced the multi-resolution signal decomposition MRSD algorithm.
Daubechies [26,33,34] adopted this approach to construct families
of compactly supported wavelets and coupled it to quadrature mir-
ror filtering.
WT has been applied successfully for signal processing in differ-
ent fields of analytical chemistry. It was applied to de-noising of
data [35], compression of signals [36], analysis of electrochemical
data [37], flow-injection analysis [38], multivariate calibration
[39] and quantitative analysis of near infrared spectra [40]. In the
field of UV–VIS spectrophotometry, Continuous Wavelet
Transform (CWT) combined either with a zero-crossing technique
[41] or ratio spectra [13] was used for simultaneous determination
of chemical species in binary and ternary mixtures [18,42–45].
ConcorÒ Plus tablets contain both Bisoprolol hemifumarate
(BIS) and Hydrochlorothiazide (HCT) for the treatment of high
blood pressure [46]. There are reported methods for the determi-
nation of BIS or HCT in different dosage forms [47–50] and in their
binary mixtures [45,51–55].
AmturnideÒ is a combination of Amlodipine besylate (AML),
Aliskiren hemifumarate (ALI) and Hydrochlorothiazide (HCT) indi-
cated for the treatment of hypertension to reduce the risk of
strokes and myocardial infarctions [56]. Many reported methods
were reported for the determination of AML, ALI or HCT in different
dosage forms [57–62]. HPLC and capillary electrophoresis methods
were reported for simultaneous determination of this ternary mix-
ture [63,64]. No spectrophotometric methods were developed for
the determination of this mixture.
In this work, ND and CWT were applied for determination of
Bisoprolol (BIS) and Hydrochlorothiazide (HCT) in their binary
mixture and for determination of Amlodipine (AML), Aliskiren
(ALI) and Hydrochlorothiazide (HCT) in their ternary mixture as
examples for complex spectra. A comparative study between the
two methods was performed to compare their resolution power
in analysis of overlapped spectra. Then the CWT was used for the
analysis of the cited mixtures in pharmaceutical formulations,
namely ConcorÒ Plus tablets and AmturnideÒ tablets.
2. Theoretical background
2.1. Numerical Differentiation (ND) [65]
For a discrete spectrum xi (i = 1, . . ., n), suppose that wi
(i = 1, . . ., n) are its sampling wavelengths, the direct-difference
method can be expressed as follows:
X iþ1  X i
yi ¼
W iþ1  W i
Here yi (i = 1, . . ., n  1) represent the series of derivatives of spec-
trum xi. If the sampling wavelength is equal, this equation can be
rewritten as yi = xi+1  xi (i = 1, . . ., n  1).
2.2. Wavelet Transform (WT): [27]
Wavelet Transform (WT) involves decomposition of a signal
function or vector (spectrum) into simpler, fixed building blocks
at different scales and positions. In WT treatment, all basis func-
tions Wa;b ðxÞ can be derived from a mother wavelet WðxÞ through
the following dilation and translation processes:
 
1 xb
Wa;b ðxÞ ¼ a2 W a; b 2 R and a–0
a
where a and b are the scale and position parameters, respectively,
expressed in real number R.
The basic idea of WT is to represent any arbitrary function f ðxÞ
as a superposition of wavelets. Decomposition of x with respect to
the wavelet function series {Wj;k ðxÞ} is described by the following
formula:
X
þ1 X
þ1
f ðxÞ ¼
ðjÞ
C k Wj;k
j¼1 k¼1
Therefore, the signal is represented by a set of coefficients {C j;k } in
the wavelet domain.
The approximate first derivative of X is assigned to be equal to
the difference between two scale coefficients at the first resolution
level as
xð1Þ  C 1;D2m  C 1;D2m~ ~
m–m
where, D2m and D2m~ represent any two Daubechies wavelet
~ being any positive integer.
functions, with m and m
 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955
3. Experimental
3.1. Apparatus
SHIMADZU dual beam UV–visible spectrophotometer
(Kyoto/Japan), model 1650 UV-PC, with matched 1-cm quartz cells
was used. The bundled software, UV-Probe personal spectroscopy
software version 2.21 (SHIMADZU) is used. The spectral bandwidth
is 2.0 nm and wavelength scanning speed is 2800 nm/min with
0.1 nm interval.
3.2. Chemicals and reagents
3.2.1. Pure samples
- BIS, HCT and AML; kindly supplied by Al-Hekma pharmaceuti-
cal Company, Cairo, Egypt, their purity was certified to be
99.7 ± 0.8%, 100.6 ± 0.6% and 99.9 ± 0.5%, respectively.
- ALI; kindly supplied by Novartis Pharmaceuticals Corporation,
USA, its purity was certified to be 100.6 ± 0.3%.
3.2.2. Market samples
- ConcorÒ 5 Plus labeled to contain 5(BIS)/12.5(HCT) mg, batch
number 1030039 and ConcorÒ 10 Plus labeled to contain
10(BIS)/25(HCT) mg, batch number 0795049, manufactured
by Pfizer Ltd., Cairo, Egypt.
- AmturnideÒ tablet dosage forms; labeled to contain 10
(AML)/300 (ALI)/25 (HCT) mg batch number F0005, manufac-
tured by Novartis Pharmaceuticals Corporation, USA.
3.2.3. Methanol
Spectroscopy grade (El-NASR Pharmaceutical Chemicals Co.,
Abu-Zaabal, Cairo, Egypt).
3.3. Procedures
3.3.1. Standard solutions
- BIS, HCT, AML and ALI standard stock solutions; 1000 lg /mL in
methanol.
- BIS, HCT and AML standard working solutions; 50 lg/mL in
methanol.
- ALI standard working solution; 500 lg/mL in methanol.
- Preparation of laboratory prepared mixtures; aliquots of stan-
dard working solutions were transferred into two series of
10-mL measuring flasks, completed to volume with methanol
to prepare binary mixture of BIS and HCT and ternary mixture
of AML, ALI and HCT containing different ratios of the drugs.
3.3.2. Spectral characteristics of the drugs
The zero-order (D0) absorption spectra of 4 lg/mL BIS and
10 lg/mL HCT were recorded against methanol as a blank over
the range of 200–400 nm.
The zero-order (D0) absorption spectra of 4, 120 and 10 lg/mL
solutions of AML, ALI and HCT, respectively, were recorded against
methanol as a blank over the range of 200–400 nm.
3.3.3. Construction of calibration curves
Aliquots equivalent to 20–400 lg BIS, 5–250 lg HCT, 10–350 lg
AML and 150–2500 lg ALI were accurately transferred from their
respective standard working solutions into separate series of
10-mL volumetric flasks then completed to volume with methanol.
The spectra of the prepared standard solutions were scanned from
200 to 400 nm and stored in the computer.
947
3.3.3.1. Numerical Differentiation method (ND).
Binary mixture
For determination of BIS, the 3rd and 4th derivative spectra were
obtained by ND method with Dk = 8 nm and scaling factor = 1000.
The peak amplitude was measured at 283.6 and 287.5 nm,
respectively.
For determination of HCT, the first derivative spectra were
obtained by ND method with Dk = 4 nm and scaling factor = 100.
The peak amplitude was measured at 272.3 nm.
The calibration curves were constructed relating the peak
amplitudes at the specified wavelengths to the corresponding
concentrations of BIS and HCT.
Ternary mixture
For determination of AML, the first derivative spectra were
obtained by ND method with Dk = 4 nm and scaling factor = 10.
The peak amplitude was measured at 386.8 nm.
For determination of ALI, the fourth derivative spectra were
obtained by ND method with Dk = 8 nm and scaling factor = 1000.
The peak amplitude was measured at 287.5 nm.
For determination of HCT, the second derivative spectra were
obtained by ND method with Dk = 8 nm and scaling factor = 100.
The peak amplitude was measured at 332.1 nm.
The calibration curves were constructed relating the peak
amplitudes at the specified wavelengths to the corresponding
concentrations of AML, ALI and HCT.
3.3.3.2. Continuous Wavelet Transform method (CWT).
Binary mixture
The first derivative spectra were obtained by CWT method with
Morlet family (morl) and scaling = 200. The calibration curves were
constructed relating the peak amplitudes at 278.3 and 283.6 nm to
the corresponding concentrations of BIS and HCT, respectively.
Ternary mixture
For determination of AML, the first derivative spectra were
obtained by CWT method with Gaussian-1 family (Gaus-1) and
scaling = 200. The peak amplitude was measured at 387.2 nm.
For determination of ALI, the first derivative spectra were
obtained by CWT method with Gaussian-8 family (Gaus-8) and
scaling = 200. The peak amplitude was measured at 315.6 nm.
For determination of HCT, the first derivative spectra were
obtained by CWT method with Coiflet-3 family (Coif-3) and
scaling = 200. The peak amplitude was measured at 289.1 nm.
The calibration curves were constructed relating the peak
amplitudes at the specified wavelengths to the corresponding
concentrations of AML, ALI and HCT.
3.3.4. Application of the proposed methods for the determination of
laboratory-prepared mixtures
The spectra of laboratory prepared mixtures were scanned from
200 to 400 nm and stored in the computer. The same procedure
under construction of calibration curves was applied and the con-
centrations of the drugs were calculated from the corresponding
regression equations.
3.3.5. Application of the proposed methods for the determination of the
drugs in dosage forms
Ten tablets of both ConcorÒ 5 Plus, ConcorÒ 10 Plus were accu-
rately weighed, finely powdered and an amount of the powder
948 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955
equivalent to 10 mg BIS was weighed. Five tablets of AmturnideÒ
were accurately weighed, finely powdered and amount of the pow-
der equivalent to 300 mg ALI was weighed. The powder was dis-
solved in methanol by shaking in ultrasonic bath for about
30 min, then they were filtered and transferred quantitatively into
separate 100-ml volumetric flasks and the volume was completed
to the mark with methanol. Ten ml aliquots were transferred into
separate 50-ml volumetric flasks and the volume was completed to
the mark with methanol. Then necessary dilutions were made to
reach concentrations in the linearity range. The spectra of these
solutions were scanned from 200 to 400 nm and stored in the com-
puter. Spectra obtained were analyzed by the proposed methods.
4. Results and discussion
Signal processing techniques use different algorithms in trans-
forming the spectra into new graphs where the spectra of different
compounds are resolved. Derivative Spectrophotometry (DS) is the
most commonly used processing technique for resolution of over-
lapped spectra. Lately, Continuous Wavelet Transform algorithm
(CWT) started to be used as an alternative to Numerical
Differentiation (ND) for calculation of DS.
The aim of this work was to establish a comparative study
between the two signal processing techniques. For proper presen-
tation of the comparison, two different mixtures were chosen to
show the resolution power of the methods. The first mixture was
a combination of BIS and HCT as an example of binary mixtures.
The spectra of BIS and HCT (Fig. 1) show severe overlap as the peak
of HCT (kmax 226 nm) completely coincide with that of BIS (kmax
225 nm) showing the same features. This overlap, together with
the high absorptivity of HCT which is also found in higher level
in ConcorÒ Plus tablets, prevents direct determination of BIS. In
previous work, the transformed ratio spectra had to be combined
with zero-crossing technique for analysis of BIS in this mixture
[45]. The second mixture composed of AML, ALI and HCT was used
as an example of ternary mixtures. The absorption spectra of AML,
Absorbance
Wavelength (nm)
Fig. 1. Zero order absorption spectra of 4 lg/mL BIS (—) and 10 lg/mL HCT (- - -) using methanol as blank.
ALI and HCT show highly overlapped spectral band in the region
200–350 nm (Fig. 2). The difficulty of BIS determination in the bin-
ary mixture and the severe overlap in the ternary mixture nomi-
nated these combinations to be ideal mixtures presenting the
comparative study to show the advantages and disadvantages of
both algorithms in analysis of overlapping spectra. HCT in the bin-
ary mixture and AML in the ternary mixture can be directly deter-
mined in the zero order spectra at their kmax, 316 and 360 nm
respectively, without interference from other components, but
they were determined by ND and CWT for investigational
purposes.
For using ND method, the parameters of derivative calculation,
wavelength increment over which the derivative is obtained (Dk)
and the scaling factor, were optimized. Different Dk (2, 4 and 8)
and scaling factors of 10, 100 and 1000 were tested, and the best
parameters, in terms of spectral shapes, linearity and recovery,
are mentioned under procedure.
For analysis of chosen mixtures using CWT algorithm, different
wavelet families were tested such as Haar, Daubechies (db),
Symlets (sym), Coiflets (coif), Meyer (meyr), Gaussian (gaus),
Mexican hat (mexh) and Morlet (morl). The only parameter that
should be optimized after choosing the proper wavelet is the scal-
ing parameter. No need for scaling factor as CWT has the advantage
of signal amplification compared to ND.
4.1. Binary mixture
In this section, we will focus on determination of BIS being a
challenge in presence of highly absorptive HCT. Thus, the choice
of derivative order and optimum parameters is mainly considered
for determination of BIS.
For analysis of BIS in this mixture using ND algorithm, the 1st,
2 , 3rd and 4th order derivatives were obtained (Fig. 3). The 1st
nd
order did not show any possible zero-crossing points that can be
used for BIS determination (Fig. 3a), while the wavelengths 241.0
and 270.0 nm did not show a real zero-crossing. Possible
 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955
Absorbance
Wavelength (nm)
Fig. 2. Zero order absorption spectra of 4 lg/mL AML (—), 120 lg/mL ALI (....) and 10 lg/mL HCT (- - -), using methanol as blank.
zero-crossing points were noticed in higher order derivatives,
which are 230.3 and 278.5 nm in 2nd order, 233.3 and 283.6 nm
in 3rd order and 229.5, 235.3 and 287.5 nm in 4th order. In contrast
to 1st order derivative, true zero-crossing points were detected in
2nd (278.5 nm), 3rd (283.6 nm) and 4th (287.5 nm) orders
(Fig. 3b–d). The zero-crossing point in 2nd order showed poor lin-
earity and recovery of laboratory prepared mixtures, while the
points in 3rd and 4th orders resulted in good recoveries in some
laboratory-prepared mixtures, where BIS was found in higher ratio
than HCT. For determination of HCT, the 1st order derivative was
enough to get zero-crossing point (272.3 nm) and good recoveries
in all ratios of laboratory-prepared mixtures were obtained
Table 1.
For analysis of BIS using CWT algorithm, the families were
tested in different orders and with different scaling parameters
and the best results, regarding linearity and recovery% in
laboratory-prepared mixtures, were obtained with 1st derivative
calculated using morl family and scaling = 200 (Fig. 4) at
278.3 nm. Different wavelengths were detected as possible
zero-crossing points at 255.3, 265.7, 278.3, 292.2 and 306.3 nm.
The points at 255.3 and 265.7 nm were not real zero-crossing
points and the last two wavelengths showed bad linearity and sen-
sitivity compared to 278.3 nm. HCT could be determined with the
same parameters at 283.6 nm.
4.2. Ternary mixture
For analysis of this mixture using ND algorithm, the 1st order
derivative spectra were obtained where a zero-crossing point
(386.8 nm) was detected for AML determination (Fig. 5a). But in
the same order spectra, ALI and HCT were not resolved, so the
2nd order derivatives were calculated and a zero-crossing point
(332.1 nm) was noticed for determination of HCT (Fig. 5b). For
determination of ALI, the 3rd and 4th order derivatives were tried.
While no zero-crossing points were observed in the 3rd order spec-
tra (Fig. 5c), the 4th order showed a zero-crossing point (287.5 nm)
that was used for ALI determination (Fig. 5d).
949
For analysis of the ternary mixture using CWT algorithm, differ-
ent families were tried in different orders and scaling. The best
results regarding presence of zero-crossing points, linearity and
selectivity were obtained with 1st order derivatives using gaus-1,
gaus-8 and coif-3 families and scaling = 200 for determination of
AML, ALI and HCT at 387.2, 315.6 and 289.1 nm, respectively
(Fig. 6).
The selectivity of the proposed ND and CWT methods was
assessed by the analysis of laboratory prepared mixtures contain-
ing different ratios of the drugs (Tables 1 and 2). The proposed
CWT method was applied for the determination of BIS and HCT
in ConcorÒ Plus tablets and AML, ALI and HCT in AmturnideÒ
tablets and the validity was further assessed by applying the stan-
dard addition technique (Tables 3 and 4). Results obtained by CWT
for analysis of the dosage forms were statistically compared to
those obtained by the reported HPLC methods [51,63]. The results
showed no significant differences between the proposed method
and the reported ones as presented in Table 5. Validation was done
according to ICH recommendations and the validation parameters
are shown in Table 6.
4.3. CWT versus ND
ND is the simplest method of derivative calculation. The major
problem is the choice of Dk. Wide Dk values can be used to sup-
press noise and increase SNR, but they degrade resolution as well.
Finding the optimum Dk to compromise between the high resolu-
tion and good signal to noise ratio (S/N) is considered a challenge.
Also the values of S/N obtained by ND are very small, so they are
usually multiplied by scaling factor to be considerable.
The 1st order derivative fails to resolve severely overlapped
spectra, so we use higher order derivatives. This approach can help
in case the components to be determined are present in a reason-
able ratio and have comparable absorptivities. In the case of BIS in
its binary mixture with HCT, the zero order spectrum of BIS is com-
pletely enclosed with that of HCT showing very similar features
and low absorptivity, so 1st order derivative failed to present a
950 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955

a b

HCT BIS

c
d

BIS

BIS

Fig. 3. Different order absorption spectra of 40 lg/mL BIS (–) and 1–22 lg/mL HCT (- - -) using ND method showing zero-crossing points for determination of BIS and HCT. (a)
1st order (b) 2nd order (c) 3rd order (d) 4th order.

Table 1
Determination of BIS and HCT in laboratory prepared mixtures by ND and CWT methods.

Concentration (lg/mL) Recovery%a

BIS HCT
BIS HCT ND (3rd order) ND (4th order) CWT ND (1st order) CWT
b b
4 10 54.7 115.1 99.5 101.8 101.3
6 6 89.9b 103.9b 102.4 101.8 99.4
6 12 65.7b 107.8b 98.4 101.1 100.6
6 15 64.6b 109.7b 100.2 100.6 100.2
6 18 53.7b 92.2b 101.3 100.2 99.1
12 6 103.0 100.4 100.6 101.8 100.7
12 15 76.0b 92.6b 98.1 101.3 99.8
15 6 102.9 100.5 101.1 102.5 99.9
15 12 85.4b 99.7 99.0 101.9 100.2
15 15 77.3b 94.3b 98.5 100.9 99.3
18 6 101.6 97.3 101.6 102 99.0
Mean ± SD 102.5 ± 0.8 99.5 ± 1.5 100.1 ± 1.5 101.4 ± 0.7 100.0 ± 0.7
a
Average of three determinations.
b
Excluded according to rejection rule [67].

zero-crossing point for its determination. The 2nd, 3rd and 4th order
derivatives showed zero-crossing points for BIS determination, the
recoveries in laboratory prepared mixtures indicated that BIS can-
not be determined in some mixtures contained higher levels of
HCT. This may be attributed to the poor absorptivity of BIS, at
the wavelengths where zero-crossing points were observed, in
addition to the fact that higher order derivatives diminish the sig-
nal to noise ratio (S/N).
In contrast to ND, CWT could resolve BIS from HCT in different
ratios by 1st order derivative using morl family. Using CWT algo-
rithm offers the flexibility of using several wavelet families, which
results in transforming the spectra into 1st order derivative in dif-
ferent shapes. This fact increases the possibility of finding
zero-crossing points in the 1st order derivatives for compounds
with very similar spectra. The CWT algorithm also has the advan-
tage of signal amplification compared to ND [66]. In addition to
 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955 951

Amplitude

HCT

BIS

Wavelength (nm)

st
Fig. 4. 1 order absorption spectra of 40 lg/mL BIS (—) and 1–22 lg/mL HCT (- - -) using CWT method (morl) showing zero-crossing points for BIS and HCT determination.

a b
AML
HCT

c d

ALI

Fig. 5. Different order absorption spectra of 35 lg/mL AML (–), 250 lg/mL ALI (....) and 25 lg/mL HCT (- - -) using ND method showing zero-crossing points for determination
of AML, ALI and HCT. (a) 1st order (b) 2nd order (c) 3rd order (d) 4th order.

the fact that calculating higher order derivatives is not necessary as
the 1st order derivative in CWT is enough to get zero-crossing
points. Thus CWT maintains high S/N in contrast to ND calculation,
which is the reason why CWT showed higher sensitivity as shown
from LOD and LOQ values in Table 6. These facts can explain why
zero-crossing points for determination of BIS was not observed in
ND 1st order derivative, in contrast to the 1st order morl wavelet.
The higher S/N of CWT spectra allowed the determination of BIS
in mixtures where HCT is found in higher ratios, while 2nd, 3rd
and 4th order derivatives obtained by ND failed.
952 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955

AML

ALI

HCT

Fig. 6. First order absorption spectra of 35 lg/mL AML (–), 250 lg/mL ALI (....) and 25 lg/mL HCT (- - -), using CWT method showing zero-crossing point for determination of
AML, ALI and HCT. (a) gaus-1 (b) gaus-8 (c) coif-3.

The severe overlap between the spectra of AML, ALI and HCT in
the region 200–350 nm prevented the resolution of the drugs in 1st
order derivative in this region. So ALI and HCT were resolved only
in the 4th and 2nd order derivatives, respectively. AML was excep-
tion being determined in 1st order derivative at 386.8 nm where
ALI and HCT show no interference in their original spectra. Upon
using CWT families, gaus-1, gaus-8 and coif-3, the 1st order deriva-
tives were enough to resolve the three drugs.
The advantages of CWT over ND in calculating derivatives can
be summarized in the following:
 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955 953

Table 2
Determination of AML, ALI and HCT in their ternary mixture by ND and CWT methods.

Concentration (lg/mL) Recovery%a

AML ALI HCT
AML ALI HCT ND CWT ND CWT ND CWT
1 30 2.5 79.6b 102.6 99.0 98.5 102.2 99.8
2 60 5 101.0 102.0 100.0 99.2 98.5 100.9
4 120 10 101.5 102.3 99.6 99.2 99.6 101.3
5 30 25 97.6 102.3 101.7 100.0 100.6 100.6
6 180 15 98.3 102.2 99.1 100.9 98.5 101.2
12 30 5 100.2 101.8 102.0 98.9 100.7 98.1
15 15 15 100.5 100.9 94.8b 105.0b 101.0 99.4
20 20 20 99.9 100.4 95.0b 101.1 98.0 99.7
35 30 5 98.5 100.0 102.0 98.1 99.3 100.5
35 70 5 99.1 100.4 99.3 99.6 97.0 99.3
Mean ± SD 99.6 ± 1.3 101.5 ± 0.9 100.3 ± 1.3 99.5 ± 1.0 99.5 ± 1.6 100.1 ± 1.0
a
Average of three determinations.
b
Excluded according to rejection rule [67].

Table 3
Determination of BIS and HCT in ConcorÒ Plus tablets by the proposed CWT and the reported HPLC method [51] with application of standard addition technique.

Product Drug CWT Reported methoda Standard addition CWT

Takenb (lg/mL) Added (lg/mL) Found (lg/mL) Recovery%c
ConcorÒ Plus 5/12.5 BIS 99.1 ± 0.4 99.4 ± 0.9 6 5 5.04 100.8
6 5.96 99.3
7 7.06 100.9
Mean ± SD 100.3 ± 0.9
HCT 99.7 ± 0.4 100.2 ± 0.8 10 8 7.97 99.6
10 9.94 99.4
12 12.11 100.9
Mean ± SD 100.0 ± 0.8
ConcorÒ Plus 10/25 BIS 100.5 ± 0.7 100.7 ± 0.4 6 5 4.96 99.2
6 6.03 100.5
7 6.97 99.6
Mean ± SD 99.8 ± 0.7
HCT 99.5 ± 0.4 99.9 ± 0.9 10 8 7.95 99.4
10 9.98 99.8
12 12.03 100.3
Mean ± SD 99.8 ± 0.5
a
RP-HPLC using C18 and 0.1 M potassium dihydrogen phosphate buffer and acetonitrile (70:30, v/v) as mobile phase. The UV detection was carried out at 228 nm at a flow
rate of 1.0 mL/min.
b
Standard addition were done on two different dilutions of the dosage form for the drugs to be within the linearity range.
c
Average of three determinations.

Table 4
Determination of AML, ALI and HCT in AmturnideÒ tablets by the proposed CWT and the reported HPLC method [63] with application of standard addition technique.

Product Drug CWT Reported methoda Standard addition CWT

Takenb (lg/mL) Added (lg/mL) Found (lg/mL) Recovery%c
Ò
Amturnide 10/300/25 AML 100.6 ± 1.1 100.4 ± 0.9 6 4 3.98 99.5
6 6.09 101.5
8 8.03 100.4
Mean ± SD 100.5 ± 1.0
ALI 99.9 ± 0.6 100.3 ± 0.8 120 110 109.31 99.4
120 121.55 101.3
130 130.73 100.6
Mean ± SD 100.4 ± 1.0
HCT 100.4 ± 0.7 99.9 ± 0.7 10 5 5.01 100.2
10 10.11 101.1
15 14.88 99.2
Mean ± SD 100.2 ± 0.9
a
RP-HPLC using acetonitrile: methanol: 50 mM phosphate buffer (20:50:30% by volume) adjusted to pH 3 with orthophosphoric acid and flow rate of 1.0 mL/min at
239 nm.
b
Standard addition were done on two different dilutions of the dosage form for the drugs to be within the linearity range.
c
Average of three determinations.
954 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955

Table 5
Statistical comparison for the results obtained by the proposed CWT and the reported HPLC [51,63] methods for the analysis of ConcorÒ Plus and AmturnideÒ tablets.

Value ConcorÒ Plus tablets AmturnideÒ tablets

a
CWT Reported method CWT Reported methodb
BIS HCT BIS HCT AML ALI HCT AML ALI HCT
Mean 99.8 99.6 100.0 100.0 100.6 99.9 100.4 100.4 100.3 99.9
RSD% 0.9 0.4 1.0 0.8 1.1 0.6 0.7 0.9 0.8 0.7
n 6 6 6 6 6 6 6 6 6 6
Variance 0.832 0.140 0.903 0.582 1.329 0.347 0.496 0.755 0.617 0.499
Student’s t testc 0.474 1.116 – – 0.405 0.845 1.228 – – –
(2.228)
F value c 1.085 4.157 – – 1.76 1.777 1.007 – – –
(5.05)
a
RP-HPLC using C18 and 0.1 M potassium dihydrogen phosphate buffer and acetonitrile (70:30, v/v) as mobile phase. The UV detection was carried out at 228 nm at a flow
rate of 1.0 mL/min.
b
RP-HPLC using acetonitrile: methanol: 50 mM phosphate buffer (20:50:30% by volume) adjusted to pH 3with orthophosphoric acid and flow rate of 1.0 mL/min at
239 nm.
c
The values in the parenthesis are the corresponding theoretical values of t and F at P = 0.05.

Table 6
Assay validation sheet of ND and CWT methods for the simultaneous determination of the mixtures under study.

Parameter Binary mixture Ternary mixture

BIS HCT AML ALI HCT
ND (3rd ND (4th CWT ND CWT ND CWT ND CWT ND CWT
order) order)
Accuracya 102.0 ± 0.3 99.7 ± 2.2 100.0 ± 0.6 99.9 ± 1.6 99.8 ± 0.5 100.6 ± 1.2 100.8 ± 1.2 99.9 ± 1.0 100.3 ± 0.9 98.5 ± 0.8 99.7 ± 1.0
(Mean ± SD)
Precision
– Repeatabilityb 1.3 1.2 0.6 0.7 0.6 0.9 0.9 1.0 0.9 0.7 0.8
– Intermediate 1.8 1.6 0.9 1.2 0.8 1.2 1.3 1.2 1.1 1.0 1.1
precisionc
Robustnessd 1.5 1.5 0.8 1.4 0.6 1.1 0.8 1.3 1.3 1.6 1.6
e
LOD (lg/mL) 1.42 1.41 0.57 0.44 0.22 0.57 0.26 4.50 4.45 1.48 0.59
LOQe (lg/mL) 4.75 4.69 1.90 1.48 0.73 1.89 0.88 15.01 14.84 4.93 1.95
Linearity
– Slope 0.0291 0.0086 0.0115 0.1743 0.1062 0.0049 0.1644 0.0113 0.0392 0.0027 0.1507
– Intercept 0.0021 0.0116 0.0014 0.0049 0.0258 0.0011 0.0030 0.3070 0.0070 0.0001 0.0190
– Correlation 0.9998 0.9997 0.9999 0.9998 0.9999 0.9999 1.000 0.9999 0.9998 0.9998 0.9999
coefficient (r)
– Range (lg/mL) 5.0–40.0 5.0–40.0 2.0–40.0 2.0–22.0 1.0–22.0 2.0–35.0 1.0–35.0 15.0– 15.0– 5.0–25.0 2.0–25.0
250.0 250.0
a
The accuracy (n = 3), average of three concentrations of each drug.
b
The intraday (n = 3), average of three concentrations of each drug repeated three times within day.
c
The interday (n = 3), average of three concentrations each drug repeated three times in three days.
d
Robustness (n = 3), average of three concentrations each drug analyzed using 75% and 70% methanol.
e
LOD and LOQ were calculated using the following equations: LOD = 3Sr and LOQ = 10Sr, r is the standard deviation of residuals and S is the slope.

1. Flexibility in calculating 1st order derivatives in different shapes similar spectra without the need of higher order derivative
using different wavelet families, that allows CWT to provide calculation.
zero-crossing points for complex mixtures.
2. Signal amplification provided by the algorithm together with its
flexibility in 1st order derivatives gives CWT high sensitivity References
compared to ND. [1] E. Dymond, Math. Proc. Cambridge Philos. Soc. 22 (1924) 405–408.
[2] V. Hammond, W. Price, J. Opt. Soc. Am. 43 (1953) 924–930.
[3] J. Morrison, J. Chem. Phys. 21 (1953) 1767–1772.
[4] A.T. Giese, C.S. French, Appl. Spectrosc. 9 (1955) 78–96.
5. Conclusion [5] A. Martin, Spectrochim. Acta A 14 (1959) 97–103.
[6] J.R. Morrey, Anal. Chem. 40 (1968) 905–914.
[7] A. Savitzky, M.J. Golay, Anal. Chem. 36 (1964) 1627–1639.
Using CWT offered many advantages over traditional ND. One [8] G. Talsky, Derivative Spectrophotometry: Low and High Order, VCH, California,
advantage is that CWT algorithm maintains high S/N in contrast 1994.
to ND calculation. In addition to the fact that calculating higher [9] N. Funasaki, S. Hada, S. Neya, J. Phys. Chem. B 103 (1999) 169–172.
[10] B. Ebeshi, E. Vaikosen, D. Tuma, Int. J. Biol. Chem. Sci. 7 (2014) 2591–2599.
order derivatives is not necessary in CWT as the 1st order deriva-
[11] Y. Ni, J. Bai, Talanta 44 (1997) 105–109.
tive is enough to get zero-crossing points for severely overlapped [12] A. Molina, M. López-Tenés, C. Serna, J. Electrochem. Soc. 147 (2000) 3429–
spectra. The flexibility of CWT algorithm allowing the use of sev- 3435.
eral wavelet families in different orders, results in transforming [13] M. Blanco, J. Gene, H. Iturriaga, S. Maspoch, J. Riba, Talanta 34 (1987) 987–993.
[14] F. Singleton, G. Collier, J. Appl. Chem. 6 (1956) 495–510.
the spectra into different shapes of 1st order derivatives, which [15] W. Snelleman, T. Rains, K. Yee, H.D. Cook, O. Menis, Anal. Chem. 42 (1970)
increases the possibility of finding zero-crossing points in very 394–398.
 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955
[16] T.C. O’Haver, Modulation and Derivative Techniques in Luminescence
Spectroscopy, Springer, New York, 1976.
[17] A.Y. El-Sayed, N.A. El-Salem, Anal. Sci. 21 (2005) 595.
[18] C.B. Ojeda, F.S. Rojas, Microchem. J. 106 (2013) 1–16.
[19] M. Korany, A. Wahbi, M. Elsayed, S. Mandour, Farmaco Ed. Prat. 39 (1984) 243.
[20] M. Korany, A. Wahbi, M. Elsayed, S. Mandour, Anal. Lett. 17 (1984) 1373–1389.
[21] M.A. Korany, A.A.S. El-Din, N.A. Abael-salam, Anal. Lett. 17 (1984) 483–495.
[22] F. Salinas, J. Nevado, A. Mansilla, Talanta 37 (1990) 347–351.
[23] J. Berzas, C. Guiberteau, F. Salinas, Talanta 39 (1992) 547.
[24] E. Dinc, F. Onur, Anal. Chim. Acta 359 (1998) 93–106.
[25] E. Dinç, D. Baleanu, Spectrochim. Acta A 63 (2006) 631–638.
[26] I. Daubechies, Ten Lectures on Wavelets, SIAM, Pennsylvania, 1992.
[27] A.K.M. Leung, F.T. Chau, J.B. Gao, Anal. Chem. 70 (1998) 5222–5229.
[28] X.D. Dai, B. Joseph, R. Motard, Introduction to Wavelet Transform and Time-
frequency Analysis, Springer, New York, 1994.
[29] R.R. Coifman, G. Weiss, Bull. Am. Math. Soc. 83 (1977) 569–645.
[30] A. Calderón, A. Zygmund, Stud. Math. 46 (1973) 297–299.
[31] A. Grossmann, J. Morlet, SIAM J. Math. Anal. 15 (1984) 723–736.
[32] S.G. Mallat, IEEE Trans. Acoust. 37 (1989) 2091–2110.
[33] I. Daubechies, IEEE Trans. Inf. Theory 36 (1990) 961–1005.
[34] I. Daubechies, SIAM J. Math. Anal. 24 (1993) 499–519.
[35] D.L. Donoho, I.M. Johnstone, C.R. Acad. Sci. I Math. 319 (1994) 1317–1322.
[36] V. Barclay, R. Bonner, I. Hamilton, Anal. Chem. 69 (1997) 78–90.
[37] X. Zou, J. Mo, Anal. Chim. Acta 340 (1997) 115–121.
[38] M. Bos, E. Hoogendam, Anal. Chim. Acta 267 (1992) 73–80.
[39] H.W. Tan, S.D. Brown, J. Chemom. 16 (2002) 228–240.
[40] U. Depczynski, K. Jetter, K. Molt, A. Niemöller, Chemom. Intell. Lab. Syst. 47
(1999) 179–187.
[41] T. O’haver, G. Green, Anal. Chem. 48 (1976) 312–318.
[42] E. Dinç, D. Baleanu, J. Pharm. Biomed. Anal. 31 (2003) 969–978.
[43] E. Dinç, A. Özdemir, D. Baleanu, Talanta 65 (2005) 36–47.
[44] E. Dinç, M. Kanbur, D. Baleanu, Die Pharm. 60 (2005) 892–896.
[45] E.S. Elzanfaly, S.A. Hassan, M.Y. Salem, B.A. El-Zeany, Spectrochim. Acta A 140
(2015) 334–343.
[46] W.H. Frishman, B.S. Bryzinski, L.R. Coulson, V.L. DeQuattro, N.D. Vlachakis, W.J.
Mroczek, G. Dukart, J.D. Goldberg, D. Alemayehu, K. Koury, Arch. Internal Med.
154 (1994) 1461–1468.
955
[47] A.D. Gudruman, N. Bibire, G. Tantaru, M. Apostu, M. Vieriu, V. Dorneanu, Rev.
Chim. 64 (2013) 393–396.
[48] H.W. Darwish, S.A. Hassan, M.Y. Salem, B.A. El-Zeany, Spectrochim. Acta A 122
(2013) 744–750.
[49] M. Emanual, J. Chromatogr. Sep. Tech. (2012).
[50] H.W. Darwish, S.A. Hassan, M.Y. Salem, B.A. El-Zeany, Int. J. Pharma Bio Sci. 4
(2013) 345–356.
[51] S.J. Joshi, P.A. Karbhari, S.I. Bhoir, K. Bindu, C. Das, J. Pharm. Biomed. Anal. 52
(2010) 362–371.
[52] A. Singh, T. Sivakumar, D. Jain, U. Mishra, A. Yadav, T. Sharma, Int. J. Pharm.
Res. Dev. 3 (2011) 11–22.
[53] J. Shi, Y. Chen, Y.P. Long, Cent. South Pharm. 11 (2012) 000.
[54] S. Kurbanoglu, P.R. San Miguel, B. Uslu, S.A. Ozkan, Chromatographia (2013) 1–
7.
[55] P.N. Bhoya, E.M. Patelia, J. Chromatogr. Sep. Tech. 4 (2013) 163–166.
[56] Y. Lacourcière, S. Taddei, G. Konis, H. Fang, T. Severin, J. Zhang, J. Hypertens. 30
(2012) 2047–2055.
[57] F. Belal, M. Walash, N. El-Enany, S. Zayed, J. Chromatogr. B 933 (2013) 24–29.
[58] H.W. Darwish, S.A. Hassan, M.Y. Salem, B.A. El-Zeany, Spectrochim. Acta A 113
(2013) 215–223.
[59] H.W. Darwish, S.A. Hassan, M.Y. Salem, B.A. El-Zeany, Spectrochim. Acta A 104
(2013) 70–76.
[60] M.I. Ezzeldien, E. Shokry, M.A. Fouad, R.I. Elbagary, Int. J. Adv. Chem. 1 (2013)
13–20.
[61] H.W. Darwish, S.A. Hassan, M.Y. Salem, B.A. El-Zeany, Int. J. Spectrosc. 2013
(2013) 8.
[62] H.W. Darwish, S.A. Hassan, M.Y. Salem, B.A. El-Zeany, Int. J. Pharma Bio Sci. 4
(2013) 230–243.
[63] G.K. Swamy, P. Sravanthi, M.L. Surekha, J.K.J. Rao, Int. J. ChemTech Res. 4
(2012) 1666–1673.
[64] M.M. Salim, W.M. Ebeid, N. El-Enany, F. Belal, M. Walash, G. Patonay, J. Sep. Sci.
37 (2014) 1206–1213.
[65] F.T. Chau, Y.Z. Liang, J. Gao, X.G. Shao, Chemometrics: From Basics to Wavelet
Transform, John Wiley & Sons, New Jersey, 2004.
[66] E. Dinç, M. Kanbur, D. Baleanu, Spectrochim. Acta A 68 (2007) 225–230.
[67] F.J. Anscombe, Technometrics 2 (1960) 123–146.
 Journal of Electroanalytical Chemistry 827 (2018) 51–57

Contents lists available at ScienceDirect

Journal of Electroanalytical Chemistry

journal homepage: www.elsevier.com/locate/jelechem

MWCNT/CdSe quantum dot modified glassy carbon electrode for the T

determination of clopidogrel bisulfate in tablet dosage form and serum
samples
⁎
Goksu Ozcelikay, Sevinc Kurbanoglu, Burcin Bozal-Palabiyik, Bengi Uslu , Sibel A. Ozkan
Ankara University, Faculty of Pharmacy, Department of Analytical Chemistry, 06560 Ankara, Turkey

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

Keywords: In this study, clopidogrel (CLP), which is an antiplatelet agent widely used in the prevention of ischemic stroke,
Clopidogrel was analyzed using functionalized multi-walled carbon nanotubes (fMWCNTs) and CdSe quantum dots (QDs)
Carbon nanotube modified glassy carbon electrode. The dependence of intensities of currents and potential on pH, concentration,
Quantum dots scan rate, nature of the buffer was further investigated. The oxidation process exhibited mixed diffusion under
Voltammetry
adsorption controlled process depending on pH. For the analytical application, adsorptive stripping differential
Determination
pulse voltammetric (AdSDPV) technique was used and all operational parameters were optimized. The plot of
peak current vs CLP concentration consisted of a linear segment in the concentration range between
2.0 × 10−6 M and 4.0 × 10−5 M and the detection limit (3σ) was obtained 7.55 × 10−8 M in pH 2.14 phosphate
buffer solution. Moreover, in order to evaluate the analytical applicability of the proposed method, it was ap-
plied to the determination of CLP in serum samples with linear range of 2.5 × 10−6–1.5 × 10−5 M and the
detection limit value of 2.99 × 10−7 M. Determination of CLP in tablet dosage form Atervix® was also proposed
with acceptable recovery values.

1. Introduction
Clopidogrel (CLP, (S)-Methyl 2-(2-chlorophenyl)-2-(6,7-dihy-
drothieno[3,2-c]pyridin-5(4H)-yl)acetate, Scheme 1), a thienopyridine
derivative drug, is a prodrug that prevents adenosine diphosphate-in-
duced platelet aggregation selectively and irreversibly; there is no di-
rect antiplatelet activity in the in vitro environment [1,2]. It transforms
into active form by hepatic cytochrome P-450 3A4 and 3A5 [3]. CLP
reduces the incidence of atherosclerotic events (myocard infarctus,
stroke and vascular death) at people with atherosclerosis and at those
who has the risk for having MI, stroke, acute coronary syndrome or
peripheral artery disease. Especially, it is more effective than aspirin for
peripheral artery disease [2–5].
Nano-sized materials can be used by mixing them with electrode
materials or they can be used via direct application on the electrode
surface. Electrodes, modified with nanomaterials, can be fabricated by
chemical vapor deposition, electrochemical deposition, chemical at-
tachment with or without crosslinking agents, and mixing of nano-
particles with electrode materials [6]. Among all nanomaterials, carbon
nanotubes (CNTs) are one of the most preferred materials for electrode
modification. Dated from the discovery of CNTs in 1991 by Sumio
Iijima, this modifier attracted scientists' attention because of its po-
tential in biological and biomedical applications [7–11]. CNTs, im-
portant carbon allotropes, are all‑carbon hollow and well-ordered gra-
phitic nanomaterials with a high aspect ratio, having lengths from
several hundred nanometers to several micrometers. CNTs are cylind-
rical nanostructures obtained by rolling up single or multiple sheets of
graphene to give single-walled carbon nanotubes (SWCNTs) and multi-
walled carbon nanotubes (MWCNTs). SWCNTs have 0.4–2 nm dia-
meters and MWCNTs have 2–100 nm coaxial diameters [7,9,12–18].
They show unique properties such as electrical conductivity, large
surface area, high electrocatalytic activity and strong chemical stability,
for being an electrode modifier [19]. CNTs can be functionalized with
some groups such as eCOOH, eOH, eNH2 and eSH to improve the
adsorption capacities [20] and solubility [7–11]. On the other hand, in
recent years quantum dots (QDs) have also been presented as an at-
tractive electrode modification agent. QDs are semiconductor nano-
particles with size range of 1–100 nm nanocrystal particle
[14–16,21–24]. These materials having distinct electronic states are
widely preferred recently because of their highly alterable properties.
Having been used as electron transfer catalysts, QDs have electron hole
pairs on which redox reactions occur in a liquid interface. The reason

⁎
Corresponding author.
E-mail address: buslu@pharmacy.ankara.edu.tr (B. Uslu).

https://doi.org/10.1016/j.jelechem.2018.09.005
Received 24 July 2018; Received in revised form 31 August 2018; Accepted 3 September 2018
Available online 04 September 2018
1572-6657/ © 2018 Elsevier B.V. All rights reserved.
G. Ozcelikay et al.
Scheme 1. Chemical structure of CLP.
Fig. 1. AdSDPV of 1 × 10−5 M CLP at the bare GCE (1, black), CdSeQDs/GCE
(2, pink), fMWCNT/GCE (3, red) and fMWCNT/CdSeQDs/GCE (4, blue) in
pH 2.14 phosphate buffer solution. (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this
article.)
for good electron-catalytic activity of the QDs is their small particle size
and its quantum cavity. The band gap of the QDs determines the che-
mical energy which is present at the QDs under their excited status
[25]. As a semiconductor, the CdSe have a three-dimensional quantum-
size effect which is the result of electron and hole localization over a
confined space. The outcome of this process is the quantization of
electron energy levels, which leads to an increase in the optical band
gap of the CdSe [26].
Various methods have been developed for CLP determination, in-
cluding high performance liquid chromatography [27–37], liquid
chromatography with tandem mass spectrometry [38–41], ultra-per-
formance liquid chromatography with tandem mass spectrometry [42],
high performance thin layer chromatography [43], spectrophotometry
[44–46], potentiometry [30] and voltammetry [20,31–33]. Although
chromatographic and spectrophotometric methods have been widely
performed for various analyses with high sensitivity and good se-
lectivity; they have some limitations, such as costly equipment, large
organic solvent consumption and time-consuming pre-treatment re-
quirements.
The voltammetric methods for analysis can be performed with rapid
response, no pre-treatment and low organic solvent consumption [19].
In addition to these superiorities, electrode modification with nano-
sized materials also contributes to sensitivity and selectivity of vol-
tammetric methods and these modified electrodes are widely used for
52
Journal of Electroanalytical Chemistry 827 (2018) 51–57
detection and determination of many biological and chemical com-
pounds [6,50–53].
In this research, fMWCNT and CdSeQDs modified GCE was devel-
oped for sensitive determination of CLP from tablet dosage forms and
synthetic serum samples. For this purpose, AdSDPV technique was de-
veloped and required validation studies were also realized according to
ICH Guidelines [54]. The developed sensor showed high sensitivity and
desired reproducibility for CLP. The advantages of developed sensor
and technique were compared with the voltammetric methods in the
literature.
2. Experimental
2.1. Apparatus
AUTOLAB-PGSTAT100N (Eco Chemie, Utrecht, The Netherlands)
electrochemical instrument was used for performing cyclic voltam-
metry, differential pulse voltammetry (DPV) and AdSDPV, which is
controlled by NOVA 2.1 software. Electrochemical measurements were
realized using three-electrode system including bare GCE (BASi; ø:
3 mm, diameter) or fMWCNT/CdSeQDs/GCE as the working electrode,
platinum wire (BASi) as the counter electrode, and Ag/AgCl (BASi; 3 M
KCl) as the reference electrode. pH measurements were carried out by a
pH meter Model 538 (WTW, Austria) with a combined electrode (glass
electrode–reference electrode). The modified electrodes were char-
acterized by SEM and energy dispersive X-ray (EDX) analysis using
ZEISS EVO 40 (Merlin, Carl Zeiss). Atomic force microscopy (AFM)
analysis was also performed for characterization using a NT-MDT
Solver Pro Instruments. The images were obtained using the soft tap-
ping mode of AFM probes, model NSG11/Pt.
2.2. Reagents and chemicals
Clopidogrel bisulphate (CLP) and its dosage form, Atervix® tablets, were
kindly provided by Biofarma İlaç (Istanbul, Turkey). Required amount of
CLP was weighed and dissolved in methanol for preparing 1.0 × 10−3 M
stock solution and it is stored in the refrigerator. 20% (v/v) methanol per-
cent was kept as constant in all pH values and buffers to avoid the solubility
problems. eCOOH group functionalized multiwalled carbon nanotubes
(fMWCNT) and CdSeQDs (DRP-QDCORE550: λem = 550 nm;
concentration = 21 ± 2 μM; diameter Ø = 2.52 ± 0.03 nm) were pur-
chased from DropSens (Llanera, Spain).
For electrochemical measurements different buffer solutions such as
acetate (pH 3.7–5.7), phosphate (pH 2.0; 3.0; 6.0–8.0) and Britton-
Robinson (BR) buffer (pH 2.0–11.0) were used. Reagents were of ana-
lytical grade and buffer solutions were prepared by using distilled
water. CH3COOH, H3PO4, methanol were obtained from
Sigma–Aldrich. NaH2PO4·2H2O and Na2HPO4 were purchased from
Riedel-de Haen. N,N-dimethylformamide (DMF) was obtained from
Merck.
2.3. Preparation of modified fMWCNT/CdSeQDs/GCE
Homogenous and stable fMWCNT suspension was prepared in DMF
in the concentration of 0.5 mg/mL and sonicated for 2 h. Bare GCE
surface was cleaned mechanically with alumina slurry on a polishing
pad, rinsed with distilled water and dried in air.
fMWCNT/CdSeQDs suspension was prepared by mixing 200 μL
fMWCNT and 50 μL CdSeQDs (dispersed in chloroform by commer-
cially) under ultrasonic stirring during 30 min. The volume of
fMWCNT/CdSeQDs suspension (4:1) was dropped as 5 μL onto the
cleaned GCE and the modified electrode was kept at room temperature
for 2 h before use. The modified electrode was cleaned electro-
chemically with cyclic voltammetry (6 cycles) between measurements
and the electrode was prepared daily.
G. Ozcelikay et al.
Fig. 2. SEM images of bare GCE (A) and CdSeQDs/GCE (B). fMWCNT/GCE (C). fMWCNT/CdSeQDs/GCE (D) Scale bars of SEM images are in 300 nm. SEM-EDX
images of fMWCNT/CdSeQDs/GCE (E).
2.4. Tablet and recovery assay procedure
Five Atervix® tablets were taken and crushed into a fine powder in a
mortar. A required amount of this powder, which is enough for pre-
paring 1.0 × 10−3 M stock solution, was accurately weighed, dissolved
in methanol and sonicated. Solutions for analyzing were prepared by
taking aliquots of the clear supernatant of tablet solution and diluting
with pH 2.14 phosphate buffer solutions. In order to investigate the
tablet excipients effect, the accuracy of the technique, known amounts
of the pure drug was added into the analyzed tablet solution.
2.5. Determination of CLP in synthetic serum sample
In order to analyze CLP in synthetic serum sample, 3.6 mL serum
sample was mixed with 5.4 mL of acetonitrile and 1 mL of 1 × 10−4 M
CLP standard sample. The sample was centrifuged for 20 min at
53
Journal of Electroanalytical Chemistry 827 (2018) 51–57
3000 rpm. From this human serum stock, desired amount of CLP was
added to the pH 2.14 phosphate buffer solution while keeping the 20%
methanol amount as constant. The calibration curve related with the
increasing concentrations of CLP was obtained and the recovery studies
were achieved using the obtained calibration data.
3. Results and discussion
3.1. Effect of modification
AdSDP voltammograms of 1 × 10−5 M CLP in pH 2.14 phosphate
buffer solution for the bare GCE, CdSeQDs/GCE, fMWCNT/GCE and
fMWCNT/CdSeQDs/GCE were obtained to investigate the electro-
chemical behavior of CLP (Fig. 1). Modification by using only CdSeQDs
did not enhance the signal of CLP when compared to bare GCE. But
modification only with fMWCNT increased the peak current of CLP and
G. Ozcelikay et al. Journal of Electroanalytical Chemistry 827 (2018) 51–57

Fig. 4. Effects of pH on the peak current of 1 × 10−5 M CLP ((♦): phosphate

buffer; (■): BR buffer; (Δ): acetate buffer).

Fig. 3. AFM images of fMWCNT/CdSeQDs/GCE.

also showed electrocatalytic effect. Using both nanomaterials

(fMWCNT/CdSeQDs) at the same time made an additional increase in
the peak current of CLP compared to fMWCNT/GCE.
Using fMWCNT/CdSeQDs/GCE, related with electrocatalytic effect
of the nanomaterials, the oxidation peak of the CLP was moved to
~928 mV, greatly enhanced with ~100 mV shift. When only fMWCNTs
used the peak current enhanced ~11 times in AdSDPV as shown in
Fig. 1. The peak current of 1 × 10−5 M CLP in pH 2.14 was approxi-
mately 23 times higher at the fMWCNT/CdSeQDs/GCE compared to the
bare GCE under the same experimental conditions by AdSDPV. For the
analytical application using designed nanosensor, firstly operational
parameters have been optimized. The dependence of intensities of
currents and potential on pH, concentration, scan rate, nature of the Fig. 5. Effect of accumulation potential (A) and accumulation time (B) on the
buffer studies was further investigated. peak current using AdSDPV for 1 × 10−5 M CLP.

54
G. Ozcelikay et al. Journal of Electroanalytical Chemistry 827 (2018) 51–57

of fMWCNT and CdSeQDs (Fig. 2D) cover the surface of the GCE
completely. Fig. 2D represents the SEM images of fMWCNT, CdSeQDs
and rough, compact morphology of fMWCNT:CdSeQDs (4:1). In this
figure, fMWCNT solution is more intense than CdSeQDs solution and
fMWCNT covers CdSeQDs surface. The prepared electrodes were also
characterized by SEM EDX analyses. According to Fig. 2E, the EDX
results confirmed that the carbon amount in the surface originates from
CNTs, Cd and Se peaks clearly demonstrate the existence of the
CdSeQDs.
AFM images fMWCNT-CdSeQDs composite films are shown in Fig. 3
which indicated that CdSeQDs were located on the fMWCNTs and the
size distribution of CdSeQDs was quite uniform.

3.2. Influence of the pH on the peak currents

The electrooxidation of CLP was studied over pH range 2.0–11.0 for

AdSDPV in different buffer media, including acetate, phosphate and
Britton-Robinson buffers. Peak current (ip) vs pH and, peak potential
(Ep) vs pH graphs were obtained using the related results of pH studies
(Fig. 4). The voltammetric peak potential was shifted to less positive
values indicating that protons involved in the electrochemical reaction.
The plot of the Ep vs pH showed the straight lines with the equation
given below:

Ep (mV) = 924.79–13.34 pH; r = 0.993 (pH 2.14–6.33) at f MWCNT

/CdSeQDs/GCE using AdSDPV

The best peak shape and highest current value were obtained when
phosphate buffer solution at pH 2.14 was used. For further studies
pH 2.14 phosphate buffer solution was chosen as the supporting elec-
trolyte.

3.3. Influence of scan rate on fMWCNT/CdSeQDs/GCE

The relationship between peak current and scan rate was in-
vestigated to obtain information about electrochemical mechanism in-
tensely by cyclic voltammetry. The peak currents were followed within
the ranges of 5 to 1000 mV s−1 on f MWCNT/CdSeQDs/GCE in phos-
Fig. 6. Increasing concentrations of CLP for bulk solutions (A): (1) blank; (2) phate buffer solution with pH 2.14 for 1 × 10−5 M CLP.
4.00 × 10−6 M; (3) 6.00 × 10−6 M; (4) 8.00 × 10−6 M; (5) 1.00 × 10−5 M
It is found that the anodic current value related with CLP was linear
and for serum solutions (B): (1) blank; (2) 7.50 × 10−6 M; (3) 1.00 × 10−5 M;
to the scan rate applied with the equation below:
(4) 1.25 × 10−6 M in pH 2.14 phosphate buffer solution.
i p (μA) = 0.0126 v (mVs−1) + 0.4345; r = 0.997 (n = 10) at f MWCNT
Table 1 /CdSeQDs/GCE
Regression data of the calibration curves for quantitative determination of CLP
by AdSDPV. The logarithm of oxidation peak currents versus scan rates exhibited
a linear relationship with a slope of 0.751 which can be expressed by
Standard solution Serum
mixture of diffusion and adsorption controlled process [55] according
Measured Potential (V) 0.93 0.92 to following equation:
Linearity range (M) 2 × 10−6–4 × 10−5 2.50 × 10−6–1.5 × 10−5
Slope (μA M−1) 2.21 × 104 2.77 × 104 log i p (μA) = 0.7512 log v (mVs−1) − 1.1962, r
Intercept (μA) 0.01 −0.06
Correlation coefficient 0.99 0.99 = 0.990 (n = 10) at f MWCNT/CdSeQDs/GCE
LOD (M) 7.55 × 10−8 2.99 × 10−7
LOQ (M) 2.51 × 10−7 9.97 × 10−7
Within day precision of peak 0.34 1.04 3.4. Accumulation time and potential optimization
current (RSD%)a
Within day precision of peak 0.38 0.40 Accumulation potential (Eacc) and accumulation time (tacc) values
potential (RSD%)a
Between days precision of 0.94 1.19
for AdSDPV were optimized in order to obtain best conditions for the
peak current (RSD%)a analyses of CLP using fMWCNT/CdSeQDs/GCE. The effect of the Eacc on
Between days precision of 0.77 0.93 the peak current was studied for 1 × 10−5 M CLP between 0 V and
peak potential (RSD%)a 0.6 V while the tacc was 60 s. According to Fig. 5, although the highest
response was obtained at 0.15 V, Eacc value was selected as 0 V because
a
Obtained from five measurements.
of repeatability. The effect of the tacc on the peak current was studied
for 1 × 10−5 M CLP between 5 s and 90 s and the best response was
In order to modify the surface of GCE, fMWCNTs and CdSeQDs were
obtained as 60 s. Further voltammetric analysis were carried out in
mixed up in different ratios before dropping. It can be clearly seen from
pH 2.14 phosphate buffer solution using 0 V as accumulation potential
the SEM images that CdSeQDs (Fig. 2B), fMWCNT (Fig. 2C) and mixture
and 60 s as accumulation time.

55
G. Ozcelikay et al. Journal of Electroanalytical Chemistry 827 (2018) 51–57

Table 2
Comparison of voltammetric studies for the determination of CLP.
Electrode Electrochemical behavior Technique Linearity range (μM) LOD (μM) Reference

Glassy carbon electrode Oxidation DPV 80–1000 19.1 [47]

Gold electrode Oxidation SWV 757–2230 279.8 [48]
Bi2O3-Poly-p-aminophenol/GCE Oxidation DPV 3–1000 2.5 [49]
Hanging mercury drop electrode Reduction CSV 0.1–2.5 0.009 [37]
fMWCNT/CdSe QDs/GCE Oxidation AdSDPV 2–40 0.0755 This work

SWV: Square wave voltammetry; CSV: Cathodic stripping voltammetry.

Table 3 4. Conclusion
Results obtained for CLP determination and recovery from Atervix® tablet and
spiked serum by AdSDPV. In this study, CLP was analyzed using fMWCNTs and CdSeQDs
Atervix® Serum modified GCE. The volume of fMWCNT/CdSeQDs suspension ratio was
found as 4:1, and from this suspension 5 μL was used to modify the GCE.
Labeled claim (mg) 75.00 – The obtained results show that the fMWCNTs/CdSeQDs/GCE exhibited
Amount found (mg)a 74.40 –
good electrocatalytic performance toward CLP oxidation. The oxidation
RSD% 1.09 –
Bias% 0.85 – process exhibited mixed diffusion under adsorption controlled process
Added (mg) 20.00 4.00 depending on pH. The designed nanosensor was applied for the de-
Found (mg)a 19.87 3.61 termination of CLP in human serum and pharmaceutical dosage forms,
Average recovery % 99.33 90.20 successfully. The suggested electrochemical analysis enabled to analyze
RSD% of recovery 0.57 0.26
Bias% 0.67 9.80
CLP from blood serum, although the pH of human blood is 7.4 and the
analyses can be conducted at pH 2.14 phosphate buffer. In Table 1,
a
Obtained from three measurements. obtained results were compared with the literature studies according to
electrochemical behavior, linearity range and LOD values. The designed
3.5. Analytical characterization and validation of the method fMWCNTs/CdSeQDs/GCE sensor gave 0.075 μM level LOD value which
is lower than the other published voltammetric techniques in anodic
Under the optimum conditions, quantitative evaluation of CLP was way.
achieved based on the linear correlation between the peak current and
concentration. Fig. 6A showed some of the voltammograms from linear References
range between 2.0 × 10−6 and 4.0 × 10−5 M CLP and the linearity was
obtained with the following equation; [1] S.R. Mehta, S. Yusuf, R.J. Peters, M.E. Bertrand, B.S. Lewis, M.K. Natarajan,
K. Malmberg, H. Rupprecht, F. Zhao, S. Chrolavicius, I. Copland, K.A. Fox, Effects of
i p (μA) = 2.187 × 10 4C (M) + 0.0175; r = 0.999 (n = 7) pretreatment with clopidogrel and aspirin followed by long-term therapy in patients
undergoing percutaneous coronary intervention: the PCI-CURE study, Lancet 358
Validation of the analytical procedures for the quantitative assay of the (2001) 527–533 http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&
CLP was examined via evaluation of the LOD, LOQ, repeatability, re- form=6&dopt=r&uid=11520521.
[2] D.S. Aschenbrenner, S.J. Venable, Drug Therapy in Nursing, third ed., Lippincott
producibility, recovery, precision and accuracy. LOD and LOQ were calcu- Williams & Wilkins, Philadelphia, 2009.
lated from the equations of LOD = 3 s/m and LOQ = 10s/m [56,57] using [3] W.H. Frishman, A. Cheng-Lai, J. Nawarskas, Current Cardiovascular Drugs, third
the standard deviation of the response and the slope of the calibration curve ed., Current Medicine, New York, 2000, https://doi.org/10.1007/978-1-4615-
6767-7.
as 7.55 × 10−8 M and 2.51 × 10−7 M, respectively. Characteristics and the [4] P. Savi, J.M. Pereillo, M.F. Uzabiaga, J. Combalbert, C. Picard, J.P. Maffrand,
details of the validation parameters were reported in Table 1. M. Pascal, J.M. Herbert, Identification and biological activity of the active meta-
It can be clearly seen from the Table 2 that, the LOD value of the bolite of clopidogrel, Thromb. Haemost. 84 (2000) 891–896 (doi:00110891 [pii]).
[5] H. Schühlen, A. Kastrati, J. Dirschinger, J. Hausleiter, S. Elezi, A. Wehinger,
designed fMWCNTs/CdSeQDs/GCE nanosensor is quite low, 0.075 μM J. Pache, M. Hadamitzky, A. Schömig, Intracoronary stenting and risk for major
level, lower that the other voltammetric techniques in anodic way. The adverse cardiac events during the first month, Circulation 98 (1998) 104–111,
linearity range is quite broad as it is compared with the other studies https://doi.org/10.1161/01.CIR.98.2.104.
[6] M. Pasandideh-Nadamani, A. Omrani, M.R. Sadeghi-Maleki, A. Samadi-Maybodi,
especially the study achieved by Alghamdi [37] where the nearly same Application of CdS quantum dots modified carbon paste electrode for monitoring
LOD value was obtained. the process of acetaminophen preparation, Anal. Biochem. 502 (2016) 36–42,
https://doi.org/10.1016/j.ab.2016.02.015.
[7] W. Yang, P. Thordarson, J.J. Gooding, S.P. Ringer, F. Braet, Carbon nanotubes for
3.6. Determination of CLP in tablet dosage form and spiked serum sample
biological and biomedical applications, IOP Publ. Nanotechnol. Nanotechnol. 18
using the fMWCNT/CdSeQDs/GCE (2007) 412001–412012, https://doi.org/10.1088/0957-4484/18/41/412001.
[8] R.H. Baughman, A.A. Zakhidov, W.A. de Heer, Carbon nanotubes–the route toward
applications, Science 297 (2002) 787–792, https://doi.org/10.1126/science.
The developed AdSDPV technique was applied to the direct de-
1060928.
termination of CLP in tablet dosage form Atervix® tablet and spiked [9] C. Yang, M.E. Denno, P. Pyakurel, B.J. Venton, Recent trends in carbon nanoma-
serum by AdSDPV, using the related calibration equations. terial-based electrochemical sensors for biomolecules: a review, Anal. Chim. Acta
For human serum applications a calibration graph was described 887 (2015) 17–37, https://doi.org/10.1016/j.aca.2015.05.049.
[10] J. Wang, Carbon-nanotube based electrochemical biosensors: a review,
using the serum stock solution of 1 × 10−4 M. Between 2.5 × 10−6 and Electroanalysis 17 (2005) 7–14, https://doi.org/10.1002/elan.200403113.
1.5 × 10−5 M CLP, the linearity was obtained with the equation below [11] P. Pandey, M. Datta, B.D. Malhotra, Prospects of nanomaterials in biosensors, Anal.
and selected voltammograms of different concentrations were given in Lett. 41 (2008) 159–209, https://doi.org/10.1080/00032710701792620.
[12] H. Fecht, K. Brühne, P. Gluche, Carbon-based Nanomaterials and Hybrids, (2014),
Fig. 6B. https://doi.org/10.1201/b15673.
[13] Carbon-based nanomaterials, Essentials Nanosci. Nanotechnol, John Wiley & Sons,
i p (μA) = 2.77 × 10 4C (M)–0.055; r = 0.993 (n = 6)
Inc, Hoboken, NJ, USA, 2016, pp. 189–236, , https://doi.org/10.1002/
9781119096122.ch5.
The recovery studies, which were also applied for both determina- [14] A. Chen, S. Chatterjee, Nanomaterials based electrochemical sensors for biomedical
tion of CLP in spiked serum and tablet dosage form by AdSDPV, were applications, Chem. Soc. Rev. 42 (2013) 5425, https://doi.org/10.1039/
reported in Table 3. The recovery % and Bias % were all in acceptable c3cs35518g.
[15] B. Perez-Lopez, A. Merkoçi, Nanomaterials based biosensors for food analysis
ranges.

56
G. Ozcelikay et al.
applications, Trends Food Sci. Technol. 22 (2011) 625–639, https://doi.org/10.
1016/j.tifs.2011.04.001.
[16] J. Wang, Nanomaterial-based electrochemical biosensors, Analyst 130 (2005) 421,
https://doi.org/10.1039/b414248a.
[17] G. Hong, S. Diao, A.L. Antaris, H. Dai, Carbon nanomaterials for biological imaging
and nanomedicinal therapy, Chem. Rev. 115 (2015) 10816–10906, https://doi.org/
10.1021/acs.chemrev.5b00008.
[18] B.S. Murty, P. Shankar, B. Raj, B.B. Rath, J. Murday, Applications of Nanomaterials,
(2013), https://doi.org/10.1142/9789814340571_0009.
[19] Z. Liu, M. Jin, J. Cao, J. Wang, X. Wang, G. Zhou, A. Van Den Berg, L. Shui, High-
sensitive electrochemical sensor for determination of Norfloxacin and its metabo-
lism using MWCNT-CPE/pRGO-ANSA/Au, Sensors Actuators B 257 (2018)
1065–1075, https://doi.org/10.1016/j.snb.2017.11.052.
[20] J. Xu, Z. Cao, Y. Zhang, Z. Yuan, Z. Lou, X. Xu, X. Wang, A review of functionalized
carbon nanotubes and graphene for heavy metal adsorption from water: prepara-
tion, application, and mechanism, Chemosphere 195 (2018) 351–364, https://doi.
org/10.1016/J.CHEMOSPHERE.2017.12.061.
[21] Y. Tu, Q. Xu, Q.-J. Zou, Z.-H. Yin, Y.-Y. Sun, Y.-D. Zhao, Electrochemical behavior of
levodopa at multi-wall carbon nanotubes-quantum dots modified glassy carbon
electrodes, Anal. Sci. 23 (2007) 1321–1324 https://www.jstage.jst.go.jp/article/
analsci/23/11/23_11_1321/_pdf , Accessed date: 29 January 2018.
[22] H. Kang, L. Wang, M. O'Donoghue, Y.C. Cao, W. Tan, Nanoparticles for biosensors,
Opt. Biosens, 2008, https://doi.org/10.1016/B978-044453125-4.50017-6.
[23] C.N.R. Rao, A. Mu, The Chemistry of Nanomaterials, (2004), https://doi.org/10.
1002/352760247X.
[24] X. Yao, P. Yan, Q. Tang, A. Deng, J. Li, Quantum dots based electro-
chemiluminescent immunosensor by coupling enzymatic amplification for ultra-
sensitive detection of clenbuterol, Anal. Chim. Acta 798 (2013) 82–88, https://doi.
org/10.1016/j.aca.2013.08.029.
[25] S. Hooshmand, Z. Es'haghi, Simultaneous quantification of arginine, alanine, me-
thionine and cysteine amino acids in supplements using a novel bioelectro-nano-
sensor based on CdSe quantum dot/modified carbon nanotube hollow fiber pencil
graphite electrode via Taguchi method, J. Pharm. Biomed. Anal. 146 (2017)
226–235, https://doi.org/10.1016/j.jpba.2017.08.034.
[26] Y.M. Aniskevich, M.V. Malashchonak, P.V. Chulkin, G.A. Ragoisha, E.A. Streltsov,
Cadmium underpotential deposition on CdSe and CdS quantum dot films: size de-
pendent underpotential shift, Electrochim. Acta 220 (2016) 493–499, https://doi.
org/10.1016/j.electacta.2016.10.132.
[27] V. Sivarama Krishna, D.R. Kumar, K. Balamuralikrishna, C. Rambabu, Development
and validation of stability indicating RP-HPLC method for the determination of
clopidogrel bisulphate in bulk and its dosage forms, Der Pharma Chem. 6 (2014)
366–374 http://derpharmachemica.com/archive.html , Accessed date: 29 January
2018.
[28] J. Sippel, L.L. Sfair, E.E. Schapoval, M. Steppe, New high-performance liquid
chromatographic method for determination of clopidogrel in coated tablets, J.
AOAC Int. 91 (2008) 67–73.
[29] M.K. Javed, Z. Iqbal, A. Khan, A. Khan, Y. Shah, L. Ahmad, Development and va-
lidation of HPLC-UV method for the determination of clopidogrel in pharmaceutical
dosage form and human plasma, J. Liq. Chromatogr. Relat. Technol. 34 (2011)
2118–2129, https://doi.org/10.1080/10826076.2011.585482.
[30] A.L. Saber, M.A. Elmosallamy, A.A. AmIn, H.M.A. Killa, Liquid chromatographic
and potentiometric methods for determinations of clopidogrel, J. Food Drug Anal.
16 (2008) 11–18 http://search.proquest.com/openview/
686fa68d7f1a1117bfbb55e59ee327e7/1?pq-origsite=gscholar&cbl=906352.
[31] F. Belal, A. El-Brashy, M. Eid, J.J. Nasr, Stability-indicating micellar liquid chro-
matographic method for the determination of clopidogrel. Application to tablets
and content uniformity testing, J. Liq. Chromatogr. Relat. Technol. 32 (2009)
2993–3008, https://doi.org/10.1080/10826070903320764.
[32] A. Mitakos, I. Panderi, A validated LC method for the determination of clopidogrel
in pharmaceutical preparations, J. Pharm. Biomed. Anal. 28 (2002) 431–438,
https://doi.org/10.1016/S0731-7085(01)00620-3.
[33] U.C. Mashelkar, S.D. Renapurkar, A LCMS compatible stability-indicating HPLC
assay method for clopidogrel bisulphate, Int. J. ChemTech Res. 2 (2010) 822–829
http://sphinxsai.com/s_v2_n2/CT_V.2No.2/ChemTech_Vol_2No.2_pdf/CT=
10(822-829).pdf , Accessed date: 29 January 2018.
[34] N.A. Alarfaj, Stability-indicating liquid chromatography for determination of clo-
pidogrel bisulfate in tablets: application to content uniformity testing, J. Saudi
Chem. Soc. 16 (2012) 23–30, https://doi.org/10.1016/j.jscs.2010.10.016.
[35] S. Sahu, S.P. Sarangi, H.B. Sahoo, Development and validation methods for the
estimation of Clopidogrel in bulk and pharmaceutical dosage forms, Int. J. Res.
Pharm. Sci. 3 (2012) 224–227 https://www.researchgate.net/profile/Dr_
Himanshu_Sahoo/publication/268187036_Development_and_validation_methods_
for_the_estimation_of_Clopidogrel_in_bulk_and_pharmaceutical_dosage_forms/links/
547824960cf293e2da284645/Development-and-validation-methods-for , Accessed
date: 29 January 2018.
[36] H. Kumar Jain, D.D. Deore, Bioanalytical method development and validation for
57
Journal of Electroanalytical Chemistry 827 (2018) 51–57
estimation of clopidogrel bisulfate in human plasma by RP-HPLC, Int. J. Appl.
Pharm. 8 (2016) 18–21, https://doi.org/10.22159/ijap.2016v8i4.13521.
[37] A.F. Alghamdi, Electrochemical and chromatographic studies of clopidogrel using
cathodic stripping voltammetry and HPLC under new experimental conditions and
its determination in the preparation tablet, urine and plasma samples, J. Chem.
Pharm. Res. 7 (2015) 1023–1030 http://www.jocpr.com/articles/electrochemical-
and-chromatographic-studies-of-clopidogrel-using-cathodic-stripping-voltammetry-
and-hplc-under-new-exper.pdf (accessed January 16, 2018).
[38] N. Bibire, M. Vieriu, G. Tantaru, M. Apostu, L. Agoroaei, A.D. Panainte,
A. Znagovan, A. Vlase, A new and sensitive LC-MS/MS method for the determina-
tion of clopidogrel in human plasma, Rev. Chim. 65 (2014) 807–810 http://www.
revistadechimie.ro , Accessed date: 1 February 2018.
[39] B. Shin, S. Yoo, Determination of clopidogrel in human plasma by liquid chroma-
tography/tandem mass spectrometry: application to a clinical pharmacokinetic
study, Biomed. Chromatogr. 21 (2007) 883–889, https://doi.org/10.1002/bmc.
850.
[40] A. Robinson, J. Hillis, C. Neal, A.C. Leary, The validation of a bioanalytical method
for the determination of clopidogrel in human plasma, J. Chromatogr. B 848 (2007)
344–354, https://doi.org/10.1016/J.JCHROMB.2006.10.076.
[41] L. Silvestro, M.C. Gheorghe, I. Tarcomnicu, S. Savu, S.R. Savu, A. Iordachescu,
C. Dulea, Development and validation of an HPLC-MS/MS method to determine
clopidogrel in human plasma. Use of incurred samples to test back-conversion, J.
Chromatogr. B 878 (2010) 3134–3142, https://doi.org/10.1016/j.jchromb.2010.
09.022.
[42] X. Yang, S. Liu, J. Sun, X. Liu, Y. Sun, Z. He, UPLC for the determination of clo-
pidogrel in dog plasma by tandem quadrupole MS: application to a pharmacokinetic
study, Chromatographia 70 (2009) 259–263, https://doi.org/10.1365/s10337-009-
1109-9.
[43] H. Agrawal, N. Kaul, A. Paradkar, K. Mahadik, Stability indicating HPTLC de-
termination of clopidogrel bisulphate as bulk drug and in pharmaceutical dosage
form, Talanta 61 (2003) 581–589, https://doi.org/10.1016/S0039-9140(03)
00364-3.
[44] H.E. Zaazaa, S.S. Abbas, M. Abdelkawy, M.M. Abdelrahman, Spectrophotometric
and spectrodensitometric determination of clopidogrel bisulfate with kinetic study
of its alkaline degradation, Talanta 78 (2009) 874–884, https://doi.org/10.1016/J.
TALANTA.2008.12.064.
[45] S. Gurav, R. Venkatamahesh, Development and validation of derivative UV-spec-
tropotometric methods for quantitative estimation of clopidogrel in bulk and
pharmaceutical dosage form, Int. J. ChemTech Res. 4 (2012) 497–501 http://www.
sphinxsai.com/2012/chemAJ/CHEM/CT=05[497-501]AJ12.pdf , Accessed date:
29 January 2018.
[46] L. Antypenko, S. Gladysheva, S. Vasyuk, Development and validation of clopidogrel
bisulphate determination in bulk by UV spectrophotometric method, Scr. Sci.
Pharm. 3 (2016) 17–22.
[47] S. Dermiş, E. Aydoǧan, Electrochemical study of the antiplatelet agent clopidogrel
and its determination using differential pulse voltammetry in bulk form and
pharmaceutical preparations with a glassy carbon electrode, Pharmazie 65 (2010)
175–181, https://doi.org/10.1691/ph.2010.9123.
[48] A.R. Mladenović, V.M. Jovanović, S.D. Petrović, D.Ž. Mijin, S.Ž. Drmanić,
M.L. Avramov Ivić, Determination of clopidogrel using square wave voltammetry at
a gold electrode, J. Serb. Chem. Soc. 78 (2013) 2131–2140, https://doi.org/10.
2298/JSC130913116M.
[49] Z.R. Dizavandi, A. Aliakbar, M. Sheykhan, Electrocatalytic determination of clopi-
dogrel using Bi2O3-Pp-AP/GCE by differential pulse voltammetry in pharmaceu-
tical productions, J. Electroanal. Chem. 805 (2017) 24–31, https://doi.org/10.
1016/J.JELECHEM.2017.10.013.
[50] R. Greef, R. Peat, L.M. Peter, D. Pletcher, Instrumental Methods in Electrochemistry,
Ellis Horwood, 1990.
[51] S.A. Ozkan, B. Uslu, From mercury to nanosensors: past, present and the future
perspective of electrochemistry in pharmaceutical and biomedical analysis, J.
Pharm. Biomed. Anal. 130 (2016) 126–140, https://doi.org/10.1016/j.jpba.2016.
05.006.
[52] J. Wang, Analytical Electrochemistry, Third (2006), https://doi.org/10.1002/
0471790303.
[53] C.M.A. Brett, Electrochemistry. Principles, Methods and Applications, Oxford
University Press, 1993.
[54] ICH Expert Working Group, ICH guideline Q1A(R2) stability testing of new drug
substances and products, Int. Conf. Harmon. 24 (2003), https://doi.org/10.1136/
bmj.333.7574.873-a.
[55] D.K. Gosser, Cyclic Voltammetry, VCH, 1994.
[56] S.A. Ozkan, Electroanalytical Methods in Pharmaceutical Analysis and Their
Validation, first ed., HNB Pub, New York, NY, 2012.
[57] S.A. Ozkan, J.-M. Kauffmann, P. Zuman, Electroanalysis in Biomedical and
Pharmaceutical Sciences, first ed., Springer, Berlin Heidelberg, 2015, https://doi.
org/10.1007/978-3-662-47138-8.
Contributing Editor
ABDULLAH A. AL-BADR

Founding Editor
KLAUS FLOREY