Electrochemical behavior of folic acid on mercury meniscus modified silver solid amalgam electrode

Electrochimica Acta 56 (2011) 2411–2419

Contents lists available at ScienceDirect

Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta

Electrochemical behavior of folic acid on mercury meniscus modified silver solid amalgam electrode a ˇ sovská a , T. Navrátil b,∗ , J. Chylková ´ L. Bandˇzuchová a , R. Seleˇ a b

University of Pardubice, Faculty of Chemical Technology, Institute of Environmental and Chemical Engineering, Studentská 573, 532 10 Pardubice, Czech Republic J. Heyrovsk´ y Institute of Physical Chemistry of the AS CR, v.v.i., Dolejˇskova 3, 18223 Prague 8, Czech Republic

a r t i c l e

i n f o

Article history: Received 21 July 2010 Received in revised form 23 October 2010 Accepted 29 October 2010 Available online 19 November 2010 Keywords: Folic acid Vitamins Voltammetry Mercury meniscus modified silver solid amalgam electrode Hanging mercury drop electrode

a b s t r a c t Voltammetric behavior of folic acid and folates has been investigated using direct current voltammetry, differential pulse voltammetry and adsorptive stripping differential pulse voltammetry at a mercury meniscus modified silver solid amalgam electrode (m-AgSAE). The optimum conditions have been found for their determination in a 1:9 mixture of methanol and aqueous acetate buffer, with the limit of detection about 0.5 nmol L−1 . RSD at this concentration level amounted to less than 4%. Practical applicability of the newly developed method was verified by analysis of three vitamin preparations and of two multivitamin juices. © 2010 Elsevier Ltd. All rights reserved.

1. Introduction Folic acid (FA) (CAS: 59-30-3, pteroyl-l-glutamatic acid, folacin, Fig. 1) and its derivates, are very important essential compounds, which belong to the group of vitamins B – vitamin B9 (in some cases is denoted as vitamin M). FA and folates (if not specified otherwise, it is not differentiated between “folic acid” and its salts – folates in this manuscript) are enzymatically transformed in the small intestine into their biologically active and very important form tetrahydrofolate (H4 -folate) [1]. This reaction is catalyzed by enzyme tetrahydrofolate dehydrogenase (dihydrofolate reductase, EC 1.5.1.3) [1,2]. H4 -folate participates in transferring one-carbon groups which are necessary for DNA and RNA biosynthesis, red blood cells formation and methylation processes in organism. H4 -folate also reduces homocysteine to cysteine [3,4]. Sufficient dietary intake of FA prevents neural tube defects (e.g., anencephaly or spina bifida) and certain types of anemia [1,5]. The protective effect of FA was also observed for treatment of some diseases like stroke, ischaemic heart disease [6] or colorectal cancer [7]. The recommended daily intake of FA is 0.2 mg (0.453 ␮mol) for adults and 0.4 mg (0.906 ␮mol) for pregnant women [8]. Many methods for the FA determination have been developed due to its biological significance. These methods include ELISA

∗ Corresponding author. Fax: +420 286 582 307. E-mail address: [email protected] (T. Navrátil). 0013-4686/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2010.10.090

(enzyme-linked immunosorbent assays) [9], HPLC (high performance liquid chromatography) with UV [10,11] or diode array detection [12], liquid chromatography with tandem mass spectroscopy (LC/MS/MS) [13] or with electrospray ionisation mass spectrometry [14], microemulsion electrokinetic chromatography [15], capillary electrophoresis [16], spectrophotometry after coupling reaction with specific compounds [17] or biosensor-based determination [18]. Many electrochemical techniques have been developed, because FA is electrochemically active compound. Electrochemical behavior of FA was studied at first at mercury electrodes. Jocobsen and Bjørnsen applied a.c. polarography with DME (dropping mercury electrode) as a working electrode for the determination of FA in pharmaceutical preparations. FA provided a well defined wave in medium of acetate buffer (pH 5.5) at the DME [19]. Electrochemical behavior of FA using AdSV (adsorptive stripping voltammetry) at hanging mercury drop electrode (HMDE) as working electrode was also examined [20–22]. The limit of detection was 1 × 10−11 mol L−1 after 10 min accumulation [20]. Villamil et al. successfully used AdSV at HMDE for the determination of FA and riboflavin in vitamins preparations [23]. Cathodic stripping voltammetry in conjunction with HMDE was successfully applied for the determination of FA in sea water [24,25]. Besides HMDE, modified electrodes or carbon nanotube electrodes have been used recently. Voltammetric behavior of FA at glassy carbon electrode modified with phosphomolybdic-polypyrrole film (PMo12 -PPy/GC) is discussed in the paper [26]. This electrode

2412

L. Bandˇzuchová et al. / Electrochimica Acta 56 (2011) 2411–2419

OH

N

H N

O

OH

NH

O

NH 2 N

N O

HN O Fig. 1. Structure of folic acid (FA).

showed good sensitivity and stability for the determination of FA. The detection limit of 1.0 × 10−10 mol L−1 of FA was estimated [26]. Vaze et al. [27] studied possibilities of carbon paste electrode modified with p-tert-butyl-calix[n]arenes (where n = 4, 6 or 8) for the determination of FA in a variety of samples (serum, asparagus, spinach, oranges and multivitamin preparations). Carbon paste electrode modified with p-tert-butyl-calix[6]arene (CME-6) was found to be more sensitive than other two tested modifications. The detection limit of 1.24 × 10−12 mol L−1 was obtained using adsorptive stripping voltammetry at calixarene based chemically modified electrodes [27]. 2-Mercaptobenzothiazole self-assembled gold electrode (MBT/SAM/Au) was also used for voltammetric monitoring behavior of FA and its determination in pharmaceuticals. The detection limit of 4.0 × 10−9 mol L−1 was estimated using CV (cycling voltammetry) [28]. A single-walled carbon nanotube (SWNT) paste coated glassy carbon electrode (GCE), which had been prepared by using room temperature ionic liquid (1-octyl-3methylimidazolium hexafluorophosphate, OMIMPF6 ) as a binder was also used for studying electrochemical behavior of FA and its determination in real samples (wheat flour, fruit juice and milk). The detection limit of 1.0 × 10−9 mol L−1 was estimated [29]. The above mentioned solid modified electrodes or carbon nanotube electrodes exhibit high sensitivity of FA determination, but their construction and preparation for measurement are often rather complicated, the prices of modifying substances are usually higher (e.g., calix[n]arenes, SWNT) and their surface stabilities and reproducibilities are relatively lower in comparison with amalgam electrodes, which applicability for this purpose is described in this manuscript. More precisely, the utilization of m-AgSAE (mercury meniscus modified silver solid amalgam electrode) for FA determination is discussed in this paper. Voltammetry and amperometry belong among the most frequently used electroanalytical methods. They are especially suitable for large scale environmental monitoring of electrochemically active pollutants in various types of matrices, because they are inexpensive, extremely sensitive, suitable for speciation and they present an independent alternative to so far prevalent spectrometric and separation techniques [30]. Nevertheless at present, the number of used polarographic, voltammetric and similar devices, utilized in commercial laboratories, is relatively lower in comparison with spectral and other analytical devices (atomic absorption spectroscopy, ICP-MS etc.). The crucial and fundamental problem in their practical application is the choice of a suitable working electrode [31]. The main reason consists in traditional electrode material, i.e., mercury. Strict ecological and safety rules introduced in the world as well as popular prejudices, fears and faults essentially complicate the use of mercury or liquid mercury containing electrodes (including hanging mercury drop electrode (HMDE)). The research is focused on the material, which could successfully and adequately replace the liquid mercury, however, which would be non-toxic, friendly toward the environment and thus compatible with the concept of so called “green analytical chemistry”. In

consequence of this situation, attention has been recently devoted to the development of solid or paste electrodes (carbon paste electrodes [32], solid composite electrodes [33–37], and many others). Several years ago new types of electrodes based on amalgamation of soft metal powder (MeSAE – metal solid amalgam electrodes) were designed by Prague research group [38]. At the same time, the Trondheim research group developed a solid dental amalgam electrode, which was prepared simply by placing the dental amalgam paste in a cavity of the electrode holder [39]. So far, this group focuses on the analysis of heavy metals [39,40]. The electrodes prepared by the Prague group feature a variety of metals used for amalgam preparation (e.g., silver, copper, gold, and thallium). They can be used either as mercury-free electrodes after polishing of solid amalgam disc (p-MeSAE) or after modification of their surfaces by mercury film (MF-MeSAE) or mercury meniscus (mMeSAE). The meniscus covering the m-AgSAE is stable; pH value does not affect its dissolution inside the working potential window. The results, achieved using these solid amalgam electrodes are fully comparable with those obtained using HMDE or DME [41–47]. Although the solid amalgam electrodes do not reach the quality of HMDE, in many cases they approach it, as it was proved by a variety of analytical applications, including the voltammetric determination of heavy metal cations and many inorganic anions as well as organic species (e.g., [48]). The value of the hydrogen overvoltage at amalgam electrodes is as high as at liquid mercury. Therefore, the field of their application is similar to this of HMDE or DME (e.g., [31,41–46,48–55]). Such universality and easiness of their surface renovation and regeneration belong to advantages of amalgam electrodes. The reaction mechanism of FA at mercury electrode has been studied in detail (e.g., [21,22,56]). All available sources conclude that the reduction process consists of three distinct steps in acidic media and only a single reduction step in alkaline media [56]. In acidic medium the reduction process can be described as follows: the molecule of FA is converted to a transient 5,8-dihydro-FA, some of which tautomerizes to 7,8-dihydro-FA. The remainder undergoes a proton dependent non-electrochemical cleavage of the para-aminobenzoylglutamate side chain. The second step involves electrochemical cleavage of the C(9)–N(10) chemical bond of the 7,8-dihydro-FA while the final reduction converts the 6-methyl7,8-dihydropterin derivative generated in the second step to a 6-methyl-5,6,7,8-tetrahydropterin. In alkaline solutions, a single 2e− /2H+ reduction step is observed only, because the proton dependent tautomerization process is slow [56]. On the other hand, in anodic way of polarization only one oxidation step was observed in acidic medium as well as in alkaline media. It was concluded that this peak corresponds to the oxidation of dihydro-FA, originated in the first reduction step [56].

2. Experimental 2.1. Materials All chemicals used for preparation of the supporting electrolyte, standard solutions and other stock solutions were of p.a. purity. Britton–Robinson (B–R) buffer of pH value from 3 to 12 was prepared from alkaline component 0.2 mol L−1 NaOH (Lachema, Brno, Czech Republic) and acidic component consisting of 0.04 mol L−1 H3 PO4 , 0.04 mol L−1 H3 BO3 and 0.04 mol L−1 CH3 COOH (Lachema, Brno, Czech Republic). 0.5 mol L−1 acetate buffer of pH 5 was prepared by mixing the stock solutions of acetic acid and sodium acetate (Lachema, Brno, Czech Republic). 0.01 mol L−1 borate buffer was prepared by mixing the stock solutions of boric acid and sodium tetraborate (borax) (Lachema, Brno, Czech Republic). Methanol was purchased from the Penta Com-

L. Bandˇzuchová et al. / Electrochimica Acta 56 (2011) 2411–2419

pany, Chrudim, Czech Republic. “Folic acid” (Sigma–Aldrich) was dissolved in 0.01 mol L−1 solution of NaOH (Lachema, Brno, Czech Republic), and was then stored in the dark. Its measured solutions diluted from its 0.01 mol L−1 stock solution were prepared daily. All solutions were prepared in bidistilled water. It is well known that FA is practically insoluble in acidic medium. However, the solubility dramatically increases with increasing pH. The pH-values used in this paper (5–8) are very close to the physiological pH values in the human body. On the basis of published results and data (e.g., [21,22]), it is possible to suppose that FA is completely dissolved in such very low concentrations, which are investigated in this manuscript, at mentioned pH values. GS Mamavit tablets originated from Green-Swan Pharmaceuticals, Czech Republic, and tablets “Folic Acid Super” and “Folic Acid Forte” were purchased from Naturvita, Czech Republic, “TOMA multivitamin juice” from General Bottlers CR, Czech Republic and Hello Viva Juice from Linea Nivnice, Czech Republic. 2.2. Instrumentation Voltammetric measurements were performed with the computer controlled Eco-Tribo Polarograph PC-ETP (Polaro-Sensors, Prague, Czech Republic), equipped by POLAR.PRO software for Windows XP version 5.1. Hanging mercury drop electrode (HMDE) and mercury meniscus modified silver solid amalgam electrode (mAgSAE) (Polaro-Sensors, Prague, Czech Republic) served as working electrodes, Ag/AgCl/saturated KCl as reference and platinum wire as auxiliary electrode (both from Monokrystaly, Turnov, Czech Republic). The measurements were performed at laboratory temperature (23 ± 2 ◦ C). Oxygen was removed from the measured solutions by bubbling with nitrogen (purity class 4.0; Linde, Prague, Czech Republic) for 5 min, and voltammograms at m-AgSAE or HMDE were recorded. The values of pH were measured using pH-meter Hanna 221 and solutions of FA tablets were prepared applying ultrasonic bath Bandelin Sonorex (both from Fisher Scientific, Pardubice, Czech Republic). 2.3. Procedures 2.3.1. Preparation and pretreatment of the silver solid amalgam electrode Silver solid amalgam electrode (AgSAE), prepared as described, e.g., in [38,41,46], was firstly ground on a soft emery paper and then polished using polishing kit (Electrochemical Detectors, Turnov, Czech Republic) consisting of polishing polyurethane stock, Al2 O3 suspension (particle size 1.1 ␮m), and soft polishing Al2 O3 powder (particle size 0.3 ␮m). The preparation process can be stopped by this step and such electrode is called polished AgSAE (p-AgSAE). Mercury meniscus at the surface was prepared by immersion of the p-AgSAE into liquid mercury for 15 s (while swirling the flask with mercury) and this way prepared electrode is called m-AgSAE. The meniscus was renewed once a week by re-dipping the electrode in liquid mercury. The electrochemical regeneration of amalgam electrodes is generally realized in two different levels: activation and regeneration by cycling. There are some differences between these two operations: (1) the activation is realized outside of the working potential window of the electrode. On the other hand, the cycling is realized inside this window. (2) The activation is performed before starting measurements with the amalgam electrode, as well as after every pause longer than 1 h. The cycling is realized before each measurement usually. Generally, the “activation” is much more robust and effective. It is performed at the relatively negative potential (more than 2000 mV). At such potential in KCl solution, the compounds dissolved in amalgam are reduced and released, and the adsorbed compounds are desorbed from its surface. Simultane-

2413

ously, hydrogen from water is reduced at the electrode surface, its small bubbles are released, and thereby the undesired substances adsorbed at the amalgam surface are removed. Potassium cations from KCl solution are partly reduced and potassium amalgam is formed, which process helps to purify the amalgam from undesired compounds. In described experiments, the electrode surface was activated in the solution of 0.2 mol L−1 KCl by applying the potential of −2200 mV for 300 s while stirring the solution (before starting of the measurements with the amalgam electrode, as well as after every pause longer than 1 h) (as it was recommended in [38,41,46,57]). The regeneration process, which was realized before each measurement, was composed of 30 polarizing cycles between two regeneration potentials (E1,reg = 0 mV and E2,reg = −1200 mV, regeneration time (treg ) of application of each of them 300 ms). The amalgam electrode used exhibited the disc diameter 0.54 mm and the meniscus surface 0.46 mm2 . For comparison, the surface of the mercury drop formed at the end of the glass capillary (HMDE) (with inner diameter about 40 ␮m) was about 1.0 mm2 . It follows therefrom that the active working surface of the m-AgSAE amounted to about one half of the HMDE surface and therefore the signal registered at the m-AgSAE can be lower. All current values in the manuscript were recalculated to the current densities.

2.3.2. Voltammetric measurements Direct current voltammetry (DCV) was used for the first set of studies of voltammetric behavior of FA at mercury meniscus modified silver solid amalgam electrode and at hanging mercury drop electrode (dependences on pH and on scan rate). Differential pulse voltammetry (DPV), with pulse duration of 100 ms, sampling time of 20 ms beginning 80 ms after the onset of the pulse and interval between pulses of 100 ms, pulse height −50 mV for cathodic scans and with the scan rate 20 mV s−1 , was used for the determination of FA using both tested working electrodes. An accumulation step was inserted into the measuring sequence for an increase of sensitivity by adsorptive DPV (AdDPV). Differential pulse voltammetry with HMDE, as well as with m-AgSAE, was used for most of the measurements with the following optimized parameters: initial potential Ein = −200 mV, final potential Efin = −1000 mV, potential of accumulation Eacc = −200 mV and time of accumulation (tacc ) corresponding to the FA concentration. B–R buffer with pH 8 was used as a supporting electrolyte for measurements with HMDE and 0.05 mol L−1 acetate buffer (pH 5) with the addition of 10% of methanol served as a supporting electrolyte for measurements with m-AgSAE. The role of methanol in described experiments has not been elucidated yet. Probably, it improves the solubility of FA. It is known that methanol is highly toxic. On the other hand, the used solvent must be mixable with water and its adsorbability must be as low as possible. Methanol fulfills both of them. Less toxic ethanol is miscible with water too, however, its adsorbability is higher, and therefore, the detection limit would be affected. Prior to measurements purified nitrogen had been passed through the solution for a period of 300 s and then nitrogen atmosphere was maintained above the solution in the cell. All registered curves were measured 3 times. The heights of DCV waves (Ip ) were evaluated from the extrapolated linear portion of the voltammogram before the onset of the peak. DPV and AdSDPV peaks were evaluated from the straight line connecting the minima before and after the peak (tangent to the curve joining beginning and end of the given peak). The limits of decision (CC ), quantification (LD ) and detections (LQ ) were calculated as described in [58,59] using the software QC Expert v. 2.5 (Trilobyte, Czech Republic). The parameters of calibration curves (e.g., slope, intercept) were obtained using software Excel 2003 (Microsoft, USA).

2414

L. Bandˇzuchová et al. / Electrochimica Acta 56 (2011) 2411–2419

2.3.3. Elimination voltammetry with linear scan This method is often used for elucidation of reaction processes on electrode surfaces and for increase of current sensitivity. Elimination voltammetry with linear scan (EVLS) is based on elimination of some particular currents from linear voltammetric curves recorded at various scan rates [60–62]. In elimination voltammetry it is assumed that the measured current is the sum of individual current contributions, such as the charging current (Ic ), the diffusion controlled current (Id ), the kinetic current (Ik ) and the s.c. irreversible current (Iir ), and that each particular current must be expressible in form (1): Ij = Wj (v) · Yj (E) = vx Yj (E)

(1)

where Ij is a particular current (the number of all particular currents is n and 1 ≤ j ≤ n), is the scan rate, Wj (v) is a function of the scan rate v and Yj (E) is a function of the potential E. On the basis of these postulates s.c. elimination coefficients for elimination equations were calculated and only one current was conserved and the others were eliminated [61]. The product of charge transfer coefficient and number of electrons transferred in reaction can be calculated from the difference between peak potential recorded on chosen elimination curve and peak potential registered on original curve [60]. The average value of this product was calculated from the results obtained from 4 or 5 different elimination equations. 2.3.4. Preparation of real sample solution One tablet was powdered in a grinding mortar, transferred into a 100 mL standard flask, and dissolved in 0.01 mol L−1 NaOH. After sonication it was filtered. After that, a suitable aliquot of the clear filtrate was diluted with proper electrolyte to prepare appropriate sample solution. Voltammograms were recorded as in standard FA solution. Similarly, the sufficient amount of the analyzed juice was only filtered and the appropriate volume of filtrate was added to the supporting electrolyte and then analyzed. To eliminate the interferences of surface active substances and other compounds likely to be present in analyzed material, the standard addition method was used for the analysis of FA in real samples.

The peak potentials, corresponding to the reduction processes, were shifted approximately linearly with increasing pH-values with both types of electrodes (the linear dependence of the most negatively situated peak is party disturbed by decomposition of the supporting electrolytes in pH below 3). The statistical parameters of these dependences are summarized in Table 1. The slopes of peak position dependences on pH are equal in case of peak 1 recorded with both electrodes (p < 0.05). Much more complicated are the dependences of peak currents on pH with both electrodes, which differs mutually for all peaks. They were not monotonous in wider pH ranges. It follows from the dependences depicted in Fig. 3A and B that the only peak, which can be recorded in the wide range of pH (from 3 to 11 at HMDE and from 3 to 8 at m-AgSAE) is the peak corresponding to the first twoelectron reduction. The most intensive peak (denoted as 1 in Fig. 2A and C) was recorded at pH about 8 at HMDE and at pH about 5 at m-AgSAE. By testing different supporting electrolytes it was found that the acetate buffer is much more suitable for the determination of FA with m-AgSAE, more than the B–R buffer, which can be used for the measurements with HMDE. The peaks, denoted as 2 in Fig. 2A and C decrease with increasing pH with both m-AgSAE and HMDE. The most negative peak at HMDE (denoted as 3 in Fig. 2C) decreases with increasing pH exponentially (inset in Fig. 3A). The peak corresponding to the same process cannot be recorded at pH values higher than 5 with m-AgSAE. This peak cannot be used for analytical purposes due to its overlapping with supporting electrolyte decomposition.

3. Results and discussion

3.1.2. Dependence on the scan rate and EVLS From the analytical point of view, the most interesting is the most positively situated peak. Therefore, this peak was investigated in more detail under optimal pH values (i.e., pH 5 in acetate buffer with m-AgSAE and pH 8 in B–R buffer with HMDE). For elucidation of the reaction processes, the dependences of the peak heights on the scan rates were investigated. The recorded data were evaluated using EVLS. The dependence of the peak height on the scan rate was linear in a wide range (from 10 to 170 mV s−1 ) with m-AgSAE in acetate buffer, pH 5, and could be described by Eq. (2) (p = 0.05).

3.1. Voltammetric behavior of FA

Jp [nA mm−2 ] = (−0.1583 ± 0.0048)  [mV s−1 ] − (5.93 ± 0.48);

3.1.1. Dependence on pH Three well separated peaks were recorded with HMDE on the cathodic curve (Fig. 2A, peaks 1, 2, 3) and one peak on the anodic curve (Fig. 2A, peak 1’) of the voltammograms in acidic medium. It corresponded to the above mentioned theory that the reduction process at the mercury surface is realized in three subsequent steps in the solution of given composition and of given pH. The curves recorded at m-AgSAE were similar, i.e., three reduction peaks, but only one oxidation peak in acidic medium (Fig. 2C). The reason, why these peaks were less intensive, could probably consist in the smaller surface area of the m-AgSAE in comparison with HMDE (the ratio 1:2). Furthermore, the potential window available at mAgSAE is a bit narrower than at HMDE [48]. It is the reason, why the most negatively situated peak was partly overlapped by the signal of supporting electrolyte decomposition. Similarly, in correspondence with the theory, one oxidation peak was recorded in acidic medium at this electrode, which corresponded to the oxidation of dihydro-FA to FA. In alkaline solutions, only one reduction and one oxidation peak were recorded at HMDE (Fig. 2B), which corresponded to the fist reduction step, from FA to dihydro-FA. On the contrary, no peak was observed in such solutions with m-AgSAE (Fig. 2D) (FA concentration 1 × 10−5 mol L−1 in B–R buffer solution).

r = −0.9985

(2)

However, the same dependence recorded at HMDE in B–R pH 8 proved to be linear from 10 to 250 mV s−1 and could be described by Eq. (3) (p = 0.05). Jp [nA mm−2 ] = (−0.2375 ± 0.0031)  [mV s−1 ] − (7.64 ± 0.46); r = −0.9995

(3)

The linear forms of these dependences can bring us to the conclusion that the processes recorded with both electrodes are controlled by adsorption. However, the controlling process has been affected by pH of the solution and by the supporting electrolyte with the m-AgSAE. While only very small, practically negligible signal could be recorded in B–R buffer of pH 8, in borate buffer an evaluable was peak recorded at m-AgSAE. However, the height of the peak was linearly dependent on the square root of the scan rate at pH 8 in the above mentioned interval. Therefore it could be concluded that the diffusion controls the transport of FA to the electrode surface at this pH value. More detail information was obtained by the EVLS (Table 2). By application of this method, the information achieved using “more traditional” procedures was confirmed. According to the EVLS, the most positively situated peaks (denoted 1 and 1’) correspond to

L. Bandˇzuchová et al. / Electrochimica Acta 56 (2011) 2411–2419

2415

Fig. 2. Cyclic voltammograms of FA, scan rate 100 mV s−1 : (A) 5 × 10−6 mol L−1 FA in 0.05 mol L−1 B–R buffer (pH 5) at HMDE; (B) 5 × 10−6 mol L−1 FA in 0.01 mol L−1 B–R buffer (pH 8) at HMDE; (C) 1 × 10−5 mol L−1 FA in 0.05 mol L−1 B–R buffer (pH 5) at m-AgSAE, (D) 1 × 10−5 mol L−1 FA in 0.01 mol L−1 B–R buffer (pH 8) at m-AgSAE. The arrows indicate the directions of the potential scans. Table 1 Parameters of linear dependences of peak potentials on pH values of analyzed solutions (p = 0.05).

Peak 1 Peak 2 Peak 3

m-AgSAE Slope [mV/pH unit]

m-AgSAE intercept [mV]

Correl. coefficient

HMDE Slope [mV/pH unit]

−60 ± 19 −78 ± 25 −64.0 ± 4.4

−40.2 ± 3.3 −340 ± 130 −648 ± 21

−0.999 −0.972 −0.999

−60.8 ± 1.3 −92.7 ± 5.3 −33.7 ± 9.9

the reduction in adsorbed state by cathodic polarization of the electrode and to oxidation in weakly adsorbed state by reverse polarization, respectively. As it was proved, the height of the peak 1 can be increased by accumulation of the analyte at the electrode surface. The other two peaks (denoted 2 and 3) corresponded to the reduction processes in weakly adsorbed state, too. If the pH values were close to the optimal values stated above, i.e., to pH 5 with m-AgSAE or to pH 8 with HMDE, the influence of the adsorption was more pronounced or even the reduction processes were realized in fully adsorbed states. The effect of adsorption was almost negligible in case of all three reduction and one oxidation peaks in solution prepared from borate buffer of pH 8

HMDE intercept [mV] −159 ± 10 −354 ± 27 −959 ± 59

Correl. coefficient −0.999 −0.999 −0.987

with m-AgSAE, where the registered process was controlled by diffusion. From the differences of the peak potentials recorded in different elimination voltammograms and in original voltammogram, the product ˛n could be evaluated for the most positively situated peak (where ˛ denotes the charge transfer coefficient and n number of exchanged electrons per molecule of reactant involved in the rate-determining step) [60]. This product amounted to 0.83 on average (from 0.71 to 0.92 for four different elimination equations) for the most positively situated peak. Therefrom it could be concluded that the number of exchanged electrons is equal to 2 and the charge transfer coefficient to 0.41. The supposed reaction mech-

Fig. 3. Dependences of peak current densities of reduction peaks of FA on pH, scan rate 100 mV s−1 , supporting electrolyte: 0.05 mol L−1 B–R buffer of given pH (3–11), Ein = 0 mV, Efin = −1000 mV. (A) 5 × 10−6 mol L−1 FA at HMDE; (B) 1 × 10−5 mol L−1 FA at m-AgSAE.

2416

L. Bandˇzuchová et al. / Electrochimica Acta 56 (2011) 2411–2419

Table 2 The controlling processes of the reduction and oxidation respectively, of FA and their products with m-AgSAE and HMDE in solution of different pH values (diff. contr. = diffusion controlled process). Buffer, electrode

Acetate, pH5 m-AgSAE Borate, pH8 m-AgSAE Acetate, pH5 HMDE Borate, pH8 HMDE

Direction

Cathodic Anodic Cathodic Anodic Cathodic Anodic Cathodic Anodic

Controlling process Peak 1

Peak 2

Peak 3

Reduction in weakly adsorbed state Oxidation in weakly adsorbed state Diffusion controlled reduction Diffusion controlled oxidation Reduction in adsorbed state Oxidation in adsorbed state Reduction in adsorbed state Oxidation in adsorbed state

Reduction in weakly adsorbed state – Diffusion controlled reduction – Reduction in adsorbed state – Reduction in adsorbed state –

Reduction in very weakly adsorbed state – Diffusion controlled reduction – Reduction in adsorbed state – Reduction in adsorbed state –

anism described in the literature (e.g., [56]) could be confirmed. For comparison, the number of transferred electrons was calculated from the difference of the peak potentials of peaks recorded in cathodic and in anodic directions too. The difference recorded at HMDE at pH 5 amounted to 27.0 mV and at pH 8 to 29.4 mV, respectively. From these data was calculated the number of transferred electrons, which amounted to 2.19 at pH 5 and 2.01 at pH 8. Consequently, the transfer of 2 electrons was confirmed at both pH-values. Because the surface of m-AgSAE is formed by mercury meniscus, i.e., of similar material as in case of HMDE, it can be supposed that the number of exchanged electrons at m-AgSAE is equal to 2 too. The difference of voltammetric peak potentials at m-AgSAE was higher (about 140 mV) as at HMDE and because this difference increased with increasing scan rate, it can be concluded that quasireversibility of the reaction was confirmed. The surface of the m-AgSAE, studied in this paper, is formed by a mercury meniscus, which contains mercury. Therefore it can be supposed that the reaction mechanism would be similar to those observed with HMDE or with DME. During the time, the solid amalgam is solved in the liquid mercury and the meniscus solidifies, and these processes can affect the rates of observed reactions (i.e., controlling processes). Nevertheless, the silver present in the meniscus does not influence the reaction mechanism. This finding is in good agreement with other measurements, realized with this type of electrode [63].

(4)) (r = 0.999, p = 0.05): Jp [nA mm−2 ] = (−0.0618 ± 0.0017) · tacc [s] + (−0.545 ± 0.052) (4) In wider interval of accumulation times (2–180 s), the current signal increases non-linearly and the recorded dependence can be adequately approximated by an adsorption isotherm (Eq. (5)) (r = 0.993, p = 0.05): Jp [nA mm−2 ] =

(5) Similar dependences were obtained in the concentration of FA 1 nmol L−1 in B–R buffer, pH 8 with HMDE, i.e., its part from 20 to 60 s could be approximated by a line (r = 0.999, p = 0.05) (Eq. (6)): Jp [nA mm−2 ] = (−0.07096 ± 0.00092) · tacc [s] + (−0.23 ± 0.15) (6) and in the interval from 20 to 180 s by an isotherm (r = 0.979, p = 0.05) (Eq. (7)): Jp [nA mm−2 ] =

3.1.3. Optimization of the determination parameters Due to the fact that the most positively situated reduction peak can be registered in a wide range of pH, it exhibits relatively high symmetry, its height can be increased by accumulation at the electrode surface, etc., only this peak was studied in detail with the aim of its analytical application. The parameters for DPV determination of FA (initial potentials (Ein ), accumulation potentials (Eacc ) and accumulation time (tacc ) were optimized with both electrodes and compared mutually. It was found that neither initial potential nor accumulation potential influences the peak height substantially. The current signals were almost constant (±1.8%) in the initial potential interval (from 0 to −350 mV) in acetate buffer, pH 5 at m-AgSAE. On the other hand, this signal was slightly decreasing (about 15%) in the initial potential interval (from 0 to −500 mV) in B–R, pH 8 with HMDE. Similarly, almost no influence of the accumulation potential on the peak current was registered at both electrodes (from 0 to −400 mV). On the basis of these experiments, the optimal parameters were obtained, which were applied in the following measurements Ein = Eacc = −200 mV. The dependence of the peak height on the accumulation time was linear for m-AgSAE in interval from 2 to 60 s at concentration level of FA 10 nmol L−1 in acetate buffer, pH 5 with m-AgSAE (Eq.

(−11.50 ± 0.35) · (0.00920 ± 0.00048) · tacc [s] 1 + (0.00920 ± 0.00048) · tacc [s]

(−2.28 ± 0.71) · (0.035 ± 0.022) · tacc [s] 1 + (0.035 ± 0.022) · tacc [s]

(7)

These results correspond to those achieved using EVLS, which suggested influence of adsorption on electrode reaction. Therefore, AdDPV can be used to increase the sensitivity. To ensure the ideal reproducibility of the registered current signals with m-AgSAE, it was necessary to optimize the parameters of the regeneration process. Such procedure consists of application of either constant potential for given time or of application of a few tens of polarizing cycles, representing the switching of the electrode potential from E1,reg to E2,reg for tens or hundreds of ms. E1,reg is usually about 50–100 mV more negative than the potential of the anodic dissolution of the electrode material, E2,reg is about 50–100 mV more positive than the potential of the hydrogen evolution in the given supporting electrolyte. The way of their finding was similar as it was described e.g., in [48,53,57]. The optimized regeneration process included the application of 30 polarizing cycles, where E1,reg amounted to 0 mV and E2,reg amounted to −1200 mV, regeneration time of application of each of theses regeneration potentials was 300 ms. Therefore, one regeneration process takes about 20 s only. The appropriate values of the potential and the time of regeneration were inset and modified in the controlling software of the used computer-controlled instrument and regeneration of m-AgSAE could thus be carried out automatically. This regeneration procedure was realized directly in the analyzed solutions.

L. Bandˇzuchová et al. / Electrochimica Acta 56 (2011) 2411–2419

2417

Table 3 Comparison of added and determined levels of FA and RSDs on four different concentration levels achieved with m-AgSAE and with HMDE. m-AgSAE

HMDE

Added [nmol L−1 ]

Determined [nmol L−1 ]

50 20 10 5

50.71 20.50 10.29 5.130

± ± ± ±

1.4 0.60 0.31 0.20

RSD [%]

Added [nmol L−1 ]

Determined [nmol L−1 ]

2.2 2.3 2.4 3.1

5.00 2.00 1.00 0.50

5.030 2.069 1.030 0.505

± ± ± ±

0.22 0.036 0.035 0.017

RSD [%] 0.5 1.4 2.7 3.4

The achieved results are very similar and the achieved statistical parameters are summarized in Table 4. They are about ten times higher with m-AgSAE than with HMDE (tacc = 20 s). Fig. 4. DP voltammograms of FA in 0.05 mol L−1 acetate buffer with methanol (9:1), pH 5, working electrode: m-AgSAE, Ein = Eacc = −200 mV, tacc = 15 s, E1,reg = 0 mV, t1,reg = 0.3 s, E2,reg = −1200 mV, t2,reg = 0.3 s, v = 20 mV s−1 , c(FA) [nmol L−1 ]: (1) 0; (2) 5; (3) 10; (4) 15; (5) 20; (6) 25. Inset: the calibration straight-line.

Nevertheless, the decrease of the current signal sensitivity with increasing number of repeated analysis was not improved by the insertion of the above described procedure (probably due to nonidealities in surface cleaning), and it was necessary to perform the activation procedure more frequently than usual. This situation was significantly improved by addition of the small amount of methanol to the supporting electrolyte. It was experimentally proved that the presence of this compound does not affect the recorded signal until 10% of the total volume. The higher amount causes exponential decrease of the recorded signal (e.g., by 20% concentration of methanol the peak signal was decreased to 60% of original height). The peak position was not significantly influenced by methanol addition (to about 15 mV). Therefore, 9:1 seems to be the optimal ratio of acetate buffer to methanol, which was applied in determinations with both tested electrodes (m-AgSAE and HMDE, respectively). In Fig. 4, there is depicted the concentration dependence of FA from 0 to 25 nmol L−1 in 0.05 mol L−1 acetate buffer with methanol (9:1), pH 5, recorded with m-AgSAE, under the proposed experimental conditions. 3.2. Analysis of FA in different solutions 3.2.1. Analysis of FA in model solutions Under application of the above mentioned optimized experimental and regeneration parameters the achieved relative standard deviations (RSD) of the voltammetric signals of FA with m-AgSAE were fully comparable with those achieved with HMDE (where the surface regeneration was realized by removing of the old and formation of the new drop). The RSD of the 11times repeated measurements with the same m-AgSAE surface in the same analyzed solution (FA concentration 5 × 10−8 mol L−1 ) amounted to 1.2% (RSD achieved in the same solution with HMDE amounted to 0.7%). The repeatability of the determinations was tested by 5 times repeated determination of FA at four different concentration levels (Table 3). The added levels of FA were included in confidence intervals of determined results in model solutions (p = 0.05) in all cases. It is possible to conclude that the RSDs are almost equal for both electrodes (e.g., at concentration level 5 nmol L−1 the achieved RSDm-AgSAE amounted to 3.1% and the RSDHMDE 0.5 %) and did not exceed 3.5% of the value. We utilized four different methods (Direct method of signal (IUPAC), combined method (Ebel, Kamm), Method 3*Sigma (from regression), Method 3*Sigma (ACS)) of evaluation of the statistical limits of FA determination in aqueous model solutions [58,59,64].

3.2.2. Analysis of FA in real samples The applicability of the mercury meniscus modified silver solid amalgam electrode for differential pulse voltammetric determination of FA was verified by analysis of real samples of 3 different nutrition supplements: “GS Mamavit” tablets (declared content of FA 400 ␮g/tablet (=0.906 ␮mol/tablet)), “Folic Acid Super” (declared content of folic acid 400 ␮g/tablet (=0.906 ␮mol/tablet)) and “Folic Acid Forte” (declared content of folic acid 200 ␮g/tablet (=0.451 ␮mol/tablet)) (Table 5). The analyzed solutions were prepared as it was described above (in Section 2.3.4). Since the DPV technique is less robust and more prone to interferences from surface active substances and other compounds likely to be present in analyzed material, the standard addition method was used for the analysis of all substances. The results are summarized in Table 5. In cases of all three supplements it was statistically verified (p < 0.05) that the results achieved with m-AgSAE were equal to those achieved with HMDE and were in fair agreement with the content of the folic acid declared by producer (Table 5, Fig. 5). The other components (vitamins, trace elements, filling materials, etc.), present in the analyzed tablets, did not disturb the determination. RSDs, calculated from five times repeated analysis of every vitamin supplement, did not exceed 4% with any electrode. Analogically, two different multivitamin juices were analyzed (each analysis was 5 times repeated): “Hello Viva Juice” and “TOMA multivitamin juice”. The declared level of folic acid in these drinks was higher than 400 ␮g L−1 (it corresponds to 0.906 ␮mol L−1 ). The samples of juices were filtered on paper filters and the appropriate amount of the filtrated juice (50–200 ␮L) was added to 10 mL of supporting electrolyte. In the first step the applicability of the above described method was tested by addition of known amount of FA to juice samples (i.e., the electrolyte with juice sample was taken

Fig. 5. Determination of FA in three different nutrition supplements and two different samples of juices with m-AgSAE and with HMDE. The error-bars correspond to the calculated confidence bars on the level p = 0.05. In case of both juices the level declared by producer represents the lowest guaranteed content of FA.

2418

L. Bandˇzuchová et al. / Electrochimica Acta 56 (2011) 2411–2419

Table 4 Statistical parameters obtained with m-AgSAE and with HMDE under optimized conditions. Method of calculation

m-AgSAE

HMDE −1

CC [nmol L Direct method of signal (IUPAC) Combined method (Ebel, Kamm) Method 3*Sigma (from regression) Method 3*Sigma (ACS) Slope of the calibration line [nA L nmol−1 ] Correlation coefficient

]

0.26 0.25 0.26 0.10 0.0719 ± 0.0015 0.9998

−1

LD [nmol L

]

0.51 0.51 0.52 0.21

−1

LQ [nmol L

]

0.75 0.76 0.75 0.78

CC [nmol L−1 ]

LD [nmol L−1 ]

LQ

0.019 0.019 0.018 0.019 0.00514 ± 0.00011 0.9999

0.036 0.036 0.036 0.036

0.052 0.053 0.052 0.054

Table 5 Content of folic acid in 3 different nutrition supplements and two different juices (p = 0.05). Declared [␮mol/tablet]

Founda – m-AgSAE [␮mol/tablet]

RSD [%]

Founda – HMDE [␮mol/tablet]

RSD [%]

GS Mamavit Folic acid Super Folic Acid Forte

0.906 0.906 0.453 [␮mol L−1 ]

0.904 ± 0.027 0.906 ± 0.027 0.452 ± 0.011 [␮mol L−1 ]

2.3 2.4 1.9 [%]

0.904 ± 0.043 0.902 ± 0.020 0.453 ± 0.013 [␮mol L−1 ]

3.9 1.8 2.3 [%]

Hello Viva Juice (juice as blank) TOMA multivitamin juice (juice as blank) Hello Viva Juice TOMA multivitamin juice

10.0 10.0 >0.906 >0.906

10.46 ± 0.47 (recovery 99.9–109.3%) 10.76 ± 0.81 (recovery 99.5–115.7%) 2.06 ± 0.11 2.280 ± 0.084

3.6 6.1 4.1 3.0

10.48 ± 0.51 (recovery 99.7–109.9%) 10.15 ± 0.22 (recovery 99.4–103.7%) 2.14 ± 0.11 2.348 ± 0.076

4.0 1.7 4.2 2.6

a

(added) (added)

Average of 5 determinations.

as blank and the content of added FA to the final concentration of 10.0 ␮mol L−1 (4.41 mg L−1 ) was determined). It is clear from the results in Table 5 that the confidence band of results recorded in both juices with both electrodes contained the level of FA, which was added to the juice samples. The determined levels of FA in both juices were more than two times higher than the content of FA declared by producer (Table 5). The error band (p = 0.05) of results achieved with mAgSAE overlapped the error band of those results achieved with HMDE (p = 0.05) in case of both juices (Fig. 5). Therefore, it can be concluded that the results achieved using both electrodes are equivalent, i.e., the results are independent of the type of the used electrodes.

4. Conclusions It has been shown that mercury meniscus modified silver solid amalgam electrode in combination with DCV or DPV is a suitable sensor for the determination of FA in subnanomolar concentrations. It provided stable and reproducible responses during long-time measurements. The limits of detection and quantification of FA proved to be sufficient for the determination of this analyte in waters, food supplements, nutrition supplements, etc., because the levels of this analyte are usually much higher than that of quantifications limit. The limits of quantification are about one order of magnitude higher than those reached with differential pulse voltammetry with HMDE, which disadvantage seems to be less important due to generally higher levels of FA in real samples. Moreover, m-AgSAE is more robust, can be characterized by sufficiently broad potential window, by acceptable signal-to-noise ratio, its mechanical stability enables its application in flowing systems, and it is resistant toward passivation. The reaction mechanism seems to be similar at m-AgSAE as wall as at HMDE. The parameters of reproducibility of repeated records (RSD = 1.2%) as well as of the complete determinations (RSD < 3.2%) achieved with m-AgSAE are fully comparable with those reached with HMDE (0.7% and 2.8%, respectively). The applicability of the mercury meniscus modified silver solid amalgam electrode for differential pulse voltammetric determination of folic acid was verified by analysis of real samples of 3 different nutrition supplements containing folic acid in different

levels. In all cases it was statistically verified (p < 0.05), by the use of differential pulse voltammetry, that the results achieved with mercury meniscus modified silver solid amalgam electrode were equal to those achieved with HMDE and were in fair agreement with the content of the folic acid declared by producer. The developed procedure realized with m-AgSAE was successfully applied also for FA determination in samples of two different multivitamin juices. The found amount was in fair agreement with added amount of FA as well as with its content found with HMDE. The possible utilization of this electrode for the determination of this analyte in body fluids (urine) is under investigation. The research described in this manuscript demonstrated that the m-AgSAE can be considered as an environmentally friendly tool and interesting alternative for mechanistic studies and the analytical determination of FA and analogically other similar pharmaceutical compounds.

Acknowledgements The authors gratefully acknowledge the financial support by the Ministry of Education, Youth and Sports of the Czech Republic by the Research Centre No. LC06035, by the project MSM 0021627502, and by the Grant Agency of Academy of Sciences of the Czech Republic (Project No. IAA400400806). References [1] R.K. Murray, K.D. Granner, P.A. Mayes, V.W. Rodwell, Harper’s Biochemistry, Appleton and Lange, Stamford, 1996. [2] Anonymous, Dihydrofolate reductase, http://www.expasy.org/enzyme/1.5.1.3 (01.05.2010). [3] T. Navratil, M. Petr, Z. Senholdova, K. Pristoupilova, T.I. Pristoupil, M. Heyrovsky, D. Pelclova, E. Kohlikova, Physiol. Res. 56 (2007) 113. [4] T. Navratil, E. Kohlikova, M. Petr, M. Heyrovsky, D. Pelclova, K. Pristoupilova, T.I. Pristoupil, Z. Senholdova, Food Chem. 112 (2009) 500. [5] A.E. Czeizel, E. Puho, Nutrition 21 (2005) 698. [6] D.S. Wald, J.K. Morris, M. Law, N.J. Wald, Br. Med. J. 333 (2006) 1114. [7] C. La Vecchia, E. Negri, C. Pelucchi, S. Franceschi, Int. J. Cancer 102 (2002) 545. [8] Anonymous, What you need to know about folate, http://www.eufic.org/article/en/artid/about-folate/ (02.05.2010). [9] J. Alaburda, A.P. de Almelda, L. Shundo, V. Ruvieri, M. Sabino, J. Food Comp. Anal. 21 (2008) 336. [10] L.A. Kozhanova, G.A. Fedorova, G.I. Baram, J. Anal. Chem. 57 (2002) 40. [11] U. Holler, C. Brodhag, A. Knobel, P. Hofmann, V. Spitzer, J. Pharm. Biomed. Anal. 31 (2003) 151. [12] H.B. Li, F. Chen, J. Sep. Sci. 24 (2001) 271.

L. Bandˇzuchová et al. / Electrochimica Acta 56 (2011) 2411–2419 [13] M. Rychlik, S. Monch, Ernahr. -Umsch 56 (2009) 270. [14] J.D.M. Patring, J.A. Jastrebova, J. Chromatogr. A 1143 (2007) 72. [15] R.H.F. Cheung, J.G. Hughes, P.J. Marriott, D.M. Small, Food Chem. 112 (2009) 507. [16] S.L. Zhao, H.Y. Yuan, C. Xie, D. Xiao, J. Chromatogr. A 1107 (2006) 290. [17] P. Nagaraja, R.A. Vasantha, H.S. Yathirajan, Anal. Biochem. 307 (2002) 316. [18] M.B. Caselunghe, J. Lindeberg, Food Chem. 70 (2000) 523. [19] E. Jacobsen, M. Wiesebjornsen, Anal. Chim. Acta 96 (1978) 345. [20] J.M.F. Alvarez, A.C. Garcia, A.J.M. Ordieres, P.T. Blanco, J. Electroanal. Chem. 225 (1987) 241. [21] B. Prokopova, M. Heyrovsky, Bioelectrochem. Bioenerg. 41 (1996) 209. [22] M. Heyrovsky, B. Prokopova, Collect. Czech. Chem. Commun. 62 (1997) 172. [23] M.J.F. Villamil, A.J.M. Ordieres, A.C. Garcia, P.T. Blanco, Anal. Chim. Acta 273 (1993) 377. [24] A.C. Le Gall, C.M.G. Vandenberg, Anal. Chim. Acta 282 (1993) 459. [25] M. Plavsic, Environ. Int. 30 (2004) 761. [26] H.X. Guo, Y.Q. Li, L.F. Fan, X.Q. Wu, M.D. Guo, Electrochim. Acta 51 (2006) 6230. [27] V.D. Vaze, A.K. Srivastava, Electrochim. Acta 53 (2007) 1713. [28] Q.J. Wan, N.J. Yang, J. Electroanal. Chem. 527 (2002) 131. [29] F. Xiao, C. Ruan, L. Liu, R. Yan, F. Zhao, B. Zeng, Sens. Actuators B: Chem. 134 (2008) 895. [30] J. Wang, Analytical Electrochemistry, Wiley-VCH, New York, 2006. [31] T. Navratil, J. Barek, Crit. Rev. Anal. Chem. 39 (2009) 131. [32] I. Svancara, K. Vytras, K. Kalcher, A. Walcarius, J. Wang, Electroanalysis 21 (2009) 7. [33] T. Navratil, S. Sebkova, M. Kopanica, Anal. Bioanal. Chem. 379 (2004) 294. [34] S. Sebkova, T. Navratil, M. Kopanica, Anal. Lett. 37 (2004) 603. [35] T. Navratil, Z. Senholdova, K. Shanmugam, J. Barek, Electroanalysis 18 (2006) 201. [36] S. Sebkova, T. Navratil, M. Kopanica, Anal. Lett. 36 (2003) 2767. [37] S. Sebkova, T. Navratil, M. Kopanica, Anal. Lett. 38 (2005) 1747. [38] L. Novotny, B. Yosypchuk, Chem. Listy 94 (2000) 1118. [39] O. Mikkelsen, K. Schroder, Anal. Lett. 33 (2000) 3253. [40] O. Mikkelsen, K.H. Schroder, Electroanalysis 16 (2004) 386. [41] B. Yosypchuk, L. Novotny, Crit. Rev. Anal. Chem. 32 (2002) 141. [42] R. Selesovska-Fadrna, M. Fojta, T. Navratil, J. Chylkova, Anal. Chim. Acta 582 (2007) 344.

2419

[43] R. Selesovska-Fadrna, T. Navratil, M. Vlcek, Chem. Anal. (Warsaw) 52 (2007) 911. [44] R. Fadrna, Anal. Lett. 37 (2004) 3255. [45] R. Fadrna, B. Yosypchuk, M. Fojta, T. Navratil, L. Novotny, Anal. Lett. 37 (2004) 399. [46] B. Yosypchuk, J. Barek, Crit. Rev. Anal. Chem. 39 (2009) 189. [47] P. Cizkova, T. Navratil, I. Sestakova, B. Yosypchuk, Electroanalysis 19 (2007) 161. [48] J. Barek, J. Fischer, T. Navratil, K. Peckova, B. Yosypchuk, J. Zima, Electroanalysis 19 (2007) 2003. [49] T. Navratil, B. Yosypchuk, J. Barek, Chem. Anal. (Warsaw) 54 (2009) 3. [50] K. Peckova, T. Navratil, B. Yosypchuk, J.C. Moreira, K.C. Leandro, J. Barek, Electroanalysis 21 (2009) 1750. [51] J. Fischer, J. Barek, B. Yosypchuk, T. Navratil, Electroanalysis 18 (2006) 127. [52] J. Barek, D. Cabalkova, J. Fischer, T. Navratil, K. Peckova, B. Yosypchuk, Environ. Chem. Lett. (2010), doi:10.1007/s10311-009-0250-x. [53] A. Danhel, K. Peckova, K. Cizek, J. Barek, J. Zima, B. Yosypchuk, T. Navratil, Chem. Listy 101 (2007) 144. [54] B. Yosypchuk, T. Navratil, A.N. Lukina, K. Peckova, J. Barek, Chem. Anal. (Warsaw) 52 (2007) 897. [55] J. Barek, E. Dodova, T. Navratil, B. Yosypchuk, L. Novotny, J. Zima, Electroanalysis 15 (2003) 1778. [56] R.C. Gurira, C. Montgomery, R. Winston, J. Electroanal. Chem. 333 (1992) 217. [57] J. Fischer, L. Vanourkova, A. Danhel, V. Vyskocil, K. Cizek, J. Barek, K. Peckova, B. Yosypchuk, T. Navratil, Int. J. Electrochem. Sci. 2 (2007) 226. [58] J.N. Miller, J.C. Miller, Statistics and Chemometrics for Analytical Chemistry, Pearson Education, Harlow, 2005. [59] M. Meloun, J. Militky, M. Forina, Chemometrics for Analytical Chemistry, vol. 1: PC-Aided Statistical Data Analysis, vol. 2: PC-Aided Regression and Related Methods, Ellis Horwood, Chichester, 1992. [60] O. Dracka, J. Electroanal. Chem. 402 (1996) 19. [61] J. Skopalova, T. Navratil, Chem. Anal. (Warsaw) 52 (2007) 961. [62] L. Trnkova, R. Kizek, O. Dracka, Electroanalysis 12 (2000) 905. [63] R. Selesovska, L. Bandzuchova, T. Navratil, J. Chylkova, Electroctroanalysis (2011), doi:10.1002/elan.201000440. [64] M. Meloun, J. Militky, Statistical Analysis of Experimental Data, Academia, Prague, 2006.