Computer-controlled Potentiostat Models 263A and 273A-PAR (Princeton Applied Research, Oak Ridge, TN, USA) with the software 270/250-PAR were used for the voltammetric measurements. A micro-electrolysis cell consisting of C-2 stand with an electrode body (BASi Model MF-2010), an Ag/AgCl/3M KCl reference electrode (BASi Model MF-2079), and a platinum wire counter electrode were used. A magnetic stirrer with a Teflon-coated magnet was used to provide the convective transport during the preconcentration step. An Eppendorf centrifuge (Model 5417 C, Hamburg, Germany) was used for separation of precipitated proteins from human serum and plasma samples prior to the assay. A micropipette (Eppendorf-Multipette^Ò plus) was used for transferring the analyte solutions throughout the experimental work. De-ionized water (6.0 ×  10⁶ Ω) was used throughout the present work.

Distilled water (AR Grade, S. D. Fine Chemicals Ltd., Mumbai, India) and Whatman filter paper no. 41 (Whatman International Ltd., England) were used in the study. A stock standard solution of 1.0 × 10⁻³ mol L⁻¹ bulk DXT.HCl was prepared in distilled water and then stored at 4˚C. Working solutions DXT.HCl (1.0 × 10⁻⁸ to 1.0 × 10⁻⁴ mol L⁻¹) were prepared daily by appropriate dilution with distilled water just before use. Cymbalta^® tablets (obtained from a local pharmacy) labeled to contain 30 mg DXT.HCl per tablet were quantitatively weighed and the average mass per tablet was determined. The tablets were then grounded (20 tablet) in a mortar to a homogeneous fine powder. A quantity of this powder equivalent to the weight of one tablet was accurately transferred into a 100 mL volume calibrated flask containing 70 mL distilled water. The content of the flask was sonicated for about 10 min and then filled up with distilled water. The solution was then filtered through a 0.45 μm Mille-pore filter (Gelman, Germany). The desired concentrations of the DXT.HCl were obtained by accurate dilution with distilled water. Six serum samples of three healthy subjects (two samples from each subject) were stored frozen until assay. Into each of 10 centrifugation tubes containing a certain concentration of DXT.HCl (1.0 × 10⁻⁸ to 1.0 × 10⁻⁴ mol L⁻¹), 1 mL-volume of the human serum was transferred, and then mixed well with 1 mL of ethanol to denature and precipitate proteins. The solutions were centrifuged for 3.0 min at 14,000 rpm to separate out the precipitated proteins. The clear supernatant layers of the solutions were filtered through 0.45 μm Mille-pore filters to produce protein-free human serum samples spiked with various concentrations of DXT.HCl (1.0 × 10⁻⁸ to 1.0 × 10⁻⁴ mol L⁻¹). An aliquot of this solution (0.01 - 0.1 ml) was diluted to 10 ml with Britton-Robinson universal buffer or phosphate buffer, and then transferred into a dark voltammetric cell.

A series of phosphate buffers of pH values 6.2 to 7.5 and universal buffers of pH values 2 to 12 were prepared and used as a supporting electrolyte. A pH-meter (Crison, Barcelona, Spain) was used for the pH measurements. Deionized water was supplied from a Purite-Still Plus de-ionizer connected to an AquaMatic double-distillation water system (Hamilton Laboratory Glass LTD., Kent, UK).

The carbon paste (CP) was prepared by mixing an amount (5.0 g) of graphite powder (1 - 2 µm, Aldrich, Milwaukee, WI, USA) and 1.8 mL Nujol oil (Sigma, density = 0.84 g mL⁻¹) uniformly by milling in a small agate mortar. The body of the electrode was a Teflon rod with an end cavity (BASi Model MF-2010, 3.0 mm diameter and 1.0 mm deep) bored at one end for paste filling. Contact was made with a copper wire through the center of the Teflon rod. An amount of the prepared CP was pressed into the end cavity of the electrode body and leveled off with a spatula. Surface of the constructed CPE was manually smoothed by polishing on clean paper before use.

Randles-Sevcik formula can be used to calculate the electro-active area of the electrode using cyclic voltammetric technique and 1.0 × 10⁻³ mol L⁻¹ K₃Fe (CN)₆ as a probe at different scan rates in 0.1 mol L⁻¹ KCl as supporting electrolyte. At T  =  298 K and for a reversible process, the Equation (1) is as follows:

I p = ( 2.69 × 10 5 ) n 3 / 2 A 0 D R 1 / 2 ν 1 / 2 C 0 (1)

In Equation (1), for 1.0 × 10⁻³ mol L⁻¹ K₃Fe (CN)₆ and 0.1 mol L⁻¹ KCl as supporting electrolyte, I_p refers to the anodic peak current, n is the number of electron transferred during the electrode reaction equal to 1. A₀ is the surface area of the electrode, D_R is the diffusion coefficient equal to 7.6  ×  10⁻⁶ cm² s⁻¹, υ is the scan rate, and C₀ is the concentration of K₃Fe(CN)₆. From the slope of the plot of I_p vs. υ^1/2, the area of the electrode surface was calculated to be 0.032  ±  0.0018 cm².

The electrochemical behavior of 1.0 × 10⁻⁴ mol L⁻¹ DXT.HCl at a CPE was studied by cyclic voltammetry at a 100 to 500 mVs⁻¹ in the B-R universal buffer of various pH values (2 - 12), Figure 1.

The oxidation process of DXT.HCl involved mainly its secondary amino group oxidation from the molecule structure. The peak potential E_p of the main peak shifted to less positive values with the increase of pH, which denotes that the protons are involved in the electrode reaction process and the proton-transfer reaction precedes the electron transfer process [19] . A linear relationship of the peak potential (E_p) vs. (pH) was obtained; via the corresponding regression equation: E_p (V) = −0.069 pH + 1.537, (r = 0.989 and n = 5). Its slope value equals 69 mV = {(59/α). ( Z H + / n a } [20] , where n_a and Z H + are the numbers of electrons

and protons involved in the rate determining step, respectively, and α is the symmetry transfer coefficient. It is well know that two electrons (n_a = 2) and one proton ( Z H + = 1) are involved in the rate-determining step of the electro-oxidation process of secondary amino group, i.e., the ratio ( Z H + / n a ) = 1/2. Accordingly, α- value of 0.42 was estimated from slope value (69 mV/pH) of the (E_p) vs. (pH) plot, indicating the irreversible nature of the electrode process of DXT.HCl at a CPE [19] .

Also, the irreversible nature of the electrode reaction was also identified from the shift of peak potential E_p to less positive values upon the increase of scan rate v (100 - 500 mV s⁻¹) at different pH values [21] [22] [23] [24] , as shown in Equation (2)

E p = E 0 + ( R T / 2 α n a F ) ln ( R T k s / α n a F ) − ( R T / 2 α n a F ) ln v (2)

Values of αn_a (product of symmetry transfer coefficient α and number of electrons n_a transferred in the rate-determining step) of 0.027 - 0.42 were estimated from slope values of the obtained (E_p) vs. (lnv) plots according to following equations of the totally irreversible electrode reaction [21] [22] [23] [24] , Equation (3)

Δ E p / Δ ln v = 0.02569 / 2 α n a (3)

The most probable values of transfer coefficient α (0.34 - 0.47) were estimated at various pH values, for the number of electrons (n_a equals 2) transferred in the rate-determining step for the electro-oxidation of the secondary amino group of the analyte [19] , this is confirmed again the irreversible nature of the electrode reaction of DXT.HCl at a CPE.

On the basis of the above results, a mechanism can be described. In the first step, a protonation-deprotonation equilibrium step, followed by a slow step in which 2e⁻ and H⁺ are involved in the electro-oxidation of the secondary amino group. Foramation of formaldehyde has been confirmed and developed for precise quantification of formaldehyde in aqueous samples at trace levels by using TLC method. The analytical method of formaldehyde analysis based on the quantitative spot test technique. The violet-red spots of formaldehyde with chromotropic acid were developed on TLC. Chromotropic acid has been widely used as an analytical reagent in organic as well as in biological chemistry after the discovery of violet spots with formaldehyde in 1937 [25] [26] . The color densities of the spots were measured with simple software by taking the image of spotty TLC in computer subsequent to scanning. TLC plate (Merck, Aluminum sheet, Silica gel 60F₂₅₄, 3 × 5 cm) was used to develop spots since paper could not be employed (burnt because of concentrated sulfuric acid). One micro-liter drop of each, chromotropic acid, sulfuric acid and formaldehyde solution taken from voltammetric cell was employed with micropipette (Pipettman) one over the other. The order of application of reagents and formaldehyde solution was altered to check the effect on color density of the spot. Later, the TLC was placed in an oven for 2 - 20 min at 60˚C to notice the effect of time. Violet color spots intensifying on cooling were obtained. This indicates the formation of formaldehyde and removal of methyl group in voltammetric cell is shown in Scheme 2.

The interfacial adsorptive affinity of DXT.HCl onto a CPE surface was also designated by recording the cyclic voltammograms of 2.0 × 10⁻⁵ mol L⁻¹ DXT.HCl at 200 mV s⁻¹ in the phosphate buffer of pH 6.8 following its preconcentration by adsorptive accumulation onto a CPE, under open circuit conditions (Figure 2,

Scheme 2. Electrode reaction mechanism of the oxidation process of DXT.HCl at a CPE.

curve a), and then at preconcentration potential (E_acc) of 0 V (vs. Ag/AgCl/KCl_s) for 10 s (Figure 2, 1^st cycle, curve b and 2^nd cycle, curve c). As shown in Figure 2, an enhanced peak current magnitude was observed following preconcentration of the analyte by adsorptive accumulation onto a CPE (1^st cycle, curve b) compared to that recorded following accumulation of the drug at open circuit (curve a) confirmed the interfacial adsorptive character of DXT.HCl onto a CPE. Whereas, in the 2^nd cycle (curve c) the voltammogram exhibited a very small peak current compared to that of the 1^st cycle (curve b), which may be attributed to desorption of DXT.HCl from the CP electrode surface. Furthermore, the electrode surface coverage (Γ^o (mol cm⁻²)) of DXT.HCl in the phosphate buffer of pH 6.8 was estimated using the Equation (4) [21] .

Γ o = Q / n F A (4)

where Q (C) is the charge consumed by the surface process which estimated by the integration of the area under the peak corrected to the residual current, n is the total number of electrons consumed in the oxidation process (n = 2), F is the Faraday constant (96,487 C) and A is the surface area of the working electrode (0.032 cm²). On dividing the amount of charge (Q) consumed by the surface process, 29.97 × 10⁻⁶ C, by the conversion factor nFA (6175.168 C mol⁻¹ cm²), a monolayer surface coverage of 4.85 × 10⁻⁹ mol cm⁻² was estimated. Each adsorbed DXT.HCl molecule thus occupied an area of 0.0342 nm². Furthermore, the logarithm of the peak current (i_p) vs. logarithm of scan rates (ν) (100 to 500 mV s⁻¹) was a linear relationship following the regression equation: logi_p = 0.81 logν + −1.43 (r = 0.980 and n = 5). The slope value of 0.81 is very close to the expected theoretical value of 1.0 for an ideal reaction of surface species [22] .

Stripping voltammetric methods were optimized for trace determination of DXT.HCl applying linear sweep and square wave potential-waveforms. Stripping voltammograms of bulk DXT.HCl in the phosphate buffer (pH 6.2 to 7.5) recorded by linear sweep and square wave voltammetry following its preconcentration onto a CPE by adsorptive accumulation for 20 s exhibited a well-defined single irreversible anodic peak with a better enhanced peak current magnitude at pH 6.8. Therefore, a phosphate buffer of pH 6.8 was chosen as a supporting electrolyte in the rest of study.

The described LS-AdASV and SW-AdASV methods were successfully applied to

(A) Using the calibration curve method and (B) Using the standard addition method. The theoretical values of F and t-test at 95% confidence limit for n₁ = 5 and n₂ = 5 (n = Number of replicated measurements) are 6.39 and 2.3, respectively.

direct determination of DXT.HCl in the commercial “Cymbalta^® tablets” (as labeling to contain 30 mg DXT.HCl per tablet as an individual drug) without the necessity for samples pretreatment and/or time-consuming extraction steps prior to the analysis. The mean percentage recoveries of DXT.HCl, based on the average of five replicate measurements using the described methods are reported and statistically compared with those obtained by reported potentiometric method (Table 3) [13] . Since the calculated F-value did not exceed the theoretical values, which means that there was no significant difference between the proposed and reported methods with respect to the reproducibility [32] . Also, no significant differences were noticed between the proposed and reported methods regarding accuracy and precision as revealed by t-test [32] .

A quantitative assay of DXT.HCl spiked in human serum was carried out by the described LS-AdASV and SW-AdASV methods without the necessity for sample pretreatment and/or time consuming extraction or evaporation steps prior to the analysis. Representative SW-AdAS voltammograms of DXT.HCl spiked in human serum following preconcentration onto a CPE by adsorptive preconcentration are shown in Figure 5. As shown in Figure 5 (curve a), no interfering peaks were observed in the blank human serum within the studied potential range. The peak current (i_p) of SW-AdASV method showed a linear dependence on the concentrations of DXT.HCl over the range 7.0 × 10⁻⁸ to 1.0 × 10⁻⁶ mol L⁻¹ DXT.HCl (Figure 5), depending on the preconcentration time, following the regression equation: (i_p (µA) = 5.055 C (µmol L⁻¹) + 1.19; r = 0.980 and n = 6). Detection limit of 2.1 × 10⁻⁸ mol L⁻¹ and quantitation limit of 7.0 × 10⁻⁸ mol L⁻¹ DXT.HCl was achieved by the described SW-AdASV method. A similar behavior was recorded in case of LS-AdASV method, the peak current (i_p) of LS-AdASV method showed a linear dependence on the concentrations of DXT.HCl over the range 2.0 × 10⁻⁷ to 5.0 × 10⁻⁶ mol L⁻¹ DXT.HCl, depending on the preconcentration time, following the regression equation: (i_p (µA) = 1.368 C (µmol L⁻¹) + 0.514; r = 0.940 and n = 7). Detection limit of 6.0 × 10⁻⁸ mol L⁻¹ and quantitation limit of 2.0 × 10⁻⁷ mol L⁻¹ DXT.HCl was achieved by the described LS-AdASV method.

The obtained mean percentage recoveries and relative standard deviations by means of described LS-AdASV and SW-AdASV methods were 98.12 ± 2.23 and 98.64 ± 1.74, respectively. The results indicated that the described voltammetric methods are sensitive enough for assay of DXT.HCl in spiked human serum. Moreover, effects of various foreign species that are likely to be in biological samples on analysis of DXT.HCl spiked in human serum were also evaluated. This was performed by analysis of standard solutions of 5 × 10⁻⁶ mol L⁻¹ DXT.HCl spiked with various excess amounts of some foreign species such as some common metal ions, excipients, and co-administrated drugs (e.g., Na⁺, K⁺, Ca²⁺, Mg²⁺, Zn²⁺, Cu²⁺, Fe³⁺, uric acid, glucose, sucrose, starch, gelatin, lactose, hypromellose, hydroxypropyl methylcellulose acetate succinate and sodium lauryl sulfate) under the optimum operational conditions (Table 4). The tolerance limit for foreign species was taken as the largest amount yielding a signal error of 5% for determination of DXT.HCl. The results in (Table 4) are good evidence for no competition between DXT.HCl species and added metal ions for surface sites indicating that the accumulation of the drug by adsorption at the electrode surface is the favor process. It was also found that the interferences from the other foreign species to the determination of DXT.HCl under the optimized experimental conditions were negligible indicating again the reliability of the optimized methods for the trace assay of DXT.HCl in spiked human serum.

Characteristics of the calibration curves and the achieved limits of detection (LOD) and quantitation (LOQ) by means of the developed stripping voltammetric methods are reported in Table 5.

*For 5% signal error.

One of the strengths of the present study is that it reports a novel square wave, linear sweep and cyclic voltammetric methods that are quick, sensitive, accurate, precise and in expensive compared to most of the reported methods as potentiometric methods for the quantitation of DXT.Hcl in human plasma of real samples and for pharmacokinetic studies. One shortcoming of the present study may be noted as the method was qualified and validated. However, a full validation was not the aim of the current pilot study, but to qualify the method and apply it for pharmacokinetic studies in human serum. The significance of the current study is that the method qualification results obtained may provide useful information to assess the feasibility of expanding the process to larger studies in the future where the method is fully validated by using modifier carbon paste electrodes as carbon nanotubes.