where n is the number of moles of electrons per mol of analyte, F is the constant of Faraday, A is the electrode surface area, D is the diffusion coefficient, c is the concentration, and t is the sampling time or time of measurement of the current. Although the already mentioned latest techniques have displaced the linear sweep voltammetry, still they show several potentialities. A large number of inorganic and organic substances can be quantified using this technique, if the experimental conditions are guaranteed, being possible to lower detection limits by preconcentration [1, 3]. The simultaneous quantification and the possibility of further speciation are some of the potentialities of the linear sweep voltammetric method. The determination of the number of electrons involved in an electrochemical reaction, the diffusion coefficient, the stability constants of coordination compounds, and the calculation of thermodynamic magnitudes are also possible [1, 3, 4]. It also allows, from the analysis of the voltammograms, establishing working conditions for the development of other electrochemical analysis methods, such as potentiometry, amperometry, and others. The voltammetric sensors still make useful the technique of linear sweep, most of all with quantitative purposes [1, 3, 5].

The equipment used for the development of the voltammetry is the potentiostat. It allows imposing a program of potential over time, and monitoring the current, throwing the voltammogram. Although its price is lower than other equipment for instrumental analysis, such as spectrophotometers, even the potentiostats are inaccessible to many laboratories. Other authors have reported research results using non-commercial equipment for the development of voltammetry methods [4, 6, 7].

Total Microscale Chemistry is based on the use of equipment and measuring instruments made of materials of easy local procurement, including plastics, metal bars, wires, inexpensive electronic components, etc. Microscale methods allow to spend less solvent and water, generating smaller amounts of waste, having lower operating costs, requiring less storage space, their equipment reparation is inexpensive, and allowing fieldwork [8-11]. In this paper, a minimal instrumentation micropolarograph was built, and its metrological characterization was performed from the oxidation of potassium hexacyanoferrate (II). Voltammograms for oxidation of potassium iodide and ascorbic acid were additionally obtained.

Experimental

Construction of the Minimal Instrumentation Micropolarograph

It causes a potential fall through the working electrode and the auxiliary electrode, permitting to measure the electrolysis current. The circuit was assembled in a plastic box as support. Multimeters Steren were used for measurements of the potential and of the current. In order to check the electric circuit according to Ohm's law, a fictitious cell was built employing two resistances, simulating a system of three electrodes.

A plastic bucket of 12 mL was used as electrolytic cell. Working and auxiliary carbon electrodes were used. A copper wire immersed in an aqueous solution of Cu (II), and separated from the solution using a permeable cotton (Cu|Cu²⁺||) [12], was used as reference electrode.

Redox systems studied

Cathodic potential sweep from 0 to -1500 mV, and anodic sweep from 0 to 1800 mV were performed to obtain the electroactivity window. The oxidation of K4[Fe(CN)6] was studied using the same anodic sweep. The analyte solutions were prepared in the supporting 1 mol L^-1 KNO3 electrolyte. Voltammograms for oxidation of analyte solutions at 0.02, 0.04, 0.06, 0.08 and 0.1 mol L^-1 were obtained to study the linear relationship between the limiting current and the concentration. Linear correlation and determination coefficients were determined.

The voltammograms for oxidation of potassium iodide and ascorbic acid were obtained. In the first case, 1 mol L^-1 KNO3 was used as supporting electrolyte; in the second case, the supporting electrolyte was a buffer pH 4 of 1 mol L^-1 sodium acetate/acetic acid. The same anodic sweep was used. The sampling time (τ) in all the experiments was 20 s.

The limiting current was determined as the average of the diffusion plateau. The half-wave potential (E1/2) was mathematically determined, fitting a straight line to the linear region of the sigmoid curve, substituting the value of the half of the limiting current in the equation obtained from the simple linear regression and computing the half-wave potential with a path ordered to the origin.

Statistical processing of results and metrological analysis Ten voltammograms for oxidation of 0.08 mol L^-1 K4[Fe(CN)6] were obtained. The coefficient of variation was calculated. A comparison of the curves was made using Kruskal-Wallis and Mood tests [13, 14].

The uncertainties associated with the atomic weights of the elements were taken from the biannual publications of the IUPAC [16]. The repeatability of the multimeters was determined using a DCV Power Supply stabilized employing a voltage regulator 7812.

Finally, the limiting current is reported as [ilim ± U].

Statgraphics Centurion, Origin 8.0 and Microsoft Office Excel 2010 were used for the statistical processing of the results.

Results and discussion

Construction of the MIMP

It is possible to see that the correlation and determination coefficients were 0.9999. This shows a linear relationship between the current and potential in simulated conditions of a three electrode system, showing the correct operation.

Study of the system [Fe(CN)6]^4-/[Fe(CN)6]^3-

The reaction associated with curve 1 is the oxidation of the solvent, according to the following equation:

The other curves (2-6) have a sigmoid form, showing how the current increases with the potential to reach the limiting current; in that moment, the speed of the electrochemical process is limited by the maximum diffusion of the analyte to the electrode surface [1, 2]. The increase of the limiting current with the analyte concentration is observed in accordance with Cottrell equation. The first branch of oxidation in the curves corresponds to the following electrochemical process:

The linear regression coefficients and determination are shown. The first shows that there is a high correlation between the two variables, and the coefficient of determination shows that the mathematical model satisfactorily describes the linear relationship between the limiting current and concentration.

This linear relationship yields a quantitative value for the voltammetric results obtained using the MIMP.

Repeatability and uncertainty of measurement of the limiting current

The most important coefficient of variation refers to the limiting current, because this is the parameter utilized in the quantitative determinations. The differences between the voltammograms do not have a great influence in the values of the limiting currents, because these are calculated as the average of the points in the diffusion plateau.

The coefficient of variation associated with the half-wave potential is acceptable, while the coefficient associated with the limiting current shows a good repeatability of this parameter on the curves. The results are satisfactory, considering that the coefficients of variation associated with voltammetric methods have been reported up to 4% [1, 18]. The expanded uncertainty associated with the limiting current to 0.08 mol L^-1 was 1.9 μA. The repeatability of the method is the most significant source of uncertainty, contributing with 99.7 % to the combined uncertainty. The limiting current is reported as [165.0 ± 1.9] μA, for a relative uncertainty of 1.1%.

Levene's test to verify the homogeneity of variance demonstrated that there are significant differences between the variances associated with the limiting current of the curves (p-value = 2.07×10^-7; α = 0.05). This is because in some curves the plateau current remains substantially constant, while in others, it tends to increase.

Another fact that contributes to differences between variances is that the number of current samples (points on the plateau) in all curves is not the same. Kruskal- Wallis and Mood tests comparing the averages did not show statistically significant differences between the ten plateaus (p-value = 0.767772 and p-value = 0.765193, respectively; α = 0.05). These tests confirm that the limiting currents of the ten curves do not statistically differ, checking the repeatability of the voltammograms.

The linear correlation coefficient between potential and time is an important parameter, considering that the potential sweep was manually done using a rheostat, so it may present deviations from linearity. However, there is no influence of sweep linearity in limiting current in the range of correlation coefficients of 0.9488 to 0.9989, considering that there are not differences between the limiting currents.

Other electroactive systems

In this case, the half-wave potential was 557 mV, being in correspondence with Baeza and Garcia [19], reporting a value of 567 mV under similar experimental conditions. The associated electrochemical reaction is:

The curves for the electrochemical reaction of the analyte and the buffer are shown.

The branch observed in the voltammogram corresponds to the oxidation of ascorbic acid to dehydroascorbic acid, according to the following reaction [19]: