From Batch to Flow Stripping Analysis with Screen-Printed Electrodes: a Possible Way to Decentralize Trace Inorganic Analysis

Home , Reference electrode, Voltammetry, Working electrode

chemosensors

Article From Batch to Flow Stripping Analysis with Screen-Printed Electrodes: A Possible Way to Decentralize Trace Inorganic Analysis

Carlo Dossi 1,* ID , Damiano Monticelli 2, Andrea Pozzi 2 ID and Sandro Recchia 2 ID 1 Dipartimento di Scienze Teoriche ed Applicate, Università degli Studi dell’Insubria, Via Dunant, 3, 22100 Varese, Italy 2 Dipartimento di Scienza ed Alta Tecnologia, Università degli Studi dell’Insubria, Via Valleggio, 11, 21100 Como, Italy; [email protected] (D.M.); [email protected] (A.P.); [email protected] (S.R.) * Correspondence: [email protected]; Tel.: +39-031-2386235

Received: 31 July 2018; Accepted: 3 September 2018; Published: 5 September 2018

Abstract: Decentralization of on-site and in-site trace metal analysis has been a key topic over the last 30 years, owing to the increasing need for environmental protection as well as industrial and health-based field applications. In trace (and ultratrace) metal analysis, electrochemical stripping analysis with mercury (or bismuth) screen-printed film electrodes has shown a fast growth in popularity thanks to the good limits of detection, the ease of application in the field, and the low cost. Moreover, the availability of new wall-jet flow cells has opened the opportunity for their use in in situ industrial monitoring. The analytical figures of merit in stripping voltammetry with screen-printed electrodes (SPEs) under decentralized conditions and/or with sensor arrays are heavily affected by some analytical factors, primarily the presence of a pseudo-reference electrode, the efficiency of mass transport during the preconcentration step, and the need for external calibration. A careful model investigation of the analytical parameters for an efficient use of SPEs in decentralized conditions has been undertaken and discussed. Different instrumental approaches were investigated, comparing optimized batch conditions and flow cell operation under either continuous flow or stopped-flow sample injection. The stripping efficiency under wall-jet flow conditions was found to be high and comparable to that in batch conditions, leading to sub-ppb (µg/L) limit of detection (LOD) figures. Finally, external calibration in stripping voltammetry was studied as a viable alternative to conventional standard addition quantitation. Results showed, indeed, that external calibration was demonstrated to be reliable for quantitation of Pb and Cd in real water samples.

Keywords: stripping voltammetry; screen-printed electrodes; trace inorganic analysis; electroanalysis; batch vs ﬂow cells

1. Introduction Anodic stripping voltammetry (ASV) has always been regarded as an extremely sensitive electroanalytical technique and is often the greatest competition for plasma spectroscopies, particularly ICP-MS, for trace metal analysis [1]. These methods exploit the properties of certain electrodic materials to preconcentrate analytes—not just inorganic but also electroactive (bio)organic molecules—on their surface and analyze them by stripping back to the bulk of solution. Despite its toxicity, mercury still plays a major role in electrochemistry thanks to its physicochemical properties and its good analytical ﬁgures in electroanalysis, particularly robustness, reproducibility, and sensitivity for heavy metal determination [2–4]. The use of screen-printed electrodes (SPEs) was a second important step in stripping electroanalysis [5]. Apart from the low cost and excellent analytical performance in organic

Chemosensors 2018, 6, 37; doi:10.3390/chemosensors6030037 www.mdpi.com/journal/chemosensors Chemosensors 2018, 6, 37 2 of 13 and bioanalytical chemistry [6,7], the microminiaturization of SPEs dramatically reduced the required amounts of metallic mercury, to the point of having it predeposited as salt on the working electrode, significantly reducing its environmental impacts [8]. In accordance with the European directive related to the restriction of metallic mercury, its substitution in favor of alternative metal coatings, such as bismuth or antimony, is an important research topic [9,10]. However, the need to investigate trace analytes under in-the-field and/or decentralized applications with potentially complex sample matrixes prompted us to continue using mercury(II) as the best choice for such study, thanks to the profound knowledge of its chemistry and electrochemical behavior [2–4]. Doubtless, bismuth(III) will be soon a viable alternative in this field, provided that a better understanding of its complex aqueous chemistry is reached [9]. The preconcentration step is the most critical aspect for in situ or in-the-field use of screen-printed electrodes in stripping analysis at trace levels. In fact, very low limits of detection and analytical sensitivities can only be reached by providing an optimized and reproducible mass transport to the electrode during the preconcentration step in batch conditions, as well as using a solvent exchange approach when complex matrixes are involved [11]. A different instrumental approach appeared in the ‘80s, involving the use of flow injection analysis for the stripping determination of trace analytes. It gained interest in the scientific community, with good analytical figures of merit and claimed patents [5,12], but it was apparently undergoing a slow oblivion. In recent years, however, this instrumental approach has resurfaced [13,14]. The availability of new, commercial low-volume wall-jet flow cells for screen-printed electrodes may be one of the main reasons for the revitalization of this technique. In organic chemistry, interesting results have already been reported [15]; more work seems, instead, to be needed for trace inorganic analysis. A recent preliminary report has shown how continuous flow operation seemed to be unsuited for their high noise, whereas flow injection was proposed as a reliable approach, with good limits of detection and reproducibility [16]. In this manuscript, a careful investigation of the analytical parameters for efficient use of SPEs in potential decentralized conditions is undertaken and discussed. Different instrumental approaches are investigated, comparing optimized batch conditions and flow cell operation, either under continuous flow or stopped-flow sample injection. Finally, external calibration in stripping voltammetry is studied, as a viable alternative to conventional standard addition quantitation.

2. Materials and Methods

2.1. Instrumental Setup Screen-printed electrodes (SPEs) were bought from DropSens (DropSens S.L, Asturias, Spain). Electrodes were printed on a 33 mm × 10 mm × 0.5 mm ceramic alumina substrate under the code DRP-C110, with carbon electrodes for working and counter, and silver for the reference electrode. The surface area of the working carbon electrode is 16 mm2. In the presence of chloride in the sample solution, a film of AgCl is grown; the presence of chloride ions is therefore a necessary requirement for using SPEs in stripping analysis. For electroanalytical experiments, an Amel 4330 versatile electroanalytical instrument (Amel Srl, Milano, Italy) was used. The Amel 4330 instrument proved very effective for stripping analysis because of its integrated voltammetric cell, which enables automatic stirring and degassing throughout all electroanalytic steps. Moreover, a Perimax 12 peristaltic pump (Spetec, Erding, Germany) for stopped-flow and flow injection operations can be easily interfaced. The Amel4330 instrument, the batch cell, and the peristaltic pump are fully handled and controlled by the Windows-based VApeak software. This software has been developed by us to allow full instrumental control, as well as full data analysis (peak finding, peak height and area calculations), and quantitation via external calibration or standard additions without requiring any external modules. All quantitative calculations were done as peak height. The conventional use of linear baselines can, however, result in underestimating peak heights in voltammetry [17]. The use of spline baselines Chemosensors 2018, 6, x FOR PEER REVIEW 3 of 13

Chemosensors 2018, 6, 37 3 of 13 All quantitative calculations were done as peak height. The conventional use of linear baselines can, however, result in underestimating peak heights in voltammetry [17]. The use of spline inbaselines VApeak in was VApeak suitably was exploited suitably toexploited improve to the improve reliability the ofreliability the calibration of the calibration calculations. calculations. Analytical sensitivityAnalytical wassensitivity defined was as thedefined slope as of the the slope best-fit of calibration the best‐fit line. calibration Limits ofline. detection Limits (LOD)of detection were determined(LOD) were followingdetermined IUPAC following guidelines IUPAC [18 guidelines] as 3 σ/m, [18] where as 3 σσ /m,was where determined σ was bydetermined running tenby replicaterunning analysesten replicate on a standardanalyses sampleon a standard near the sample limit of detection,near the limit and m of is detection, the analytical and sensitivity. m is the Limitsanalytical of quantitation sensitivity. Limits (LOQ) of were quantitation then calculated (LOQ) as were 10 σ then/m. calculated as 10 σ/m. All the PTFE beakers and cells, round-bottomedround‐bottomed flasksflasks in polypropylene, polypropylene tubes, and pipette tips used for preparation and storage of samples and solutions were rinsed with copious amounts of double-distilleddouble‐distilled water.water. Hydrochloric andand nitricnitric acid acid (TraceSelect (TraceSelect from from Fluka) Fluka) were were purified purified by by sub-boiling sub‐boiling distillation distillation in ain quartz–Teflon a quartz–Teflon apparatus. apparatus. Sodium Sodium chloride chloride (Trace (Trace Select) Select) was was purchased purchased from from Fluka. Fluka. Cadmium and leadlead standardstandard stockstock solutionssolutions (1000(1000 mg/L)mg/L) were from Sigma-AldrichSigma‐Aldrich and diluted as required.required. Mercury stockstock solutionsolution (80(80 mg/L)mg/L) was prepared from mercury(II) chloride (CarloErba, Milan Italy) in 1.3 M hydrochlorichydrochloric acid.acid. High-purityHigh‐purity MilliQ water (from MilliPore, Burlington, MA,MA, USA) was used for all solutions and standards, exceptexcept ifif otherwiseotherwise indicated.indicated.

2.2. Cells and Flow System For performingperforming stripping stripping voltammetry voltammetry under under batch conditions,batch conditions, a specially a designed, specially plastic-made designed, cellplastic (mod.4330-PX)‐made cell (mod.4330 was supplied‐PX) by was AMEL supplied (#1 in Figure by AMEL1A) and (#1 assembled in Figure on1A) an and Amel4330 assembled instrument on an (FigureAmel43301B). instrument (Figure 1B).

Figure 1.1. ((A)) TheThe batchbatch cellcell withwith screen-printedscreen‐printed electrode (SPE)(SPE) (A-1);(A‐1); (B) The assembled cell on an Amel4330Amel4330 instrument:instrument: SPE cablecable connectorconnector (B-1);(B‐1); palettepalette toptop stirrerstirrer (B-2);(B‐2); externalexternal electrodeselectrodes (B-3). (B‐3).

The screen‐printed electrode was placed on the bottom of the cell, kept in place, and sealed by a The screen-printed electrode was placed on the bottom of the cell, kept in place, and sealed by Viton O‐ring. A palette top stirrer (#2 in Figure 1B) was then placed above the active area of the a Viton O-ring. A palette top stirrer (#2 in Figure1B) was then placed above the active area of the electrode, and moved by a stepped motor in the range 0–800 rpm under computer control. In this electrode, and moved by a stepped motor in the range 0–800 rpm under computer control. In this way, way, mass transport to the electrode under stirred conditions was optimized, preventing, at the mass transport to the electrode under stirred conditions was optimized, preventing, at the same time, same time, water or acid vapors from degrading the electric contacts to the SPE. All experiments water or acid vapors from degrading the electric contacts to the SPE. All experiments were conducted were conducted at 400 rpm. Moreover, this cell allowed the use of an external 3 M KCl Ag/AgCl at 400 rpm. Moreover, this cell allowed the use of an external 3 M KCl Ag/AgCl reference electrode in reference electrode in order to reference all potentials (#3 in Figure 1B). In those cases where the order to reference all potentials (#3 in Figure1B). In those cases where the internal pseudo-reference internal pseudo‐reference Ag wire was used, a 0.01 M chloride concentration had to be kept constant Ag wire was used, a 0.01 M chloride concentration had to be kept constant in the cell solution, and in the cell solution, and potentials were referred to the reference electrode. potentials were referred to the reference electrode. The analytical step of analyte redissolution was initially done by differential pulse in the The analytical step of analyte redissolution was initially done by differential pulse in the potential potential range between −1000 and −250 mV in order to account for all chloride activity. However, range between −1000 and −250 mV in order to account for all chloride activity. However, under

Chemosensors 2018, 6, 37 4 of 13

Chemosensors 2018, 6, x FOR PEER REVIEW 4 of 13 our actual conditions, the stripping scan could be stopped at −500 mV, to save analysis time. Chemosensors 2018, 6, x FOR PEER REVIEW 4 of 13 The differentialunder our actual pulse conditions, profile was the performed stripping scan with could pulse be heightstopped of at 50 −500 mV mV, and to asave voltage analysis step time. of 5 mV. The differential pulse profile was performed with pulse height of 50 mV and a voltage step of 5 mV. A finalunder cleaning our actual step conditions, was done atthe− stripping250 mV forscan 10 could s. be stopped at −500 mV, to save analysis time. A final cleaning step was done at −250 mV for 10 s. The differential flow system pulse was profile home-made was performed in our laboratorywith pulse height using of a Perimax50 mV and 12 a peristalticvoltage step cell, of 5 a mV. six-port The flow system was home‐made in our laboratory using a Perimax 12 peristaltic cell, a six‐port injectionA final valve cleaning (Omnifit, step was Cambridge, done at −250 UK) mV for for the 10 s. loading (“LOAD”) and the injection (“INJECT”) of injection valve (Omnifit, Cambridge, UK) for the loading (“LOAD”) and the injection (“INJECT”) of samples inThe flow flow injection system was analysis home‐made (FIA) in experiments, our laboratory and using a low-volumea Perimax 12 peristaltic wall-jet Plexiglas cell, a six‐port flow cell samples in flow injection analysis (FIA) experiments, and a low‐volume wall‐jet Plexiglas flow cell injection valve (Omnifit, Cambridge, UK) for the loading (“LOAD”) and the injection (“INJECT”) of DRP-FLWCLDRP‐FLWCL from from Dropsens, Dropsens, Spain. Spain. The The fluidic fluidic cell cell waswas run at at 1.0 1.0 mL/min mL/min flow flow of eluent of eluent in order in order to to samples in flow injection analysis (FIA) experiments, and a low‐volume wall‐jet Plexiglas flow cell havehave best best conditions conditions inside inside the the wall-jet wall‐jet cell. cell. DRP‐FLWCL from Dropsens, Spain. The fluidic cell was run at 1.0 mL/min flow of eluent in order to The fullThe instrumentalfull instrumental scheme scheme for flow for flow stripping, stripping, for both for both stopped-flow stopped‐flow and and flow flow injection injection analysis, have best conditions inside the wall‐jet cell. is depictedanalysis, in is Figure depicted2. in Figure 2. The full instrumental scheme for flow stripping, for both stopped‐flow and flow injection analysis, is depicted in Figure 2.

FigureFigure 2. Instrumental 2 Instrumental scheme scheme of of the the flow flow system system to perform perform voltammetric voltammetric stripping stripping analyses analyses under under stopped‐flow (SF‐SV) or under flow injection analysis (FIA‐SV). stopped-flowFigure 2 Instrumental (SF-SV) or under scheme flow of the injection flow system analysis to perform (FIA-SV). voltammetric stripping analyses under stoppedFor stopped‐flow (SF‐flow‐SV) stripping or under flowvoltammetry injection analysis (SF‐SV), (FIA the‐SV). peristaltic pump was operated in the Forpreconcentration stopped-flow and stripping cleaning step, voltammetry whereas it was (SF-SV), switched the off peristaltic during the pump stripping was analytical operated step. in the For stopped‐flow stripping voltammetry (SF‐SV), the peristaltic pump was operated in the preconcentrationThe complete andflow cleaningtiming is described step, whereas in Figure it was 3. switched off during the stripping analytical step. preconcentration and cleaning step, whereas it was switched off during the stripping analytical step. The completeThe complete flow flow timing timing is describedis described in in Figure Figure3 3..

Figure 3. Timing diagram of stopped‐flow operation in the stripping voltammetric analysis of Pb and Cd. Steps: (a) optional preanalysis degassing under open‐cell condition; (b) preconcentration; Figure 3. Timing diagram of stopped‐flow operation in the stripping voltammetric analysis of Pb and Figure(c) 3. strippingTiming analysis; diagram (d) ofcleaning. stopped-ﬂow Lines: Green, operation solid line: in theapplied stripping cell potential; voltammetric Green, dashed analysis line: of Pb Cd. Steps: (a) optional preanalysis degassing under open‐cell condition; (b) preconcentration; open‐cell conditions, with no current flowing. Yellow: sample flow; Red: injection valve position; and Cd.(c) stripping Steps: (analysis;a) optional (d) cleaning. preanalysis Lines: degassing Green, solid under line: applied open-cell cell condition; potential; Green, (b) preconcentration; dashed line: Blue: N2 purge. (c) strippingopen‐cell analysis; conditions, (d )with cleaning. no current Lines: flowing. Green, Yellow: solid line: sample applied flow; cell Red: potential; injection Green,valve position; dashed line:

open-cellBlue: conditions,N2 purge. with no current ﬂowing. Yellow: sample ﬂow; Red: injection valve position; Blue:

N2 purge.

Chemosensors 2018, 6, 37 5 of 13

For flow injection analysis stripping voltammetry (FIA-SV), the standard and sample solutions were injected via the Omnifit valve equipped with a 0.5 mL loop. The flow timing procedure was kept similar to that reported in Reference [16]. The setup of the fluidic system was performed using potassium ferricyanide (50 mM in 1 M KNO3/0.1 M KCl solution) reduction as a model reaction at 0 mV and amperometric detection. For both flow injection and stopped-flow experiments, the fluidic setup was kept unchanged in order to allow a full comparison of the analytic performance in the two types of operation. Obviously, the Omnifit injection valve was operated only for flow injection analyses.

2.3. Measurement Procedures Prior to the analysis, screen-printed electrodes were cleaned in ethanol, rinsed in water, and left in a 0.01 M hydrochloric acid solution. No problems of lead contamination were found. Screen-printed electrodes were used in a standard 3-wire configuration. For comparison purposes, an external Ag/AgCl (3 M KCl) electrode was used as a reference in the batch cell instead of the internal Ag electrode of the SPE. The chloride ion activity had to be kept constant throughout all experiments. Prior to any analysis, a mercury film was grown on the working electrode using a plating solution of 80 mg/L HgCl2 in 1.3 M HCl at −1000 mV under a stirring (or flowing) time of 200 s. The analysis can theoretically be performed on this film without any other addition of the mercury salt. However, it was found that the addition of a 100-fold diluted plating solution to standards and samples greatly increased the reproducibility of the analysis. This is well known in the literature, since the diluted mercury solution helped to stabilize the mercury film and made the amalgam formation easier. − −2 In flow analysis, a 0.8 mg/L HgCl2 solution at pH = 2 and aCl = 10 M was used as the blank and eluent solution. Between each sample, rinsing of the cell and the SPE surface was done with MilliQ water acidified to pH = 2 with hydrochloric acid. In all solutions, if the chloride activity was different from 10−2 M, sodium chloride or nitric acid were used for adjustment, and stripping potentials were corrected for chloride activity. For standard addition quantitation, the sample was analyzed first. Then, three additions of a multistandard containing both Cd and Pb to reach a final standard concentration of around 15 and 25 µg/L were performed and the effects analyzed. External calibration was done before the samples were analyzed by using a linear calibration line. A series of five standard solutions containing Pb and Cd in a range between 0 and the limit of linearity (LOL) were prepared in the same matrix as the samples (in our case, an acidified synthetic water). Normally, calibration is performed between 0 and 25 µg/L. The five standard solutions were sequentially analyzed, and the best fit was computed to obtain the calibration line. The samples were then analyzed and the two analytes were quantified using the parameters of the best-fit calibration line.

3. Results and Discussion

3.1. From Batch to Stopped-Flow Differential Pulse Stripping Voltammetry One the most critical aspects of screen-printed electrodes is the intrinsic pseudo-reference character of the silver electrode, since potentials are dependent on the halide activity in the analyte solution. The Cd and Pb signals in batch stripping analysis on a water solution of around pH = 2 with HCl were compared using either the internal silver electrode of the screen-printed electrode or an external 3 M KCl Ag/AgCl electrode. This was made possible thanks to the versatility of the AMEL batch cell, which allows both reference electrodes to be used and easily selected. A signal shift of about 150 mV towards more negative potential is observed using the internal Ag electrode of the SPE, in agreement with a quasi-Nernstian behavior. A more detailed investigation of the SPE at different chloride ion Chemosensors 2018, 6, 37 6 of 13 Chemosensors 2018, 6, x FOR PEER REVIEW 6 of 13 differentactivities ischloride shown inion Figure activities4. The is red shown trace in is theFigure blank 4. solution, The red with trace following is the blank additions solution, of 5 µwithg/L followingeach of Cd additions and Pb standard of 5 μg/L solution. each of Cd and Pb standard solution. Chemosensors 2018, 6, x FOR PEER REVIEW 6 of 13

different chloride ion activities is shown in Figure 4. The red trace is the blank solution, with following additions of 5 μg/L each of Cd and Pb standard solution.

Figure 4. Stripping analysis of Pb and Cd at 15 μµg/L at pH = 2 and chloride ion activity of 10−2 M.−2 Figure 4. Stripping analysis of Pb and Cd at 15 g/L at pH = 2 and chloride ion activity of 10 EffectM. Effect of the of theaddition addition of chloride of chloride ions ions on stripping on stripping signals signals using using the theinternal internal pseudo pseudo-reference‐reference Ag electrode.Ag electrode.Figure 4. Stripping analysis of Pb and Cd at 15 μg/L at pH = 2 and chloride ion activity of 10−2 M. Effect of the addition of chloride ions on stripping signals using the internal pseudo‐reference Ag Interestingly,electrode. the potential shift with changing chloride ion activity is accompanied by a change Interestingly, the potential shift with changing chloride ion activity is accompanied by a change in the analytical sensitivity of Cd and Pb in particular. This observation is not unexpected, because of in the analyticalInterestingly, sensitivity the potential of Cd andshift Pbwith in changing particular. chloride This observationion activity is is accompanied not unexpected, by a change because of the different speciation chemistry of the two metals. However, two important consequences can be the differentin the analytical speciation sensitivity chemistry of Cd and of the Pb twoin particular. metals. This However, observation two importantis not unexpected, consequences because of can be seen. On one side, deposition potentials always have to be corrected for the actual chloride activity; seen.the On different one side, speciation deposition chemistry potentials of the always two metals. have However, to be corrected two important for the consequences actual chloride can activity;be on the other side, the chloride concentration has to be taken as constant during all calibration and on theseen. other On side,one side, the deposition chloride concentrationpotentials always has have to beto be taken corrected as constant for the actual during chloride all calibration activity; and quantitation steps. quantitationon the other steps. side, the chloride concentration has to be taken as constant during all calibration and Thequantitation analytical steps. behavior of the stripping analysis of cadmium and lead between 0 and 15 μg/L The analytical behavior of the stripping analysis of cadmium and lead between 0 and 15 µg/L was comparedThe analytical between behavior batch (Figure of the stripping 5A) and analysis stopped of‐flow cadmium conditions and lead (Figure between 5B). 0 and 15 μg/L was comparedwas compared between between batch batch (Figure (Figure5A) 5A) and and stopped-ﬂow stopped‐flow conditions conditions (Figure (Figure 5B).5B).

Figure 5. Cont.

Chemosensors 2018, 6, 37 7 of 13 Chemosensors 2018, 6, x FOR PEER REVIEW 7 of 13

Figure 5. Comparison of stripping analysis signals of Pb and Cd between 0 and 15 μg/L with a Figurepreconcentration 5. Comparison time of of stripping 60 s. (A analysis) Using batch signals cell; of ( PbB) andusing Cd stopped between‐flow 0 conditions. and 15 µg/L Inset: with a preconcentrationCalibration lines time for of Cd 60 (in s. (blue)A) Using and Pb batch (in red). cell; Reference (B) using electrode: stopped-flow internal conditions. pseudo‐reference. Inset: Calibration lines for Cd (in blue) and Pb (in red). Reference electrode: internal pseudo-reference. It is immediately evidenced that the intensity of the stripping signals is similar within a factor of 2 in the batch and flow cells, indicating that the efficiency of mass transfer to the working electrode is Itsimilar is immediately in both cases. evidenced However, that the use the of intensity the flow ofcell the leads stripping to a higher signals noise is contribution, similar within although a factor of 2 in theits typical batch and spike flow appearance cells, indicating may be removed that the efficiencyby moving of median mass transferfiltering tofollowed the working by low electrode‐order is similarSavitsky–Golay in both cases. smoothing However, [19]. the Both use features of the flow are already cell leads integrated to a higher into the noise VApeak contribution, software althoughand its typicalcan be spikeperformed appearance automatically, may beif necessary. removed Accordingly, by moving the median analytical filtering figures followed of merit calculated by low-order Savitsky–Golayfor the two calibration smoothing curves [19]. Both(Table features 1) show arethat already the sensitivity integrated and intercepts into the VApeak are quite software similar, and can bewhereas performed limits automatically,of detection and if of necessary. quantitation Accordingly, are lower for the the analytical batch cell. figuresInterestingly, of merit the LODs calculated for thefor twoPb and calibration Cd in the curvesbatch cell (Table are well1) show under that 0.1 μ theg/L, sensitivity approaching and those intercepts classically are reported quite similar,for mercury‐film‐coated glassy carbon electrodes; the advantages of SPEs in terms of cost and versatility whereas limits of detection and of quantitation are lower for the batch cell. Interestingly, the LODs are now fully demonstrated. for Pb and Cd in the batch cell are well under 0.1 µg/L, approaching those classically reported for mercury-film-coatedTable 1. Figures glassy of merit carbon of Cd electrodes; and Pb analysis the advantagesunder differential of SPEs pulse in stripping terms of voltammetry. cost and versatility are now fullyComparison demonstrated. between batch and stopped‐flow conditions.

Cell Type Figure of Merit Cd Pb Table 1. Figures of merit of Cd and Pb analysis under differential pulse stripping voltammetry. Sensitivity (μA × L × μg−1) 0.2831 0.1681 Comparison between batch and stopped-flow conditions. Intercept (μA) −0.306 0.098 Batch Limit of Detection (μg/L) 0.17 0.19 Cell Type Figure of Merit Cd Pb Limit of Quantitation (μg/L) 0.58 0.64 −1 SensitivitySensitivity (µA × (LμA× ×µ Lg × μ) g−1) 0.16350.2831 0.1202 0.1681 InterceptIntercept (µA) (μA) −0.1890.306 0.279 0.098 Batch Stopped‐Flow µ Limit ofLimit Detection of Detection ( g/L) (μg/L) 0.58 0.17 0.62 0.19 Limit of Quantitation (µg/L) 0.58 0.64 Limit of Quantitation (μg/L) 1.98 2.06 Sensitivity (µA × L × µg−1) 0.1635 0.1202 Intercept (µA) 0.189 0.279 ThisStopped-Flow difference in the analytical sensitivities between batch and stopped‐flow conditions can likely be ascribed to restrictedLimit diffusion of Detection inside (µ g/L)the low‐volume 0.58flow cell, since 0.62the speciation µ chemistry is kept unchanged.Limit However, of Quantitation it is harmless ( g/L) for analytical 1.98 purposes, provided 2.06 that the fluidodynamic conditions are kept constant and reproducible. ThisAn difference important in thepoint analytical to be investigated sensitivities relating between to screen batch‐ andprinted stopped-flow electrodes under conditions stirred can or likely be ascribedflow conditions to restricted is the diffusion repeatability inside between the low-volume different electrodes flow cell, and since the thedurability speciation of the chemistry active is kept unchanged. However, it is harmless for analytical purposes, provided that the fluidodynamic conditions are kept constant and reproducible. Chemosensors 2018, 6, 37 8 of 13

An important point to be investigated relating to screen-printed electrodes under stirred or flow conditionsChemosensors is the 2018 repeatability, 6, x FOR PEER REVIEW between different electrodes and the durability of the active8 electrode of 13 surfaces. A comparison among three different screen-printed electrodes shows how the standard deviationselectrode for surfaces. the analytical A comparison sensitivity among are alwaysthree different below 10%screen for‐printed both analytes. electrodes shows how the Thestandard durability deviations of the for electrode the analytical was sensitivity then investigated are always by below repeating 10% for standard both analytes. calibration analysis. The analyticalThe durability sensitivity of the slowly electrode decreases was then with investigated the increasing by repeating number standard of analysis calibration rounds, analysis. reaching a The analytical sensitivity slowly decreases with the increasing number of analysis rounds, reaching 10% decrease from the initial value after 150 analyses for Cd and over 200 analyses for Pb. After 200 a 10% decrease from the initial value after 150 analyses for Cd and over 200 analyses for Pb. After analyses, the electrode surface is significantly modified, which is particularly evident for the working 200 analyses, the electrode surface is significantly modified, which is particularly evident for the and counterworking electrodes and counter (Figure electrodes6). (Figure 6).

Figure 6. Optical microscopy pictures of the surface of the screen‐printed electrode before (left) and Figure 6. Optical microscopy pictures of the surface of the screen-printed electrode before (left) and after (right) performing 200 stripping voltammetric analyses. Experimental conditions as in Figure 4. after (right) performing 200 stripping voltammetric analyses. Experimental conditions as in Figure4. The degradation of the counter electrode was not found to be relevant, since the use of an Theexternal degradation glassy carbon of the counterelectrode electrode as counter was notdid foundnot improve to be relevant, analytical since figures the use of of anmerit. external glassyConsequently, carbon electrode the degradation as counter of the did analytical not improve sensitivity analytical with increasing figures of analysis merit. number Consequently, is to be the degradationattributed of to the surface analytical modifications sensitivity of the with working increasing electrode analysis only. number is to be attributed to surface The main difference between batch and flow techniques is the methodology used for the modifications of the working electrode only. quantitative analysis. Under batch conditions, quantitation is well known to be easily done via Thestandard main additions, difference and betweenno further batch investigation and flow has techniquesbeen done. Accordingly, is the methodology acidified tap used water for the quantitativesamples from analysis. the town Under of Como batch were conditions, analyzed quantitationafter being spiked is well at 5 μ knowng/L; recoveries to be easily in the done 95– via standard105% additions,range were andfound no for further both lead investigation and cadmium. has been done. Accordingly, acidified tap water samples fromUnder the flow town conditions, of Como wereinstead, analyzed standard after addition being spikedis quite at clumsy 5 µg/L; to recoveriesperform and in thealmost 95–105% rangeimpossible were found if a forcontinuous both lead flow and of sample cadmium. is analyzed for automated decentralized analysis. External Undercalibration flow was conditions, then found instead, to be particularly standard suited addition for isstopped quite‐flow clumsy stripping, to perform either andfor the almost impossiblesimplicity if of a changing continuous standards flow ofand sample samples is or analyzedfor the reproducibility for automated that is decentralized guaranteed by analysis.the Externalinstrumental calibration setup. was The then VApeak found software to be particularly was particularly suited efficient for stopped-flow in this respect stripping, since it has either a for dedicated section for external calibration in stripping analysis. the simplicity of changing standards and samples or for the reproducibility that is guaranteed by The calibration under stopped flow was performed with five standards of Pb and Cd between 0 the instrumentaland 25 μg/L. A setup. series Theof synthetic VApeak water software samples was was particularly then analyzed; efficient results in are this shown respect in Table since 2. it has a dedicated section for external calibration in stripping analysis. The calibrationTable 2. Reference under sample stopped analyses flow in stopped was performed‐flow stripping with voltammetry five standards with external of Pb calibration. and Cd between 0 and 25 µg/L.Flow A= 1.0 series mL/min; of synthetic Preconcentration water time samples = 60 s. was then analyzed; results are shown in Table2. The samples werePeak analyzed Height in randomCalculated order to avoidExpected error Concentration accumulation. It is noteworthy that Analyte % Recovery all samples analyzed are(μ withinA) a maximumConcentration error (μg/L) of 10%, with deviations(μg/L) from certified values mostly in the ±3% level. The analytical2.4449 reproducibility15.28 was always found15 to remain in this102% range. This result is quite0.8463 important in light of5.11 decentralizing SPE-based5 techniques in105% the field, such as Cd 4.3753 27.56 25 110% for continuous monitoring3.3041 of real streams,20.75 where addition of standards20 is really unpractical.103% 1.5945 9.87 10 99% 1.9857 14.49 15 96% 0.6657 4.57 5 91% Pb 3.4794 25.72 25 103% 2.6807 19.72 20 98% 1.3532 9.74 10 97%

Chemosensors 2018, 6, 37 9 of 13

Table 2. Reference sample analyses in stopped-ﬂow stripping voltammetry with external calibration. Flow = 1.0 mL/min; Preconcentration time = 60 s.

Calculated Concentration Expected Concentration Analyte Peak Height (µA) % Recovery (µg/L) (µg/L) 2.4449 15.28 15 102% 0.8463 5.11 5 105% ChemosensorsCd 2018, 6, x FOR 4.3753PEER REVIEW 27.56 25 110% 9 of 13 3.3041 20.75 20 103% The samples were1.5945 analyzed in random 9.87order to avoid error accumulation. 10 It is noteworthy 99% that all samples analyzed 1.9857are within a maximum 14.49 error of 10%, with deviations 15 from certified 96% values mostly in the ±3% level.0.6657 The analytical reproducibility 4.57 was always found 5 to remain in this 91% range. Pb 3.4794 25.72 25 103% This result is quite2.6807 important in light of 19.72 decentralizing SPE‐based 20 techniques in the 98% field, such as for continuous monitoring1.3532 of real streams, 9.74 where addition of standards 10 is really unpractical. 97%

3.2. From Stopped Stopped-Flow‐Flow to Flow Injection Analysis in Differential Pulse Stripping Voltammetry The major disadvantage of stopped-flowstopped‐flow operations in stripping voltammetry is the noise occurring during during the the redissolution redissolution step. step. The The source source of this of this noise noise is likely is likely to be toascribed be ascribed to two to major two causes,major causes, one being one beingrelated related to bubble to bubble formation formation on the on electrode the electrode surface surface and the and other the other to the to low the volumelow volume of the of wall the‐ wall-jetjet cell. However, cell. However, the use the of use flow of flowinjection injection during during the deposition the deposition step has step been has proposedbeen proposed to improve to improve both noise both noiseand analytical and analytical reproducibility reproducibility in flow in flowstripping stripping analysis analysis [16]. Our [16]. instrumentalOur instrumental setup setup was was then then devised devised in order in order to tomake make a adirect direct comparison comparison between between stopped stopped-flow‐flow and flowflow injection injection conditions conditions for for stripping stripping voltammetry. voltammetry. The shapeThe shape and the and reproducibility the reproducibility of injection of peaksinjection werre peaks investigated werre usinginvestigated ferricyanide using reduction ferricyanide as model reduction reaction underas model amperometric reaction detection under (Figureamperometric7). detection (Figure 7).

Figure 7. Shape and reproducibility of consecutive ﬂowflow injection analyses in potassium ferricyanide reduction at 0 mV as model reaction.

In accordance with a low‐dispersion system, the FIA peaks showed rectangular shapes with fast In accordance with a low-dispersion system, the FIA peaks showed rectangular shapes with recovery to baseline [20]. The injection peak is considered terminated within 100 s without any fast recovery to baseline [20]. The injection peak is considered terminated within 100 s without any carryover. This is important for calculating the correct timing for stripping steps in FIA‐SV. carryover. This is important for calculating the correct timing for stripping steps in FIA-SV. Moreover, Moreover, reproducibility between consecutive injections was found to be excellent. reproducibility between consecutive injections was found to be excellent. The analytical redissolution step was then performed under identical fluidic conditions, under The analytical redissolution step was then performed under identical fluidic conditions, under quiescent solution in the cell: the sample for stopped flow, or pure electrolyte for flow injection. A quiescent solution in the cell: the sample for stopped flow, or pure electrolyte for flow injection. preliminary investigation of flow‐injection stripping voltammetry of Pb and Cd between 0 and 25 A preliminary investigation of flow-injection stripping voltammetry of Pb and Cd between 0 and μg/L was very indicative of the correct implementation of the fluidic setup, giving good linearity and a relatively low background noise level. A comparison of the analytical performance between SF‐SV and FIA‐SV was done on a synthetic water sample at a Pb and Cd concentration of 15 μg/L (Figure 8).

Chemosensors 2018, 6, 37 10 of 13

25 µg/L was very indicative of the correct implementation of the ﬂuidic setup, giving good linearity and a relatively low background noise level. A comparison of the analytical performance between SF-SV and FIA-SV was done on a synthetic water sample at a Pb and Cd concentration of 15 µg/L Chemosensors(Figure8). 2018, 6, x FOR PEER REVIEW 10 of 13

Figure 8. Comparison of analytical signals of Pb and Cd at 15 μµg/L obtained obtained in in stopped stopped-ﬂow‐flow (SF (SF-SV‐SV redred trace) and in flow ﬂow injection (FIA (FIA-SV‐SV blue trace) stripping voltammetry.

ItIt is immediately evident that that the flow flow injection analysis shows significantly significantly lower noise than does stopped stopped-flow‐flow analysis. A A simplistic simplistic view view of of noise noise arising arising from from microbubble formation formation at at the electrode surface should should therefore therefore be discarded. It It appears appears plausible plausible that that the the noise noise is is related related to to the presence ofof thethe sample sample inside inside the the low-volume low‐volume flow flow cell, cell, maybe maybe influencing influencing the ideal the diffusion ideal diffusion control controlof Pb and of CdPb re-dissolvingand Cd re‐dissolving into the solution.into the solution. In any case, In any the stopped-flowcase, the stopped technique‐flow technique will always will be alwaysvery sensitive be very to sensitive the sample to the matrix. sample Consequently, matrix. Consequently, flow injection flow stripping injection voltammetry stripping voltammetry will be the willtechnique be the of technique choice particularly of choice particularly with complex with matrix complex samples, matrix since samples, the analytical since the step analytical is always step done is alwaysin pure done eluent in solution. pure eluent solution. A detailed investigation of of the analytical figures figures of merit was then undertaken, with particular emphasis onon the the analytical analytical sensibility sensibility for inter-electrodefor inter‐electrode reproducibility reproducibility and limits and oflimits detection of detection (Table3). (Table 3). Table 3. Reproducibility and figures of merit of Cd and Pb analysis under flow-injection analysis Tabledifferential 3. Reproducibility pulse stripping and voltammetry. figures of merit of Cd and Pb analysis under flow‐injection analysis differential pulse stripping voltammetry.

MeasureMeasure Type Type FigureFigure of Merit of Merit CdCd Pb Pb −1 SensitivitySensitivity (µA × (Lμ×A µ×g L ×) μg−1) 0.13820.1382 0.09095 0.09095 ElectrodeElectrode 1 1 µ − − InterceptIntercept ( A) (μA) −0.18540.1854 −0.09360.0936 −1 Sensitivity (µA × L × µg ) −1 0.1363 0.08795 Electrode 2 Sensitivity (μA × L × μg ) 0.1363 0.08795 Electrode 2 Intercept (µA) −0.3284 −0.1461 Intercept (μA) −0.3284 −0.1461 −1 Sensitivity (µA × L × µg ) −1 0.1335 0.09007 Electrode 3 Sensitivity (μA × L × μg ) 0.1335 0.09007 Electrode 3 Intercept (µA) −0.450 −0.1820 Intercept (μA) −0.450 −0.1820 Limit of Detection (µg/L) 0.32 0.35 Mean result of three electrodes Limit of Detection (μg/L) 0.32 0.35 Mean result of three electrodesLimit of Quantitation (µg/L) 1.06 1.08 Limit of Quantitation (μg/L) 1.06 1.08

The analytical sensitivities are lower than those found in stopped flow. In flow injection stripping, the preconcentration step is, in fact, related not to the deposition time, as in conventional batch or stopped‐flow stripping, but to the loop volume, which is injected during the preconcentration step. The duration of the preconcentration step is, therefore, not related to the desired enrichment of the analytes at the mercury electrode, but has to be selected in order to have

Chemosensors 2018, 6, 37 11 of 13

The analytical sensitivities are lower than those found in stopped flow. In flow injection stripping, the preconcentration step is, in fact, related not to the deposition time, as in conventional batch or stopped-flow stripping, but to the loop volume, which is injected during the preconcentration step.

TheChemosensors duration 2018 of, 6, the x FOR preconcentration PEER REVIEW step is, therefore, not related to the desired enrichment11 of of the 13 analytes at the mercury electrode, but has to be selected in order to have the injected peaks returning theto baseline injected before peaks starting returning the to analytical baseline step before of analyte starting redissolution. the analytical The step sensitivity of analyte in redissolution. this approach Theis easily sensitivity increased in bythis changing approach the is loop easily volume increased as desired, by changing and setting the theloop preconcentration volume as desired, time and as a settingfunction the of preconcentration peak duration. time as a function of peak duration. Apparently, there are fewfew problemsproblems ofof saturationsaturation ofof thethe mercurymercury ﬁlm.film. The linearity in the calibration curves for both Pb and Cd isis excellent,excellent, withwith LOLLOL valuesvalues upup toto 100100 μµg/L;g/L; the curves then tend to bend at higher concentrations, as can be expected (Figure(Figure9 9).).

Figure 9. Calibration curves in flow injection analysis stripping voltammetry for Pb and Cd between Figure 9. Calibration curves in ﬂow injection analysis stripping voltammetry for Pb and Cd 0 and 400 μg/L. Discrete points: experimental data; black lines: second‐order polynomial between 0 and 400 µg/L. Discrete points: experimental data; black lines: second-order polynomial interpolating curves. interpolating curves.

Quantitative analysis results were studied via external calibration within the limit of linearity of 100 μQuantitativeg/L. Provided analysis that samples results and were standards studied via shared external the same calibration matrix, within results the were limit found of linearity to be µ similarof 100 tog/L. those Provided reported that for samples the stopped and standards‐flow stripping shared voltammetry. the same matrix, Analytical results wereresults found for three to be acidifiedsimilar to tap those water reported samples for spiked the stopped-flow at 5 μg/L with stripping lead and voltammetry. cadmium are Analytical summarized results in forTable three 4. µ Similaracidified conclusions tap water were samples reported spiked in at a recent 5 g/L investigation with lead and [16]. cadmium are summarized in Table4. Similar conclusions were reported in a recent investigation [16]. Table 4. Stripping analysis of Cd and Pb under flow‐injection analysis differential pulse stripping Table 4. Stripping analysis of Cd and Pb under flow-injection analysis differential pulse stripping voltammetry on spiked samples at 5 μg/L with lead and cadmium. voltammetry on spiked samples at 5 µg/L with lead and cadmium. Cd Pb

Found (μg/L) Cd%Recovery Found (μg/L) Pb %Recovery Sample 1 Found4.80 (µg/L) %Recovery96% Found (4.99µg/L) %Recovery100% SampleSample 2 14.51 4.80 96%90% 4.994.60 100% 92% SampleSample 3 24.88 4.51 90%97% 4.604.81 92% 96% Sample 3 4.88 97% 4.81 96% However, it can already be estimated visually from Figure 9 that calibration curves for both Pb and Cd can be easily fitted by second‐order polynomials. Quantitation is, thus, possible up to concentration levels around a few hundreds of μg/L by exploiting the nonlinear calibration curve. This result will widen the application of flow injection stripping voltammetry since sample pretreatment and/or dilution is often unpractical for in situ analysis under a decentralized approach. The limits of detection were calculated following IUPAC guidelines; values around 0.35 μg/L for Pb for Cd were found, which are similar to those found for the stopped‐flow technique. A simple increase of the injected loop volume will immediately result in a decrease of the detection limits without any adverse effect thanks to the extended linearity of the flow injection technique.

Chemosensors 2018, 6, 37 12 of 13

However, it can already be estimated visually from Figure9 that calibration curves for both Pb and Cd can be easily fitted by second-order polynomials. Quantitation is, thus, possible up to concentration levels around a few hundreds of µg/L by exploiting the nonlinear calibration curve. This result will widen the application of flow injection stripping voltammetry since sample pretreatment and/or dilution is often unpractical for in situ analysis under a decentralized approach. The limits of detection were calculated following IUPAC guidelines; values around 0.35 µg/L for Pb for Cd were found, which are similar to those found for the stopped-flow technique. A simple increase of the injected loop volume will immediately result in a decrease of the detection limits without any adverse effect thanks to the extended linearity of the flow injection technique.

4. Conclusions Screen-printed electrodes have been demonstrated to play an increasing role in ultratrace inorganic chemistry, as already observed for organic analysis [15]. These results have been linked to a careful control of matrix chemistry, chloride ion activity in particular, and to the recent development of batch and flow cells for optimized hydrodynamic conditions during the preconcentration steps. Main results for batch, stopped-flow, and flow injection analysis are summarized in Table5.

Table 5. Analytical ﬁgures of merit for stripping analysis of Cd and Pb under different conditions.

Batch Stopped Flow Flow Injection Analysis Cd Pb Cd Pb Cd Pb Sensitivity (µA × L × µg−1) 0.2831 0.1681 0.1635 0.1202 0.1381 0.0909 Limit of Detection (µg/L) 0.17 0.19 0.58 0.62 0.32 0.35 Analytical Errors <5% <5% <10% <10% <10% <10% Analysis Time a 175 s 175 s 220 s Sample Volume 10–25 mL 1–2 mL 0.5 mL (loop volume) a preconcentration time: 60 s for batch and stopped ﬂow; 110 s at 0.5 mL loop for ﬂow injection analysis.

Batch analysis still shows the best analytical figures of merit, particularly in terms of sensitivity and limits of detection. Flow cells are, however, rapidly filling the gap; flow injection is confirmed to be superior to stopped flow, both in terms of signal-to-noise ratio and sample consumption. Moreover, sample throughput is significantly increased in flow cells since external calibration can be used as the preferred method of quantitation with respect to standard addition. Finally, the feasibility of using nonlinear calibration procedures for flow injection stripping voltammetry will open new and unexplored possibilities in screen-printed electrode/flow injection stripping voltammetry for in situ decentralized analysis in industrial environments. Provisions for mercury substitution in real working conditions are under investigation, as well as hardware and software developments to fulfill all analytical needs for decentralization.

Author Contributions: Conceptualization, C.D., D.M., A.P. and S.R.; Writing—original draft, C.D. Funding: Research was funded by Fondo di Ateneo per la Ricerca (FAR)—Università degli Studi dell’Insubria. Acknowledgments: Fabio Stropeni is gratefully acknowledged for his chemical expertise and for performing the laboratory experiments. The authors also wish to thank Bruno Giacomel and Flavio Villa of AMEL srl (Milano, Italy) for the electrochemical instrumentation and joint project collaboration, and PierPaolo Protti for scientific discussions. Conflicts of Interest: The authors declare no conflict of interest. Chemosensors 2018, 6, 37 13 of 13

References

1. Ochsenkühn-Petropoulou, M.; Ochsenkühn, K.M. Comparison of inductively coupled plasma-atomic emission spectrometry, anodic stripping voltammetry and instrumental neutron-activation analysis for the determination of heavy metals in airborne particulate matter. Fres. J. Anal. Chem. 2001, 369, 629–632. [CrossRef] 2. Wang, J. Analytical Electrochemistry, 3rd ed.; Wiley: Hoboken, NJ, USA, 2006. 3. Ozkan, S.A. Principles and Techniques of Electroanalytical Stripping Methods for Pharmaceutically Active Compounds in Dosage Forms and Biological Samples. Curr. Pharm. Anal. 2009, 5, 127–143. [CrossRef] 4. Economou, A.; Fielden, P.R. Applications, potentialities and limitations of adsorptive stripping analysis on mercury film electrodes. Trends Anal. Chem. 1997, 16, 286–292. [CrossRef] 5. Wang, J. Decentralized Electrochemical Monitoring of Trace Metals: From Disposable Strips to Remote Electrodes. Analyst 1994, 119, 763–766. [CrossRef] 6. Arduini, F.; Amine, A.; Moscone, D.; Ricci, F.; Palleschi, G. Fast, sensitive and cost-effective detection of nerve agents in the gas phase using a portable instrument and an electrochemical biosensor. Anal. Bioanal. Chem. 2007, 388, 1049–1057. [CrossRef][PubMed] 7. Arduini, F.; Cassisi, A.; Amine, A.; Ricci, F.; Moscone, D.; Palleschi, G. Electrocatalytic oxidation of thiocholine at chemically modified cobalt hexacyanoferrate screen-printed electrodes. J. Electroanal. Chem. 2009, 626, 66–74. [CrossRef] 8. Palchetti, I.; Laschi, S.; Mascini, M. Miniaturised stripping-based carbon modified sensor for in field analysis of heavy metals. Anal. Chim. Acta 2005, 530, 61–67. [CrossRef] 9. Gomez, V.S.; Barcelò, C.; Serrano, N.; Esteban, M. Antimony film screen-printed carbon electrode for stripping analysis of Cd(II), Pb(II), and Cu(II) in natural sample. Anal. Chim. Acta 2015, 855, 34–40. 10. Hocevar,ˆ S.B.; Ogorevc, B.; Wang, J.; Pihlar, B. A Study on Operational Parameters for Advanced Use of Bismuth Film Electrode in Anodic Stripping Voltammetry. Electroanalysis 2002, 14, 1707–1712. [CrossRef] 11. Romanus, A.; Mueller, H.; Kirsch, D. Application of adsorptive stripping voltammetry (AdSV) for the analysis of trace metals in brine. Fresenius J. Anal. Chem. 1991, 340, 371–376. [CrossRef] 12. Economou, A. Recent developments in on-line electrochemical stripping analysis—An overview of the last 12 years. Anal. Chim. Acta 2010, 683, 38–51. [CrossRef][PubMed] 13. Honeychurch, K.C. Screen-printed electrochemical sensors and biosensors for monitoring metal pollutants. Sci. J. 2012, 2, 1–51. [CrossRef] 14. Guell, R.; Fontas, C.; Aragay, G.; Merkoci, A.; Antico, E. Screen-printed electrodes incorporated in a flow system for the decentralized monitoring of lead, cadmium and copper in natural and wastewater samples. Int. J. Environ. Anal. Chem. 2012, 93, 1–12. [CrossRef] 15. Mozo, J.D.; Carbajo, J.; Sturm, J.C.; Nunez-Vergara, L.J.; Salgado, P.; Squella, J.A. Determination of Nifuroxazide by flow injection linear adsorptive stripping voltammetry on a screen-printed carbon nanofiber modified electrode. Electroanalysis 2010, 24, 676–682. [CrossRef] 16. Dossi, C.; Stropeni, F.; Monticelli, D. Screen Printed Electrode-flow stripping voltammetry for inorganic analysis. In Sensors-Lecture Notes in Electrical Engineering; Compagnone, D., Baldini, F., Di Natale, C., Betta, G., Siciliano, P., Eds.; Springer: Berlin, Germany, 2015; Chapter 25; Volume 319. 17. Pizeta, I.; Omanovic, D.; Branica, M. The influence of data treatment on the interpretation of experimental results in voltammetry. Anal. Chim. Acta 1999, 401, 163–172. [CrossRef] 18. Mocak, J.; Bond, A.M.; Mitchell, S.; Schollary, G. A statistical overview of standard (IUPAC and ACS) and new procedures for determining the limits of detection and quantification: Application to voltammetric and stripping techniques. IUPAC Technical Report 1997. Pure Appl. Chem. 1997, 69, 297–328. [CrossRef] 19. Jakubowska, M.; Piech, R.; Dzierwa, T.; Wcislo, J.; Kubiak, W.W. The evaluation method of smoothing algorithms in voltammetry. Electroanalysis 2003, 15, 1729–1736. [CrossRef] 20. Margoshes, M.; Burns, D.A. Treatise on Analytical Chemistry, Part 1-Theory and Practice, 2nd ed.; Elving, P.J., Mossotti, E.V.G., Kolthoff, I.K., Eds.; Wiley: New York, NY, USA, 1984; Volume 4, p. 413.

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).