The Effect of Glassy Carbon Surface Oxides in Non-Aqueous

Oct 3, 2014 - Department of Chemistry and Biochemistry, San Diego State ... data is made available by participants in Crossref's Cited-by Linking serv...
0 downloads 0 Views 878KB Size
Article pubs.acs.org/ac

The Effect of Glassy Carbon Surface Oxides in Non-Aqueous Voltammetry: The Case of Quinones in Acetonitrile Patrick A. Staley, Christina M. Newell, David P. Pullman, and Diane K. Smith* Department of Chemistry and Biochemistry, San Diego State University, San Diego, California 92182-1030, United States S Supporting Information *

ABSTRACT: Glassy carbon (GC) electrodes are well-known to contain oxygenated functional groups such as phenols, carbonyls, and carboxylic acids on their surface. The effects of these groups on voltammetry in aqueous solution are wellstudied, but there has been little discussion of their possible effects in nonaqueous solution. In this study, we show that the acidic functional groups, particularly phenols, are likely causes of anomalous features often seen in the voltammetry of quinones in nonaqueous solution. These features, a too small second cyclic voltammetric wave and extra current between the two waves that sometimes appears to be a small, broad third voltammetric wave, have previously been attributed to different types of dimerization. In this work, concentrationdependent voltammetry in acetonitrile rules out dimerization with a series of alkyl-benzoquinones because the anomalous features get larger as the concentration decreases. At low concentrations, solution bimolecular reactions will be relatively less important than reactions with surface groups. Addition of substoichiometric amounts of naphthol at higher quinone concentrations produces almost identical behavior as seen at low quinone concentrations with no added naphthol. This implicates hydrogen bonding and proton transfer from the surface phenolic groups as the cause of the anomalous features in quinone voltammetry at GC electrodes. This conclusion is supported by the perturbation of surface oxide coverage on GC electrodes through different electrode pretreatments.

G

lassy carbon (GC),1−3 along with platinum and gold, are the most commonly used solid electrode materials for electroanalytical studies. The advantages of glassy carbon are mainly that it is relatively inexpensive and has a large voltage window in aqueous solution. It is also useful for working with compounds with functional groups that strongly adsorb on platinum and gold. The latter quality is often why glassy carbon is used in organic solvents. The case in point is quinones, for which GC electrodes have become the electrode of choice in nearly every experiment since commercial GC electrodes became readily available.4−13 Although glassy carbon is thought of as relatively inert, oxygenated functional groups are well-known to exist on the electrode surface.1,2 The oxygenated groups include phenols and possibly other alcohols,14−17 ketones14,16−18 including quinones,19,20 and carboxylic acids.14 In some cases, these groups have been shown to promote electron transfer. Indeed, since the 1970s, numerous reports of different methods to “activate” GC and other carbon electrodes by purposely oxidizing the surface have appeared.14,17,19,21−25 In the 1990s and early 2000s, McCreery and co-workers did much to rationalize the maze of experimental observations, by using a variety of techniques to systematically vary GC surfaces and noting the effects that the changes had on the voltammetry of a series of test redox couples in aqueous solution.26−29 Work of a similar © 2014 American Chemical Society

nature has continued to the present decade as new types of carbon electrodes and new applications emerge.17 However, it is important to note that even without specific activation, these oxygenated functional groups are present on the surface of GC,1,2 and their presence can be expected to affect the voltammetry not just in aqueous solution but also in nonaqueous solution, particularly if acidic or basic analytes are present. While this has been well-discussed in the literature for aqueous solution, to date, there appears to have been very little discussion of the possible consequences of these surface groups being present in nonaqueous solution. The objective of this report is to show that the oxygenated functional groups on GC can indeed have a significant effect on voltammetry in nonaqueous solution, with the quinone redox couple providing an important example. In many respects, quinones are the prototypical organic redox couple. They were among the first organic redox couples studied by voltammetry30 and have been continuously re-examined throughout the development of modern electroanalytical chemistry.31,32 Quinones play essential roles in biology33−35 and have applications as pharmaceuticals,36−38dyes,39 and in the Received: August 23, 2014 Accepted: October 3, 2014 Published: October 3, 2014 10917

dx.doi.org/10.1021/ac503176d | Anal. Chem. 2014, 86, 10917−10924

Analytical Chemistry

Article

would reduce the flux of quinone from the bulk to the electrode surface at voltages in which the dianion was being formed at the electrode, leading to a decreased wave height. However, these authors also noted that they could not explain why disproportionation would not occur instead of dimerization, and they found that simulations of the CVs using this mechanism could not fully account for the experimental voltammograms. Macias-Ruvalcaba and co-workers11 suggested another mechanism to explain the unusually small second CV wave which involved an electroactive π-dimer formed by dimerization of the radical anions, Rxn 2 in Scheme 1. One electron reduction of this dimer at potentials near the reduction of the radical to the dianion will slightly decrease the current at the second wave since this process involves 1 e− per two quinones, compared to 1 e− per quinone. They used temperature dependent UV−vis spectroscopy to verify that π-dimerization occurs with 2,3-dichloro-5,6-dicyanobenzoquinone and digital simulation to show that this mechanism could explain the voltammetry for this particular quinone. The same authors found a similar mechanism operational for 5-hydroxy-naphthoquinone in acetonitrile.12 In the Macias−Ruvalcaba and Evans data, no broad CV wave between the two main waves is apparent, but if the radical dimerization was very favorable, their mechanism could also explain this wave. Indeed, Gupta and Linschitz suggested such a mechanism to explain the observation of this wave in their data. As pointed out recently by Gonzalez and co-workers,13 a similar wave is also observed when substoichiometric amounts of acids are added to quinones in nonaqueous solution. Furthermore, these workers found that a comparable CV wave was obtained upon oxidation of half-deprotonated hydroquinone. They suggested that this could be explained by the formation of a low-barrier H-bond complex between hydroquinone and the quinone dianion, which is oxidized and reduced, Rxn 2, at potentials between the two main CV waves.

manufacture of hydrogen peroxide.40 They also have been investigated for use in electronic materials and devices,41,42 including electrochemical capacitors.43 More specifically for our purposes, quinones, Q, are well-known to undergo a 2e−, 2H+ reduction in buffered aqueous solution (pH’s < ∼10) to give the hydroquinone, QH2, Rxn 1, but in aprotic solvents they generally undergo two separate 1e− reductions, first to the radical anion, Q−, and then to the dianion, Q2−, Scheme 1. The Scheme 1. Dimerization Reactions That Have Been Proposed to Explain Anomalous Features of Quinone Voltammetry in Aprotic Solvents

basicity of the quinone increases by 8−10 orders of magnitude with each electron transfer, with the result that the addition of protic guests has a dramatic (and well-studied) effect on the voltammetry of quinones in nonaqueous solvents. With very weakly acidic guests, H-bonding to the dianion leads to a positive shift in the second CV wave,4−7 but with more acidic guests proton transfer will occur, eventually leading, with strong enough acids and high enough concentrations, to a single 2e−, 2H+ reduction to the hydroquinone.4,8,9 However, even without the intentional addition of protic guests, the voltammetry of quinones in nonaqueous solvents often shows features that are not explained by the simple two-step mechanism. In particular, the second cyclic voltammetric (CV) wave is often clearly smaller than the first, even though both supposedly correspond to 1e− transfers. In addition, there is typically extra current between the two CV waves, which in some cases manifests as a small, broad CV wave in between the two main waves. Although the above discrepancies in quinone voltammetry are clearly apparent in numerous literature reports, only a few authors have actually mentioned them.4,10,11,13 Gupta and Linschitz4 appear to be the first to offer an explanation. They proposed that the small second CV wave could be due to the formation of an electroinactive dimer between the quinone dianion and the starting quinone, Rxn 1 in Scheme 1. This

Thus far, the main explanation for deviations in quinone voltammetry in nonaqueous solution, all of which have been observed at GC electrodes, is some form of dimerization. In order to explore this further, in this work we report the results of concentration dependent CV studies on several alkyl benzoquinones. If dimerization is at the root of the nonideal behavior, the CVs should become more ideal, or at least stay the same, as the quinone concentration is lowered, but, in fact, in all cases the CVs become even less ideal as the concentration is lowered, with the second wave decreasing dramatically in size and a new, broad CV wave growing in at potentials in between the two main quinone waves. Since the effect of functional groups on the electrode surface will become relatively more important at lower analyte concentrations, these results, supported by additional studies described within, argue strongly that it is actually the acidic surface groups on GC that are the main cause for the nonideal CVs for these quinones under our experimental conditions.



EXPERIMENTAL SECTION Chemicals. CH3CN was HPLC grade and dried by refluxing overnight with CaH2. Tetrabutylammonium hexafluorophosphate (NBu4PF6) was purified by recrystallization 10918

dx.doi.org/10.1021/ac503176d | Anal. Chem. 2014, 86, 10917−10924

Analytical Chemistry

Article

referenced procedure was impossible because our electrodes have Kel-F as insulation, which would melt under the literature hydrogenation conditions. Furthermore, a more in-depth study of GC hydrogenation found that to actually etch the electrode surfaceso that polishing with the deoxygenated slurry would not be neededit is necessary that the electrode reach a temperature of approximately 650 K,45 which would also melt Kel-F. In light of these concerns, hydrogenation was performed by modifying a procedure46 previously used to hydrogenate silicon in an ultrahigh vacuum chamber. In our modified version, we pumped the chamber down to 5.0 × 10−6 Torr and then backfilled with H2 to a pressure of 6.0 × 10−5 Torr, followed by running a current through a tungsten filament to heat it to approximately 1900 K while it was about 1 cm from the electrode surface. The temperature of the electrode was monitored by thermocouple, and the current to the filament was stopped when the electrode reached 50 °C; the electrode was then allowed to cool before the current was started again. The total time that the filament had current flowing through it was 30 min. This procedure is expected to produce a sufficient flux of H atoms at the electrode surface to at least partially remove the oxygenated functional groups. Digital Simulation. Simulated CVs were made using the fitting algorithm in DigiSim 3.03 (BAS). The parameters were fitted to the experimental CV using an iterative procedure by fitting the parameters in the following order: E° values, diffusion coefficients, and heterogeneous rate constants. The sequence was then repeated two more times. The transfer coefficient, α, was set to 0.5 for both electron transfers. Values were fit relative to the Ag quasi reference electrode and then rereferenced to Fc. In the end, the values attained were E° = −1.24 V vs Fc and ks = 39.0 cm/s for the first reduction, E° = −1.93 V vs Fc and ks = 0.0229 cm/s for the second reduction, D = 6.04 × 10−5 cm2/s for all species, and R = 350 Ω.

from 95% reagent alcohol three times followed by drying overnight at 100 °C under a vacuum. Vitamin K1 was used as received. 2-Naphthol and all of the benzoquinones were purified through sublimation. General Cyclic Voltammetry Procedure. Voltammetry was performed under nitrogen, in a drybox, in an undivided cell with three electrodes: 0.0407 cm2 glassy carbon (GC) working electrode, Pt wire counter electrode, and Ag wire quasi-reference electrode. The latter was housed in a septumsealed glass tube filled with electrolyte solution and connected to the rest of the cell via a vycor glass frit attached to the bottom of the tube with Teflon heat shrink tubing. For the majority of the experiments, the working electrode was polished in an atmosphere with 0.25 μm diamond polishing paste (Buehler), then 0.05 μm deagglomerated alumina slurry (Buehler), and rinsed thoroughly with D.I. water and acetone after each polish. We are aware that this procedure is not recommended for GC due to the possibility of contamination from other ingredients in the commercial polish.1,2 However, we chose to use this procedure for this study since it is the one that we have commonly used in our lab and believe it is commonly used by other laboratories when using GC electrodes for studying organic compounds in nonaqueous solution. A more rigorous polishing procedure that avoided the commercial slurries was used to prepare the “reduced” electrodes as described below. Since the same CV features were observed with these electrodes as those polished by the above standard method, these features cannot be caused solely by contaminants from the commercial pastes. The electrolyte solution for all of the studies consisted of 0.1 M NBu4PF6 in CH3CN and was made 40−60 mL at a time by adding the appropriate amount of CH3CN to NBu4PF6 by passing the CH3CN through a column of activated alumina. After preparation, the electrolyte solution was stored in the drybox until use. Once the cell was assembled, the working electrode was cycled through the potential range until the background stabilizedusually ∼2000 cycles. Analytes were added from highly concentrated stock solutions using microliter syringes. At the end of experiments, ferrocene (Fc) was added to the cell for use as an internal reference, and the CVs were later rereferenced to V vs Fc. Scan Rate and Concentration Dependence Studies. CVs were taken at multiple scan rates from 0.1 V/s to 2 V/s. Concentration dependence studies were performed between 0.010 mM and more than 1.0 mM, but most of the lower concentration scans were not used in our analysis due to an unacceptable signal-to-noise ratio and/or poor background subtractions. Electrode Oxidation and Reduction. “Oxidized” electrodes were prepared by first polishing normally, as described above. They were then placed in 0.1 M aqueous sulfuric acid and cycled 30 times between −0.52 and 1.83 V vs SCE at 0.15 V/s, followed by water and acetone rinses. “Reduced” electrodes were initially prepared by polishing with a 0.050 μm alumina powder slurried in deoxygenated cyclohexane. This procedure was reported to bring the oxygen content of the surface from 15% to 4%44 but was found by us to make the background unstable in acetonitrile, presumably due to slow, irreversible reactions with the analyte after the background scans were already completed. “Reduced” electrodes with stable backgrounds were produced by hydrogenation under high vacuum via modification of another published procedure.44 Direct use of the



RESULTS AND DISCUSSION Comparison of Pt, Au, and GC Electrodes. shows cyclic voltammograms (CVs) of 1 mM DQ NBu4PF6/CH3CN obtained using a gold disk electrode (red scan) and a Pt disk (black scan).

Figure 1 in 0.1 M working The first

Figure 1. Background-subtracted CVs of 1 mM duroquinone in 0.1 M NBu4PF6/CH3CN at 0.1 V/s. Black scan: Pt disk working electrode. Red scan: Au disk working electrode. Currents were normalized so as to give wave Ic a height of −1.0 for both electrodes. 10919

dx.doi.org/10.1021/ac503176d | Anal. Chem. 2014, 86, 10917−10924

Analytical Chemistry

Article

favoring the monomer intermediates over dimers. Moreover, the presence of large steric groups should also affect the current ratio of IIc to Ic because their presence would interfere with dimerization. Figure 3 shows an overlay of three scans of DQ taken at different concentrations in which the current has been

wave (Ic/Ia), corresponding to reduction to the radical anion, is reversible. However, in both cases the second wave (IIc/IIa), presumably corresponding to the reduction of the radical anion to the dianion, is too small and exhibits a large ΔEp, indicating very slow electron transfer kinetics. This behavior is typical for quinones with Pt or Au in nonaqueous solvents, which explains why these common electrodes are rarely used for quinones in nonaqueous solvents. In contrast to Pt and Au, use of glassy carbon (GC) working electrodes produces much more ideal CVs with quinones in nonaqueous solvents. This is seen for duroquinone (black scan) in Figure 2. The second wave (IIc/IIa) is clearly much closer to

Figure 3. Background-subtracted CVs taken with a GC electrode of different concentrations of duroquinone in 0.1 M NBu4PF6/CH3CN at 0.5 V/s. Currents have been normalized by dividing by the concentration of duroquinone in M and by the electrode area.

normalized by dividing by concentration to allow an easy qualitative comparison. What we see here is that the size of the second wave (IIc/IIa) shrinks as compared to the first wave (Ic/Ia) as the concentration decreases; this is the opposite of what should happen if dimerization of any kind were the cause of the unexpectedly small size. Moreover, the extra current between the two expected waves resolves into a third wave (IIIc/IIIa), and another completely unexpected oxidation peak (IVa) also grows in at a potential positive of wave one (Ic/Ia). Rather than dimerization, these results are consistent with something present at low concentration that is causing some of the DQ to go through redox reactions at potentials that are different from the bulk of the DQ. Hydrogen bonding and proton transfer have both been shown to change the potential of quinone electron transfer reactions,4−9 the extent to which this happens depending upon the pKa’s of the particular quinone and the proton donor and their concentrations. We will come back to this later. In order to further probe the role of dimerization in the quinone voltammetry, concentration dependent CVs were also obtained for 2,6-dimethylbenzoquinone (2,6-Me2BQ), 2,6-di-tbutylbenzoquinone (2,6-tBu2BQ), 2,5-dimethylbenzoquinone (2,5-Me2BQ), and 2,5-di-t-butylbenzoquinone (2,5-tBu2BQ). Both 2,6 isomers, as well as both 2,5 isomers, should have very similar electronic properties but different degrees of steric hindrance. The latter is apparent from DFT-calculated electrostatic potential surfaces of 2,6-Me2BQ and 2,6-tBu2BQ. These structures, provided in the Supporting Information, show that the electron surfaces of the methyl hydrogens do not protrude significantly further out than that of the π electrons and therefore do not present a significant hindrance toward π dimerization. On the other hand, the electron density associated with the t-butyl groups does extend significantly beyond that of the π electrons, so π dimerization should be

Figure 2. Background-subtracted CV taken with a GC electrode of 0.94 mM duroquinone in 0.1 M NBu4PF6/CH3CN at 0.5 V/s. Black scan: Experimental CV using a GC disk working electrode. Red scan: simulated CV assuming two sequential electron transfers. Simulation parameters are given in the Experimental Section.

ideal in this CV than those in Figure 1, explaining why GC is the electrode of choice for quinones in nonaqueous solvents. However, even in this CV the behavior is actually not completely ideal. The red scan shows the best fit simulated CV that could be obtained assuming a simple two step mechanism involving sequential electron transfer steps. Even with slower kinetics for the second electron transfer (a reasonable assumption) the second wave in the experimental CV is smaller than it should be. The current between the two waves in the experimental CV is also larger than in the simulated CV. As discussed, this behavior has been noted by several authors. Each has offered a slightly different explanation, but all of them involve some form of dimerization. What has not been discussed so far is the role that the GC electrode may be playing in this behavior. Concentration Dependence of Quinone CVs on GC. So far the explanation for the small second CV wave and the extra current between the two waves has been some version of dimerization. While dimerization has been verified experimentally in some cases, those cases have all been quinones which have hydroxyl groups or strong electron withdrawing groups or which form sigma dimers.11,12 In order to explain why this unexpected electrochemistry is present in quinones generally, the mechanism also must be present in quinones generally. If dimerization were the case, then the CV should become more idealthe current ratio between IIc and Ic should increaseas the concentration of DQ decreases, thereby 10920

dx.doi.org/10.1021/ac503176d | Anal. Chem. 2014, 86, 10917−10924

Analytical Chemistry

Article

CVs were obtained using much drier electrolyte solutions that had been stored over activated 3A molecular sieves, but this led to obvious precipitation of the dianion onto the electrode with these quinones, and hence even greater distortion of the CVs. (See the Supporting Information for an example.) The lack of precipitation in these CVs at 1 mM quinone concentrations suggests there must be at least one or two waters per quinone, which puts the lower limit of water at ∼2 mM. An attempt (see the Supporting Information) was made to estimate the actual water concentration based upon the difference between vitamin K1’s E1/2 values using Webster’s method,47 but the potential difference was outside the linear range of Webster’s equation. That gives an upper limit for the water concentration of ∼8 mM. However, as noted, the effect of water and other weak acids of similar strength on quinone voltammetry in aprotic solvents has been very well studied.4,5,7 The general trend is that upon addition of water, the second CV wave shifts positive in potential, toward the first, with very little change in height or wave shape. Thus, the presence of water cannot by itself explain the observed changes in the CV waves for quinones as the concentration decreases. As discussed, it is well-known that there are acidic functional groups such as phenols and carboxylic acids on the surface of GC as a result of polishing in air.1,2 With that in mind, proton transfer from the electrode surface to the quinone emerges as a possible explanation for the decreasing size of wave II in Figures 3 and 4 as the concentration is lowered, as well as the new redox wave (IIIc/IIIa) that grows in at lower concentrations. Since phenols are the most common protic groups on GC, CVs taken with 2-naphthol titrated into DQ solution at millimolar concentration were compared to the lower concentration DQ CVs.

hindered in the t-butyl substituted quinones compared to the methyl-substituted quinones. The results of the concentration dependent CV studies for all five quinones examined are summarized in Figure 4. The

Figure 4. Ratios of the peak current for IIc to Ic in the CVs of different benzoquinones as a function of quinone concentration. The CVs were taken with a GC electrode in 0.1 M NBu4PF6/CH3CN at 0.5 V/s. The peak currents used for this calculation are the absolute currents for the background-subtracted CVs. Each value is an average of three independent measurements; the error bars are ±1 standard deviation.

horizontal axis is the concentration in millimolarity, and the vertical axis is the ratio between the peak current of the second reduction peak (e.g., IIc in Figure 3) and the first reduction peak (e.g., Ic in Figure 3). Note that because of uncertainties in the baseline current for peak IIc due to the growth of peak IIIc, the currents used for this calculation are the absolute currents for the background-subtracted CVs. This explains why ratios greater than 1 are measured in some cases. Representative CVs used to obtain the data are provided in the Supporting Information. A dimer of any kind is less stable as the total concentration of its constituents decreases because the monomers take longer to find new partners once the dimer breaks up. This should lead the peak ratio to approach 1 as the concentration of quinone decreases if dimerization were the reason for the second wave being so small, but what actually happens is the opposite. Figure 4 shows quantitatively, and for five differently substituted quinones, that one of the salient features of Figure 3that as the concentration of quinone decreases, the ratio of peak IIc to peak Ic also decreasesis not only present in other quinones but is apparently unaffected by steric hindrance. In the case of the 2,5-isomers, there appears to be no significant difference between the methyl and t-butyl derivatives, and with the 2,6-isomers the ratio appears on average to be actually lower with the less bulky methyl substituents, which is opposite of the trend that would be expected if π dimerization was important. Put together, Figures 3 and 4 rule out dimerization as an explanation for the general trend of the quinone second reduction (IIc/IIa) being smaller than the first (Ic/Ia) and suggest that there is something present at moderately low concentration to cause some of the DQ to be reduced or oxidized at different potentials. That something might be water, since small amounts of water are present in these solutions.

Figure 5. Background-subtracted CVs taken with a GC electrode of 1.4 mM duroquinone in 0.1 M NBu4PF6/CH3CN at 0.5 V/s with additions of substoichiometric amounts of 2-naphthol.

Addition of 2-Naphthol. Figure 5 shows an overlay of high concentration DQ CV scans with 2-naphthol titrated in. The similarities between these CVs and those shown in Figure 3 of low concentrations of DQ and no added naphthol are striking. As has been described above, as the concentration of DQ is decreased (Figure 3), the second CV wave (IIc/IIa) decreases in size, and the third redox wave (IIIc/IIIa) grows in, as does a new oxidation peak (IVa). As 2-naphthol is titrated into the high concentration DQ solution (Figure 5), we also see 10921

dx.doi.org/10.1021/ac503176d | Anal. Chem. 2014, 86, 10917−10924

Analytical Chemistry

Article

the second wave (IIc/IIa) decrease and an intermediate wave (IIIc/IIIa) grow in along with a new oxidation peak (IVa) positive of the first wave. It is evident from these data that addition of substoichiometric amounts of naphthol causes very similar changes in the voltammetry of DQ in acetonitrile as does going to low concentrations of DQ in the absence of naphthol. Since the only phenol groups present under the latter conditions are those on the surface of GC, and these groups will become relatively more abundant as the solution concentration decreases, these data strongly implicate the surface phenol groups on GC as the cause of the concentration dependent changes in the CVs of DQ and the other quinones in acetonitrile. Perturbation of Surface Oxygenated Groups. If it is indeed the acidic groups on the electrode surface that are the source of the third redox wave (IIIc/IIIa, Figures 3 and 5) and an explanation for the small size of the second redox wave (IIc/IIa, Figures 3 and 5), then it should be possible to perturb both waves by reducing or oxidizing the electrode surface and thereby changing the number of acidic groups on the surface. Figure 6 shows a comparison of CVs at the same low concentration of DQ (0.040 mM) in which the electrode was prepared in three different ways: oxidized electrode surface, normal preparation, and reduced electrode surface (see the Experimental Section for details). The oxygen content of the electrode surface determines the acidic group concentration since the acidic groups on the electrode surface are oxygen based, so by controlling the oxygen content we can also control the surface coverage of acidic groups. In Figure 6b, the normally polished electrode gives a CV at this low DQ concentration which does not include wave two (IIc/IIa) but does include wave three (IIIc/IIIa) and a small, broad peak IVa. Figure 6a, in which the electrode has been oxidized, also shows no wave two and a large wave three, but in addition, peak Ia is now obviously smaller than peak Ic and peak IVa has grown substantially in size. These changes suggest further protonation, possibly because more potent acids, such as carboxylic acids, are now present. In addition, the background subtraction is substantially worse with the oxidized electrode in Figure 6a than the other cases, indicating that the electrode surface is generally more reactive and less stable after the oxidative treatment. Figure 6c, in which the electrode has been reduced, shows that redox waves two and three are both present, indicating that whatever caused DQ to be reduced at redox wave three rather than wave two has decreased in concentration along with the decrease in the oxygen and therefore protic group content. Therefore, it can be concluded that something about the oxidation state of the surface is responsible for the small size of the second quinone wave, and the most obvious candidates are hydrogen bonding and protonation from the surface oxides. Possible Mechanisms. So how do substoichiometric amounts of phenol, either added to solution or on the electrode surface, cause the observed changes in the voltammetry of the quinones? One hypothesis, based on the work of Gonzalez and co-workers with hydroquinone,13 is that the phenols are strong enough acids to protonate the quinone radical anion giving the neutral radical, which, being easier to reduce than the starting quinone, is immediately reduced by a second electron to give the hydroquinone. The hydroquinone then H-bonds to quinone radical anions formed after the available acid is consumed. Wave III would then be due to reduction of the radical anion in this H-bond complex to give their proposed

Figure 6. Background-subtracted CVs of 0.040 mM duroquinone in 0.1 M NBu4PF6/CH3CN at 0.5 V/s with different amounts of surface oxidation on the GC electrode. (a) Highly oxidized surface, (b) normally polished surface, (c) reduced GC surface. See the Experimental Section for details on the electrode preparation.

low barrier, dianionic H-bond complex as in Rxn 2. Note that this reaction does not require the protonation of the quinone radical anion to be favorable, it just needs to be fast enough that all the available acid is consumed at the potential of the first wave. The overall sequence of reactions, given by rxns 3−8, corresponds to 1 phenol consumed per quinone. This is consistent with the data in Figure 5 in which wave II is fully 10922

dx.doi.org/10.1021/ac503176d | Anal. Chem. 2014, 86, 10917−10924

Analytical Chemistry

Article

gone after 1 equiv of naphthol has been added. At less than 1 equiv of napthol, not all of the quinone radical anion would be consumed by rxns 4−8, leaving some to be reduced to the dianion in wave II. On the return scan in the CVs, the reactions in principle could simply follow the reverse reactions in rxns 8−3 to go back to the starting state. However, this would require that the reverse proton transfer from QH2 to ArO− be very fast so that this can occur at the potential of wave I. To the extent that this is not true, some of the oxidation would occur at more positive potentials, giving an explanation for the appearance of peak IVa. 2Q + 2e− = 2Q− (wave I)

(3)

Q− + ArOH = QH + ArO−

(4)

QH + e− = QH− (wave I)

(5)

QH− + ArOH = QH 2 + ArO−

(6)

QH 2 + Q− = QHQH− (formation of H bond)

(7)

QHQH− + e− = QHQH2 − (wave III)

(8)

The mechanism summarized by rxns 9−12 would occur in the context of the “wedge scheme” framework that we have recently introduced to explain the voltammetry of a phenylenediamine-based redox couple.47 A more detailed mechanistic investigation of the DQ-napthol system in this light is underway, and the results will be reported in due course. For the purpose of this study, the main point is that there are at least two mechanisms that could explain how small amounts of phenols, either added to solution or present on the electrode surface, could lead to voltammograms of quinones in which the second CV wave is too small and there is extra current in between the two main CV waves.



CONCLUSIONS This study provides compelling evidence that acidic and basic groups on the surface of glassy carbon electrodes can undergo proton transfer with electrogenerated species in solution, resulting in observable effects on the voltammetry. This point is illustrated with the case of quinones, which typically show irregularities in their CVs at GC electrodes in nonaqueous solution. In particular, it is often observed that the second CV wave is smaller than it should be, and there is extra current in between the two waves; in some cases, this extra current actually appears to be a small, third CV wave in between the two main quinone waves. The only explanation for these unusual features so far has involved some form of dimerization, but in this study we have ruled out this possibility for a series of alkylbenzoquinones since these features become more prominent rather than less as the concentration is decreased. Titrating naphthol into high concentrations of duroquinone also enhances these features, resulting in CVs that look very similar to those obtained with GC electrodes at low DQ concentrations and no added naphthol. This leads to the conclusion that proton transfer from surface phenols is the cause of the decrease in the second CV wave and growth of the intermediate wave as the quinone concentration is lowered, a conclusion that is reinforced by the effects of purposely oxidizing or reducing the GC surface. It should be noted that this study by no means rules out the possibility of dimerization occurring with some quinone radicals. Indeed, both π and σ dimerization of the radical ions of quinones11,12 and related species48,49 is well-documented. Instead, we are simply pointing out that dimerization does not explain all the irregularities observed with quinone voltammetry and that reaction with surface groups on GC electrodes is another cause of irregular behavior that should be considered. Looking at the concentration dependence of the CVs is a good way to distinguish between these since dimerization will become more important at higher concentrations, whereas reaction with surface groups will be more important at lower concentrations. The actual concentration where the latter becomes important will depend very much on the condition of the GC electrode used. This point nicely explains the large variation in both the relative heights of the first and second CV waves and the thickness between the waves that can be seen for the same quinone with GC electrodes under apparently similar experimental conditions in different publications and, in some cases, even the same publication. A final point worth mentioning is that the nature of the effect that we report in this work is fundamentally different from the effects that have been reported for the oxygenated functional groups on GC in aqueous solution. In our case, it is the acid− base character of these groups that is relevant. While this could be important in unbuffered aqueous solution as well, in the vast

Although the above mechanism explains the main features of the CVs of DQ with 2-naphthol, there are some inconsistencies. First, the mechanism corresponds to a net 1.5 e− per DQ (3 e− per 2 DQ) being transferred in wave I in the presence of 1 equiv naphthol, which means the current in wave I should increase as naphthol is titrated in. There is some increase, but it is less than that expected based on the stoichiometry. There is extra current observed past peak Ic, so it is possible that some of the missing current shows up past the peak due to slow proton transfer. More troubling however is the return scan. Breakup of the H-bond complex following oxidation (peak IIIa) gives QH2 and Q− in equal amounts. If peak IVa corresponds to oxidation of QH2 (likely H-bonded to ArO−), then it really should be of equal size to peak Ia (due to oxidation of Q−). Clearly it is much smaller, which would require the reverse reactions in rxns 6−4 to be unexpectedly rapid so that oxidation of most of QH2 occurs at the potential of wave I. A final problem with this mechanism is that for it to also be occurring with the surface phenols on GC, a high enough density of phenols would be required so that two protons could be efficiently transferred to a single DQ. An alternative mechanism, which is actually much simpler and avoids some of the issues mentioned above, is given by rxns 9−12. In this mechanism, the phenol is not sufficiently acidic to protonate the quinone radical anion, but it can H-bond with it to form ArOHQ−, which can be further reduced in wave III to ArOHQ2−. The reduced H-bond complex can break apart with the proton transferred to give QH− and ArO−, rxn 12. Peak IVa would be due to oxidation of QH−, which has not had time to recombine with ArO− to get back to ArOHQ2−. The relative heights of peak Ia and IVa would then be governed by the rate of the reverse rxn 12. Q + e− = Q− (wave I)

(9)

Q− + ArOH = ArOHQ−

(10)

ArOHQ− + e− = ArOHQ2 − (wave III)

(11)

ArOHQ2 − = ArO− + QH−

(12) 10923

dx.doi.org/10.1021/ac503176d | Anal. Chem. 2014, 86, 10917−10924

Analytical Chemistry

Article

(14) Cabaniss, G. E.; Diamantis, A. A.; Murphy, W. R.; Linton, R. W.; Meyer, T. J. J. Am. Chem. Soc. 1985, 107, 1845−1853. (15) Collier, W. G.; Tougas, T. P. Anal. Chem. 1987, 59, 396−399. (16) Ray, K. G.; McCreery, R. L. J. Electroanal. Chem. 1999, 469, 150−158. (17) Roberts, J. G.; Moody, B. P.; McCarty, G. S.; Sombers, L. A. Langmuir 2010, 26, 9116−9122. (18) Tougas, T. P.; Collier, W. G. Anal. Chem. 1987, 59, 2269−2272. (19) Engstrom, R. C.; Strasser, V. A. Anal. Chem. 1984, 56, 136−141. (20) Barbero, C.; Silber, J. J.; Sereno, L. J. Electroanal. Chem. 1988, 248, 321−340. (21) Taylor, R. J.; Humffray, A. A. J. Electroanal. Chem. 1973, 42, 347−354. (22) Blaedel, W. J.; Jenkins, R. A. Anal. Chem. 1975, 47, 1337−1343. (23) Engstrom, R. C. Anal. Chem. 1982, 54, 2310−2314. (24) Kelly, R. S.; Weiss, D. J.; Chong, S. H.; Kuwana, T. Anal. Chem. 1999, 71, 413−418. (25) Heien, M. L. A. V.; Phillips, P. E. M.; Stuber, G. D.; Seipel, A. T.; Wightman, R. M. Analyst 2003, 128, 1413−1419. (26) Chen, P.; Fryling, M. A.; McCreery, R. L. Anal. Chem. 1995, 67, 3115−3122. (27) Chen, P.; McCreery, R. L. Anal. Chem. 1996, 68, 3958−3965. (28) DuVall, S. H.; McCreery, R. L. Anal. Chem. 1999, 71, 4594− 4602. (29) DuVall, S. H.; McCreery, R. L. J. Am. Chem. Soc. 2000, 122, 6759−6764. (30) Muller, O. H.; Baumberger, J. P. Trans. Electrochem. Soc. 1937, 71, 169−180. (31) Chambers, J. Q. In Chemistry of the Quinoid Compounds; Patai, S., Rappoport, Z., Eds.; John Wiley & Sons: New York, 1988; Vol. 2, pp 719−757. (32) Quan, M.; Sanchez, D.; Wasylkiw, M. F.; Smith, D. K. J. Am. Chem. Soc. 2007, 129, 12847−12856. (33) Hirst, J. Biochem. J. 2009, 425, 327−339. (34) Hervas, M.; Navarro, J. A.; De la Rosa, M. A. Acc. Chem. Res. 2003, 36, 798−805. (35) Shearer, M. J. Lancet 1995, 345, 229−234. (36) de Castro, S. L.; Emery, F. S.; da Silva Junior, E. N. Eur. J. Med. Chem. 2013, 69, 678−700. (37) Tandon, V. K.; Chhor, R. B.; Singh, R. V.; Rai, S. J.; Yadav, D. B. Bioorg. Med. Chem. Lett. 2004, 14, 1079−1083. (38) Chen, Y.; Hu, L. Q. Med. Res. Rev. 2009, 29, 29−64. (39) Hattori, M. In Kirk-Othmer Encyclopedia of Chemical Technology, 5th ed.; Seidel, A., Ed.; John Wiley & Sons: Hoboken, NJ, 2004; Vol. 9, pp 300−349. (40) Eul, M.; Moeller, A.; Steiner, N. In Kirk-Othmer Encyclopedia of Chemical Technology, 5th ed.; Seidel, A., Ed.; John Wiley & Sons: Hoboken, NJ, 2004; Vol. 14, pp 42−51. (41) Mamada, M.; Kumaki, D.; Nishida, J.-i.; Tokito, S.; Yamashita, Y. ACS Appl. Mater. Interfaces 2010, 2, 1303−1307. (42) Letizia, J. A.; Cronin, S.; Ortiz, R. P.; Facchetti, A.; Ratner, M. A.; Marks, T. J. Chem.Eur. J. 2010, 16, 1911−1928. (43) Pognon, G.; Brousse, T.; Demarconnay, L.; Belanger, D. J. Power Sources 2011, 196, 4117−4122. (44) Kuo, T.-C.; McCreery, R. L. Anal. Chem. 1999, 71, 1553−1560. (45) Guttler, A.; Zecho, T.; Kuppers, J. Carbon 2004, 42, 337−343. (46) Kottke, M.; Neuburger, M. L.; Fee, D.; Beyer, J.; Pullman, D. P. Mol. Phys. 2012, 110, 1953−1966. (47) Hui, Y. L.; Chng, E. L. K.; Chua, L. P. L.; Liu, W. Z.; Webster, R. D. Anal. Chem. 2010, 82, 1928−1934. (48) Clare, L. A.; Pham, A. T.; Magdaleno, F.; Acosta, J.; Woods, J. E.; Cooksy, A. L.; Smith, D. K. J. Am. Chem. Soc. 2013, 135, 18930− 18941. (49) Macias-Ruvalcaba, N. A.; Evans, D. H. J. Electroanal. Chem. 2011, 660, 243−246. (50) Macias-Ruvalcaba, N. A.; Evans, D. H. J. Phys. Chem. C 2007, 111, 5805−5811.

majority, if not all, of the aqueous studies focusing on GC surface groups, either strongly acidic, basic, or buffered aqueous solutions were used. Under these conditions, the overall reaction that will occur will be determined by the bulk pH and the acid−base character of the surface becomes a moot point. Instead, it is either the electrocatalytic properties of the surface functional groups or their ability to promote adsorption onto the electrode surface that is the source of their effect on the voltammetry. While we have no evidence of these issues being important for quinones in acetonitrile, it is certainly possible that these types of effects could also occur in nonaqueous solution. In conclusion, the presence of oxygenated functional groups on the surface of GC should be kept in mind when working with GC electrodes in nonaqueous solution as well as aqueous solution. In both cases, these groups could affect interactions between dissolved species and the electrode surface resulting in increased interactions that can either enhance or hinder electron transfer. In nonaqueous solution, there is the additional possibility that H-bonding and proton transfer could occur, resulting in completely different reaction pathways, with different products.



ASSOCIATED CONTENT

S Supporting Information *

ESP surfaces for quinones, additional CVs of quinones at different concentrations, evaluation of the water content of the solvent, and the effect of superdry conditions. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS This work was supported by the National Science Foundation (CHE-1214151). REFERENCES

(1) McCreery, R. L. Chem. Rev. 2008, 108, 2646−2687. (2) McCreery, R. L. In Electroanalytical Chemistry; Bard, A. J., Ed.; Marcel Dekker: New York, 1991; Vol. 17, pp 221−374. (3) Van der Linden, W. E.; Dieker, J. W. Anal. Chim. Acta 1980, 119, 1−24. (4) Gupta, N.; Linschitz, H. J. Am. Chem. Soc. 1997, 119, 6384−6391. (5) Gomez, M.; Gonzalez, F. J.; Gonzalez, I. Electroanalysis 2003, 15, 635−645. (6) Gomez, M.; Gomez-Castro, C. Z.; Padilla-Martinez, I. I.; Martinez-Martinez, F. J.; Gonzalez, F. J. J. Electroanal. Chem. 2004, 567, 269−276. (7) Tessensohn, M. E.; Hirao, H.; Webster, R. D. J. Phys. Chem. C 2013, 117, 1081−1090. (8) Macias-Ruvalcaba, N. A.; Gonzalez, I.; Aguilar-Martinez, M. J. Electrochem. Soc. 2004, 151, E110−E118. (9) Astudillo, P. D.; Tiburcio, J.; Gonzalez, F. J. J. Electroanal. Chem. 2007, 604, 57−64. (10) Lehmann, M. W.; Evans, D. H. J. Electroanal. Chem. 2001, 500, 12−20. (11) Macias-Ruvalcaba, N. A.; Felton, G. A. N.; Evans, D. H. J. Phys. Chem. C 2009, 113, 338−345. (12) Macias-Ruvalcaba, N. A.; Evans, D. H. J. Phys. Chem. C 2010, 114, 1285−1292. (13) Astudillo, P. D.; Valencia, D. P.; Gonzalez-Fuentes, M. A.; DiazSanchez, B. R.; Frontana, C.; Gonzalez, F. J. Electrochim. Acta 2012, 81, 197−204. 10924

dx.doi.org/10.1021/ac503176d | Anal. Chem. 2014, 86, 10917−10924