Anal. Chem. 1995,67, 993-998
This Research Contribution is in Commemoration of the Life and Science of 1. M. Kolthoff (1894- 1993).
Solid Phase Extraction in Conjunction with Solution or Solid State Voltammetry as a Strategy for the Determination of Neutral Organic Compounds James A. Cox,* Kathryn S. Alber, Carrie A. Brockway, Mark E. Tess, and Waldemar Gorski Department of Chemistry, Miami University, Oxford, Ohio 45056
Solid phase extraction by a perfluorosulfonate,Ndon, of various aromatic amines and dipeptides, phenylalanine, and selected N-nitrosamineswas found to occur by both ion-exchange and nonspecific interactions. The latter was not observed at an ionomer based on sulfonated styrene. A polymeric ruthenium oxide film retained its electrocatalytic activity in the presence of a Nation overlayer. The combination of solid phase extraction and electrocatalysis provided a new strippingvoltammetric method by which such compounds as 4-nifroso-N,N-diethyladhe,1-amino2-phenylethane,cysteine-glycine, and N-nitrosodiphenylamine (NNDPhA) were determined at the submicromolar level by linear calibration curves. Both the extraction and catalysis chemistry operated for gas phase analytes, thereby allowing the design of an amperometric sensor for gases without a solution phase. Linear calibration curves at the micromole level for methanol and the nanomole level for NNDPhA in a 15-mL chamber were obtained. The development of electrochemical methods for the determination of neutral organic compounds in air or water requires addressing two problems. First, the levels of the analytes often are in a range where a preconcentration step is necessary. Second, most of these compounds are not electroactive in the potential window available at common electrodes, a situation that is exacerbated by the need to avoid the reduction of oxygen in applications such as monitoring with remote sensors. Moreover, virtually all studies on air samples have involved a transfer to the aqueous phase prior to electroanalysis. Performing a preconcentration directly on the electrode is an attractive approach to improving the sensitivity of electrochemical methods. If these analytes are in the ionic state in the sample solution, they are readily sorbed by an ionomer film on the electrode surface. Indeed, the Hoffmeister series of affinities of ions for charged sites on organic polymers suggests the sorption of organic cations by sulfonate even in the presence of high concentrations of inorganic cations. This prediction has been (1) Szentirmay, M. N.: Martin, C. R Anal. Chem. 1984,56,1898-1902. (2) Nagy, G.; Gerhardt, G. A; Oke, A. F.; Rice, M. E.; Adams, R N.: Moore, R B., III; Szentirmay, M. N.: Martin, C. R J. Electroanal. Chem. 1985,188, 85-94. (3) Capella, P.; Ghasemzadah, B.; Mitchell, IC; Adams, R N. Electroanalysis 1990,2, 175-182.
0 1995 American Chemical Society 0003-2700/95/0367-0993$9.00/0
verified on such compounds as methyl viologen and however, few applications to electroanalysis have appeared. Preconcentration of neutral organic compounds at polymercoated electrodes, a topic of the present study, has not been reported previously. A significant complication is that charge transfer considerations favor the use of ion-conducting Elms for the preconcentration medium. Therefore, solid phase extraction of neutral compounds by ionomers is required. Neutral organic compounds have been preconcentrated at bare electrodes as part of an electroanalytical procedure. The method, adsorptive stripping voltammetry, has received a great deal of study recently.4-6 In this method, the stripping current is proportional to the surface excess of electroactivecompounds on the electrode after a prescribed time of exposure to the sample. The adsorption isotherm governs the shape of the calibration curve and, more important, the maximum amount of analyte that can be sorbed to the electrode. Perhaps the greater challenge in extending stripping voltammetry to organic compounds is the requirement for electroactivity of the analyte in the sorbed state. Electrochemical activity of organic compounds very often requires a catalyst. Adsorptive stripping voltammetry using an electrode modified by a film of a catalyst has not been reported for cases where the analytes are organic compounds, and electrochemistry at surfaces that are modified by both a catalytic film and a polymeric overlayer has been the subject of few reports. The latter studies have dealt with small molecules as analytes. For example, electrocatalysisof the reduction of dioxygen by cobalt porphyrins which were coated with poly(ethy1ene oxide) or Nafion8 films has been observed, and a nitric oxide sensor which has a Nation film over a polymeric porphyrin that serves as the electrocatalysthas been developed and a ~ p l i e d .The ~ present study differs from these examples by including the preconcentration of the analyte as well as by considering larger molecules. Of particular interest to our program are N-nitrosamines. These potent carcinogens have been determined electrochemically on the basis of their reduction at a mercury electrode.lO,llAs our goal is to develop electrochemical monitors for these compounds (4) Paletek, E.; Postbieglovl, I. J Electroanal. Chem. 1986,214, 359-371. (5) mores, J. R; Smyth, M. R 1.Electroanal. Chem. 1987,235,317-326. (6) Wang, J.; Tapia, T. Bioelectrochem. Bioenev. 1988,19, 39-47. (7)Bettelheim, A.; Reed, R; Hendricks, N. H.; Collman, J. P.; Murray, R W.J Electroanal. Chem. 1987,238,259-276. (8) Harth, R; Mor, U.; Ozer, D.; Bettelheim, A. J. Electrochem. SOC.1989,136, 3863-3867. (9) Malinski, T.; Taha, Z. Nature 1992,358,676-678.
Analytical Chemistry, Vol. 67, No. 5, March 1, 1995 993
in air and water, the interference on the determination by oxygen cannot be tolerated. N-Nitrosamines are not oxidized in aqueous medium at conventional electrodes; however, we recently reported that their oxidation is mediated by a film of a mixed-valent ruthenium oxide polymer that is stabilized by cyano cross-links (mvRuCN).I2 These compounds generally are neutral; for example, the pK, of N-nitroso-n-propylamine is -0.85.13 Hence, the development of a stripping voltammetric method for N-nitrosamines requires a sigdicant extension of the reported methodolOB.
Finally, the present study includes a preliminary evaluation of a voltammetric sensor based on the ionomer-coated mvRuCN electrode for the measurement of organic compounds in air. Voltammetry in the absence of a solution phase can be accomplished as long as an ionic conductor contacts both the working and reference electrodes. Except for systems related to solid electrolyte batteries and fuel cells, such voltammetric systems have been the subject of few reports. In an example that illustrates the transition between solution and "dry" voltammetry, Chidsey et al. investigated the characteristics of interdigitated array electrodes that were coated with electroactive polymers and discussed the use of these electrodes for electrochemical studies in the absence of a solution phase (see ref 14 and citations therein). Ghoroghchian et al.I5demonstrated a faradaic response of gas phase species at an ultramicroelectrode that was separated by thin layers of glass and epoxy from a quasireference electrode in a biamperometric cell. Most systems with potential applicability to analytical chemistry have used ion-conductingpolymers as the electrolyte phase and the host for the analyte. Typical examples are films of PEO that contained dissolved metal salts73J7 and films of Nation.'* In the previous report that is most related to the present study, Kulesza and Fauknerlg demonstrated the voltammetric oxidation of methanol in a cell that used an anionexchange membrane that was loaded with a ruthenium-based catalyst as the electrolyte. The potential for this system to function as a sensor for methanol in the gas phase was mentioned, but the influence of concentration on the signal was not reported. Here, calibration data for gas phase methanol and for N-nitrosamines are shown, and a strategy for the measurement of N-nitrosamines in the presence of other oxidizable gases is described. EXPERIMENTAL SECTION
Reagents and Materials. The nitrosamines, phenylalanine, and peptides were from Sigma, and the 1-amino-2-phenylethane (AF'hA) was from Aldrich. They were used without further purification. The ruthenium salts used to prepare the solution from which the catalyst was plated, RuClyH20 and &Ru(CN)&HzO, were from Pfaltz and Bauer, Inc. and Johnson Matthey Electronics, respectively. Films were cast with Nafion (1100 EW) that was purchased from Aldrich as a 5%(wt) solution (10)Hasebe, IC; Osteryoung, J. Anal. Chem. 1975,47,2412-2418. (11) Samuelson, R;O'Dea, J.; Osteryoung, J.Anal. Chem. 1980,522215-2216. (12)Gorski, W.;Cox, J. A Anal. Chem. 1994,66,2771-2774. (13) Pulidori, F.; Borghesani, G.; Bighi, C.; Pedriali, R J. Electyoanal. Chem. 1970,27,385-396. (14)Chidsey, C. E.;Feldman, B. J.; hndgren, C.; Murray, R W. Anal. Chem. 1986,58,601-607. (15) Ghoroghchian, J.; Sarfarazi, F.; Dibble, T.; Cassidy, J.; Smith,J. J.; Russell, A; Dunmore, G.; Fleishmann, M.; Pons, S.Anal. Chem. 1986,58,22782282. (16)Oliver, B. N.;Egekeze, J. 0.; Murray, R W.J Am. Chem. SOC.1988,110, 2321-2322. (17) Reed, R A;Geng, L.; Murray, R W.J Electroanal. Chem. 1986,208,185193. (18)Parthasarathy, A.;Martin, C. R; Srinivasan, S . J Electrochem. SOC.1991, 138, 916-921. (19)Kulesza, P. J.; Faulkner, L. R /. Electrochem. Soc. 1993, 140,L66-L68. 994 Analytical Chemistry, Vol. 67, No. 5, March 1, 7995
in lower aliphatic alcohols and water (10%). Solid phase extraction experiments were performed with Nation NR50 beads from Aldrich. They were 10-35 mesh with an equivalent weight of 1250 and an ion-exchangecapacity of 0.8 mequiv/g. Prior to use, 1.0 g of the beads was soaked for 1h in 20 mL of the appropriate buffer, separated from the conditioning buffer by filtration, and rinsed with water. All other chemicals were reagent grade and were used without further purikation. House-distilled water that was puriiled with a Sybron/Bamstead NANOpure cartridge system was used throughout this work.
Experimental Techniques, Instrumentation, and Procedures. Cyclic voltammetry and chronoamperometry were performed with an EG&G PAR Model 273A potentiostat/galvanostat controlled by their Model 270 software. Electrochemical experiments (solution studies) were performed in a conventionalthreeelectrode cell with a Smm-diameter glassy carbon electrode (Bioanalytical Systems, Inc.) modified as described below. The reference electrode was Ag/AgCl, 3 M NaCl. All solution electrochemistry experiments are reported relative to that reference. Solid state electrochemistry was performed in the twoelectrode mode with an interdigitated microsensor electrode @ME)from AAI-Al3TECH (Yardley, PA). The IME was composed of two sets of 50 Pt fingers with dimensions of 10 pm x 5 mm and a spacing of 10 pm between the fingers. Prior to use, both sets of fingers were modified as described below. One set of the fingers served as a quasireference electrode and the other (that at the more positive potential during measurements) as the indicator electrode. The glassy carbon electrode was modified by depositing the catalyst and casting an overlayer of Nation. First, a film of mixedvalent ruthenium oxide with cyano cross-links (mvRuCN), the nature of which was previously elucidated,20121 was electrochemically polymerized on the surface from a 2 mM RuC13,2 mM &Ru(CN),j, 0.5 M KCl mixture at pH 2.0. The procedure, which is a minor variation of one we previously comprised an initial polishing of the surfaces, sonicating in water for 5 min, and electroplating by cyclic voltammetry for 20 cycles between 500 and 1100 mV at 50 mV s-' unless otherwise stated. The modified electrode was air-dried and coated with a film of Nafion. These films were cast with 10 p L of a 2.5%Nafion solution, a procedure that results in a thickness of about 5 pm.23 The IMEs were coated with mvRuCN by the same procedure except that the electrodes were cleaned by sequential rinses in hexane, 2-propanol, and water and were pretreated by cyclic voltammetry using 30 cycles at 50 mV s-l between -230 and 1200 mV in 0.5 M H2S04. The Nation overlayer was cast by air-drying the array, placing the entire assembly for 5 min in a closed chamber that was maintained at 80%relative humidity, and adding 10 p L of a 0.5%Nafion solution to the array surface. The estimated thickness of the Nafion layer was 0.5 pm23 The system was exposed to the ambient atmosphere for 4 min to allow the solvent to evaporate, and 0.4 mL of a 0.05 M W04,O.Ol M HzS04 mixture was place on the film. The water was removed by air-drying for 1 h. Because the mvRuCN did not cover the space between the electrode fingers, the Nafion provided the only conductance pathway across this two-electrode cell. The modified IMEs were used as sensors for methanol and N-nitrosamine vapors in a gastight glass chamber (15mL volume) that was fitted with a rubber septum and equipped with a Vaisala humidity meter (Cole Parmer). To control the possibility of an influence of humidity on the response, all of the reported (20)Cox, J. A;Kulesza, P. J. Anal. Chem. 1984,56,1021-1025. (21)Kulesza, P. J. J. Electyoanal. Chem. 1987,220,295-309. (22)Cox, J. A;Gray, T. J. Anal. Chem. 1989,61, 2462-2464. (23)Mauritz, K. A;Mora, C. J.; Hophger, A Polym. Prepr., Am. Chem. SOC. Diu. Polym. Chem. 1978,19,324-335.
experiments were performed at 96-98% relative humidity. However, it should be noted that a significant influence of humidity was not observed. A gas syringe was used to introduce controlled volumes of the vapors taken from bottles closed with a rubber septum. Solid phase extraction studies were performed by mixing 1.0 g of the preconditioned Nafion beads with 20-mL solutions of the analytes in 0.05 M buffers (PH 3, formate; pH 4-5, acetate; pH 6-9, phosphate and After 60 min, the mixtures were filtered, and the residual concentrations of the analytes in the filtrates were determined. Except for the study on the dipeptide, cysteineglycine,the determinations were made by measuring the absorbances with a Hewlett Packard Model 8452A diode array spectrophotometer and fitting the data to calibration curves prepared in buffers identical to those used in the extraction step. The dipeptide was determined by the voltammetric method developed herein. All experiments were performed at room temperature, 20 k 2 "C. Caution. N-Nitrosamines are potent carcinogens. The experimental steps with these compounds were performed in a fume hood, including weighings and sample transfers. The workers used disposable gloves which were also confined to the fume hood. Experiments on the N-nitrosamines were limited to the minimum number needed to verify conclusions.
60
A /
-
O2
4
6
8
PH Figure 1. Solid phase extraction of (A) .Q-nitroso-N,N-diethylaniline and (B) phenylalanine onto Nafion beads.
RESULTS AND DISCUSSION
The basis of the present study was our observation that such diverse compounds as dopamine, various peptides, and 4nitrosoNJ-diethylaniline (NDEA) in their protonated forms were not transported across a membrane of a sulfonated polymer of periluoroethylene (Nafion, DuPont) under conditions where they were dialyzed across a membrane consisting of a sulfonated styrene graft of poly(tetrafluoroethy1ene). We hypothesized that Nation has an unusually high affinity to sorb these compounds through nonspecific interactions, a mechanism that permits preconcentration of neutral organic compounds. The first part of this report is a veritication that solid phase extraction of such compounds occurs at Nafion. Second, this chemistry in conjunction with electrocatalysis was employed in the development of a stripping voltammetric method for neutral organic compounds. The catalytic activity of a ruthenium-based polymer sandwiched between a Nation film and a glassy carbon electrode was retained, and mobility of neutral compounds in the Nafion layer was demonstrated. These observations, along with the ionic conductivity of Nation in the absence of a solution phase, allowed the design of a solid state amperometric sensor for gases, the evaluation of which is presented in the thiid section of this report. Solid Phase Extraction Studies. Figure 1 illustrates the innuence of pH on the solid phase extraction of NDEA, which has a pKa of 4.6,2j and phenylalanine (Phe), which represent respectively an organic base and a compound that forms zwitte nons with a pK1, 5.48,26similar to the pKa of NDEA. The protonated forms of both compounds are strongly sorbed by Nafion, which is expected on the basis of the Hofheister series of ionexchange affinities. At pH values where NDEA is neutral, significant sorption still occurs, which is consistent with the hypothesis of nonspecific interaction. Near the pKa of NDEA, these data may be influenced by a shift in acid-base equilibrium at the surface of the beads; however, at pH 8.0, only about 0.4% of the NDEA is protonated. Perhaps more convincing is the fact that the sorption process is difficult to reverse with an electrolyte (24) Dean,1.A. Lunge's Handbook ofchemistry, 14th ed.;McGraw-Hill, Inc.: New York, 1992; p 8.109. (25) Gornostaev, L. M.; Skvortsov, N. IC; Belyaev, Yu. E.; Ionin, B. I. Zh. 0%. Khim. 1974, 10,2484-2486. (26)Voet, D.; Voet, J. G. Biochemistfl J. Wiley & Sons: New York, 1990; p 62.
solution. Leaching the beads, which contained the NDEA extracted from the pH 7 sample, for 30 min with 20 mL of 0.1 M HCl in water did not remove a detectable amount of NDEA Under the same conditions but with a 2 + 8 mixture (vol) of acetonitrile in water as the solvent for the HCl, 25%of the sorbed NDEA was recovered in 30 min. Further evidence that the sorption of NDEA from pH 7 solution was by nonspecific interaction with Nafion is that the distribution constant, K, is not a function of NDEA concentration. By definition, K is the ratio of the quantity of the analyte sorbed to the solid to that remaining in solution at equilibrium. The values of K determined with 5.0, 19, and 25 pM NDEA were 0.32, 0.30, and 0.31, respectively. With Phe as the analyte, significantsorption was not observed at pH values greater than the ~ K I .This may be due to repulsion effects due to the anionic nature of Phe in this region, but an alternative explanation is that the NDEA data represent an unusual case. To differentiatebetween these possibilities,the experiment described in Figure 1 was repeated on a structural analogue of phenylalanine but without a carboxylic acid group, 1-amino-2phenylethane (APhA). At sample pH values of 1.0, 4.5, 8.0, and 12.0, the sorptions were 100,100,71, and 27%,respectively. The pKa of APhA was determined to be 9.4 by acid-base titration with HCl. Clearly, the lack of sorption of Phe at pH values above the pK1 is a result of repulsion between sulfonate and carboxylate, and nonspecific interaction between these compounds, which have a common structural feature of an aromatic ring, and Nation is significant. The repulsion effect between the carboxylate and sulfonate is important with dipeptides as well. Phenylalanineleucine (Phe Leu) was not sorbed from basic 1.9 x M sample solutions; at pH 1.0,4.0, 5.5, 8.0, and 12.0, the sorptions were 100, 100, 48, 0.0, and 0.0, respectively. With 10 pM cysteineglycine (Cys-Gly) samples, sorptions of 58% (PH 1.0) and 20%(PH 9.3) were obtained. The difference in behavior between these dipeptides probably results from the fact that Cys is by far the strongest base of the amino acids in this set; the pK1 of Cys, 9.56, is more than 3 units greater than that for Phe, Leu, or Gly. Because of their toxicity, only one solid phase extraction experiment was performed with N-nitrosamines. Using a sample Analytical Chemistry, Vol. 67, No. 5, March 1, 1995
995
~~
~~~~~~
Table 1. Callbration Curves for Anodlc Strlpplng Voltammetry at an mvRuCN-Coated Qlassy Carbon Electrode wlth a Nafion Overlayer
analyte NDEA NNDPhAC APhA
0.60
0.80
1.00
E IV
APhA Cys-Gly
preconcn time,min
ldr,"pM
slope,pApM-'
9
DL,b pM
5 30 5 1od 10
4.2-240 0.51-5.1 0.54-10 0.54-81 1.0-48
0.21 f 0.01 2.20 f 0.09 1.19 0.09 0.56 f 0.02 0.16 f 0.003
0.99 0.998 0.992 0.998 0.999
0.8 0.3 0.5 0.5 0.6
Flgure 2. Anodic stripping voltammetry following solid phase extraction of 4-nitroso-N,N-diethylaniline (NDEA) at an mvRuCNcoated glassy carbon electrode with a Nafion overlayer. The displayed voltammogramsare (1) 0.05 M HCI at mvRuCN-coated glassy carbon, (2) same as 1 except with a Nafion overlayer, and (3) same as 2 except with a 10 pM NDEA sample and a 600-spreconcentration at open circuit. Scan rate, 50 mV s-l.
Linear dynamic range. * Detection limit; concentration that yields a signal 3 times the standard deviation of the blank. c The experiments with NNDPhA were with a &monolayer mvRuCN film;all other experiments were with 13 monolayers of mvRuCN. Preconcentration from pH 7 solution and stripping in 0.01 M HCl, 0.1 M KCI; all other experiments were in 0.05 M HCI.
of 49 pM N-nitrosodi-n-propylamine(NNDPA) in neutral solution, 44%sorption was observed under the Figure 1 conditions. The presence of an aromatic ring is therefore not a requirement for significant nonspecific interaction of a compound with Nafion. The results of this solid phase extraction study suggest that a variety of organic compounds can be preconcentrated at a film of Nafion on an electrode surface, which, in conjunction with the conductivity of this ionomer, allows the design of stripping voltammetric methods for such analytes. This approach is applicable to both solution and gas phase analytes. Whether the sorbed species are sufiiciently mobile in a Nafion film to sustain a faradaic current and whether a catalyst such as mvRuCN is active when coated with Nafion were identified as problems that could limit these methods. Stripping voltammetry of solution phase analytes was employed to investigate these possible limitations. Solution Voltammetric Studies. Initial experiments were performed to determine whether the mvRuCN catalyst retains activity in the presence of a Nation overlayer. Sandwich electrodes were prepared by first coating glassy carbon with mvRuCN and then casting a Nafion film. By ion exchange, protonated NDEA was preconcentrated from a 0.05 M HC1. Subsequently,a positive going linear potential scan was applied to the electrode. The resulting current-potential curve along with the voltammogram of the blank (HC1) solution is shown in Figure 2. The voltammogram of the blank is the same whether or not the Nafion is present; the processes at about 0.8 and 1.05 V are assigned to the oxidation of Ru(I1I) to R u O and the oxidation of R u O to R u m , re~pectively.'~~~~ Although the data in Figure 2 demonstrate that mvRuCN remains active when coated with Nafion, the overlayer does influence the voltammetry of protonated NDEA In the absence of the overlayer, NDEA is oxidized by R u O , but in the presence of the Nafion film, mediation of the oxidation requires the presence of R u m . This can be explained by the strong ion-exchange interaction that is expected between the protonated NDEA and sulfonate. The free energy of this reaction causes a shift in the formal potential for the oxidation of NDEA toward positive values, thereby requiring a stronger oxidizing agent to mediate the electrode process. Presuming that the oxidation is a twoelectron, two-proton process to the nitramine, the 24@mV shift in formal potential that is required for this interpretation of the voltammetry will result from an equilibrium constant larger than about 104, which is reasonable for an ionexchange reaction.' Anodic stripping voltammograms were obtained for NDEA at the sandwich electrode. A linear least-squares fit of calibration
data is shown in Table 1. Also included in Table 1 are results with APhA and Cys-Gly as the analytes. The influence of pH on the sensitivity toward APhA results from a change in the importance of ion-exchangerelative to nonspecific interactions in the preconcentration. The greater sensitivity of the determination of APhA than that of NDEA under identical, acidic conditions reflects the relative strength of these compounds as bases. When the neutral compound, N-nitrosodiphenylamine (NNDPhA), was the analyte, linear scan voltammetry following solid phase extraction into a Nafion film produced current-potential curves with the same features as those for the oxidation of NDEA (Figure 2). That is, the mediated oxidation of NNDPhA was observed at about 1.05 V under the Figure 2 conditions, except that 50 pM NNDPhA was the analyte. Linear calibration curves were obtained over the concentrationrange of 0.51-5.1 pM (Table 1); however, over a wider range, 5-300 pM NNDPhA (30 min preconcentration time), the calibration curve was nonlinear. Moreover, the sensitivity of the determination of NNDPhA was lower than that for NDEA; the signal for 17 pM NNDPhA was 2.1 & whereas it was 3.0 pA for the same concentration of NDEA This difference was expected because the sorption mechanism for NNDPhA is nonspecific interaction,whereas for NDEA in 0.05 M HCl ion-exchange as well as nonspecific interaction occurs (see Figure 1). The cause of the nonlinearity of the NNDPhA calibration curve over a wide concentration range is related to the chemistry of this analyte in the sorbed state. When the stripping current is measured for a single concentration but with a variable preconcentration time, the shape of the plot of these parameters (Figure 3) is the same as that for the calibration curve. With NNDPhA as the analyte, a comparison was made of stripping voltammetric data obtained at the Nafioncoated electrode to those obtained at the same electrode (mvRuCN-coated glassy carbon) without the Nafion overlayer. When the preconcentration was for 30 min at open circuit, the calibration curves became independent of the concentration of the analyte in solution at 300 and 1pM NNDPhA in the presence and absence of Nafion, respectively. Given the results of the solid phase extraction experiments, the greater sorption of this neutral compound in the presence of the Nation film was expected. However, these data also demonstrated that a compound sorbed to the Nafion by nonspecific interactions diffuses to the electrode at a rate sufficiently high to contribute to the stripping peak current when the scan rate is 50 mV s-'. The detection limits shown in Table 1are significantly higher than those for the analogous experiments with inorganic ions; for example, 1 x M C r O was determined by linear scan voltammetry at an electrode coated with an anion-exchange
(27) Burke, L. D.;Healy, J. F.J. ElectroanaZ. Chem. 1981,124, 327-332.
996 Analytical Chemistry, Vol. 67, No. 5, March 7, 7995
,
2.11
I
0.0
C
I
0
10
I
20
30
Time h r Figure 3. Dependence of the peak current for the stripping of NNDPhA on the preconcentration time. Electrode, glassy carbon coated with six monolayers of mvRuCN and 5 pm of Nafion; analyte, 1.1 pM NNDPhA; electrolyte, 0.05 M HCI; linear potential scan rate, 50 mV s-l. Background-corrected peak currents were measured at 1.0 v.
mvRu(CN -Nation
Figure 4. Interdigitatedmicrosensor with both the working electrode (WE) and the reference electrode (RE) coated with mvRuCN. A continuous Nafion film covers both electrodes.
polymeric film following a 6@s preconcentration.28The probable limiting factor is slower diffusion of compounds that interact with the membrane backbone than of ionic species in an ionomer. To test this hypothesis, NNDPhA was preconcentrated for 7 min by the procedure in Table 1, but the stripping was performed by potential step chronocoulometry.29 The detection limit was decreased from 0.3 pM to 50 nM, even though the preconcentration time was shortened from 30 to 7 min. Based on this result, the studies on the gas phase sensor that are described below were quantified by amperometry at constant potential. Voltammetric Sensors without a Solution Phase. The demonstration that Nation can serve as the electrolyte for voltammetry in the absence of a solution phase,18J9the ability to preconcentrate neutral organic compounds by solid phase extraction onto Nafion, the mobility of these compounds in Nation, and general advances in the field of solid state voltammetry, when considered in combination, suggest the design of voltammetric sensors for the determination of analytes in the gas phase. The design of the sensor is shown in Figure 4. Previously reported designs of voltammetric sensors for gases are quite different. Schiavon et al.30used an indicator electrode that was supported on Nafion; the indicator was exposed to a gas phase, but the Nafion was contacted to a solution electrolyte that contained the reference and counter electrodes. The previously mentioned sensor that used ultramicroelectrodes separated by glass and cannot preconcentrate the analyte. The cell design most related to the present one, an anionexchange membrane impregnated with a mixed-valent ruthenium catalyst,lgwas used only for (28) Cox, J. A; Kulesza, P. J. Anal. Chim. Acta 1983,154, 71-78. (29) Bard, A. J.; Faulkner, L. R Electrochemical Methods, Fundamentals and Applications; Wdey & Sons, Inc.: New York, 1980; pp 199-206. (30) Schiavon, G.; Zotti, G.; Bontempelli, G. Anal. Chim. Acta 1989,221, 2741.
1
I
0.50
.oo
1
E /V Figure 5. Cyclic voltammetry of mvRuCN in the presence of a Nafion overlayer in 0.05 M HCI. Reference electrode, Ag/AgCI, 3 M NaCI; scan rate, 50 mV s-l. The processes near 0.0, 0.8,and 1.O V are related to Ru(III,II), Ru(IV,III), and Ru(VI,IV), respectively.
cyclic voltammetry at a single concentration,and the catalyst was distributed throughout the membrane. The cell in Figure 4 generally was used without a solution phase, but to evaluate the potential of the quasireference, a cyclic voltammogram of the modified IME was obtained in 0.05 M HCl (Figure 5). From this voltammogram it is seen that the mvRuCN quasireference (the cathode in the twoelectrode sensor in Figure 4) will be at a potential of about 0.0 V vs Ag/AgCl. The actual value is uncertain because of the slight pH dependence of the potentials of the voltammetric peaks in Figure 5 . All data for this sensor are reported vs the mvRuCN quasireference. Initial experiments were performed using methanol vapor (200 p L injected into the chamber) as the analyte. The sensor was exposed at open circuit to the vapor for 1-10 min, after which cyclic voltammograms were obtained. The results generally agreed with those reported by Kulesza and Faulkner.lg In the present study, the potential of the onset of the oxidation of methanol was 0.7 V vs the quasireference electrode, whereas is was about 0.8 V vs Ag/AgCl in the cell of Kulesza and Faulkner. This potential is in a range where Rum is the mediator. The previous study was performed with a 10-min preconcentration; we found no influence of preconcentration time at open circuit on the current for the oxidation of methanol. A calibration curve was prepared by using the sensor under potentiostatic conditions (0.9 V). The current was measured at prescribed times after introduction of various quantities of the methanol vapor. Over the range 200-1000 p L of methanol injected into the chamber, a linear plot of current vs quantity was obtained. A least-squares fit of the data (n = 5) obtained with a 5s sampling time yielded the following: slope, 0.80 f 0.05 nA pL-l; P, 0.999, The sensitivity of the sensor is better evaluated by using the vapor pressure for methanoF1to calculate the actual quantity injected. The 20@pLsample corresponds to less than 2 pmol of this compound. Voltammograms were obtained with NNDPhA and N-nitrosodimethylamine (NNDMA) vapors as analytes. Consistent with the results of solution phase studies, the onset of their oxidations occurred at the potential at which R u O is present in the film. A typical voltammogram is shown in Figure 6. As in our study on methanol, calibration data were obtained under potentiostatic conditions. A typical current-time response that was seen with stepwise increases in concentration is shown in Figure 7. The response to an increase of analyte concentration is instantaneous on the time scale of seconds, and the current approaches steady state (changes less than 3%per minute) after 3 min. The latter point suggests that the Nafion-coated IME has the feature of an ultramicroelectrode of not developing time dependence in the concentration gradient at the electrode surface as long as facile mass transport occurs in the surrounding medium. (31) Jordan, T. E. Vapor fiesure of Organic Compounds; Interscience: New York, 1954; p 199.
Analytical Chemistry, Vol. 67,No. 5,March 1, 7995 997
T
I
I
0.50 1.00 AE IV
0.0
I
I
0.60 0.80 1.00
I
I
AE /V
Figure 6. Solid-statelinear scan voltammetry with the cell in Figure 4 mounted in a chamber maintained at 96-98% relative humidity. Concentrations (vol YO)of Nnitrosodimethylamine(NNDMA) vapors are (1) 0, (2) 7.5, and (3) 60. The voltammograms were recorded 2 min after injection of the NNDMA vapor to the chamber. The display contains both (A) uncorrected and (6)background-correctedvoltammograms.
;1: 4.5
+
4.5
.
9.0
n
12 ^ I
240 s
Figure 7. Current-time response at 1.00 V of the interdigitated microsensor shown in Figure 4 to injections of NNDMA vapors into the chamber. The arrows indicate the point of injection. The vol % concentration of NNDMA is shown in the figure.
A calibration curve was generated for NNDPhA using 1.1V as the applied potential and measuring the current 1 s after injections of 200-900 ,uL. A least-squares fit of the data (n = 4) yielded the following parameters: slope, 92 f 7 pApL-'; 12,0.988. Over a greater concentration range, the response is nonlinear. However, the sensor is very sensitive. Using the vapor pressure vs temperature data of an analogous compound, diphenylamine?l as a rough estimate of that for NNDPhA and assuming ideal gas behavior, the 20O;UL injection corresponds to less than 1nmol of the N-nitrosamine. Unlike methanol, NNDPhA is preconcentrated at the Nafion film from the gas phase; increasing the exposure time in Figure 6 from 2 to 10 min approximately doubles the
998 Analytical Chemistry, Vol. 67, No. 5, March 1, 1995
corrected peak current. Because of the toxicity of these compounds, the effect of time on the sensitivity was not systematically studied. Hence, an analogue of stripping voltammetry will further improve the sensitivity of the measurement. The fact that the sensor responds to both methanol and N-nitrosamines does not obviate a selective measurement of the latter. To date, the only compounds for which we and others have observed electrocatalytic oxidation by R u O but not by R u O are the N-nitrosamines (see ref 12 and citations therein). This suggests a strategy for their selective determination, namely preconcentrating at 0.8 V and stepping the potential to 1.1V for the measurement. This concept was tested in the present study. Methanol vapors do not yield a signal under this experimental design, even with preconcentration times of 10 min; however, the preconcentration of N-nitrosamines at 0.8 V results in the same signal when the electrode is stepped to 1.1V, as does an identical preconcentration at open circuit. For applications to other analytes, the sensor will need to be coupled to a separation method or selective coatings for targeted analyteswill need to be identified. The latter option is being pursued at present. Conclusion. Solid phase extraction of both neutral and cationic organic compounds occurs at Nafion. The nonspecific interactions that allow extraction of the neutral compounds is not general for sulfonated organic polymers. The presence of a Nafion coating does not change the ability of mvRuCN to serve as an electrocatalyst, so the use of these modfiers in combination increases the number of compounds that can be determined by stripping voltammetry. Moreover, when the Naiion film covers both the reference and the indicator electrode, this combination electrode allows voltammetric study of gas phase analytes without the use of a bulk solvent. ACKNOWLEDGMENT
This work was supported in part by the U S . Environmental Protection Agency through Grant 816507. The results have not been reviewed by the Agency, so an endorsement should not be inferred. We are grateful for the support of M.E.T. by the National Science Foundation through Grant CHE9322137. The authors thank T. E. Cummings for his helpful suggestions on the gas sensor. Scientzj5c Parentage of the Author. J. A. Cox, Ph.D. under A. M. Hartley, Ph.D. under J. J. Lingane, Ph.D. under I. M. Kolthoff. Received for review August 17, 1994. Accepted December 19, 1994.@ AC9408080 @
Abstract published in Advance ACS Abstracts, February 1,1995.