A Thin-Layer Amperometric Sensor for Hydrogen Sulfide: The Use of

A Thin-Layer Amperometric Sensor for Hydrogen Sulfide: The Use of Microelectrodes To Achieve a Membrane-Independent Response for Clark-Type Sensors...
0 downloads 0 Views 82KB Size
Anal. Chem. 2003, 75, 2499-2503

A Thin-Layer Amperometric Sensor for Hydrogen Sulfide: The Use of Microelectrodes To Achieve a Membrane-Independent Response for Clark-Type Sensors Nathan S. Lawrence,† Li Jiang,‡ Timothy G. J. Jones,‡ and Richard G. Compton*,†

Physical and Theoretical Chemistry Laboratory, University of Oxford, South Parks Road, Oxford, OX1 3QZ, U.K., and Schlumberger Cambridge Research, High Cross, Madingley Road, Cambridge, CB3 0EL, U.K.

An electrochemical cell design of the Clark type including a thin layer of electrolyte in contact with a microelectrode has been successfully applied for the determination of sulfide utilizing its electrochemically initiated reaction with aqueous diethyl-p-phenylenediamine. The analytical parameters obtained were independent of the membrane used to separate the inner chamber and the outside sulfide-containing solution. The independence arises since the thickness of the diffusion layer associated with the microelectrode is small and, in contrast to the conventional macroelectrode Clark electrode, does not impinge on the membrane. This provides an improvement in gas sensor design and development as it obviates the need for membrane calibration and should simplify the application of Clark cells for variable-temperature measurements. Traditional gas sensing systems are often based upon amperomeric Clark sensors.1,2 These sensors comprise a chamber in which the electrolyte and electrodes are housed. A gas-permeable membrane is used to separate the electrolyte and the outside flow, behind which the sensing electrode is placed, to obtain maximum sensitivity. The Clark cell usually contains two electrodes consisting of a cathode working macroelectrode and a suitable reference electrode.1 The working electrode is then poised at a suitable detection potential (depending on the gas to be sensed) for amperometric sensing. The mathematical treatment of such devices has been extensively studied for macro- and microdisk electrodes using a one-dimensional diffusion model. The first utilized a one-dimensional model of the transport of the gaseous species through each of the three layers (sample | membrane | electrolyte) as detailed in Scheme 1a.1 Analysis of this scheme reveals that any electrochemical signal obtained due to the gas permeating through the membrane reflects not only diffusion of the species through the electrolyte to the electrode surface but * To whom correspondence should be addressed. Tel: 01865 275413. Fax: 01865 275410. E-mail: [email protected]. † University of Oxford. ‡ Schlumberger Cambridge Research. (1) Clark, L. C. Trans. Am. Soc. Int. Artif. Organs 1956, 2, 41. (2) Hahn, C. E. W. Analyst 1998, 123, 57R-86R. 10.1021/ac0206465 CCC: $25.00 Published on Web 04/19/2003

© 2003 American Chemical Society

Scheme 1. (A) Conventional Clark Cell Design and (B) Proposed Pathway for Membrane Independence

also diffusion of the gaseous species from the outside flow through the membrane into the sensing chamber. Over the years, this model has been extensively refined; Hahn3 first described a simplified one-dimensional model for the behavior of microdisks; this was then extended further to the twodimensional cylindrical model by Linek et al.4 In this case, twodimensional diffusion of the gas in both the sample and electrolyte layers with only linear diffusion through the membrane was considered in order to mimic the real electrochemical situations occurring in sensors with a microdisk. Finally, Gavaghan et al.5-7 reworked this model using digital simulation techniques to describe the relationship of the sensor current to the microdisk size, electrolyte, and membrane layer characteristics. Unlike Linek et al.,4 the one-dimensional diffusion of the gas through the membrane was relaxed and was treated as two dimensional in each layer. In the account outlined below, the idea of a membraneindependent hydrogen sulfide gas sensor is developed. This is based upon the use of a microelectrode placed inside a thin-layer chamber. The idea is that a thin membrane is employed which (3) Hahn, C. E. W. J. Phys. E 1981, 14, 783-797. (4) Linek, V.; Vacek, V.; Snikule, J.; Benes, P. Measurements of Oxygen by Membrane-Covered Probes; Guidelines for Apllications in Chemical and Biochemical Engineering; Ellis Horwood: Chichester, U.K., 1988. (5) Gavaghan, D. J.; Rollett, J. S.; Hahn, C. E. W. J. Electroanal. Chem 1990, 325, 23-44. (6) Gavaghan, D. J.; Rollett, J. S.; Hahn, C. E. W. J. Electroanal. Chem. 1993, 348, 15-27. (7) Sutton, L.; Gavaghan, D. J.; Hahn, C. E. W. J. Electroanal. Chem. 1996, 408, 21-31.

Analytical Chemistry, Vol. 75, No. 10, May 15, 2003 2499

allows rapid equilibration of the target gas between the thin layer and the membrane exterior. The sensing is carried out at a microelectrode of a dimension such that its diffusion layer does not impinge on the membrane (Scheme 1b); accordingly, the signal observed solely reflects the transport within the thin layer and not within the membrane since in the latter equilibrium conditions prevail. Consequently, the signal is membrane independent. In the conventional Clark design, rather thick layers of solution may be used within the cell since typically the membrane transport is rate limiting and so the volume within the “thin layer” is relatively unimportant. In the device now proposed, the electrolyte thickness within the layer can be shrunk to a small dimension provided the latter is larger than the diffusion layer thickness of the microeletrodes; assuming a 5-µm electrode, a thickness of 10-20 µm should suffice. Accordingly, the proposed design is additionally in principle faster than the conventional Clark cell. Hydrogen sulfide is a gas encountered in a variety of industrial environments and whose toxicity8 has necessitated the development of fast and sensitive monitoring devices. As such, an extensive array of electrochemical techniques have been developed9-15 and have proven to be produce some of the more viable in terms of portability and sensitivity. Such sensor designs have been based upon modified amperometric Clark sensors and potentiometric devices.9-15 In the sensor design outlined below, the electrochemically initiated reaction of diethyl-p-phenylenediamine (DEPD) with sulfide is utilized as a means of detecting hydrogen sulfide,16-18 the reaction mechanism of which is depicted in Scheme 2. This detection strategy has been previously utilized in a dual flow cell sensing device whereby indicator species is flowed through one channel that is separated by a gas-permeable membrane from the effluent channel containing sulfide.19 In the report outlined below, a single microelectrode is placed in a thinlayer static pool of DEPD indicator solution behind a gaspermeable membrane. The potential is swept in the oxidative direction such that the DEPD is oxidized to its quinine-imine form. The sulfide reacts with the quinine-imine species in a Michael addition, producing an adduct that is oxidized at the microelectrode thereby producing the analytical signal. This has been termed as an EC2xE reaction process and has been shown to occur at both macroelectrodes and microelectrodes.20 (8) Lawrence, N. S.; Davis, J.; Compton, R. G. Talanta 2000, 52, 771-784. (9) Kroll, A. V.; Smorchkov, V. I.; Nazarenko, A. Y. Sens. Actuators, B 1994, 21, 97-100. (10) Schiavon, G.; Zotti, G.; Toniolo, R.; Bontempelli, G. Anal. Chem. 1995, 67, 318-323. (11) Jeroschewski, P.; Haase, K.; Trommer, A.; Grundler, P. Fresenius J. Anal. Chem. 1993, 346, 930-933. (12) Jeroschewski, P.; Haase, K.; Trommer, A.; Grundler, P. Electroanalysis 1994, 6, 769-772. (13) Jeroschewski, P.; Braun, S. Fresenius J. Anal. Chem. 1996, 354, 169-172. (14) Jeroschewski, P.; Stveukart, C.; Kuhl, M. Anal. Chem. 1996, 68, 43514357. (15) Orion 9614A Sulfide Ion Selective Electrode, Orion Research Inc., Beverley, MA. (16) Lawrence, N. S.; Davis, J.; Jiang, L.; Jones, T. G. J.; Davies, S. N.; Compton, R. G. Electroanalysis 2000, 12, 1453-1460. (17) Lawrence, N. S.; Davis, J.; Jiang, L.; Jones, T. G. J.; Davies, S. N.; Compton, R. G. Electroanalysis 2001, 13, 432-436. (18) Lawrence, N. S.; Thompson, M.; Davis, J.; Jiang, L.; Jones, T. G. J.; Compton, R. G. Electroanal. Mikrochim. Acta 2001, 137, 105-110. (19) Lawrence, N. S.; Davis, J.; Marken, F.; Jiang, L.; Jones, T. G. J.; Compton, R. G. Sens. Actuators, B 2000, 69, 189-192.

2500

Analytical Chemistry, Vol. 75, No. 10, May 15, 2003

Scheme 2. Electrochemically Initiated Reaction of Diethyl-p-phenylenediamine with Sulfide

In the case outlined below, we examine the steady-state limiting response when sulfide passes through the membrane into the thinlayer cell. The response obtained at the microdisk electrode for the above-mentioned EC2xE reaction process will be shown to be membrane independent as an equilibrium of the target gas between the membrane exterior and cell chamber will be set up prior to any measurement being recorded. EXPERIMENTAL SECTION Reagents and Equipment. All reagents were obtained from Aldrich, were of the highest grade available, and were used without further purification. All solutions and subsequent dilutions were prepared using deionized water from an Elgastat (Elga) UHQ grade water system with a resistivity of 18 MW cm. Stock sulfide solutions (0.05 M) were prepared by dissolving the sodium salt in previously degassed water and were used within 1 h of preparation to minimize losses due to aerial oxidation. All experiments were carried out at 20 ( 2 °C. Electrochemical measurements were recorded using an Autolab PGSTAT 30 computer-controlled potentiostat (Eco-Chemie) with a standard three-electrode configuration (cell design is given below). The carbon microelectrode was carefully polished prior to use, using 1.0- and 0.3-µm alumina (Buehler).21 The microelectrode was electrochemically calibrated prior to use to obtain an accurate measurement of electrode radius. This was carried out in solutions containing various concentrations of ferricyanide (2.00.25 mM, D ) 0.76 × 10-5 cm2 s-1 22) in 0.1 M aqueous KCl with linear sweep voltammograms recorded in each. The corresponding plot of limiting current against concentration allowed the electrode radius to be determined as 4.4 µm. Cell Assembly. A schematic diagram of the cell setup is shown in Figure 1. An aqueous solution of the redox indicator species (20) Brookes, B. A.; Lawrence, N. S.; Compton, R. G. J. Phys. Chem. B 2000, 104, 11258. (21) Cardwell, T. J.; Mocak, J.; Santos, J. H.; Bond, A. M. Analyst 1996, 121, 357-362. (22) Adams, R. N. Electrochemistry at Solid Electrodes; Marcel Dekker Inc.: New York, 1969.

Figure 1. Microelectrode cell setup.

Figure 2. Microelectrode cell designs.

DEPD is placed in the upper chamber, which is separated from the sulfide flow stream by a gas-permeable membrane. A carbon microelectrode is placed opposite the membrane and acts as the working electrode. A platinum counter electrode is singly coiled inside the cell to define the cell height (0.5 mm). AgCl acts as the reference electrode and is placed alongside the working electrode. Linear sweep voltammograms were recorded for measurements before, during, and after the introduction of sulfide to the flow. The microelectrode rapidly attains (∼ >0.5 s) steadystate diffusion23 so that the electrochemically initiated reaction with sulfide can occur. The use of the microelectrode in the thinlayer cell means only a small portion of the DEPD is oxidized such that there is very little depletion of the indicator species compared to what would be observed at a macroelectrode. Cell Design. A schematic diagram of the cell is shown in Figure 2. The cell was machined from Delrin (Goodfellow Advanced Materials, Cambridge Ltd.) and comprises two sections that screw together. The upper chamber (1) is the working area (23) Aoki, K. J. Electroanal. 1993, 5, 627-639.

where the electrodes are mounted. This is composed of both inlet and outlet filling holes (A and D), so that DEPD solution can be initially supplied to the chamber, and holes for the three electrodes. The platinum counter electrode (Advent Research Materials, wire diameter 0.5 mm) is located through hole E and is coiled singly to define the cell thickness (0.5 mm). The AgCl reference electrode is placed in hole C next to the working electrode and sealed using low melting point wax (50 °C, depilatory wax, WAX-A-WAY, London, England). The working microelectrode (hole B) is placed at the center of the cell and is fixed using silicone rubber (RS Components Ltd.). Section 2 completes the cell assembly and allows the flow of hydrogen sulfide toward chamber 1. The inlet and outlet are positioned immediately below the different poly(tetrafluoroethylene) (PTFE) membranes (Goodfellows Advanced Materials, Cambridge Ltd., thickness 0.085 mm, porosity 50%, and pore size 0.02 µm or thickness 0.263 mm, porosity 64%, and pore size 0.45 µm), which covers the orifice such that the flowing stream passes directly over the membrane. PTFE tubing was used throughout to connect the various inlet and waste streams. Cell Operation. A solution containing 300 µM DEPD was placed into the upper chamber after which sulfide solutions were allowed to flow through the effluent inlet. The flow velocity was manipulated by linking the solution into a capillary exit and manually adjusting the position of the feeding reservoir. First 1.0 M H2SO4 was passed through the effluent stream and the voltammetric response of the DEPD (300 µM) solution recorded. Solutions containing the test sulfide sample (25 µM to 125 µM in 1.0 M H2SO4) were added to the effluent reservoir and allowed to pass through the sensor assembly, and the voltammetric response was recorded. Once the sulfide solution was added into the 1.0 M H2SO4, gaseous H2S was generated and was passed through the membrane into the DEPD-containing chamber. Stability of the AgCl Reference Electrode. The stability of the response of the silver/silver chloride reference electrode to sulfide was tested in the absence and in the presence of KCl in pH 7 phosphate buffer by measuring the change in potential difference with respect to a saturated calomel electrode when sulfide was added to the solution. The potential difference between the two electrodes was found to be stable when sulfide was added to the cell over the examined period of one month. RESULTS AND DISCUSSION Outlined below are the voltammetric responses of DEPD to increasing sulfide concentrations inside the closed cell arrangement. First, however, the response of DEPD to sulfide was studied at the microelectrode in an open cell whereby the sulfide is injected directly into the DEPD solution. Open Cell. The linear sweep voltammetric response (scan rate 10 mV s-1) of 300 µM DEPD (pH 7, phosphate buffer, 0.05 M NaH2PO4, 0.05 M Na2HPO4) to increasing additions of sulfide is detailed in Figure 3. In the absence of sulfide, a single oxidation wave is observed at +0.15 V (vs AgCl) that reaches a steadystate limiting current at +0.25 V. This voltammetric behavior can be attributed to the two-electron oxidation of DEPD to the quinone-imine species as detailed in Scheme 2.18 It can be seen that upon the addition of sulfide to the system an increase in the limiting current occurs. This reveals that the Analytical Chemistry, Vol. 75, No. 10, May 15, 2003

2501

Figure 3. (A) Linear sweep voltammograms detailing the response of 300 µM DEPD (pH 7 phosphate buffer) to increasing additions of sulfide (25-150 µM) in the absence of a membrane. (B) Corresponding plot of limiting current against concentration of sulfide. Table 1. Effect of Varying the Concentration of DEPD on the Analytical Signal concn of DEPD (µM)

sensitivity (µA/µM cm2) [H2S]

R2

200 50

2.24 × 10-12 1.95 × 10-12

0.983 0.996

a

N

linear range (µM)

LODa (µM)

5 6

25-100 10-50

18 4

LOD, limit of detection, based on 3sb.

sulfide attacks the electrochemically oxidized DEPD species in a 1,4-Michael addition as detailed in Scheme 2. The newly formed product is then oxidized at the electrode surface thereby producing an enhancement in the oxidation peak current consistent with that observed previously for the electrochemically initiated reaction of dimethyl-p-phenylendiamine with sulfide.20 To show that the enhancement in limiting current is due to the electrochemically initiated reaction of sulfide with DEPD, the voltammetric response of 25 µM sulfide in the absence of DEPD was studied and was found to produce no voltammetric waves in the potential range studied and therefore means that the enhancement in oxidation current is due to the reaction process detailed in Scheme 2 occurring. A plot of this increase in limiting current against concentration of sulfide is depicted in Figure 3b; this shows a linear range from 25 to 100 µM sulfide and a corresponding limit of detection (based on 3sb) of 18 µM. To improve these analytical parameters, the concentration of DEPD was decreased to 50 µM. The corresponding analytical parameters obtained for the two different concentrations of DEPD to increasing sulfide concentrations are detailed in Table 1, which shows a linear range from 10 to 50 µM and limit of detection of 4 µM for sulfide detection can be achieved by varying the concentration of DEPD. These show that by increasing the concentration of DEPD the linear range can be increased, at some cost to the limit of detection. However, it should be noted that the sensitivity is virtually independent of the DEPD concentration present, with only a 10% difference observed between the two concentrations of sulfide, which is consistent with that observed previously for the electrochemically initiated reactions of phenylenediamines at macro electrodes.18 Closed Cell. In all experiments detailed below, a PTFE membrane separates the sulfide and DEPD chambers such that gaseous sulfide passes through the membrane before detection. 2502 Analytical Chemistry, Vol. 75, No. 10, May 15, 2003

Figure 4. Linear sweep voltammograms detailing the response of 300 µM DEPD (pH 7 phosphate buffer) to increasing additions of sulfide (25-75 µM) in the effluent stream separated by a gaspermeable PTFE membrane (thickness 0.085 mm, porosity 50%, and pore size 0.02 µm).

Initial characterization of the cell focused upon the electrochemical oxidation of DEPD inside the closed cell. Figure 4 depicts the linear sweep (10 mV s-1) voltammetric response for a solution (pH 7, phosphate buffer) containing 300 µM DEPD inside the thin-layer cell while 1.0 M H2SO4 was being passed through the effluent channel. This shows a single steady-state oxidation wave at +0.15 V (vs AgCl) analogous to that observed previously (see above). The reproducibility of this steady-state limiting current obtained was examined by measuring the limiting current obtained for five linear sweep voltammograms in a solution (pH 7, 0.1 M phosphate, 0.1 M KCl) containing 300 µM DEPD; the limiting currents were found to be 0.607 ( 0.017 nA giving a standard error of 2.8%, showing that this process results in reproducible baseline in the absence of sulfide and that no fouling of the electrode surface occurs. This is important for the use of the sensor in long-term on-line applications. Also depicted in Figure 4 is the steady-state linear sweep voltammograms obtained for 300 µM DEPD when aliquots of sulfide (25 µM in 1.0 M H2SO4) were introduced into the effluent stream. These reveal an increase in the limiting current consistent with gaseous sulfide passing through the PTFE membrane (thickness 0.085 mm, porosity 50%, and pore size 0.02 µm) into the DEPD chamber, at which time the electrochemically initiated reaction can occur in accordance with observations recorded previously in the absence of the membrane (see above). After the experiment was completed, a fresh solution was passed into the thin-layer chamber to produce a new steady-state baseline, corresponding to the oxidation of 300 µM DEPD. To verify that the voltammetric response to increasing additions of sulfide is membrane independent, the voltammetric signal was obtained when a second membrane was used. This has a thickness of 0.263 mm, a porosity of 64%, and a pore size of 0.45 µm. The voltammetric responses (not shown) once again revealed that upon the addition of sulfide to the effluent channel an increase in the steady-state limiting current of the DEPD oxidation wave was observed. The corresponding analytical parameters obtained for a solution containing 300 µM DEPD to both the direct injection of and addition behind different membranes of sulfide to the solution are detailed in Table 2. These results reveal that both in the absence and in the presence of different membranes analogous analytical results are observed in terms of linear ranges and, more

Table 2. Comparison of the Analytical Data Obtained at the Microelectrode for the Three Setups Described membrane setup

sensitivity (µA/µM cm2) [H2S]

R2

membrane-free membrane A membrane B

2.24 × 10-12 2.28 × 10-12 2.07 × 10-12

0.983 0.989 0.987

a

N

linear range (µM)

LODa (µM)

5 4 4

25-100 25-75 25-75

18 12 14

LOD, limit of detection, based on 3sb.

importantly, sensitivities. The fact that similar results are achieved for each situation detailed in Table 2 means that the steady-state limiting response is membrane independent where an equilibrium between the exterior flow and thin-layer cell is achieved prior to measurement.

CONCLUSIONS These results develop a novel cell design incorporating a thinlayer indicator reservoir within which a microelectrode is placed and has been successfully applied for the determination of sulfide. Further, and more importantly, the results reveal that the use of the microelectrode inside the thin-layer cell shows that a membrane-independent response is achievable. This arises from the fact that the diffusion layer of the sensing microelectrode does not impinge on the membrane. This provides an enhancement in gas sensor design as it discards the need for membrane calibration, thereby simplifying the use of Clark-type cells in variabletemperature experiments. ACKNOWLEDGMENT We thank Schlumberger Cambridge Research for financial support of this project and Dr J. Davis for helpful discussions. AC0206465

Analytical Chemistry, Vol. 75, No. 10, May 15, 2003

2503