Anal. Chem. 1996, 68, 2062-2066
A Pulse Amperometric Sensor for the Measurement of Atmospheric Hydrogen Peroxide Huiliang Huang, Purnendu K. Dasgupta,* and Zhang Genfa
Department of Chemistry and Biochemistry, Texas Tech University, Lubbock, Texas 79409-1061 Joseph Wang
Department of Chemistry, New Mexico State University, Las Cruces, New Mexico 88003
Gaseous H2O2 is sampled through a Nafion membrane diffusion scrubber while 1 mM HCl is maintained stationary in the scrubber. After a preselected preconcentration time (typically, 5-10 min), a valve is opened to allow the scrubber liquid to flow by gravity over an electrochemical H2O2 sensor for a brief period. The miniature flow-over sensor consists of a Pt/Rh wire working electrode and a Pt wire counter electrode wound respectively on separate segments of a Nafion solid polymer electrolyte tubing supported on a Ag/AgCl wire reference electrode. A simple electronic interface and a personal computer are used to control and record the electrochemical measurement. The liquid phase detection limit for this sensor is ∼30 nM H2O2 in the anodic oxidation mode. For a 9 min gas sample preconcentration period, the LOD (S/N ) 3 criterion) is 0.11 ppbv H2O2(g). Ambient H2O2 data obtained with this instrument were in excellent agreement with those obtained by an established fluorometric technique in a blind intercomparison. Hydrogen peroxide is ubiquitous in the atmosphere. Its chemistry and determination have been, and continue to be, of significant interest to atmospheric chemists.1,2 A reasonably recent review on this topic is available.3 In this laboratory, we have had an ongoing interest in the measurement of atmospheric H2O2.4-13 The two most common methods presently used for the measurement of atmospheric H2O2 both rely on an enzyme/ enzyme mimic catalyzed fluorometric chemistry. A glass coil is used in the method of Lazrus et al.,14 whereas a Nafion membrane diffusion scrubber (NMDS) is used in the method developed by (1) Buckley, P. T.; Birks, J. W. Atmos. Environ. 1995, 29, 2409-2416. (2) Dutkiewicz, V. A.; Burichard, E. G. Atmos. Environ. 1995, 29, 3281-3292. (3) Gunz, D. W.; Hoffmann, M. R. Atmos. Environ. 1990, 24A, 1601-1633. (4) Hwang, H.; Dasgupta, P. K. Environ. Sci. Technol. 1985, 19, 255-258. (5) Dasgupta, P. K.; Hwang, H. Anal. Chem. 1985, 57, 1009-1012. (6) Hwang, H.; Dasgupta, P. K. Anal. Chem. 1986, 58, 1521-1524. (7) Hwang, H.; Dasgupta, P. K. Anal. Chem. 1987, 59, 1356-1360. (8) Dasgupta, P. K.; Dong, S.; Hwang, H.; Yang H.-C.; Genfa Z. Atmos. Environ. 1988, 22, 949-963. (9) Genfa, Z.; Dasgupta, P. K.; Edgemond, W. S.; Marx, J. N. Anal. Chim. Acta 1991, 243, 207-216. (10) Zhang, G.; Dasgupta P. K.; Cheng, Y.-S. Atmos. Environ. 1991, 25A, 27172729. (11) Zhang, G.; Dasgupta P. K.; Sigg, A. Anal. Chim. Acta 1992, 260, 57-64. (12) Zhang, G.; Dasgupta P. K. Anal. Chem. 1992, 64, 517-522. (13) Sigg, A.; Neftel A.; Dasgupta, P. K. In Proceedings of the Workshop on the Development of Analytical Techniques for Atmospheric Pollutants, Rome, April 13-15, 1992; Allegrini, I., Ed.; Air Pollution Research Report 41; Commission of the European Communities; Dordrecht: Boston, 1993; pp. 111-117.
2062 Analytical Chemistry, Vol. 68, No. 13, July 1, 1996
Zhang et al.11 for collecting gaseous H2O2 in water. Both methods work well; one intercomparison15 claims the diffusion scrubber (DS) method to have advantages. Miniaturization and cost reduction are desirable attributes in any analytical system. For this reason, we have recently been attracted to electrochemical detection schemes for H2O2. H2O2 can be determined electrochemically by anodic oxidation to O2 or by cathodic reduction to H2O. In clinical chemistry, H2O2 is an important analyte as well. In biological fluids, interferences are more common in the oxidative determination mode due to the presence of easily oxidizable substances like ascorbate; reductive determination is therefore preferred. In atmospheric measurements, oxidants such as NO2 and O3, and even O2 (due to its preponderance and significant aqueous solubility), can cause problems in the reductive determination of H2O2; the oxidative mode would therefore be preferred. Kulys16 developed a flow-through amperometric sensor to directly determine hydrogen peroxide in gaseous media by collecting gaseous H2O2 in a wetted cellulose film. Anodic oxidation at a platinum working electrode was used. The limit of detection was 10 ppbv H2O2(g). This is completely unacceptable for atmospheric measurements which require limits of detection (LODs) on the order of 0.1 ppbv. Initially, we attempted to design a more sensitive sensor for H2O2(g) along the same lines as that of Kulys, where collection takes place on a wetted Nafion surface and the H2O2 is then electrochemically oxidized in situ.17 We could achieve an LOD of 40 pptv H2O2(g), but the collection efficiency was found to be extremely dependent on the sample relative humidity (rh), increasing with increasing rh. The rh dependence was judged to be too great for such a sensor to be practical. It has been established that the DS does not display any marked dependence on rh for the collection of H2O2(g), nor does it have any significant interference from common atmospheric gases.8 Considering that a dilute electrolyte solution can be used as the scrubber liquid, the direct measurement of the H2O2 in the DS effluent by anodic oxidation should be possible. A smallvolume flow-through three-electrode electrochemical cell can be readily assembled for this purpose. Personal computers (PCs) (14) Lazrus, A. L.; Kok, G. L.; Lind, J. A.; Gitlin, S. N.; Heikes, B. G.; Shetter, R. E. Anal. Chem. 1986, 58, 594-597. (15) de Serves, C.; Ross, H. B. Environ. Sci. Technol. 1993, 27, 2712-2718. (16) Kulys, J. Sens. Actuators B 1992, 9, 143-147. (17) Huang, H.; Dasgupta, P. K. Unpublished studies, Texas Tech University, 1994-95. S0003-2700(96)00143-6 CCC: $12.00
© 1996 American Chemical Society
Figure 1. PC-based electrochemical detection system. A1-A3, FET input operational amplifiers; S, sensor; C, R, and W, counter, reference, and working electrodes; F12DB, flash-12 and daughter board housing D/A and A/D units and digital output-controled relays (D-O-1, D-O-2) that govern valves V1 and V2.
are commonly used today as an integral part of most continuous measurement instrumentation for data acquisition. It is possible to use simple and affordable electronics in conjunction with a PC. Although the DS coupled arrangement is perhaps not conceptually as elegant as a direct gas phase sensor, such a system will nevertheless be an important step forward for facilitating the measurement of atmospheric H2O2. EXPERIMENTAL SECTION Reagents and Materials. A stock solution of 1 M H2O2 was prepared from 30% H2O2 (Fisher) and standardized by iodimetry.18 Working concentrations of H2O2 at the micromolar level were prepared daily. Nafion tubing (dry dimensions 0.5 mm o.d., 0.4 mm i.d.) was obtained from Perma-Pure Products (Toms River, NJ). Platinum, platinum/rhodium (30%), and silver wires were obtained from Aesar, Inc. Instrumentation. Figure 1 shows a computerized H2O2 analyzer consisting of a potentiostat and a current-voltage converter. An LF353N (National Semiconductor) dual BiFET operational amplifier (A1 and A2) is assembled into a potentiostat to provide a constant potential for conventional amperometry or a pulse potential for pulse amperometry. The A3 amplifier functions as a current-voltage converter to monitor the sensor current. In the H2O2 sensors, the reference electrode R, the counter electrode C, and the working electrode W are respectively connected to the noninverting input of A2, to the output of A1, and to the inverting input of A3. The rotary switch P selects the feedback gain resistor (10 kΩ-10 MΩ). The PC (80486 class processor operating at 33 MHz) is interfaced with the electronic circuit via a 12-bit A/D-D/A board (Flash-12 High Speed Analog & Digital I/O Card, Strawberry Tree Inc., Sunnyvale, CA) for controlling instrument operation and data acquisition. The digitalto-analog converter can generate a constant voltage for amperometric measurement or a pulse wave form for pulse amperometric measurement. The A/D converter acquires data from the A3 output. The digital outputs D-O-1 and D-O-2 provide a means for (18) Vogel, A. I. A Textbook of quantitative Inorganic Analysis, 3rd ed.; Longmans Green: London, 1961.
Figure 2. Analyzer schematic. NMDS, Nafion membrane-based diffusion scrubber; S, sensor; MFC, mass flow controller; P, air suction pump; C, H2O2 removal cartridge; V1, air switching valve; V2, on/off valve.
controlling valves (vide infra). Instrument control and operation is implemented with software written in-house in C and is available from the authors on request. The operational details are given below. Test Arrangement. The sampling arrangement is shown in Figure 2. Gas flow rate is metered by the mass flow controller MFC (Model FC-280, Tylan General, Torrance, CA). At higher flow rates, the amount collected becomes virtually independent of the flow rate, and the use of a sophisticated flow controller is not essential.8 To generate H2O2-free air, ambient air is purified by column C, containing sequential beds of silica gel and activated carbon (or granular MnO2). Either this H2O2-free air or ambient air is selected by the three-way PFA Teflon valve V1 (Fluoroware, Chaska, MN) to flow through the NMDS. The NMDS construction has been described in detail previously.10 The gas is aspirated by the pump at a flow rate determined by the flow controller (1 L/min, unless otherwise stated). An aqueous solution containing 1 mM HCl is used as the scrubber liquid for collecting H2O2(g). Normally, the two-way valve V2 is off, there is no liquid flow, and H2O2 is preconcentrated by the NMDS. The HCl solution is protected from the intrusion of H2O2 by a MnO2 trap. When valve V2 is on, the HCl solution flows by gravity through the NMDS and carries the collected H2O2 to the sensors. When valve V2 is on, the solution flows through the sensor, and the sensor current is proportional the H2O2 concentration in the solution. The system was calibrated with hydrogen peroxide generated by a membrane source4 based on Henry’s law. Except as stated otherwise, a concentration of 1 ppbv H2O2(g) was used in the experiments. Sensor Fabrication. Figure 3 shows the construction of the sensor. Two ∼10 mm long segments of Nafion tubing are slipped over a 0.5 mm diameter Ag wire, leaving a ∼1 mm segment of bare Ag surface exposed between the two tube segments. The exposed Ag is anodically chloridized for 5 min in a solution of 0.1 M HCl. This functions as the reference electrode. A 0.2 mm diameter Pt wire is coiled onto one segment of the Nafion tubing to function as the counter electrode, and a 0.25 mm diameter Pt/ Rh (30%) wire is coiled onto the other, upstream, segment of the Nafion tubing to function as the working electrode. The assembly is then put inside a PTFE tube segment ∼2.5 mm i.d. and ∼15 Analytical Chemistry, Vol. 68, No. 13, July 1, 1996
2063
Figure 3. Sensor schematic. See text for construction.
Figure 5. Cyclic voltammograms of 0.2 mM H2O2 in 0.5 mM HCl and a HCl blank with and without Nafion when using a Pt or a Pt/Rh working electrode, showing the utility of the Pt/Rh electrode and the Nafion solid polymer electrolyte.
Figure 4. Temporal protocol for the operation of the sensor and instrument. The times shown is for a 10 min analytical cycle. For other cycle times, the duration of the initial period at 0.3 V is altered.
mm long, such that the Nafion-covered ends protrude from the PTFE tube at both ends. Narrow-bore PTFE inlet/outlet tubes (0.6 mm i.d., 1.2 mm o.d.) are inserted into each end of the larger PTFE tube, one on the top and the other on the bottom of the electrode assembly to juxtapose the latter in place at an angle. Epoxy adhesive is then liberally applied to cover up the exposed ends as well as the whole assembly, with the electrode connections protruding. The cell is not operated at any significant back pressure, so the fact that there is little adherence of the epoxy on PTFE does not pose any problems. The temporal protocol for system operation is shown in Figure 4 for a 10 min analytical cycle. During the 490 s quiescent period t1 (this period is lengthened or shortened depending on the desired sampling time), the working electrode is maintained at 0.3 V. Next, the electrode is cleaned for 2 s at 0.0 V, and the electrode is pulsed high to 0.8 V for the next 108 s. Fifty-eight seconds into the high pulse period, liquid valve V2 is turned on (Figure 2). With a liquid flow rate of ∼0.6 mL/min, the peak appears within 6 s. Data are acquired for 15 s from the time V2 is turned on. Peak heights are automatically calculated and stored; peak areas are calculated by integration over this 15 s period. At the end of the data acquisition period, the air sampling valve V1 is switched to admit H2O2-free air in the system. Twenty seconds after V1 is switched, data corresponding to the blank response are collected for another 15 s for blank corrections. At the end of this period, V1 and V2 are both switched again, the working electrode is returned to 0.3 V, and sampling begins again. 2064 Analytical Chemistry, Vol. 68, No. 13, July 1, 1996
RESULTS AND DISCUSSION Sensor Parameters. Several flow cell designs and working electrode materials were investigated; the best combination is reported here. It is well known that there is a large overpotential for the oxidation or reduction of H2O2 with most electrode materials. It is also recognized that the key to a sensitive H2O2 sensor must lie in a facile means to lower this overpotential. In implantable glucose monitors that work via sensing H2O2, enzymes or enzyme mimics have been successfully used to lower the overpotential.19-23 Gorton24 reduced the overpotential by >800 mV by depositing Pd/Au on glassy carbon. An oxidized Mn coating on glassy carbon has also been found to significantly lower the overpotential of H2O2 oxidation.25 Wang and Chen26 have shown that rhodium on carbon paste is also very effective. The sensor configuration chosen by us is not compatible with glassy carbon or carbon paste electrodes; a flexible wire-based working electrode is preferred. In this work, we have discovered that commercially available Pt/Rh alloys (30% Rh) offer very attractive performance as the working electrode in the anodic detection of H2O2. The cyclic voltammogram (CV) in Figure 5 shows the marked improvement in performance when Pt/Rh is used as the working electrode, relative to Pt. Indeed, this Pt/Rh alloy offers superior performance over either pure Pt or pure Rh (not shown). In the present system, the liquid flowing through the DS has a very low ionic strength; the Nafion tubing in the sensor appears to play an important role in this system. In the presence of Nafion, the CV of 200 µM H2O2 in 0.5 mM HCl clearly shows a (19) Tatsuma, T.; Okawa, Y.; Watanabe, T. Anal. Chem. 1989, 61, 2176-2178. (20) Tatsuma, T.; Okawa, Y.; Watanabe, T. Anal. Chem. 1992, 64, 1183-1187. (21) Vreedke, M.; Maidan, R.; Heller, A. Anal. Chem. 1992, 64, 3084-3090. (22) Garguilo, M. G.; Huynh, N.; Proctor, A.; Michael, A. C. Anal. Chem. 1993, 65, 523-528. (23) Cso ¨regi, E.; Gorton, L.; Marko-Varga, G.; Tu ¨ do ¨s, A. J.; Kok, W. Th. Anal. Chem. 1994, 66, 3604-3610. (24) Gorton, L. Anal. Chim. Acta 1985, 17, 247-253. (25) Taha, Z.; Wang, J. Electroanalysis 1991, 3, 215-219. (26) Wang, J.; Chen, Q. Anal. Chem. 1994, 66, 1007-1011
Figure 6. Liquid phase response of the sensor in the flow injection mode (1.5 mL/min, 100 µL injected). The inset shows the response to 200 nM sample with a 10-point (0.5 s) running average smoothing routine applied.
well-defined redox process (Figure 5); when the sensor is constructed without the solid polymer electrolyte, none of these features can be seen. Regarding the choice of a reference electrode, Ag/AgCl is both convenient and easily miniaturized; further, it is very affordable. However, a minimum concentration of chloride is necessary for the electrode to maintain a constant reference potential. Since diffusion scrubbers are not particularly compatible with saturated salt solutions (evaporation causes saline deposits), we chose 1 mM HCl as the scrubber liquid. This solution also functions as the reference electrolyte. All potentials referred to in this paper are relative to this reference; note that this electrode has an absolute potential ∼0.2 V more positive than the Ag/AgCl electrode based on saturated KCl. Based on the CV data, we chose pulse potentials of 0.8 V for monitoring H2O2 and 0.3 V for the quiescent mode of the electrochemical cell. Liquid Phase Measurements. The sensor was first tested in the flow injection mode. Here, 1 mM HCl was gravity pumped at 1.5 mL/min through a minicolumn packed with granular MnO2 to remove any residual H2O2. H2O2 standards were prepared in 1 mM HCl, and 100 µL aliquots were injected. Over the range 200 nM-4 µM H2O2, both peak areas and heights were linearly related to the injected concentration (r2 ) 1.0000). Representative sensor response is shown in Figure 6. The estimated LOD is ∼30 nM, an order of magnitude better than that reported in a recent publication in this Journal.23 Gas Phase Experiments. (i) Dependence on Liquid Flow Rate. With 1 ppbv H2O2(g) sampled for 5 min, the liquid flow rate was varied from 0.38 to 1.20 mL/min by varying the hydrostatic head from 30 to 90 cm. While the peak shapes became broader at slower flow rates, the peak height at five different flow rates within this range was remarkably constant, 96.5 ( 2.6 nA. Predictably, peak area variation was greater, with a relative standard deviation of 7.9%. However, even this variation would be negligible as a result of variations of changes in the flow rate that can be expected in a constant head gravity-induced flow system. The constancy of the peak height is particularly
Figure 7. Comparison of ambient air data collected in Lubbock, TX, by the electrochemical instrument and a fluorometric instrument. Note that the latter instrument is more sensitive and that the midnight-7 a.m. values from the electrochemical instrument are essentially at or below the LOD of that instrument.
fortunate because the need for a pump to provide a highly constant flow rate is obviated. Further experiments were carried out with a hydrostatic head of ∼50 cm. (ii) Dependence on Gas Flow Rate. The signals observed were 75, 101, 125, and 133 nA at sample gas flow rates of 0.50, 1.00, 1.50, and 2.00 L/min, respectively. An initial increase in response with sample flow rate, followed by the attainment of a virtual plateau at higher flow rates, is typical of diffusion-based collection systems.27 A flow rate of 1 L/min was used henceforth. (iii) Performance and Intercomparison Results. For 1 ppbv H2O2, the sensor response was linearly related to the sample preconcentration period (sample gas flowing, no liquid flow) over a range of 1.5-9 min by the following equation:
peak area, nC ) 91.08 ( 1.266(tpreconcentration, min) + 240.1 ( 7.3, r2 ) 0.9994 (1)
For a constant preconcentration period of 4 min, the response was reasonably linear with the H2O2 concentration sampled over the range of 0-2 ppbv, according to the equation
peak area, nC ) 192.6 ( 17.3(H2O2, ppbv) + 172.1 ( 19.2, r2 ) 0.9683 (2)
showing some departure from linearity at levels below 0.5 ppbv. The responses for a blank and a 0.37 ppbv H2O2 standard were 204.2 ( 5.3 and 228.0 ( 4.0 nC, respectively, from which an LOD of 0.24 ppbv can be estimated. With an increased preconcentration period of 9 min (analytical cycle time of 10 min), the LOD improves proportionally to 0.11 ppbv. This LOD and time resolution are acceptable for most stationary measurement needs (27) Dasgupta, P. K. ACS Adv. Chem. Ser. 1993, 232, 41-90.
Analytical Chemistry, Vol. 68, No. 13, July 1, 1996
2065
for H2O2. Therefore, an in-house intercomparison of the fluorometric method11 was conducted with a 10 min analytical cycle time. The two instruments were calibrated from a common source. After the calibration, an individual experimenter was in charge of each instrument, and each made ambient air measurements using a common manifold in our laboratories in Lubbock, TX (lat. 33°35′ N, long. 101°50′ W, altitude 988 m MSL), over several days in the winter of 1995-96. Each experimenter was deliberately kept blind to the other data set. Representative data from the two instruments are shown for one day in Figure 7. The absolute maximum corresponds well with wintertime values previously measured at this location, and the two disparate measurement techniques clearly correlate very well.
2066
Analytical Chemistry, Vol. 68, No. 13, July 1, 1996
ACKNOWLEDGMENT This work was partially supported by the Office Of Naval Research, through ONR Grant No. N00014-94-1-0295. This manuscript has not been subject to review by ONR and no endorsements should be inferred.
Received for review February 14, 1996. Accepted April 10, 1996.X AC960143X X
Abstract published in Advance ACS Abstracts, May 15, 1996.
