Amperometric determination of nitrogen dioxide in air samples by flow injection and reaction at a gas-liquid interface

A number of methods for the determination of NO2(g) in air by use of spectrophotometric [1 and 2], fluorometric [3], chemiluminescent [4], and electrochemical [5], detection have been reported. These methods require trapping and preconcentration of NO2 as nitrite, a process with unreliable stoichiometry [6], and are.either time consuming, or use expensive instrumentation. The method proposed here, while not as sensitive as some of the above, does not require expensive instrumentation nor pretrapping steps and thus allows for determination of real-time NO concentrations. The determination is based on the NO oxidation of tris 1,10-phenanthroline)-iron (I I) [ferroin] to its iron(III) analogue at a gas-liquid interface. Two continu-ous-flow manifolds for sample processing are described and compared. One incorporates a gas diffusion step in which a gas-permeable membrane separates a donor (sampling) stream from an acceptor (detecting) stream. Some of the air sample containing NO diffuses across the membrane into the detecting stream and there causes ferroin oxidation. In the other manifold, the NO-air sample is directly intercalated into the ferroin-containing liquid carrier; oxidation occurs at the gas-liquid interface. The extent of ferroin oxidation is amperometrically monitored at a carbon paste electrode and is proportional to the NO2 concentration in the air sample. The ferroin/ferriin redox couple is a reversible electrochemical system in which oxidation/reduction occurs without affecting the chemical structure of the 1,10-phenanthrol-ine ligand. Some information on the electrochemistry of the ferroin/ferriin couple can

A number of methods for the determination of NO2(g) in air by use of spectrophotometric [1 and 2], fluorometric [3], chemiluminescent [4], and electrochemical [5], detection have been reported. These methods require trapping and preconcentration of NO2 as nitrite, a process with unreliable stoichiometry [6], and are.either time consuming, or use expensive instrumentation. The method proposed here, while not as sensitive as some of the above, does not require expensive instrumentation nor pretrapping steps and thus allows for determination of real-time NO concentrations.
The determination is based on the NO oxidation of tris 1,10-phenanthroline)-iron (I I) [ferroin] to its iron(III) analogue at a gas-liquid interface. Two continuous-flow manifolds for sample processing are described and compared. One incorporates a gas diffusion step in which a gas-permeable membrane separates a donor (sampling) stream from an acceptor (detecting) stream.
Some of the air sample containing NO diffuses across the membrane into the detecting stream and there causes ferroin oxidation. In the other manifold, the NO-air sample is directly intercalated into the ferroin-containing liquid carrier; oxidation occurs at the gas-liquid interface. The extent of ferroin oxidation is amperometrically monitored at a carbon paste electrode and is proportional to the NO2 concentration in the air sample. The ferroin/ferriin redox couple is a reversible electrochemical system in which oxidation/reduction occurs without affecting the chemical structure of the 1,10-phenanthroline ligand. Some information on the electrochemistry of the ferroin/ferriin couple can be found in the literature [7]. An advantage of the reversibility of this redox process is that the ferriin produced during the oxidation with NO can be electrochemically reduced back to ferroin in a reagent reservoir and thereby can be used in a closed-loop system [8]. A similar direct intercalation procedure for the determination of SO2(g) has been reported [8,9] in which SO2(g) reduces di--hydroxo-bis[bis(1,10-phenanthroline)iron(III)] at a gas-liquid interface. The product of the reduction, ferroin, can be photometrically monitored at 510 nm [8] or amperometically at 965 mV (versus an Ag/AgC1 reference electrode) [9].

Results and discussion
Determinations by gas diffusion The effect of varying the detection potential from 700 to 900 mV vs SCE is shown in table 1. These measurements were made with an NO concentration (v/v) of 100 ppm, a 70-tl sample volume, a ml/min flow rate, a 25 cm length of reactor tubing from the gas diffusion cell to the detector, and a 1"00 x 10 .5 M ferroin solution (pH 5"00). From the table it is clear that the optimum detection potential, with respect to signal size, is between 780 and 900 mV. It was found, however, that a substantial increase in base-line noise resulted when potentials in excess of 800 mV were used. Consequently all subsequent measurements were made at an applied potential of 780 mV. At this potential, the base-line current corresponds to the oxidation of ferroin to ferriin. The direct intercalation system is shown in figure 2. A 5 cm length of Gore-Tex microporous PTFE tubing (1 mm i.d.) of 3"5 gm pore size (Anspec, Ann Arbor, Michigan, USA) was used as a debubbler [11]. This tubing functioned best in the flow system when placed vertically with the solution flowing downwards. The tube between the gas diffusion cell and detector in figure l(a) and injector to detector in figure 2 had a single-beadstring-reactor configuration produced by incorporation of glass beads 0"6 mm in diameter inside mm i.d. tubing. Such reactors enhance the mixing performance in flow injection systems [8 and 12]. The signals recorded retlect the decrease in the current as a result of ferroin oxidation by the NO2 contained in the intercalated sample.
Determination of NO2(g) based on the detection of the oxidation product (ferriin) is possible at potentials less than 630 mV, however, the sensitivity and precision of such measurements was poor when compared to measurements at 780 mV (ferroin detection). Figure 3 shows the effect of varying the length of reactor tubing from the gas diffusion cell to the detector. The optimum length, at a flow rate of ml/min, fell in the range of 20-25 cm. Shorter lengths resulted in insufficient time for signal realization and greater lengths produced substantial dispersion (dilution). A systematic study of the effects of ferroin concentration and hydrogen ion concentration led to the choice of 1"00 x 10 .5 M ferroin and pH 5"00. Figure 4 shows typical signal profiles using the conditions chosen as best. The equation for calibration curves derived from these signals was: (current in na) 0"3 [NO2(g)] 0"2 with the NO2(g) concentration in ppm (v/v) and a correlation coefficient of 0-9994. The limit of detection (defined as the average of blank signals plus three times its standard deviation) was found to be 12 ppm. Ten successive measurements of 50 ppm NO gave a relative standard deviation of 2"1%. Determination could be made at a sampling frequency of 35/h. The use of air carrier instead of a liquid one resulted in smaller signals, presumably because of a larger dispersion (dilution) taking place between the point of intercalation and the gas diffusion cell.

Determinations by direct intercalation
Ferroin concentration, pH, detection potential, flow rate, and sample volume were the same as those used in the gas diffusion determinations. A 1.0 m length of tubing was used from intercalator to detector. Reactor length (cm) ppm, B 2"8 ppm, C 4"6 ppm, D 5"8 ppm, E 7"6 ppm, F 10"6 ppm. Figure 5 shows typical flow injection profiles recorded in the range of 1"0 to 10"6 ppm by the direct intercalation procedure. The negative peaks partially cut by positioning of the recorder pen occurred because the gas sample reached the debubbler and disturbed the flow continuity.
No flow throughout the detector was sustained during the exit of gas through the debubbler. Amperometric detection is known to be sensitive to flow fluctuations 14]. The following equation describes the calibration curve derived from signals obtained by direct intercalation: (Current in na) 10"2 [NO2(g)] 0"3 with a correlation coefficient of 0"9998. The limit of detection was found to be 0"5 ppm. Ten successive determinations of 4"5 ppm NO2 (v/v) gave relative standard deviation of 2"9%. Determinations could be made at a frequency of 25 samples/h.
This intercalation procedure is more sensitive than the gas diffusion one for three reasons: (1) there is a longer contact time between the air sample and the ferroin; (2) the contact area for the NOg-ferroin reaction is larger; and (3)  THE CHEMICAL SENSORS CLUB The Chemical Sensors Club has been formed to ensure that all organizations in the UK concerned with the exploitation of chemical sensor technology are as well co-ordinated as possible.
The advantages of Club membership are: (1) The Club provides a formn through which member organizations can make known their capabilities and needs.
(2) Two general meetings are organized each year to enable members to discuss particular aspects of chemical sensor technology and to establish personal contacts with each other.
(3) A members' handbook contains printed information generated or compiled by the Club covering current areas of interest of member organizations, available hardware, UK research activities, details ofconsultancy services and any other more detailed reports on specific aspects of sensors.
(4) Members also receive a regular newsletter covering Club activities and topical items ofinterest in the sensors field.
(5) Data-bases are being established to allow members to gain access to collated information on user needs, available sensor systems and research activities in progress.
(6) Club membership may entitle members to reduced subscription rates for other DTI funded groups active in the sensors field.
(7) Through membership of the Club, organizations will be assisted to join together to form consortia, specific to their interests, to develop new chemical sensors and sensor applications.
Details of membership from Mr D. G. Porter, Laboratory of the Government Chemist, Cornwall House, Waterloo Road, London SE1 8XY.