Validation of a diVusive sampler for NO2† Annika Hagenbjo�rk-Gustafsson,*a Roger Lindahl,a Jan-Olof Levina and Doris Karlssonb aNational Institute for Working Life, Department of Chemistry, P.O. Box 7654, S-907 13 Umea° , Sweden bDepartment of Environmental Health, Umea° University, S-901 87 Umea° , Sweden Received 13th April 1999, Accepted 27th May 1999 A diVusive sampler for NO2, Willems badge, was validated in laboratory experiments and field tests.The collecting reagent for NO2 in the sampler is triethanolamine, and the analysis is based on a modified colorimetric method, the Saltzman method. The analysis was performed by a flow injection analysis (FIA) technique. The sampling rate for the sampler was determined to be 40.0 ml min-1. There was no eVect of NO2 concentration or relative humidity on sampling rate, and the influence of sampling time was found to be small.The detection limit was 4 mg m-3 for a 24 h sample. The capacity is high enough to allow sampling of 150 mg m-3 for 7 days, which is twice the recommended Swedish short-term (24 h) guideline value as a 98-percentile over 6 months. In field tests, the sampler performed well, even at wind speeds higher than 2 m s-1, and at low temperatures.The overall uncertainty of the method was 24%. The sensitivity and capacity of the method also make it suitable for personal sampling for 2–8 h in working environments. the spatial variation. They are also valuable for the determi- Introduction nation of background levels over longer periods of time. Nitrogen oxides (NO and NO2) are produced as by-products Palmes et al.4 developed the first diVusive sampler for NO2, in various combustion processes, especially at high tempera- the Palmes tube, with a rather low sampling rate.A badgetures. Primarily NO is produced, but it oxidizes in the atmos- type sampler for personal sampling, with higher uptake rate, phere to form the more harmful nitrogen dioxide.The main was developed by Yanagisawa and Nishimura.5 In 1998, Ferm sources of nitrogen oxides in urban areas are motor traYc and Svanberg3 introduced another design of the badge-type emissions and burning of fossil fuels in power plants. High sampler with a rather large opening-to-length ratio, resulting indoor concentrations of nitrogen dioxide can be found in in a device more sensitive than the tube-type sampler.At the association with the use of gas stoves and unvented gas and University of Wageningen in the Netherlands, Willems and kerosene space heaters.1 Hofschreuder6 developed a diVusive sampler with a geometry A number of studies have indicated that human exposure similar to Ferm’s, originally for ammonia measurements, and to nitrogen dioxide is associated with increased susceptibility later adapted for measuring NO2.The Willems badge was used in the European PEACE (Pollution EVects on Asthmatic to airway infections and impaired lung function. Correlations Children in Europe) study to measure concentrations of nitro- between NO2 exposure and respiratory illness and increased gen dioxide inside and outside the homes of asthmatic severity of asthma, as well as increased response to inhaled children.7 allergens in asthmatics, have been found.2 In a recent study, van Reeuwijk et al.8 used the badge for There are diVerent techniques for the determination of measuring the 2-week average NO2 concentrations in three nitrogen dioxide in ambient air.These techniques can be European areas within the EU SAVIAH (Small Area divided into active or passive sampling.The active methods Variations in Air Quality and Health) project. The Willems include impinger methods, where NO2 is collected by bubbling badge was compared with the tube-type sampler and a refer- air through a solution in which NO2 is reduced to NO2- and ence method. The validation was, however, insuYcient as the analysed colorimetrically.A modification is the impregnated comparison between the reference method and the diVusive sintered glass filter technique.3 Another active technique of sampler only included a total of nine measuring points. In monitoring NO2 is the chemiluminescence technique in which addition, no sampling rate was determined. emission from excited NO2* is detected photo-electrically.The No extensive laboratory test of the performance of the chemiluminescence analyser is a very sensitive technique, which sampler has been carried out. This paper describes the vali- oVers hourly time-weighted averages of the NO2 concentration. dation of the Willems badge for NO2 measurements. The aim The analysers are, however, expensive, need electricity, caliwas to experimentally determine the sampling rate of the bration and specialist maintenance, and are for that reason diVusive sampler in laboratory studies, and to investigate the not suitable to assess spatial variation.eVects of sampling time, concentration of NO2 and relative DiVusive (passive) samplers are based on the molecular humidity on sampling rate. Another objective was to confirm diVusion of the gas to a collector medium.DiVusive samplers the results from the laboratory tests in field studies, in con- are ideal for the purpose of monitoring ambient air. They are ditions of low temperature, which are common in Northern small, cheap, easy to handle and enable measurements to be Europe during the winter season. conducted at remote places, as they require no pump or electricity.They allow the measurement of NO2 levels at Experimental various places in a town at the same time, in order to analyse DiVusive sampler The Willems badge (Fig. 1) consists of a cylinder of polystyrene. The absorption filter, a Whatman GF-A glass fibre †Presented at AIRMON ’99, Geilo, Norway, February 10–14, 1999. J. Environ. Monit., 1999, 1, 349–352 349of lowest concentrations (blank, 0.03, 0.05, 0.2 mg l-1 NO2).The limit of quantification (12 mg m-3 for 24 h; 2 mg m-3 for 7 days) was determined as 10 times the mean standard deviation for the same samples.10 The repeatability (1.5%) was determined as the relative standard deviation for six replicates of the samples that exceeded the limit of quantification (0.2, 1.0, 3.0 and 5.0 mg l-1 NO2).The reproducibility (4%) of the method was determined as the relative standard deviation of a control sample, run together with normal sample analysis seven times during a period of 8 months. Fig. 1 Expanded view of the Willems badge: a, sampler base of polystyrene; b, absorption filter; c, spacer ring; d, Teflon filter; DiVusive sampler laboratory tests e, fixation ring; f, cap of polyethylene.Generation of standard atmospheres of nitrogen dioxide. filter, is placed at the bottom of the cylinder. By placing a Known concentrations of nitrogen dioxide were generated in Teflon filter on a distance ring of polystyrene (6 mm), a region an exposure chamber (70×48×900 mm) shown in Fig. 2. without turbulence is created. The Teflon filter (Schleicher & Nitrogen dioxide (23.3 ppm±3% NO2 in nitrogen; AGA, Schuell TE 38, 5 mm) is secured with a polystyrene ring of Stockholm, Sweden) from a gas cylinder regulated by a mass 3 mm.A polyethylene cap closes the badge. The absorption flow controller (0–500 ml min-1) was diluted by clean, humidi- filter is impregnated with a solution of triethanolamine–acetone fied air, controlled by a mass flow controller (0–200 l min-1) as a collector for NO2.Sampling is initiated by removing the and mixed before the entrance to the exposure chamber. In cap from the sampler, which is exposed with the open end order to produce diVerent relative humidities, the air was down. In field studies, the sampler is attached to an angled passed through one to four gas-dispersion bottles containing aluminium plate which shelters the badge from rain and snow.water. By modifying the number of bottles by means of valves, After a sampling period, the absorption filter is removed from the relative humidity of the air could be varied from 20 to the badge, extracted and analysed for nitrite. 90%. The air flow through the exposure chamber was The filters and sampler components were cleaned before use 40 l min-1, whicgave an air velocity in the chamber of 0.3 as follows.The sampler components were immersed in 96% m s-1. The exposure chamber has been described in detail ethanol. Teflon filters were dipped twice in a solution of 96% previously.11 ethanol–distilled de-ionized water 151 (v/v), dried and then immersed in 96% ethanol.Before coating, the glass fibre filters Experimental design. The sampling rate was experimentally were boiled in water for a few minutes to remove loose glass determined by exposing samplers in the exposure chamber. To fibres, dried, immersed twice in a solution of acetone and then investigate the eVects of sampling time, concentration of NO2 dried again. Both Teflon filters and glass fibre filters can be and relative humidity on the sampling rate of the diVusive stored for about 3 months in a closed vessel after cleaning. sampler, a laboratory test with a factorial design was per- The filters were coated by dipping in a fresh solution of formed.Six samplers were exposed simultaneously to nitrogen triethanolamine (Riedel de Hae�n, p.a.) in acetone (Merck, dioxide levels from 5 to 150 mg m-3, with exposure times from p.a.), 1550 (v/v), and then dried for about 1 min in a ventilated 1 day to 7 days.The relative humidity was varied between 20 desiccator equipped with an entrance filter coated with tri- and 80%. The temperature was 20 °C. Two laboratory blanks ethanolamine as absorbent for NO2. The samplers were were collected in each experiment, and the mean value was mounted directly after drying the filters. Samplers loaded with subtracted from the amount of nitrite in the exposed samples. coated filters were stored in a refrigerator (+8 °C) for up to As a reference method, NO2 was measured with a chemilumi- 1 month.nescence instrument (ECO PHYSICS CLD 700 AL med, Du� rnten, Switzerland), with the inlet placed in a port at the Analysis centre of the exposure chamber.A daily calibration of the analyser was performed using a certified gas of NO in nitro- Absorbed nitrogen dioxide on the filter was determined colorigen (AGA). metrically as nitrite. The analysis is based on a modified colorimetric method, the Saltzman method,9 and was per- Field validation formed by the flow injection analysis (FIA) technique.The absorption filter of a badge was soaked in 5 ml of 0.005 M Two studies of outdoor simultaneous measurements with NaOH (Eka Nobel, Goteborg, Sweden), shaken for 30 min passive samplers and a chemiluminescence instrument were and centrifuged at 4000 rpm for 10 min. After centrifugation, performed during the winter season. Both studies included one the centrifugation tubes were placed in an autosampler in the sampling period of 6 days and three sampling periods of 2 FIA analyser (FIA Tecator Star 5010 analyser, Controller days. Six parallel samplers were placed on an outer wall of a 5032, Sampler 5027, Ho�gano� s, Sweden).The sample was injected into a carrier stream of water with a flow of 1.5 ml min-1. On the addition of sulfanilamide (10 g l-1, Merck p.a.; flow, 0.6 ml min-1), a diazo compound is formed which then reacts with N-(1-naphthyl )ethylenediamine dihydrochloride (1.0 g l-1, Merck) provided from another reagent stream (flow, 0.6 ml min-1).An azo dye is formed and the colour intensity is measured in a 10 mm flow cell at 540 nm. To validate the FIA and to determine the detection limit, limit of quantification, repeatability and reproducibility of the method, seven samples of diVerent concentrations were run in six replicates.The detection limit (4 mg m-3 for 24 h; 0.6 mg m-3 for 7 days) was determined as three times the mean Fig. 2 Equipment for the generation of standard atmospheres of NO2. MFC, mass flow controller. standard deviation for the mass on the filters of four samples 350 J.Environ. Monit., 1999, 1, 349–352building. To protect the samplers from rain and snow, they The influence of sampling time, concentration of NO2 and relative humidity on the sampling rate of the diVusive sampler were attached to the underside of an angled aluminium plate. Blanks were included in the same manner as in the laboratory was statistically analysed by multiple regression.14 As shown in Table 2, there was a small, but statistically significant (P= experiments.The inlet of the chemiluminescence instrument was placed close to the diVusive samplers. Temperature, wind 0.05) negative eVect of sampling time, but no eVects of relative humidity or concentration of NO2 on sampling rate. The small velocity and relative humidity were recorded by a datalogger every 30 s during the sampling periods.In the first study, the influence of the sampling time does not aVect the usefulness of the method, and this influence is included in the relative mean nitrogen dioxide concentration ranged from 2 to 9 mg m-3. The mean temperature varied between -5.2 and standard deviation of 22%. -6.4 °C, the relative humidity between 48 and 75% and the mean wind velocity in the four sampling periods ranged from Field validation 0.5 to 1.7 m s-1.During the second series, the mean NO2 The field validation was performed during the winter when concentration in the four sampling periods varied between 8 the temperature was low, and the possibility to record increased and 20 mg m-3, the mean temperature between -2 and nitrogen dioxide concentrations was high, due to inversion -17.6 °C, the mean relative humidity between 73 and 82% and burning of fossil fuels.However, no inversion occurred. and the mean wind velocity range was 1.2 to 1.8 m s-1. As can be seen from Fig. 3, there was a good correlation between the concentrations of NO2 found by diVusive sampling Results and discussion and by the chemiluminescence instrument, with a coeYcient of correlation of 0.95.The mean ratio between the concen- DiVusive sampling trations obtained by Willems badge and chemiluminescence was 1.08, and the relative standard deviation for six samplers According to a simplification of Fick’s law, the concentration (C) of an analyte in air can be calculated if the uptake rate of in eight runs was 7%, as shown in Table 3.The field data given in Table 3 were submitted to multiple regression analy- the sampler, the amount of analyte (m) in the sampler and the sampling time (t) are known sis.14 No statistically significant eVects on the badge sampling rate of the exposure time, concentration of NO2, relative humidity, wind velocity or temperature were found.The wind m t =DA C L =SC speed varied within a wide range during the field experiments. Some of the wind speed variations are shown in Table 4. Run where D is the diVusion coeYcient (cm2 s-1), A is the cross- 5 covers the 6 days of measuring in runs 6, 7 and 8. As can sectional area of the badge (cm2), L is the length of the be seen from Table 4, wind speeds in excess of 2 m s-1 occurred diVusion path (cm), m is the amount of analyte on the filter 20 and 31% of the time, respectively, in runs 6 and 7.Although (g), t is the sampling time (s) and C is the concentration not statistically significant, this could be a reason for the (mg cm-3). increased ratio between the diVusive sampler and the reference The theoretical sampling rate is given by DA/L (cm3 s-1) method in runs 6 and 7 (Table 3).In run 8, where the wind and can be calculated from the geometry of the sampler. speed was <0.3 m s-1 for 38% of the time, the ratio was 0.93. However, the sampling rate of a diVusive sampler must be A wide range of temperatures was also covered in the field verified experimentally in accordance with existing stanexperiments, and the sampler performed well, although no dards.12,13 The Willems badge was partly evaluated according corrections for temperature were performed on the sampling to the draft European standard for the validation of diVusive rate.Since no significant factors were identified, temperature samplers for ambient air measurements.13 Laboratory tests Table 2 Multiple regression analysis of the influence of sampling time, concentration and relative humidity; S, significant; NS, not significant; The results obtained in the factorial laboratory study are Rh, relative humidity shown in Table 1.Thempling rate was 40.0 ml min-1 with a relative standard deviation for the 42 experiments of Variable Parameter estimate Standard error 22%. The concentrations measured by the reference method Sampling rate 40.0 2.2 (chemiluminescence instrument) were taken as the true values, Time -8.6 2.8 S and the uptake rate of the sampler was determined based on Concentration -4.8 2.8 NS these concentrations.Rh 1.8 2.4 NS Table 1 Sampling rates of Willems badge (n=6) at diVerent sampling times, relative humidities and NO2 concentrations in laboratory tests with chemiluminescence as reference method.RSD, relative standard deviation Mean Relative reference Sampling humidity concentration/ RSD Sampling time/days (%) mg m-3 (%) rate/ml min-1 7 80 190 13 27 1 20 182 9 45 1 80 180 12 54 7 20 173 8 33 2 50 147 12 40 7 20 5 8 38 7 80 5 12 43 Mean 11 40 Fig. 3 Relationship between concentrations of NO2 obtained by RSD (%) 22 Willems badge and chemiluminescence instrument in field studies.J. Environ. Monit., 1999, 1, 349–352 351Table 3 Results from field studies. Ratio between Willems badge measurements (n=6; sampling rate, 40 ml min-1) and chemiluminescence. Rh, relative humidity; RSD, relative standard deviation Mean reference Mean Mean Ratio between Sampling concentration/ RSD Mean wind Rh temperature/ Willems badge and Experiment time/days mg m-3 (%) velocity/m s-1 (%) °C reference method 1 6 15 3 1.5 78 -11.2 1.05 2 2 8 6 1.8 73 -2 1.11 3 2 16 6 1.4 82 -14 1.03 4 2 20 11 1.2 80 -17.6 1.06 5 6 8 3 1.3 64 -5.6 1.04 6 2 7 10 1.5 48 -5.2 1.32 7 2 3 12 1.7 69 -5.3 1.11 8 2 12 6 0.5 75 -6.4 0.93 Mean 7 1.08 Table 4 Variation of wind speed during field experiments 5–8, Acknowledgements expressed as percentage of total time For financial support and valuable discussions we would like Percentage of time (%) to thank Dr Bertil Forsberg, Department of Environmental Health, Umea° University, Sweden.Wind speed/m s-1 Run 5 Run 6 Run 7 Run 8 <0.3 13 1 1 38 References 0.3–2 70 79 68 61 >2 17 20 31 1 1 J. Quackenboss, J. Spengler, M. Kanarek, R. Letz and C. DuVy, Environ.Sci. Technol., 1986, 20, 775. 2 R. Helleday, PhD Thesis and references therein, Umea° University, included, the overall uncertainty (OU) of the method accord- Sweden, 1995, ISBN 91-7191-017-4. 3 M. Ferm and P-A. Svanberg, Atmos. Environ., 1998, 8, 1377. ing to existing European standards was determined to be 24% 4 E. D. Palmes, A. F. Gunnison, J. DiMattio and C. Tomczyk, Am. based on the eight series of field measurements, which is below Ind.Hyg. Assoc. J., 1976, 37, 570. the required 30%.13,15 5 Y. Yanagisawa and H. Nishimura, Environ. Intern., 1982, 8, 235. 6 J. J. H. Willems and P. Hofschreuder, in A Passive Monitor for Measuring Ammonia, ed. I. Allegrini, A. Febo and C. Perrino, Air Conclusions Pollution Research Report Nr 37, Commission of the European Communities, Brussels, 1991, pp. 113–121. The diVusive sampler validated in this study was designed for 7 A. Hagenbjo� rk-Gustafsson, B. Forsberg, G. Hestvik, D. Karlsson, short term (24 h) sampling in ambient air. It has been validated S. Wahlberg and T. Sandstro�m, Analyst, 1996, 121, 1261. in laboratory studies as well as in field tests for measurements 8 H. van Reeuwijk, P. H. Fischer, H.Harssema, D. J. Briggs, of NO2. The experimental sampling rate was determined to K. Smallbone and E. Lebret, Environ. Monit. Assess., 1992, 50, 37. be 40.0 ml min-1 with a relative standard deviation of 22%. 9 B. E. Saltzman, Anal. Chem., 1954, 26, 1949. There were no eVects of the relative humidity or concentration 10 Analytical Methods Committee, Analyst, 1987, 112, 119. 11 R. Lindahl, PhD Thesis and references therein, Umea° University, of NO2, but a small eVect of the sampling time on the sampling Sweden, 1997, ISBN 91-7191-355-6. rate was noted, which however does not influence the useful- 12 CEN, European Committee for Standardisation, Workplace ness of the method. Atmospheres—Requirements and Test Methods for DiVusive The sampling rate determined in the laboratory studies was Samplers for the Determination of Gases and Vapours, EN 838, confirmed in field studies, and the mean diVerence in the CEN, Brussels, 1995.concentrations obtained with Willems badge and the chemi- 13 CEN, European Committee for Standardisation, Ambient Air Quality—DiVusive Samplers for the Determination of luminescence reference method was less than 10%. The sampler Concentration of Gases and Vapours—Requirements and Test performed well in wind speeds higher than 2 m s-1 as well as Methods. Part 1 and Part 2, Draft EN, CEN/TC264/WG11, at low temperatures. The overall uncertainty of the method Brussels, 1998. was 24%. 14 R. Carlsson, Design and Optimization in Organic Synthesis (Data The sampler is useful for monitoring background levels of Handling in Science and Technology, Vol. 8), Elsevier, ambient nitrogen dioxide as the sensitivity for a 48 h sampling Amsterdam, 1992. 15 CEN, European Committee for Standardisation, Workplace period is 2 mg m-3. The capacity is high enough to allow Atmospheres—General Requirements for the Performance of sampling of 150 mg m-3 for 7 days, which is twice the rec- Procedures for Measurements, EN 482, CEN, Brussels, 1994. ommended Swedish short-term (24 h) guideline value as a 16 P. Hofschreuder, W. van der Meulen, P. Heeres and S. Slanina, 98-percentile over 6 months. J. Environ Monit., 1999, 1, 143. In summary, the sampler is suitable for monitoring nitrogen dioxide in urban air as well as in remote areas with sampling Paper 9/02937K times from 1 to 7 days. The sensitivity and capacity of the method also make it suitable for personal sampling for 2–8 h in working environments. 352 J. Environ. Monit., 1999, 1, 349&ndash