PERFORMANCES AND APPLICATION OF A PASSIVE SAMPLING METHOD FOR THE SIMULTANEOUS DETERMINATION OF NITROGEN DIOXIDE AND SULFUR DIOXIDE IN AMBIENT AIR

Transcription

1 PERFORMANCES AND APPLICATION OF A PASSIVE SAMPLING METHOD FOR THE SIMULTANEOUS DETERMINATION OF NITROGEN DIOXIDE AND SULFUR DIOXIDE IN AMBIENT AIR H. PLAISANCE 1,I.SAGNIER 2,J.Y.SAISON 2,J.C.GALLOO 1 and R. GUILLERMO 1 1 Laboratoire Central de Surveillance de la Qualité de l Air, Ecole des Mines de Douai, Département Chimie et Environnement, 941 rue Charles Bourseul, Douai, France; 2 A.R.E.M.A Lille Métropole, 5 boulevard de la Liberté, Lille, France ( author for correspondence, plaisance@ensm-douai.fr.) (Received 22 May, 2001; accepted 30 November, 2001) Abstract. The performances and applicability of a diffusion tube sampler for the simultaneous measurements of NO 2 and SO 2 in ambient air were evaluated. SO 2 and NO 2 are collected by the passive sampler using triethanolamine as trapping agent and are determined as sulphate and nitrite with ion chromatography. The detection limit (2.3 µg m 3 of NO 2 and 4.2 µg m 3 of SO 2 for two weeks sampling) is adequate for the determination of concentrations in urban and industrial areas. Precision of the method as RSD is in mean 5% for NO 2 and 12% for SO 2 at the concentration levels in urban areas. Calibration of the method was performed in the field conditions by comparison between the responses of sampler and the concentrations measured by the continuous monitors. High degree of linearity (correlation coefficients > 0.8) is found between the passive sampler tube and the continuous monitor data for both NO 2 and SO 2. To reduce the wind velocity influence on passive sampling of diffusion tubes, a protective shelter was tested in this study. The overall uncertainty of one measure for the optimised method is estimated at 5 µg m 3 for NO 2 and 6 µg m 3 for SO 2. Suitability of this passive sampling method for air pollution monitoring in urban areas was demonstrated by the results shown in this paper on a campaign carried out in the French agglomeration. 1. Introduction Nitrogen dioxide and sulfur dioxide are the most monitored pollutants in ambient air because of their effects on the human respiratory system, their contributions to the acidification of the ecosystems and their roles in the formation of photochemical oxidants. In France, air quality is dealt with a law which has required the setting-up of air monitoring networks in the cities over 100,000 inhabitants on January 1st, 1998 and planned to spread the survey to the whole national territory on January 1st, 2000 (Loi no , 1997). In order to help to the screening of sites for the new automated monitoring stations and also to widen the monitoring survey in areas meriting further attention, it is necessary to use the simple instruments allowing to equip many sites. These requirements can be overcome by the use of a passive sampling methods. This mode of sampling is based on the principle of passive diffusion of a pollutant through an air layer to an absorbing medium. The Environmental Monitoring and Assessment 79: , Kluwer Academic Publishers. Printed in the Netherlands.

2 302 H. PLAISANCE ET AL. mean concentration of pollutant is calculated from the collected amount integrated over the sampling time. Neither pumping of air nor electricity is needed in passive sampling. The classical passive sampler is the tube-type sampler first introduced by Palmes and Gunnison (1973). Although simple to use, the Palmes tube requires a long process of optimization and validation. Indeed, the passive samplers must collect efficiently and selectively air pollutants while being resistant to meteorological factors, such as temperature, humidity, sunlight, precipitation and wind. The Palmes tube takes advantage of the small area-to-length ratio to minimise the influence from turbulence in the front of the tube. Nevertheless, there are conflicting results from a number of some investigations on possible sampling problems caused by wind conditions. Although several workers have provided evidence that the error is negligible to the quantification of NO 2 (Palmes et al., 1976; Atkins and Quirino, 1992), Campbell et al. (1994) have revealed that the Palmes tubes overestimate relative to chemiluminescence monitors by a factor of about 1.3 which is variable from site to site. This was attributed to a shortening of the effective diffusion length along the tube caused by wind-induced turbulence at the face of the sampler. A similar conclusion was reached by Hargreaves (1989) from laboratory wind-tunnel experiments and by Gair and Penkett (1995) who found in their field experiments a reduction in the diffusion length between 7% and 38% which was dependent on the increase of the wind velocity at the site. To overcome this problem of wind sensibility, the passive samplers can be fitted with a porous membrane barrier (De Santis et al., 1997). In practice, the membrane presents a resistance to diffusion which must be measured experimentally. Therefore, the sampling device depends entirely on a calibration against mean ambient concentrations obtained with a standard active method. This resistance can also vary under the low-wind speed effects and fluctuations in meteorological conditions such as temperature and humidity at the sampling site. Another option is to use a protective shelter for the passive diffusion tubes, as recommended by Monn and Hangartner (1989). Very little information is available concerning the effects of shelter on the measurements by passive samplers. This paper describes a passive sampling method for the simultaneous measurements of NO 2 and SO 2 in ambient air. In order to improve the quality of these measurements, some investigations, like the optimization of absorbing solution for SO 2 and NO 2 sampling, and the test of protective shelter for the samplers, are led through many fieldwork studies. Accuracy, precision and limits of these methods are evaluated and discussed. Finally, we report on the maps of spatial distributions of NO 2 and SO 2 in the agglomeration of northern France, obtained by measurements made with this passive sampling technique.

3 PERFORMANCES AND APPLICATION OF A PASSIVE SAMPLING METHOD The Diffusion Tube and Its Operating Principles During the sampling time, a concentration gradient set up for each of the pollutants along the length of the tube from the absorber to the ambient air at the open end of the sampler. The pollutant diffuses up the tube where it is trapped and accumulated in the form of absorbed products which are subsequently analysed in laboratory. NO 2 and SO 2 are simultaneously sampled in a same Palmes tube during a sampling period of several days. As described previously (Shooter et al., 1997), this diffusion tube sampler consists of an acrylic tube, 7.1 cm long and 1.1 cm in internal diameter. At one end, three stainless steel meshes coated with an aqueous triethanolamine (TEA) solution, an absorbing medium, are held in place with a polythene coloured cap; the other end of the sampler is sealed with the colourless cap which is removed during the period of sampling. TEA absorbs NO 2 from the air in the form of nitrite ions (Palmes et al., 1976). For sulfur dioxide, both sulfites and sulfates are formed in this absorbing solution at the stainless steel meshes (Krochmal and Kalina, 1997). After extraction, these absorbed products are analysed in ion chromatography. According to Fick s first law, the quantity of absorption products in the samplers is proportional to the concentration in the air outside the tube, the sampling time, the diffusion coefficient and the dimensions of the sampler. Concentrations are determined from the quantity of absorption products measured in the sampler with the following formula: C = Q d A t D (1) where C concentration measured by passive sampling tube (µg m 3 ) Q quantity of absorption products present in the sampler (µg) d diffusion length (m) A cross-sectional area (m 2 ) t sampling time (s) D diffusion coefficient (m 2 s 1 ) For more information on the theoretical background of diffusion tube sampling, see Palmes et al. (1976) and Gair et al. (1991). The diffusion coefficient of m 2 s 1 for NO 2 and m 2 s 1 for SO 2 is used in concentration calculation. The efficiency of TEA to remove these two pollutants is assumed to be 100%, according to the results of Gair et al. (1991) and Scheeren et al. (1994).

4 304 H. PLAISANCE ET AL. 3. Experimental Procedure 3.1. SAMPLER PREPARATION To minimize background contamination, all the components of diffusion tubes are soaked before using in demineralized water batch maintained under roughness during 6 h. The water of batch is regularly changed. Then, each component is passed under a nozzle of demineralized water and dried with compressed air. The diffusion samplers are assembled at the same time as the cleaning of components. The three stainless steel meshes are impregnated with a 10% v/v aqueous solution of triethanolamine (TEA). 300 µl of Brij-35 (a wetting agent) is added to 100 ml of the aqueous TEA solution making it viscous, and therefore adding the coating of the meshes. 30 µl of absorbing solution is directly injected onto the meshes. Then, the tube is immediately assembled and stored in a fridge at 4 C SAMPLER EXPOSURE Samplers are mounted vertically, with the cap containing the coated meshes uppermost. Sampling is started by the removal of the colourless cap which is retained to be replaced at the end of the sampling period. The diffusion tubes are exposed on the sites allowing unrestricted movement of air around the samplers. Two sampling modes are tested in ambient conditions on several urban sites: In the first mode of exposure the most currently used, the diffusion tube is fixed by a clip to a stake without protection at least 1.5 metres above the ground. This clip holds the sampler to distance from the surface of stake. In the second mode, the diffusion tube is placed in a protective shelter to minimise the wind turbulence effects on passive sampling. This protective device in form of cylinder has two inlets/outlets which warrants a good circulation of air without turbulence around the tubes. It is likewise fixed by a clip to a stake. To guarantee a sufficient quantity of pollutants collected by the tube, the exposure time is fixed to fourteen days. In each field experiment, samplers are simultaneously exposed with unopened samplers being used as blanks. The exposed samplers and blanks are stored in a fridge at 4 C prior to analysis to minimise background contamination EXTRACTION AND ANALYSIS OF PASSIVE SAMPLER Before extraction, the sampler is disassembled, the body of the tube is cleaned with deionised water and dried with compressed air. This cleaning is necessary to avoid the contamination by the deposition of atmospheric particles rich in sulfur on the inner surface of the tube during the exposure period. For extraction, a volume of 2,5 ml of deionised water is added directly in the tube. The extract is completed with 10 µl of 35% (v/v) H 2 O 2 to oxidise completely

5 PERFORMANCES AND APPLICATION OF A PASSIVE SAMPLING METHOD 305 Figure 1. Chromatograms of extracts from a diffusion tube exposed at 50 µgm 3 of NO 2 and 29 µg m 3 of SO 2 during two weeks (a) and an unexposed tube (b). the products of absorption of SO 2 to sulphates. The alkaline ph of this extract (about 10) due to the presence of triethanolamine prevents oxidation of nitrites to nitrates (Krochmal and Kalina, 1997). Then, the extract contained in the tube is gently shaken during about 2 minutes. SO 2 and NO 2 are respectively determined as sulphate and nitrite with analysis of extract by ion chromatography in a same run. A Dionex DX300 ion chromatograph

6 306 H. PLAISANCE ET AL. TABLE I Results of tests on the precision of passive sampling method for the NO 2 and SO 2 measurements Number of Number of Range of concentrations RSD (%) batches exposed tubes (µgm 3 ) NO ( ) SO ( ) Minimum and maximum of RSD values are given in parentheses. with an Ion Pac AS4A column, suppressor ASRS-I and conductivity detector is used for analysis. Analytical conditions for this system is as follows: eluent 1.8 mmol L 1 Na 2 CO 3 / 1.7 mmol L 1 NaHCO 3 ; 2 ml min 1 flow rate and 200µL sample size. Chromatographs of an exposed tube and a blank are presented in Figure 1. A low concentration of sulphates is found in unexposed tube, whereas the nitrite peak is not detected. The possible source of this weak contamination is the trace of SO 2 and particulate sulfate remained along the tube walls after the cleaning (Kasper-Giebl and Puxbaum, 1999). The masses of NO 2 and SO2 4 collected by the tube (m NO 2 and m SO 2 )are 4 systematically corrected with the blank values. The corresponding quantities of NO 2 and SO 2 are calculated as follows: Q NO2 =m NO 2 Q SO2 = m SO 2 4 (2) (3) Concentrations of NO 2 and SO 2 in µg m 3 are determined by applying the formula (1) previously presented. 4. Results and Discussion 4.1. PERFORMANCES OF PASSIVE SAMPLER The detection limits for a two-week sampling are estimated at 2.3 µg m 3 for NO 2 and 4.2 µg m 3 for SO 2 calculated as the threefold standard deviation of the mean area of the blank determined by twenty measurements of blank peaks. These detection limits seem adequate for the determination of NO 2 and SO 2 in urban and industrial areas.

7 PERFORMANCES AND APPLICATION OF A PASSIVE SAMPLING METHOD 307 TABLE II Results of regression lines for the field comparisons between the diffusion tube and monitor measurements Pollutant Sampling mode Number of Equation of the regression line Correlation comparisons coefficient NO 2 Without the protective device 52 Tube = 0.97 [±0.13] Monitor [± 4.05] 0.91 With the protective device 52 Tube = 0.98 [±0.08] Monitor 1.59[± 2.63] 0.96 SO 2 Without the protective device 86 Tube = 0.98[± 0.13] Monitor 1.11[± 1.43] 0.86 With the protective device 86 Tube = 0.86[± 0.10] Monitor 2.20[± 1.14] 0.88 Confidence intervals at 95% are given in parentheses for the slopes and intercepts of regression lines. In order to evaluate the precision, the replicate measurements were carried out on a site at the Mines School of Douai. Eight batches of diffusion tubes are exposed to outside air on the same site. Each batch consists of six exposed tubes and two unopened tubes (blanks). Precision of the diffusion tube technique expressed as RSD values of replicates is reported in Table I. A good repeatability is obtained for the measurements of NO 2 (RSD 5%) and medium for SO 2 (RSD 12%). Accuracy of this passive method is investigated by the comparison between diffusion tubes and the continuously measuring instruments, the chemiluminescent nitrogen oxide analyser and the fluorescent sulfur dioxide analyser, considered as being the reference methods for the measurements of NO 2 and SO 2 in ambient air. Measurements were performed in urban areas at four monitoring stations for ten months from March 1998 to January The two sampling modes (with and without protective device) are tested. A least squares regression analysis is performed on passive sampler readings versus the integrated values of analysers. The regression equations of comparisons between diffusive samplers and continuous monitors are given in Table II. The corresponding regression lines are presented in Figures 2a d. High degrees of linearity (close to 0.9) are found between the passive sampler tube and the continuous monitor data for both NO 2 and SO 2. The passive sampler readings closely approximate the integrated values of monitors, except in the case of SO 2 measurements with the protective device where a discrepancy is found. Comparing the results of linear regressions obtained with the two sampling modes, it appears that the correlation coefficient is increased for the diffusion tubes placed in a protective device. This shelter seems to limit the wind effects on the diffusion

8 308 H. PLAISANCE ET AL. Figure 2. Comparisons between the diffusion tube with and without protective device and the continuous monitor for the measurements of NO 2 (a,b) and SO 2 (c,d). path of the tube. The slope value inferior to 1 obtained on the regression line of SO 2 measurements with the protective device suggests that the diffusion tubes systematically underestimate the SO 2 concentration. The reason can be the capture of SO 2 by the surface of shelter in PVC. It seems necessary to apply the linear relationship previously established in order to correct the SO 2 measurements of passive tubes with the protective device. The overall uncertainty of one measure of diffusion tube is estimated from the prediction interval at 95% (Saporta, 1990), established for each regression line between the responses of diffusion tubes and those of the monitor. This statistical interval traduces the scattering of point values around the calibration line (see

9 PERFORMANCES AND APPLICATION OF A PASSIVE SAMPLING METHOD 309 TABLE III Estimation of the overall uncertainties for the SO 2 and NO 2 measurements performed by diffusion tube with and without the protective device Pollutant Mode of exposure Overall uncertainty NO 2 without the protective device [NO 2 ] ± 6 µgm 3 with the protective device [NO 2 ] ± 5 µgm 3 SO 2 without the protective device [SO 2 ] ± 8 µg/m 3 with the protective device [SO 2 ] ± 6 µgm 3 Figures 2a d). The monitor value is taken as reference. The measure error of monitor is considered as low with regard to the one of passive sampler. The overall uncertainties for the SO 2 and NO 2 measurements performed by a diffusion tube with and without protective device are presented in Table III. These uncertainties appreciably decrease with the use of protective shelter. Nevertheless, this passive method seems to give non negligible uncertainties that are important to take into consideration in the data interpretation. Accordingly, this passive method is adapted to measure in ambient air of urban and industrial areas, with an overall uncertainty below 30% for the concentration levels of NO 2 and SO 2 above 20 µg m APPLICATION OF DIFFUSION TUBES TO MAP THE LEVELS OF NO 2 AND SO 2 CONCENTRATIONS IN A CITY To demonstrate the use of this passive method for the measurements in urban areas, a winter campaign was organised in collaboration with the local air monitoring network (AREMA-LM) on one two-week sampling period from 03/03 to 17/03/1999. The agglomeration covered by sampling consists of three main cities (Lille, Roubaix and Tourcoing). 145 simultaneous measurements of NO 2 and SO 2 distributed over an area of about 300 km 2 were obtained, respecting the following sampling plan: one tube per km 2 in the urban centres and one tube per four km 2 in the suburban and rural areas. Uniform criteria for site selection was accepted, all the measurement points were situated at least 50 m from significant sources of air pollution, such as busy traffic roads. The scope is to have access to the background concentration levels that are the most representative of long term human exposure. Isoconcentration curves were calculated by interpolation from the measurements obtained by diffusive sampling. The ordinary kriging was used as interpolation method. The calculations and the graphical representation of the isopleths were performed on a PC by means of Surfer 6.04 software completed by Variowin

10 310 H. PLAISANCE ET AL. Figure 3. Experimental variograms for the concentrations of NO 2 (a) and SO 2 (b) and the fitted models. Number of pairs of measures is indicated upon each point.

11 PERFORMANCES AND APPLICATION OF A PASSIVE SAMPLING METHOD 311 Figure 4a. Figure 4a b. Spatial distributions of concentrations of NO 2 (a) and SO 2 (b) in the agglomeration.

12 312 H. PLAISANCE ET AL. Figure 5. Compass card rose of a meteorological station of the agglomeration for the sampling period from 03/03 to 17/03/ software (Kanevsky et al., 1995) for the analysis of variogram. The theory of ordinary kriging was described by Armstrong (1998). The kriging grid chosen consisted of cells from 100m 100m. Analysis with Variowin software has allowed to select two omnidirectional models which best fit the experimental variograms of NO 2 and of SO 2. Figures 3a b show the experimental variograms calculated for NO 2 and SO 2 data and their fitted models. The distribution maps of NO 2 and SO 2 concentrations for the sampling period are given in Figures 4a b. The distribution of concentrations is represented by isopleths with increments of 2 µgm 3 for NO 2 and 4 µg m 3 for SO 2. The locations of the sampling sites are indicated by the crosses and the main industrial emissions of pollutant by the triangles on these two maps. Note that the frequent anticyclonic conditions with the fluxes from northeastern sector during this campaign (see the compass card rose in Figure 5) were favourable to the accumulation of atmospheric pollutants. Analysis of the distribution maps supplies the following information: Map of NO 2 For NO 2, the background pollution from 40 to 46 µgm 3 in the urban agglomeration is observed. To this background pollution is added a pollution from automotive dense traffic which appears in the centres of Lille and Roubaix [A and B] and in the main intersections of the agglomeration [C, D, E and F] where the levels were

13 PERFORMANCES AND APPLICATION OF A PASSIVE SAMPLING METHOD 313 rising up to 46 µg m 3. The effects of industrial NOx emissions only appear in the residential areas [C and G] outside the urban agglomeration with the levels reaching values up to 44 µg m 3. The average NO 2 concentration calculated for all sites is 42.8 µg m 3.This value is close to 43 µg m 3 which is the average of NO 2 concentrations measured at the same period by the analysers equipping the urban background stations of the AREMA-LM air monitoring network. This displays that data of AREMA-LM stations supply a good indication on the background pollution of NO 2 in this agglomeration. This level of NO 2 concentrations remain medium in comparison with those found in the winter campaigns of Brussels (De Saeger et al., 1995), Paris (Atkins and Quirino, 1992), Madrid (De Saeger et al., 1991) and Toulouse (Della Massa et al., 1994) where the average NO 2 concentrations for the background sites were respectively to 53.9, 49.3 and 48.3 and 37.1 µg m 3. By superimposing this NO 2 map, upon the density of population (census 1990) in the agglomeration, we have estimated that during this period, 73% of townmen are in areas where the levels in outside air exceeded 40 µg m 3 and only 10% in areas with the outside concentrations above 50 µgm 3. Map of SO 2 The concentration levels for SO 2 are weak. The background pollution in the urban agglomeration amounts to about 14 to 22 µgm 3. At this winter period, these weak levels are mainly due to the emissions from domestic heating. Many hot spots with levels above 22 µgm 3 are recorded in the centre of Lille [A], area where the urban frame is the densest, and in three industrial areas of the agglomeration [B, C and D]. The industrial emissions appear as the main contribution to the highest concentrations recorded in this agglomeration. The mean concentration obtained from the AREMA-LM SO 2 analysers present in the studied area is 16.2 µg m 3, that is also close to the mean value for all the tubes (15.5 µg m 3 ). Nevertheless, the levels of SO 2 concentrations remain too weak to constitute a real risk for the human health. 5. Conclusions This paper describes the performances of a passive sampling method for the simultaneous measurements of NO 2 and SO 2 with a same diffusion tube of the Palmes type. This method uses triethanolamine as trapping agent and ion chromatography as analytical technique. For an exposition time of two weeks, the detection limit is estimated at 2.3 µg m 3 for NO 2 and 4.2 µg m 3 for SO 2 and the precision expressed as RSD is on average 5% for NO 2 and 12% for SO 2 at concentration levels found in urban areas. In the field conditions, a good agreement is found in the intercomparisons between the responses of passive sampler and those of reference active methods

14 314 H. PLAISANCE ET AL. with correlation coefficients close to 0.9. These tests also show that the use of a protective shelter reduces the wind velocity influence on passive sampling of diffusion tube and thereby allows to limit the overall uncertainty of one measure estimated at 5 µgm 3 for NO 2 and 6 µg m 3 for SO 2. Then, this passive sampling method was applied to determine the spatial distributions of NO 2 and SO 2 over the French agglomeration. The maps of air pollution levels made by an geostatistic method, called kriging, clearly highlight the sources responsible to the high concentrations recorded in this agglomeration. The highest concentrations of NO 2 exceeding 50 µgm 3 are essentially due to a pollution from automotive traffic, whereas the industrial emissions appear as the main contribution to the highest concentrations of SO 2 above 20 µg m 3 recorded in the agglomeration. The frequent anticyclonic conditions with the fluxes from northeastern sector during this campaign were favourable to the accumulation of atmospheric pollutants. The good agreement found between the averages of tube data and those of local network analysers reveals the implantation of stations is adequate to provide a right information on the background levels of two pollutants in the agglomeration. This example shows that this passive sampling method is well suited for the simultaneous measurements of NO 2 and SO 2 in a region of complex emissions like an urban area. Acknowledgements The authors would like to thank the staff of AREMA-LM and Ecole des Mines de Douai which have taken part in this study. We are greateful to the French Ministry of Environment and the environmental agency (ADEME) for financial support towards this program. References Armstrong, M.: 1998, Basic Linear Geostatistics, Springer-Verlag, Berlin, Heidelberg. Atkins, D. H. F. and Quirino, I.: 1992, A Survey of Nitrogen Dioxide in Paris, Report of Commission of the European Communities EUR EN, Bruxelles. Campbell, G. W., Stedman, J. R. and Stevenson, K.: 1994, A survey of nitrogen dioxide concentrations in the United Kingdom using diffusion tubes, Atmos. Environ. 28, Della Massa, J. P., Gerbolès, M. and Payrissat, M.: 1994, Etude de la distribution du dioxyde d Azote dans l agglomeration Toulousaine par la méthode des échantillonneurs passifs, Report of Commission of the European Communities EUR FR, Bruxelles. De Saeger, E., Gerbolès, M. and Payrissat, M.: 1991, La surveillance du dioxyde d Azote á Madrid au moyen d échantillonneurs passifs, Report of Commission of the European Communities EUR FR, Bruxelles. De Saeger, E., Gerbolès, M., Pérez Ballesta, P., Amantini, L., Payrissat, M.: 1995, Air Quality Measurements in Brussels ( ) NO 2 and BTX Monitoring Campaigns by Diffusive Samplers, Report of Commission of the European Communities EUR EN, Bruxelles.

15 PERFORMANCES AND APPLICATION OF A PASSIVE SAMPLING METHOD 315 De Santis, F., Allegrini, I., Fazio, M. C., Pasella, D. and Piredda, R.: 1997, Development of a passive sampling technique for the determination of nitrogen dioxide and sulphur dioxide in ambient air, Anal. Chim. Acta 346, Gair, A. J., Penkett, S. A. and Oyola, P.: 1991, Development of a simple passive technique for the determination of nitrogen dioxide in remote continental locations, Atmos. Environ. 25A, Gair, A. J. and Penkett, S. A.: 1995, The effects of wind speed and turbulence on the performance of diffusion tube samplers, Atmos. Environ. 29, Hargreaves, K. J.: 1989, The Development and Application of Diffusion Tubes for Air Pollution Measurement, Ph.D. thesis, University of Nottingham. Kanevsky M., Chernov, S. and Demyanov, V.: 1995, Upfile User s Guide, COMETT 95, Lausanne University of Lausanne. Kasper-Giebl, A. and Puxbaum, H.: 1999, Deposition of particulate matter in diffusion tube samplers for the determination of NO 2 and SO 2, Atmos. Environ. 33, Krochmal, D. and Kalina, A.: 1997, A method of nitrogen dioxide and sulphur dioxide determination in ambient air by use of passive samplers and ion chromatography, Atmos. Environ. 31, Loi no sur l air et l utilisation rationnelle de l énergie: 1997, Journal Officiel de la République Française du 1er Janvier 1997, Paris: Monn, C. and Hangartner, M.: 1989, Ozone Passive Sampling Tubes for Ambient Monitoring, 82th Annual APCA Meeting, Anaheim CA, Paper Palmes, E. D. and Gunnison, A. F.: 1973, Personal monitoring device for gaseous contaminants, Amer. Indust. Hyg. Assoc. J. 34, 78. Palmes, E. D., Gunnison, A. F., Dimattio, J. and Tomezyk, C.: 1976, Personal sampler for nitrogen dioxide, Amer. Indust. Hyg. Assoc. J. 37, Saporta, G.: 1990, Probabilités analyse des données et statistique, Editions Technip, Paris. Scheeren, B. A., Santis, F. de, Allegrini, I. et Heeres, P.: 1994, Monitoring SO 2 with passive samplers: a laboratory evaluation of Na 2 CO 3 and triethanolamine as absorbing media, Int. J. Environ. Anal. Chem. 56, Shooter, D., Brimblecombe, P., Shooter, J., Lowe, D., Day, P. J. and Du, S.: 1997, Some characteristics and applications of nitrogen dioxide passive samplers, Environ. Technol. 18,