The influence of geometry and draught shields on the performance of passive samplers Peter Hofschreuder,* Wobbe van der Meulen, Paul Heeres and Sjaak Slanina Meteorology and Air Quality Group, Agricultural University Wageningen, Duivendaal 2, 6701 AP Wageningen, The Netherlands Received 26th November 1998, Accepted 23rd February 1999 Passive samplers provide an excellent opportunity to perform indicative measurements or establish a dense network of measuring sites. A drawback compared with conventional active measuring methods is the larger spread of results.This variation can, to a large extent, be attributed to the influence of temperature, sampler geometry and wind on sampling results. A proper design of sampler geometry and optimum choice of draught shield can reduce the influence of wind velocity on a badge type sampler to less than 10%.Wire mesh screens prove to be inadequate in damping turbulence. Filters give good results. Attention should be paid to the size and isolation value of the walls of the sampler to prevent thermal updrafts occurring within the sampler. Tube type samplers are less influenced by wind, provided that turbulence is prevented from influencing diVusion within the sampler.where m=mass (mg m-3); f=correction factor for temperature Introduction and pressure; A=surface area of sampler (m2); D=diVusion Passive samplers are attracting increasing attention as a tool coeYcient (m2 s-1); r=concentration of pollutant (mg m-3) in obtaining outdoor air quality data in remote locations far (1=at entrance of the sampler and 2=at the adsorption from electrical connections, and for intensive studies such as surface); t=exposure time (s); and l=diVusion length within epidemiological studies and model validation.The use of sampler (m). passive samplers is strongly supported by the European Union. The correction factor f for temperature and pressure is In Council Directive 96/62/EC of 27 September 1996 on Air proportional to T n, where T=absolute temperature (K), and Quality Assessment and Management, a framework is set for 0.5<n<1.0.6 The correction to STP in the range 263–308 K preliminary assessments of air quality, optimisation of station amounts up to a maximum of 10%.The uncertainty in this siting, supporting generalisation measurements and evalua- correction has a maximum of 6%.tion of existing measurements. In the Guidance Report on The resistance against uptake of the pollutant is defined by Supplementary Assessment under EC Air Quality Directives (1997), it is concluded that the low cost and easy operation Rd= l D of diVusive sampling techniques make them an ideal tool for large scale air pollution surveys with a high spatial resolu- where Rd is resistance (s m-1).This Rd only constitutes the tion. The diVusive sampler is also of particular interest as an resistance within the sampler. Introduction of a draught screen indicative technique. The maximum uncertainty is estimated gives an additional resistance against uptake of pollutant (Rf). in an EC report1 to be about 30%. This uncertainty is mainly At low wind speeds, the possibility exists that there will be a caused by the influence of meteorology on the sampler.This development of a stagnant layer of air in front of the draught paper deals with possibilities of reducing the uncertainty by screen. This layer exhibits another resistance against uptake careful design and reduction of the influence of meteorology (Re).For a badge type sampler (and a tube type sampler with a draught screen) we end up with a total resistance (Rt) against on the sampler. uptake of pollutants of Tube type passive samplers such as the Palmes tube are constructed with a relatively long diVusion distance from the Rt= 1 Rd+Rf+Re entrance of the tube to the sorbent compared with the diameter of the tube. This geometry is chosen to avoid too much The last uncertain factor in the calculation of the sampled influence of external turbulence on the eVective diVusion mass is the equilibrium concentration of the pollutant just length.2 A drawback of this geometry is the low sampling rate above the absorbing material.This may be calculated from of the tube type sampler. To increase the sampling rate of a chemical data.When this concentration plays a role (0), we passive sampler, badge type samplers were developed3 and in obtain a decrease in the concentration gradient during sam- other cases the diVusion length of the tube type sampler was pling and should apply an integrated collection function. In shortened.4 Shortening of the diVusion length increases the practice, using absorbing media with a low equilibrium gas influence of turbulence and wind speed on the sampler and concentration and application of excess absorbing material to makes the entrance geometry more critical.To mitigate probmaintain a low concentration of pollutant in the absorbing lems of this kind, draught shields were introduced.5 material circumvent this situation. The mass of a pollutant that is collected by a passive sampler in a steady state situation during a certain time is derived Aim of the study from Fick’s first law of diVusion and given by the equation14 It can be concluded that the largest source of uncertainty in passive sampling is the uncertainty in the value of Rt. Lowering m= fAD(r1-r2)t l the uncertainty in Rt was the aim of this study.J. Environ. Monit., 1999, 1, 143–147 143When we define the sampling rate to be Theory on the resistance against diVusion The concept of a total resistance (Rt) against uptake of s= A Rt pollutants by diVusion by a passive sampler with a draught shield was presented in the Introduction. To be able to interpret where s=sampling rate (m-3 s-1), it will be clear that we will results, this concept will be extended.obtain a decreasing sampling rate with decreasing wind velocity The resistance of the draught shield can be approximated and that this decrease is more pronounced for badge type by the equation samplers. Rf= e DP Testing passive samplers General requirements for testing passive samplers are described where Rf=resistance of draught shield (s m-1); e=thickness in the draft CEN Standard on Passive Samplers Part 19 and of draught shield (m); and P=porosity.will not be treated here. In the draft CEN Standard on Passive For a filter, the thickness and porosity should be derived Samplers Part 3,10 attention is paid to the influence of environ- from data from the manufacturer. Typical values for Rf for a mental factors on sampler performance and the protection glass fibre filter and a membrane filter are 15–20 s m-1 (e= from adverse environmental conditions. 0.35 mm, P=0.9) and 1–4 s m-1 (e=0.025–0.075 mm, P= In general, samplers are protected against rain by simple 0.85), respectively. When a wire mesh screen is used as a metal profiles, plastic funnels and boxes that are more or less draught shield, the porosity can be calculated by closed but well ventilated by having no bottom or having drilled holes.9 All protective devices should be evaluated with P= L2 (L+d)2 respect to inertia of the material against the gas to be measured and possible interfering emissions from the construction where L=distance between wires (m) and d=diameter of material.wire (m). Shielding against rain almost automatically means protec- Typical values for Rf for commercial wire mesh screens with tion against high wind speeds.Problems may arise at the lower mesh range between 8 and 25 are 35–65 s m-1. end of the wind velocity spectrum. Most draught shields will A value for the external boundary layer resistance (Re) is lower the wind velocity and especially during very stable and diYcult to estimate.This resistance will be very dependent on very unstable atmospheric conditions (at night and on sunny sampler geometry and meteorological conditions. days in summer), this may cause the formation of an external For a flat surface and turbulent conditions, Monteith and boundary layer in front of the sampler entrance and/or even Unsworth7 estimated the average boundary layer thickness a too low refreshment rate in front of the sampler causing independent of wind speed: ‘starvation’ of the sampler.11 Another problem that may arise is the influence of protective devices on the temperature of the d#Z0.8 sampler.Strong insulation during the day or radiative cooling where d=boundary layer thickness (m) and Z=length of during the night may increase respectively decrease the sampler surface (m). Ballesta et al.8 suggested a slight dependence on temperature with respect to the ambient temperature.Hence the wind velocity: the albedo and the isolation of the walls of the protective device should also be considered. The operating temperature d=0.37 An uB0.2 Z0.8 should be known for the conversion of the sampled mass to an average ambient concentration at standard temperature where n=kinematic viscosity of air (1.485×10-5 m2 s-1) and and pressure.6,9 The same considerations as mentioned for u=wind velocity (m s-1).protective devices apply to the passive samplers themselves. When we take Z=0.785z (z=diameter of badge/tube) for Four samplers were used in the experiments. The first the average length of the surface, we can calculate the thickness sampler, which was used to demonstrate the influence of wind of the boundary layer and the corresponding resistance.The speed on the sampling rate, was the original design of the results are presented in Table 1. Willems badge12 (Fig. 1). The total resistance as the sum of diVusional, filter, and The badge is made of polystyrene, being hydrophobic, inert boundary layer resistances for a Willems badge with a glass and having a low permeability for gases.A metal construction fibre draught shield and a modified Palmes tube with a Teflon with high reflectivity could be considered, causing the temperamembrane filter as a draught shield is presented in Table 2. ture of the badge to be as close as possible to ambient A conclusion that can be drawn from these tables is that conditions, but has the disadvantage of a low isolation value.for a tube type sampler the resistance of the draught shield As the badge was primarily used for sampling ammonia, the and the boundary layer resistance are of minor importance drawback of condensation of water vapour on a metal badge and constitute only a few per cent of the total resistance was considered to be more important than a slight diVerence against uptake of the pollutant.For badge type samplers the in temperature. The badges were stuck by means of Velcro to boundary layer resistance can amount up to tens of per cent an angled aluminium plate in an upside down position. The use of a reaction filter (or stainless steel grid) and a spacer of the total resistance.Table 1 Boundary layer thickness (mm) and corresponding resistance according to the equation of Ballesta et al.8 Wind speed/m s-1 Sampler Parameter 0.1 0.2 0.5 1.0 2.0 4.0 10 Willems badge d/mm 3.0 2.6 2.2 1.9 1.7 1.4 1.2 Re/s m-1 131 114 95 83 72 63 52 Palmes tube d/mm 1.5 1.3 1.1 0.96 0.83 0.73 0.60 Re/s m-1 67 58 48 42 37 32 26 144 J.Environ. Monit., 1999, 1, 143–147Table 2 Total resistance (Rt/s m-1) as a function of wind velocity for a Willems badge with a 6 mm diVusion length and a glass fibre filter and a modified Palmes tube with membrane filter as a draught shield Wind speed/m s-1 Sampler 0.1 0.2 0.5 1.0 2.0 4.0 10 Willems badge 411 394 375 363 352 343 332 Palmes tube 1867 1858 1848 1842 1837 1832 1826 sampler allowed eight samples taken from the centre and 12 samples taken from the part of the absorption filter close to the wall of the sampler.For the medium sized sampler one and seven samples were taken, respectively. A fourth sampler was developed, taking the results of the experiments with the Willems badge and the draught shields on the large samplers into account.To avoid the development of an external boundary layer as much as possible, a modifi- cation of the Palmes tube was constructed with a modified draught shield at the entrance (Fig. 3). The materials (polyethylene and FEP-Teflon) are again hydrophobic. No polystyrene is used because of the use of sulfuric acid instead of tartaric acid. Owing to the lower uptake rate, this tube is more appropriate for long measurement periods.13,15 An improvement with respect to the design of the Willems Fig. 1 Diagram of a cylindrical badge (Willems badge). badge is the construction of the draught shield. The Teflon filter is held in place by a tight fitting cap with a centre hole. ring that also keeps the reaction filter in place is obvious. The The distance between the draught shield and the rim of the Teflon membrane is used as a draught shield, and is also tube is no more than the thickness of the cap (0.2 mm).The hydrophobic. It is kept in place by a fixation ring. The sampler external boundary layer is reduced to a minimum in this way. is closed during storage and after exposure by means of a cap. This design can also be used for the badge type samplers.It is important to note that there is a distance between the A drawback of the system is that the transport/closing cap draught shield and the outer rim of the badge of about 7 mm. has to slip over the cap with the centre hole. If this transport The ventilation rate of this external (possible) boundary layer cap fits too tight it will loosen the cap that fixes the entrance is very dependent on meteorological conditions (wind speed filter upon removal.A looser cap is, however, less restrictive and turbulence intensity). It may give rise to an additional in blocking the uptake of ammonia. With this type of cap, the external resistance against uptake of the reacting species. uptake rate of a capped tube was determined to be 0.01% of To study the eVect of draught shields on damping turbulence, the uptake rate of an open tube (with draught shield ).When it was decided to use large samplers in the large wind tunnel the draught shield and cap with centre hole were discarded to have a better view on the influence of turbulence within the and replaced by a cap similar to the cap holding the impregsampler. Two types of large badges were used: a badge of nated stainless steel grids, the uptake rate was only 0.001% of 0.3 m diameter and a variable diVusion length of 0.01–0.23 m that of an open sampler.This is probably mainly due to and a medium sized badge of 0.11 m diameter and a variable diVusion of ammonia through the walls of the sampler. To diVusion length of 0.01–0.1 m. Five types of draught shields avoid passive loading during transport and storage, it was were investigated; a flat PVC plate in front of the large decided to place the closed tubes individually in polyethylene sampler, a flat plate with edges rounded to the inside for the containers with screw caps.medium size sampler (Fig. 2), a coarse wire mesh screen, a fine wire mesh screen and filter-paper. Results and discussion When diVusion of the ammonia within the sampler takes place in still air, the concentration on the adsorption filter The eVect of wind speed on the sampling rate of the original should be uniform.Wall eVects would introduce a systematic Willems badge was studied. Fig. 4 shows the results of theoretidi Verence between the centre and outer edge of the adsorption filter. Turbulence inside the sampler would cause large variations in concentration from place to place on the filter.To be able to study these eVects, circular samples were taken from the adsorption filter in a systematic way. The large Fig. 3 Diagram of a modified Palmes tube. 1=cap for retaining adsorption grids; 2=two stainless steel adsorption grids; 3=FEP Fig. 2 Diagram of large sampler with flat draught shield and medium Teflon tube of 41 mm length; 4=Teflon entrance filter 5 mm pore size; 5=retaining cap with 10 mm diameter centre hole; and 6=transport sized sampler with rounded draught shield in front of the badge opening.guard cap. J. Environ. Monit., 1999, 1, 143–147 145The eYciency of a draught shield Results for the large sampler are given in Table 4 and for the medium sized sampler in Table 5.The results are remarkable when we realise that the standard deviation of the analytical method is only about 0.01 and with calibration samples 0.02–0.04. The tables show that air currents within the samplers lead to no uniform deposition of ammonia on the absorption filter. These currents may result from turbulence entering the sampler or thermal updrafts within the sampler caused by temperature diVerences.The decrease in standard deviation from a situation with a turbulence damping plate and no filter to the situation with a fine wire mesh screen or a glass fibre filter as a draught screen indicates that turbulence may enter the sampler when no proper shielding is used. A second cause of the larger standard deviation compared with that of the analytical method applied is the eVect of the sampler walls on the deposition pattern. Analysis of the results indicates that there is a systematic lower Fig. 4 Theoretical and practical sampling rate for a Willems badge with a diVusion space of 2 mm. absorption next to the walls than in the middle of the sampler. This result can be explained by the non-uniform geometry of the diVusion process next to the wall and probable deposition on the sampler walls.Coating of the walls with acid increased cal calculations and of the experimental determination of the the diVerence in the amount of absorbed ammonia in the sampling rate, s, for the Willems badge (all for STP). middle and next to the walls of the sampler. From Fig. 4 it can be derived that the sampling rate of a The ratio Rt/Rm indicates that air movements within the badge type sampler with a small (2 mm) diVusion space may sampler increase the sampling rate in comparison with the vary by a factor of 2.5 owing to variations in the wind velocity.calculated sampling rate for still air. When maximum pre- The variation in sampling rate will decrease when increasing cautions are applied to prevent turbulence entering the sam- the diVusion length inside the sampler decreases the ratio of pler, the uptake rate is still a factor of 2–3.5 higher than that external resistance to the sum of diVusion and filter resistance.calculated from theory. This factor is less for the large sampler This was the reason for increasing the diVusion length of the having a wall thickness of 5 mm than for the small sampler Willems badge from the original 2 mm to 6 mm.Table 3 gives having a wall thickness of 3 mm. calculated and measured external resistance’s for the original Willems badge with a 2 mm internal diVusion length. Table 4 Ratio of calculated and measured resistance against uptake The external resistance reaches a constant value for wind of ammonia Rt/Rm and standard deviation s of absorption filter velocities larger than 2 m s-1, which is smaller than the samples for a 0.3 m badge type sampler.This ratio and s are presented theoretical value. The high rim (11 mm) of the sampler may as a function of l, S and Rf a introduce enhanced turbulence above the filter, decreasing the boundary layer thickness. In a low wind velocity situation a l S Rf Rt/Rm s thicker boundary layer may develop because of the rim, giving 0.01 — 40 3.9 6.0 rise to a higher boundary layer thickness than derived from 0.01 — 50 2.6 4.4 theory. 0.18 — 40 15.2 2.4 The 6 mm version of the Willems badge has a rim of 7 mm 0.18 — 50 6.9 1.5 instead of 11 mm and a resistance due to the diVusion length 0.18 0.015 50 3.6 0.8 and filter of 280 s m-1 instead of 105 s m-1.This will reduce 0.23 — 40 24.6 3.8 0.23 — 40 2.2 0.4 the variation in sampling rate from a factor of 2.5 to a factor 0.23 — 50 4.1 0.7 of 1.5, assuming the same boundary layer resistance. In reality, 0.23 — 50 2.6 0.5 this factor may even be lower than 1.5 when we take into 0.23 0.01 40 — 0.6 account that the maximum boundary layer thickness due to 0.23 0.015 40 2.2 0.4 the rim at low wind velocity will be less. 0.23 0.01 50 2.6 0.5 Assuming a linear relationship between rim height and 0.23 0.015 17 2.6 0.5 boundary layer thickness at u=0.2 m s-1, the variation in al=diVusion distance inside sampler (m); S=distance between draught sampling rate is estimated to be about 1.3.plate and sampler entrance (m); Rf=resistance of draught shield It will be obvious that a construction of the draught shield (s m-1) [40=coarse wire mesh (mesh size 17), 50=fine wire mesh (mesh size 50) and 17=glass fibre filter]; Rt/Rm=ratio of calculated for the Willems badge similar to that of the modified Palmes total resistance and measured total resistance; and s=standard devi- tube will largely avoid problems with a variable sampling rate ation of amount of ammonia on filter samples.and reduce the variation close to the theoretical value of 1.1. Table 5 Ratio of calculated and measured resistance against uptake Table 3 Theoretical and measured external resistances for a 2 mm of ammonia Rt/Rm and standard deviation s of absorption filter samples for a 0.11 m badge type sampler.This ratio and s are Willems badge as a function of wind velocity presented as a function of diVusion distance l, distance between turbulence damping plate and sampler base S and type of draught Wind speed/m s-1 screen, indicated by its resistance Rf Parameter 0.1 0.2 0.5 1.0 2.0 4.0 10 l S Rf Rt/Rm s Theoretical 131 114 95 83 72 63 52 Re/s m-1 0.08 0.01 — 17.0 7.0 0.08 0.01 40 3.5 3.5 Experimental — 232 137 76 38 38 — Re/s m-1 0.08 — 40 3.5 3.5 146 J.Environ. Monit., 1999, 1, 143–147The results of the experiments were used in the design of the modified Palmes tube. This tube should experience a minimum influence of ambient conditions on the sampling rate. Fig. 5 shows the results with the modified Palmes tube compared with the ECN wet rotating denuder (AMOR) as a reference sampler in a field experiment in Bavaria, Germany.15 In this case the calculated concentrations match the reference concentration very well.There is no indication of internal circulation of air within the sampler or influence of a boundary layer resistance. Conclusions Wind velocity influences the uptake rate of passive samplers. The influence is less for tube samplers than for badge type samplers. With a proper design of badge type samplers, the influence of wind velocity on the uptake rate can be diminished to 10% at a wind velocity of 0.2 m s-1.Draught shields are vital for proper functioning of badge Fig. 5 Linear relationship between passive sampler data and reference type samplers. The draught shield can be made of fine wire concentration for ammonia based on 2 week averages for six parallel mesh screen or a filter.samplers. Turbulence damping plates in front of a proper draught shield are not needed. 7 J. L. Monteith and M. H. Unsworth, Principles of Environmental Physics, Edward Arnold, London, 1990. A large air space within a passive sampler may give rise to 8 P.Pe� rez Ballesta, E. G. Ferradas and A.M. Aznar, Environ. Sci. thermal circulation, enhancing the sampling rate of the instru- Technol., 1993, 27, 2031. ment in an unpredictable way. 9 Ambient Air Quality; DiVusive Samplers for the Determination of Gases and Vapours—Requirements and Test Methods, Part 1. General Requirements, Draft Report, CEN/TC 264/WG11, 1998. References 10 Ambient Air Quality; DiVusive Samplers for the Determination of Gases and Vapours—Requirements and Test Methods.Part 3. Guide 1 R. van Aalst, L. Edwards, T. Pulles, E. de Saeger, M. Tombrou for Selection, Use and Maintenance, Draft Report, CEN/TC and D. Tonnesen, Guidance Report on Supplementary Assessment 264/WG11, 1998. Under EC Air Quality Directives. Second Draft, European 11 R. H. Brown, Pure Appl. Chem., 1993, 65, 1859. Commission, Brussels, 1997. 12 J. J. H. Willems and P. Hofschreuder, in A Passive Monitor for 2 E. D. Palmes, A. F. Gunnison, J. Di Matteo and C. Tomczyk, Am. Measuring Ammonia, ed. I. Allegrini, A. Febo and C. Perrino. Air Ind. Hyg. Assoc. J., 1976, 37, 570. Pollution Research Report 37, 1991, CEC, Brussels, pp. 113–121. 3 R. G. Lewis, J. D. Mulik, R. W. Coutant, G. W. Wooten and 13 G. P. Wyers, A. Wayers, J. J. Moels, Th. R. Thysse, J. H. Puyzer, C. R. McMillin, Anal. Chem., 1985, 57, 214. H. L. M. Verhagen and J. W. Erioman, Karakterisering van de 4 K. J. Hargreaves and D. H. F. Atkins, The Measurement of Regionale Concentratieveldevan Ammoniak: Monstername en Ammonia in the Outdoor Environment Using Passive DiVusion Tube Meetstrategie, ECN Report C-95-109, 1995. Samplers, Report AERE R 12568, Harwell Laboratory, Culham, 14 J. D. Posner and G. Moore, Am. Ind. Hyg. Assoc. J., 1985, 46, 277. 15 M. Kirchner, S. Braeutigam, M. Ferm, M. Haas, M. Hangartner, 1987. P. Hofschreuder, A. Kasper-Giebel, H. Roemmelt, J. Striedner, L. 5 D. B. Orr, J. C. Hipfner, W. H. Chan, M. A. Lusis and J. E. Hunt, de Temmerman, W. Terzer, L. Thoeni, H. Werner and R. Atmos. Environ., 1987, 21, 1473. Zimmerling, to be published. 6 T. Hafkenscheid, Comments on Document CEN/TC 264/11 n 50, 1998, (Concerning Correction for Pressure and Temperature for Passive Samplers), CEN/TC 264/WG11 N69, 1998. Paper 8/09269I J. Environ. Monit., 1999, 1, 143–147 147