A PASSIVE SAMPLER FOR HYDROGEN SULFIDE D. SHOOTER,* S. E WATTS1 and A. J. HAYES Chemistry Department, The University of Auckland, Private Bag 92019, Auckland, New Zealand 1 School of Biological and Molecular Sciences, Oxford Brookes University, Headington, Oxford, OX3 0BP, U.K.
(Received: December 1994; revised: May 1995)
Abstract. The silver nitrate/fluoresceinmercuric acetate fluorimetricmethodfor the measurement of atmospherichydrogensulfidehas been adapted to passive sampling.Standardsamplershave been tested and used in both indoor and outdoorenvironments. Samplerperformancewas not dependent on constructionmaterials or sunlightintensityand gave similarresults to active sampling. Two case studies were carried out, one in the HornimanMuseumand its associatedstorage and studybuilding, London, UK, and the other in the vicinity of a pulp and paper mill and geothermal area North Island, New Zealand. The detectionlimit of the samplers (50 ppt average for a one-weekexposure) provides the opportunityto make measurementsin a varietyof locations providedexposuretimes are sufficientlylong, i.e., up to one monthin areas of low hydrogensulfideconcentration.
1. Introduction Hydrogen sulfide (H2S) is one of the main reduced sulfur compounds cycling through the atmosphere (Brown and Bell, 1986). There is considerable uncertainty in the overall budget, but it is likely that it may be important in local sulfur budgets and in particular environments (Andreae and Andreae, 1988; Steudler and Peterson, 1984). Table I demonstrates the difficulty of budget estimation. It is clear that H2S production varies diurnally, seasonally and geographically. An overall budget of 0.5-1.2 g S m -2 a -1 has been estimated (Aneja, 1986; Aneja et aL, 1981; Moller, 1984a), with most of this being terrestrial in origin (0.4-0.7 g S m -2 a-i). Although the main sources of H2S to the atmosphere are geothermal and biogenic, it is the anthropogenic sources which cause the most concern because of the toxicity and deleterious effect of H2S on air quality as well as artefacts (Bates et aI., 1992). In addition, H2S has a very unpleasant odour which one can detect at levels of 500 ppt (Siegel et al., 1986). Urban and anthropogenic concentrations vary considerably. Some examples are 430 ppt during a local pollution episode at Boulder, Colorado (Natusch et al., 1972); 590 ppt in French air samples (Servant and Delapart, 1982); 1320 ppt in Frankfurt (Jaeschke and Herrmann, 1981). At these levels HzS is not known to be harmful (although may induce headaches and non-specific symptoms), but it can cause odour complaints (Furr, 1990; Siegel et al., 1986). Consequently, quantification of atmospheric HzS concentrations has been seen as important and there are a number of measurement methods available. Colourimetric methods have been in use to quantitate H2S in laboratory air as well as natural waters for about 70 years (Almy, 1925), although they have changed * To whomcorrespondence shouldbe addressed. Environmental Monitoring and Assessment 38:11-23, 1995. (~) 1995 Kluwer Academic Publishers. Printed in the Netherlands.
12
D. SHOOTERET AL. TABLEI Estimated fluxes(if) of hydrogensulfidefrom various sources to the troposphere [mg S m-2
50-2000 0.9-70 26-35 2320 0--79 0.4-88 139 500 29-260
500-17 300 1000--10 000 300
a -1]
Source
Reference
Steudler and Peterson (1984) Goldan et al. (1987) Wetlands Cooper et al. (1987) Intertidal mudflats Bates et al. (1992); Jaeschke(1978) Tidal Creeks Steudler and Peterson (1984) Estuaries Jorgenson and Okholm-Hansen(1985) Aerated soils Adams et al. (1979); Delmas et aI. (1980) Servant and Delapart (1982) Saline marsh Adams et al. (1979); Bates et al. (i992) Tropical Forests Delmas et al. (1980) Delmas and Servant (1983) Andreae and Andreae (1988) Volcanic eruption Estimated'~ from Bandyet al. (1982); Jaeschke and Herrmann(1981); Kodoskyet al. (1991) Geothermal systems Estimated~ from Cope (1981); D'Amore and Panichi (1980); Siegelet al. (1986) Estimatedb fromMoiler (1984a) Anthropogenic/Industry Spartina Alterniflora
a Eruption frequencyfor all calculationsassumedto be 0.04 a-z and durationfor whole year. Total crater/fumarolearea for each systemassumedto be 1 kmz. 3 Tg S a -1 (Moiler, 1984a) dividedby an area of 100 km2.
considerably in that time. Modem methods (Balasubramanian and Kumar, 1990; Koh e t al., 1990; Leggett et al., 1981), involve trapping the H2S in a solution with a chromophore and quantifying of an addition product (sometimes using flow injection techniques). The best of these techiques has an absorption efficiency of >98% and a detection limit of about i 300 000 ppt with sampling periods ranging from 20 to 360 rain (Balasubramanian and Kumar, 1990). A more sensitive variant of this type of methodology is the colourimetric tube. Air is sucked through robes which have been packed with silica gel impregnated with a silver-gelatin complex, with the formation of Ag2S being observed. Colourimetric tubes have been used to sample urban air, but have been more widely employed in the safety industry. Detection limits are in the range 300-30 000 ppt (Bhatia, 1988; Pal et al., 1986), and sampling is usually over a period of 1 to 10 rain. One problem identified with this method is that sunlight (or artificial lights) can photoreduce the silver-gelatin complex (Pal e t al., 1986). Chemically treated papers are also widely used (Hayes e t al., 1990; LaRue e t al., 1987; Natusch e t al., 1972). Air is drawn through a treated filter paper forming a stain in proportion to the H2S concentration. Such dosimeters have been used in
A PASSIVE SAMPLER FOR HYDROGEN SULFIDE
13
remote areas and urban environments as well in the laboratory. Typical sampling periods are in the range of 15-60 min, and detection limits are of the order of 50 000 ppt (LaRue et al., 1987). A more sensitive method with a detection limit of about 50 ppt and air sampling volumes in the vicinity of 3000 dm 3 (sampling time about 2-3 h) utilises the quenching of fluorescein mercuric acetate (FMA) by the Ag2S formed from the H2S (Axelrod et al., 1969; Natusch et al., 1974). Interferences by other trace gases that might becollected using Ag-impregnated filters, e.g., 03, OCS; CS2; SO2; CH3SH and NO2 have been shown to be minimal (Farwell etal., 1987; Jaeschke and Herrmann, 1981; Natusch etal., 1974), although OCS may be a problem in situations of low H2S (Cooper et al., 1987). Gas chromatography is also used to quantify reduced sulfur gases including H2S (Turner and Liss, 1985). Modem systems involve preconcentration onto an absorbent (e.g., Tenax or gold wool), often at room temperature, and subsequent quantification with a flame photometric or chemiluminesence detector (Gibson et al., 1994; Sze and Kho, 1980). Although detection limits are down at the 0.1 ppt level, these systems usually require constant operator presence. Passive samplers are used for a number of trace gases, e.g., NO2; 03; SO2 (Brown, 1993; Gair et al., 1991; Monn and Hangamer, 1990). They consist of a small tube with one end sealed. Transfer of an ambient trace gas from the open end to an absorber held at the closed end relies on molecular diffusion of the gas along the tube with Fick's Law being used to calculate the average concentration of the gas. The relatively low sampling rate of these passive devices means that sampling times vary, e.g., for NO2, in urban situations, the sampling time may be as little as 12 h, whereas in remote (non-polluted) regions 5 days may be required for adsorption of sufficient material to enable quantification. Passive samplers have several advantages over conventional sampling techniques; the sampler is small and not functionally complex. Passive samplers employ a passive method and hence do not require power or other services; their low cost means theY can be deployed simultaneously at a large number of sites; the results are in the form of time averaged concentrations (useful for establishing the distribution of pollution for policy and planning purposes). There are also some disadvantages: de facto this method does not generate instantaneous concentrations and their sensitivity can be limited by short exposure times. Also, there are some inherent limitations to the performance of passive samplers; differences between individual samplers can lead to relative standard deviations of up to -4- 20%, and -4- 10% is common. However, their accuracy has been tested by comparison with active methods of sampling and good agreement has been found (see for example Gair et al., 1991). Provided they are used with an awareness of these limitations, they are a pratical method of trace gas monitoring. This paper describes the validation and field trials of a passive sampler which is specific for H2S. There are numerous locations where a H2S passive sampler could prove useful. For example, the measurement of indoor and outdoor air pollution and in the
14
D. SHOOTER ET AL. polyethylene cap
clear acrylic tube
(71 mm Ion(J, 11mm internal diameter)
t:~ / ~ i ~ l l
\
~
r
.o,oro,n°.,°.o.
, ~
or stalnfll~Ps~eapo;'lmesh
polyethylene cap (removed to expose the silver nitrate absorbing solution to the atmosphere)
Fig. 1. Expanded view of the Palmes type passive sampler used in this study showing meshes, caps and acrylic tube.
vicinity of traffic, geothermal areas and industrial complexes. Samplers might also be useful in the measurement of emissions from plants and micro-organisms and even unusual uses such as the measurement of the build-up of HzS in ship's bilges.
2. Methodology Full details of the methodology have appeared elsewhere (Shooter, 1993). However, briefly, samplers used in this study consisted of an acrylic tube, 7.1 cm long and 1.2 cm in diameter, sealed at one end with a clear polythene cap and an absorbent supported at the other end on either stainless steel meshes or filter paper held in a red polythene cap, see Figure 1. Samplers were prepared as follows. Stainless steel meshes. Two were placed in the red polythene cap and coated with 50 microlitres of a 1% aqueous solution of AgNO3 containing 20% absolute ethanol and 10% glycerol. Filter paper in 1.2 cm diameter circles were cut from Whatman No. 4 filter paper and soaked for 2 h in a solution of 0.01 mol 1-1 nitric acid containing 2% AgNO3, 20% absolute ethanol and 2% glycerol before insertion in the red polythene cap. All samplers had their open end sealed with a colourless cap when not in use and samplers were either assembled as quickly as possible to minimise contamination by laboratory air or assembled in a nitrogen gas bag. Plastic forceps and gloves were used to minimise contamination of the samplers.
A PASSIVE SAMPLER FOR HYDROGEN SULFIDE
15
For the fieldwork studies samplers were atached in groups of 5 to a wall or other vertical surface either by adhesive tape or putty. The samplers were mounted vertically with the mesh end uppermost. Trace gas collection was initiated by removal of the colourless polythene cap and terminated by replacing the cap. Some samplers were left unopened to obtain their 'blank' response. FMA was prepared by the method of Axelrod et al. (1969), dried over molecular sieve 5A, recrystallised and storedin the dark. Sodium cyanide and sodium hydroxide were analytical grade and used as supplied. Analytical grade AgNO3 was recrystallised prior to use. Sodium sulfide was washed with deionised water followed by absolute ethanol. It was allowed to dry at room temperature for 10 rain before being weighed and dissolved in 0.10 mol dm -3 sodium hydroxide. After exposure samplers were stored in the dark until analysed with storage times varying from 5 to 48 h. The samplers were opened and the sulfide extracted from the meshes or filter papers by the addition of 2.5 cm 3 of 0.10 tool dm -3 sodium cyanide in 0.10 mol dm -3 sodium hydroxide. The extract was transferred to a stoppered quartz spectrofluorimeter cell, 0.10 cm 3 of the FMA solution added, and the fluorescence measured using either Shimadzu RF-540 or a Cecil spectrofluorimeter. Both the extraction and the cell filling was carried out in a nitrogen atmosphere (gas bag) to minimise the oxidation of the FMA. After an extensive study Farwell et al. (1987) concluded that aqueous sulfide can be used as a standard for the AgNO3 method although there is a risk of underestimating the precision of the method. Sulfide standards were used in this study and prepared by dilution of a 1.00 x 10 .3 mol dm -3 solution of sodium sulfide using 0.10 mol dm -3 sodium cyanide solution. This produced standards in the range 0 to 1.5 x 10 .7 tool dm -3. Fresh sulfide standards were prepared for each set of samplers analysed. As indicated by Figure 2 the mass of sulfide analysed using our methodology has to be less than 20 ng. Active (as opposed to passive) atmospheric sampling was carried out using AgNO3 impregnated filter paper held in a 50 mm polycarbonate in-line filter holder. A 12 V pump and battery exposed the filter papers to a flowrate of 2000 dm 3 h -1. Blanks values were obtained by not exposing several of the prepared filter papers. Exposure chamber tests showed that passive samplers with a AgNO3 absorber responded linearly to H2S concentration. A value for the diffusion coefficient of H2S of 0.160 cm 2 s -1 has been reported by Gudzhedzhiani (1978). Using this value, the effective sampling rate of the sampler for H2S in air is calculated to be 0.077 dm 3 of air per hour.
16
D. SHOOTER ET AL.
500
400.
300 fluorescence intensity
v
I
I
o
s
1'o
l's
20
mass S" [ng]
Fig. 2. Calibration curve for the AgNO3 method for H2S. FMA (fluorescein mercuric acetate) fluorescence intensity vs. ng sulfide in a 3.0 cm3 10 mm quartz cell.
3. Method Development The 'precarious' nature of this method has been noted previously (Farwell et al., 1987). We observed the instability of FMA in solution and traced the cause to dissolved oxygen. The deterioration of FMA fluorescence was prevented by sparging all solutions with nitrogen prior to their contact with FMA and by keeping all FMA solutions in the dark. When these precautions were taken the spectrofluorimeter readings were stable over a period of several hours. Tests of absorbent holder material revealed little difference in sampler response between stainless steel mesh and filter paper. Both these materials were found to be satisfactory, with sonification not being required for the extraction of the sulfide from the filter paper. The sampler response did not appear to depend on the tube material. Samplers with glass and acrylic bodies exposed together near the Kawerau geothermal field (see Case Study 1 below) showed no significant difference in response. Six acrylic samplers gave a H2S concentration of 20 600 + 400 ppt while the glass samplers gave a concentration of 22 200 + 1700 ppt. The accuracy of the samplers was tested by a comparison of passive with active sampling. In a field trial 9 passive samplers were exposed for 3.5 days and the
A PASSIVESAMPLERFOR HYDROGENSULFIDE
17
active sampling apparatus was used to collect H2S over the same period with the AgNO3 impregnated filter papers being renewed every 12 h. The 9 samplers gave an average HzS" concentration of 660 4- 90 ppt for the period, while for the active sampling the 7 filter papers gave a comparable mean value of 610 ppt. (Pal, et al., 1986), observed that while direct sunlight did not affect the response of a AgNO3/gelatin complex in aqueous media it did reduce the silver in a silica gel column with a high concentration of AgNO3. Our comparison of insolated and blacked out samplers in a sampling situation revealed no difference in response to H2S. An excess of Ag + was present in our samplers. Assuming exposure for one week at an average H2S concentration of 5 000 ppt, approximately one half of the initial amount of Ag+ would remain after the exposure. Samplers which were kept capped during an exposure (blanks) gave a response that was much less than the uncapped samplers. The detection limit of the samplers was calculated as the mean of blank response plus three standard deviations. This calculation gave a detection limit of 1 ng H2S, or approximately 50 ppt for a oneweek exposure. Sampler detection limits are often expressed in ppb/h or ppt/h. In these terms the detection limit of the samplers is 8 400 ppt/h. Exposure times for samplers in a variety of situations can be estimated using this value. As mentioned above, previous studies have not found any interference by other atmospheric trace gases through collection on the Ag-impregnated filter papers and stibsequent fluorescence quenching of the FMA. Consequently this aspect of samplers performance was not examined. Although temperature, humidity and air velocity are also potential variables in passive sampling, they do not lead to significant errors for well designed samplers such as those used in this study (Brown, 1993).
3.1.
CASE STUDY 1 (INDOOR)
Museums can contain a range of materials sensitive to air pollutants. While a number of atmospheric trace gases can cause problems, HzS is likely to cause specific problems such as tarnishing and corrosion of metal (Pope et al., 1968). To obtain information regarding museum storage and display environments HzS levels were measured using both active and passive sampling in the Horniman Museum, South London and the Dreadnought Study Collection Centre. The Study Centre, situated in an industrial area of South London, UK, is a Victorian school building now used for artefact storage and study by the Horniman Museum. Indoor and outdoor concentrations were measured at both locations in September 1992. Several measurements were also made on the HzS emission characteristics of museum exhibits. Passive samplers, in groups of five, were exposed for one week (3-10 September 1992) at both sites, in locations chosen to indicate the H2S outdoor/indoor concentration ratio and concentration variation within the buildings. Active sam-
18
D. SHOOTERET AL.
piing was used for short-term measurement of the outdoor/indoor concentration ratio. Both sampling methods were used for exhibit emission checks. Levels of H2S in the Homiman Museum were low, neither museum patrons nor the local South London environment provided sufficient amounts of H2S to exceed the passive sampler detection limit of 50 ppt (one week's exposure). However, 60 ppt HzS was measured by active sampling (15 minutes, 0.24 m 3 air) in the North Hall of the Museum while it was open to the public. Outdoor HzS in the vicinity of the Homiman Museum was not detectable by either sampling method. In contrast to the Homiman Museum environment, outdoor levels in the vicinity of the Dreadnought Study Centre were high, Passive sampling gave a weekly average of 710 ppt while active sampling gave a 15-min average (measured at midday) of 1810 ppt. The area surrounding the Study Centre is subject to heavy diesel and automobile traffic and contains several food processing industries, all of which may have contributed to the local H2S levels. Catalyst-equipped vehicles are known to emit hydrogen sulfide (Fried et al., 1992). Our results suggest that time-averaged concentrations of HzS provided by passive sampling could be useful as a means of investigating concerns regarding these H2S emissions. Passive samplers positioned at various distances from the entrace to the Dreadnought Study Centre showed that the concentration of HzS decreased towards the centre of the building, presumably through adsorption or chemical reaction. The highest concentration of 350 ppt was seen in the stairwell near the entrace to the Centre (between levels 1 and 2). Levels decreased with distance from the entrace (110 pt Level 1,140 ppt Level 2 and 66 ppt Level 3). This decrease was also seen in the indoor/outdoor ratios. A ratio of 0.45 was observed near the entrace but decreased to 0.13 further into the building. Several display cases in the Horniman Museum were checked for H2S levels using passive samplers (one week exposure). The only case with detectable levels (660 ppt) displayed stuffed birds. Feathers contain quantities of sulfur-containing protein and are thought to be the source of the H2S. All other display case contained
(Received: December 1994; revised: May 1995)
Abstract. The silver nitrate/fluoresceinmercuric acetate fluorimetricmethodfor the measurement of atmospherichydrogensulfidehas been adapted to passive sampling.Standardsamplershave been tested and used in both indoor and outdoorenvironments. Samplerperformancewas not dependent on constructionmaterials or sunlightintensityand gave similarresults to active sampling. Two case studies were carried out, one in the HornimanMuseumand its associatedstorage and studybuilding, London, UK, and the other in the vicinity of a pulp and paper mill and geothermal area North Island, New Zealand. The detectionlimit of the samplers (50 ppt average for a one-weekexposure) provides the opportunityto make measurementsin a varietyof locations providedexposuretimes are sufficientlylong, i.e., up to one monthin areas of low hydrogensulfideconcentration.
1. Introduction Hydrogen sulfide (H2S) is one of the main reduced sulfur compounds cycling through the atmosphere (Brown and Bell, 1986). There is considerable uncertainty in the overall budget, but it is likely that it may be important in local sulfur budgets and in particular environments (Andreae and Andreae, 1988; Steudler and Peterson, 1984). Table I demonstrates the difficulty of budget estimation. It is clear that H2S production varies diurnally, seasonally and geographically. An overall budget of 0.5-1.2 g S m -2 a -1 has been estimated (Aneja, 1986; Aneja et aL, 1981; Moller, 1984a), with most of this being terrestrial in origin (0.4-0.7 g S m -2 a-i). Although the main sources of H2S to the atmosphere are geothermal and biogenic, it is the anthropogenic sources which cause the most concern because of the toxicity and deleterious effect of H2S on air quality as well as artefacts (Bates et aI., 1992). In addition, H2S has a very unpleasant odour which one can detect at levels of 500 ppt (Siegel et al., 1986). Urban and anthropogenic concentrations vary considerably. Some examples are 430 ppt during a local pollution episode at Boulder, Colorado (Natusch et al., 1972); 590 ppt in French air samples (Servant and Delapart, 1982); 1320 ppt in Frankfurt (Jaeschke and Herrmann, 1981). At these levels HzS is not known to be harmful (although may induce headaches and non-specific symptoms), but it can cause odour complaints (Furr, 1990; Siegel et al., 1986). Consequently, quantification of atmospheric HzS concentrations has been seen as important and there are a number of measurement methods available. Colourimetric methods have been in use to quantitate H2S in laboratory air as well as natural waters for about 70 years (Almy, 1925), although they have changed * To whomcorrespondence shouldbe addressed. Environmental Monitoring and Assessment 38:11-23, 1995. (~) 1995 Kluwer Academic Publishers. Printed in the Netherlands.
12
D. SHOOTERET AL. TABLEI Estimated fluxes(if) of hydrogensulfidefrom various sources to the troposphere [mg S m-2
50-2000 0.9-70 26-35 2320 0--79 0.4-88 139 500 29-260
500-17 300 1000--10 000 300
a -1]
Source
Reference
Steudler and Peterson (1984) Goldan et al. (1987) Wetlands Cooper et al. (1987) Intertidal mudflats Bates et al. (1992); Jaeschke(1978) Tidal Creeks Steudler and Peterson (1984) Estuaries Jorgenson and Okholm-Hansen(1985) Aerated soils Adams et al. (1979); Delmas et aI. (1980) Servant and Delapart (1982) Saline marsh Adams et al. (1979); Bates et al. (i992) Tropical Forests Delmas et al. (1980) Delmas and Servant (1983) Andreae and Andreae (1988) Volcanic eruption Estimated'~ from Bandyet al. (1982); Jaeschke and Herrmann(1981); Kodoskyet al. (1991) Geothermal systems Estimated~ from Cope (1981); D'Amore and Panichi (1980); Siegelet al. (1986) Estimatedb fromMoiler (1984a) Anthropogenic/Industry Spartina Alterniflora
a Eruption frequencyfor all calculationsassumedto be 0.04 a-z and durationfor whole year. Total crater/fumarolearea for each systemassumedto be 1 kmz. 3 Tg S a -1 (Moiler, 1984a) dividedby an area of 100 km2.
considerably in that time. Modem methods (Balasubramanian and Kumar, 1990; Koh e t al., 1990; Leggett et al., 1981), involve trapping the H2S in a solution with a chromophore and quantifying of an addition product (sometimes using flow injection techniques). The best of these techiques has an absorption efficiency of >98% and a detection limit of about i 300 000 ppt with sampling periods ranging from 20 to 360 rain (Balasubramanian and Kumar, 1990). A more sensitive variant of this type of methodology is the colourimetric tube. Air is sucked through robes which have been packed with silica gel impregnated with a silver-gelatin complex, with the formation of Ag2S being observed. Colourimetric tubes have been used to sample urban air, but have been more widely employed in the safety industry. Detection limits are in the range 300-30 000 ppt (Bhatia, 1988; Pal et al., 1986), and sampling is usually over a period of 1 to 10 rain. One problem identified with this method is that sunlight (or artificial lights) can photoreduce the silver-gelatin complex (Pal e t al., 1986). Chemically treated papers are also widely used (Hayes e t al., 1990; LaRue e t al., 1987; Natusch e t al., 1972). Air is drawn through a treated filter paper forming a stain in proportion to the H2S concentration. Such dosimeters have been used in
A PASSIVE SAMPLER FOR HYDROGEN SULFIDE
13
remote areas and urban environments as well in the laboratory. Typical sampling periods are in the range of 15-60 min, and detection limits are of the order of 50 000 ppt (LaRue et al., 1987). A more sensitive method with a detection limit of about 50 ppt and air sampling volumes in the vicinity of 3000 dm 3 (sampling time about 2-3 h) utilises the quenching of fluorescein mercuric acetate (FMA) by the Ag2S formed from the H2S (Axelrod et al., 1969; Natusch et al., 1974). Interferences by other trace gases that might becollected using Ag-impregnated filters, e.g., 03, OCS; CS2; SO2; CH3SH and NO2 have been shown to be minimal (Farwell etal., 1987; Jaeschke and Herrmann, 1981; Natusch etal., 1974), although OCS may be a problem in situations of low H2S (Cooper et al., 1987). Gas chromatography is also used to quantify reduced sulfur gases including H2S (Turner and Liss, 1985). Modem systems involve preconcentration onto an absorbent (e.g., Tenax or gold wool), often at room temperature, and subsequent quantification with a flame photometric or chemiluminesence detector (Gibson et al., 1994; Sze and Kho, 1980). Although detection limits are down at the 0.1 ppt level, these systems usually require constant operator presence. Passive samplers are used for a number of trace gases, e.g., NO2; 03; SO2 (Brown, 1993; Gair et al., 1991; Monn and Hangamer, 1990). They consist of a small tube with one end sealed. Transfer of an ambient trace gas from the open end to an absorber held at the closed end relies on molecular diffusion of the gas along the tube with Fick's Law being used to calculate the average concentration of the gas. The relatively low sampling rate of these passive devices means that sampling times vary, e.g., for NO2, in urban situations, the sampling time may be as little as 12 h, whereas in remote (non-polluted) regions 5 days may be required for adsorption of sufficient material to enable quantification. Passive samplers have several advantages over conventional sampling techniques; the sampler is small and not functionally complex. Passive samplers employ a passive method and hence do not require power or other services; their low cost means theY can be deployed simultaneously at a large number of sites; the results are in the form of time averaged concentrations (useful for establishing the distribution of pollution for policy and planning purposes). There are also some disadvantages: de facto this method does not generate instantaneous concentrations and their sensitivity can be limited by short exposure times. Also, there are some inherent limitations to the performance of passive samplers; differences between individual samplers can lead to relative standard deviations of up to -4- 20%, and -4- 10% is common. However, their accuracy has been tested by comparison with active methods of sampling and good agreement has been found (see for example Gair et al., 1991). Provided they are used with an awareness of these limitations, they are a pratical method of trace gas monitoring. This paper describes the validation and field trials of a passive sampler which is specific for H2S. There are numerous locations where a H2S passive sampler could prove useful. For example, the measurement of indoor and outdoor air pollution and in the
14
D. SHOOTER ET AL. polyethylene cap
clear acrylic tube
(71 mm Ion(J, 11mm internal diameter)
t:~ / ~ i ~ l l
\
~
r
.o,oro,n°.,°.o.
, ~
or stalnfll~Ps~eapo;'lmesh
polyethylene cap (removed to expose the silver nitrate absorbing solution to the atmosphere)
Fig. 1. Expanded view of the Palmes type passive sampler used in this study showing meshes, caps and acrylic tube.
vicinity of traffic, geothermal areas and industrial complexes. Samplers might also be useful in the measurement of emissions from plants and micro-organisms and even unusual uses such as the measurement of the build-up of HzS in ship's bilges.
2. Methodology Full details of the methodology have appeared elsewhere (Shooter, 1993). However, briefly, samplers used in this study consisted of an acrylic tube, 7.1 cm long and 1.2 cm in diameter, sealed at one end with a clear polythene cap and an absorbent supported at the other end on either stainless steel meshes or filter paper held in a red polythene cap, see Figure 1. Samplers were prepared as follows. Stainless steel meshes. Two were placed in the red polythene cap and coated with 50 microlitres of a 1% aqueous solution of AgNO3 containing 20% absolute ethanol and 10% glycerol. Filter paper in 1.2 cm diameter circles were cut from Whatman No. 4 filter paper and soaked for 2 h in a solution of 0.01 mol 1-1 nitric acid containing 2% AgNO3, 20% absolute ethanol and 2% glycerol before insertion in the red polythene cap. All samplers had their open end sealed with a colourless cap when not in use and samplers were either assembled as quickly as possible to minimise contamination by laboratory air or assembled in a nitrogen gas bag. Plastic forceps and gloves were used to minimise contamination of the samplers.
A PASSIVE SAMPLER FOR HYDROGEN SULFIDE
15
For the fieldwork studies samplers were atached in groups of 5 to a wall or other vertical surface either by adhesive tape or putty. The samplers were mounted vertically with the mesh end uppermost. Trace gas collection was initiated by removal of the colourless polythene cap and terminated by replacing the cap. Some samplers were left unopened to obtain their 'blank' response. FMA was prepared by the method of Axelrod et al. (1969), dried over molecular sieve 5A, recrystallised and storedin the dark. Sodium cyanide and sodium hydroxide were analytical grade and used as supplied. Analytical grade AgNO3 was recrystallised prior to use. Sodium sulfide was washed with deionised water followed by absolute ethanol. It was allowed to dry at room temperature for 10 rain before being weighed and dissolved in 0.10 mol dm -3 sodium hydroxide. After exposure samplers were stored in the dark until analysed with storage times varying from 5 to 48 h. The samplers were opened and the sulfide extracted from the meshes or filter papers by the addition of 2.5 cm 3 of 0.10 tool dm -3 sodium cyanide in 0.10 mol dm -3 sodium hydroxide. The extract was transferred to a stoppered quartz spectrofluorimeter cell, 0.10 cm 3 of the FMA solution added, and the fluorescence measured using either Shimadzu RF-540 or a Cecil spectrofluorimeter. Both the extraction and the cell filling was carried out in a nitrogen atmosphere (gas bag) to minimise the oxidation of the FMA. After an extensive study Farwell et al. (1987) concluded that aqueous sulfide can be used as a standard for the AgNO3 method although there is a risk of underestimating the precision of the method. Sulfide standards were used in this study and prepared by dilution of a 1.00 x 10 .3 mol dm -3 solution of sodium sulfide using 0.10 mol dm -3 sodium cyanide solution. This produced standards in the range 0 to 1.5 x 10 .7 tool dm -3. Fresh sulfide standards were prepared for each set of samplers analysed. As indicated by Figure 2 the mass of sulfide analysed using our methodology has to be less than 20 ng. Active (as opposed to passive) atmospheric sampling was carried out using AgNO3 impregnated filter paper held in a 50 mm polycarbonate in-line filter holder. A 12 V pump and battery exposed the filter papers to a flowrate of 2000 dm 3 h -1. Blanks values were obtained by not exposing several of the prepared filter papers. Exposure chamber tests showed that passive samplers with a AgNO3 absorber responded linearly to H2S concentration. A value for the diffusion coefficient of H2S of 0.160 cm 2 s -1 has been reported by Gudzhedzhiani (1978). Using this value, the effective sampling rate of the sampler for H2S in air is calculated to be 0.077 dm 3 of air per hour.
16
D. SHOOTER ET AL.
500
400.
300 fluorescence intensity
v
I
I
o
s
1'o
l's
20
mass S" [ng]
Fig. 2. Calibration curve for the AgNO3 method for H2S. FMA (fluorescein mercuric acetate) fluorescence intensity vs. ng sulfide in a 3.0 cm3 10 mm quartz cell.
3. Method Development The 'precarious' nature of this method has been noted previously (Farwell et al., 1987). We observed the instability of FMA in solution and traced the cause to dissolved oxygen. The deterioration of FMA fluorescence was prevented by sparging all solutions with nitrogen prior to their contact with FMA and by keeping all FMA solutions in the dark. When these precautions were taken the spectrofluorimeter readings were stable over a period of several hours. Tests of absorbent holder material revealed little difference in sampler response between stainless steel mesh and filter paper. Both these materials were found to be satisfactory, with sonification not being required for the extraction of the sulfide from the filter paper. The sampler response did not appear to depend on the tube material. Samplers with glass and acrylic bodies exposed together near the Kawerau geothermal field (see Case Study 1 below) showed no significant difference in response. Six acrylic samplers gave a H2S concentration of 20 600 + 400 ppt while the glass samplers gave a concentration of 22 200 + 1700 ppt. The accuracy of the samplers was tested by a comparison of passive with active sampling. In a field trial 9 passive samplers were exposed for 3.5 days and the
A PASSIVESAMPLERFOR HYDROGENSULFIDE
17
active sampling apparatus was used to collect H2S over the same period with the AgNO3 impregnated filter papers being renewed every 12 h. The 9 samplers gave an average HzS" concentration of 660 4- 90 ppt for the period, while for the active sampling the 7 filter papers gave a comparable mean value of 610 ppt. (Pal, et al., 1986), observed that while direct sunlight did not affect the response of a AgNO3/gelatin complex in aqueous media it did reduce the silver in a silica gel column with a high concentration of AgNO3. Our comparison of insolated and blacked out samplers in a sampling situation revealed no difference in response to H2S. An excess of Ag + was present in our samplers. Assuming exposure for one week at an average H2S concentration of 5 000 ppt, approximately one half of the initial amount of Ag+ would remain after the exposure. Samplers which were kept capped during an exposure (blanks) gave a response that was much less than the uncapped samplers. The detection limit of the samplers was calculated as the mean of blank response plus three standard deviations. This calculation gave a detection limit of 1 ng H2S, or approximately 50 ppt for a oneweek exposure. Sampler detection limits are often expressed in ppb/h or ppt/h. In these terms the detection limit of the samplers is 8 400 ppt/h. Exposure times for samplers in a variety of situations can be estimated using this value. As mentioned above, previous studies have not found any interference by other atmospheric trace gases through collection on the Ag-impregnated filter papers and stibsequent fluorescence quenching of the FMA. Consequently this aspect of samplers performance was not examined. Although temperature, humidity and air velocity are also potential variables in passive sampling, they do not lead to significant errors for well designed samplers such as those used in this study (Brown, 1993).
3.1.
CASE STUDY 1 (INDOOR)
Museums can contain a range of materials sensitive to air pollutants. While a number of atmospheric trace gases can cause problems, HzS is likely to cause specific problems such as tarnishing and corrosion of metal (Pope et al., 1968). To obtain information regarding museum storage and display environments HzS levels were measured using both active and passive sampling in the Horniman Museum, South London and the Dreadnought Study Collection Centre. The Study Centre, situated in an industrial area of South London, UK, is a Victorian school building now used for artefact storage and study by the Horniman Museum. Indoor and outdoor concentrations were measured at both locations in September 1992. Several measurements were also made on the HzS emission characteristics of museum exhibits. Passive samplers, in groups of five, were exposed for one week (3-10 September 1992) at both sites, in locations chosen to indicate the H2S outdoor/indoor concentration ratio and concentration variation within the buildings. Active sam-
18
D. SHOOTERET AL.
piing was used for short-term measurement of the outdoor/indoor concentration ratio. Both sampling methods were used for exhibit emission checks. Levels of H2S in the Homiman Museum were low, neither museum patrons nor the local South London environment provided sufficient amounts of H2S to exceed the passive sampler detection limit of 50 ppt (one week's exposure). However, 60 ppt HzS was measured by active sampling (15 minutes, 0.24 m 3 air) in the North Hall of the Museum while it was open to the public. Outdoor HzS in the vicinity of the Homiman Museum was not detectable by either sampling method. In contrast to the Homiman Museum environment, outdoor levels in the vicinity of the Dreadnought Study Centre were high, Passive sampling gave a weekly average of 710 ppt while active sampling gave a 15-min average (measured at midday) of 1810 ppt. The area surrounding the Study Centre is subject to heavy diesel and automobile traffic and contains several food processing industries, all of which may have contributed to the local H2S levels. Catalyst-equipped vehicles are known to emit hydrogen sulfide (Fried et al., 1992). Our results suggest that time-averaged concentrations of HzS provided by passive sampling could be useful as a means of investigating concerns regarding these H2S emissions. Passive samplers positioned at various distances from the entrace to the Dreadnought Study Centre showed that the concentration of HzS decreased towards the centre of the building, presumably through adsorption or chemical reaction. The highest concentration of 350 ppt was seen in the stairwell near the entrace to the Centre (between levels 1 and 2). Levels decreased with distance from the entrace (110 pt Level 1,140 ppt Level 2 and 66 ppt Level 3). This decrease was also seen in the indoor/outdoor ratios. A ratio of 0.45 was observed near the entrace but decreased to 0.13 further into the building. Several display cases in the Horniman Museum were checked for H2S levels using passive samplers (one week exposure). The only case with detectable levels (660 ppt) displayed stuffed birds. Feathers contain quantities of sulfur-containing protein and are thought to be the source of the H2S. All other display case contained
