O 75 O 70 65 60 55 50

FIGURE 11.36 Zonal average concentrations of CO in (a) the Northern Hemisphere and (b) the Southern Hemisphere (adapted from Novelli et al., 1998b).

Southern Hemisphere. The seasonal trends observed in both hemispheres reflect an anticorrelation with OH that removes CO from the atmosphere. CO in the NH, for example, peaks in March/April and is a minimum in July.

Detection techniques. As shown in Table 11.1, common techniques for measuring SOz include UV fluorescence, DOAS, and a wet chemical (pararosaniline) technique in which S02 undergoes the Schiff reaction with pararosaniline, HCHO, and HC1 to form a red-violet product whose absorbance at 560-580 nm is measured (West and Gaeke, 1956; Dasgupta and Gupta, 1986). Filter packs (e.g., Ferek and Hegg, 1993), diffusion scrubbers (e.g., Lindgren and Dasgupta, 1989), and denuders (e.g., see Pui et al., 1990) have also been used for S02.

Infrared techniques, including TDLS (Schiff et al., f994b) and matrix isolation FTIR (Griffith and Schuster, f987), have been applied to S02, with detection limits of 500 and fO ppt, respectively.

Several chemiluminescence methods have also been developed for S02. For example, S02 has been shown to enhance the chemiluminescence signal from the luminol-N02 reaction so that the enhancement can be used as a measure of S02 at a fixed N02 concentration (Zhang et al., 1985). Another method involves the formation of SO in a hydrogen flame followed by reaction with 03 to generate electronically excited S02 whose emission is followed at 340 nm (e.g., see Benner and Stedman, 1990).

Gas chromatography with various detectors such as flame photometry have been applied in a number of studies; flame photometry involves the generation of

FIGURE 11.36 Zonal average concentrations of CO in (a) the Northern Hemisphere and (b) the Southern Hemisphere (adapted from Novelli et al., 1998b).


Year electronically excited S2 in a hydrogen-rich flame, and the emission at 394 nm is followed. The signal is proportional to the square of the concentration of the sulfur compound. Since all compounds containing sulfur will form S2 in the flame, a separation technique such as GC must be used to differentiate S02 from other sulfur compounds.

Another GC method, isotope dilution GC-MS, involves the addition of an isotopomer of the analyte of interest to the sampling manifold (e.g., see Bandy et al., 1993; and Blomquist et al., 1993). In the case of S02, where the ambient S02 consists mainly of 16 O and 32 S, S02 containing the 34 S isotope is used. This labeled S02 at mass 66 is used as internal standard and has a number of additional advantages such as minimizing the loss of the analyte in the sampling system (Bandy et al., 1993). The air with the added isotopomer is trapped cryogenically and then sampled into a GC-MS for analysis.

The results of intercomparison studies suggest that measuring S02, particularly at low concentrations found in remote areas, is difficult. For example, Gregory et al. (1993) evaluated five techniques for measuring S02 in air: GC with flame photometric or mass spectrometric detection, a chemiluminescence method using reaction with KMn04, and a filter method with either the KMn04 chemiluminescence or ion chromatographic detection. Above ~200 ppt, agreement between pairs of techniques varied from about 30% to several orders of magnitude. At concentrations below this, there was no correlation between the measurements made using these five methods! Kok et al. (1990) compared a filter pack, diffusion denuder, and a commercial pulsed fluorescence analyzer over the range of O.f-f ppb. The results for this intercomparison were also quite scattered, although they did show some correlation with each other.

Typical ambient levels. Concentrations of S02 in remote areas are quite low, ~ 10-50 ppt (e.g., Bandy et al., 1993), since the only source is oxidation of biogeni-cally produced organic sulfur compounds such as dimethyl sulfide (see Chapter 8.E). In rural-suburban areas, concentrations of ~l-20 ppb (e.g., Luria et al., 1987; Boatman et al., 1988) are common and in polluted urban areas, levels up to several hundred ppb are observed (e.g., Bennett et al., 1986).

e. NMHC and VOC

As discussed in detail in Chapters 6 and 16, organic compounds play a key role in the formation of ozone, particles, and other species of interest. While some of the individual species are of concern from the point of view of health effects (e.g., the carcinogens benzene and 1,3-butadiene), for most VOC it is because of their central role in the formation of ozone and associated pollutants that their emissions are regulated.

As discussed in Chapter 16.B, methane reacts relatively slowly with OH (&29XK = 6.3 X 10" 15 cm3 molecule ~1 s~'), with a lifetime with respect to this reaction of about 5 years at an OH concentration of 1 X 106 cm-3. It is because of this long lifetime in the troposphere that methane is the only hydrocarbon to survive long enough to reach the stratosphere and participate in chemistry there. However, in terms of generating R02 radicals that convert NO to N02 on time scales of a few hours to days, CH4 is not important compared to larger hydrocarbons (see Chapter 16.B for a discussion of reactivity and data in Table 16.8). This is the case even though its concentrations in air are much higher than those of other organics, 1.8 ppm versus a few ppb or less. Because of this, the regulatory focus has traditionally been on "non-methane hydrocarbons" (NMHC). (See Chapter 14 for a discussion of the role of CH4 in global climate.)

However, it is not only hydrocarbons, but a wide variety of organics, that participate in the chemistry. Indeed, as discussed in Chapter 16, some of these such as the carbonyl compounds can be much more reactive than the parent hydrocarbon. As a result, the term "volatile organic compounds" (VOC) is often used to encompass all of the individual reactive organics in the troposphere, and we shall use this term here. Other terms used synonymously are ROG ("reactive organic gases") and NMOG ("non-methane organic gases").

There are two approaches to measuring VOC: measuring the total without segregating into individual species, or measuring the individual organics that contribute to VOC. The former is commonly used for routine monitoring purposes whereas the latter is clearly of interest from the point of view of understanding the details of atmospheric reactions.

(1) Sampling and measurement techniques for VOC

Total NMOC. Total non-methane organics (NMOC) as a group (i.e., nonspeciated) are commonly measured by cryotrapping the air sample, e.g., in liquid argon, which does not trap CH4. The contents of the trap can then be thermally desorbed directly into a flame ionization detector (FID) (McElroy et al., 1986).

In a second approach, CH4 is separated and measured independently and NMOC obtained by difference. This can be accomplished using gas chromatography to separate the CH4 or, alternatively, oxidizing all of the organics except CH4 to C02 + H20 and measuring the CH4 remaining by FID, which does not respond to C02 and water vapor. [Note, however, that water affects the GC baseline so that care must be taken in integrating peaks on top of this baseline (McElroy et al., 1986).]

For urban data, total NMOC or NMHC is commonly reported in terms of ppm or ppb carbon, ppmC or ppbC, respectively. This is defined as follows:

ppbC = ppb of the organic

X number of carbon atoms per molecule.

For example, 70 ppb of n-butane would be reported as 280 ppbC. (For nonurban data, mixing ratios (ppb) of the individual compounds are often used instead, so the reader is cautioned to determine how such data are presented.)

The relative response of FIDs per carbon atom is quite constant across a spectrum of hydrocarbons typically found in ambient air, with some exceptions such as acetylene (e.g., see Apel et al., 1994, 1995). Hence, calibrations are carried out using a known concentration of a reference hydrocarbon such as propane in air, and the results expressed as ppbC referenced to propane.

Individual VOC. The term VOC is commonly used to describe speciated measurements of individual organ-ics. The almost universal approach to the identification and measurement of individual VOC is GC with either FID or mass spectrometry (MS). GC-MS is used to establish the identity of a particular compound through the combination of retention times and mass spectra and, of course, can also be used for quantification. However, for a given type of air mass, GC-FtD is commonly used for more extensive quantitative measurements after the individual peaks have been identified. For reviews of various aspects of sampling and measurement of VOC in air, see Westberg and Zimmerman (1993), Apel et al. (1994), Klemp et al. (1994), Sacks and Akard (1994), and Dewulf and Van Langen-hove (1997).

Because most organics are present at ppb-ppt concentrations in air, sample preconcentration is required. This is typically accomplished by direct cryotrapping of air, trapping on a solid sorbent, or sampling into an evacuated canister followed by cryotrapping the canister contents prior to injection onto the GC column. Combinations of these, such as cryotrapping followed by transfer to a solid sorbent, have also been used (e.g., Shepson et al., 1987).

Cryotrapping involves pumping air through a cooled loop, which usually contains an inert material such as glass beads to improve the trapping efficiency. Liquid argon (bp - 186°C) or nitrogen (bp - f 96°C) are commonly used and will trap all but CH4, which is too volatile even at these temperatures (vapor pressure is about 9 Torr even in liquid nitrogen, ( —196°C)) to be trapped. The trap is then warmed and the mixture of organics injected onto the head of a GC column. A scheme for carrying this out without the use of valves to minimize the formation of artifacts is described by Stephens (1989).

A common problem in trapping VOC is the presence of significant quantities of water vapor. Trapping this water can plug loops used in direct cryotrapping or in the cryotrapping of samples collected in canisters prior to injection into the GC. In the case of solid sorbents, adsorption of water can reduce the efficiency of trapping of the organics; as might be expected, the amount of water trapped depends on the particular sorbent used, with Tenax, for example, adsorbing relatively small amounts (less than 3 mg of water per gram of adsorbent) and Carboxen 569 adsorbing as much as 400 mg of water per gram of adsorbent (Helmig and Vierling, 1995). Water can also cause shifting baselines and changing retention times, as well as problems with various detectors (McElroy et al., 1986; Helmig and Vierling, 1995). As a result, removing water, or at least decreasing its concentration before it is trapped, is important.

As a result, water traps are usually placed ahead of the cryotrap (e.g., see Goldan et al., 1995). Alternatively, a two-trap system is used in which water and the organics are trapped in the first trap, which is then slowly warmed to desorb organics, but not most of the water, into a second trap (e.g., Greenberg et al., 1996).

A variety of systems have been used to remove water prior to trapping the organics. For example, passing the airstream through a semipermeable membrane that allows water to diffuse through where it is removed by a flow of inert gas is common. However, depending on the preconditioning of the membrane material, some loss and rearrangement of some organics can occur. For example, Gong and Demerjian (1995) reported that Nafion that was preconditioned by heating caused depletion of some C4-Cft alkenes such as isoprene and rearrangements of others. Alternate approaches include the use of multisorbent tubes, purging with a dry gas such as He or N2 to remove some water (e.g., see McClenny et al., 1995), and using a combination of cryotrapping and adsorption (Shepson et al., f987).

Trapping of organics by passing the air through solid sorbents is an alternative to cryotrapping. The organics are then recovered either by thermal desorption or by solvent extraction of the sorbent. Typical sorbents used include carbon-based sorbents such as activated charcoal, porous carbon, Carbosieves, Carbotraps, and Car-bopacks and polymers such as Tenax, Chromosorbs, polyurethane foams (see Chapter 10), and XAD resins. A description of solid sorbents and their application to ambient air measurements are discussed in detail by

Camel and Caude (1995) and Matisova and Skrabakova (1995).

As might be expected, these sorbents can have different efficiencies for the uptake of individual organics, so that results from trapping on one sorbent may not be the same as that on another one. For example, a comparison of Tenax-TA with a combination of Carbo-traps B and C on a test atmosphere containing a number of compounds of atmospheric interest showed good agreement for most compounds; however, ethanethiol was not well retained by the Carbotrap, whereas these carbon sorbents were more effective for chlorodifluoromethane and 1,2-dichlorotetrafluoro-ethane (McCaffrey et al., 1994). A combination of adsorbents in one trap is often used to overcome such problems (e.g., see Helmig and Greenberg, 1994). The sorbents are arranged so that lightest organics pass through to the last layer and are trapped there, so that the most highly absorbing sorbent is the last one exposed to the airstream. The direction of the flow is reversed during thermal desorption (Matisova and Skrabakova, 1995).

Even for compounds that are efficiently trapped by a particular sorbent, care must be taken not to pass so much sample through the adsorbent that it becomes overloaded and hence "breakthrough" of the organic occurs (e.g., see Brown and Purnell (1979) for data on a variety of organics using Tenax-GC). Care must also be taken to investigate and, where possible, avoid artifacts when such sorbents are used. These can arise from desorption of organics such as benzene, either from the sorbent itself or from uptake of organics on it during storage (e.g., see Cao and Hewitt, 1994). To overcome such problems, the sorbents are cleaned prior to use, which is commonly done either by heating under a flow of gas or by extracting the solid with a solvent or combination of solvents followed by storage in a cold, organic-free container (Helmig, 1996).

Artifacts can also arise from reaction of the sorbent with components of the airstream, such as 03 and N02. For example, Clausen and Wolkoff (1997) observed a number of products from the reactions of 03 with Tenax, such as acetophenone, decanal, and benozic acid. Interestingly, 2,6-diphenyl-p-hydroquinone was generated when limonene was also present and was suggested to arise from the reaction with the Tenax of the radicals or the Criegee biradical generated in the ozone-limonene reaction. They also observed reactions of N02 with the Tenax sorbent.

Reaction of adsorbed organics with other components of the airstream during sampling, particularly ozone, is well known, leading to artificially low levels of the organic and, in some cases, the appearance of new species that are products of these artifact reactions (see

Helmig (1997) for a review of these interferences). This is particularly the case for reactive compounds such as the unsaturated biogenics. It should be noted that such interferences due to ozone also occur with cryotrapping (Helmig, 1997). In this case, reactions between ozone and the organics can occur as the trap is warmed to desorb the organics onto the GC column. For example, Goldan et al. (1995) examined the effects of 100 ppb 03 on a synthetic mixture of n-pentane, isoprene, 2,2-dimethylbutane, benzene, ra-xylene, and a-pinene. Figure 11.37 shows the chromatograms of this mixture in the absence of 03, in the presence of 100 ppb 03 with no ozone trap, and finally, with a Na2S03 scrubber for ozone upstream of the cryotrap. The presence of 100 ppb 03 essentially completely removes a-pinene and most of the isoprene, and new peaks corresponding to oxidation products (methacrolein and methyl vinyl ketone) appear. However, this is not a problem if an ozone trap is used (Fig. ff.37c). Similar results have been reported for a series of terpenes, where 03 at 8 ppb had no effect but at 61-125 ppb reduced the measured terpene concentrations and decreased the precision of the measurements (Larsen et al., 1997).

C with 100 ppb ozone and o3 trap

FIGURE 11.37 Chromatograms of a synthetic mixture of n-pen-tane, isoprene, 2,2-dimethylbutene, benzene, m-xylene, and a-pinene in synthetic air which are cryotrapped (a) in the absence of 03, (b) in the presence of 100 ppb 03, and (c) same as (b) but with an ozone scrubber upstream (from Goldan et al., 1995).

The severity of these problems is quite sensitive to the nature of the particular organic, as expected, and likely to the particular sampling configuration and conditions as well. For example, Calogirou et al. (1997) showed that saturated oxygenated terpenes adsorbed on Tenax were not affected by the presence of ~ 100 ppb 03 whereas as much as 80-90% of the most reactive, unsaturated compounds reacted. Indeed, no a-terpinene was observed when ozone was present. On the other hand, Koppmann et al. (1995) report no significant interference problems with 03 for the C2-C4 hydrocarbons which were sampled using a heated stainless steel inlet line, which destroys as much as half of the initial ozone, and then cryotrapped.

Because of these problems, removal of ozone prior to trapping the organics is highly desirable. A variety of approaches have been used, including the use of heated inlet lines as just described (Koppmann et al., 1995), annular denuders coated with inorganics that react rapidly with 03 such as KI (e.g., Williams and Grosjean, 1990), scrubbers containing 03-reactive substances such as KI, crystalline Na2S03, Na2C03, or Mn02, or polymeric sulfur scrubbers such as polyphenylene sulfide in which 03 oxidizes the -S-group to =S(0)- without the formation of gas-phase products that could interfere with the analysis (e.g., Calogirou et al., 1997). Titration of the 03 with NO (e.g., Sirju and Shepson, 1995) has also been used. These and other techniques are reviewed in detail by Helmig (1997).

Collection of air samples in stainless steel canisters whose surfaces have been passivated is another common collection technique for VOCs. (Aluminum has also been used but the stability of polar organics in them is poor; Gholson et al., 1990.) Indeed, this method is used not only for sampling air but in medical applications as well, where they have been used to sample organics in a single breath (Pleil and Lindstrom, 1995). Passivation of the canisters is often carried out using a process called SUMMA and hence referred to as "SUMMA canisters." The canisters also have to be thoroughly cleaned before use; an example of one such procedure is described by Blake et al. (1994). The sample is then typically preconcentrated by transfer to a cold trap prior to injection onto the GC column (e.g., see Blake et al., 1994).

Loss of some organics to the walls of the canisters can occur, and these can subsequently desorb. Such negative changes depend on the nature, past use, and pretreatment of the canister surface, the nature of the compound, the canister pressure, the storage temperature, and interestingly, the water vapor present. The loss of organics to the walls is generally less in the presence of water, which has been attributed to water occupying active sites on the metal surface that would otherwise adsorb organics (e.g., Pate et al., 1992; Apel et al., 1994). The nature of the compound is particularly important, as might be expected, since highly reactive organics can react with other species such as ozone in the canister itself (Apel et al., 1994). In addition, characteristics of the compound such as its vapor pressure, solubility in water, and polarity determine how readily it is taken up onto the canister surfaces or into a thin film of water on the surface. In fact, positive changes in some VOCs have been observed in canisters and attributed to uptake into a film of water on the canister surface. As samples are removed from the canister and the pressure reduced, the amount of surface water decreases, causing a release of dissolved organic into the gas phase. This appears then as a positive change with time.

Zielinska et al. (1996) and Kelly and Holdren (1995) have summarized the stability in canisters of organics, some of which are U.S. EPA designated HAPs (hazardous air pollutants). Kelly and Holdren propose that for compounds whose stability in canisters is not known, estimates can be made based on species of similar physical and chemical characteristics. These characteristics include their vapor pressure, polarizabil-ity, water solubility, Henry's law coefficient in water, and estimated lifetimes with respect to reactions in air and in the aqueous phase.

While these methods are most commonly used for sampling and analysis of ambient air, others have been applied as well, particularly in laboratory studies or source sampling. For example, flexible chambers made from thin films of Teflon or Tedlar are often used to store organics prior to sampling. The advantage is that as sample is removed, they collapse so that their pressure remains at atmospheric while the volume decreases. However, again some losses of organics can occur in these chambers, either on the walls or on the attached sampling hardware (e.g., see Wang et al., 1996).

Passive samplers are used for specific applications such as for indoor air environments or as passive dosimeters. In this approach, the air containing the organic diffuses to and adsorbs on a solid sorbent without active pumping. The organics are subsequently thermally desorbed or extracted from the sorbent using a solvent (e.g., see Shields and Weschler, 1987).

(2) Intercomparison studies Because of the importance of accurate measurement of individual organics in air, there have been several intercomparison studies carried out in which a number of different laboratories have analyzed common samples, either synthetic or ambient air. In one such study (Apel et al., 1994), a

0 0

Post a comment