Environ. Sci. Technol. 2001, 35, 170-177
Atmospheric Mercury Speciation: Laboratory and Field Evaluation of a Mist Chamber Method for Measuring Reactive Gaseous Mercury WILMER J. STRATTON* Earlham College, Richmond, Indiana 47374 STEVEN E. LINDBERG Environmental Sciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831 CHRISTOPHER J. PERRY University of Aberdeen, Aberdeen, Scotland AB9 2UE
Knowledge of atmospheric mercury speciation is critical to modeling its fate. Thus there is a crucial need for reliable methods to measure the fraction of gaseous atmospheric Hg which is in the oxidized Hg(II) form (termed reactive gaseous mercury, RGM). We have developed a novel method for measurement of RGM using a refluxing mist chamber, and we recently reported the results of sampling campaigns for RGM in Tennessee and Indiana. In general, measured RGM levels were about 3% of total gaseous mercury (TGM), and our results support prevailing hypotheses about the nature and behavior of RGM in ambient air. Because its use for RGM is growing, we now report in more detail the development and testing of the mist chamber method. Several styles of mist chambers have been investigated. The most versatile design employs a single nebulizer nozzle and can operate at flows of 15-20 L/min. The water-soluble Hg is collected in ca. 20 mL of absorbing solution, which is then analyzed for Hg(II) by SnCl2 reduction and CVAFS. One-hour samples (ca. 1 m3 of air) generally contain 50-200 pg of RGM. The method detection limit for 1-h samples is approximately 6-10 pg/m3. Thus short sample times can reveal temporal variations in RGM that would not otherwise be observable. The efficiency of collecting RGM in mist chambers is highly dependent on Clconcentration in the absorbing solution, in keeping with equilibrium calculations. Artifact formation of Hg(II) by oxidation of Hg0 under ozone ambient conditions appears to be sufficiently slow so as to be negligible for the short (ca. 1 h) runs that are typically employed. We observed no significant error from cosampled particles or aerosols in rural nonimpacted air samples. We have developed a simple approach to analyzing mist chamber samples in the field using an automated Hg sampler.
Introduction One of the major unanswered questions in the biogeochemical cycling of mercury is the extent to which gaseous Hg(II) * Corresponding author phone: (540)745-6193; e-mail: wilmers@ earlham.edu. 170
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001
species exist in ambient air (1). Such species have been postulated to exist but only as a small fraction of the total atmospheric mercury. Nevertheless, this small fraction is important to the geochemical cycling of mercury (2). Modeling results to date have indicated that modeled fluxes are highly sensitive to the assumed fraction of total mercury that is Hg(II) (3, 4). Thus, measurement of gaseous Hg(II) is crucial to an understanding of the atmospheric transport and deposition of mercury. This gaseous Hg(II) fraction, termed reactive gaseous mercury (RGM), is generally assumed to consist primarily of HgCl2 but might also include HgO, Hg(OH)2, and/or other mercury halides. The assumption that RGM is largely HgCl2 is now open to question in light of recent studies at the University of North Dakota Energy and Environmental Research Center where investigators have positively identified Hg(NO3)2‚H2O as the primary Hg(II) species in certain simulated flue gases (5). These investigators conclude that, because of the volatility and unusual stability of Hg(NO3)2‚H2O, it may be the major RGM species in coal combustion flue gas, especially in plants using low-chlorine coals. Reliable mesurement of RGM in ambient air is difficult for two reasons: the concentrations are extremely low and RGM is a very reactive substance. In recent papers (2, 6), we described a novel method for measuring RGM, using a highflow refluxing mist chamber (MC). This technique, combined with dual gold trap atomic fluoresence analysis, is sufficiently sensitive to permit ca. 1-h samples of ambient air. Our results showed RGM to be 2-4% of the total gaseous mercury (TGM). Extensive field sampling for RGM using the MC method revealed that RGM is correlated with air temperature, solar radiation, O3, SO2, and TGM; shows regional trends related to emission density; exhibits a strong diel cycle with a daytime maximum; is efficiently removed from the air by precipitation; and is rapidly dry-deposited to vegetation (2, 7). This paper describes in more detail the MC technique for measuring RGM, along with our attempts to validate the method, including tests of mist chamber design, solution composition, collection efficiency, and artifact formation of RGM. It must be emphasized that the method development and our understanding of atmospheric processes for Hg are necessarily intertwined and have evolved simultaneously.
Materials and Methods Laboratory and Field Facilities. Most of the studies reported here utilized facilities at Oak Ridge National Laboratory (ORNL) in eastern Tennessee and at Earlham College (EC) in Richmond, IN. One study was done at Frontier Geosciences in Seattle. Each laboratory was equipped with class 100 clean air benches and Brooks Rand CVAFS mercury analyzers. Field sites in both locations have been described (2, 6). Mist Chambers. These have been described (6). Their use in atmospheric studies of various water-soluble gases has been described by Talbot and others (8). They consist of a custom-fabricated Pyrex glass body with an all-Teflon conical unit attached to the downstream side of the glass body (Figure 1). The mist chambers are partially filled with a small volume (10-25 mL) of an appropriate RGM-absorbing solution (see later discussion). Air (15-20 L/min) is drawn in through the bottom. One or more nebulizer nozzles (orifice ca. 1 mm) produce a fine mist of the absorbing solution, with aerosol-sized droplets. In this fashion, water-soluble species are extracted from the air stream. A hydrophobic Teflon membrane filter mounted in the Teflon top retains the water droplets so that the solution drains back and is continuously refluxed. The range of droplet sizes produced by the nebulizer 10.1021/es001260j CCC: $20.00
2001 American Chemical Society Published on Web 11/16/2000
FIGURE 1. Diagram of the mist chamber apparatus (not to scale). is affected by even small changes in mist chamber design and fabrication. Some droplets are easily visible, do not travel far, and are assumed to be of minor importance in the airliquid exchange. The aerosol mist, as measured in two of our mist chamber designs, contains droplets predominantly in the range of 3-7 µm diameter (9). (It should be noted that other investigators have used mist chambers with different operating parameters, e.g. up to 90 SLPM and with larger volumes of absorbent solution.) Several modifications of the basic MC design have been utilized in our studies to date. (a) Most of our early studies used mist chambers of the type we reported previously (6). This design has four nebulizer nozzles in a glass envelope of ca. 35 mm diameter. The combined air volume of mist chamber plus Teflon top is 75 mL. The working range of the absorbing solution volume is 12-25 mL. The combined critical orifice of the four nebulizers permitted air flows up to ca. 25 L/min. (b) Much of the work reported here used a single-nozzle design developed by Keene and Maben at the University of Virginia (10) and later modified with addition of an impact bead as a flow spoiler to provide more efficient mist formation (11). The glass chamber has a working volume of ca. 30 mL and a maximum flow rate of ca. 20 L/min. (c) Within the last year we have evaluated and deployed in the field a further modification of the Keene/Maben design that was developed and fabricated by the University Research Glass Corporation (URG, 118 E. Main St., Carrboro, NC 27510). This model uses standard air-sampling components for the Teflon upper portion, including an ultrasmooth Teflon cone that gives low sampling blanks and minimal memory effect plus a stainless steel membrane support with a proprietary coating to permit use of unsupported Teflo membranes. All three mist chamber designs have a side port for introduction and removal of the absorbent solution, and all three utilize a Teflon top assembly containing a 47 mm Zefluor or Teflo membrane, 2 µm pore size. As discussed in a later section, most of our studies were conducted without a filter on the MC inlet. This contrasts with the MC sampling configurations used by other investigators, nearly all of which incorporate an upstream filter to remove aerosols. However, we feel that prefilters may compromise the RGM sample through artifacts related to reactions on the filter surface (discussed further in a later section).
General Design of Sampling Equipment. The equipment configuration is shown schematically in Figure 1. Mist chambers were usually clamped to a support structure with the inlet ca. 1.5 m above ground. On some occasions the MCs were shielded from light by wrapping in aluminum foil, but tests showed no significant difference in RGM with or without light exposure. The MC outlets were connected to 1 cfm pumps via inline mass flow meters (Sierra Instruments). “Drierite” cannisters (6 cm dia. × 20 cm) filled with 4-6 mesh soda lime were placed inline between the MCs and the mass flow meters to protect the latter from acid vapors. Mass flow meters were calibrated perioidically against a Gilibrator bubble meter, corrected to STP. Mist chambers were filled and emptied by means of 20 mL all-polypropylene Luer-Lok syringes (used without precleaning but replaced daily) attached, via either a Teflon Minnert valve or a small polyethylene stopcock, to a 1/16 in. Teflon tube connected to the bottom of the chamber. Slow evaporation of water (2-6 mL/h, depending on the ambient temperature, relative humidity, flow rate, and vacuum produced by the pump) required periodic replenishment with high-purity deionized water which added as needed to maintain the desired volume. Blanks are discussed below. Absorbing Solutions. Mist chamber solutions of varying HCl and NaCl concentration were prepared daily in glass or Teflon bottles, using Hg-free deionized water (Barnsted Nanopure or Millipore MilliQ). Hydrochloric acid was Fisher trace-metal grade or EM Suprapur, tested to confirm low Hg content. Sodium chloride was Fisher reagent grade, baked at 500 °C prior to use. Sample Containers. These were either (a) 40 mL glass vials (acid washed with 1% BrCl, rinsed well, and baked at 500 °C) along with acid-cleaned Teflon-lined caps or (b) disposable polypropylene centrifuge tubes with caps, Falcon #2070, used as received in sealed packages. Container blanks were negligle for both types of containers. Sampling Protocol. Prior to the start of each set of measurements, the mist chambers were rinsed repeatedly in the field location using short runs with the working solution. The mist chambers were then filled with 15-20 mL of the working solution, with the syringes left in place. Sample times were most commonly ca. 1 h in duration, with an air flow of 15-20 SLPM, representing about 1 m3 of air sampled. During special studies, some samples were 2 h in duration, and a few samples (e.g. the nighttime portion of diel studies) were as long as 5 h in duration (∼5 m3 of air). At the end of the sampling time, the solution was withdrawn into the syringe and transferred to a sample container. To rinse the mist chamber, the same syringe was used to add deionized water (typically 8-10 mL). After a 30-s aspiration, the rinse solution was withdrawn and added to the sample container. Samples were stored capped and out of direct light. One or two additional rinses with DI water or the working solution were discarded before the next sample or blank was collected. Field Sampling Blanks. A blank correction was used for all samples. Ideally, this should be a “dynamic blank” in which zero-RGM air is sampled for the same time intervals as the samples. Unfortunately, we have not yet been able to devise any suitable method to produce zero-RGM air at a flow of 15-20 L/min. In the absence of true dynamic blanks, we resorted to a procedure in which field blanks were obtained by filling the mist chamber with the absorbing solution and aspirating for ca. 30 s. In effect this was a “rinse” of the MC. The solution was then collected, stored, transported, and analyzed in the same fashion as the samples. The values for the blanks obtained in this fashion were a combination of (1) the solution blank, (2) the container, syringe, and trip blanks, (3) the combined analytical blank for the bubbler and CVAFS system, and (4) any residual Hg in the mist chamber itself. The numerical values for these VOL. 35, NO. 1, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
171
combined blanks varied from day to day and were affected by such things as the composition of the absorbing solution, the purity of the reagents, the condition of the mist chambers, the holding time and transport of samples, and the laboratory analytical blank. Although the magnitude of these combined blanks varied, we can report the precision of the blanks within individual data sets. For 46 1-day data sets that each included four or more values for combined blanks (including over 300 individual values), the mean of the standard deviations (smean) within the individual sets was 5 pg (range 3-8 pg). Test for Possible Loss of HCl by Volatilization during Sampling. To test whether the MC solution composition changes because of HCl volatilization during sampling, two l-h MC samples were collected, and the solutions were analyzed for both H+ and Cl-. The results showed a negligible loss of HCl (less than 4%) after 1 h of air sampling (ca. 1 m3 of air) for both low (0.042 M) and high (0.25 M) HCl concentrations. Analytical Method. In general, samples and blanks were analyzed within 6 h of collection. Hg was determined by cold vapor atomic fluorescence (CVAFS) and dual gold-sand trap amalgamation (12, 13). MC solutions and blanks were measured for easily reducible Hg by reduction with SnCl2 and purging to gold traps. Quantitation was by peak area using a chromatographic integrator. Analyses were carried out at our two laboratories in Tennessee and Indiana, with satisfactory intercalibration results between them (agreement within (5%). Calibration was done either with aqueous standards (Indiana laboratory) or with saturated Hg vapor (Tennessee laboratory) (14). The analytical limit of detection (bubbler + CVAFS) was ca. 4 pg Hg (3 × SD of bubbler blanks). The limit of detection for RGM samples is discussed in the Results section. Recently, we have tested an alternate analytical procedure for the MC samples utilizing a Tekran model 2537A automated mercury analyzer (15). The Tekran analyzer, which utilizes gold trap amalgamation and cold vapor atomic fluorescence, is now widely used for atmospheric Hg studies. Its operation has been described in detail elsewhere (16). Use of this instrument for RGM analysis is particularly attractive for field studies in locations without direct access to a laboratory. It also allows us to monitor possible contamination of MCs or solutions during field studies. In this procedure, a bubbler is connected (via a 15 cm × 9 mm ID soda lime trap) to the Tekran air inlet; the air flow is adjusted to 0.8-1.0 L/min; and samples are purged directly into the Tekran analyzer. Ambient air entering the bubbler is made Hg-free with a small iodated charcoal trap. A four-way stopcock on the bubbler allows Hg-free air to bypass the bubbler or to purge the solution. When an acceptable baseline has been achieved, the stopcock is turned to bypass; sample and SnCl2 are added to the bubbler; the stopcock is turned to purge at the start of a Tekran cycle; and the bubbler is purged for at least two 5-min Tekran cycles. With this configuration, two cycles were adequate for most RGM samples. A similar Tekran purging method for dissolved Hg0 in aquatic samples has been described in detail (16).
Results To validate the mist chamber technique for measuring RGM in ambient air, we have subjected the equipment and procedures to a variety of tests. The nature of these tests, the questions addressed, and our sampling protocols have evolved over time. In this section we first comment on the general operation and behavior of mist chambers for RGM and then describe the various tests. Tests of Mist Chamber Design. Of the two basic mist chamber designs (four-nozzle and one-nozzle), we found that both types worked well and gave similar results for RGM. In a comparison study using side-by-side 1-h field samples 172
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001
with a four-nozzle MC and a one-nozzle MC, the average ratio of RGM values was 1.00 ( 0.10 (n ) 9, RGM range 50-150 pg/m3). We conclude that there is no significant difference in the sampling efficiency for the two types. The single-nozzle design has the advantage of being simpler and less expensive to fabricate, but one modest disadvantage to single-nozzle MCs is their higher resistance to flow, which means both lower flow rates and faster solution evaporation. Limit of Detection. Using an average standard deviation for the field sampling blanks of 5 pg (see Materials and Methods section), we can estimate an LOD for MC solutions at 95% confidence of ∼10 pg (LOD ) 1.96 × SD of blanks). For typical 1-h samples (ca. 1 m3 of air) this gives an estimated LOD for RGM of 10 pg/m3. Alternatively, using the method of Keene et al. (10), based on paired measurements for lowRGM samples, we estimate an LOD at 95% confidence of 6 pg/m3. Keene et al. acknowledge that their approach is not a rigorous statistical formulation, but they conclude that it does provide a useful estimate of the LOD. For RGM studies, it has the further advantage of being based on data for actual sampling times of ca. 1 h rather than on the short (
