MEASUREMENT
OF PARTICULATE AND REACTIVE GASEOUS MERCURY, RGM, IN THE AMBIENT AIR: METHODS
DEVELOPMENT.
Mary
M. Lynam* J. Timothy Dvonch and Gerald J. Keeler
University
of Michigan Air Quality Laboratory, Ann Arbor, MI 48109;
Matthew
S. Landis, US EPA National Exposure Research Laboratory, Research Triangle
Park,NC 27711; Robert K. Stevens, Florida Department of Environmental
Protection at USEPA, Research Triangle Park, NC 27711
This contribution describes our efforts in the development and optimization of methods for accurate and reliable measurement of particulate and reactive gaseous mercury in the ambient air. Particulate levels measured in during summer of 1999 in Ann Arbor, MI were in the range of 1-40 pg/m3 while RGM concentrations in the Florida Everglades during winter of 1999 ranged from 3-54 pg/m3. A new thermoreductive method was developed for the analysis of particulate mercury. Results of fine particulate samples (PM2.5) show good agreement between the new thermal method and a conventional acid digestion method. However, the agreement between the two methods for total suspended particulate (TSP) was not comparable presumably because higher temperatures are required. Sampling of RGM using manually operated annular denuders was in excellent agreement with an automated unit and will permit simple, mobile and inexpensive monitoring of RGM to be conducted in a variety of environs.
Mercury is an inessential nutrient and is unique among the heavy metals because of its significant contribution from gas phase species as it cycles through the environment. Mercury cycles through the environmental compartments as a consequence of natural and anthropogenic activities. Atmospheric mercury species of environmental concern exist largely as the gaseous Hg0 form (95-99%), with the remainder being attributed to HgII which is present in ultra-trace amounts (pg/m3). Atmospheric HgII compounds may be associated with particles or occur as gases e.g. HgCl2, or HgO. Gaseous HgII species, termed Reactive Gaseous Mercury, RGM, are 105 more soluble than Hg0 a fact which strongly influences the extent of removal from the atmosphere and subsequent deposition to the biosphere (Lindberg and Stratton,1998). Once present in the biosphere, mercury can be methylated by microorganisms and bioaccumulate in the food chain. Although present in trace amounts, HgII compounds may control overall deposition of mercury thereby making it imperative to reliably and precisely determine their concentrations in the environment. The need for information on mercury speciation is particularly pressing so that its fate and transport in the environment can be adequately assessed. The University of Michigan Air Quality Laboratory (UMAQL) is involved in numerous ongoing studies to contribute to further knowledge regarding atmospheric mercury speciation. Owing to the fact that particulate and gaseous forms are present in ultra-trace amounts in the atmosphere their accurate collection and anlysis is a continuing challenge. The present contribution describes futher efforts in developing and optimizing methods for atmospheric measurement of particulate and reactive gaseous mercury.
Sampling for
particulate mercury was carried out from June-October 1999 at the University of
Michigan, Ann Arbor on the rooftop of
the Space Research Building at an elevation of approximately 12 meters. Ultra–clean sampling and analysis techniques
were used (Keeler et al, 1995). Sampling
equipment comprising Teflon filter packs, forceps and petri dishes were
acid-cleaned prior to sampling. Filters
(glass fiber and quartz) were baked at 500°C for 1 hour prior to sampling. Particle-free gloves were used during the
sampling procedures.
Total suspended
particulate (TSP) mercury was collected using open-faced filter packs onto 47
mm quartz filters (Whatman) for 24 hours or greater at a flow rate of 30 L min-1. Fine particulate mercury (<2.5 mm)
was collected (collocated with TSP) onto 47 mm glass fiber filters using Teflon
coated aluminum cyclones (URG, Carboro, NC) to remove larger particles upstream
of the filter. After sampling, the
filters were placed into acid-cleaned petri dishes, the dishes were sealed with
teflon tape and stored at –40°C until analysis. Flow rates were measured with calibrated rotameters, and sample
volumes were determined using in-line calibrated dry test meters.
Two
methods of analysis were used; a thermoreductive method in which the filter was
placed in a quartz pyrolyser and subsequently heated under nitrogen to 800°C in
a tube furnace. The furnace was
connected to a mercury analyzer unit and the mercury liberated during the
heating process was analyzed by means
of Cold Vapor Atomic Fluorescence Spectroscopy, (CVAFS), using a Tekran 2735
unit. The second method of analysis for
the filters consisted of acid digestion in a microwave followed by detection
using CVAFS. The acid digestion
involved extraction of each filter with 20 mLs of a 10% dilution of
concentrated HNO3 (1.6M) followed by digestion of the filter in a
teflon vessel for 20 minutes at 160°C (70 psi) using a CEM MDS-200 computer
controlled microwave unit, as described by Keeler et al (1995). The mercury forms in the acid solution were
oxidized by BrCl and left overnight.
They were subsequently reduced with SnCl2 and purged out of solution
and collected onto a gold trap which was analyzed using CVAFS. A calibration curve was generated by spiking
vessels containing blank filters with known amounts of a working standard. The method detection limit is 1 pg/m3. Field and storage blanks were collected
with the samples . They were collected
by loading acid-cleaned filter packs with a glass fiber or quartz filter and
placing the filter packs in the sampling box for two minutes without drawing
air through the system. Samples were
collected as part of an ongoing effort
to improve existing analysis methods as well as develop new methods of analysis
for particulate mercury. In particular
an overarching aim of this study was the development of a reliable and precise
thermal method of analysis for particulate mercury. The conventional acid digestion analysis method is very time
consuming whereas pyrolysis is much more rapid and has a lower risk of
contamination because no reagents are added (Lu and Schroeder, 1999).
Sampling and Analysis of Reactive Gaseous Mercury (RGM)
Sampling
and analysis of RGM was carried out during the Florida Everglades Dry Deposition
Study, FEDDS, an intensive field campaign in February-March 1999. RGM was collected by means of quartz annular
denuders, and both manual and automated methods were used. In general, denuders sample reactive polar
gases which diffuse along a surface and are trapped by an adsorbent. In the case of mercury sampling, the sorbent
chosen must sorb only the reactive form, i.e. HgII and not elemental
mercury. The sorbent of choice is KCl
as it has a high deliquescence point (Sommar et al, 1999). Annular denuders are found to be more
suitable for mercury monitoring efforts since the enhanced surface area can
facilitate larger sampling times thereby permitting the collection of
sufficient amounts of a sample (Slanina, de Wild and Wyers ,1992). The denuder coating process was as
follows; denuders were rinsed with
milliQ water, methanol and milliQ water and shaken to dispense any water
droplets. A saturated KCl solution was
made using KCl previously baked at 500°C to remove any traces of mercury. The solution was drawn up and down the
interior of a denuder four times in order to coat all available inner surfaces.
The denuder was removed from the solution and dried by flowing mercury free
nitrogen gently through it. The ends
of the denuder should not contain any KCl and should be dipped in milliQ water
and dried . The denuder was then heated
at 550°C for 1 hour in a stream of mercury-free nitrogen in order to remove any
traces of mercury associated with the coating procedure. The ends of the denuder were capped and
coated with teflon tape until commencement of sampling.
Manual
denuders were collocated vertically in a sampling box. A glass fine particle (< 2.5 mm) inlet which was coupled to the quartz
denuder. Air was pulled through at a
flow rate of 10 Lpm at sample durations of two hours. Some longer sampling times were also used. If overnight sampling was carried, out
heating tape was wrapped around each denuder to prevent water vapor from
dissolving the KCl coating. After
sampling was completed, the denuder was removed from the sampling box for
analysis. Analysis was achieved by
heating the denuder to 500°C for 15 minutes.
The desorbed RGM is carried in a mercury-free Argon stream into a CVAFS
analyzer for detection.
Automated sampling was carried out using a prototype 1130 Tekran speciation module coupled to a Tekran CVAFS unit. Ambient air was sampled at five minute intervals at a flow rate of 10 Lpm through the 1130 module, the RGM fraction is trapped on the KCl coating while elemental mercury passes through the denuder and is subsequently analyzed using CVAFS. After 24 five-minute sampling intervals (total two hour RGM sample integration), the denuder was heated for three five-minute periods to ensure complete desorption.
PARTICULATE PHASE MERCURY
Table 1 shows that mercury is
associated with both the coarse and fine particulate material sampled and is
detected by both the thermal and acid digestion methods. The range of values found in this study is
similar to that seen in other studies, 1-100 pg/m3 in the Great
Lakes Region with annual average
particle-phase mercury concentration of 21.9 pg/ m3 in Ann Arbor, MI
(Keeler et al, 1995 ) and 21 pg/m3 in Dexter, MI (Burke, 1998). It should be noted that the average values
for particulate mercury in this present study are less than those observed in
previous studies. Particulate mercury
levels in Michigan were found to display seasonal behavior at rural sites and
tend to be higher during the winter and
early spring months (Keeler et al, 1995).
Since our sampling campaign
was in the summer and fall the measured levels would be lower resulting in a
lower average and are therefore consistent with results of previous studies. In Keeler et al (1995), particulate mercury
measurements for June through October were in the range 8-24 pg/m3. Examination of the results in Table 1 for
TSP show significant differences in the mean values for each method.
Table 1. Particle Phase Hg measurements in Ann Arbor, MI (pg/m3)
Type Analysis Method N Median Mean Min Max Std Dev
TSP Acid 20 15 15 5 37 8
TSP Thermal 18 6 7 1 23 5
PM2.5 Acid 18 9 10 3 18 4
PM2.5 Thermal 22 7 9 2 40 8
This
apparent inability of the thermal method to detect higher mercury values may be
due to the presence of a crustal mercury component which may require
temperatures greater than 800°C to be liberated. Results of the two analysis methods for fine particulate matter
are in better agreement and are a reflection of the fact that PM2.5
is primarily of anthropogenic origin and due to combustion processes and may
exhibit greater particle homogeneity.
Based on these results thermal analysis shows promise for use in
analyzing PM2.5 samples.
Further method refinements such as the use of higher temperatures would
be required to get more precise measurements for TSP.
REACTIVE
GASEOUS MERCURY
Table
2 shows RGM measurements using manual and automated denuder-based sampling
techniques and shows the two methods to be in excellent agreement. Reactive gaseous mercury concentrations in the Everglades were found
to exhibit a diel cycle which is in agreement with observations by other
researchers. RGM concentrations tend to
be lower at night and reach a peak in the afternoon to early evening
hours. A possible cause of nighttime
decreases in RGM due to dissolution in dew is plausible and is a documented
mechanism for it removal (Pleijel and Munthe, 1995).
Table 2. RGM Concentrations In The
Florida Everglades
Type N Mean
(pg/m3) Min (pg/m3) Max (pg/m3) Std Dev
____________________________________________________________________________________________________________________________________________________________
Manual 45 15.4 2.5 54.3 12
Automated 44 14.7 2.5 53.6 11
We graciously
acknowledge the field efforts of Frank Marsik, Jim Barres, and Elizabeth
Malcolm. This work was partially funded
by both the USEPA and the Florida DEP.
1. Lindberg S, Stratton W. J., (1998),
Environ. Sci. Technol. 32: 49-57
2. Sommar, J, Xinbin F, Gårdfeldt K,
Lindqvist O, (1999), J. Environ. Monit. 1: 435-439.
3. Slanina J, de Wild P J, Wyers G P,
(1992), in: Gaseous Pollutants: Characterization and Cycling. (JO Nriagu, Editor), New York, Wiley & Sons pp
129-154.
4. Lu J, Shroeder WH, (1999), Talanta, 49:
15-24.
5. Keeler J, Glinsorn G, Pirrone N, (1995),
Water, Air and Soil Pollution, 80: 159-168.
6. Burke J, (1998), Doctoral dissertation,
The University of Michigan.
7. Pleijel K, Munthe J, Atmospheric
Environment. (1995) 29: 1441-1457.