1.1. Discovery of GOM
Mercury (Hg) exists in the atmosphere as three forms: gaseous elemental (GEM), gaseous oxidized (GOM), and particulate-bound (PBM). Often, GOM and PBM concentrations are combined and collectively described as reactive Hg (RM). In the beginning, the atmospheric Hg research community focused on development of methods for GEM and did not know GOM existed. Now, GOM is known to be emitted from anthropogenic point sources and formed by atmospheric oxidation reactions of GEM with ozone (O3
), hydroxyl radical (OH·
), nitrate (NO3
), hydrogen peroxide (H2
), and/or halogen-containing compounds (Cl.
, ClO, BrO, ClBr) [1
]. A more recent paper by Saiz-Lopez et al. [3
] provides an update on current thinking regarding our understanding with respect to reactions and points out, using a global model based on bromine-induced GEM oxidation that other oxidation mechanisms are needed in the troposphere to explain observations.
In 1979, Fogg and Fitzgerald [4
] postulated that since GEM is not highly water soluble, concentrations measured in precipitation could not be explained by GEM alone. Kothny (1973) [5
] suggested Hg adsorbed to aerosols was the Hg form present in precipitation. Brosset (1983) [6
], based on equilibrium coefficients developed by Iverfeldt, noted that HgCl2
HgCl could explain observed concentrations. A mechanism for oxidation was proposed by Iverfeldt and Lindqvist [7
] that entailed oxidation of GEM in water by ozone.
At the time, atmospheric Hg measurements were made using gold traps with a glass wool filter upstream to capture the particulate component. However, there were inconsistent results with the particulate filter. Research then focused on collecting GEM using gold surfaces such as gold-coated denuders [8
] and gold-coated sand traps [10
]. Currently, gold-coated sand traps are the standard method for measurement of GEM; there is still controversy as to whether this is a measurement of GEM or total gaseous Hg (TGM). The Global Mercury Observation System standard operating procedure states that a soda lime trap in front of the Tekran 2537 removes GOM, though this has not been adequately tested.
In 1996, in a critical review paper on Hg speciation in flue gases associated with coal combustion, Galbreath and Zygarlicke [12
] pointed out that a variety of RM compounds should exist, including Cl-, O-, and S-based compounds. They also reported Hg(II) (oxidized) forms did exist in the flue gas, based on measurements using USA Environmental Protection Agency (EPA) Method 29, EPA Method 101A, and the modified Method 101A and laboratory tests. Lindberg et al. [13
] suggested that if such oxidized forms of gaseous Hg persisted in ambient air, they had the potential to be significant contributors to Hg deposition.
1.2. Early Development of Methods
In 1995, a landmark paper was published that described the use of a mist chamber method for measuring RM and provided the first measurements of RM in ambient air [15
]. A similar type of method had been attempted earlier by Brosset and Lord [16
] using bubblers and long sampling times. Brosset and Lord [16
] concluded that measured GOM was an artifact and better approaches were needed. The mist chamber used a single nebulizer nozzle, operated at a flow rate of 15 to 20 L min−1
, and collected samples in 20 mL of solution [15
]. Stratton and Lindberg [15
] reported that one-hour samples contained 50 to 200 pg RM. The mist chamber was deployed at two locations, Tennessee and Indiana, and concentrations of 50 to 150 pg m−3
were reported; similar trends were observed under field conditions at the two sites, leading to the conclusion that the method provided reasonable results [15
]. The main concerns with this method were artifacts associated with O3
and the presence of aerosols, which were extensively tested [15
]. Artifact formation was considered sufficiently slow relative to sampling times. Data collected using the mist chamber method was significantly correlated with temperature, solar radiation, O3
, and total gaseous Hg [18
]. Additional work using the sampling system further demonstrated the utility of the method and the limited effect of artifacts on the measurements [18
]. Two known drawbacks of the system were that it was not calibrated, and potential for artifacts could vary by sampling location.
At this same time, researchers were also testing the use of membranes for both PBM and GOM capture. Ebinghaus et al. [19
] applied Teflon disc filters, Whatman quartz filters, and quartz wool plugs, or Au traps preceded by Au denuders for PBM, and ion exchange membranes for GOM measurements. PBM measured by the different methods ranged from 5 to 100 pg m−3
with the highest concentrations observed on gold traps after a denuder. Ion exchange membranes measured concentrations of 40 to 95 pg m−3
, higher than denuder methods by 10 to 20 pg m−3
that were determined after liquid extraction. Munthe et al. [20
] explored the use of microquartz fiber filters, cellular acetate, glass fiber, and Teflon filters for measurement of PBM; these results were quite variable.
Denuder methods for measuring GOM in ambient air were first pioneered by Oliver Lindqvist and his collaborators (e.g., Xiao et al. [8
]; Feng et al. [21
]). Their method utilized a KCl-coated tubular denuder, with GOM quantified using a liquid extraction. Comparison of the tubular and annular denuders showed similar recoveries in two studies in which the tubular denuder was liquid extracted and the annular denuder desorbed (Munthe et al. [20
]; Nacht et al. [22
]); Sommar et al. [23
] reported lower GOM concentrations for annular denuders. In the Munthe et al. [20
] intercomparison, mist chamber measurements were made, and concentrations agreed with those measured by the denuders. Nacht et al. [22
] worked in a highly Hg contaminated location, reported RM concentrations of up to 75,000 pg m−3
with the highest values being above mine tailings.
In 2000, Steffen et al. [24
] reported on the use of a cold regions pyrolysis unit manufactured by Tekran to allow for measurement of total gaseous mercury, while simultaneously measuring GEM. Their measurements were conducted during a Hg depletion event in the Arctic at Alert, Nunaurt, Canada. They observed that 48% of the converted GEM was measured as RM with the pyrolyzer unit and the rest deposited to snow.
Landis et al. [25
] was the first to report on the use of an annular denuder in an automated system from which GOM could be thermally desorbed repeatedly to improve temporal resolution. During the period of denuder development, Landis et al. [25
] and Xiao et al. [26
] tested the efficiency of KCl denuders to collect permeated HgCl2
, with the latter testing CH3
HgCl as well. Neither study was conclusive; for example, Xiao et al. [26
] utilized clean air, and the spiked GOM concentrations are not reported; while Landis et al. [25
] data were limited (n = 2) and spike concentrations were one-to-two orders of magnitude higher than reported ambient concentrations (c.f. Valente et al. [27
]). Feng et al. [21
] reported limited laboratory tests of a tubular denuder loaded with hundreds to 1200 pg in three tests to determine breakthrough; however, the air used for the tests was not made clear. They used thermal desorption of the KCl denuders instead of liquid extraction The authors suggested that if a pyrolyzer was not used after desorption of the denuder that volatile or semi-volatile compounds trapped in the denuder would be released and deposit on the surface of the gold trap, risking passivation. Feng et al. [21
] also recommended a denuder desorption temperature of 900 °C, due to the presence of a dual peak that they suggested was not a Hg compound, but volatile organic compounds that interfered with the analysis of Hg. No interference testing was reported in these studies. Feng et al. [21
] commented on the fact that if water vapor condensed on the denuder, the sampling efficiency would decrease. Landis et al. [25
] suggested that the temperature of the denuder be maintained at 50 °C to prevent hydrolysis of the KCl coating.
2537/1130/1135 speciation system (Tekran system manufactured by Tekran, Toronto, Canada) was first introduced in 2002, and collects GEM, GOM, and PBM, respectively [25
]. Ambient air entering the Tekran system first passes through an elutriator used to prevent coarse particles (>2.5 µm) from moving into the system; the flow rate of the system determines the particle cut size and must be routinely monitored and adjusted. Air then passes through the KCl denuder (1130 module, GOM capture) and subsequently through a quartz fiber filter (1135 module, PBM capture). Downstream of these modules is a pyrolyzer, packed with quartz chips, used to reduce GOM and PBM to GEM at predetermined intervals. Lastly, the air enters the 2537 module, which collects GEM by way of amalgamation on one of two gold-coated sand traps; the two traps are used to alternately collect and desorb Hg, allowing for continuous collection at 2.5 + min resolution (commonly 5 min). GEM is desorbed from the cartridges at 325 to 370 °C, then carried by argon to a quartz cell where Hg is quantified using cold vapor atomic fluorescence spectroscopy (CVAFS). The method detection limit for GEM is 0.1 ng m−3
. While GEM is being measured, GOM and PBM are collected over 1 to 2 h. These operationally defined fractions are then sequentially thermally desorbed at 550 and 700 °C for GOM and PBM, respectively. GOM and PBM concentrations are quantified in Hg-free air after three flushing cycles without heating (system blank check), then one cycle of pyrolyzer heating, three cycles for desorbing the particulate filter, three cycles for desorbing the denuder, and two flushing cycles without heating to allow the system to cool. Desorbed GOM and PBM compounds pass through the pyrolyzer and are measured as GEM by the 2537. A soda lime trap is typically installed inline directly upstream of the 2537 inlet to prolong the life of the gold traps and is changed monthly. Typically, the 2537 module is calibrated every 24 h using an internal GEM permeation source, and less regularly using manual injections from an external GEM permeation source. It is noteworthy that calibrated 2537 units sampling the same air can generate concentrations that are up to 28% different (c.f., Gustin et al. [28