Introduction
Mercury (Hg) emitted in the gas phase can remain in the Earth's atmosphere
for many months and be transported around the globe (Lindberg et al.,
2007). Atmospheric Hg pollution is a global problem, and regulation of Hg
emissions exist at the state, national, and international levels (Selin,
2009). Hg pollution arises from a variety of natural and anthropogenic point
and nonpoint sources (Gustin et al., 2008; Seigneur et al., 2004). Hg can
exist in the atmosphere as elemental Hg (Hg0), or as various oxidized
Hg compounds (HgII) (Lyman et al., 2010a). Most mercury
is emitted to the atmosphere as Hg0 (Pacyna et al., 2006),
but it can be oxidized to HgII in the atmosphere, and HgII can be
reduced to Hg0 (Hedgecock and Pirrone, 2004). HgII can be
found in both the particulate-bound (HgpII) and gaseous forms
(HggII) (Sprovieri et al., 2010) and is water-soluble
and semi-volatile (Gustin et al., 2008; Lindberg et al., 2007). As a
result, aerosols and clouds readily absorb HggII, and it is also
readily dry deposited (Holmes, 2012; Lyman et al., 2007). Lyman
and Jaffe (2012) and others (Gratz et al., 2015; Slemr et al., 2009;
Talbot et al., 2007) report that the upper troposphere and lower
stratosphere are depleted in Hg0 and enriched in HgII, and
oxidation of Hg0 to HgII has also been shown to occur in the
marine boundary layer (Wang et al., 2014) and the Arctic during
springtime (Steffen et al., 2008). The location and timing of
Hg deposition to ecosystems depends on atmospheric chemistry and the form of
Hg in the air (Gustin et al., 2013b; Holmes et al., 2010; Lin et al.,
2006; Lyman and Gustin, 2008; Lyman et al., 2010a).
In 1974, Johnson and Braman (1974) suggested that oxidized Hg
might consist of HgO and/or Hg halides, and most current studies still echo
this hypothesis (Ariya et al., 2009; Holmes et al., 2010; Hynes et al.,
2009; Lin et al., 2006). HgO could be produced by the reaction of Hg0 with
ozone (Pal and Ariya, 2004b), OH (Pal and Ariya,
2004a), or NO3 (Sommar et al., 1997). Evidence exists for the
involvement of NO3 in formation of HgII (Peleg et al., 2015).
Some have argued that HgO is likely to exist only as HgpII
(Calvert and Lindberg, 2005; Shepler and Peterson, 2003) and that
oxidation of Hg0 by ozone may produce an HgO3 intermediate which
could decompose to HgO on particles (Calvert and
Lindberg, 2005) or react with water to form Hg(OH)2 (Tossell, 2006). Seigneur et al. (1994) suggested
that Hg(OH)2 could be produced from the reaction of Hg0 with
H2O2.
Reactive halogen species, especially bromine species, have received attention
as potential Hg0 oxidants since the discovery of Hg depletion events
associated with “bromine explosions” during Arctic spring (Schroeder et
al., 1998; Steffen et al., 2008), and HgBr2 and/or HgBrOH have been
hypothesized as products (Holmes et al., 2006). Halogen radicals have also
been implicated as potential Hg oxidants in the marine boundary layer
(Laurier et al., 2003), the Dead Sea (Obrist et al., 2011), and the free troposphere and stratosphere (Holmes et al., 2006).
HgII emitted from combustion facilities is generally thought to be
HgCl2 (Galbreath and Zygarlicke, 2000; Wilcox and Blowers, 2004).
Some oxidized Hg in the atmosphere may be methylmercury, but less than 3 % of oxidized Hg in rainwater is methylated (Lindberg et al.,
2007), and it is likely that most atmospheric HgII is inorganic.
Hg0, HgpII, and HggII (operationally defined) are
measured routinely at dozens of locations around the world. Current
measurement methods for HggII have been shown to be biased low,
however (Gustin et al., 2013a; McClure et al., 2014), and ozone and water
vapor have been implicated as interferences (Huang and Gustin,
2015; Lyman et al., 2010b). These measurements have only been calibrated
rarely, and development and regular deployment of a field-deployable
calibrator has been called for (Gustin et al., 2013b; Jaffe et al.,
2014). Current HgII instrumentation is also not species-specific, so it
does not provide information about the individual compounds that make up
measured HgII (Jaffe et al., 2014). Different species will have
different deposition rates, solubility, and bioavailability (Eagles-Smith
and Ackerman, 2014; Peterson et al., 2015), so determining the chemical
nature of HgII is a critical research priority (Gustin et al.,
2013b; Jaffe et al., 2014; Malcolm and Keeler, 2007). Huang et al. (2013b) have
developed a thermal desorption system to provide chemical
information about HggII, but other, complimentary methods are also
needed.
Unlike the atomic fluorescence method commonly used for measurement of
atmospheric Hg, mass spectrometric methods can be used to identify the
chemical composition of Hg compounds (Deeds et al., 2015). Further, gas
chromatography/mass spectrometry (GC/MS) can allow for separation of
individual Hg compounds and separation of Hg compounds from non-mercury
components of ambient air samples (Babko et al., 2001; Olson et al., 2002). Inductively
coupled plasma mass spectrometry (ICP-MS) is routinely used for measurement of Hg
and Hg isotope ratios, but the method is only useful for elemental analysis
(dos Santos et al., 2009). GC analysis is routinely used for analysis of
organic Hg in various media, and for analysis of HgII in water after
alkylation of the inorganic Hg compounds (Cavalheiro et al., 2014).
However, alkylation destroys the native structure of Hg compounds and thus
does not provide information about their original identity.
Babko et al. (2001) showed that GC/MS could be used to
separate and identify Hg halides. They injected a solution of HgCl2 in
acetone into a GC/MS and found good recovery and consistent peaks. The
detection limits, however, were much higher than would be practical for
ambient air analysis. Olson et al. (2002) used GC/MS to identify HgII
generated by an MnO2 sorbent in simulated flue gas. They used an
impinger to trap the Hg in acetonitrile, then evaporatively concentrated the
solution before injecting into a GC/MS. They injected HgCl2 in
acetonitrile and observed mass spectra that were clearly indicative of
HgCl2. They also injected Hg(NO3)2 and
Hg(NO3)2× H2O and saw a similar peak and mass
spectra to what they observed in the gas that passed through the MnO2
sorbent.
We have developed a GC/MS-based system to quantify and chemically identify
Hg compounds. We describe this analytical system in detail, provide first
results, and discuss remaining challenges.
Materials and methods
The GC/MS-based HgII detection system consisted of a sample collector
to concentrate Hg compounds from the ambient atmosphere, a sample desorber
to introduce collected compounds into the gas phase, a cryogenic
preconcentrator (cryotrap) to focus and inject Hg compounds, a gas
chromatograph to separate Hg compounds from each other and from possible
interferents, and an ultra-sensitive mass spectrometer to definitively
determine the chemical speciation of Hg compounds (Fig. 1). It also
incorporated a permeation system, pyrolyzer, and Hg0 detector to
introduce a consistent, quantifiable amount of various Hg compounds to the
system in the gas phase. All wetted parts of the system were kept at at
least 160 ∘C (except the sample desorber, which was sometimes
cooler), and all wetted parts except the GC columns and VICI GC valve rotors
were composed of deactivated fused silica-coated stainless steel. The VICI
GC valve rotors were composed of Valcon E (a polyaryletherketone/PTFE
composite).
Diagram of GC/MS system used to identify Hg compounds, and a
pyrolyzer and Tekran 2537 system to quantify the amount of Hg compounds
generated by the permeation oven.
Sample collector
Four collection materials were tested for suitability to concentrate
volatile Hg compounds in air samples. Collection materials tested were nylon
membranes, polydimethylsiloxane (PDMS) sorption tubes, quartz wool-filled
tubes, and deactivate fused silica-coated stainless steel. Nylon membranes
were Cole-Parmer nylon polyamide membranes (47 mm round, 0.2 µm thick,
P/N: EW-36229-04). The PDMS sorption tubes were those for a Gerstel Thermal
Desorption Unit (TDU), and were filled with conditioned PDMS foam (P/N:
013758-105-00). Quartz wool-filled tubes were made from 22 cm long × 1.3 cm diameter
perfluoroalkoxy (PFA) tubing that was washed with
soap and water, soaked for 24 h in a 10 % nitric acid bath, rinsed
with 18.2 MΩ cm-1 water, then dried in a particle-free
environment. The tubing was then filled with quartz wool that had been baked
at 800 ∘C for 2 h to ensure no contamination. Deactivated
fused silica-coated stainless steel was a 1 cm length of 0.3 cm tubing.
Selection of the best collection surface was based on presence of
identifiable Hg peaks on the GC/MS with the least amount of signal
interference. Hg was introduced to the sample collectors either from the
permeation oven or by passing outdoor ambient air through the collectors.
Ambient air samples were collected from Peavine Peak (latitude 39.590,
longitude -119.929) near Reno, Nevada; at the University of Nevada, Reno
campus in Reno, Nevada (latitude 39.537, longitude -119.805); and Grizzly
Ridge (latitude 40.738, longitude -109.484) near Vernal, Utah, using quartz
wool-filled tubes and nylon membranes. Quartz wool-filled tube ambient air
samples were collected by pulling air through the tubes at 30 L min-1
for 3 h. Some of the tubes were at ambient temperature during
collection, while others were kept at 0 ∘C. Nylon membranes were
collected by pulling air through the membranes at 1 L min-1 for 2
weeks. More information about nylon membrane methods is available from
Huang et al. (2013a). All ambient air samples were collected during
summer months.
Sample desorber
A thermal desorption module was used to reintroduce collected compounds into
the gas phase. We constructed this module by connecting a lab oven to an
adjustable digital temperature controller. Membrane samples were placed
within a sample desorption chamber inside the oven, while sample collection
tubes were connected directly to the desorption flow path. The desorption
chamber was stainless steel coated with deactivated fused silica. A constant
flow of ultra-high purity (UHP) Helium (He) acted as a carrier gas to pass
volatilized compounds to the cryotrap. The flow of UHP He was controlled at
a rate of 30 mL min-1. Desorption temperatures in the range of
80–160 ∘C were used.
Cryogenic preconcentrator
The cryogenic preconcentrator (cryotrap) was used to focus desorbed
compounds prior to introduction into the GC. A Scientific Instrument
Services Model 961 GC Cryo-Trap was used with liquid nitrogen as the
cryogen. The cryotrap works by enclosing a portion of the GC column in a
small metal cylinder. A flow of liquid nitrogen is passed through the small
cylinder at a rate determined by a digital temperature controller. Volatile
compounds are retained on the cooled column. After collection, the metal
cylinder rapidly heats up via a nichrome wire heating coil, volatilizing
concentrated compounds and allowing them to pass into the GC/MS. During this
step, the trap temperature is able to increase by 14 ∘C per
second. Cryotrap cooling temperatures tested ranged from -50
to 30 ∘C, and heating temperatures tested ranged from
170 to 240 ∘C. The cryotrap was housed in a lab oven.
Lab oven temperatures tested ranged from 160 to 220 ∘C. A VICI six-port GC valve (Model 4C6WT) was housed within the lab oven and
controlled flow of sample to the cryotrap, and from the cryotrap to the
GC. A heated line with a deactivated fused silica guard column (0.25 mm
internal diameter) was used to connect the cryotrap to the GC.
Gas chromatograph with an ultra-sensitive mass spectrometer
A Shimadzu GC-2010 Plus gas chromatograph was used to separate Hg compounds
from each other and from possible interferents. Different GC column types
and lengths were used to test optimum conditions for Hg compound separation.
A 30 m low-polarity Restek Rxi-5Sil MS column (5 % diphenyl/95 %
dimethyl polysiloxane), a 60 m ultra-low-polarity Supelco SPB-Octyl
fused silica capillary column (50 % n-Octyl/50 % methyl siloxane), and a
30 m non-polar Restek Rxi-1ms column (100 % dimethyl polysiloxane)
were tested. GC oven temperatures tested ranged from 140 to
220 ∘C. After passing through the GC column, the compounds of
interest moved to a Shimadzu QP2010 Ultra mass spectrometer for detection of
unique chemical signatures of Hg compounds in samples. The MS was operated
in high-sensitivity electron impact ionization mode, and included a direct
probe inlet. Small quantities of solid-phase Hg compounds were added
directly to the MS via to the direct probe inlet to determine representative
mass spectra for the compounds.
Permeation system, pyrolyzer, and Hg detector
Permeation tubes were made to generate four HgII compounds (HgBr2,
HgCl2, Hg(NO3)2, and HgO; Sigma-Aldrich, purity 99.9 % or
greater). These compounds were packed in permeation tubes constructed of
thin-wall 0.3 cm diameter FEP tubing with solid polytetrafluoroethylene (PTFE)
plugs in both ends. The permeable length of each tube was
approximately 1 mm. Larger permeation tubes (1.3 cm diameter × 15 cm permeable length) were also tested for Hg(NO3)2 and HgO.
Permeation tubes were enclosed within 0.5 cm inner diameter deactivated
fused silica-coated stainless steel tubing. UHP He flowed at 30 mL min-1
through the stainless steel tubing and over the permeation tubes.
The tubes were house in an oven consisting of an insulated metal box heated
to 100 ± 0.1 ∘C. A VICI multiport GC valve (Model CSF6)
selected among four available permeation tubes or passed permeation flow to
vent. The multiport valve was housed within the same lab oven as the
cryotrap.
A pyrolyzer was used to verify permeation rates of Hg compounds. The
pyrolyzer consisted of a 2.5 cm diameter × 18 cm length quartz tube
packed with quartz wool. The quartz tube was wrapped with nichrome wire that
was used with a variable voltage controller to control the temperature of
the tube. The pyrolyzer was kept at 800 ∘C to convert Hg compounds
to Hg0 as they passed from the permeation oven through the quartz tube.
Hg0 concentrations were measured downstream of the pyrolyzer using a
Tekran 2537 mercury vapor analyzer.
Chromatogram of a mixed standard of HgCl2 and HgBr2 in
acetone (0.5 mug µL-1 for each compound) (top), mass
spectra for HgCl2 (b, middle) and HgBr2 (b, bottom). Solvent delay
was 2 min.
Picograms of HgII and Hg0 recovered when HgBr2 was
permeated into mercury-free air and passed through a 15 cm section of 0.3 cm
diameter tubing constructed of the materials indicated. Tubing was kept at
110–115 ∘C, except when indicated otherwise. “Siltek” indicates
stainless steel tubing coated with Siltek-brand deactivated fused silica.
Whiskers represent 95 % confidence intervals.
Hg compound transmission tests
We permeated HgBr2 into a 1 cm diameter PFA manifold to test the
ability of different materials to transmit Hg compounds (Fig. S1 in the Supplement). The
manifold was heated to 100 ∘C. Air scrubbed of Hg via an activated
carbon cartridge was drawn through the manifold at 10 L min-1. A tee
pulled a 1 L min-1 subset of air from this manifold into a Tekran
2537/1130 speciation system with a KCl-coated denuder, which measured
Hg0 and HgII. The tee to the denuder was 100 cm downstream of the
point where HgBr2 was added to the manifold air. One of several 15 cm
long × 0.3 cm diameter tubes was placed between the manifold and
the Tekran speciation system to test the ability of these tubes to transmit
HgBr2. These 15 cm tubes were constructed of different materials,
including stainless steel, PFA (as a control, since the entire manifold was
PFA), PEEK, and deactivated fused silica-coated stainless steel. Fittings
used to secure the 15 cm tubes were of the same materials as the tubes. The
15 cm stainless steel, PFA, PEEK, and deactivated fused silica-coated
stainless steel tubes were each tested for 24 h. The Tekran system collected
one measurement every 1.5 h, so approximately 16 samples were obtained
for each experimental condition.
Results and discussion
We initially identified HgBr2 and HgCl2 via the method utilized by
Babko et al. (2001), which was to dissolve the compounds in
acetone and inject the solution into a splitless inlet at 200 ∘C.
We used a non-polar 100 % PDMS column at 160 ∘C for separation.
We were able to separate HgCl2 from HgBr2 with this system (Fig. 2).
The relative abundance of Hg isotopes in Fig. 2 is similar to isotopic
abundances reported by IUPAC (de Laeter et al.,
2004), confirming the identification. The detection limit of HgCl2
analyzed by this method, calculated as 3 times the standard deviation of
seven injections near the detection limit, was 9 ng, much too high for
ambient air detection of mercury compounds. Replicate injections showed a
high degree of variability (relative standard deviation of 30 to 45 %).
The manual syringe used for injections became permanently contaminated with
HgBr2 and HgCl2 after use, so we expect that much of the observed
variability was due to retention and interactions on the syringe walls, as
well as the walls of the injection port.
To eliminate the variability created by liquid injections of these very
reactive compounds, we developed the system described in Methods. We tested
a number of different configurations and materials to determine the method
most likely to allow for identification of HgII in ambient air.
System materials and temperatures
As a first step in the development of the system, we tested the ability of
stainless steel, deactivated fused silica-coated stainless steel (Siltek
brand), PFA, and PEEK tubing to transmit gas-phase HgBr2. The lowest
HgII recovery (and highest Hg0) was observed with stainless steel
tubing, followed by PEEK (Fig. 3). Deactivated fused silica-coated stainless
steel performed better than PFA, with more HgII and less Hg0
observed. The amount of total Hg recovered was the same whether deactivated
fused silica-coated stainless steel was heated to 135 or
110–115 ∘C (p= 0.39; the manifold used for these tests was not
capable of achieving more than 135 ∘C), but the percentage of
recovered Hg that was HgII increased from 83 to 98 %. It is not
clear why the 15 cm deactivated fused silica-coated stainless steel tube at
135 ∘C resulted in such an improved HgII : Hg0 ratio
relative to the 15 cm PFA tubing, since a relatively short length of coated
stainless tubing was used in a manifold that was otherwise constructed
entirely of PFA, and some decomposition of HgII to Hg0 would be
expected in the remainder of the manifold even if the 15 cm tube did not
lead to any decomposition. More study of Hg compound decomposition in the
presence of different materials is warranted.
GC/MS results from permeation of HgBr2 with varying GC oven,
valve oven, and line temperatures. Each iteration was performed at least in
duplicate to verify the consistency of results. Peak area and height are for
m/z 362.
GC oven
Retention
Peak
Peak
temp. (∘C)
time
area
height
140
5.91
11 794
802
160
4.52
14 676
1077
180
3.87
17 037
1158
200
2.66
22 892
1235
Valve oven and
line temp. (∘C)
180
4.32
36 058
1989
200
4.19
40 646
1998
It appears from Fig. 3 that cooler temperatures and less suitable tubing
materials led to the conversion of HgBr2 to Hg0, perhaps due to
reactions facilitated by or on the materials themselves. Others have
reported that higher temperatures allow for increased transmission of Hg
halides without significant decomposition to Hg0 (Lyman et
al., 2010b). Wilcox and Blowers (2004) determined a theoretical
temperature-dependent rate constant for the decomposition of HgCl2
(HgCl2+ M ↔ HgCl + Cl + M), and compared those
results to rate constants developed from experimental data (Widmer
et al., 2000). Their theoretical rate equation predicts that HgCl2 will
be stable up to about 850 ∘C, while the rate equation determined
from experimental data predicts HgCl2 stability up to 450 ∘C
(stable = less than half of initial HgCl2 decomposed within 60 s). At
240 ∘C (the maximum temperature reached by any part of our GC/MS
system), both rate equations predict ≪ 1 % decomposition
within 60 s. In addition, L'Vov (1999) showed minimal decomposition of HgO up
to 450 ∘C.
We constructed the plumbing of our GC/MS system using deactivated fused
silica-coated stainless steel where possible. After the system was
constructed, we tested different system temperature settings, including
temperatures of the oven that housed the VICI valves and the cryotrap, the
temperature of the transfer line from the cryotrap to the GC, and the GC
oven temperature. Table 1 shows that higher peak area and peak height were
observed for higher GC oven, valve oven, and transfer line temperatures, up
to 200 ∘C. These temperatures were optimized for HgBr2, and
additional work is needed to determine optimal system temperatures for other
Hg compounds.
Cryogenic preconcentrator
Results of cryotrap cooling temperature tests are given in Table 2. While
cryotrap cooling temperatures of 0 and -25 ∘C
resulted in similar peak areas, peak heights were greater with 0 ∘C, probably because the higher temperature allowed for more rapid desorption
when the cryotrap was heated. Peak area and shape deteriorated at cryotrap
cooling temperatures above 0 ∘C. When analyzing ambient air
samples, cryotrapping temperatures slightly above 0 ∘C may be
ideal, since they would allow for efficient trapping of Hg compounds, but
allow water to pass through. While HgBr2 peaks were smaller when the
cryotrap was cooled to -50 ∘C, this lower temperature allowed the
cryotrap to efficiently collect Hg0 as well (data not shown).
GC/MS results from permeation of HgBr2 with varying cryotrap
cooling and desorption temperatures. Each iteration was performed at least
in duplicate to verify the consistency of results. Peak area and height are
for m/z 362.
Cryotrap cooling
Retention
Peak
Peak
temp. (∘C)
time
area
height
-50
4.21
104 386
4789
-25
4.22
137 471
7206
0
4.14
134 462
8623
5
4.14
117 864
8272
10
4.12
113 516
7589
30
4.13
39 526
1583
Cryotrap desorb
temp. (∘C)
170
3.35
83 198
8582
200
4.14
101 757
8557
220
3.36
173 059
17 501
240
3.377
175 723
17 567
Hotter cryotrap desorption temperatures resulted in better peak areas for
HgBr2 (Table 2), with a desorption temperature of 240 ∘C
resulting in the best peak area and peak shape. Hotter desorption
temperatures likely resulted in more rapid volatilization of HgBr2 from
the cryotrap, leading to improved peak shape.
Chromatographic columns
Of the three columns we tested for transmission of HgBr2, we only
observed HgBr2 peaks with the Rxi-5Sil MS column or the SPB-Octyl
column. We observed consistent HgBr2 peaks with the Rxi-1ms column.
Babko et al. (2001) observed HgCl2 peaks with a low-polarity column similar to the RXi-5Sil MS column we used (DB-5; 5 %
diphenyl/95 % dimethyl polysiloxane). Olson et al. (2002) only
observed Hg compound peaks when using a 30 m non-polar phase column (DB1;
100 % dimethyl polysiloxane) similar to the Rxi-1ms column we used.
Olson et al. (2002) were not able to observe Hg peaks with more
polar columns that had polyethylene glycol, cyanopropyl phenyl, or
trifluoropropyl phases.
Chromatogram (above) and mass spectrum at 2.92 min (below) of
HgBr2 emitted from a permeation tube. The Y axis shows the mass
spectrometer's ion current intensity. The box on the bottom right compares the
expected HgBr2+ mass spectrum (from de
Laeter et al., 2004) to the observed spectrum.
Hg compound detection
The permeation rate for HgBr2, determined by the pyrolyzer and Tekran
analyzer system, was 37 pg s-1. After incorporating the optimizations
reported above, and after further optimizing the parameters of the mass
spectrometer, permeation of 22.2 ng of HgBr2 resulted in a peak area of
1 788 451 and a peak height of 143 238 when the MS was operated in selected
ion mode for m/z 362 (Fig. 4). The HgBr2 detection limit for the
optimized system in selected ion mode, calculated as 3 times the
standard deviation of replicate low-concentration samples, was 90 pg. The
detection limit in scan mode was 300 pg.
A small, poorly formed Hg0 peak was often observed prior to Hg halide
peaks in chromatograms (see m/z 202 trace in top of Fig. 4). This Hg0
peak was probably the result of breakdown of Hg halides to Hg0 within
the chromatographic column, and its poor shape can be explained by continued
breakdown as Hg halides moved through the column. The small size of the peak
relative to the Hg halide peak is an indicator that Hg halide decomposition
in the column was limited. Hg0 chromatographic peaks during Hg halide
injections were not likely the result of Hg0 emitted from permeation
tubes, since any Hg0 emitted or formed prior to the cryotrap would have
passed through the cryotrap at its typical collection temperature of
0 ∘C.
While Hg0 was detected from Hg(NO3)2 and HgO-containing
permeation tubes (when the cryotrap cooling temperature was lowered to
-50 ∘C), we were unable to observe unequivocal Hg(NO3)2
or HgO mass spectra when analyzing the output of these permeation tubes with
the GC/MS. A consistent peak with a prominent 218 m/z signal was observed at
a retention time of about 10 min when Hg(NO3)2 or HgO was
permeated (Fig. 5). While m/z 218 is the most abundant expected mass for
HgO+, the observed isotope pattern did not indicate Hg. Instead, the
mass spectrum for this peak was similar to mass spectra for siloxanes,
indicating column or tubing degradation as the source. The absence of Hg in
this peak was confirmed by the lack of an Hg+ signal at m/z 202 (Fig. 5).
Lyman et al. (2009) constructed HgO permeation tubes and found
that, along with Hg0, an Hg compound was emitted from these tubes that
could be collected and analyzed using KCl-coated denuders or cation-exchange
membranes. Huang et al. (2013a) showed that, when collected on nylon
membranes, the Hg compound emitted from HgO permeation tubes exhibited a
thermal desorption profile that was different from those exhibited by
HgCl2, HgBr2, and Hg0. They showed that Hg(NO3)2
permeation tubes also emit a reactive Hg compound. It is not known, however,
whether the Hg compound emitted from HgO permeation tubes is HgO. Huang
et al. (2013a) proposed that the emitted compound could be Hg2O.
Regardless of the chemical identity of the compound, it is possible that we
were not able to detect it because it degraded within the system tubing,
valves, or the chromatographic column. The GC/MS did not report any Hg
signal, including Hg0, when analyzing the output from HgO and
Hg(NO3)2 permeation tubes with a cryotrap temperature of
0 ∘C. This indicates either (1) the Hg compound emitted from these
permeation tubes degraded prior to or within the cryotrap, allowing Hg0
to pass through the cryotrap and out of the system; (2) the emitted Hg
compound is too volatile to collect on a cryotrap at 0 ∘C, or (3) the
emitted Hg compound becomes permanently bound to some part of the
analytical system and does not degrade. System temperatures and materials
were optimized for HgBr2, and different materials and/or a different
temperature regime may improve detection for compounds emitted from HgO and
Hg(NO3)2 permation tubes.
Chromatogram (above) and mass spectrum at 10.10 min (below)
generated from analysis of 34 ng Hg emitted from an HgO permeation tube. The
box on the bottom right compares the expected HgO+ mass spectrum (from
de Laeter et al., 2004) to the observed spectrum.
While the prominent mass peak was m/z 218, the observed isotope pattern does
not indicate HgO+.
We introduced small quantities of HgO and Hg(NO3)2 (separately)
into the direct injection probe on the MS to determine a mass spectrum for
these compounds. Figure 6 shows these mass spectra. Only Hg+ was
observed from the direct injection of HgO, and no significant signal was
observed around m/z 218 (HgO+) or m/z 420 (Hg2O+). This could
be because (1) HgO is not volatile enough to produce enough vapor in the
ionization chamber of the MS to result in a detectable signal, so only
off-gassed Hg0 was observed (as Hg+) or (2) the ionization energy
of the MS was so strong that all HgO was broken down to Hg+ within the
ionization chamber. HgO+ was observed, however, when Hg(NO3)2
was inserted into the direct probe, probably as a breakdown product. This
provides evidence that the MS could detect HgO as HgO+ if it did indeed
exist in the gas phase. The fact that it was not detected when solid HgO was
inserted into the direct probe may indicate that HgO does not have an
appreciable gas phase. If this is the case, it is not clear what is emitted
from HgO permeation tubes that can be collected on nylon membranes and
KCl-coated denuders (Huang et al., 2013a).
The mass spectrum for Hg(NO3)2 showed Hg(NO3)2+
(m/z 326) and several breakdown products, including HgNO3+ (m/z
264), and HgO+ (m/z 218). A cluster of peaks around m/z 343 was also
observed and could be interpreted as the monohydrate of Hg(NO3)2.
Olson et al. (2002) dissolved Hg(NO3)2 in acetonitrile
and interpreted resultant mass spectra to be caused by reaction of
Hg(NO3)2 with column material, producing CH3HgCl. Our direct
probe results suggest the existence of gas-phase Hg(NO3)2 and do
not match the spectra presented by Olson et al. (2002).
Laboratory tests of sample collection materials
We permeated Hg compounds onto nylon membranes and into PDMS foam-filled
tubes, quartz wool-filled tubes, and deactivated fused silica-coated tubes
and used the sample desorption oven to transmit collected Hg compounds onto
the cryotrap and then into the GC/MS. When we heated nylon membranes to
100 ∘C or greater in the desorption oven, the prominent observed
peak was consistent with dodecanoic acid, a potential degradation product of
the nylon material (Carraher, 2014). Dodecanoic acid exhibits prominent
peaks at and near m/z 200, and the dodecanoic acid peak intensity was so
high as to obscure any Hg peaks when loaded in the laboratory or when
sampling ambient air. Huang et al. (2013a) used nylon membranes to
collect Hg compounds and desorbed those compounds into a pyrolyzer and
atomic fluorescence analyzer. The atomic fluorescence instrument only
detects Hg0, however, so interference from nylon breakdown products was
not an issue.
Mass spectrum for Hg(NO3)2 (above) and HgO (below)
derived from direct probe injection into the MS. The boxes at top left in
each pane compare the expected indicated mass spectrum (from
de Laeter et al., 2004) to the observed spectrum.
Like nylon membranes, PDMS foam-filled tubes also had too much interference
to allow for detection of Hg compounds, even when loaded with as much as 60
ng HgBr2.
Quartz wool-filled PFA tubes exhibited less interference and clearly
identifiable Hg halide signals (Fig. 7). However, chromatograms from quartz
wool-filled tubes had poorly shaped peaks and substantial non-Hg signal. The
cause of the poorly shaped peaks is not known. Quartz particles could have
accumulated in the cryotrap, causing a slow desorption of Hg compounds
during cryotrap heating. If this occurred, the quartz particles were
apparently cleaned out after each analysis, since we performed injections of
Hg halides directly from the permeation oven onto the cryotrap subsequent to
quartz wool analyses and observed normal chromatographic peaks.
Chromatogram (above) and mass spectrum at 5.6 min (below) of
HgCl2 permeated onto quartz wool-filled PFA tubing, then desorbed into
the GC/MS system.
While HgCl2 was loaded onto a quartz wool-filled tube for the analysis
shown in Fig. 7, the figure shows some HgBr2 signal in the mass
spectrum. The quartz wool-filled tube had previously been loaded with
HgBr2, and some residual HgBr2 apparently remained in the tube.
Additionally, a cluster of masses centered at m/z 316, corresponding with HgBrCl,
can be clearly observed in the mass spectrum. The same cluster of masses can
be seen in part B of Fig. 2 and in Fig. 4. HgBrCl could exist because of (1) a
reaction between HgCl2 and HgBr2 in the acetone solution in Fig. 2
or on the quartz wool in Fig. 7, or (2) it could be a contaminant in the
commercial Hg halide compounds used in this study.
The deactivated fused silica-coated stainless steel tube collected
HgBr2 with no discernable non-Hg interference, but the amount of
HgBr2 collected was low, probably because of high breakthrough.
Deactivated fused silica-coated tubing may be a viable HgBr2 collection
device if a device with larger surface area is used and/or if the tubing is
cooled to a temperature that limits HgII breakthrough.
Ambient air sample collection
The location, duration, and methods used for ambient air sample collection
are given in Methods. None of the ambient air samples collected using quartz
wool-filled tubes or nylon membranes resulted in any detectable Hg
compounds. Low HgII concentrations in ambient air could have led to a
sampled HgII mass below the detection limits of the GC/MS. 10.1 m3
of ambient air was sampled by each nylon membrane, and 5.4 m3 of air
was sampled by each quartz wool-filled tube. These sampling volumes are
adequate to collect 100 pg HgII (within the detection limit of the
GC/MS) if ambient HgII was 10 and 18 pg m-3, respectively. Not all
ambient air sample collections were associated with alternative HgII
measurements, but for quartz wool-filled tube collection at the University
of Nevada, Reno, HgII (measured by a Tekran 2537/1130/1135 speciation
system as the sum of Hg collected on the system's denuder and particulate
filter) was 38 ± 9 pg m-3 (mean ±95 % confidence
interval). However, some breakthrough may have occurred through the sample
collection devices, and the non-Hg interference caused by the sample
collection devices likely increased the actual detection limit for these
samples. Alternatively, it is possible that the Hg compounds in sampled
ambient air were not Hg halides and were undetectable by the GC/MS system.
Future work
Improvements to the MS used in this work may increase its sensitivity for Hg
compounds and decrease the detection limits of our system. Deeds et al. (2015)
reported detection limits of 6–40 pg for HgCl2 and HgBr2
with an atmospheric pressure chemical ionization MS, and pointed out that
chemical ionization is likely to decrease detection limits relative to
electron impact ionization. The system used by Deeds et al. (2015),
however, suffered from interference from non-Hg compounds found in ambient
air because it did not utilize chromatographic separation. Our GC system,
coupled with chemical ionization MS, may be able to achieve improved ambient
air detection limits while maintaining the ability to separate individual Hg
compounds and separate Hg compounds from non-Hg atmospheric constituents.
Interference in mass spectra created by collection materials likely limited
our ability to detect Hg compounds in ambient air. Testing of additional
collection materials is needed. Deeds et al. (2015) used shredded Teflon
packed in tubes to collect Hg halides, and they did not note any
interference from these materials. We found no interference from deactivated
fused silica-coated stainless steel tubing. Highly inert surfaces like these
are ideal because they do not result in off-gasing that may interfere with
mass spectra. However, HgII may not collect efficiently on these
surfaces unless they are cooled to 0 ∘C or lower. In addition,
Lyman et al. (2010b) reported that ozone reduces Hg halides
collected on uncoated quartz traps to Hg0, and highly inert surfaces
may also leave HgII exposed to reaction with ozone or other atmospheric
constituents.
We observed poor results from tubes packed with PDMS foam, but saw little
interference from the PDMS-coated chromatographic column (for Hg halides).
PDMS in chromatographic columns is cross-linked to stabilize it and is less
likely to decompose. PDMS denuders have been used successfully to
preconcentrate a wide variety of compounds (Burger et al., 1991; Dudek et
al., 2002), including semivolatiles (Rowe and Perlinger, 2010).
PDMS may also shield analytes from atmospheric oxidants that have low
affinity for the PDMS phase (possibly including ozone) (Rowe and
Perlinger, 2010).
While the GC/MS system in this study was able to separate and quantitatively
analyze Hg halides, we have not yet shown that it can detect non-halide Hg
compounds, including Hg(NO3)2 and HgO. HgO may not exist in the
gas phase, but Hg(NO3)2 is likely to be non-polar, as are Hg
halides (Goodsite et al., 2004), and likely can exist in the gas phase.
HgBrX compounds, including HgBrOH, may exist in the atmosphere
(Weiss-Penzias et al., 2015). HgBrOH, unlike HgBr2, has an
appreciable dipole moment (Goodsite et al., 2004) and may have different
reactivity and volatility than HgBr2. Our system is able to detect
HgBrCl, so it could possibly identify other bromine-containing Hg compounds,
but this has not been tested. System temperatures and materials may need to
be optimized for individual compounds or groups of compounds. Different Hg
compounds may perform better with different columns, as shown by
Babko et al. (2001). Finally, replacement of VICI valves
(which have Valcon rotors that may react with Hg compounds) with
all-stainless steel valves that can be coated with deactivated fused silica
may improve system performance for Hg compounds that are more reactive than
Hg halides.