Overview of Atmospheric Mercury
Measurement Uncertainties
Mae Gustin and Jiaoyan Huang
Department of Natural Resources
and Environmental Science
Acknowledgments
• Funding sources –
– U. S. National Science Foundation
– Electric Power Research Institute –EPRI
• Thanks to Gustin lab graduate and
undergraduate students
• Thanks to the many field site operators who
have helped us
Progress on Understanding Atmospheric Mercury Hampered
by Uncertain Measurements
Daniel A. Jaffe,*,†,‡ Seth Lyman,§ Helen M. Amos,∥ Mae S. Gustin,⊥ Jiaoyan Huang,⊥
Noelle E. Selin,# Leonard Levin,∇ Arnout ter Schure,○ Robert P. Mason,◆ Robert Talbot,¶
Andrew Rutter,∞ Brandon Finley,† Lyatt Jaeglé,‡ Viral Shah,‡ Crystal McClure,‡ Jesse Ambrose,†
Lynne Gratz,† Steven Lindberg,$ Peter Weiss-Penzias,⊗ Guey-Rong Sheu,∀ Dara Feddersen,⧓
Milena Horvat,◘ Ashu Dastoor,Я Anthony J. Hynes,@ Huiting Mao,Π Jeroen E. Sonke,★ Franz Slemr,⧖
Jenny A. Fisher,∫ Ralf Ebinghaus,∮ Yanxu Zhang,× and Grant Edwards⪫
• “at present there is no consensus on what the chemical
form(s) of GOM is(are),nor any reliable method to
identify the chemical form(s) in the atmosphere. It is
likely that more than one form of Hg(II) exists in the
atmosphere, depending on its source.” and the
oxidants in the air.
• “protocols do not provide a way to calibrate for GOM
or PBM, quantify collection efficiency or quantify
measurement interferences.”
• “There is some uncertainty as to whether current
unspeciated measurements capture total gaseous
mercury (TGM) or GEM.”
EST, 2014
Tekran system
-Has been useful for
making progress for
understanding
atmospheric Hg
-Being used in networks
world wide
--KCl-coated denuder
and particulate filter not
calibrated or checked for
interferences
Talk outline
• Focus on GOM uncertainties
• Brief summary of methods
• Chronological discussion of investigation of
GOM uncertainties
• Note on PBM– instrument demonstrated to not collect all PBM
(Talbot et al., 2011)
– PBM likely measures GOM not collected by the
denuder (Gustin et al., 2013)
GOM Passive samplers
Easily deployed by
regular people
Dry deposition
No electricity needed
Demonstrated use in Florida, California, Nevada,
New Mexico, Oklahoma, Texas…….Maryland
Lyman et al., 2009 and 2010 EST AE
Huang et al., 2014 Environ. Sci.: Processes Impacts
Concentrations
UNR active system
Tekran 1135
Tekran 1130
Dry gas meter
Flow controller
Vacuum pump
00014m3
Tekran
1130/2537
Developed for measurement of GOM concentrations and
chemistry
Huang et al. 2013
UNR laboratory manifold
Ozone analyzer at
1.2 Lpm
Tekran® at 4 or
7Lpm
DI water
source
Impactor
Port 4: 1 Lpm
Port 3: 1 Lpm
Port 2: 1 Lpm
Zero air tank
Port 1: 1 Lpm
2.5-5.5 Lpm
RH and Temp
Manifold
Exhaust
GOM source
0.05-0.1 Lpm
Zero air filter
8-13 Lpm
Developed for loading and calibrating
membranes and calibrating the denuder
Ambient air inlet
Main pump
Huang et al., 2013
Florida TMDL Study
Assume the passive sampler data are correct and the Tekran data
are wrong
Gustin et al. 2013 ACP
Peterson et al. 2012 STOTEN
Passive samplers in Florida-GOM
SS data corrected should be
higher by 0.2 ng m-2 h-1
Recent work suggests CEC
reflects natural surfaces
(Huang and Gustin 2014)
Measured deposition always
higher than modeled
Passive samplers and Tekran
not always correlated
Passive sampler uptake and
SS deposition not always
correlated
Different deposition velocity
of the different forms will
influence uptake
Peterson et al., 2012 STOTEN
Conclusions
 Peterson et al. 2012 Investigate the utility of passive sampling systems to
record spatial and temporal patterns of atmospheric Hg
 Samplers do record spatial and temporal variation
 Variation does not match that measured with the Tekran
 Data suggests different forms of GOM across space and time
 Gustin et al. 2012 based on combining criteria air pollutant data and
meteorology
 At OLF
 natural background dry deposition 0.03 ng m-2 hr-1
 no significant influence of the EGP
 LRT 0.11 ng m-2 hr-1 GOM dry deposition derived from N-NW
 At Tampa
 Mobile sources 0.2 ng m-2 hr-1
 LRT in the spring 0.08 ng m-2 hr-1
 Davie
 Local point sources 0.1 ng m-2 hr-1
 LRT in the fall and spring 0.1 ng m-2 hr-1
Remember the dry deposition values were corrected and should be
higher by 0.2 ng m-2 hr-1
-Limitations of current methods to
measure atmospheric Hg need to be
systematically addressed
Reno Atmospheric Mercury
eXperiment – RAMIX
August to September 2011
Tekran Uncertainties Field-RAMIX
In manifold
GEM/TGM
Response
Weeks 1 to 3
[ng m-3]
Spike
Concentration
5 to 25
Response/Recovery (%)
mean std dev max
min
72
13
93
40
55
6
62
43
Instrument
r2
Spec 1
Spec 2
Spec 2
adjusted
0.88
0.92
n
hourly
20
16
0.92
16
0.96
47
76
9
86
60
76
12
97
35
UNR 2537
23
24
13
15
48
61
9
9
33
22
85
13
Spec 1
Spec 2
Spec 2
adjusted
16
16
RM Response
Week 3
[pg m-3]
~660
Estimated GEM
Recovery
Week 4 [ng m-3]
6 to 8
76
7
-
-
Spec 1 – Spec
2 adjusted
3
Estimated RM
Recovery
Week 4 [pg m-3]
340 to 780
17
3
24
11
Spec 1 – Spec
2 adjusted
18
16
Gustin et al., 2013 EST
RM comparison between Tekran and
DOHGS instruments
DOHGS system higher RM concentration and measured form(s)
of RM not detected by Tekran
Tekran RM [pg m
100
500
400
300
200
50
100
0
00:00
Sep-09
0
12:00
00:00
Sep-10
12:00
00:00
Sep-11
Environmental Science and Technology v 47 Issue 13
Environmental Measurement Methods
Finley et al. Ambrose et al. Gustin et al. 2013
12:00
00:00
Sep-12
-3]
-3]
Spec 1
DOHGS
DOHGS BDL
Spec 2 adjusted
DOHGS RM [pg m
150
Tekran Uncertainties
Laboratory manifold
2
6
HgBr2
HgO
5
1
HgCl2, y = 1.6 (0.1) x + 0.002 (0.02)
r2 = 0.97, n = 12
HgBr2, y = 1.7 (0.1) x - 0.01 (0.02)
r2 = 0.99, n = 10
HgO, y = 1.8 (0.05) x - 0.02 (0.03)
r2 = 0.99, n = 8
GOM measured by CEM [ng]
GOM measured by nylon membrane [ng]
HgCl2
4
3
2
HgCl2, y = 2.4 (0.8) x + 0.1 (0.5)
r2 = 0.58, n = 9
HgBr2, y = 1.6 (0.4) x + 0.2 ( 0.1)
r2 = 0.86, n = 5
1
HgO, y = 3.7 (0.2) x + 0.1 (0.1)
r2 = 0.99, n = 6
0
0
0
1
2
0
1
2
3
4
5
6
GOM measured by automatic KCl-coated denuder [ng]
Huang et al., 2013 EST
0
25 100
Precip
TEMP
RH
20
60
40
20
0
5
-20
0
80
15
60
40
15
10
20
10
0
-20
Precipitation [mm]
MBL
06-10-12
0
100
05-27-12
20
05-13-12
40
04-29-12
A
B B B B
04-15-12
A A A
80
04-01-12
B
03-18-12
Tekran
Nylon
CEM
RM concentration [pg m-3]
120
03-04-12
100
04-17-12
HI
04-03-12
03-20-12
03-06-12
B
02-21-12
A
02-07-12
B
01-24-12
A A A
01-10-12
80
12-27-11
60
12-13-11
RM concentration [pg m-3]
80
11-29-11
11-15-11
Temp [C] and RH [%]
06-10-12
05-27-12
05-13-12
04-29-12
04-15-12
04-01-12
03-18-12
03-04-12
04-17-12
04-03-12
03-20-12
03-06-12
02-21-12
02-07-12
01-24-12
01-10-12
12-27-11
12-13-11
11-29-11
11-15-11
Field measurements
120
100
AI
B
60
40
20
25
20
5
0
Huang et al., 2013 EST
GOM compounds
Temp [C]
40
60
80
100
120
140
160
Laboratory
20
180
200
HgCl2 (n = 9)
20
HgBr2 (n = 9)
HgO (n = 6)
GEM (n = 3)
15
10
10
5
5
0
HI
Nov and Dec 2011 (n = 6)
Jan and Feb 2012 (n = 9)
April 2012 (n = 3)
10
5
10
5
0
15
AI
Aug 2011 (n = 4)
Jun 2012 (n = 6)
0
15
10
10
5
5
0
25
MBL
Mar 2012 (n = 9)
Apr 2012 (n = 3)
0
25
20
20
15
15
10
10
5
5
0
0
40
60
80
100
120
140
160
180
200
Temp [C]
Huang et al., 2013 EST
GOM/RM release percentage [%]
0
GOM/RM release percentage [%]
15
Laboratory study- Ozone impacts the
denuder
Lyman et al. 2010
Laboratory study-relative humidity
impacts the denuder
RH <35% collection efficiency is 21 + 9% lower n=8
RH > 35% collection efficiency is 35 + 18 % lower n=9
TGM versus GEM
• Inlet configuration will significantly influence
result
• RAMIX insight- covered line and temperature
drops results in RM deposition
• Uncovered line results in better transmission
of TGM
• Some systematic tests are needed to better
understand this
Conclusions
• KCl denuder does not collect different forms
of GOM with equal efficiency
• The measurement is biased low and varies as
a function of the different forms in air.
• There are interferences with water vapor and
ozone that result in GOM being biased low.
Conclusions
• Surrogate surfaces are useful for understanding
potential dry deposition
• Passive samplers are useful for understanding relative
concentrations
• These samplers may be applied across broad spatial
and temporal scales
• The passive sampler method needs to be refined
• Additional laboratory tests are needed to understand
and calibrate
• Huang et al 2014 Critical Review Environ. Sci.:
Processes Impacts, 16, 374-392.
Conclusions
• GOM compounds vary across space and time
• Source of Hg are global it’s the oxidants
present that will influence the production of
GOM and the GOM chemistry
What are current issues?
Determine limitations of the denuder so we
can interpret data collected in the past
Additional tests needed for passive samplers
Unknown GOM compounds in air
 Address uncertainties of PBM measurements
Lack of world-wide measurement of long term
spatial and temporal variation in Hg
How to move forward
• Active system appears to work well for
quantifying GOM collected on CEM
• Thermodesorption is a first step at trying to
understand presence of different forms in air
• Passive samplers are useful for measuring
deposition and assessing concentrations
across space and time
Needs
• “Develop calibration methods for GOM and provide routine
calibrations for field instrumentation;
• Conduct detailed investigations to quantify interferences in
the existing GOM methods and develop new
methodologies to measure it; and
• Conduct fundamental research on the chemistry, reaction
kinetics and chemical identity of the compounds that
makeup GOM and PBM in the atmosphere.
• We believe these items should be given high priority by the
mercury scientific community. To do otherwise impedes
scientific progress and environmental monitoring efforts.”
Exact words from Environmental Science & Technology
Viewpoint
B dx.doi.org/10.1021/es5026432