implications of ambient ammonia on aerosol acidity and ... · the rst project investigates di...
TRANSCRIPT
Implications of Ambient Ammonia on Aerosol Acidity andReactive Nitrogen Measurements
by
Phillip K. Gregoire
A thesis submitted in conformity with the requirementsfor the degree of Master of ScienceGraduate Department of Chemistry
University of Toronto
c© Copyright 2013 by Phillip K. Gregoire
Abstract
Implications of Ambient Ammonia on Aerosol Acidity and Reactive Nitrogen
Measurements
Phillip K. Gregoire
Master of Science
Graduate Department of Chemistry
University of Toronto
2013
This study describes two projects involving recent research on atmospheric ammonia.
The first project investigates differences in modelling techniques of aerosol acidity using
data from two recent field campaigns. Our results show that allowing or disallowing
gas-particle partitioning in the Extended Aerosol Inorganic Model (E-AIM) changed the
average modelled aerosol activity of H+ from one campaign by seven orders of magnitude
and that disallowing gas-particle partitioning may not accurately represent the chemical
state of the aerosols.
The second project investigates the interference of reduced nitrogen in commercial
chemiluminescent nitrogen oxide monitors with molybdenum oxide catalytic converters.
This phenomenon is strongly dependent on the temperature of the catalytic converter.
Our results show these instruments can have high conversion efficiencies of gaseous NH3
and NH4+
salts to NO at typical reported converter temperatures, but conversion efficiency
varies between instruments and may be the result of uncertainty in reported converter
temperature.
ii
Acknowledgements
Jennifer Murphy, my supervisor, has been inspirational in my development as an
academic and as a person. Her guidance has left me optimistic about the prospects of
bond between the physical sciences and the greater society.
Thanks also go to my committee, Jon Abbatt and Jamie Donaldson, who have both
instructed me and engaged with my academic interests. A special thanks to Jon Abbatt
for reviewing and commenting on my thesis.
Jeff Geddes, Milos Markovic, and Trevor Vandenboer facilitated my learning with
patience and enthusiasm that I hope I can share with others. Without Jeff’s instruction
on data processing and Milos’ and Trevor’s assistance with the AIM-IC, I cannot imagine
that this thesis would be possible in its current form.
Greg Wentworth and Alex Tevlin, my CONTACT-2012 colleagues, have truly put
time, energy, and passion into their work. Without their efforts, CONTACT-2012 would
not have existed. Carol Cheyne, Rachel Hems, and Geoff Stupple were also crucial to this
campaign.
I also thank IACPES for funding and creating a unique environment to experience
the research from my peers and be exposed to the social and political ramifications of
scientific research.
Finally, I want to thank Angela Hong for personally supporting me throughout my
research.
iii
Contents
List of Tables vi
List of Figures vii
1 Constraining Aerosol Acidity 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Accumulation and coarse mode aerosols . . . . . . . . . . . . . . . 3
1.1.2 Methods of characterizing aerosol acidity . . . . . . . . . . . . . . 4
1.1.3 E-AIM: model description . . . . . . . . . . . . . . . . . . . . . . 8
1.1.4 E-AIM and instrumental uncertainty . . . . . . . . . . . . . . . . 11
1.2 Experimental methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.2.1 Datasets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.2.2 CONTACT-2012 instrumentation . . . . . . . . . . . . . . . . . . 14
1.2.3 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.3.1 Particle-only E-AIM analysis for CONTACT-2012 . . . . . . . . . 22
1.3.2 Gas-particle partitioning E-AIM analysis for CONTACT-2012 . . 27
1.3.3 Ad hoc two mode separation from CalNex 2010 . . . . . . . . . . 35
1.3.4 24-hour integration analysis . . . . . . . . . . . . . . . . . . . . . 43
1.3.5 pH distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
1.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
1.5 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
1.6 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2 Interference in Nitrogen Oxide Monitors 57
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.2.1 Analytical instrumentation . . . . . . . . . . . . . . . . . . . . . . 60
iv
2.2.2 Converter temperature ramping with ammonia gas . . . . . . . . 62
2.2.3 Converter temperature ramping with salt particles . . . . . . . . . 63
2.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.3.1 Ammonia gas conversion efficiency trends in commercial nitrogen
dioxide analyzers . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.3.2 Ammonia gas conversion efficiency trends in total reactive nitrogen
oxide analyzer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
2.3.3 Salt conversion efficiency trends in total reactive nitrogen oxide
analyzer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
2.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
2.6 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
2.7 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
A AIM-IC calibration and instrumental information 75
A.1 Gradient eluent chromatographic programs . . . . . . . . . . . . . . . . . 75
A.2 Non-linear organic acids and ammonium . . . . . . . . . . . . . . . . . . 76
A.3 Significant linear inorganic species . . . . . . . . . . . . . . . . . . . . . . 78
A.4 Limits of detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
B E-AIM correlation data 81
v
List of Tables
2.1 Summary of selected literature gaseous NH3 MoOx-CLD conversion effi-
ciencies and associated Tconv . . . . . . . . . . . . . . . . . . . . . . . . . 60
A.1 Cation IC gradient eluent program set points. Flow rate is set to 1.0
mL/min and column temperature is set to 30. . . . . . . . . . . . . . . . 75
A.2 Anion IC gradient eluent program set points. Flow rate is set to 1 mL/min
and column temperature is set to 30 . . . . . . . . . . . . . . . . . . . . 76
A.3 Summary of calibration information for Acetic Acid from the CONTACT-
2012 field campaign for the AIM-IC . . . . . . . . . . . . . . . . . . . . . 76
A.4 Summary of calibration information for Formic Acid from the CONTACT-
2012 field campaign for the AIM-IC . . . . . . . . . . . . . . . . . . . . . 76
A.5 Summary of calibration information for Oxalic Acid from the CONTACT-
2012 field campaign for the AIM-IC . . . . . . . . . . . . . . . . . . . . . 77
A.6 Summary of calibration information for NH4+
from the CONTACT-2012
field campaign for the AIM-IC . . . . . . . . . . . . . . . . . . . . . . . . 77
A.7 Summary of calibration information for Cl-
from the CONTACT-2012 field
campaign for the AIM-IC . . . . . . . . . . . . . . . . . . . . . . . . . . 78
A.8 Summary of calibration information for NO2-
from the CONTACT-2012
field campaign for the AIM-IC . . . . . . . . . . . . . . . . . . . . . . . . 78
A.9 Summary of calibration information for NO3-
from the CONTACT-2012
field campaign for the AIM-IC . . . . . . . . . . . . . . . . . . . . . . . . 79
A.10 Summary of calibration information for SO42-
from the CONTACT-2012
field campaign for the AIM-IC . . . . . . . . . . . . . . . . . . . . . . . . 79
A.11 Summary of calibration information for SO42-
from the CONTACT-2012
field campaign for the AIM-IC . . . . . . . . . . . . . . . . . . . . . . . . 80
B.1 CONTACT-2012 gas-particle partitioning correlation data . . . . . . . . 81
vi
List of Figures
1.1 Diagram of E-AIM gas-particle partitioning model . . . . . . . . . . . . . 9
1.2 Parameterized particle study . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3 Layout of campaign field site at CARE . . . . . . . . . . . . . . . . . . . 16
1.4 Photo of campaign field site . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.5 Example calibration curve of particle-phase SO42-
. . . . . . . . . . . . . . 19
1.6 Example calibration curve of gas-phase acetic acid . . . . . . . . . . . . . 21
1.7 Case study of charge balance in CONTACT-2012 . . . . . . . . . . . . . 23
1.8 Particle-only study in situ pH trends in CONTACT-2012 . . . . . . . . . 24
1.9 pH and measured H+strong in CONTACT-2012 . . . . . . . . . . . . . . . . 25
1.10 Case study of particle-only pH in CONTACT-2012 . . . . . . . . . . . . 26
1.11 Time series of NH3(g) particle-only AIM-II predictions and measurements
during CONTACT-2012. . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.12 Relationship between in situ pH and the output H+strong . . . . . . . . . . 28
1.13 Diurnal in situ pH coloured for relative humidity in CONTACT-2012 with
gas-particle partitioning enabled . . . . . . . . . . . . . . . . . . . . . . . 29
1.14 Full time series of in situ pH coloured for relative humidity in CONTACT-
2012 with gas-particle partitioning enabled . . . . . . . . . . . . . . . . . 29
1.15 Case study of in situ pH estimated by disallowing and allowing gas-particle
partitioning and including organic acids with gas-particle partitition enabled
in E-AIM in CONTACT-2012 . . . . . . . . . . . . . . . . . . . . . . . . 30
1.16 Differences between partitioning allowed and disallowed in CONTACT-2012 31
1.17 Relationship between in situ pH and XH2O in the gas-particle partitioning
study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.18 Relationship between in situ pH and XH2O in the particle-only study . . 32
1.19 Partitioning enabled NH3 predictions . . . . . . . . . . . . . . . . . . . . 33
1.20 Case study of particle-only pH in CONTACT-2012 . . . . . . . . . . . . 35
1.21 AMS/PILS ratio in CalNex 2010 . . . . . . . . . . . . . . . . . . . . . . 37
vii
1.22 HNO3(g) predictions from optimized sea salt mode and accumulation mode
aerosols from CalNex 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . 38
1.23 NH3(g) predictions from optimized sea salt mode and accumulation mode
aerosols from CalNex 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . 38
1.24 HCl(g) predictions from optimized sea salt mode and accumulation mode
aerosols from CalNex 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . 39
1.25 NO3(aq)-
predictions from optimized sea salt mode and accumulation mode
aerosols and input and modelled ratio of sea salt to accumulation mode
from CalNex 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
1.26 NH4(aq)+
predictions from optimized sea salt mode and accumulation mode
aerosols and input and modelled ratio of sea salt to accumulation mode
from CalNex 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
1.27 Cl(aq)-
predictions from optimized sea salt mode and accumulation mode
aerosols and input and modelled ratio of sea salt to accumulation mode
from CalNex 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
1.28 Optimized pH values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
1.29 24-hour data integration methodologies from CONTACT-2012 . . . . . . 44
1.30 pH histogram of recent continental North American field campaigns . . . 45
2.1 Schematic of TSI NOx analyzers (adapted from TSI 42i Manual) . . . . . 61
2.2 Conversion efficiencies of NH3(g) in TSI NOx analyzers . . . . . . . . . . 64
2.3 NH3(g) profiles in AQD NOxy . . . . . . . . . . . . . . . . . . . . . . . . . 65
2.4 Conversion efficiencies of NH3(g) and HNO3(g) in AQD NOxy with simulta-
neous AIM-IC monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.5 Conversion efficiencies of nitrogen-containing salts in AQD NOxy with
simultaneous AIM-IC monitoring . . . . . . . . . . . . . . . . . . . . . . 67
B.1 Optimized separation comparison between modelled and input NO3(aq)-
. 81
B.2 Optimized separation comparison between modelled and input NH4(aq)+
. 82
B.3 Optimized separation comparison between modelled and input Cl(aq)-
. . . 82
viii
Chapter 1
Constraining Aerosol Acidity
through Ambient Gaseous
Measurements
1.1 Introduction
Atmospheric aerosols are small particles that exist either as suspended solids or liquids in
the gas phase. Aqueous aerosols often contain high concentrations of acids and bases, which
define the acidity of the aerosols. The implications of aerosol acidity include impacts on
both inorganic and organic aerosol formation and effects on the cardio-pulmonary system
and respiration in humans [1–3]. The relationship between aerosol formation, modification
and acidity has become an important area of research for atmospheric chemists in the
past decade as studies have demonstrated new methodologies of characterizing aerosol
acidity and its effects [4, 5].
Aerosol pollution has been known to impact public health, but research has also
investigated whether aerosol acidity exacerbates this effect. Several studies have attempted
to correlate public health data with ambient aerosol acidity. For example, Ostro et al. [3]
noted a significant positive correlation between aerosol acidity and self-reported medication
use and coughing in asthmatics. Another study by Gwynn et al. [6] noted a significant
relationship between periods of elevated aerosol acidity and respiratory hospital admissions
and mortality. Work by Mao et al. [7] determined that inflammation of mucous membranes
in school children was correlated with aerosol acidity. Dockery et al. [2] established that
children living in cities with high particulate acidity were significantly more likely to report
episodes of bronchitis than children living in areas with less acidic aerosols. Raizenne
1
Chapter 1. Constraining Aerosol Acidity 2
et al. [8] reported that the same populations that were observed by Dockery et al. [2]
were more likely to have reduced lung function, growth, and development if individuals
were chronically exposed to high acidity aerosols. As aerosol pollution throughout the
world remains a serious public health concern, understanding how acidity impacts human
health will likely continue to be a relevant area of research.
Another area of aerosol acidity research has focused on halogen activation through
acid-catalyzed pathways in aerosols as an important source of reactive halogen radicals in
the marine boundary layer [9–11]. Brimblecombe and Clegg [12] demonstrated that as
trace acids (e.g. HNO3 and H2SO4) titrate the alkilinity of fresh sea salt particles, the
Cl-
is displaced and volatilized as HCl. HCl can react with gas-phase OH to produce
Cl radicals, and these radicals are important for ozone production and destruction [13].
Recent research by Roberts et al. [14] has demonstrated that increased in situ acidity of
sea salt aerosols may increase the uptake of ClNO2 and subsequent yield of Cl2 gas upon
reaction with particle-phase Cl-. Cl2 may be photolyzed to form Cl radicals. Mozurkewich
[15] and Vogt et al. [16] proposed that dehalogenation of Cl2 and two other important
ozone related reactants, Br2 and BrCl, could occur through acid-catalyzed reactions
with between Cl-/Br
-and HOBr/HOCl. Keene et al. [17] utilized an aerosol chemistry
model and found that modelled aerosols with pH values of 3 and 5.5 produced higher
concentrations of gaseous Br2 and Cl2 than aerosols at pH 8. In order to understand the
extent to which aerosol acidity affects the production of ozone and the general composition
of the atmosphere, it is important to have the ability to produce accurate estimates of
aerosol acidity.
Formation and growth of secondary organic aerosol (SOA) through particle-phase
acid-catalyzed reactions has also driven recent research in aerosol acidity. SOA arises from
the condensation of oxidized volatile organic compounds (VOCs) emitted from the surface
and can comprise significant fractions of total particulate atmospheric organic carbon
[1, 18]. According to Hoyle et al. [1], current mechanisms for describing the partitioning
of atmospheric organics to aerosols does not adequately describe SOA formation without
considering the effects of aerosol acidity on condensed phase formation. Acid-catalyzed
reversible and irreversible reactions can significantly affect the expected growth of particles
that would otherwise be solely defined by thermodynamic partitioning and condensation
[19].
Characterizing aerosol acidity is challenging, and there are differing approaches to
defining and estimating aerosol acidity depending on the purpose of the metric. This work
will describe current methods of determining aerosol acidity as well as test the validity of
different methods with ambient data collected in recent field campaigns.
Chapter 1. Constraining Aerosol Acidity 3
1.1.1 Accumulation and coarse mode aerosols
Understanding the processes that generate aerosols is crucial to the characterization
of aerosol acidity. Between 25% and 75% of particulate matter less than 2.5 µm in
aerodynamic diameter (PM2.5) is comprised of inorganic salts which primarily consist
of Na+
, Cl-, SO4
2-, NO3
-, and NH4
+[20–23]. The remaining fractions are made up of
organics and crustal species. There are several different dominant particle size-composition
distribution regions (modes) in ambient aerosols including the ultrafine region, which
consists of particles < 0.01 µm in diameter (Dp), accumulation mode (0.01 µm < Dp <
2.5 µm), and coarse mode (Dp > 2.5 µm) [24]. Measurements of PM2.5, therefore, collect
ultrafine and accumulation mode particles and some fraction of the coarse mode due to
the overlap of the modes.
This work focuses on the inorganic component of coarse and accumulation mode
aerosols. Due to their small size, ultrafine aerosols have relatively low atmospheric mass
loadings compared to the larger mode particles. The inorganic component of accumulation
mode aerosols is dominated by SO42-
, NO3-, and NH4
+[20, 21, 25, 26]. Sea salt aerosols
(composed of mostly Na+
and Cl-) are a significant fraction of PM2.5 and make up the
majority of coarse mode aerosols in oceanic environments [27, 28].
Sea salt aerosols are produced through wave action and resulting sea spray [29–31]. As
the sea water droplets are lofted into the lower troposphere they reach equilibrium with
the ambient relative humidity and may decrease in size as the water partitions between
the gas and particle phases. Several modelling studies suggest that the seawater alkalinity
that is transferred to the aerosols is quickly titrated by atmospheric trace acids and results
in acidic aerosols [10, 11, 32].
Unlike sea salt aerosols, secondary aerosols in the accumulation mode (sometimes
referred to as sulphate aerosols) are produced through condensation of low volatility
molecular species in the atmosphere. H2SO4 gas is produced through oxidative reactions of
anthropogenic SO2 or biogenic reduced sulphur compounds like dimethylsulphide (DMS)
[33]. SO2 is produced primarily by coal power in northern mid-latitude regions [13, 33].
H2SO4 has a low vapour pressure and readily condenses to existing particles and may be
neutralized with gaseous NH3, which is produced primarily by agricultural emissions [13].
As the particles grow, they scavenge H2O and HNO3, an important atmospheric pollutant
that is produced in part through anthropogenic emissions [34]. Sulphate aerosols grow
throughout their lifetime but typically remain in the submicron Dp range [35].
In both particle types, the total water content of the particles is determined by the
ambient relative humidity, but also by the hygroscopicity of the constituent species [36].
Chapter 1. Constraining Aerosol Acidity 4
1.1.2 Methods of characterizing aerosol acidity
The ability to determine aerosol acidity has evolved over the past 30 years. There are
multiple definitions of aerosol acidity that rely on different instrumental and analytical
techniques. Some techniques have been more fully characterized than others, but the
difficulty in creating universal definitions is that each of the techniques has different
advantages and often the disadvantages are poorly understood. Truly direct measurements
are challenging due to the small size and non-ideality of chemical species in solvated
aerosols. Indirect methods that rely on modelling suffer from uncertainty of whether
theoretical estimations accurately represent the physical state of the aerosol.
The in situ aerosol acidity is defined as the pH of the aqueous phase and is calculated
by equation 1.1, where AH+ is defined as the activity of H
+in the aqueous phase. This
definition most accurately represents the chemical state of the aerosols, but it is only
since the advent of modelling and direct measurement techniques that this definition has
been employed [4, 5]
in situ pH = −log(AH+) (1.1)
Indirect measurements
In ambient environments, the acidity must be determined from sampled particles. Early
studies relied on the charge balance of the particulate phase which describes a metric
that this work will refer to as “strong acidity” (H+strong) [17, 37–40]. H
+strong is defined in
equation 1.2 for aerosols by subtracting the sum of the positively charged species (e.g.
NH4+
and Na+
) from the sum of negatively charged species (e.g. SO42-
, NO3-, and Cl
-)
[20, 38, 41].
H+strong = 2 ∗ [SO2−
4 ] + [NO−3 ] + [Cl−]− [NH+4 ]− [Na+] (1.2)
SO42-
, NO3-, etc. are the molar concentrations of the inorganic species per unit volume
of air (nmol m−3). The individual species are often characterized by ion chromatography
(IC) and may also be characterized by other analytical methods including aerosol mass
spectrometry (AMS) [20, 42, 43].
Assuming that all acids are fully deprotonated and bases fully protonated, the value of
H+strong defines what concentration of H
+or OH
-is necessary to maintain charge neutrality.
H+strong>0 signifies excess H
+relative to OH
-in the particle and that the acids in the
particles are not fully neutralized by the bases (primarily NH3). If H+strong is zero or
negative, the acids are considered to be neutralized [13, 44, 45]. This technique has many
Chapter 1. Constraining Aerosol Acidity 5
limitations and may not adequately describe the acidity of the particles. The technique
does not represent total concentration of free H+
in solvated aerosols, in part because
it ignores the equilibrium between H+
and HSO4-
at low pH. Even in the cases where
HSO4-
is not important, the strong acidity refers to the molar concentration of H+
per
unit volume of air, whereas the pH requires information of the activity of H+
in the
aqueous phase. Therefore, to use measurements of strong acidity to determine particle
pH, one must obtain information on the particle liquid water content and the activity
coefficient for H+
. Another problem is that there may be species contributing to the ion
charge balance that are not represented using the dominant acids and bases in equation
1.2. Work by Keene et al. [17] demonstrates that another limitation of the technique is
that the estimation of H+strong is highly reliant on homogeneity of the composition of the
population of particles. Keene et al. [17] further note that dynamic partitioning of volatile
species between sea salt and sulphate aerosols serves to buffer the H+strong between particles
of differing composition. As a result, bulk H+strong estimates may not effectively capture
differences in the chemistry of differing particle types. Despite these drawbacks, some
researchers still use this technique to characterize particle acidity, although it is more
common for this metric to be used when researchers probe broad atmospheric implications
or public health effects, rather than specific chemical pathways [6, 7, 46].
Another method of estimating the strong acidity of aerosols is to extract particle filters
in small volumes of water and perform acid or base titrations to determine the excess
H+
or OH-
in solution [37, 47, 48]. This technique also provides a concentration of H+
as
number of moles per volume of air sampled. A similar technique to this method utilizes a
pH probe rather than acid-base titration to estimate the H+
concentration [49–52]. The
U.S. Environmental Protection Agency (EPA) considers extraction of filtered particles
into 1 mL of water followed by pH probe analysis to be the standardized method for
estimating aerosol H+strong [53]. These techniques are significantly limited by the difference
in equilibrium concentrations of species in diluted particles from their initial concentrated
state. Keene [54] discusses the potential for orders of magnitude difference between the
concentration of free H+
as a result of the buffering effect of SO42-
and HSO4-
that changes
with dilution. This technique also neglects the activity of the constituents which would
be different when the particles are diluted into solutions many orders of magnitude higher
in water volume than the particles would naturally contain.
Another indirect estimation of acidity is defined as the degree of stoichiometric
neutralization of species in the particle phase, which is the ratio of measured NH4+
molar
concentrations (NH4 meas+
) to the molar concentration of NH4+
needed to fully neutralize
the acidic species in the aerosol (NH4 neu+
) [20]. While equation 1.2 relies on the difference
Chapter 1. Constraining Aerosol Acidity 6
between anions and cations, this approach uses a ratio, which may be more robust
when measurements are subject to systematic errors, or are close to the detection limit.
Equation 1.3 shows the relationship between acids and bases. The inorganic species on
the right side of the equation were again defined as their respective molar concentrations
(per volume of air).
NH4+meas
NH4+neu
=[NH+
4 ]
2 ∗ [SO2−4 ] + [NO−3 ] + [Cl−]
(1.3)
Direct measurements
Recent work by Li and Jang [5] characterizes a novel method of evaluating in situ aerosol
acidity using colorimetry reflectance UV-visible (C-RUV) spectroscopy. The technique
consists of a sampling pump connected to an indicator-dyed filter. Ambient air is pulled
through the filter and a UV-visible spectrum is taken of the filter. Acidic particles
react with the indicator and produce a signal proportional to the H+
concentration
and mass loading of inorganic particles and inversely proportional to relative humidity.
This technique is advantageous over extraction-based filtration methods because it offers
shortened analysis times, although low mass loadings require longer integration times. It
also is the most direct method available for measuring acidity as it requires no dilution.
An intercomparison study between modelled predictions and results from C-RUV showed
very close correlation across a range of relative humidities. The C-RUV technique is reliant
on modelling to account for water content of the particles and has not been thoroughly
assessed in aerosols with low H+strong. This is a promising new technique, although it
currently is not capable of online sampling or complete independence from modelled
results.
Modelling
In order to probe particle-phase processes and estimate in situ pH, research has focused on
modelling as a methodology for describing the thermodynamic chemical state of aerosols.
Works published by Keene et al. [17] and Keene and Savoie [55] describe an early
methodology which used modelling to estimate aerosol acidity. Keene et al. [17] discusses
using a box model called the Model of Chemistry Considering Aerosols (MOCCA) for
marine boundary layer aerosols. In the study, the model was run five times under different
input conditions with arbitrary steady-state atmospheric concentrations of ambient species.
While this method is not useful for the application of sampled particle data, it did establish
the possibility of using thermodynamically derived data to estimate aerosol pH. Keene
Chapter 1. Constraining Aerosol Acidity 7
et al. [17] then developed a model for characterizing aerosol pH using measured gas
and particle-phase concentrations. This system relied on the gas-particle partitioning
equilibria of volatile species. The authors rationalized that, given Henry’s law constants
(KH) and acid-base dissociation constants (Kd) of the species in solution, an expression
could be derived to calculate the pH required for a certain particle-phase concentration of
a volatile species. The authors derived aerosol pH by the set of equations 1.4.
AHCl(aq) = KHHCl∗ [HCl(g)] (1.4a)
AHCl(aq) =AH+
(aq)∗ ACl−
(aq)
KaHCl
(1.4b)
ACl−(aq)
=[Cl−(aq)] ∗ γCl−
LWC(1.4c)
By rearranging the above equations, the following AH+ may be calculated
AH+(aq)
=KaHCl
∗KHHCl∗ [HCl(g)]
[Cl−(aq)
]∗γCl−
LWC
(1.4d)
Equation 1.4a relates the atmospheric concentration of gaseous HCl ([HCl(g)]) to the
aqueous activity of HCl (AHCl(aq)) with the Henry’s law constant (KHHCl). Equation
1.4b shows the relationship between AHCl(aq) and the activities of H+(aq) and Cl−(aq) with
the acid dissociation constant (KaHCl). Finally equation 1.4c describes the relationship
between ACl−(aq)
, the activity coefficient (γCl−), measured Cl−(aq) concentration ([Cl−(aq)],
in mol m−3) and the liquid water content (LWC). Liquid water content was estimated
by taking the volume difference between median measured “wet” aerodynamic particle
diameter and the theoretical diameter of a spherical dry salt particle with an assumed
density. The in situ pH was then calculated from AH+(aq)
using equation 1.1. A similar
derivation can be made for HNO3(g)/NO3(aq)-
and other volatile constituents of particles
that participate in gas-particle partitioning.
Other, more complex, modelling methodologies also rely on measurements of particulate
and sometimes gaseous species; the models ISORROPIA, SCAPE2 and E-AIM have all
been used to produce estimations of aerosol acidity [4, 36, 56, 57]. Work by Yao et al.
[4] compares results from the three previously mentioned thermodynamic models using
ambient data. The first method uses E-AIM to determine in situ pH using only particle-
phase thermodynamics without considering gas-particle partitioning. The second approach
was similar to Keene and Savoie [55] and used the relative measured concentrations of
Chapter 1. Constraining Aerosol Acidity 8
particulate and gaseous NH4+
/NH3, NO3-/HNO3, and Cl
-/HCl to calculate the required
pH for a given ratio to occur. The third approach used the thermodynamic gas-particle
partitioning models, ISORROPIA and SCAPE2, which rely on total gas and particle
concentrations of species which are then repartitioned according to Henry’s law and
dissociation constants. The study concluded that E-AIM produced superior results to
those of the other two approaches, since E-AIM alone was able to reproduce the measured
H+strong [4].
Recent research involving aerosol acidity has focused on using E-AIM as the preferred
estimation technique [4, 20, 44, 45, 58–60]. Input to E-AIM can be restricted to particle
composition only (i.e. the approach used in Yao et al. [4]) or allow both gas and particle
constituents and treat the particles as being in dynamic equilibrium with the gas-phase.
Many current studies have followed the recommendation of Yao et al. [4] and model
aerosol acidity with gas-particle partitioning disabled in E-AIM [20, 44, 45, 58–60]. The
validity of this methodology has not been thoroughly established, and this study will
examine this recommendation with modelled results from different regimes of the E-AIM
model compared with measured data.
In this study, E-AIM was implemented with data taken from recent field campaigns
in order to understand the biases and effects of uncertainty on the different permutations
of the model’s operating parameters. The goal of this research is to examine the quality
of the predictive power of the model with regards to acidity. E-AIM was chosen over
other models because of its widespread usage, flexibility, and global free energy solution
rather than the computationally less expensive, assumption-based methodologies of other
models.
1.1.3 E-AIM: model description
The E-AIM model finds a global minimum free energy state that is defined by the
chemical state of the system. The system involves the dissolution of solid salts, gas-
particle partitioning, acid-base dissociations, and partitioning between the aqueous particle
phase and an organic particle phase. The system may be modified to exclude certain
equilibria depending on assumptions made regarding the nature of the chemical state. In
our study, all modelling analysis of aerosol acidity was performed with E-AIM. The model
has several versions and in this study both the second (AIM-II) and fourth (AIM-IV)
were used [36, 61, 62]. AIM-II processes user defined organic and standardized inorganic
ions including particle-phase H+
, NH4+
, NO3-, and SO4
2-and gaseous organic acids and
inorganic NH3 and HNO3. AIM-IV includes particulate Na+
, Cl-, and gaseous HCl in
Chapter 1. Constraining Aerosol Acidity 9
addition to the aforementioned species in AIM-II. Relative humidity and temperature are
also included as inputs in both model versions, although AIM-IV is limited to inputs with
high relative humidity (>60%) when all available input species are utilized. The model
then partitions the species using the aformentioned thermodynamic equilibria to produce
moles, mole fractions, and activity coefficients of particulate and gaseous species. The
model also estimates liquid water content based on the hygroscopicity of the particles
and finds activities of all chemical species, including dissociated and undissociated acids
and bases. The in situ pH is estimated from the outputs of the model through equation
1.5. [20, 36].
pH = −log(fH+(aq) ∗ xH
+(aq)) (1.5)
fH+(aq) is the mole fraction-based activity coefficient of aqueous H+, xH+
(aq) is the
mole fraction of aqueous H+ in the particle as predicted by E-AIM. A simple diagram
of the gas-particle partitioning framework is depicted in Fig. 1.1; aqueous/hydrophobic
phase interaction and solid formation are not depicted.
NH3 NH4
+
Aqueous Particle
HNO3 HNO3
HCl HCl
Gas Phase
H2SO4
HSO4- SO4
2-
NH3
Cl-
NO3-
H2SO4
RH
RH (organic acid)
KH
KH KH
KH
KH
Ka
Ka
Ka Ka
Kb
R- Ka
Figure 1.1: Diagram of E-AIM gas-particle partitioning model
All models currently rely on a number of assumptions that vary slightly between
each. In E-AIM there are three major assumptions that will be addressed below. These
Chapter 1. Constraining Aerosol Acidity 10
assumptions highlight the difficulty with modelling aerosol acidity. The first assumption
is that aerosols are internally mixed, which means that composition among the analyzed
particles is uniform. The second assumption is that the particles are at or near equilibrium
with surrounding ambient air. Finally, in our use of the model, the particles were assumed
to be deliquesced (i.e. the aerosols were present as aqueous solutions). An organic particle
phase was not incorporated into the model as the data required for such an analysis was
beyond the scope of the available instrumentation. However, ignoring the organic phase
could impact the hygroscopicity and equilibrium relationships of particulate constituents.
Assumptions relating to size, surface tension, particle phases and others were not addressed
in this work, but are considerations which are ongoing areas of research.
The first assumption is that the aerosols are internally, rather than externally mixed.
Internal mixing implies that the composition of the particles is homogeneous whereas
external mixing indicates that the particles are not uniform in composition. This as-
sumption will be discussed in further detail later in this work. Essentially, E-AIM takes
bulk concentration inputs and repartitions the species to produce uniform composition
particles. If particles are known to not be uniform, segregation of particle types must
occur before the data are input into the model. This assumption is of interest because
ambient particle populations are often externally mixed and research has shown that
particles of differing composition tend to become more uniform as they age due to the
equilibrium dynamics among particles through the gas-phase [10, 55].
Work by Meng and Seinfeld [63] discusses the assumption of equilibration time as it
relates to marine and continental aerosols. Submicron aerosols equilibrate very quickly
but larger aerosols may never reach equilibrium and rather reside in nonequilibrium
states [63]. Dassios and Pandis [64] estimate that the equilibration time for submicron
NH4NO3 particles to be on the order of minutes. SO4(aq)2-
and H2SO4(g) are less important
for gas-particle equilibrium considerations than the TNO3 (HNO3(g) and NO-3(aq)) and
TNHx (NH3(g) and NH+4(aq)) because the vapour pressure of H2SO4(g) is very low and
does not partition significantly into the gas phase. Similarly, Ca2+
and Na+
do not
partition significantly into the gas phase. Followup work by Cruz et al. [65] observed a
relatively small decrease in the rate of equilibration due to organic films on dry, inorganic
aerosols. However, work by Shiraiwa et al. [66] suggests that organic coatings may have
a more significant effect than previously thought. Sun and Wexler [67] established that
the equilibration time for H+
is short enough that in systems that are at near acid-
base neutrality, the particulate H+
concentration can be assumed to be in equilibrium
with respect to ambient gaseous species. Work by Zhang et al. [68], determined that the
assumption of thermodynamic equilibrium was met reasonably well over 5 minute sampling
Chapter 1. Constraining Aerosol Acidity 11
intervals using comparisons of ISORROPIA and experimental data. However, Yao et al.
[4] determined that based on comparisons between the predicted gas concentrations from
the particle-only regime of E-AIM and measured results, the system was not always at
equilibrium.
Finally, the model requires users to decide whether or not to assume that the aerosols
are deliquesced. The deliquescence relative humidity (DRH) is the ambient relative
humidity required for a given solid salt to transition from the solid phase to the aqueous
phase. In cases where the ambient relative humidity is higher than the DRH, the model
assumes that a given species will be deliquesced. However, in cases where the DRH is
higher than the ambient RH, the model provides users with two options. The first option
is to allow the model to leave the salt as a solid while the other allows the particle to exist
in a metastable deliquesced state. The latter option makes the implicit assumption that
the ambient particle had at one point been deliquesced, and that due to the hysteresis
of solubilized particles, the particle remained in the aqueous phase even after the RH
dropped below the DRH. The efflorescence RH (ERH), the RH at which a deliquesced salt
returns to the solid phase, is much lower than the DRH [69]. Furthermore, theoretical and
experimental evidence shows that the DRH of salt mixtures is lower than the DRH of any
of the constituent salts [23, 70, 71]. It is often operationally impossible to determine if the
particles are deliquesced and researchers must acknowledge the limitation of the model
with regards to this assumption. In this work, particles were assumed to be deliquesced
due to the high ambient relative humidities taken throughout the datasets mentioned in
the following section.
The most significant problem with modelling aerosol acidity is that there is no certainty
as to whether the estimates truly correspond with the actual state of the system. If any
of the above assumptions are not met, the model will produce unphysical results, but it
may not be possible to identify which of the assumptions has not be met.
1.1.4 E-AIM and instrumental uncertainty
When gas-particle partitioning is disallowed in E-AIM (called the particle-only mode
of E-AIM), the estimated in situ pH is driven by the balance between acids and bases
in the bulk composition measurements. Particles with excess acid will produce low pH
values, while particles with excess base will produce high pH values. However, since
ambient aerosols are often at or near the equivalence point (i.e. H+strong=0), relatively
small instrumental errors in the measurements of indivual anions and cations can result
in pH values that lie on either side of the pH titration curve.
Chapter 1. Constraining Aerosol Acidity 12
Although the particle-only mode of E-AIM has been widely used for understanding
aerosol acidity, work by Zhou et al. [44] illustrates the limitations of current acidity
estimations using this technique. Zhou et al. [44] collected hourly measurements of
particle-phase soluble inorganics. Approximately one third of their dataset showed
negative H+strong (i.e. excess NH4
+), and these datapoints were rejected for analysis due to
the authors’ assertion that the model was unable to produce valid output data. Work
by Behera et al. [45] utilized a similar rationale to Zhou et al. [44] for discarding 2/3 of
their data set before processing in E-AIM. Behera et al. [45] argued that when H+strong<0,
the aerosols may be considered neutralized and not considered acidic. When there is an
excess of positively charged species in solution (i.e. H+strong<0), E-AIM tends to produce
results that suggest the particle phase has high concentrations of neutral NH3 in solution.
However, based on Henry’s law, the amount of uncharged NH3 predicted by the model
would not be expected if the NH3 were allowed to partition into the gas phase. Datapoints
with excess acids present a less significant problem for the model because the vapour
pressure of undissociated sulphuric acid is low and the sulphate/bisulphate system acts
as a buffer for free H+
. The average particle-only E-AIM in situ pH of non-neutralized
aerosols in numerous studies has been predicted to be below 2.5 [44, 45, 58, 72]. Since
these studies do not account for neutralized particle data, these results may not represent
the true acidity of the particles due to the limitations of the particle-only regime of
E-AIM.
One reason why researchers might observe aerosols where H+strong is less than zero
is due to instrumental uncertainty. To illustrate the significance of uncertainty that
arises in particle-only E-AIM analysis, a representative parameterized particulate system
was formulated and processed in the particle-only version of E-AIM. In this system the
balance of NH4+
and the sum of NO3-
and SO42-
was allowed to vary from an H+
deficit
to surplus. The modeled activity coefficient and mole fraction of H+
were input into
equation 1.5 to estimate pH. The results of the study are displayed as pH versus H+strong
in Fig. 1.2. As H+strong approaches zero, indicating neutralization, the pH passes through
the titration equivalence point. An absolute concentration difference between the sum of
the acids and bases of 1.0 neq m−3 could result in a difference of free H+
concentration of
approximately 8 orders of magnitude. Given that studies often find ambient aerosols near
charge neutrality, uncertainty in measurements could be strongly affecting the in situ pH
estimates from the particle-only version of E-AIM [20, 44, 45].
This work will endeavor to elucidate the problems with the predictive ability of the
model and provide evidence for why certain modelling methodologies are susceptible to
producing model outputs that are not representative of the measurements. This work will
Chapter 1. Constraining Aerosol Acidity 13
10
8
6
4
pH
-2 -1 0 1 2
H+
strong (neq m-3
)
Figure 1.2: Parameterized particle with constant SO42-
(2.5 neq mol−1) and NO3-
(5.0 neqmol−1) with linearly increasing NH4
+(8.0-12.0 neq mol−1)
.
also examine the assumption that the particle-only version of E-AIM is the most accurate
representation of the particle phase by investigating whether gas-particle partitioning can
improve the in situ pH estimations. Several studies have suggested that the pH of aerosols
is highly buffered by the gas-particle partitioning effects of expulsion of neutralized volatile
species, [10, 55]. This study will attempt to explore the validity of modelling regimes that
do or do not support this hypothesis.
1.2 Experimental methods
1.2.1 Datasets
The first dataset used in this work was collected during the Characterizing Ontario
Nitrogen Transport and Chemical Transformation (CONTACT-2012) field campaign and
is described in detail in the following subsection of this chapter. The site was located
at the Centre for Atmospheric Research Experiments (CARE, 4413’51”N 7946’58”W)
Chapter 1. Constraining Aerosol Acidity 14
which is an Environment Canada research site near Egbert, ON.
The second dataset used in this work was collected during the California Research at the
Nexus of Air Quality and Climate Change (CalNex 2010) at the Pasadena, CA, ground site
(348’16”N, 1187’34”W). A compilation of research papers relating to this study may be
found at http://www.esrl.noaa.gov/csd/projects/calnex/papers/. For the compiled data
in this work, a quantum cascade tunable infrared laser diode absorbance spectrometer
(QC-TILDAS) provided ambient NH3 mixing ratios, particle into liquid sampler ion
chromatography (PILS-IC) provided PM2.5 mass loadings of inorganic anions and cations,
AMS provided inorganic anion and cation mass loadings of submicron particulate matter,
and negative ion proton transfer chemical ionization mass spectrometry (NI-PT-CIMS)
provided mixing ratios of gaseous inorganic acids. Meteorological data, including relative
humidity and temperature, was also provided from weather station monitoring. Although
certain instruments were run extensively for weeks during the campaign, only 78 hours of
semi-continuous hourly-averaged data were processed for this study due to instrumental
downtime.
1.2.2 CONTACT-2012 instrumentation
The CONTACT-2012 field campaign relied on instrumentation from the Murphy Research
Group at the University of Toronto, a mobile atmospheric measurement laboratory (the
Mobile Analysis of ParticuLates in the Environment (MAPLE)) from the Southern Ontario
Centre for Atmospheric Aerosol Research (SOCAAR) for transportation and instrumental
housing, and publicly available meteorological data provided by Environment Canada.
The instrumentation included ambient ion monitor ion chromatography (AIM-IC), a
multiple uniform orifice deposition impactor (MOUDI), a sonic anemometer, and QC-
TILDAS, but only AIM-IC data will be discussed in this work. Data were collected from
July 19th, 2012 until October 2nd, 2012.
AIM-IC is a sample analysis and collection system developed by University Research
Glassware Corporation (URG Corp, Chapel Hill, NC). The AIM-IC Model 9000D simul-
taneously measures water-soluble particulate matter and trace gases with hourly time
resolution. The operational details of the instrument are described below and closely
follow those described by Markovic et al. [73]. The particulate species quantified during
the campaign were SO42-
, NO3-, NO2
-, Cl
-, PO4
3-, Li
+, Na
+, NH4
+, K
+, Mg
2+, Ca
2+, acetate,
formate, and oxalate. The trace gases were SO2, HNO3, HONO, HCl, H3PO4, NH3, acetic
acid, formic acid, and oxalic acid.
The modified AIM-IC 9000D in this study consists of three separate components: inlet
Chapter 1. Constraining Aerosol Acidity 15
system, control box, and ion chromatographs. The inlet draws ambient air at 3.0 L min−1
through an impactor with a size cutoff of 2.5 µm (PM2.5). All debris and particulate
matter larger than PM2.5 is deposited on the impactor frit which was replaced periodically
throughout the campaign to minimize offgassing of volatile constituents that had impacted
on the frit. Prior to the impactor, there is a raincap which blocks precipitation from
entering the inlet. Air is then drawn into a continually renewed parallel plate wet denuder
(PPWD) which dissolves gaseous acids and bases and oxidizes SO2 to SO2−4 . The remaining
flow, which contains non-soluble gases and the atmospheric particles, enters the denuder
into the particle super saturation chamber (PSSC) and the particles hygroscopically grow
and are collected in a cyclone. The remaining air is drawn through a diaphragm pump
and exits the system as exhaust.
The dissolved species from the denuder and PSSC are drawn through 19.2 m of Teflon R©
(FEP) tubing to the control box by two respective sets of two 5 mL gastight syringes
(Hamilton Co, Reno, NV). The liquid from the denuder and PSSC first passes through
two respective 6-port injection valves (Rheodyne LLC, Rhonert Park, CA) before entering
the gastight syringes. The syringes pull liquid for 55 minutes per cycle. The two valves
switch configuration when the syringes are filled to allow injection to four separate 6-port
injection valves (i.e. one valve for each syringe). Two of these valves allow the flow to be
loaded onto two concentrator columns for the gas (denuder) channel, and the other two
valves allow the flow to be loaded onto two concentrator columns for the particle (PSSC)
channel. The injections of the gastight syringes into the four concentrator columns takes
5 minutes; the total collection and injection cycle of the gastight syringes takes one hour.
The system contains one IC which is set up to detect anionic samples and the other for
cationic samples. The particle analytes in the concentrator columns are injected onto the
chromatographic columns of the anion and cation ICs at the beginning of each hour. The
particle samples run for 26 minutes and at 30 minutes after the hour, the gas samples are
injected onto the chromatographic columns and also run for 26 minutes. The total cycle
for the ICs to run both gas and particle samples is one hour to maintain the synchronicity
with the control box gastight syringe cycle.
The denuder solution used in the field campaign contains 2 mM H2O2 in deionized
water (>18 MΩ, DIW, Barnstead, Dubuque, IA and ELGA, High Wycombe, UK), which
was optimized by finding the lowest concentration of H2O2 to oxidize SO2 to SO42-
for
reasonable ambient concentrations.
In this campaign, the PPWD utilized two parallel nylon denuder membranes. Previous
research performed by Markovic et al. [73] demonstrated the dramatic improvement in
response time of nylon membranes to NH3 compared with cellulose membranes. Nylon
Chapter 1. Constraining Aerosol Acidity 16
QC-TILDAS and Sonic
Anemometer
MAPLE
AIM-IC Inlet
MOUDI
27.5 m
18 m 6.5 m
14.5 m
Figure 1.3: Layout of campaign field site at CARE
membranes also have longer lifetimes and do not need the frequent replacements that
plagued cellulose membranes. The denuder configuration was implemented such that the
denuder solution entered the bottom of the first membrane plate, rose to the top, passed
through a short piece of tubing to the bottom of the second plate, and finally rose to the
top of the second membrane plate and exited the denuder toward the control box. The
denuder setup is an example of countercurrent exchange that is designed to maximize the
diffusion gradient as air flows through the system.
While the PC, control box, and ICs were installed in MAPLE, the inlet was mounted
on a 3 m tower located 6.5 m from MAPLE. The entrance to the inlet was located 2.75
m above ground. The position of the inlet with relation to other instrumental inlets and
MAPLE is shown below in Fig. 1.3. A photo of the AIM-IC inlet and MAPLE is included
in Fig. 1.4.
Chapter 1. Constraining Aerosol Acidity 17
Figure 1.4: Photo of campaign field site at CARE. The inlet is housed in the aluminumbox in the foreground and MAPLE is in the background
Chapter 1. Constraining Aerosol Acidity 18
The two IC systems used for the AIM-IC were Dionex ICS-2000 (Dionex Inc. Sunnyvale,
CA). Both ICs interface with a PC with the Dionex Chromeleon R© 6.8 software that allows
software commands to be transmitted to the ICs and live recording of the chromatographic
runs. Appendix A lists the specific chromatographic componentry used throughout
CONTACT-2012.
The flow rate, column temperature, and gradient eluent programming of the IC
systems were optimized prior to deployment during the campaign. The gradient eluent
program for both systems is tabulated in Table A.1 and Table A.2. The maximum eluent
concentration and eluent flow rate defined by the eluent program is also necessary to
choose the suppressor current. According to the SRS-300 Dionex manual, the suppressor
current is calculated in equation 1.6.
Current (mA) = flow rate (mL min−1) ∗max. eluent conc. (mN) ∗ suppressor factor(1.6)
The suppressor factor for the ASRS-300=2.47 and CSRS-300=2.94
The meteorological data was provided by the Environment Canada, National Climate
Data and Information Archive which is found at climate.weatheroffice.gc.ca. The temper-
ature, relative humidity, and atmospheric pressure are recorded as hourly averages and
posted shortly after acquisition. Since the research site for CONTACT-2012 was at an
Environment Canada research facility, the climate station was located in the immediate
vicinity (<500 m) of the chemical instrumentation.
1.2.3 Data analysis
The methodology for calibrations for AIM-IC are discussed in Markovic et al. [73] and
VandenBoer et al. [74]. Calibrations were run three times on the AIM-IC during the
course of the campaign. The standards were prepared by serial dilution from the inorganic
Dionex 6-cation IC standard and 7-anion IC standard (Sigma-Aldrich, St. Louis, MO).
The 6-cation standard contains Li+, Na+, NH+4 , K+, Mg2+, and Ca2+ while the 7-anion
standard contains F−, Cl−, NO−2 , Br−, NO−3 , SO2−4 , and PO3−
4 (in order of chromatographic
elution). Organic standards were not included in the multiple inorganic standards and were
instead produced from single standards (Sigma-Aldrich, St. Louis, MO) of each. Acetic
acid/acetate, formic acid/formate, and oxalic acid/oxalate were detected and calibrated
for at the same intervals as the inorganic standards. IC dilutions were carefully performed
using powder-free latex gloves and micropipettes with disposable tips. The standard vials
and volumetric flasks were all made from ETFE to reduce risk of contamination. Before
Chapter 1. Constraining Aerosol Acidity 19
12
10
8
6
4
2
0
Pea
k A
rea
(µS
min
)
14x10-9121086420
Amount Injected (mol)
Coefficient values ± one standard deviation
slope =7.87*108 ± 1.1*10
7
Figure 1.5: Example calibration curve of particle-phase SO42-
usage, the vials and flasks were thoroughly rinsed eight times and dried by mechanical
action (tapping). All steps required to reduce contamination were regularly performed
including working on clean surfaces, ensuring the DIW was of high purity (>18 MΩ), and
that all injection syringes and transfer pipettes were thoroughly cleaned.
The ICs were calibrated with the following offline methodology. The lines from the
control box sample syringes were disconnected from the 6-port injection valves. The
standards and blanks were injected into the vacated ports with 5 mL gastight syringes
which were identical to the control box sample syringes. 5 mL of each standard was
injected at the rate of 1 mL min−1 in order to mimic the conditions of the hourly sample
syringe injections. The chromatographic program that was used for the online AIM-IC
sampling was used for offline calibrations. Blanks were injected prior to the standards
and after the standards in order to ensure minimal background interference. The peak
areas of the standards were quantified by manual integration in Chromeleon R©.
The calibration slopes for most species were linear. The slopes of the linear calibration
curves were determined using linear regression in IgorPro 6. Weak acids and bases tend
to have non-linear calibration slopes as a result of acid-base equilibria in the conductivity
detector, as described by VandenBoer et al. [74]. Calibration for non-linear analytes is
discussed below. The linear analytes were Li+
, Na+
, K+
, Mg2+
, Ca2+
, F-, Cl
-, NO2
-, Br
-,
NO3-, SO4
2-, and PO4
3-. An example of the offline calibration curve from SO4
2-is plotted in
Fig 1.5.
Chapter 1. Constraining Aerosol Acidity 20
Since the online system may change the background signal detected by the ICs, an
online background was taken by overflowing zero air (Praxair, Mississauga, ON) directly
into the denuder for 24 hours while instrument was in online operation. The average of
this measurement was taken as the online background since it accounted for impurities
and interferences from the denuder membrane, denuder solution, or other impurities in
the lines or instrumentation. The average peak area from the online background was
subtracted from each ambient data point before using the calibration slope to calculate the
moles of analyte. The limit of detection was calculated by using the standard deviation
of background measurement with equation 1.7. The limits of detection for all quantified
species are tabulated in Table A.11. The collection efficiency of the online system must
also be taken into account when converting peak area values into concentrations. Cellulose
membranes are known to have collection efficiencies at or around 90%, and preliminary
work from the our research group and URG [75] indicates that nylon membranes also
have very high collection efficiency (>99% for SO2), so an assumption of 100% was used
in this work [73].
Limit of Detection = 3 ∗ σbackground (1.7)
While the calculation of concentration from peak area was well established for the linear
species, determination of concentration of weak acids and bases was more complicated.
Work by VandenBoer et al. [74] established a methodology for quantifying weak acids and
bases which was implemented during CONTACT-2012. While it is possible to fit a curve
to the calibration data with exponential functions supplied by IgorPro, VandenBoer et al.
[74] showed that, based on the acid dissociation constants of the species, the physical
process is more accurately represented with a quadratic function and is briefly discussed
below.
The AIM-IC conductivity detector is only able to detect ions, but with weak acids
and bases, the protonated acids or deprotonated bases are invisible to the detector. After
the eluent is removed by the suppressor, weak acids and bases are often concentrated
enough to prevent either full deprotonation (for acids) or protonation (for bases) which
results in neutralized molecules. Equation 1.8a shows the derived relationship between
ntotal analyte (the sum of the number of moles of detected ion and undetected uncharged
species) and nanalyte ion (the detected analyte ion). In equation 1.8a, v is a scalar to
convert between nanalyte ion and the detected peak area. A = KD
2and B = KD, where KD
is the acid or base dissociation constant. volume is the theoretical volume of solution that
would result in a given peak area of a homogeneous solution of analyte in the detector.
Equation 1.8b simplifies the equation for practical usage with IgorPro’s curve fitting
Chapter 1. Constraining Aerosol Acidity 21
12
10
8
6
4
2
0
Pea
k A
rea
(µS
min
)
200x10-9150100500
Amount Injected (mol)
Coefficient values ± one standard deviationA'=2.37 ± 0.82
B'=7.03*108 ± 9.6*10
7
Figure 1.6: Example calibration curve of gas-phase acetic acid
application. In equation 1.8b A′ is the product of A and v while B′ is the product of B, v,
and the reciprocal of volume. Equation 1.8b was used for organic acid calibrations while
equation 1.8a was used for the ammonium calibrations. Equation 1.8a is advantageous
in instances where a single estimated volume can be constrained as a constant between
multiple calibrations and result in reasonable curve fitting, because it contains a single
variable rather than two in equation 1.8b. An example of the organic acid calibrations is
shown is Fig. 1.6.
Peak area = v ∗ nanalyte ion = v ∗[(−A) +
√A2 +B ∗ ntotal analyte
volume
](1.8a)
Peak area = (−A′) +√A′2 +B′ ∗ ntotal analyte (1.8b)
One difficulty with non-linear calibrations that may increase the uncertainty of con-
centration values is caused by differences between the calibration methodology and actual
ambient sampling. The methodology assumes that the background signal of the standards
(i.e. the DIW signal) is the same as the background signal of the ambient samples taken
with zero air flowing into the inlet. However, comparing the background signals of DIW
injections during calibration and zero-air samples for the ambient background reveals
substantial differences. Non-linear calibrations are complicated since the subtraction of a
single background value from each of the standard points on the curve improperly implies
that the background signal is constant at all concentrations. For the purposes of this
Chapter 1. Constraining Aerosol Acidity 22
study, since both background signals were relatively small for NH3 and NH4+
, the impact
of the non-linear background effect was minimal. The organic acids had significantly
higher zero air background peak areas by about one order of magnitude, and thus the
calibrations may be less reliable than those of other species.
Three calibrations throughout the campaign served to constrain the degradation of the
concentrators over time. Unlike simple sample loops, concentrators have a limited number
of binding sites which degrade over time and thus change the ability of the concentrators
to retain analytes. Between each calibration run, the percent difference between slopes
of linear species and the differences in curve shape were evaluated to establish if the
concentrators were failing. After the second calibration, the anion gas concentrator was
replaced and calibrated prior to usage for the remainder of the campaign. In order to
account for the variation in calibration curves, the calibration curve data of the linear
species and NH3 were linearly interpolated between calibration runs and concentration
values were calculated using the interpolated calibrations. The concentration values for
non-linear species were calculated using each of the calibration curves and then weighting
the concentration values depending on the sample’s position between two calibration runs.
For example, an online concentration value taken a quarter of the way between the first
and second calibration would be composed of 75% of the concentration calculated from the
first calibration and 25% of the concentration calculated from the second calibration. The
difference between the two interpolating methods was necessitated by the way non-linear
regression program allowed A′ and B′ to vary.
The overall relative uncertainty for all species in this analysis was estimated as 15%
and the limit of detection for each respective quantified species. Markovic et al. [73]
and Zhou et al. [44] estimated the relative uncertainty of AIM-IC istrumentation to
be approximately 10%, but work by this research group indicates that the uncertainty
may be somewhat higher. Further work is needed to ensure the best represenatation of
instrumental uncertainty with this technique.
A summary of calibration information, including limits of detection, is provided in
Appendix A.
1.3 Results and discussion
1.3.1 Particle-only E-AIM analysis for CONTACT-2012
The composition of the particles collected during the CONTACT-2012 campaign consisted
mostly of SO42-
, NH4+
, and NO3-. H
+strong was calculated for the CONTACT-2012 dataset
Chapter 1. Constraining Aerosol Acidity 23
using the mole equivalents of the dominant particulate species throughout the campaign. A
case study of charge balance in the campaign shown in Fig. 1.7 illustrates the comparison
between the near equivalent concentration of anions and cations in the aerosols. At certain
times, there are excess cations, but much of the campaign features excess NH4+
. The
average H+strong for the entire campaign was -3.84±5.80 nmol m−3.
200
150
100
50
0
Con
cent
ratio
n (n
eq m
-3)
3:00 PM8/24/2012
6:00 AM8/25/2012
9:00 PM 12:00 PM8/26/2012
3:00 AM8/27/2012
6:00 PM 9:00 AM8/28/2012
Date
4020
0
H+
stro
ng (
nmol
m-3
)
Cl-
NO3-
SO42-
NH4+
Figure 1.7: Case study of charge balance in CONTACT-2012.
For CONTACT-2012 data analysis, AIM-II was used to obtain the mole fraction and
activity coefficient of H+
required for calculated in situ pH as described in equation 1.5.
The particles were assumed to be internally mixed and consist of only one deliquesced,
aqueous phase. Since the field campaign found only low levels of HCl/Cl-, Na
+and other
detectable inorganic ions, it was unnecessary to utilize the more inflexible AIM-IV. In
order to obtain the activity coefficient and mole fraction of H+
needed for estimating pH,
concentrations of particulate NO3-, SO4
2-, and NH4
+and, in some cases, gaseous NH3 and
HNO3 were input into the batch mode of AIM-II. Any of the hourly data that did not
contain all of the essential measurements was removed from the finalized dataset that was
input into the model. Negative concentration data points below the limit of detection
were set to zero to avoid model errors. A small number of other hourly measurements
were discarded due to the concentrations of individual species exceeding the parameters
of the model.
Chapter 1. Constraining Aerosol Acidity 24
The first iteration of the model experiments involved disallowing the gas phase to
interact with the particle phase. This model configuration (called the “particle-only
study”) simulates other studies which only collected particle data and were unable to
allow gas-particle partitioning [20, 59, 60]. Based on the near-zero average of H+strong
throughout the campaign and the results from previous studies, we expect the in situ pH
calculated with this method to fluctuate between acidic and basic values [44, 45].
The in situ pH values in the particle-only study exhibited a bimodal trend with the
majority of modeled values predicted to be >9 or <5. Fig. 1.8 shows the time series of
the in situ pH estimations and a histogram of number of values at each pH unit. The
average in situ pH was 10.45±2.91.
14
12
10
8
6
4
2
0
pH
7/21/2012 8/10/2012 8/30/2012 9/19/2012Date
6004002000Number of Values
Figure 1.8: Particle-only study in situ pH trends in CONTACT-2012
The results from this approach may be explained by acid-base titration of the species
in solution. Since this approach does not permit the buffering effect of neutralized species
partitioning into the gas phase, the acidity of the particles is controlled by the measured
balance between acids and bases. This effect is clearly demonstrated in Fig. 1.9, which
shows the in situ pH estimated by AIM-II plotted against the H+strong. As expected from
the parameterized particle study (Fig. 1.2), the in situ pH is never basic when H+strong is
positive and never acidic when H+strong is negative.
Fig. 1.10 shows a subset of the overall time series with associated error. Error was
estimated by rerunning the model twice as sensitivity runs. The sensitivity runs were
created by estimating the maximum H+strong due to instrumental uncertainty as increasing
the anions by 15% and the detection limit while decreasing the cations by 15% and the
detection limit and then performing the reverse for the low estimate of H+strong. As can be
seen from this subset of the data, the relatively small error in measurement can result in
Chapter 1. Constraining Aerosol Acidity 25
14
12
10
8
6
4
2
0
pH
40200-20-40
H+
strong (nmol m-3
)
Figure 1.9: pH and measured H+strong in CONTACT-2012
several orders of magnitude error in free H+
concentration. The bimodal trend in Fig.
1.8 may be the result of the instrumental uncertainty interpreted by AIM-II as moving
across the equivalence point of the acid-base titration. Qualitatively, the balance of acids
and bases (signified as H+strong) plotted in Fig. 1.10 suggests that periods when the NH4
+
has neutralized or nearly neutralized the acids results in strongly basic aerosols, while
the periods of increased anions relative to cations result in strongly acidic aerosols. One
interpretation of Fig. 1.10 is that the precision and accuracy of the particle constituent
measurements from the AIM-IC is not sufficient to determine with certainty if the strong
acidity is different than zero. As a result, a broad distribution of values is possible at
each time point.
Chapter 1. Constraining Aerosol Acidity 26
12
8
4
0
pH
12:00 AM8/25/2012
12:00 AM8/26/2012
12:00 AM8/27/2012
Date
806040200
-20H
+st
rong
(nm
ol m
-3)
Particle-only Mode
H+
strong
Figure 1.10: Case study of particle-only pH in CONTACT-2012
E-AIM is also able to provide theoretical trace gas concentrations predicted to be
in equilibrium with measured particulate species that would be required to produce the
measured aerosol values. In order to assess the physical plausibility of the calculated
strong acidity, the modelled atmospheric concentrations of the gases were compared
with the concentrations measured with the AIM-IC. As can be seen if Fig. 1.11, the
modelled mixing ratios were frequently orders of magnitude higher than the measured
values. The mean measured and predicted mixing ratios of NH3 were 2.42±1.98 ppb
and 5.76 ∗ 107 ± 7.20 ∗ 107 ppb, respectively. This discrepancy is a result of the model
predicting excess aqueous NH3 due to the near equivalency of acids and bases throughout
the campaign. The model is unable to protonate the excess NH3 due to insufficient
quantities of acids and is forced to leave unprotonated NH3 at concentrations which could
only partition into the condensed phase at very high ambient gas concentrations. The
prediction of unrealistically high gaseous NH3 suggests that particle-only inputs are not
sufficient to constrain the pH predictions.
Chapter 1. Constraining Aerosol Acidity 27
10-4
10-2
100
102
104
106
108
Mix
ing
Rat
io (
ppb)
7/21/2012 8/10/2012 8/30/2012 9/19/2012Date
Measured NH3(g)
AIM-II Predicted NH3(g)
Figure 1.11: Time series of NH3(g) particle-only AIM-II predictions and measurementsduring CONTACT-2012.
1.3.2 Gas-particle partitioning E-AIM analysis for CONTACT-
2012
Since the particle-only regime produces in situ pH values that are inconsistent with
observed gas-phase species, gas-particle partitioning was enabled in AIM-II (called the
“partitioning study”) to establish if this regime could better characterize the in situ pH and
chemical state of the system. The model was allowed to repartition volatile species and
was then evaluated to determine if the thermodynamic predictions of both particle and
gas phase species matched measurements. The sensitivity of the model to instrumental
uncertainty was tested as in the previous section. The model was run with inorganic
species followed by inclusion of the organic acids.
It was determined that allowing gas-particle partitioning in E-AIM dramatically
changes the model predictions of gas and particle phase concentrations. In addition, E-
AIM produced in situ pH outputs with a reduced range. Fig. 1.12 shows the relationship
between the repartitioned (output) H+strong and in situ pH. The results show that unlike
the relationship in Fig. 1.9, the output H+strong values are never negative, and the pH
values are higher than 6 or lower than 2. Fig. 1.13 shows the diurnal trend of pH averaged
over the field campaign, and Fig. 1.14 shows the time series of in situ pH over the course
of the field campaign. The pH increased at night with the increased relative humidity.
The liquid water content of the particle was similarly correlated with periods of elevated
pH. During the day, as the relative humidity decreases and the temperature increases, the
Chapter 1. Constraining Aerosol Acidity 28
pH decreased. The campaign average for the pH was 3.84±0.56 and no evidence of basic
aerosols was observed. This result matches those predicted by Keene and Savoie [55] and
suggests that the acidity of the particle is continually buffered by the volatility of NH3.
5.5
5.0
4.5
4.0
3.5
3.0
2.5
pH
1.6x10-91.41.21.00.80.60.40.20.0
Output H+
strong (nmol m-3
)
0.90.80.70.60.5Mole Fraction of H2O
Figure 1.12: Relationship between in situ pH and the output H+strong
Fig. 1.15 depicts a subset of the in situ pH data and illustrates the differences between
particle-only and partitioning studies. As can be observed, the uncertainty of the pH
in the partitioning study was very small compared to the uncertainty of the pH in the
particle-only study. The in situ pH estimated for the partitioning study was generally
within the uncertainty of the particle-only study.
Chapter 1. Constraining Aerosol Acidity 29
5.5
5.0
4.5
4.0
3.5
3.0
2.5
pH
00:00 06:00 12:00 18:00Time of Day
pH Mean Median0.90.80.70.60.50.4
Relative Humidity
Figure 1.13: Diurnal in situ pH coloured for relative humidity in CONTACT-2012 withgas-particle partitioning enabled
6
5
4
3
2
pH
7/21/2012 8/10/2012 8/30/2012 9/19/2012Date
4003002001000Number of Values
0.90.80.70.60.50.4Relative Humidity
Figure 1.14: Full time series of in situ pH coloured for relative humidity in CONTACT-2012with gas-particle partitioning enabled
Chapter 1. Constraining Aerosol Acidity 30
14
12
10
8
6
4
2
0
pH
12:00 AM8/25/2012
12:00 PM 12:00 AM8/26/2012
12:00 PM 12:00 AM8/27/2012
Date
80
40
0
-40
H+
stro
ng (
nmol
m-3
)
Particle-only Study Partitioning Study Partitioning Study w/ Organics
H+
strong
Figure 1.15: Case study of in situ pH estimated by disallowing and allowing gas-particlepartitioning and including organic acids with gas-particle partitition enabled in E-AIM inCONTACT-2012
However, at times the uncertainty of the measurements was unable to account for the
differences between the particle-only pH and the partitioning allowed pH predictions. In
Fig. 1.16, the particle-only pH followed the extremes of H+strong. It is possible that the
instrumental uncertainty was underestimated, which could result in a maximum H+strong
sensitivity run that still has a negative H+strong value. Also, periods where the pH values
were significantly different often occured during hours with low ambient mass loadings,
which could indicate that the limit of detection estimates had been low. Ultimately,
these results indicate that the in situ pH estimated by the gas-particle partitioning study
was far less sensitive to instrumental uncertainty and was generally not significantly
different than the pH estimated by the particle-only study due to its high sensitivity to
instrumental uncertainty.
Chapter 1. Constraining Aerosol Acidity 31
14
12
10
8
6
4
2
0
pH
12:00 AM8/27/2012
12:00 PM 12:00 AM8/28/2012
12:00 PM 12:00 AM8/29/2012
Date
40
20
0
-20H+
stro
ng (
nmol
m-3
) Particle-only Study Partitioning Study Partitioning Study w/ Organics
H+
strong
Figure 1.16: Case study of significant difference between gas-particle partitioning allowedand disallowed regimes in CONTACT-2012
Another important figure for qualitative analysis is the relationship between aqueous
mole fraction and in situ pH. Theoretically, for equivalent mass loadings of solutes, as
the particle becomes more dilute, the pH should trend towards higher values. Fig. 1.17
depicts the relationship between calculated in situ pH in the gas-particle partitioning
study and the predicted XH2O. The pH of particles at a given XH2O varies depending on
composition of the particle but generally increases with increasing XH2O (slope=2.89±0.05,
r2 =0.660). However, in the particle-only study, there was no relationship between pH and
XH2O as seen if Fig. 1.18, which further suggests that the results from the particle-only
regime are not physically representative of the ambient particles.
Chapter 1. Constraining Aerosol Acidity 32
5.5
5.0
4.5
4.0
3.5
3.0
2.5
pH
1.00.90.80.70.60.50.40.3Mole fraction of H2O
Figure 1.17: Relationship between in situ pH and XH2O in the gas-particle partitioningstudy
14
12
10
8
6
4
2
0
pH
0.80.60.4Mole fraction of H2O
Figure 1.18: Relationship between in situ pH and XH2O in the particle-only study
Predictions of volatile species versus measured results using AIM-II gas-
particle partitioning
In order to assess the quality of the predicted gas-particle repartitioning, the predicted
gaseous and particulate concentrations of volatile species were compared to the measured
Chapter 1. Constraining Aerosol Acidity 33
values. As seen in the particle-only regime, dramatic differences between the predicted
and measured values would indicate that the model is failing to replicate the conditions
measured during the campaign. However, the modelled and measured values for most
species closely correlated for the gas-particle partitioning enabled model. Fig. 1.19 shows
a subset of the NH3(g) time series with modeled and measured concentrations of both
particulate and gaseous species. Predicted NH3(g) and NH4(aq)+
values closely correlate to
measured concentrations. Correlation data for all major species are tabulated in Table
B.1.
200
150
100
50
0Con
cent
ratio
n (n
mol
m-3
)
12:00 AM8/25/2012
12:00 AM8/26/2012
12:00 AM8/27/2012
12:00 AM8/28/2012
Date
400
300
200
100
0
Con
cent
ratio
n (n
mol
m-3
)
NH3(g) Measured
NH4+
(aq) Measured Modelled Concentration
Figure 1.19: NH3(g) predictions from AIM-II upon allowing gas-particle partitioning andmeasured concentrations in CONTACT-2012
AIM-II predictions of NO3(aq)-
and HNO3(g) were also relatively well correlated with the
measured concentrations, although outliers significantly changed the r2 values. Generally
the modelled concentration was different in absolute value than the measured concentration.
Fig. 1.20 shows that while the trends for both of these species were matched by the
predictions, the absolute values tend to lie outside of the instrumental uncertainty. The
Chapter 1. Constraining Aerosol Acidity 34
reason for the discrepancy is possibly explained by the underestimation of uncertainty.
Since the absolute concentrations of NO3(aq)-
and absolute instrumental uncertainties were
much lower than those of NH4(aq)+
or SO4(aq)2-
, the sensitivity runs will be dominated by
the changes in NH4(aq)+
or SO4(aq)2-
. Small deviations between predicted and measured NH4(aq)+
or SO4(aq)2-
may result in relatively large deviations in NO3(aq)-
. The difference between
measured and modelled NH4(aq)+
and NH3(g) was 5.78±8.23 nmol m−3 and -5.53±12.05
nmol m−3, respectively, while the difference between measured and modelled NO3(aq)-
and
HNO3(g) was 1.07±6.08 nmol m−3 and -1.06±6.08 nmol m−3. These results indicate that
while the relative difference between modelled and measured NO3(aq)-
and HNO3(g) is
large compared to NH4(aq)+
and NH3(g), the model is producing results that vary from the
measured value by roughly the same absolute amount for both species.
The discrepancy between modelled and measured values may also be explained by the
ambient aerosols not meeting the assumptions of the model. The aerosols may have been
composed of phases other than the simple deliquesced aqueous phase utilized in the model.
For example, an organic phase could have impacted the ability of the particles to uptake
water and changed the rate at which the particles equilibrated with the gas phase. The
particles also may have been externally mixed and some of the NO3(aq)-
may have been
associated with coarse mode particles or the system may have not reached equilibrium.
In the following section we investigate E-AIM analysis with externally mixed aerosols.
Overall, the evidence of these studies suggests that the particle-only version of E-AIM
may not accurately portray the in situ pH and that enabling gas-particle partitioning in
the model results in an improved relationship between the modelled and measured gas
and particulate composition.
Chapter 1. Constraining Aerosol Acidity 35
50
40
30
20
10
0Con
cent
ratio
n (n
mol
m-3
)
12:00 AM8/25/2012
12:00 AM8/26/2012
12:00 AM8/27/2012
12:00 AM8/28/2012
Date
50
40
30
20
10
0Con
cent
ratio
n (n
mol
m-3
) HNO3(g) Measured
NO3-(aq) Measured
Modelled Concentration
Figure 1.20: Case study of particle-only pH in CONTACT-2012
1.3.3 Ad hoc two mode separation from CalNex 2010
In order to probe whether the assumption of internal mixing is leading to the poor
HNO3(g)/NO3(aq)-
predictions, data from CalNex 2010 were used to examine the plausibility
of bimodal acidity and chemical system prediction improvement on results from Ellis
et al. [76] (in prep). The PILS-IC (PM2.5) and AMS (PM1.0) used during CalNex 2010
provided the opportunity to examine differences in particle acidity between sea salt and
accumulation mode aerosols. The coarse mode was assumed to be dominated by sea
salt particles since Pasadena, CA is located less than 30 km from the Pacific Ocean
[31]. In this study, PM2.5 data were segregated as the “accumulation mode” and “sea
salt mode.” It was assumed that the AMS measurements reflect the composition of the
accumulation mode. For reasons discussed later in this section, the AMS data were used
only for qualitative analysis, while the PILS-IC (PM2.5) data were segregated based on
assumptions about the composition of each mode.
The particles were separated into sea salt and accumulation mode by inputting a
Chapter 1. Constraining Aerosol Acidity 36
uniform percentage of each species measured with PILS-IC across all hourly datapoints
into AIM-IV (e.g. if 40% of measured particulate NH4+
was input as the sea salt mode,
60% was input as accumulation mode). Since both particle populations were assumed to
be in equilibrium with the gas phase, the gas-phase concentration results from the “sea
salt” and “accumulation” model runs were compared with the measured values as a way
of checking whether the mode segregation had adequately represented the system.
In the simplest separation scenario, the particles were treated as being internally
mixed and run in AIM-IV with allowed gas-particle partitioning. Next, the PILS-IC
particulate concentration data were separated to produced a sea salt aerosol which was
defined as being purely Na+
and Cl-
while the accumulation mode was defined as NH4+
,
NO3-, and SO4
2-. The model was then run iteratively to produce optimized input sea salt
and accumulation mode particles which were set percentages of all five species. Studies
have shown that SO42-
is almost exclusively found in accumulation mode aerosols, while
Na+
is found primarily in sea salt aerosols [35, 77]. Ultimately an accumulation/sea salt
aerosol input ratio was optimized from PILS-IC data at a sea salt value of 30% NH4+
,
100% Na+
, 10% SO42-
, 60% NO3-, and 90% Cl
-and all remaining particulate composition
was designated as accumulation mode.
One significant assumption of creating this ad hoc separation of sea salt and accumu-
lation mode aerosols from bulk data is that the relative ratios of each species between
the two modes is constant throughout the dataset. An approximate assessment of this
assumption was made by plotting the relative ratio throughout the campaign of PILS-IC
and AMS data (see Fig. 1.21). Since the two instruments collect different size fractions,
the data were evaluated to determine whether there was a constant relative amount of each
species in each mode throughout the campaign. Ideally, the two instruments would have
been able to provide mode resolved composition, but this is not possible due, most likely,
to the variability of the size of populations of particles and problems with the accuracy
of the instrumentation. In Fig. 1.21 the ratio of AMS to PILS-IC concentrations of
sulphate remained above 1.5 throughout the campaign. Since the PILS-IC samples PM2.5
while the AMS samples approximately PM1.0, the result indicates a possible calibration
error as the theoretical ratios should always remain at or below 1.0 [78]. This is likely
the result of the uncertainty in the collection efficiency of the AMS, a known and well
documented problem [73, 79]. In addition, the uncertainty of the AMS measurements was
estimated at 30% and the PILS-IC uncertainty was estimated to be 13% [80]. However,
despite the calibration problem, the ratio remains relatively constant throughout the
campaign. Although the data does not provide exact answers as to the relative amounts
between particle populations, it does suggest the relative composition of the modes remain
Chapter 1. Constraining Aerosol Acidity 37
approximately the same throughout the campaign.
2.0
1.5
1.0
0.5
0.0
Rat
io [A
MS
]/[P
ILS
]
6/9/2010 6/10/2010 6/11/2010 6/12/2010 6/13/2010 6/14/2010
Date
NH4+
Cl-
NO3-
SO42-
Figure 1.21: AMS/PILS ratio in CalNex 2010
The input values and modelled results from the optimized separation study demonstrate
that the technique was able to produce modelled gas concentrations from the two different
particle types that reasonably matched the measured value. Figs. 1.22, 1.23 and 1.24 show
the measured and modelled gaseous concentrations of the three dominant volatile gases,
HNO3, NH3, and HCl, from the optimized sea salt and accumulation mode experiments.
The modelled values are generally within the uncertainty of the measurements for all
of the species. Figs. 1.25, 1.26 and 1.27 show the comparison between the optimized
estimations of the total sea salt and accumulation mode concentrations and the AIM-IV
modelled predictions as well as the differences between the input and modelled ratio of sea
salt and accumulation mode. Figs. B.1, B.2 and B.3 show the relationship between the
individual optimized input and output particle-phase data for each mode. Although the
results are often outside of the instrumental uncertainty of the PILS-IC, the concentration
trends were generally captured. Given the uncertainty of the uniformity of the sea
salt/accumulation mode ratio, the results, while imperfect, do appear to be representing
the chemical state of the particles. The deviations in particle concentrations are likely
significantly affected by the inaccuracy of the fixed percentage input method, since the
relative amounts of any given species likely varies throughout the campaign.
Chapter 1. Constraining Aerosol Acidity 38
250x10-9
200
150
100
50
0
Con
cent
ratio
n (m
ol m
-3)
6/9/2010 6/10/2010 6/11/2010 6/12/2010 6/13/2010 6/14/2010Date
Measured HNO3(g)
Modelled HNO3(g) from Accumulation Mode Modelled HNO3(g) from Sea Salt Mode
Figure 1.22: HNO3(g) predictions from optimized sea salt mode and accumulation modeaerosols from CalNex 2010
200x10-9
150
100
50
0
Con
cent
raito
n (m
ol m
-3)
6/9/2010 6/10/2010 6/11/2010 6/12/2010 6/13/2010 6/14/2010Date
Measured NH3(g)
Modelled NH3(g) from Accumulation Mode Modelled NH3(g) from Sea Salt Mode
Figure 1.23: NH3(g) predictions from optimized sea salt mode and accumulation modeaerosols from CalNex 2010
Chapter 1. Constraining Aerosol Acidity 39
160x10-9
140
120
100
80
60
40
20
0
Con
cent
ratio
n (m
ol m
-3)
6/9/2010 6/10/2010 6/11/2010 6/12/2010 6/13/2010 6/14/2010Date
Measured HCl(g)
Modelled HCl(g) from Accumulation Mode Modelled HCl(g) from Sea Salt Mode
Figure 1.24: HCl(g) predictions from optimized sea salt mode and accumulation modeaerosols from CalNex 2010
200x10-9
150
100
50
0
Con
cent
ratio
n (m
ol m
-3)
6/9/2010 6/10/2010 6/11/2010 6/12/2010 6/13/2010 6/14/2010Date
6
4
2
0Rat
io [S
ea S
alt]/
[Acc
umul
atio
n]
Modelled Sea Salt Mode NO3-(aq)
Modelled Accumulation Mode NO3-(aq)
Measured Total NO3-(aq)
Input Concentration Ratio Modelled Output Concentration Ratio
Figure 1.25: NO3(aq)-
predictions from optimized sea salt mode and accumulation modeaerosols and input and modelled ratio of sea salt to accumulation mode from CalNex 2010
Chapter 1. Constraining Aerosol Acidity 40
250x10-9
200
150
100
50
Con
cent
ratio
n (m
ol m
-3)
6/9/2010 6/10/2010 6/11/2010 6/12/2010 6/13/2010 6/14/2010Date
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0Rat
io [S
ea S
alt]/
[Acc
umul
atio
n]
Modelled Sea Salt Mode NH4+
(aq)
Modelled Accumulation Mode NH4+
(aq)
Measured Total NH4+
(aq)
Input Concentration Ratio Modelled Output Concentration Ratio
Figure 1.26: NH4(aq)+
predictions from optimized sea salt mode and accumulation modeaerosols and input and modelled ratio of sea salt to accumulation mode from CalNex 2010
Chapter 1. Constraining Aerosol Acidity 41
50x10-9
40
30
20
10
0Con
cent
ratio
n (m
ol m
-3)
6/9/2010 6/10/2010 6/11/2010 6/12/2010 6/13/2010 6/14/2010Date
12
10
8
6
4
2
0Rat
io [S
ea S
alt]/
[Acc
umul
atio
n]
Modelled Sea Salt Mode Cl-(aq)
Modelled Accumulation Mode Cl-(aq)
Measured Total Cl-(aq)
Input Concentration Ratio Modelled Output
Concentration Ratio
Figure 1.27: Cl(aq)-
predictions from optimized sea salt mode and accumulation modeaerosols and input and modelled ratio of sea salt to accumulation mode from CalNex 2010
Fig. 1.28 shows the optimized and bulk pH values calculated for each hourly datapoint
throughout the campaign. Despite the significant differences in input composition of the
two aerosol modes the magnitude of the difference in pH was relatively small throughout
the campaign. This result agrees with findings by Keene and Savoie [55] and Erickson et al.
[10] who concluded that partitioning drives the pH of sea salt and accumulation modes
toward a common value. Our results show that the average pH values for the optimized
accumulation and sea salt modes were 3.95±0.27 and 4.21±0.26, respectively, while the
modelled bulk pH was 4.10±0.27. Keene et al. [81] found that for size resolved marine
aerosols with Dp <1 µm were approximately 1 pH unit lower at pH=3, on average, than
supermicron aerosols ranging up to 10 µm at pH=4. Despite the dramatically different
composition of the particles, the pH in our study between the two input modes was not
significantly different.
This study illustrates a proof of concept regarding the use of the E-AIM to characterize
the pH of different size fractions of aerosols. However, since the size fractions were not
physically separated during the study, the results cannot be independently verified. As
a consequence, it is possible that the model found the “optimized” modes to produce
Chapter 1. Constraining Aerosol Acidity 42
5.5
5.0
4.5
4.0
3.5
pH
6/9/2010 6/11/2010 6/13/2010Date
Accumulation Mode Sea Salt Mode Bulk
Figure 1.28: The pH values for the accumulation and sea salt modes of the optimizedseparation and the bulk pH from CalNex 2010
predictions that approximately matched the arbitrarily defined modes, but the possibility
that the model found results that coincidentally matched the inputs cannot be precluded.
Furthermore, if the aerosols are not at equilibrium, the ad hoc separation could produce a
system that is attempting to replicate conditions that do not exist. A more sophisticated
model that accounts for the effect of size and composition dependence of mass transfer
rates on equilibration time could be developed to produce more accurate results. Given
the uncertainty of the instrumentation and the inability to know the fractionated con-
centrations independently, the uncertainty in the fractionated segregation should not be
understated.
This analysis demonstrates that the pH difference between the two modes is strongly
buffered by gas-particle partitioning. However, the bimodal system simplifies the reality
that atmospheric aerosol composition is often very heterogeneous. For example, small,
unaged, nucleating particles found in coal power plant exhaust plumes may be highly
acidic since they have not yet been neutralized by ambient concentrations of NH3 [82].
However, since their mass is significantly less than the rest of the accumulation mode,
their contribution to the acidity of the particles may be underrepresented.
The study also used a relatively approximative methodology of estimating the frac-
tionation of the individual aerosol constituents. Analysis time could be shortened by
Chapter 1. Constraining Aerosol Acidity 43
automating the model processing to enable rapid optimization in future studies. This
would allow for more exact segregation of species into the two modes that could be
performed on each datapoint rather than inputting a constant fractionation throughout
the campaign.
1.3.4 24-hour integration analysis
Numerous aerosol acidity studies have been conducted that integrate particle measurements
over 24-hour or longer sampling periods [4, 59, 60, 83]. One of the drawbacks of this
method is that the processes that affect particulate composition may change dramatically
over the course of 24 hours. Temperature, relative humidity, mass loading, and ambient
gas concentration all vary on relatively short timescales [84]. As a result, long integration
times may not allow for the necessary resolution to accurately estimate acidity. Since
the CONTACT-2012 dataset provided hourly gas and particle measurements, model
experiments were performed to determine the extent to which 24-hour integration impacted
acidity predictions.
In order to probe the question of whether differing integration will impact in situ
pH, four model runs were performed in AIM-II. The first run averaged the concentration
values of particulate and gaseous species for 24 hours (starting at 00:00). This run was
called the “bulk average.” The temperature and relative humidity were averaged over the
same time period. The model was run in the particle-only configuration with gas-particle
partitioning disallowed. The second run used the particle-only AIM-II regime to process
hourly measurements and find hourly activities of H+
. The AH+ values were averaged over
24 hours (also starting at 00:00) and the pH calculated from each 24-hour AH+ integration
was called the “model average.” pH values from the bulk and model averages were plotted
in Fig. 1.29. Both model and bulk averaging methodologies were also applied to the
gas-particle partitioning regime of AIM-II and the results are also plotted in Fig. 1.29.
All 24-hour periods that lacked complete hourly measurements were disregarded.
Several averaged 24-hour periods produced errors in the model and were unable to be
processed. These errors were the result of conditions being outside of the ranges that
the model was designed to operate at and usually were caused by very low relative
concentrations of NH4+
to other constituents.
The results showed that there were substantial differences between the pH values
predicted by the two averaging methods in the particle-only experiments, while the
gas-particle partitioning experiments showed a close relationship between bulk and model
averages. The results demonstrate that the variability throughout a 24-hour period may
Chapter 1. Constraining Aerosol Acidity 44
strongly influence the particle-only predictions. An explanation for this effect is that
if a single hour from the model average has a pH that falls on the acidic (pH<7) side
of the pH titration curve, the model average will be strongly influenced by this single
value since the scale is logarithmic. If particles generally are on the pH>7 side of the pH
titration curve, the bulk average will be less influenced by a single hour with pH<7 since
the overall 24-hour balance between acid and base may still produce excess base.
12
10
8
6
4
pH
7/21/2012 7/31/2012 8/10/2012 8/20/2012 8/30/2012 9/9/2012 9/19/2012 9/29/2012Date
Bulk Average (Particle-only) Model Average (Particle-only) Bulk Average (Partitioning allowed) Model Average (Partitioning allowed)
Figure 1.29: 24-hour data integration methodologies from CONTACT-2012
1.3.5 pH distribution
The pH values throughout the CONTACT-2012 campaign were plotted against other
datasets which apply the same E-AIM gas-particle partitioning pH estimation in Fig.
1.30. These studies suggest that the aerosol pH measured in the boundary layer at urban
and rural sites in continental North America lies approximately in the pH 3-5 range.
These studies also confirm the buffering effect observed by Keene and Savoie [55] and
Erickson et al. [10] that suggests the pH of ambient aerosols is strongly regulated by the
partitioning of acids and bases and that the particle-gas equilibrium exerts a dominating
effect on acidity.
Chapter 1. Constraining Aerosol Acidity 45
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
Nor
mal
ized
Fre
quen
cy o
f Obs
erva
tion
7654321
pH
Location n= Pasadena 475 Bakersfield 547 BAQS-Met 309 Toronto 139 CONTACT 1394
Figure 1.30: pH histogram of recent continental North American field campaigns, where“n” is the number of measurements taken at each site
1.4 Conclusions
These results demonstrate the importance of constraining aerosol acidity measurements
with ambient gaseous data. While uncertainty regarding Henry’s law partitioning constants
is avoided by disabling gas-particle partitioning, the errors in the particulate measurements
can have magnified, deleterious effects on the model’s accuracy. Aerosol acidity methods
which rely solely on sampled particle concentrations and thermodynamic modelling
may be inadequate for providing accurate pH estimates. Model predictions may not
be representative of a real atmospheric system. Future studies that use E-AIM and
disallow partitioning should evaluate the discrepancies between modelled and measured
ambient gas concentrations to establish if the pH estimates are reliable. This analysis
is not commonly done in recent literature and may explain the findings of papers that
observe dramatic fluctuations of in situ pH and H+strong [44, 45, 85]. Despite inadequacies
in the predictive powers of HNO3/NO3-
from the gas-particle partitioning study from
the CONTACT-2012 campaign, the model is generally able to produce repartitioned
results that closely correlate with the measured values of both gas and particle species.
Furthermore, size-resolved analysis may aid the model’s ability to accurately estimate pH
of chemically dissimilar aerosols by reducing the effects of the assumption of particulate
Chapter 1. Constraining Aerosol Acidity 46
internal mixing. Future research that collects size-resolved concentration data of inorganic
species may verify the findings of this study.
The results from this study also suggest that the difference in pH between the sea
salt and the accumulation mode may be more similar and more invariable than has been
previously suggested [85]. The results support the assertions from Keene [54] and Erickson
et al. [10] that pH values of sea salt and accumulation mode tend to be similar due to the
buffering effect of intermode partitioning.
Laboratory studies, especially those relating to recent SOA experiments, must constrain
the acidity of the particles. The current methodologies of evaluating SOA growth as a
function of H+strong or the in situ pH estimated from the particle-only mode of E-AIM may
fail to accurately probe the aerosol system. Furthermore, the results from this and other
recent field studies found average in situ pH values that are higher than those required by
previous laboratory studies to produce significant VOC uptake [19, 50, 51]. Ambient air
quality and public health studies may find novel implications of aerosol acidity if they are
able to correlate in situ pH with respiratory and pulmonary illness rather than H+strong.
1.5 Acknowledgements
CalNex 2010 NH3(g) data was collected and processed by Raluca Ellis using the QC-
TILDAS. Rodney Weber provided PILS-IC data and Patrick Hayes provided AMS data.
Raluca Ellis also compiled the dataset prior to the E-AIM analysis and provided guidance
regarding the model’s usage. Greg Wentworth contributed to much of the AIM-IC data
collection and analysis. Alex Tevlin provided CONTACT-2012 QC-TILDAS data as well
as E-AIM support. Jennifer Murphy contributed much insight into theoretical background
of the analysis; she also compiled pH data from recent campaigns. Carol Cheyne, Rachel
Hems, and Geoff Stupple also helped with instrumentation throughout CONTACT-2012.
Finally, the author thanks the NSERC CREATE program, IACPES, for continued funding
throughout this research.
1.6 Bibliography
[1] C. R. Hoyle, M. Boy, N. M. Donahue, J. L. Fry, M. Glasius, A. Guenther, A. G.
Hallar, K. Huff Hartz, M. D. Petters, T. Petaja, T. Rosenoern, and A. P. Sullivan.
A review of the anthropogenic influence on biogenic secondary organic aerosol.
Atmospheric Chemistry and Physics, 11(1):321–343, January 2011. ISSN 1680-7324.
doi: 10.5194/acp-11-321-2011.
Chapter 1. Constraining Aerosol Acidity 47
[2] D. W. Dockery, J. Cunningham, A. I. Damokosh, L. M. Neas, J. D. Spengler,
P. Koutrakis, J. H. Ware, M. Raizenne, and F. E. Speizer. Health effects of acid
aerosols on North American children: respiratory symptoms. Environmental Health
Perspectives, 104(5):500–505, May 1996. ISSN 0091-6765. doi: 10.1289/ehp.96104500.
[3] B. D. Ostro, M. J. Lipsett, M. B. Wiener, and J. C. Selner. Asthmatic responses to
airborne acid aerosols. American Journal of Public Health, 81(6):694–702, June 1991.
ISSN 0090-0036. doi: 10.2105/AJPH.81.6.694.
[4] X. Yao, T. Yan Ling, M. Fang, and C. K. Chan. Comparison of thermodynamic
predictions for in situ pH in PM2.5. Atmospheric Environment, 40(16):2835–2844,
May 2006. ISSN 13522310. doi: 10.1016/j.atmosenv.2006.01.006.
[5] J. Li and M. Jang. Aerosol Acidity Measurement Using Colorimetry Coupled With
a Reflectance UV-Visible Spectrometer. Aerosol Science and Technology, 2012.
[6] R. C. Gwynn, R. T. Burnett, and G. D. Thurston. A time-series analysis of acidic
particulate matter and daily mortality and morbidity in the Buffalo, New York,
region. Environmental Health Perspectives, 108(2):125–133, 2000.
[7] I.-F. Mao, C.-H. Lin, C.-J. Lin, Y.-J. Chen, F.-C. Sung, and M.-L. Chen. Exposure
of acid aerosol for schoolchildren in metropolitan Taipei. Atmospheric Environment,
43(35):5622–5629, November 2009. ISSN 13522310. doi: 10.1016/j.atmosenv.2009.07.
054.
[8] M. Raizenne, L. M. Neas, A. I. Damokosh, D. W. Dockery, J. D. Spengler,
P. Koutrakis, J. H. Ware, and F. E. Speizer. Health effects of acid aerosols on
North American children: pulmonary function. Environmental Health Perspectives,
104(5):506–514, May 1996. ISSN 0091-6765. doi: 10.1289/ehp.96104506.
[9] W. C. Keene, M. A. K. Khalil, D. J. Erickson, A. McCulloch, T. E. Graedel, J. M.
Lobert, M. L. Aucott, S. L. Gong, D. B. Harper, G. Kleiman, P. Midgley, R. M.
Moore, C. Seuzaret, W. T. Sturges, C. M. Benkovitz, V. Koropalov, L. A. Barrie,
and Y. F. Li. Composite global emissions of reactive chlorine from anthropogenic
and natural sources: Reactive Chlorine Emissions Inventory. Journal of Geophysical
Research, 104(D7):8429, April 1999. ISSN 0148-0227. doi: 10.1029/1998JD100084.
[10] D. J. Erickson, C. Seuzaret, W. C. Keene, and S. L. Gong. A general circulation
model based calculation of HCl and ClNO 2 production from sea salt dechlorination:
Chapter 1. Constraining Aerosol Acidity 48
Reactive Chlorine Emissions Inventory. Journal of Geophysical Research, 104(D7):
8347, April 1999. ISSN 0148-0227. doi: 10.1029/98JD01384.
[11] W. L. Chameides and A. W. Stelson. Aqueous-phase chemical processes in deliques-
cent sea-salt aerosols: A mechanism that couples the atmospheric cycles of S and sea
salt. Journal of Geophysical Research, 97(D18):20565, 1992. ISSN 0148-0227. doi:
10.1029/92JD01923.
[12] P. Brimblecombe and S. L. Clegg. The solubility and behaviour of acid gases in
the marine aerosol. Journal of Atmospheric Chemistry, 7(1):1–18, July 1988. ISSN
0167-7764. doi: 10.1007/BF00048251.
[13] D. J. Jacob. Introduction to Atmospheric Chemistry. Princeton University Press,
1999.
[14] J. M. Roberts, H. D. Osthoff, S. S. Brown, and A. R. Ravishankara. N2O5 oxidizes
chloride to Cl2 in acidic atmospheric aerosol. Science, 321(5892):1059, August 2008.
ISSN 1095-9203. doi: 10.1126/science.1158777.
[15] M. Mozurkewich. Mechanisms for the release of halogens from sea-salt particles
by free radical reactions. Journal of Geophysical Research: Atmospheres, 100(D7):
14199–14207, 1995.
[16] R. Vogt, P. Crutzen, and R. Sander. A mechanism for halogen release from sea-salt.
Nature, 383:327–331, 1996.
[17] W. C. Keene, R. Sander, A. A. Pszenny, R. Vogt, P. J. Crutzen, and J. N. Galloway.
Aerosol pH in the marine boundary layer. Journal of Aerosol Science, 29(3):339–356,
March 1998. ISSN 00218502. doi: 10.1016/S0021-8502(97)10011-8.
[18] J. C. Cabada, S. N. Pandis, and A. L. Robinson. Sources of atmospheric carbonaceous
particulate matter in Pittsburgh, Pennsylvania. Journal of the Air Waste Management
Association 1995, 52(6):732–741, 2002.
[19] M. Jang, N. M. Czoschke, S. Lee, and R. M. Kamens. Heterogeneous atmospheric
aerosol production by acid-catalyzed particle-phase reactions. Science (New York,
N.Y.), 298(5594):814–7, October 2002. ISSN 1095-9203. doi: 10.1126/science.1075798.
[20] Q. Zhang, J. L. Jimenez, D. R. Worsnop, and M. Canagaratna. A Case Study of Urban
Particle Acidity and Its Influence on Secondary Organic Aerosol. Environmental
Chapter 1. Constraining Aerosol Acidity 49
Science & Technology, 41(9):3213–3219, May 2007. ISSN 0013-936X. doi: 10.1021/
es061812j.
[21] J. Heintzenberg. Fine particles in the global troposphere: A review. Tellus B, 41(2):
149–160, 1989.
[22] W. C. Malm, J. F. Sisler, D. Huffman, R. A. Eldred, and T. A. Cahill. Spatial and
seasonal trends in particle concentration and optical extinction in the United States.
Journal of Geophysical Research, 99(D1):1347–1370, 1994.
[23] S. Potukuchi and A. S. Wexler. Identifying solid-aqueous-phase transitions in atmo-
spheric aerosols. II. Acidic solutions. Atmospheric Environment, 29(22):3357–3364,
November 1995. ISSN 13522310. doi: 10.1016/1352-2310(95)00212-H.
[24] J. H. Seinfeld and S. N. Pandis. Atmospheric Chemistry and Physics: from air
pollution to climate change. John Wiley and Sons, 2006.
[25] W. C. Malm. Spatial and monthly trends in speciated fine particle concentration in
the United States. Journal of Geophysical Research, 109(D3):D03306, 2004. ISSN
0148-0227. doi: 10.1029/2003JD003739.
[26] R. W. Pinder, P. J. Adams, and S. N. Pandis. Ammonia Emission Controls as a
Cost-Effective Strategy for Reducing Atmospheric Particulate Matter in the Eastern
United States. Environmental Science & Technology, 41(2):380–386, January 2007.
ISSN 0013-936X. doi: 10.1021/es060379a.
[27] A. Russell and G. Cass. Acquisition of regional air quality model validation data
for nitrate, sulfate, ammonium ion and their precursors. Atmospheric Environment
(1967), 1984.
[28] W. John, S. Wall, and J. Ondo. A new method for nitric acid and nitrate aerosol
measurement using the dichotomous sampler. Atmospheric Environment (1967),
1988.
[29] D. C. Blanchard, A. H. Woodcock, and R. J. Cipriano. The vertical distribution of
the concentration of sea salt in the marine atmosphere near Hawaii. Tellus B, 36B
(2):118–125, April 1984. ISSN 02806509. doi: 10.1111/j.1600-0889.1984.tb00233.x.
[30] J. W. Fitzgerald. Marine aerosols: A review. Atmospheric Environment. Part
A. General Topics, 25(3-4):533–545, January 1991. ISSN 09601686. doi: 10.1016/
0960-1686(91)90050-H.
Chapter 1. Constraining Aerosol Acidity 50
[31] S. L. Gong, L. A. Barrie, J. M. Prospero, D. L. Savoie, G. P. Ayers, J.-P. Blanchet,
and L. Spacek. Modeling sea-salt aerosols in the atmosphere: 2. Atmospheric
concentrations and fluxes. Journal of Geophysical Research, 102(D3):3819, February
1997. ISSN 0148-0227. doi: 10.1029/96JD03401.
[32] A. M. Fridlind and M. Z. Jacobson. A study of gas-aerosol equilibrium and aerosol
pH in the remote marine boundary layer during the First Aerosol Characterization
Experiment (ACE 1). Journal of Geophysical Research, 105(D13):17325, July 2000.
ISSN 0148-0227. doi: 10.1029/2000JD900209.
[33] M. Pham, J.-F. Muller, G. P. Brasseur, C. Granier, and G. Megie. A three-dimensional
study of the tropospheric sulfur cycle. Journal of Geophysical Research, 100(D12):
26061, 1995. ISSN 0148-0227. doi: 10.1029/95JD02095.
[34] E. Holland, F. Dentener, B. Braswell, and J. Sulzman. Contemporary and pre-
industrial global reactive nitrogen budgets. Biogeochemistry, 46(1-3):7–43, July 1999.
doi: 10.1023/A:1006148011944.
[35] B. L. Lefer and R. W. Talbot. Summertime measurements of aerosol nitrate and
ammonium at a northeastern U.S. site. Journal of Geophysical Research, 106(D17):
20365, September 2001. ISSN 0148-0227. doi: 10.1029/2000JD900693.
[36] S. L. Clegg, P. Brimblecombe, and A. S. Wexler. Thermodynamic Model of the
System H+-NH+4 -SO2−
4 -NO−3 -H2O at Tropospheric Temperatures. The Journal of
Physical Chemistry A, 102(12):2137–2154, March 1998. ISSN 1089-5639. doi: 10.
1021/jp973042r.
[37] P. Koutrakis, J. M. Wolfson, and J. D. Spengler. An improved method for measuring
aerosol strong acidity: Results from a nine-month study in St Louis, Missouri and
Kingston, Tennessee. Atmospheric Environment, 22(1):157–162, January 1988. ISSN
00046981. doi: 10.1016/0004-6981(88)90308-3.
[38] W. C. Keene, A. A. P. Pszenny, D. J. Jacob, R. A. Duce, J. N. Galloway, J. J.
Schultz-Tokos, H. Sievering, and J. F. Boatman. The geochemical cycling of reactive
chlorine through the marine troposphere. Global Biogeochemical Cycles, 4(4):407–430,
December 1990. ISSN 08866236. doi: 10.1029/GB004i004p00407.
[39] Z. Meng, J. H. Seinfeld, P. Saxena, and Y. P. Kim. Contribution of Water to
Particulate Mass in the South Coast Air Basin. Aerosol Science and Technology, 22
(1):111–123, January 1995. ISSN 0278-6826. doi: 10.1080/02786829408959731.
Chapter 1. Constraining Aerosol Acidity 51
[40] R. K. Pathak, X. Yao, A. K. Lau, and C. K. Chan. Acidity and concentrations of
ionic species of PM2.5 in Hong Kong. Atmospheric Environment, 37(8):1113–1124,
March 2003. ISSN 13522310. doi: 10.1016/S1352-2310(02)00958-5.
[41] X. Yao, P. J. Rehbein, C. J. Lee, G. J. Evans, J. Corbin, and C.-H. Jeong. A study
on the extent of neutralization of sulphate aerosol through laboratory and field
experiments using an ATOFMS and a GPIC. Atmospheric Environment, 45(34):
6251–6256, November 2011. ISSN 13522310. doi: 10.1016/j.atmosenv.2011.06.061.
[42] R. K. Pathak, X. Yao, and C. K. Chan. Sampling Artifacts of Acidity and Ionic
Species in PM 2.5. Environmental Science & Technology, 38(1):254–259, January
2004. ISSN 0013-936X. doi: 10.1021/es0342244.
[43] S. Takahama, C. I. Davidson, and S. N. Pandis. Semicontinuous Measurements of
Organic Carbon and Acidity during the Pittsburgh Air Quality Study: Implications
for Acid-Catalyzed Organic Aerosol Formation. Environmental Science & Technology,
40(7):2191–2199, April 2006. ISSN 0013-936X. doi: 10.1021/es050856+.
[44] Y. Zhou, L. Xue, T. Wang, X. Gao, Z. Wang, X. Wang, J. Zhang, Q. Zhang, and
W. Wang. Characterization of aerosol acidity at a high mountain site in central
eastern China. Atmospheric Environment, 51:11–20, May 2012. ISSN 13522310. doi:
10.1016/j.atmosenv.2012.01.061.
[45] S. N. Behera, R. Betha, P. Liu, and R. Balasubramanian. A study of diurnal
variations of PM2.5 acidity and related chemical species using a new thermodynamic
equilibrium model. Science of The Total Environment, 452-453:286–295, 2013. doi:
10.1016/j.scitotenv.2013.02.062.
[46] L. D. Ziemba, E. Fischer, R. J. Griffin, and R. W. Talbot. Aerosol acidity in rural New
England: Temporal trends and source region analysis. Journal of Geophysical Re-
search, 112(D10):D10S22, April 2007. ISSN 0148-0227. doi: 10.1029/2006JD007605.
[47] A.-M. N. Kitto and R. M. Harrison. Processes affecting concentrations of aerosol
strong acidity at sites in eastern England. Atmospheric Environment. Part A.
General Topics, 26(13):2389–2399, September 1992. ISSN 09601686. doi: 10.1016/
0960-1686(92)90369-V.
[48] P. Winkler. Observations on acidity in continental and in marine atmospheric aerosols
and in precipitation. Journal of Geophysical Research, 85(C8):4481, 1980. ISSN
0148-0227. doi: 10.1029/JC085iC08p04481.
Chapter 1. Constraining Aerosol Acidity 52
[49] J. Ludwig and O. Klemm. Acidity of size-fractionated aerosol particles. Water, Air,
and Soil Pollution, 49(1-2):35–50, January 1990. ISSN 0049-6979. doi: 10.1007/
BF00279508.
[50] J. D. Surratt, J. H. Kroll, T. E. Kleindienst, E. O. Edney, M. Claeys, A. Sorooshian,
N. L. Ng, J. H. Offenberg, M. Lewandowski, M. Jaoui, R. C. Flagan, and J. H. Seinfeld.
Evidence for Organosulfates in Secondary Organic Aerosol. Environmental Science &
Technology, 41(2):517–527, January 2007. ISSN 0013-936X. doi: 10.1021/es062081q.
[51] J. H. Offenberg, M. Lewandowski, E. O. Edney, T. E. Kleindienst, and M. Jaoui.
Influence of aerosol acidity on the formation of secondary organic aerosol from
biogenic precursor hydrocarbons. Environmental Science & Technology, 43(20):
7742–7, October 2009. ISSN 0013-936X. doi: 10.1021/es901538e.
[52] M. N. Chan, J. D. Surratt, A. W. H. Chan, K. Schilling, J. H. Offenberg,
M. Lewandowski, E. O. Edney, T. E. Kleindienst, M. Jaoui, E. S. Edgerton, R. L.
Tanner, S. L. Shaw, M. Zheng, E. M. Knipping, and J. H. Seinfeld. Influence of
aerosol acidity on the chemical composition of secondary organic aerosol from β-
caryophyllene. Atmospheric Chemistry and Physics, 11(4):1735–1751, February 2011.
ISSN 1680-7324. doi: 10.5194/acp-11-1735-2011.
[53] EPA. Compendium of methods for the determination of inorganic compounds in
air. Technical Report EPA/625/R-96/010a Method IO-4.1, U.S. Environmental
Protection Agency, Center for Environmental Research Information, 1999.
[54] W. C. Keene. Variation of marine aerosol acidity with particle size. Geophysical
Research Letters, 29(7):1101, 2002. ISSN 0094-8276. doi: 10.1029/2001GL013881.
[55] W. C. Keene and D. L. Savoie. The pH of deliquesced sea-salt aerosol in polluted
marine air. Geophysical Research Letters, 25(12):2181–2184, June 1998. ISSN
00948276. doi: 10.1029/98GL01591.
[56] Z. Meng, J. H. Seinfeld, P. Saxena, and Y. P. Kim. Atmospheric Gas-Aerosol
Equilibrium: IV. Thermodynamics of Carbonates. Aerosol Science and Technology,
23(2):131–154, January 1995. ISSN 0278-6826. doi: 10.1080/02786829508965300.
[57] Nenes, A, Pandis, SN, Pilinis, and C. ISORROPIA: A new thermodynamic equilib-
rium model for multiphase multicomponent inorganic aerosols. Aquatic Geochemistry,
4(1):123–152, 1998. doi: 10.1023/A:1009604003981.
Chapter 1. Constraining Aerosol Acidity 53
[58] X. Yao, T. Y. Ling, M. Fang, and C. K. Chan. Size dependence of in situ pH in
submicron atmospheric particles in Hong Kong. Atmospheric Environment, 41(2):
382–393, January 2007. ISSN 13522310. doi: 10.1016/j.atmosenv.2006.07.037.
[59] K. He, Q. Zhao, Y. Ma, and F. Duan. Spatial and seasonal variability of PM2.5
acidity at two Chinese megacities: insights into the formation of secondary inorganic
aerosols. Atmospheric Chemistry & Physics, 12(1377-1395):2012, 2012.
[60] R. Pathak, W. Wu, and T. Wang. Summertime PM2. 5 ionic species in four major
cities of China: nitrate formation in an ammonia-deficient atmosphere. Atmos. Chem.
Phys, 2009.
[61] A. S. Wexler. Atmospheric aerosol models for systems including the ions H+, NH+4 ,
Na+, SO2−4 , NO−3 , Cl−, Br−, and H2O. Journal of Geophysical Research, 107(D14):
4207, 2002. ISSN 0148-0227. doi: 10.1029/2001JD000451.
[62] E. Friese and A. Ebel. Temperature Dependent Thermodynamic Model of the System
H+-NH+4 -Na+-SO2−
4 -NO−3 -Cl−-H2O. The Journal of Physical Chemistry A, 114(43):
11595–11631, October 2010. ISSN 1520-5215. doi: 10.1021/jp101041j.
[63] Z. Meng and J. H. Seinfeld. Time scales to achieve atmospheric gas-aerosol equilibrium
for volatile species. Atmospheric Environment, 30(16):2889–2900, August 1996. ISSN
13522310. doi: 10.1016/1352-2310(95)00493-9.
[64] K. G. Dassios and S. N. Pandis. The mass accommodation coefficient of ammonium
nitrate aerosol. Atmospheric Environment, 33(18):2993–3003, August 1999. ISSN
13522310. doi: 10.1016/S1352-2310(99)00079-5.
[65] C. N. Cruz, K. G. Dassios, and S. N. Pandis. The effect of dioctyl phthalate films on
the ammonium nitrate aerosol evaporation rate. Atmospheric Environment, 34(23):
3897–3905, January 2000. ISSN 13522310. doi: 10.1016/S1352-2310(00)00173-4.
[66] M. Shiraiwa, M. Ammann, T. Koop, and U. Poschl. Gas uptake and chemical aging of
semisolid organic aerosol particles. Proceedings of the National Academy of Sciences
of the United States of America, 108(27):11003–8, July 2011. ISSN 1091-6490. doi:
10.1073/pnas.1103045108.
[67] Q. Sun and A. S. Wexler. Modeling urban and regional aerosols near acid neutral-
ityapplication to the 2425 June scaqs episode. Atmospheric Environment, 32(20):
3533–3545, September 1998. ISSN 13522310. doi: 10.1016/S1352-2310(98)00060-0.
Chapter 1. Constraining Aerosol Acidity 54
[68] J. Zhang, W. L. Chameides, R. Weber, G. Cass, D. Orsini, E. Edgerton, P. Jongejan,
and J. Slanina. An evaluation of the thermodynamic equilibrium assumption for fine
particulate composition: Nitrate and ammonium during the 1999 Atlanta Supersite
Experiment. Journal of Geophysical Research, 108(D7):8414, 2002. ISSN 0148-0227.
doi: 10.1029/2001JD001592.
[69] T. B. Onasch, R. L. Siefert, S. D. Brooks, A. J. Prenni, B. Murray, M. A. Wil-
son, and M. A. Tolbert. Infrared spectroscopic study of the deliquescence and
efflorescence of ammonium sulfate aerosol as a function of temperature. Journal
of Geophysical Research, 104(D17):21317, September 1999. ISSN 0148-0227. doi:
10.1029/1999JD900384.
[70] S. Potukuchi and A. Wexler. Identifying solid-aqueous phase transitions in atmo-
spheric aerosolsI. Neutral-acidity solutions. Atmospheric Environment, 1995.
[71] I. N. Tang and H. R. Munkelwitz. Composition and temperature dependence of
the deliquescence properties of hygroscopic aerosols. Atmospheric Environment.
Part A. General Topics, 27(4):467–473, March 1993. ISSN 09601686. doi: 10.1016/
0960-1686(93)90204-C.
[72] R. K. Pathak, P. K. Louie, and C. K. Chan. Characteristics of aerosol acidity in Hong
Kong. Atmospheric Environment, 38(19):2965–2974, June 2004. ISSN 13522310. doi:
10.1016/j.atmosenv.2004.02.044.
[73] M. Z. Markovic, T. C. VandenBoer, and J. G. Murphy. Characterization and
optimization of an online system for the simultaneous measurement of atmospheric
water-soluble constituents in the gas and particle phases. Journal of Environmental
Monitoring, 14(7), 2012. doi: 10.1039/c2em00004k.
[74] T. C. VandenBoer, M. Z. Markovic, A. Petroff, M. F. Czar, N. Borduas, and J. G.
Murphy. Ion chromatographic separation and quantitation of alkyl methylamines
and ethylamines in atmospheric gas and particulate matter using preconcentration
and suppressed conductivity detection. Journal of chromatography. A, 1252:74–83,
August 2012. ISSN 1873-3778. doi: 10.1016/j.chroma.2012.06.062.
[75] URG. Performance and collection efficiency of the urg 9000 ambient ion monitor
system. Technical report, University Research Glassware Corporation, 116 S. Merritt
Mill Road, Chapel Hill, NC 27516, 2012.
Chapter 1. Constraining Aerosol Acidity 55
[76] R. Ellis, J. Murphy, P. Hayes, M. Cubison, A. Ortega, J. Jiminez, J. Liu, R. Weber,
P. Veres, A. Cochran, and J. Roberts. Gas-particle partitioning of ammonia at the
calnex-la ground site and the influence of aerosol ph. in prep.
[77] P. Winkler. Relations Between Aerosol Acidity and Ion Balance: Vol. 6. NATO
Advanced Science Instituted Series G Ecological Sciences, New York, N.Y., 1986.
[78] M. R. Canagaratna, J. T. Jayne, J. L. Jimenez, J. D. Allan, M. R. Alfarra, Q. Zhang,
T. B. Onasch, F. Drewnick, H. Coe, A. Middlebrook, A. Delia, L. R. Williams, A. M.
Trimborn, M. J. Northway, P. F. DeCarlo, C. E. Kolb, P. Davidovits, and D. R.
Worsnop. Chemical and microphysical characterization of ambient aerosols with the
aerodyne aerosol mass spectrometer. Mass spectrometry reviews, 26(2):185–222, 2007.
ISSN 0277-7037. doi: 10.1002/mas.20115.
[79] A. M. Middlebrook, R. Bahreini, J. L. Jimenez, and M. R. Canagaratna. Evaluation
of Composition-Dependent Collection Efficiencies for the Aerodyne Aerosol Mass
Spectrometer using Field Data. Aerosol Science and Technology, 46(3):258–271,
March 2012. ISSN 0278-6826. doi: 10.1080/02786826.2011.620041.
[80] R. Ellis. Using high resolution measurments and models to investigate the behaviour
of atmospheric ammonia, 2011.
[81] W. C. Keene, M. S. Long, A. A. P. Pszenny, R. Sander, J. R. Maben, A. J. Wall,
T. L. O’Halloran, A. Kerkweg, E. V. Fischer, and O. Schrems. Latitudinal variation
in the multiphase chemical processing of inorganic halogens and related species over
the eastern North and South Atlantic Oceans. Atmospheric Chemistry and Physics,
9(19):7361–7385, 2009. ISSN 1680-7316.
[82] R. G. Stevens, J. R. Pierce, C. A. Brock, M. K. Reed, J. H. Crawford, J. S. Holloway,
T. B. Ryerson, L. G. Huey, and J. B. Nowak. Nucleation and growth of sulfate aerosol
in coal-fired power plant plumes: sensitivity to background aerosol and meteorology.
Atmospheric Chemistry and Physics, 12(1):189–206, January 2012. ISSN 1680-7324.
doi: 10.5194/acp-12-189-2012.
[83] P. Koutrakis, K. M. Thompson, J. M. Wolfson, J. D. Spengler, G. J. Keeler, and J. L.
Slater. Determination of aerosol strong acidity losses due to interactions of collected
particles: Results from laboratory and field studies. Atmospheric Environment.
Part A. General Topics, 26(6):987–995, April 1992. ISSN 09601686. doi: 10.1016/
0960-1686(92)90030-O.
Chapter 1. Constraining Aerosol Acidity 56
[84] W. Nie, T. Wang, X. Gao, R. K. Pathak, X. Wang, R. Gao, Q. Zhang, L. Yang, and
W. Wang. Comparison among filter-based, impactor-based and continuous techniques
for measuring atmospheric fine sulfate and nitrate. Atmospheric Environment, 44(35):
4396–4403, November 2010. ISSN 13522310. doi: 10.1016/j.atmosenv.2010.07.047.
[85] X. Zhu, J. M. Prospero, F. J. Millero, D. L. Savoie, and G. W. Brass. The solubility of
ferric ion in marine mineral aerosol solutions at ambient relative humidities. Marine
Chemistry, 38(1-2):91–107, June 1992. ISSN 03044203. doi: 10.1016/0304-4203(92)
90069-M.
Chapter 2
Reduced Nitrogen Interference in
Chemiluminescent Nitrogen Oxide
Monitors
2.1 Introduction
Reactive nitrogen oxides have wide-ranging implications for air quality by influencing par-
ticulate matter formation, ozone production and the oxidative capacity of the troposphere
[1, 2]. Nitrogen dioxide (NO2) is designated as a criteria pollutant by the United States
Environmental Protection Agency (EPA) and Environment Canada due to its health and
environmental effects and is regulated according to national air quality standards (CFR 40
part 50.11 and CEPA Schedule 1, 63). NO2 readily interconverts with NO, and the sum
of both species is defined as NOx. NOx is primarily formed as a byproduct of combustion,
e.g. from fossil fuel use by mobile sources and in energy production [3]. The lifetime of
NOx in the atmosphere is governed by its oxidation to higher order oxides, which are
subsequently deposited and lost from the atmosphere. The sum of NOx and more highly
oxidized products such as N2O5, peroxyacyl nitrates (PANs), nitric acid (HNO3), nitrous
acid (HONO), alkyl nitrates (RONO2) and particulate nitrate (NO3-), is defined as total
reactive nitrogen oxides (NOy). Non-NOy nitrogen-containing atmospheric species include
nitriles, ammonia, amines, N2O, and HCN.
Due to the interconnected chemistry of NO2 and other nitrogen oxides, accurate NOx
and NOy measurements have remained important elements of understanding processes that
affect air quality. Air quality models rely on accurate and precise in situ measurements
of NO2, and accurate measurements are also necessary for the validation of satellite
57
Chapter 2. Interference in Nitrogen Oxide Monitors 58
measurements and for some public health epidemiological studies [4, 5].
Publicly funded air quality networks, including the Canadian Air and Precipitation
Monitoring Network (CAPMoN) and National Air Pollution Surveillance Program (NAPS)
in Canada and the California Air Resources Board (CARB) and the State or Local Air
Monitoring Stations (SLAMS) in the U.S.A., often use commercial NOx or NOy analyzers
which rely on the chemiluminescent reaction of NO with O3 [6–8]. The primary channel
in these instruments draws ambient air and an O3 stream into the reaction chamber to
quantify the NO concentrations by the light emitted from the chemiluminescent reaction
of NO and O3. The secondary channel reduces oxidized reactive nitrogen species to NO
before detection in the reaction chamber as either NOx, if the converter only reduces NO2,
or NOy, if the converter also oxidizes higher order nitrogen oxides.
Most chemiluminescent NOy instruments use heated molybdenum oxide (MoOx) cata-
lysts to convert nitrogen oxides to NO, and many chemiluminescent NOx instruments use
the same catalysts in an altered configuration to convert NO2 to NO [7, 9, 10]. Since both
NOx and NOy are operationally defined by these instruments as the concentration of NO
found through chemiluminescence in the converter channel, other species that degrade
or convert to NO through non-specific conversion will interfere with the desired signal
[4]. MoOx-based chemiluminescence detectors (MoOx-CLD) have been shown to system-
atically overestimate NOx concentrations when compared with collocated spectroscopic
instruments [9, 11–13]. Both NOx and NOy instruments have also been shown to convert
non-NOy species—including NH3, HCN, and N2O—to NO and such interferences have
been extensively characterized over the past four decades [10, 14–18].
The interference of gaseous ammonia (NH3) and particulate ammonium (NH4+
) from
oxidation within the converter is poorly understood. NH3 is a trace gas that is emitted
from both natural and agricultural sources and NH4+
results from the gas to particle
partitioning of NH3 in the presence of atmospheric trace acids [3]. The interferences of
MoOx-CLD instruments have been considered to be relatively less significant in cities and
other high-NOx environments due to the relatively small concentrations of interfering
species compared with NOx and NOy [9, 19]. However, Bishop et al. [20] showed that
modern three-way catalytic converters may be changing the relative amounts of reduced
and oxidized nitrogen emitted from mobile sources [20]. The implications of increasing
NHx relative to NOx and NOy could mean that in urban environments, NHx interferences
may be more significant than previously thought.
Some studies have suggested a minimal interference of NHx [14, 15], while others have
reported conversions of NH3 as high as 38% under normal operating conditions [21, 22].
Unfortunately, the confusion over the findings has led some researchers to either claim
Chapter 2. Interference in Nitrogen Oxide Monitors 59
that NHx interference is negligible based on prior literature or to neglect to investigate its
role when considering biases in reactive nitrogen oxide measurements [9, 23? ]. Several
studies have observed a gap between measurements from MoOx-CLD measurements and
the sum of individual NOy constituents [24–26]. In NOy chemiluminescent analyzers
with gold catalytic converters, NHx interference has been observed to account for 85%
of the “NOy gap” and a similar phenomenon could be occurring in molybdenum-based
instruments if significant NHx is converting to NO [27].
The variability of NH3 conversion of has been considered a consequence of converter
age—the result of increased oxidation of the molybdenum surface—and converter tem-
perature (Tconv) [10, 21]. The oxidative aging of the converter is a known problem that
instrument manufacturers recommend resolving using a reductive regeneration procedure.
However, the current characterization of the relationship between Tconv and conversion
efficiency is incomplete. Typically, Tconv in NO2 instruments is set at 325C, while Tconv
for NOy is higher, usually up to 350C. However, these values fluctuate throughout the
literature. For example, the Canadian Council of Ministers of the Environment recom-
mends that Tconv for NO2 be set at 325C, while the Environment Canada CAPMoN
program sets Tconv at 325C for NOy instruements [28, 29]. Breitenbach and Shelef [22]
noted that ammonia conversion increased in a molybdenum-carbon composite converter
as they increased the Tconv [22]. Xue et al. [21] observed a decrease in NH3 MoOx-CLD
conversion efficiency from 38% to 11% when lowering Tconv from 350C to 325C [21].
Fehsenfeld et al. [30] developed a MoOx-CLD technique to measure NHx as the difference
between MoOx-CLD measurements taken with a converter at 350 and another at 450
[30]. Their results demonstrate that at a reasonably low Tconv, a MoOx converter can
quantitatively convert NH3 and NH4+
to NO.
Despite the evidence of a strong temperature dependence on the NHx interference,
a survey of the literature results in dramatically different conversion efficiencies as a
function of reported converter temperatures. Work by Fitz et al. [10] attempted to isolate
this problem by first adjusting the Tconv of 14 new analogous converters to a minimal
value while still maintaining a high conversion efficiency of NO2 [10]. The researchers
noted that not only did the optimal temperature for NO2 conversion vary from 327C
to 370C between 14 analogous converters, but the NH3 conversion rate at the optimal
temperature ranged from 2.5% to 26.8% with an average of 11.2% and was not correlated
with the Tconv [10]. Conversion efficiencies from selected previous studies with reported
MoOx Tconv are shown in Table 2.1. The relationship of temperature and NH3 conversion
remains unclear, and the influence of NH4+
also has not been thoroughly studied.
In this work we investigate the relationship between Tconv and NH3 and NH4+
with
Chapter 2. Interference in Nitrogen Oxide Monitors 60
Table 2.1: Summary of selected literature gaseous NH3 MoOx-CLD conversion efficienciesand associated Tconv
Reference Tconv (C) Conversion EfficiencyXue et al. [21] 350 3 converters: 0-14%
350 1 New converter: 38%325 Same new converter: 11%
Fitz et al. [10] 327-370: 14 Converters 2.5%-26.8%(no correlation with Tconv)
Fehsenfeld et al. [30] 350 negligible450 100%
Williams et al. [15] 350: Instrument 1 0%375: Instrument 2 5%340: Instrument 3 8%
Dickerson et al. [31] 425 <0.1%Williams et al. [32] 400 37%Fehsenfeld et al. [14] 400 2%
500 Substantial ConversionBreitenbach and Shelef [22] 525 7%? ] Unreported <1%Minarro and Ferradas [33] Unreported <1%Dunlea et al. [11] Unreported No correlation between ambient
NH3 and excess NOx
three MoOx-CLD instruments. Our objective is to elucidate the causes of the confusion
in the existing literature regarding this problem and offer practical recommendations to
minimize the effects of the interference.
2.2 Methods
2.2.1 Analytical instrumentation
The experimental setup consisted of an Ambient Ion Monitor-Ion Chromatograph (AIM-
IC, URG Corp., Chapel Hill, NC), a custom-built NO chemiluminescence analyzer with a
2-channel NOx and NOy inlet system (AQD NOxy, Air Quality Design Inc., Wheat Ridge,
CO) and two commercial chemiluminescence NO-NO2-NOx analyzers (TSI NOx, Thermo
Scientific, Model 42i, Franklin, MA). AIM-IC provided simultaneous measurements of gas
and particle phase NHx and other water-soluble species while the AQD NOxy and TSI
NOx analyzer were utilized to characterize molybdenum converter response to NH3 and
Chapter 2. Interference in Nitrogen Oxide Monitors 61
MoOx Output
Ozone Generator
NO2 to NO Converter
Flow Sensor
NO Channel
NO2 Channel
Pressure Transducer
Dry Air
Sample
Exhaust
Reaction Chamber
Filter
PMT
Pump Ozone Scrubber
Flow
Flow
Flow
Figure 2.1: Schematic of TSI NOx analyzers (adapted from TSI 42i Manual)
NH4+
.
The AQD NOxy inlet consists of two inlet channels that separately measure NOx and
NOy by conversion to NO. The NOx channel utilizes a photolytic converter (Blue Light
Converter, Air Quality Design, Wheat Ridge, CO) to selectively convert NO2 to NO,
while the NOy channel relies on a MoOx converter to reduce NOy species to NO. The
converted air flows at 1.5 standard L min−1 through the ozone reaction chamber and
subsequent chemiluminescence is detected by a PMT. The TSI NOx instruments (Fig.
2.1) are functionally analogous to the NOy channel of the NOxy as ambient air enters
the MoOx converter and is detected in an ozone reaction chamber. The flow through
each channel in the TSI NOx instruments was 0.6 volumetric L min−1. The Air Quality
Design and Thermo Scientific instruments utilized equivalent, commercially available
MoOx converters (PN: 9269, Thermo Environmental, Inc.).
AIM-IC simultaneously measures atmospheric gases, including NH3, SO2, HNO3, and
HONO, and particulate species, including NH4+
, SO42-
, NO3-, and NO2
-, at 3.00 volumetric
L min−1 with hourly time resolution. A parallel-plate wet denuder strips water-soluble
gases from the air sample and oxidizes SO2 to SO42-
with dilute (1 mmol) H2O2 solution.
The remaining particles and non-soluble gases are drawn into a particle supersaturation
chamber (PSSC) and condense into a liquid phase. The effluents from the denuder and
Chapter 2. Interference in Nitrogen Oxide Monitors 62
PSSC are drawn into two respective sets of 5 mL syringes over the course of an hour
before injection onto cation and anion ion chromatographs.
Permeation sources of HNO3 and NH3 (Type HRT for HNO3 and EL-SRT-2 for NH3,
Kin-Tek, La Marque, TX) were utilized to produce constant signals for assessing interfer-
ences. The permeation rate for the NH3 source may have varied between experiments and
it was necessary to evaluate the permeation output using AIM-IC to quantify conversion
efficiency. The methodology of quantitation is described in further detail in Markovic
et al. [34].
2.2.2 Converter temperature ramping with ammonia gas
The first set of experiments was designed to isolate the relationship of Tconv and NH3
conversion. Temperature ramping experiments were performed on the AQD NOxy by
diluting NH3 from a permeation source into zero air and observing the relationship between
the NOy mixing ratios and converter temperature. Tconv was increased stepwise between
275C and 375C and was held constant at each step for approximately 30 minutes. This
experiment was replicated five times with the aged converter which had been originally
installed on the AQD NOxy and once with a previously unused replacement converter from
one of the TSI NOx instruments. Between each set of experiments, the physical placement
of the thermocouple and heating unit changed due to heating insulation being unpacked
and repacked. The AIM-IC was only implemented to simultaneously monitor the NH3
gas concentration during the final experiment as it became clear that the permeation rate
of NH3 had not remained constant between the first five sets of temperature ramping
experiments.
A similar temperature ramping experiment was performed once on both TSI NOx
instruments. In these instruments Tconv is electronically limited between 310C and
340C. Tconv was increased stepwise by 5C and held at each step for approximately 30
minutes. Both commercial NOx analyzers (henceforth denoted as TSI NOx–A and –B)
were purchased approximately one year prior to the experiments, but TSI NOx–A had
been operated continuously over that time while TSI NOx–B had been operated for <1
week prior to the ramping experiments. While the AIM-IC was not used during this
experiment to simultaneously measure NH3 mixing ratios, the NH3 permeation source was
calibrated shortly before use and was assumed to have remained at the same permeation
rate.
Chapter 2. Interference in Nitrogen Oxide Monitors 63
2.2.3 Converter temperature ramping with salt particles
In order to analyze the conversion efficiency of particle NH4+
and NO3-, solutions of
(NH4)2SO4, NH4NO3 and Ca(NO3)2, were atomized (Aerosol Generator ATM 226, Topas
GmbH, Dresden, Germany) and dried through a drying tube (DDU 570/L, Topas GmbH,
Dresden, Germany) packed with silica beads (P077.1, Carl Roth GmbH, Karlsruhe,
Germany) to produce polydisperse, internally mixed particles with a maximum number
concentration between 300 and 500 nm (Topas ATM 226 Manual). These salts were
chosen to isolate conversion behaviour between salts with NOy and non-NOy particulate
nitrogenous species. In addition, high purity deionized water (DIW, >18 MΩ, Easypure
RoDI II, Barnstead Inc. Dubuque, IA) was atomized to provide a background measurement.
The particles were directed into a teflon T-junction and split through PFA tubing toward
the AQD NOxy and AIM–IC. A temperature ramp was performed on the AQD NOxy
while the AIM-IC monitored the mass loadings of particulate species. No particle size
selection device was implemented on either the AQD NOxy or AIM-IC. 80% particle
collection efficiency was estimated for the AIM-IC due to high mass loadings of particles
and based on measurements taken from the exhaust of the inlet.
2.3 Results
2.3.1 Ammonia gas conversion efficiency trends in commercial
nitrogen dioxide analyzers
The temperature ramping experiments for NH3 showed a strong positive linear correlation
between conversion efficiency and Tconv in the TSI NOx instruments (Fig. 2.2). Within the
electronically limited temperature range of 310-340C, there was also significant difference
between the absolute conversion efficiencies of the two instruments. TSI NOx–A, which
had been operated continuously for a year, had an elevated conversion efficiency compared
to TSI NOx–B, which had been operated for less than a week. Furthermore, immediately
after TSI NOx–B was first calibrated after purchase, it was noted that addition of ammonia
resulted in negligible conversion. Halfway through the temperature ramping experiment,
TSI NOx–B experienced an unrelated instrumental failure and was unavailable for further
analysis.
Chapter 2. Interference in Nitrogen Oxide Monitors 64
0.30
0.25
0.20
0.15
0.10
Con
vers
ion
Effi
cien
cy
340335330325320315310Converter Temperature (°C)
TSI NOx-A TSI NOx-B
Figure 2.2: Conversion efficiencies of NH3(g) in TSI NOx analyzers
2.3.2 Ammonia gas conversion efficiency trends in total reactive
nitrogen oxide analyzer
The temperature conversion experiment was repeated six times with the AQD NOxy and
each experiment is plotted as a function of the fraction of the maximum signal against
the reported converter temperature in Fig. 2.3. The sigmoidal curve shape observed
in these experiments contrasts with the linear relationship in the TSI NOx instruments.
This result may be explained by the relatively narrow temperature range in the TSI NOx
instruments that does not allow observation of the non-linear relationship at the extremes
of Tconv.
Throughout the AQD NOxy temperature ramping experiments, the sigmoidal shape
of the conversion curve remained the same while the temperatures that corresponded to
given fractions of maximum signal varied. We hypothesize that these results are due to the
irreproducibility of position of the converter’s temperature probe and local environment
upon physical disturbance of the converter and its housing unit due to adjustments to the
insulation in the converter housing throughout the experiments. Due to the fluctuation
of the permeation source over the course of the six temperature ramping experiments, the
fraction of maximum detected signal (rather than conversion efficiency) as a function of
reported temperature in the AQD NOxy is displayed in Fig. 2.3. The February 7, 2012
Chapter 2. Interference in Nitrogen Oxide Monitors 65
temperature ramp experiment was performed with a previously unused converter taken
from the TSI NOx–B instrument while the rest of the runs were performed with the aged
converter from the AQD NOxy. The AIM-IC was implemented during the April, 16 2012
temperature ramp experiment to give the estimated conversion efficiency as a function
of reported converter temperature as seen in Fig. 2.4. We also noted that there was no
temperature dependence of HNO3—an important NOy species—within the Tconv range,
as shown in Fig. 2.4.
1.0
0.8
0.6
0.4
0.2
0.0
Fra
ctio
n of
Max
imum
Sig
nal
380360340320300280260Reported Converter Temperature (°C)
June 27, 2011 January 20, 2012 January 23, 2012 February 7, 2012 February 24, 2012 April 16, 2012
Figure 2.3: NH3(g) profiles in AQD NOxy
2.3.3 Salt conversion efficiency trends in total reactive nitrogen
oxide analyzer
The conversion efficiency of NO3-
and NH4+
salts were plotted as functions of reported
temperature in Fig. 2.5. The (NH4)2SO4 curve exhibits a similar conversion profile
to that of NH3; at the lowest temperatures the conversion efficiency is very low, but
reaches near unity at the highest temperatures. Ca(NO3)2 exhibits 50% conversion at
low temperatures followed by a linear increase to near unity at the highest temperatures.
Finally, NH4NO3 also exhibits the sigmoidal trend, but the conversion efficiency begins at
50% at the lowest temperature. Due to the low degradation temperature of NH4NO3, the
high initial conversion may be due to the complete conversion of NO3-
at low temperatures
Chapter 2. Interference in Nitrogen Oxide Monitors 66
1.0
0.8
0.6
0.4
0.2
0.0
Con
vers
ion
Effi
cien
cy
360340320300280Reported Converter Temperature (°C)
HNO3(g)
NH3(g)
Figure 2.4: Conversion efficiencies of NH3(g) and HNO3(g) in AQD NOxy with simultaneousAIM-IC monitoring
followed by the increasing conversion of NH4+
with temperature [35]. The difference
between the low-temperature conversion of particulate NO3-
in NH4NO3 and Ca(NO3)2
may be due to the relative susceptibility of NH4NO3 to thermal degradation compared
Ca(NO3)2 [36, 37].
Chapter 2. Interference in Nitrogen Oxide Monitors 67
1.0
0.8
0.6
0.4
0.2
Con
vers
ion
Effi
cien
cy
360340320300280Reported Converter Temperature (°C)
NH4NO3
Ca(NO3)2
(NH4)2SO4
Figure 2.5: Conversion efficiencies of nitrogen-containing salts in AQD NOxy with simul-taneous AIM-IC monitoring
2.4 Discussion
There are several major results which highlight the importance of understanding of
the NHx interference in MoOx-CLD reactive nitrogen instruments. The first is that
the absolute conversion efficiencies between two effectively identical commercial NOx
analyzers at any given temperature were significantly different. This result is consistent
with previous studies that reported variations between operationally similar instruments.
While the exact cause of the variations in conversion efficiency could not be isolated, we
speculate that the converter age, local environment in or around the converter, or the
calibration of the temperature probe could all be responsible.
The second major result is that the trend of increasing NH3 conversion efficiency with
increasing temperature was consistently observed. However, the AQD NOxy exhibited
substantial variability in the temperature required to produce maximum conversion
between the runs. This internal variability may be explained by the high sensitivity of
conversion efficiency to small changes in temperature between 300C and 360C. Error
in the temperature readout (perhaps caused by changes to the local environment of the
converter) could create the broad sensitivity range that allowed the AQD NOxy to convert
between 20% and >95% of the maximum signal at a reported temperature of 325C (a
Chapter 2. Interference in Nitrogen Oxide Monitors 68
typical operating temperature). Although uncertainty in Tconv may lead to the differences
in reported conversion efficiencies, it is also possible that manufacturing inadequacies that
lead to so-called “hot spots” within the converter or converter aging is impacting the
absolute conversion efficiency of the instrument [14]. This study does not preclude the
possibility of these factors but rather suggests that the effect of all of these problems may
only be accounted for by direct measurement of NH3 conversion efficiency.
The results obtained in this study contradict the interpretation of prior literature that
the conversion efficiency of NH3 is not temperature dependent [21]. Although prior studies
have observed dramatically different conversion efficiencies across a range of Tconv, this
study illustrates a critical shortcoming of the approach of reporting a single conversion
efficiency value at a given Tconv. If the position of the conversion efficiency curve on the
Tconv axis varies between instruments or converters due to bias or uncertainty in the readout
temperature, then previous studies may have observed accurate conversion efficiencies, but
either the absolute converter temperature was incorrect or converter aging had changed the
converter’s oxidizing ability. Since no study systematically characterized the relationship
between conversion efficiency and Tconv of a single converter and instrument, the results
could justifiably conflict.
Finally, NH4+
also exhibited similar conversion efficiency trends to NH3. Fehsenfeld
et al. [30] indirectly observed NH4+
interference, but this study directly confirms the
NH4+
interference and suggests that NH4+
must be considered along with NH3 to fully
characterize NHx interference.
2.5 Conclusions
Modern photolytic converters are replacing MoOx converters for chemiluminescent NOx
instrumentation and have been shown to closely correlate with spectroscopic NO2 measure-
ments via cavity ring-down spectroscopy (CARDS), laser-induced fluorescence (LIF), and
differential absorption spectroscopy (DOAS) [38–40]. However, many molybdenum con-
verter NOx instruments remain in usage and on the market, and MoOx-CLD instruments
are the principle method for determining NOy. Consequently, understanding the influence
of interferences is crucial to obtaining accurate measurements from these instruments.
Conventional analysis of field measurements taken by MoOx chemiluminescent instru-
ments characterizes MoOx conversion of NHx as being at or around a fixed percentage.
However, this approach may cause researchers to underestimate the true variability of NHx
conversion efficiency through neglecting to individually characterize their instruments.
Our research demonstrates the importance of regular characterization of NHx interference
Chapter 2. Interference in Nitrogen Oxide Monitors 69
and keeping the converter temperature as low as possible while still maintaining high
conversion efficiency of desired species (e.g. NO2, HNO3, etc.). Furthermore, the high
sensitivity of the conversion efficiency of NHx and uncertainty of instrumental reported
temperature helps explain the disagreements in prior literature. For NOx analyzers,
intoducing particle filters and NH3 denuders may effectively reduce the NHx interference,
however these devices may be impractical for NOy instrumentation since they may supress
particulate NO3-
and gas-phase HNO3. Further studies should be performed to determine
the best methods for reducing the amount of NHx entering the analyzers.
2.6 Acknowledgements
The author wishes to acknowledge the Abbatt and Evans research groups at the University
of Toronto for usage of the TSI NOx instruments. Jeff Geddes processed and analyzed all
of the AQD NOxy data and Jennifer Murphy provided guidance and support throughout
the experiments. The authors also wish to thank Jon Wang, Alex Tevlin, Greg Wentworth,
Angela Hong, and Stephanie Pugliese for assistance with various instrumental operations
throughout the experiment. Finally, the author acknowledges the NSERC CREATE
program, IACPES, for continued financial support.
2.7 Bibliography
[1] W. Chameides and J. C. G. Walker. Photochemical theory of tropospheric ozone. Jour-
nal of Geophysical Research, 78(36):8751–8760, 1973. doi: 10.1029/JC078i036p08751.
[2] P. Crutzen. Discussion of chemistry of some minor constituents in stratosphere
and troposphere. Pure and Applied Geophysics, 106(5-7):1385–1399, 1973. doi:
10.1007/bf00881092.
[3] E. Holland, F. Dentener, B. Braswell, and J. Sulzman. Contemporary and pre-
industrial global reactive nitrogen budgets. Biogeochemistry, 46(1-3):7–43, July 1999.
doi: 10.1023/A:1006148011944.
[4] W. C. McClenny. Recommended methods for ambient air monitoring of NO, NO2,
NOy, and individual NOz species. Technical Report EPA/600/R-01/005, U.S. En-
vironmental Protection Agency, National Exposure Research Laboratory, Research
Triangle Park, NC, 2000.
Chapter 2. Interference in Nitrogen Oxide Monitors 70
[5] C. Ordonez, A. Richter, M. Steinbacher, C. Zellweger, H. Nuss, J. Burrows, and A. Pre-
vot. Comparison of 7 years of satellite-borne and ground-based tropospheric NO2
measurements around Milan, Italy. Journal of Geophysical Research-Atmospheres,
111(D5):D05310, MAR 9 2006. ISSN 0148-0227. doi: 10.1029/2005JD006305.
[6] R. Cohen. Analysis of satellite measurements to improve California’s models of O3
and PM. Technical Report ARB 06-328, California Air Resources Board, Unveristy
of California-Berkeley, Berkeley, CA 94720, 2010.
[7] N. Watkins and R. Baldauf. Near-road NO2 monitoring technical document. Technical
Report EPA-454/B-12-002, U.S. Environmental Protection Agency, Ambient Air
Monitoring Group, Research Triangle Park, NC, 2012.
[8] NAPS. National air pollution surveillance network quality assurance and quality con-
trol guidelines. Technical Report AAQD 2004-1, Environment Canada, Environmental
Technology Centre, Ottawa, ON K1A 0H3, 2004.
[9] M. Steinbacher, C. Zellweger, B. Schwarzenbach, S. Bugmann, B. Buchmann, C. Or-
donez, A. S. H. Prevot, and C. Hueglin. Nitrogen oxide measurements at rural sites
in Switzerland: Bias of conventional measurement techniques. Journal of Geophysical
Research-Atmospheres, 112(D11), 2007. doi: 10.1029/2006jd007971.
[10] D. R. Fitz, K. Bumiller, and A. Lashgari. Measurement of NOy during the SCOS97-
NARSTO. Atmospheric Environment, 37, 2003. doi: 10.1016/s1352-2310(03)00385-6.
[11] E. J. Dunlea, S. C. Herndon, D. D. Nelson, R. M. Volkamer, F. San Martini, P. M.
Sheehy, M. S. Zahniser, J. H. Shorter, J. C. Wormhoudt, B. K. Lamb, E. J. Allwine,
J. S. Gaffney, N. A. Marley, M. Grutter, C. Marquez, S. Blanco, B. Cardenas,
A. Retama, C. R. Ramos Villegas, C. E. Kolb, L. T. Molina, and M. J. Molina.
Evaluation of nitrogen dioxide chemiluminescence monitors in a polluted urban
environment. Atmospheric Chemistry and Physics, 7(10):2691–2704, 2007.
[12] H. Suzuki, Y. Miyao, T. Nakayama, J. K. Pearce, Y. Matsumi, K. Takahashi, K. Kita,
and K. Tonokura. Comparison of laser-induced fluorescence and chemiluminescence
measurements of NO2 at an urban site. Atmospheric Environment, 45(34), 2011. doi:
10.1016/j.atmosenv.2011.07.065.
[13] G. Villena, I. Bejan, R. Kurtenbach, P. Wiesen, and J. Kleffmann. Interferences
of commercial NO2 instruments in the urban atmosphere and in a smog chamber.
Atmospheric Measurement Techniques, 5(1), 2012. doi: 10.5194/amt-5-149-2012.
Chapter 2. Interference in Nitrogen Oxide Monitors 71
[14] F. C. Fehsenfeld, R. R. Dickerson, G. Hubler, W. T. Luke, L. J. Nunnermacker, E. J.
Williams, J. M. Roberts, J. G. Calvert, C. M. Curran, A. C. Delany, C. S. Eubank,
D. W. Fahey, A. Fried, B. W. Gandrud, A. O. Langford, P. C. Murphy, R. B. Norton,
K. E. Pickering, and B. A. Ridley. A ground-based intercomparison of NO, NOx,
and NOy measurement techniques. Journal of Geophysical Research-Atmospheres, 92
(D12):14710–14722, 1987. doi: 10.1029/JD092iD12p14710.
[15] E. J. Williams, K. Baumann, J. M. Roberts, S. B. Bertman, R. B. Norton,
F. C. Fehsenfeld, S. R. Springston, L. J. Nunnermacker, L. Newman, K. Olszyna,
J. Meagher, B. Hartsell, E. Edgerton, J. R. Pearson, and M. O. Rodgers. Inter-
comparison of ground-based NOy measurement techniques. Journal of Geophysical
Research-Atmospheres, 103(D17), 1998. doi: 10.1029/98jd00074.
[16] A. M. Winer, J. W. Peters, J. P. Smith, and J. N. Pitts. Response of commer-
cial chemiluminescent NO-NO2 analyzers to other nitrogen-containing compounds.
Environmental Science and Technology, 8(13), 1974. doi: 10.1021/es60098a004.
[17] A. Fontijn, A. J. Sabadell, and R. J. Ronco. Homogeneous chemiluminescent measure-
ment of nitric oxide with ozone - implications for continuous selective monitoring of
gaseous air pollutants. Analytical Chemistry, 42(6), 1970. doi: 10.1021/ac60288a034.
[18] D. Grosjean and J. Harrison. Response of chemi-luminescence NOx analyzers and
ultraviolet ozone analyzers to organic air-pollutants. Environmental Science and
Technology, 19(9), 1985. doi: 10.1021/es00139a016.
[19] EPA. Code of federal regulations of the clean air act air regulations. Technical
Report 40, Part 50, Appendix F, U.S. Environmental Protection Agency, 2007.
[20] G. A. Bishop, A. M. Peddle, D. H. Stedman, and T. Zhan. On-road emission mea-
surements of reactive nitrogen compounds from three california cities. Environmental
Science and Technology, 44(9):3616–3620, 2010. doi: 10.1021/es903722p.
[21] L. Xue, T. Wang, J. Zhang, X. Zhang, C. Poon, A. Ding, X. Zhou, W. Wu, J. Tang,
Q. Zhang, et al. Source of surface ozone and reactive nitrogen speciation at mount
waliguan in western china: New insights from the 2006 summer study. Journal of
Geophysical Research-Atmospheres, 116(D7), 2011. doi: 10.1029/2010jd014735.
[22] L. Breitenbach and M. Shelef. Development of a method for analysis of NO2 and NH3
by NO-measuring instruments. Journal of the Air Pollution Control Association, 23
(2):128–131, 1973.
Chapter 2. Interference in Nitrogen Oxide Monitors 72
[23] W. T. Luke, P. Kelley, B. L. Lefer, J. Flynn, B. Rappenglueck, M. Leuchner, J. E.
Dibb, L. D. Ziemba, C. H. Anderson, and M. Buhr. Measurements of primary trace
gases and NOy composition in Houston, Texas. Atmospheric Environment, 44(33),
2010. doi: 10.1016/j.atmosenv.2009.08.014.
[24] V. Aneja, D. Kim, M. Das, and B. Hartsell. Measurements and analysis of reactive
nitrogen species in the rural troposphere of southeast United States: Southern
oxidant study site SONIA. Atmospheric Environment, 30(4):649–659, FEB 1996.
ISSN 1352-2310. doi: 10.1016/1352-2310(95)00294-4.
[25] T. Thornberry, M. Carroll, G. Keeler, S. Sillman, S. Bertman, M. Pippin, K. Ostling,
J. Grossenbacher, P. Shepson, O. Cooper, J. Moody, and W. Stockwell. Observations
of reactive oxidized nitrogen and speciation of NOy during the PROPHET summer
1998 intensive. Journal of Geophysical Research-Atmospheres, 106(D20):24359–24386,
OCT 27 2001. ISSN 0747-7309. doi: 10.1029/2000JD900760.
[26] D. Day, M. Dillon, P. Wooldridge, J. Thornton, R. Rosen, E. Wood, and R. Co-
hen. On alkyl nitrates, O3, and the “missing NOy”. Journal of Geophysi-
cal Research-Atmospheres, 108(D16):4501, AUG 20 2003. ISSN 0148-0227. doi:
10.1029/2003JD003685.
[27] R. Harrison, J. Grenfell, S. Yamulki, K. Clemitshaw, S. Penkett, J. Cape, and G. Mc-
Fadyen. Budget of NOy species measured at a coastal site. Atmospheric Environment,
33(26):4255–4272, NOV 1999. ISSN 1352-2310. doi: 10.1016/S1352-2310(99)
00176-4.
[28] CCME. Ambient Air Monitoring Protocol for PM2.5 and Ozone Canada Wide
Standards. Technical report.
[29] CAPMoN. Nitrogen measurements, Retrieved April, 2013. URL http://www.ec.gc.
ca/rs-mn/default.asp?lang=En&n=752CE271-1.
[30] F. C. Fehsenfeld, L. G. Huey, E. Leibrock, R. Dissly, E. Williams, T. B. Ryerson,
R. Norton, D. T. Sueper, and B. Hartsell. Results from an informal intercomparison
of ammonia measurement techniques. Journal of Geophysical Research-Atmospheres,
107(D24):4812, 2002. doi: 10.1029/2001jd001327.
[31] R. R. Dickerson, A. C. Delany, and A. F. Wartburg. Further modification of a
commercial NOx detector for high-sensitivity. Review of Scientific Instruments, 55
(12):1995–1998, 1984. doi: 10.1063/1.1137694.
Chapter 2. Interference in Nitrogen Oxide Monitors 73
[32] E. J. Williams, S. T. Sandholm, J. D. Bradshaw, J. S. Schendel, A. O. Langford, P. K.
Quinn, P. J. Lebel, S. A. Vay, P. D. Roberts, R. B. Norton, B. A. Watkins, M. P.
Buhr, D. D. Parrish, J. G. Calvert, and F. C. Fehsenfeld. An intercomparison of 5
ammonia measurement techniques. Journal of Geophysical Research-Atmospheres,
97(D11):11591–11611, 1992. doi: 10.1029/92jd00721.
[33] M. D. Minarro and E. G. Ferradas. Performance evaluation of two commercial
chemiluminescence NOx analysers according to European Standard EN 14211. Journal
of Environmental Monitoring, 14(2):383–390, 2012. doi: 10.1039/c1em10601e.
[34] M. Z. Markovic, T. C. VandenBoer, and J. G. Murphy. Characterization and
optimization of an online system for the simultaneous measurement of atmospheric
water-soluble constituents in the gas and particle phases. Journal of Environmental
Monitoring, 14(7), 2012. doi: 10.1039/c2em00004k.
[35] B. J. Wood and H. Wise. Acid catalysis in the thermal decomposition of ammonium
nitrate. The Journal of Chemical Physics, 23:693, 1955.
[36] K. Brower, J. Oxley, and M. Tewari. Evidence for homolytic decomposition of
ammonium-nitrate at high-temperature. Journal of Physical Chemistry, 93(10):
4029–4033, MAY 18 1989. ISSN 0022-3654. doi: 10.1021/j100347a033.
[37] C. Ettarh and A. K. Galwey. A kinetic and mechanistic study of the thermal
decomposition of calcium nitrate. Thermochimica Acta, 288(1-2):203–219, 1996.
[38] H. D. Osthoff, S. S. Brown, T. B. Ryerson, T. J. Fortin, B. M. Lerner, E. J. Williams,
A. Pettersson, T. Baynard, W. P. Dube, S. J. Ciciora, and A. R. Ravishankara.
Measurement of atmospheric NO2 by pulsed cavity ring-down spectroscopy. Journal
of Geophysical Research-Atmospheres, 111(D12), 2006. doi: 10.1029/2005jd006942.
[39] J. A. Thornton, P. J. Wooldridge, R. C. Cohen, E. J. Williams, D. Hereid, F. C.
Fehsenfeld, J. Stutz, and B. Alicke. Comparisons of in situ and long path measure-
ments of NO2 in urban plumes. Journal of Geophysical Research-Atmospheres, 108
(D16), 2003. doi: 10.1029/2003jd003559.
[40] H. Fuchs, S. M. Ball, B. Bohn, T. Brauers, R. C. Cohen, H. P. Dorn, W. P. Dube, J. L.
Fry, R. Haseler, U. Heitmann, R. L. Jones, J. Kleffmann, T. F. Mentel, P. Musgen,
F. Rohrer, A. W. Rollins, A. A. Ruth, A. Kiendler-Scharr, E. Schlosser, A. J. L.
Shillings, R. Tillmann, R. M. Varma, D. S. Venables, G. V. Tapia, A. Wahner,
R. Wegener, P. J. Wooldridge, and S. S. Brown. Intercomparison of measurements
Chapter 2. Interference in Nitrogen Oxide Monitors 74
of NO2 concentrations in the atmosphere simulation chamber SAPHIR during the
NO3Comp campaign. Atmospheric Measurement Techniques, 3(1):21–37, 2010.
Appendix A
AIM-IC calibration and
instrumental information
The following lists the models of various Dionex IC components that were used specifically
for the AIM-IC operation. The eluent is produced from eluent generator cartridges (EGC)
and is recycled through the suppressor and continuously regenerated trap column (CR-TC).
In CONTACT-2012, both anion and cation systems used the EGC model III. The eluent
in the cation IC was methanesulphonic acid (MSA) while the anion IC used potassium
hydroxide (KOH). The cation concentrators, guard columns, and analytical columns
were models TCC-ULP1, CG17, and CS17, respectively, while the anion analogues were
TAC-ULP1, AG19, and AS19, respectively. The suppressor model was the respective
anion and cation versions of the self-regenerating suppressor (ASRS-300 and CSRS-300).
A.1 Gradient eluent chromatographic programs
Table A.1: Cation IC gradient eluent program set points. Flow rate is set to 1.0 mL/minand column temperature is set to 30.
Time (min) MSA Concentration (mM)0.0 2.009.0 2.0011.0 6.0016.0 10.0022.0 10.0027.0 2.00
75
Appendix A. AIM-IC calibration and instrumental information 76
Table A.2: Anion IC gradient eluent program set points. Flow rate is set to 1 mL/minand column temperature is set to 30
Time KOH Concentration (mM)0.0 1.0012.1 20.0018.0 85.0025.0 1.00
A.2 Non-linear organic acids and ammonium
Table A.3: Summary of calibration information for Acetic Acid from the CONTACT-2012field campaign for the AIM-IC
Date A’ value A’ σ B’ value B’ σ χ2
Gas StandardsJuly 24 1.06 0.257 6.56 ∗ 108 3.7 ∗ 107 0.0523September 7 2.37 0.82 7.03 ∗ 108 9.6 ∗ 107 0.164October 10 2.11 0.96 4.68 ∗ 108 1.4 ∗ 108 0.0123Particle StandardsJuly 24 1.75 0.30 8.26 ∗ 108 4.3 ∗ 107 0.0424September 7 2.27 1.01 8.02 ∗ 108 1.32 ∗ 108 0.309October 10 1.76 0.27 7.64 ∗ 108 6.0 ∗ 107 0.00394
Table A.4: Summary of calibration information for Formic Acid from the CONTACT-2012field campaign for the AIM-IC
Date A’ value A’ σ B’ value B’ σ χ2
Gas StandardsJuly 24 31.94 1.02 1.80 ∗ 1010 3.6 ∗ 108 0.0219September 7 47.53 13.9 2.29 ∗ 1010 4.71 ∗ 109 1.23October 10 475.1 497 1.51 ∗ 1011 1.63 ∗ 1011 0.0759Particle StandardsJuly 24 52.97 5.23 2.64 ∗ 1010 1.86 ∗ 109 0.146September 7 55.68 17.8 2.97 ∗ 1010 6.75 ∗ 109 1.74October 10 77.03 43.8 3.66 ∗ 1010 1.94 ∗ 1010 0.0245
Appendix A. AIM-IC calibration and instrumental information 77
Table A.5: Summary of calibration information for Oxalic Acid from the CONTACT-2012field campaign for the AIM-IC
Date A’ value A’ σ B’ value B’ σ χ2
Gas StandardsJuly 24 64.74 7.12 7.11 ∗ 1010 6.37 ∗ 109 0.0330September 7 52.22 3.89 6.00 ∗ 1010 3.5 ∗ 109 0.0207October 10 38.46 8.54 4.39 ∗ 1010 8.98 ∗ 109 0.00206Particle StandardsJuly 24 44.38 2.66 4.90 ∗ 1010 2.25 ∗ 109 0.0135September 7 46.756 3.02 5.41 ∗ 1010 2.68 ∗ 109 0.0174October 10 41.64 11.1 4.697 ∗ 1010 1.16 ∗ 1010 0.00257
Table A.6: Summary of calibration information for NH4+
from the CONTACT-2012 fieldcampaign for the AIM-IC
Date v v σ χ2
volume = 2.41mLGas StandardsJuly 24 4.33 ∗ 105 6.46 ∗ 103 0.02214September 7 4.25 ∗ 105 6.00 ∗ 103 0.0190October 10 3.82 ∗ 105 3.62 ∗ 103 0.0070Particle StandardsJuly 24 4.11 ∗ 105 2.63 ∗ 103 0.00367September 7 4.14 ∗ 105 1.90 ∗ 103 0.00192October 10 3.85 ∗ 105 3.95 ∗ 103 0.00828
Appendix A. AIM-IC calibration and instrumental information 78
A.3 Significant linear inorganic species
Table A.7: Summary of calibration information for Cl-
from the CONTACT-2012 fieldcampaign for the AIM-IC
Date slope (µS min mol−1) slope σ (µS min mol−1) r2
Gas StandardsJuly 24 4.46 ∗ 108 3.21 ∗ 107 0.985September 7 4.33 ∗ 108 4.34 ∗ 107 0.971New ConcentratorSeptember 10 4.31 ∗ 108 4.60 ∗ 107 0.967October 10 2.41 ∗ 108 9.91 ∗ 106 0.995Particle StandardsJuly 24 4.21 ∗ 108 8.08 ∗ 107 0.900September 7 4.21 ∗ 108 2.88 ∗ 107 0.986October 10 4.22 ∗ 108 7.88 ∗ 106 0.999
Table A.8: Summary of calibration information for NO2-
from the CONTACT-2012 fieldcampaign for the AIM-IC
Date slope (µS min mol−1) slope σ (µS min mol−1) r2
Gas StandardsJuly 24 5.50 ∗ 108 4.54 ∗ 106 0.9998September 7 4.20 ∗ 108 3.24 ∗ 106 0.9998New ConcentratorSeptember 10 4.01 ∗ 108 7.14 ∗ 106 0.999October 10 2.50 ∗ 108 4.90 ∗ 106 0.999Particle StandardsJuly 24 3.90 ∗ 108 1.19 ∗ 107 0.997September 7 3.98 ∗ 108 1.11 ∗ 107 0.998October 10 3.48 ∗ 108 1.50 ∗ 106 0.9999
Appendix A. AIM-IC calibration and instrumental information 79
Table A.9: Summary of calibration information for NO3-
from the CONTACT-2012 fieldcampaign for the AIM-IC
Date slope (µS min mol−1) slope σ (µS min mol−1) r2
Gas StandardsJuly 24 4.50 ∗ 108 2.43 ∗ 106 0.9999September 7 4.36 ∗ 108 6.13 ∗ 105 0.99999New ConcentratorSeptember 10 4.40 ∗ 108 1.01 ∗ 106 0.99998October 10 4.09 ∗ 108 1.11 ∗ 106 0.99998Particle StandardsJuly 24 2.86 ∗ 108 2.11 ∗ 106 0.9998September 7 4.31 ∗ 108 3.93 ∗ 106 0.9998October 10 4.13 ∗ 108 1.65 ∗ 106 0.99995
Table A.10: Summary of calibration information for SO42-
from the CONTACT-2012 fieldcampaign for the AIM-IC
Date slope (µS min mol−1) slope σ (µS min mol−1) r2
Gas StandardsJuly 24 8.56 ∗ 108 1.00 ∗ 107 0.9996September 7 8.17 ∗ 108 5.56 ∗ 106 0.9999New ConcentratorSeptember 10 7.90 ∗ 108 2.68 ∗ 107 0.998October 10 7.86 ∗ 108 1.07 ∗ 107 0.9994Particle StandardsJuly 24 6.81 ∗ 108 7.70 ∗ 106 0.9996September 7 8.00 ∗ 108 9.28 ∗ 106 0.9996October 10 7.95 ∗ 108 9.45 ∗ 106 0.9996
Appendix A. AIM-IC calibration and instrumental information 80
A.4 Limits of detection
Table A.11: Summary of calibration information for SO42-
from the CONTACT-2012 fieldcampaign for the AIM-IC
Species Limit of Detection (pptv)
Gas StandardsNH3 200.HCl 12.1HONO 2.79HNO3 8.43SO2 2.68Acetic Acid 566Formic Acid 488Oxalic Acid 7.39
Species Limits of Detection (ng m−3)
Particle Standards
Na+
150.
NH4+
25.0
K+
3.57
Mg2+
17.1
Ca2+
527Cl
-8.42
NO2-
3.52NO3
-36.1
SO42-
38.1Acetate 145.Formate 27.0Oxalate 5.43
Appendix B
E-AIM correlation data
Table B.1: CONTACT-2012 gas-particle partitioning correlation data
Species Slope r2
NH3 0.969 0.975
NH4+
1.031 0.940HNO3 0.974 0.349NO3
-0.429 0.288
SO42-
1.11 0.979
120x10-9
100
80
60
40
20
Con
cent
ratio
n (m
ol m
-3)
6/9/2010 6/10/2010 6/11/2010 6/12/2010 6/13/2010 6/14/2010Date
80x10-9
60
40
20
0Con
cent
ratio
n (m
ol m
-3) Input Accumulation Mode
Modelled Sea Salt Mode Input Sea Salt Mode Input Accumulation Mode
Figure B.1: Optimized separation comparison between modelled and input NO3(aq)-
81
Appendix B. E-AIM correlation data 82
160x10-9
120
80
40
Con
cent
ratio
n (m
ol m
-3)
6/9/2010 6/10/2010 6/11/2010 6/12/2010 6/13/2010 6/14/2010Date
100x10-9
80
60
40
20
0 Con
cent
ratio
n (m
ol m
-3)
Input Accumulation Mode Input Sea Salt Mode Modelled Accumulation Mode Modelled Sea Salt
Figure B.2: Optimized separation comparison between modelled and input NH4(aq)+
20x10-9
15
10
5
0Con
cent
ratio
n (m
ol m
-3)
6/9/2010 6/10/2010 6/11/2010 6/12/2010 6/13/2010 6/14/2010Date
50x10-9
40
30
20
10
0Con
cent
ratio
n (m
ol m
-3)
Input Accumulation Mode Input Sea Salt Mode Modelled Accumulation Mode Modelled Sea Salt
Figure B.3: Optimized separation comparison between modelled and input Cl(aq)-