PUBLICATIONS

Journal of Geophysical Research: Atmospheres

RESEARCH ARTICLE Sulfuric acid nucleation: An experimental study
10.1002/2014JD022730
of the effect of seven bases
Key Points: W. A. Glasoe1, K. Volz1, B. Panta1, N. Freshour1, R. Bachman1, D. R. Hanson1, P. H. McMurry2, and C. Jen2
• Sulfuric acid nucleation with bases
strongly depend on the type of base 1
Chemistry Department, Augsburg College, Minneapolis, Minnesota, USA, 2Department of Mechanical Engineering,
• When both an amine and ammonia
are present, nucleation is enhanced University of Minnesota, Minneapolis, Minnesota, USA
• Only weak effects were observed
with amides
Abstract Nucleation of particles with sulfuric acid, water, and nitrogeneous bases was studied in a flow
reactor. Sulfuric acid and water levels were set by flows over sulfuric acid and water reservoirs, respectively,
Supporting Information:
• Readme and the base concentrations were determined from measured permeation rates and flow dilution ratios.
• Sections S1–S4, Tables S1 and S2, Particle number distributions were measured with a nano-differential-mobility-analyzer system. Results indicate
and Figures S1–S4
that the nucleation capability of NH3, methylamine, dimethylamine, and trimethylamine with sulfuric acid
increases from NH3 as the weakest, methylamine next, and dimethylamine and trimethylamine the strongest.
Correspondence to:
D. R. Hanson, Three other bases were studied, and experiments with triethylamine showed that it is less effective than
[email protected] methylamine, and experiments with urea and acetamide showed that their capabilities are much lower than
the amines with acetamide having basically no effect. When both NH3 and an amine were present, nucleation
Citation: was more strongly enhanced than with just the amine present. Comparisons of nucleation rates to predictions
Glasoe, W. A., K. Volz, B. Panta, N. Freshour, and previous experimental work are discussed, and the sulfuric acid-base nucleation rates measured here are
R. Bachman, D. R. Hanson, P. H. McMurry,
and C. Jen (2015), Sulfuric acid nucleation:
extrapolated to atmospheric conditions. The measurements suggest that atmospheric nucleation rates are
An experimental study of the effect of significantly affected by synergistic interactions between ammonia and amines.
seven bases, J. Geophys. Res. Atmos., 120,
1933–1950, doi:10.1002/2014JD022730.

Received 18 OCT 2014 1. Introduction

Accepted 31 JAN 2015
Accepted article online 6 FEB 2015
Sulfuric acid (SA) is a primary driver of atmospheric particle nucleation [McMurry et al., 2005; Kuang et al.,
Published online 11 MAR 2015 2008; Sipila et al., 2010], and an important mechanism involves acid-base reactions of SA with ammonia and
amines [Almeida et al., 2013; Coffman and Hegg, 1995; Kurten et al., 2008; Weber et al., 1998]. Nucleation is
important because it can enhance concentrations of cloud condensation nuclei that can significantly affect
climate [Intergovernmental Panel on Climate Change, 2013]. Small particles in the atmosphere have also been
shown to have an effect on human health [Nel, 2005]. However, the quantitative impacts of SA nanoparticle
formation on health and climate are uncertain.
Binary nucleation between SA and H2O has been well studied and serves as a useful point of comparison
[Kirkby et al., 2011; Zollner et al., 2012], yet it cannot explain observations [Kuang et al., 2008; Weber et al., 1996]
in the atmospheric boundary layer. In this region of the atmosphere, other species must participate to
stabilize SA particles, such as biogenic amines, invoked to explain nanoparticle measurements [Weber et al.,
1998], and ammonia, shown in a theoretical study to promote nucleation [Coffman and Hegg, 1995].
Several experimental studies have corroborated that ammonia and alkyl amines strongly influence
nucleation of sulfuric acid particles [Almeida et al., 2013; Ball et al., 1999; Benson et al., 2011; Berndt et al., 2010,
2014; Kirkby et al., 2011; Zollner et al., 2012]. There is also laboratory evidence for a synergistic effect for
ammonia and an amine [Yu et al., 2012]. Quantum chemical calculations indicate that nucleation with amines
is energetically favorable over nucleation with ammonia [Kurten et al., 2008; Nadykto and Yu, 2011; Ortega
et al., 2012]. Furthermore, recent field studies with gas-phase amine detection have shown a correlation
between new particle formation and the presence of amines [Jen et al., 2014; Chen et al., 2012; Freshour et al.,
2014; Hanson et al., 2011].
Yet laboratory experiments with an emphasis on systematic variations of SA and base abundances are
needed to understand mechanisms, to provide predictive schemes, and to estimate cluster thermochemical
information. Also, to discern the atmospheric roles played by other nucleation mechanisms, such as the
binary system [Clarke et al., 1998; Kazil et al., 2007], ions [Lovejoy et al., 2004; Yu and Turco, 2000], and oxidized
organic compounds [Metzger et al., 2010; Riccobono et al., 2014; Schobesberger et al., 2013], acid-base
nucleation rates need to be quantified as well as possible.

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1933
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

The nucleation flow reactor (NFR) described by Zollner et al.

[2012] was used here to explore the effects of ammonia and
the small alkyl amines (methylamine, dimethylamine, and
trimethylamine) on nucleation with SA. A double or triple
dynamic dilution system was used to control the amount of
amine entering the system from permeation tube sources
[Freshour et al., 2014]. NH3 was introduced in amounts
that resulted in levels of ~10 parts per trillion by volume
(pptv) to several hundred pptv, and the amines were
studied for abundances of ~1 to many tens of pptv. The
permeation tubes, their quantification, and the dynamic
dilution systems are described in Freshour et al. [2014].
Particles were detected using a butanol ultrafine
condensation particle counter (UCPC) [Stolzenburg and
McMurry, 1991] as well as a state of the art nanoparticle
sizing and detection system [Jiang et al., 2011] using a
diethylene glycol UCPC [Iida et al., 2009].
Particle concentrations were measured as a function of SA
and base abundances. Power dependencies can reveal the
molecular content of the cluster that is a nascent stable
particle. This “crucial” cluster can be the critical cluster
[Oxtoby, 1994] when classical nucleation theory applies or
the “bottleneck” cluster(s) for cluster-based nucleation
theories [Chen et al., 2012; Jen et al., 2014; McGrath et al.,
2012]. There are limitations to deriving crucial cluster
content from power dependencies [Vehkamaki et al., 2012]
(also see Zollner et al. [2012] for deriving the H2O content
of the crucial cluster in the H2SO4-H2O binary system).
Nonetheless, comparison of power dependencies between
experiments has value for determining how results may be
affected by experimental conditions. For example, if the
loss of small clusters is significant before they grow to a
detectable size, power dependencies will be affected.
Observed new particle formation (NPF) rates can
significantly differ from nucleation rates [McMurry and
Friedlander, 1979; McMurry, 1983; Weber et al., 1997; Kazil
and Lovejoy, 2007]. Arguments are presented that the NPF
Figure 1. Schematic diagram of the NFR experimental
rates measured here closely reflect nucleation rates, i.e., the
apparatus. net production rates of the smallest stable clusters.

2. Experimental Procedure
The nucleation experiments were performed using the Zollner et al. experimental apparatus with the
addition of a state-of-the-art nanoparticle detection and sizing system, based on diethylene glycol (DEG)
[Iida et al., 2009] condensation onto size-selected nanoparticles (the DEG-mobility particle sizer (MPS) system).
Important details of the experiment and the DEG-MPS are described in sections 2.1 and 2.2 below.
2.1. NFR and Reactants
A schematic of the experiment is depicted in Figure 1. A 100 cm length of an ~120 cm long, 5 cm ID, glass flow
reactor is maintained at 296 K with a circulating fluid. Atop the flow reactor sits another ~25 cm, 5 cm ID, flow
reactor, the top 20 cm of which is kept at 308(±1) K, where sulfuric acid vapor and water vapor mix together in
~1 atm of N2 (liquid N2, Minneapolis Oxygen). Amines were added to the flow through an ~6 mm ID port located
in the transition region between the 308 K mixing region and the 296 K particle formation region. The 1/8″ OD
exit tube of the dynamic dilution system [Freshour et al., 2014] extends through the port so that its end is flush

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1934
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

with the wall of the flow reactor and ~60 cm3 min
1 STP (273 K and 1 atm) of nitrogen carries amine (typically
~1 nmol/mol) into the main flow (6 standard liters per minute (sLpm)). About 20 cm below the mixing region and
~10 cm into the 296 K region is the top of the nucleation zone (NZ) where the majority of particle formation
occurs, as determined from measurements in the binary system [Zollner et al., 2012; Panta et al., 2012].
SA concentrations were set by the N2 flow rate (QB4) over a sulfuric acid reservoir held at 35°C (30°C in Zollner
et al.), and relative humidity was set by the N2 flow rate over a water reservoir at 23°C. Relative humidity was
28% for the experiments. The N2 flow rate through the SA reservoir was varied over the range of 0.1 to
1 sLpm, resulting in SA concentrations of approximately 109 to 1010 cm
3 in the nucleation region. Zollner
et al. showed that an excellent linearity exists between the flow rate through the reservoir (QA in Zollner et al.)
and SA measured at the bottom of the reactor. An additional N2 flow (QB2) of 3.4 to 4.3 sLpm was needed
along with the 1.7 sLpm humidified flow (QB3) to get a total flow rate of 6 sLpm. This gas (QB2) was preheated
to 37–40°C so that the gas in the mixing region approximates the wall temperature of 35°C. As shown in
Panta et al. [2012], the gas in the NZ is warmer than the 296 K wall temperature, ~300 K.
Ammonia and amine mixing ratios are reported assuming no loss for the base: the no loss mixing (NLM) ratio
values [Freshour et al., 2014]. They are calculated from the permeation rate of the base, the flows in the
dilution system, and the final dilution into the 6 sLpm main flow. Although base concentrations are relatively
high in the sidearm flow of ~60 cm3 min
1 STP, they are quickly diluted inside the flow reactor as this plume
of gas disperses. Base is lower than NLM, well downstream of the addition point and near the wall because it is
lost on the SA-coated walls. The NLM-quoted values are useful for discussing and comparing results even
though particle formation occurs over a range of base and SA concentrations.
2.2. Particle Detection: DEG System and UCPC
Nanoparticles of 1.5 to 4 nm mobility diameter (henceforth, all diameters refer to mobility diameters, which
are about 0.3 nm larger than particle mass diameter [Larriba et al., 2011]) are formed in the experiment, and
their concentrations and sizes were obtained with the DEG-MPS [Jiang et al., 2011]. This instrument sampled
the middle of the flow near the bottom of the NFR (Figure 1) about 120 cm below the base-addition port.
Particles are sized by sending them through a bipolar charger and then a nano-differential-mobility analyzer
(TSI 3085, nano-MPS). The particle charger (2.5 cm ID × 10 cm long) contains a radioactive source of 210Po
(strength of 80 to 12 μCi over the course of the experiments) that ionizes a fraction of the particles. For a few
experiments, either a fresh (~450 μCi) 210Po source or a 6 μCi 241Am source were used, and there was no
detectable difference in results, consistent with the results of Jiang et al. [2014]. The nano-MPS had a sheath
flow of 13.2 Lpm and an aerosol flow of 2.2 Lpm, and the voltage on the inner rod was stepped every ~15 s,
and particle counts were recorded for 10 s, allowing for 5 s of settling time (positive voltages, e.g., 3, 4.5, 8, and
12 V—negatively charged particles were mobility selected). Note that the transport flow to the nano-MPS
[Jiang et al., 2011] was not used here. The monodispersed aerosol flow entered the DEG UCPC with the
saturator and condenser temperatures typically at 56 and 20°C, respectively. The large supersaturation of
DEG vapor in the condenser activates particles down to ~1 nm diameter, they grow to ~20 nm, and then they
can be counted with a model 3760 TSI CPC.
Particle count rates as a function of nano-MPS voltage (i.e., size and Dp), Nm(Dp), were corrected for charging
efficiency, DEG activation efficiency, and particle losses through all the sampling tubes and the nano-MPS [Jiang
et al., 2011]. These corrected measurements, Nm,corr(Dp), were fit to lognormal distributions, where values for the
peak number density, Ap, mode diameter, Dpeak, and width (lnσ) were assigned. The majority of the particle
distributions were well described by a single lognormal distribution with lnσ = 0.27 ± 0.03, although at times,
a second distribution at larger particle sizes was evident. This second distribution contained less particles: 10%
or less of the main distribution. Number size distributions as dN/dlnDp, are directly related to the Nm,corr plots by
the factor of a*(Qc/Qa), where a* is
dln(Zp)/dln(Dp) (Zp is the mobility, a* = 2.0 for the small particles here),
and Qc and Qa are the nano-MPS sheath (13.2 Lpm) and aerosol (2.2 Lpm) flows, respectively. Integrating the
lognormal distribution yields Np, the total number concentration of particles
Np ¼ Ap a*ðQc =Qa Þ lnσ ð2π Þ0:5 : (1)

With the fitted values for lnσ and Ap, Np was calculated from equation (1).
While particles smaller than 2 nm were detected, uncertainties in their number density are high. For example,
the data below 1.5 nm mobility can be influenced by clusters containing only a few SA molecules that may be

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1935
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

generated in the charger due to the presence of

sulfuric acid and amine vapors. Therefore, to
be conservative, data are only presented for
particles larger than ~1.8 nm. For the amounts of SA
used here, most of the newly nucleated particles
formed in these experiments attained this size.
Additional details on particle sampling and data
analysis, including the contribution of ~1.8 nm
particle measurements on Np, can be found in
section S2.1 in the supporting information.
The n-butanol UCPC [Stolzenburg and McMurry,
1991] used by Zollner et al. [2012] was attached to
the flow reactor just above the DEG-MPS sampling
port for some experiments. It was primarily used to
monitor particle formation in the binary system to
Figure 2. Size distributions from experiments with 11 and
assess the status of the NFR for nucleation baseline
27 pptv NH3 with QB4 = 0.4 sLpm. Two lognormal functions
are shown for each distribution, one for the main mode of conditions (NBC). At a flow rate of 1 sLpm through
particles and the second for the larger mode particles (about the SA reservoir and 27% relative humidity (RH),
twice the mode diameter). The number of particles in the UCPC particle concentrations were periodically
larger size distribution is about 10% of that in the smaller checked to ensure that the system was functioning
one. The mode diameters and lnσ are 3.4 nm and 0.25 and normally: a certain particle formation rate is
3.8 nm and 0.27, for the 11 and 27 pptv NH3 experiments,
respectively.
expected for these conditions. A reproducible NBC
result shows that (i) the system has a constant
cleanliness level over time and (ii) the SA level,
flows, temperatures, and other experimental conditions are in their normal state. The UCPC data had little
value during the amine addition experiments as Np typically exceeded the maximum concentration limit for
the UCPC. However, a few experiments with low amounts of NH3 added (10–15 pptv) gave Np less than
105 cm
3, and those UCPC results were compared to Np from the DEG-MPS system.

3. Results
3.1. Particle Detector Comparison
Data from the DEG-MPS system are depicted in Figure 2, a plot of Nm,corr(Dp)a*Qc/Qa, which is equal to
dN/dlnDp, versus Dp. Results are shown from two experiments with ammonia at NLM = 11 and 27 pptv
and a fixed amount of SA (QB4 = 0.4 sLpm, gives [SA] 3–4 × 109 cm
3 in the nucleation region). Lognormal
distributions are also shown in Figure 2, and the integrated distributions give Np = 4.0 × 104 and 7.0 × 103 cm
3,
for 27 and 11 pptv NH3, respectively. The butanol UCPC was also used in these experiments, and measured
particle concentrations were 2.7 × 104 cm
3 (live time corrected and a live time, 44%) and 5.7 × 103 cm
3
(live time, 86%.) The two particle counters are in reasonable agreement (UCPC low by 19 and 32%) for these
conditions with the better agreement at the lower Np, where the live time fraction is high (less particle overlap)
and the UCPC functions best. The agreement between the two counters verifies the analysis of the DEG
system data for these conditions.
In the results depicted in Figure 2, a secondary distribution of particles at larger sizes arises that has a total
particle number of about 10% or less of the primary distribution. This may represent particles that had
taken a path down the flow reactor that had a greater exposure to SA than the particles formed in the
main parts of the flow. The amount of particles in this secondary distribution was variable but always less
than 10% of Np.
The UCPC was also used for periodic NBC particle number measurements. These data (plotted in the supporting
information) show that the stability of the system is quite good over a long time period. It also shows distinct
perturbations to the system due to severe events such as a contamination episode [Zollner et al., 2012] and
a recent flow reactor rinsing. A significant change to the system that is reflected in the NBC data is a change in
the temperature of the SA reservoir from 30 to 35°C in late summer 2012. This resulted in a factor of 3 to 6
increase in the expected Np for NBC, probably due to an increase in sulfuric acid for a given QB4.

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1936
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

Figure 3 shows the DEG particle measurements

versus ln(Dp/nm) for two experiments at
QB4 = 0.4 sLpm: one with 220 pptv NH3 and one
with 4 pptv dimethylamine (DMA). Lognormal
distributions are also shown, and they represent
the data well. The Np are calculated from lognormal
fits to the DEG particle distribution data, and for
these two experiments, the Np are 2.3 × 106 and
2.6 × 106 cm
3, respectively.
The lognormal fits to the DEG data yield the
peak number, Ap, mode diameter (Dpeak), and
width (lnσ). It was found that the mode diameter
and QB4 were well correlated. This is consistent
with particle growth being linearly dependent
on the amount of sulfuric acid present, as SA
is proportional to QB4. A plot of Dpeak versus
Figure 3. Size distributions from experiments at QB4 is shown in section S2.2 in the supporting
QB4 = 0.4 sLpm: one with 220 pptv for NH3 and one with 4 pptv
information for several ammonia, methylamine,
for dimethylamine. The lognormal functions for each set of
measurements are shown with mode diameters and lnσ of and DMA additions. A linear fit to the data
3.9 nm and 0.25, respectively, for the NH3 experiment and represents quite well the variation of particle
3.4 nm and 0.24, respectively, for the DMA experiments. sizes with SA, indicating that SA vapor is
the primary growth agent in the present
experiments. The ammonia data, as shown in Figure 3, has somewhat larger Dpeak than the DMA
data at a given QB4. This may indicate that ammonia is contributing to growth, consistent with the
relatively large amounts of ammonia used in the nucleation experiments. For single-digit pptv amine
experiments, the distributions when trimethyl amine (TMA) was added were on average slightly
larger than the DMA results, while distributions with MA added tended to be slightly smaller than
those with DMA added. These tendencies in size distributions are slight, but they are consistent with
the changes in molecular size of the amines.
3.2. Variation of SA at Constant Base
3.2.1. Ammonia
Figure 4a shows the Np, the total number of particles, as a function of QB4 and [SA] for NH3 additions of
6, 10, 55, 110, 220, and 2200 pptv. The amount of NH3 added was held steady for each set of measurements.
Individual power law fits to these data show dependencies upon SA that scatter from 3.4 to 5.4 with
an average of about 4. Similar experiments where QB4 was fixed while NH3 was varied result in power
dependencies on [NH3] of 1.5 to 2.2. Note that NH3 at a level of 50–100 pptv gives nucleation rates that are
about 1 million times that of the binary system (green dashed line); this is in rough agreement with a
preliminary measurement [Zollner et al., 2012], where a factor of about 105 over the binary system was
reported for 45 pptv NH3. As expected, the power dependency of particle numbers on SA of ~4.4 ± 1 is less
than the SA power dependency of 6 ± 1 for the binary system at 27% RH. The power dependency for the binary
system was obtained for QB4 that ranged from 0.4 to 1.5 sLpm: consideration of the expected increase in the
binary SA dependency at low QB4 (0.15 to 0.6 sLpm for the ammonia measurements) further highlights the
effect of NH3 on decreasing nucleation’s SA dependency.
With an estimate of [SA] = 3.5 × 109 cm
3 at a QB4 = 0.4 sLpm, an overall fit to the data in Figure 4a gives


Np cm
3 ¼ 710
50 ½SA4:2 ½NH3 1:6 (2)

with [SA] and [NH3] in cm
3 (NLM was used to calculate [NH3]). The overall power dependency fit represents
the data well over the range of SA and NH3 concentrations in the experiment. The power dependencies can
be related to the crucial cluster composition assuming (i) that the spatial variations in SA or NH3 concentrations
do not depend on the other component and (ii) that the loss of particles between the crucial cluster and the
minimum detectable size does not depend on reactant concentrations. With these assumptions, the crucial
cluster contains 4 or 5 SA molecules and 1 to 2 ammonia molecules.

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1937
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

9
3
Figure 4. Np plotted versus QB4 flow or [SA] for four different bases. At QB4 = 0.4 sLpm, [SA] is estimated to be 3–4 × 10 cm .
The symbols across the plots are coded for base level in pptv (approximately): diamond, 20–50; plus sign, 8–10; square, 4–6;
cross sign, 2–3; and circle, 1–2. One million times the expected value for the binary system is shown as the green dashed
box [Zollner et al., 2012] in the plots. (a) Ammonia NLM values from 6 to 2000 pptv. Experimental SA power dependencies
average about 4 and power dependency on NH3 is about 2. (b) Methylamine nucleation results. Experimental power
dependencies on SA of 4.5, 4, and 3.5 in the three bins centered around MA = 3, 6, and 17 pptv. (c) Results for DMA at 1 to

3
32 pptv. Fits to the data below Np = 107 cm show power dependencies on SA of ~2.5 and on DMA of about 2. (d) TMA results
for 1 to 14 pptv, NLM. Individual power dependencies on SA range from 1.6 up to ~3.5 as TMA decreases from 14 to 3 pptv.

Extrapolating the overall fit (equation (2)) to the conditions of other experimental results is uncertain because
of gradients in [SA] and [base] within the NZ. Panta et al. [2012] showed that SA is relatively constant across the
NZ, varying about 20% in this region; however, its absolute value is estimated to be uncertain to +100/
50%
(see section S1.2 in the supporting information). On the other hand, the base concentration in the experiment has
large gradients (see supporting information in Freshour et al. [2014] and in Panta et al. [2012]). The [base] is taken
to be NLM in equation (3) even though the base in the NZ can differ significantly from NLM (see next paragraph).
Nucleation rates JNH3 are calculated from equation (2), assuming an average nucleation time, tnuc : J = Np/tnuc.
The value for tnuc is estimated to be 7 s (+100/
50%), and JNH3 for our experimental conditions is


J NH3 cm
3 s
1 ¼ 10
50 ½SA4:2 ½NH3 1:6 (3)

with [SA] and [NH3] in cm
3. This expression can be used for rough comparisons between experiments
with the understanding that the expression is based on acid and base concentrations that are uncertain.
Recent computational fluid dynamic simulations (D. R. Hanson and B. Panta, A Computational Fluid Dynamics
Study of a Flow Reactor: Formation of Sulfuric Acid Clusters with NH3 and Dimethylamine, manuscript in
preparation, 2014) based on the model presented in Panta et al. [2012] present details of the nucleation time

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1938
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

and reactant distribution and show that equation (3) yields reasonable results over a wide range of conditions,
including SA down to 107 cm
3.
3.2.2. Methylamine
Figure 4b shows the change in Np as [SA] is changed, while the amount of methylamine (MA) is held steady;
the data are grouped into NLM bins. MA is much more effective at nucleation than NH3 for a given amount of
base present: using 106 times the binary system as a reference, single-digit pptv MA levels yield rates that are
comparable to what NH3 levels of 50–100 pptv generate. There is quite a bit more scatter in the particle
measurements with MA present than with NH3 (Figure 4a). This can be due in part to increased uncertainties
in NLM values at the large dilutions needed for the amine experiments (i.e., fmol/s is required to achieve pptv
NLM). Also, in a study describing the amine permeation tubes [Freshour et al., 2014], MA was observed to
have a large interaction with surfaces, including perfluorocarbon tubing. Surface interactions could cause
variability in the amount of MA delivered to the flow reactor. This is discussed further in section S3.1.1 in the
supporting information.
The power dependencies of Np on SA and on MA are about 4 and 2, respectively, for 2 to 21 pptv MA and
the range of SA shown in Figure 4b: these power dependencies are comparable to what was found for NH3.
This suggests a similar crucial cluster size for ammonia and methylamine. Less MA than NH3 is needed for
comparable nucleation rates which probably indicates that the evaporation rate of MA (=b) from the SA4b2,
SA3b2, SA3b1, etc. clusters is lower than the evaporation of NH3 when b = NH3. An increased stabilization of
the dimer of SA for methylamine over that for NH3 was demonstrated in measurements presented in Jen et al.
[2014]. Note that the results at 40 pptv MA show a decreased dependence of particle number on SA which is
due at least in part to coagulation at these high particle number densities (see section 4.3 below).
3.2.3. Dimethylamine
Figure 4c is a plot of Np versus QB4 and [SA] for several DMA addition experiments. Again, DMA promotes
nucleation very well with single-digit pptv levels of DMA giving more than a millionfold increase in particle
numbers with respect to the neat binary system. The scatter in the data is large and is in part due to the
difficulty in delivering small amounts of amine to the flow reactor, as discussed above for MA. Other potential
causes for the large scatter are the uncertainty in Np from the derived lognormal fits and day-to-day
variations in the DMA permeation rate. For 4 pptv and lower DMA, the power dependency of Np upon SA is
around 2.5, and the power dependency on [DMA] is about 2 (see section 3.3). The overall power relationship
between Np and concentrations is


Np cm
3 ¼ C exp ½SA2:5 ½DMA2 ; (4)

where Cexp = 2 × 10
33 cm10.5 and [SA] and [DMA] in cm
3. This is an approximate relationship but useful for
discussing the results. This relationship applies for Np that ranges between 105 and 107cm
3. Nucleation rates can
be calculated from equation (4) in the same way that equation (3) was derived from equation (2):


J DMA cm
3 s
1 ¼ 3  10
34 ½SA2:5 ½DMA2 : (5)

The power dependency of Np on SA of 2.5 is much less than those found for the ammonia, methylamine, and
the binary system, suggesting a small crucial cluster of about 2–3 SA molecules and 2 DMA molecules. The
dependence on DMA and on SA appears to decrease at DMA NLM levels of 8 pptv and above. Depletion of SA
and base monomers due to uptake onto particles and coagulation could play significant roles when Np
reaches 1 × 107 cm
3 and larger; these are discussed below.
The precision of JDMA can be taken to be the scatter for a particular NLM data set in Figure 4c, which is about
100%/
50%. Added quadratically to the factor of 2 uncertainty in the nucleation time, the uncertainty of
equation (5) is approximately a factor of 3. Uncertainties in experimental parameters and extrapolated
nucleation rates are discussed further in section S1.2 in the supporting information and section 4.4 below. It is
concluded that the nucleation rates given by equations (3) and (5) are dominated by the uncertainty in SA
due to the power dependency of J on SA. Equations (3) and (5) have overall accuracies of about a factor of 16
and 6, respectively, dominated by the factor of 2 uncertainty in [SA] (24 and 22.5, respectively).
3.2.4. Trimethylamine
Figure 4d is a plot of Np versus QB4 and [SA] for several TMA addition experiments. The effect of adding TMA in
the single-digit pptv range leads to an enhancement over the binary case of a factor of over a million.

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1939
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

Comparing Figures 4c and 4d, it appears that

TMA is not quite as efficient as DMA in
promoting particle formation especially at low
QB4. Also, a lower level of Np is reached at
saturation levels of SA and TMA, QB4~0.4 sLpm,
and TMA = ~10 pptv. The scatter in the data is
somewhat less than that for DMA.
The power dependency of Np upon SA is
comparable to that for DMA: at 3 and 5 pptv, the
power dependency is ~3 on SA, while at 9 pptv, it
decreases to about 2. For experimental conditions
with Np values <5 × 106 cm
3, Np has an overall
power dependency of 2.8 for SA and 2 on TMA.
3.2.5. Triethylamine, Urea, and Acetamide
A few experiments were performed with
triethylamine (TEA), from a permeation tube,
and urea, using its vapor pressure directly,
and acetamide using its vapor pressure and
a diffusion tube. The amounts of urea and
Figure 5. Variation of Np versus mixing ratio of MA (plus sign), DMA acetamide were quantified with ambient
(asterisk), and TMA (black symbols) at a constant QB4 = 0.4 sLpm. pressure proton transfer mass spectrometer
The fit has a dependence on base to the second power at low base
mixing ratios. At this level of SA, the nucleation capability of these
(AmPMS) (Freshour et al.) or an AmPMS-type
three amines is similar. system [Jen et al., 2014]. The vapor pressure of
urea was found to be about 2 × 10
8 atm, close
to an estimate of 1.6 × 10
8 atm [Jones, 1960].
The effect of adding TEA at 36 pptv had an effect that is about equal to the effect of adding ~2 pptv of DMA
(Figure 4c), while the effect of adding 75 pptv urea gave results that are about equal to those of 10 pptv of
ammonia in Figure 4a. At QB4 = 0.4 sLpm, acetamide added at 30 pptv had no discernible effect on particle
numbers above background count rates of the DEG system. At QB4 = 1 sLpm, an effect due to acetamide was
observed: there was an increase in Np of about a factor of 50 over the binary case upon addition of 30 pptv
acetamide. In comparison to the amines, where millionfold increases were observed at pptv levels, acetamide
is a very weak particle formation partner with H2SO4. Plots of these results and more discussion are presented
in section S3.2 in the supporting information.
TEA is effective at forming particles, but it is less efficient than the other amines studied here. This could be
due to a steric effect since the gas-phase basicity of TEA is comparable to that of DMA and TMA. Urea does
increase the particle formation rate over the binary system, but it seems to be even less effective than
ammonia. Amides have large gas-phase basicities and are detected efficiently with AmPMS; however, amides
are very weak acids in solution, and this weakness, for acetamide and for urea at least in part, continues down
to the level of very small clusters.
3.3. Variation of Amine at Constant SA
A series of measurements, where amine levels were varied at a fixed QB4 = 0.4 sLpm, are shown in Figure 5, a plot of
Np versus NLM for MA, TMA, and DMA (QB4 = 0.4 sLpm is [SA] = 3.5 × 109 cm
3 in the NZ). The TMA data were taken
over a period of many days and show that reproducibility for this amine can be quite good. The midrange of the
data from Figure 4c for the DMA experiments (blue asterisk) are shown along with recent measurements. Although
the DMA data may be somewhat higher than the TMA results, within the scatter exhibited in Figures 4 and 5,
TMA and DMA act similarly for these conditions. Methylamine data (red plus sign) are also not significantly
different than the TMA results at this level of SA, considering the scatter in the data exhibited in Figure 4.
Although the difference between these amines is clearly seen in Figure 4, it seems that the action of these
amines on SA nucleation is similar when SA is sufficiently high. This insensitivity to the type of amine suggests
that the nucleation rate is dominated by the SA collision rate for these conditions. Note that the data for
10 pptv NH3 at QB4 = 0.4 sLpm is about a factor of 300 lower than the amine data, and ammonia is a weak
partner in SA nucleation for these conditions.

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1940
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

The value of Np depends on TMA2

at low TMA, but it has a decreased
dependence above 10 pptv TMA. The
behaviors of the MA and DMA data
are similar to that of TMA. The dashed
line is a fit to the TMA data according
to (c/TMA2 + d/TMA + e)
1, where c,
d, and e are the fitted parameters.
These parameters are not ascribed a
physical meaning as coagulation of
particles and decreases in SA monomer
are both likely to be important at high
Np. For example, at an Np of 107 cm
3
and a volume-averaged diameter
of 4 nm, each particle has ~200 SA
molecules; the total number of SA
molecules in particles is 2 × 109 cm
3,
which is comparable to the initial [SA]
in the NZ. Coagulation loss rates are
estimated to be ~0.1 s
1 for 4 nm
diameter particles at these number
densities (see section 4.3), which is
significant when considering their ~20 s
Figure 6. Variation of particle number with NLM amine at a constant SA residence time in the flow reactor.
(QB4 = 0.2 sLpm) and the synergistic effect when NH3 is also present. For
7
3
Np < 1 × 10 cm and MA and DMA only, Np goes as amine to the 2.7 A similar set of experiments at a
and 2 power, respectively. With 200 pptv NH3 present, these powerQB4 = 0.2 sLpm for MA and DMA
dependencies drop to ~1. There was little dependence on NH3 for the
additions were conducted, and the
MA data (NH3 varied from 200 to ~600 pptv).
results are shown as the open symbols
in Figure 6. Plotted are the Np values as a
function of amine NLM value for either MA or DMA added. These results are similar to those depicted in
Figure 5 but were conducted at QB4 = 0.2 sLpm, which is equivalent to about 2 × 109 cm
3 SA in the NZ.
At this level of SA, MA and DMA do not act similarly to form particles. For low amine levels, the power
dependency on DMA is about 2, and the power dependency on MA is about 2.7.
Comparing Figures 5 and 6, the power dependency on DMA is similar for these two sets of QB4 (0.2 and
0.4 sLpm) even though particle volume density is a factor of ~20 larger at the higher QB4 for similar DMA
levels. The measured power dependency does not depend on the amount of particulate volume or SA over
this range. Assuming that the DMA power dependency should increase if it changes at all with decreases in
SA, the independence on SA level suggests that the depletion of DMA by uptake into particles is probably
not a dominant loss process for DMA. This is true also for the other amines by extension, since they are
less active.

3.4. Synergistic Effect

Also shown in Figure 6 are Np (filled symbols) as a function of amine NLM value with 200 pptv NH3
present. The MA results also include higher NH3 levels up to 630 pptv: surprisingly, there is no discernible
trend with NH3 level. For both MA (triangles) and DMA (squares), there is a significant effect due to the
presence of NH3: at 2 pptv MA, there is about a hundredfold increase in particle number due to 200 pptv
NH3, and at 2 pptv DMA, there is about a twentyfold increase in particle number due to 200 pptv NH3.
Note that the expected Np for these conditions for NH3 alone is ~1 × 105 cm
3 (see Figure 4a.) Also,
the particle diameters were observed to increase slightly due to the presence of 200 pptv NH3, from
~2 nm to ~2.5 nm. This may represent growth due to the high abundance of ammonia, in accord with
the results of the NH3 alone experiments: the mode diameter for 220 pptv NH3 at QB4 = 0.2 sLpm was
also about 2.5 nm.

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1941
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

The dependency of particle number on [amine] decreased due to the presence of ammonia: the power
dependencies decrease to about 1 for both DMA and MA. This probably indicates that an ammonia molecule
stabilizes some single amine-containing clusters better than an additional amine molecule would for these
conditions. The dependence of particle number on QB4 was studied for DMA over the range of 0.2 to
0.3 sLpm, and the SA power dependency (PD) was low, 1.1 and 2, for two experiments. Three experiments
varying QB4 with both MA and NH3 present resulted in PDs of Np upon SA of 1.6, 1.7, and 2.3. These are
substantial decreases from the SA power dependencies without ammonia present, which were 2.5 for DMA
alone and 4 for MA alone.

4. Discussion
4.1. Power Dependencies
For QB4 ≤ 0.3 sLpm ([SA] < 3 × 109 cm
3) and a given [base], the nucleation rates for NH3 and MA are lower
than those of DMA and TMA. Overall, particle number densities increase, and their power dependencies
on SA decrease for the series NH3, MA, TMA, and DMA, respectively. These observations can be interpreted
using the first nucleation theorem [Oxtoby, 1994]. The power dependency trend suggests that more SA is
needed in clusters that contain NH3 or MA than for clusters that contain DMA or TMA: particle formation
begins earlier and more strongly with DMA and TMA than with NH3 and MA. For example, the NH3 power
dependence of 4 upon SA indicates that 4 SA molecules are present in the crucial clusters. In comparison,
2 or perhaps 3 SA molecules are present in the crucial clusters that contain DMA. The presence of DMA
better stabilizes particles than does NH3 because it more effectively reduces SA evaporation rates.
Interestingly, the presence of NH3 and DMA together further stabilizes SA clusters; the nucleation rate
increases while the power dependence on SA decreases. This is also the case for NH3 present with MA
where increases in Np were met with decreases in PDs on MA and on SA, and on NH3, for that matter.
The power dependence of Np upon base is about 2 for these four bases over a range of base mixing ratios
(NLM): for ammonia up to the ppbv level, for MA up to about 20 pptv, and for DMA and TMA up to about
10 pptv. These results indicate that more than 1 base molecule is required to form stable particles for these
conditions. Two (or more) base molecules also appear to be needed for the synergistic effect: the power
dependence on the amine decreases to about 1 when NH3 was present (presumably 1 or more NH3 were in
the clusters). Thus, an ammonia molecule stabilizes the crucial cluster as would additional amine molecules
and perhaps better for these conditions.
Above the tens of pptv levels for the amines, Np exceeds 107 cm
3, and the particle formation rates seem
to saturate, but this could be due to coagulation (see section 4.3). Coagulation or other losses may play a
role for the synergistic effect too when Np approaches 107 cm
3. Finally, we assume in this discussion that
the nucleation theorem applies. If information becomes available on limitations of the applicability of the
nucleation theorem for our conditions, the relation of power dependencies to the composition of the crucial
cluster may have to be re-evaluated [Vehkamaki et al., 2012].
Chen et al.’s [2012] model assumes that 4 SA molecules comprise a new particle—this model was further
developed in two ways by Jen et al. [2014], retaining a nonnegligible evaporation of the 3 SA containing
cluster in one of them. These models are consistent with the power dependency of the present results:
Chen et al.’s model has a power dependency on SA of about 2.4 over the [SA] and DMA ranges of the
present results, 1–4 × 109 cm
3 and single-digit pptv, respectively. This SA power dependency agreement
is due to the growth rate of the trimer by SA addition being comparable to or larger than its empirical
evaporation rate.
The NFR is much cleaner than the Teflon film chamber used by Chen et al. [2012], enabling far superior
chemical characterization and repeatability. While the acid-base reaction model for chemical nucleation
described by Chen et al. provides a useful framework, the experimental results described here and the results
of Jen et al. [2014] are quantitatively more reliable than those of Chen et al.
Chen et al.’s framework explicitly includes only a single base molecule in the nucleation rate expressions,
although additional base molecules were not excluded from playing a role. A simple dependence on base
concentration was assumed because there was no information on the base content of clusters. Jen et al.’s
measurements and model refinements focused on the abundance of the SA dimer, which could be explained

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1942
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

by the presence of just 1 base molecule with 2 SA molecules. Therefore, the PDs here that indicate 2 base
molecules are involved in nucleation adds to the Jen et al. measurements.
The supporting information contains a table of the observed power dependencies and nucleation rate
expressions for NH3, MA, DMA, and TMA and the synergistic effect for MA and DMA with NH3.

4.2. Residual Amines in Binary System

There have been recent discussions [Almeida et al., 2013; Kirkby et al., 2011; Berndt et al., 2010, 2014]
concerning the effect of potential contaminants on nucleation results for the neat binary system; one study
[Olenius et al., 2014] has postulated that fractions to single-digit pptv levels are common contaminant levels
of amines. Our results suggest very large effects if contaminants are at this level: nucleation results will
deviate from the binary system by as much as a factor of 106 (see Figure 4c) if 1 pptv amine is present and
perhaps off by as much as a factor of 104 (see Figure 4a) if 1 pptv ammonia is present. We conclude that
amine contaminants in our flow reactor studies are far lower than the ~0.1–1 pptv level.
Zollner et al. [2012] discussed the possibility that very small amounts of contaminant amines could have
influenced the binary system measurements. Using bulk thermodynamic calculations [Wexler and Clegg, 2002;
Ge et al., 2011], they concluded that levels of 1 to 10 fmol/mol (i.e., 10
14 mole fraction) could have an influence
on the H2SO4-H2O system results. The results here substantiate that: if the PDs for NH3 and DMA can be
extrapolated to the ppqv level (fmol/mol), then about 10 ppqv NH3 or 1 ppqv DMA would result in particle
numbers consistent with the Np observed for the putative neat binary system reported here (supporting
information) and in Zollner et al. [2012]. The extrapolation of this empirical finding suggests therefore
that contaminant levels of DMA in the flow reactor are 1 ppqv or less. See the supporting information for
further discussion of these points.

4.3. Cluster Loss Rates: Growth, Wall Loss, and Coagulation

The power dependencies suggest that very small clusters of 2 to 4 SA and 1–2 base molecules are the crucial
clusters. These are smaller than the smallest-sized particles that were monitored with the particle detector,
about ~1.8 nm mobility diameter, which corresponds to ~10 SA molecules assuming a bulk composition
and density for 27% RH (about 3.5 H2O to 1 SA). The growth of the crucial clusters to the minimum detected
size may be interrupted by their loss to the flow reactor wall or via scavenging. If this loss is significant,
the experimental Np divided by the nucleation time may not be simply related to the production rate of
the smallest stable cluster, taken to be the true nucleation (TN) rate. Fundamentally, this is the reason for
distinguishing observed new particle formation rates, NPF, and nucleation rates [McMurry, 1983]. If losses are
significant for clusters between the crucial cluster and the minimum detected size, NPF is less than TN. TN is
equal to the formation rate of the smallest stable cluster, perhaps the crucial cluster or a cluster just a
molecule or two larger.
One way to evaluate the factors that can contribute to the difference between NPF and TN is to compare
the growth and loss rates of the individual clusters. For the present work, loss to the reactor wall can be
estimated from the first-order loss rate constant for diffusion in laminar flow, 3.6 Dc/r2, which results in rates of
0.05, 0.026, and 0.016 s
1 for the SA monomer, trimer, and hexamer, respectively, using the diffusion coefficients
Panta et al. [2012]. Growth rate constants of the SA trimer and hexamer due to addition of an SA molecule at
QB4 = 0.2 sLpm ([SA] = 2 × 109 cm
3) are 1.1 and 1.3 s
1, respectively. Loss of trimer and hexamer clusters to the
wall is 1 to 2% of that due to growth, and thus Np at 2 nm mobility diameter (about 15 SA molecules) closely
reflects (within 20%) the nucleation rate of clusters with 3 to 4 SA molecules. A greater than 80% survival
probability is estimated based on 10 growth steps needed to get to the 2 nm mobility diameter.
Loss of clusters via scavenging by stable particles can be significant under certain conditions. For example, at
a QB4 of 0.4 sLpm, the value of Np can be up to 3 × 107 cm
3 and with a surface-area-weighted mode diameter
of ~4 nm, the surface area per unit volume is 1.5 × 10
5 cm
1. For these conditions, the pseudo-first-order
loss rate coefficient for the SA trimer (multiplying by ω/4 = 2700 cm/s for a3b1(H2O)10) is 0.04 s
1. Application
of a Hamaker constant [Hamaker, 1937] of ~3 for this interaction gives a trimer loss rate of ~0.12 s
1. This
amounts to a significant loss over a given length of the flow reactor, whether the ~30 cm length of the NZ or
the ~110 cm distance to the particle detector. Alternatively, applying the steady state model of Weber et al.
[1997, equation (8)], with these conditions, the survival probability that 1 nm particles will make it to 3 nm

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1943
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

a
Table 1. Comparison Between Experimental Nucleation Rate and Nucleation Rates From Previous Experiments for NH3

3
1
3
1
Reference Note Jref (cm s ) T (K) NH3 (pptv) SA (pptv) Jequation (3) (cm s ) Jequation (3)/Jref
b
Kirkby et al. [2011] Jgcr, chamber 1 292 100 6 0.24 0.24
c
6 292 190 8.8 3.3 0.55
d 4 4
Berndt et al. [2010] FR 2 × 10 293 4000 30 7.5 × 10 4
e 4 4
Benson et al. [2011] FR 500 288 2 × 10 12 2.1 × 10 42
Zollner et al. [2012] FR 125 298 20 120 5300 42
FR 5 298 2 120 130 27
4
FR 875 298 45 120 1.9 × 10 22
Ball et al. [1999] FR 15 298 2 120 130 9
10
12
Coffman and Hegg [1995] LD model 1 × 10 298 5 10 0.017 1.7 × 10
6
4
11
Napari et al. [2002] LD model 1.7 × 10 298 100 1 1.3 × 10 7.5 ×

5 4 9
Merikanto et al. [2009] LD model 1 × 10 294 1000 40 3 × 10 3 × 10
f
McGrath et al. [2012] ACDC, 5 298 8 120 1200 240
g
6
7
ACDC, 1 × 10 298 40 0.4 6.3 × 10 0.6
a
FR is flow reactor; LD is liquid drop model (classical nucleation).
b
Data from their Figure 4a.
c
Data from their Figure 5.
d 5
3
3 1 4
Data from their Figure 12: 30 L/min is a residence time of 32 s, Np = 6 × 10 cm , thus, J (cm s ) = 2 × 10 .
e
Data from their Figure 1a.
f
McGrath et al.’s [2012] cluster-based model (their Figure 1 extrapolated to present experimental conditions).
g
Extrapolation of present data to the midrange conditions of the ACDC model calculations.

diameter is about 60%. Scavenging of small clusters can play a significant role in flattening the power
dependencies at high Np especially for the larger-sized particles formed at QB4 = 0.4 sLpm.
For experimental conditions that result in Np ≤ 107 cm
3, the scavenging loss drops to 0.04 s
1 or less, which is
≤2% of the SA growth rate at QB4 = 0.4 sLpm. Scavenging of clusters by the 2 nm diameter particles formed at
QB4 = 0.2 sLpm will be 0.01 s
1 at Np = 107 cm
3. Therefore, at low Np or low QB4, scavenging losses are negligible.
Coagulation of the detectable particles as they travel down the flow reactor can also be important.
A coagulation loss term can be represented by dN/dt =
kcoagN2, where kcoag is the coagulation loss
rate coefficient which is about 6 × 10
9 cm3 s
1 for 2 to 4 nm sized particles ([Mcmurry, 1980] using a
Hamaker constant of 3 [Hamaker, 1937]). With a residence time of 10 s, coagulation can be a significant
limitation to increases in Np beyond 107 cm
3. Because of scavenging and coagulation, equations (2)–(5) and
the comparison of results are limited to results when Np < 1 × 107 cm
3.
Finally, using a bulk source for SA vapor and its loss to the wall means that a particle has limited exposure to SA
and an overall growth limit arises. Therefore, for most of the particles to reach a detectable size, there is a
minimum value to which QB4 should be set. The mode diameter of the distributions ranged from 1.8 to 2.5 nm
for QB4 = 0.2 sLpm, and this is a practical lower limit for QB4 for most conditions. Data below QB4 = 0.2 sLpm are
reported in Figure 4 at the highest NH3, DMA, and TMA levels because the size distributions had a significant
number of particles of 2 nm diameter and larger. A few DMA experiments at 4 pptv at low QB4 are also
included in Figure 4c even though most of the particles were smaller than 2 nm. These data lie within the
general trend of the higher QB4 data and do not depart significantly from equations (4) and (5).

4.4. Comparison to Previous Work

Nucleation rates have been reported for sulfuric acid with added amines or ammonia over a wide range of
conditions and for a variety of experimental techniques. Because of the differences in experimental conditions
and measurement techniques, 1:1 comparisons are typically not possible. On the whole, the studies all show
that the presence of amines with sulfuric acid greatly increases the appearance rate of particles. Table 1 presents
a comparison of previous experimental and model results for NH3 (equation (3)), and Table 2 presents the
comparisons for DMA (equation (5)). Before discussing these comparisons, we present an uncertainty analysis
of the present results.
4.4.1. Uncertainties
It was stated above that the uncertainty in equation (3) is about a factor of 16. The uncertainty in SA is
dominant for equation (5) as well so the uncertainty in equation (5) is primarily due to the factor of 2

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1944
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

Table 2. DMA and TMA Comparison Between Present Nucleation Rates and Rates From Previous Studies

3
1
3
1 a
Reference Note J (cm s ) Amine (pptv) SA (pptv) Jequation (5) (cm s ) Jequation (5)/Jmeas
b
Almeida et al. [2013] DMA 300 10 0.5 10 0.04
c 4 7
Yu et al. [2012] DMA 1 × 10 4100 1.6 3.2 × 10 3200
d 5
Erupe et al. [2011] TMA 3000 480 1.5 4 × 10 125
e 5
Berndt et al. [2014] DMA 1250 40 8 2 × 10 160
5 5
This work TMA 1.3 × 10 5 60 4 × 10 3
f 6 5
ACDC model >1 × 10 4 40 1 × 10 <0.1
ACDC model 10 0.4 0.4 0.01 0.001
a
3
1 2.5 2
Cast into pptv at 298 K and 1 atm: equation (5) is J/(cm s ) = 0.7 * SApptv DMApptv .
b
They predict little temperature dependence for J over the range of 255 to 300 K; 278 K and with 10 pptv NH3.
c
From their Figure S3 in the supporting information, 288 K and 28 % RH.
d
Their Figure 2a, 288 K and 31% RH.
e
Their Figure 6, 25% RH, 32 s residence time, 293 K.
f
McGrath et al. [2012]. Note that the stability of an important cluster in their model was revised, and model rates are likely to decrease.

uncertainty in SA: equation (5) is accurate to about a factor of 6 (×2±2.5). For equation (6) (see below),
however, the power dependency on SA is diminished somewhat, and the uncertainty was determined by
propagating all the factors in quadrature: a factor of 21.5–2 due to uncertainty in SA, the factor of ~2 scatter in the
measured Np, and the factor of 2 uncertainty in the nucleation time. The overall uncertainty in equation (4)
is then on the order of a factor of 5 (× 5±1).
It is not clear how to estimate additional uncertainties induced by extrapolating these equations
outside the range of the present experimental conditions. The statement above that equation (3) gives
reasonable results over a wide range of conditions also applies to equation (5) for the DMA results
where equation (5) was compared to a box model run at atmospheric conditions (D. R. Hanson and B.
Panta, manuscript in preparation, 2014). The synergistic effect has not yet been modeled; thus, the
same statement cannot be made about the uncertainty of extrapolating those results to lower SA. A
power dependency on SA of 2 is assumed in this extrapolation, which is within the range of those
reported in section 3.4.
4.4.2. The NH3/H2SO4 System
For nominally equal experimental settings, the present ammonia data are about a factor of 25 to 50 larger than
those of Zollner et al. [2012]. The discrepancy in the MA results is greater, a factor of 100 to 1000 larger. Three
experimental considerations could contribute significantly to this discrepancy: (i) the DEG-MPS system was
not used in the previous study, (ii) changes in the temperatures of the mixing region and H2SO4 saturator, and
(iii) the base dilution system apparatus and protocols have undergone development since our previous study.
It is possible that the conditions of the previous study led to decreased particle formation rates because (ii) less
SA might be present at a given QB4 and (iii) the previous dilution system’s reliability is not known, and thus,
ammonia or MA could have been less in the previous study. The ammonia data of Ball et al. [1999] lie about an
order of magnitude below the present results, which may be due to (i) and also to large uncertainties in
ammonia levels: NH3 was introduced at the top of the mixing region, and its estimated loss from there to the NZ
is highly uncertain.
Previous experimental work on nucleation in the NH3-SA system is largely at SA levels less than the present
study. The NH3 results from the Cosmics Leaving Outdoor Droplets (CLOUD) chamber [Kirkby et al., 2011]
show rates (galactic cosmic ray conditions) that are within a factor of 3 of that given by equation (3), which
is good quantitative agreement. However, considering the differences in experimental conditions between
the two experiments, the agreement might improve (temperature of 292 K versus 298 K) or worsen (NPF
versus TN rates can increasingly diverge as SA decreases). Also, the power dependencies of Kirkby et al.’s data
do not agree with the current work: the NPF rate J1.7 nm goes as [NH3]0.7 at constant [SA] = 1.5 × 108 cm
3
and J1.7 nm goes as [SA]2.4 at 190 pptv NH3. These are significantly less than the 1.6 and 4, respectively,
reported here. Two flow reactor studies that used photolytic generation of H2SO4 presented nucleation rates
that are factors of 5 to 50 below that of equation (3), which is decent quantitative agreement, especially
considering how readily NPF can diverge from TN at low SA. The most recent study out of the CLOUD
consortium [Almeida et al., 2013] presented NH3 measurements at a temperature (278 K) about 20 K lower

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1945
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

than the present measurements. Because nucleation rates in the NH3-H2SO4 system are temperature
dependent [Kirkby et al., 2011; Almeida et al., 2013], a more sophisticated approach is needed for a detailed
comparison with the present results.
The early theoretical studies of Coffman and Hegg [1995] and Napari et al. [2002] applied classical nucleation
theory (“liquid drop” model) to the ammonia-sulfuric acid-water system. That approach yielded nucleation
rates that are many (10 to 12) orders of magnitude greater than the present results. Later classical approaches
[Merikanto et al., 2009] result in predicted nucleation results that are much lower than the present results,
(9 to 10 orders of magnitude). Recent theoretical models based on quantum chemical calculations better
agree with the experimental results. For example, McGrath et al. [2012] presented a model (Atmospheric
Cluster Dynamics Code (ACDC)) that applied quantum chemical energies to calculate nucleation rates. These
rates are about a factor of 300 below equation (3) for [SA] and [NH3] at the lower boundaries of the present
experimental conditions, while at the midpoint of the calculated conditions, the ACDC ammonia prediction
is in good agreement with the extrapolation of the results according to equation (3).
More detailed comparisons among experiments and with theories depend on a number of factors that
can influence the reported nucleation rates, i.e., temperature, cluster wall loss, and scavenging. Some of
these are discussed for the ammonia system in a forthcoming publication (D. R. Hanson and B. Panta,
manuscript in preparation, 2014).
4.4.3. The Dimethylamine/H2SO4 System
Previously published results for DMA and TMA are shown in Table 2 for the typical conditions of each study.
Because the TMA results and the DMA results (see Figures 4c and 4d) for the present experiment are close,
TMA results for previous studies are included in the Table. Using equation (5) as a point of comparison
with which to extrapolate to other conditions, the present results lie about a factor of 25 below the
DMA results of Almeida et al. [2013] for 9 pptv DMA and 0.5 pptv SA. This might be due to the presence of
10 pptv NH3 and a lower temperature (278 K) in Almeida et al.’s experiments, both of which would tend
to increase particle formation rates over the neat DMA system and 298 K. Note that the model results
presented in Almeida et al. suggests that there is little temperature dependence to the nucleation rate in the
DMA-SA system.
Both the present results and those of Almeida et al. suggest that there exists a saturation level with
respect to DMA: above ~3 pptv levels of DMA (~10 for the present results), the production rate of
new particles no longer increases with DMA concentration. This saturation level for the present
experimental results is attributed to scavenging and coagulation losses. Saturation in the Almeida et al.’s
experiment was attributed to the availability of H2SO4, and similarly, Kirkby et al. [2011] had postulated
that the saturation behavior observed with NH3 was due to achieving a 1:1 base to acid ratio in the
clusters/particles. Equation (5) does not take into account saturation with respect to DMA. However,
it does give a nucleation rate in good agreement with Almeida et al. for some conditions (e.g., J = 0.02 cm3 s
1
at 0.08 pptv SA and 3 pptv DMA) despite the temperature differences. Yet equation (5) has a marked
dependence on DMA at the low SA levels of Almeida et al., which is not exhibited in Almeida et al.’s
experimental values (see their Figure 2a).
Comparisons are complicated because Almeida et al.’s nucleation rates are for 1.7 nm diameter particles
(clusters that contain roughly 8 each SA and DMA molecules), while equation (5) is the nucleation rate for
clusters at ~3 SA molecules. Because SA and thus cluster growth rates are low in Almeida et al.’s experiment,
loss to the chamber wall can cause NPF to decrease significantly from the TN. On the other hand, the lower
temperature (278 K versus 298 K) and the presence of NH3 in Almeida et al. would tend to increase the
TN over that given by equation (5). These effects may cancel semiquantitatively for a set of conditions and
lead to the rough agreement at 3 pptv DMA cited in the previous paragraph.
The predicted nucleation rate using equation (5) is several thousand times the flow reactor results for the
high DMA levels present in Yu et al.’s experiments. In addition to the wide disparity in DMA levels, the
experiments may differ in the mixing and loss of base molecules. Also, wall loss of the small nucleated
clusters can be important in Yu et al.’s [2012] experimental setup due to low growth rates. With SA levels
in the 107 cm
3 range, first-order growth rates will be in the 0.01 s
1 range, which is comparable to or
less than the estimated laminar-flow diffusion-limited (0.01–0.03 s
1) wall loss rates. The NPF rates of
Yu et al. can be quite a bit lower than the nucleation rate of clusters that contain as little as 3 SA molecules.

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1946
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

TheTMA results of Erupe et al. [2011], with a similar setup as Yu et al., are in better agreement with
equation (5), and this may be due to TMA levels being closer to the present conditions. The potential
disparity between NPF and TN, which depends on experimental details, may be part of the explanation for
the remaining discrepancy with equation (5).
Comparison with the flow reactor DMA results of Berndt et al. [2014] show a factor of ~100 disparity from
equation (5). A likely contributing factor to this difference is the loss of clusters to the wall before they are
counted by the particle counter. Also, a nucleation time was not provided, and the flow reactor residence
time was used to estimate J from the reported particle number density.
The model-predicted results of ACDC [McGrath et al., 2012] are within an order of magnitude of
equation (5) for conditions that nearly overlap (4 pptv DMA and 40 pptv SA: just outside the range of
the calculations and close to the present experimental conditions). On the other hand, equation (5)
extrapolated to the midrange values of the ACDC calculations (0.4 pptv for SA and DMA) gives a much
lower TN than ACDC predicts, although revised cluster thermodynamics are likely to lower ACDC rates [Ortega
et al., 2013]. Nonetheless, there is different dependence of TN on reactants SA and DMA between the
experiment and in ACDC.
The empirical model of Chen et al. [2012] gives nucleation rates that are nearly 2 orders of magnitude lower
than the present results. As discussed above, it was developed from measurements at much higher base levels
and much lower acid levels than the present measurements and assumed that ammonia and amines affect
nucleation equally. The modifications of this model presented in Jen et al. [2014] are in better agreement with
the current results; see Table 3 and section 4.5. Note that Jen et al.’s measurements and model were developed
independently from the present measurements and that the decomposition rate for all SA trimers (all clusters
that contain 3 SA molecules) of Chen et al. was used in the model.
4.4.4. Synergistic Effect of NH3 With an Amine
There is an increase in particle formation when NH3 was added along with single-digit pptv levels of MA
or DMA. As seen in Figure 6, for 2 pptv of the amine and 200 pptv of ammonia, there was about a factor of
100 increase over the 2 pptv MA-only case and about a factor of 10 increase over the 2 pptv DMA-only
case. A synergistic effect of NH3 and DMA was also noted by Yu et al. [2012], who observed particle
numbers to increase about a factor of 10 when ~500–800 pptv NH3 was present with 1000–1500 pptv
DMA (compared to 1500–1800 pptv DMA alone). Yu et al. [2012] showed that this effect is important for
atmospheric levels of SA, and our observations show that this synergism is important for atmospheric
levels of ammonia and the amines.
The synergistic effect has important implications for other experimental work, for mechanisms, and
for atmospheric observations. This effect might be a partial explanation for the lack of an amine
dependence above 3 pptv DMA reported by Almeida et al. [2013] if the synergistic effect of NH3 is
important down to the 10 pptv level. Mechanistically, we observed that the power dependencies
on amine decrease to about 1 and the power dependency on SA decreases to 1.5 to 2 when NH3
was present. In addition, our limited set of measurements showed little dependence on ammonia
level over the range of 180 to 600 pptv. More measurements are needed to better quantify the
ammonia dependence.
The synergistic effect may be very important for nucleation in the atmosphere. Atmospheric levels of
ammonia are typically hundreds to thousands of pptv, while amines can be tens of pptv [Hanson et al., 2011;
Freshour et al., 2014]. For comparisons, and for preliminary application in atmospheric models, the following
equation (with SA and DMA in cm
3) gives nucleation rates for the synergistic effect using the power
dependencies we observe.


J DMA; syn cm
3 ¼ 110
20 SA2 DMA: (6)

This is a preliminary formulation of the synergistic effect that can be applied to atmospheric conditions and
for NH3 abundances of 200 pptv. Atmospheric conditions involve a large extrapolation in SA from those in
our experiment, and a power dependency of 2 is assumed. Also, it is possible that at low SA, a dependence
upon NH3 will be important, which may significantly affect the nucleation rate. Future work is needed on this
synergistic effect for a range of experimental conditions, e.g., at different temperatures and at lower SA. Also,

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1947
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

7
3
Table 3. Nucleation Rates, TN, at 298 K and [SA] = 10 cm and [DMA] better mechanistic information can
8
3a
= 10 cm

3
1
be gained from the dependence of
Reference Note J(cm s ) the nucleation rate on the SA and
This work equation (5) 1 the base levels.
b
Chen et al. [2012] DMA 0.01
with 400 pptv NH3
c
0.8 4.5. Comparison of Models and
Jen et al. [2014] DMA 20 the Atmosphere
d 4
ACDC DMA 1 × 10
e Nucleation rates using various
Field work 10–100
Synergistic, this work with 200 pptv NH3
f
100 models at 298 K and [SA] and
a [DMA] levels of 107 and 108 cm
3,
Last row considers also the presence of hundreds of pptv of NH3.
b
DMA alone. Value for κ3 includes a typical wall loss rate coefficient, respectively, typical of the atmosphere,

1
0.04 s . are compared in Table 3. The results
c 10
3
Assuming that NH3 acts the same as DMA, with [base] = 10 cm . of the present work extrapolated to
d
See table note f in Table 2.
e 7
These values cover the range of J reported at an [SA] = 10 cm .

3 these conditions, both equation (5)
Field data as reported in Jen et al. [2014] and Chen et al. [2012]. Also for DMA alone and equation (6) for
consistent with the J1.5 nm reported in Kulmala et al., 2013. the synergistic effect, and the model
f
Extrapolated from Figure 6 using equation (6).
of Jen et al. [2014] predict rates
that are within the realm of the
field-measured nucleation rates. The model of Chen et al. [2012] is several orders of magnitude low for
4 pptv (108 cm
3) DMA, but the model is in rough agreement with the field observations if 400 pptv NH3 is
considered an efficient base also (which had been assumed in Chen et al.) The ACDC model [McGrath
et al., 2012] for 4 pptv of DMA predicts very high rates: 100 to 1000 times larger than those observed in
the atmosphere. Note that model rates will be affected by the change in thermochemistry presented
in Ortega et al. [2013]. It is not clear what effect NH3 at 400 pptv will have on predicted nucleation
rates in the ACDC model.
The synergistic effect as demonstrated here reveals that hundreds of pptv of NH3 in addition to an amine has
a large effect on the nucleation rate compared to the amine alone. Assuming an SA2 dependence and a linear
dependence on amine, it is predicted to play a large role in forming new particles in the atmosphere. The
effect of an amine and ammonia together with SA needs more study to fully understand its mechanism and
to assess the accuracy of equation (6) at the low levels of SA found in the atmosphere.

5. Conclusions
The effectiveness of promoting nucleation with SA increases as the base changes from NH3 to MA then to
TMA and DMA. At high enough SA, the difference in effectiveness between the amines diminishes. At the
upper range of SA explored here, tens of pptv of these amines produce so many particles that coagulation
and scavenging limited their number density. Data with particle number densities less than 107 cm
3 were
fit with power dependency relationships on [SA] and on [base]. These relations for NH3 or DMA with SA and
for both present with SA were used to extrapolate the present results to other experimental conditions,
particularly low [SA]. In addition, PDs of J on base and SA were used to discuss molecular level information on
the composition of the crucial cluster.
Three other bases, a large tertiary amine, TEA, and two amides, urea and acetamide, were found to
be much less effective than TMA and DMA at promoting nucleation with SA. In fact, the amides urea
and acetamide were particularly inefficient in comparison to the amines. This finding is in accord with
the low basicity of amides compared to amines. It is likely that a whole class of compounds, amides,
can be eliminated from nucleation considerations in the atmosphere. As shown for pyridine by Berndt
et al. [2014], our results for TEA show that the structure of the amine plays a role in addition to its
gas-phase basicity.
The synergistic effect of 100s of pptv ammonia present along with single-digit pptv levels of MA or DMA led
to enhanced nucleation rates and changes in power dependencies compared to the amines alone results.
Since the atmosphere typically has much higher levels of ammonia than it does amines, this effect is
particularly relevant to the atmosphere. A crude relationship for the DMA results presented here shows
that it is likely to play a significant role in atmospheric nucleation.

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1948
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

Acknowledgments References
The National Science Foundation awards
AGS1338706, AGS1068201, and Almeida, J., et al. (2013), Molecular understanding of sulphuric acid-amine particle nucleation in the atmosphere, Nature (London, U. K.),
AGS0943721 provided funds for this 502(7471), 359–363, doi:10.1038/nature12663.
work. The authors would like to thank Ball, S. M., D. R. Hanson, F. L. Eisele, and P. H. McMurry (1999), Laboratory studies of particle nucleation: Initial results for H2SO4, H2O, and NH3
Michael Sienkowski and Juan Tigre-Lazo vapors, J. Geophys. Res., 104(D19), 23,709–23,718, doi:10.1029/1999JD900411.
for collecting a portion of the MA Benson, D. R., J. H. Yu, A. Markovich, and S.-H. Lee (2011), Ternary homogeneous nucleation of H2SO4, NH3, and H2O under conditions
nucleation data and Natasha Clark and relevant to the lower troposphere, Atmos. Chem. Phys., 11(10), 4755–4766, doi:10.5194/acp-11-4755-2011.
Oscar Martinez for collecting data to help Berndt, T., et al. (2010), Laboratory study on new particle formation from the reaction OH + SO2: Influence of experimental conditions, H2O
assess the experimental apparatus. All vapour, NH3 and the amine tert-butylamine on the overall process, Atmos. Chem. Phys., 10(15), 7101–7116, doi:10.5194/acp-10-7101-2010.
data in this work are available upon Berndt, T., et al. (2014), Enhancement of atmospheric H2SO4 / H2O nucleation: Organic oxidation products versus amines, Atmos. Chem. Phys.,
request to [email protected]. 14, 751–764, doi:10.5194/acp-14-751-2014.
Chen, M., et al. (2012), Acid/base chemical reaction model for nucleation rates in the polluted atmospheric boundary layer, Proc. Natl. Acad.
Sci. U.S.A., 109(46), 18,713–18,718, doi:10.1073/pnas.1210285109.
Clarke, A. D., J. L. Varner, F. Eisele, R. L. Mauldin, D. Tanner, and M. Litchy (1998), Particle production in the remote marine atmosphere: Cloud
outflow and subsidence during ACE 1, J. Geophys. Res., 103, 16,397–16,409, doi:10.1029/97JD02987.
Coffman, D. J., and D. A. Hegg (1995), A preliminary study of the effect of ammonia on particle nucleation in the marine boundary layer,
J. Geophys. Res., 100(D4), 7147–7160, doi:10.1029/94JD03253.
Erupe, M. E., A. A. Viggiano, and S.-H. Lee (2011), The effect of trimethylamine on atmospheric nucleation involving H2SO4, Atmos. Chem.
Phys., 11, 4767–4775, doi:10.5194/acp-11-4767-2011.
Freshour, N., K. K. Carlson, Y. A. Melka, S. Hinz, B. Panta, and D. R. Hanson (2014), Amine permeation sources characterized with acid
neutralization and sensitivities of an amine mass spectrometer, Atmos. Meas. Tech., 7, 3611–3621, doi:10.5194/amt-7-3611-2014.
Ge, X., A. S. Wexler, and S. L. Clegg (2011), Atmospheric amines - Part II. Thermodynamic properties and 5 gas/particle partitioning, Atmos.
Environ., 45, 561–577.
Hamaker, H. C. (1937), The London—van der Waals attraction between spherical particles, Physica, 4, 1058–1072.
Hanson, D. R., P. H. McMurry, J. Jiang, D. Tanner, and L. G. Huey (2011), Ambient pressure proton transfer mass spectrometry: Detection of
amines and ammonia, Environ. Sci. Technol., 45(20), 8881–8888.
Iida, K., M. R. Stolzenburg, and P. H. McMurry (2009), Effect of working fluid on sub-2 nm particle detection with a laminar flow ultrafine
condensation particle counter, Aerosol Sci. Technol., 43(1), 81–96.
Intergovernmental Panel on Climate Change (2013), Climate Change 2013: IPCC 5th Assessment Report (AR5), edited by T. F. Stocker et al.,
Cambridge Univ. Press, Cambridge, U. K.
Jen, C., P. H. McMurry, and D. R. Hanson (2014), Stabilization of sulfuric acid dimers by ammonia, methylamine, dimethylamine, and
trimethylamine, J. Geophys. Res. Atmos., 119, 7502–7514, doi:10.1002/2014JD021592.
Jiang, J., M. Chen, C. Kuang, M. Attoui, and P. H. McMurry (2011), Electrical mobility spectrometer using a diethylene glycol
condensation particle counter for measurement of aerosol size distributions down to 1 nm, Aerosol Sci. Technol., 45(4), 510–521,
doi:10.1080/02786826.2010.547538.
Jiang, J., C. Kim, X. Wang, M. R. Stolzenburg, S. L. Kaufman, C. Qi, G. J. Sem, H. Sakurai, N. Hama, and P. H. McMurry (2014), Aerosol charge
fractions downstream of six Bipolar chargers: Effects of ion source, source activity, and flow rate, Aerosol Sci. Technol., 48(12), 1207–1216,
doi:10.1080/02786826.2014.976333.
Jones, A. H. (1960), Sublimation pressure data for organic compounds, J. Chem. Eng. Data, 5, 196–200.
Kazil, J., and Lovejoy, E. R. (2007), A semi-analytical method for calculating rates of new sulfate aerosol formation from the gas phase, Atmos.
Chem. Phys., 7, 3447–3459, doi:10.5194/acp-7-3447-2007.
Kazil, J., Lovejoy, E. R., Jensen, R. J., and Hanson, D. R. (2007), Is aerosol formation in cirrus clouds possible?, Atmos. Chem. Phys., 7, 1407–1413,
doi:10.5194/acp-7-1407-2007.
Kirkby, J., et al. (2011), Role of sulphuric acid, ammonia and galactic cosmic rays in atmospheric aerosol nucleation, Nature (London, U. K.),
476(7361), 429–433, doi:10.1038/nature10343.
Kuang, C., P. H. McMurry, A. V. McCormick, and F. L. Eisele (2008), Dependence of nucleation rates on sulfuric acid vapor concentration in
diverse atmospheric locations, J. Geophys. Res., 113, D10209, doi:10.1029/2007JD009253.
Kulmala, M., et al. (2013), Direct observations of atmospheric aerosol nucleation, Science, 339, 943–946.
Kurten, T., V. Loukonen, H. Vehkamaki, and M. Kulmala (2008), Amines are likely to enhance neutral and ion-induced sulfuric acid-water
nucleation in the atmosphere more effectively than ammonia, Atmos. Chem. Phys., 8(14), 4095–4103, doi:10.5194/acp-8-4095-2008.
Larriba, C., C. J. Hogan, M. Attoui, R. Barrajo, J. Fernandez-Garcia, and J. Fernandez de la Mora (2011), The mobility-volume relationship below
3 nm examined by tandem mobility-mass measurement, Aerosol Sci. Technol., 45, 453–467.
Lovejoy, E. R., J. Curtius, and K. D. Froyd (2004), Atmospheric ion induced nucleation of sulfuric acid and water, J. Geophys. Res., 109, D08204,
doi:10.1029/2003JD004460.
McGrath, M. J., T. Olenius, I. K. Ortega, V. Loukonen, P. Paasonen, T. Kurtén, M. Kulmala, and H. Vehkamäki (2012), Atmospheric cluster
dynamics code: A flexible method for solution of the birth-death equations, Atmos. Chem. Phys., 12, 2345–2355, doi:10.5194/
acp-12-2345-2012.
McMurry, P. H. (1980), Photochemical aerosol formation from sulfur dioxide: A theoretical analysis of smog chamber data, J. Colloid Interface
Sci., 78(2), 513–527, doi:10.1016/0021-9797(80)90589-5.
McMurry, P. H. (1983), New particle formation in the presence of an aerosol: Rates, time scales and sub-0.01 μm size distributions, J. Colloid
Interface Sci., 95(1), 72–80.
McMurry, P. H., and S. K. Friedlander (1979), New particle formation in the presence of an aerosol, Atmos. Environ., 13, 1635–1651.
McMurry, P. H., et al. (2005), A criterion for new particle formation in the sulfur-rich atlanta atmosphere, J. Geophys. Res., 110, D22S02,
doi:10.1029/2005JD005901.
Merikanto, J., I. Napari, H. Vehkamaki, T. Anttila, and M. Kulmala (2009), Correction to “New parameterization of sulfuric acid-ammonia-water
ternary nucleation rates at tropospheric conditions”, J. Geophys. Res., 114, D09206, doi:10.1029/2009JD012136.
Metzger, A., et al. (2010), Evidence for the role of organics in aerosol particle formation under atmospheric conditions, Proc. Natl. Acad. Sci.
U.S.A., 107(15), 6646–6651, doi:10.1073/pnas.0911330107.
Nadykto, A., and F. Yu (2011), Amine’s in the Earth’s atmosphere: A density functional, Entropy, 13, 554–569, doi:10.3390/e13020554.
Napari, I., M. Noppel, H. Vehkamäki, and M. Kulmala (2002), Parametrization of ternary nucleation rates for H2SO4-NH3-H2O, J. Geophys. Res.,
107(D19), 4381, doi:10.1029/2002JD002132.

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1949
 21698996, 2015, 5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2014JD022730, Wiley Online Library on [04/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2014JD022730

Nel, A. (2005), Air pollution-related illness: Effects of particles, Science, 308(5723), 804–806, doi:10.1126/science.1108752.
Olenius, T., T. Kurtén, O. Kupiainen-Määttä, H. Henschel, I. K. Ortega, and H. Vehkamäki (2014), Effect of hydration and base contaminants on
sulfuric acid diffusion measurement: A computational study, Aerosol Sci. Technol., 48(6), 593–603, doi:10.1080/02786826.2014.903556.
Ortega, I. K., et al. (2012), From quantum chemical formation free energies to evaporation rates, Atmos. Chem. Phys., 12(1), 225–235,
doi:10.5194/acp-12-225-2012.
Ortega, I. K., et al. (2013), Corrigendum to "From quantum chemical formation free energies to evaporation rates", Atmos. Chem. Phys., 13,
3321–3327, doi:10.5194/acp-13-3321-2013.
Oxtoby, D. W. (1994), A general relation between the nucleation work and the size of the nucleus in multicomponent nucleation, J. Chem.
Phys., 100(10), 7665–7671, doi:10.1063/1.466859.
Panta, B., W. A. Glasoe, J. H. Zollner, K. K. Carlson, and D. R. Hanson (2012), Computational fluid dynamics of a cylindrical nucleation flow
reactor with detailed cluster thermodynamics, J. Phys. Chem. A, 116, 10,122–10,134, doi:10.1021/jp302444y.
Riccobono, F., et al. (2014), Oxidation products of biogenic emissions contribute to nucleation of atmospheric particles, Science, 344, 717,
doi:10.1126/science.1243527.
Schobesberger, S., et al. (2013), Molecular understanding of atmospheric particle formation from sulfuric acid and large oxidized organic
molecules, Proc. Natl. Acad. Sci. U.S.A., 110(43), 17,223–17,228, doi:10.1073/pnas.1306973110.
Sipila, M., et al. (2010), The role of sulfuric acid in atmospheric nucleation, Science, 327(5970), 1243–1246, doi:10.1126/science.1180315.
Stolzenburg, M. R., and P. H. McMurry (1991), An ultrafine aerosol condensation nucleus counter, Aerosol Sci. Technol., 14(1), 48–65,
doi:10.1080/02786829108959470.
Vehkamaki, H., et al. (2012), Rethinking the application of the first nucleation theorem to particle formation, J. Chem. Phys., 136, 094107,
doi:10.1063/1.3689227.
Weber, R. J., J. Marti, P. H. McMurry, F. L. Eisele, D. J. Tanner, and A. Jefferson (1996), Measured atmospheric new particle formation rates:
Implications for nucleation mechanisms, Chem. Eng. Commun., 151, 53–64.
Weber, R. J., J. J. Marti, P. H. McMurry, F. L. Eisele, D. J. Tanner, and A. Jefferson (1997), Measurements of new particle formation and ultrafine
particle growth rates at a clean continental site, J. Geophys. Res., 102(D4), 4375–4385, doi:10.1029/96JD03656.
Weber, R. J., et al. (1998), A study of new particle formation and growth involving biogenic and trace gas species measured during ACE 1,
J. Geophys. Res., 103(D13), 16,385–16,396, doi:10.1029/97JD02465.
+ + + 2




Wexler, A. S., and S. L. Clegg (2002), Atmospheric aerosol models for systems including the ions H , NH4 , Na , SO4 , NO3 , Cl , Br and
H2O, J. Geophys. Res., 107(D14), 4207, doi:10.1029/2001JD000451.
Yu, F., and R. Turco (2000), Ultrafine aerosol formation via ion-mediated nucleation, Geophys. Res. Lett., 27, 883–886, doi:10.1029/1999GL011151.
Yu, H., R. McGraw, and S. Lee (2012), Effects of amines on formation of sub-3 nm particles and their subsequent growth, Geophys. Res. Lett., 39,
L02807, doi:10.1029/2011GL050099.
Zollner, J. H., W. A. Glasoe, B. Panta, K. K. Carlson, P. H. McMurry, and D. R. Hanson (2012), Sulfuric acid nucleation: Power dependencies,
variation with relative humidity, and effect of bases, Atmos. Chem. Phys., 12(10), 4399–4411, doi:10.5194/acp-12-4399-2012.

GLASOE ET AL. ©2015. American Geophysical Union. All Rights Reserved. 1950