Subscriber access provided by - Access paid by the | UCSB Libraries
Article
Direct measurements of gas/particle partitioning and mass accommodation coefficients in environmental chambers Jordan Edward Krechmer, Douglas A. Day, Paul J. Ziemann, and Jose L. Jimenez Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.7b02144 • Publication Date (Web): 31 Aug 2017 Downloaded from http://pubs.acs.org on September 3, 2017
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 31
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
Environmental Science & Technology
Direct measurements of gas/particle partitioning and mass accommodation coefficients in environmental chambers Jordan E. Krechmer%, Douglas A. Day, Paul J. Ziemann, Jose L. Jimenez* Cooperative Institute for Research in Environmental Sciences (CIRES) and Department of Chemistry and Biochemistry, Boulder, Colorado 80309, United States Corresponding author: * Jose-Luis Jimenez University of Colorado; UCB 216, Boulder, CO 80309-0216 Phone: 303-492-3557 Fax: 303-492-1149
[email protected] Keywords: Aerosols, SOA, environmental chambers, gas/particle partitioning Abstract Secondary organic aerosols (SOA) are a major contributor to fine particulate mass and wield substantial influences on the Earth’s climate and human health. Despite extensive research in recent years, many of the fundamental processes of SOA formation and evolution remain poorly understood. Most atmospheric aerosol models use gas/particle equilibrium partitioning theory as a default treatment of gas-aerosol transfer, despite questions about potentially large kinetic effects. We have conducted fundamental SOA formation experiments in a Teflon environmental chamber using a novel method. A simple chemical system produces a very fast burst of lowvolatility gas-phase products, which are competitively taken up by liquid organic seed particles and Teflon chamber walls. Clear changes in the species time evolution with differing amounts of seed allow us to quantify the particle uptake processes. We reproduce gas- and aerosol-phase observations using a kinetic box model, from which we quantify the aerosol mass accommodation coefficient (α) as 0.7 on average, with values near unity especially for low volatility species. α appears to decrease as volatility increases. α has historically been a very difficult parameter to measure with reported values varying over three orders of magnitude. We use the experimentally-constrained model to evaluate the correction factor (Φ) needed for chamber SOA mass yields due to losses of vapors to walls as a function of species volatility and particle condensational sink. Φ ranges from 1 – 4.
1 ACS Paragon Plus Environment
Environmental Science & Technology
Page 2 of 31
37
Introduction
38
Gas/particle partitioning (GPP) is a fundamental physical process describing the interaction
39
between atmospheric organic gases and particles. Originally treated as an adsorptive process,1,2
40
most atmospheric models currently treat GPP of organic compounds as an absorptive process.3–5
41
In this interpretation, semivolatile organic gases condense into and evaporate from particles until
42
quickly reaching a volatility- and available organic aerosol mass-dependent equilibrium.3,4
43
Semivolatile equilibrium partitioning is often parameterized using a volatility basis set, which
44
lumps the amount of condensable material in the air in multiple bins spaced typically by a decade
45
of the saturation mass concentration of the compound.5 GPP has been incorporated into
46
numerous box, regional, and global models6–10 as a default treatment of organic aerosol
47
formation and evolution.
48
Still, modeling of GPP by rapid absorptive equilibrium remains controversial. There are
49
now several real-time analysis techniques capable of measuring the gas/particle distribution of
50
trace organic compounds at relatively high time resolution in the field.11–14 Many of these
51
measurements have shown substantial disagreements with semivolatile equilibrium partitioning
52
theory and one another, however.15,16 Some recent works have even suggested that semivolatile
53
equilibrium partitioning is not reached under certain conditions due to strong kinetic limitations,
54
and more detailed treatments are required to match atmospheric observations.17–21 GPP model treatment is expected to have a large influence on secondary organic aerosol
55 56
(SOA)
mass
concentrations
predicted
by
atmospheric
models.
Most
models
use
57
parameterizations of the SOA yield, i.e. the mass of SOA formed per unit mass of volatile
58
organic compound (VOC) precursor reacted.22 Almost all experiments quantifying SOA mass
59
yields that have been used in models were performed in Teflon-walled environmental
2 ACS Paragon Plus Environment
Page 3 of 31
Environmental Science & Technology
60
chambers.22–26 However, it has recently been reported that Teflon chambers suffer from losses of
61
semivolatile gaseous compounds to chamber walls.27–30 Wall losses can have strong effects on
62
aerosol mass yields,30 and could also affect conclusions about chemical composition and
63
processes gleaned from chamber experiments. The question even arises: if wall losses are
64
substantial for a certain range of compounds, and fast enough to be on similar time scales as the
65
approach to equilibrium GPP, is it possible to observe and accurately characterize GPP kinetics
66
and equilibrium in environmental chambers?
67
In mathematical representations of the kinetics of GPP, the mass accommodation
68
coefficient (α; also known as the sticking coefficient; and equal to the evaporation coefficient at
69
equilibrium) is a critical parameter that defines the fraction of gas/particle collisions that result in
70
a colliding species being taken up by a particle. α for organic species was historically assumed to
71
be 1, mostly from liquid evaporation measurements.31–33 More recent measurements of mass
72
accommodation coefficients for aerosol-phase compounds34–36 range across two orders of
73
magnitude. A very wide range of values (as low as 0.001) has been used in models and data
74
fitting, sometimes as a tuning parameter which might obscure other model structural
75
limitations.30,37–40 Comparing these literature values is complicated by the different methods used
76
to derive α. Some models34,41 and measurements34,35 conflate particle diffusion kinetics and
77
surface accommodation as one effective mass accommodation coefficient. Other models have
78
attempted to specifically separate mass accommodation and particle-phase diffusion
79
coefficients.36,42,43
80
Given the discrepancies that persist between ambient SOA measurements and
81
models,44,45 the substantial fraction of organic material in the atmosphere which is thought to be
82
semivolatile,46 and the difficulty of quantifying GPP in field studies,16 it is important to closely
3 ACS Paragon Plus Environment
Environmental Science & Technology
Page 4 of 31
83
examine in detail the kinetics of approach to equilibrium GPP. . It is also important to further
84
evaluate the impact that losses of vapors to Teflon chamber walls may have in SOA yield
85
measurements. Herein, we conduct fundamental gas/particle partitioning experiments in an
86
environmental chamber with a particular focus on GPP kinetics. A well-characterized simple
87
chemical system is used to produce low-volatility organic compounds very rapidly, which are
88
taken up by liquid organic seed particles and/or the Teflon chamber walls. We use liquid organic
89
seed aerosol specifically to establish an experimental method under which particle-phase kinetic
90
limitations are not present. Both gas-phase products and total aerosol volume and surface area
91
concentrations are continuously monitored. Analysis of these experiments allow us to thoroughly
92
analyze the gas-phase and accommodation kinetics of GPP in the presence of walls for semi- and
93
low-volatility gas-phase compounds. A box model is used to quantify c*and α. We use liquid
94
aerosol particles as seeds, which removes particle-phase diffusion as a limiting parameter and
95
allows us to infer the true value of α. We discuss the implications for quantifying gas/particle
96
partitioning under more complex conditions.
97 98
MATERIALS AND METHODS
99
Environmental chamber experiments
100
A schematic outlining the chamber and instruments used in this work is shown in Figure
101
1. We conducted all of our experiments in the one of the two chambers in the CU Environmental
102
Chamber (CUEC) Facility. The CUEC bag was constructed of FEP Teflon with a maximum
103
volume of 20 m3. The enclosure temperature was maintained at 26 ± 1° C for all experiments.
104
The chambers were filled with clean, dry (< 1% RH) air from two AADCO Model 737-15 clean
105
air generators (Cleaves, OH, USA). The chambers were equipped with an automated flushing
4 ACS Paragon Plus Environment
Page 5 of 31
Environmental Science & Technology
106
system that alternately evacuated and refilled the bag when it was not in use (~5% of full volume
107
every 5 minutes), while maintaining the bag volume at ~95% of maximum volume using
108
differential pressure between the bag and climate-controlled enclosure. Running the automated
109
flushing sequence overnight before each experiment ensured that each experiment started with
110
aerosol concentrations of 0 particles cm-3 as measured with an ultrafine condensation particle
111
counter (UCPC, TSI model 3776 with a diameter cutoff of 2.5 nm). The experimental procedure is similar to the one detailed in Krechmer et al.27 to measure
112 113
vapor wall losses to Teflon walls, but with the addition of seed aerosol. Dioctyl sebacate (DOS)
114
seed was chosen for this set of experiments because it has been shown19 to be reliably liquid
115
under the experimental conditions used here and should not impose particle-phase diffusion
116
limitations on time scales relevant to the experiment. DOS also has a low vapor pressure (2 × 10-
117
6
118
reaction, thus not affecting GPP through particle-phase reactions.
119
Pa)47 and, as a diester, should be unreactive with any of the products formed in the OH˙
First, we injected dioctyl sebacate (DOS; >97%; Sigma-Aldrich; saturation concentration
120
at 298 K (c*) < 0.1 µg m-3)48 liquid seed aerosol into the chamber using an evaporation-
121
condensation generator modeled after Sinclair and Lamer.49 The aerosol generator was
122
configured to seed the chamber to 100 µg m-3 of DOS in ~15 minutes. Particle-phase wall losses
123
occurred on a ~10 hr timescale, but it was not necessary to account for them explicitly in this
124
model since the time-dependent aerosol size distribution measurements were used to constrain
125
the box model. If needed, we diluted the chamber with clean air until the DOS seed surface area
126
reached a desired level. '-condensation generator was tuned to produce aerosol with a lognormal
127
surface area distribution centered at 200 nm. A scanning mobility particle sizer (SMPS)
128
comprised of a TSI model 3080 electrostatic classifier and model 3775 condensation particle
5 ACS Paragon Plus Environment
Environmental Science & Technology
129
counter monitored aerosol number, surface area, and volume concentrations. The SMPS scan
130
was programmed to last 120 s, with 15 s to return to the starting voltage, resulting in a time
131
resolution of 135 s. We provide a summary of different experiments and seed concentrations in
132
Table S1.
133
Page 6 of 31
The chemical system used here was the oxidation of long-chain alcohols by the OH
134
radical under high-NO conditions. This system is the same used by Krechmer et al.27 because the
135
oxidation products are simple and well-characterized (and with consistent chemical functionality
136
in a progression of carbon numbers) as opposed to those of other gaseous precursors such as α-
137
pinene or isoprene.
138
After the seed aerosol concentration stabilized, we injected 50 µL of 1-hexanol, 40 µL of
139
1-octanol, 25 µL of 1-nonanol, 18 µL of 1-decanol, and 12 µL of 1-dodecanol (all >99% purity;
140
Sigma-Aldrich) into the chamber by gently heating the liquids into a clean nitrogen stream (UHP
141
liquid nitrogen evaporate) flowing into the chamber. Injection took ~5 minutes. This resulted in a
142
total VOC concentration in the chamber of 1.4 ppm. We fully mixed the bag contents by turning
143
on a Teflon-coated fan inside the bag for 1 min to mix the contents. Finally, we injected 4 ppm
144
of NO (99%, Matheson Tri-gas) and 4 ppm of methyl nitrite (MeONO, synthesized after the
145
method of Taylor et al.50) into the chamber as pure gases using a small glass bulb and mixed the
146
contents again for 1 min.
147
To initiate chemical production, we turned on the chamber UV black lights (~300-400
148
nm) at 100% intensity (JNO2 = 0.01 s-1) for precisely 10 s via computer-control. MeONO formed
149
OH radicals, which then oxidized the alkanols forming ppt-levels of several oxidized products,27
150
including hydroxynitrates (HN), dihydroxynitrates (DHN), trihydroxynitrates (THN), and
151
carbonyl dihydroxynitrates (CDHN). A list of the compounds, along with their exact masses and
6 ACS Paragon Plus Environment
Page 7 of 31
Environmental Science & Technology
152
estimated saturation concentrations (c*) is provided in Table S2. We did not observe new particle
153
formation. We were able to continuously monitor the latter three types of products with a
154
chemical ionization mass spectrometer (CIMS). This “rapid burst” method, developed
155
previously,27 is critical to these experiments because it produces compounds over a wide range of
156
volatilities over the entire volume of the chamber much faster than the vapors can equilibrate
157
with the particles and walls. After each burst, the system was left to come to equilibrium for at
158
least 1 hr.
159
The amount of aerosol produced by each photooxidation burst comprised an average of
160
7% of the total aerosol volume in the chamber (Figure S1). Therefore, the system can be
161
modeled assuming that the gas-phase compounds are partitioning into pure DOS particles and
162
not into a product compound mixture. The specific molecules studied here are estimated to form
163
~ 20 ppt and account for < ~5% of the SOA. The majority of the aerosol formed is likely to be
164
semivolatile hydroxynitrates, which are not detectable by the nitrate CIMS.27 The commercial
165
iodide CIMS, which would likely detect the majority of aerosol-forming compounds, suffers
166
from a significant time delay for semivolatile products due to gas-wall partitioning in the
167
instrument source.27 This makes it unsuitable for a study such as this where it is critical to
168
measure the gas-phase decays as close to real time as possible.
169
NO3-CIMS
170
A chemical ionization time-of-flight mass spectrometer equipped with a nitrate ion
171
ionization source (NO3-CIMS) monitored gas-phase product concentrations. The instrument51,52
172
and nitrate anion source53 have been described extensively in previous works. We used the
173
instrument in a configuration similar to the one in Krechmer et al.27. Notably here, we placed the
174
instrument inside the CUEC enclosure directly adjacent to the reactor bag. This allowed us to use
7 ACS Paragon Plus Environment
Environmental Science & Technology
Page 8 of 31
175
a 0.6 m long electropolished stainless steel inlet to isothermally bring 10 standard L min-1
176
(defined as 25° C, 1 Atm) of sample air directly from the bag to the instrument without dilution.
177
The short inlet with a sub-sampled center flow minimizes wall losses and results in a residence
178
time of sampled air in the instrument inlet and source of < 4 s. The NO3-CIMS acquired spectra
179
at 1 Hz. We then processed instrument data using Tofware (Tofwerk, AG and Aerodyne, version
180
2.5.8) toolkit within IGOR Pro 6 (Wavemetrics, Inc.) at the same time resolution. Throughout
181
this work we report CIMS results in units of ions s-1 because we are only concerned with relative
182
changes in gas-phase concentration and the CIMS response has previously been shown to be
183
linear.54,55 The NO3-CIMS detected gaseous product molecules as clusters with the nitrate ion,
184
but the NO3- prefix has been removed from reported formulas throughout this paper for clarity.
185
An example mass spectrum of some of the product compounds is provided in Figure S2.
186
Box model
187
A chemical kinetic box model was used to simulate the behavior of gas- and aerosol-
188
phase compounds, accounting for partitioning and evaporation of each oxidized products to walls
189
and particles. A schematic of the model is shown in Figure S3. The model was solved using
190
KinSim v3.24 within IGOR Pro.56
191
The vapor wall loss or condensation rate coefficient, kc was determined using the
192
procedure of Krechmer et al.,27 the results of which are shown in Figure S4. Under these
193
conditions (c*
