advances.sciencemag.org/cgi/content/full/6/22/eaay4945/DC1
Supplementary Materials for
Size-dependent influence of NO
x
on the growth rates of organic aerosol particles
C. Yan*, W. Nie, A. L. Vogel, L. Dada, K. Lehtipalo*, D. Stolze
nburg, R. Wagner, M. P. Rissanen, M. Xiao, L. Ahonen,
L. Fischer, C. Rose, F. Bianchi, H. Gordon, M. Simon, M. Heinri
tzi, O. Garmash, P. Roldin, A. Dias, P. Ye,
V. Hofbauer, A. Amorim, P. S. Bauer, A. Bergen, A.-K. Bernhamme
r, M. Breitenlechner, S. Brilke, A. Buchholz,
S. Buenrostro Mazon, M. R. Canag
aratna, X. Chen, A. Ding, J. Do
mmen, D. C. Draper, J. Duplissy, C. Frege, C. Heyn,
R. Guida, J. Hakala, L. Heikkinen, C. R. Hoyle, T. Jokinen, J.
Kangasluoma, J. Kirkby, J. Kontkanen, A. Kürten,
M. J. Lawler, H. Mai, S. Mathot, R. L. Mauldin III, U. Molteni,
L. Nichman, T. Nieminen, J. Nowak, A. Ojdanic, A. Onnela,
A. Pajunoja, T. Petäjä, F. Piel, L. L. J. Quéléver, N. Sarnela,
S. Schallhart, K. Sengupta, M. Sipilä, A. Tomé, J. Tröstl,
O. Väisänen, A. C. Wagner, A. Ylisirniö, Q. Zha, U. Baltensperg
er, K. S. Carslaw, J. Curtius, R. C. Flagan, A. Hansel,
I. Riipinen, J. N. Smith, A. Virtanen, P. M. Winkler, N. M. Don
ahue, V.-M. Kerminen, M. Kulmala, M. Ehn, D. R. Worsnop
*Corresponding author. Email: chao.yan@helsinki.fi (C.Y.); katr
ianne.lehtipalo@helsinki.fi (K.L.)
Published 27 May 2020,
Sci. Adv.
6
, eaay4945 (2020)
DOI: 10.1126/sciadv.aay4945
This PDF file includes:
Supplementary Materials and Methods
Figs. S1 to S8
Tables S1 and S2
References
Supplementary
Materials and Methods
1.
Measurements
Measurement of sulfuric acid and gas
-
phase HOMs
Concentrations of sulfuric acid and highly oxygenated molecules (HOMs) are measured with
a nitrate
-
ion based chemical ionization atmospheric pressure interface time
-
of
-
flight mass
spectrometer (CI
-
APi
-
TOF)
(
17, 34, 35
)
, which has been deployed in the previ
ous CLOUD
experiments and described in details in previous publications
(
12, 14
)
. The nitrate ions are
produced by exposing nitric acid (HNO
3
)
-
containing sheath flow to soft x
-
ray radiation.
These nitrate ions charge the analyte (e.g., H
2
SO
4
or HOMs) in th
e drift tube, with a reaction
time of 200 ms. After that, the sample flow enters the mass spectrometer, where it is focused
in the APi module and analyzed in the TOF chamber based on the ion mass
-
to
-
charge ratio.
The raw data were processed with the MATLAB
tofTools package (version 603)
(
36
)
. The
data were analyzed in high resolution mode that allowed peaks with different elemental
formulae but under the same unit mass to be evaluated separately. Identified peaks were
further grouped by their carbon number o
r nitrogen number (See Fig. 2).
To quantify sulfuric acid and HOMs, we
calibrate the system with H
2
SO
4
and correct the
mass
-
dependent transmission. First, a general calibration coefficient is obtained for sulfuric
acid. The OH concentration in the chamber
is determined from an independent run using
1,3,5
-
trimethylbenzene. The production rate of sulfuric acid can be then calculated based on
the OH and SO
2
concentrations in the chamber. Together with the well
-
characterized total
sulfuric acid loss rate, the c
oncentration
is
derived and the calibration coefficient can be
obtained. We use this calibration coefficient for quantifying HOMs as well. In addition,
corrections for the size
-
dependent ion transmission
(
37
)
and inlet sampling loss are applied.
The overal
l uncertainty in the sulfuric acid and HOMs concentrations is estimated to be ca. 40
%, assuming a unit charging probability. However, recently studies have shown that the
charging of HOMs by nitrate may vary considerably, depending on the functionality of
HOMs
(
38, 39
)
. Therefore, this method gives the lower bound of the HOM concentration.
The uncertainty for relatively less oxidized HOMs needs further investigation.
Measurement
of
particle
-
phase
products
and
their
thermal
desorption
Chemical
analysis
of
the
bulk
particle
-
phase
composition
is
performed
using
the
Filter
Inlet
for
Gases
and
Aerosols
(FIGAERO)
(
40
)
coupled
to
an
iodide
-
based
chemical
ionization
time
-
of
-
flight
mass
spectrometer
(CI
-
ToF
-
MS)
(Tofwerk,
HTOF).
The
measurement
cycle
involves
particl
e
sampling
for
50
minutes
at
8
SLPM
(standard
liters
per
min)
sample
flow,
followed
by
a
thermal
desorption
ramp
(at
the
maximum
heating
rate)
that
reaches
100
°C
after
three
minutes,
130
°C
after
five
minutes
and
140
°C
after
ten
minutes.
This
heating
pro
file
is
sufficient
for
quantitative
evaporation,
as
the
ion
-
traces
for
low
-
volatility
dimers
are
close
to
background
levels
after
the
ten
-
minute
heating
program.
The
filter
is
replaced
between
experiments,
approximately
every
48
hours,
(Zefluor
membrane,
2
.0
μm
pore
size,
25
mm
diameter,
PALL,
USA).
The
thermal
desorption
gas
flow
is
2
SLPM
of
ultrapure
dry
nitrogen.
The
iodide
ions
are
produced
by
first
passing
dry
nitrogen
over
a
methyliodide
permeation
source
(VICI
International,
USA)
kept
at
30
°C.
The
flow
of
2.2
SLPM
of
the
methyliodide/nitrogen
reagent
gas
mixture
then
passes
through
a
radioactive
charge
conditioner
(210Po,
370
MBq,
Model
P
-
2021,
NRD
Inc.,
USA).
The
reagent
gas
is
kept
dry
in
order
to
maximize
sensitivity
toward
organic
acids
(17)
.
Th
e
IMR
body
is
heated
to
approximately
50
°C
to
avoid
condensation
of
low
-
volatility
compounds;
an
active
pressure
control
device
(Aerodyne
Inc.,
USA)
maintained
the
pressure
at
800
mbar.
The
instrument
is
calibrated
with
a
mixture
of
organic
acid
standards
,
but
the
data
shown
here
are
only
qualitative.
Data
analysis
is
performed
with
Tofware
version
2.5.10_FIGAERO
on
10
seconds
average
spectra.
A
post
-
acquisition
mass
calibration
is
conducted
based
on
the
ions
I
−
,
I(H
2
O)
−
,
I(HNO
3
)
−
and
I
3
−
.
Peak
identificat
ion
of
the
particle
phase
products
is
conducted
by
manual
peak
identification
that
allowed
nC:
1
-
20,
nH:
2
-
50,
nO:
0
-
20.
Based
on
different
experimental
conditions
(both
presence
of
NO
x
and
SO
2
),
the
allowed
number
of
nitrogen
and
sulfur
is
0
-
2
and
0
-
1,
re
spectively.
Only
elemental
combinations
of
the
neutral
molecules
having
an
integer
double
bond
equivalent
are
considered.
Overfitting
of
the
peaks
is
avoided
by
allowing
≤
4
molecular
compositions
per
nominal
mass.
The
appearance
of
organic
nitrates
and
or
ganic
di
-
nitrates
is
closely
observed
with
the
increase
of
the
NO
x
levels
during
the
experiment.
The
gas
-
phase
adsorption
on
the
filter
is
corrected
by
subtracting
the
adsorbed
fraction
from
the
measured
intensity
during
the
early
stages
of
the
experiment,
when
the
gas
-
phase
is
in
steady
-
state
while
particle
mass
was
still
building
up.
Gas
-
phase
concentrations
of
the
most
of
the
oxidation
products
are
too
low
throughout
all
experiments
to
be
monitored
in
real
-
time
by
the
gas
-
phase
sampling
position
of
the
F
IGAERO
inlet.
In
the
thermos
-
desorption
measurement
for
the
T
max
of
HOM
monomer,
we
noticed
strong
influence
from
the
thermal
decomposition
of
dimeric
compounds,
which
also
has
been
shown
by
Stolzenburg
and
co
-
workers
(
41
)
.
We
commonly
observed
a
bimodal
character
of
the
thermograms
of
several
monomeric
compounds
(e.g.
C
10
H
16
O
4
,
see
Fig.
S4).
In
this
study
(monoterpenes,
NO
x
,
SO
2
),
it
turns
out
that
the
complexity
is
increased,
and
we
cannot
rule
out
that
thermal
decomposit
ion
of
organic
sulfates
and/or
organic
nitrates
during
the
evaporation
adds
another
dimension
of
complexity
to
the
interpretation
of
thermograms.
Therefore,
we
focused
on
the
region
of
dimeric
mass
regions,
in
which
we
can
exclude
to
introduce
a
bias
that
is
caused
by
thermal
decomposition.
Measurement
of
monoterpenes
The
concentrations
of
monoterpenes
and
other
volatile
organic
compounds
(e.g.
1,3,5
-
trimethylbenzene)
are
measured
with
a
newly
-
developed
prototype
of
the
proton
transfer
reaction
time
-
of
-
flight
mass
spectrometer
(PTR
-
TOF
-
MS;
model:
PTR3).
Compared
with
the
pr
evious
model,
the
PTR3
significantly
improves
the
detection
of
low
volatility
compounds
by
reducing
the
sampling
loss
and
increasing
the
sensitivity.
More
details
about
the
instrument
can
be
found
elsewhere
(
42
)
.
Measurement
of
Particle/ion
number
concentr
ation
and
size
distribution
The
particle/ion
concentration
and
number
size
distribution
in
the
chamber
are
measured
with
several
independent
instruments.
The
particle
size
magnifier
(PSM,
Airmodus
Ltd.)
(
43
)
,
coupled
with
a
condensation
particle
counter
(CPC)
is
used
to
determine
the
number
concentration
of
the
smallest
particles.
The
PSM
uses
diethylene
glycol
(DEG)
as
working
fluid
at
constant
or
varied
supersaturation
conditions
achieved
by
tuning
the
mixi
ng
ratios
between
heated
DEG
-
saturated
air
and
the
sample
flow,
which
determines
the
instruments
lower
cut
-
off
size.
The
external
CPC
is
used
to
grow
particles
further
with
butanol
and
to
determine
their
number
concentration.
In
this
study
we
use
the
scann
ing
PSM
to
determine
the
particle
concentration
at
several
different
cut
-
off
sizes
(1.7
nm
used
for
nucleation
rates),
and
the
number
size
distributions
between
about
1.3
and
3
nm
(
44
)
.
In
addition,
several
PSMs
at
fixed
cut
-
off
sizes
and
a
conventional
bu
tanol
ultra
-
fine
CPC
(TSI
3776)
with
a
cut
-
off
size
of
ca.
2.5
nm
is
used
for
comparison.
The
size
distribution
of
particles
in
the
1.9
-
7
nm
size
range
was
measured
with
the
DMA
-
train
(
45
)
.
This
is
a
recently
developed
instrument,
consisting
of
six
pairs
of
differential
mobility
analyzers
(DMAs)
and
CPCs
operated
in
parallel
at
fixed
sizes.
This
method
achieves
a
100
%
measurement
duty
cycle
that
allows
exploitation
of
the
full
counting
statistics
at
all
six
sizes,
enabling
high
time
-
resolution
and
high
se
nsitivity
to
low
particle
concentrations.
Measurement
of
larger
particles
were
conducted
with
a
commercial
nano
-
SMPS
(TSI
3938)
coupled
with
a
water
-
CPC
(TSI
3788),
and
a
custom
-
built
SMPS,
consisting
of
a
TSI
X
-
ray
source,
a
long
DMA
and
a
CPC
(TSI
3010).
Overall,
the
measured
size
distribution
spans
the
range
from
about
1
to
500
nm.
A
neutral
cluster
and
air
ion
spectrometer
(NAIS,
Airel
Ltd.)
(
46
)
was
deployed
to
measure
the
ion
concentrations
and
size
distributions.
It
simultaneously
determines
the
numb
er
size
distribution
of
positive
and
negative
ions
in
the
range
of
0.75
–
45
nm
(ion
mobility
diameter)
with
two
cylindrical
mobility
spectrometers
in
parallel,
one
for
each
polarity.
Additionally,
a
corona
unipolar
charger
is
periodically
switched
on
to
c
harge
the
particles
to
enable
measurement
of
the
total
particle
size
distribution
over
the
2
–
45
nm
size
range.
Measurement
of
other
trace
gases
Gas
monitors
are
used
to
measure
the
concentration
of
sulfur
dioxide
(SO
2
,
ThermoFisher
Scientific,
Inc.,
model:
42i
-
TLE)
and
ozone
(O
3
,
ThermoFisher
Scientific,
Inc.,
model:
49C).
In
addition,
an
accurate
measurement
of
low
-
level
nitrogen
oxide
(NO)
concentrations
is
achieved
with
an
advanced
NO
monitor
(ECO
PHYSICS,
model:
CLD
780
TR),
which
has
a
detection
limit
of
ca.
3
pptv
for
a
1
min
integration
time.
In
all
experiments
reported
here,
NO
is
sampled
from
the
middle
of
the
chamber
in
the
same
way
as
the
other
gases.
The
NO
mixing
ratio
close
to
its
injection
port
is
measured
as
about
a
factor
of
5
higher.
Howeve
r,
the
space
with
highly
concentrated
NO
should
be
very
limited
compared
to
the
overall
chamber
volume;
thus,
the
effect
of
this
NO
hotspot
is
neglected
in
this
study.
The
concentration
of
nitrogen
dioxide
(NO
2
)
is
measured
with
a
cavity
-
attenuated
phase
-
s
hift
nitrogen
dioxide
monitor
(CAPS
NO
2
,
Aerodyne
Research
Inc.).
The
baseline
signal
is
monitored
periodically
by
flushing
the
inlet
line
with
synthetic
air.
The
NO
2
concentration
is
found
to
be
similar
when
sampling
from
different
ports,
suggesting
that
the
NO
2
concentration
is
homogeneous
in
the
chamber.
2.
Determination
of
particle
nucleation
rate
and
growth
rate
The
nucleation
rates
(
J
)
are
calculated
from
the
time
derivative
of
the
total
particle
concentration
and
corrected
for
the
particle
losses
in
the
chamber
using
the
full
size
distribution.
퐽
=
푑푁
푑푡
+
푆
푑푖푙
+
푆
푤푎푙푙
+
푆
푐표푎푔
(cm
-
3
s
-
1
)
(1)
where
N
is
the
particle
concentration
above
a
certain
cut
-
off
s
ize
(
d
p
)
to
which
the
nucleation
rate
is
calculated.
The
dilution
correction
S
dil
arises
from
the
fact
that
the
chamber
is
constantly
flushed
with
synthetic
air
to
account
for
the
instruments’
sample
flows.
푆
푑푖푙
=
푁
·
푘
푑푖푙
(cm
-
3
s
-
1
)
(2)
where
k
dil
= 1.437
10
-
4
s
-
1
for CLOUD 10.
Diffusional losses to the chamber walls (
S
wall
) are determined empirically by observing the
decay of the sulfuric acid
monomer concentration in the chamber. The wall loss rate is
inversely proportional to the particle size:
푘
푤푎푙푙
(
푑
푝
′
,
푇
)
=
2
.
116
·
10
−
3
·
(
푇
푇
푟푒푓
)
0
.
875
·
(
푑
푝
,
푟푒푓
푑
푝
′
)
(s
-
1
) (3)
where
d
p
’
is the mobility diameter of the particle,
d
p,r
ef
is the mobility diameter of the sulfuric
acid monomer (= 0.82 nm),
T
ref
= 278 K, and
T
is the actual chamber temperature. Thus the
total wall loss for particles larger than
d
p
is:
푆
푤푎푙푙
(
푑
푝
,
푇
)
=
∑
푁
(
푑
푝
′
)
·
푘
푤푎푙푙
(
푑
푝
′
,
푇
)
푑
푝
,
푚푎푥
′
푑
푝
′
=
푑
푝
(cm
-
3
s
-
1
) (4)
Coagulation losses to the surface of larger aerosol particles (
S
coag
) are calculated from the
measured number size distribution of particles present in the chamber
푆
푐표푎푔
(
푑
푝
,
푘
)
=
∑
∑
훿
푖
,
푗
∙
퐾
(
푑
푝
,
푖
,
푑
푝
,
푗
)
∙
푁
푖
∙
푁
푗
푑
푝
,
푚푎푥
푑
푝
,
푗
=
푑
푝
,
푖
푑
푝
,
푚푎푥
푑
푝
,
푖
=
푑
푝
,
푘
(cm
-
3
s
-
1
)
(5)
where
K(d
p,i
,d
p,j
)
is the coagulation coefficient for particles of size
d
p,i
and
d
p,j
,
N
i
and
N
j
are
the number densities of particles in a size bins i and j, and
δ
i,j
= 0.5, if i = j and
δ
i,j
= 1, if i ≠ j.
The nucleation rates at 1.7 nm (
J
1.7
) are calculated from the scanning PSM and verified
against the values calculated from the two other PSMs and the butanol CPC at fixed cut
-
off
sizes. It should be noted that ther
e is an uncertainty of about 0.5 nm in the cut
-
off size of the
particle counters due to the effect of composition and charge on the detection efficiency
(
47
)
.
To account for this, we verify the cut
-
off size of the PSM for each chemical system in the
chambe
r by comparing the concentration and rising time of the PSM at different saturator
flow rates against the different size bins of the NAIS, which has been shown to be very
accurate in determining the ion mobility. The
J
value given for each experiment is th
e median
value after reaching stable conditions
The
growth
rates
are
calculated
using
the
appearance
time
(
48
)
,
as
shown
in
Fig.
1b
(the
white
dots).
The
appearance
time
at
each
particle
size
is
defined
as
the
time
when
particle
concentration
at
a
certai
n
size
reaches
half
of
its
maximum
concentration.
To
exclude
the
systematic
difference
between
these
instruments,
the
appearance
time
from
different
instruments
are
checked
for
consistency
in
overlapping
size
regions,
including
the
scanning
PSM
(1.3
–
2.3
nm),
the
DMA
-
train
(1.9
–
3.5
nm
and
3.5
–
7
nm),
NAIS
(2
–
40
nm),
the
nano
-
SMPS
(4
–
30
nm),
and
the
SMPS
(>30
nm).
In
principle,
the
growth
rate
can
be
determined
at
any
size,
but,
in
practice,
we
fit
the
growth
rate
using
the
data
from
individual
instr
uments
at
six
fixed
size
ranges;
these
are
1.3
–
2.3
nm,
1.9
–
3.5
nm,
3.5
–
7
nm,
7
–
20
nm,
20
–
30
nm
and
>30
nm.
It
should
be
noted
that
the
growth
rates
in
the
smallest
size
ranges
are
difficult
to
define,
and
different
methods
might
differ
from
each
other
depending
non
-
linearly
on
the
environmental
conditions
(
49
)
.
3.
Deduced
HOM
formation
pathways
in
the
presence
of
NO
x
NO
x
can affect the HOM formation in many ways. We need to firstly consider how NO
x
influenced the main oxidants of monoterpenes. The concentration of O
3
can be influenced in
three channels: First, NO can reduce O
3
concentration by directly reacting with it; Second,
NO
2
can produce O
3
via its photolysis, especially with the presence of R
O
2
or HO
2
radicals
that recycle NO back to NO
2
; Third, NO
2
can also react with O
3
to form NO
3
radical, another
important oxidant of monoterpenes
(
50, 51
)
. Regarding to the O
3
concentration, Channel 1
was the most important, as we observed a net reduction o
f O
3
, which was almost equivalent to
the net production of NO
2
. Although only a tiny amount of NO
3
was formed via Channel 3 in
comparison to NO and NO
2
, it had a considerable contribution to the production of HOMs, as
we will discuss below. The OH concentr
ation can also be influenced in multiple ways: First,
both NO and NO
2
directly react with OH, leading to a reduction of OH concentration;
Second, NO can react with HO
2
the produce OH; Third, the OH production via monoterpene
ozonolysis is influenced via af
fecting the O
3
concentration. The overall influence on OH
concentration can be inferred from the change of H
2
SO
4
concentration at the constant SO
2
concentration. In the conditions we focused on in this study, i.e., zero and 1.9 ppbv NO
x
,
the
concentrations
of O
3
and OH were not significantly affected: when 1.9 ppbv NO
x
were
injected, the O
3
and OH concentrations decreased by about 3 % and 10 %, respectively
(Table
S1)
.
After excluding a substantial decrease in OH and O
3
concentrations, the reduction of
C
x
H
y
O
z_di
can be mainly attributed to the reaction between NO
x
and RO
2
(from the
monoterpene oxidized by
O
3
and OH
)
(
19
)
, competing with the suggested dimer formation via
the accretion reaction of two RO
2
(
52, 53
)
. Meanwhile, this reaction also leads to the
f
ormation of C
x
H
y
O
z
N
_mono
. Under most of our chamber conditions, NO, NO
2
and NO
3
were
positively correlated,
so their relative contribution cannot be easily distinguished with a single
experiment. Because of this,
we compare
the HOM composition in two runs
(Run1752 and
Run1768, see Table S1)
–
the latter had a similar level of NO
2
but no NO. The concentration
of
C
x
H
y
O
z
N
_mono
formed in the Run1768 was much lower (~ 30 %) than that in the Run1752
(see Table S1), suggesting that the reaction of NO +
RO
2
is a mo
re efficient pathway of
forming
organic nitrate monomers. This means the traditional knowledge of RO
2
chemistry
still applies to these highly oxygenated RO
2
, that
NO +
RO
2
forms stable organic nitrates but
NO
2
only reacts with acylperoxy radicals forming peroxy nitrates (PANs) that are thermally
unstable
(
54
)
.
In addition, the NO
3
-
initiated oxidation had a considerable contribution, indicated by the
formation of
C
x
H
y
O
z
N
2_mono
, C
x
H
y
O
z
N
_di
, and C
x
H
y
O
z
N
2_di.
Mean
while, this pathway may also
lead to the formation of C
x
H
y
O
z
N
1_mono
,
which however,
cannot be distinguished from the
products via NO
x
+RO
2
pathways. Similar to our recent findings at the reference station
(
55
)
,
we attribute the formation of C
x
H
y
O
z
N
_di
and C
x
H
y
O
z
N
2_di
to the RO
2
+ RO
2
reactions, with
one or both parent RO
2
being produced by the NO
3
-
initiated oxidation. In addition, we also
attribute the formation of C
x
H
y
O
z
N
2_mono
, exclusively observed as C
10
H
16
O
x
N
2
,
to a
combination of NO
3
-
initiated oxidatio
n followed by a termination reaction with NO. This
formation pathway explains two features of these molecules: 1) the two nitrogen atoms are
introduced by NO
3
oxidation and NO termination, respectively, and 2) there is no hydrogen
gain or loss, as the numb
er of hydrogen atoms in these products is the same as in the
monoterpene precursors. It seems that NO
2
is not involved in the formation of C
x
H
y
O
z
N
2_mono
(Fig. S
7
), most likely because the acylperoxy radicals are rarely formed from the NO
3
-
initiated oxidati
on
(
56
)
. As previously shown, the HOM yields from the NO
3
-
initiated
oxidation of different monoterpenes may vary significantly
(
51
)
, and the formation of highly
oxygenated RO
2
is favored in NO
3
oxidizing
-
3
-
carene but disfavored in the
-
pinene case
(
56
)
.
Therefore, the NO
3
-
related HOMs in this study might mainly come from the oxidation
of
-
3
-
carene. Overall, in this chosen experiment, at 1.9 ppbv NO
x
, NO
3
-
initiated oxidation
contributes to at least about 10% (Fig. 2b, the sum of fractions of C
x
H
y
O
z
N
1
-
2_d
i
and
C
x
H
y
O
z
N
2_mono
, but disregarding the plausible fraction of C
x
H
y
O
z
N
_mono
) of the total HOMs
and about 30% of the HOM dimers (sum of C
x
H
y
O
z
N
1
-
2_di
fractions over the total dimer
fraction in Fig. 2b).
4.
Volatility estimation
As shown in the manuscript, t
he previously reported volatility parameterization
(
14
)
can be
extended to include all types of HOMs by using the O:C
eff
. This parameterization is based on
the volatility estimation of individual HOM molecule with the SIMPOL model
(
57
)
. There
also exists a few other approaches to estimate the HOM volatility, including the
EVAPORATION group contribution method
(
58, 59
)
, COSMOTherm
(
60
)
, and a method to
derive volatility from the thermogram measured with the FIGAERO inlet
(
61
)
. Significan
t
differences have been revealed in the absolute volatilities estimated between the different
models. For example, it has been suggested that the EVAPORATION and SIMPOL models
may underestimate the volatility due to the negligence of intramolecular H
-
bonds
, whereas
the COSMOTherm model seems to overestimate volatility
(
60
)
. The thermogram method, on
the other hand, gives even lower volatilities relative to the SIMPOL model, likely due to the
thermal decomposition of the particle
-
phase oligomers
(
29
)
. In the
analysis of the
thermograms in this work, we also found that thermal decomposition of HOM dimers
interfered with the T
max
of monomers.
Another important uncertainty of HOM volatility distribution comes from the use of different
parameterizations. For exa
mple, different parameterizations, both based on SIMPOL model,
were used in this work and by Stolzenburg and co
-
workers
(
41
)
, meanwhile, another
parameterization was derived by Li and co
-
workers based on the EVAPORATION model
(
61
)
. We show in Fig. S8 the c
omparison of HOM volatility distributions estimated by these
three parameterizations under both zero and 1.9 ppbv NO
x
conditions. Despite of very
different parameterizations, the HOM volatility distributions derived in this work and by
Stolzenburg and co
-
w
orkers are largely similar. On the other hand, the parameterization based
on EVAPORATION gives lower volatility by about 3
-
4 orders of magnitude than what are
predicted by the other two parameterizations. However, the fact that NO
x
can cause the HOM
volati
lity distribution shift to higher volatility can be seen from all parameterizations (Fig. S8
bottom panel). Therefore, the main finding of this work that adding NO
x
substantially shifts
the volatility to higher values is not affected by the parameterizatio
n chosen.
Supplementary figures
Fig. S1. Size
-
segregated particle growth rates suppressed by NO
x
. Similar to the bottom panel of Fig.1,
growth rates are normalized to those measured at the zero NO
x
condition, so that the ratio shows the
degree of suppression by NO
x
. The original data are shown in Table S1.