Seasonal variation in aerosol composition

Environmental
Science
Processes & Impacts
View Article Online
PAPER View Journal
Seasonal variation in aerosol composition and

Published on 24 January 2019. Downloaded by University of Winnipeg on 1/31/2019 9:45:00 AM.
Cite this: DOI: 10.1039/c8em00471d

concentration upon transport from the outdoor to
indoor environment†
ac a *ab
Anita M. Avery, Michael S. Waring and Peter F. DeCarlo
Outdoor-originated aerosols are an important component impacting indoor air quality. Since outdoor
aerosols vary over short (diurnal) and long (seasonal) timescales, we examined how the variation in
outdoor aerosol concentration and composition impact indoor aerosol. Measurements of both indoor
and outdoor aerosol composition in real time in an urban classroom in winter and summer seasons
were performed using an aerosol mass spectrometer (AMS), aethalometer, and a suite of gas phase
instruments. Factor analysis of the organic aerosol components identified three factors in common
between seasons, including hydrocarbon-like, cooking, and oxidized organic aerosol (HOA, COA, and
OOA). Since sulfate is non-volatile, we report a sulfate-normalized indoor–outdoor ratio (I/O)i/SO4 for
measured aerosol i components, allowing us to estimate aerosol component-based effects of seasonal
and other variations in ventilation and HVAC operation, indoor emission sources, and chemically-based
loss processes between outdoor and indoor environments. These chemical loss processes are
interpreted in terms of changes in temperature and relative humidity (RH) between environments,
which fluctuate on a daily and seasonal basis. The degree to which any effect is observed depends on
the particular outdoor aerosol population and the magnitude of temperature or RH change. In
wintertime, when aerosols were warmed upon transport indoors and loss of volatile components is
favored, median (I/O)i/SO4 values for nitrate, total organics, HOA, and BC were smaller (0.35, 1.00, 1.24,
and 1.18, respectively) than summertime values (0.75, 1.17, 1.96, and 1.80). For COA and OOA, however,
(I/O)i/SO4 values were higher in the winter than in summer. Calculated aerosol liquid water (ALW), which
is a function of temperature and RH and the relative contribution of hygroscopic components, varied
significantly by season. Summertime ALW indoors provides a medium for aqueous processing, which is
Received 15th October 2018
Accepted 23rd January 2019
necessary for some hydrophilic gas phase reaction products that are important to indoor air quality and
occupant exposure. This work describes the linkages between seasonal variability in aerosol
DOI: 10.1039/c8em00471d
composition outdoors and the subsequent chemically-specific variation observed when that aerosol is
rsc.li/espi brought indoors.
Environmental signicance
Outdoor aerosols are regulated by governments worldwide to reduce exposure and protect human health. In the developed world, however, people spend the vast
majority of their time indoors. Consequently, exposure to outdoor aerosols predominantly occurs in the indoor environment. This study aims to quantify the
impact of temperature and relative humidity gradients on the composition and transport of aerosol particles from the outdoor to the indoor environment in
both the summer and winter seasons when temperature gradients between indoors and outdoors are reversed. The results presented here can be used to inform
models of indoor air quality and provide insight into how outdoor aerosol is modied when transported indoors.
1. Introduction
1.1 Seasonal-dependence of outdoor-originated aerosol
High concentrations of ambient aerosols have been linked to
a
Department of Civil, Architectural, and Environmental Engineering, Drexel University, adverse respiratory and cardiovascular conditions.1 Humans
USA. E-mail: [email protected] spend most of their time indoors.2 However, for practical and
b
Department of Chemistry, Drexel University, USA legislative reasons, we monitor and regulate outdoor air quality
c
Now at Center for Aerosol and Cloud Chemistry, Aerodyne Research, Inc, USA
with improving human health as one of the aims. Seasonal-
† Electronic supplementary information (ESI) available. See DOI:
dependent uctuations in outdoor aerosol concentration and
10.1039/c8em00471d
This journal is © The Royal Society of Chemistry 2019 Environ. Sci.: Processes Impacts
View Article Online
Environmental Science: Processes & Impacts Paper
composition, and by extension, exposure to them is dependent order parameter to normalize for mechanical losses (e.g. deposi-
on meteorological conditions (e.g. natural boundary layer height tion and ltration) and demonstrate the impact of additional
variability), seasonal processes (photochemical variation with processes on the indoor aerosol population. These additional
season), and variability in source emissions (e.g. heating-specic processes include indoor emissions as well as gains or losses due
emissions in winter). However, long-term in-depth characteriza- to physiochemical transformation. Implicitly, the (I/O)i/SO4
tion of outdoor aerosols is difficult because most eld work is time assumes an internally mixed aerosol population or similar
constrained with typical sampling durations on the order of one mechanical loss rates for externally mixed populations. Externally
month. The IMPROVE network3 has investigated chemical speci- mixed aerosol populations that differ in size and composition
ation via integrated lter samples at 110 locations, some operating complicate the use of (I/O)i/SO4, but trends and interpretations are
since 1985, but are limited to rural areas and predominantly in still useful for understanding outdoor-to-indoor transport. Use of
National Parks and Wilderness Areas in the United States. This the (I/O)i/SO4 creates a basis for comparison across seasons when
limits application to exposure between the indoor and outdoor building ventilation differs, and provides a standardization for
environments. Other monitoring networks including the EPA air other regions, climates, and building operation.
quality monitoring stations provide more local level data, but are
similarly limited by time integrated lters. Recent instrument
developments like the Aerosol Chemical Speciation Monitor 1.3 Factors affecting chemical transformation: temperature,
(ACSM),4 which can sample independently and continuously for humidity, liquid water, and hygroscopic components
weeks to months, has made long-term inter-seasonal and inter- In addition to mechanical loss processes (e.g. ltration and
annual measurements more common in urban areas.5,6 deposition) affecting the I/O ratio, chemical transformation of
More than a decade of aerosol mass spectrometer (AMS) aerosol upon transport from outdoors to indoors is governed by
measurements, providing minute-to-hour integrated real-time chemical-specic physiochemical properties, such as volatility.
submicron aerosol composition,7,8 across the world have resul- For a mechanically ventilated building, at positive temperature
ted in several chemically-specic inter-seasonal analyses of mega (T) gradients (in>out, as during midlatitude winter), the heat-
cities.9–13 Despite continental, latitudinal, annual, and intra-city ing, ventilation, and air conditioning (HVAC) systems operating
differences, trends in aerosol composition and chemical pro- to heat outdoor air can drive semi-volatile components from the
cessing have emerged, contributing greatly to our understanding condensed phase to the gas phase and decrease the (I/O)i/SO4; at
of tropospheric outdoor aerosols. By bringing the AMS negative gradients (in < out, as during midlatitude summer),
measurement methods to the indoor environment, we provide condensation of semi-volatile species from the gas phase to the
new insights to aerosol exposure indoors, utilizing the knowl- particle phase can increase the (I/O)i/SO4. The positive T gradient
edge gained from outdoor sampling with the AMS.14,15 This work scenario has been studied extensively outdoors using a ther-
provides a direct comparative analysis for outdoor aerosol modenuder coupled with an AMS;24 however, the T gradients in
species transported to the indoor environment and the inuence these studies are oen much greater than for outdoor-to-indoor
of sources and processes occurring in the indoor environment. environments and lack high T resolution in temperature at low
DT. Thermodenuder work has examined some of the complexity
of organic aerosol (OA) from various sources25 or positive matrix
1.2 Indoor–outdoor ratio factorization (PMF) result types.26 Temperature gradients
Chemical species from the outdoors make their way indoors via between indoors and outdoors on organics using the volatility
several important mechanisms (e.g. ventilation, inltration),16,17 basis set (VBS)27 on relationship of indoor, outdoor and
and emissions in each environment and processing between personal air (RIOPA) study data28,29 highlighted the geo-specic
them are important parameters of exposure. For environments relationship between outdoor aerosol volatility and indoor
with minimal indoor sources, the inuence of outdoor aerosol aerosol observations. However, the negative T gradient case (in
on the indoor environment can be succinctly described with the < out, summer) has not been experimentally investigated.
indoor/outdoor (I/O) ratio:18 the concentration of indoor aerosol While relative humidity (RH) and T are linked, the role of
normalized by that of outdoor. For environments with indoor humidity in contributing to aerosol liquid water (ALW) uptake
sources including cooking,19 cleaning,20,21 smoking,22 or other from hygroscopic components and subsequent re-partitioning
emission sources, the I/O ratio can exceed unity, but trends of of water soluble aerosol components with the gas phase is an
the I/O ratio over time provide insights into the sources of important consideration in I/O ratios.30 ALW mass globally is
aerosols to the indoor environment. The I/O ratio has been approximately twice that of dry aerosol,31 indicating the ubiq-
widely used to describe penetration and inltration across uitous availability of an aqueous phase for processing of both
building types, environmental conditions, and outdoor emis- organics and inorganics,32 and making organic aerosol depen-
sion types,23 but without the ability to compare directly across dent on temperature, humidity,33,34 and inorganic hygroscopic
these important operating conditions, and has been used with components.35 Recent eld studies, especially the southern
limited discussion as to the diverse characteristics of the oxidant and aerosol study (SOAS),36 have highlighted the
outdoor-originated aerosols. importance of humidity and aqueous-phase processing in high-
Johnson et al.14 described a sulfate-normalized I/O ratio as the humidity environments. Long-term measurements in Beijing
I/O ratio of any aerosol chemical component i divided by the I/O also showed effects seasonal-dependence of aqueous process-
ratio of non-volatile sulfate, (I/O)i/(I/O)SO4 ¼ (I/O)i/SO4, as a rst ing with humidity across oxidized PMF factors.37
Environ. Sci.: Processes Impacts This journal is © The Royal Society of Chemistry 2019
View Article Online
Paper Environmental Science: Processes & Impacts
The impact of hygroscopic growth or contraction of aerosols, laser off as a high resolution time of ight (HR-ToF) AMS.7 The
dependent on chemical composition and changes in humidity AMS measured bulk and size-resolved chemical composition of
across a building envelope can be analogized to hygroscopicity- non-refractory, submicron particles including organics, nitrate,
tandem differential mobility analyzer (HTDMA) work, which sulfate, ammonium, and chloride. Data was processed using
has explored the expansion of aerosol size distribution with standard AMS soware packages including high resolution
humidication, dependent on aerosol size, hygroscopic compo- analysis. A relative ionization efficiency of 1.4 was assumed for
nents, and humidity.38–41 Further combination of analytical all organic species. A collection efficiency of 0.5 was applied as
methods in a volatility-hygroscopicity tandem DMA (VH-TDMA)42 less than 8% of the winter outdoor data, and none of the indoor
have shown the hygroscopicity of volatile compounds to be high, winter or either environment in summer met the criteria for
in agreement with growth factor measurements and Zdanovskii, a higher efficiency.43 We discussed whether to apply this and are
Stokes, and Robinson (ZSR) predictions. Growth factor calcula- concerned that modifying the outdoor CE, but not the indoor
tions based on hygroscopic components and humidity link these CE may potentially introduce artifacts in the data as well.
properties with ALW and size distribution outcomes. In the Chloride was used in the ALW calculations, but is otherwise
indoor environment, all of these properties are expressed in their excluded from analysis here. Ammonium was also excluded
(I/O)i/SO4, dependent on the environmental differences and pro- from I/O analysis since it is a function of sulfate and nitrate, and
cessing between environments. trends in its I/O ratio are due to variations in sulfate and nitrate.
Predicted versus measured ammonium analysis, describing the
aerosol acid/base neutralization state,44 can be found in
2. Methods Fig. S1.† This analysis predicts the required ammonium
2.1 Sampling methods concentration to fully neutralize the measured sulfate, nitrate,
Real-time, near simultaneous measurements with the AMS and and chloride, and compares it to the measured ammonium by
other instrumentation were taken on outdoor air and the indoor the AMS. CO2 measured by the Picarro CRDS (described below)
air in a classroom. Sampling took place in winter (Jan 31-Mar 2) was used for mass spectra correction at m/z 44 in the frag-
and summer (July 13-Aug 13) 2016, at Drexel University in mentation table.45 Elemental ratios of C/H/O/N PMF factors
Philadelphia. Outdoor measurements sampled air including were calculated from high resolution mass spectra.46,47
emissions from regional and local sources including the nearby 2.2.2 Aethalometer. The AE-33 (Magee Scientic) aethal-
I-76 freeway, the major transportation hub of 30th Street Station, ometer48 measured light absorption at seven wavelengths.
and Center City Philadelphia. Nearby, a row of food trucks However, only one wavelength (880 nm) was used for this
served the university on weekdays (7 AM–6 PM) with limited analysis. Mass absorption coefficients and other calibrations,
service until lunch on Saturday and no service on Sunday. The including automatic loading artifact correction, were provided
outdoor inlet was a dedicated inlet from the roof directly above by the manufacturer.
the laboratory, next to the air intake for the HVAC zone serving 2.2.3 Picarro CRDS gas analyzer. The G2401 analyzer for
the classroom where indoor air was sampled. CO2, CO, CH4, H2O (Picarro, Inc.) used cavity ringdown spec-
The classroom was 148 m3, with tiled oors and painted troscopy (CRDS) to measure given gases at 0.25 Hz. It was cali-
cinder block walls, a chalkboard and mobile desk seating for brated with a calibration tank of CO2, CO, and CH4, was used to
about 25 students. There were doors on each of two adjacent calibrate the instrument at 3 dilution levels each week of
walls, but one door led to another recessed room, so no direct operation.
air ow currents between the two doors were expected. The 2.2.4 Other measurements. Indoor T and RH were
adjacent hallways were not mechanically ventilated. The room measured with an Elgato Eve indoor sensor system. A Vaisala
was occupied regularly for classes only during the winter Automatic Weather Station 310 (AWS310) was located outdoors
season, but the number of occupants and occupant activities on Drexel campus for outdoor meteorological data, including T
were not recorded. The human contribution to the indoor and RH, wind speed (WS), wind direction (WD), pressure (P),
aerosol composition during the wintertime measurements will solar radiation, and precipitation.
be discussed in detail in a future publication. The HVAC zone
included adjacent faculty and student offices but no other 2.3 Lag time and air exchange rates
classrooms. Sampling for each inlet was controlled by a custom
valve-switching device to alternate sampling for all instruments Sulfate measurements were used to determine and time-shi
between indoor and outdoor inlets every 4 minutes; a bypass for the time delay (or “lag”) for the changes in outdoor-
line was utilized to ensure continuous ow at a constant rate in originated aerosols to be reected in the indoor aerosol
each inlet. Sample air was dried to below 30% RH using concentration. This shi value for each dataset captured
a Naon Dryer (MD-700 series, Perma Pure Inc.) prior to differences in building air ows in different seasons. This time
measurement by the aerosol instrumentation. was determined with a time-shiing regression analysis, which
has previously been described in detail.14 Briey, the peak
correlation between outdoor and indoor sulfate concentration
2.2 Instrumentation across several inlet cycles was calculated for all continuous 24
2.2.1 Soot-particle aerosol mass spectrometer (SP-AMS). hour segments for each dataset. The aggregate of these peak
The SP-AMS (Aerodyne, Inc.) used here was operated with the correlation values resulted in a distribution of shi values, the
View Article Online
peak of which was chosen as the single lag value for the entire each season, no detection cutoff was used for separating
dataset. All aerosol I/O ratios presented here use the lag-time organic from inorganic, as suggested in Kiendler-Scharr
correction. These were different between seasons, with indoor (2016).50
data reecting a 29 minute lag for winter and 37 minute lag for
summer. However, the effect of this analysis on the ratios 2.5 Positive matrix factorization
described here was minimal, as shown in Table S1.†
Positive matrix factorization (PMF) was used to analyze the
Of note, the aerosol lag shi time in this room with venti-
organic aerosol (OA) in each season. PMF51 is a receptor model
lation and recirculation air exchange, and subject to ltration,
that mathematically decomposes a total signal into a linear
is distinct from air exchange rate as measured by gas phase
combination of factors. This technique can identify the relative
species. The ventilation rate is dened as the rate of exchange of

contributions of OA types for AMS data.52,53 Determination of
indoor air with outdoor air, and is distinct from the room air
the nal PMF result for each season included comparisons
exchange rate (AER) which includes recirculation from other
between seasons and with published spectra, which will be
rooms within the HVAC zone, in addition to the ventilation rate.
discussed in the results section. For each season, PMF was
The ventilation rate was calculated from the discretized solu-
applied to the organic mass spectral matrix using the data from
tion of a simple CH4 model using measured indoor and outdoor
both indoor and outdoor inlets collectively. Additionally, indoor
concentrations, as described in eqn (1):18
only and outdoor only datasets were analyzed separately to
dC conrm the consistency of the PMF solutions between the
¼ lCout lC (1)
dt individual (indoor-only and outdoor-only) and the combined
where the change in time (t) of indoor concentration (C) of CH4 dataset. We present only the results of the combined dataset
is described by the air exchange rate (l) and the outdoor here.
concentration (Cout).
The room AER was calculated from the average decay of CO2 2.6 Aerosol liquid water from Köhler theory
aer releases (performed multiple times per season). In each The aerosol liquid water content in each environment was
season, the calculated lag time (converted to an air exchange calculated as a function of relative humidity and their chemical
rate) fell between ventilation rate and room AER. In winter, the composition. Petters and Kreidenweis54 proposed a single value
ventilation and room AERs were 0.39 and 3.5 h1 and in to account for the hygroscopic properties of a chemical species
summer, 0.17 and 4.1 h1. Seasonal differences are reective of using k, dened by the relationship between the volume of the
operational changes in the HVAC system. dry particle VS, the volume of water VW, and the water activity aw
as:

2.4 Organic and inorganic nitrate aerosol VW 1
ki ¼ 1 (3)
VS aw
Nitrate has been separated into contributions from inorganic
nitrate (iNO3) and organic-nitrate (oNO3) based on the ratio of assuming the Zdanovskii–Stokes–Robinson (ZSR) rule of mixing
high resolution NO2+ to NO+ ions.49,50 Measured nitrate frag- applies, a multi-component mixture kmix can be described as
ments NO+, and NO2+ can originate from organic (i.e. RONO2) or the sum of the k of individual components, i, weighted by the
inorganic (i.e. NH4NO3) molecules. The ratio R of NO2+ to NO+ volume fraction (3) of that component:
fragments used in eqn (2) are dened as follows: X
kmix ¼ 3i ki (4)
2.4.1 RoNO3. Organic nitrates, was given a value of 0.1, i
based on published measurements of organic nitrates.49
2.4.2 Rcalib. For inorganic nitrate, is calculated as a static
For use with AMS data, the form of the measured species (i.e.
value from ammonium nitrate calibrations. The value varies in
SO4 as H2SO4 or (NH4)2SO4, etc.) was based on a pairing scheme
each AMS instrument and its history. As a result, the value of
for inorganics55 and the fraction of signal at m/z 44 (f44) for
Rcalib was slightly different between winter (0.79) and summer
organics.40
(0.70).
The growth factor (GF) at any relative humidity is described
2.4.3 Rmeasured. The time-varying ratio of NO2+ to NO+ ion
by the ratio of the diameter of a wet particle at that relative
fragments measured in the AMS during sampling.
humidity, DRH, and dry diameter D0, or GF ¼ DRH/D0. Rear-
Using these values, the fraction of organic nitrate signal of
ranging and converting the diameters to spherical volumes and
the total nitrate signal can be calculated as using the formula:50
using eqn (3) and (4), the growth factor of a multicomponent
1 þ RoNO3 ðRmeasured Rcalib Þ mixture at a given RH (here aw used as equivalent to fractional
oNO3;frac ¼ (2)
ð1 þ Rmeasured Þ RoNO3 Rcalib RH), is calculated by:
!1=3
X 3
The fraction of inorganic NO3 is therefore the difference GFmix ¼ 3i ðGFi Þ (5)
i
between 1 and oNO3,frac, and the mass of iNO3 and oNO3 is the
fraction multiplied by the total nitrate signal. While the detec- and the aerosol liquid water (ALW) of an aerosol at that RH can
tion limits of individual fragments were veried throughout be calculated by:
View Article Online
ALW ¼ VS(GFmix3 1) (6) indoor RH is expected from thermally well-controlled environ-

ments, including classrooms and offices. Comparisons of
With a known aerosol composition via the AMS and aethal- environmental conditions between indoors and outdoors as
ometer, the dry volume can be calculated. Of note, this calcu- shown in Fig. 1 indicate that the HVAC system in both winter
lation does not explicitly require a known aerosol diameter or and summer was largely independent of weekday/weekend
size distribution, and does not take into account hysteresis variations, and mostly independent of outdoor conditions
effects which could be important in this dataset, since T and RH with the exception of the very cold outdoor period in winter
extremes in each season will occur in the HVAC system. We do lowering indoor T by several degrees.
not have measurements of T and RH in the HVAC, and calcu- To understand the effects of both environments, the T and
lations of aerosol liquid water therefore reect the conditions of RH gradients between outdoors and indoors were used to
the indoor classroom or outdoors. investigate compositional changes between environments. The
temperature gradient is dened as (indoor T outdoor T), and
3. Results and discussion the RH gradient dened as (outdoor RH indoor RH). These
were dened so that for both gradients, positive values indicate
3.1 Meteorological conditions
conditions that promote volatilization or thermodynamic
Fig. 1 shows the measured indoor and outdoor temperatures for aerosol losses, and the negative values are conditions that favor
the two measurement campaigns discussed here. Seasonal condensation or thermodynamic aerosol addition. This
differences in Philadelphia are clearly observed in temperature gradient difference was not prescribed as in thermodenuder
differences and variability (e.g. Taverage standard deviation). experiments, but a result of natural outdoor variations and
Outdoor T was 27 3 C during summer and 4 6 C during HVAC control of indoor conditions. The two-dimensional T and
winter. Extremes in winter temperatures were coincident with RH gradients for both winter and summer measurements are
storms with a low of 14 to a high of +18 C. Summer displayed in Fig. 2. The grid areas in Fig. 2 are colored by the
temperatures uctuated almost exclusively in diurnal patterns. number of measurements corresponding to that particular
RH varied over a similar range in each season, summer 62 combination of T and RH gradients. Together, the two seasons
15% and winter 56 17%, but again diurnally dependent in span RH gradients from 18 to +62%, with signicant overlap
summer with winter RH driven by larger scale meteorological between seasons. The T gradients span from 18 to +29 C and
inuences. The winter season experienced episodes of higher do not overlap between seasons. This set provides interesting
wind speeds (>6 m s1) corresponding to winter storms that analytical continuity in humidity, but with seasonal-based
brought snow while summer thunderstorms were not associ- distinctions in temperature. An important note is the largest
ated with high wind. Occasional summer thunderstorms T gradient (+25 C) corresponds to the lowest T conditions in
brought changes in T and RH but not in measured aerosol both environments (10 C outdoors, 15 C indoors).
species. Full meteorological data can be found in Fig. S2.† Additionally, we note that the indoor temperature in summer
Indoor T in both seasons varied very little: virtually was 4 C cooler than the average indoor temperature in winter.
unchanging (18 1 C) in summer, and in winter (22 3 C),
somewhat inuenced by a period of extreme cold (below 10 C
outdoors corresponding to the indoor minimum) with 3.2 PMF results
measurable warming while occupied. The RH, however, varied PMF results from both seasons identied three organic aerosol
greatly during the winter (26 9%), roughly following (OA) factors in common: oxygenated (OOA), hydrocarbon-like
temperature uctuations outdoors, and was constant in (HOA), and cooking (COA) aerosol. In summer, an additional
summer (55 2%), regardless of outdoor RH. This constant factor contained signicant contribution of reduced-nitrogen
Fig. 1 Measured outdoor and indoor temperature (red lines) and relative humidity (blue lines) in winter and summer seasons.
View Article Online
correlates very well with previously published wintertime

spectra (R2 ¼ 0.95),61 but inter-seasonal correlation is the lowest
of the three at R2 ¼ 0.79. Identication of COA in the
summertime was more dependent on the overall fragment
signature of mixed hydrocarbon and oxidized series above m/z
50 and a diurnal pattern corresponding to the nearby food
vendors. Summertime O/C 0.21 is high (and H/C 1.41 low) for
COA, and may be a result of some spectral mixing with the OOA
factor. In general, O/C ratios for all factors were higher in
summer than winter.

Fig. 2 Heat map showing the temperature and relative humidity
differences between the indoor and outdoor environment. Pixel color 3.3 Seasonal trends in outdoor aerosol composition
shows the number of observations of a specific temperature and RH
difference, white color indicates no observation. During winter, Total submicron outdoor aerosol concentrations were similar
temperature and RH differences were all greater than 0, but during between seasons with an average of 6.96 mg m3 in winter and
summer, the temperature gradient was negative and the humidity 9.21 mg m3 in summer, and ranged from 2.58 to 25.9 in the
gradient varied. However, the number of points at each temperature
summer, and 0.63 to 24.7 mg m3 in the winter. Compositional
and humidity gradient varied greatly, even sometimes for gradients
near each other. differences were notable between seasons. Outdoor organics,
ammonium sulfate, and black carbon were slightly higher in
summer, while ammonium and nitrate were much higher in
species linked to third hand smoke (THS). This factor was found winter. Average concentrations for all species can be found in
to be predominantly indoors (median I/O ¼ 7.6) and has been Table 1. Summer was characterized by consistent and stable
discussed separately.15 Full mass spectra for each factor and high organic aerosol, comprising 74% of AMS-measured
each season are shown in the SI Fig. S3.† While the PMF results species. In wintertime, that fraction was reduced to 52%,
here are common between seasons, each season was investi- closer to other urban areas in humid continental climate
gated and the appropriate PMF solution was determined sepa- zones.11–13,62 Winter was characterized by storm meteorology-
rately. Therefore, there are some seasonal differences between driven highs and lows, as well as short duration spikes in
them. Diurnal variations discussed in the next section further organic aerosol concentration, likely from local sources.
support the factor identication for each season. Correlation Complete time series of the outdoor concentrations in this work
with published spectra was generally higher for wintertime are displayed in Fig. 3. Stagnation and high RH in winter
factors than summertime, so for convenience, the correlation produced prominent peaks in ammonium nitrate, with
values listed here are only the lower of the two seasons. concentrations exceeding the organic aerosol concentration.
In both seasons, OOA was more similar to low volatility High nitrate (20 to 30%) is common in wintertime traffic-
oxygenated OA (LV-OOA, R2 > 0.9) than semi-volatile oxygenated dominated cities61 due to its semi-volatile tendency to prefer-
OA (SV-OOA, R2 > 0.72)56,57 published factors. However, between entially partition to the particle phase at ambient winter
less-oxidized oxygenated OA (LO-OOA) and more-oxidized temperatures, but to the gas phase at summertime tempera-
oxygenated OA (MO-OOA), the OOA correlated slightly better tures. In summer, overnight (10 PM to 2 AM) spikes in sulfate
with LO-OOA (R2 > 0.92) than MO-OOA (R2 > 0.89).58 Between
winter and summer, the two OOA factors correlated very well (R2
Table 1 For each season and aerosol component, the campaign-
¼ 0.97). However, wintertime OOA was less oxidized and
average concentration, and quartile statistics for (I/O)i/SO4 as visualized
exhibited a lower O/C ratio and higher H/C ratio (winter O/C in Fig. 6
0.43 and H/C 1.31; summer O/C 0.59 and H/C 1.23).
HOA was characterized by prominent mass fragments at Winter Summer Percentiles for
unsaturated and saturated hydrocarbon chain pairs m/z 41 and conc. conc. Percentiles for (I/O)i/SO4
(mg m3) (mg m3) (I/O)i/SO4 winter summer
43 (C3H5 and C3H7), and m/z 55 and 57 (C4H7 and C4H9), which Species or
are representative of lubricating oils and incomplete combus- factor Out In Out In 25th 50th 75th 25th 50th 75th
tion and.59,60 HOA was highly consistent between seasons (R2 ¼
0.93) and well correlated with other urban HOA spectra (R2 > SO4 0.86 0.28 1.40 0.43 — — — — — —
NH4 0.63 0.07 0.44 0.05 — — — — — —
0.90).56,58,61 The ratios of H/C 1.83 and 1.84, O/C 0.1 and 0.05 for
NO3 1.35 0.13 0.23 0.04 0.24 0.35 0.49 0.59 0.75 0.93
summer and winter respectively further indicate the robustness iNO3 1.11 0.09 0.11 0.01 0.17 0.28 0.41 0.43 0.65 0.91
of this factor mass spectrum across seasons. oNO3 0.24 0.04 0.11 0.03 0.43 0.57 0.72 0.64 0.79 0.96
COA was characterized by both oxidized and unoxidized Org 3.24 1.14 5.98 2.22 0.88 1.00 1.17 1.07 1.17 1.29
fragments at m/z 43 (C2H3O and C3H7) and m/z 55 (C3H3O and HOA 0.69 0.29 0.46 0.26 0.94 1.24 1.77 1.36 1.96 2.89
C4H7), due to aliphatic acids from cooking oils and meat, and COA 0.69 0.33 1.69 0.43 1.00 1.24 1.70 0.66 0.75 0.88
OOA 1.85 0.53 3.72 0.63 0.71 0.82 0.91 0.41 0.51 0.59
was somewhat different between seasons, due to signicant BC 0.62 0.25 0.74 0.41 0.93 1.18 1.62 1.29 1.80 2.80
contribution from more oxidized fragment components C2H3O+ THS — — 0.14 0.88 — — — — — —
and CO2+ in summer. Wintertime COA (O/C 0.11, H/C 1.61) Total 6.96 1.94 9.21 3.25 — — — — — —
View Article Online

Fig. 3 Full stacked time series of AMS and aethalometer data for outdoor and indoor in winter (top pair) and summer (bottom pair).
and corresponding ammonium from an unidentied source the largest inorganic contributor to aerosol mass. During
were common. episodic buildup of aerosol mass, some peaks exceeded 4 mg
3.3.1 Outdoor aerosol species trends and diurnal patterns. m3 which were 3 the study average (see Fig. 3). The diurnal
Summary average diurnal patterns of all measured aerosol pattern of nitrate in the winter (Fig. 4c) shows the highest
species can be found in Fig. 4. Outdoor sulfate concentrations concentrations at night, with the minimum during early aer-
(panels a, b) showed little variation in diurnal pattern and noon hours. In winter, only 15% of total nitrate is organic
similar concentrations across seasons near 1 mg m3 indicative nitrate. The organic nitrate diurnal pattern is different from
of regional (non-local) steady emissions. Ammonium is associ- inorganic nitrate, showing a slight increase in concentration
ated with both semi-volatile nitrate and non-volatile sulfate, around 7 AM, and peak concentration during the daytime
and its diurnal pattern reects that combined effect. hours. In the summertime dataset, total nitrate is a weak
Aerosol nitrate shows very different seasonal trends. In the contributor to aerosol mass with a summertime average
winter dataset, aerosol nitrate is predominately inorganic and is concentration at only 11% of the average winter concentration
View Article Online

Fig. 4Median diurnal patterns of each measured and calculated aerosol species in winter (left) and summer (right) as outdoor (solid) and indoor
(dashed) lines. Sulfate, ammonium, and aerosol liquid water (a, b), total nitrate and its organic and inorganic fractions (c, d), total organics and the
PMF components (e, f), and traffic-related components (g, h) of BC, CO, and HOA. Summertime nitrate (d) is magnified by 5 for visual
enhancement, and the PMF factor THS was only observed in summer (f) and not winter (e).
of nitrate. Most of this seasonal difference comes from much outdoors and 0.53 0.24 indoors. This mass concentration is
lower inorganic nitrate concentrations, which in summer is similar in concentration to all of the inorganic species
only 7% of that in winter, while summer organic nitrate is combined. The ALW concentration follows the diurnal trends in
a more comparable 37% of winter organic nitrate concentra- RH and nitrate as described above, but is much more consistent
tion. This observation is consistent with inorganic nitrate being across seasons. The implications of indoor ALW are discussed
more volatile as compared to organic nitrate, and consequently further in Section 3.4.
less likely to be in the aerosol phase. In summer, organic and COA diurnal patterns (Fig. 4e and f) show different patterns
inorganic contributions to total nitrate signal are roughly equal. by season. Both summer and winter show COA concentrations
Aerosol liquid water (ALW) in winter has an average outdoor that start to increase around 9 AM and reach a maximum
and indoor concentration (standard deviation) of 2.6 3.9 mg around lunch time. This trend is likely strongly inuenced by
m3 outdoors and only 0.11 0.06 mg m3 indoors. In summer, the nearby food truck emissions as seen previously.14 Food
the decrease in concentrations upon transport indoors is much trucks end service around 6–7 PM, and in the winter dataset the
smaller, and similar to aerosol species at 2.7 2.5 mg m3 decrease in COA diurnal concentration follows that trend. The
View Article Online
summertime diurnal pattern demonstrates a decrease in COA inorganic nitrate and RH, suggesting possible dependency on
concentration aer lunch followed by a second peak in COA an aqueous phase, while the opposite is observed in summer.
concentration during evening hours (6 to 8 PM). This second Black carbon, CO, and HOA (panels g, h) exhibit diurnal
peak in summertime COA may be indicative of inuence from patterns consistent with traffic emissions in each season, with
additional cooking sources in the larger area around the CO and HOA having the most similar pattern. In contrast with
sampling site. Outdoor cooking and grilling is common CO and HOA, concentrations of BC were higher in summer than
throughout the city and much more common in the summer- winter, resulting in a higher HOA/BC ratio during the colder
time than other seasons. winter measurements as compared to the summer. HOA/BC
Seasonal differences in the OOA diurnal cycle (Fig. 4e and f) ratios have been discussed at length previously65 and will be
are clearly apparent. Summer OOA has a diurnal pattern which compared between indoors and outdoors below.
peaks during the early aernoon similar to ozone indicating 3.3.2 Indoor aerosol species trends and diurnal patterns.
photochemical production as has oen been seen.63,64 However, Indoor aerosol concentrations (dashed lines in Fig. 4) in both
the wintertime OOA diurnal pattern decreases during the day, seasons are characterized by low nitrate, ammonium, COA,
likely due to boundary layer dilution (see Fig. 5 for meteoro- HOA, and ALW. All measured aerosol components follow the
logical and gas-phase indicators of boundary layer). This temporal trends of their outdoor counterparts, albeit at lower
decrease in concentration also follows trends in aerosol concentrations (the one exception is the third hand smoke
Fig. 5 Average diurnal patterns of meteorological data and measured gas-phase species in winter (left) and summer (right). Temperature and
humidity (a, b), solar radiation and ozone (c, d), methane and CO (e, f), CO2 (g, h). With the exception of CO2 (indoor on left axis and outdoor on
right axis), outdoor are shown in solid lines and indoor in dashed lines.
View Article Online
factor (THS) which described in depth elsewhere15). Nitrate is inuenced by a small unidentied indoor source. Peaks in
especially low, even when it is high outdoors in winter, and outdoor CO and CH4 linger indoors long aer the outdoor
barely contributes indoors with average concentrations of concentrations return to background values, especially in
organic and inorganic nitrate < 0.1 mg m3 in both seasons. In summer when the ventilation rate was lower. This effect is
addition to the lower concentrations observed indoors, each observed in Fig. S5.† The CO2 concentration indoors uctuated
aerosol component shows a smoothed timeseries, and lags during weekdays due to occupants in both in the classroom
behind outdoor concentration uctuations due to ventilation itself and the rooms on the same HVAC zone. In summer, when
rate and mixing considerations. Importantly, some components the room (but not the HVAC zone) was generally unoccupied (no
like sulfate are similar across seasons. OOA in each season does scheduled classes), the peak CO2 reached 872 ppm, compared
vary with time of day, but less dramatically than outdoors. Black with wintertime concentration peaking at 1110 ppm when
carbon consistently follows outdoor levels in both seasons as scheduled classes were common. This seasonal difference is
expected for a non-volatile, outdoor primary aerosol source. also observed in the daily patterns of CO2 concentrations.
Indoor HOA in summer did not track the outdoor HOA well Summertime CO2 followed a smooth pattern of gradual rise in
during the daytime similar to previous observations.14 HOA/BC the morning and fall in the evening due to no direct emissions
values in both seasons show an increase indoors compared to in the classroom and the well mixed recirculated air smoothing
out. In the winter measurements, the HOA/BC value increased the impact of occupancy throughout the HVAC zone. However,
from 1.12 (R2 ¼ 0.66) outdoors to 1.46 (R2 ¼ 0.53) indoors. in winter, the direct inuence of classroom occupant emissions
Summertime HOA/BC values increased from 0.52 (R2 ¼ 0.33) is visible with CO2 concentrations showing steep increases and
outdoors to 1.54 (R2 ¼ 0.14) indoors, although the correlation at decreases multiple times in a single day, and the impact of
low loading was poor. Since BC is a non-volatile tracer and evening classes as an additional bump in the winter CO2
generally associated with diesel emissions in the absence of diurnal. Ozone concentration is nearly always close to detection
strong biomass signal, and HOA is a volatile species,24,26 varia- limit of 1 ppb indoors, due to indoor deposition and reaction
tions in the HOA/BC ratio are indicative of mass addition of mechanisms that have been widely described previously.66–68
chemical species similar to HOA. The outdoor HOA/BC ratio is That said, the summer indoor concentration does show
inversely correlated with outdoor temperature in both seasons increases with outdoor photochemically produced O3 in mid-
(R2 ¼ 0.24). These relationships are shown in more detail in aernoon. Products of ozone reactions, especially with occu-
Fig. S4.† pants, will be examined in future work.
3.3.3 Outdoor gas-phase trends and diurnal patterns.
Measured gas phases species showed seasonally-dependent
trends. Meteorological and gas-phase diurnal average patterns 3.4 SO4 normalized I/O ratios
are shown in Fig. 5. All measured gas-phase species (CO2, CO, The sulfate-normalized indoor/outdoor ratio, denoted herein as
CH4, O3) demonstrated a similar pattern between seasons, but (I/O)i/SO4 (see Section 1.2), of each species and PMF factor from
diurnal variations are less pronounced, and concentrations are AMS and aethalometer data is displayed in Fig. S6,† and
generally higher in winter than summer. Measured CH4 summarized in Fig. 6 and Table 1. Nitrate and the OOA factor all
concentrations were slightly higher in the wintertime (panels e, have median (I/O)i/SO4 ratios less than unity, indicating an
f). Summertime CH4 diurnal trends showed a more pronounced additional loss mechanism compared to sulfate in both
cycle, 0.05 ppm lower at the daytime minimum than in seasons. In contrast, black carbon, HOA, and COA have median
wintertime. CO was elevated during traffic rush hours along (I/O)i/SO4 ratios greater than unity, indicating either less loss
with enhanced concentrations of BC and HOA, as shown in than sulfate from ltration and deposition, or an additional
Fig. 4. CO2 followed the combined trends CH4 and traffic- indoor source. Inter-seasonal trends reveal more about the
associated CO. While the diurnal pattern of ozone in the behavior of each species.
summer is more pronounced, the peak ozone values were Observed trends in nitrate are consistent with its semi-
similar between seasons. However, as Fig. 5 shows, ozone volatile nature. The heating of buildings well over the outdoor
outdoors in summer rose and fell in specically diurnal temperatures in winter drives more nitrate components into the
patterns, while ozone in winter was relatively constant and gas phase than in the summer when the outdoor-to-indoor
elevated (except for times of CO, CO2, and CH4 plumes). temperature gradient is reversed. In both seasons, (I/O)NO3/SO4
Concurrent plumes of CH4, CO, and CO2 exceeding 0.5 ppm CO is less than unity, with median value of 0.35 in the winter and
and 2.6 ppm CH4, coinciding with negligible ozone, lasting 12 0.75 in the summer, indicating no source indoors or conden-
hours occurred in winter. A full time series of these species can sation of nitrate components from the gas phase. Even with
be found in Fig. S5.† cooler indoor temperatures compared to outdoors in the
3.3.4 Indoor gas-phase trends and diurnal patterns. summer, nitrate still shows losses in addition to those attrib-
Indoors, concentrations of CO and CH4 indicate a little to no utable to mechanical loss by ltration and deposition. The
contribution from indoor sources, and are predominantly (I/O)i/SO4 of organic (oNO3) and inorganic (iNO3) nitrate in each
controlled by air exchange with the outdoors. The diurnal season follows a similar trend with higher (I/O)i/SO4 ratios in the
pattern of CH4 (Fig. 5) indicates that the average CH4 is higher summer than winter (seasonal differences in NO2+ to NO+ ratio
indoors than outdoors, which is the result of low ventilation are shown in Fig. S7†). Fig. 6 illustrates that overall oNO3 have
rate (0.4 h1 in winter and 0.17 h1 summer) and/or potentially consistently higher (I/O)i/SO4 ratios than iNO3 in both seasons.
View Article Online

Fig. 6 The sulfate-normalized I/O ratio ((I/O)i/SO4) for each chemical component in each season. Values less than one indicate additional losses
such as volatile losses, while greater than one indicates fewer losses or an indoor source.
This trend is likely reective of volatility differences between effective size of accumulation mode particles in the summer-
oNO3 and iNO3. Note that the normalized I/O ratios for iNO3 in time, and subsequent higher ltration efficiency may explain
summer span a very wide range due, in part, to its low the higher (I/O)i/SO4 values for BC and HOA in that season.
concentrations and associated error when taking ratios of low Fig. S8† shows that the size dependent ltration efficiency of m/
concentrations. z 57 (a proxy for traffic particles) has a more similar indoor/
Both HOA and BC show signicantly higher concentrations outdoor ratio than sulfate at smaller sizes and a higher ratio
outdoors than indoors, and their indoor concentrations are than sulfate at larger sizes.70,71 This is more dramatic for the
likely due to outdoor sources in this classroom environment. summer data than for winter, consistent with HVAC-associated
Black carbon and HOA trend together and have similar (I/O)i/SO4 losses of hygroscopic components. However, the I/O ratios of
distributions in each season (BC and HOA median 1.18 and hygroscopic sulfate are comparable (0.35 in winter and 0.31 in
1.24, respectively, in winter, and 1.80 and 1.96, respectively, in summer).
summer). Their (I/O)i/SO4 ratios are typically above unity, and A shi in the proportional contribution of different sources
vary similarly with a large spread across a wide range of values, or age of BC between seasons could also contribute to this
indicating similar sources. In contrast with sulfate, which has seasonal trend. Given the differences in slope and regression of
a at diurnal pattern indicative of a regional source, HOA and the HOA/BC ratio between seasons (see Fig. S4†), there are likely
BC are strongly linked to traffic emissions with strong local and some differences in emission sources between seasons. Addi-
diurnal character. Their (I/O)i/SO4 ratios are also typically above tionally, wintertime build-ups of aerosol mass could have aged
unity, potentially because black carbon and HOA may have less BC and made it more internally mixed with regional aerosol.
indoor mechanical losses than sulfate aerosol, and the (I/O)i/SO4 It is also important to note here that, previous work by
ratio implicitly assumes internal mixing and associated similar Johnson et al. (2017)14 found an enhancement of HOA indoors
mechanical losses indoors. without a corresponding increase of BC. Using the same
It is likely that external mixing of aerosol populations may be procedure to calculate BC-associated HOA indoors using the
the reason for this trend, since on short timescales when mixing observed outdoor ratio shows that for this study most of the
freshly emitted traffic emissions into the regional plume there HOA indoors can be linked to BC and therefore due to outdoor-
will be external mixing of aerosol populations (e.g. Rissler to-indoor transport (see Fig. S4†). While measurements for the
et al.69). Since the (I/O)i/SO4 ratio implicitly assumes internal previous study were done in the same building, they were per-
mixing and similar mechanical loss, and this is not valid for formed in a separate part of the building with a different HVAC
external mixtures, values in both seasons above unity may be system and 100% outdoor air operation (no recirculation).
due to less mechanical (e.g. ltration and depositional) losses of Differences between these studies highlight the heterogeneity
these traffic particles. of indoor spaces even within the same building.
Additionally, since summertime (I/O)i/SO4 values of BC and For OOA, the (I/O)OOA/SO4 value has a different seasonal
HOA are higher than values in wintertime, variations in the relationship, with summer showing lower sulfate normalized
HVAC operation may explain that difference. Summertime RH ratios than winter. Seasonal differences in the OOA diurnal
levels and HVAC cooling coil operation will serve to increase patterns indicate that much of the summertime OOA is formed
water content of hygroscopic accumulation mode particles, locally through photochemical production due to its similar
potentially activating them to droplets, compared to the non- diurnal patterns as outdoor ozone, whereas the wintertime OOA
hygroscopic traffic related particles.72 This could increase the appears to be more regional and therefore aged, and varies
View Article Online
inversely with boundary layer height. With this trend in mind, it are provided in Table 2. Additionally, a non-parametric LOESS
is likely that the summer time OOA is more volatile than the t as a function of T and RH are provided in Table S2† which
wintertime OOA.26,63 Additional losses may also be due to HVAC provides a stepwise linear t to the data with no assumption of
operational differences by season. Summertime air condi- functional form. We provide the various regression analysis and
tioning condenses water and gas-phase water-soluble materials t parameters so that researchers who model the indoor envi-
in the aqueous lms of the system cooling coil. We hypothesize ronment may be able to choose a scheme to implement in
that this process would lead to losses of water-soluble compo- indoor models of aerosol chemical species. Of note is that all
nents as that water is drained and removed from the system and linear regressions agree in sign with those discussed in Johnson
then replenished. Many types of OOA and their gas-phase et al.,14 for a similar single-season analysis in a different loca-
counterparts are water-soluble. Depletion of gas-phase species tion, except for those reported with effectively no correlation (R2
through loss to aqueous lms will lead to the need to re- < 0.1). As seen in Fig. 7 and 8, the overall directional trend of
establish equilibrium when the air leaves the HVAC system, each species and PMF factor (I/O)i/SO4 are the same for both T
resulting in loss of particle based OOA components to the gas and RH gradients. This is expected since T and RH gradients are
phase, and likely contribute to the observation that (I/O)OOA/SO4 broadly interrelated, with increases in T gradients having cor-
are less than unity in both seasons.73 responding decreases in RH gradients. When split by seasons,
COA exhibits the largest inter-seasonal variation in (I/O)i/SO4 summer linear ts are better for the summer data than for
in both median value and range. This variability is likely due to winter (except for BC). In general, the regressions are low due to
an indoor source of COA or an indoor source chemically similar invariance in the (I/O)i/SO4 ratio over the season rather than
to COA. Unlike other species where (I/O)i/SO4 is large (including scatter (i.e. slopes < 0.01 correspond to R2 < 0.6).
HOA), inspection of the COA time series reveals several The behavior of total aerosol nitrate is shown in Fig. 7a. Total
instances of indoor concentrations exceeding outdoor concen- nitrate is lost in both seasons (almost all I/O ratio data is less
trations, with the (I/O) and (I/O)i/SO4 both greater than unity in than unity), with the exponential t better capturing the at-
the wintertime. Potential explanations for this include a cook- tening out at (I/O)NO3/SO4 of 0.37 for positive (in > out) T gradi-
ing source from recirculated office air (the smell of toast was ents. This attening indicates that potentially some limiting
observed indoors) or differences in occupancy with regular factor could keep all of the nitrate from being completely
classes in the winter, but not in the summer. This seasonal removed. This effectively full removal at low positive T gradients
difference is important to note while discussing the seasonal could be why the regression for winter only is poor. This trend is
gradients. also consistent with thermodenuder data showing nitrate
remaining in the aerosol phase at temperatures exceeding
100 C, albeit at higher values.26 Inorganic nitrate (Fig. 7b) and
3.5 Dependence of (I/O)i/SO4 ratios on T and RH gradients organic nitrate (part c) have similar responses to the T gradient.
By combining winter and summer datasets, the trends in However, the asymptote at high indoor vs. outdoor T of oNO3 is
(I/O)i/SO4 ratios as a function of temperature and relative higher than iNO3, at 0.60 versus 0.31, respectively (see Table 2),
humidity can be investigated and extend the single season which is consistent for a less volatile species. For RH gradients,
analysis by Johnson et al.14 Data from each season was a linear regression matches the trend for total nitrate, oNO3,
combined and T and RH bins were created by placing an equal and iNO3 very well (total nitrate R2 ¼ 0.95). Note that the y-
number of points in each bin instead of an evenly-spaced intercepts 0.61, 0.52, and 0.72 for total, iNO3, and oNO3
temperature or humidity difference. This unfortunately respectively, are each less than unity. If T or RH gradients were
creates non-uniform bin spacing (more pronounced in T data), the only loss mechanism beyond the mechanical losses
but does not suffer from over-weighing sparse data. For observed for sulfate, the y-intercept from the linear regression
example, there are very few measurements made for a T for aerosol nitrate species should be unity. This lower y-inter-
gradient between 5 and +5 C but there are dozens of cept indicates that there is an additional loss process occurring
measurements made for a temperature gradient of 10 to 5 C in addition to mechanical, and T or RH gradient related losses.
bin (see Fig. 2). Since the T differences are, by nature, segregated Trends in (I/O)Org/SO4 versus T and RH are shown in 7d and
by season (winter positive, summer negative), and the RH are mostly all slightly above unity, with a weak decrease in T and
differences are not (winter positive, summer both positive and RH gradients. The variation in (I/O)Org/SO4 values is small
negative), and the RH range is much larger, examining the compared to nitrate with the box-whisker plots showing tighter
ratios as a function of both T and RH provides important distributions in the summer (negative T gradients). The sepa-
information. ration of organics into components via PMF and the subsequent
The (I/O)i/SO4 values as a function of T and RH gradients T and RH gradient analysis is shown in Fig. 8. This gure shows
between indoors and outdoors are shown in Fig. 7 and 8. Each considerable differences in the various organic component
plot shows the statistical distribution of (I/O)i/SO4 data as a box- behaviors. These changes largely offset each other, and explain
whisker placed at the center of each T or RH bin. Regression why the observed total organic ratio (I/O)Org/SO4 is largely
analysis on median values for each (I/O)i/SO4 are provided for invariable.
linear (black lines) and exponential (grey lines) ts to the full Trends over the whole range of T and RH gradients for OOA
datasets, while linear regression for each season separately is are shown in Fig. 8, and indicate additional losses ((I/O)i/SO4 less
demonstrated in red lines. The t parameters for each analysis than unity) for both T and RH gradients. The variations,
View Article Online

Fig. 7 Dependence of (I/O)i/SO4 on differences in temperature and humidity between environments for (a) nitrate, (b) inorganic nitrate (c) organic
nitrate, and (d) organics data have been binned by temperature or humidity differences, and the regression is of the median value of each bin.
Both temperature and humidity gradients favor volatilization in the positive direction. Complete statistics of the regression are listed in Table 2.
however, are not t well and may be due to chemical differences slope while the winter is at, similar to that observed with
in the OOA factor between seasons as discussed during the temperature gradients.
diurnal analysis. The T gradient plot (again, naturally separated For COA, the T and RH gradients in (I/O)COA/SO4 more clearly
by season) shows somewhat different trends between seasons, shows disparities between seasons. In summer (negative T
with a negative slope in summer (negative gradient values) and gradients) there appears to be a slight decrease in median
at in winter. The positive trend between seasons is likely (I/O)COA/SO4 similar to volatility-based removal, and all median
explained by the differences in OOA volatility and formation values are less than unity. However, in winter (positive T
differences between the seasons as previously discussed in gradients) the trend is reversed and the box and whiskers cover
section 3.4. This potential difference in OOA between seasons a much broader range. This again indicates a COA source
complicates the RH gradient analysis where seasons overlap; indoors in winter that is not related to T or RH gradient. This is
this is observed in the broad distribution of ratios seen for the likely due to high occupancy in the winter with occupants and
overlapping seasonal data in the 10 to 30% RH difference range. associated activities in adjacent spaces (e.g. use of toasters and
Fig. S9† shows the season-separated binning and regression other office cooking products). The human contribution to the
lines, indicating that the summer regression has a negative indoor aerosol composition during the wintertime measure-
ments will be discussed in a future publication. In summer,
View Article Online

Fig. 8 Dependence of (I/O)i/SO4 on differences in temperature and humidity between environments for (a) OOA, (b) COA, (c) HOA, and (d) black
carbon (BC). Data have been binned by temperature or humidity differences, and the regression is of the median value of each bin. Both
temperature and humidity gradients favor volatilization in the positive direction. Complete statistics of the regression are listed in Table 2.
there were no scheduled classes and overall student activities a combination of the following: additional gas to particle par-
were reduced. (I/O)COA/SO4 trends with RH gradient show an titioning of HOA vapors at lower indoor T, enhanced loss of
increase in value with increasing RH gradient, but this is accumulation mode particles in relation to traffic related
complicated by seasonal overlap. For RH gradients observed in particles as a function of T, or an indoor source of HOA that is T
both seasons (bins encompassing +10 to 30% gradient), the dependent. Comparing the characteristics of HOA and BC as
(I/O)COA/SO4 values in the box whisker plot show broader described below help to elucidate what mechanisms are
distributions. dominant. From the results of the work discussed here, it is
For HOA, the (I/O)HOA/SO4 trends are quite large in the clear that there is evidence of external mixing and size
summer (negative T gradient) and low RH gradient (<15%). dependent differences in I/O ratios for the AMS traffic related
Similar to the general seasonal analysis (I/O)HOA/SO4 is nearly signal at m/z 57 (see Fig. S8†). Further work will need to be
always greater than unity with a clear trend in T and RH performed with additional sampling locations to address
gradients. As discussed above, the (I/O)HOA/SO4 value greater whether loss rate or potential sources (and location of those
than unity is likely due to external mixing of HOA containing sources e.g. HVAC, room, etc.) of HOA are more important.
traffic particles and sulfate containing accumulation mode While the total (I/O)HOA/SO4 and (I/O)BC/SO4 statistics for each
particles. The strong dependence on T may be one or season are similar between HOA and BC (see Fig. 6), the trends
View Article Online
Table 2 For each AMS species or PMF-derived organic component, changes between environments and between seasons: winter is
summary statistics of regressions for I/O ratios of components i higher than summer, and indoor is higher than outdoor in both
normalized by the I/O ratio of sulfate (I/O)i/SO4 against indoor-to-
seasons. However, the correlation in summer is poor compared
outdoor temperature differences (DTin–out) and humidity differences
(DRHout–in) with winter (R2 ¼ 0.33 versus R2 ¼ 0.66 outdoors), such that the
relative contribution of different sources of HOA and BC are
Linear t of (I/O)i/SO4 different. The outdoor HOA/BC ratio is correlated to outdoor T
against multi-season Exponential t of (I/O)i/SO4 (slope ¼ 0.009, R2 ¼ 0.69), across the almost 40 C range of
DTin–out against multi-season DTin–out
Species or observed outdoor T, indicating volatility-based differences
factor Slope y-int R2 y0 A s between BC and HOA contribute to differences between seasons
and environments. However, the range of values in summer is

NO3 0.012 0.61 0.77 0.37 0.56 9.9 considerably higher than winter (see Fig. S4†). Similarly, the I/O
iNO3 0.011 0.52 0.66 0.31 0.58 7.04
ratio of HOA/BC (of note, not normalized by sulfate), shows
oNO3 0.007 0.72 0.53 0.6 0.36 6.58
Org 0.007 1.10 0.82 0.29 0.91 119.1 a weak decrease with T gradient (slope ¼ 0.007, R2 ¼ 0.24),
OOA 0.010 0.61 0.74 1.11 0.65 45.1 with what appears to be a downward slope in each season
COA 0.020 0.95 0.87 346 345 1.7 104 independently. There is no correlation with the I/O ratio per RH
HOA 0.032 1.76 0.73 1.2 1.58 6.8 difference (slope ¼ 0.0001, R2 < 0.01).
BC 0.007 1.72 0.03 7.32 5.51 1.87 1010
Linear t of (I/O)i/SO4
against multi-season Exponential t of (I/O)i/SO4
DRHout–in against multi-season DRHout–in 3.6 Aerosol liquid water and implications for the indoor
Species or environment
factor Slope y-int R2 y0 A s
Aerosol liquid water (ALW) calculated from Köhler theory has
NO3 0.012 0.81 0.95 0.23 1.27 66.6 important implications for indoor chemistry and dynamics.30
iNO3 0.013 0.73 0.92 0.013 0.73 0.92 Outdoors, over the lifetime of aerosols, ALW inuences parti-
oNO3 0.007 0.84 0.84 0.007 0.84 0.84 tioning and chemical transformation of both organic and
Org 0.003 1.16 0.81 0.93 0.29 67.2
OOA 0.004 0.55 0.58 67.8 67.31 1.59 104
inorganic aerosol components. The impact of ALW on organics
COA 0.010 0.77 0.64 467 466 4.62 104 outdoors is not completely understood, but it has been shown
HOA 0.023 2.10 0.77 1.06 1.74 24.8 to participate in both reversible and irreversible uptake of
BC 0.021 1.98 0.74 1.17 1.54 18.2 organics.74,75 In this work, the most water-soluble organic frac-
tion, OOA, trends diurnally with ALW in winter, but the reverse
Linear t of (I/O)i/SO4 Linear t of (I/O)i/SO4
against summer DTin–out against winter DTin–out is true in the summer, when outdoor OOA concentrations are
Species or strongly driven by photochemistry.
factor Slope y-int R2 Slope y-int R2 Indoors, an aqueous phase or surface is also an important
part of understanding chemical behavior including partition-
NO3 0.034 0.42 0.95 0.002 0.36 0.04
iNO3 0.044 0.23 0.97 0.007 0.20 0.44 ing, transformation, and subsequent exposure. Multiple studies
oNO3 0.029 0.53 0.94 0.006 0.49 0.22 have examined the importance of humidity on ozone-terpene
Org 0.004 1.14 0.22 0.004 1.06 0.12 reactions indoors, increasing mass loading and particle
OOA 0.010 0.41 0.52 0.004 0.86 0.20 formation.76–78 For this study, ALW indoors here is seasonally-
COA 0.016 0.61 0.82 0.021 0.94 0.60
dependent, with both seasons showing lower ALW indoors
HOA 0.122 0.93 0.88 0.004 1.28 0.02
BC 0.030 1.62 0.40 0.093 0.10 0.52 compared to outdoors.
ALW is driven strongly by the mass fraction of aerosol that is
hygroscopic,54 typically the inorganic salt components. In the
versus T and RH gradients reveal differences. While (I/O)HOA/SO4 wintertime, the average nitrate concentration was higher than
shows a decrease across T gradients from about 2.5 to just above summer, and additional build-up events lead to extremely high
unity, BC shows little variation across T gradients with nitrate concentrations, and enhanced ALW outdoors. Upon
(I/O)BC/SO4 ratios near 2 throughout. The consistently high transport to the indoor environment, the combination of
(I/O)i/SO4 values (much greater than unity) for both HOA and BC heating and lower RH drives the nitrate out of the aerosol phase,
are likely explained by relative loss rates of traffic and accumu- and signicantly reduces the ALW of indoor aerosol in the
lation mode particles, but there is little change over T or RH wintertime. In wintertime, the ALW difference between indoor
gradients for non-volatile BC as compared to the volatile HOA. and outdoor is dramatic with high (2.6 3.9 mg m3) outdoors
Since (I/O)HOA/SO4 has a stronger dependence on T than (I/O)BC/SO4 and extremely low (0.11 0.06 mg m3) indoors. In summer, the
for the summer season, the most likely explanation for the lower-nitrate but humid environment outdoors gives similar
observed HOA dependence is T driven partitioning. RH gradients but more consistent ALW 2.7 2.5 mg m3. In contrast, summer
for (I/O)BC/SO4 indicate some dependence of BC similar to that indoor ALW is much higher than during winter indoors at 0.53
observed for HOA but not to the same magnitude. 0.24 mg m3. As ALW is calculated from dry mass concen-
HOA is known to be a semi-volatile OA component24,26 while trations, the mechanical removal of mass upon transport to the
BC is not. As described in Section 3.3.1, the HOA/BC ratio indoor environment contributes to the lower ALW
View Article Online
indoor and outdoor air quality. The outdoor seasonal analysis

shows trends similar to those expected of an urban area, as
organics dominated outdoor aerosol in both seasons, and
nitrate was a major component in winter but not summer.
However, most human exposure to pollutants occurs indoors,
and outdoor aerosols are transformed before reaching indoor
spaces, and resulting indoor concentrations are a result of the
contribution of individual aerosol components, the HVAC
system inuence, and the environmental conditions (e.g. T,
RH). The sulfate normalized I/O ratio, called (I/O)i/SO4, was used
to isolate chemically-specic changes to other aerosol species,
beyond aerosol mechanical losses. The broad range of condi-
Fig. 9 Distribution of liquid water normalized by the measured dry tions sampled here, across several weeks in summer and winter
aerosol mass for each season and environment. Winter and summer
seasons in a highly controlled indoor space, provides
have similar distributions outdoors (although slightly larger distribution
in winter), but in winter indoors, minimal ALW is observed, compared a comprehensive parameterization for analysis of these effects.
with summer. The (I/O)i/SO4 of nitrate, total organics, HOA, and BC are higher
in the summer season than winter, and the (I/O)i/SO4 are
described as a function of outdoor and indoor T and RH
concentration indoors. Therefore, the following section will
gradients. These trends can be applied to indoor models to
discuss ALW as normalized to dry mass.
better inform indoor aerosol dynamics and exposure. Aerosol
Fig. 9 shows the distribution of ALW per dry aerosol mass in
liquid water (ALW) inuences aerosol processing and is gener-
each season and environment. Distributions of ALW are highly
ally available outdoors in both seasons, but ALW is nearly
variable outdoors since neither T or RH are controlled.
nonexistent in winter indoors, and higher but rather invariable
Maximum wintertime ALW are larger than in summer, but
in summer indoors. This trend in ALW can have important
wintertime increases in hygroscopic nitrate are mitigated by
consequences for indoor chemistry. Connecting outdoor
higher humidity in summer, with the result of very similar ALW
pollutants to exposure indoors is a key challenge for human
outdoors between seasons. In the indoor environment, ALW
health, and this work provides direct measurements of indoor
occupies a much smaller range of values due to the less variable
and outdoor aerosols, and parameterizations directly applicable
T and RH conditions of the classroom. In winter there is no
to future analyses.
overlap between the indoor and outdoor environment, whereas
the summer time shows the indoor ALW varying roughly
between the 25th and 50th percentiles of the outdoor values. Conflicts of interest
Of importance to this discussion is also the role the HVAC
The authors have no conicts of interest to declare.
system plays in both regulating the indoor classroom environ-
ment and providing an extreme of T and RH which changes
seasonally. Wintertime aerosol passing through the HVAC Acknowledgements
system as part of the air intake or recirculation will be heated to
Funding for this work was provided by the US National Science
temperatures of about 40 C. These are the highest tempera-
Foundation Award #: CBET-1437916.
tures and lowest RH values the indoor air will experience, and
will effectively dry out or effloresce aerosols. In summer, the
HVAC system works to cool the air and exposes air in the system References
to temperatures of approximately 10 C and generally
1 F. Dominici, R. D. Peng, M. L. Bell, L. Pham, A. Mcdermott,
increasing RH to 100%, driving deliquescence of aerosol parti-
S. L. Zeger and J. M. Samet, Fine Particulate Air Pollution and
cles and condensation of water on chiller coils. The ALW pre-
Hospital Admission for Cardiovascular and Respiratory
sented in Fig. 9 show the calculated values for the room
Diseases, J. Am. Med. Assoc., 2006, 295, 1127–1134.
conditions where the particles spend the vast majority of their
2 N. E. Klepeis, W. C. Nelson, W. R. Ott, J. P. Robinson,
time, but these values will be much higher for indoor aerosols
A. M. Tsang, P. Switzer, J. V. Behar, S. C. Hern and
in summer in the chilling portion of the HVAC and lower in
W. H. Engelmann, The National Human Activity Pattern
winter in the heating section. Differences in aerosol composi-
Survey (NHAPS): A resource for assessing exposure to
tion and chemistry at various stages of transport to the indoor
environmental pollutants, J. Exposure Anal. Environ.
environment are beyond the scope of this study, but should be
Epidemiol., 2001, 11, 231–252.
investigated in future work.
3 W. C. Malm, J. F. Sisler, D. Huffman, R. A. Eldred and
T. A. Cahill, Spatial and seasonal trends in particle
4. Conclusions concentration and optical extinction in the United States, J.
Geophys. Res., 1994, 99, 1347.
This investigation of seasonal trends of outdoor and indoor 4 N. L. Ng, S. C. Herndon, A. Trimborn, M. R. Canagaratna,
aerosols has improved our understanding of the link between P. L. Croteau, T. B. Onasch, D. Sueper, D. R. Worsnop,
View Article Online
Q. Zhang, Y. L. Sun and J. T. Jayne, An Aerosol Chemical 16 Q. Y. Meng, B. J. Turpin, H. L. Jong, A. Polidori, C. P. Weisel,
Speciation Monitor (ACSM) for routine monitoring of the M. Morandi, S. Colome, J. Zhang, T. Stock and A. Winer, How
composition and mass concentrations of ambient aerosol, does inltration behavior modify the composition of
Aerosol Sci. Technol., 2011, 45, 780–794. ambient PM 2.5 in indoor spaces? An analysis of RIOPA
5 Y. L. Sun, Z. F. Wang, W. Du, Q. Zhang, Q. Q. Wang, P. Q. Fu, data, Environ. Sci. Technol., 2007, 41, 7315–7321.
X. L. Pan, J. Li, J. Jayne and D. R. Worsnop, Long-term real- 17 D. L. Liu and W. W. Nazaroff, Particle penetration through
time measurements of aerosol particle composition in building cracks, Aerosol Sci. Technol., 2003, 37, 565–573.
Beijing, China: Seasonal variations, meteorological effects, 18 W. J. Riley, T. E. McKone, A. C. K. Lai and W. W. Nazaroff,
and source analysis, Atmos. Chem. Phys., 2015, 15, 10149– Indoor particulate matter of outdoor origin: Importance of
10165. size-dependent removal mechanisms, Environ. Sci.

6 S. Takahama, A. Johnson, J. Guzman Morales, L. M. Russell, Technol., 2002, 36, 200–207.
R. Duran, G. Rodriguez, J. Zheng, R. Zhang, D. Toom-Sauntry 19 L. A. Wallace, S. J. Emmerich and C. Howard-Reed, Source
and W. R. Leaitch, Submicron organic aerosol in Tijuana, Strengths of Ultrane and Fine Particles Due to Cooking
Mexico, from local and Southern California sources during with a Gas Stove, Environ. Sci. Technol., 2004, 38, 2304–2311.
the Calmex campaign, Atmos. Environ., 2013, 70, 500–512. 20 W. W. Nazaroff and C. J. Weschler, Cleaning products and
7 P. F. DeCarlo, J. R. Kimmel, A. Trimborn, M. J. Northway, air fresheners: Exposure to primary and secondary air
J. T. Jayne, A. C. Aiken, M. Gonin, K. Fuhrer, T. Horvath, pollutants, Atmos. Environ., 2004, 38, 2841–2865.
K. S. Docherty, D. R. Worsnop and J. L. Jimenez, Field- 21 B. C. Singer, H. Destaillats, A. T. Hodgson and
Deployable, High-Resolution, Time-of-Flight Aerosol Mass W. W. Nazaroff, Cleaning products and air fresheners:
Spectrometer, Anal. Chem., 2006, 78, 8281–8289. Emissions and resulting concentrations of glycol ethers
8 M. R. Canagaratna, J. T. Jayne, J. L. Jimenez, J. D. Allan, and terpenoids, Indoor Air, 2006, 16, 179–191.
M. R. Alfarra, Q. Zhang, T. B. Onasch, F. Drewnick, H. Coe, 22 N. E. Klepeis, M. G. Apte, L. A. Gundel, R. G. Sextro and
A. Middlebrook, A. Delia, L. R. Williams, A. M. Trimborn, W. W. Nazaroff, Determining Size-Specic Emission
M. J. Northway, P. F. DeCarlo, C. E. Kolb, P. Davidovits and Factors for Environmental Tobacco Smoke Particles,
D. R. Worsnop, Mass Spectrom. Rev., 2007, 26, 185–222. Aerosol Sci. Technol., 2003, 37, 780–790.
9 Q. Zhang, J. L. Jimenez, D. R. Worsnop and 23 C. Chen and B. Zhao, Review of relationship between indoor
M. R. Canagaratna, A case study of urban particle acidity and outdoor particles: I/O ratio, inltration factor and
and its inuence on secondary organic aerosol, Environ. penetration factor, Atmos. Environ., 2011, 45, 275–288.
Sci. Technol., 2007, 41, 3213–3219. 24 C. D. Cappa and J. L. Jimenez, Quantitative estimates of the
10 V. A. Lanz, A. S. H. Pŕevôt, M. R. Alfarra, S. Weimer, C. Mohr, volatility of ambient organic aerosol, Atmos. Chem. Phys.,
P. F. DeCarlo, M. F. D. Gianini, C. Hueglin, J. Schneider, 2010, 10, 5409–5424.
O. Favez, B. D'Anna, C. George and U. Baltensperger, 25 K. R. Kolesar, Z. Li, K. R. Wilson and C. D. Cappa, Heating-
Characterization of aerosol chemical composition with Induced Evaporation of Nine Different Secondary Organic
aerosol mass spectrometry in Central Europe: An overview, Aerosol Types, Environ. Sci. Technol., 2015, 49, 12242–12252.
Atmos. Chem. Phys., 2010, 10, 10453–10471. 26 J. A. Huffman, K. S. Docherty, A. C. Aiken, M. J. Cubison,
11 C. L. Martin, J. D. Allan, J. Crosier, T. W. Choularton, H. Coe I. M. Ulbrich, P. F. DeCarlo, D. Sueper, J. T. Jayne,
and M. W. Gallagher, Seasonal variation of ne particulate D. R. Worsnop, P. J. Ziemann and J. L. Jimenez,
composition in the centre of a UK city, Atmos. Environ., Chemically-resolved aerosol volatility measurements from
2011, 45, 4379–4389. two megacity eld studies, Atmos. Chem. Phys., 2009, 9,
12 J. Park, S. Lee, M. Kang, H. joo Cho, K. Lee and K. Park, 7161–7182.
Seasonal characteristics of submicrometer organic aerosols 27 N. M. Donahue, A. L. Robinson, C. O. Stanier and
in urban Gwangju, Korea using an aerosol mass S. N. Pandis, Coupled partitioning, dilution, and chemical
spectrometer, Atmos. Environ., 2013, 80, 445–454. aging of semivolatile organics, Environ. Sci. Technol., 2006,
13 W. Hu, M. Hu, W. Hu, J. L. Jimenez, B. Yuan, W. Chen, 40, 2635–2643.
M. Wang, Y. Wu, C. Chen, Z. Wang, J. Peng, L. Zeng and 28 N. Hodas and B. J. Turpin, Shis in the gas-particle
M. Shao, Chemical composition, sources, and aging partitioning of ambient organics with transport into the
process of submicron aerosols in Beijing: Contrast indoor environment, Aerosol Sci. Technol., 2014, 48, 271–281.
between summer and winter, J. Geophys. Res., 2016, 121, 29 M. S. Waring, Secondary organic aerosol in residences:
1955–1977. Predicting its fraction of ne particle mass and
14 A. M. Johnson, M. S. Waring and P. F. DeCarlo, Real-time determinants of formation strength, Indoor Air, 2014, 24,
transformation of outdoor aerosol components upon 376–389.
transport indoors measured with aerosol mass 30 S. Duncan, K. G. Sexton and B. Turpin, Oxygenated VOCs,
spectrometry, Indoor Air, 2017, 27, 230–240. aqueous chemistry, and potential impacts on residential
15 P. F. DeCarlo, A. M. Avery and M. S. Waring, Thirdhand indoor air composition, Indoor Air, 2018, 28, 198–212.
smoke uptake to aerosol particles in the indoor 31 Z. Meng, J. H. Seinfeld, P. Saxena and Y. P. Kim,
environment, Sci. Adv., 2018, 4, 1–8. Contribution of Water to Particulate Mass in the South
Coast Air Basin, Aerosol Sci. Technol., 1995, 22, 111–123.
View Article Online
32 B. Ervens, B. J. Turpin and R. J. Weber, Secondary organic 42 N. K. Meyer, J. Duplissy, M. Gysel, A. Metzger, J. Dommen,
aerosol formation in cloud droplets and aqueous particles E. Weingartner, M. R. Alfarra, A. S. H. Prevot, C. Fletcher,
(aqSOA): A review of laboratory, eld and model studies, N. Good, G. Mcggans, A. M. Jonsson, M. Hallquist,
Atmos. Chem. Phys., 2011, 11, 11069–11102. U. Baltensperger and Z. D. Ristovski, Analysis of the
33 J. F. Pankow, Organic particulate material levels in the hygroscopic and volatile properties of ammonium sulphate
atmosphere: Conditions favoring sensitivity to varying seeded and unseeded SOA particles, Atmos. Chem. Phys.,
relative humidity and temperature, Proc. Natl. Acad. Sci., 2009, 9, 721–732.
2010, 107, 6682–6686. 43 A. M. Middlebrook, R. Bahreini, J. L. Jimenez and
34 M. Shiraiwa, M. Ammann, T. Koop and U. Poschl, Gas M. R. Canagaratna, Evaluation of composition-dependent
uptake and chemical aging of semisolid organic aerosol collection efficiencies for the Aerodyne aerosol mass
particles, Proc. Natl. Acad. Sci., 2011, 108, 11003–11008. spectrometer using eld data, Aerosol Sci. Technol., 2012,
35 N. Hodas, A. P. Sullivan, K. Skog, F. N. Keutsch, J. L. Collett, 46(3), 258–271.
S. Decesari, M. C. Facchini, A. G. Carlton, A. Laaksonen and 44 N. Takegawa, Y. Miyazaki, Y. Kondo, Y. Komazaki,
B. J. Turpin, Aerosol liquid water driven by anthropogenic T. Miyakawa, J. L. Jimenez, J. T. Jayne, D. R. Worsnop,
nitrate: Implications for lifetimes of water-soluble organic J. D. Allan and R. J. Weber, Characterization of an
gases and potential for secondary organic aerosol Aerodyne Aerosol Mass Spectrometer (AMS):
formation, Environ. Sci. Technol., 2014, 48, 11127–11136. Intercomparison with Other Aerosol Instruments, Aerodyne
36 T. K. V Nguyen, M. D. Petters, S. R. Suda, H. Guo, R. J. Weber Aerosol Mass Spectrometer, 2005, 39, 760–770.
and A. G. Carlton, Trends in particle-phase liquid water 45 J. D. Allan, A. E. Delia, H. Coe, K. N. Bower, M. R. Alfarra,
during the Southern Oxidant and Aerosol Study, Atmos. J. L. Jimenez, A. M. Middlebrook, F. Drewnick,
Chem. Phys., 2014, 14, 10911–10930. T. B. Onasch, M. R. Canagaratna, J. T. Jayne and
37 W. Xu, T. Han, W. Du, Q. Wang, C. Chen, J. Zhao, Y. Zhang, D. R. Worsnop, A generalised method for the extraction of
J. Li, P. Fu, Z. Wang, D. R. Worsnop and Y. Sun, Effects of chemically resolved mass spectra from Aerodyne aerosol
aqueous-phase and photochemical processing on mass spectrometer data, J. Aerosol Sci., 2004, 35, 909–922.
secondary organic aerosol formation and evolution in 46 A. C. Aiken, P. F. DeCarlo, J. H. Kroll, D. R. Worsnop,
Beijing, China, Environ. Sci. Technol., 2017, 51, 762–770. J. A. Huffman, K. S. Docherty, I. M. Ulbrich, C. Mohr,
38 B. Jing, S. Tong, Q. Liu, K. Li, W. Wang, Y. Zhang and M. Ge, J. R. Kimmel, D. Sueper, Y. Sun, Q. Zhang, A. Trimborn,
Hygroscopic behavior of multicomponent organic aerosols M. Northway, P. J. Ziemann, M. R. Canagaratna,
and their internal mixtures with ammonium sulfate, T. B. Onasch, M. R. Alfarra, A. S. H. Prevot, J. Dommen,
Atmos. Chem. Phys., 2016, 16, 4101–4118. J. Duplissy, A. Metzger, U. Baltensperger and J. L. Jimenez,
39 M. L. Pöhlker, C. Pöhlker, F. Ditas, T. Klimach, I. H. De O/C and OM/OC ratios of primary, secondary, and ambient
Angelis, A. Araújo, J. Brito, S. Carbone, Y. Cheng, X. Chi, organic aerosols with high-resolution time-of-ight aerosol
R. Ditz, S. S. Gunthe, J. Kesselmeier, T. Könemann, mass spectrometry, Environ. Sci. Technol., 2008, 42, 4478–
J. V. Lavrič, S. T. Martin, E. Mikhailov, D. Moran-Zuloaga, 4485.
D. Rose, J. Saturno, H. Su, R. Thalman, D. Walter, J. Wang, 47 M. R. Canagaratna, J. L. Jimenez, J. H. Kroll, Q. Chen,
S. Wolff, H. M. J. Barbosa, P. Artaxo, M. O. Andreae and S. H. Kessler, P. Massoli, L. Hildebrandt Ruiz, E. Fortner,
U. Pöschl, Long-term observations of cloud condensation L. R. Williams, K. R. Wilson, J. D. Surratt, N. M. Donahue,
nuclei in the Amazon rain forest - Part 1: Aerosol size J. T. Jayne and D. R. Worsnop, Elemental ratio
distribution, hygroscopicity, and new model measurements of organic compounds using aerosol mass
parametrizations for CCN prediction, Atmos. Chem. Phys., spectrometry: Characterization, improved calibration, and
2016, 16, 15709–15740. implications, Atmos. Chem. Phys., 2015, 15, 253–272.
40 J. Duplissy, P. F. De Carlo, J. Dommen, M. R. Alfarra, 48 L. Drinovec, G. Močnik, P. Zotter, A. S. H. Prévôt,
A. Metzger, I. Barmpadimos, A. S. H. Prevot, C. Ruckstuhl, E. Coz, M. Rupakheti, J. Sciare, T. Müller,
E. Weingartner, T. Tritscher, M. Gysel, A. C. Aiken, A. Wiedensohler and A. D. A. Hansen, The ‘dual-spot’
J. L. Jimenez, M. R. Canagaratna, D. R. Worsnop, Aethalometer: An improved measurement of aerosol black
D. R. Collins, J. Tomlinson and U. Baltensperger, Relating carbon with real-time loading compensation, Atmos. Meas.
hygroscopicity and composition of organic aerosol Tech., 2015, 8, 1965–1979.
particulate matter, Atmos. Chem. Phys., 2011, 11, 1155–1165. 49 D. K. Farmer, A. Matsunaga, K. S. Docherty, J. D. Surratt,
41 E. Swietlicki, H. C. Hansson, K. Hämeri, B. Svenningsson, J. H. Seinfeld, P. J. Ziemann and J. L. Jimenez, Response of
A. Massling, G. Mcggans, P. H. Mcmurry, T. Petäjä, an aerosol mass spectrometer to organonitrates and
P. Tunved, M. Gysel, D. Topping, E. Weingartner, organosulfates and implications for atmospheric
U. Baltensperger, J. Rissler, A. Wiedensohler and chemistry, Proc. Natl. Acad. Sci., 2010, 107, 6670–6675.
M. Kulmala, Hygroscopic properties of submicrometer 50 A. Kiendler-Scharr, A. A. Mensah, E. Friese, D. Topping,
atmospheric aerosol particles measured with H-TDMA E. Nemitz, A. S. H. Prevot, M. Äijälä, J. Allan, F. Canonaco,
instruments in various environments - A review, Tellus, Ser. M. R. Canagaratna, S. Carbone, M. Crippa, M. Dall Osto,
B: Chem. Phys. Meteorol., 2008, 60, 432–469. D. A. Day, P. De Carlo, C. F. Di Marco, H. Elbern,
A. Eriksson, E. Freney, L. Hao, H. Herrmann,
View Article Online
L. Hildebrandt, R. Hillamo, J. L. Jimenez, A. Laaksonen, 60 M. R. Canagaratna, J. T. Jayne, D. A. Ghertner, S. Herndon,

G. McFiggans, C. Mohr, C. O'Dowd, R. Otjes, Q. Shi, J. L. Jimenez, P. J. Silva, P. Williams, T. Lanni,
J. Ovadnevaite, S. N. Pandis, L. Poulain, P. Schlag, F. Drewnick, K. L. Demerjian, C. E. Kolb and
K. Sellegri, E. Swietlicki, P. Tiitta, A. Vermeulen, D. R. Worsnop, Chase Studies of Particulate Emissions
A. Wahner, D. Worsnop and H. C. Wu, Ubiquity of organic from in-use New York City Vehicles, Aerosol Sci. Technol.,
nitrates from nighttime chemistry in the European 2004, 38, 555–573.
submicron aerosol, Geophys. Res. Lett., 2016, 43, 7735–7744. 61 M. Crippa, P. F. DeCarlo, J. G. Slowik, C. Mohr,
51 P. Paatero and U. Tapper, Positive Matrix Factorization: A M. F. Heringa, R. Chirico, L. Poulain, F. Freutel, J. Sciare,
non-negative factor model with optimal utilization of erro J. Cozic, C. F. Di Marco, M. Elsasser, J. B. Nicolas,
estimates of data values, Environmetrics, 1994, 5, 111–126. N. Marchand, E. Abidi, A. Wiedensohler, F. Drewnick,
52 V. A. Lanz, M. R. Alfarra, U. Baltensperger, B. Buchmann, J. Schneider, S. Borrmann, E. Nemitz, R. Zimmermann,
C. Hueglin and A. S. H. Prévôt, Source apportionment of J. L. Jaffrezo, A. S. H. Prévôt and U. Baltensperger,
submicron organic aerosols at an urban site by factor Wintertime aerosol chemical composition and source
analytical modelling of aerosol mass spectra, Atmos. Chem. apportionment of the organic fraction in the metropolitan
Phys., 2007, 7, 1503–1522. area of Paris, Atmos. Chem. Phys., 2013, 13, 961–981.
53 I. M. Ulbrich, M. R. Canagaratna, Q. Zhang, D. R. Worsnop 62 F. Freutel, F. Drewnick, J. Schneider, T. Klimach and
and J. L. Jimenez, Interpretation of organic components S. Borrmann, Quantitative single-particle analysis with the
from Positive Matrix Factorization of aerosol mass Aerodyne aerosol mass spectrometer: Development of
spectrometric data, Atmos. Chem. Phys., 2009, 9, 2891–2918. a new classication algorithm and its application to eld
54 M. D. Petters and S. M. Kreidenweis, A single parameter data, Atmos. Meas. Tech., 2013, 6, 3131–3145.
representation of hygroscopic growth and cloud 63 J. L. Jimenez, M. R. Canagaratna, N. M. Donahue,
condensation nucleus activity, Atmos. Chem. Phys., 2007, 7, A. S. H. Prevot, Q. Zhang, J. H. Kroll, P. F. DeCarlo,
1961–1971. J. D. Allan, H. Coe, N. L. Ng, A. C. Aiken, K. S. Docherty,
55 M. Gysel, J. Crosier, D. O. Topping, J. D. Whitehead, I. M. Ulbrich, A. P. Grieshop, A. L. Robinson, J. Duplissy,
K. N. Bower, M. J. Cubison, P. I. Williams, M. J. Flynn, J. D. Smith, K. R. Wilson, V. A. Lanz, C. Hueglin, Y. L. Sun,
G. B. McFiggans and H. Coe, Closure study between J. Tian, A. Laaksonen, T. Raatikainen, J. Rautiainen,
chemical composition and hygroscopic growth of aerosol P. Vaattovaara, M. Ehn, M. Kulmala, J. M. Tomlinson,
particles during TORCH2, Atmos. Chem. Phys., 2007, 7, D. R. Collins, M. J. Cubison, E. J. Dunlea, J. A. Huffman,
6131–6144. T. B. Onasch, M. R. Alfarra, P. I. Williams, K. Bower,
56 C. Mohr, P. F. DeCarlo, M. F. Heringa, R. Chirico, Y. Kondo, J. Schneider, F. Drewnick, S. Borrmann,
J. G. Slowik, R. Richter, C. Reche, A. Alastuey, X. Querol, S. Weimer, K. Demerjian, D. Salcedo, L. Cottrell, R. Griffin,
R. Seco, J. Peñuelas, J. L. Jimenez, M. Crippa, A. Takami, T. Miyoshi, S. Hatakeyama, A. Shimono,
R. Zimmermann, U. Baltensperger and A. S. H. Prévôt, J. Y. Sun, Y. M. Zhang, K. Dzepina, J. R. Kimmel,
Identication and quantication of organic aerosol from D. Sueper, J. T. Jayne, S. C. Herndon, A. M. Trimborn,
cooking and other sources in Barcelona using aerosol L. R. Williams, E. C. Wood, A. M. Middlebrook, C. E. Kolb,
mass spectrometer data, Atmos. Chem. Phys., 2012, 12, U. Baltensperger and D. R. Worsnop, Evolution of organic
1649–1665. aerosols in the atmosphere, Science, 2009, 326, 1525–1529.
57 M. Crippa, I. El Haddad, J. G. Slowik, P. F. DeCarlo, C. Mohr, 64 E. C. Wood, M. R. Canagaratna, S. C. Herndon, T. B. Onasch,
M. F. Heringa, R. Chirico, N. Marchand, J. Sciare, C. E. Kolb, D. R. Worsnop, J. H. Kroll, W. B. Knighton,
U. Baltensperger and A. S. H. Prévôt, Identication of R. Seila, M. Zavala, L. T. Molina, P. F. DeCarlo,
marine and continental aerosol sources in Paris using high J. L. Jimenez, A. J. Weinheimer, D. J. Knapp, B. T. Jobson,
resolution aerosol mass spectrometry, J. Geophys. Res.: J. Stutz, W. C. Kuster and E. J. Williams, Investigation of
Atmos., 2013, 118, 1950–1963. the correlation between odd oxygen and secondary organic
58 A. Setyan, Q. Zhang, M. Merkel, W. B. Knighton, Y. Sun, aerosol in Mexico City and Houston, Atmos. Chem. Phys.,
C. Song, J. E. Shilling, T. B. Onasch, S. C. Herndon, 2010, 10, 8947–8968.
D. R. Worsnop, J. D. Fast, R. A. Zaveri, L. K. Berg, 65 R. Chirico, A. S. H. Prevot, P. F. DeCarlo, M. F. Heringa,
A. Wiedensohler, B. A. Flowers, M. K. Dubey and R. Richter, E. Weingartner and U. Baltensperger, Aerosol
R. Subramanian, Characterization of submicron particles and trace gas vehicle emission factors measured in
inuenced by mixed biogenic and anthropogenic a tunnel using an Aerosol Mass Spectrometer and other
emissions using high-resolution aerosol mass on-line instrumentation, Atmos. Environ., 2011, 45, 1282–
spectrometry: results from CARES, Atmos. Chem. Phys., 2192.
2012, 12, 8131–8156. 66 C. J. Weschler and H. C. Shields, Experiments probing the
59 Q. Zhang, M. Rami Alfarra, D. R. Worsnop, J. D. Allan, inuence of air exchange rates on secondary organic
H. Coe, M. R. Canagaratna and J. L. Jimenez, aerosols derived from indoor chemistry, Atmos. Environ.,
Deconvolution and quantication of hydrocarbon-like and 2003, 37, 5621–5631.
oxygenated organic aerosols based on aerosol mass 67 M. S. Waring and J. A. Siegel, Indoor secondary organic
spectrometry, Environ. Sci. Technol., 2005, 39, 4938–4952. aerosol formation initiated from reactions between ozone
View Article Online
and surface-sorbed d -limonene, Environ. Sci. Technol., 2013, hygroscopicity and morphology during ageing of diesel
47, 6341–6348. soot, Environ. Res. Lett., 2011, 6, 1–10.
68 D. Rim, J. Il Choi and L. A. Wallace, Size-Resolved Source 73 A. G. Carlton and B. J. Turpin, Particle partitioning potential
Emission Rates of Indoor Ultrane Particles Considering of organic compounds is highest in the Eastern US and
Coagulation, Environ. Sci. Technol., 2016, 50, 10031–10038. driven by anthropogenic water, Atmos. Chem. Phys., 2013,
69 J. Rissler, E. Z. Nordin, A. C. Eriksson, P. T. Nilsson, 13, 10203–10214.
M. Frosch, M. K. Sporre, A. Wierzbicka, B. Svenningsson, 74 M. M. H. El-Sayed, Y. Wang and C. J. Hennigan, Direct
J. Löndahl, M. E. Messing, S. Sjogren, J. G. Hemmingsen, atmospheric evidence for the irreversible formation of
S. Lo, J. H. Pagels and E. Swietlicki, Effective density and aqueous secondary organic aerosol, Geophys. Res. Lett.,
mixing state of aerosol particles in a near-traffic urban 2015, 42, 5577–5586.

environment, Environ. Sci. Technol., 2014, 48, 6300–6308. 75 M. M. H. El-Sayed, D. Amenumey and C. J. Hennigan,
70 P. F. DeCarlo, J. G. Slowik, D. R. Worsnop, P. Davidovits and Drying-Induced Evaporation of Secondary Organic Aerosol
J. L. Jimenez, Particle Morphology and Density during Summer, Environ. Sci. Technol., 2016, 50, 3626–3633.
Characterization by Combined Mobility and Aerodynamic 76 S. Yousse and M. S. Waring, Indoor transient SOA
Diameter Measurements. Part 1: Theory, Aerosol Sci. formation from ozone+a-pinene reactions: Impacts of air
Technol., 2004, 38, 1185–1205. exchange and initial product concentrations, and
71 P. Massoli, E. C. Fortner, M. R. Canagaratna, L. R. Williams, comparison to limonene ozonolysis, Atmos. Environ., 2015,
Q. Zhang, Y. Sun, J. J. Schwab, A. Trimborn, T. B. Onasch, 112, 106–115.
K. L. Demerjian, C. E. Kolb, D. R. Worsnop and J. T. Jayne, 77 X. Chen and P. K. Hopke, A chamber study of secondary
Pollution gradients and chemical characterization of organic aerosol formation by limonene ozonolysis, Indoor
particulatematter from vehicular traffic near major Air, 2010, 20, 320–328.
roadways: Results from the 2009 queens college air quality 78 Å. M. Jonsson, M. Hallquist and E. Ljungström, Impact of
study in NYC, Aerosol Sci. Technol., 2012, 46, 1201–1218. humidity on the ozone initiated oxidation of limonene, D3-
72 T. Tritscher, Z. Jurnyi, M. Martin, R. Chirico, M. Gysel, carene, and a-pinene, Environ. Sci. Technol., 2006, 40, 188–
M. F. Heringa, P. F. DeCarlo, B. Sierau, A. S. H. Prévt, 194.
E. Weingartner and U. Baltensperger, Changes of

Seasonal variation in aerosol composition

Uploaded by

Copyright:

Available Formats

Seasonal variation in aerosol composition

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Seasonal variation in aerosol composition

Uploaded by

Copyright:

Available Formats

Environmental

Seasonal variation in aerosol composition and

Cite this: DOI: 10.1039/c8em00471d

Environmental Science: Processes & Impacts Paper

Paper Environmental Science: Processes & Impacts

Environmental Science: Processes & Impacts Paper

species. The ventilation rate is dened as the rate of exchange of

Paper Environmental Science: Processes & Impacts

ALW ¼ VS(GFmix3 1) (6) indoor RH is expected from thermally well-controlled environ-

Environmental Science: Processes & Impacts Paper

correlates very well with previously published wintertime

summer than winter.

Paper Environmental Science: Processes & Impacts

Environmental Science: Processes & Impacts Paper

Paper Environmental Science: Processes & Impacts

Environmental Science: Processes & Impacts Paper

Paper Environmental Science: Processes & Impacts

Environmental Science: Processes & Impacts Paper

Paper Environmental Science: Processes & Impacts

Environmental Science: Processes & Impacts Paper

Paper Environmental Science: Processes & Impacts

and environments. However, the range of values in summer is

Environmental Science: Processes & Impacts Paper

indoor and outdoor air quality. The outdoor seasonal analysis

Paper Environmental Science: Processes & Impacts

10165. size-dependent removal mechanisms, Environ. Sci.

Environmental Science: Processes & Impacts Paper

Paper Environmental Science: Processes & Impacts

L. Hildebrandt, R. Hillamo, J. L. Jimenez, A. Laaksonen, 60 M. R. Canagaratna, J. T. Jayne, D. A. Ghertner, S. Herndon,

Environmental Science: Processes & Impacts Paper

mixing state of aerosol particles in a near-traﬃc urban 2015, 42, 5577–5586.

You might also like