Atmospheric Research 218 (2019) 257–268

Contents lists available at ScienceDirect

Atmospheric Research
journal homepage: www.elsevier.com/locate/atmosres

Temporal distribution and source apportionment of PM2.5 chemical T

composition in Xinjiang, NW-China
Yusan Turapa, Dilinuer Talifua, , Xinming Wangb, Abulikemu Abulizia, Mailikezhati Maihemutia,
⁎

Yalkunjan Tursuna, Xiang Dingb, Tuergong Aierkena, Suwubinuer Rekefua

a
Key Laboratory of Coal Clean Conversion & Chemical Engineering Process, College of Chemistry and Chemical Engineering, Xinjiang University, Urumqi 830046, China
b
State Key Laboratory of Organic Geochemistry, Guangzhou Institute of Geochemistry Chinese Academy of Sciences, Guangzhou 510640, China

ARTICLE INFO ABSTRACT

Keywords: Daily fine particulate matter samples were collected in Dushanzi district within four months from September
Fine particulate matter 2015 to August 2016 and represent the four seasons. The samples were determined for major chemical com-
Chemical composition ponents in PM2.5, including elements, water-soluble ions (WSIs) and the organic/elemental carbon (OC/EC). The
Hysplit trajectory model results indicated that the annual mean PM2.5 concentration was 62.85 ± 43.5 μg m−3 in the Dushanzi district,
Sources apportionment
with the highest seasonal average in winter (95.47 ± 61.7 μg m−3) and the lowest in summer
(33.22 ± 17.7 μg m−3). The crustal elements were the most abundant elements and accounted for 96.51% of
the total analyzed elements. Carcinogenic metals, such as Cr, Pb, As and Cd, originated from human activity,
especially during winter. The highest total WSI concentration was 68.99 μg m−3 in winter, followed by autumn
(16.32 μg m−3), spring (10.23 μg m−3) and summer (7.06 μg m−3). SO42−, NO3− and NH4+ were the most
abundant WSIs in Dushanzi. Ion balance calculations showed that PM2.5 in winter was acidic; in autumn and
spring alkaline; and in summer nearly neutral. Total carbonaceous aerosol (TCA) accounted for 34% of the
PM2.5. The chemical mass closure (CMC) indicated that minerals and WSIs were the major fraction, accounting
for 33.58% and 23.17% of PM2.5 mass concentration, respectively. Dushanzi was controlled by four major air
masses, and the relative contributions of these air masses differ by season. Positive matrix factorization (PMF)
analysis identified six sources including vehicle emission, biomass burning, coal combustion, industrial pollu-
tion, secondary aerosols and soil dust, with annual mean contributions of 9.43%, 10.86%, 18.45%, 12.15%,
18.26% and 30.85%, respectively. Moreover, the relative contributions of these identified sources varied sig-
nificantly with the changing seasons.

1. Introduction formation mechanism of PM2.5 in some megacities of China, such as

Atmospheric particulate matter (PM) pollution has become a serious
environmental problem, especially in developing countries, owing to its
severe threat to human health and climate stability (Cesari et al., 2016;
Ming et al., 2017). Particularly, airborne PM2.5 (mean aerodynamic
diameter ≤ 2.5 μm) is extremely harmful because the tiny particles are
composed of various carcinogenic chemicals that can enter into the
human respiratory system and even penetrate into the circulatory
system, leading to cell dysfunction, inflammation and cardio-pul-
monary disease (Semmler et al., 2008; Yu et al., 2016). Numerous
epidemiological and toxicological studies found that 3.45 million pre-
mature deaths worldwide were related to PM2.5 pollution in 2007
(Zhang et al., 2017).
Earlier studies, chemical compositions, potential sources and the
Beijing, Nanjing, Shanghai and Guangzhou, have been investigated
(Huang et al., 2014; Ming et al., 2017; Wang et al., 2006). However,
very few investigations of PM2.5 have been conducted in northwest
China, especially in the Xinjiang Uygur Autonomous Region. Xinjiang is
located in the central post of the largest continent, the Eurasian con-
tinent; it is an important strategic area of the Belt and Road and is going
through rapid economic and industrial development. In the last decade,
Xinjiang has suffered from a high level of air pollution. The source of
PM in Xinjiang gradually shifted from dust storms to a mixture of ve-
hicle emissions, industrial emissions, biomass combustion and dust
storms.
Dushanzi district is an enclave of Karmay city, which is a central city
in Xinjiang and is in an upstream area of Asia dust. It is located in the
northwestern margin of China's second largest basin, Junggar basin,

⁎
Corresponding author.
E-mail address: dilnurt@xju.edu.cn (D. Talifu).

https://doi.org/10.1016/j.atmosres.2018.12.010
Received 4 February 2018; Received in revised form 14 November 2018; Accepted 10 December 2018
Available online 11 December 2018
0169-8095/ © 2018 Elsevier B.V. All rights reserved.
Y. Turap et al.
with a total residential area of 26 km2 and a population of 90 thousand;
most of the area is covered by desert and Gobi (Li et al., 2016).
Dushanzi district faces the Gurbantongute desert in the east and Ta-
kelamakan desert in the south. The geomorphological type of the
Dushanzi district is predominantly the Gobi desert; the district has a
typical temperate continental arid desert climate. The ratio of eva-
poration to precipitation of rainwater is > 27.6 (Pan et al., 2012). The
poor ecological environment is uniquely susceptible to atmospheric
pollution. In addition, Dushanzi district is an industrial area char-
acterized by large-scale petrochemical oil refinery industries and power
plants. In recent years, Dushanzi has undergone rapid economic de-
velopment and population growth, which have increased the emissions
of air pollutants, thus raising widespread health concerns in local re-
sidents. However, there have been few reports on air pollutants in
Dushanzi. Systematic knowledge of chemical composition and sources
apportionment of PM2.5 is still lacking for Dushanzi compared to other
cities.
The objectives of this study were to elucidate the seasonal variations
and potential sources of PM2.5 by continuously collecting the daily
PM2.5 samples across Dushanzi for four months. The primary aims of
this study are to critically: (1) identify the chemical species in the PM2.5
samples in an arid and petrochemical industrialized area of Xinjiang;
(2) analyze the seasonal variations of the chemical components and
source contributions; and (3) quantitatively assess the potential sources
of PM2.5 using a PMF model coupled with chemical mass closure and a
Hysplit Trajectory Model. The results gained from this study will be
useful for understanding the temporal distribution of PM2.5 chemical
compounds in northwest China and will contribute to the development
of more effective control strategies for PM2.5 in arid and petrochemical
areas.
2. Materials and methods
2.1. PM2.5 sampling
Air samples were collected during typical periods in a residential
and commercial area of the Dushanzi district (44°19′33″N, 84°53′0″E)
(Fig. 1), which was located in the southwestern part of the petro-
chemical refinery industries and thermal power plant. The sampling
Fig. 1. Location of the sampling site.
258
Atmospheric Research 218 (2019) 257–268
equipment was installed on the roof of a residential building (20 m
above ground level). The topography around the site was flat and open.
PM2.5 was collected by a high-volume sampler (Wuhan Tianhong
Instruments Co., Ltd., TH-1000) at a controllable flow rate of
1.05 m3 min−1. The filters were changed every 22 h (from 10:00 a.m. to
8:00 a.m. the next day). Field blank filters were also collected without
ambient air passing through the instrument during the sampling per-
iods. In total, 83 PM2.5 samples were collected during the following
periods: September 14 to October 5, 2015 (autumn, 20 samples), De-
cember 5 to 28, 2015 (winter, 22 samples), April 19 to May 16, 2016
(spring, 22 samples), and July 1 to 22, 2016 (summer, 23 samples).
PM2.5 samples were collected on quartz-fiber filters (Whatman,
203 × 254 mm), which were pre-heated at 450 °C for 4 h before use,
with the aim to reduce the residual carbon levels that were associated
with the filters. The filters were then equilibrated at a temperature
between 20 ± 2 °C, and the relative humidity was between 35% and
45% for 48 h. Each filter was wrapped in Al-foils, sealed in polyethylene
bags and stored in a refrigerator at −20 °C until analysis.
2.2. Chemical analysis
The weighing of simples and blank filters were performed at least
three times both before and after sampling using an electronic micro-
balance (Metller Toledo AB204-S) with a precision of 0.1 mg, and the
differences of the filter weights were divided by the sampling volume to
obtain PM2.5 mass concentrations.
An area of 5.06 cm2 of each filter, including a blank sample, was cut
to analyze elements. First, each filter was soaked in a mixture acid
(4 mL 67–70% HNO3 and 1 mL 47–51% HF) and was extracted by mi-
crowave digestion (XT-9912, Shanghai). Then, the sample was heated
by an electric stove and was evaporated until 1 mL of residual was left.
After cooling to room temperature, the sample was transferred to a test
tube, and 12 mL ultrapure water was added. Subsequently, inductively
coupled plasma-mass spectrometry (ICP-MS) was used to determine the
concentrations of the elements. The major elements including Na, Mg,
Al, K, Ca, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Rb, Sr, Cd, Ba, Tl and Pb
were measured. The method detection limit (MDL) of the elements
ranged from 0.1 to 1 ng m−3, and the uncertainties were < 5%.
A small portion of each filter (area of 5.06 cm2) was cut to analyze the
Y. Turap et al.
water-soluble ions. The filters were extracted ultrasonically using approxi-
mately 15 g ultrapure water (specific resistivity =18.5 MΩ cm−1) for
20 min and were centrifugalized for 10 min. Then, 4 mL supernatant solu-
tion was taken in a clean bottle. This process was carried out twice. Next,
1 mL solution was drawn out by using a syringe and was injected into the
ion chromatograph (BIC-plus-883). Before analysis of the water-soluble
ions, the standard solutions were detected three times and low relative
standard deviations were observed. Finally, water-soluble ions were iden-
tified by the solution flowing through the cation and anion columns. Field
blanks were used to correct the concentrations of ionic species. The detec-
tion limits of these ions were within the range of 0.01 to 0.03 μg m−3, and
the uncertainties were < 5%.
A 0.5 cm2 punch area from each filter was measured by DRI Model
2015 (thermal/optical carbon analyzer) for OC and EC, following the
Interagency Monitoring of Protected Visual Environment (IMPROVE).
The analysis consisted of two stages: In the first stage, a non-oxidizing
helium atmosphere of four OC fractions were produced. In the second
stage, the samples were heated in helium containing 2% oxygen, which
yielded three EC fractions. In this process, pyrolyzed carbon was de-
termined by the reflected laser light and returned to the initial value
after oxygen was added to the analysis atmosphere. The detection limits
were 0.18 ± 0.06 and 0.04 ± 0.01 μg m−3 for OC and EC, respec-
tively, with uncertainties < 5%.
2.3. Quality assurance and quality control
All analytical procedures were monitored with strict quality assur-
ance (QA) and quality control (QC) measures. Field and laboratory
blanks were analyzed in the same manner as the filter samples. Both
pre- and post-sampling and component analysis strictly guaranteed that
the quartz fiber filter was intact and undamaged, and the cracked filters
were excluded. The background contamination was regularly mon-
itored by blank tests, which were used to validate and correct the
corresponding data. Blanks and duplicate sample analyses were con-
ducted for nearly 10% of the samples. Certified reference materials
(CRMs, produced by National Research Center for Certified Reference
Materials, China) were used for QA/QC. The pretreatment procedure,
chemical analysis and QA/QC were described in detail in other related
studies (Liu et al., 2016; Zhang et al., 2015; Zhao et al., 2013).
2.4. Chemical mass closure
In this study, PM2.5 mass was reconstructed by taking into account
SO42−, NO3−, NH4+, particulate organic matter (POM), EC, trace ele-
ment oxide (TEO) and mineral dust. SO42−, NO3−, and NH4+ were the
major WSIs and their concentrations were directly measured (Chen
et al., 2017; Tao et al., 2013). The POM mass was estimated by the
organic carbon (OC) mass concentration multiplied by a conversion
factor of 1.6, as in the following equation:
[POM] = 1.6 × OC (1)
The factor has been adopted in many other investigations, because
the latest result shows an OM/OC ratio that averaged at 1.59 ± 0.18 in
PM2.5 across the China (Wang et al., 2016; Zhang et al., 2013).
The mineral dust component is often estimated through several
formulas (Zhang et al., 2013). A straightforward method conventionally
used in estimating dust aerosols from Al was used in the present study,
as follows:
[Mineral] = Al/0.07 (2)
where 0.07 is the average Al content (7%). A similar estimation had
been applied previously (Hsu et al., 2010; Zhang et al., 2013).
For TEO, the contribution of heavy metals as metal oxides was es-
timated by the following equation (Chen et al., 2017):
259
Atmospheric Research 218 (2019) 257–268
TEO = 1.3 × [0.5 × (Sr + Ba + Mn + Co + Rb + Ni + V)
+ 1.0 × (Cu + Zn + Mo + Cd + Tl + Pb + As + Se)] (3)
2.5. Sources identification methods
Backward trajectories with a Hybrid Single Particle Lagrangian
Integrated Trajectory (HYSPLIT) model were used to determine the
transport trajectory of air parcels at the sampling site. We calculated
48 h air mass back trajectories arriving at sampling site during the
sampling period by using the National Oceanic and Atmospheric
Administration HYSPLIT 4 (Li et al., 2017). The six hourly final archive
data were generated from the National Center for Environmental Pre-
diction's Global Data Assimilation System (GDAS) wind field reanalysis.
More details about the HYSPLIT model can be found at https://ready.
arl.noaa.gov/archives.php (NOAA Air Resources Laboratory). The
model was run four times per day at starting times of 04:00, 10:00,
16:00, and 22:00 UTC (12:00, 18:00,00:00, and 06:00 LT - local time,
respectively). The arrival level was set at 100 m above ground level
(AGL).
The PMF, which was developed by the Environmental Protection
Agency, is an effective source apportionment receptor model that does
not require source profiles prior to analysis and has no limitation on the
numbers of source (Lee et al., 2003; Zhang et al., 2013). The principles
of PMF can be found elsewhere, in detail (Chen et al., 2017; Hsu et al.,
2016; Paatero and Tapper, 1994; Yu, 2013). In this study, PMF 5.0 was
employed to measure components, including: PM2.5, OC, EC, Cl−,
NO3−, SO42−, NH4+, Na,Mg, Al, K, Ca, V, Cr, Mn, Fe, Co, Ni, Cu, Zn,
As, Rb, Sr, Cd, Ba, Tl and Pb. WSIs such as Na+, K+, Mg2+ and Ca2+
were excluded due to their duplication in elemental analysis. The un-
certainty (Unc) is calculated by using the following equation, which is
based on the error fraction and the method detection limit (MDL). If the
concentration is greater than the MDL provided, the calculation is based
on Eq. (4), by contrast, if it is less than or equal to MDL, the Unc is
calculated using Eq. (5). In this study, all species were categorized as
strong due to the signal–to-noise ratio (S/N) > 2, and PM2.5 was spe-
cified as a total variable (Wang et al., 2016).
Unc = (Error Fraction × Concentration) 2 + (0.5 × MDL)2 (4)
Unc = 5 6 MDL (5)
3. Results and discussion
3.1. The characteristics of PM2.5 and chemical compositions
3.1.1. PM2.5 mass concentration
Continuous meteorological parameters, including wind speed, wind
direction, temperature (T) and the relative humidity (RH), were re-
corded simultaneously at the sampling site, which is also the environ-
mental observation station that was set by the Dushanzi environmental
monitoring station. Fig. 2 shows the time series of the meteorological
parameters during the sampling period. The annual wind speed was
0.73 m s−1, with the highest wind speed occurring in spring
(0.96 m s−1) and the lowest occurring in winter (0.39 m s−1). The
temperature was highest in summer (26.51 °C) and lowest in winter
(−8.7 °C); in contrast, the relative humidity was highest in winter
(93.10%) and lowest in summer (43.51%). The annual average con-
centration of PM2.5 was 62.85 ± 43.5 μg m−3 at Dushanzi, which was
a factor of 2 higher than the Chinese National Ambient Air Quality
Standards' (NAAQS) Grade II of 35 μg m−3, implying serious PM2.5
pollution in Dushanzi. The seasonal variations were evident: PM2.5 level
was highest in winter (95.47 ± 61.7 μg m−3), followed by autumn
(70.22 ± 25.3 μg m−3), spring (56.14 ± 37.1 μg m−3) and summer
(33.22 ± 17.7 μg m−3) (Fig. 2 and Table 1). The seasonality of PM2.5
Y. Turap et al. Atmospheric Research 218 (2019) 257–268

Fig. 2. Time series of meteorological records and PM2.5 during the sampling time.

in Dushanzi was similar to the typical pollution character of northwest
China, with the maxima in winter and the minima in summer (Chen
et al., 2017; Wang et al., 2016). The difference in the mean con-
centrations between winter and summer was 62.45 μg m−3, which
suggested large seasonal variability. This probably was caused by a
combination of factors including favorable wind direction and stronger
atmospheric vertical mixing in the summer and higher emissions from
coal-fired thermal plants and stronger stable atmospheric boundary
layer in the winter.
The annual mean concentration of PM2.5 in Dushanzi was approxi-
mately 32% lower than that in Lanzhou (93.7 μg m−3) (Wang et al.,
2016) and 49% lower than in Delhi (122 μg m−3) (Sharma et al., 2016)
because the lower humidity (the annual average humidity was 57%)
does not favor the growth of aerosol particuls and PM2.5 formation. The
electrical low-pressure impactor (ELPI) and scanning electron micro-
scope (SEM) measurements showed that the atmospheric particulates
were an amorphous solid state at low humidity (lower than 70%) and
that the chemical reactions were impeded in glassy or highly viscous
aerosol particles because the mass transport of reactants within the
aerosol particle bulk became a rate-limiting step. For the same reasons,
the chemical reactions in solid particles were typically surface-limited;
that is, heterogeneous reactions occurred only on the particle surfaces
(Maria et al., 2004; Virtanen et al., 2010).
However, 62.85 ± 43.5 μg m−3 was higher than that of most mega-
cities in southern China that are located in the subtropical monsoon climate
zone, such as Guangzhou (42.4 μg m−3), Zhuhai (33.8 μg m−3), and
Shenzhen (32.0 μg m−3) (Ming et al., 2017), as well as that of other
countries, such as Korea (Choi et al., 2012), and Greece (Samara et al.,
2014). The higher PM2.5 concentration was likely caused by the high level
of precursor gas from regional industrial (such as power plants and petro-
chemical oil refinery industries) and transportation sources. The other
reason was probably that high frequency of dust storms could have con-
tributed markedly to the PM2.5 concentration in Dushanzi due to the
southwest and northwest wind (Fig. 2) brought dust from the Gobi, Gur-
bantongute and Takelamakan deserts (Fig. 1) to Dushanz. The composition
and morphology of the dust particles are subject to transformation by ad-
sorbing gaseous species, surface reactions, and coagulation with other
particles during transport and will lead to high levels of geological aerosols;
the same results are demonstrated in other studies (Di et al., 2016; Wang
et al., 2016; Wei et al., 2005; Zhang et al., 2015). Therefore, local emissions
and poor ecological conditions (most of the area is covered by the desert
and Gobi) could both play important roles in the PM2.5 level in Dushanzi.

Table 1
The seasonal and annual mean concentration of PM2.5 and metal elements.
Species Autumn(n = 19) Winter(n = 21) Spring(n = 21) Summer(n = 22) Annual(n = 83)

μg m−3
PM2.5 70.22 ± 25.3 95.47 ± 61.7 56.14 ± 37.1 33.22 ± 17.7 62.85 ± 43.5
Na 1.54 ± 1.0 3.61 ± 1.5 2.09 ± 0.2 3.12 ± 1.1 1.67 ± 0.8
Mg 0.85 ± 0.5 0.55 ± 0.2 1.12 ± 0.6 0.88 ± 0.3 0.84 ± 0.5
Al 3.08 ± 1.7 0.51 ± 0.3 3.24 ± 1.6 1.59 ± 0.8 2.1 ± 1.2
K 1.46 ± 0.9 0.26 ± 0.1 1.14 ± 1.0 0.56 ± 0.3 0.78 ± 0.4
Ca 2.25 ± 1.1 0.41 ± 0.2 1.98 ± 1.7 1.33 ± 0.9 1.51 ± 1.3
Fe 2.24 ± 1.2 0.13 ± 0.8 2.97 ± 1.6 1.21 ± 0.8 1.45 ± 1.4

ng m−3
V 6.39 ± 3.8 0.59 ± 0.3 5.58 ± 3.8 3.12 ± 2.2 4.02 ± 3.0
Cr 5.28 ± 2.8 3.17 ± 2 11.47 ± 8.6 9.4 ± 6.2 7.26 ± 5.7
Mn 58.84 ± 39.4 14.58 ± 9.9 59.84 ± 43.6 39.7 ± 22.6 44.05 ± 37.2
Co 0.92 ± 0.5 0.47 ± 0.3 0.83 ± 0.7 0.52 ± 0.3 0.7 ± 0.5
Ni 10.44 ± 9.6 1.29 ± 0.8 17.61 ± 14.8 16.34 ± 12.6 12.11 ± 9.9
Cu 6.73 ± 2.6 6.22 ± 4.1 5.97 ± 3.0 4.24 ± 3.6 6.07 ± 3.5
Zn 50.91 ± 28.5 24.73 ± 12.7 55.31 ± 30.1 21.45 ± 10.5 38.63 ± 24.8
As 1.91 ± 0.8 4.14 ± 2.6 1.56 ± 0.7 1.48 ± 1.1 2.39 ± 1.9
Rb 5.79 ± 3.4 1.02 ± 0.7 4.8 ± 3.4 2.22 ± 1.8 3.58 ± 2.1
Sr 29.88 ± 15.8 3.48 ± 1.8 27.95 ± 25.3 17.04 ± 11.1 20.85 ± 16.4
Cd 0.27 ± 0.1 0.35 ± 0.1 0.29 ± 0.1 0.11 ± 0.1 0.24 ± 0.1
Ba 37.6 ± 20.8 3.84 ± 2.0 30.09 ± 21.3 17.4 ± 11.0 22.75 ± 18.2
Tl 0.25 ± 0.1 0.18 ± 0.1 0.11 ± 0.1 0.16 ± 0.1 0.15 ± 0.1
Pb 15.59 ± 8 17.51 ± 10.1 14.54 ± 6.2 15.28 ± 11.3 15.28 ± 10.5
SUM/PM2.5 0.16 ± 0.06 0.04 ± 0.03 0.21 ± 0.05 0.23 ± 0.06 0.16 ± 0.09

260
Y. Turap et al.
3.1.2. Metal elements
The annual and seasonal mean concentrations of airborne metal ele-
ments are listed in Table 1. For the annual mean concentration, the ele-
ments contributed to 16.01 ± 9.0% of the PM2.5 mass, higher than
Lanzhou (8.8 ± 2.9%) (Wang et al., 2016) and Heze (12.5%) (Liu et al.,
2017). The most abundant elements are crustal elements (e.g., Na, Mg, Al,
K, Ca and Fe), which account for 96.51 ± 1.3% of the total analyzed
elements, and this percentage was higher than that found for Lanzhou
(93.6 ± 4.5%) (Wang et al., 2016), and Xinxiang (90.8%) (Feng et al.,
2016). Crustal elements waxed in the spring and autumn and waned in the
winter. Such levels of seasonal mean concentrations and seasonality are
consistent with those that were observed in previous studies (Wang et al.,
2016; Xu et al., 2016), and are related to dust storms and anthropogenic
fugitive dust. The trace elements (V, Cr, Mn, Ni, Cu, Zn, As, Rb, Sr, Ba and
Pb) are present at levels between 2.4 and 44.0 ng m−3. The seasonal levels
of elements decreased in the following order: nearly equal in spring
(11.68 ± 8.8 μg m−3) and autumn (11.63 ± 6.0 μg m−3) > summer
(9.54 ± 3.4 μg m−3) > winter (2.51 ± 1.0 μg m−3), and contributing to
the PM2.5 mass concentrations were 21.12 ± 5.0%, 23.21 ± 6. 1%,
16.06 ± 6.0% and 4.14 ± 3.2% for spring, summer, autumn and winter,
respectively. Shi et al., 2017 reported that the crustal elements mainly
originated from re-suspension, and snow can reduce the re-suspension. In
this study, the wind speed was the lowest in winter (Fig. 2), and Na and
Cocker, 2009 reported that dry, hot weather could facilitate re-suspension
of soil-related material by wind; and crustal element concentrations were
higher by approximately a factor of 2 for September and October compared
with December and January. Therefore, the re-suspension was reduced and
the concentrations of crustal elements were the lowest in winter. However,
heavy metals including As, Pb, Cu and Zn were the highest in winter due to
coal burning and thermal power plants, which demonstrated that coal
burning was an important source of As and Pb (Wang et al., 2016). The
other trace elements in PM2.5 in Dushanzi can be traced back to vehicle
exhaust, industrial emissions and dust storms (due to the large standard
deviation).
Enrichment factor (EF) is a quantity that distinguishes anthro-
pogenic sources from natural process and assesses the degree of ele-
ments pollution; it is estimated by the Eqs (Hsu et al., 2016; Shao et al.,
2016).
EF = (Ci /Cr )sample /(Ci /Cr )crust (6)
where (Ci/Cr)sample represent the ratio of the target element to the re-
ference element mass concentration in aerosol and (Ci/Cr)crust represent
the ratio of the target element to the reference element in the upper
continental crust (Clemens Reimann and Caritat‡, 2000; Taylor and
Mclennan, 1995). The crustal element concentration was extracted
from previous studies that investigated a region near the Dushanzi
district (Luo et al., 2012; Zhao and Fan, 1994). The typical reference
elements include Al, Fe, Ti and Mg, and there is still no well-established
rule for the choice of a reference element (Khare and Baruah, 2010). In
this study, Al was chosen for the reference element due to its abundance
in the earth's crust and its stability to most anthropogenic contaminants
(Hsu et al., 2016).
Fig. 3 shows the seasonal EF values; elements such as Al, Fe, Mn, Rb,
K, Ga were lower than 10 during the sampling period, which indicated
that these elements originated mostly from mineral dust. In contrast, in
autumn, the EF values for Tl, Zn, Pb and Cd were above 10. The EF
values were higher than 10 in winter for Cr, Cu, Tl, Zn, As Pb and Cd, in
spring for Cr, Pb, Ni, Zn and Cd and in summer for As, Tl, Ni, Cu, Cr, Zn,
Pb and Cd. On average, trace metals such as Zn, Pb and Cd have high EF
values in all seasons, which suggested that they were mostly from an-
thropogenic sources (Hsu et al., 2016). Generally, EF values could be
used to source tracers (Hsu et al., 2016). For instance, high EF values of
Cr, Pb, As and Cd were associated with coal combustion, high levels of
Cu and Zn have demonstrated a link to vehicle emission and abundant
Ni and V originated from oil combustion (Zhai et al., 2014; Zhao et al.,
2015). Moreover, Li et al., 2008 collected the surface soil samples
261
Atmospheric Research 218 (2019) 257–268
(which could have occurred from the deposition during the early years)
on the ground of Xinjiang (including road dust and the agriculture
field), which could be transported into the atmosphere, and found that
the surface soil contains a higher concentration of Pb, As and Cd
compared to their crustal abundance, which indicated that these ele-
ments in the aerosols might come not only from the re-suspended road
dust but also the from outside transported soil from the Jungger Basin.
In this study, trace elements including Cd, Pb, As and Zn, have the
highest EF values in each season, especially in winter (276.30, 129.81,
57.72 and 54.24), which was probably due to coal combustion for
domestic heating promoted the emission of metals (such as Pb, As, Sb
and Cr) in winter. In other seasons, it might have originated from
mixing the local anthropogenic aerosol with the transported soil dust.
Hence, these elements in PM2.5 in Dushanzi were related to human
activities, such as coal combustion, vehicle emission, biomass burning
and industry, and were also impacted by dust particles transported from
the desert.
3.1.3. Water-soluble ions
Table 2 shows the mean concentrations of the WSIs in four seasons
in Dushanzi. The mass concentrations of them followed the order
SO42− > NO3− > NH4+ > Ca2+ > Na+ > Cl− > K+ > Mg2+
during all seasons, and the total WSIs accounted for 20.12%, 64.20%,
21.17% and 16.18% of PM2.5 in autumn, winter, spring and summer,
respectively. It is worth noting that the WSIs contributions in winter
were approximately 3 times higher than that in other seasons. This
might be in north China where coal-based heating systems and others
combustion in winter have become the dominant sources of SO2, NOx
and primary PM (Song et al., 2015), and the domestic heating in Xin-
jiang occurs earlier (usually from 15 October to 15 March the following
year) than in other cities in China, and the six months of heating ac-
cumulated the precursor gas of PM. The stable atmosphere (low wind
speed) and low temperature (Fig. 2) in winter might also increase the
nucleation and growth, as well as accelerate the adsorption of aerosol in
Dushanzi, these results were consistent with those of earlier studies
(Chen et al., 2017; Ming et al., 2017; Zhang et al., 2013).
Among these, WSIs such as NH4+, NO3− and SO42− were the
dominant species and accounted for 78.55 ± 13.3% of the total ionic
concentration in Dushanzi. The annual mean concentration of NH4+,
NO3− and SO42+ were 5.03 ± 8.9, 6.75 ± 10.3 and
11.94 ± 21.1 μg m−3, respectively, which constituted 5.67 ± 6.0%,
8.07 ± 7.6% and 14.0 ± 12.6% of the PM2.5 mass respectively, which
was lower than in Beijing (Zhang et al., 2013), Shanghai (Ming et al.,
2017) and Chongqing (Chen et al., 2017). However, the levels of NH4+,
NO3− and SO42− in winter were factors of 8 to 10 higher than those in
other seasons. Other WSIs, such as Ca2+, Mg2+ and Na+, were ob-
served in higher concentrations in spring and autumn than in other
seasons, due to dust storms frequently occurring in spring and autumn
in Dushanzi (Wang et al., 2016; Wei et al., 2005; Xu et al., 2016). The
highest K+ and Cl− concentrations were observed in the autumn,
which was likely due to agricultural burning after the harvest around
Dushanzi, which was also found by some previous studies in Beijing (He
et al., 2017; Zhang et al., 2013). Generally, the NO3−/SO42− ratios
were used as a relative measure of the importance of mobile versus
stationary pollution emission sources in many studies (Xu et al., 2016).
The high NO3−/SO42− ratio was found in vehicle-predominant regions.
However, lower ratios < 1 have been found, usually in China, because
of the wide use of sulfur-containing coal (Feng et al., 2009). In the
present study, the NO3−/SO42− ratios ranged from 0.14 to 1.12, and
the annual mean ratio was 0.61 ± 0.4, which was lower than that in
cities with heavy traffic density, such as Guangzhou (0.79) (Tan et al.,
2009), but higher than that in coal-dominated cities, such as Guiyang
(0.14) and Qingdao (0.35) (He et al., 2017), suggesting that the
stronger influences come from stationary pollution emission sources,
such as coal combustion; however, the mobile sources should not be
ignored.
Y. Turap et al. Atmospheric Research 218 (2019) 257–268

Fig. 3. Enrichment factors of elements in four seasons.

Understanding the acidity of the PM is significantly important due

to it being a key parameter that affects heterogeneous reactions, hy-
groscopic growth, and toxicity of atmospheric PM, as well as the neu-
tralizing process of acid rain (Shi et al., 2017; Tao et al., 2013).
Ion balance calculations were useful for studying the acid-base
balance of aerosol. The ratios of AE (anion equivalent) and CE (cation
equivalent) could also be used to indicate the acidity of atmospheric
aerosol. The calculations of particulate anion and cation equivalent are
as follows (Shi et al., 2017; Tao et al., 2013):

AE = [NO3 ]/62 + [SO4 2 ]/48 + [Cl ]/35.5 (7)

CE = [NH 4+]/18 + [Ca2 +]/20 + [K+]/39 + [Mg2 +]/12 + [Na+]/23 (8)

where [NO3−], [SO42−], [Cl−], [NH4+], [Ca2+], [K+], [Mg2+] and

[Na+] represent the mass concentrations of these ion species in the
PM2.5 samples. Previous studies on an AE/CE ratio that was larger than
1 had been used to indicate the acidic particle condition, and a ratio
lower than 1 suggests an alkaline condition (Shi et al., 2017; Tao et al.,
2013), Fig. 4. Scatter plots of AE (anion equivalent) vs CE (cation equivalent) in four
In the present study, strong correlations (R2 = 0.98) between AE seasons.
and CE were found for all of the samples (Fig. 4), and the average AE/

Table 2
The seasonal and annual mean concentration of water-soluble ions (μg m−3).
Species Autumn(n = 19) Winter(n = 21) Spring(n = 21) Summer(n = 22) Annual(n = 83)

Na+ 0.48 ± 0.2 0.51 ± 0.3 0.34 ± 0.2 0.34 ± 0.2 0.47 ± 0.5
NH4+ 1.73 ± 1.2 15.32 ± 12.1 1.15 ± 0.8 0.47 ± 0.2 5.03 ± 8.9
K+ 0.30 ± 0.2 0.17 ± 0.2 0.11 ± 0.1 0.10 ± 0.02 0.17 ± 0.1
Mg2+ 0.13 ± 0.1 0.02 ± 0 0.08 ± 0.01 0.06 ± 0.03 0.07 ± 0.1
Ca2+ 2.02 ± 0.9 0.09 ± 0.06 1.27 ± 1.0 0.74 ± 0.5 1.02 ± 1
Cl− 0.75 ± 0.6 0.05 ± 0.1 0.25 ± 0.3 0.10 ± 0.1 0.29 ± 0.5
NO3− 4.45 ± 3.5 17.97 ± 14.1 2.47 ± 2.1 0.58 ± 0.3 6.75 ± 10.3
SO42− 4.38 ± 1.8 33.63 ± 31.5 4.25 ± 2.6 2.47 ± 0.8 11.94 ± 21.1
NO3−/SO42− 0.92 ± 0.4 0.66 ± 0.3 0.65 ± 0.5 0.24 ± 0.1 0.61 ± 0.4
SUM/PM2.5 0.20 ± 0.04 0.64 ± 0.2 0.21 ± 0.1 0.16 ± 0.03 0.31 ± 0.2

262
Y. Turap et al. Atmospheric Research 218 (2019) 257–268

Table 3
The seasonal and annual mean concentration of carbonaceous species in PM2.5.
μg m−3 Autumn(n = 18) Winter(n = 19) Spring(n = 18) Summer(n = 18) Annual(n = 73)

OC 5.67 ± 2.1 8.76 ± 4.3 6.46 ± 2.6 7.53 ± 3.0 7.13 ± 3.3
EC 1.73 ± 0.8 1.37 ± 0.6 0.90 ± 0.4 0.97 ± 0.5 1.25 ± 0.7
TCA 10.81 ± 3.9 15.39 ± 7.2 11.23 ± 4.4 13.01 ± 5.0 12.55 ± 5.6
OC/EC 3.48 ± 0.8 6.27 ± 1.2 7.47 ± 1.5 8.45 ± 2.5 6.44 ± 2.5
SOC 0.98 ± 0.8 3.37 ± 2.1 1.87 ± 1.0 2.19 ± 1.2 2.0 ± 1.6
SOC/OC 0.25 ± 0.1 0.35 ± 0.2 0.30 ± 0.1 0.32 ± 0.2 0.31 ± 0.2
TCA/PM2.5 0.16 ± 0.03 0.35 ± 0.1 0.29 ± 0.1 0.44 ± 0.2 0.34 ± 0.2

CE for all samples was 0.87 ± 0.2, which indicated that the collected
PM was slightly alkaline. This finding agreed with that of other studies
in northern China (Shi et al., 2017; Tao et al., 2013). On the other hand,
the anion and cation equivalents were strongly correlated in four sea-
sons (autumn: R2 = 0.91, winter: R2 = 0.99, spring: R2 = 0.94 and
summer: R2 = 0.95). The AE/CE ration revealed a typical seasonality
with larger than 1 for winter and lower than 1 for other seasons. For
autumn and spring, the AE/CE ration was almost equal to the 1:1 line,
and the particles were closer to neutral because of the significant in-
crease in alkaline compounds from the sand storms. In summer, almost
all of the samples were under the 1:1 line, which might have been re-
lated to carboxylic and phosphate that were not detected in this study,
and would be higher with higher air temperature and a stronger pho-
tochemical reaction (He et al., 2017). In winter, 74% samples were
above the 1:1 line, which illustrated that the particles were acidic as a
result of the highest levels of acid compounds (NO3− and SO42−).

3.1.4. Carbonaceous aerosols

Table 3 provides a summary of the seasonal and annual average level of
carbonaceous aerosols in Dushanzi. The annual mean concentrations of OC
and EC in PM2.5 were 7.13 ± 3.3 and 1.25 ± 0.7 μg m−3, which account-
for PM2.5 17.47 ± 10% and 2.72 ± 2.7%, respectively. The levels of OC
and EC were consistent with Shanghai (9.89 ± 8.9 μg m−3 for OC and
1.63 ± 1.5 μg m−3 for EC) (Ming et al., 2017). Seasonal levels of OC con-
centrations decreased in the following order: winter (8.76 ± 4.3 μg m−3)
> summer (7.53 ± 3.0 μg m−3) > spring (6.46 ± 2.6 μg m−3) > autumn
(5.67 ± 2.1 μg m−3). However, the seasonal variation of EC was different,
the EC concentration was the highest for autumn (1.73 ± 0.8 μg m−3), and
the levels were similar for spring (0.90 ± 0.4 μg m−3) and summer
(0.97 ± 0.5 μg m−3). The seasonal variation of EC was different from other
cities in China, such as Lanzhou (the highest for winter) (Tan et al., 2017),
possibly due to the big farm in southwest Dushanzi and the biomass burning
after the harvest in autumn influencing the EC emission. It was demonstrated
that OC can either release primary OC or secondary origins (formed through
photochemical conversion), while EC is essentially a primary pollutant from
combustion (Tan et al., 2017). Therefore, the discrepancy of OC and EC in
Dushanzi had much to do with their seasonal sources, chemical transforma-
tions and meteorology.
A number of scholars have investigated the OC/EC ratio as a useful
tool for identifying the sources of carbonaceous aerosols (Chen et al.,
2017; Koch, 2001; Wang et al., 2016; Watson et al., 2001). As shown in
Table 3, the OC/EC ratios in Dushanzi were higher in summer
(8.45 ± 2.5) than those in spring (7.47 ± 1.5) winter (6.27 ± 1.2)
and autumn (3.84 ± 10.5), indicating the source variations of carbo-
naceous aerosols were vague because the ratios partially overlapped.
Strong correlation, however, were shown between OC and EC (Fig.5),
with an R2 ranging from 0.80 to 0.88, which suggested that OC and EC
had common sources, such as coal combustion, vehicle emission and
biomass burning. Chow et al., 1996 demonstrated that OC/EC ratios
exceeding 2.0 may indicate the presence of secondary organic matter
(SOC). Interestingly, in this study, the ratios were higher than the
standard. It is difficult to quantify and separate primary and secondary
OC due to the vast number of organic compounds with a variety of
chemical and physical properties. However, it is possible to use an
Fig. 5. Scatter plot between OC and EC in four seasons in Dushanzi.
indirect method to estimate the SOC, the EC tracer method formation
with the following equation proposed by Turpin and Huntzicker (1995),
Castro et al. (1999):
SOC = OC (OC/EC) pri × EC (9)
The minimum OC/EC ratio during the sampling periods is often
assumed as the primary OC/EC ratio ((OC/EC)pri). The estimated SOC
concentrations in this study are presented in Table 3. The annual mean
concentration of SOC in PM2.5 was 2.0 ± 1.6, with the seasonal var-
iation in winter > summer > spring > autumn. The seasonality may be
governed by the variability in emission strengths (OC was the lowest
and EC was the highest in autumn) and meteorology. The SOC/OC
ratios were higher in winter (0.35 ± 0.2) and summer (0.32 ± 0.2)
than in spring (0.30 ± 0.1) and autumn (0.25 ± 0.1). These results
were consistent with previous observations in other cities in China,
which showed high values of SOC and SOC/EC in winter, due to the
coal combustion accelerating the emission of carbon aerosols and the
low temperatures increasing the adsorption of volatile organic com-
pounds (VOCs) and SOC yields (Tao et al., 2013; Wang et al., 2017;
Wang et al., 2016).
The total carbonaceous aerosol (TCA) was calculated by the sum of
EC and organic matter, which was estimated by multiplying the amount
of OC by 1.6 to assess the contribution of carbonaceous aerosols to the
PM mass (Ming et al., 2017). TCA accounted for an average of 34% of
PM2.5, with the seasonal rank of winter > summer > spring > autumn.
The ratios were higher than those of other cities, such as Tianjin (16%),
Zhejiang (17%), Guangdong (23%) and Sichuan (20%) (Zhou et al.,
2016). It was suggested that carbonaceous aerosol was an important
fraction of PM2.5 in Dushanzi.
3.2. Chemical mass closure
The PM2.5 mass was reconstructed on a seasonal and annual basis by

263
Y. Turap et al.
Fig. 6. Correlation between reconstructed and measured and chemical mass closure of PM2.5 in four seasons and annually.
the methods described in Sect. 2.3. The gravimetric PM2.5 mass con-
centration was compared with the reconstructed PM2.5 mass con-
centration, as shown in Fig. 6(a), which show a good correlation
(higher than 0.90) with one another in each season and throughout the
year. The proportions of all components in PM2.5 together with the
unidentified constituents as a whole are schematically illustrated by
five pie charts for the four seasons and for the whole year.
Collectively, PM2.5 in the Dushanzi district was composed of mi-
neral dust 33.58%, TEO 0.20%, POM 14.87%, EC 1.56%, NH4+ 4.58%,
NO3− 6.18%, SO42− 11.07% and an unidentified constituent 27.95%
(Fig. 6(f)). The major components were mineral dust, secondary in-
organic aerosols (a combination of SO42+, NO3− and NH4+) and POM,
each of which was > 15%, albeit with seasonal variations. Specifically,
the proportions of mineral dust showed a notable seasonal change, with
the maximum in spring (50.17%), the minimum in winter (only 7.14%)
and an intermediate (40.90% and 41.56%) in autumn and summer,
which were much higher than those in Beijing (23.5%) (Zhang et al.,
2013), Chongqing (8.2%) (Chen et al., 2017) and Lanzhou (30.3%)
264
Atmospheric Research 218 (2019) 257–268
(Wang et al., 2016), which indicated more dust pollution in Dushanzi.
The relatively high level of mineral dust was likely, because it was
surrounded by the desert and Gobi, and a low vegetation cover. Dust
was lower in the winter because of the reduction of dust emissions by
snow (Shi et al., 2017). The secondary inorganic species (SO42−, NO3−
and NH4+) have the largest proportion in winter (53.25%) and lower
percentage during other seasons. The maximal value of SO42−, NO3−
and NH4+ were all noted in winter 26.24%, 14.58% and 12.43%, re-
spectively. The POM fraction largely varied as follows: summer
(25.40%) > winter (12.97%) > spring (12.70%) > autumn (9.26%).
EC and TEO exhibited the largest proportion 2.02% and 0.28% in
summer, which were lower than those in Beijing and Chongqing (Chen
et al., 2017; Zhang et al., 2013). The unidentified components reached
27.95% of the total PM2.5. They showed seasonal variability with the
smallest (22.64%) in summer and the largest (32.20%) in autumn. Such
a high unidentified component was demonstrated in autumn in Beijing
(33.9%) due to the volatilization of some components of PM2.5, espe-
cially in autumn and summer during the storage of the weighted
Y. Turap et al.
Fig. 7. Backward trajectory clusters and their chemical components by season, at 100 m height located at 44.19°N, 84.53°E.
samples before extraction, which may have led to negative biases in the
specific components (Zhang et al., 2013).
Above all, the seasonality of increased proportions of PM2.5 in
Dushanzi was as follows: mineral in autumn and spring; secondary in-
organic species in winter; POM, TEO and EC in summer.
3.3. Sources apportionment analysis
Previous studies have reported that the regional sources and transport of
air pollutants exert a profound impact on the local air quality (Wang et al.,
2004). To address this issue, the Hybrid Single Particle Lagrangian In-
tegrated Trajectory (HYSPLIT) model was used to determine the transport
trajectory of air parcels at the sampling site. The 48 h back trajectories were
conducted every 6 h, at 100 m above the ground. Fig. 7 shows that the
PM2.5 concentrations were related to the size of the pie charts and the re-
lative chemical components by season.
In autumn, six major air masses influenced the Dushanzi district. Air
masses 1 and 3 originated from the southwest (SW) direction and air
mass 3 consisted of 71% of the trajectories. The SW cluster was further
differentiated into two types; i. e., fast (SWf, air mass 1) and slow (SWs,
air mass 3), according to the distance of air parcels. Air masses 2 and 6
arrived from the northwest direction, from Kazakhstan. Among them,
air mass 6 carried the highest PM2.5 mass concentration (83.24 μg m−3)
and the highest mineral concentration (60.79%), which demonstrated
that the northwest might be a potential origin of the crustal dust and
the long pathway promotes the absorptivity of PM2.5. Air masses 4 and
5 arrived from north direction, providing the lowest PM2.5
265
Atmospheric Research 218 (2019) 257–268
concentration. All of the chemical compounds except mineral had ap-
proximately similar distributions in each air mass.
In winter, three major air masses influenced the Dushanzi district,
and they originated mainly from two areas, including the northwest,
south and southwest. Air mass 1 carried the highest concentrations of
PM2.5 (138.16 μg m−3), including a high proportion of SO42−, NO3−
and NH4+ (which probably originated from coal combustion and ve-
hicle emission), and arrived via a short pathway from Kuitun city. Air
mass 2 reached the sampling site via a long pathway from the south
direction, thus loading the highest SO42−. Air mass 3 originated from
the neighboring countries, and the concentration of loaded PM2.5 was
78.57 μg m−3, which suggested that neighboring countries also influ-
enced the air quality of the Dushanzi district.
In spring, the trajectories originated from two areas of the Dushanzi
district. The majority arrived on air mass 1 (consisting of 86% the
trajectories), which originated from the southwest direction. Air mass 2
started from Kazakhstan, which loading the highest mineral con-
centration (54%) and was consistent with that of autumn. Other com-
pounds that were in proportion in this air mass was similar to those of
air mass 1, which indicated that the concentrations of elements in PM2.5
in the Dushanzi district were greatly impacted by dust particles that has
been transported from the desert and were in line with Section 3.1.2.
In summer, three air masses impacted the Dushanzi district atmo-
sphere. The highest PM2.5 concentration (36.04 μg m−3) was loaded by
air mass 1. Air mass 2 and air mass 3 started from northwest and
southwest direction, respectively. The PM2.5 concentrations loaded by
three air masses were less than or close to the NAAQS limit, indicating
Y. Turap et al. Atmospheric Research 218 (2019) 257–268

that the transport air masses were relatively clean or loaded low pol- Table 4
lution compounds. In addition the three air masses were approximately Relative contributions from six identified sources of PM2.5 in Dushanzi within
similar in their distributions of chemical components. the one-year and four-season periods.
In short, the Dushanzi district atmosphere was influenced by four air Source Autumn Winter Spring Summer Annual
masses (SWf, SWs, W and NW), which accounted for 38.17%, 24.73%,
19.35% and 17.75%, respectively. Kazakhstan (via long pathway vehicular 4.76% 13.94% 9.21% 8.22% 9.43%
biomass 18.85% 8.26% 18.93% 15.20% 10.86%
transport) impacted the mineral distribution, and Kuitun city impacted
coal 21.35% 36.67% 17.59% 7.79% 18.45%
the WSI in the PM2.5. Most of the air mass came from the desert and industry 15.43% 5.47% 9.24% 18.35% 12.15%
Gobi areas, which indicated that the desert significantly influenced the secondary 20.68% 30.90% 6.62% 23.45% 18.26%
PM2.5 concentration in Dushanzi, and part of the air masses via a short soil 18.94% 4.81% 38.23% 26.98% 30.85%
pathway transport arrived at the sampling site, which further indicated
that local pollutant emissions mixed with the mineral aerosols on the
pathway were the main source of the PM in Dushanzi. natural gas engines in Xinjiang. It was also worth noting that the con-
By utilizing the PMF model with the obtained full data set as the centrations of Ca and Mn were relatively high, which suggested that
input date, six main sources were identified: vehicular emission, bio- road dust could be an important contributor to this factor due to the
mass burning, coal combustion, industrial pollution, secondary aerosols extremely dry climate in Dushanzi. Therefore, the factor can be re-
and soil dust. Their mean contributions to PM2.5 in the Dushanzi district garded as a mixed source. The contribution of vehicular emission to
were 9.43%, 10.86%, 18.45%, 12.15%, 18.26% and 30.85%, respec- PM2.5 was 9.43%, with a maximum in winter (13.94%) and a minimum
tively. Fig. 8 shows the source profiles, and Table 4 summarizes the in autumn (4.79%). The result is similar to the contribution of traffic
source apportionment results of the relative contributions of the in- emissions to the PM2.5 mass in Chongqing (Chen et al., 2017).
dividual source to the PM2.5 on both seasonal and annual basis in the The second factor represented biomass burning characterized by a
Dushanzi district. high proportion of K and Tl. K is an excellent tracer of biomass burning
The first factor was characterized by high proportions of Cu, Zn, Cd, aerosols (Zhang et al., 2013). On average, this factor accounted for
Pb, Mg and OC in PM2.5 featured by vehicular emission. Previous stu- 10.86% of the PM2.5 mass concentration in the Dushanzi district, which
dies have reported that traffic-generated aerosol particles are rich in was higher than Lanzhou (Wang et al., 2016). Table 4 shows that bio-
OC, Cu, Zn and Pb (Chen et al., 2017). OC is the main pollutants from mass burning has higher contributions in autumn, spring and summer
gasoline and diesel combustion, Zn is used as an additive in lubricating than in winter. Zhang et al., 2010 reported that the contribution of
oil and Pb is linked to metal brake wear (Yao et al., 2016). In this factor, biomass varied with seasons; however, they were based on data for only
the proportion of Cd and Mg were relatively high, and may be emitted a few months at a limited number of sites and were further restricted by
from natural gas engine combustion due to many cars whit gasoline/ uncertainties in biomass burning emissions and tracer concentrations.

Fig. 8. Factor profiles (% of species total) obtained from the PMF analysis in Dushanzi.

266
Y. Turap et al.
In this study, the contribution of biomass was not different in autumn,
spring and summer, which was probably due to open field burning of
agricultural waste after harvesting in autumn and cultivation in spring
and summer enhanced the unorganized burning in agriculture (Zhang
et al., 2013). Moreover, there is a large farm in the southwest sampling
site, and the transport trajectories (Fig. 7) also indicated that the
southwest area was a potential source region.
The third factor was coal combustion, which was characterized by
high level of Na, V, As, OC and Cl−, and it was reported that As was
proposed to be a more useful tracer of coal combustions (Hsu et al.,
2016). In addition, there was a large coal-fired power plant in Dush-
anzi, and a large consumption coal every year strengthened the atmo-
spheric As and V emission. Different seasons have significantly different
contributions of coal combustion to PM2.5, which ranged from 7.79 to
36.67%, with a mean contribution of 18.45%. The portion of coal
combustion was higher in winter and lower in summer because do-
mestic heating would aggravate the emission in wintertime. Such sea-
sonality was also found by some previous studies (Chen et al., 2017;
Wang et al., 2016).
The fourth factor was characterized by the high level of Ni Co and
OC. Some previous studies found that Ni could be from oil burning in
industrial oil boilers or no-ferrous smelters (Chen et al., 2017; Dall'Osto
et al., 2013; Hsu et al., 2016). Although Co and OC represented the coal
combustion and vehicle emission, Zhang et al. (2013) reported that coal
was the primary energy source that is commonly used industrially in
China, and carbonaceous aerosols were also a vital source of the in-
dustrial pollution. There are large oil refineries and petrochemical in-
dustries in Dushanzi; therefore, industrial pollution was a main factor.
On average, industrial pollution accounted for 12.5% of the PM2.5 mass,
with a higher percentage in summer (18.35%) due to the volatilization
of chemical solvent in high temperature. Moreover, the trajectories
(Fig.7) showed that the airflow originated from the north oil refinery
and petrochemical industry area.
The fifth factor was recognized as secondary aerosols, as it had a
high content of SO42−, NO3−, NH4+ and OC. Secondary products were
formed by photochemical or other chemical reaction processes. Their
levels were affected by precursor gases (e.g., SO2 and NOx) and atmo-
spheric conditions (e.g., humidity, temperature, solar radiation and
insolation). This factor contributed to 18.26% during the sampling
period. Seasonality percentages of the secondary aerosols were the
highest in wintertime (30.90%) and were the lowest in springtime
(6.62%). These finding were distinct from those of other previous stu-
dies, in which summer often had the highest percentage of secondary
aerosols (Chen et al., 2017; Zhang et al., 2013). However, the results
were consistent with Sec. 3.1.3, which had the highest concentration of
NH4+, NO3− and SO42+ in winter due to the higher emission of pre-
cursor gas in winter.
The sixth factor was relevant to soil dust, which was typically
characterized by remarkable crustal elements, such as Al, Ca, Mg, Fe
and K. The Ca and Mn contents were higher than Al, which indicated a
Ca- and Mn-rich dust. It was reported that Ca was the indicator element
for construction dust (Zhang and Iwasaka, 1999), and construction
activities were prevalent in Dushanzi in recent years. Interestingly,
there was also the presence of V, Cr, Rb, Sr, Ba and Cl−, of which the
contents were close to 30%. Both Rb and Sr with low EF values were
identified as nature sources in Sec. 3.1.2 (Fig. 4). Because the location
of our study was near to the Gurbantongute desert and Takelamakan
desert and the geomorphological type of the Dushanzi district is pre-
dominantly Gobi desert, the heavy wind-blown soil dust from the desert
likely contributed to these crustal elements during the sampling period.
Therefore, this source possibly was mixed with resuspended road dust,
fugitive dust, anthropogenic construction dust and desert/Gobi dust. On
average, this factor accounted for 30.85% of the PM2.5 mass con-
centration in the Dushanzi district, which was higher than in other ci-
ties in China (Chen et al., 2017; Wang et al., 2016; Zhang et al., 2013).
Previous studies reported that dust storms are a common phenomenon
267
Atmospheric Research 218 (2019) 257–268
in arid and semi-arid areas (Ganor and Foner, 2001; Ganor et al., 2010),
and dust is a major source of global atmospheric particulate matter
(PM) (Shao et al., 2011). In this study, the seasonal contribution of soil
dust was the highest in spring (38.23%), which was consistent with the
variation of mineral in Sec. 3.2 and the results of the backward tra-
jectory clusters of air mass originated from the desert direction and
frequent occurrence of dust storms in Xinjiang.
4. Conclusions
Ambient PM2.5 and its major components were measured in four
seasons during the period 2015–2016 in Dushanzi, northwest China.
The annual mean concentration of PM2.5 was 62.85 ± 43.5 μg.m−3,
which was much higher than that established for Grade II (35 μg.m−3)
of the Chinese National Ambient Air Quality Standards' (NAAQS). The
seasonal averages of PM2.5 in winter were a factor of 3 higher than
those in summer. Distinctive seasonality occurred for various aerosol
species. The highest and lowest concentrations of dominant chemical
compositions were observed in winter and summer, and the highest
mineral elements were observed in spring, summer and autumn. The
enrichment factor showed that elements such as Ni, Mg, Cr, Cu, Ba, Zn,
As, Pb and Cd were attributable to anthropogenic emission. SO42−,
NO3−and NH4+ were the dominant species, which accounted for
78.55% of the total ionic concentration, and the ion balance indicated
that the collected PM in Dushanzi was slightly alkaline. Carbonaceous
aerosol was an important component in PM2.5 in Dushanzi, with the
TCA accounting for 34.12% of PM2.5. CMCs were successfully con-
structed on a seasonal basis, although with an unidentified fraction that
was averaged at 27.95%. The major aerosol speciation that was con-
sidered in this study included mineral dust, POM, and SIA. The tra-
jectory clustering method and the PMF model showed that the mixing
of the anthropogenic aerosol sources (vehicle emission, biomass
burning, coal combustion, industrial pollution, secondary aerosols)
with the transported soil dust were the main sources of PM in Dushanzi.
Acknowledgments
This work was supported by the National Natural Science
Foundation of China (No. 41465007) and State Key Laboratory of
Organic Geochemistry, GIGCAS (No. SKLOG-2016201624).
References
Castro, L.M., Pio, C.A., Harrison, R.M., Smith, D.J.T., 1999. Carbonaceous aerosol in
urban and rural European atmospheres: estimation of secondary organic carbon
concentrations. Atmos. Environ. 33, 2771–2781.
Cesari, D., Donateo, A., Conte, M., Merico, E., Giangreco, A., Giangreco, F., Contini, D.,
2016. An inter-comparison of PM2.5 at urban and urban background sites: Chemical
characterization and source apportionment. Atmos. Res. 174-175, 106–119.
Chen, Y., Xie, S.-d., Luo, B., Zhai, C.-z, 2017. Particulate pollution in urban Chongqing of
southwest China: Historical trends of variation, chemical characteristics and source
apportionment. Sci. Total Environ. 584-585, 523–534.
Choi, J.K., Heo, J.B., Ban, S.J., Yi, S.M., Zoh, K.D., 2012. Chemical characteristics of PM
2.5 aerosol in Incheon. Korea. Atmos. Environ. 60, 583–592.
Chow, J.C., Watson, J.G., Lu, Z., Lowenthal, D.H., Frazier, C.A., Solomon, P.A., Thuillier,
R.H., Magliano, K., 1996. Descriptive analysis of PM 2.5 and PM 10 at regionally
representative locations during SJVAQS/AUSPEX. Atmos. Environ. 30, 2079–2112.
Clemens Reimann, A., Caritat‡, P.D., 2000. Intrinsic flaws of element enrichment factors
(EFs) in environmental geochemistry. Environ. Sci. Technol. 34, 5084–5091.
Dall'Osto, M., Querol, X., Amato, F., Karanasiou, A., 2013. Hourly elemental concentra-
tions in PM2.5 aerosols sampled simultaneously at urban background and road site
during SAPUSS – diurnal variations and PMF receptor modelling. Atmos. Chem. Phys.
13, 4375–4392.
Di, A., Xue, Y., Yang, X., Leys, J., Jie, G., Mei, L., Wang, J., She, L., Hu, Y., He, X., 2016.
Dust aerosol optical depth retrieval and dust storm detection for xinjiang region using
indian national satellite observations. Remote Sensing 8, 702.
Feng, Y., Chen, Y., Guo, H., Zhi, G., Xiong, S., Li, J., Sheng, G., Fu, J., 2009.
Characteristics of organic and elemental carbon in PM2.5 samples in Shanghai.
China. Atmos. Res. 92, 434–442.
Feng, J., Yu, H., Su, X., Liu, S., Li, Y., Pan, Y., Sun, J.-H., 2016. Chemical composition and
source apportionment of PM 2.5 during Chinese Spring Festival at Xinxiang, a heavily
polluted city in North China: Fireworks and health risks. Atmos. Res. 182, 176–188.
Y. Turap et al.
Ganor, E., Foner, H.A., 2001. Mineral dust concentrations, deposition fluxes and de-
position velocities in dust episodes over Israel. J. Geophys. Res. 106, 18431–18438.
Ganor, E., Osetinsky, I., Stupp, A., Alpert, P., 2010. Increasing trend of African dust, over
49 years, in the eastern Mediterranean. J. Geophys. Res. Atmos. 115, 2232.
He, Q., Yan, Y., Guo, L., Zhang, Y., Zhang, G., Wang, X., 2017. Characterization and
source analysis of water-soluble inorganic ionic species in PM 2.5 in Taiyuan city.
China. Atmos. Res. 184, 48–55.
Hsu, S.C., Liu, S.C., Arimoto, R., Shiah, F.K., Gong, G.C., Huang, Y.T., Kao, S.J., Chen, J.P.,
Lin, F.J., Lin, C.Y., 2010. Effects of acidic processing, transport history, and dust and
sea salt loadings on the dissolution of iron from Asian dust. J. Geophys. Res. Atmos.
115, 1485–1490.
Hsu, C.-Y., Chiang, H.-C., Lin, S.-L., Chen, M.-J., Lin, T.-Y., Chen, Y.-C., 2016. Elemental
characterization and source apportionment of PM10 and PM2.5 in the western
coastal area of central Taiwan. Sci. Total Environ. 541, 1139–1150.
Huang, G., Cheng, T., Zhang, R., Tao, J., Leng, C., Zhang, Y., Zha, S., Zhang, D., Li, X., Xu,
C., 2014. Optical properties and chemical composition of PM2.5 in Shanghai in the
spring of 2012. Particuology. 13, 52–59.
Khare, P., Baruah, B.P., 2010. Elemental characterization and source identification of PM
2.5 using multivariate analysis at the suburban site of North-East India. Atmos. Res.
98, 148–162.
Koch, D., 2001. Transport and direct radiative forcing of carbonaceous and sulfate
aerosols in the GISS GCM. J. Geophys. Res. Atmos. 106, 20311–20332.
Lee, P.K., Brook, J.R., Dabek-Zlotorzynska, E., Mabury, S.A., 2003. Identification of the
major sources contributing to PM2.5 observed in Toronto. Environ. Sci. Technol. 37,
4831.
Li, J., Zhuang, G., Huang, K., Lin, Y., Xu, C., Yu, S., 2008. Characteristics and sources of
air-borne particulate in Urumqi, China, the upstream area of Asia dust. Atmos.
Environ. 42, 776–787.
Li, Y., Yao, N., Sahin, S., Appels, W.M., 2016. Spatiotemporal variability of four pre-
cipitation-based drought indices in Xinjiang. China. Theor. Appl. Climatol. 1–18.
Li, M., Hu, M., Du, B., Guo, Q., Tan, T., Zheng, J., Huang, X., He, L., Wu, Z., Guo, S., 2017.
Temporal and spatial distribution of PM 2.5 chemical composition in a coastal city of
Southeast China. Sci. Total Environ. 605–606.
Liu, B.S., Song, Na, Dai, Q.L., Mei, R.B., Sui, B.H., Bi, X.H., Feng, Y.C., 2016. Chemical
composition and source apportionment of ambient PM2.5 during the non-heating
period in Taian. China. Atmos. Res. 170, 23–33.
Liu, B.S., Wu, J., Zhang, J., Wang, L., Yang, J., Liang, D., Dai, Q., Bi, X., Feng, Y., Zhang,
Y., Zhang, Q., 2017. Characterization and source apportionment of PM 2.5 based on
error estimation from EPA PMF 5.0 model at a medium city in China. Environ. Pollut.
222, 10–22.
Luo, Y., Zheng, C., Jiang, P., Xu, Y., Wu, H., 2012. Assessment of Ecological Risk of Heavy
Metals in Soils in Kuitun. Xinjiang. Chinese J. Soil Sci. 43, 1247–1252 (In Chinese).
Maria, S.F., Russell, L.M., Gilles, M.K., Myneni, S.C., 2004. Organic Aerosol Growth
Mechanisms and Their Climate-Forcing Implications. Science 306, 1921–1924.
Ming, L., Jin, L., Li, J., Fu, P., Yang, W., Liu, D., Zhang, G., Wang, Z., Li, X., 2017. PM2.5
in the Yangtze River Delta, China: Chemical compositions, seasonal variations, and
regional pollution events. Environ. Pollut. 223, 200–212.
Na, K., Cocker, D.R., 2009. Characterization and source identification of trace elements in
PM2.5 from Mira Loma, Southern California. Atmos. Res. 93, 793–800.
Paatero, P., Tapper, U., 1994. Positive matrix factorization: A non-negative factor model
with optimal utilization of error estimates of data values. Environmetrics 5, 111–126.
Pan, F., Zhou, W., Chen, F., 2012. Evaluation of ecological sensitivity in Karamay,
Xinjiang. China. J. Geogr. Sci. 22, 329–345.
Samara, C., Voutsa, D., Kouras, A., Eleftheriadis, K., Maggos, T., Saraga, D., Petrakakis,
M., 2014. Organic and elemental carbon associated to PM10 and PM 2.5 at urban
sites of northern Greece. Environ. Sci. Pollut. R. 21, 1769–1785.
Semmler, M., Seitz, J., Erbe, F., Mayer, P., Heyder, J., Oberdörster, G., Kreyling, W.G.,
2008. Long-term clearance kinetics of inhaled ultrafine insoluble iridium particles
from the rat lung, including transient translocation into secondary organs. Inhal.
Toxicol. 16, 453–459.
Shao, Y., Wyrwoll, K.H., Chappell, A., Huang, J., Lin, Z., Mctainsh, G.H., Mikami, M.,
Tanaka, T.Y., Wang, X., Yoon, S., 2011. Dust cycle: an emerging core theme in earth
system science. Aeolian Res. 2, 181–204.
Shao, L., Hou, C., Geng, C., Liu, J., Hu, Y., Wang, J., Jones, T., Zhao, C., Bérubé, K., 2016.
The oxidative potential of PM 10 from coal, briquettes and wood charcoal burnt in an
experimental domestic stove. Atmos. Environ. 127, 372–381.
Sharma, S.K., Mandal, T.K., Jain, S., 2016. Source apportionment of PM2.5 in Delhi, India
Using PMF Model. B. Environ. Contam. Tox. 97, 1–8.
Shi, G., Xu, J., Peng, X., Xiao, Z., Chen, K., Tian, Y., Guan, X., Feng, Y., Yu, H., Nenes, A.,
Russell, A.G., 2017. pH of aerosols in a polluted atmosphere: source contributions to
highly acidic Aerosol. Environ. Sci. Technol. 51, 4289–4296.
Song, W., Chang, Y., Liu, X., Li, K., Gong, Y., He, G., Wang, X., Christie, P., Zheng, M.,
Dore, A.J., Tian, C., 2015. A multiyear assessment of air quality benefits from China's
emerging shale gas revolution: Urumqi as a case study. Environ. Sci. and Technol. 49,
2066–2072.
Tan, J.H., Duan, J.C., Chen, D.H., Wang, X.H., Guo, S.J., Bi, X.H., Sheng, G.Y., He, K.B.,
Fu, J.M., 2009. Chemical characteristics of haze during summer and winter in
268
Atmospheric Research 218 (2019) 257–268
Guangzhou. Atmos. Res. 94, 238–245.
Tan, J., Zhang, L., Zhou, X., Duan, J., Li, Y., Hu, J., He, K., 2017. Chemical characteristics
and source apportionment of PM2.5 in Lanzhou. China. Sci. Total Environ. 601–602.
Tao, J., Zhang, L., Engling, G., Zhang, R., Yang, Y., Cao, J., Zhu, C., Wang, Q., Luo, L.,
2013. Chemical composition of PM2.5 in an urban environment in Chengdu, China:
Importance of springtime dust storms and biomass burning. Atmos. Res. 122,
270–283.
Taylor, S.R., Mclennan, S.M., 1995. The geochemical evolution of the continental crust.
Rev. Geophys. 33, 293–301.
Turpin, B.J., Huntzicker, J.J., 1995. Identification of secondary aerosol episodes and
quantification of primary and secondary organic aerosol concentrations during
SCAQS. Atmos. Environ. 29, 3527–3544.
Virtanen, A., Joutsensaari, J., Koop, T., Kannosto, J., Yli-Pirila, P., Leskinen, J., Makela,
J.M., Holopainen, J.K., Poschl, U., Kulmala, M., Worsnop, D.R., Laaksonen, A., 2010.
An amorphous solid state of biogenic secondary organic aerosol particles. Nature
467, 824–827.
Wang, Y.Q., Zhang, X.Y., Arimoto, R., Cao, J.J., Shen, Z.X., 2004. The transport pathways
and sources of PM10 pollution in Beijing during spring 2001, 2002 and 2003.
Geophys. Res. Lett. 31, L14110.
Wang, X., Bi, X., Sheng, G., Fu, J., 2006. Chemical composition and sources of PM10 and
PM2.5 aerosols in Guangzhou. China. Environ. Monit. Assess. 119, 425–439.
Wang, Y., Jia, C., Tao, J., Zhang, L., Liang, X., Ma, J., Gao, H., Huang, T., Zhang, K., 2016.
Chemical characterization and source apportionment of PM 2.5 in a semi-arid and
petrochemical-industrialized city. Northwest China. Sci. Total Environ. 573,
1031–1040.
Wang, Q., Jiang, N., Yin, S., Li, X., Yu, F., Guo, Y., Zhang, R., 2017. Carbonaceous species
in PM 2.5 and PM 10 in urban area of Zhengzhou in China: Seasonal variations and
source apportionment. Atmos. Res. 191, 1–11.
Watson, J.G., Chow, J.C., Houck, J.E., 2001. PM2.5 chemical source profiles for vehicle
exhaust, vegetative burning, geological material, and coal burning in Northwestern
Colorado during 1995. Chemosphere 43, 1141.
Wei, W., Zhou, H., Shi, Y., Abe, O., Kai, K., 2005. Climatic and Environmental Changes in
the Source Areas of Dust Storms in Xinjiang, China, during the Last 50 Years. Water,
Air, & Soil Pollution: Focus 5, 207–216 (In Chinese).
Xu, H., Cao, J., Chow, J.C., Huang, R.J., Shen, Z., Chen, L.W.A., Ho, K.F., Watson, J.G.,
2016. Inter-annual variability of wintertime PM 2.5 chemical composition in Xi'an,
China: Evidences of changing source emissions. Sci. Total Environ. 545-546,
546–555.
Yao, L., Yang, L., Yuan, Q., Yan, C., Dong, C., Meng, C., Sui, X., Yang, F., Lu, Y., Wang, W.,
2016. Sources apportionment of PM2.5 in a background site in the North China Plain.
Sci. Total Environ. 541, 590–598.
Yu, L., 2013. Characterization and Source Apportionment of PM2.5 in an Urban
Environment in Beijing. Aerosol Air Qual. Res. 13, 574–583.
Yu, Q., Gao, B., Li, G., Zhang, Y., He, Q., Deng, W., Huang, Z., Ding, X., Hu, Q., Huang, Z.,
Wang, Y., Bi, X., Wang, X., 2016. Attributing risk burden of PM 2.5 -bound polycyclic
aromatic hydrocarbons to major emission sources: Case study in Guangzhou, south
China. Atmos. Environ. 142, 313–323.
Zhai, Y., Liu, X., Chen, H., Xu, B., Zhu, L., Li, C., Zeng, G., 2014. Source identification and
potential ecological risk assessment of heavy metals in PM 2.5 from Changsha. Sci.
Total Environ. 493, 109–115.
Zhang, D., Iwasaka, Y., 1999. Nitrate and sulfate in individual Asian dust-storm particles
in Beijing, China in Spring of 1995 and 1996. Atmos. Environ. 33, 3213–3223.
Zhang, X., Hecobian, A., Zheng, M., Frank, N.H., Weber, R.J., 2010. Biomass burning
impact on PM 2.5 over the southeastern US during 2007: integrating chemically
speciated FRM filter measurements, MODIS fire counts and PMF analysis. Atmos.
Chem. Phys. 10, 6839–6853.
Zhang, R., Jing, J., Tao, J., Hsu, S.C., 2013. Chemical characterization and source ap-
portionment of PM2.5 in Beijing: seasonal perspective. Atmos. Chem. Phys. 13,
7053–7074.
Zhang, J., Tian, W., Long, X., Tian, H., Huang, Q., Pingping, X.U., Yang, Q., Zhang, J.,
2015. Fact and simulation of dust aerosol transported to stratosphere during a strong
dust storm in south Xinjiang. Plateau Meteorology 14, 313–318.
Zhang, Q., Jiang, X., Tong, D., Davis, S.J., Zhao, H., Geng, G., Feng, T., Zheng, B., Lu, Z.,
Streets, D.G., Ni, R., Brauer, M., van Donkelaar, A., Martin, R.V., Huo, H., Liu, Z., Pan,
D., Kan, H., Yan, Y., Lin, J., He, K., Guan, D., 2017. Transboundary health impacts of
transported global air pollution and international trade. Nature 543, 705–709.
Zhao, G., Fan, Z., 1994. Study on soil Element Background Values of the Taklamakan
Desert. Arid Zone Research 2, 35–40 (In Chinese).
Zhao, X.J., Zhao, P.S., Xu, J., Meng, W., Pu, W.W., Dong, F., He, D., Shi, Q.F., 2013.
Analysis of a winter regional haze event and its formation mechanism in the North
China Plain. Atmos. Chem. Phys. 13, 5685–5696.
Zhao, M., Huang, Z., Qiao, T., Zhang, Y., Xiu, G., Yu, J., 2015. Chemical characterization,
the transport pathways and potential sources of PM 2.5 in Shanghai: Seasonal var-
iations. Atmos. Res. 158–159, 66–78.
Zhou, J., Xing, Z., Deng, J., Du, K., 2016. Characterizing and sourcing ambient PM2.5
over key emission regions in China I: Water-soluble ions and carbonaceous fractions.
Atmos. Environ. 135, 20–30.