Speciation of Iron in Atmospheric Particulate Matter by
EXAFS
Wang Yinsong1, Li Aiguo1, Zhang Yuanxun1, Xie Yaning2, Li Delu1, Li Yan1,
Zhang Guilin1
1
Shanghai Institute of Applied Physics, The Chinese Academy of Sciences, Shanghai 201800.
2
Institute of High Energy Physics, The Chinese Academy of Sciences, Beijing 100049.
Abstract. The speciation of iron in atmospheric particulate matter (APM) was investigated by EXAFS. APM
samples with different particle sizes and from different sampling sites were collected in Shanghai, China. The
chemical components of iron in the samples were calculated by the regression analysis of their EXAFS
spectra. The results show that the iron in all samples mainly consists of Fe2O3, Fe3O4 and Fe2(SO4)3, however,
their proportions are different in different samples.
Keywords: atmospheric particulate matter, speciation, regression analysis, iron, EXAFS
PACS: 92.60.Sz, 61.10.Ht
INTRODUCTION
EXPERIMENTS AND DATA ANALYSIS
In atmospheric particulate matter (APM), fractions
with a particle size of less than 10 μm or even less than
2.5 μm in aerodynamic diameter are called PM10 and
PM2.5, respectively. They can be inhaled into the human
body and cause serious health problems. The influence
of the APM on health is not only related to elemental
concentrations, but also to its speciation, that is, the
chemical state and component.
XAFS is a powerful technique to study speciation.
This is a non-destructive technique and is suitable for
the investigation of elemental speciation in APM.
Using least square fitting of XAFS, the chemical
components in samples can be calculated, but only little
information has been reported to date.
In the present work, the PM10 and PM2.5 samples
were collected in Shanghai, China. The chemical
components of iron in the samples were calculated by
regression analysis of EXAFS. The results show that
iron in all samples consist of Fe2O3, Fe3O4 and
Fe2(SO4)3 mainly, but their proportions are different in
different samples. The regression curve can be fitted
well with the experimental curve. That means the
results of the calculation conform to those of the
experiments.
Sampling
The samples of PM10 and PM2.5 were collected from
three sampling sites in Shanghai, China. The sites are
“gy” (an iron and steel industrial area), “gc” (in
downtown of Shanghai), and “nh” (in suburban of
Shanghai).
Fe2O3, Fe3O4, Fe2(SO4)3 and FeCl3 were chosen as
reference materials. In order to observe the influence of
chemical components on the EXAFS spectra and check
the method, a series of mixed reference samples were
prepared (according to atomic percentage of iron).
Measuring XAFS Spectra
The XAFS spectra of APM and reference samples
were measured at the XAFS station of Beijing
Synchrotron Radiation Facility (BSRF). The electron
energy in the storage ring was about 2.2 GeV with a
current of about 100 mA.
Data Analysis
We assume that the EXAFS of a mixture is the sum
of that of each component. If an element has n
components in the mixture, its EXAFS can be indicated
in a manner of linear combination:
χ(k) = C1χ1(k)+ C2χ2(k)+ …
+ Cnχn(k),
(1)
RESULTS AND DISCUSSION
Figure 1 shows the EXAFS spectra of APM samples
and reference materials. In the figure the EXAFS
spectrum of iron trichloride is obviously different from
that of the APM samples, so we conclude that there is
no iron trichloride in the APM samples. The spectra of
APM samples are different from spectrum of any
reference material, which shows that APM samples do
not consist of a single reference material, but possibly,
of several reference materials.
0.10
100%Fe2O3
80%Fe2O3 - 20%Fe3O4
50%Fe2O3 - 50%Fe3O4
100%Fe3O4
0.05
EXAFS
where, χ(k) is EXAFS spectrum of the mixture, χ1(k),
χ2(k),… χn (k) are EXAFS spectra of each component,
and C1, C2,… Cn are their relative concentrations,
respectively.
For this system, a regression analysis based on least
square fitting can be performed to calculate the relative
concentrations of each component.
The EXAFS spectra were obtained using the
software NSRLXAFS developed by National
Synchrotron Radiation Laboratory, China University of
Science and Technology. Then the EXAFS data were
deposited in an Excel worksheet, and the relative
concentrations were calculated with the regression
analysis in the Excel program.
0.00
-0.05
-0.10
-0.15
2
4
6
8
10
12
14
k
FIGURE 2. The EXAFS spectra of mixed reference samples
with different ratios of Fe2O3 and Fe3O4.
The components of mixed reference samples are
known, so EXAFS spectra of these samples can be
calculated with equation (1). In Fig. 3, the calculated
spectrum of a mixed reference sample is compared with
the experimental spectrum. They are nearly identical,
indicating that the calculated results using equation (1)
matches the experiments.
0.10
Experiment
Calculation
0.05
gc10
gc2.5
gy10
gy2.5
nh10
nh2.5
FeCl3
Fe3O4
Fe2O3
Fe2(SO4)3
0.8
EXAFS
0.6
0.4
0.2
0.0
EXAFS
1.0
0.00
-0.05
-0.10
-0.15
2
4
6
8
10
12
14
k
-0.2
2
4
6
8
10
12
14
k
FIGURE 3. Comparison of calculated spectrum and
experimental spectrum for a mixed reference sample.
FIGURE 1. The EXAFS spectra of APM samples and
reference materials.
The components of mixed reference samples were
determined through the regression analysis of their
EXAFS spectra. By comparing the determined values
with the prepared values, we can check the precision of
the method. Table 1 lists the determined values and the
prepared values of the mixed reference samples. The
determined values are very close to the prepared ones,
so the method has good precision and can be used to
estimate the chemical components of the samples.
Figure 2 is the EXAFS spectra of mixed reference
samples with different ratios of Fe2O3 and Fe3O4. The
EXAFS spectra correspondingly change with the
changes of the ratios.
.
No.
TABLE 1. Determined values and prepared values of mixed reference samples.
Prepared values (%)
Determined values (%)
Fe2(SO4)3 Fe2O3
Fe3O4
Fe2(SO4)3
Fe2O3
Fe3O4
1
0
50
50
0.9±0.5
49.6±0.7
49.5±0.6
2
0
80
20
1.7±0.6
77.7±0.9
20.6±0.7
3
0
90
10
0.7±1.0
85.7±1.5
13.5±1.3
4
10
90
0
5.0±0.6
92.8±1.0
2.3±0.8
5
20
80
0
15.6±0.6
81.7±0.8
2.6±0.7
6
10
70
20
7.0±0.6
75.4±0.9
17.5±0.7
7
20
70
10
13.7±0.8
73.7±1.2
12.3±0.1
8
10
80
10
8.2±0.5
81.9±0.7
9.9±0.6
We assume that the main components of iron in the
APM samples were Fe2O3, Fe3O4 and Fe2(SO4)3. The
components were calculated through regression
analysis of their EXAFS spectra. Table 2 shows the
percentages of each component in different samples.
TABLE 2. Components of iron in APM samples.
Samples
components (%)
Fe2O3
Fe3O4
Fe2(SO4)3
Gy10
Gy2.5
gc10
Gc2.5
Nh10
31.9±1.4
32.8±2.3
29.8±3.0
45.6±5.1
38.5±2.8
19.6±2.1
8.0±3.4
28.8±4.6
25.4±7.7
32.2±4.2
48.5±1.8
59.2±2.9
41.4±3.9
29.0±5.9
29.3±3.6
Nh2.5
53.0±3.1
21.1±4.6
26.0±3.9
0.08
0.06
Experiment
Regression
0.04
0.02
0.00
EXAFS
Table 2 shows that the iron contents in all the APM
samples consists mainly of Fe2O3, Fe3O4 and Fe2(SO4)3,
but their relative proportions are different in different
samples. The relative concentrations of Fe3O4 in the
samples of the iron and steel industrial area (gy) are
higher than those of other areas. In the samples of
downtown (gc) and suburban (nh), Fe2(SO4)3 in PM2.5
are higher than those in PM10, suggesting that Fe2(SO4)3
is likely to get enriched in fine particles.
The regression curves can be obtained using the
relative concentrations in Table 2 and the EXAFS
spectra of reference materials. Fig. 4 shows the
regression curve and experimental curve for sample gy
10.
The two curves fit well, showing that the results are
in agreement with the experiments.
-0.02
-0.04
-0.06
-0.08
-0.10
-0.12
2
4
6
8
10
12
14
k
FIGURE 4. Comparison
experimental curve
of
regressive
curve
and