Continental China has been recognized as one of the most important sources of atmospheric mineral dust particles (called Kosa in Japan, which literally means yellow sand). Many investigators have pointed out the importance of study of the long-range transport of mineral dust particles and their modifications in this process even during the nondust storm periods. Because of these modifications, particles can change their radiative properties and their ability to be a condensation nucleus. Therefore it is important to examine the composition of individual mineral particles in their source region and compare these particles with those after long-range transport. A number of investigations have been carried out on the subject; however, the amount of data is still insufficient. Samples of aerosol particles were collected in Dunhuang, China, in different seasons in 2001 and 2002 during the ACE-Asia campaign. The collected particles were examined using a scanning electron microscope equipped with an energy dispersive X-ray analyzer. The particles in all the samples were mainly mineral particles. Similar types of mineral particles were found in the free troposphere over Japan. A number of differences were found between the particles collected in China and those collected over Japan, and these differences can be explained by chemical modifications that occurred in the particles during their transport from China to Japan. Approximately 40–45% of mineral particles mixed internally with sulphate during their transport in the troposphere. Also, the particles collected over Japan were found to be different from those obtained in ground-based measurements in Nagasaki, Nagoya, and Fukuoka, Japan (reported by other research groups). The portion of mineral particles that mixed internally with sea salt and sulphates was considerably smaller than for the samples obtained in Japan near the ground. It is important to take this fact into account while investigating the impact of mineral particles on the biogeochemical cycle and climate.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Most results for particles after transport were obtained in ground-based measurements. However, the composition of particles near the ground is different from that in the troposphere through which the particles are transported. To enable better understanding of temporal variation in the composition of mineral particles and their modification during long-range transport, several samples were collected in Dunhuang, China, near the source region of these particles. The collected particles were compared with those collected in spring over Japan. The measurements in both studies were performed using identical techniques. Some interesting similarities and differences between the two groups of particles were found. The results were also compared with previous results obtained in ground-based measurements in China and Japan by other research groups. This comparison revealed significant differences between the free-tropospheric mineral particles observed in this study and those obtained in ground-based measurements in Japan.
 Aerosol particles were collected in Dunhuang (40°09′N; 94°41′E), China, in 2001 and 2002 (Figure 1). The sampling sites were the campus of the Meteorological Bureau of Dunhuang City (17 August 2001; 29 April 2002) and the area around Mogao Grots located approximately 30 km from Dunhuang (18 August 2001, 18 October 2001 and 13 January 2002). A two-stage low-volume impactor was used for the particle collection. Aerosol samples were collected on a carbon-coated nitrocellulose film on the surface of copper or nickel grids. The jet diameter of the first stage of the impactor was 1.3 mm and jet diameter of the second stage −0.4 mm. The flow rate of air was about 1 L/min. The 50% cut-off diameter for the mineral particles in the first and second stages was about 1.0 and 0.1 μm, respectively, assuming a mineral particle density of 2.6 g/cm3 [Ishizaka and Ono, 1982]. Identical techniques were used for the particle collection in the aircraftborne measurements (the flight courses are shown in Figure 1) in the spring of 2000 and 2001 (Table 1). A total of 15 samples were collected at different altitudes. The samples were collected in spring because this season believed to be the time of the most active transport of Asian dust. All the flights were performed over the western side of Honshu Island (Japan) to avoid the effects of “domestic” gases and aerosols. For safety considerations, the samples for the aircraftborne measurements were collected only on calm weather days, but the samples for the ground-based measurements were collected under different weather conditions (Table 2). During the aircraftborne sample collection, samplers were mounted on a Cessna-404 aircraft and air was introduced into the aircraft cabin through an isokinetic decelerator, and thereafter it was distributed to a number of instruments, such as samplers, optical particle counters, gas analyzers, and others.
Table 1. Sampling Altitudes for Aircraftborne Measurements
Fine, No Dust Storm Events Reported by Meteorological Observatories, Japan
Fine, High Visibility
Dust Storm, Low Visibility
10 March 2000
11 March 2000
29 April 2000
21 March 2001
17 Aug. 2001
18 Aug. 2001
18 Oct. 2001
13 Jan. 2002
29 April 2002
 The morphology of individual aerosol particles and their elemental composition were examined with a scanning electron microscope (Hitachi, S-3000N) equipped with an energy dispersive X-ray (EDX) analyzer (Horiba, EMAX-500). The particle size was determined by measuring the particles' maximum length if their shape was irregular. The accelerating voltage for the X-ray analysis was 20 kV, and the beam current was 0.1 nA. The counting time was typically 50 s. It should be noted that the current EDX analysis method cannot detect lighter elements (such as C, N, and O) as well as Cu and Ni because of their background.
 We found that the particles were primarily composed of the following elements: Na, Mg, Al, Si, S, Cl, K, Ca, Ti, and Fe. However, a number of other elements, such as P, V, Cr, Mn, Zn, and Rh, were occasionally detected. The particles were classified into subgroups using criteria suggested by Okada and Kai . First, for each particle, we determined the element with the largest relative weight (the largest weight ratio). We then put each particle into one of “X”-rich groups, where “X” was the element's name. Next, for particles within each group, we determined several additional elements having large weight ratios, the total weight ratio of which was larger than 65% (P(X1) + P(X2) + … + P(XN) > 0.65), where P(X) is the ratio suggested by Okada and Kai :
 Using the 65% threshold we divided the particles into subgroups and calculated the average weight ratio and standard deviation for each subgroup.
3. Results and Discussion
3.1. Presence of Mineral Particles
 For the particles collected over Japan, the results indicated the presence of background mineral, sea salt, and sulphate aerosol particles [Trochkine et al., 2002]. Mineral particles were found to constitute the most significant portion of coarse-mode (d > 1 μm) particles. Although mineral particles were also detected in fine-mode (d < 1 μm) particles, in the free troposphere over Japan, the portion of fine-mode mineral particles was negligible compared to that of sulfur particles, such as ammonium sulphate, ammonium hydrosulphate, and sulphuric acid [Trochkine et al., 2002]. This made the drawing of conclusions concerning fine-mode mineral particles in the free troposphere over Japan difficult, and so the results of the present study on mineral particles over Japan, which are presented below, are for coarse-mode particles (d > 1 μm).
 Particles collected in Dunhuang were primarily mineral, and the compositions of coarse- (d > 1 μm) and fine- (0.1 μm < d < 1 μm) mode particles were relatively similar, so the results of the present study on mineral particles collected in Dunhuang, which are presented below, are for both fine- and coarse-mode particles (d > 0.1 μm).
3.2. Origin of Mineral Particles Collected in the Troposphere Over Japan
 Isentropic backward trajectory analysis was used to estimate the origin of air masses. Global objective analysis data provided by the Japan Meteorological Agency (GAPLX) were used for the estimations. These data included the geopotential height, horizontal wind, temperature, and humidity (below the pressure level of 300 hPa) measured at 18 pressure levels with a horizontal resolution of 1.25°. Air parcels were traced backward on the isentropic surface by using the fourth-order Runge-Kutta scheme [Sakai, 2001]. The resulting backward trajectories obtained over a 5-day period are shown in Figure 2. In all the sampling points, the air parcels originated from the Asian continent.
 The wind speed during the aircraftborne measurements and the average wind speed during the spring of 2000 were estimated from analytical grid data provided by the Japan Meteorological Agency (GAPLX) (Table 3). The longitudinal component of the winds was predominantly westerly at all altitudes. The winds were the strongest in March, and they gradually weakened toward May. The latitudinal component of the winds was more northerly in early spring, and it shifted slightly to southerly by May. Although the deviations were large for both components of the winds (the strength and direction of the winds varied daily), the main vector of the wind direction remained westerly. The mean wind speeds measured during the aircraftborne measurements (Table 3) were close to the mean monthly wind speed observed in March. Because three out of the four aircraftborne measurements were done in March and the remaining one in April, it can be said that the measurements were done on typical spring days. Thus the presented data on the state of mineral particles in the troposphere over Japan can be said to represent the background state of mineral particles.
Table 3. Comparison of Seasonally and Monthly Averaged Wind Speeds (km/h) Over Japan (36°N, 135°E, 10:00 JST) in Spring (March, April, and May 2000) and Average Wind Speed During the Periods of the Aircraftborne Measurements
 Both mineral particles collected in Dunhuang and those collected over Japan had an irregular shape. However, some of the mineral particles collected over Japan also included soluble materials on their surface such as liquid-phase sulphate; something that was not observed in the Dunhuang particles. Some of the particles collected over Japan had a distinctive mixed structure, so that different layers could be clearly identified (e.g., Figure 3a). The composition of these different layers was determined by EDX analysis (Figure 3b). As can be seen in the figure, the central part of that particle has mineral peaks together with S peak, while the surrounding areas have only S peak.
4. Particle Composition
4.1. Comparison of Particles Collected in Dunhuang and Those Collected Over Japan
 The detection frequencies of Na, Mg, Al, Si, S, Cl, K, Ca, Ti, and Fe in the mineral particles collected in both Dunhuang and over Japan are shown in Figures 4a and 4b, respectively, where the percentages indicates the relative number of particles in which the corresponding element was detected. In both cases, the distributions of frequencies for Na, Mg, Al, Si, K, Ca, Ti, and Fe were very similar, while those for S and Cl were different. This may be because S and Cl were added to the mineral particles during their transport in the atmosphere.
 For the particles collected in Dunhuang, the portion of Si-rich particles varied from 46% for the sample collected on 17 August 2001, to 77% for the sample collected on 13 January 2002 (hazy weather). For the sample collected during a dust storm (on 29 April 2002), this value was 72%. The portion of Ca-rich particles varied from 13% for the sample collected on 13 January 2002, to 41% for the sample collected on 17 August 2001, while that for the sample collected during the dust storm was 16%. The portion of Fe-rich particles was 3–10%. Si-rich, Ca-rich, and Fe-rich particles were present in all the samples collected in Dunhuang, and they contribute more than 90% to all mineral particles. Also, Mg-rich, Ti-rich, K-rich, and Cl-rich particles were detected as well with the portions of 0–7%, 0–3%, 0–1%, and 0–1%, respectively.
 We classified the particles into subgroups according to the relative weight of elements detected in these particles (Table 4) and obtained the substances that constituted the main part of the particle content. In the Si-rich particle group (particles with the largest weight ratio of Si), the Si-dominant-type particles were mainly composed of quartz, and the Si + Al-type particles were mainly aluminosilicate. In the Ca-rich particle group, the Ca-dominant-type particles were mainly composed of calcite, the Ca + Mg-type particles were composed of dolomite, and the Ca + S-type particles were mainly gypsum. In the Fe-rich particle group, the Fe-dominant-type particles were mainly iron oxide-based. Other substances detected in the particles in all the samples (shown in bold in Table 4) were mainly mixtures of the substances described above. The presence of these substances in the Chinese desert areas was reported by many scientific groups on the basis of chemical analysis.
Table 4. Comparison of Particle Types for Particles Collected in Dunhuang, China, and in the Free Troposphere Over Japan, Similar Types Particle
Type of Particles
Averaged Major Elements, Average wt %
Averaged Major Elements, Average wt %
Si(81 ± 13), Al(15 ± 11)
Si(86 ± 10), Al(7 ± 9)
Si + Al
Si(56 ± 5), Al(23 ± 3), K(10 ± 6)
Si(54 ± 6), Al(22 ± 5)
Si + Fe + Al
Si(41 ± 6), Fe(24 ± 5), Al(19 ± 3)
Si(38 ± 4), Fe(20 ± 5), Al(20 ± 3)
Si + Fe
Si(41 ± 6), Fe(24 ± 5), Al(19 ± 3)
Si(48 ± 9), Fe(24 ± 6), Al(17 ± 5)
Si + Ca + Al
Si(39 ± 2), Ca(23 ± 3), Al(20 ± 2)
Si(38 ± 4), Ca(20 ± 5), Al(19 ± 2)
Si + Al + Mg
Si(42 ± 4), Al(20 ± 3), Mg(18 ± 3)
Si(43 ± 2), Al(17 ± 1), Mg(17 ± 3)
Si + Mg
Si(57 ± 3), Mg(35 ± 12)
Si(54 ± 5), Mg(19 ± 2)
Si + Ca
Si(54 ± 8), Ca(32 ± 5)
Si(40 ± 2), Ca(31 ± 5)
Si + Ca + Mg
Si(36 ± 5), Ca(22 ± 1), Mg(19 ± 3)
Si(35 ± 4), Ca(20 ± 4), Mg(14 ± 3)
Si + Mg + Fe
Si(36), Mg(23 ± 2), Fe(22 ± 1)
Si(41), Mg(13), Fe(22)
Si + K
Si(58 ± 2), K(24 ± 2)
Si(59 ± 1), K(25 ± 1)
Ca(86 ± 10)
Ca(86 ± 12)
Ca + Mg
Ca(54 ± 4), Mg(38 ± 7)
Ca(44 ± 1), Mg(30 ± 4)
Ca + Si
Ca(53 ± 8), Si(26 ± 7)
Ca(56 ± 8), Si(24 ± 10)
Ca + S
Ca(55 ± 7), S(39 ± 3)
Ca(52 ± 7), S(35 ± 10)
Fe(82 ± 11)
Fe(86 ± 12)
Fe + Si
Fe(49 ± 9), Si(30 ± 6), Al(15 ± 5)
Fe(50 ± 6), Si(26 ± 1), Al(13 ± 1)
Ti (98 ± 3)
Ti (86 ± 14)
Ti + Si
Ti (43 ± 5), Si (31 ± 1)
Ti (52), Si (32), Ca (17)
 The main types of mineral particles collected in Dunhuang are shown in bold in Table 4. These types of Si-rich, Ca-rich, and Fe-rich particles were present in all the samples. A close examination of the particle types showed that all significant particle types present in the surface air of Dunhuang were also present in the free troposphere over Japan (Table 4 shows that the main elements of the particles collected in Dunhuang and other Japan, their average weight and standard deviation are similar for particles collected in Dunhuang and other Japan). However, in Japan, these particles represented a smaller percentage of all collected mineral particles (approximately 70% compared to 90% in China). The total percentage of particles collected in Dunhuang does not add up to 100% because some types of particles collected in Dunhuang were not detected over Japan. Particles with unusual composition were occasionally present in the samples in very low numbers. These particles may not have been detected in the samples collected over Japan simply because of the relatively small number of particles in each sample or because of the technical difficulties in the aircraftborne measurement [Mori et al., 1999; Trochkine et al., 2002]. However, the difference in the percentages of the same types of particles (90% and 70%, respectively) can be attributed primarily to the presence of particles that mixed internally with sulphate. Such particles, which were significantly modified as a result of mixing with sulphate, were detected in almost all the samples collected over Japan. Also, particles mixed internally with sea salt and anthropogenic contaminants were detected over Japan (using a method suggested by Niimura et al. ), but their proportion was negligible; these particles were detected only in some samples, and they constituted only about 3% of all the mineral particles.
 The distribution of relative weight ratios of the most significant mineral elements (Al, Si, and Ca) represented by a triangular diagram (Figures 5a, 5b, 5c, 5d, 5e, and 6) shows that different mineral particles had different combinations of these elements. For example, Ca-rich components (calcite, dolomite, and gypsum) mixed internally with quartz and aluminosilicate in some of the particles (shown by dots inside the triangles), and externally in others (shown by dots near the Ca corner and along the Al-Si line). This distribution is typical for natural dust [e.g., Okada and Kai, 1995; Zhou et al., 1996] from Chinese desert areas. For each sample collected in Dunhuang, the distributions of Al, Si, and Ca had some characteristic features resulting from the different proportions of particles of different types in the sample (these proportions often varied significantly). However, overall, the distributions were similar not only for different samples collected in Dunhuang (Figures 5a–5e), but also for samples collected in the free troposphere over Japan (Figure 6). This means that the mixing of particles with sulphate results only in a proportional decrease of the relative weight ratios of different mineral elements, but does not change the relationship between these elements.
 Comparing the distributions of the relative weight ratios of Al, Ca, and S we can see significant differences between the samples collected in Dunhuang and the samples collected over Japan (Figures 7a, 7b, and 7c). Only a few particles containing S that were collected in Dunhuang were not gypsum particles, (Figures 7a and 7b), compared to a large number of such particles collected in the free troposphere over Japan (Figure 7c). It can be easily seen that many mineral particles collected over Japan contained S without Ca (shown by dots along the Al-S line), and for particles contained both S and Ca many of them had S/Ca ratio large than those of 0.8; the value typical for gypsum (shown by dots on the left of the dashed line).
 The percentage of modified particles can exceed 20% (a value estimated from the difference in the percentages of mineral particles of the same type in Table 4), because of a very small weight ratio of sulfur for a number of particles, which not changes the particle type. In order to estimate the portion of modified particles, we compared the portions of Ca- and S-containing particles, and the S/Ca weight ratios. In Dunhuang, only 20% of the particles containing Ca also contained S, but for the particles collected over Japan, this value was 56%. Therefore we think that in some particles, calcite (CaCO3) chemically changes to gypsum (CaSO4 · 2H2O); however, this cannot explain all the modified particles. The portion of S-containing particles over Japan was 55%, and in Dunhuang it was only 8.6% (Figure 8). The portion of particles containing S without Ca was 2.4% and 28.3% for Dunhuang and the free troposphere over Japan, respectively (Figure 8). Therefore for a large number of mineral particles collected over Japan the presence of S cannot be explained by the presence of gypsum. In Dunhuang, only 16% of the particles containing both S and Ca had a S/Ca ratio larger than 0.8, which is a typical ratio for gypsum, while for the particles collected over Japan, this value was 47%. Thus, in spring, a total of 40–45% of background mineral particles were in a mixing state in the atmosphere over Japan.
4.2. Comparison of Our Results With Those Obtained for Particles Collected Near the Ground in China and Japan
 If we compare our results with the results of ground-based measurements, the following picture emerges.
 First, the chemical composition of individual particles collected in Dunhuang was similar to that of unmodified dust particles collected in China and investigated by other scientific groups [e.g., Okada and Kai, 1995; Zhou et al., 1996]. In these studies, the particles were collected and examined using similar techniques. Note, however, that the similarities in composition between individual particles do not mean similarities between the bulk samples, because the proportions of different types of particles were different. For example, the number of Ca-rich particles in some of the samples collected in Dunhuang was higher than that of particles collected in Zhangye [Okada and Kai, 1995], but the types of particles observed in those samples were very similar in terms of the average weights of the elements composing these particles. Besides, in all cases the distributions of the relative weight ratios of Al, Ca, and S were similar, and the number of particles containing significant amounts of S not in the CaSO4 form was very low.
 Second, in this study, the distributions of the relative weight ratios of Al, Ca, and S in the samples collected over Japan were different from those obtained from the ground-based measurements of modified particles in Japan (e.g., Okada et al.  for Nagoya and Nagasaki; Zhou et al.  for Fukuoka). In the present study, we found many unmodified mineral particles, in the samples collected over Japan that were chemically very similar to the particles collected in Dunhuang and other Chinese areas. In contrast, in earlier ground-based measurements in Japan almost all particles were found to have been modified as a result of mixing with sulphate and sea salt. Also, in the present study, about 55% of the mineral particles collected in the troposphere over Japan contained S, while in previous ground-based measurements, the percentage of particles obtained in Japan that contained S was significantly larger (more than 80% for Nagasaki [Okada et al., 1990], more than 85% for Fukuoka (dust particles without sea salt [Zhou et al., 1996]), and more than 90% for Nagoya [Okada et al., 1990].
 Third, all studies investigating mineral particles near the ground in Japan indicated the presence of dust particles that mixed internally with sea salt [e.g., Okada et al., 1987; Niimura et al., 1994; Zhou et al., 1996]. Fan et al.  found that the portions of mineral particles mixed internally with sea salt that were measured on two different days in Nagasaki, Japan, were 20% and 60%. Zhou et al.  found that the portion of mineral particles mixed internally with sea salt in Fukuoka was 48%. These values are considerably larger than the value measured in the troposphere in the present work (about 3%). This may be because the internal mixing of dust particles with sea salt mainly occurs in cloud processes and at the boundary layer atmosphere during the particles deposition, but not in the cloud-free troposphere.
 Mineral particles make up the most significant portion of aerosols in the atmosphere over Chinese desert areas throughout the whole year. During the measurements in Dunhuang in 2001 and 2002, about 46–77% of mineral particles were Si-rich particles (mainly composed of quarts and aluminosilicate), and about 13–41% were Ca-rich particles composed of calcite (CaCO3), dolomite (CaMg(CO3)2), and gypsum (CaSO4 × 2H2O). Also, Fe-rich particles composed of iron oxides were detected in significant portions (3–10%). A number of particles were composed of internal mixtures of the above substances.
 Similar types of mineral particles were found in the free troposphere other Japan on the basis of comparison between different particle types, but some of the mineral particles collected over Japan were significantly modified as a result of mixing with sulphate, sea salt and anthropogenic contaminants. The chemical composition of mineral particles changed as a result of their long-range transport in the free troposphere. Only a few particles collected near their source region showed the presence of sulfur (except calcium sulphate (gypsum); the present study and previous results), compared to about 55% of particles obtained in the troposphere over Japan (the present study) and more than 80% of particles collected in Japan near the ground (previous results). Also, in the present study, the portion of mineral particles that mixed internally with sea salt was considerably smaller than for the samples obtained in Japan near the ground (previous results). The total portion of modified mineral particles in the free troposphere over Japan was estimated to be about 40–45%.
 Further research is needed to determine the behavior of mineral dust particles in different seasons (especially in spring), under different weather conditions, and at different altitudes (using balloon borne sampling) to clarify the temporal and spatial variation of mineral aerosols near their source region and their modification during the long-range transport from China to the west Pacific region.
 This work was partly supported by a grant from the Japanese Ministry of Education, Culture, Sports, Science and Technology (Grant-in-Aid for Specially Promoted Research, 10144104). We are grateful to Gang Li and Bin Chen from the Institute of Atmospheric Physics, Chinese Academy of Science and Kazuo Osada, Chiharu Nishita, Yayoi Inomata, Takayoshi Nezuka, and Mizuka Kido from Solar-Terrestrial Environment Laboratory, Nagoya University for their kind help.