ANALYTICAL SCIENCES JUNE 2003, VOL. 19 835 2003 © The Society for Analytical Chemistry

Enrichment Adsorption of a Labile Substance to the Surface of Particular Mineral Particles in Water as Investigated by SEM-EDX and Dilute-Acid Extraction/ICP-MS

Tomohiro KYOTANI*,**† and Satoshi KOSHIMIZU***

*Bio Nanotec Research Institute Inc. (BNRI), Mitsui & Co., Ltd. Nanotech Park, 2-1, Koyadai, Tsukuba, Ibaraki 305Ð0074, Japan **Japan Science and Technology Corporation (JST), 4-1-8, Honmachi, Kawaguchi, Saitama 332Ð0012, Japan ***Earth Science Division, Yamanashi Institute of Environmental Sciences, 5597-1, Kenmarubi Kamiyoshida, , Yamanashi 403Ð0005, Japan

The selective enrichment behavior of a labile substance, such as hydroxides, to the surface of particular mineral particles in river water was clarified by scanning electron microscopy/energy dispersive X-ray microanalysis (SEM-EDX). Individual particles other than diatom collected on a 0.45 µm filter from the Fuji and Sagami , central Japan, were analyzed by SEM-EDX and classified into seventeen groups according to the chemical composition and shape. Phosphorus, sulfur, chlorine, manganese and copper detected in each particle collected on the 0.45 µm filter could be successfully used as effective indicators of labile substance secondarily formed and adsorbed afresh in river water, because the detection frequencies of such elements are quite low, or negligible, in fresh mineral particles derived from igneous rocks. The labile substance adsorbed on mineral particles collected on the 0.45 µm filter was also evaluated by dilute-acid leaching, followed by inductively coupled plasma mass spectrometry (ICP-MS). Almost all parts of the manganese detected in individual particles were those adsorbed afresh as hydroxides together with iron and aluminum. Also, anionic elements, such as phosphorus, sulfur and chlorine, formed complexes with the hydroxides and/or were incorporated in them. Mg and/or Ca-rich aluminosilicate groups were the most effective adsorbers of such labile species. However, Si-rich and Na-, K- and Na-Ca rich aluminosilicates did not significantly adsorb the labile substance. Consequently, the remarkable selectivity was clarified in the adsorption process of labile substance to individual mineral particles in river water.

(Received October 4, 2002; Accepted March 26, 2003)

One of the processes for determining the dissolved elemental EPMA studies. concentration in river water is the interaction between the liquid The aim of this study was to clarify the factors affecting the and solid phases, that is, the elemental fractionation between dissolved elemental concentrations in river water in terms of the dissolved species and suspended particles.1 However, the chemical composition of insoluble individual particles. In this chemistry of river water has been long discussed concentrating study, the chemical composition of individual particles collected on the dissolved species only obtained from 0.45 µm filtrate.2,3 on a 0.45 µm filter from river water was determined by Consequently, information4Ð7 on the chemical composition of scanning electron microscopy/energy dispersive X-ray suspended particles in river water is not very much. The main microanalysis (SEM-EDX). Phosphorus, sulfur, chlorine, removal processes of dissolved elements in natural water are the manganese and copper detected in individual particles could be formation of metal-fumic complexes and coprecipitation with used as effective indicators of a labile substance secondarily hydroxides.8Ð10 Recently, the former has been intensively formed in river water. Consequently, the selective enrichment discussed based on speciation studies of trace elements by many behavior of labile substance to the surface of particular mineral researchers.8Ð10 Although the latter has been mainly discussed particles in river water was first clarified. using iron in a colloidal fraction from 0.45 µm filtrate,10 the action of the particles collected on the 0.45 µm filter has generally been neglected. While, electron probe X-ray Experimental microanalysis (EPMA) provides useful information on the chemical composition of individual particles collected on a 0.45 Geological characteristics of the study area µm filter.11Ð19 However, the role of individual mineral particles Water sampling was carried out at each of four sites in Fuji in the removal process of dissolved elements in river or lake and Sagami rivers, , central Japan. The water has not yet been clarified sufficiently by conventional details of the sampling location are shown in Fig. 1. starts from areas around Shosenkyo and Mt. Yatsugatake, which † To whom correspondence should be addressed. consist of mainly granitic and andesitic rocks, respectively,3 E-mail: [email protected] runs through Kofu basin and flows into . Sites F1 836 ANALYTICAL SCIENCES JUNE 2003, VOL. 19

Fig. 2 Analytical procedures of labile substance adsorbed on suspended particles in river water. Dilute-acid soluble fraction from the particles collected on the 0.45 µm filter was evaluated by B Ð A.

Fig. 1 Sampling locations.

the residual 500 ml portion of the river-water sample, and the and F4 in tributaries of Fuji river reflect approximately the acidified sample was kept as a 0.02 M nitric acid solution for 4 effects from andesitic rocks and sediments, respectively.3 Sites h. Thereafter, the acidified sample was filtered through a 0.45 F2 and F3 in the main stream of Fuji river reflect complex µm membrane filter and the filtrate was subjected to ICP-MS. effects from granitic, andesitic and sedimentary rocks. Sagami The residue on the filter was not analyzed in this study. The river starts from the foothills on the northern area of Mt. Fuji, elemental concentrations (B) obtained mean the total sum of the for example, “Oshinohakkai”, turns east the direction at Otsuki original dissolved components and those extracted from all of and flows into . The foot of Mt. Fuji consists of the particles. Therefore, the dilute-acid soluble fraction from mainly basaltic rocks.2,3,20 Mt. Fuji is the largest basaltic the particles collected on the 0.45 µm filter was evaluated by stratovolcano from the quaternary period in Japan, which B Ð A. spewed tremendous amounts of volcanics since the prehistorical age.20 Sites S1, S2 and S3 in the main stream of Single-particle analysis by SEM-EDX reflect the effects from basaltic rocks.3 Site S4 reflects the The particles on the filter were detached by both sides- effects from andesitic rocks.3 Although both rivers are fed by a adhesive carbon tape, and were fixed on an aluminum sample number of tributaries, it can be considered approximately that stand for SEM-EDX.19 Thereafter, the particles were carbon the Fuji and Sagami rivers show the effects from granitic and coated (∼200 Å) and the specimens were subjected to an SEM andesitic, and basaltic rocks, respectively. These sampling (Hitachi S-3000N)-EDX (Horiba EMAX550) analysis. About locations provide a good situation for geochemical studies of 200 particles other than diatoms in each specimen (F1 Ð F4 and river water, because this area covers extremely different S2) were analyzed under the condition of an accelerating geologies in spite of the narrow area. voltage of 20 kV for an X-ray intensity measurement and 25 kV for an SEM image observation, using a probe current of 0.3 nA River water sampling and pretreatment and a dead time of 20 Ð 30%. In the X-ray intensity The nitric acid used was of ultrapure grade. High-purity water measurements, an area analysis of individual particles was (Millipore, Milli-Q, Bedford, MA, USA) was used throughout. carried out as follows. The electron beam was usually Figure 2 shows the analytical scheme. River-water samples irradiated at the central part of an individual particle to obtain were collected in polypropylene containers (1 L) at each four the average composition, and the data were accumulated for 100 site in the Fuji (F1 Ð F4) and Sagami (S1 Ð S4) rivers on March s. The characteristic intensities of the detected elements were 1, 2000. The samples were shaken thoroughly prior to use. normalized to the total sum of the characteristic X-ray Firstly, a 500 ml portion was immediately filtered through a intensities of elements with an atomic number over 11 to avoid membrane filter (Millipore, Omunipore JH, Hydrophilized the particle-size effect.13,15,16,18 Obvious single particles with PTFE, 0.45 µm in pore size and 47 mm in diameter). The sizes approximately between 1 and 10 µm were classified into particles on the filter were used for an SEM-EDX analysis. sixteen groups according to the relative X-ray intensities of the Next, a leaching experiment was also carried out in order to major elements detected, as shown in Table 1. Aggregates or estimate the easily soluble components from the particles biomineral-like substances21 with sizes of approximately collected on the 0.45 µm filter, as follows. Above 0.45 µm between 20 and 30 µm were firstly discriminated from other filtrate was acidified (as 0.02 M nitric acid) with ultra-pure particles by their peculiar shapes, and then re-classified using nitric acid, and subjected to inductively coupled plasma mass the relative X-ray intensities of the constituents. The method of spectrometry (ICP-MS). The elemental concentrations (A) particle-type classification proposed in this study is not affected obtained mean the total sum of the original dissolved by minor and trace elements. The abundances of particles of components and those extracted from particles passed through interest were calculated as the relative ratios of the number of the 0.45 µm filter. While, ultra-pure nitric acid was added to the particles to the total number of particles measured in each ANALYTICAL SCIENCES JUNE 2003, VOL. 19 837

Table 1 Classification rules of individual particles in river water by SEM-EDX

Particle type Classification rules based on the relative X-ray intensity A Si-rich Si ≥ 80% B Aluminosilicates Si < 80% and has the highest relative X-ray intensitya B1 pure aluminosilicates 1) and Al + Si ≥ 90% B2 other aluminosilicates 1) and Al + Si < 90%b  B2-1 Na-K-Ca rich type 2) and Na + K+ Ca > Mg + Fec  B2-1-1 Na-rich 3) and Na > K + Ca  B2-1-2 K-rich 3) and K > Na + Ca  As the results, Na + Mg  B2-1-3 Ca-rich 3) and Ca > Na + K + Al + Si + K + Ca + Ti  + Fe > ca. 85% B2-1-4 Na-Ca other than B2-1-1,2,3 and Ca > K  B2-1-5 Na-K other than B2-1-1,2,3 and K > Ca  B2-2 Mg-Fe rich type 2) and Na + K + Ca < Mg + Fe  B2-3 Ca-Fe rich type 2) and Na + K + Ca = Mg + Fe  B2-4 Ca-Ti rich type 2) and Ca + Ti > 30%, Ti > 10%  C P-rich P > Si D Al-rich Al > Si E Fe-rich Fe > Si F Mn-rich Mn > Fe G S-Fe rich S + Fe > 50%, S/Fe = 2.7 Ð 3.4 as the result H Soot Only Si is detected with high background. I Aggregates or biomineral-like Discriminated from the shape

a, b and c indicate the particle types B, B2 and B2-1, respectively.

specimen.

Determination of dilute-acid soluble components by ICP-MS Twenty elements in two (A and B) 0.45 µm filtrates were determined by ICP-MS. Standard solutions for calibration were prepared by diluting a multi-element mixing standard solution (ICP Multi Element Standrad IV, 1 mg mlÐ1 each, Merck, Darmstadt, Germany) and several 1 mg mlÐ1 single element standard solutions (Wako Pure Chemistry Co., Japan) with high-purity water. An HP4500 ICP-MS was used for the determination of multi-elements. The plasma conditions used were the following: an r.f. power of 1.2 kW was applied to the plasma and flow rates of 15.4, 1.0 and 1.2 l/min were used for the coolant, auxiliary and nebulizer gases, respectively. The sample uptake was 0.25 ml/min. The data acquisition was as follows: the conditions of 300 ms/channel, 3 points/peak and 3 were used for the dwell time, data points and number of scans, respectively. Fig. 3 Abundances of the particles containing phosphorus, sulfur, chlorine, manganese and copper in river waters (a) and the several source materials (b). Results and Discussion

Phosphorus, sulfur, chlorine, manganese and copper as effective indicators of labile substances adsorbed on individual the detection limits19 of SEM-EDX are not enough to detect particles collected on a 0.45 µm filter these elements, having contents below approximately 0.01 Ð A “Labile” substance in this study was defined as one 0.1%. Thus, individual mineral particles from igneous rocks secondarily formed in river water and adsorbed afresh on being the major source material of large suspended particles in mineral particles; that is, it was not detected significantly in river water do not contain significant amounts of these original or fresh mineral particles from igneous rocks being the elements. Therefore, most of them detected in insoluble major source material of large suspended particles in river individual particles from river water can be considered to not be water. In this study, phosphorus, sulfur, chlorine, manganese the original constituents of aluminosilicates, but to be labile and copper were used as effective indicators of labile substances adsorbed afresh, or secondarily formed in water. substances. Figure 3 shows the abundances of phosphorus, Original phosphorus-, sulfur- and manganese-rich particles do sulfur, chlorine, manganese and copper detected in individual not affect the abundances, as described below. It was reported particles from river water and several source materials. Figure that individual particles in natural water have a surface layer 3(a) expresses all of the particles measured other than diatom. enriched in organic matter and manganese.5 The present result As shown in Fig. 3(b), the detection frequencies of these suggests that anionic elements, such as phosphorus, sulfur and elements in individual particles from igneous rocks were quite chlorine, besides manganese, are also enriched in the surface low or negligible compared to those from river water, because layer. 838 ANALYTICAL SCIENCES JUNE 2003, VOL. 19

(a) Fuji river 20 µm

Fig. 4 Concentrations of dilute-acid soluble elements from the particles collected on the 0.45 µm filter. (b) Sagami river 10 µm

Relationship between the insoluble labile substance adsorbed on individual particles and the dilute-acid soluble component Figure 4 shows the concentrations of dilute-acid soluble elements from the particles collected on a 0.45 µm filter. In the twenty elements (sodium, magnesium, aluminum, potassium, calcium, vanadium, chromium, iron, manganese, cobalt, copper, zinc, gallium, strontium, molybdenum, cadmium, caesium, barium, tungsten and lead) measured, ten elements, as shown in Fig. 4, were easily extracted by dilute-acid. Other elements were not detected. The concentrations of the dilute-acid soluble elements were predominantly high at sites F2 and F3 from Fuji river. Especially, manganese is one of the predominant elements together with iron and aluminum. The regional distribution in the number of individual particles containing manganese (Fig. 3a) corresponds well to it (Fig. 4) of the dilute- acid soluble manganese concentration; that is, the main stream of the Fuji river shows their higher values, although site S2 Fig. 5 SEM images of type-I particles (aggregates or biomineral- shows a high detection frequency of particulate manganese in like substances) from Fuji (a) and Sagami (b) rivers. spite of the low soluble component amounts. The components which can be released easily from a particle by such a dilute- acid are not oxides or silicates, but hydroxides.4,10,22,23 There is a possibility that such components dominate the biogeochemical Classification of particle type based on the relative X-ray behavior of dissolved trace elements.4,5 Also, the content of intensity humic acid in particles larger than 0.45 µm is usually low.24 In order to estimate the detail adsorption behavior of labile Therefore, these results suggest that most of the manganese substances to mineral particles in river water, individual detected in individual particles collected on a 0.45 µm filter particles measured by SEM-EDX were classified into seventeen exists as easily soluble forms, and forms mainly hydroxides groups, as described in the experimental section and Table 1. together with iron and aluminum. Figure 5 shows examples of SEM images of type-I particles Thus, the labile manganese forms mainly the hydroxides (aggregates or biomineral-like substances), discriminated from together with labile iron and aluminum and adsorbs on the the shape. Particle of types A to H were dealt with as single surface of individual mineral particles. Aluminum, iron and particles, even if components adsorbed secondarily existed on manganese are more effective adsorbers of metals than organic the surfaces of the particles. Figure 6 shows a comparison matter.5 Therefore, it is suggested that not only heavy metals, among the relative abundances at each site of the different but also anionic elements, such as phosphorus, sulfur and particle types, as defined in Table 1. Furthermore, the type-I chlorine, form complexes with the hydroxides and/or are particle discriminated from the shape firstly was also re- incorporated in them. Judging from these points, water from the classified using the relative X-ray intensity, as shown in Fig. 7. Fuji river, which is mainly characterized by andesitic and As overall characteristics, the aluminosilicate particle types granitic rocks, may have a higher removal efficiency of dominated in most samples. This result indicates that the dissolved species than those from the Sagami river which is contribution from anthropogenic materials to the present study characterized by basaltic rocks. In conventional EPMA area was not very much. Therefore, the particle type studies11Ð18 of individual particles from river or lake water, such classification method (Table 1) proposed in this study is mainly dilute-acid leaching experiment was not performed. based on a detailed classification of aluminosilicates. The abundances of Si-rich and Na- and K-rich aluminosilicates are ANALYTICAL SCIENCES JUNE 2003, VOL. 19 839

Fig. 7 Re-classification results of the type I particles (aggregates or biomineral-like substances) by the relative X-ray intensity. F2 and F3, Fuji river; S2, Sagami river.

Fig. 6 Relative abundances of the different particle types other than diatom detected in Fuji (F1 Ð F4) and Sagami (S2) rivers.

higher in Fuji river, and those of Ca- and Fe-rich aluminosilicates are higher in Sagami river. Trace elements like vanadium, chromium, nickel and zinc were detected in only the type E (Fe-rich) from Sagami river. In general, granitic and basaltic rocks contain predominantly felsic minerals of Si- and Na-K-Ca rich (e.g. quartz, plagioclase, alkali-feldspars, etc.), and mafic minerals of Mg-Fe rich (e.g. olivine, pyroxene, amphibole, biotite, etc.), respectively.25 Therefore, broadly speaking, the difference in the particle type composition in both rivers seems to reflect the geological characteristics. A type-G particle (S-Fe rich), like pyrite, which may reflect the redox states of river water, was detected only in the main stream of Fuji river, although the particle abundance was quite low. The regional differences in the abundances of particle types G and I may relate to the generation efficiency of hydroxides in river water and the adsorption behavior to mineral particles. The predominant components in the Fuji and Sagami rivers are type B2-2 (Mg-Fe rich type of aluminosilicates) and type I, Fig. 8 Abundances of labile phosphorus, sulfur, chlorine, respectively. The chemical compositions of type I in Fuji and manganese and copper by the particle type in river water. The types C, G and F were excluded from the calculation of the abundances for Sagami rivers mainly correspond to type B2-2 and type B2-1-3 phosphorus, sulfur and manganese, respectively. (Ca-rich aluminosilicates), respectively, as a result of re- classification of type I in terms of the chemical compositions. Therefore, the highest abundance component in Sagami river can be considered to be type B2-1-3, although the particle type B2-2 and type I, where type I mainly corresponds to type B2-2 distribution in Fuji river is not changed. However, the result in terms of the chemical composition, as shown in Fig. 7. The that the Mg-Fe rich aluminosilicates (type B2-2) are the highest detection frequencies in Sagami river were high in type I, where abundance component in Fuji river considerably contradicts the type I mainly corresponds to type B2-1-3 in terms of the general characteristics25 of igneous rocks. The reason can be chemical composition, as shown in Fig. 7. However, although considered to be as follows. The classification given in Fig. 6 is Si-rich (A), and Na- (B2-1-1), K- (B2-1-2) and Na-Ca (B2-1-4) an expedient one, as has been done by many researchers up to rich aluminosilicates, such as quartz, alkali feldspars or now, because the contributions of labile substances adsorbed plagioclase, are major minerals in Fuji river, labile elements secondarily on individual mineral particles are not yet presently were rarely detected in them. Therefore, the particles which considered. Therefore, the selective adsorption behavior of predominantly adsorb labile substances are type B2-2 and type labile substance to individual mineral particles and the effect on B2-1-3 for Fuji and Sagami rivers, respectively. However, as the particle type classification are discussed as follows. described above, type B2-2 especially from Fuji river is an expedient existence. The real shape of the particles classified as Abundance of labile substances by the particle type Mg-Fe rich aluminosilicates was clarified as follows. Figure 8 shows the abundances of labile phosphorus, sulfur, chlorine, manganese and copper by the particle type. These Mg-Fe rich aluminosilicates elements can be considered to be effective indicators of labile Figure 9 shows the typical X-ray spectras of type B2-2 from substances, as described above. The detection frequencies of Fuji and Sagami rivers. (1) The excitation efficiency of the these elements in Fuji river were predominantly high in type characteristic X-ray is similar for manganese and iron. 840 ANALYTICAL SCIENCES JUNE 2003, VOL. 19

in the individual particles in spite of similar concentrations of dilute-acid soluble elements. However, although Si-rich, and Na-, K- and Na-Ca rich aluminosilicates, such as quartz, alkali feldspars or plagioclase, are major minerals in Fuji river, they do not significantly adsorb labile substances. Therefore, mafic minerals, and felsic minerals such as quartz, albite, orthoclase or plagioclase, do not significantly adsorb labile substances. The lower concentrations of dilute-acid soluble elements in Sagami river may be related to the geological nature, which is characterized by basaltic rocks containing larger amounts of mafic minerals. However, felsic mineral, such as anorthite, is an effective adsorber of labile substances, especially in Sagami river, although an effective adsorber in Fuji river contains significantly not only calcium, but also magnesium, as the major constituents of aluminosilicate. Thus, the adsorption behavior of labile substances to individual mineral particles shows remarkable selectivity. Consequently, these results indicate that Mg and/or Ca-rich aluminosilicates in river water play important roles as scavengers of hydroxides adsorbing dissolved trace elements, and affect the elemental fractionation between the liquid and solid phases. However, further research is needed in order to clarify the difference in the types of effective adsorbers in both rivers and the relation with the generation efficiency of Fig. 9 Comparison among the typical X-ray spectra of the type B2- hydroxides. 2 particle (Mg-Fe rich aluminosilicates) from Fuji (a) and Sagami (b) rivers. References

Therefore, their relative X-ray intensity ratio directly 1. B. Salbu and E. Steinnes (ed.), “Trace elements in natural corresponds to the concentration ratio. (2) Fe/Mn from the waters”, 1995, CRC Press, London. dilute-acid soluble components, which suggest a labile 2. Y. Sakai, K. Ohshita, S. Koshimizu, and K. Tomura, J. substance adsorbed afresh in river water, is 2.3 Ð 5.8 and is Radioanal. Nuc. Chem., 1997, 216, 203. similar to the relative X-ray intensity ratio of them in type B2-2 3. S. Koshimizu and K. Tomura, in “Groundwater Updates”, from Fuji river, as shown in Fig. 9(a). (3) True Mg-Fe rich ed. K. Sato and Y. Iwasa, 2000, Springer-Verlag, Tokyo, aluminosilicates, like olivine, are originally predominant 171 Ð 176. components in basaltic rocks.20,25 (4) Manganese was rarely 4. T. Akagi, K. Fuwa, and H. Haraguchi, Anal. Chim. Acta, detected not only in fresh mineral particles from igneous rocks, 1985, 117, 149. but also in olivine, being typical Mg-Fe rich aluminosilicates, as 5. T. Akagi and H. Haraguchi, Geochem. J., 1994, 28, 81. shown in Fig. 3(b). The result also corresponds well to the 6. M. Sugiyama, Bunseki Kagaku, 1996, 45, 667. characteristics of the original Mg-Fe rich aluminosilicates from 7. A. Itoh, T. Nagasawa, and H. Haraguchi, Anal. Sci., 2001, Sagami river, which is characterized by basaltic rocks, as shown 17 (Supplement), i13. in Figs. 8 and 9(b). Namely, original Mg-Fe rich 8. P. K. Appelblad, D. C. Baxter, and J. O. Thunberg, J. aluminosilicates, like olivine, do not significantly adsorb labile Environ. Monit., 1999, 1, 211. substance. All things considered, almost all parts of the 9. H. Haraguchi, A. Itoh, S. Ji, and T. Yabutani, Environ. Sci., manganese and most of the iron in type B2-2, especially from 2000, 13, 313. Fuji river, can be regarded not as the original constituents of 10. S. Kawakubo, K. Tachikawa, and M. Iwatsuki, J. Environ. aluminosilicates, but as secondarily hydroxides formed and Monit., 2002, 4, 263. adsorbed afresh in water. Namely, most of type B2-2 from Fuji 11. J. Jedwab, Geochim. Cosmochim. Acta, 1979, 43, 101. river classified in Fig. 6 should be considered to be essentially 12. S. Ramamoorthy and A. Massalski, Environ. Geol., 1979, Mg and/or Ca rich aluminosilicates having low iron contents. 2, 351. Therefore, the abundance of true Mg-Fe rich aluminosilicates is 13. N. Kaufherr and D. Lichtman, Environ. Sci. Technol., 1984, higher in Sagami river than Fuji river. This idea does not 18, 544. contradict the distribution25 of minerals in igneous rocks. 14. K. Norrish, H. Rosser, and L. J. Warren, Appl. Geochem., 1986, 1, 117. Selective enrichment of labile substances to the surface of 15. J. Osán, Sz. Török, K. Török, L. Németh, and J. L. Lábár, particular mineral particles X-ray Spectrom., 1996, 25, 167. The abundances of labile elements, such as phosphorus, 16. W. Jambers and R. Van Grieken, Environ. Sci. Technol., sulfur, chlorine, manganese and copper, by the particle type are 1997, 31, 1525. predominantly high, not in Fe-rich aluminosilicates, but in Mg 17. B. L. W. Höflich, M. Wentzel, H. M. Ortner, S. Weinbruch, and/or Ca rich (for Fuji river) and Ca-rich (for Sagami river) A. Skogstad, S. Hetland, Y. Thomassen, V. P. Chaschin, aluminosilicates. It is attributed to differences in the and E. Nieboer, J. Environ. Monit., 2000, 2, 213. abundances of Ca-rich type aluminosilicate groups that the main 18. L. Xu, K. Okada, Y. Iwasaka, K. Hara, Y. Okuhara, Y. stream of Sagami river and the tributaries of Fuji river show Tsutsumi, and G. Shi, Atmos. Environ., 2001, 35, 3145. extremely different detection frequencies of the labile elements 19. T. Kyotani and S. Koshimizu, Bull. Chem. Soc. Jpn., 2001, ANALYTICAL SCIENCES JUNE 2003, VOL. 19 841

74, 723. 639. 20. N. Miyaji, J. Geol. Soc. Jpn., 1988, 94, 433. 24. S. Kawakubo, S. Ichikawa, and M. Iwatsuki, Anal. Sci., 21. R. Asada and K. Tazaki, J. Geol. Soc. Jpn., 2000, 106, 597. 2002, 18, 651. 22. T. Kyotani and M. Iwatsuki, Anal. Sci., 1998, 14, 741. 25. N. Imai, S. Terashima, S. Itoh, and A. Ando, Geochem. J., 23. T. Kyotani and M. Iwatsuki, Atmos. Environ., 2002, 36, 1995, 29, 91.