Monchegorsk ecology of beautiful tundra








Leaching of nickel and copper from soil contaminated by metallurgical dust

Valery Barcan
Lapland Biosphere Reserve, Green Lane (Zeleny), 8, 184505 Monchegorsk, Murmansk, Russia Received 1 April 2001;
accepted 14 December 2001


Abstract

The paper presents the results of the laboratory percolation experiment simulated soil contamination by emissions from a Ni-Cu smelter. Humus (Ao horizon) columns were transferred to lysimeters from an illuvial, humic, ferriferous forest podzol site. Fine metallurgical dust containing Ni and Cu was layered on the columns and irrigated with sulphuric acid solutions at pH 3, 4, 5, and 6. Irrigation for 19 months indicated that the leaching of metals down the humus column was greatest at pH 6. Calculations indicated that it would take 160-270 years for complete leaching of Ni from the Ao layer, and 100-200 years for Cu, depending on the dust composition. Natural decontamination of affected soils will take centuries.
© 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Nickel; Copper; Leaching; Dust; Soil; Contamination

1. Introduction

Industrial airborne heavy-metal contamination of landscapes around nonferrous smelters is a well-known and widely spread phenomenon (Nriagu, 1988). Emissions of metallurgical dust are spread according to the wind direction and particle size (Rutherford and Bray, 1979; Hazlett et al., 1984; Steinnes, 1992; Barcan and Kovnat-sky, 1998). Soil is the main depot of heavy metals in dry land (Dobrovolsky, 1997; Alloway, 1995). Dust emissions from smelters using sulfide copper-nickel ores are similar, independent of their location (Kola Peninsula, Taymir Peninsula, Canada, Australia, Finland), owing to the fact that the same raw materials are used in similar metallurgical processes (Polpherov, 1978).

The main metal-containing compounds deposited on the landscape in the form of dust emissions from the smelter are pentlandite (Ni,Fe)9S8, pyrrotite Fe7Sg(Nix), chalcopyrite CuFeS2, chalcosite Cu2S, covellite CuS, cuprite Cu20, tenorite CuO, and metal copper and nickel. The fine fractions of the dust are enriched with lead, arsenic, and zinc. The quantity and composition of dust derived from different sources (metallurgical processes) varies according to the raw materials and the condition of the gas cleaning systems.

Airborne industrial emissions are distributed among the three media—air, water, and soil. Gaseous pollutants, such as sulfur dioxide, nitrogen oxides, and chlorine, are mainly present in the ambient air, although they are partly absorbed by soil and water. Particles of metallurgical dust, greater than 1 -2 \xm are deposited on flat surfaces such as soil and water. Particles less than 1-2 ^m behave as gases, and remain in the air for indefinite periods and can be transported in the atmosphere for thousands of kilometers (Kryuchkov and Makarova, 1989; Malakhov and Zyrin, 1988).

The level of environmental contamination by emissions of Severonickel Smelter Complex is described as follows. The area about 4000 km2 around the plant is found contaminated by nickel and copper by the factor 6-1500 times more than European background, dependently on the distance and direction from the pollution source (Barcan and Silina, 1996). The soils sampled from the center of this airborne pollution, including the inhabited territory of the town of Monchegorsk, were found to be contaminated by nickel and copper to levels 450 and 250 times higher in comparison to the background, respectively—9000-4000 mg/kg of soil (Barcan and Kovnatsky, 1998). By this reason, the plants (i.e., berries and mushrooms) growing over an area of at least 3000 km2 around the smelter complex contain toxic element nickel—well-known carcinogen (Duffus and Worth, 1996)—and berries and mushrooms are unsuitable for human consumption due to the elevated nickel concentration (Barcan et al., 1998). The urban and suburban areas at the Kola Peninsula, for example, near the towns of Monchegorsk and Nikel have become hot spots of environmental pollution. The contamination of plants and soils by toxic metals is dangerous itself, but it is not yet a calamity—it is an indication of disastrous pollution of environment. The industrial environmental pollution does not remain constant—after smelter activity is stopped the pollutant emissions are stopped too and the process of soil self-purification is beginning.

The heavy-metal leaching from a soil is the slow process— from decades to centuries (Bergkvist, 1989; Dumon-tet et al., 1990; Iimura, 1977; Tyler, 1978; Asche and Beese, 1985; Brammer, 1986; Jorgensen, 1991). Many researchers studied the behaviour of already ionized metals (Berggren, 1992; Rasmussen et al., 1988; Myrlyan et al., 1996).

The task of conducted investigation was to estimate the total velocity of leaching nickel and copper from the forest litter, contaminated by the metallurgical dust. The study of the physical and chemical transformations of metal compounds in soil was not included in the research. The laboratory lysimeter experiments have been carried out to reach this aim.



Table 1
Content of some chemical elements in the metallurgical dusts used for the lysimeter experiments
Content (%)
Dust Ni Cu Co Fe S As Se Si Ca Mg Al
No. 1 (copper bearing) 11.0 71.7 0.5 3.2 15.7 0.01 n.d. 0.2 traces traces traces
No. 2 (nickeliferous)8.4 3.2/td> 0.2 33.8 3.2 0.08 0.01 7.1 2.0 3.4 1.6


2. Materials and methods

The soil columns used in the experiments consisted of the organic layer (Ao, mor) from a ferriferous illuvial humus podzol in a mixed (pine/birch) forest. Roots and other large materials were removed from the humus samples. The humus was then homogenized, that is, the soil with a disturbed structure was used in the experiment.

The moist substrate was placed in a 1-L column having 10 cm diameter with a perforated bottom. A known amount of dust was then added to the soil surface. The weight of humus in the column was 141 g, and the height of the column 12 cm. Dust no. 1 was added to the top of the column at a dose of 1 g/column, corresponding to 850 mg Ni/kg humus and 5120 mg Cu/kg humus. Dust no. 2 was added to the top of the column at a dose of 2 g/column, corresponding to 1270 mg Ni/kg humus and 500 mg Cu/kg humus. These levels corresponded to the amounts measured in the organic layer at a forest site at a distance of 10-15 km from the smelter complex in the direction of the emission plume (Barcan and Kovnatsky, 1998).The columns were irrigated twice a week with distilled water adjusted to pH values ranging from 3 to 6 by the addition of sulphuric acid. The pH values of water used for the leaching simulations were chosen on the basis of precipitation pH (Kryuchkov and Makarova, 1989; Alexeyev, 1990; Status of environment and ecology problems at the Kola Peninsula, 2000)—pH 3-6, though the natural limit pH is 4-5. The total amount of irrigation was approximately equal to the volume of precipitation falling during the snowless period, that is, 100 mm/month (Yakovlev, 1961). The top of the columns was open. The experiments were conducted in closed premises and no special efforts were undertaken to support the stable air temperature; it was 20-23 0Ñ. Two or three drops of chloroform were added to the solution to prevent microbial activity. The leachate from the columns was collected each month and total Ni and Cu solution concentrations were determined.



Table 2
Some chemical properties of forest litter (horizon Ao) used for experiments
Total carbon (%) 73
Total nitrogen (%) 0.8
Alkaline-hydrolyzable nitrogen (mg/100 g) 79
Mobile phosphorus (mg/100 g) 20
Mobile potassium (mg/100 g) 71
pH (H20) 4.3
pH (KC1) 3.0
Exchangeable acidity (meq/100 g) 80
Exchangeable calcium (meq/100 g) 11
Exchangeable magnesium (meq/100 g) 1.6
Base saturation (%) 37
Total nickel (mg/kg) 73
Total copper (mg/kg) 34


The comparative experiments were conducted the quartz sand (0.1-1 mm) washed with hydrochloric acid and distilled water was used as the neutral substrate to evaluate the dynamics of leaching metals from the dust not complicated by its interaction with the soil.

The quart sand is a neutral carrier. The metallurgical dust superficialed on the sand reacts with acid, metal compounds are ionized and turn into water-soluble form. The metal leaching from dust in soil is a more complicated process. Interaction between precipitation acid and the dust, the sorption of sulfates by the soil, and subsequent metal desorption from the soil are possible. A direct interaction between dust and soil acids, and subsequent desorption of formed compounds, are possible, too.

The duration of the experiments "litter+dust" and "sand + dust" was 19 months, and repetitions of experiments were 3. The total number of columns was 51.

The metallurgical dust ( - 63 |im) for the experiments was taken from the gas flow near the smoke ducts used in the smelting of the Ni-Cu ore and copper raw materials.

The chemical analyses of the dust were conducted after extracting of the dust samples with a mixture of concentrated HC1 and HN03 (3:1, aqua regia). The total metal concentrations in the solution were determined as follows: Ni, Cu, Co, Fe, Zn, Pb by flame atomic absorption spectrometry (AAS) on a Perkin Elmer 3030B spectrometer; Al, Cr, Mn, Se, V, W, Ti, Sr, Mo, Ca, Mg on an AtomScan 25 emission ICP spectrometer (Thermo Jarrell Ash); and Cd and As on a Perkin Elmer 3030B spectrometer using a furnace cuvette. The total sulphur concentration in the dust was determined on a LECO CS-444 instrument by ignition in a current of oxygen and sulphur dioxide determination by infrared detection (LECO CS-444, 1991). The total silicon concentration in the dust was determined by a standard method involving alkaline melting (Ponomaryov, 1955).



Table 3
Rate of dissolution of Ni and Cu from the dust in the control lysimeter
Dust Amount indust before experiment(mg) Amount after the end of experiment (mg) Rate of dissolution of metals (mg/month) Period of complete dissolution (years)
Ni Cu Ni Cu Ni Cu Ni Cu
No. 1 110 717 47 301 0.1 0.5 78 100
No. 2 1696513428 0.10.122347


The leachate collected from the columns was treated with a mixture of nitric and sulphuric acids in order to destroy the organic matter. The humus sections from the columns were ashed in a muffle furnace at 450-500 °C, and the ash extracted with a mixture of hydrochloric, nitric, and sulphuric acids. AAS was used for Ni and Cu determination in solutions. A standard colorimetric method was also used for determination of Ni and Cu in solution: nickel by dymethylglyoxyme and copper by lead diethyldithyocarbamate (Hillebrand and Lundell, 1953; Sandell, 1959).

The described methods of chemical analyses are widely used.

The main criteria to choose methods were sufficient sensibility and accuracy, minimum labour, and money expenditures.

3. Calculation and statistics

The standard statistic treatment of results was used. Linear regression was determined by the method of least squares. Nonorthogonal regression with minimization of the sum of the square roots of standard deviation on ordinate (function) was used. The type of regression function accepted depended on the calculated correlation coefficient, that is, after the calculation of 16 functions, we chose the one that had the maximum correlation coefficient and the minimum dispersion.

4. Results

Table 1 presents the chemical composition of the dust, used in the experiment. Table 2 presents some chemical soil properties.





Fig. 1. Results of lysimetric experiments with quartz sand as substrate. (A) Metal concentration in the leachate. (B) Leaching of metals from the dust, the cumulative proportion (%).






Fig. 2. Cumulative leaching of Ni from the humus during the lysimetric experiment (%).


5. Leaching of metallurgical dust on sand (neutral substrate)

During the course of the experiment, the Ni and Cu concentrations in the leachate and the cumulative amount of metals removed in the leachate both increased hyperboli-cally (Fig. 1).

Nickel in the experiment was extracted at a rate 0.1 mg per month from both species of the dust, copper—0.5 mg per month from dust no. 1 (copper bearing), and 0.1 mg per month from dust no. 2 (nickeliferous).

Table 3 provides estimates of the time required for the release of the Ni and Cu from the dust (control experiments on sand). Release of all the Ni will take 80-200 years, and all of the Cu 50-100 years.

6. Total leaching of Ni and Cu from the humus columns

Leaching of metals from the columns treated with either type of dust increased in proportion to the duration of the experiment (Figs. 2 and 3).

The metal amount increased with increasing pH of the irrigation water.

The experiment with dust no. 1 and pH 3 presented the exception. The leaching of Cu showed a similar trend (without exceptions), although the absolute values were less than for Ni, the total leaching of Ni to the end of experiments as percentage of the initial content was 3-5.5% from the soil with dust no. 1, and 1-2% from the soil with dust no. 2. Extraction of copper was 1-2.0% and 1.5-3.5%) for dust nos. 1 and 2, respectively (Fig. 4).

Copper is known to be more strongly bound in soils than Ni (Bergkvist, 1989; Dumontet et al, 1990; Berggren, 1992). Competition for absorption sites may occur when equal amounts of metals are present in the soil. This may explain why the leaching of Ni from the columns with dust no. 1 (Cu:Ni = 6) was markedly greater than that from the columns with dust no. 2 (Cu:Ni = 0.4).





Fig. 3. Cumulative leaching of Cu from the humus during the lysimetric experiment (%).






Fig. 4. The level of Ni and Cu leaching to the end of the experiment (%).


7. Discussion

7.1. The residue time for Ni and Cu

It is possible to make a rough estimate of the time required for the complete dissolution of Ni and Cu from the metallurgical dust deposited on the organic layer of forest soils. Table 4 gives the results of the calculations made for the different treatments. Ni from the dust no. 2 (nickeliferous) will be completely dissolved after 55 years, and the same from dust no. 1 (copper bearing) after 70-90 years. The corresponding values for Cu were 90 years (dust no. 2) and 140-250 years (dust no. 1).

The release of metals from the humus does not depend on their release from the dust. The metals are rapidly leached from the dust but are then absorbed in the soil, from where they are subsequently slowly released (e.g., 50 years). In reality, 100-200 years are required for the complete decomposition of the main mass of the dust. In practice, this means that when the last remnants of the metals from the dust are leached from the humus layer, then most of the remaining metals are derived. The time required for the complete removal of all the Ni from the Ao horizon is 160 years for dust no. 1 (copper bearing) and 270 years for dust no. 2 (nickeliferous), and for Cu 200 and 100 years, respectively.

These results coincide with ones known from the references: metals fixed in the soil are released and leached at a very slow rate—tens and even hundreds of years (Iimura et al., 1977; Tyler, 1978; Asche and Beese, 1975; Abrahamsen and Stuanes, 1986; Bergkvist, 1986; Rasmus-sen et al., 1988).

The removal of metals from the organic layer does not mean removal from the soil profile, but merely displacement down into the underlying horizons. The illuvial horizons (B and ÂÑ) of forest podzols accumulate heavy metals leaching from organic horizon. Ni and Cu in podzolic soils in the vicinity of the Severonickel Smelter Complex, polluted by heavy-metal emissions during the last 50 years distribute as follows: 79-84% of the Ni and 81-92% of the Cu have accumulated in the Â, ÂÑ, and Ñ horizons, and only 7-12% of the Ni and 5—12% of the Cu in the organic layer (Barcan and Kovnatsky, 1998). Further leached metals pass in ground waters, that is, soil self-purification does not mean environment purification—only the pollutant pass from the media to media.



Table 4
Rate of release of Ni and Cu from the forest humus
Substrate and load of Ni and Cu Irrigation pH Rate of release of metals (% / year) Duration of complete release of metals (years)
Ni Cu Ni Cu
Humus + dust no. 135.30.819125
Ni = 0.923 g/kg41.10.491250
Cu = 5.158g/kg51.40.771143
61.30.477250
Humus + dust no.231.80.756143
Ni-1.343 g/kg41.81.15691
Cu = 0.538g/kg51.81.15691


The pH of precipitation is often considered to have an effect on the mobility of heavy metals in soil; an increase in acid precipitation may decrease the soil solution pH, thus promoting the metal leaching from the soil (Brummer, 1986; Freiesleben and Rasmussen, 1986; Nedbaev, 1987; Brown, 1987; Rasmussen et al, 1988; Hogg et al., 1993; Bertelsen et al, 1994). At the same time, the published information suggests that the leaching of heavy metals from the organic soil horizon depends more on the formation and leaching of organic compounds (Schnitzer, 1969; Aleshchukin, 1972; Bergkvist, 1986; Elpatievsky and Lutsenko, 1990; Jorgen-sen, 1991; Varskog et al, 1994; Ladonin, 1996; Myrlyan et al., 1996; Derome and Lindroos, 1998).

The described lysimeter experiment has shown that the most leaching rate of nickel and copper was observed when irrigation was of slightly acid conditions (pH 5-6). This seems to be the essential evidence that formation and leaching of organic metal compounds are the determinants for leaching of heavy metals from soils, but not the precipitation acidity. Derome and Lindroos (1998) have shown in their field lysimeter experiments that nickel and copper are completely combined in organic-metal complex compounds in coloured soil solutions.

The complete, natural decontamination of soils polluted by metallurgical dusts will take centuries once pollution emissions have stopped. At the present time, the annual deposition of Ni and Cu on the soil around the Severonickel Smelter Complex will delay natural decontamination by tens of years.

8. Conclusion

Complete removal of heavy metals from the organic horizon of humus podzolic soils polluted by metallurgical dusts will take from 160 to 270 years for Ni and from 100 to 200 years for Cu, depending on the type of metallurgical dust. An increase in the irrigation water pH increased the amount of metals leached from the soil.

Acknowledgments

The authors wish to thank Mr. John Derome, Senior Scientist from Finish Forest Research Institute, for discussions and help in English language, and Ms. H. Kruglikova for the help in English.

References

Abrahamsen G, Stuanes ÀÎ. Lysimeter study of effects of acid deposition on properties and leaching of gleyed dystric Brunisolic soil in Norway. Water, Air, Soil Pollut 1986;31:865-78.

Aleshchukin L. Geochemistry of copper, nickel and iron in soils of Murmansk trans-polar area. In: Dobrovolsky V, editor. Materials to geochemistry of Kola Peninsula landscapes. Moscow: Moscow Teacher's Training Institute, 1972. pp. 69-84 (Russian).

Alexeyev VA, editor. Forest ecosystems and air pollution. Leningrad: Nau-ka (Science), 1990. 200 pp. (Russian).

Alloway BJ, editor. Heavy metals in soils. 2nd ed. London: Blackie Academic and Professional, 1995. 370 pp.

Asche N, Beese F. Untersuchungen zur Schwermetalladsorption in einem sauren Waldboden. Z Pflanzenernaehr Bodenkd 1985;149(2):172-80.

Barcan V, Silina A. The appraisal of snow sampling for environment pol¬lution valuation. Water, Air, Soil Pollut 1996;89:49-65.

Barcan V, Kovnatsky E. Soil surface geochemical anomaly around the copper-nickel metallurgical smelter. Water, Air, Soil Pollut 1998; 103: 197-218.

Barcan V, Kovnatsky E, Smetannikova M. Absorption of heavy metals in wild berries and edible mushrooms in an area affected by smelter emis¬sions. Water, Air, Soil Pollut 1998;103:173-95.

Berggren D. Speciation of copper in soil solutions from podzols and cam-bisols of S. Sweden. Water, Air, Soil Pollut. 1992;lll-23.

Bergkvist B. Leaching of metals from a spruce forest soil as influenced by experimental acidification. Water, Air, Soil Pollut 1986;31(3-4): 901-16.

Bergkvist B. Fluxes of Cu, Zn, Pb, Cd, and Ni in temperate forest ecosys¬tems. Water, Air, Soil Pollut 1989;47:217-86.

Bertelsen BO, Ardal L, Steinnes E. Mobility of heavy metals in pine forest as influenced by experimental acidification. Water, Air, Soil Pollut 1994; 73:29-48.

Brown KA. Chemical effects of ph3 sulphuric acid on a soil profile. Water, Air, Soil Pollut 1987;32(1-2):201-18.

Brummer GW. Heavy metal species, mobility and availability in soils. In: Bernhard M, Brinckman FE, Sadler PJ, editors. The importance of chemical "speciation" in environmental processes. Berlin-Heidelberg: Dahlem Konferenzen, Springer-Verlag, 1986. pp. 169-92.

Derome J, Lindroos A-J. Copper and nickel mobility in podzolic forest soils subjected to heavy metal and sulphur deposition in Western Finland. Chemosphere 1998;36(4-5):1131-6.

Dobrovolsky VV. Biospheric cycles of heavy metals and regulatory role of soil. Eurasian Soil Sci 1997;30(4):431-41.

Duffus JH, Worth HGJ, editors. Fundamental toxicology for chemists. Cambridge: The Royal Society of Chemistry, 1996. 327 pp.

Dumontet S, Levesque M, Mathur SP. Limited downward migration of pollutant metals (Cu, Zn, Ni and Pb) in acidic virgin peat soils near a smelter. Water, Air, Soil Pollut 1990;49:329-42.

Elpatievsky PV, Lutsenko TN. Role of water-soluble organic compounds in transportation of emission origin metals upon a profile of mountain brown soil. Pochvovedenie (Soil Science) 1990;6:30-42 (Russian).

Freiesleben NEV, Rasmussen L. Effects of acid rain on ion leaching in a Danish forest soil. Water, Air, Soil Pollut 1986;31(3-4):965-8.

Hazlett PW, Rutherford GK, Van Loon GW. Characteristics of soil profiles affected by smelting of nickel and copper at Coniston, Ontario, Canada. Geoderma 1984;32:273-85.

Hillebrand WF, Lundell GEE Applied inorganic analysis. London: Wiley and Sons, 1953. 267 pp. (New York).

Hogg DS, McLaren RG, Swift RS. Desorption of copper from New Zealand soils. Soil Sci Soc Am J 1993;57:361-6.

Iimura K, Ito H, Chino M, Morishita T, Hirata H. Behaviour of contaminant heavy metals in soil-plant system. Proc Inst SEM. Tokyo: SEFMIA, 1977. p. 357.

Jorgensen SS. Mobility of metals in soils. Folia Geogr Danica 1991;XIX: 104-14.

Kryuchkov VV, Makarova TD. Airborne affect on ecosystems of the Kola North Russia: Apatity, 1989. 96 pp. (Russian).

Ladonin DV. The study of transformation by soil of technogenically caused copper and zink in the simulation experiment. Agric Chem 1996;1: 94-9 (Russian).

LECO CS-444. Instruction manual, 1991.

Malakhov ST, Zyrin HG. Calculations of limit tolerable emissions by the heavy metal soil content. In: Heavy metals on environment and nature protection. Proceedings of ÀÍ-Union Conference, Moscow University, Dec 1987. Moscow University Publisher House, 1988. pp. 93-9 (Russian).

Myrlyan NF, AUard B, Hakansson L, Nykora VI. Laboratory simulation of copper and zink migration in the soil-water system with use of suc¬cessive leaching. Ecol Chem 1996;5(3):171-9 (Russian).

Nedbaev NP. The use of lysimeters in study of the behavior of mobile forms of Fe, Mn, Zn and Cu. Abstr Conf "Experimental and mathematical simulation in study of forest and swamp biogeoceonoses." Moscow, 1987. pp. 253-5 (Russian).

Nriagu JO. A silent epidemic of environmental metal poisoning? Environ Pollut 1988;50:139-61.

Polpherov DV. Geology, geochemistry and origin of deposits of copper-nickel sulfide ores. "Nedra" (Depths) Leningrad, 294 pp. (Russian).

Ponomaryov AI. Methods of chemical analysis of minerals and rocks. Moscow: Academy of Science USSR, 1961 (Russian).

Rasmussen L, Freiesleben NEV, Irgensen PV. Leaching of ions from a forest typic udipsamment by acidified throughfall. Geoderma 1988; 43(I):33-47.

Rutherford GK, Bray CR. Extent and distribution of soil heavy metal con¬tamination near a nickel smelter at Coniston, Ontario. J Environ Qual 1979;8(2):219-22.

Sandell EB. Colorimetric determination of traces of metals. New York: Wiley and Sons, 1959. pp. 398-403 and 598-9.

Schnitzer M. Reactions between fulvic acid, a soil humic compound and inorganic soil constituents. Soil Sci Soc Am Proc 1969;33(1):75-81.

Status of environment and ecology problems at the Kola Peninsula (1999 year), 2000. Murmansk. 192 pp. (Russian).

Steinnes E. Heavy metal contamination of terrestrial ecosystems in northern Norway from smelters on the Kola Peninsula. Finn Air Pollut Prev News 1992;3:19-22.

Tyler G. Leaching rates of heavy metal ions in forest soil. Water, Air, Soil Pollut 1978;9(2):137-48.

Varskog P, Flaten TP, Steinnes E. Relation between elemental concentra¬tions and soil organic matter content in Norwegian forest soils. In: Allan JO, Nriagu JO, editors. Heavy Metals in the Environment. Proc Int Conf Toronto, vol. 2 1994.

Yakovlev B. Climate of Murmansk prov. Murmansk: Murmansk Book Publisher House, 1961. 199 pp.



V. Barcan / Environment International 28 (2002) 63-68

Tel.: +7-815-36-5-07-36, +7-815-36-5-80-18; fax: +7-815-36-5-71-99.
E-mail address: lapland@monch.mels.ru (V. Barcan).

0160-4120/02/$ - see front matter © 2002 Elsevier Science Ltd. All rights reserved. ÐÏ: S0160-4120(02)00005-3