VZJ Download to Citation Manager
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (7)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Laegdsmand, M.
Right arrow Articles by Keiding, K.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Laegdsmand, M.
Right arrow Articles by Keiding, K.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Laegdsmand, M.
Right arrow Articles by Keiding, K.
Related Collections
Right arrow Batch Studies
Right arrow Organic Compounds
Published in Vadose Zone Journal 3:451-461 (2004)
© 2004 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA

SPECIAL SECTION: COLLOIDS AND COLLOID-FACILITATED TRANSPORT OF CONTAMINANTS IN SOILS

Pyrene Sorption to Water-Dispersible Colloids

Effect of Solution Chemistry and Organic Matter

M. Laegdsmand*,a, L. W. de Jongeb, P. Moldrupa and K. Keidingc

a Aalborg University, Dep. of Life Sciences, Environmental Engineering Section, Sohngaardsholmsvej 57, DK-9000 Aalborg, Denmark
b Danish Institute of Agricultural Sciences, Dep. of Agroecology, P.O. Box 50, DK-8830 Tjele, Denmark
c Aalborg University, Dep. of Life Sciences, Chemistry Section, Sohngaardsholmsvej 57, DK-9000 Aalborg, Denmark
* Corresponding author (mette.laegdsmand{at}agrsci.dk).

Received 9 July 2003.



   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 APPENDIX
 REFERENCES
 
Chemical sorption to mobile soil colloids is a controlling factor for colloid-facilitated chemical transport in the vadose zone and groundwater. We investigated sorption of pyrene to soil colloid suspensions originating from soils differing in organic matter content for different solution chemistries. Colloids were obtained from two soils with different organic matter contents but similar geological histories by three different methods: (i) chemical dispersion, (ii) mechanical dispersion in water, and (iii) spontaneous release in water. Batch sorption experiments were conducted at five pyrene concentrations, in either pure water or at two different concentrations of K+ and Ca2+. Generally, K+ addition enhanced pyrene sorption, whereas Ca2+ addition decreased sorption. The chemically dispersed colloids exhibited the highest pyrene sorption capacity and had the most nonlinear sorption isotherms, whereas whole soil had the most linear isotherm. Model calculations of the potential amounts of leachable pyrene illustrated the importance of including both colloid- and dissolved organic matter (DOM)-facilitated transport in risk assessment models when dealing with pyrene transport. The leaching potential of dissolved pyrene (with no DOM- and colloid-facilitated transport) was 5% of the leaching potential when both DOM- and colloid-sorbed pyrene was included.

Abbreviations: DOC, dissolved organic C • DOM, dissolved organic matter • DW, deionized ultrapurified water • HOC, hydrophobic organic compound • PAH, polyaromatic hydrocarbon • SOM, soil organic matter • SWDC, spontaneous water-dispersible colloids • TC, total colloids • TOC, total organic C • WDC, water-dispersible colloids


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 APPENDIX
 REFERENCES
 
SORPTION OF HYDROPHOBIC ORGANIC COMPOUNDS (HOC) to soil colloidal material is an important process when considering colloid-facilitated transport. While colloids released from soil generally contain the same minerals and type of organic matter as the clay fraction of the soil, the quantitative proportions of the different mineral and organic phases often vary (Kaplan et al., 1997; Kretzschmar et al., 1999). These differences may relate to the conditions under which the colloids are released from soil. Natural soil colloids are released from soil either by in situ mobilization from soil aggregates, raindrop impact, or surface erosive flow. In situ mobilization from soil aggregates results from moderate chemical dispersion due to lowering of the ionic strength by the exchange of soil water with rainwater. Release of colloids from the soil by raindrop impact or due to erosive flow results from a combination of mild chemical dispersion by low ionic strength rainwater and mechanical stress. Strong chemical dispersion, by monovalent cations and high pH, generally does not occur in agricultural soils. When investigating the sorption properties of the fine fraction of the soil, the normal procedure is to disperse the soil by strong chemical dispersion, isolate the clay fraction, and conduct sorption experiments on this fraction. This procedure ignores the fact that sorption properties of the chemically dispersed clay fraction may vary from the sorption properties of the mobile colloids released by mild chemical dispersion or mechanical stress. Colloids with high stability are often leached preferentially from a soil (Kretzschmar et al., 1995; Kaplan et al., 1997). These colloids may have gained their stability from an increased surface charge due to organic coatings (Kretzschmar et al., 1995; Kaplan et al., 1997) or by an expanded Stern layer due to specific adsorption of monovalent cations (e.g., Singh and Uehara, 1999). The preferential leaching of certain colloid types may lead to sorption characteristics of the mobile colloids differing from those of bulk soil and the clay fraction. This key aspect of HOC sorption to colloids and, thus, colloid-facilitated transport, has not been investigated in the past.

The amount of HOC sorbed to soil material depends mainly on the organic matter content of the soil material (foc) (Means et al., 1980). The quality of the organic matter is a secondary determinant for the amount of HOC sorbed to the soil material (Chiou et al., 1998; Gauthier et al., 1987; Xing and Pignatello, 1997). Bed sediment and soil organic matter (SOM) have been shown to have different sorption capacities (Kile et al., 1995). Hydrophobic organic compounds may have a higher affinity for certain particle size fractions. Wilcke et al. (1996) and Müller et al. (2000) found that polyaromatic hydrocarbon (PAH) had a higher affinity for the silt fraction of the soil. This was explained by a higher relative concentration of aromatic organic compounds in this fraction.

Soil organic matter consists of mainly of humic materials that may be sorbed onto the mineral phase. Soil organic matter may have a more or less pronounced flexible nature. Humic material seems to be more condensed toward the center (Hayes et al., 1989) and have a diffuse outer boundary. The expanded and condensed parts of SOM may be described as rubbery and glassy polymer, respectively (Xing and Pignatello, 1997; Leboeuf and Weber, 1997). Sorption in the rubbery part of SOM is considered to be a phase-partitioning process by solid phase dissolution. In the glassy part of SOM the sorption process is surface adsorption by hole filling (sorption in nanovoids of the rigid SOM structure). Xing and Pignatello (1996) found an inverse relationship between the nanovoid volume and the Freundlich sorption isotherm exponent (N). This means that glassy SOM tends to exhibit nonlinear sorption of HOC, and rubbery SOM more linear HOC sorption.

High ionic strength may condensate SOM due to a decrease in the double layer interactions (Ghosh and Schnitzer, 1980). The binding of HOC to dissolved humic materials decreases with increasing ionic strength (Schlautman and Morgan, 1993; Jones and Tiller, 1999) resulting from this condensation. The presence of di- or polyvalent cations may also condensate the humic materials in the soil because of cross-linking (Pignatello, 1998; Murphy et al., 1994). Divalent cations have been observed to both increase and decrease the sorption of HOC compared with univalent ions. Murphy et al. (1994) and Jones and Tiller (1999) found that the sorption of HOC decreased with divalent cations, and Schlautman and Morgan (1993) found that it increased. The experiments of Schlautman and Morgan (1993) were conducted on river humic material and hence may not have been representative of the humic material in soils. Murphy et al. (1994), Jones and Tiller (1999) and Laor et al. (1998) found that the configuration of humic acids was altered and that the sorption capacity was generally lowered, when sorbed onto a mineral surface. Different minerals have different effects on the HOC sorption characteristics of the sorbed humic acids. The HOC sorption capacity of the sorbed humic materials also decreased with higher ionic strength and divalent cations (Murphy et al., 1994; Jones and Tiller, 1999). The water phase chemistry may also affect sorption, as changed polarity of the water phase changes the solubility of the HOC (Schwarzenbach et al., 1993).

We investigated the sorption properties of three different colloidal suspensions (representing chemical dispersion, mechanical release, and spontaneous release in water) from two agricultural soils. The objectives were to (i) determine if different release procedures of colloidal particles result in different HOC (pyrene) sorption properties, (ii) determine if particles released in a soil with high organic matter content have different HOC (pyrene) sorption characteristics from those released in a soil with low organic matter content, (iii) investigate the effect of solution chemistry on sorption of HOC (pyrene) to soil colloids, and (iv) to evaluate the effect of these differences on the facilitated transport of HOC (pyrene).

Three different dispersion methods were used to isolate the colloid fractions: (i) chemical dispersion and mechanical disruption (for total colloids [TC]) releasing most of the clay fraction, (ii) mechanical disruption (water-dispersible colloids [WDC]) resembling overland erosive mobilization of colloids (Miller and Barahuddin, 1986), and (iii) spontaneous release (spontaneous water dispersible colloids [SWDC]) resembling the mobilization of colloids when water infiltrates into the subsurface environment. It is assumed that the low mechanical breakdown and mild chemical dispersion in the SWDC method resemble the processes of in situ release of natural colloids in soil. Pyrene was chosen as the model substance since it has low degradability and is exclusively sorbed to the organic fraction. Pyrene is a PAH with four rings and has a relatively low water solubility (0.14 mg L–1). The major sources of pyrene contamination are tar, fossil fuel combustion, and forest and agricultural fires. Hence, pyrene also has practical relevance for colloid-facilitated transport in soil.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 APPENDIX
 REFERENCES
 
Soils
Two geologically similar soils were used in the experiments. The soils originated from the same geological site but with different management histories and organic matter contents. Both soils were sandy loams and Glossic Phaozems according to the FAO scheme (Krogh and Greve, 1999). One soil was organic matter enriched from an organically operated dairy farm with grass in the rotation ("manured soil"). The other was organic matter depleted, having been under annual cereal crops for the last 20 yr ("depleted soil"). The soils were sampled at three different sampling points in each of two neighboring fields in the northeastern part of Denmark (Schjønning et al., 2002). Table 1 shows some basic soil properties of the two soils.


View this table:
[in this window]
[in a new window]
  		Table 1. Soil physical and chemical properties (from Schjønning et al., 2002).

 
Colloid Suspensions
Three different colloid isolation methods were used: (i) TC, released by both chemical dispersion and mechanical disruption of the aggregates; (ii) WDC, released by mechanical disruption of aggregates, and (iii) SWDC with only spontaneous release (Table 2). After sieving air-dry soil (8-mm sieve), 45 g was packed into stainless-steel cylinders (50-mm diam., 30 mm long) with a 20-µm nylon mesh strapped to the bottom. The soil samples were then slowly prewetted with an artificial soil water solution (0.652 mM NaCl, 0.026 mM KCl, 2.44 mM CaCl2, and 0.255 mM MgCl2) and drained to –30 hPa. After 3 d of saturation and 4 d of subsequent drainage, the samples were immersed in deionized water (DW) or 0.1 M Na2CO3. The dry soil/liquid weight ratio was 1:8. Different degrees of mechanical stress were applied to the soil liquid mixtures (Table 2). The samples were left to settle in a sedimentation device. The sedimentation cut-off value was calculated using Stokes Law to particle size <2 µm (assuming spherical particles with a specific density of 2.65). The soil particles remaining in the sedimentation device from the TC suspension were resuspended in 0.1 M Na2CO3 and left to settle again five times. Larger particulate organic matter floating on the surface (light fraction) was removed from the suspensions by passing the suspension through a 20-µm nylon mesh after sedimentation. The TC suspensions were washed to remove excess Na+ by centrifugation (4180 g, 1 h, cut-off value >0.04 µm assuming spherical particles and specific density of 2.65), after which the supernatant was removed and the remainder resuspended in DW. The procedure was repeated five times, the final suspension being resuspended in artificial soil water. The colloid suspensions were not sterilized since commonly used sterilization methods (heating and addition of toxins) would change the physiochemical properties of the colloid suspensions. For the sorption experiments the colloid suspensions were diluted to approximately the same concentrations of organic matter.


View this table:
[in this window]
[in a new window]
  		Table 2. Colloid isolation procedures to obtain the three types of colloid suspensions.

 
Characterization of Colloid Suspensions
The undiluted colloid suspensions were characterized with respect to colloid concentration, total organic C (TOC), and dissolved organic C (DOC). The colloid concentration was measured by filtering 70 mL of diluted colloid suspension through a preheated and weighed glass-fiber filter (pore size 0.4 µm) and drying it at 105°C for 24 h. The filter was weighed after drying. Even though the filtering procedure does not capture all colloids in the 0.04- to 2-µm size, most of the colloid mass is captured on the filter since colloid mass varies with particle size to a power of three. The TOC and DOC of the colloid suspensions were measured on a Shimadzu TOC-5000A (Kyoto, Japan) equipped with a suspended particles kit. The DOC samples were centrifuged (4180 g, 1 h) and DOC measured on 3 mL of the supernatant. The fraction of organic matter (foc) was calculated using

[1]
where DM is the amount of dry matter in the suspension. (A complete list of variables used in the paper is given in the Appendix.)

The zeta potential was measured on the diluted suspensions using a Zetamaster (Malvern Instruments, Worcestershire, UK). Turbidity was measured using a Spectrophotometer UV-1601 at wavelength 650 nm (Shimadzu, Hiroshima, Japan). Particle size was measured using a Microtrac II particle size analyzer 7997-10 detection limits [0.7;700] µm (Leeds & Northrup Int., Sumney Pike, North Wales, PA) on the diluted suspensions. After centrifugation (400 g, 2 min) particle size was measured on the supernatant using the Zetamaster. The separation limit of this centrifugation procedure was 1 µm. Turbidity was also measured on the supernatant as well as the diluted suspensions. Turbidity was assumed to be proportional to the number of particles in the suspension. Total organic C was also measured on the centrifuged samples.

Pyrene
Radiolabeled pyrene (4,5,9,10,14C) with a specific activity of 58.7 mCi mmol–1 and nonlabeled pyrene with purity 99% were obtained from Sigma Chemical Co. (St. Louis, MO). A stock solution was prepared by adding radiolabeled pyrene in acetone and nonlabeled pyrene in acetone solution to deionized water and stirring for 24 h while slowly evaporating the acetone. The initial amount of acetone in the stock solution was 1:1000 before evaporation. The octanol–water partitioning coefficient, Log(Kow), was measured for pyrene using OECD guidelines (OECD, 1995). The water-phase corresponds to the electrolytic solutions used in the experiments (0.001 and 0.01 M CaCl2, 0.003 and 0.03 M KCl, and DW) using four replicates. Log(Kow) varied between 5.18 and 5.20 (in DW and 0.03 M KCl, respectively). These small differences in Kow suggest that reduced solubility of pyrene in the electrolytes has only minor effects on sorption. Experimentally determined Kow values vary from 5.1 to 5.3 (Nikunen et al., 1990; Karickhoff et al., 1979; Hansch and Leo, 1985).

Sorption Experiments
The experiments were conducted as 24-h nonequilibrium batch sorption experiments (five concentrations of pyrene ranging over two orders of magnitude and four replicates). The batches contained two different major cations (Ca2+ and K+) and had three levels of ionic strength: 0.001 M CaCl2 and 0.003 M KCl (0.003 meq L–1), 0.01 M CaCl2 and 0.03 M KCl (0.03 meq L–1), and deionized ultrapurified water (DW; 1 x 10–7 meq L–1). The maximum concentration of pyrene in the batches did not exceed 50% of total pyrene water solubility. Preliminary sorption kinetics experiments revealed that pyrene sorption to the water dispersible colloid fractions in 0.01 M CaCl2 of the two soils were in the linear part of the sorption process after 24 h, but not in equilibrium (Fig. 1) . A reaction time of 24 h was used to ensure that the degradation of pyrene was negligible. Desorption isotherms in 0.03 M KCl, 0.01 M CaCl2, and DW were investigated on the WDC fraction from the manured soil. Sorption isotherms using whole soil were determined on both soils in 0.01 M CaCl2.



View larger version (19K):
[in this window]
[in a new window]
  		Fig. 1. Kinetics of the adsorption process using the water-dispersible colloids (WDC) suspensions of the two soils.

 
The laboratory glassware used for the sorption experiments were soaked in 20% H2SO4 for 4 h, washed six times with deionized water, and heated to 500°C for 4 h before use. The sorption experiments with colloid suspensions were conducted in 10-mL centrifuge tubes. The tubes were filled with 1.5 mL diluted colloid suspension, 0.5 mL salt solution, and 3 mL pyrene solution. The colloid concentration in the batches varied from 38 to 103 mg L–1 among the different suspensions. The tubes were sealed with Teflon-lined caps (the caps were washed with acetone) and rotated end-over-end for 24 h. The dissolved pyrene was separated from the sorbed pyrene by centrifugation (4180 g, 1 h), and the activity of 3 mL of supernatant was measured by Liquid Scintillation Counter Packard Tri-Carb 2250 CA, Packard Instrument Company (Perkin Elmer, Boston, MA) using Ultima Gold from Packard Bioscience (Groningen, The Netherlands) as the scintillation cocktail. The glass vessels and Teflon caps of the centrifuge tubes sorbed considerable amounts of the pyrene (about 10%). This was included in the calculations using a linear isotherm with an intercept of zero (R2 = 0.95) for describing the sorption to glass and Teflon liners (the linear isotherm was based on 120 measurements with no colloids). Aluminum-lined caps were also tested, but no statistical difference was found between the mass loss to Teflon and the aluminum caps. We believe that the very high mass loss occurred because the sorbent was only present in very low concentrations in the suspension sorption experiments (0.2–0.5 mg batch–1). The actual amount of sorbed pyrene (Msorb) was calculated using

[2]
where Mtotal is the amount of pyrene added to the batch, C is dissolved concentration, V is the volume, and Mcorr = KcorrC is the amount of pyrene sorbed to the glassware and Teflon caps, where Kcorr is the slope of the linear correction isotherm.

The sorption experiments with whole soil were conducted in 50-mL Pyrex bottles with Teflon-lined caps. After sieving (2-mm sieve), 60 mg of air-dry soil was added to each bottle. The soil was equilibrated with 15 mL of DW for 24 h before the experiments. A 30-mL pyrene solution and 5 mL of salt solution were added and the bottles were rotated end-over-end for 24 h. Subsequently, 5 mL was extracted from each bottle with a glass pipette and centrifuged before measuring 3 mL of supernatant by liquid scintillation counting. The mass loss in the whole soil sorption experiments was <1%.

A sorption experiment was conducted on the WDC fraction of the manured soil with 0.01 M Ca2+ with a mixture of CaCl2 and Ca(OH)2 resulting in six different pH values between 6.07 and 8.55 (results not shown). There was no significant difference in sorption between the different pH values. Therefore, pH effects on pyrene sorption were not considered in the main sorption experiments where pH varied between 6 and 7.

Desorption experiments were conducted in 10-mL acid washed centrifuge tubes with Teflon lining. The mixtures in the batches were identical to those of the sorption experiments. After 24 h of end-over-end rotation, the desorption test tubes were centrifuged (4180 g, 1 h) and 3 mL of supernatant replaced with colloid-free suspension. The tubes were again rotated end-over-end for 24 h, and the activity of the supernatant was measured by liquid scintillation counting and replaced with colloid-free solution. This procedure was repeated five times. Colloid-free solution was prepared by mixing 1.5 mL colloid suspension, 0.5 mL salt solution, and 3 mL DW for 24 h. The mixture was centrifuged, and 3 mL supernatant was extracted.

Sorption Parameter Estimation
The data for sorbed pyrene concentration (corrected for sorption to glass and Teflon lining) as a function of dissolved pyrene concentration were fitted to Freundlich isotherms (Eq. [3]) using a generalized linear model (the GENMOD procedure in the statistical software package SAS [SAS Institute, 1999] with a logarithmic link and {gamma} distribution):

[3]
where S is the amount of pyrene sorbed to the solid phase (mg kg–1), C is the water phase concentration (mg L–1) Kf,a is the Freundlich sorption coefficient, Na is the Freundlich exponent describing nonlinearity of the sorption process, and subscript "a" refers to adsorption.

A organic matter normalized Freundlich equation was also used to compare the different isotherms statistically,

[4]
where Soc is the sorbed pyrene relative to the organic C content (mg kg–1 organic C) and Kf,oc,a is the Freundlich coefficient normalized with respect to organic C content (L kg–1). Different combinations of normalized isotherms were studied using the generalized linear model GENMOD (SAS Institute,1999) to test if factors such as soil type, chemistry, and suspension type could be removed from the estimation of the entire isotherm, or from the estimation of Na, without significant loss of accuracy (statistical significance level 0.05).

To evaluate the sorption capacity of the colloids at a dissolved pyrene concentration within the experimental range, Kd,C (Eq. [5]) and Koc,C (Eq. [6]) values were calculated at C = 0.01 mg L–1 pyrene using the estimated values of Kf,a and Na and the measured values of foc as follows

[5]

[6]
where Kd,C is the distribution coefficient of pyrene between the solid and water phases at a given dissolved concentration of pyrene, C (here 0.01 mg L–1), and Koc,C is the distribution coefficient between pyrene sorbed to organic C and dissolved in water at dissolved pyrene concentration C.

Karickhoff et al., (1979) suggested that the Koc in soils may be estimated from Kow using

[7]
where Koc is the distribution coefficient of pyrene between organic C and water and Kow is the octanol–water distribution coefficient.

The data from the desorption experiment were also fitted to a Freundlich isotherm,

[8]
where Kf,d is the Freundlich desorption coefficient and Nd is the Freundlich desorption exponent. The sorption nonsingularity index ({omega}) as proposed by Ma et al. (1993) was calculated as

[9]
where subscripts "a" and "d" denote adsorption and desorption isotherms, respectively.

Model Calculations
The leaching potential of pyrene in the two soils may be evaluated using the results from the sorption experiments. In soil, pyrene may be in solution, sorbed onto the stationary soil, sorbed to mobile colloids, or sorbed to dissolved organic matter. The mass balance of pyrene in soil is

[10]
where X is the total pyrene concentration of the soil (mg pyrene kg–1 soil), Ssoil is the concentration of pyrene sorbed to stationary soil (mg kg–1), Scoll is the concentration of pyrene sorbed to mobile colloids (mg kg–1), SDOC is the concentration of pyrene sorbed to DOC (mg kg–1), C is the dissolved concentration of pyrene (mg L–1), {theta} is soil water content (m3 m–3), {rho}bulk is the bulk density of the soil (kg m–3), coll is the amount of dispersed colloids (kg kg–1 soil), DOC is the amount of soluble DOC (kg kg–1 soil), and 103 L m–3 is a unit conversion factor.

We assumed that the amount of pyrene sorbed to mobile colloids and stationary soil may be described also using Freundlich isotherms,

[11]

[12]
where Kf,coll (L kg–1) and Kf,soil (L kg–1) are the Freundlich coefficients for the mobile colloids and stationary soil, respectively, and Na,coll and Na,soil are the Freundlich exponents.

The amount of pyrene sorbed to DOC may be described by a linear isotherm (Raber et al., 1998),

[13]
where KDOC is the partitioning coefficient of pyrene between organic C and water.

The potential amount of leachable pyrene, Cpot (mg L–1) consists of the colloid-bound and DOM-bound pyrene as well as dissolved pyrene, and is calculated as

[14]
using Eq. [10] through [13].


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 APPENDIX
 REFERENCES
 
Suspension Characteristics
Dispersible colloid concentrations and foc of the total colloidal fraction varied as a result of the different fractionation methods and soils (Fig. 2) . The TC fractionation method released the highest amount of colloids (76 and 87% of the clay fraction for the manured and depleted soils, respectively) with the lowest mass fraction of organic C. The SWDC fractionation method released <1% of the clay fraction. Colloid concentrations of the SWDC suspensions were in the same range as found in soil effluent (Villholth et al., 2000; Jacobsen et al., 1997; Laegdsmand et al., 2000; Kjaergaard et al., 2004). The lower amounts of colloids generally released from the manured soil compared with the depleted soil were probably a result of organic matter stabilization of the aggregates. The colloids from the manured soil had a higher mass fraction of organic C relative to the depleted soil.



View larger version (26K):
[in this window]
[in a new window]
  		Fig. 2. Dispersible colloids and organic matter content of three different suspensions from two soils. SWDC, spontaneous water-dispersible colloids; TC, total colloids; WDC, water-dispersible colloids.

 
The electrical conductivity of the diluted suspensions used in the sorption experiments was between 22 and 104 µS cm–1, with the highest electrical conductivity found for the TC suspensions and the lowest for the WDC suspensions (Table 3). The electrical conductivity was generally lower for the depleted soil than for the suspensions from the manured soil. pH values varied between 6.16 and 6.96, and the WDC generally had the higher pH. The zeta potential of the suspensions differed among suspensions and was lowest for the WDC suspensions. No differences were found between the two soils, except for the SWDC suspensions. The SWDC suspension of the depleted soil had significantly lower zeta potential than that of SWDC of the manured soil.


View this table:
[in this window]
[in a new window]
  		Table 3. pH, electrical conductivity, and zeta potential of the six different diluted suspensions.{dagger}

 
The mean particle size of the colloids measured with the Microtrac particle counter (the lowest detectable particle size was 0.7 µm) was around 10 µm and not significantly different among the different suspensions and soils (not shown). The particle size of the centrifuged suspensions (representing the colloids with a particle size <1 µm) fell in two distinct particle size classes: one above 0.3 µm (medium colloids) and one below 0.3 µm (fine colloids) (Fig. 3) . The particle size class below 0.3 µm, however, was absent in the SWDC suspensions (Table 4). In the TC and WDC suspensions the amount of medium size colloids was very low, and the amount of fine colloids was high. It seems that the chemical and mechanical dispersion converted medium colloids to fine colloids.



View larger version (19K):
[in this window]
[in a new window]
  		Fig. 3. The particle size distribution of the water-dispersible colloids (WDC) suspension of the depleted soil.

 

View this table:
[in this window]
[in a new window]
  		Table 4. Percentage of particle in particle size class 0.03 to 0.3, 0.3 to 1, and >1 µm for the suspensions.

 
The organic C content of the diluted suspensions was similar, but the partitioning to particles was very different (Fig. 4) . The TC suspensions had more C associated with the larger particles or flocs (particle size >1 µm). Most of these colloids were probably not single particles but stable microaggregates. The SWDC suspensions had high amounts of DOC and organic C associated with colloids <1 µm. The amount of organic C associated with the larger colloids or flocs were generally higher in the suspensions from the depleted soil. The SWDC suspensions contained no fine colloids (0.03–0.3 µm), and almost all colloid-associated C was present in the colloids <1 µm. The TC suspensions contained a distinct class of colloids smaller than 300 nm, and most of the organic matter was associated with particles >1 µm. This may be a result of chemical breakdown of microaggregates during preparation of the TC suspensions and subsequent release of organic matter and single clay particles. The remaining organic matter in the TC suspensions associated with the larger colloids or flocs was probably more resistant toward chemical breakdown. This different association of organic matter with different particle size classes probably reflects a difference in the nature of the colloid-associated organic matter between the different suspensions.



View larger version (29K):
[in this window]
[in a new window]
  		Fig. 4. Organic C in the different suspensions (SWDC, spontaneous water-dispersible colloids; TC, total colloids; WDC, water-dispersible colloids). DOC is dissolved organic C, POC < 1 µm is the organic C bound to particles smaller than one micrometer, and POC > 1 µm is the C bound to particles larger than one micrometer.

 
Sorption Isotherms
The adsorption data were described well using Freundlich isotherms (Fig. 5) . The estimated log(Kf,a) and log(Kf,a,oc) of the different experiments are listed in Table 5, and the estimated Na values are listed in Table 6. The organic C normalized adsorption isotherms for the different suspensions and whole soil in 0.01 M CaCl2 are shown in Fig. 6 . The isotherms from the two soils were not significantly different in TC and WDC suspensions and whole soil, thus indicating that the organic matter exposed in the whole soil was similar for the two soils. However, the sorption isotherms of the SWDC suspensions were significantly different for the two soils. This was also true for the other chemistries as well. A statistical analysis showed that soil type could be removed from the estimation of the whole soil, TC, and WDC isotherms, but not from the SWDC isotherms, without loss of accuracy.



View larger version (26K):
[in this window]
[in a new window]
  		Fig. 5. Sorbed vs. dissolved pyrene concentration for the water-dispersible colloid fraction in 0.01 M CaCl2, 0.03 M KCl, and deionized ultrapurified water (DW).

 

View this table:
[in this window]
[in a new window]
  		Table 5. The fitted log(Kf) and log(Koc) of the pyrene sorption experiments (using Eq. [3] and [4]). Standard error of estimate between 0.03 and 0.13.{dagger}

 

View this table:
[in this window]
[in a new window]
  		Table 6. The estimated values of the Freundlich nonlinearity parameter (Na) of the pyrene sorption experiments. The statistical analysis was made on the differences of the suspension types within one soil and one chemistry (e.g., manured soil in 0.001 M CaCl2).{dagger}

 


View larger version (35K):
[in this window]
[in a new window]
  		Fig. 6. Sorbed vs. dissolved pyrene concentration for the different suspension types (SWDC, spontaneous water-dispersible colloids; TC, total colloids; WDC, water-dispersible colloids) and whole soil in 0.01 M CaCl2. The sorbed pyrene is normalized toward organic matter.

 
Karickhoff et al. (1979) found that the ratio between Koc and Kow for soil and sediment samples was about 0.63. This was also true for our two whole soils investigated when the solution was 0.01 M CaCl2; the manured soil had a ratio of 0.63 and the depleted soil 0.62. The colloid suspensions generally exhibited a higher ratio, varying between 0.62 and 1.3 (Fig. 7) . This indicates that the colloids were enriched in hydrophobic organic matter compared with the whole soil. The TC colloids had the highest Koc/Kow ratio, indicating that these colloids, released during total breakdown of microaggregates, had more hydrophobic organic matter than the more loosely attached SWDC colloids. Addition of KCl also increased the Koc/Kow ratios. This is consistent with reconfiguration of the organic matter into a more open structure due to exchange of some of the cross-linking di- and polyvalent ions in the organic material with monovalent K+ (Pignatello, 1998), thereby exposing more of the organic matter interior to the solution. The sorption capacity of colloid-associated SOM increased with increasing ionic strength when K+ was present. This finding appears contradictive to Murphy et al. (1994), who found that mineral-bound humic acids had lower HOC sorption capacities when the ionic strength of monovalent cations increased. Also, fulvic and humic acids were condensed with increasing ionic strength of monovalent ions (Ghosh and Schnitzer, 1980; Murphy et al., 1994). The colloid-associated SOM had an apparently different response to the ionic strength of monovalent cations than mineral-bound humic acids. However, the colloids used in the experiments were retrieved from agricultural soils with Ca2+ dominating the exchange complex; the effect of K+ may be a result of ion exchange of Ca2+ with K+. When the ionic strength of KCl increases, an increasing amount of Ca2+ will be exchanged with K+ and sorption will increase. When Ca2+ was the dominating cation in the solution, the ionic strength had little or no effect on sorption. This was also observed by Murphy et al. (1994) for mineral-bound humic acids.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 7. Ratio between Koc,0.01 and Kow. Koc,0.01 was calculated from isotherm data according to Eq. [5] and [6], and Kow was measured on the relevant electrolytic solution. SWDC, spontaneous water-dispersible colloids; TC, total colloids; WDC, water-dispersible colloids.

 
All isotherms were nonlinear (Fig. 8) . For the two whole soils, the Freundlich exponent Na was about 0.9, but for the colloid suspensions it was smaller, ranging between 0.65 and 0.82 (Table 6). The TC suspensions of both soils exhibited more nonlinear isotherms than whole soil. Following Pignatello (1998), this is consistent with hole-filling due to the glassy structure of SOM. Hence, the SOM associated with TC suspensions had a more glassy structure than the SOM in whole soil. In TC suspensions, the microaggregates were dispersed; only the resistant organic matter remained on the colloids. In the manured soil, the Na values were significantly lower for the TC suspensions than for the SWDC suspensions, regardless of the chemistry (Table 6). For the depleted soil the isotherms had more similar nonlinearity (only significantly different between TC and SWDC suspensions in 0.03 M KCl and DW). The SWDC isotherms of the depleted soil were more nonlinear than the manured soil, suggesting that the easily detachable colloids from the organic matter depleted soil contain glassier SOM than the easily detachable colloids from the organic matter enriched soil.



View larger version (21K):
[in this window]
[in a new window]
  		Fig. 8. Estimated Freundlich exponent (Na) for the three different suspensions (SWDC, spontaneous water-dispersible colloids; TC, total colloids; WDC, water-dispersible colloids) from two soils.

 
Addition of Ca2+ seemed to have increased the Na value of the isotherms compared with K+ and pure water. For the manured soil, the effect of chemistry on Na was significant. For all of the suspensions of the depleted soil, however, it was possible to remove the effect of solution chemistry from the estimation of Na in the statistical model without significant loss of accuracy. This means that there was no effect of chemistry on Na in the suspensions from the depleted soil.

Desorption experiments were conducted on WDC suspension in DW, 0.01 M CaCl2, and 0.03 M KCl. Sorption nonsingularity was observed in 0.03 M KCl and pure water (DW) (Fig. 9) . The nonsingularity was more pronounced in pure water than in KCl. With addition of Ca2+, sorption was singular ({omega} {approx} 0). The nonsingularity of the sorption process was probably due to sorption nonequilibrium rather than resistant sorption due to the relatively short duration of the sorption and desorption experiments. The WDC suspension of the manured soil in pure water had a low sorption capacity and exhibited both nonlinear and nonsingular sorption. At low ionic strength, SOM will have a relatively open configuration but with some cross-linking of the organic material by naturally occurring polyvalent ions. Addition of Ca2+ increased the linearity and reversibility of the sorption process and produced a lower sorption capacity—this could be due to condensation and fixation of humic material by Ca2+, which reduce the apparent (nonequilibrium) sorption capacity by hindering diffusion into the interior of SOM. Upon addition of K+, the sorption was nonlinear and slightly nonsingular with a high sorption capacity. K+ increased the amount of available sorption sites compared with Ca2+ in all suspensions. When naturally occurring polyvalent cations are exchanged with K+ in SOM, it will expand, exposing more of the interior and leading to nonlinear and nonsingular sorption with high sorption capacity.



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 9. Pyrene adsorption and desorption isotherms for the water-dispersible colloids (WDC) fraction of the manured soil with 0.01 M CaCl2, 0.03 M KCl, and deionized ultrapurified water (DW).

 
Model Calculations
The introduction of colloid- and DOM-facilitated transport processes in risk assessment models is a future challenge. Figure 10 shows the results of simple model calculations of the leaching potential of pyrene (Cpot) (Eq. [14]). For the standard scenario (Scenario A), the measured amount of spontaneously dispersible colloids (SWDC) was used as an estimate for coll and the measured amount of DOC in the SWDC suspensions as an estimate for DOC in the equations (Table 7). The estimated isotherm parameters from the experiments with the SWDC suspensions and the whole soil were for the colloid isotherm and the whole soil isotherm, respectively (Eq. [11] and [12]). The SWDC dispersion method was used since it resembles the colloid mobilization process dominating the natural leaching from soil. Raber et al. (1998) found a log(Koc) of 4.6 for sorption of pyrene to natural soil derived DOC. This value was used as an estimate for KDOC in Eq. [13]. Soil water content was set to 0.3 m3 m–3 and soil bulk density to 1200 kg m–3. The different calculation scenarios are described in Table 8. The leaching potential of the standard scenario (Scenario A) using data from the depleted soil was higher than with data from manured soil (Fig. 10) because of a higher amount of dispersible colloids. The leaching potential of pyrene with no facilitation (Scenario C) was <4% of the leaching potential when both colloid- and DOM-sorbed pyrene was included on both soils (Scenario A). If only dissolved and DOM-sorbed pyrene were considered (Scenario B), the leaching potential was less than half of the leaching potential with colloid-facilitated transport for both soils. Thus, it is important to include both colloid and DOM-facilitated transport when evaluating the leaching potential of strongly sorbing compounds such as pyrene. The amount of leachable pyrene increased when colloid sorption characteristics of the experiments with 0.03 M KCl were used instead of 0.01 M CaCl2 (Scenario D). The amount of leachable pyrene was more affected by solution chemistry when the amount of pyrene in the soil was low.



View larger version (28K):
[in this window]
[in a new window]
  		Fig. 10. Model calculations: pyrene leaching potential calculated using the scenarios listed in Table 8.

 

View this table:
[in this window]
[in a new window]
  		Table 8. Scenarios used in the model calculations.{dagger}

 

View this table:
[in this window]
[in a new window]
  		Table 7. Parameters used for the model calculations. Measured in the spontaneous water-dispersible colloids (SWDC) suspensions.

 
If the Freundlich parameters of the WDC or TC suspensions were used as an estimate for the colloid sorption parameters (Scenario E and F, Fig. 10), the leaching potential was essentially the same as the standard scenario. However, the manured soil Scenario F showed more leaching with a low pyrene concentration in the soil than the standard scenario. Using the Freundlich parameters from whole soil resulted in underestimation of leaching on both soils (Scenario G). Whereas the whole soil could not be used as a model sorbent for the mobile colloids, the WDC could be used for this purpose with caution. If the colloid and soil isotherms were replaced with linear isotherms using Kd,0.01 (Scenario H, Fig 10), no change occurred in the leachable amount of pyrene. If the distribution coefficient calculated at a lower concentration was used (Kd,0.001, Scenario I), the leaching was overestimated. If the distribution coefficient calculated at a higher concentration was used (Kd,0.05, Scenario J), the leaching was underestimated. Caution should be exercised when using linear isotherms with the mobile colloids, since the sorption to the colloids is highly nonlinear in nature.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
APPENDIX
 REFERENCES
 
The different release processes—chemical dispersion (TC); mechanical disruption (WDC); spontaneous release (SWDC)—resulted in different sorption properties of the suspensions. The organic C normalized sorption capacity of the chemically dispersed colloid suspensions (TC) was higher than for the other two types of suspensions and the whole soil. The TC suspensions showed the most nonlinear isotherms. This indicates that the TC suspensions contained glassier SOM. The remaining organic matter associated with the colloids after the chemical dispersion was probably more condensed and resistant.

The pyrene sorption isotherms of the two soils were similar when normalized against organic C. Also, the isotherms of the chemically (TC) and mechanically dispersed (WDC) fractions of the colloids were similar when normalized against organic C content. The spontaneously released colloids of the two soils had different sorption isotherms when normalized against organic C. The sorption capacity of the spontaneously dispersed colloids was higher for the manured soil than the depleted soil, while the sorption was more linear on the manured soil.

In deionized water, the sorption properties of the colloidal suspensions were nonlinear. Addition of K+ to the sorption batches increased the sorption capacity significantly for all suspensions. Higher ionic strength increased the sorption capacity. This was probably due to expansion of the organic matter due to exchange of polyvalent ions with monovalent K+. Addition of Ca2+ reduced the sorption capacity and made the isotherms more linear. This is consistent with Ca2+ condensing the organic matter and blocking the diffusion paths to the interior glassy SOM.

Model calculations showed that more than 95% of the leaching potential of pyrene from the two soils was due to colloid and DOM sorption. Although the different fractionation methods gave different sorption properties, the potentially leachable amount of pyrene was almost the same when using the Freundlich estimates from TC, WDC, or SWDC. Whole soil Freundlich estimates gave considerably lower potential leaching. The model calculations also showed that caution should be exercised when using linear isotherms in the colloid sorption calculations.


   APPENDIX
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 APPENDIX
REFERENCES
 
C, Concentration of dissolved pyrene (mg L–1)

S, Concentration of sorbed pyrene (mg kg–1)

Scoll, Pyrene sorbed to colloids (mg pyrene kg–1 colloids)

Ssoil, Pyrene sorbed to soil (mg pyrene kg–1 soil)

SDOM, Pyrene sorbed to dissolved organic matter (mg pyrene kg–1 OC)

Kf, Freundlich coefficient (index a denotes adsorption and d desorption) (L kg–1)

Kf,oc, Freundlich coefficient of organic carbon normalized isotherm (subscript "a" denotes adsorption and "d" desorption) (L kg–1)

N, Freundlich exponent (index a denotes adsorption and d desorption)

Kd,C, Distribution coefficient at dissolved concentration C (L kg–1)

Koc,C, Organic C normalized distribution coefficient at dissolved concentration C (L kg–1)

Kow, Octanol–water partitioning coefficient

foc, Mass fraction of organic C of the colloids (kg OC kg–1 colloids)

{omega}, Nonsingularity index

coll, Mobile colloids (kg colloids kg–1 soil)

DOC, Dissolved organic matter (kg OC kg–1 soil)

X, Total pyrene concentration of soil (mg pyrene kg–1 soil)

{rho}bulk, Bulk density (kg m–3)

{theta}, Soil water content (m3 m–3)

Kf,coll, Freundlich coefficient (mobile colloids) (L kg–1)

Kf,soil, Freundlich coefficient (stationary soil) (L kg–1)

KDOC, Distribution coefficient (DOC) (L kg–1)

Na,coll, Freundlich exponent (mobile colloids)

Na,soil, Freundlich exponent (stationary soil)


   ACKNOWLEDGMENTS
 
This work was financed by the Danish FREJA program (Female Researchers in Joint Action) of the Danish Research Council.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 APPENDIX
 REFERENCES
 




This article has been cited by other articles:


Home page
 Vadose Zone JHome page
T. G. Poulsen, P. Moldrup, L. W. de Jonge, and T. Komatsu
Colloid and Bromide Transport in Undisturbed Soil Columns: Application of Two-Region Model
Vadose Zone J., May 26, 2006; 5(2): 649 - 656.
[Abstract] [Full Text] [PDF]

Home page
 Vadose Zone JHome page
L. W. de Jonge, C. Kjaergaard, and P. Moldrup
Colloids and Colloid-Facilitated Transport of Contaminants in Soils: An Introduction
Vadose Zone J., May 1, 2004; 3(2): 321 - 325.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow