JEQ Journal of Natural Resources and Life Sciences Education
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 Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (28)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Clausen, L.
Right arrow Articles by Fabricius, I.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Clausen, L.
Right arrow Articles by Fabricius, I.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Clausen, L.
Right arrow Articles by Fabricius, I.
Related Collections
Right arrow Agricultural Pesticides
Right arrow Water Pollution
Right arrow Ground Water Quality
Journal of Environmental Quality 30:858-869 (2001)
© 2001 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America

TECHNICAL REPORT
Ground Water Quality

Atrazine, Isoproturon, Mecoprop, 2,4-D, and Bentazone Adsorption onto Iron Oxides

Liselotte Clausen and Ida Fabricius

Department of Geology and Geotechnical Engineering, Technical University of Denmark, Building 204, DK-2800 Lyngby, Denmark

Corresponding author (igglc{at}pop.dtu.dk)

Received for publication January 7, 2000.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 ENVIRONMENTAL IMPLICATIONS
 CONCLUSIONS
 REFERENCES
 
Iron oxides are important components influencing the adsorption of various inorganic and organic compounds in soils and sediments. In this study the adsorption on iron oxides of nonionic and ionic pesticides was determined as a function of solution pH, ionic strength, and pesticide concentration. The investigated iron oxides included two-line ferrihydrite, goethite, and lepidocrocite. Selected pesticides comprised atrazine (6-chloro-N2-ethyl-N4-isopropyl-1,3,5-triazine-2,4-diamine), isoproturon [3-(4-isopropylphenyl)-1,1-dimethylurea)], mecoprop [(RS)-2-(4-chloro-2-methylphenoxy)propionic acid], 2,4-D (2,4-dichlorophenoxyacetic acid), and bentazone [3-isopropyl-1H-2,1,3-benzothiadiazin-4-(3H)-one 2,2-dioxide]. The adsorption of the nonionic pesticides (atrazine and isoproturon) was insignificant, whereas the adsorption of the acidic pesticides (mecoprop, 2,4-D, and bentazone) was significant on all investigated iron oxides. The adsorption capacity increased with decreasing pH, with maximum adsorption reached close to the pKa values. The addition of CaCl2 in concentrations from 0.0025 to 0.01 M caused the adsorption capacity to diminish. The adsorption of bentazone was significantly lower than the adsorption of mecoprop and 2,4-D, illustrating the importance of a carboxyl group in the pesticide structure. The adsorption capacity on the iron oxides increased in the order: lepidocrocite < goethite < two-line ferrihydrite. The maximum adsorption capacities of mecoprop and 2,4-D on goethite were found to be equivalent to the site density of singly coordinated hydroxyl groups on the faces of the dominant {110} form, suggesting that singly coordinated hydroxyl groups are responsible for adsorption. Differences in adsorption capacities between iron oxides can be explained by differences in the surface site density of singly coordinated hydroxyl groups. The maximum measured adsorption capacity of mecoprop on two-line ferrihydrite was equivalent to 0.2 mol/mol Fe.

Abbreviations: HPLC, high performance liquid chromatography • PZC, point of zero charge • SEM, scanning electron micrograph • TEM, transmission electron micrograph • TOC, total organic carbon


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 ENVIRONMENTAL IMPLICATIONS
 CONCLUSIONS
 REFERENCES
 
ADSORPTION from aqueous solution to solid surfaces is one of the key processes determining the concentration and rate of transport of pesticides in aquifers. The adsorption properties of soils and sediments are influenced strongly by constituents that have high specific surface area and highly reactive surfaces (Bailey and White, 1970). In sediments containing a significant amount of organic matter the sorption of pesticides is therefore as a rule controlled by the organic carbon content, due to the porous nature and large surface of the humic substances, where a variety of functional groups are present (Bailey and White, 1970; Stevenson, 1976; Chiou et al., 1979). However, in aquifer sediments, where the organic carbon content is low, the association of pesticides with mineral surfaces may become significant (Stevenson, 1976; Brownawell et al., 1990; Schwarzenbach et al., 1993; Celis et al., 1996, 1999). Of the various inorganic aquifer components, clay minerals and oxides have the greatest potential for adsorption of pesticides due to the large surface area and the functional groups of these constituents (Bailey and White, 1970). Many studies have been done on the adsorption of pesticides to clays (Frissel and Bolt, 1962; Aly and Faust, 1964; Terce and Calvet, 1978; Borggaard and Streibig, 1988; Laird et al., 1992; Worrall et al., 1996; Sannino et al., 1997; Moreau-Kervévan and Mouvet, 1998; Celis et al., 1999; Clausen et al., 2001), whereas less information is available on the adsorption of pesticides to iron oxides. The objective of this study is therefore to investigate the relative importance of sorption of pesticides to common iron oxides in aquifer sediments. In order to identify the influence of a mineral phase, the sorbent and solution conditions have to be well defined, and we have therefore used well-characterized, synthetic iron oxides, free of organic carbon.

The pesticides selected for the work are atrazine, isoproturon, mecoprop, 2,4-D, and bentazone. The selected pesticides have been or are being used as herbicides in Europe in amounts over 500 Mg/yr (Fielding et al., 1991), and they have all been detected in surface waters and ground waters in Denmark.

Earlier investigations on the adsorption of atrazine to ferrihydrite (Celis et al., 1997; Moreau-Kervévan and Mouvet, 1998) and goethite (Borggaard and Streibig, 1988) have shown an insignificant adsorption of atrazine to these iron oxides. Investigations of the ionic pesticide 2,4-D have shown a significant adsorption to goethite (Whatson et al., 1973; Kavanagh et al., 1977; Sticher and Agustoni Phan, 1977), ferrihydrite (Sticher and Agustoni Phan, 1977; Celis et al., 1999), lepidocrocite (Sticher and Agustoni Phan, 1977; Madrid and Diaz-Barrientos, 1991), and hematite (Sticher and Agustoni Phan, 1977). Celis et al. (1999) found a maximum adsorption capacity on ferrihydrite equivalent to one 2,4-D species sorbed per 0.75 nm2, and Watson et al. (1973) obtained a similar result for the adsorption on goethite, apparently indicating a similar adsorption capacity of different iron oxides. By contrast, Sticher and Agustoni Phan (1977) found a significantly higher adsorption capacity per square meter of ferrihydrite than of goethite. Watson et al. (1973) demonstrated that the adsorption of 2,4-D on goethite is strongly dependent on pH and ionic strength and hypothesized that the adsorption of 2,4-D occurs due to favorable Coulombic interaction between the sorbing anion and the positively charged surface of the iron oxides (Watson et al., 1973; Kavanagh et al., 1977; Celis et al., 1999).

Most earlier studies involve only one pesticide and one iron oxide, with the work done by Sticher and Agustoni Phan (1977) as an exception: they investigated four different iron oxides. In this study we have chosen to work with nonionic pesticides as well as ionic pesticides in order to investigate whether the adsorption on iron oxides is limited to ionic species. We have chosen three different iron oxides as sorbents to investigate whether the iron oxides have a similar adsorption capacity when the results are normalized to specific surface area, or whether iron oxides have different adsorption capacities according to differences in chemical and morphological properties. Goethite was chosen because it is the most common iron oxide in sediments (Cornell and Schwertmann, 1996). Lepidocrocite was also chosen. It occurs in anaerobic, noncalcareous sediments of temperate regions and is less widespread than goethite (Cornell and Schwertmann, 1996). Lepidocrocite is, however, far from rare. In some Danish aquifer sediments it is the second most important iron oxide as coatings on quartz, feldspars, amphiboles, and pyroxenes (Postma and Brockenhuus-Schack, 1987). To obtain a variation in crystallinity and thereby surface area of the investigated iron oxides, two-line ferrihydrite was also included as sorbent. Ferrihydrite is not often detected in sediments, but the abundance of ferrrihydrite in sediments may well be underestimated because of detection problems (Jambor and Dutrizac, 1998). Because of its high reactivity and large specific surface area, ferrihydrite is believed to be one of the most important adsorbents in near-surface and ground water systems (Davis and Kent, 1990).

The specific goals of the study were to (i) quantify the contributions to adsorption from three selected iron oxides, (ii) determine the effect of pH, (iii) determine the effect of CaCl2 as a background electrolyte (Ca2+ and Cl- are common ions in aquifers), (iv) determine the maximum adsorption capacity on the different iron oxides, and (v) compare adsorption characteristics of the selected pesticides.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 ENVIRONMENTAL IMPLICATIONS
 CONCLUSIONS
 REFERENCES
 
Iron Oxides: Preparation and Characteristics
The two-line ferrihydrite and goethite were synthesized according to the procedure of Schwertmann and Cornell (1991). The two-line ferrihydrite was prepared by neutralizing a 0.2 M solution of Fe(NO3)3·9H2O with 1 M KOH. The product was washed by successive centrifugations and decantations with Millipore (Molsheim, France) water. Dialysis was not used, because preliminary experiments had shown that dialysis tubes could contaminate the sample with organic carbon. The final suspension of the two-line ferrihydrite contained less than 0.9 mM NO-3. The goethite was synthesized by controlled-air oxidation of 45 mM FeSO4·7H2O in a 0.1 M bicarbonate buffered solution. The suspension was washed and centrifuged to less than 0.12 mM SO2-4. Lepidocrocite was synthesized by a modified procedure based on Schwertmann and Cornell (1991). A 0.2 M FeSO4·7H2O, 0.4 M NaCl solution was oxidized by controlled dispension of atmospheric air. 1 M NaOH neutralized protons produced during oxidation–hydrolysis of Fe2+. The suspension was washed and centrifuged to less than 0.11 mM SO2-4 and less than 0.22 mM Cl-. For characterization purposes, parts of the iron oxide suspensions were freeze-dried for X-ray etc., but the bulk of the iron oxide suspensions were stored at 10°C and subsequently used for adsorption experiments as aqueous suspensions. The original suspensions contained 17.4 g/L ferrihydrite, 23.5 g/L goethite, and 13.8 g/L lepidocrocite, respectively. These suspensions were diluted to stock solutions containing 0.13 g/L ferrihydrite, 2.48 g/L goethite, and 1.17 g/L lepidocrocite, respectively.

The specific surface areas of the iron oxides were determined by multipoint N2–BET analysis using a Gemini III 2375 surface area analyzer (Micromeritics Instruments Corp., Norcross, GA). Prior to the measurements the samples were outgassed on a FlowPrep 060 degasser (Micromeritics Instruments Corp.). In order to minimize phase changes the samples were outgassed at room temperature for 19 h because these conditions had provided consistent results (Clausen and Fabricius, 2000). The organic carbon content in the iron oxide suspensions was measured on a DOC analyzer (Dohrmann [Cincinnati, OH] DC-180) (Table 1).


View this table:
[in this window]
[in a new window]
  		Table 1. Iron oxide characteristics.

 
The iron oxides were further examined by scanning electron micrograph (SEM) (Philips [New York, NY] XL20, platinum-coated grid, 30 kV beam), X-ray diffraction (CoK{alpha}, Siemens [New York, NY] D5000), IR spectroscopy (PerkinElmer [Norwalk, CT] 2000, KBr pressed pellet technique) and Mössbauer spectroscopy (conventional constant acceleration spectrometer; samples measured at temperatures between 20 and 298 K using an APD closed-cycle He cryostat.). Measurements by X-ray diffraction, Mössbauer spectroscopy, and BET were performed on the freshly precipitated oxides and again at the end of the experimental period (approximately 1 yr) on the aged material after storage of the suspensions at 10°C.

Two-Line Ferrihydrite
The surface area of the two-line ferrihydrite determined by the BET method (Table 1) is in the same range as found by other workers with this method (Dzombak and Morel, 1990). The specific surface was determined without heating the ferrihydrite to minimize phase changes during the measurements (Clausen and Fabricius, 2000). From the BET measurements an equivalent mean spherical diameter of 6.6 nm of the two-line ferrihydrite crystals can be calculated, which is in the same range as found by other workers by transmission electron microscopy (TEM) (Saraswat et al., 1980; Cornell and Schwertmann, 1996; Weidler and Stanjek, 1998). The SEM image of the two-line ferrihydrite shows 35- to 50-nm-sized ill-defined shapes (Fig. 1A), so what we see on the SEM image is therefore probably rounded aggregates instead of single crystals. Because the calculated equivalent mean spherical diameter is in agreement with the crystal size often found by TEM, we believe that the BET value gives a reliable measure of the surface area of two-line ferrihydrite.



View larger version (89K):
[in this window]
[in a new window]
  		Fig. 1. Scanning electron micrograph (SEM) images of iron oxides. (A) Two-line ferrihydrite, (B) goethite, and (C) lepidocrocite. The scale bar in the SEM images is 200 nm (Clausen and Fabricius, 2000).

 
From X-ray diffraction (CoK{alpha}) (Fig. 2), IR spectroscopy, and Mössbauer spectroscopy no crystalline impurities were indicated in the freshly precipitated two-line ferrihydrite. However, ferrihydrite is unstable and will transform to hematite and goethite on aging (Schwertmann and Cornell, 1991; Jambor and Dutrizac, 1998), and after storage of the suspension for 1 yr, X-ray diffraction on the aged material showed a significant amount of goethite (Fig. 2). From Mössbauer spectroscopy this impurity was quantified to be between 5 and 15%. However, no detectable change in surface area was measured on the aged ferrihydrite.



View larger version (29K):
[in this window]
[in a new window]
  		Fig. 2. X-ray (CoK{alpha}) diffractograms of synthesized iron oxides. Fresh denotes that the measurements were performed on freshly precipitated material, and aged denotes that the measurements were performed on the same material after storage in suspension for more than 1 yr. G denotes goethite, L denotes lepidocrocite.

 
Goethite
The synthesized goethite has a low crystallinity, with a surface area of 60 m2/g, and low crystalline goethite prepared from a Fe(II) system is often similar to goethites from various natural environments (Schwertmann and Cornell, 1991). The SEM image of the goethite sample shows small aggregates of up to 0.3-µm-long acicular crystals (Fig. 1B). X-ray diffraction indicates that the goethite sample contains some lepidocrocite (Fig. 2). From Mössbauer spectroscopy this impurity was quantified to less than 5%. X-ray diffraction and BET measurements confirmed that no transformation of the goethite sample occurred during the time span of the adsorption experiments.

Lepidocrocite
The synthesized lepidocrocite has a specific surface area of 117 m2/g (Table 1), within the range (15 to 260 m2/g) reported by Cornell and Schwertmann (1996). The crystals are lath-like platelets (Fig. 1C). Their two major dimensions appear to be near 0.30 and 0.03 µm, respectively. X-ray diffraction indicates that the lepidocrocite sample contains goethite (Fig. 2). Mössbauer spectroscopy revealed that the lepidocrocite contains between 20 and 25% goethite. However, no goethite was observed in the SEM image, so the goethite may be overgrown by lepidocrocite. Lepidocrocite can transform to goethite on aging, but the transformation is extremely slow (Schwertmann and Cornell, 1991). X-ray diffraction and BET measurements indicated no changes in the lepidocrocite sample during the time span of adsorption experiments.

Sorbates and Solutions
The structural formulae and the relevant properties of the five selected herbicides are shown in Fig. 3 and Table 2.



View larger version (12K):
[in this window]
[in a new window]
  		Fig. 3. Structural formulae of herbicides used in sorption studies.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Properties of herbicides used in sorption studies. Data from Shiu et al. (1990), Tomlin (1994), and Hornsby et al. (1996).

 
The experiments were carried out with unlabeled pesticides and 14C-labeled pesticides. The unlabeled pesticides were used as supplied (Riedel-de Haën, Seelze, Germany) with the following purities expressed in mass percent: atrazine, 99.4%; isoproturon, 99%; mecoprop, 99%; 2,4-D, 99%; and bentazone, 99.9%. The specific activity and purity of the 14C-labeled pesticides are: atrazine, 695 GBq/mol, >95%; isoproturon, 914 GBq/mol, >97%; mecoprop, 1828 GBq/mol, >98.4%; 2,4-D, 673 GBq/mol, >98%; and bentazone, 514 GBq/mol, >98%. In bentazone, the 14C label is located in the carbonyl group and in the other selected pesticides the 14C label is located in the ring structure. The labeled 14C pesticides were purchased from Sigma Chemical Company (St. Louis, MO; atrazine and 2,4-D), Amersham International (Buckinghamshire, UK; isoproturon and mecoprop), and International Isotope (Munich, Germany; bentazone).

Stock solutions of pesticides were prepared in methanol and kept refrigerated. The methanol was removed under N2 flow before dilution in Millipore water, resulting in a methanol concentration of less than 0.0003 m3/m3. The procedure resulted in no detectable loss of pesticides. In experiments with controlled ionic strength, the pesticides were diluted in a CaCl2 solution. All solutions were sterile-filtered (0.2 µm) before use.

Methods for Adsorption Experiments
Adsorption studies were performed using a batch equilibrium procedure. Sticher and Agustoni Phan (1977) reported that 90% of adsorption occurs within 4 h, but that the total adsorption process may take several days. Hence, all suspensions were allowed to equilibrate for 96 h. The experiments were done at 10 ± 1°C in the dark. Glass tubes (10 mL) with Teflon caps were used. Preliminary tests showed no detectable adsorption on those materials. All glassware used in experiments was baked at 550°C to remove organic contamination. We added 4 mL of the iron oxide stock suspension to the vials, giving the following amounts of solid: ferrihydrite, 0.52 mg; goethite, 9.9 mg; and lepidocrocite 4.7 mg. Then, 1 mL of pesticide solution was added, and each test tube placed in a vertical rotator. After equilibration for 96 h, the supernatant was separated from the suspension using a 0.2-µm PTFE filter (Advantec [Pleasanton, CA]/MFS 13 HP). Preliminary experiments demonstrated no detectable retention in this filter type. We mixed 1 mL of the separated supernatant with 10 mL OptiPhase HiSafe 3 (Wallac, Turku, Finland) scintillation cocktail and the amount of radioactivity was determined by counting for 20 min in a 1414 WinSpectral liquid scintillation counter (Wallac). The specific radioactivities in the experiments were 16.7 Bq/mL, while the background counting was 0.7 Bq/mL. Reference samples were prepared without iron oxides but otherwise handled identically. Both reference samples and test samples were prepared in triplicates. All radioactivity measurements were corrected for background counting, and the concentration of pesticide in the supernatant solution was then calculated from the difference between the radioactivity measurements in the supernatant solution and the radioactivity measurements in the reference solution.

The pH was controlled by adding HCl or NaOH to the stock iron oxide suspensions. Experiments with a controlled pH were carried out in the pH range from 3.5 to 8. This corresponds to an ionic strength in the range 1 x 10-8 to 3 x 10-4 M. No background electrolyte was added in order to avoid competition for the surface sites. Atomic absorption spectroscopy showed no detectable dissolution of the iron oxides. In experiments with a controlled ionic strength the stock iron oxide suspensions were prepared in a CaCl2 solution. The iron oxide suspensions were equilibrated with CaCl2 for at least 24 h before the pesticide solution was added. The CaCl2 concentration varied between 0.0025 and 0.01 M, which gives an ionic strength in the range 7.5 x 10-3 to 3.0 x 10-2 M. In experiments with 0.01 M CaCl2, three times more solid was used, due to the low sorption in these experiments. Experiments with a controlled pH and CaCl2 concentration were carried out with a single initial pesticide concentration of 0.25 mg/L (1.04–1.21 µmol/L). The isotherm experiments were carried out with 10 concentrations ranging from 0.05 to 150 mg/L. High performance liquid chromatography (HPLC) was used to check that no degradation occurred during the time the adsorption experiments were running. The following HPLC conditions were used: 5-µm Hypersil (Cheshire, UK) ODS column of 250 mm length and 2.0 mm i.d.; C18 column packing; 0.3 mL/min flow rate; 10:90 (v/v) acetonitrile to 0.001 M ammonium acetate eluent system; 200 nm UV detection.

Data Analysis
The maximum adsorption capacity on the iron oxides was calculated from the adsorption isotherm using a Langmuir equation:

where Cs is the equilibrium concentration of sorbate associated with the sorbent (mol/m2), Ce is the equilibrium concentration of sorbate in solution (mol/L), Cs,max is the maximum concentration of sorbate associated with the sorbent (mol/m2), and KL is the Langmuir constant (L/mol). The Langmuir parameters were calculated by nonlinear least squares (NLLS) analysis as described by Kinniburgh (1986) using TableCurve software (Jandel Scientific, 1989).

From the isotherm data a linear adsorption coefficient was calculated in the solution equilibrium concentration range where the following criteria were fulfilled: (i) the residuals are normally distributed about zero, (ii) the residuals are less than two times the standard deviation, (iii) the error on the residuals are less than 6%, and (iv) the correlation coefficient is larger than or equal to 0.99. The linear adsorption coefficient was calculated by the least-squares technique using log-transformed equilibrium concentrations. The intercept was assumed to be 0 in the linear system and thereby we enforced a slope of 1 in the log system. By log-transforming the data before least square analysis, we assumed a constant relative error on the isotherm data (Kinniburgh, 1986), which is an appropriate assumption according to our measurements expressed in percent.

Data from experiments with a single initial pesticide concentration were obtained in the linear equilibrium concentration range and therefore analyzed assuming a linear isotherm:

where Kd is the adsorption distribution coefficient. To compare the adsorption capacity on the different iron oxides, the Kd values were normalized to the solid surface area. Accordingly, Kd in the present study is given in liters per square meter. The adsorption was measured in percentage from the liquid scintillation counting, and the Kd values (L/m2) were calculated from:

where A is the adsorption in percent, V0 is the initial volume of aqueous phase in contact with the mineral (L), and Smineral is the surface area of the mineral phase (m2). The standard deviations of Kd were calculated from the formula of propagation error.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
ENVIRONMENTAL IMPLICATIONS
 CONCLUSIONS
 REFERENCES
 
Screening Experiments
Preliminary experiments with a single initial pesticide concentration of 0.25 mg/L (1.04–1.21 µmol/L) and pH adjusted to 4.6 demonstrate that the adsorption of the nonionic pesticides (isoproturon and atrazine) is insignificant on all three studied iron oxides. Atrazine is here classified as a nonionic pesticide, because the compound, due to a high pKb value (Table 2), will behave as a nonionic pesticide except at extremely acid conditions. The lack of adsorption of atrazine on iron oxides is in agreement with results from earlier studies (Borggaard and Streibig, 1988; Celis et al., 1997; Moreau-Kervévan and Mouvet, 1998). Whereas no adsorption was measured for the nonionic pesticides, a significant adsorption of the acidic pesticides (mecoprop, 2,4-D, and bentazone) was detected on all three studied iron oxides. For these pesticides, adsorption experiments with variable pesticide concentration were therefore carried out at near the same pH as in the preliminary experiments.

Langmuir Isotherms
Isotherm experiments with mecoprop, 2,4-D, and bentazone were performed with an initial pesticide concentration range from 0.05 to 150 mg/L (0.2–700 µmol/L). The results were analyzed according to the Langmuir equation (Fig. 4, Table 3). The pH of the adsorption isotherm experiments was not specifically controlled with an electrolyte, and this caused a decrease in pH with increasing initial concentration of the acidic pesticides during the isotherm experiments. As demonstrated later, the effect of the pH variations on the calculated adsorption parameters (Tables 3 and 4) is small.



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 4. Langmuir adsorption isotherms shown on a linear scale for the sorption of 2,4-D, mecoprop, and bentazone on iron oxides. The corresponding Langmuir parameters are listed in Table 3. Error bars indicate standard deviation (may be covered by the symbols).

 

View this table:
[in this window]
[in a new window]
  		Table 3. Regression estimates of Langmuir isotherm parameters.

 

View this table:
[in this window]
[in a new window]
  		Table 4. Linear sorption distribution coefficients (Kd values) calculated from isotherm experiments.

 
The adsorption data for all three acidic pesticides on two-line ferrihydrite and goethite fit the Langmuir equation with a high correlation coefficient (Fig. 4, Table 3). On lepidocrocite a correlation to the Langmuir equation was only possible for bentazone. The reason for the lack of correlation to the Langmuir equation for the isotherm data for mecoprop and 2,4-D on lepidocrocite is probably that the investigated equilibrium concentration range was too limited.

Previous studies have shown that the adsorption isotherm for 2,4-D on lepidocrocite (Madrid and Diaz-Barrientos, 1991) and on ferrihydrite (Celis et al., 1999) has a shape corresponding to the S-class of Giles et al. (1960). However, Watson et al. (1973) obtained adsorption isotherms for 2,4-D on goethite of the conventional Langmuir type. In our experiments, we also find the S-shape in the isotherms on all three iron oxides below an equilibrium concentration of approximately 100 µmol/L. However, the S-shape behavior of the lower part of the isotherms is masked in Fig. 4 by the larger range of equilibrium concentrations. Kavanagh et al. (1977) suggested the existence of Van der Waals interactions between the adsorbed ions themselves, and as noted by Madrid and Diaz-Barrientos (1991), this may explain the observed S-shaped isotherms.

Linear Distribution Coefficients
From the isotherm experiments (Fig. 4) linear adsorption coefficients were calculated in the linear part of the solution equilibrium concentration (Table 4). In order to accommodate all data in the linear interval, the results are shown on a log–log scale in Fig. 5, where the lines from the linear regression (with a slope equal to 1) also are shown. The linear adsorption range varies significantly between the different adsorption isotherms (Table 4). However, all isotherms are linear when the solution equilibrium concentration is below 2 µmol/L, and well-defined Kd values can be obtained. The screening data all fall within the linear interval and the Kd values determined from the screening experiments with two-line ferrihydrite and goethite are in good agreement with the Kd values calculated from the isotherm data (Fig. 5), indicating a good reproducibility of the experimental results. The Kd values determined from the screening experiments with lepidocrocite are lower than the Kd values determine in the linear equilibrium concentration range (Fig. 5), due to the difference in pH between the screening experiments and the isotherm experiments (see below).



View larger version (25K):
[in this window]
[in a new window]
  		Fig. 5. Linear adsorption isotherms shown on a log scale for the sorption of 2,4-D, mecoprop, and bentazone on iron oxides, and adsorption distribution coefficients (Kd values) obtained from screening experiments with an initial pesticide concentration of 250 µg/L (1.04–1.21 µmol/L). Error bars indicate standard deviation (may be covered by the symbols).

 
Effect of pH
Experiments with variable pH were carried out with a single initial pesticide concentration of 0.25 mg/L (1.04–1.21 µmol/L), resulting in Cs values within the linear range determined in the isotherm experiments. Earlier studies have demonstrated that the adsorption mechanism is the same at different pH levels (Watson et al., 1973; Madrid and Diaz-Barrientos, 1991), and we expect therefore approximately linear isotherms at each pH at the given initial pesticide concentration. The Kd values calculated at variable pH (Fig. 6) demonstrate that the adsorption of mecoprop, 2,4-D, and bentazone on iron oxides exhibit behavior typical for anionic sorbates on positively charged oxide minerals, giving a strong adsorption at low pH and a decreasing adsorption as pH increases (Watson et al., 1973; Davis and Kent, 1990; Dzombak and Morel, 1990; Cornell and Schwertmann, 1996). This behavior is a consequence of electrostatic interactions between the oxide surface and the anionic part of the pesticide molecule. At pH values above the point of zero charge (PZC) of the minerals, the surfaces have a net negative charge and adsorption of anions is restricted due to electrostatic repulsion, whereas at pH values lower than PZC, adsorption of anions is promoted due to electrostatic attraction to the positively charged surface. The adsorption of 2,4-D and bentazone is insignificant at the PZC of the iron oxides, whereas a significant adsorption of mecoprop was measured at the PZC on all iron oxides (Table 1, Fig. 6). The pH range where acidic pesticides adsorb is therefore larger for mecoprop than for 2,4-D and bentazone. At pH values above 4.7, the largest adsorption was measured for mecoprop on all three iron oxides. Sorption of 2,4-D is insignificant above pH 7, but the pH curve shows a steeply increasing adsorption for pH values below 6. On two-line ferrihydrite, the adsorption of 2,4-D reaches the same level as the adsorption of mecoprop, and on goethite the adsorption of 2,4-D even exceeds the adsorption of mecoprop at pH 4.7. The adsorption of bentazone is significantly lower than the adsorption of the other acidic pesticides (Fig. 6), and at pH values above 6 the adsorption is negligible.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 6. Adsorption distribution coefficients (Kd values) obtained at an initial pesticide concentration of 250 µg/L (1.04–1.21 µmol/L) for 2,4-D, mecoprop, and bentazone on iron oxides as a function of pH in solution. Error bars indicate standard deviation (may be covered by the symbols).

 
Several investigations of the adsorption of organic acids on oxide minerals have shown that the adsorption behavior is influenced by the acidity of the acids, because the maximum extent of surface binding occurs close to the pKa value of the organic acid (Whatson et al., 1973; Sticher and Agustoni Phan, 1977; Kummert and Stumm, 1980; Evanko and Dzombak, 1998). This is a result of two effects: (i) at pH values close to the PZC, the adsorption is limited by the number of positive sites on the surface; and (ii) at pH values close to the pKa value of the acids, the adsorption is limited by the relative concentration of the anionic form of the ionic pesticides. The maximum adsorption occurs then where the product of the number of active surface sites and the concentration of anionic species is highest (Cornell and Schwertmann, 1996). In our experiments, the adsorption of mecoprop and bentazone reached a plateau close to the pKa value of these compounds. The maximum adsorption of 2,4-D was not reached, probably because the pKa for 2,4-D is lower than the investigated pH range (Fig. 6, Table 2). Because the isotherm experiments for mecoprop and bentazone were performed close to the pKa values of these compounds, the observed decrease in pH during the isotherm experiments will only have a limited effect on the calculated Langmuir parameters (Table 3) and the calculated linear adsorption coefficients (Table 4). In the isotherm experiments for 2,4-D, the equilibrium concentration range where a linear adsorption coefficient could be defined was so limited that the variation in pH is insignificant. The calculation of the maximum adsorption capacity for 2,4-D (Table 3) is only affected to a limited degree because the calculation is controlled by the last three to four adsorption points in the large concentration range, and within these data points the pH variation is small (<0.4).

Effect of Calcium Chloride as a Background Electrolyte
In the experiments with variable CaCl2 concentrations, atrazine and isoproturon were included as sorbates to determine whether the CaCl2 concentration would promote an adsorption of nonionic pesticides. This was not the case and the adsorption of atrazine and isoproturon remained insignificant in all experiments. The CaCl2 concentration, however, strongly affected the adsorption of the acidic pesticides (Fig. 7). The linear adsorption coefficients decrease with increasing CaCl2 concentration. This effect was also found in experiments done by Watson et al. (1973). The relationship between the decrease in adsorption of the acidic pesticides and the increasing CaCl2 concentration found in this study, however, is not linear. A large decrease in adsorption was observed between the experiments without CaCl2 and the experiments with 0.0025 M CaCl2. After this decrease the adsorption of 2,4-D and bentazone was nearly constant in the examined CaCl2 concentration range. The adsorption of mecoprop was constant in the CaCl2 concentration range between 0.0025 and 0.0078 M, but at 0.01 M CaCl2 the adsorption was significantly lower on all three iron oxides.



View larger version (21K):
[in this window]
[in a new window]
  		Fig. 7. Adsorption distribution coefficients (Kd values) obtained at an initial pesticide concentration of 250 µg/L (1.04–1.21 µmol/L) for 2,4-D, mecoprop, and bentazone on iron oxides as a function of the CaCl2 concentration. The solution conditions are: temperature = 10°C and pH 4.6 ± 0.1. Error bars indicate standard deviation (may be covered by the symbols).

 
At first sight the decreasing adsorption with increasing CaCl2 concentration may seem contrary to the expected trend because an increase in ionic strength causes an increasing positive charge on the surface (Dzombak and Morell, 1990). However, there are several contrary effects. First, with regard to increasing chloride, the anionic pesticide ratio leads to more efficient competition for surface positive sites, because chloride is known to adsorb on iron oxides (Cornell and Schwertmann, 1996). Furthermore, Ali and Dzombak (1996) have demonstrated that Ca2+ is capable of forming complexes with simple organic acids, so adding CaCl2 to the experiments leads to a possible complexation between the anionic pesticides and Ca2+, which results in nonsorbing Ca2+–pesticide solution complexes.

Adsorption Capacity and Site Density
The adsorption distribution coefficients (Table 4, Fig. 5–7) and the maximum adsorption capacities (Cs,max in Table 3, Fig. 4) illustrate that the three different iron oxides have significantly different adsorption capacities. For all three acidic pesticides the adsorption increases in the order: lepidocrocite < goethite < two-line ferrihydrite. This sequence was also found in a study on the adsorption of 2,4-D to iron oxides done by Sticher and Agustoni Phan (1977). These authors explain the difference in adsorption capacity on goethite and lepidocrocite by a change in surface area of lepidocrocite during the adsorption experiments, because they measured a 25% decrease in surface area after adsorption of 2,4-D. Therefore, we measured the BET surface area of all three iron oxides before and after an adsorption experiment with 2,4-D at pH 4.5, but no detectable change in the surface area due to adsorption of 2,4-D was found. The difference in adsorption capacity of the iron oxides is therefore not an effect of a change in surface area, but probably an effect of different site densities on the iron oxides.

The Langmuir sorption isotherm is derived assuming that only one type of adsorption site is involved (Stumm, 1992). A close fit to a Langmuir isotherm was observed for the adsorption of all three acidic pesticides on two-line ferrihydrite and on goethite, and for bentazone on lepidocrocite. Crystallographic considerations indicate that the surface hydroxyl groups on iron oxides may be coordinated to one, two, or three Fe atoms (Barrón and Torrent, 1996). Earlier studies indicate that the adsorption of ions involves only single coordinated hydroxyl groups, whereas double and triple coordinated groups appear to be comparatively unreactive (Russell et al., 1974; Torrent et al., 1990). In the goethite crystals, it is the faces of the {110} form and to a lesser extent the {021} form that constitute the surface. The site densities of singly coordinated groups on these faces are 3.0 and 8.2 per nm2, respectively (Barrón and Torrent, 1996), corresponding to one singly coordinated hydroxyl group per 0.33 and 0.12 nm2, respectively. The maximum adsorption capacities for mecoprop and 2,4-D on goethite are 5.7 and 5.5 µmol/m2, respectively. This corresponds to an adsorption density of approximately one species per 0.30 nm2 (Table 3), which is close to the area occupied by one singly coordinated hydroxyl group on the faces of the dominant {110} form. These results therefore indicate that it is the singly coordinated hydroxyl groups that are responsible for the adsorption on goethite. Watson et al. (1973) found a maximum adsorption of 2,4-D on goethite in 0.01 M NaCl solution, equivalent to one 2,4-D anion adsorbed per 0.70 nm2, and these authors pictured therefore the adsorption of 2,4-D as bridging across two singly coordinated hydroxyl groups via the –COO- group. However, our results suggest that the reason for the lower adsorption of 2,4-D in the study done by Watson et al. (1973) is the competition with the electrolyte anion for the adsorption sites rather than the formation of bidentate complexes.

On the predominant {010} faces of lepidocrocite there are only double coordinated hydroxyl groups (Cornell and Schwertmann, 1996), and as a consequence the overall site density of singly coordinated hydroxyl groups is less than on goethite. The differences in adsorption capacity between goethite and lepidocrocite can therefore be explained by the difference in site densities of singly coordinated hydroxyl groups. The large impurity (20–25%) of goethite in the lepidocrocite powders could in principle account for the observed adsorption of pesticides on the lepidocrocite. However, we believe that the adsorption does take place on the lepidocrocite for two reasons: (i) no goethite crystals were noted on the SEM image, indicating that the goethite is overgrown by lepidocrocite; and (ii) the adsorption curves for goethite are not similar in shape to those of lepidocrocite (Fig. 4–7).

The small grain size of ferrihydrite and its poor crystallinity have thwarted attempts at a direct determination of its crystal structure. Crystallographic considerations of the densities of groups that outcrop on the surface are therefore difficult. However, the low Fe to OH ratio of ferrihydrite indicates that OH on the surface most probably is singly coordinated. The maximum adsorption capacity for mecoprop on ferrihydrite is 15 µmol/m2, which corresponds to an adsorption density of one species per 0.11 nm2 (Table 3). This site density normalized to the Fe content is equivalent to 0.2 mol/mol Fe. Site densities for ferrihydrite obtained from different adsorption studies with different cations and anions as sorbates have shown that the site density normalized to the Fe content varies between 0.05 and 0.24 mol/mol Fe, with a mean value of 0.17 mol/mol Fe and a standard deviation of 0.05 mol/mol Fe (Dzombak and Morel, 1990). The maximum adsorption capacity of mecoprop on ferrihydrite is therefore in close agreement with results from other adsorption studies.

The maximum adsorption capacity on two-line ferrihydrite seems to be relatively constant for strongly sorbing adsorbates, indicating that two-line ferrihydrite has a rather constant crystal size with a well-defined number of active sites of approximately 9 singly coordinated hydroxyl groups per nm2.

Adsorption Mechanistic Interpretation
In all experiments the adsorption of bentazone was significantly lower than the adsorption of mecoprop and 2,4-D, illustrating that the carboxyl groups in the two latter pesticides promote adsorption. The adsorption of bentazone is negligible at the PZC of the iron oxides, and because the adsorption of bentazone is strongly affected by the CaCl2 concentration, this pesticide probably is weakly attached to the surface through outer-sphere complexation. Mecoprop and 2,4-D have the same functional group in the pesticide structure, and in the experiments without electrolyte at pH 4.6, the adsorption of mecoprop and 2,4-D on ferrihydrite and goethite was similar (Fig. 4–6). However, in the experiments with lepidocrocite and when CaCl2 was added to the experiments, the sorption of mecoprop was significantly larger than the adsorption of 2,4-D (Fig. 7), suggesting that 2,4-D is more weakly adsorbed than mecoprop, probably through outer-sphere complexation. A possible explanation for the larger sorption of mecoprop is the formation of inner-sphere complexes as well as outer-sphere complexes. This would also explain the significant adsorption of mecoprop at PZC, where the adsorption of 2,4-D and bentazone is insignificant. In the experiments with variable CaCl2 concentration (Fig. 7), the first decrease in the adsorption of mecoprop between the experiment without electrolyte and the experiment with 0.0025 M CaCl2 may be explained by a decrease in outer-sphere adsorbed mecoprop, which is strongly affected by the electrolyte concentration. The constant level of adsorbed mecoprop between 0.0025 and 0.0075 M CaCl2 may then be due to inner-sphere complexation. The relative proportion of inner-sphere complexes between mecoprop and the iron oxides is larger on lepidocrocite and ferrihydrite than on goethite. The decrease in the adsorption curves for mecoprop between the experiments with 0.0075 M and 0.01 M CaCl2 (Fig. 7) may tentatively be explained by the formation of nonsorbing Ca–pesticide complexes.


   ENVIRONMENTAL IMPLICATIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 ENVIRONMENTAL IMPLICATIONS
CONCLUSIONS
 REFERENCES
 
In a related study, the adsorption of atrazine and isoproturon was demonstrated to be insignificant on quartz, calcite, and {alpha}-alumina (Clausen et al., 2001), and we found the same for iron oxides. A significant adsorption was, however, found on the clay mineral kaolinite (Clausen et al., 2001). The adsorption of the nonionic pesticides was unaffected by pH and ionic strength. These results imply that the clay content probably controls the sorption of nonionic pesticides in low-total organic carbon (TOC) aquifer sediments. The Kd values for the acidic pesticides mecoprop, 2,4-D, and bentazone on pure iron oxides measured in this study were one to three orders of magnitude larger than Kd values for the same pesticides on quartz, calcite, kaolinite, and {alpha}-alumina measured in Clausen et al. (2001). These results imply that in low-TOC aquifer sediments, quartz, calcite, and clay minerals only play a minor role in the adsorption of acidic pesticides compared with iron oxides. Significant adsorption to iron oxides, however, would be limited to low-pH conditions. To demonstrate the orders of size of Kd values in natural sediments expected from this study, we have made a simple calculation of the retardation factor for mecoprop. We assumed that we have an aquifer sediment, containing 0.5% Fe and composed of quartz, with an insignificant TOC content and a surface area of 1.2 m2/g. Furthermore, we assumed that all the iron is present in goethite, with a surface area of 60 m2/g. From these data we get an accessible surface area for adsorption of 0.48 m2/g, or 40% of the total surface area. By assuming a pH of 4.6 and an ionic strength comparable with 0.01 M CaCl2, we get a Kd value for the aquifer sediment of 0.23 mL/g (by using the determined Kd value for mecoprop on goethite shown in Fig. 7). This is actually very close to Kd values for mecoprop found in Danish aquifers with low pH values (Madsen et al., 2000). The calculated Kd value for a quartz sediment is equivalent to a retardation factor of 2.2, by assuming a bulk density of 1.6 g/cm3, equivalent to a porosity of 0.3.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 ENVIRONMENTAL IMPLICATIONS
 CONCLUSIONS
REFERENCES
 
(i) The investigated acidic pesticides (mecoprop, 2,4-D, and bentazone) adsorb to surfaces of iron oxides from aqueous solutions, whereas the adsorption of the investigated nonionic pesticides (atrazine and isoproturon) is insignificant.

(ii) The adsorption of bentazone is significantly lower than the adsorption of mecoprop and 2,4-D, illustrating the significant role in adsorption of a carboxyl group in the pesticide structure.

(iii) The adsorption of the acidic pesticides is strongly affected by the pH and the CaCl2 concentration in solution. The adsorption is increasing with decreasing pH with maximum adsorption close to the pKa values of the acids. The addition of CaCl2 to the solution causes the adsorption to diminish. This effect is more pronounced for the adsorption of 2,4-D and bentazone than for mecoprop, which leads to the suggestion that mecoprop partly adsorbs by outer-sphere complexes and partly as inner-sphere complexes, whereas 2,4-D and bentazone are only weakly adsorbed through outer-sphere complexes.

(iv) Iron oxides have different adsorption capacities. The adsorption increases in the order: lepidocrocite < goethite < two-line ferrihydrite, probably due to the surface site density of the singly coordinated hydroxyl groups. The maximum adsorption capacities of mecoprop and 2,4-D on goethite were found to be equivalent to the site density of the singly coordinated hydroxyl groups on the faces of the dominant {110} form of goethite. The maximum adsorption capacity of mecoprop on two-line ferrihydrite was found to be equivalent to 0.2 mol/mol Fe, which is in close agreement with results found from other adsorption studies with a variety of different sorbates, indicating that two-line ferrihydrite has a rather constant crystal size with a well-defined number of active sites of approximately 9 singly coordinated hydroxyl groups per nm2.


   ACKNOWLEDGMENTS
 
The present work was supported financially by the Technical University of Denmark (DTU) and by the Ground Water Research Centre. The authors wish to thank Christian Bender Koch (The Royal Veterinary and Agricultural University) for providing the X-ray diffraction, IR-spectroscopy, and Mössbauer-spectroscopy data. The technical assistance from Bente Frydenlund is gratefully acknowledged. Vibeke Knudsen is thanked for editing the figures, and Flemming Rasmussen is thanked for the construction of the vertical rotator. The authors also wish to thank Henrik Spliid (Department of Mathematical Modelling, DTU) for advice concerning the statistic data analysis. Carbon-14 labeled pesticides were kindly provided by GEUS (Geological Survey of Denmark and Greenland). The helpful suggestions and comments from Dieke Postma (DTU), Lene Madsen (GEUS), and Bo Lindhardt (GEUS) are also gratefully acknowledged.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 ENVIRONMENTAL IMPLICATIONS
 CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 Waste Manag ResHome page
L. Thorneby, L. Mathiasson, L. Martensson, and W. Hogland
The performance of a natural treatment system for landfill leachate with special emphasis on the fate of organic pollutants
Waste Management Research, April 1, 2006; 24(2): 183 - 194.
[Abstract] [PDF]

Home page
 Soil Sci.Home page
S. E. Johnson and R. H. Loeppert
Role of Organic Acids in Phosphate Mobilization from Iron Oxide
Soil Sci. Soc. Am. J., January 6, 2006; 70(1): 222 - 234.
[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 Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (28)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Clausen, L.
Right arrow Articles by Fabricius, I.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation