JEQ Grow Your Career With ASA
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 (8)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Kollist-Siigur, K.
Right arrow Articles by Kirso, U.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Kollist-Siigur, K.
Right arrow Articles by Kirso, U.
Agricola
Right arrow Articles by Kollist-Siigur, K.
Right arrow Articles by Kirso, U.
Related Collections
Right arrow Organic Compounds
Right arrow Soil Pollution
Right arrow Water Pollution
Journal of Environmental Quality 30:526-537 (2001)
© 2001 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America

TECHNICAL REPORT
ORGANIC COMPOUNDS IN THE ENVIRONMENT

Sorption of Polycyclic Aromatic Compounds to Humic and Fulvic Acid HPLC Column Materials

Katrin Kollist-Siigura,b, Torben Nielsenb, Christian Grønb, Poul Erik Hansenc, Christian Helwegb, Kristoffer E.N. Jonassenb, Ole Jørgensenb and Uuve Kirsoa

a National Inst. of Chemical Physics and Biophysics, Akadeemia tee 23, EE-12618 Tallinn, Estonia
b PBK 313, Risø National Lab., P.O. Box 49, DK-4000, Roskilde, Denmark
c Dep. of Life Sciences and Chemistry, Roskilde Univ., P.O. Box 260, DK-4000, Roskilde, Denmark

Corresponding author (torben.nielsen{at}risoe.dk)

Received for publication May 7, 1999.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSION
 REFERENCES
 
Two different humic acids (HA) and a fulvic acid (FA) were chemically immobilized to a high performance liquid chromatography (HPLC) silica column material. The immobilization was performed by binding amino groups in HA/FA to the free aldehyde group in glutardialdehyde attached to the silica gel. The HPLC column materials were compared with a blank column material made by applying the same procedure but without immobilizing HA or FA. Also, a column was made by binding carbonyl groups in HA to amino groups attached to the silica gel. The humic substances were selected to secure appropriate variation of their structural features. The retention factors of 45 polycyclic aromatic compounds (PAC) to the four columns were determined by HPLC. The advantage of the technique is a large number of compounds can easily be studied. The binding procedure does not appear to cause a drastic selection between the HA molecules. The k' values obtained for the two Aldrich HA columns agree in general reasonably. The retention or sorption of the compounds increased with the size of the PAC and the number of lipophilic substituents, but decreased when polar substituents were present. The PAC retention was much stronger to the two HA columns than to the FA and blank column, both for hydrophobic polycyclic aromatic hydrocarbons (PAH) and the polar PAC. Other factors impacting the PAC binding may be specific interactions with HA and the ionic strength of the aqueous phase. The technique has been applied to do direct determinations of Koc.

Abbreviations: HA, humic acid • FA, fulvic acid • HPLC, high performance liquid chromatography • PAC, polycyclic aromatic compounds • PAH, polycyclic aromatic hydrocarbons • tr, retention time • t0, dead volume • k' = /t0, capacity coefficient • OC, organic carbon • Koc, organic carbon partition coefficient • Kow, octanol-water partition coefficient • NMR, nuclear magnetic resonance


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSION
 REFERENCES
 
AMONG the organic pollutants in the environment, the polycyclic aromatic compounds (PAC) have caused major concern because many are considered to be carcinogenic (IARC, 1983). Besides, PAC with nitrogen atoms in the structure (N-PAC) have recently been demonstrated to be phytotoxics (Gissel-Nielsen and Nielsen, 1996). A number of different PAC has been identified in the environment. The best known group is the hydrophobic polycyclic aromatic hydrocarbons (PAH), but also N-, S-, O-PAC, a number of different oxygenated derivatives, phenols, quinones, ketones, aldehydes and carboxylic acid derivatives, and nitro and chloro substituted PAC have been found (Blanco et al., 1992; Haglund et al., 1987; Nielsen et al., 1983; Youngblood and Blumer, 1975).

Transport, fate, and bioavailability of organic pollutants in the aquatic and terrestrial environment depend on the partitioning of these compounds between water, dissolved humic matter, and soil or sediment humic matter (Calvet, 1989; Schwarzenbach et al., 1993; Totsche et al., 1997). Binding of nonionic organic compounds to soils and sediments depends on the concentration of organic carbon (OC) in the solid (de Maagd et al., 1994; Means et al., 1980). The distribution coefficient, Koc (mol kg-1 OC/mol L-1 water), is a measure for the binding of the compound. Generally, a simple linear relationship between binding and total OC of the solid is assumed, but recently, a dependency also upon the properties of the solid organic matter has been demonstrated (Chiou et al., 1998; Gauthier et al., 1987). Humic substances constitute a large fraction of the solid organic matter in soils and sediments. The molecular size appears to be an important factor determining the binding of solids to organic matter (Chiou et al., 1986; Engebretson et al., 1996; Engebretson and Wandruszka, 1997). Gauthier et al. (1987) found that Koc for pyrene binding to four different humic acids increased from 0.9 x 105 to 2.0 x 105 L kg-1 as the humic acid aromatic fraction varied from 0.20 to 0.34 of total C. One explanation for the changes in binding with the properties of the solid organic matter is that the binding may be either by adsorption or by partitioning into the surface organic phase of the humic acids (LeBoeuf and Weber, 1997; Murphy et al., 1990). The process appears to shift from partitioning to adsorption with the diagenetic alteration of the materials and thus with decreasing O/C and increasing H/C elemental ratios (Huang and Weber, 1997).

The complexity and inhomogenity of humic substances precludes determination of their exact structures for evaluation of the impact of their structural variability upon the binding capacity for organic compounds. The substances exhibit a great variation in their content of aromatic and aliphatic C, N, and S, and phenolic and carboxylic acid groups (Aiken et al., 1996; Garcia et al., 1994; Gauthier et al., 1987; Malcolm and MacCarthy, 1986; Schulten, 1995; Wershaw, 1986). The inhomogenity of the humic samples is also reflected in their broad molecular weight distributions and also the chemical structures of the different molecular size fractions vary (Tanaka and Senoo, 1995). As an example, the fraction of aromatic C varied from 0.14 to 0.51 and the content of carboxylic C from 0.06 to 0.30 for humic substances isolated from Danish ground water systems (Grøn et al., 1996).

Humic acid chemically bonded to HPLC column material has been used as a new and cost efficient method for the study of the binding of organic compounds to solid organic matter (Nielsen et al., 1997; Szabo and Bulman, 1994). In this investigation the effects of different humic (HA) and fulvic acids (FA) have been studied with this technique. Two different binding procedures were compared for the Aldrich HA. The impact of the humic substance properties on the binding of 45 PAC was determined and compared, as were the ionic strength and cosolvent effects on the binding. The method was tested for its applicability in direct Koc determinations. To span as broad a range of compounds as possible, 45 PAC were evaluated, representing PAH, substituted PAH, N-, O- and S-PAC, neutral as well as charged compounds.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSION
 REFERENCES
 
Chemicals
The test compounds listed in Table 1 along with results discussed later were dissolved in methanol (Lichrosolv 99.8%, Merck, Darmstadt) with a typical concentration of 0.04 g L-1. The preparation of the bromopyrenes is described elsewhere (Hansen and Berg, 1981; Nielsen et al., 1997). N-methylquinolinium iodide was prepared from mixing quinoline and methyl iodide in ethanol. 9-anthracenecarboxamide, -carboxylic acid methyl ester, 9-formyl-, 9-methoxy-, 9-nitro- and 9-cyanoanthracene were provided by Ole Hammerich and Tove Thomsen, University of Copenhagen. All other compounds were obtained from commercial sources.


View this table:
[in this window]
[in a new window]
  		Table 1. PAC log Kow{dagger} and capacity coefficients (k') on the three humic acid coated columns and the blank column. Mobile phase (pH = 8): Methanol (MeOH) + buffered water

 
The humic acid was extracted from ground water in Western Jutland (Fjand) and the fulvic acid from ground water in Northern Jutland (Skagen), Denmark. Ground water was pumped through 0.3-µm filters (200-35-AAH borosilicate glass microfibers with inorganic resin binder from Balston, UK). The samples were stored and transported in glass containers to the laboratory for isolation of the humic substances according to the method applied by Krog and Grøn (1995). In short, humic acids were isolated from the Fjand ground water by precipitation with nitric acid, purification by dialysis, and hydrogen saturation by cation exchange. Fulvic acids were isolated by concentration from the acidified Skagen ground water on XAD-8 column material, further purified and desalted by repeated XAD-8 column treatments and finally hydrogen saturated by cation exchange. The humic substances were lyophilized and stored dry until used.

Glassware was cleaned with detergents, followed by nitric acid rinse, several rinses with low OC laboratory water (MilliQ water purification system from Millipore) and was heated to 450°C for 18 h while flushing with air filtered through activated C. For Teflon tubing, rinses with 0.01 M sodium hydroxide, low OC water, 0.1 M hydrochloric acid, and again low OC water were used.

Preparation of the Chemically Bonded Humic and Fulvic Acid Silica Gel for HPLC Columns [Amino (NH2) Procedure]
The silica gel with chemically bonded humic substances was prepared in a number of stages as described elsewhere (Nielsen et al., 1997; Szabo and Bulman, 1994). The blank column material was the same as the FA and the HA column materials, the only exception being that neither HA nor FA was applied in the synthesis. The method in short is as follows: A suspension of the silica gel (Nucleosil-Si-50-10, surface area [BET] 450000 m2 kg-1, Macherey-Nagel) was refluxed with 3-aminopropyl triethoxysilane (Lancaster, 97%) dissolved in toluene. A suspension of the purified aminopropyl silica gel was stirred at room temperature with a 5% aqueous solution of glutardialdehyde under argon. The activated silica gel was added to a solution of humic or fulvic acid (HA/FA) in water. Finally, the HA/FA silica gel was deactivated with a 0.1 M aqueous solution (pH = 7.5) of 2-aminoethanol and isolated. The gels were packed in methanol at 35 MPa into the HPLC columns. The dimensions for the Aldrich HA, Fjand HA, and the blank column were 12.5 cm by 4.6 mm (i.d.) and for the Skagen FA column 5 cm by 4.0 mm (i.d.).

The immobilization of macromolecules to silica gel or glass beds by means of (3-aminopropyl)triethoxysilane and glutardialdehyde have been applied for several years within enzymology (Andres and Narayanaswamy, 1995; Suleiman et al., 1993). Analogous methods have also been used for the preparation of chiral HPLC column materials (Haginaka et al., 1994). In the first step the silanol groups in the silica gel are alkylated with 3-aminopropyl triethoxysilane:

[1]

In the second step, a 10-fold excess of glutardialdehyde implies that only one of the carbonyl groups is transformed to an imine:

[2]

In the third step humic or fulvic acid is immobilized to the silica gel by imine formation between the aldehyde group and one or more amino groups in the HA/FA molecules:

[3]

Only amino groups are expected to be attached. Residual aldehyde groups are inactivated by imine formation with 2-aminoethanol in the fourth step:

[4]

The latter, N-5-(2-hydroxyethylimino)pentyleno-3-aminopropyl, is the structure of the substituted group in the blank column material. Thus, the blank column contains OC.

Preparation of the Chemically Bonded Aldrich Humic Acid Silica Gel for HPLC Columns [Carbonyl (C=O) Procedure]
This silica gel with chemically bonded humic acid was prepared in a number of stages modified from the way listed above, leaving the stage with glutardialdehyde out and using formaldehyde instead of 2-aminoethanol in the last step.

A suspension of the silica gel (Nucleoprep-Si-300-20, surface area (BET) 100000 m2 kg-1, Macherey-Nagel) was stirred under N2 with 3-aminopropyl triethoxysilane (Merck, p.a.) dissolved in toluene. The product was filtered, washed and dried, and the purified aminopropyl silica gel was added to a solution of humic acid (Aldrich) in water and stirred at room temperature under N2 and finally deactivated with 0.1 M aqueous solution of formaldehyde. The gel was packed in methanol at 35 MPa into a 12.5 cm by 4.6 mm (i.d.) HPLC column.

The first step is the same as that in the amino procedure. In the second step, humic acid is immobilized to the silica gel by imine formation between the amino group and carbonyl groups in the HA molecules:

[5]

The basic residual amino groups are inactivated by imine formation with formaldehyde:

[6]

Determination of Reactive Primary Amino-Groups in Aldrich Humic Acid
A 2.0 mM crotonaldehyde (Fluka, tech.) solution was prepared by dissolving 0.2 mmol (16.5 µL) crotonaldehyde in 1 mL methanol (Lab-Scan, HPLC-grade) and adding purified water (MilliQ) to a total volume of 100 mL. A suspension of 0.5 g Aldrich Humic Acid in 50 mL of the crotonaldehyde solution was stirred under N2 at room temperature for 7.5 h.

Two aliquots of the suspension, each of a few milliliters, were centrifuged for 15 min at 500 x g, and filtrated through a 45-µm PTFE filter before analyses.

The concentration of crotonaldehyde in the solution before and after the reaction was determined by HPLC, using a C-18 HPLC column [25.0 cm by 4.6 mm (i.d.) column packed with Nucleosil 100-5 C18 (Macherey-Nagel) at 35 MPa]. A mixture of equal amounts (volume) of purified water (MilliQ) and methanol (Lab-Scan, HPLC-grade) was used as mobile phase. The crotonaldehyde was measured at 214 nm on a photodiodearray detector.

Mobile Phases
Methanol (Lichrosolv 99.8% from Merck) and low OC laboratory water (MilliQ water purification system from Millipore) were used as mobile phase components. The eluent water (phosphate buffer 0.01 M, pH = 7) to methanol ratio was adjusted depending of the properties of the applied humic and fulvic acids to keep approximately the same retention times of the PAC. Thus, the methanol fraction in the eluent was 0.65 for the Aldrich HA column, 0.75 for the Fjand HA column, and 0.50 for the Skagen FA column and the blank column. pH in the methanol water mixtures was about 8 (Roses et al., 1996).

The fraction of phosphate buffer varied from 0.35 to 0.95 when investigating the influence of increasing water content in the mobile phase to the capacity coefficients of the bicyclic aromatic compounds. The pH in the methanol water mixtures was 7.

When investigating the salt effect on the retention of tricyclic aromatic compounds on the Aldrich HA column, an eluent fraction of 0.01 M Na2B4O7 in water of 0.65 and of methanol of 0.35 was used. The salt (Ca(NO3)2) concentration was changed from 0 to 0.07 M.

HPLC Instrumentation
A low pressure gradient Shimadzu LC-10 HPLC system with photodiodearray detector, thermostatted (30°C) column oven and autoinjector was used. Appropriate mixtures of the compounds were injected and the retention time of each compound was recorded. The injected amount was typical 2 µg of each component. The correct identification of the peaks was controlled by means of the UV spectrum. All measurements were repeated at least two times. The dead time (t0) of the system, used for calculating the capacity coefficient , was determined by injecting and chromatographing water six times. The reproducibility of t0 was 1.3%, and that of the retention time, tr, for the 45 PAC was 0.9 ± 0.6% for the Aldrich HA column as determined by the expression (100 x {sum}(t - tmean)/(n x tmean)). The reproducibilities of t0 and tr were 0.5% and 0.6 ± 0.5% for the Fjand HA column, 0.5% and 0.5 ± 0.3% for the Skagen FA column, and 0.1% and 1.4 ± 1.3% for the blank column.

Dialysis Experiments
The influence of pH on the sorption of quinoline and acridine on dissolved humic acid was also studied with the dialysis method. The method is described in details elsewhere (Nielsen et al., 1997). In short, the equilibrium dialysis experiments to measure the binding of N-PAC to HA were performed by placing 5 mL of a HA solution in dialysis tubing and clamping the ends. The dialysis bag (Spectra Por 6; molecular weight cut off of 1000) was then placed in a 100-mL glass bottle containing a buffered (0.01 M phosphate) solution of the N-PAC. The bottle was shaken in the dark for 24 h at 25°C. Control experiments demonstrated that this was sufficient to achieve a steady state. The described procedure was done for a series of different HA concentrations. In each series two blank experiments without HA were performed to correct for sorption of the N-PAC to the surfaces. The N-PAC concentration in the outer solution was determined by HPLC. The Koc value was derived from the relationship between the ratio of the concentration of the N-PAC in the blank experiment to that in the sorption experiments and the HA total OC concentration at different pHs in the range 2 to 12.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSION
 REFERENCES
 
Immobilization of Humic Substances to Silica Gel HPLC Column Material
Humic substances are a complex mixture of different molecules and hardly two identical humic acid molecules exist in nature. The same is true for fulvic acid. Humus compounds are made up of a skeleton of aromatic rings and aliphatic chains. The skeleton may contain different substituents, e.g., phenolic OH and carboxylic acid. Some groups of the aromatic systems (phenols) are partially oxidized to quinones. The humic substances also contain residues of biomolecules, proteins, and carbohydrates. The amino procedure for preparing HPLC column materials utilizes the content of protein residues. The advantage of the technique is that it does not change the aromatic structures in HA. However, the content of amino groups may not be homogeneous in the humic substance material. Therefore, the binding procedure may modify the humic substance composition. Unless otherwise stated, the Aldrich HA column mentioned here and in the following sections refers to the column material prepared by the amino procedure. The relative loading of the column materials can be assessed from their C content. The total content of C in the blank column material was 125 ± 2 g kg-1, in the Skagen FA column material 137 ± 0.7 g kg-1, in the Aldrich HA column material 132 ± 0.5 g kg-1, and in the Fjand HA column material 146 ± 3 g kg-1. Thus, the loading appears to follow the range: Fjand HA > Skagen FA > Aldrich HA. An approximate value of the loading of the column material with humic substance OC can be achieved by comparing the column OC content with that of the blank column. This implies that the content of the C from the binding material, the substituted aminopropyl part, in the HA and FA columns should be approximately 125 g kg-1. The error in doing this should in the worst case be the following: Fjand HA 3.6%, Aldrich HA 1.5%, and Skagen FA 0.55%. Thus, the content of humic OC is estimated to be the following with the number in parentheses indicating the worst case maximum value: Fjand HA 21 (26) g kg-1, Aldrich HA 7 (9) g kg-1, and Skagen FA 12 (13) g kg-1. The worst case number of aldehyde groups bonded to HA amino groups was estimated by comparing the OC content of the HA-loaded column with that of the blank column and accounting for the C/N ratio of the HA using the data in Table 2. The estimation requires that all the N in the three humic/fulvic acids are amino N and that all the amino groups have reacted with a carbonyl group. However, this does not appear plausible. The amount of hydrolyzable amino acids was 0.098 g N kg-1 (0.7% of total N) in Fjand HA and 1.8 g N kg-1 (16%) in Skagen FA (Grøn et al., 1996). These amounts appear to represent an upper limit for the part of N being amino groups (Anderson et al., 1989). In Aldrich HA the amount of reactive amino groups was determined by its ability to remove crotonaldehyde and determination of the crotonaldehyde concentration before and after reaction as described in Materials and Methods:

[7]


View this table:
[in this window]
[in a new window]
  		Table 2. Properties of the free and the bonded humic and fulvic acids

 
The reduction of the crotonaldehyde concentration corresponded to a content of reactive amino groups in Aldrich HA of 1.04 ± 0.03 g N/kg or 26% of the total content of N. In the Aldrich HA carbonyl column material, the content of OC from humic acid was 19 g kg-1 and that from the post- and pretreatment was 5 g kg-1 (Jonassen et al., 1999). The Skagen FA had a narrow molecular weight distribution below 5000 D, while the Fjand HA had broad molecular weight distributions up to 100000 D (Grøn et al., 1996). The Aldrich HA also had a broad molecular weight distribution but with contributions >100000 D (Tanaka and Senoo, 1995).

Column Composition
Table 2 shows further details of the composition of the free and bounded humic and fulvic acids. The free acids have a medium range of O/C ratio and a low H/C ratio, indicating that partitioning should be an important mechanism for the binding of nonionic organic compounds to these humic substances according to Huang and Weber (1997). Characteristics for the Fjand HA column material are a medium aliphatic and high aromatic content, the Aldrich HA material showed high aliphatic and low aromatic contents, and the Skagen FA material low aliphatic and high aromatic content. Fourier Transformation Infrared Spectroscopy of the bonded Fjand HA and Skagen FA showed a strong peak at 1610 cm-1 and a high content of aromatic C, but higher in the bonded Fjand HA than in Skagen FA. The C composition of the bonded acids was also estimated by means of 13C NMR (see Table 2). These numbers should be interpreted with caution, as it is not possible to do an exact correction for the C originating from the pre- and posttreatment procedure, i.e., non-HA/FA C, (see the end of the section on preparation of the chemically bonded humic and fulvic acid silica gel for HPLC columns [amino (NH2) procedure]). The numbers are based on the assumption that this procedure contributes with the same amount of C in the three humic substance columns as in the blank column. The silica gel used for the Aldrich HA (amino) column was the same product but not the same batch as that used for the other columns (Fjand HA, Skagen FA, and blank), but the different batches should be very similar in their properties. The difficulties doing an exact estimate for the pre- and posttreatment C is also illustrated by the fact that the carbonyl C fraction in the bonded Aldrich HA (amino) gave a negative value. Thus, the sum of the fraction of the other types of C is higher than 1.0 (see Table 2). The bonded Fjand HA has a high content of aromatic C and a medium content of carboxylic C. Both appeared to be reduced compared with the free humic acid, while the content of aliphatic C had increased. The bonded Aldrich HA (amino) has a lower content of aromatic C than the free humic acid, whereas the amount of carboxylic C appeared to be the same. The bonded Skagen FA appeared to be enriched in aromatic C and depleted in carboxylic C compared with the free acid.

Variation of the Capacity Coefficients on Different Columns
The HPLC capacity coefficient, k' = /t0, is a relative measure for the distribution coefficient, K = , of the compound, where Cs = sorbed compound and Cw = dissolved compound. The expression (tr - t0)/tr is a measure for the proportion of the compound being adsorbed to the column material and t0/tr is a measure for the part being dissolved in the eluent (Antworth et al., 1989). The capacity coefficients of all 45 PAC on the Aldrich HA, the Fjand HA, the Skagen FA and the blank columns are presented in Table 1 and compared with Kow (octanol–water partition coefficients). In addition, results for the Aldrich HA using the carbonyl procedure are also presented. A higher amount of methanol was used in the eluent for Fjand HA and Aldrich HA than for Skagen FA and the blank column to keep the retention times approximately the same on the five columns (see Table 1). An increase of the methanol content causes k' to decrease. This should be borne in mind comparing the log k' for the five columns. The need for different methanol content implied that Fjand HA was the strongest sorbent (highest affinity binding) and Skagen FA and the blank column was the weakest ones. The Fjand and Aldrich HA results do not appear to have been significantly affected by the eventual swelling of the column material with increasing methanol content as discussed later, although some minor effects cannot be ruled out.

The column material is probably linked to residues of carbohydrates and aromatic quinones in the humic substance when prepared by the carbonyl procedure and to peptide residues by the amino procedure. Thus, a comparison of the retention on the two types of column materials should make it possible to determine if the attachment procedure causes major changes in the humic acid composition. Figure 1 shows a comparison of the log k' values on the two types of column materials applying the same eluent composition. In general, there is a good accordance between the results. However, as discussed later, e.g., in the section on retention of substituted PAH, there are significant differences for some compounds. The retention of a compound is dependent both on the column surface area of the column material (BET) and its humic acid content, as expressed as humic acid organic carbon (HAOC):



View larger version (14K):
[in this window]
[in a new window]
  		Fig. 1. The relationship between log k' for the two Aldrich HA columns prepared by the amino procedure (NH2) and the carbonyl procedure (C=O). The methanol fraction in the eluent was 0.65 (see Table 1)

 
The ratio C=O/NH2 = / = 0.60 is very close to the slope of 0.62 ± 0.02 for the regression line plotting k'C=O vs. k'NH2 (data not shown). Figure 1 shows the corresponding log k' to log k' plot. The log BET x HA ratio of 0.94 is also reasonably close to the slope in Fig. 1 of 1.10 ± 0.05 considering the negative intercept value. Thus, in general terms there was fairly good agreement in the properties of the humic acid bonded by the two methods. This also means that the properties of the bonded humic acid should not be very different from those of the free humic acid, at least for the Aldrich humic acid.

The Kow value of a compound is often applied to estimate its Koc value. Figure 2 uses an approach corresponding to that suggested by Seth et al. (1999), applying log Kow as a reference indicator (x axis) and the expression log (k'/Kow) as the y axis. This is a way to show if other factors than hydrophobicity contribute to the partitioning. The regression line for the hydrophobic PAH is incorporated in the figures as a reference line. Data for N-methylquinolinium iodide is excluded in these plots, because its behavior is very different from those of the other compounds. The N-methylquinolinium is a cation in contrast to the other neutral compounds. In general the figure indicates a good correlation between the log k' values [plotted as log (k'/Kow)] and log Kow values. However, some systematic variations can be observed comparing Fig. 2 with the values in Table 1. These variations are most pronounced for the humic acid columns, Fjand and Aldrich (both the amino and the carbonyl version). The sorption to the column materials is relatively stronger for the polar compounds (N-PAC, substituted N-PAC and polar substituted PAH), when the Kow values are used as a relative reference. The relative increase is particularly strong for those polar PAC having low Kow values. Thus, the comparison indicates that for polar compounds chemical interactions contribute to the partitioning process in addition to the hydrophobicity. The sorption of PAH with lipophilic substituents, especially the three dibromopyrenes, is relatively weaker than the sorption of the unsubstituted PAH. The reasons for this are not known, but an explanation might be that the substituents cause some steric hindrance in the sorption process to the column material. The same type of deviations from the PAH line was observed for columns prepared by either the amino method or the carbonyl method, although the PAC containing polar substituents are adsorbed relatively stronger on the carbonyl column than on the amino column.



View larger version (19K):
[in this window]
[in a new window]
  		Fig. 2. The relationship between log (k'/Kow) and log Kow for the humic acids Fjand HA and Aldrich HA, a fulvic acid, Skagen FA (all attached using amino-groups for the binding), a blank column using the same procedure for attachment without adding HA or FA and Aldrich HA (carbonyl) using carbonyl in HA for the attachment. The methanol content in the eluent was as shown in Table 1. PAH are polycyclic aromatic hydrocarbons, X-PAH-substituted PAH, N-PAC-azaarenes, and X-N-PAC- substituted azaarenes

 
Retention Dependency of the Mobile Phase Composition
The capacity factors (log k') of bicyclic aromatic compounds were determined changing the methanol/water ratio in the mobile phase (see Fig. 3) . For most of the bicyclic aromatic compounds the retention increased with increasing eluent water content. When the methanol content of the mobile phase increases, the polarity of the mobile phase and the retention of the solute decreases. The increase in retention with decreasing methanol content is found for all other compounds than N-methylquinolinium iodide as shown in Fig. 3. The elution order of the other compounds was independent of the methanol content in the mobile phase on the Skagen FA and the blank columns as indicated by the lines on Fig. 3 not crossing. On the Aldrich HA (amino) and the Fjand HA columns 2,6-dimethylquinoline and naphthalene changed their elution order at high water content (for the Aldrich HA column >0.85 and for the Fjand HA column >0.75 of water). On the Skagen FA column there was no significant change in the capacity coefficient of N-methylquinolinium iodide when the water fraction of the eluent was changed from 0.45 to 0.85. A different behavior is observed for the Aldrich HA and the Fjand HA columns. With increasing water content in the mobile phase, the retention of N-methylquinolinium iodide decreases slightly. This decrease may be caused by a decrease in the size of the humic acid or an increase in the rigidity of the structure of the humic structure. When water fraction reaches ~0.50 (Fjand HA column) and ~0.70 (Aldrich HA column) the capacity coefficient started to increase. However, the increase is not large compared with those for the other compounds.



View larger version (38K):
[in this window]
[in a new window]
  		Fig. 3. The influence of the water content of the mobile phase on the retention of bicyclic aromatic compounds on the Aldrich HA (amino procedure), Skagen FA, Fjand HA, and blank column

 
Values of log k' for seven of the eight bicyclic aromatics were linearly dependent on the eluent water concentration for the Fjand HA, Aldrich HA, and Blank column, although there are some minor deviations from the linearity in the low end of the eluent water content (Fig. 3). This suggests that changes in the eluent composition did not cause major changes in the sorption mechanism. The ranges of the correlation coefficients were as follows: r = 0.995 to 0.999 (Fjand HA), r = 0.995 to 0.9998 (Aldrich HA), and r = 0.982 to 0.999 (Blank). For Skagen FA a better relation between log k' and the eluent water concentration was achieved using a second-order relation: log k' = ax2 + bx + c, where x is the eluent water proportion, r = 0.987 to 0.998.

The k' data was extrapolated to an eluent composition of pure buffered water (water fraction = 1.0) for seven of the eight bicyclic aromatics (not N-methylquinolinium iodide). Figure 4 compares the extrapolated k' values on the three HA/FA columns with that on the blank column. The relative extrapolated k' values are Fjand HA 3.5, Aldrich HA 1.8, Blank 1.0, and Skagen FA 0.6. The comparison suggests a decrease in the sorption in the range Fjand HA > Aldrich HA > Blank ~ Skagen FA. The retention of the bicyclic aromatics was greater on the Skagen FA column than on the blank column when the eluent water fraction was in the range 0.55 to 0.85, but lower when the proportion was 0.45 or 0.95 (see Fig. 3). The major difference between Skagen FA and Fjand HA and Aldrich HA is the size of the molecules. Thus, the results indicate that the molecule size of the humic substances is the dominant factor affecting the sorption. The higher sorption capacity of the Fjand HA column relative to the Aldrich HA column appears to be caused by its higher loading with the humic acid. As discussed in the section Direct Determination of Koc, the determined Koc values (which account for loading) of bicyclic aromatics are almost the same for the Fjand and the Aldrich HA columns. The Fjand HA and Aldrich HA column materials have very different C composition. Thus, our results did not corroborate the observations of Gauthier et al. (1987), that the type of the humic acid C is an important factor affecting the binding of PAC.



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 4. Comparison of the capacity coefficient, k' (achieved by extrapolation), for bicyclic aromatics at buffered water (without methanol) of the three humic/fulvic acid bounded columns, Fjand HA, Aldrich HA (amino), and Skagen FA, with the corresponding k' values for the blank column. Legends: QNO, quinoline-N-oxide; 2HOQ, 2-hydroxyquinoline; iQ, isoquinoline; Q, quinoline; 6MQ, 6-methylquinoline; 2,6-dMQ, 2,6-dimethylquinoline; N, naphthalene

 
Retention of Unsubstituted PAH and N-PAC on Different Columns
The capacity coefficients (k') of unsubstituted PAH and N-PAC increased with the molecular size of the PAH and N-PAC for all columns as exemplified by comparison of napthalene, anthracene, benz[a]anthracene, and benzo[a]pyrene (Table 1); likewise for quinoline, acridine, and 10-azabenzo[a]pyrene. The N-PAC has smaller k' than the corresponding PAH. For both types of compounds the increase with molecular weight was larger in the range Fjand HA > Aldrich HA > Blank ~ Skagen FA. Thus, for the Fjand HA column the capacity coefficients varies by a factor of 119 from the bicyclic naphthalene to the pentacyclic dibenz[a,c]anthracene. For the Aldrich HA column the increase was a factor of 74, for the Skagen FA column the factor was 57 and for the blank column without any humic/fulvic acid it was 65. For the corresponding N-PAC, isoquinoline and dibenz[a,c]acridine, the increase of k' on the four columns was by a factor of 178 for the Fjand HA column, 113 for the Aldrich HA column, 68 for the Skagen FA column, and 92 for the blank column. For all the five column materials, k' as well as log Kow for the various dibenzacridines decreased in the order: dibenz[c,h]acridine > dibenz[a,c]acridine, {approx} dibenz[a,h]acridine > dibenz[a,j]acridine, reflecting the shielding of the N atom by the benzene rings. The shielding causes the behavior of the shielded N-PAC to be more like that of PAH (Helweg et al., 1997a; Nielsen et al., 1997). The relatively stronger sorption of polar compounds on Aldrich HA (carbonyl) column material is perhaps the reason that the order of sorption of the four dibenzacridines is the opposite on this column.

Retention of Substituted PAH Compounds on Different Columns
Generally the capacity coefficients, k', increased with the addition of lipophilic substituents and decreased with the addition of polar substituents; Kow and k' increased for the substituted anthracenes on the five columns as seen in Table 3. Processes other than partitioning contribute to the sorption, as the order is not the same on the five different columns as in Kow. The largest difference is the position of anthraquinone (AQ) in these rows. Anthraquinone elutes relatively faster in the order: Blank > Kow ~ Skagen FA > Aldrich HA (amino) > Fjand HA > Aldrich HA (carbonyl). A similar pattern can be seen for cyano substituent, although the changes are smaller. 9-Cyanoanthracene is adsorbed stronger on Fjand HA and Aldrich HA than one should expect, if the sorption process solely was a partitioning phenomenon. Two factors are likely to contribute to a high k': (i) partitioning, and (ii) specific interactions with the humic acid. Examples of specific interactions are H bonding and charge transfer or dipole–dipole interactions. In the case of anthraquinone probably both types of specific interactions are active. In the case of 9-cyanoanthracene, only charge–transfer or dipole–dipole interactions seems likely. 9-Nitroanthracene shows also variations in its positions in the elution sequences. Its position in the blank column sequence compared with others indicates that H bonding is the dominant specific interaction. The specific interactions between some of the 9-substituted anthracenes and humic acid are plausible, considering that humic acid is assumed to consist of quinoid and phenolic structures.


View this table:
[in this window]
[in a new window]
  		Table 3. Ranking of the effect of substituents on the sorption of anthracene derivatives on five different column materials and the octanol–water partition coefficient, Kow.{dagger}

 
Retention of Substituted N-PAC on Different Columns
The most significant variation of capacity coefficients on the three columns can be noticed in the case of polar substituted N-PAC. Figure 4 compares the extrapolated capacity coefficients for the three HA/FA columns with those of the blank column with buffered water (no methanol). The capacity coefficients increased in the order:

The sorption of N-methylquinolinium ion is very dependent of the type of humic substance, as it increases a factor of 20 from the Skagen FA column to the Fjand HA column (Fig. 3). The eluent composition has also some influence on its sorption, but the influence is much less than that on the neutral bicyclic aromatics (see Fig. 3). On the blank column no peak of N-methylquinolinium iodide was observed in the chromatograms. The reason for this is not known. The N-methylquinolinium ion is expected to bind to negatively charged groups on the column material. This may be carboxylate and phenolate ions or residual silanols having a pKa of about 5 to 7 in the silica surface column material (Dorsey and Cooper, 1994). The residual silanols do not, however, appear to be important as k'Fjand and k'Aldrich for N-methylquinolinium ion is much higher than k'Skagen. The number of residual silanols is probably almost the same in the two HA materials as in the Skagen FA material as all columns initially are treated the same way (see Materials and Methods).

Experiments with N-methylquinolinium iodide at varying pH showed a pH effect on the sorption of N-methylquinolinium iodide (see Fig. 5) with a linear dependency of pH in the range 5.5 to 7.5. The sorption of quinoline and acridine to the free Aldrich HA and Fjand HA at varying pH (see Fig. 6) also clearly illustrates the effect of a binding between a positively charged N-PAC and negatively charged carboxylate group in the humic substance as discussed previously in Nielsen et al. (1997). The sorption coefficients had a maximum at pH = 4 to 5 and Koc decreased strongly when pH fell below 4. However, the retention of N-methylquinolinium ion on the three HA/FA columns may also be more complex than a simple ionic interaction. The sorption of the ion is stronger to the Aldrich HA than the Skagen FA column even though the Skagen FA column material has a higher content of carboxylic C.



View larger version (18K):
[in this window]
[in a new window]
  		Fig. 5. The influence of pH on the retention of N-methylquinolinium iodide on Skagen FA (eluent: methanol fraction 0.25 + buffered water fraction 0.75), Aldrich HA (amino procedure) (eluent: methanol 0.35 + buffered water 0.65), Fjand HA columns (eluent: methanol 0.45 + buffered water 0.55). pH was determined in the eluent mixtures (see Roses et al., 1996)

 


View larger version (15K):
[in this window]
[in a new window]
  		Fig. 6. The influence of pH on the binding of acridine to dissolved Fjand humic acid and quinoline to dissolved Aldrich HA from Nielsen et al., 1997

 
The sorption order of 2-hydroxyquinoline, quinoline-N-oxide and quinoline was different on the two Aldrich columns, confirming that polar PAC are relatively stronger adsorbed to the carbonyl column than to the amino column. Thus 2-hydroxyquinoline and quinoline-N-oxide eluted later than quinoline on the carbonyl column.

Direct Determination of Koc
The HPLC method can be applied to do direct determination of Koc in case of a linear relationship between log k' and the eluent water content as found for the Fjand and Aldrich HA (amino) columns. The results for this part of the study are presented in Table 4. This appears to be an important aspect considering the variety of humic acid at different locations and the variety of organic pollutants. Although the bonded humic acid has been modified compared with the free one, the achieved Koc values should be more useful than the Kow values (Seth et al., 1999). Differences between Koc values for the Fjand HA and the Aldrich HA were minor, indicating that sorption was influenced more by the size (molecular weight) of the HA molecules than by the content of aromatic C (Table 2). The major uncertainty in the Koc determination is the assessment of the content of humic acid C on the two columns. However, in the worst case this should only affect the log Koc values with an uncertainty of 0.1. For Fjand HA there is a good agreement with the Koc HPLC value for quinoline and the estimated one for quinoline using the Koc acridine data from the batch experiment (see Fig. 6). The direct log Koc values for naphthalene and quinoline for the Aldrich HA column deviate with 0.2 to 0.6 from the literature batch values, but this is also a good agreement considering the different techniques. Work is in progress to test a large number PAC for direct Koc determinations on Aldrich HA (carbonyl) (Jonassen et al., 1999). The present study was limited because the relation between log k' and the water eluent only was investigated for the bicyclic PAC. The method for the direct determination was not applied for the results on the Skagen FA column, as the k' values were almost the same as for the blank column.


View this table:
[in this window]
[in a new window]
  		Table 4. The Koc of bicyclic aromatics determined directly from k' on the bonded HA columns or indirectly and literature values (Nielsen et al., 1997)

 
Influence of Addition of Salt [Ca(NO3)2] to the Mobile Phase
The method using bonded HA is also useful and fast in the study of salt effects. The influence of salt addition on retention of bi- and tricyclic PAC to the Aldrich humic acid column was investigated, because inorganic ions may compete with organic compounds on sorption sites or change the structure of humic substances. Figure 7 depicts the salt effect on a few representative compounds, but the total data set includes all the bi- and tricyclic compounds mentioned in Table 1. Addition of 0.007 M Ca(NO3)2 to the mobile phase decreased the retention slightly, but if the salt concentration was increased further, little or no change on the retention was noticed. The salt effect seems to be strongest for substituted PAH. Increasing the salt concentration in the mobile phase from 0 to 0.007 M decreased the retention for substituted PAH (nine anthracene derivatives) 1.17 ± 0.13 times; for PAH (four compounds) 1.10 ± 0.05 times; for O, S-PAC (two compounds) 1.09 ± 0.04 times; for N-PAC (eight compounds) 1.07 ± 0.04 times; and for substituted N-PAC (three compounds) 1.00 ± 0.03 times. The results from a representative of the four former classes are depicted in Fig. 7. The negative salt effect on the sorption of quinoline to a surface-bound humic acid, of phenanthrene to Aldrich humic acid, and of anthracene to Suwannee River humic and fulvic acid was observed by others (Chorover et al., 1999; Lassen and Carlsen, 1997; Schlautman and Morgan, 1993). The possible causes of electrolyte effects on tr include conformation of the bonded humic acid, e.g., increased coiling with increasing Ca2+ concentration (Tombacz and Meleg, 1990) and decreasing PAC solubility with increasing electrolyte concentration (Schwarzenbach et al., 1993). If the only mechanism was a salting out effect, it is estimated that tr for anthracene should have increased 1.01 times at the 0.007 M Ca2+ concentration and 1.05 times at the concentration on 0.07 M. The estimate is based on data from Xie et al. (1997). It is also interesting that the effects in general almost counterbalanced each other at salt concentrations above 0.007 M.



View larger version (18K):
[in this window]
[in a new window]
  		Fig. 7. The influence of salt [Ca(NO3)2] concentration (M) on the retention time, tr (min) of selected tricyclic aromatic compounds on the Aldrich HA column (amino procedure)

 

   CONCLUSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSION
REFERENCES
 
A technique involving bonding of humic substances to HPLC columns allowed rapid determination of relative binding affinity of a wide range of PAC to HA. The Koc values determined for selected compounds on humic acid by this method were comparable to values determined on the free (nonbonded) humic acid and to values reported in the literatue. In addition, similar results were achieved for columns when Aldrich HA was bonded by two different methods. Thus, the possible modification of the humic substances during the bonding was in general not a problem, although it is not negligible. Column studies on two HAs and a FA showed that sorption of PAC was determined by the size (molecular wt.) of the humic molecules. Data indicated that the predominant mechanism for PAC bonding to HA and FA was partitioning, but with contributions of other mechanisms such as H bonding, ionic bonding (for ionic PAC), dipole–dipole interactions, or steric effects. Sorption of PAC to humic substances increased with size (molecular wt.) of the PAC. Lipophilic substituents at the PAC molecule caused sorption to increase while substitution of polar substituents caused sorption to decrease. However, the sorption of PAC with polar substituents is stronger than indicated by their Kow values. This is especially distinct for nitro and cyano-substituents or the presence of a quinone structure. Sorption of PAC was also affected by aqueous Ca(NO3)2 concentration in the eluent, particularly at Ca2+ between 0 and 0.007 M.


   ACKNOWLEDGMENTS
 
The project was supported by funds from the Centers of Ecotoxicological Research and of Biological Processes in Contaminated Soil and Sediment under the Danish Environmental Research Program, the Danish Natural Science Research Council, and the Estonian Science Foundation. The 13C-NMR spectra were made by Prof. Elisabeth Shabanova, University of Copenhagen, and the element analysis of Prof. Gunnar Hessellus, Micro Chemistry AB, Uppsala, Sweden. One of the referees is gratefully acknowledged for his many valuable comments and suggestions to the manuscript.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSION
 REFERENCES
 




This article has been cited by other articles:


Home page
 Poult. Sci.Home page
C. J. van Rensburg, C. E. J. Van Rensburg, J. B. J. Van Ryssen, N. H. Casey, and G. E. Rottinghaus
In vitro and in vivo assessment of humic acid as an aflatoxin binder in broiler chickens.
Poult. Sci., September 1, 2006; 85(9): 1576 - 1583.
[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 (8)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Kollist-Siigur, K.
Right arrow Articles by Kirso, U.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Kollist-Siigur, K.
Right arrow Articles by Kirso, U.<