Professional Documents
Culture Documents
E
nhanced primary production in surface water (i.e.,
Factors such as pH, solution ion composition, and the presence
eutrophication) has become a worldwide problem that
of natural organic matter (NOM) play a crucial role in the
effectiveness of phosphorous adsorption by iron oxides. The results in algal blooms, the development of harmful
interplay between these factors shows a complicated pattern cyanobacteria, fish kills, smell nuisance, and other detrimen-
and can sometimes lead to controversial results. With the help tal consequences (Carpenter et al., 1998). In the past decades,
of mechanistic modeling and adsorption experiments, the net considerable reduction of P loss has been achieved thanks to
macroscopic effect of single and combined factors can be better
measures to control industrial release and sewage water treat-
understood and predicted. In the present work, the relative
importance of the above-mentioned factors in the adsorption of ment. As a result, the relative importance of P release from
phosphate was analyzed using modeling and comparison between agricultural fields has increased in recent years (Barberis et al.,
the model prediction and experimental data. The results show 1996; Chardon and Koopmans, 2005; Koopmans et al., 2007;
that, under normal soil conditions, pH, concentration of Ca, and Sharpley et al., 1994).
the presence of NOM are the most important factors that control
Various strategies and measures have been proposed, tested,
adsorption of phosphate to iron oxides. The presence of Ca not
only enhances the amount of phosphate adsorbed but also changes and implemented to reduce release of P from agricultural fields,
the pH dependency of the adsorption. An increase of dissolved including input reduction and leaching prevention (Kronvang
organic carbon from 0.5 to 50 mg L−1 can lead to a >50% decrease et al., 2005; Sharpley et al., 1994; Withers et al., 2000). In
in the amount of phosphate adsorbed. Silicic acid may decrease agriculture fields, to guarantee sufficient nutrient supply for
phosphate adsorption, but this effect is only important at a very
crop production, a solution concentration of above 3 mg L−1
low phosphate concentration, in particular at high pH.
is required, and P concentrations as high as 30 mg L−1 can be
found in soil runoff (Withers et al., 2000). The P concentra-
tion in leachate from agricultural fields is much higher than the
target value for P in surface water (<0.1 mg L−1). One attrac-
tive approach to reduce leaching is to use P-fixing materials
that are directly mixed with soil or used to build a reactive
barrier (Codling et al., 2000; McDowell et al., 2008; Stout
et al., 2000). Iron-rich materials are good candidates to build
P-removing filters (McDowell et al., 2008; Zeng et al., 2004;
Zhang et al., 2009). Such materials can be metal iron (e.g.,
zerovalent iron) (McDowell et al., 2008; Sakadevan and Bavor,
1998) and iron oxides in various forms (Boujelben et al., 2008;
Pan et al., 2010; Zeng et al., 2004).
Phosphorus in soil drainage water consists of orthophos-
phate, organic P, and P associated with mineral colloidal particles
(adsorbed or precipitated). A permeable reactive filter/barrier
containing iron oxides can, in principle, remove all three forms
of P. The mineral colloidal P can be intercepted by filtration
effects, and orthophosphate and organic P can be immobilized
by adsorption reactions. The efficiency and effective lifespan
of a P barrier depends on various factors, including the basic
Copyright © 2012 by the American Society of Agronomy, Crop Science Society characteristic of the adsorptive material itself (e.g., the porosity,
of America, and Soil Science Society of America. All rights reserved. No part of reactive surface area, P-binding affinity, and binding capacity)
this periodical may be reproduced or transmitted in any form or by any means, and environmental factors (e.g., pH, ion composition, the pres-
electronic or mechanical, including photocopying, recording, or any information
storage and retrieval system, without permission in writing from the publisher. ence of other compounds, and water flow). The reactive groups
present on the surface of iron oxides can bind various cations
J. Environ. Qual. 41:628–635 (2012)
doi:10.2134/jeq2011.0250
Posted online 13 Jan. 2012. Dep. of Soil Quality, Wageningen Univ., P.O. Box 47, 6700 AA, Wageningen, The
Received 14 July 2011. Netherlands. Assigned to Associate Editor Ray Bryant.
*Corresponding author (liping.weng@wur.nl).
© ASA, CSSA, SSSA Abbreviations: DOC, dissolved organic carbon; FA, fulvic acid; LCD, ligand and
5585 Guilford Rd., Madison, WI 53711 USA charge distribution; NOM, natural organic matter.
628
and anions. The chemical reactions at the surface become more The goethite material was prepared based on the procedure
difficult to predict the more the ions and compounds interact. described by Hiemstra et al. (1989). The specific surface area
The complete set of chemical reactions that may take place in the of the goethite material was 94 m2 g−1, as determined using
field situation leads to a very complex system. the BET-N2 adsorption method (Weng et al., 2005). Fulvic
In adsorption reactions, when equilibrium is maintained, acid (FA) was used to simulate NOM, and the FA material
the relation between solution concentration of phosphate (C, was purified from a peat soil (Strichen Soil Assoc., Scotland)
mol L−1) and the corresponding loading on the surface (Γ, mol following the International Humic Substance Society proce-
kg−1) can be expressed using the Langmuir model: dures. The purified FA material contains 43% C, and its aver-
age molar mass is 0.68 kDa, which was measured using size
K 'C
Γ = Γ 'max [1] exclusion chromatography.
1 + K 'C This study included three series of treatments: (i) no FA,
where Γ′max is the apparent maximum binding capacity (mol (ii) 150 mg L−1 FA (64.5 mg C L−1), and (iii) 300 mg L−1 FA
kg−1) and K′ the conditional affinity constant. For a reactive (129 mg C L−1). In all series, the suspension contained final
barrier, the increase of the P loading equals the net amount of total concentrations of 2.7 g L−1 goethite, 0.01 CaCl2, and 0.48
P added per unit adsorptive material; this is equal to V(C0 – mmol L−1 NaH2PO4. The solutions were added immediately
C), where V (L kg−1) is the volume of solute drained expressed after each other in the order CaCl2, goethite, NaH2PO4, and
per kg of adsorptive material, and C0 and C are the input and FA. For all series, acid or base solution (0.01 mol L−1 NaOH
output P concentrations, respectively. If the latter is small, the and 0.05 mol L−1 HCl) was added to the suspension to adjust
average loading of the reactive material can be approached by the pH to selected values in the range of pH 3 to 7. During
VC0. Under this condition, Eq. [1] can be written as the preparation, the gas-tight vessels used in the experiments
were flushed with N2 gas to minimize the influence of CO2.
K 'C
Γ = VC 0 = Γ 'max [2] The prepared suspensions were shaken at 20°C for 4 d and
1 + K 'C centrifuged at 18,000 g for 30 min. The final pH in the super-
Rearrangement of Eq. [2] results in natant was measured using a pH meter. The P concentration
in the supernatant was measured with ICP–MS. The Ca con-
VC 0 centration was measured with ICP–AES. The concentration of
C= [3]
(Γ 'max −VC0 ) K ' dissolved organic carbon (DOC) was measured with a TOC
analyzer. The amounts of P, Ca, and FA adsorbed were derived
To maintain the C under a certain target value, the maxi- from the difference between the treatments and the blank (no
mum amount of drainage that can be treated depends strongly goethite). The small contribution of organic P (P in FA) to the
on both Γ′max and K′. In a mechanistic model, the maximum total soluble P measured is corrected using P content in FA
binding capacity depends on the amount and type of iron measured in a FA solution (0.16 μg mg−1) and DOC. The final
oxides in the material, the particle size (surface area), and acces- DOC concentration ranged from 15 to 75 mg L−1.
sibility of the surface, i.e., Γmax is characterized by the mate-
rial. When this type of Langmuir model is derived empirically, Modeling
the apparent Γ′max also depends on factors that may promote
adsorption (like Ca2+) or decrease the adsorption due to com- CD-MUSIC Model
petition with natural organic matter (NOM). The conditional Iron (hydr)oxides can occur in nature in various crystal forms,
affinity constant K′ depends on the type of reactive groups such as goethite (α-FeOOH), hydrous ferric oxide, or ferrihy-
present on the surface (a character of the material) and is also drite, hematite (α-Fe2O3), and lepidocrocite (γ-FeOOH). All
strongly affected by the solution chemistry (e.g., pH and pres- these solid phases have a strong affinity for phosphate. From
ence of other adsorbing compounds). a more fundamental point of view, much more is known with
The objective of this study was to apply the most recent respect to the adsorption of phosphate to goethite, whereas
developments in the understanding of ion and NOM adsorp- much less information is available for the other phases men-
tion to iron oxides to gain insight into the most relevant factors tioned. Goethite with varying crystal morphology shows similar
that influence the performance of P-filters under field condi- binding properties for phosphate (Torrent et al., 1990). It has
tions. The relative importance of pH, major cations (Ca2+ and been shown (Hiemstra et al., 2010a) that phosphate desorption
Mg2+), anions (HCO3− and SO42−), neutral species (H4SiO40), data for various soils can be described well by assuming that the
and NOM in phosphate adsorption were analyzed using mech- reactive metal oxide phase in soils can be interpreted based on
anistic modeling and the model prediction were compared the well calibrated adsorption model for goethite, although the
with data of an adsorption experiment. We show that modern process may involve several metal oxide phases and clay miner-
surface complexation models such as the CD-MUSIC model als (Devau et al., 2009, 2011).
(Hiemstra and Van Riemsdijk, 2006) can be applied to explain The surface of the iron oxides consist of a few different types
and predict the net effects of various factors on the quality and surface oxygen groups that differ in their degree of coordina-
capacitance of a barrier. tion with the iron in the solid. These oxygens have a variable
number of protons associated with them. Due to breaking
Materials and Methods of bonds, surface groups with a coordination number that is
Batch experiments were performed to study the effects less than the maximum can occur. The formal charge of the
of pH, Ca, and NOM on phosphate adsorption to goethite. various surface groups can be derived using a simple Pauling
Fig. 2. The surface composition of goethite in equilibrium with K+, Na+, Ca2+, Mg2+, H4SiO40, Cl−, and HCO3− at concentrations representative for
natural soil solutions (see text) as a function of the phosphate concentration in solution for pH 4, 6, and 8 at PCO2 = 4 x 10−4 bar. Compositions
have been calculated with the CD-MUSIC model using the parameters of Table 1 (Hiemstra et al., 2007, 2010a). In the systems of Fig. 2a and 2b, no
natural organic matter is present, whereas in Fig. 2c and 2d, the natural organic matter loading is 1.66 μmol m−2. Figure 2a and 2b refer to the same
calculation, but the loading is given at the logarithmic and linear scales, respectively. Figure 2c and 2d refer to the same solution composition,
except a 10-fold higher H4SiO4 concentration (1.0 mmol L−1) was used in Fig. 2d than in 2c.
Fig. 3. Effect of the presence of fulvic acid (FA) (0, 150, and 300 mg L−1)
on the concentration of phosphate in a 0.01 mol L−1 CaCl2 solution
in a system with goethite (2.7 g L−1) having initially 0.48 mmol L−1
PO4 (experimental data points). The effects of FA on phosphate Fig. 4. Effect of dissolved organic carbon (DOC) (0.5–50 mg L−1) on
adsorption were calculated using the ligand and charge distribution the phosphate loading of the natural oxides in 0.01 mol L−1 CaCl2.
approach (lines) applying the CD-MUSIC model parameters of Table The isotherms have been calculated using Eq. 9, which has been cali-
1. For comparison, the calculated soluble phosphate concentration brated on a large set of top soils with variable phosphate equilibrium
is given (thin line) for a system with 0.01 mol L−1 NaCl as background concentrations (~1–30 μmol L−1), DOC (~5–30 mg L−1), and pH (~4–7)
(no Ca and FA). in 0.01 mol L−1 CaCl2.