Environ. Sci. Technol. 2005, 39, 5372-5377
Modeling Salt-Dependent Proton Binding by Organic Soils with the NICA-Donnan and Stockholm Humic Models J O N P E T T E R G U S T A F S S O N * ,† A N D DAN BERGGREN KLEJA‡ Department of Land and Water Resources Engineering, KTH (Royal Institute of Technology), 100 44 Stockholm, Sweden, and Department of Soil Sciences, Swedish University of Agricultural Sciences, Box 7014, 750 07 Uppsala, Sweden
Models are available for simulations of proton dissociation and cation binding by natural organic matter; two examples are the NICA-Donnan and Stockholm Humic (SHM) models. To model proton and metal binding, it is necessary to properly account for the ionic strength dependence of proton dissociation. In previous applications of the models for soils it was assumed that the electrostatic interactions for solid-phase humic substances were the same as in solution; this assumption was recently challenged. Therefore, we reanalyzed previously published acidbase titrations of acid-washed Sphagnum peat, and we produced additional data sets for two Sphagnum peats and two Spodosol Oe horizons. For the soil suspensions, the original NICA-Donnan and SHM models, which were developed for dissolved humic substances, underestimated the observed salt dependence considerably. When a fixed Donnan volume of 1 L kg-1 for humic substances in the solid phase was used, the NICA-Donnan model fits were much improved. Also for SHM, slight changes produced improved model fits. The models also produced acceptable simulations of the dissolved Ca, Mg, and Cd concentrations, provided that cation selectivity was introduced. In conclusion, the proposed extensions to the NICA-Donnan and SHM models were shown to predict the salt dependence of solidphase humic substances more satisfactorily than earlier model versions.
Introduction To simulate acid buffering and metal binding for soils that contain natural organic matter, it is necessary to use correct descriptions for the acid-base properties of the solid-phase organic matter (1, 2). It is known that humic and fulvic acids (HA and FA) express ionic-strength-dependent dissociation behavior, a phenomenon referred to as the “salt effect”. This results from electrostatic interactions between neighboring charged groups. With increasing ionic strength, the individual charges are more increasingly shielded, permitting a greater overall charge at the same pH (3). The salt effect is described in different ways in today’s equilibrium models. In the model of De Wit et al. (4, 5), later * Corresponding author phone: +46-8-7908316; fax: +46-84110775; e-mail:
[email protected]. † KTH (Royal Institute of Technology). ‡ Swedish University of Agricultural Sciences. 5372
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005
refined by Avena et al. (3), the fulvic or humic acid is a charged, rigid impermeable sphere. The resulting potential gradient was described with the Poisson-Boltzmann equation. The magnitude of the salt effect was related to the radius of the HA or FA, which was treated as an adjustable parameter. Another approach is taken in the NICA-Donnan model (6, 7), in which the HA or FA is seen as a permeable gel, where a Donnan potential regulates the partitioning of ions between the bulk solution and the gel phase. Model VI (8) uses the empirical parameter P to describe the salt effect. The Stockholm Humic model, SHM (9), is similar in structure to Model VI (8). The SHM differs primarily in its electrostatic submodel, which uses an approach similar to that of De Wit et al. (4, 5) except that the model contains extra Boltzmann terms to account for the shielding effect of counterions inside the gel. In the SHM, the radius is set to a constant value; instead the proportion of the HA and FA that can be considered as a gel phase (the gf value) is an adjustable parameter. Although many previous studies focused on the influence of ionic strength on the proton dissociation of dissolved humic substances (see compilation in ref 10), there are very few reports dealing with solid-phase organic matter. Bloom and McBride (11) and Marinsky et al. (12) studied the proton dissociation of acid-washed peat at different ionic strengths. When using their rigid impermeable-sphere model for the peat of Marinsky et al. (12), De Wit et al. (4) found that the salt effect was much stronger than for dissolved humic substances. Recently, Smith et al. (13) performed acid-base titrations of one acid-washed peat sample at different ionic strengths, using Model VI in an effort to interpret the results. Again it was shown that the salt effect was much stronger than for dissolved humic substances. To be able to correctly simulate these and the older (11, 12) results, the electrostatic submodel in Model VI with its empirical parameter P was abandoned for humic substances in the solid phase. Instead a Donnan model with a fixed Donnan volume VD was introduced. The revised model was called Model VI-FD. If these observations are true for soil organic matter in general, it would mean that not only the original Model VI but also the NICA-Donnan and SHM models provide incorrect simulations of proton and metal binding when applied at different ionic strengths, as the electrostatic interactions are currently simulated using averaged parameters for dissolved HA and FA (2, 14-16). Based on the results of Smith et al. (13), it can certainly be hypothesized that the electrostatic submodels of not only the Model VI-FD but also the NICA-Donnan and SHM models should be different for solid-phase HA and FA than for dissolved HA and FA, to reflect the stronger salt effect for the former. With this background, we reanalyzed Marinsky’s data set for acid-washed Sphagnum peat (12) with the NICA-Donnan and SHM models, and we carried out additional acid-base titrations on some organic soils to provide more data for the ionic strength dependence of proton dissociation. Our primary objective was to determine if the ionic strength dependence of proton binding by organic soils could be simulated using a consistent set of parameters for the electrostatic interactions.
Materials and Methods Soils. Earlier studies focusing on ionic-strength dependence of proton dissociation have used acid-washed peats (1113). To avoid losses of reactive fulvic acid (FA) fractions in the acid wash, we decided to use fresh unaltered soil material. 10.1021/es0503332 CCC: $30.25
2005 American Chemical Society Published on Web 06/08/2005
Therefore, we chose four organic soils with relatively low concentrations of bound Al, Fe, Ca, and Mg, to limit the effect of bound cations on the developed negative charge. Two Sphagnum peat samples from central Sweden were collected from the surface horizon at a depth of between 10 and 25 cm. One of these, Paskalampa Oi, had a low degree of humification, with a value between H2 and H3 on the Von Post scale (17) and classified as a Sphagnofibrist (18); Sphagnum fuscum was the dominant species. The second peat sample (Torrvedsmossen Oi) was a slightly more humified (H4) Sphagnofibrist in which the dominant vegetation was Sphagnum fallax. The two other organic soils were Oe horizons collected from Spodosols in central Sweden: Risbergsho¨jden Oe and Korsmossen Oe. The former soil was used in earlier investigations (2, 19). The vegetation on both sites was dominated by Scots Pine (Pinus sylvestris). Key soil characteristics are shown in Table S1, Supporting Information. The samples were kept in their field-moist state at 2 °C for a maximum of 2 months before experiments. We included one previously published data set for acidwashed Sphagnum peat (12) to model the proton dissociation at different ionic strengths. This sample was titrated with NaOH in 0.001, 0.01 and 0.1 M NaCl electrolytes. Experimental Procedures. In batch experiments, 1 or 2 g field-moist soil (1.00 g for the Spodosol Oe horizons, 2.00 g for the Sphagnofibrists) was suspended in 30 mL solutions of variable composition in polypropylene centrifuge tubes and shaken for 7 days in a shaking-water bath at 8 °C. Because the experiment was originally designed for studying ionic-strength-dependent Cd binding, the solutions contained approximately 20 µM Cd as Cd(NO3)2. Acid or base (as HNO3 or NaOH) was added, so that each system was studied at a number of (4-8) different pH values, each analyzed in duplicate. Additional NaNO3 salt was added to produce three data sets with different ionic strengths. The concentrations of acid, base, and salts added to the suspensions are listed in Table S2, Supporting Information. Further experimental details are available from the senior author on request. The final ionic strength of the data set with the highest NaNO3 addition was very close to 0.1 M in all samples. The set with intermediate NaNO3 additions was termed ‘0.01 M’, but the final ionic strength (i.e., after equilibration of the samples) differed between approximately 0.005 and 0.011 M, with the lowest values at the highest pH. Finally, for the third data set termed ‘0.001 M’, the final ionic strength was between ∼0.0001 and 0.002 M, with the lowest values at the highest pH. Under these conditions, the calculated ionic strength was model-dependent, cf. the Discussion section. As an example, the addition of 1.3 mM NaOH + 0.02 mM Cd(NO3)2 to the Torrvedsmossen Oi suspension (the highest pH point in the 0.001 M series, see Table S2, Supporting Information) caused the calculated ionic strength to be 0.06 mM and 0.23 mM for the NICA-Donnan and SHM models, respectively. After equilibration, the suspensions were centrifuged at 5000g for 20 min, their pHs were measured (with a Radiometer glass combination electrode), and they were filtered (Acrodisc PF 0.2 µm) at the experimental temperature. About half of each filtered sample was acidified with 0.5% HNO3 prior to the analysis of metals. Samples were analyzed for dissolved organic C within 24 h using a Shimadzu TOC-5000 Analyzer. We determined several elements (Ca, Mg, Cd, Mn, K, Fe, Al, Si) by plasma emission spectroscopy using a Jobin-Yvon JY24 ICP instrument. Chloride and sulfate, which were present in very low concentrations (
