Measuring and Modeling Zinc and Cadmium

concentrations in purified humic acid solutions using the recently developed Donnan membrane ... Present address: National Institute for Integrated In...
15 downloads 4 Views 91KB Size
Download PDF
Anal. Chem. 2002, 74, 856-862

Measuring and Modeling Zinc and Cadmium Binding by Humic Acid Leonard A. Oste,† Erwin J. M. Temminghoff,* Theo M. Lexmond, and Willem H. Van Riemsdijk

Subdepartment of Soil Quality, Department of Environmental Sciences, Wageningen University, PO Box 8005, 6700 EC Wageningen, The Netherlands

Free metal ions in aqueous and terrestrial systems strongly influence bioavailability and toxicity. Most analytical techniques determine the total metal concentration, including the metal ions bound by dissolved organic matter. Ion activity can be measured with ion-specific electrodes (ISEs) for some metals, but an electrode for Zn is not commercially available. As a result, very few data are available on Zn binding by natural dissolved organic matter. The aim of this study is to determine free Zn concentrations in purified humic acid solutions using the recently developed Donnan membrane technique. However, several analytical aspects of the Donnan membrane technique had to be clarified before reliable data could be composed. Cd was chosen for validation. This study shows that free Cd concentrations as measured by the Donnan membrane technique agreed well with Cd ISE measurements. It is also shown that the Donnan membrane technique could be used at high pH. The Donnan membrane technique provided consistent results in a range of p[Cd2+] ) 3-9 and p[Zn2+] ) 3-8 at pH 4, 6, and 8. Metal speciation in humic acid solutions was also calculated with the consistent NICA-Donnan model using generic parameters. The model could excellently describe the experimental data without adjusting any of the parameters (R2Cd ) 0.971, R2Zn ) 0.988). Bioavailability of trace metals in the environment is influenced by both environmental and physiological factors. The environmental conditions (e.g., pH, total metal content in the soil or sediment, number of binding sites, competing ions, etc.) strongly influence the speciation in solution. The uptake by plants and dermal uptake by soil and aquatic organisms was thought to be determined by the free ion activity.1,2 However, there is increasing evidence that also labile metal complexes (e.g., metal chloride or sulfate complexes, but also metal ions bound to organic ligands) influence metal uptake or toxicity.3-6 It is not yet clear whether * Corresponding author. Fax: +31 317 483766. E-mail: [email protected]. † Present address: National Institute for Integrated Inland Water Management and Waste Water Treatment, P.O. Box 17, 8200 AA Lelystad, The Netherlands. (1) Bingham, F. T.; Sposito, G.; Strong, J. E. J. Environ. Qual. 1984, 13, 7174. (2) Morel, F. M. M., Ed. Principles of Aquatic Chemistry, 1st ed.; John Wiley: New York, 1983; 300-308. (3) Lindsay, W. L.; Hodgson, J. F.; Norvell, W. A. In Soil Chemistry and Soil Fertility. Meeting of Commissions II and IV of the International Society of Soil Science; Jacks, G. V., Ed.; University Press: Aberdeen, 1967; 305-316.

856 Analytical Chemistry, Vol. 74, No. 4, February 15, 2002

this is caused by a fast dissociation or by direct uptake of metal complexes. Even if a number of species determine metal toxicity or uptake, the free metal ion is still the key parameter for speciation calculations, and therefore, is crucial in connecting soil chemistry to biological responses. It is, however, complicated to determine the free ion concentration in solutions containing dissolved organic matter (DOM). Hence, speciation models have been used extensively to calculate free ion concentrations in solution on the basis of the total dissolved metal concentration, which could be easily measured. Different models have been developed to describe metal binding by DOM. Dobbs et al.7 developed a continuous multiple site ligand model that was incorporated in MINTEQA2. Tipping8 derived model VI, a discrete site/electrostatic model, which accounts for chemical heterogeneity, competition, ionic strength effects, and proton-metal exchange. The same aspects were accounted for in a model used by Kinniburgh et al.9 This is the consistent nonideal competitive adsorption equation combined with an electrostatic Donnan model to correct for the variable charge nature, which is referred as the NICA-Donnan model. Instead of discrete sites, as in model VI, the NICA-Donnan model uses a continuous affinity distribution. All models have one thing in common: they need reliable data to obtain the model parameters. Parameters for Zn binding by humic acid in model VI are based on 17 data points extracted from two papers written in 1965 and 1971 covering a very limited concentration range.8 As stated, direct measurement of free metal ions is difficult. Common analytical equipment (e.g., atomic absorption spectometry (AAS), inductively coupled plasma combined with optic emission spectometry (ICP-OES) or mass spectometry (ICPMS)) measures total concentrations in a sample. Polarographic techniques determine not only the free ion concentration, but also labile metal complexes. Ion-specific electrodes (ISEs) have been used frequently to measure free Cd, Ca, or Cu activities.10,11 (4) Smolders, E.; McLaughlin, M. J. Soil Sci. Soc. Am. J. 1996, 60, 14431447. (5) Van Leeuwen, H. P. In In Situ Monitoring Aquatic Systems. Chemical Analysis and Speciation; Buffle, J., Horvai, G., Eds.; John Wiley: New York, 2000; 253-277. (6) Zhang, H.; Zhao, F. J.; Sun, B.; Davison, W.; Mcgrath, S. P. Environ. Sci. Technol. 2001, 35, 2602-2607. (7) Dobbs, J. C.; Susetyo, W.; Knight, F. E.; Castles, M. A.; Carriera, L. A.; Azarraga, L. V. Anal. Chem. 1989, 61, 483-488. (8) Tipping, E. Aquat. Geochem. 1998, 4, 3-48. (9) Kinniburgh, D. G.; Van Riemsdijk, W. H.; Koopal, L. K.; Borkovec, M.; Benedetti, M. F.; Avena, M. J. Colloids Surf. A 1999, 151, 147-166. 10.1021/ac0105080 CCC: $22.00

© 2002 American Chemical Society Published on Web 01/16/2002

Although an ionophore for Zn was introduced,12 this never led to the development of a commercially available electrode for Zn. The lack of a Zn ISE is probably the main reason for the very limited number of published data on Zn binding to humic substances. Recently, new methods to determine free ion concentrations in natural solutions have been developed. Holm et al.13 applied a method using a cation-exchange resin. The exchange with the resin highly depends on the ionic strength, the cation composition, and the pH of the sample, which is taken into account by a reference experiment. The reference solution should contain exactly the same electrolyte solution, but no complexing ligands. The method appeared to be very sensitive to [Ca2+] in the range from 0.001 to 0.005 M. Particularly in this range, the free Ca concentration in the sample and reference solution should be equal; otherwise, a correction has to be made on the basis of an empirical standard curve.13 Davison and Zhang14 introduced the DGT technique (diffusion gradients in thin films) based on metal fluxes from the aqueous phase through a highly porous hydrogel to an exchange resin, which functions as a sink. The exchange resin functions well between pH 5 and 9. The diffusion rate of metals through the gel depends on the thickness of the gel layer, the metal ion itself, the metal concentration in the bulk solution, and the diffusion coefficient.15 The diffusion coefficient depends on the properties of the gel, the ionic strength of the solution, and the type of ligand. Not only free ions, but also metal complexes can diffuse through the gel, depending on the size of the ligands and the type of gel.15 By using at least two DGT devices with different gel compositions, the speciation in solution can be estimated.16 The method is suitable for in situ measurement in natural waters,14 but it is complicated for soils. Soil pore water is not very well mixed, which might result in a concentration gradient toward the membrane filter.15 Moreover, at moisture contents lower than field capacity, the contact between pore water and gel surface can be disturbed, complicating the diffusion surface.17 Recently, a new method has been developed by Temminghoff et al.18 and Weng et al.19 The so-called Wageningen Donnan membrane technique (DMT) is based on the research of Lampert20 and Fitch and Helmke.21 The Donnan membrane cell consists of a donor side flushed with aqueous solution containing free metal ions and metal complexes (e.g., metal-DOM), and an acceptor (10) Bresnahan, W. T.; Grant, C. L.; Weber, J. H. Anal. Chem. 1978, 50, 16751679. (11) Milne, C. J.; Kinniburgh, D. G.; De Wit, J. C. M.; Van Riemsdijk, W. H.; Koopal, L. K. J. Colloid Interface Sci. 1995, 175, 448-460. (12) Kojima, R.; Kamata, S. Anal. Sci. 1994, 10, 409-412. (13) Holm, P. E.; Christensen, T. H.; Tjell, J. C.; McGrath, S. P. J. Environ. Qual. 1995, 24, 183-190. (14) Davison, W.; Zhang, H. Nature 1994, 367, 546. (15) Davison, W.; Fones, G.; Harper, M.; Teasdale, P.; Zhang, H. In In Situ Analytical Techniques of Aquatic Systems; Buffle, J., Horvai, G., Eds.; IUPAC: 2000. (16) Zhang, H.; Davison, W. Anal. Chem. 2000, 72, 4447. (17) Hooda, P. S.; Zhang, H.; Davison, W.; Edwards, A. C. Eur. J. Soil Sci. 1999, 50, 285-294. (18) Temminghoff, E. J. M.; Plette, A. C. C.; Van Eck, R.; Van Riemsdijk, W. H. Anal. Chim. Acta 2000, 417, 149-157. (19) Weng, L.; Temminghoff, E. J. M.; Van Riemsdijk, W. H. Eur. J. Soil Sci. 2001, 52, in press. (20) Lampert, J. K. Measurement of trace catin activities by Donnan membrane equilibrium and atomic adsorption analysis. Ph.D. dissertation, University of Wisconsin, Madison, WI, 1982. (21) Fitch, A.; Helmke, P. A. Anal. Chem. 1989, 61, 1298-1300.

side flushed with electrolyte solution with approximately the same salt level as the donor solution. The two sides are separated by a negatively charged semipermeable cation exchange membrane that is completely deprotonated above pH 2. Cations can easily pass the membrane, whereas the transport rate of negatively charged compounds in solution, for example, humic acids and Chloride, is practically zero. When equilibrium is reached, the free metal ion concentration in the acceptor and the donor will become identical if ionic strength and ionic composition of both solutions are equal. A simple correction can be made if the donor and acceptor solution are different.19 Temminghoff et al.18 showed that DMT measurement of free Cd and Cu concentrations in the presence of various (in)organic complexing agents (chloride, EDTA, humic acid) agreed very well with speciation calcuations. They performed experiments in a background electrolyte ranging from 1 to 10 mM Ca(NO3)2, whereas pH ranged from 4 to 6. The objective of this paper comprises the reliable determination of free Zn concentrations in solutions containing purified forest soil humic acid (FSHA). Data were collected over a large Zn range and at different pH values. Furthermore, we model the obtained data with the NICA-Donnan model using a generic set of parameters as determined by Milne et al.22,23 Some specific experimental conditions in this study had not been evaluated by Temminghoff et al.18 and Weng et al.19 Therefore, we evaluated three analytical aspects of the Donnan membrane technique. First, the influence of the background electrolyte was studied. Metal binding data were ideally measured without other ions that compete for the binding sites of organic matter, such as Ca. We evaluated the possibilities of NaNO3 as a background electrolyte, because only Ca(NO3)2 had been used until now. Second, Temminghoff et al.18 compared DMT measurement with speciation calculations, but we compare DMT with ISE measurement. Third, the Donnan membrane technique had been validated for metal complexes with synthetic complexing ligands at pH 5.18 We evaluate the behavior of the system at pH values up to 8, when CO2 diffusion can cause the formation of significant amounts of metal carbonate complexes in the system. After the analytical aspects were studied, we measured Zn. THEORY Model. The NICA-Donnan model is based on the NICA equation, an analytical isotherm for multicomponent adsorption to heterogeneous surfaces.24 The NICA model was extended by incorporating nonspecific electrostatic binding25 and by achieving thermodynamic consistency.9 Eq 1 shows the bimodal form of this equation, since it is well-known from heterogeneity analysis that there are two broad distributions related to carboxylic and phenolic sites.26 (22) Milne, C. J.; Kinniburgh, D. G.; Tipping, E. Environ. Sci. Technol. 2001, 35, 2049-2059 (23) Milne, C. J.; Kinniburgh, D. G.; et al.. Environ. Sci. Technol. Submitted. (24) Koopal, L. K.; Van Riemsdijk, W. H.; De Wit, J. C. M.; Benedetti, M. F. J. Colloid Interface Sci. 1994, 166, 51-60. (25) Kinniburgh, D. G.; Milne, C. J.; Benedetti, M. F.; Pinheiro, J. P.; Filius, J. D.; Koopal, L. K.; Van Riemsdijk, W. H. Environ. Sci. Technol. 1996, 30, 1687-1698. (26) Nederlof, M. M.; Van Riemsdijk, W. H.; Koopal, L. K. Environ. Sci. Technol. 1992, 26, 763-771.

Analytical Chemistry, Vol. 74, No. 4, February 15, 2002

857

Qi )

ni,1 nH,1

Qmax,1

[

(K ˜ i,1cD,i)ni,1

∑(K˜ j

ni,2

Qmax,2 nH,2

nj,1 p1

]

+

1+[

j,1cD,j)

j,1cD,j)

j

×

nj,1

∑(K˜

∑(K˜

nj,1 p1

j,1cD,j)

]

j

(K ˜ i,2cD,i)ni,2

∑(K˜

nj,1

j,2cD,j)

[

∑(K˜

nj,2 p2 j,2cD,j) ]

j

×

1+[

j

∑(K˜

(1) nj,2 p2

j,2cD,j)

]

j

where Qi is the adsorbed quantity of component i (in mol kg-1 organic matter), Qmax is the maximum adsorption capacity of the organic matter for protons (in mol kg-1 organic matter), K ˜ j is the median of the affinity constants, Kj, of component j, where j includes component i, nj is the the nonideality parameter of component j, p is the the intrinsic heterogeneity parameter of the organic matter, subscript 1 is “carboxylic”-like types of groups, subscript 2 is “phenolic”-like types of groups, and cD,j is the concentration of the component j in the Donnan phase (mol L-1). cD,j is related to the concentration in the bulk solution, cj, according to eq 2,

cD,j ) e-zjΨD/kTcj

(2)

where z is the valence of the ion, ΨD is the Donnan potential, k is the Boltzmann’s constant, and T is the temperature (in K). It is assumed in the Donnan model that there is a uniform distribution of potential within the Donnan volume or “phase”. The Donnan model does not require any particular assumption about the geometry of the humic particles. This contrasts with the diffuse double-layer model in which the potential varies in a systematic way with distance from the interface and, therefore, is dependent on a particular particle geometry. The Donnan volume is related to the ionic strength (I) according to the following empirical relation (eq 3),27

log VD ) b(1 - log I) - 1

(3)

in which the coefficient b varies with the type of humic substance. There is a close relation between the Donnan volume and the Donnan potential: for a given net charge on the humic particle, a smaller Donnan volume implies a higher concentration of counterions in the Donnan phase, a higher Boltzmann factor, and therefore, a higher absolute value of the Donnan potential. (27) Benedetti, M. F.; Van Riemsdijk, W. H.; Koopal, L. K. Environ. Sci. Technol. 1996, 30, 1805-1813.

858 Analytical Chemistry, Vol. 74, No. 4, February 15, 2002

Figure 1. Overview of the DMT equipment.

EXPERIMENTAL PROCEDURES Equipment. A personal-computer-based automatic titrator, developed by Kinniburgh et al.,28 was used to conduct the ISE experiments. We used four motorized burets (Metrohm 665 Dosimat) to add metal solutions to the reaction vessel (10-3 M and 10-1 M metal nitrate solutions) and 0.1 M HNO3 and 0.1 M NaOH to adjust the pH. All experiments were conducted in a temperature-controlled room at 20 °C, and the solution was kept under a nitrogen atmosphere. The ISE measurements were performed according to Milne et al.11 We used a double-junction saturated KCl reference electrode. The outer junction of the reference electrode was filled with a mixture of 0.125 M NaNO3 and 0.875 M KNO3. The pH electrode was a standard glass half-cell that was calibrated using two buffer solutions of pH 4 and 7. The Cd electrode (Metrohm 6.0502.110) was calibrated by titrating a Cd solution in 0.03 M NaNO3 with ethylenediamine.11 Emf readings produced calibration curves that were linear for p[Cd2+] between 11 and 3 (slope ) -29.94 ( 0.12 mV/p[Cd2+] unit; intercept ) -60.66 ( 1.31; R2 ) 0.99983-0.99996; n ) 5). Donnan membrane experiments were conducted in a system according to Figure 1. Details of the DMT equipment are presented by Temminghoff et al.18 The two sides of the Donnan cells were separated by a negatively charged cation exchange membrane (BDH, Catalog no. 55165 2U). On the basis of work done by Weng et al.,19 we calculated that in our system, up to 12% of the free metal ions in the system were bound by the membrane, but the free ion concentration was only a small fraction of the total metal in experiments with a high DOM concentration. The loss of metal ions to the membrane was compensated by dissociation of metal complexes. The ratio Me2+:MeNOM is normally far below 1:10 except for extremely high metal concentrations. The error as a result of metal binding by the membrane is generally less than 1% of the total metal present in the system. It is important to note that a loss of metal ions decreases the total dissolved metal concentration in the system, but it need not to be quantified to calculate the ratio between the free and complexed metal concentration so long as both acceptor and donor solution are analyzed (see Figure 2). (28) Kinniburgh, D. G.; Milne, C. J.; Venema, P. Soil Sci. Soc. Am. J. 1995, 59, 417-422.

Figure 2. Schematic representation of a DMT cell.

The DMT cells and the membranes were washed in 0.1 M HNO3 and rinsed with ultrahigh purity water (UPW). UPW was produced by using the Elgastat maxima-HPLC unit (Elga, Bucks, U.K.) to further reduce impurities, such as carbon, and inorganic elements in demineralized water. The membranes were then saturated with 1 M Ca(NO3)2 for 2 h, and washed three times for 1 h with acceptor solution to reach a faster equilibrium with the acceptor solution. The pretreatment procedure prevents cation exchange in the membranes which strongly influences the solution composition. The donor side of the cell was connected to the reaction vessel that was also used for the ISE experiments. The donor side of the cell was kept under nitrogen to prevent formation of carbonate complexes in the solution at higher pH values. The acceptor side was connected to a 10-mL ICPMS test tube (resulting in an acceptor volume of 17 mL, ∼10% of the total volume in the system). A CO2 free acceptor solution requires extra experimental efforts. The pHacceptor at pHdonor ) 8 will decrease because of CO2 diffusion if it is not kept under nitrogen atmosphere. If a difference in pH between both sides occurs, the donor and acceptor strive toward equilibrium in proton activity by transporting protons from the acceptor to the donor solution. These two processes might result in a different pH in the acceptor, but the very low proton concentrations at near-neutral pH do not significantly contribute to the ionic strength. Although there is not necessarily Donnan equilibrium for protons over the membrane, this will not affect the Donnan equilibrium for the metal ions. The advantage of the rather low pH in the acceptor is that metal carbonate complexes might be insignificant. We studied the pH in the acceptor at pHdonor ) 8 and evaluated whether it was necessary to keep the acceptor side under nitrogen atmosphere. After 24 h, a 0.5-mL sample was taken from the donor solution, and the 10-mL test tube containing acceptor solution was changed. The pH was measured immediately in both donor and acceptor solution (combined pH electrode, pHC2005-7, Radiometer). After a 10-fold dilution of the donor, a sample was taken to measure dissolved organic carbon (DOC). The DOC concentration in the acceptor solution was always below the detection limit of 0.5 mg/ L. Na, Ca, and Cd or Zn were analyzed in all samples that were determined by ICP-OES (Spectro Analytical Instruments). If concentrations were below the ICP-OES detection limit, samples were measured again by ICPMS (Perkin-Elmer, Elan 6000). Figure 2 schematically presents the situation in the DMT cells. At equilibrium, the concentration of the free metal ion is equal in the donor and acceptor in the case that there is no difference in ionic strength between donor and acceptor. Figure 3 shows a stepwise calculation scheme, including a correction equation, if salt differences exist between donor and acceptor. The total concentrations (Na, Ca, and Cd or Zn) in the acceptor solution

Figure 3. Scheme to calculate the free and complexed metal (Me) concentration in a DMT cell using measured total concentration in the donor and acceptor and a speciation program (e.g., ECOSAT) to calculate the contribution of inorganic complexes.

were converted into free ion activities using the ECOSAT computer code accounting for inorganic complexes, such as ZnOH+(aq).29 Differences in salt levels over the membrane, which might occur during the experiment, were corrected using the distribution ratio of Na in the donor and acceptor solution. This resulted in the free ion activity in the donor solution. Next, the inorganic species in the donor solution were calculated. Finally, the DOM-bound metal concentration in the Donnan solution was calculated from the total metal concentration, the free metal concentration, and the concentration of inorganic complexes. Effect of the Background Electrolyte in Donnan Membrane Cells. This preliminary experiment had a very simple design without using DOM. A 0.5-L portion of a solution containing 1.78 µM Cd(NO3)2 in a background electrolyte of 3 mM NaNO3 or 1 mM Ca(NO3)2 was connected to the donor side. The acceptor solution (3.3% of the total volume) contained the same concentration of background electrolyte but no Cd. The donor and the acceptor were sampled every 24 h for 7 days. Titration Experiments. We used FSHA as dissolved organic matter (consisting of 50% carbon). The material was extracted from forest floor material. The extraction and purification procedure are described elsewhere.30 All metal titration experiments (measured by ISE or DMT) had an initial donor solution volume of 150 mL containing 1.67 g L-1 FSHA. The FSHA concentration (29) Keizer, M. G.; Van Riemsdijk, W. H. A Computer Program for the Equilibrium Calculation of Speciation and Transport in Soil-Water Systems (ECOSAT 4.7); Wageningen University: The Netherlands, 1999. (30) Temminghoff, E. J. M.; Van der Zee, S. E. A. T. M.; De Haan, F. A. M. Environ. Sci. Technol. 1997, 31, 1109-1115.

Analytical Chemistry, Vol. 74, No. 4, February 15, 2002

859

Table 1. Generic NICA-Donnan Parameters for Purified Humic Acids22,23 parameters

site 1

b (eq 4) Qmax p log K ˜H nH log K ˜ Ca nCa log K ˜ Cd nCd log K ˜ Zn nZn

3.15 0.62 2.93 0.81 -1.37 0.78 -0.20 0.73 0.11 0.67

site 2 0.49 2.55 0.41 8.00 0.63 -0.43 0.75 2.37 0.54 2.39 0.27

was much higher than DOM concentrations commonly found in nature. The total metal concentrations were also higher than in most natural systems, ranging from 1.7 × 10-6 to 1.7 × 10-3 M. As a result, the free Cd and Zn concentrations normally found in nature were within the range measured in our experiments. We chose to use higher DOM and metal concentrations, because it improved the analytical accuracy and reliability of the measurements, which is important if the data are to be used to determine model parameters. The equilibrium between bound and free metal concentration is not affected by the higher DOM concentration. The DOM concentration we used is a common value in studies determining binding constants for metal-DOM complexation.8,11 The donor solution was adjusted to a specified pH and kept at this pH by adding 0.1 M HNO3 or 0.1 M NaOH. The Cd or Zn concentration was increased stepwise after each measurement. First, a Cd titration (p[Cd2+] ) 3-9) was conducted in 0.03 M NaNO3 at pH 4 and 8. [Cd2+] was measured by an ISE. Second, a similar experiment was performed at pH 4, 6, and 8, but the background electrolyte contained a mixture of 0.03 M NaNO3 and 0.001 M Ca(NO3)2. Third, the second experiment was repeated, but measured with the DMT (Figure 1) instead of ISE. The Donnan membrane technique requires 24 h per measurement. For this reason, the number of data points of the DMT is less than the data obtained by ISE. Fourth, Zn binding to FSHA was determined by DMT, since an ISE is not available for Zn (p[Zn2+] ) 3-8). The experimental conditions were similar to the Cd experiment, except for the NaNO3 concentration at pH 4 and 6, which was 0.003 M. Modeling. The parameter determination of complex multicomponent models, such as the NICA-Donnan model, requires that a number of decisions be made. Milne et al.22,23 derived two sets of generic parameters. One was based on a large number of literature data for humic acids and another on data for fulvic acids. We used purified forest soil humic acid in our experiments and, therefore, used the generic parameters for humic acids (Table 1). Apart from our data, which were also included in this database, 7 data sets (494 data points) were used for Cd, whereas only one other set (15 data points) was used for Zn. We predicted the Cd or Zn bound to FSHA using the NICA-Donnan model with generic parameters. The measured free ionic concentrations of Na, Ca, H, Cd, Zn, and the FSHA concentration were used as input variables. Model results could be compared with the measured concentrations. 860 Analytical Chemistry, Vol. 74, No. 4, February 15, 2002

Figure 4. [Cd2+] in the donor and acceptor side as a function of the background electrolyte during 7 days. Closed symbols represent measurement in the donor solution; open symbols represent measurement in the acceptor solution. Circles, in the presence of 0.001 M Ca(NO3)2; diamonds, in the presence of 0.003 M NaNO3.

RESULTS AND DISCUSSION Further validation of the DMT comprised three aspects: (1) effect of the background electrolyte, (2) comparison with ISE measurement, and (3) measurement at high pH. Subsequently, we will pay attention to modeling aspects of Cd binding by humic acid. Finally, we will present experimental data and modeling results of Zn binding to FSHA. Analytical Aspects of the DMT: (1) Background Electrolyte. Metal-binding data should ideally be measured in the absence of strongly competing ions, but the competition of protons is inevitable. In nature, other (mono-, di-, or trivalent) ions will play a role. We tested 0.003 M NaNO3 and 0.001 M Ca(NO3)2 as background electrolytes. The results are presented in Figure 4. If no Ca is present in the system, Cd does not pass the membrane during the time span of the experiment. This can be explained by the binding capacity of the cation-exchange membrane. The negatively charged membrane can bind positively charged ions and prefers divalent to monovalent ions.19 In 0.003 M NaNO3 and for the chosen donor volume and Cd concentration the membrane binds all Cd that is initially present in the donor solution. When the background electrolyte contains Ca, this divalent ion acts as a competitor, reducing Cd binding to the membrane. The Cd concentration in the donor solution in the presence of Ca nevertheless decreased somewhat during the first day (Figure 4) as a result of the binding of some Cd in the membrane and a small loss of Cd to the acceptor solution. A loss of Cd to the membrane and acceptor solution results in a concentration drop in both donor and acceptor. On the basis of the results of the first experiment, we decided to use a mixture of 0.03 M NaNO3 and 0.001 M Ca(NO3)2 as the background solution in DMT experiments. NaNO3 defines the ionic strength, and Ca(NO3)2 ensures a sufficient metal transport rate across the membrane in the other experiments. Although the presence of Ca approaches the natural environment more, it makes our system more complex, since Ca will affect the metal binding by organic matter. The data points in Figure 5 show the effect of Ca in the background electrolyte, as measured by ISE. Clear differences in Cd adsorption can be observed when Ca is present or absent, particularly at pH 8. The same trend in experimental data was observed by Kinniburgh et al.9 At high pH,

Figure 5. Cd bound to humic acid versus free Cd (log/log) with and without Ca in the background electrolyte as measured by ISE. Closed symbols and solid lines represent data and predictions in 0.03M NaNO3 and 0.001 M Ca(NO3)2; open symbols and dotted lines are data and predictions in 0.03 M NaNO3.

Figure 6. Free Cd versus bound Cd (log/log) as measured by DMT and ISE. [, pH 4, DMT; 9, pH 6, DMT; 2, pH 8, DMT; +, pH 4, ISE; O, pH 6, ISE; and ×, pH 8, ISE.

the surface charge of organic matter is more negative. Part of the negative charge is compensated by counterions in the Donnan volume. The role of the Donnan volume is greater at higher pH values. Ca is the most important ion in the Donnan volume. As a result, the effect of Ca will be greater at higher pH values. (2) Comparison with ISE. Figure 6 shows the Cd data measured by ISE and DMT. The comparison between the two analytical techniques is extremely good. Slight differences can be observed only at high Cd concentrations at pH 4. Under these conditions, most of the Cd is present as a free ion. A small error in the measurement of the free concentration may result in a relatively large error in the adsorbed concentration because of the mass balance calculation to obtain MeNOM (Figure 3). As a remark, we would like to report that three Cd-selective electrodes failed during use. One of them was newly purchased. A fourth electrode (of a different brand) worked properly during all of the experiments. Milne et al.11 also reported the failure of two of their electrodes. Apparently, a Cd-ISE is a very sensitive instrument. (3) Measurements at High pH. Our measurements at pH 8 could be performed without a nitrogen atmosphere in the acceptor side. The pH in the acceptor solution was ∼6.5 at pHdonor

) 8. Apparently, the increase in protons by CO2 diffusion and the decrease in protons by transport to the donor side (details in Experimental Procedures) compensate each other, creating a steady-state pH. Carbonate complexes are negligible at pH 6.5, and the donor is kept under nitrogen atmosphere, but a difference in pH between donor and acceptor requires pH measurement in both donor and acceptor. If the pH in the donor is assumed to be equal to the pH in the acceptor, this will lead to an erroneous calculation of free metal ions because of the correction for carbonate and hydrolysis complexes in the acceptor, which are not present in reality. The higher volume of the acceptor, compared to original design20,21 is an advantage of the DMT cells. Measurement of pH, but also multielement analysis are easier if a more sample solution is available. Modeling Cd Binding by FSHA. The competition between Ca and Cd complicates the determination of ion specific model parameters. The experimental data in Figure 5 were calculated by the NICA-Donnan model using the generic model constants (Table 1) for humic acid, which have been determined recently.22,23 The free Ca, Na, H, and Cd concentrations, as measured by DMT, and the FSHA concentration were used as the input variables to calculate the amounts of Cd bound to FSHA. The average effect of Ca as predicted by the model using generic parameters is quite reasonable. The modeling results (lines in Figure 5) also show a larger effect of Ca at pH 8. Generally, the effect of Ca at high pH is underestimated, especially at low Cd concentrations. The predictions could be improved if specific parameters for this humic acid were used. Concentrations of both Cd and Ca were measured. As in Figure 5, we can compare the measured free Ca with the Ca bound to FSHA. The Ca bound to FSHA is slightly overestimated by the model (data not shown), but the trends were predicted well. The predicted log Cabound is, on average, 0.3 log units higher, as compared to the measured value. Both the influence of Ca competition on Cd binding and the comparison of measurements and predictions of Ca bound by FSHA indicate that the effect of Ca is reasonably taken into account. The lines in Figure 7a represent the model predictions as calculated by the NICA-Donnan model using the generic parameters presented in Table 1. The model is very capable of describing the Cd data measured in competition with Ca without adjusting any parameter (RMSE ) 0.210; R2 ) 0.971). The Cd bound to FSHA was slightly overestimated, particularly at low Cd concentrations. Zinc Binding by FSHA Measured by the DMT and Modeled with NICA-Donnan. Zinc binding by FSHA is presented in Figure 7b. The trends are similar to Cd: lower binding at lower pH values. Zn is bound slightly more strongly by FSHA, as compared to Cd (Figure 7a). The detection limit for Zn (ca. 2 × 10-8 M) is higher than that for Cd (ca. 3 × 10-9 M). The detection limit for Zn is mainly determined by contamination in the laboratory, whereas laboratory contamination of Cd is generally no problem. The calculation of Zn bound by FSHA is represented by the lines in Figure 7b. The model predicts the Zn data even better than Cd (RMSE ) 0.090; R2 ) 0.988), but this is no surprise, since our data formed the major part of the Zn database used by Milne et al.,22,23 which is the basis for the generic parameters. Analytical Chemistry, Vol. 74, No. 4, February 15, 2002

861

Figure 7. Cd (a) and Zn (b) bound to humic acid versus free Cd (a) and Zn (b) (log/log) at 3 different pH values in a mixed background electrolyte of 0.03 M NaNO3 and 0.001 M Ca(NO3)2: [, pH 4; 9, pH 6; and 2, pH 8. Lines represent model calculations: solid line, pH 4; dashed line, pH 6; and dotted line, pH 8.

CONCLUSIONS The DMT is an excellent method to determine free metal ion concentrations in complex solutions, without any interference of labile complexes. The results for Zn illustrate that this technique provides possibilities for measuring free ion concentrations if no ISE is available. Sufficient time for equilibration is required, and an abundant divalent ion, such as Ca, is needed as a background electrolyte in the case that divalent metal ions are to be measured at low concentrations. The method provides reliable results over a large metal and pH range and is suitable for multielement analysis.

Climate. We thank Chris Milne (BGS, Wallingford, U.K) for providing the generic NICA-Donnan parameters, Rein van Eck for performing part of the experimental work, and Jeroen Filius for his valuable comments on the manuscript. We also thank the reviewers for their comments to improve the paper.

ACKNOWLEDGMENT This research was supported by the EU project FAMEST (ENVU-CT97-0554) via the program DG XII Environment and

Received for review May 3, 2001. Accepted November 9, 2001.

862 Analytical Chemistry, Vol. 74, No. 4, February 15, 2002

SUPPORTING INFORMATION AVAILABLE The data presented in Figure 7 (Cd and Zn binding by FSHA) are available as Supporting Information. This material is available free of charge via the Internet at http://pubs.acs.org.

AC0105080