Environ. Sci. Technol. 2007, 41, 8104–8110
Comparing WHAM 6 and MINEQL+ 4.5 for the Chemical Speciation of Cu2+ in the Rhizosphere of Forest Soils BENOÎT CLOUTIER-HURTEAU,† SÉBASTIEN SAUVÉ,‡ AND F R A N Ç O I S C O U R C H E S N E * ,† Departments of Geography and of Chemistry, Université de Montréal, P.O. Box 6128, succursale Centre-Ville, Montréal, Québec, Canada H3C 3J7
Received April 10, 2007. Revised manuscript received August 28, 2007. Accepted September 12, 2007.
Metal speciation data calculated by modeling could give useful information regarding the fate of metals in the rhizospheric environment. However, no comparative study has evaluated the relative accuracy of speciation models in this microenvironment. Consequently, the present study evaluates the reliability of free Cu ion (Cu2+) activity modeled by WHAM 6 and MINEQL+ 4.5 for 18 bulk and 18 rhizospheric soil samples collected in two Canadian forested areas located near industrial facilities. The modeling of Cu speciation was performed on water extracts using pH, dissolved organic carbon (DOC), major ions, and total dissolved Al, Ca, Cu, Mg, and Zn concentrations as input data. Four scenarios representing the composition of dissolved organic substances using fulvic, humic, and acetic acids were derived from the literature and used in the modeling exercise. Different scenarios were used to contrast soil components (rhizosphere vs bulk) and soil pH levels (acidic vs neutral to alkaline). Reference Cu2+ activity values measured by an ion-selective electrode varied between 0.39 and 41 nM. The model MINEQL+ 4.5 provided good predictions of Cu2+ activities [root-mean-square residual (RMSR) ) 0.37], while predictions from WHAM 6 were poor (RMSR ) 1.74) because they overestimated Cu complexation with DOC. Modeling with WHAM 6 could be improved by adjusting the proportion of inert DOC and the composition of DOC (RMSR ) 0.94), but it remained weaker than predictions with MINEQL+ 4.5. These results suggested that the discrepancies between speciation models were attributed to differences in the binding capacity of humic substances with Cu, where WHAM 6 appeared to be too aggressive. Therefore, we concluded that chemical interactions occurring between Cu and DOC were key factors for an accurate simulation of Cu speciation, especially in rhizospheric forest soils, where high variation of the DOC concentration and composition are observed.
Introduction It is widely recognized that the environmental impact of metals is not dependent on the total soil metals but mediated * Corresponding author phone: 514-343-8027; fax: 514-343-8008; e-mail:
[email protected]. † Department of Geography. ‡ Department of Chemistry. 8104
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 23, 2007
through dissolved metal species (1, 2). In that context, studying the chemical speciation of metals in a soil solution helps us to understand their potential environmental impacts. This concern is especially true for rhizosphere (Rz) environments that control the transfer of elements from soil to plant. Unfortunately, because many metal species are difficult to measure, the main tool used to study the speciation of metals in a soil solution is computer modeling. Among the speciation models available, WHAM 6 and MINEQL+ 4.5 are some of the most cited, they are well established, and they integrate the metal complexation capacity of dissolved organic matter differently. Indeed, WHAM 6 is a discrete site/electrostatic model that simulates the interactions of metals with humic substances using sites presenting two types of affinities for cations (types A and B) and three types of binding capacities (monodentate, bidentate, and tridentate) (3, 4). Conversely, MINEQL+ 4.5 is a classical chemical equilibrium model that integrates all inorganic and organic ligands, mass balance, and thermodynamic reactions to model chemical speciation (5, 6). Unfortunately, few studies have evaluated the reliability of these speciation models in soils and even more so in rhizospheric soils. Because the free ionic form of Cu (Cu2+) can be easily measured, Cu seems to be an ideal metal candidate to compare the reliability of these two speciation models. Therefore, the aim of this study was to evaluate the performance of modeling the Cu2+ activity in bulk (Bk) and in Rz forest soils by MINEQL+ 4.5 and WHAM 6.
Experimental Section Sampling and Analyses. Forest soils were sampled in two Canadian areas covering a range of chemical conditions (soil pH and Cu concentrations) but that were similar in terms of topography, vegetation, soil order, texture, and mineralogy. Soils were collected on three sites in the Rouyn-Noranda area (48°,14′N and 78°,58′W) in August 2004, at 0.5, 2, and 8 km downwind from a Cu smelter (identified respectively as RN 1–RN 3) and on three sites in the Monteregian area in August 2005, near a cement plant (45°,16′N and 73°,28′W), a steel-work (45°,51′N and 73°,13′W), and an airport (45°,31′N and 73°,23′W) (identified respectively as MTG 1–MTG 3). At each of the six sites, the Rz and Bk soil components were sampled under three trembling aspens (Populus tremuloides Michx.) from the B horizons of Luvisol profiles, as described by Séguin et al. (7). Briefly, the trees were uprooted, and the fine roots ( 6.00 5.03 40 CuFA22pH < 6.00 3.95 41 pH > 6.00 4.03 41 ZnFA pH < 6.00 4.78 42 pH > 6.00 4.50 42 Metals Complexed with Humic Acid (HA)a AlHA+ 6.62 39, 43, 44c CaHA 3.60 45 Ca2HA2+ 6.70 45 CuHA 9.70 46 CuHA225.40 47 MgHA 5.50 25 ZnHA 4.93 48 Metals Complexed with Acetic Acid (AA)b AlAA2+ 2.02 49 AlAA2+ 3.50 49 CaAA+ 1.18 23 CuAA+ 2.61 50 CuAA2 3.40 23 CuAA33.94 23 2+ CuHAA 4.59 50 HAA 4.76 23 KAA -0.19 23 MgAA+ 1.27 23 NaAA -0.18 23 ZnAA+ 1.58 23 ZnAA2 2.64 23 a Conditional association constants integrated in MINEQL+ 4.5 model. b Conditional association constants integrated in MINEQL+ 4.5 and WHAM 6 models. c Based on interpolation made from references.
Results and Discussion Properties of Water Extracts. The properties of water extracts and of soil components are presented for each site in Tables 3 and 4. The laboratory replicates of these properties were used, for each soil component and under each tree, as input data in the speciation models. The soil solution pH spanned up to 2 orders of magnitude between sampling sites, with sites of RN and MTG 3 considered as acidic and sites MTG 1 and 2 having near-neutral conditions (Table 3). Higher concentrations of water-soluble Al, Cu, Zn, K+, and NH4+ were observed at the RN sites (Tables 3 and 4) and were attributed to higher anthropogenic deposition in this area and to more acidic soil pH levels. On the other hand, higher values of EC, Cl–, F–, Na+, NO3–, Ca, and Mg were measured at the MTG sites (Tables 3 and 4). They are linked to the mineralogy and texture of the parent material in this area. For DOC and SO42– concentrations, no clear patterns were noted. The water extracts of the Rz soil were enriched in DOC, Al, Cu, Zn, Cl–, Na+, K+, Ca, and Mg, but were depleted in NO3– as compared to Bk soils (Tables 3 and 4). No trend existed for SO42– and NH4+. These observations generally agreed well with previous studies (27–30) except for some studies conducted under controlled environments (31). For example, in our Rz samples, acidification was observed in acidic soils and alkalinization was recorded in near-neutral 8106
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 23, 2007
soils (Table 3). This pattern differed from the results of other studies (32) and could be attributed to the adaptation of tree specie to pH-contrasted environments. Finally, large variations in the pH values and water-soluble Cu concentrations between sampling sites appeared as the main factors responsible for the wide range of measured Cu2+ activities (0.39–41 nM) reported in Table 5. These activities were higher in the Rz soils compared to the Bk soils and were comparable to previous determinations in forest soils (29, 33). Yet, these contrasts in absolute Cu2+ activities differed from the results of Legrand et al. (29), where no trend was noted between soil components. They also found that the fraction of water-soluble Cu present as Cu2+ was systematically higher in Bk soils as compared to the Rz component (29). The differences in tree species, soil orders, and soil textures between studies probably explain these discrepancies. Estimation of Cu2+ Activity Using MINEQL+ 4.5. The Cu2+ activities measured by an ISE and the values calculated with MINEQL+ 4.5 generally differed by less than 0.5 order of magnitude, except for some soils in scenarios 2–4, where the disparity reached up to 0.75 order of magnitude (Figure 1). MINEQL+ 4.5 provided excellent estimates of Cu2+ activities as demonstrated by the low RMSR value (RMSR ) 0.37). Despite its good prediction of Cu2+ activities, we noted that MINEQL+ slightly overestimated the measured Cu2+ activities for all DOC scenarios. Many factors can be invoked to explain these discrepancies. One of them is the error associated with the free-ion measurements, which could be expected to be larger at lower Cu2+ activities. All of our measurements were, however, well above the detection limits of the method estimated at 10-14 M (8). Moreover, because the errors between measured and estimated Cu2+ activities were constant for the whole pCu2+ range, this source of discrepancy was viewed as minor for our data set. The disparities may also be associated with the scenarios developed to represent the DOC composition or with the conditional stability constants used to describe the complexation of metals by organic substances (Tables 1 and 2). However, in both cases the data used in the models were gathered from the most recent publications on these topics. It remains that the literature on DOC composition in the Rz soil is sparse, especially for forest soils and for a wide range of pH levels. We submit that our limited knowledge of the composition of DOC in the Rz soil could contribute to the explanation of the slight but systematic discrepancies observed between measured pCu2+ values and those modeled with MINEQL+ (Figure 1). In contrast, the selection of complexation constants for metals was based on an abundant and critically evaluated literature. We thus assumed that the errors due to these complexation constant values were relatively minor compared to those associated with the scenarios for DOC composition. Estimation of Cu2+ Activity Using WHAM 6. In contrast to MINEQL+ 4.5, WHAM 6 did not provide good estimates of Cu2+ activities (RMSR ) 1.74; Figure 2). For most samples, WHAM 6 underestimated the activity of Cu2+ in water extracts when compared to measured values, mostly because it overestimated the complexation of Cu by organic ligands (Table 5). This observation suggests that predicted interactions of metals with humic substances in WHAM 6 are too aggressive as compared to measured data, mostly for nearneutral soils and for Bk acidic soils. The role of competition between metals for organic ligands could partly help explain this observation. Recent simulations performed with WHAM 6 revealed that the competition between Al and Cu for DOC binding sites was highest at pH values below 5.5, when Al concentrations increased markedly (34). Our results agreed with this finding and showed that Al competition at pH values below 5.5 had increased the proportion of Cu2+ activities in
TABLE 3. Chemical Variables and Concentrations of Dissolved Metals in Water Extracts for the Bk and Rz Soils of the Six Sampling Sites DOCc (mg L-1)
pHb site
a
d
RN 1 RN 2 RN 3
MTG 1 MTG 2 MTG 3
d
d
Alwsc (µmol L-1)
ECc (µS cm–1)
d
d
d
d
SD
Cuwsc (µmol L-1)
mean
SD
mean
SD
meand
SDd
Area 40.0 47.4 31.2 44.9 36.2 58.8
8.8 35.0 7.1 17.9 21.9 16.9
0.36 0.86 0.85 1.63 0.29 0.56
0.11 0.34 0.63 0.47 0.04 0.03
0.70 1.18 0.31 0.56 0.08 0.15
0.08 0.09 0.24 0.27 0.04 0.01
2.8 9.0 2.96 15.5 4.2 6.7
0.11 0.16 0.08 0.13 0.04 0.08
0.03 0.05 0.01 0.04 0.01 0.02
0.12 0.07 0.05 0.09 0.07 0.09
0.09 0.09 0.01 0.06 0.02 0.02
mean
SD
mean
SD
mean
Bk Rz Bk Rz Bk Rz
5.01 4.98 5.40 5.34 5.44 5.35
0.04 0.02 0.09 0.12 0.16 0.08
26.8 78.1 48.9 75.7 64.1 82.1
4.5 15.0 23.5 3.5 3.0 4.8
Rouyn-Noranda 20.1 3.0 33.6 1.5 12.3 4.4 25.6 3.6 11.9 1.6 23.7 3.3
Bk Rz Bk Rz Bk Rz
6.79 6.93 7.03 7.06 5.54 5.43
0.15 0.21 0.11 0.18 0.22 0.16
60.1 90.2 44.7 55.2 51.0 91.8
19.3 36.2 4.1 4.9 8.5 7.0
Monteregian Area 66.3 28.6 11.8 98.7 33.9 22.0 61.4 6.6 8.51 87.9 14.6 15.3 43.8 7.0 23.4 56.6 7.0 29.6
d
Znwsc (µmol L-1)
d
d
a See text for full descriptions of site codes. b Soil solution pH measured in a 1:10 soil–water extract. c Dissolved organic carbon, electrical conductivity, and total concentrations of water-soluble Al, Cu, and Zn. d Mean and standard deviations of field triplicates.
TABLE 4. Cations and Anions Concentrations (µmol L–1) in the Water Extracts of the Bk and Rz Soils of the Six Sampling Sites Cl– a
c
site
RN 1 RN 2 RN 3
M
Bk Rz Bk Rz Bk Rz
MTG 1 Bk Rz MTG 2 Bk Rz MTG 3 Bk Rz
10.2 28.7 6.61 22.9 8.41 13.9
32.2 67.2 27.1 73.9 26.3 68.5
NO3–
F– c
SD
2.2 4.7 3.08 7.2 2.41 2.6
21.4 15.1 10.7 17.4 14.8 21.9
c
M
