| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Division of Ecosystem Sciences, Department of Environmental Science, Policy, and Management, 151 Hilgard Hall, Univ. of California, Berkeley, CA 94720-3110
* Corresponding author (doner{at}nature.berkeley.edu)
Received for publication December 27, 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Arsenic is found in -3, 0, +3, and +5 oxidation states in nature and is most commonly present as As(III) in reducing environments (at <100 mV) and As(V) in oxic environments. The most common inorganic forms are arsenite (H3AsO3) and arsenate (H2AsO-4 or HAsO2-4). Various organic forms of As also occur, primarily methylated species formed through microbial processes. Arsenite is generally thought to be the more mobile and toxic of the inorganic species, and mobilization of As under reducing conditions has been observed by numerous researchers (Amrhein et al., 1993; Masscheleyn et al., 1991; McGeehan and Naylor, 1994). McGeehan and Naylor (1994) and Masscheleyn et al. (1991) emphasized the importance of both lower adsorption of arsenite and the dissolution of adsorbing phases such as iron oxides for the higher solubility of As under reducing conditions. However, other researchers have shown that arsenite is more strongly adsorbed on iron oxides than arsenate at higher pH values (pH > 6) (Jain and Loeppert, 2000; Sun and Doner, 1998).
Vanadium may exist in +2, +3, +4, and +5 valence states. V(III) is generally only present in extreme reducing environments, while V(IV) and V(V) species are dominant under moderately reducing and aerobic conditions, respectively. In general, the solubility of V increases as V is oxidized from V(III) to V(V). For instance, V(III) is commonly precipitated as an oxide or oxyhydroxide, V(IV) is commonly found as a vanadyl cation [VO2+, VO(OH)+] and V(V) is most often present as a vanadate oxyanion 
(Wanty and Goldhaber, 1992). Vanadyl is strongly sorbed by solid phases, including organic and oxide–oxyhydroxide phases (Wanty and Goldhaber, 1992; Wehrli and Stumm, 1989). Adsorption of the V anions is much lower than the cations; however, VO2+ solubility may be greatly increased through complexation with organic matter (Wanty and Goldhaber, 1992; Wehrli and Stumm, 1989). While V(IV) is not thermodynamically stable above pH 7, complexation by various organic and inorganic species may greatly increase its stability (Wanty and Goldhaber, 1992; Wehrli and Stumm, 1989). In a sediment incubation study, Amrhein et al. (1993) found that under reducing conditions V concentrations slowly decreased in the solution phase; when samples were exposed to oxidizing conditions, V concentrations in the solution phase dropped to essentially 0. They attributed the loss of V from solution under reducing conditions to the precipitation of VO(OH)2.
Molybdenum is generally found in two oxidation states in nature, Mo(IV) and Mo(VI). The reduced state may be present as MoS2, although the formation of MoS2 may be kinetically limited (Arutyunyan and Khurshudyan, 1966; Helz et al., 1996). Under oxidizing conditions, Mo(VI) dominates, and it is commonly present as the oxyanion molybdate 
. Amrhein et al. (1993) found that Mo was lost from solution under reducing conditions and resolubilized under oxidizing conditions. They hypothesized that MoS2 formed in their system. Helz et al. (1996) have proposed that Mo is incorporated into the solid phase of sediments under reducing conditions via sulfur bridges (formation of thiomolybdate) and listed both iron and aluminum phases and organic matter as important sinks for Mo.
This study was undertaken in order to gain a greater understanding of the fate of the trace elements As, Mo, and V in the sediments of a constructed wetland. The adsorption behavior of these elements on many important soil components has been studied in the laboratory, generally under aerobic conditions (Bibak and Borggaard, 1994; Doner and Zavarin, 1997; Goldberg et al., 1996; Mikkonen and Tummavuori, 1994; Zhang and Sparks, 1989). Sequential extraction procedures have also been used in order to fractionate soil into various components and their corresponding sorbed trace elements; however, these procedures suffer from problems of nonspecificity, and fractions must be operationally defined.
In this experiment a method was used that involves placing a pure mineral or known soil into a permeable bag, which is then inserted into the flooded soil of a constructed wetland for a specified time period. This allows us to assess changes in trace element concentrations for each mineral or soil under field conditions. Similar methods have been used by others to determine geochemical or soil-forming processes (Ranger et al., 1991) as well as chemical responses of soil to experimental acidification in forest soils (David et al., 1990; Mitchell et al., 1994). In this study we chose goethite-coated sand and calcite as our model minerals. Goethite has been shown to sorb large quantities of trace elements, including As, Mo, and V (Bibak and Borggaard, 1994; Goldberg et al., 1996; Sun and Doner, 1998). Calcite is an important component of this calcareous soil, and carbonates have been shown to play an important role in trace element cycling at alkaline pH values (Goldberg and Glaubig, 1988a, b; Zavarin, 1999). Many carbonates are enriched in both cation and anion (e.g., selenite) trace elements (Doner and Zavarin, 1997).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
Half of all samples were removed after about 3 mo (102 d, September 1998) and the remaining samples were removed after about 1 yr (348 d, June 1999). The samples were stored on dry ice for transport back to the laboratory where they were air-dried, ground to less than 2 mm, and homogenized prior to sequentially extracting 1-g subsamples into two or three fractions. Three subsamples from each bag were first extracted with 10 mL of deionized water for 4 h. The soil and goethite-coated sand samples were then extracted with 10 mL of 0.1 M K2HPO4–KH2PO4 at pH 8 to remove specifically adsorbed oxyanions. The final extraction for the goethite-coated sand samples consisted of the addition of 20 mL of 6 M HCl, mixed overnight, followed by the addition of 10 mL of 6 M HCl heated to 55°C for 2 h. The soil was digested with a HNO3–H2SO4–ammonium oxalate digestion in Teflon beakers, as described by Huang and Fugii (1996). The calcite samples were fully dissolved in 10 mL of 3 M HCl. All samples were then analyzed by inductively coupled plasma (ICP–AES, Thermo Jarrell Ash [Franklin, MA] IRIS) for As, Mo, and V.
The concentrations for the three subsamples were averaged; variation between subsamples was less than 10%. Student's t tests were performed to determine changes in trace element concentrations on the minerals or soil over time, as well as to compare concentrations in Cells 4 and 5. In order to determine whether the mineral bags accumulated greater levels of trace elements at the sediment surface or buried 2.5 cm below the surface, a paired t test was performed for each cell. For this test, the difference between the buried and surface concentrations at each location within the cell (i.e., for each grid section) was calculated and t tests were performed on the differences. The paired t test allows one to determine the effect of depth on trace element concentration, which may otherwise be masked by high spatial variability within each cell. For all analyses, the probability level was set to 0.05 unless otherwise noted. All values reported in the text and on graphs are reported as average ± standard error. All of the data, including concentrations that were below the detection limit, were used in statistical analyses in order to prevent bias associated with omitting or setting such concentrations to zero. However, differences were not considered significant if more than a few samples fell below the detection limit.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
 | [1] |
According to a model proposed by Helz et al. (1996), Mo may form covalent bonds to iron through sulfur bridges (e.g., –Fe–S–Mo–) in sulfidic waters. Only sulfate reduction is necessary for this mechanism of Mo accumulation. The following reaction illustrates the formation of tetrathiomolybdate by sulfate reduction:
 | [2] |
Thiomolybdate can readily bond to Fe-containing minerals. Thus, at lower redox potential and pH, H2S availability is higher and greater Mo accumulation on goethite-coated sand may result. In addition, Erickson and Helz (2000) noted that the conversion of MoO2-4 to MoS2-4 is acid-catalyzed. Factors such as vegetation type, total organic matter content, and microbial activity may all influence pH in the field. Although our data does not provide direct evidence for distinguishing between the two mechanisms of Mo accumulation, Helz et al. (1996) indicated that in their study, the thiomolybdate pathway was kinetically more favorable than the formation of MoS2. The pH of the water extracts was lower than field pH values and the pH in water extracts of soil mineral bags, particularly after 12 mo. This was most likely due to oxidation of iron sulfides (e.g., pyrite, FeS2) that may have formed. This point will be discussed further below.
The pore size of the nylon mesh bags (25 µm) was large enough to allow the entrance of various microorganisms as well as organic and inorganic particles < 25 µm. While a fraction of the total accumulated Mo may be associated with these materials, evidence from the calcite mineral bags suggests that this "passive accumulation" was not a large source of additional Mo.
As shown in Table 3, significant accumulations of As occurred in Cells 4 and 5. In Cell 4, significantly more As was present at 12 mo compared with 3 mo; however, in Cell 5 the concentrations at 3 and 12 mo were not significantly different from one another. Similar levels of As were found on samples from Cells 4 and 5, indicating that accumulation of As on goethite-coated sand is not as strongly affected by differences in water depth and redox status as is Mo. Significant accumulations of V occurred in goethite-coated sand bags in Cell 4, with more V present after 12 mo. Significant accumulations of V also occurred in Cell 5; however, V concentrations at 3 and 12 mo were not significantly different.
There was a large variation in the total iron content of the goethite-coated sand after sampling. The total Fe contents ranged from 648 to 3960 mg Fe kg-1 sand with a mean of 2629 and a standard deviation of 755 mg Fe kg-1 sand. The goethite-coated sand initially contained approximately 4000 mg Fe kg-1 sand. Slightly less Fe was recovered from samples in Cell 4 (2266 ± 683 mg Fe kg-1 sand) than from samples in Cell 5 (2766 ± 768 mg Fe kg-1 sand). The changes in Fe content were most likely due to the reducing conditions of the wetlands. These conditions may either dissolve or transform goethite to other Fe-containing mineral or precipitate forms. Many of the goethite-coated sand samples, particularly those sampled after 12 mo, were darker in color than they were initially and it was more difficult to completely dissolve the Fe coating (some samples retained a slight brown-orange color even after acid extraction). This suggests that at least some of the Fe may have been in another form (e.g., FeS or FeS2). Pyrite, FeS2, is insoluble in hydrochloric acid, while other iron sulfides, such as FeS or Fe3S4, will dissolve in hydrochloric acid (Goldhaber and Kaplan, 1982). However, the total Fe content of the goethite-coated sand was not correlated with trace element concentrations in any fraction, with the possible exception of those samples containing less than approximately 1400 mg Fe kg-1 sand, which had low Mo concentrations. There were only a few samples in this category. Oxidation of Fe sulfides lowers the pH of samples (Nordstrom, 1982). It is likely that some oxidation occurred during air-drying, as evidenced by the low pH of the water extracts, particularly in samples from Cell 4, where the average water extract pH was 5.97.
When surface and buried goethite-coated sand samples were compared, some depth trends were apparent. Greater accumulations of Mo and As occurred at the sediment surface for Cell 4, but not for Cell 5. For instance, after 12 mo 27.14 ± 7.01 mg Mo kg-1 sand more Mo and 3.94 ± 1.43 mg As kg-1 sand more As had accumulated at the surface in Cell 4. However, there were no detectable differences in V concentration with depth in the goethite-coated sand bags. The As and Mo results may be explained by a combination of factors, including proximity to the trace element source (overlying water column) and the presumably greater levels of sulfate reduction occurring at the sediment surface in Cell 4 compared with Cell 5. In many flooded soils and lake sediments, the redox potential decreases as you move down the sediment profile; an oxidized layer is commonly present at the sediment surface and is underlain by anaerobic sediment (Ponnamperuma, 1972; Reddy and Patrick, 1983; Tian-ren, 1985). However, evidence indicates that in this system the opposite is true. Gao et al. (2000) reported an increase in redox potential from -170 mV in the top 5 cm to -90 mV at 5 to 10 cm and up to 300 mV below 15 cm at this study site. The sediment–water interface is a zone of highly active reduction, where both higher levels of organic matter and microbial activity lead to strong reducing conditions; this zone is underlain by a more oxidized, unsaturated zone where microbial activity is lower (Terry, 1998; Gao et al., 2000). Both As and Mo Fe–sulfide complexes may exist in strongly reducing environments (Helz et al., 1996; La Force et al., 2000; Reynolds et al., 1999). In addition, arsenite adsorption on Fe oxides is greater than arsenate adsorption at pH > 6 or 7 (Jain and Loeppert, 2000; Sun and Doner, 1998).
Soil
Table 4 lists concentrations of As, Mo, and V in soil bags taken from Cells 4 and 5. The soil mineral bags accumulated significant levels of Mo in Cells 4 and 5 throughout the experimental period (compared with initial concentrations). In Cell 4, the Mo concentrations increased from 3 to 12 mo; however, the concentrations did not change from 3 to 12 mo in Cell 5. The drying of the outlet end of Cell 5, as well as the differences in redox potential and pH may explain why no additional accumulation was seen after 12 mo in Cell 5. While Mo concentrations in Cells 4 and 5 were similar at 3 mo, at 12 mo Cell 4 accumulated significantly more Mo.
|
For both As and V, the only significant change in total concentrations over time was a loss from Cell 4 after 3 mo; after 12 mo As and V levels increased to preflooded levels in Cell 4. No change over time was detected for Cell 5. However, when samples were fractionated into water-soluble, phosphate-extractable, and acid-extractable fractions, some differences were noted. At 3 mo the As concentrations were greater in Cell 5 compared with Cell 4 in all three fractions; however, at 12 mo the As concentrations in Cell 5 were greater only in the water fraction.
In both Cells 4 and 5, As and V decreased in the acid fraction and increased in the more bioavailable water and phosphate fractions. Either the trace elements redistributed and reached steady state, in which case the concentrations in the soil will remain the same and no additional As or V will be released from the acid fraction, or the trace elements were being slowly released from the acid fraction, resulting in a net depletion. Other researchers have reported increased As solubility under low redox conditions (Amrhein et al., 1993; Masscheleyn et al., 1991; McGeehan and Naylor, 1994; Reynolds et al., 1999). However, Reynolds et al. (1999) found that while overall As solubility increased, some As was partitioned into the solid phase under anaerobic conditions, namely as arsenopyrite (FeAsS). This process may be important in our system due to the high levels of sulfate present. If mobilization of As occurred in this system, longer time periods may be necessary to detect the loss of As from the soil. The data from Cell 4 certainly indicate that the concentrations of As were not in steady state, but instead changed over time. The loss after 3 mo (September) followed by the increase after 12 mo (June) may reflect dynamic changes in the As fractionation as it reaches steady state or seasonal variations in the wetland. La Force et al. (2000) found that As speciation and partitioning varied seasonally in a natural wetland receiving As-contaminated waters and attributed the changes to seasonal variation in both temperature and water depth. In our study, water depth was held constant, but temperature and plant growth varied seasonally.
The concentrations of V were slightly higher in Cell 5 than Cell 4 (significant at 
= 0.06). A number of factors may affect the solubility of V in such a system, including solid and dissolved organic matter, dissolution of adsorbent phases, dissolution of minerals containing V in their crystal lattice, reduction of V(V) to V(IV) and V(III), and uptake by plants. Some of these processes will increase V solubility, while others may decrease V solubility. For instance, reduction of V(V) will generally decrease V solubility, unless large quantities of dissolved organic compounds are present; these organic molecules strongly sorb VO2+ and VO(OH)+ (Wanty and Goldhaber, 1992; Wehrli and Stumm, 1989).
Differences between surface and buried soil bags were noted. Significantly more Mo accumulated at the surface in Cell 5 after 3 mo (3.85 ± 1.80 mg Mo kg-1 soil more, on average), but no depth distribution was noted in Cell 4. This is the opposite of what was found for the goethite-coated sand mineral bags, where more accumulation occurred at the surface in Cell 4, but not in Cell 5. In Cell 4, significantly more V was present at the sediment surface (10.49 ± 3.02 mg V kg-1 soil more at surface) after 12 mo. Similarly, 6.91 ± 2.77 mg V kg-1 soil more V was present at the surface in Cell 5 after 3 mo. There were no detectable differences between surface and buried samples in Cell 4 after 3 mo or Cell 5 after 12 mo. In Cell 4 there appeared to be slightly more As present at the sediment–water interface than buried, on average 3.09 ± 1.05 mg As kg-1 soil more was present at the surface after 12 mo. While our data show that higher levels of trace elements generally occurred at the sediment surface, both for goethite-coated sand and soil bags, the depth distributions were not consistent enough across cells and over time to quantitatively determine the degree of surface versus buried accumulation or release.
The greater concentrations of Mo at the sediment surface reflect a greater accumulation of Mo at the surface compared with buried samples. Molybdenum accumulation occurred where low redox potential and a Mo source (overlying water column) coincide. However, the greater surface concentrations of As and V reflect a net loss in the deeper sediment (buried samples) rather than an accumulation at the surface. As stated earlier, the sediment surface in this system was more reducing than the deeper sediment. More extensive reduction at the surface may have converted As(V) to As(III) and V(V) to V(IV) and possibly V(III). Vanadium (IV) cations are more strongly sorbed by organic matter and Fe and Al oxides (Wanty and Goldhaber, 1992; Wehrli and Stumm, 1989), and V(III) is often precipitated as V oxides and oxyhydroxides (Wanty and Goldhaber, 1992). If most of the organic-sorbed V was sorbed by solid-phase organic matter, and not by dissolved organic matter, then V would be conserved at the surface. Arsenic may be conserved in sulfide phases (e.g., realgar [AsS], arsenopyrite [FeAsS], and orpiment [As2S3]) under strongly reducing conditions (McCreadie et al., 2000; Reynolds et al., 1999). Jain and Loeppert (2000) and Sun and Doner (1998) found that sorption of arsenate and arsenite are roughly equal at neutral pH (6–7), and at higher pH arsenite is sorbed at greater levels. Both As and V may be reduced by H2S in solution (Rochette et al., 2000; Wanty and Goldhaber, 1992) and As(V) may also be reduced by anaerobic bacteria (Zobrist et al., 2000). It is possible that at 2.5 cm below the sediment surface, the redox potential was low enough for dissolution of adsorbent phases, but not low enough for formation of sulfides or V(III) oxides, resulting in a net loss of As and V.
Calcite
Low levels of Mo accumulation occurred on the calcite recovered from bags in Cell 4 after both 3 and 12 mo. At 3 mo, 1.00 ± 0.19 mg Mo kg-1 calcite was present, increasing significantly (at = 0.06) to 1.81 ± 0.37 after 12 mo. In Cell 5 similar accumulations of Mo occurred after 3 and 12 mo, reaching 1.67 ± 0.22 and 1.08 ± 0.24 mg Mo kg-1 calcite, respectively. There was no discernable difference in accumulation of Mo between cells.
The low levels of Mo recovered from calcite bags may have resulted primarily from residual Mo in the pore waters. In Cells 4 and 5, the water-soluble Mo due to pore waters was estimated to be 0.34 to 0.69 mg Mo kg-1. Calcite was not an important sink for Mo under the conditions present in this wetland. Our data supports the laboratory results of Goldberg et al. (1996) who found negligible adsorption of Mo on calcite and calcareous soils over a broad pH range (3–10). However, some researchers have stated that Mo sorption on calcite increases with liming and Mo may form precipitates and coprecipitates with calcite (Adriano, 1986; Kabata-Pendias and Pendias, 1992).
Results from the calcite mineral bags may help in the interpretation of results from the soil mineral bags. Low Mo concentrations in the calcite samples (1.00–1.81 mg Mo kg-1 calcite) suggest that the passive accumulation of Mo in the mineral bags was low. Passive accumulation includes diffusion of small (<25 µm) organic and inorganic particles into the bags, growth of algae in the bags, and enrichment of Mo in sediment pore waters compared with the overlying water column. Algal growth on the outside surface of some mineral bags was observed. Algal growth inside the bags was not obvious, however some growth may have occurred. This passive diffusion represented a low portion (16–30%) of the water-soluble Mo in the soil mineral bags.
No significant accumulation of As was noted on calcite. There was a low but significant accumulation of V on the calcite mineral bags for both cells. Cell 4 accumulated 2.07 ± 0.19 mg V kg-1 calcite after 3 mo. After 12 mo, the V levels dropped to 1.18 ± 0.16 mg V kg-1 calcite. The behavior of V in Cell 5 was similar, accumulating 2.84 ± 0.84 and 0.91 ± 0.12 mg V kg-1 calcite after 3 and 12 mo, respectively. These low levels of V accumulation on calcite indicate that the importance of calcite as a sink for V was relatively low when compared with other adsorbent phases such as goethite-coated sand.
Due most likely to the generally low levels of trace element accumulation on calcite, a comparison of surface and buried samples revealed no significant variation with depth.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
In the soil bags, nearly half of the Mo was present in the water fraction and it is unlikely that the water-soluble Mo resulted solely from Mo in entrained pore waters. Instead, it appears that nearly half of the total accumulated Mo was transformed to a water-soluble form upon air-drying. This has important implications for systems undergoing periodic changes in redox status and for the possible decommissioning of this and similar constructed wetlands. Upon drying, large fluxes of Mo may be mobilized, possibly contaminating ground water or runoff.
Mineral bags were used successfully in this experiment to determine in situ changes in trace element concentrations on specific minerals and soil over time. This method is unique in that accumulations on specific mineral phases can be determined directly, without the problems associated with extrapolating between lab and field conditions (as in sorption experiments) or the problems associated with typical extraction procedures that target specific, but operationally defined, phases of soil. Differences with depth were difficult to detect with this method and results were mixed. This may reflect actual conditions in the field (i.e., fairly uniform concentrations in the top 2 to 3 cm), difficulties in determining and placing samples at exact depths in the field, or the high spatial variability in the field. We suspect that depth variations would be easier to detect if the buried samples were placed further down in the soil profile (6 to 8 cm). However, at greater depths inserting mineral bags becomes more difficult and causes a greater disturbance in the wetland.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  P. M. Fox and H. E. Doner Accumulation, Release, and Solubility of Arsenic, Molybdenum, and Vanadium in Wetland Sediments J. Environ. Qual., November 1, 2003; 32(6): 2428 - 2435. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Vadose Zone Journal | Journal of Plant Registrations | ||||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal |
