Oxidative Degradation of Glyphosate and Aminomethylphosphonate

We report the first evidence for an abiotic pathway of glyphosate and AMPA degradation under environmentally realistic conditions. Both glyphosate and...
0 downloads 0 Views 148KB Size
Download PDF
Environ. Sci. Technol. 2005, 39, 9223-9228

Oxidative Degradation of Glyphosate and Aminomethylphosphonate by Manganese Oxide K. A. BARRETT AND M. B. MCBRIDE* Department of Crop and Soil Science, Cornell University, Ithaca, New York 14853

Glyphosate (N-(phosphonomethyl)glycine), the most commonly used herbicide worldwide, degrades relatively rapidly in soils under most conditions, presumably by microbial processes. The most frequently detected degradation product in soil and water is AMPA (aminomethylphosphonic acid). We report the first evidence for an abiotic pathway of glyphosate and AMPA degradation under environmentally realistic conditions. Both glyphosate and AMPA degraded at 20 °C in dilute aqueous suspensions of birnessite, a manganese oxide common in soils, as evidenced by the accumulation of orthophosphate in solution over a period of several days. It is concluded that the abiotic degradation involved C-P bond cleavage at the Mn oxide surface, although evidence for C-N bond cleavage in the case of glyphosate and sarcosine, a likely degradation product of glyphosate, was found. The degradation of glyphosate was faster than that of AMPA, and higher temperature (50 °C) resulted in faster degradation of both glyphosate and AMPA. The addition of sulfate to the solution had no marked effect on the reaction rate, although Cu2+ addition inhibited degradation. As this metal ion complexes strongly with glyphosate, the inhibition can be attributed to the ability of Cu2+ to limit glyphosate coordination to reactive oxidation sites at the Mn oxide surface. Using a similar experimental design, we were unable to detect glyphosate degradation in an equimolar solution of MnCl2 (0.5 mM). However, we demonstrated that the oxidation of Mn2+ is enhanced both in solution and on an inert surface, in the presence of glyphosate (4:1 Mn-glyphosate molar ratio). This result suggests that the oxidative breakdown of glyphosate in the presence of Mn2+ may ultimately occur following the spontaneous oxygen-mediated oxidation of manganese.

Introduction Globally, glyphosate is one of the most widely applied herbicidal agents. Once in the environment, degradation of this herbicide is generally observed to be rapid, although in some instances relatively long residence times have been reported (e.g., 1, 2). The environmental degradation of glyphosate, a phosphonic acid, is considered to be mediated almost exclusively by microorganisms, with the contribution of abiotic mechanisms being largely dismissed in the literature (e.g., 3). However, Nowack and Stone (4) have described numerous instances where phosphonate ligands chelate metal ions and are subsequently degraded in aqueous solution. In addition, Cordeiro et al. (5) proposed a general* Corresponding author phone: (607)255-1728; fax: (607)255-8615; e-mail: [email protected]. 10.1021/es051342d CCC: $30.25 Published on Web 10/21/2005

 2005 American Chemical Society

ized scheme for the oxidative breakdown of phosphonates involving coordination to a metal oxide via a free radical pathway. More recently, a number of studies have shown that mineral phases of Mn(III, IV) (e.g., manganite, birnessite) may be involved in the degradation of xenobiotics including chlorophenols (6) and triclosan (7). Manganese deficiency is a common constraint to U.S. agricultural production. While the benefits of soil application of MnSO4 are observed to be low (8), a single foliar application of 0.1 kg Mn/ha has been shown to effectively increase yields (9). Although the application of a Mn salt-glyphosate mixture is favored for practical reasons, an antagonistic effect of divalent manganese on the herbicidal activity of glyphosate has been reported. Given the tendency for glyphosate to form stable complexes with transition metals, it has been hypothesized that the formation of insoluble Mn-glyphosate salts may explain this observation (10). However, this reasoning has the problem that Mn2+-glyphosate complexes are relatively weak. Conversely, as it is known that various forms of Mn are involved in the degradation of organic compounds, including cleavage of the C-P bond present in phosphonates, we hypothesize that interaction of glyphosate and its primary degradation product, aminomethylphosphonic acid (AMPA), with Mn in higher oxidation states may result in the oxidation of these phosphonates, at the C-P bond, and the subsequent release of orthophosphate as a degradation product. Thus, with the interest of identifying potentially significant abiotic pathways of glyphosate and AMPA degradation, the interaction of the two molecules with divalent Mn and a Mn(III, IV) oxide (i.e., K+-birnessite) in batch reaction vessels was explored.

Experimental Section A preliminary study assessed the potential for glyphosate degradation catalyzed by divalent Mn at three pH levels. In a 125 mL Erlenmeyer flask, a 100 mL solution of glyphosate reagent (96% purity) containing 10.5 mg/L glyphosate and 0.5 mM MnCl2 was prepared in a background electrolyte solution of 0.01 M NaNO3. On a molar basis, Mn was present at approximately 8 times the concentration of glyphosate. The effect of solution pH was assessed by adjusting the pH to 5.0, 6.0, or 7.0 with NaOH. Controls were prepared with no MnCl2. Each vessel was duplicated, sealed with Parafilm, and placed on a rotary shaker (150 rpm) at room temperature. Aliquots were taken at 20 min and 1, 3, 5, and 24 h, and the appearance of orthophosphate was determined colorimetrically using the phosphomolybdate blue reaction. Degradation of both glyphosate and AMPA (99% purity) in the presence of the manganese oxide, birnessite, was also investigated. In a 125 mL Erlenmeyer flask, a 100 mL solution containing either 0.59 mmol/L glyphosate or 0.90 mmol/L AMPA and 0.05 g of K+-birnessite was prepared in a background electrolyte solution of 0.01 M KNO3. The K+saturated birnessite was synthesized with the HCl-KMnO4 procedure developed by McKenzie (11). As reported by McBride (12), the oxide has a zero point of charge (ZPC) of 3.6 and a surface area of 25.7 m2/g (determined by BET analysis of the N2 adsorption). A preliminary study, monitoring the release of orthophosphate in K+-birnessite-glyphosate suspensions, indicated that adjusting the initial pH to 5, 6, or 7 had only a small effect on the overall rate or extent of the degradation reaction. Therefore, for the remaining experiments, each suspension was adjusted to pH 5 using KOH. In addition to the glyphosate and AMPA reaction with birnessite, three modified reaction conditions were tested. VOL. 39, NO. 23, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

9223

First, to assess the impact of strongly adsorbing and glyphosate-complexing cations on birnessite-catalyzed degradation, Cu(NO3)2 was added to both glyphosate and AMPA suspensions in a 1:1 molar ratio with these phosphonates (22 °C). Second, the effect of anions that might compete with glyphosate or phosphate sorption on the birnessite was investigated with the addition of K2SO4 at 0.01 M (22 °C). Finally, the glyphosate and AMPA reactions with birnessite were run at high temperature (50 °C). This high temperature was used to confirm that any degradation reactions were chemical and not microbial; chemical reaction rates are expected to increase with temperature, whereas 50 °C is inhibitory to the growth of most microorganisms. Reaction flasks for all treatments were prepared in duplicate, sealed with Parafilm, and placed on a rotary shaker at 150 rpm. Controls included reagent solutions incubated in the absence of birnessite (to detect any microbial degradation of the phosphonates) as well as birnessite in 0.01 M KNO3 with no phosphonates. At each sampling time, the flasks were removed from the shaker and stirred rapidly with a Teflon-coated magnetic stir bar so that a homogeneous ∼7 mL aliquot of suspension could be removed without significantly altering the birnessite-glyphosate solution ratio. Aliquots were filtered using a mixed cellulose ester membrane filter (0.2 µm) and analyzed for total orthophosphate using the phosphomolybdate blue method. During this experiment, a significant interference with this colorimetric method, particularly by AMPA, was noted in the form of color suppression. Therefore, calibration curves were developed to correct for AMPA and glyphosate interference in those samples where concentrations of these phosphonates were high enough to create a significant error. Between five and seven sampling times (up to 96 h) were chosen. At selected time intervals, the pH was measured and total dissolved P and Mn in filtrates were determined by ICP spectrometry. In cases where Cu2+ was added to the reaction systems, dissolved Cu2+ was measured by flame atomic absorption spectrophotometry. To test semiquantitatively for the degradation of glyphosate and AMPA to primary amine products, filtrates from selected birnessite reaction systems were analyzed by the buffered (pH 5.5) ninhydrin colorimetric method. In initial tests, freshly prepared glyphosate standards (0.59 mM) gave nonreproducible weakly to strongly positive tests for primary amines based on purple color development with ninhydrin. The positive test result is difficult to explain as glyphosate is a secondary amine and not expected to form the intensely colored Ruhemann’s purple upon reaction with ninhydrin. Sarcosine, a secondary amine and a possible initial decomposition product of glyphosate, consistently gave a negative ninhydrin test, whereas AMPA gave a strongly positive test in buffer. Concentrations of primary amines in solution were estimated by measuring visible absorbance at 570 nm after development of the color, using glycine solutions as standards. To enhance detection of amine breakdown products over a 48 h reaction period, conditions were changed from those previously described, using a 5-fold increase of both birnessite and glyphosate or AMPA concentration and 50 °C reaction temperature. As the above-described glyphosate reactions with birnessite were all conducted under fully aerated conditions, which may have allowed reduced Mn to reoxidize, an additional replicated experiment was conducted under anoxic conditions by continuous purging of the reaction flasks with pure N2, taking samples of suspension at several time intervals up to 48 h and immediately filtering using 0.2 µm membrane filters. The reaction conditions were otherwise as described for the aerated systems, and N2-purged controls (birnessite without glyphosate added) were run simultaneously. The filtrates were analyzed for dissolved Mn by flame atomic 9224

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 23, 2005

absorption, orthophosphate by the phosphomolybdate blue method, and primary amines by the ninhydrin test. The potential for divalent Mn to oxidize in both MnSO4 and Mn2+-glyphosate mixed solutions was tested both in solution at several pH values and upon drying of these solutions on a nonreactive (glass) surface. The MnSO4 solution was prepared at a concentration of 100 mM and adjusted to pH 7.0 with KOH. The Mn2+-glyphosate was similarly prepared from MnSO4 and reagent-grade glyphosate and adjusted to pH 7, with a final concentration of 100 mM Mn and 25 mM glyphosate. Glyphosate solution (25 mM) without Mn was prepared and adjusted to pH 7.0 and used as a control for the oxidation experiments. To measure the tendency for Mn2+ as both a sulfate salt and in mixture with glyphosate to oxidize on exposure to air and ambient light, 1 mL volumes of the MnSO4, Mn-glyphosate, and glyphosate (control) solutions were pipetted into glass Petri dishes and allowed to dry either in full sunlight or in the dark for 6 h. A 3 mL volume of freshly prepared 0.04% tetramethylbenzidine (in water) was then added to each dish. The solution was allowed to react briefly with the contents of the dish to develop color and then transferred into a cuvette for measurement of absorbance at 636 nm using a spectrophotometer. Tetramethylbenzidine is oxidized by Mn in the +3 or +4 oxidation state to benzidineblue, providing a measure of reactive Mn in the oxidized state. The benzidine-blue color intensity at 636 nm was calibrated to reactive Mn oxide by measuring absorbance as a function of the quantity of birnessite added to the tetramethylbenzidine reagent. This allowed the reactivity of oxidized Mn to be expressed as benzidine-reactive Mn (BRMn) in units of µg of birnessite-equivalent.

Results Degradation of Glyphosate with Divalent Mn. No significant glyphosate degradation via C-P bond cleavage was noted in 0.5 mM aqueous MnCl2 at pH 5-7 in equilibrium with atmospheric O2 as evidenced by lack of phosphate in solution over time (data not shown). In contrast, using similar experimental conditions, Nowack and Stone (4) observed the rapid degradation of the polyphosphonate nitrilotris(methylene)phosphonic acid (NTMP), reporting a half-life of 10 min at a buffered pH of 6.5, in equilibrium with 0.21 atm O2. Our result suggests that strong coordination to Mn(II) is a necessary preliminary step to degradation. Indeed, Nowack and Stone (4) reported a log K value of 11.9 for the Mn(II) complex of NTMP while a value of 5.5 has been proposed by Motekaitis and Martell (13) for the glyphosate-Mn(II) chelate (both reported log K values for the 1:1 chelate with fully deprotonated ligand). On the basis of the stability constant for Mn-glyphosate complexation (13), speciation calculations show that less than 1% of the glyphosate would have been complexed with Mn2+ at pH 5, increasing to nearly 10% at pH 7. Degradation of Glyphosate and AMPA with Mn Oxide. The production of orthophosphate was relatively rapid in the glyphosate-birnessite suspensions, but slower in the AMPA-birnessite suspensions, as shown in Figures 1 and 2. Negligible orthophosphate was produced in the glyphosate and AMPA controls, indicating that microbial decomposition was not significant within the time frame of the experiments. Because cleavage of the phosphonates occurs at the C-P bond, the generation of orthophosphate was interpreted to stoichiometrically represent the degradation of both glyphosate and AMPA, with one mole of orthophosphate released for each mole of glyphosate or AMPA degraded. However, the presence of the oxide surface, and therefore the possibility of readsorption of released orthophosphate, prevented accurate monitoring of the total production of the degradation product. On the basis of ICP analysis of total

FIGURE 1. Accumulation of orthophosphate in solution over time as glyphosate (0.59 mM) is reacted at 22 °C with birnessite (50 mg) in 100 mL of 0.01 M KNO3 at an initial pH of 5. The effects of SO42-, Cu2+, anoxia, and high temperature (50 °C) on the accumulation of orthophosphate are also shown. The dissolved PO43- predicted for 50% decomposition of glyphosate by C-P bond cleavage is indicated by the horizontal line.

FIGURE 3. Change in pH during the reaction of AMPA and glyphosate with birnessite. The reaction conditions and treatments are described in Figures 1 and 2.

FIGURE 2. Accumulation of orthophosphate in solution over time as AMPA (0.9 mM) is reacted at 22 °C with birnessite (50 mg) in 100 mL of 0.01 M KNO3 at an initial pH of 5. The effects of SO42-, Cu2+, and high temperature (50 °C) on the accumulation of orthophosphate are also shown. The dissolved PO43- predicted for 50% decomposition of glyphosate by C-P bond cleavage is indicated by the horizontal line. solution P (both phosphonate and phosphate), it could be confirmed that at least 98% of the orthophosphate generated from glyphosate and AMPA reaction with birnessite remained unadsorbed to the Mn oxide, as total dissolved P changed little as the reaction progressed and orthophosphate accumulated in solution. This result may indicate that adsorbed glyphosate inhibited orthophosphate adsorption as it was generated and confirms that the monitoring of dissolved orthophosphate provided a good measure of the extent of phosphonate degradation. As the degradation reaction proceeded, an ∼2 unit increase in pH was measured in all of the glyphosate reaction systems. Therefore, during the 96 h experiment the pH approached 7.4 in all of the glyphosate treatments but peaked at lower pH in the AMPA treatments (Figure 3). The increase in pH is likely to result from the birnessite-mediated oxidation of the phosphonates which may consume H+ as suggested by the proposed glyphosate degradation reaction scheme in Figure 4. The observed pH plateau in both glyphosate and AMPA systems probably reflects termination of the degradation reaction as a result of limited birnessite surface area. In contrast with all other reaction systems tested, the pH was observed to become more acidic over the course of the experiment in the AMPACu2+ system (Figure 3). This is possibly due to the exchange of H+ on the surface of the birnessite by the Cu2+ cation,

FIGURE 4. A possible degradation reaction scheme for glyphosate adsorbed on manganese oxide. which was largely adsorbed in this system. This trend was not observed in the glyphosate-Cu2+ systems, presumably because the metallic cation is held in a stronger tridentate complex with glyphosate, relative to the bidentate AMPACu2+ chelate. ICP analysis detected no increase of Mn2+ in solution during reaction (data not shown). This observation could be due to the rapid readsorption of the cation to the surface of birnessite. In an attempt to determine whether Mn oxide behaved as a true catalyst in these systems, being regenerated with O2 after each oxidation, a polarographic O2 electrode and sealed reaction cell was used to measure molecular oxygen uptake in the glyphosate-birnessite reaction system. No O2 consumption was detected; however, the reaction may be too slow to permit measurement with this method. However, when glyphosate was reacted with birnessite under VOL. 39, NO. 23, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

9225

N2 purging, dissolved Mn increased to concentrations higher than those measured in the birnessite control (pH in both treatment and control suspensions was near 6.3 after 48 h of reaction), exceeding 3 mg/L at the 24 h of reaction time, more than 10 times higher than the concentration in the control. Nevertheless, there was less than stoichiometric release of Mn based on the quantity of glyphosate oxidized, and the observed darkening in color of the birnessite as reaction with glyphosate progressed suggested a change in oxidation state of the structural Mn. Earlier research has shown very little dissolution of Mn2+ from K+-birnessite during the oxidation of dihydroxybenzenes until approximately 20-40 µmol had been oxidized by 50 mg of the birnessite (14). Although the anoxic condition did not prevent glyphosate oxidation by birnessite, it did limit the rate of oxidation as evidenced by a nearly 50% lower orthophosphate measured after 24 h of reaction under N2 compared to aerated conditions (see Figure 1). The results suggest that Mn oxide behaves catalytically to some degree in the aerated systems, as Mn2+ released during glyphosate oxidation is adsorbed and reoxidized. Overall, birnessite-mediated degradation was much more rapid and complete for glyphosate than for AMPA (Figures 1 and 2). In the glyphosate system, 0.031 mmol (53%) of the glyphosate had been oxidized within 50 h, based on the quantity of dissolved orthophosphate generated. The reaction did not proceed significantly further over the following 46 h (Figure 1). In the AMPA systems, a steady, linear accumulation of orthophosphate was noted with 0.014 mmol (16%) of AMPA degraded by 96 h (Figure 2). The difference in the rate of glyphosate and AMPA degradation may be related to the affinity of each molecule for the surface of the birnessite, as coordination to the surface Mn of the oxide is necessary for the transfer of electrons from the phosphonate to Mn(III) or Mn(IV) and subsequent cleavage of the C-P bond. For both glyphosate and AMPA, the high temperature (50 °C) treatment had a dramatic effect on both the rate and extent of the chemical degradation reaction, demonstrating the reaction to be chemical and not biological (Figures 1 and 2). Although degradation and phosphate release at 50 °C reached a plateau more rapidly for glyphosate (∼26 h) than for AMPA (∼48 h), approximately the same number of total moles (i.e., 0.042 mmol) of both were degraded by the birnessite. On the basis of a calculation using the surface area of the birnessite and the molecular size of glyphosate, the ratio of glyphosate or AMPA to birnessite in the reaction systems was such that, at most, 10% of the molecules in solution could be sorbed at one time. This was consistent with the total dissolved P analysis by ICP (data not shown), which indicated that over the course of the experiment, less than 2% of the total P was sorbed in either the glyphosate or AMPA suspensions. While reduced Mn may be reoxidized by molecular oxygen, as suggested by the N2-purging experiments, and therefore the degradation of both glyphosate and AMPA may be theoretically completed with a small amount of Mn oxide, the reaction nonetheless seemed to be limited to degrading 0.042 mmol of the phosphonates (approximately 71% of the glyphosate and 47% of the AMPA), as was observed during the 50 °C temperature incubations. Thus, it seems more likely that the complete degradation of the phosphonates was prevented by changes to the surface of the oxide as the Mn was progressively reduced, as anionic degradation products occupied potential glyphosate and AMPA binding sites, and as the increasing pH influenced both the binding of the phosphonates to the oxide and the oxidation potential of the system. As expected, the addition of Cu2+, a metal cation with high affinity for both the phosphonates and the oxide, had 9226

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 23, 2005

a suppressive effect on the rate and extent of the degradation reactions (Figures 1 and 2). This effect is presumably a result of a competitive complexation reaction that prevents the interaction of the glyphosate and AMPA phosphonate moiety with the surface of the oxide. From the reaction conditions used in this experiment, and the known stability constants for the Cu2+-glyphosate and Cu2+-AMPA complexes (13), it was calculated that 96% of the glyphosate in solution would have been complexed at the initial pH of 5. Since the pH increased to well above 6 during reaction, >99% of dissolved glyphosate would have been complexed after reaction times of a few hours. Analysis of the solution phase for dissolved Cu2+ showed that, after 24 h of reaction, 64% of the Cu2+ was sorbed by the birnessite. Therefore, Cu2+ may have inhibited the glyphosate decomposition reaction by complexing with both dissolved and sorbed glyphosate (forming a ternary complex on the surface) or by occupying reactive surface sites of the oxide. AMPA complexes less strongly with Cu2+ than glyphosate, with speciation calculations showing that only about 11% of the dissolved AMPA in the reaction systems would have been complexed with Cu2+ at the initial reaction pH of 5. This would increase to 64% at pH 6. However, because the Cu2+-AMPA-birnessite system became more acidic with time, approaching pH 4 (Figure 3), it is unlikely that the suppressive effect of Cu on AMPA decomposition is attributable to complexation of AMPA in solution. Even at this low pH, however, Cu2+ sorbed readily on the birnessite, with analysis of the solution phase for Cu2+ revealing that 92% of the total Cu was sorbed after 24 h of reaction time. Thus, Cu inhibition of AMPA decomposition was probably caused primarily by complexation with sorbed AMPA or occupation of reactive surface sites. Given the limited reactive surface area of the birnessite, excess SO42- at approximately a 10:1 molar ratio with the phosphonates was tested for its ability to competitively sorb and inhibit both glyphosate and AMPA degradation. However, reduced degradation rates were not observed with SO42- present, and the trend of accumulating orthophosphate in solution was nearly identical for both the glyphosate and glyphosate-SO42- treatments (Figure 1). In the AMPA reactions, slightly more orthophosphate was detected in solution when SO42- was present (Figure 2). This result suggests that SO42effectively displaces bound orthophosphate, which may sorb more readily to the surface of birnessite in the AMPA systems relative to that in the glyphosate systems, as AMPA is thought to have a lower affinity for the oxide surface. Therefore, greater orthophosphate in solution with SO42- present is not interpreted as an enhanced degradation rate of AMPA. As phosphate accumulation in solution increased nearly linearly in the initial stages of glyphosate and AMPA decomposition without any lag time, it is reasonable to infer that C-P bond cleavage is the initial step of degradation. However, Nowack and Stone (15) found evidence for Mn oxide-promoted C-N as well as C-P bond cleavage during the oxidative decomposition of nitrilotrismethylenephosphonate, a molecule chemically similar to glyphosate and AMPA but with three phosphonate groups. To test for the possibility of C-N bond cleavage during the birnessitepromoted decomposition of glyphosate and AMPA, the ninhydrin test for primary amines was run on filtrates collected from the reaction systems at several time intervals. Glyphosate, which is not a primary amine, nevertheless produced a variably positive ninhydrin test but weaker than that of glycine or AMPA. During glyphosate reactions with birnessite under both anoxic and oxic conditions, the ninhydrin test on the solution phase produced a more intense color after 24-48 h of reaction, indicating some formation of the primary amines, presumably AMPA or glycine, from C-N bond cleavage. The test results further suggest that sarcosine, a known decomposition product of glyphosate,

FIGURE 5. Quantity of oxidized benzidine-reactive Mn measured when 1 mL of 100 mM glyphosate (Gly) Mn sulfate or Mn-glyphosate (Mn-Gly) complex was exposed to full sunlight. could not have accumulated as a main decomposition product, as sarcosine tests negatively with ninhydrin. A separate experiment was conducted in which 5.6 mM sarcosine was reacted at 50 °C with 250 mg of birnessite in 100 mL of 0.01 M KNO3 at an initial pH of 5. The birnessite promoted rapid degradation of sarcosine to a primary amine, possibly glycine, as indicated by a strongly positive ninhydrin test within 2 h. This result indicates that, if sarcosine had been the initial product of glyphosate degradation via C-P bond cleavage, it may have decomposed in the presence of birnessite via C-N bond cleavage. However, under the initial reaction conditions used in the glyphosate-birnessite systems (50 mg of birnessite reacted at 20 °C with 0.59 mM glyphosate at an initial pH of 5), the ninhydrin test showed no degradation. This suggests that birnessite-catalyzed C-N bond cleavage is strongly temperature-dependent like C-P bond cleavage. In summary, the ninhydrin test revealed that C-N bond cleavage is a significant mechanism of glyphosate and sarcosine degradation by birnessite, although the specific reaction pathway(s) could not be deduced from this test. Oxidation of Mn(II) in the Absence and Presence of Glyphosate. As shown in Figure 5, there was measurable oxidation of Mn when 100 mM MnSO4 was dried on a glass surface, as detected by benzidine-reactive Mn (BRMn). Oxidation was greater when the Mn salt was exposed to full sunlight, and the presence of glyphosate further enhanced Mn oxidation in both the dark and light. The light-enhanced oxidation may be explained by the fact that UV radiation photooxidizes Mn2+ forming birnessite in aqueous solution (16), and glyphosate may further promote Mn oxidation by stabilizing the Mn(III) oxidation state (4, 17). In this experiment a 4:1 Mn/glyphosate mole ratio was tested, as in field applications typical tank mixtures also contain an excess of Mn. For example Bernards et al. (18) employed an application ratio of approximately 17:1 Mn/glyphosate. A similar enhancement of Mn oxidation was measured for Mn-glyphosate relative to MnSO4 solutions at pH 4.5 and pH 6.5, with oxidation being greater at the higher pH (data not shown). These results confirm that oxidation of Mn2+ occurs spontaneously in Mn salt solutions dried on surfaces, particularly when glyphosate is present. In solution, oxidation of Mn2+ is also enhanced by the presence of glyphosate and can occur even at relatively low pH, conditions that are normally unfavorable for Mn2+ oxidation by molecular oxygen.

Discussion Although the polyphosphonates studied by Nowack and Stone (4, 17) could be rapidly decomposed in aqueous

solution in the presence of Mn2+, under similar experimental conditions we were unable to measure glyphosate degradation. The cleavage of the C-P bond in polyphosphonates by soluble Mn2+ is assumed to be pH- and O2-dependent, and is initiated by complex formation between the phosphonate and Mn2+. Nowack and Stone (4, 17) have proposed two possible mechanisms for the subsequent oxidative degradation of phosphonates: (1) oxidation of Mn2+ to Mn3+ (by O2) followed by electron transfer from the organic molecule to Mn3+, and (2) activation of O2 by phosphonate-bound Mn2+ without any change in the oxidation state of Mn2+. Therefore, the fact that the glyphosate C-P bond is not highly susceptible to this degradation pathway can be attributed to the relatively weak glyphosate complex with Mn2+. The relatively facile degradation of glyphosate by birnessite, involving C-P and C-N bond cleavage, almost certainly requires bonding of the molecule to the oxide surface, presumably by innersphere coordination to Mn(III) or Mn(IV). Electrons are then transferred to Mn, and the reaction can proceed to some degree in the absence of dissolved O2. In the process, both Mn2+ and OH- are generated, although release of Mn2+ into solution may not be detected in aerated systems as the Mn2+ is readsorbed or reoxidized by O2 (12). Since glyphosate and AMPA are not completely oxidized, the Mn oxide may become sufficiently reduced so that it is no longer an effective oxidant. The slower degradation of AMPA, despite the fact that the identical C-P bond is cleaved in both AMPA and glyphosate, suggests that AMPA does not coordinate as strongly to the Mn oxide as does glyphosate. This weaker coordination can be attributed to the lack of the carboxylate functional group in AMPA and suggests that surface complexation of glyphosate by the oxide involves more than Mn-phosphonate bond formation. Although the results of this study do not prove that the field-observed antagonistic effect of Mn on glyphosate herbicidal efficacy in tank mixtures could be explained by Mn2+-promoted degradation in the tank, our results with MnSO4 and Mn-glyphosate solutions dried and exposed to oxygen and light strongly suggest that Mn oxides would be formed on plant leaves from the application of divalent Mn salts, and this may be enhanced when Mn2+-glyphosate mixtures are sprayed onto crops. Mn oxides formed on the leaf surfaces could sorb and degrade simultaneously or subsequently added glyphosate, reducing its herbicidal activity. In most soils, Mn is a trace metal with a typical concentration of about 300-1000 mg/kg, so that glyphosate is more likely to sorb on soil constituents other than Mn oxides. The greatest potential for Mn-mediated degradation of glyphosate is probably in subsoils, where microbial activity is limited.

Literature Cited (1) Nomura, N. S.; Hilton, H. W. The adsorption and degradation of glyphosate in sugarcane soils. Weed Res. 1977, 17, 113-121. (2) Bo¨rjesson, E.; Torstensson, L. New methods for determining glyphosate and (aminomethyl)phosphonic acid in water and soil. J. Chromatogr., A 2000, 886, 207-216. (3) Rueppel, M. L.; Brightwell, B. B.; Schaefer, J.; Marvel, J. T. Metabolism and degradation of glyphosate in soil and water. J. Agric. Food Chem. 1977, 25, 517-527. (4) Nowack, B.; Stone, A. T. Degradation of nitrilotris(methylenephosphonic acid) and related (amino)phosphonate chelating agents in the presence of manganese and molecular oxygen. Environ. Sci. Technol. 2000, 34, 4759-4765. (5) Cordeiro, M. L.; Pompliano, D. L.; Frost, J. W. Degradation and detoxification of organophosphonates: cleavage of the carbon to phosphorus bond. J. Am. Chem. Soc. 1986, 108, 332-334. (6) Ukrainczyk, L.; McBride, M. B. The oxidative dechlorination reaction of 2,4,6-trichlorophenol in dilute aqueous suspensions of manganese oxides. Environ. Toxicol. Chem. 1993, 12, 20052014. VOL. 39, NO. 23, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

9227

(7) Zhang, H.; Huang, C. H. Oxidative transformation of triclosan and chlorophene by manganese oxides. Environ. Sci. Technol. 2003, 37, 2421-2430. (8) Gettier, S. W.; Martens, D. C.; Hallock, D. L.; Stewart, M. J. Residual Mn and associated soybean yield response from MnSO4 application on a sandy loam. Plant Soil 1984, 81, 101-110. (9) Mascagni, H. J.; Cox, F. R. Effective rates of fertilization for correcting manganese deficiency in soybeans. Agron. J. 1985, 77, 363-366. (10) Bailey, W. A.; Poston, D. H.; Wilson, H. P.; Hines, T. E. Glyphosate interaction with manganese. Weed Technol. 2002, 16, 792799. (11) McKenzie, R. M. The reaction of cobalt with manganese dioxide minerals. Aust. J. Soil Res. 1970, 8, 97-106. (12) McBride, M. B. Oxidation of dihydroxybenzenes in aerated aqueous suspensions of birnessite. Clays Clay Miner. 1989, 37, 341-347. (13) Motekaitis, R. J.; Martell, A. E. Metal chelate formation by N-phosphonomethylglycine and related ligands. J. Coord. Chem. 1985, 14, 139-149.

9228

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 23, 2005

(14) McBride, M. B. Oxidation of 1,2- and 1,4-dihydroxybenzene by birnessite in acidic aqueous suspension. Clays Clay Miner. 1989, 37, 479-486. (15) Nowack, B.; Stone, A. T. Homogeneous and heterogeneous oxidation of nitrilotrismethylenephosphonic acid (NTMP) in the presence of manganese (II, III) and molecular oxygen. J. Phys. Chem. B 2002, 106, 6227-6233. (16) Anbar, A. D.; Holland, H. D. The photochemistry of manganese and the origin of the banded iron formations. Geochim. Cosmochim. Acta 1992, 56, 2595-2603. (17) Nowack, B.; Stone, A. T. Manganese-catalyzed degradation of phosphonic acids. Environ. Chem. Lett. 2003, 1, 24-31. (18) Bernards, M. L.; Thelen, K. D.; Penner, D. Glyphosate efficacy is antagonized by manganese. Weed Technol. 2005, 19, 27-34.

Received for review July 11, 2005. Revised manuscript received September 22, 2005. Accepted September 22, 2005. ES051342D