Using 19F NMR spectroscopy to determine trifluralin binding to soil

Article abstract of DOI:10.1021/es0403110

Trifluralin is a widely used herbicide for the control of broad leaf weeds in a variety of crops. Its binding to soil may result in significant losses in herbicidal activity and a delayed pollution problem. To investigate the nature of soil-bound trif lur

Full text of DOI:10.1021/es0403110

Environ. Sci. Technol. 2004, 38, 6645-6655
mended that trifluralin be injected under the soil surface (1,
Using 19F NMR Spectroscopy to
12). However, subsurface application may not necessarily
prevent losses in herbicidal activity, since as demonstrated
by Grover et al. (1), in wet soils of high organic matter
contents, reduced forms of trifluralin can become incorpo-
rated into soil organic matter. The incorporation, in turn,
may evolve into a delayed pollution problem in case of future
release of bound trifluralin residues under favorable condi-
tions.
To date, relatively little has been done to investigate the
mechanism of trifluralin binding despite the progress in
determining other xenobiotic interactions with soil (13-15).
Recent research in this area involved contaminants labeled
with both 14C and stable isotopes to facilitate the analyses
of xenobiotic/soil organic matter complexes by NMR tech-
niques. In the studies of Strynar et al. (16) using [15N]-2,4,6-
trinitrotoluene (TNT), NMR evidence suggested that reducing
the nitro groups under anoxic conditions led to the formation
of very stable covalent bonds with soil and compost humic
materials. Among other studies on binding interactions that
involved xenobiotics labeled with stable isotopes were those
using [15N]-aniline (17), [13C]-2,4-dichlorophenol (18), and
[13C]-cyprodinil (19).
Determine Trifluralin Binding to Soil
M A R K S T R Y N A R , † J E R Z Y D E C , * , ‡
A L A N B E N E S I , § A . D A N I E L J O N E S , §
R O D E R I C K A . F R Y , § A N D
J E A N - M A R C B O L L A G ‡
National Exposure Research Laboratory, U.S. EPA,
Research Triangle Park, North Carolina 27711,
Laboratory of Soil Biochemistry, Penn State Institutes of the
Environment, 107 Research Building C, University Park,
Pennsylvania 16802, and Department of Chemistry,
Eberly College of Science, The Pennsylvania State University,
University Park, Pennsylvania 16802
Trifluralin is a widely used herbicide for the control of
broad leaf weeds in a variety of crops. Its binding to soil
may result in significant losses in herbicidal activity and a
delayed pollution problem. To investigate the nature of soil-
bound trifluralin residues, 14C-labeled herbicide was incubated
for 7 weeks with four soils under anoxic conditions. As
determined by radiocounting, trifluralin binding ranged
between 10 and 53% of the initial 14C depending on the
soil tested. 19F NMR analyses of the methanol extracts and
different fractions of the extracted soil suggested that
bound residue formation largely involved reduced metabolites
of the herbicide. A 2,6-diamino product of trifluralin
reduction with zero-valent iron (Fe-TR), and the standard
of a 1,2-diaminotrifluralin derivative (TR6) formed covalent
bonds with fulvic acid (FA), as indicated by the 19F NMR
spectra taken periodically over a 3-week contact time. At
short contact times, TR6 and Fe-TR formed weak physical
bonds with FA, as the respective spin-lattice relaxation
times (T1) decreased from the range 1300-1831 ms for
TR6 or Fe-TR analyzed in the absence of FA to the range 150-
410 ms for TR6/FA or Fe-TR/FA mixtures. In general, the
results indicated that trifluralin immobilization involved a
variety of mechanisms (covalent binding, adsorption,
sequestration), and with time it became increasingly stable.
Presently, 19F emerges as a useful tracer for research in
this area. Benefits of using 19F over the use of stable isotopes
include lack of background 19F signal in soils, increased
sensitivity and spectral range relative to 15N and 13C, and
perhaps most important, no need to synthesize 19F-labeled
xenobiotics as 19F is 100% abundant in fluorinated com-
pounds and constitutes an elemental tracer. A limitation is
that only F-containing xenobiotics may be investigated,
though Chien et al. (20) used atrazine derivatized with fluorine
for their investigation.
Previously, 19F NMR has been used to study interactions
between fluorinated atrazine (20, 21) or 4′-fluoro-1′-ac-
etoonaphthone (22) and various humic materials. Other
studies used this technique to investigate fipronil degradation
pathways (23), characterize photolysis products of 3-triflu-
oromethyl-4-nitrophenol (24), detect the chlorodifluoroacetic
acid ions (25), and evaluate sorption of hexafluorobenzene
to soil organic matter (26), sediments, polymers, or activated
carbon (27). Mabury and Crosby (11) appear to be the only
researchers who investigated trifluralin in depth by 19F NMR,
mainly to identify photodegradation products that could react
with soil and form bound residues.
In this study, 19F NMR was applied to investigate the
formation of soil-bound residues of trifluralin under anoxic
metabolic conditions. The hypothesis was that, as was the
case for TNT (16), the nitro groups of the trifluralin molecule
would be reduced to amino groups which would then react
with soil organic matter. In an analogous manner, aromatic
amines reacted with humates via carbonyl or quinone
moieties (28). On the basis of previous reports (1, 3), bound
residues of trifluralin were assumed to retain the trifluoro-
methyl (CF3) group. No study, to date, has suggested
mineralization of the CF3 group, at least in the short term (3
yr). There are indications, however, that eventually the CF3
group may be oxidized to the carboxyl group (3).
Introduction
Trifluralin (R,R,R-trifluoro-2,6-dinitro-N,N-dipropyl-p-tolui-
dine) is a dinitroaniline pre-emergence herbicide used for
the control of broadleaf weeds in a wide variety of crops
including cotton, soybeans, and alfalfa (1, 2). It is among the
top five herbicides produced annually in the United States
(24 000 t yr-1) with $300 million in annual sales (1) despite
being a suspected carcinogen. Previous studies on trifluralin
have included its fate and transport in soils (3, 4), the
transformation under low redox or anoxic conditions (5-7),
the co-metabolism in sewage systems (8), and trifluralin
photodecomposition (9-11). To maintain the viability of the
herbicide and prevent photodecomposition, it was recom-
The major goal of this study was to investigate the
mechanism of trifluralin binding to soil. Despite the 100%
abundance of 19F in the molecule of test compound, the
concentration of trifluralin had to be 3 orders of magnitude
higher than the recommended application rate to ensure
obtaining clear NMR signals. However, elevating pesticide
concentrations to study binding mechanisms is an acceptable
approach that was applied in several previous investigations
(14, 17, 19).
* Corresponding author phone: (814)863-0843; fax: (814)865-7836;
e-mail: jdec@psu.edu.
† U.S. EPA.
‡ Penn State Institutes of the Environment.
§ The Pennsylvania State University.
9
10.1021/es0403110 CCC: $27.50
Published on Web 11/12/2004
2004 American Chemical Society
VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 6645

TABLE 1. Chemicals Used for Experiments
Experimental Setup. Fifty-gram samples of four uncon-
taminated local soils (designated Pope, Hagerstown, Chagrin,
and Carlisle) were thoroughly mixed with 10 mg of trifluralin
and 0.23 µCi of uniformly ring-labeled [14C]-trifluralin dis-
solved in methanol to result in an initial herbicide concen-
tration of 200 mg kg-1 and an initial radioactivity of 4.6 µCi
kg-1. The recommended application rate for this herbicide
is 0.25-0.5 mg kg-1 (1); however, for analytical purposes, a
much higher trifluralin concentration was necessary. Table
2 lists physicochemical properties of the soils. Soils were
collected 1 week prior to use and stored at 4 °C. Soil
characteristics were determined by The Pennsylvania State
University Agricultural Analytical Testing service using
Experimental Section
Chemicals. Unlabeled and 14C-labeled trifluralins as well as
five standards of reduced trifluralin derivatives were donated
by Dow AgroSciences (formerly Dow-Elanco) (Indianapolis,
IN). To obtain two additional trifluralin reduction products,
1 mg of trifluralin in 20:80 water:methanol was treated with
50 mg of zero-valent iron nanoparticles or zero-valent iron
nanoparticles with palladium (both donated by Dr. Thomas
Mallouk, The Pennsylvania State University) and centrifuged.
The supernatant was passed through a 0.45-µm nylon filter
and analyzed by LC-MS and 19F NMR. The chemical names,
structures of trifluralin, and its transformation products are
listed in Table 1.
9
6646 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004

TABLE 2. Physicochemical Properties of Soil
Pope
Hagerstown
Chagrin
Carlisle
OM %
3.3
2.7
2.8
7.4
sand %
silt %
55.9
29.9
27.7
38.0
34.2
clay loam
17.0
0.15
7.6
35.9
28.9
35.2
25.4
18.7
46.2
clay %
24.0
texture
CEC (mequiv/100 g)
N %
sandy loam
loam
clay loam
9.8
9.2
19.9
0.21
0.15
0.31
pH
class.
5.1
6.2
5.1
Fluventic Dystrochrept
Typic Hapludalf
Dystric Fluventic Eutrochrept
Typic Medisaprist
standard soil testing procedures. The samples were distrib-
uted into 100-mL serum bottles, amended with an anaerobic
microbial consortium from swine manure slurry (2.0 mL)
and glucose (3 mL of a 5% glucose solution w/v), flushed
with N2 gas to create anoxic conditions, crimp-sealed with
a butyl rubber stopper, and incubated in darkness for 7 weeks
at 25 °C.
The MS analysis of the methanol extracts was performed
on a Quattro-II mass spectrometer (Micromass, Beverly, MA)
that was coupled to a Shimadzu LC-10ADvp pump (Shimadzu
Instruments, Columbia, MD). The spectra were generated
using atmospheric pressure chemical ionization (APCI)
operating in positive-ion mode. Trifluralin transformation
products were separated on a Waters Spherisorb ODS1 4.6
mm × 250 mm column with a 5 µm particle size (Waters,
Milford, MA) and a Security Guard C18 guard column 4.0
mm length × 3.0 mm i.d. (Phenomenex, Torrance, CA). The
mobile phase was delivered at 1.0 mL/min in a gradient mode
that began at 55:45 acetonitrile:water and ramped to 100:0
acetonitrile:water over 10 min, followed by holding for 5 min
and then ramping back down to 55:45 acetonitrile:water over
5 min.
For 19F NMR analyses, methanol extracts (1.0 mL) were
evaporated to dryness and redissolved in methanol-d4 (0.5
mL), and fulvic acid samples (10-25 mg) were dissolved in
50% solution of methanol-d4 in deionized water. Liquid-state
19F NMR spectra were acquired on a Bruker DPX-300
spectrometer operating at 282.13 MHz with 1H decoupling,
typically with 64-1024 scans and a relaxation delay of 0.5-1
s. All liquid-state 19F spectra were referenced to CFCl3 (0.000
ppm). Solid-state 19F NMR spectra for humic acid were
obtained using an 11.7-T Varian/Chemagnetics Infinity
spectrometer operating at 470.081 MHz. All solid-state spectra
were referenced to sodium trifluoroacetate (-79.13 ppm).
Solid trifluralin (5 mg) or trifluralin (5 mg) mixed with control
Chagrin HA (50 mg) with mortar and pestle were used as
controls. Control HA from Chagrin soil was processed as
above.
Soil Fractionation. On the completion of the incubation,
soil samples were extracted by shaking for 24 h with 50 mL
of methanol, centrifuged at 10000g for 1 h, and rinsed 3 times
by resuspending in 10 mL of methanol and shaking for 1 h.
The supernatants were combined, and the total extract was
prepared for analysis by radiocounting, thin-layer chroma-
tography (TLC), high-performance liquid chromatography
(HPLC), mass spectrometry (MS), and 19F NMR spectroscopy.
The solids were air-dried, extracted by a 24-h shaking with
N2-purged 0.1 M NaOH (75 mL), centrifuged for 1 h at 10000g,
and rinsed 5 times with 30 mL of 0.1 M NaOH. The resulting
humin fraction was analyzed by radiocounting (after com-
bustion to 14CO2 and trapping in scintillation cocktail). The
NaOH extract (combined with the washings) was acidified
with 5 M HCl to pH <1.0, stored overnight in a refrigerator
to facilitate the precipitation of humic acid (HA), and
centrifuged. The pellets of humic acid from Carlisle and
Chagrin soils were rinsed 5 times with 0.1 M HCl, redissolved
in 0.1 M NaOH, analyzed by radiocounting, and prepared for
subsequent 19F liquid-state or solid-state NMR analyses by
dialysis against water using a 6-8-kDa membrane and freeze-
drying. The washings from Chagrin humic acid were com-
bined with the supernatant (fulvic acid fraction, FA), extracted
with chloroform to remove any free residues of trifluralin
and also analyzed by radiocounting and 19F NMR for fulvic
acid-bound trifluralin residues.
Analyses of Coupling Products by 19F NMR and Spin-
Lattice Relaxation Time (T1) Measurements. Fulvic acid
isolated from a control sample of Chagrin soil was dissolved
in a 50% solution of methanol-d4 in deionized water and
spiked with the reduced trifluralin products (TR6, Fe-TR, or
Pd-Fe-TR) immediately before the19F NMR analysis or spin-
lattice relaxation time measurements (both on a Bruker DPX-
300 spectrometer). The NMR spectra were taken at selected
time intervals from time zero up to 3 weeks. All spectra were
referenced to CFCl3 with chemical shift set to 0.000 ppm.
Typical spectra were acquired with 1024 scans, a relaxation
Analyses of Soil Fractions by Radiocounting, TLC, MS,
and 19F NMR. Radiocounting was performed on a Beta Trac
6895 liquid scintillation counter (Tracor Analytic, Elk Grove,
IL) using Eco-Scint cocktail (National Diagnostics, Atlanta,
GA). The extracted solids (0.1 g) were prepared for radio-
activity measurements by combustion in a R. J. Harvey OX600
biological oxidizer (Hillsdale, NJ) with 14CO2 trapping in 15
mL of R. J. Harvey Carbon-14 Cocktail. Liquid samples (0.5
mL, methanol extracts, FA, HA) were analyzed in 15 mL of
scintillation cocktail (Ecoscint, National Diagnostics, Atlanta,
GA).
1
delay of 1.0 s, and H decoupling.
At the completion of the 3-week experiment, the reaction
mixtures were evaporated to dryness under gentle N2 stream,
redissolved in 1 mL of deionized water, and extracted three
times with 1 mL of chloroform. The combined chloroform
extracts were evaporated to dryness under N2 gas. The
extracted aqueous solution of FA was freeze-dried. The dry
residues were redissolved in 50% solution of methanol-d4 in
deionized water and analyzed by 19F NMR as described above.
Spin-lattice (T1) relaxation time experiments were per-
formed with 1H decoupling using TR6 and Fe-TR. The
compounds were analyzed with and without FA. Relaxation
data were obtained using the inversion recovery pulse
sequence. T1 values were determined for all peaks that were
Prior to TLC analysis, aliquots of the methanol extracts
(2.0 mL) were evaporated to dryness under a gentle stream
of N2, redissolved in 100 µL of methanol, and applied as
5-mm spots to a 0.25 mm silica gel 60 F254 precoated TLC
plates (E. Merck, Darmstadt, Germany). A spot of 14C-labeled
standard of trifluralin was used as a reference for each TLC
plate. The plates were developed using an 90/10 hexane/
acetone solvent system according to Jacobson et al. (8), then
air-dried, and imaged on a Bioscan System 200 imaging
scanner (Washington, D.C.). Rf values (retention factor) were
determined by dividing the distance moved by the product
spot by the distance moved by the solvent front.
9
VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 6647

FIGURE 1. Fractionation of [14C]-trifluralin after a 7-week anoxic
incubation.
clearly defined in the spectra. Multiple peaks were chosen
in “broad resonances” that probably represented bound
metabolites. NMR parameters (90°pulse )10.85, 180°πpulse
) 21.7, sw ) 5.164 ppm, and o1p ) -63.307 ppm) with 16
delay values ranging from 1 ms to 16 s were used in the T1
experiments.
Results
Trifluralin Binding. At the end of the 7-week incubation
period, the recoveries of the added 14C ranged from 89.8%
for Pope soil to 96.8% for Chagrin soil (Figure 1). The majority
of extractable radioactivity was found in the methanol-
extractable fraction. The Pope soil showed the highest
percentage of extractable radioactivity (80.9%), with little
radioactivity present in the HA, FA, and humin fractions (9.0%
combined). The remaining three soils (Hagerstown, Chagrin,
and Carlisle) showed significant amounts of unextractable
radioactivity, ranging from 31.1% in Hagerstown soil to 53.1%
in Carlisle soil. In most instances, the percentage of bound
radioactivity decreased in the following order: humin > HA
> FA (5-35%, 2-15%, and 4-6%, respectively).
TLC analysis of the methanol extracts showed three major
radioactive peaks, one of which represented unaltered
herbicide, as indicated by its Rf value (∼0.9) that matched
that for the [14C]-trifluralin standard (Figure 2). The Pope
soil extract showed only the trifluralin peak. No trifluralin
peak, however, was generated by the Chagrin soil extract
that, on the other hand, generated two peaks at Rf 0.0 and
0.5, representing transformation products. Methanol extracts
from the remaining soils (Carlisle and Hagerstown) showed
three spots at Rf 0.0, 0.5, and 0.9 for transformation products
and residual trifluralin, respectively.
FIGURE 2. Radioscanning of thin-layer chromatography (TLC) plates
for methanol extracts from different soils.
Both the Carlisle and Chagrin samples showed clear fluorine
signals in the form of large broad resonances at -60.8 and
-61.2 ppm, respectively. The signals were centered between
several much smaller signals that might be spinning side-
bands resulting from weak interactions between trifluralin
and humic acid. A chloroform-extracted FA fraction from the
Chagrin soil analyzed by liquid-state 19F NMR showed two
broad resonances at -60.598 and -61.135 ppm (Figure 6).
19F NMR Analysis of Fe-TR and Pd-Fe-TR and Their
Coupling Products with Fulvic Acid. As tentatively deter-
mined by mass spectrometry, the reaction with zero-valent
iron resulted in the formation of 2,6-diamino analogue
trifluralin (Fe-TR) (m/z 276) with a 19F signal at -62.653 ppm
and an unidentified product at -62.612 ppm (data not
shown). These signals did not match any of the19F resonances
collected for methanol extracts from the incubated soils,
probably because Fe-TR-like products were involved in
bound residue formation. The same two products were also
formed, although in lesser quantities, in the presence the
palladium zero-valent iron; however, additionally a product
was formed with a chemical shift at -60.879 ppm that closely
matched 19F signals shown in the spectra of methanol extracts
from Hagerstown, Chagrin, and Carlisle soils (-60.874,
-60.870, and -60.868 ppm, respectively) (Figure 3). As
determined by mass spectrometry, this product (Pa-Fe-TR)
had the same mass (m/z 276) as the zero-valent iron product
(Fe-TR), and was tentatively identified as a 2,6-diamino
derivative of trifluralin with an intramolecular hydrogen bond
between two of the three nitrogen atoms (see chemical
structure in Table 1).
19F NMR Analysis of the Methanol Extracts. The 19F liquid-
state NMR spectra of the soil methanol extracts are shown
in Figure 3. The chemical shift for trifluralin relative to CFCl3
(0.000 ppm) was -61.994 ppm (Figure 4). The Pope soil 19
F
NMR spectrum showed only one peak (-61.989 ppm) that
represented unaltered herbicide (Figure 3A). Resonances for
residual trifluralin also appeared in the spectra for Hager-
stown soil (-61.995 ppm, Figure 3B) and Carlisle soil (-61.990
ppm, Figure 3D), but the spectrum for the Chagrin soil did
not show a trifluralin resonance (Figure 3C). All spectra,
except that for Pope soil, showed additional resonances that
represented trifluralin metabolites. Several common reso-
nances appeared at -60.868 to -60.874, -62.703 to -62.710,
and -60.699 to -60.706 ppm (Figure 3B-D). However, none
of these resonances matched the 19F peaks generated by the
five standards from Dow AgroSciences (Table 2 and Figure
4).
19F NMR Analysis of Humic and Fulvic Acids. The 19F
solid-state NMR spectra of humic acids isolated (after
incubation) from the Chagrin and Carlisle soils are presented
in Figure 5 along with the spectra of trifluralin and the mixture
of trifluralin with a control humic acid (from Chagrin soil).
When Fe-TR was mixed with FA, a 19F signal at -62.676
ppm gradually disappeared over a 3-week contact time, giving
rise to an emergence and a gradual increase in the size of a
broad signal at -61.150 (Figure 7A). Similarly, a gradual
disappearance of an 19F signal (at -61.450 ppm) and a gradual
buildup of a new 19F signal (at -60.215 ppm) was observed
9
6648 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004

FIGURE 3. 19F NMR liquid-state spectra of methanol extracts from soils incubated for 7 weeks with trifluralin under anoxic conditions:
(A) Pope soil, (B) Hagerstown soil, (C) Chagrin soil, and (D) Carlisle soil.
for the Dow AgroSciences standard 2,6-diaminotrifluralin
(TR6) when the latter was mixed with FA (Figure 7B).
In the absence of FA, T1 values for both Fe-TR (1192 ms)
and TR6 (1831 ms) were much greater than those for Fe-TR
(151.9 ms) or TR6 (248 to 410 ms) mixed with FA (Figure 8).
Chloroform extracts from FA complexes generated 19F three
signals at -60.166, -60.284, and -61.347 ppm (for Fe-TR/
FA) and seven signals at -60.269 to -60.856 ppm (for TR6/
FA) but did not generate any signals corresponding to free
Fe-TR or TR6 (data not shown). The chloroform-extracted
FA solutions showed some changes in their 19F NMR features
(data not shown) as compared to 19F signals collected before
extraction. The changes included reduced sizes of the major
TR6/FA peaks (centered around -61.438 and -60.213 ppm)
and, in the case of the Fe-TR/FA, the appearance of new
small signals (-60.789 and -61.525 ppm) on either side of
the main peak (-61.191 ppm).
of the initial 14C) remained unaltered, and only 9% occurred
in the bound form at the completion of the 7-week incubation
(Figure 1). In contrast, the Chagrin soil showed no free
trifluralin (Figure 2), but it produced considerable amounts
of methanol-extractable transformation products (58%) and
bound material (40%) (Figure 1). Similarly, as a result of
considerable transformation and binding (Figure 1), only
small amounts of free trifluralin were extracted from the
Carlisle soil (Figure 2). The differences may have been a result
of different soil properties (Table 2). Carlisle soil that had the
highest organic matter content (7.4%), clay content (35.2%),
and cation exchange capacity (19.9 mequiv/100 g) showed
the greatest amount of bound residues (53%) despite the
presence of some unaltered herbicide (Figure 2). In contrast,
Pope soil, which had less organic matter (3.3%), more sand
(55.9%), less clay (18.7%), and smaller cation exchange
capacity (9.8 mequiv/100 g) than Carlisle soil, was the least
effective in trifluralin transformation and binding (Figures
1 and 2). Chagrin soil with its relatively low organic matter
content (2.8%) and high pH (7.6) was the most effective in
trifluralin transformation, but not as effective in binding as
Carlisle soil (probably due to increased microbial activity as
compared to other soils). The complete disappearance of
Discussion
The soils under investigation differed significantly in their
ability to transform and immobilize trifluralin under anoxic
incubations. On the basis of radioscanning of the TLC plates
(Figure 2), most of trifluralin incubated with Pope soil (80.9%
9
VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 6649

FIGURE 4. 19F NMR liquid-state spectra of trifluralin and five metabolite standards dissolved in methanol-d4.
trifluralin in Chagrin soil might be due to increased microbial
activity as a result of an optimal pH (7.6) (29).
groups and a heterocyclic benzimidazole (3). In the study of
Golab et al. (3), the same compounds were also formed,
although in smaller amounts, under oxic growth conditions,
and it was believed that some of these compounds were
involved in the formation of bound residues.
The 19F liquid-state NMR analysis confirmed TLC plate
findings of the presence of only unaltered trifluralin (with
chemical shift at -61.989 ppm) in the methanol extract from
Pope soil (Figure 3A) and the absence of trifluralin (no
corresponding peak in the NMR spectrum) in the Chagrin
soil extract (Figure 3C). Trifluralin signals for Hagerstown
and Carlisle soils (-61.995 and -61.990 ppm, respectively)
(Figure 3B and D) were considerably reduced relative to that
for Pope soil (Figure 3A), which is in agreement with TLC
radioscans (Figure 2).
The chemical shifts of metabolite standards (listed as TR2,
TR6, TR15, 4-ABT, and 2-NITRO in Table 2) (Figure 4) did
not match any of the resonances shown by the methanol
extracts. This suggested that, if any of them were formed,
they underwent covalent binding and thus could not be
extracted. The possibility of covalent binding was supported
by solid-state 19F NMR analysis. The NMR spectra were
acquired with magic angle spinning (MAS) one-pulse experi-
ments, without cross polarization or proton decoupling.
This was to obtain signals for all 19F present in the samples
experiments with cross polarization magic angle spinning
(CP-MAS) would only show 19F close enough to H to be cross
polarized. Humic acid samples from Carlisle and Chagrin
soils, analyzed by 19F solid-state NMR, generated large, broad
resonances at -60.8 and -60.2 ppm (Figure 5). Prior to NMR
analysis, the samples were exhaustively extracted with
methanol and subjected to alkaline hydrolysis and 48-h
dialysis; therefore, the resonances could be attributed to
trifluralin residues covalently associated with organic matter.
An additional indication of covalent interactions was the
width of NMR signals (Figure 5). In the absence of humic
acid, trifluralin standard showed a sharp well-resolved peak
at -63.7 ppm (Figure 5 top left). When the standard was just
mixed with the control humic acid, trifluralin peak (at -63.4
ppm) remained sharp and well-resolved (Figure 5 top right).
In contrast, humic acid samples extracted from the Chagrin
and Carlisle soils with incorporated reduced forms of
trifluralin showed broad resonances (Figure 5 bottom). Line
broadening is in general a strong indication of covalent
In NMR spectra (Figure 3), methanol-extractable trans-
formation products with Rf values of 0 and 0.51, 0.56, or 0.61
(as determined by radio-scanning of the TLC; Figure 2)
showed many signals representing at least 14 different
compounds with chemical shifts ranging from -60.656 to
-62.721 ppm. The signals were very narrow and well resolved,
suggesting that they represented free trifluralin metabolites
rather than trifluralin metabolites complexed with methanol-
soluble components of soil organic matter. Golab et al. (3)
reported 31 extractable trifluralin metabolites formed in soil
under field conditions, 28 of which were identified. The
identified products belonged to four categories: (i) products
with the dealkylated amino group and/or reduced nitro
groups, (ii) heterocyclic benzimidazole products, (iii) azoxy
and azo dimeric products, and (iv) miscellaneous products
of oxidation/hydroxylation (3). The latter category is probably
not represented by any of the 19F signals collected in this
study for methanol extracts, because the experiments were
carried out under anoxic conditions. The anoxic incubation
in the study of Golab et al. (3), created by flooding the soil,
resulted in the formation of three major products including
compounds with one or two nitro groups reduced to amino
9
6650 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004

FIGURE 5. 19F NMR solid-state spectra of trifluralin, trifluralin mixed with the dialyzed control HA (from Chagrin soil) at a ratio 1:10 (5
mg/50 mg), and the dialyzed HA from Carslisle and Chagrin soils containing bound residues of trifluralin.
interactions between xenobiotics and humic materials (30).
Due to the heterogeneous nature of humic acid, bound
xenobiotic molecules assume a multitude of slightly different
structural conformations, showing broad, composite NMR
signals, as was the case for bound residues of trifluralin in
this study.
broadening (by paramagnetic substances) of this signal as
compared to the trifluralin peak collected in the absence of
humic acid and the appearance of the semi-symmetrical
sidebands next to this signal suggested weak interactions
between the herbicide molecule and humic acid (e.g., van
der Waals forces or hydrogen bonding) (30).
The chloroform-extracted fulvic acid from the Chagrin
soil also generated large, broad resonances (at -60.598 and
-61.135 ppm) (Figure 6), which seemed to represent reduced
trifluralin products involved in covalent interactions with
organic matter. The presence of two major resonances, rather
than one, as was the case for humic acid, indicated that the
fulvic acid-bound residues of trifluralin existed in two
distinctly different chemical environments or underwent
binding via two distinctly different reduced metabolites of
trifluralin. It should be emphasized that the resolution of
liquid-state 19F NMR is superior to solid-state 19F NMR so
that broad resonances observed in the solid-state spectra
may represent more than one trifluralin product.
No free trifluralin products could contribute to the signals
in Figures 5 and 6, because all free material was removed
during the 48-h dialysis of humic acids or by chloroform
extraction of fulvic acid. Chloroform extraction of fulvic acid
was used to extract free residues of the herbicide and was
unlikely to remove any trifluralin residues that were covalently
bound to fulvic acid. In the study of Williams and Feil (32),
for instance, chloroform extraction of rumen microbial media
incubated with 14C-labeled trifluralin resulted in the recovery
of 88-93% of radioactivity that represented only free residues
of the herbicide. Free xenobiotics usually generate sharp and
well-resolved NMR peaks (19), similar to the signal generated
by the trifluralin standard (-63.4 ppm) that was mixed with
humic acid shortly before the analysis (Figure 5). A slight
As already mentioned, at long contact times (weeks to
months rather than hours), a given compound may react
with different functional groups present in humic acid and
generate a large number of overlapping resonances that
appear on the NMR spectra as one broad resonance (17, 19).
Two smaller broad resonances (at ∼-30 and -95 ppm) that
appeared in the humic acid spectra (Figure 5) resembled the
sidebands shown by the mixture of trifluralin with the control
humic acid, but they probably represented trifluralin me-
tabolites involved in covalent binding rather than weak
interactions. This is based on the expectation that extended
dialysis and acid-base treatment should have removed all
metabolites held by weak interactions leaving only covalently
bound metabolites.
Experiments involving reactions between reduced triflu-
ralin products (Fe-TR and TR6) and fulvic acid (Figure 7)
supported the hypothesis that the broad resonances gener-
ated by solid and liquid soil fractions resulted from covalent
binding of amino metabolites of trifluralin, probably through
nucleophilic addition to quinone moieties of humic or fulvic
acids. As already mentioned, Fe-TR was obtained by incu-
bating trifluralin with zero-valent iron nanoparticles. It
generated 19F resonances (at -62.653 and -62.612 ppm; data
not shown) that did not match any of the 19F resonances
generated by the methanol extracts (Figure 3), which
suggested that Fe-TR-like products (2,6-diamines) might be
involved in bound residue formation. On the other hand, the
9
VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 6651

FIGURE 6. 19F NMR liquid-state spectrum of the chloroform-extracted Chagrin fulvic acid containing bound residues of trifluralin.
reaction with the palladium zero-valent iron produced a 2,6-
diamine derivative that according to MS analysis had the
same molecular weight as the zero-valent iron derivative but
showed a different chemical shift (at -60.879 ppm; data not
shown), probably because of an intramolecular hydrogen
bond between two of the three nitrogen atoms (see chemical
structure in Table 1). The methanol extracts from three soils
showed the presence of trifluralin products with 19F reso-
nances (Figure 3) matching closely that of Pd-Fe-TR (-60.879
ppm), which suggested that the intramolecular hydrogen
bonding made 2,6-diamine derivatives unreactive and thus
not capable of binding.
The gradual reduction in the size of Fe-TR and TR6
resonances (at -62.446, -62.635, and -62.676 ppm and at
-61.450 ppm, respectively) and the gradual buildup of new
broad signals (at -61.150 ppm and at -60.215, -60.414,
-60.441, and -61.796 ppm, respectively) during the 3-week
incubation of Fe-TR and TR6 with fulvic acid (Figure 7)
supported the hypothesis that 2,6-diamine products of
trifluralin metabolism are formed and are bound to humic
material with time. Aromatic amines have been shown to
react quickly but reversibly with soil organic matter via imine
bonds and slower but irreversibly via nucleophilic addition
of aromatic amines to soil organic matter quinone moieties
(28). The unextractable nature of the FA complexes, formation
of broad 19F resonances, and growth of these resonances
with time are consistent with the formation of covalent bonds.
The reduced spin-lattice T1 values for Fe-TR and TR6
that were complexed with fulvic acid as compared with those
determined for Fe-TR and TR6 analyzed in the absence of
humic acid (Figure 8) are consistent with strong covalent
interactions between diamino derivatives of trifluralin and
soil organic matter. Measuring spin-lattice (T1) or spin-
spin (T2) relaxation times has been used in the past to examine
xenobiotic interactions with humic materials. Nanny et al.
(31), for instance, used 13C NMR T1 relaxation to investigate
noncovalent interactions between acenaphthenone and
fulvic acid samples in aqueous solutions. Their approach
was based on the assumption that nuclei of xenobiotics
associated with humic material through either strong linkages
(covalent binding) or weak linkages (hydrogen bonding,
sorption, sequestration) should relax faster than those of
free xenobiotics. Depending on the concentration of the
reaction components, the T1 of acenaphthenone adsorbed
to fulvic acid ranged between 24 and 26 s, whereas the T1 of
free acenaphthenone was about 30 s. In this study, the T1
measured for Fe-TR or TR6 in the absence of fulvic acid were
1192-1518 and 1831 ms, respectively, and they were reduced
to 152 and 248-410 ms, respectively, when fulvic acid was
added (Figure 8). From this observation, we can conclude
that interactions occurred between FA and reduced me-
tabolites of trifluralin, although interactions may not neces-
sarily be covalent.
Dixon et al. (22) used 19F NMR T1 and T2 relaxation
measurements to investigate interactions between 4′-fluoro-
1′-acetonaphthone and Suwannee River fulvic acid (SRFA).
Addition of 4.5 mg of SRFA mL-1 reduced T1 relaxation times
from 1781 to 555 ms, with observed broadening of resonances.
However, Dixon et al. (22) attributed interactions to sorption
either by hydrophobic interactions or hydrogen bonding. In
this investigation, similar reductions in T1 relaxation mea-
surements are reported (Figure 8).
After mixing with fulvic acid, the Fe-TR signal moved from
-62.007 ppm (T1 1192 ms) for Fe-TR alone to -61.188 ppm
(T1 152 ms) after a 66-h contact with fulvic acid and showed
considerable line broadening (Figure 8), demonstrating
9
6652 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004

FIGURE 7. 19F NMR liquid-state spectra of (A) Fe-TR metabolite and (B) TR6 metabolite involved in binding to Chagrin fulvic acid.
xenobiotic-humic associations consistent with covalent
binding. Line broadening could be due to heterogeneity of
environments in which the Fe-TR was immobilized. In the
case of TR6, the initial signals at -61.510 ppm remained in
the spectrum (at -61.438 ppm) even after the addition of
fulvic acid (Figure 8). This peak decreased and considerably
broadened with time, as new broad resonances appeared
maximizing at -60.213 and -61.793 ppm (Figure 8). It was
expected that T1 measurements for these new peaks would
be different than those for the remainder of the original peak;
however, all relaxation times were in a narrow range of 248
to 337 ms. This demonstrated that, in some cases, it is not
possible to distinguish whether changes in T1 arise from either
covalent or noncovalent binding interactions.
Chloroform extracts generated no 19F signals for free Fe-TR
or TR6 (at -62.007 and -61.510 ppm as shown in Figure 8,
panels A and C, respectively), suggesting the covalent nature
of associations between these compounds and fulvic acid.
Three chemical shifts at -60.166, -61,284, and -61.347 ppm
(for Fe-TR chloroform extract) and seven signals ranging from
-60.269 to -60.856 ppm (for TR6 chloroform extract) can
probably be attributed to Fe-TR and TR6 associated with
chloroform-soluble fulvic acid components.
The chloroform extraction of the Fe-TR/FA (Fe-TR
metabolite incubated with FA) resulted in a reduction of the
peak at -61.188 ppm (shown for the unextracted complex
in Figure 8B), and the appearance of two new peaks at -60.789
and -61.525 ppm (data not shown) that probably resulted
from changes in the structural arrangement of the chemically
bound Fe-TR molecules after the removal of the chloroform-
soluble fulvic acid complex.
In an attempt to verify the covalent nature of bonds, the
Fe-TR and TR6 complexes with fulvic acid (identified based
on T1 relaxation measurements) were extracted with chlo-
roform and again analyzed by 19F NMR (data not shown).
9
VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 6653

FIGURE 8. 19F NMR liquid-state spectra and the respective spin-lattice relaxation times (T1) of (A) Fe-TR metabolite, (B) Fe-TR with Chagrin
FA, (C) TR6 metabolite, and (D) TR6 metabolite with Chagrin FA.
As a result of chloroform extraction of the TR6/FA
complex, the large sharp peak at -61.438 ppm (shown for
the unextracted complex in Figure 8D) decreased somewhat
(more in height rather than in size or peak area); and the
center of the peak at -60.213 ppm shifted to -60.249 ppm
(data not shown). In view of this relatively minor modification,
the latter peak probably represented reduced trifluralin
molecules that were covalently attached to fulvic acid. The
peak at -61.438 ppm probably represented the unaltered
TR6 molecules that may have been retained by strong
electrostatic attraction between the negative charges of fulvic
acid and the positive charges of the protonated amino groups
and, therefore, resisted chloroform extraction.
amino derivatives that in turn interact with soil organic
matter. The differences in the location of 19F signals in the
NMR spectra could represent varying numbers and positions
of amino groups (3-, 4-, and 5- relative to the CF3 group) in
different transformation products.
The 19F NMR technique shows great potential for inves-
tigating the fate of important fluorinated pollutants in soil
environments. If combined with a sufficient number of
reference standards and traditional approaches (e.g., using
14C-labeled xenobiotics), it could quickly provide reliable
information on both transformation pathways and the
mechanisms of binding.
Acknowledgments
In general, broad resonances generated by the fulvic acid
complexes appeared in the same regions of the spectra as
those collected for the fulvic acid fraction of the incubated
soil (Figure 6). These and other NMR data (lack of extractable
amino metabolites, broad 19F resonances in FA and HA from
trifluralin/soil incubations) collected in this study can be
considered a strong confirmation of the hypothesis that
trifluralin binding in soil proceeds through the formation of
This research was supported by the Office of Research and
Development, U.S. Environmental Protection Agency (Grant
R-826646). Although this work was reviewed by the U.S. EPA
and approved for publication, it may not necessarily reflect
official Agency policy. Partial support was also provided by
Penn State Biogeochemical Research Initiative for Education
(BRIE) sponsored by NSF (IGERT) Grant DGE-9972759.
9
6654 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004

Sci. Technol. 1997, 45, 514-520.
Literature Cited
(20) Chien, Y.-Y.; Kim, E.-G.; Bleam, W. F. Paramagnetic relaxation
of atrazine solubilized by humic micellar solutions. Environ.
Sci. Technol. 1997, 31, 3204-3208.
(21) Chien, Y.-Y.; Bleam, W. F. Fluorine-19 nuclear magnetic
resonance study of atrazine in humic and sodium dodecyl sulfate
micelles swollen by polar and nonpolar solvents. Langmuir 1997,
13, 5283-5288.
(22) Dixon, A. M.; Mai, M. A.; Larive, C. K. NMR investigation of the
interactions between 4′-fluoro-1′-acetonaphthone and the
Suwannee River fulvic acid. Environ. Sci. Technol. 1999, 33,
958-964.
(23) Ngim, K. K.; Mabury, S. A.; Crosby, D. G. Elucidation of fipronil
photodegradation pathways. J. Agric. Food Chem. 2000, 48,
4661-4665.
(1) Grover, R.; Wolt, J. D.; Cessna, A. J.; Schiefer, H. B. Environmental
fate of trifluralin. Rev. Environ. Contam. Toxicol. 1997, 153, 1-64.
(2) USGS. The USGS National Water Quality Assessment and
Pesticide National Synthesis Project; 2002; http://ca.water.usgs-
.gov/pnsp/use92/triflrln.html.
(3) Golab, T.; Althaus, W. A.; Wooten, H. L. Fate of [14C] trifluralin
in soil. J. Agric. Food Chem. 1979, 27, 163-179.
(4) Malterre, F.; Pierre, J.-G.; Schiavon, M. Trifluralin transfer from
top soil. Ecotoxicol. Environ. Saf. 1998, 39, 98-103.
(5) Parr, J. F.; Smith, S. Degradation of trifluralin under laboratory
conditions and soil anaerobiosis. Soil Sci. 1973, 115, 55-63.
(6) Willis, G. H.; Wander, R. C.; Southwick, L. M. Degradation of
trifluralin in soil suspensions as related to redox potential. J.
Environ. Qual. 1974, 3, 262-265.
(7) Tor, J. M.; Xu, C.; Stucki, J. M.; Wander, M. M.; Sims, G. K.
Trifluralin degradation under microbiologically induced nitrate
and Fe(III) reducing conditions. Environ. Sci. Technol. 2000,
34, 3148-3152.
(24) Ellis, D. A.; Mabury, S. A. The aqueous photolysis of TFM and
related trifluoromethylphenols. An alternate source of trifluo-
roacetic acid in the environment. Environ. Sci. Technol. 2000,
34, 632-637.
(25) Martin, J. W.; Franklin, J.; Hansen, M. L.; Solomon, K. R.; Mabury,
S. A.; Ellis, D. A.; Scott, B. F.; Muir, D. C. G. Detection of
chlorofluoroacetic acid in precipitation: a possible product of
fluorocarbon degradation. Environ. Sci. Technol. 2000, 34, 274-
281.
(26) Kohl, S. D.; Toscano, P. J.; Hou, W.; Rice, J. A. Solid-state 19F
NMR investigation of hexafluorobenzene sorption to soil organic
matter. Environ. Sci. Technol. 2000, 34, 204-210.
(27) Cornelissen, G.; Van Noort, P. C. M.; Nachtegaal, G.; Kentgens,
A. P. M. A solid-state fluorine-NMR study on hexafluorobenzene
sorbed by sediment, polymers, and active carbon. Environ. Sci.
Technol. 2000, 34, 645-649.
(28) Parris, G. E. Covalent binding of aromatic amines to humates.
1. Reactions with carbonyls and quinones. Environ. Sci. Technol.
1980, 14, 1099-1106.
(29) Dibble, J. T.; Bartha, R. Effect of environmental parameters on
the biodegradation of oil sludge. Appl. Environ. Microbiol. 1979,
37, 729-739.
(30) Achtnich, C.; Fernandes, E.; Bollag, J.-M.; Knackmuss, H.-J.;
Lenke, H. Covalent binding of reduced metabolites of [15N3]TNT
to soil organic matter during a bioremediation process analyzed
by 15N NMR spectroscopy. Environ. Sci. Technol. 1999, 33, 4448-
4456.
(31) Nanny, M. A.; Bortiatynski, J. M.; Hatcher, P. G. Noncovalent
interactions between acenaphthenone and dissolved fulvic acid
as determined by 13C NMR T1 relaxation measurements. Environ.
Sci. Technol. 1997, 31, 530-534.
(8) Jacobson, S. N.; O’Mara, N. L.; Alexander, M. Evidence for
cometabolism in sewage. Appl. Environ. Microbiol. 1980, 40,
917-921.
(9) Crosby, D. G.; Lettis, E. The photodecomposition of trifluralin
in water. Bull. Environ. Contam. Toxicol. 1973, 10, 237-241.
(10) Lettis, E.; Crosby, D. G. Photodecomposition of trifluralin. J.
Agric. Food Chem. 1974, 22, 842-848.
(11) Mabury, S. A.; Crosby, D. G. 19F NMR as an analytical tool for
fluorinated agrochemical research. J. Agric. Food Chem. 1995,
43, 1845-1848.
(12) DOW Agrosciences. Trifluralin pesticide Fact Sheet. 2002;
http://www.dowagro.com/webapps/lit/litorder.asp?objid)-
09002f138016bbd5&filepath)/noreg.
(13) Pignatello, J. J.; Xing, B. Mechanisms of slow sorption of organic
chemicals to natural particles. Environ. Sci. Technol. 1996, 30,
1-11.
(14) Dec, J.; Bollag, J.-M. Determination of covalent and noncovalent
binding interactions between xenobiotic chemicals and soil.
Soil Sci. 1997, 162, 858-874.
(15) Sparks, D. L. Kinetics and mechanisms of chemical reactions
at the soil interface. In Soil Physical Chemistry, 2nd ed.; Sparks,
D. L., Ed.; CRC Press: Boca Raton, FL, 1999; pp 135-191.
(16) Strynar, M. J.; Dec, J.; Bollag, J.-M. Anaerobic/aerobic composting
of soil contaminated with 2,4,6-trinitrotoluene. Biorem. J. 2002,
6, 177-190.
(17) Thorn, K. A.; Pettigrew, P. J.; Goldenberg, W. S.; Weber, E. J.
Covalent binding of aniline to humic substances. 2. 15N NMR
studies of nucleophilic addition reactions. Environ. Sci. Technol.
1996, 30, 2764-2775.
(18) Hatcher, P. G.; Bortiatynski, J. M.; Minard, R. D.; Dec, J.; Bollag,
J.-M. Use of high-resolution 13C NMR to examine the enzymatic
covalent binding of 13C-labeled 2,4-dichlorophenol to humic
substances. Environ. Sci. Technol. 1993, 27, 2098-2103.
(19) Dec, J.; Haider, K.; Benesi, A.; Rangaswamy, V.; Scha¨ffer, A.;
Plu¨cken, U.; Bollag, J.-M. Analysis of soil-bound residues of the
13C-labeled fungicide cyprodinil by NMR spectroscopy. Environ.
(32) Williams, P. P.; Feil, V. J. Identification of trifluralin metabolites
from rumen microbial cultures. Effect of trifluralin on bacteria
and protozoa. J. Agric. Food Chem. 1971, 19, 1198-1204.
Received for review January 2, 2004. Revised manuscript
received September 19, 2004. Accepted September 20, 2004.
ES0403110
9
VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 6655