transport of triticonazole in homogeneous soil columns

Transport of Triticonazole in Homogeneous Soil Columns:Influence of Nonequilibrium Sorption

C. Beigel and L. Di Pietro*

ABSTRACTNonequilibrium sorption of pesticides is frequently reported to

greatly affect their transport and dissipation in soil. This study wasaimed at evaluating the performances of equilibrium and two site-tworegion nonequilibrium convective-dispersive models for describingthe sorption and decay characteristics during transport of triticonazolesystemic fungicide in water-saturated homogeneous soil. Chloride anduC-triticonazole column displacement experiments were carried outin a loamy clay soil under steady-state water flow at high pore watervelocities. The symmetrical breakthrough curves (BTC) obtained withthe conservative tracer showed no significant physical nonequilibriumand were used to estimate a dispersivity of 0.06 cm. Compared withchloride, the 14C-triticonazole BTC was strongly asymmetrical andshifted to the right, with a broad, extended tailing characteristic ofsorption nonequilibrium. Chemical analysis of the soil after elutionshowed that bound residues were rapidly formed during transport.These bound residues were accounted for as decayed in the models.The two-site model correctly described the first part of the tailing, withan estimated partition coefficient Ka of 1.5 L kg"1 for instantaneoussorption, and it predicted high values in the range of 58 d~', and 7d"1 for the sorption and decay first-order rates, respectively. However,the model failed to describe the slower, extended release of 14C-triticonazole. Nonequilibrium sorption and formation of bound resi-dues of triticonazole were attributed to the rate-limiting diffusiveprocess. It was thus concluded that use of a single first-order rateconstant for description and prediction of both nonequilibrium sorp-tion and dissipation of triticonazole in soil is not appropriate.

TRITICONAZOLE [(lRS)-(E)-5-(4-chlorophenyl-meth-ylen)-2,2-dimethyl-l-(lH-l,2,4-triazol-l-ylmethyl)-

cyclopentan-1-ol] is a new triazole systemic fungicide,developed by Rhone-Poulenc Agro (Courbevoie,France), that is used in cereal seed treatments. Tritico-nazole controls major seedborne, foliar, and straw dis-eases, thus allowing for cereal protection from seed todeveloped growth stages. The efficacy of systemic pesti-cides applied in seed treatments depends closely on theirdissipation, localization, and availability in the soil pro-file in relation to their uptake by the plant root system.The transport and fate of these molecules in soils is,therefore, of crucial interest for an optimal utilization.

Solute transport of organic chemicals in soil dependson the soil's structural and hydraulic properties, and itis controlled by sorption and degradation, which bothlimit the mobility of the pesticides in soil. These majordissipation processes have been extensively studied(Graham-Bryce, 1981; Weber and Miller, 1989; Weber,1991). The difficulty of accurately accounting for thesetwo interacting processes with simple input parametersis one of the main problems encountered in predictingand modeling solute transport in soil (Calvet, 1995).

C. Beigel and L. Di Pietro, Unite de Science du Sol, I.N.R. A., DomaineSt Paul, Site Agroparc, 84914 Avignon Cedex 9, France. Received 30June 1998. *Corresponding author ([email protected]).

Published in Soil Sci. Soc. Am. J. 63:1077-1086 (1999).

Sorption of hydrophobic organic compounds (HOC)has been related to soil organic matter (OM) throughnonspecific interaction mechanisms (Hamaker andThompson, 1972; Ainsworth et al., 1989; Barriuso andCalvet, 1992). Soil sorption isotherms of HOC at dilutedconcentrations are usually linear (Calvet, 1989), and asoil-solution partition coefficient Kd (in L kg^1 forsorbed concentration 5e divided by solution concentra-tion Ce at equilibrium) is frequently used to characterizeHOC sorption on a particular soil. These coefficients areexperimentally determined in batch systems, assumingreversible, quasi-instantaneous sorption. On the otherhand, there is increasing evidence of nonreversible andtime-dependent sorption of most pesticides in soil (Leh-mann et al., 1990; Scribner et al., 1992, Barriuso et al.,1992). Sorption kinetics for HOC usually exhibit a two-stage approach to equilibrium, with a short initial phaseof rapid uptake followed by an extended period of muchslower uptake (Lee et al., 1988; Gaston and Locke,1995).

The degradation of pesticides in soil is mainly due tobiological transformations and is thus controlled by theavailability of the organic chemical and by the activityof the soil microflora (Torstensson, 1987). It is usuallycharacterized by first-order rate constants (k) calculatedfrom mineralization or dissipation data, using the first-order relation C = C0 e~kt, where C0 and C (mg L"1)are the substrate concentrations in soil at time 0 andt, respectively. However, degradation may not alwaysfollow simple first-order kinetics if the pesticide is sub-ject to significant abiotic degradation. Also, the degrad-ing capacity of the soil microflora may vary with time,because the growth and activity of the degrading micro-organisms are extremely sensitive to environmental con-ditions such as temperature and humidity (Walker etal. 1992, Veeh et al. 1996), and because adaptation phe-nomena may occur (Felsot and Shelton, 1993; Ou et al.,1993). Furthermore, changes in the pesticide availabilitywith time that are due to nonequilibrium sorption mayalso affect the degradation rate.

The sorption and degradation characteristics of triti-conazole systemic fungicide in a loamy soil of Grignon,France, (Typic Eutrochrept) have been examined inprevious studies (Beigel et al., 1997,1999). The degrada-tion of triticonazole was essentially due to microbial,cometabolic transformations that could be adequatelycharacterized by first-order mineralization rate con-stants ranging from 0.3 X 10~3 to 0.6 X 10~3 d"1, de-pending on the initial dose applied. Batch equilibriumstudies showed that triticonazole equilibrium sorption

Abbreviations: BTC, breakthrough curves; CDE, convective-dispersive equation; CDeq, CXTFIT 2.0 deterministic equilibriumCDE model; CDnoneq, CXTFIT 2.0 deterministic two-site, two-region nonequilibrium model; HOC, hydrophobic organic com-pounds; LEA, local equilibrium for solute adsorption; LSC, liquidscintillation counting; OM, organic matter; STD, standard deviation.

1077

1078 SOIL SCI. SOC. AM. J., VOL. 63, SEPTEMBER-OCTOBER 1999

was related to soil organic matter content and could beapproximated by a linear isotherm, with a measured KAof 4.35 L kg"1; however, it appeared that triticonazolesorption into Grignon soil during incubation and diffu-sion experiments was strongly time-dependent and in-creased as the time of contact with the soil increased.A time-dependent, apparent partition coefficient wasmeasured, which increased from 2.5 to 10 L kg"1 duringa 130-d incubation. Rate-limiting desorption, and for-mation of methanol nonextractable, bound residueswere observed, which resulted in a decreasing apparentdiffusion rate of triticonazole in soil. This was attributedto rate-limiting intrasorbent (organic matter) diffusionto restricted sorption sites. Use of the batch equilibriumKa to characterize triticonazole sorption and desorptionduring transport would thus prove erroneous, and non-equilibrium sorption conditions need to be accounted for.

The objective of the present study was to test simpleconceptual coupled-process models to describe the one-dimensional transport of triticonazole in homogeneoussaturated soil columns. We evaluated the deterministicequilibrium convective-dispersive equation (CDE) andthe two-site, two-region, deterministic, nonequilibriumCDE models using the computer program CXTFIT 2.0of Toride et al. (1995), which is an updated version ofthe CXTFIT code of Parker and van Genuchten (1984a).

MATERIALS AND METHODSSoil and Chemicals

The soil used in this study was a loamy clay (Typic Eutro-chrept), sampled in the surface layer (0-20 cm) of a continuous

WATER TRACER(CaClj or "C-triliconazole)

Supply tanks

Waste coll

Mi

*/ &1 3-way valve

£

iz:6.5 cm n

Ah i

Soil column

\7effluent collector

SYSTEM OF REGULATION OF THE INFILTRATION RATEBY CONSTANT PRESSURE HEAD Ah:

O reservoir inlet© reservoir outlet© overflow outletO column inlet© column outlet

wheat experimental plot located at Grignon, France. It had apH in water of 8.2, with the composition: 29.1% clay, 54.0%silt, 14.5% sand, and 1.04% organic C. Soil samples were airdried and passed through a 2-mm sieve. To avoid problemsof clogging while packing the columns and during elution, thefiner particles were removed by further sieving at 0.5-mm.Soil water content (kg kg^1) was 0.043%.

Carbon-14-U-benzyl-labeled triticonazole (specific activity:1184 MBq minor1; radiopurity >98%) was provided byRhone-Poulenc Agrochemicals, Lyon, France. Triticonazolewater solubility is 8.4 mg L"1 and 10.6 mg L"1 at 20 and 22°C,respectively, and the distribution coefficient between octanoland water is 1950. A solution of I4C-triticonazole at 5.0 mgL'1 and 8.332 Kbq mLr1 (0.224 (jiCi mL"1) was preparedfor input tracer solution by adding 14C-triticonazole methanolstock solution to a saturated water solution of triticonazole(analytical standard, purity >92%) and adjusting the concen-tration with MilliQ water (Millipore, Saint-Quentin, France).The solution concentrations were measured at 262.5 nm witha UV-visible spectrophotometer Lambda V (Perkin-Elmer,Uberlingen, Germany).1

Chloride was used as a nonreactive, conservative tracerto determine the soil hydrodynamic dispersion coefficient atdifferent water velocities. A solution of CaCl2 with Cl~ at1 g L"1 was prepared for input tracer solution by dilutingCaCl2,2H2O (analytical reagent >98% purity; R.P. Normapur,Prolabo, Paris, France) in the proper amount of MilliQ water.

Miscible Displacement ExperimentsThe ETC for CaCl2 and triticonazole were measured in

water saturated, isotropic homogeneous soil columns. Thesystem consisted of PVC columns 65 mm in length (L) and55 mm diam. A stainless steel porous filter of 50-|jim meshwas used as bed support on the bottom of the column. At thetop of the column, a void volume 5 mm deep was providedas a mixing cell to allow for the formation of a piston-likewater front. The columns were packed under water saturationconditions in a water bath by adding successive layers of soil toestablish uniform bulk density and water content. The columnswere then covered to avoid evaporation and allowed to equili-brate in the water bath for 24 h at 22 ± 2°C. Measured in 10test columns, mean gravimetric water content (Qg), volumetricwater content (6V), and bulk density (Mg irT3) were 0.65 ±0.01, 0.60 ± 0.01, and 1.07 ± 0.01, respectively. The porevolume (V0) was calculated as the product of the columnvolume and 0» at 92.7 cm3. Soil saturated hydraulic conductivity(A"Sat), measured in similar soil conditions, was of 97.2 cm d"1.

A constant pressure head (A/z) was applied on the soilcolumns for steady-state flow conditions, using the systemshown in Fig. 1. The desired pore water velocities were ob-tained by varying the level of the column relative to the over-flow outlet. The elution fluxes at the columns outlet (Q, cm3

d"1) were measured during the elution experiments byweighing the amount of solution eluted at regular time inter-vals. The Darcy's velocities (q, cm d"1) were evaluated bydividing Q by the cross sectional area of the soil column (23.76cm2). We observed a good agreement between the measuredand theoretical calculated Darcy's velocities for different Aft.All the elution experiments were performed in duplicate at22 ± 2 °C. For the chloride-miscible displacement studies,three different q of 125,160, and 212.5 cm d"1, correspondingto A/* of 2, 4, and 8 cm were established, respectively. Theaverage pore water velocities (v) for the duplicates were calcu-lated from v = <7/6v at 208, 263, and 354 cm d"1, respectively.

Fig. 1. Schematic layout of the experimental system for solute trans-port with constant pressure head, AA.

1 Trade names and company names are included for the benefitof the reader and do not imply endorsement or preferential treatmentof the product listed by INRA.

BEIGEL & DI PIETRO: INFLUENCE OF NONEQUILIBRIUM SORPTION ON TRANSPORT IN SOIL 1079

The elution of triticonazole was performed at the lowest chlo-ride Darcian velocity of 125 cm d~'. The columns were firstsupplied with MilliQ water until steady-state flow conditionswere attained at the desired infiltration rate. A pulse of thetracer solution corresponding to 0.5 pore volume was thenapplied at the same rate, and the sytem was then switchedback to the water reservoir for elution periods of 3 h (chlorideBTC), and 11 h (triticonazole ETC). At regular time intervals,aliquot samples of the leachate were collected, weighed, andstored at —20°C until chemical analysis. At the end of theelution time course, sequential sampling of the triticonazolesoil columns was performed by extruding and slicing the soilin 10 incremental disks of 6-mm sections using a soil extrudingscrew procedure described in a previous paper (Beigel et al.1997).

Chemical AnalysisChloride concentration in the effluent samples was mea-

sured by capillary analysis by direct UV detection with asulfate-OFM-CHES electrolyte, using a CIA analyzerequipped with a AccuSep 75 (jum X 60 cm capillary and Millen-ium 2.1 software (Waters, Milford, MA).

Triticonazole concentration in the effluent samples wasmeasured by liquid scintillation counting (LSC). Aliquots ofthe elution samples (0.5 mL) were pipetted and put in scintilla-tion vials. Four milliliters of Ultima Gold XR LSC Cocktail(Packard Instrument, Meriden, CT) were added, and theamount of radioactivity was measured by LSC using a KontronBetamatic V counter (Kontron Instrument, Montigny le Bret-onneux, France).

Total 14C-triticonazole residues remaining in the soil sam-ples were measured by LSC of the 14CO2 evolved after combus-tion of triplicate 300-mg aliquots of air-dried and finely groundsoil with a Sample Oxidizer 307 (Packard Instrument, Meri-den, CT). Extractable residues were analyzed after exhaustiveextraction with methanol. The soil samples were extractedtwice with 50 mL of methanol. After 24 h of shaking, thesamples were centrifuged for 15 min at 5000 rpm (8000 g ) ,and the radioactivity content in the supernatant was measuredby LSC as previously described. After the second extraction,the soil pellets were air dried, and the remaining radioactivityin the soil was measured by combustion of triplicate 300-mgsoil aliquots as previously described.

ModelsCXTFIT 2.0 (Torride et al., 1995) is a program presenting

a number of analytical solutions for one-dimensional transportmodels based on the convection-dispersion equation (CDE).Assuming steady-state flow in a homogeneous soil and first-order transformation kinetics with uniformly distributed non-growing biomass, the equilibrium CDE may be written as

e dt dx [1]

where t is time (d), x is depth (cm), p is soil bulk density (gcm"3), 9 is soil volumetric water content (cm3 cm~3), C is theconcentration of the liquid phase (mg L~'), S is the concentra-tion of the adsorbed phase (mg kg"1), v is the average porewater velocity (cm d"1), D is the hydrodynamic dispersioncoefficient (cm2 d"1), and |julit| is a first-order decay coefficientfor degradation in the liquid phase (d~'). Depending on theequilibrium-nonequilibrium assumptions about S, we usedCXTFIT 2.0 deterministic equilibrium CDE model (Mode 1),further noted as CDeq, or CXTFIT 2.0 deterministic two-site,two-region nonequilibrium model (Mode 2), further notedas CDnoneq.

The CDeq model assumes local equilibrium (LEA) for sol-

ute adsorption and that sorption can be described by a singlelinear isotherm, 5e = KA Ce, where 5e and Ce are the concentra-tions in sorbed and liquid phases at equilibrium, and KA is theequilibrium partition coefficient (L kg"1).

The two-site, two-region bicontinuum model has been for-mulated to account for either sorption-related or transport-related nonequilibrium during solute transport. The two-sitenonequilibrium concept assumes that sorption sites in soils canbe classified into two fractions. In the first fraction, sorption isinstantaneous and is described by an equilibrium sorptionisotherm (Type 1, equilibrium). In the second fraction, sorp-tion is time-dependent and follows first-order kinetics (Type2, kinetic). In this case, the rate-limiting step for Type 2 siteswould be either chemical (chemisorption), or diffusive (intra-particle or intrasorbent diffusive mass transfer), as discussedby Brusseau et al. (1991). The two-region approach assumesthat the liquid phase can be partitioned into mobile (flowing,macropore domain) and immobile (stagnant, matrix, or micro-pore domain) regions. The exchange between the two liquidregions is modeled by a first-order kinetic equation. Flowoccurs only in the mobile region. Sorption is assumed to beinstantaneous on all sorption sites, and the sorption rate islimited here by the diffusion of the solutes to the exchangesites in the stagnant phase. If dimensionless parameters areemployed, then the two-site and two-region models reduce tothe same dimensionless form:

dC, dC,

[3]

where T = vtIL, Z = x/L, P is the Peclet number P = vLID,R is the retardation factor, defined as R = 1 + p/9 Kt. Thesubscripts 1 and 2 refer to equilibrium and nonequilibriumsites respectively, (3 is a fraction factor, and o> is a dimen-sionless, mass transfer coefficient. The various dimensionlessparameters have different meanings for the two-site and two-region models, which are defined in the CXTFIT 2.0 code(Toride et al., 1995).

For the two-site model, p and co are defined as

6 + fpKd p(9 +'y

- 9

co = , a = vcoRL(l - p)

[4]

[5]

where /is the fraction of Type 1 sites, and a (d"1) is the first-order rate for the kinetic, Type 2 sites. The dimensionlessdecay terms jo,], for degradation in the liquid phase only, re-duces to

M-i = ——3v

For the two-region model, (3 and to are defined as

9m + fpKa

e

[6]

[7]

where 9m is the volumetric water content of the mobile region,and/is the fraction of sorption sites in the mobile region; and

oL[81

where a is a first-order mass coefficient governing the rate ofsolute exchange between the mobile and immobile liquidregions.


Parameters EstimationThe hydrodynamic dispersion coefficient D of the soil was

estimated from the chloride data with both models using thenonlinear least-squares parameter optimization method (in-verse problem). The solute transport parameters v and D atthe three infiltration rates were first estimated from the chlo-ride ETC (pooled results from the duplicate columns) withthe CDeq model with R set to 1. The two-region approach fromthe CDnoneq was then used to check for potential physicalnonequilibrium. Nonequilibrium transport parameters D, a,and P were estimated for chloride nonreactive tracer, with Rset to 1, and v set to the values obtained from CDeq. Thevalues of D obtained from the chloride ETC were used toestimate the soil dispersivity, A. = Div. The dispersivity wasused to calculate the value of D for triticonazole ETC, whichwas then introduced in CDeq to estimate the retardation factorR, the dimensionless parameter for linear sorption, and thefirst-order degradation rate |ji,iq from the triticonazole ETC.CDnoneq was used to estimate the retardation factor, andthe nonequilibrium dimensionless parameters p, co, and (A fortriticonazole transport and decay.

Both models were used under the conditions of semi-infinitesystem, first-type (flux) boundary conditions, pulse input, lin-ear sorption and first-order degradation in the liquid phaseonly. All these conditions were assumed to reasonably applyto our elution column experimental setup (see Parker and vanGenuchten, 1984b) and to triticonazole sorption and degrada-tion characteristics. The corresponding initial and boundaryconditions, as well as some analytical solutions are detailedin the CXTFIT 2.0 code (Toride et al., 1995).

RESULTSChloride Breakthrough Curves

The ETC of chloride conservative tracer measuredat the three water velocities were almost identical (Fig.2). The effluent curves appeared symmetrical and sig-moidal, with invariant frontal and distal portions. No

significant retardation occured, as chloride was detectedin the effluent after application of 0.8 V/VQ of water,and relative concentration peaks were measured at 1.25

The curve fitting results of CDeq and of the two-region CDnoneq to the experimental chloride data arelisted in Table 1. The CDE fitted the observed resultswell, as indicated by the high correlation coefficient rvalues obtained (>0.97). Estimates of v with CDeq forthe duplicate columns at the three infiltration rates wereclose to the measured velocities; however, the estima-tion of the dispersion coefficients D was not satisfactory,as the standard deviation (STD) values were higher thanthe D values. Use of CDnoneq with the CDeq fittedvalues for v proved more efficient for the estimation ofD, as STD were much lower. Estimated values for Dwere very low and increased from 16 to 25 cm2 d^1

with increasing water velocity. As D appeared directlyproportional to v, the dimensionless Peclet number P =Lv/D, and the dispersivity X = Div were almost invari-ant, with mean values of 86.7 ± 5.5 for P, and 0.06 ±0.02 cm for X, respectively. The estimated parameter (3for physical nonequilibrium appeared dependent uponthe water velocity, with (3 decreasing from 1 to 0.92 withestimated v values decreasing from 355 to 217 cm d"1.

Triticonazole Breakthrough CurvesComparison of the breakthrough of chloride and triti-

conazole measured at a water velocity of 208 cm d^1

for a pulse of 0.5 V0 is shown in Fig. 3. TriticonazoleETC appeared shifted to the right. The relative concen-tration of triticonazole in the effluent was comparativelymuch lower than the conservative tracer concentration.A delayed concentration peak at a low C/C0 of 0.16 wasobtained after elution of ~3 pore volumes. Triticonazole

V/V0

Fig. 2. Effect of pore water velocity on the measured (symbols) and simulated (lines) breakthrough curves in Grignon soil for displacementthrough Grignon soil of a pulse of 0.5 K0 of chloride conservative tracer at 1 g L"1.


Table 1. Comparison of CDeq and CDnoneq parameter values estimated from the chloride data at three pore water velocities.Experimental conditions CDeq fit CDnoneq two-region fit

Column Measured v Pulse Pt

12

3

4

5

6

cmd-'202

215

252

275

362

346

v/va0.51

0.54

0.54

0.58

0.58

0.52

cm d ' cm2 d * cm2 d '

217 ± 54.5 15.6 ± 54.5 0.974 16.6 ± 0.1 0.92 ± 0.05 10~6 0.977

269 ± 54.1 22.0 ± 54.1 0.970 21.6 ± 0.1 0.98 ± 0.04 10~6 0.986

355 ± 42.3 24.6 ± 42.3 0.974 24.5 ± 0.1 1.00 ± 0.03 lO'6 0.988

t Estimated value ± standard deviation.

ETC was strongly asymetrical in shape (Fig. 4). Thedistal part of the elution peak appeared skewed andbiphasic, showing a rapid release in the early tailingfollowed by a much slower release in the extendedtailing.

Results of CDeq and CDnoneq fits with v set to 208cm d"1, and D = \v set to 12.48 cm2 d"1 are summarizedin Table 2. The equilibrium model proved inefficient indescribing the experimental data (Fig. 4). CDeq couldnot account for any of the asymmetry and tailing, andit could only describe the position of the delayed concen-tration peak; i.e. the retardation due to instantaneousequilibrium sorption. The CDeq fit without decay re-sulted in a peak twice as high as the observed data. Theintroduction of a first-order decay constant (sink term)improved the fit, as it allowed better description of thespreading of the elution peak. Use of the two-site non-equilibrium model considerably improved the fit to theexperimental ETC (r > 0.93). CDnoneq accounted forthe early tailing asymmetry of the concentration peak,but it still failed to account for the extended tailing of the

ETC (Fig. 5). The model predicted that all the appliedradioactivity would be recovered in the effluent afterelution of 12 pore volumes of water, whereas significantamounts of 14C were still detected in the effluent afterelution of the 16 pore volumes. CDnoneq without decayslightly overpredicted the width of the concentrationpeak. Introduction of the decay term improved the fit(/• of 0.97), as it allowed better description of the spread-ing of the elution peak. The dimensionless parameterfor decay ̂ allows for the calculation of the first-orderdegradation rate in the soil liquid phase, ̂ . The esti-mated value of (jL|iq for the degradation of triticonazolewas 7 d"1.

In both models, instantaneous sorption is character-ized by the dimensionless retardation factor R, basedon the mean position of the BTC (first moment). Thisallows for the calculation of a partition coefficient KAfor instantaneous sorption. Estimated values of KA were1.4 and 1.3 L kg"1 from CDeq, and 1.5 and 1.7 L kg-1

from CDnoneq.In the two-site model, the nonequilibrium dimen-

——Chloride

Triticonazole

""̂ 'Uii10 12 14 16

V/V0

Fig. 3. Comparison of chloride (line) and 14C-triticonazoIe (symbols) measured breakthrough curves for displacement through Grignon soil ofa pulse of 0.5 V0 and a pore water velocity of 208 cm d~'. Initial concentrations of the tracers input solutions were 1 g L"1 for chloride and5 mg L~' for triticonazole.


0.4

0.35

0.3

0.25

g °'2

0.15

0.1

0.05

Measured

~~ • CDeq / no decay

CDeq / first-order decay

10 12 14 16

V/V0

Fig. 4. Measured and CDeq-simulated breakthrough curves for displacement through Grignon soil of a pulse of 0.5 V',, of 14C-triticonazole at 5mg L"1. Simulated curves were calculated with the model parameters v and D set to 208 cm d~" and 12.48 cm2 d"1, respectively, withoutdecay (dotted line) or with first-order decay in the liquid phase (solid line).

sionless parameters (J and w were used for calculationof the fraction of Type 1 sites, /, and the first-order ratea (d^1) for the kinetic, Type 2 sites. The estimated valuesfor the parameters / and a. were different for the fitswith or without degradation term. If degradation is ne-glected, then the fraction of sites at equilibrium (/ =0.06) would indicate that the majority of the sites areof the kinetic type, whereas if a sink term is introduced,then the proportion of kinetic sites would be much lower(57%). The estimates of the first-order rates for thesorption kinetics are extremely high (>58 d'1), and theyincrease if degradation is not accounted for.

Mass balance of the amount of triticonazole appliedwas obtained from the total amount of radioactivityrecovered from the columns, and this showed that=95%of the total amount was detected in the effluent. Never-theless, results of the extraction and analysis of the soilsections at the end of the experiments showed that a

Table 2. Comparison of CDeq and CDnoneq parameter valuesestimated from the 14C-triticonazoIe data, with v and D set to208 cm d~' and 12.48 cm2 d~', respectively.___________

Model for parameter estimationCDeq CDnoneq

KfKt (L kg-')Pt

it

H* («> ')r

Degradation Two-site, Two-site,in solution no degradation in

No degradation phase degradation solution phase3.51 ± 0.06 3.26 ± 0.03 3.64 ± 0.01 3.99 ± 0.01

1.41 1.27 1.48 1.680.32 ± 0.05 0.57 ± 0.02

0.06 0.435.33 ± 0.05 3.12 ± 0.02

68.9 58.20 0.91 ± 0.03 0 0.22 ± 0.01

29.1 7.0-0.953 0.577 0.930 0.971

t Estimated value ± standard deviation.

fraction of 5.3% was recovered as methanol-extractable(2.7% of total applied) and bound 14C residues (2.6%of total applied). The distribution of the extractable andbound 14C residues in the soil profiles (Fig. 6) show thatthe amount of extractable residues increased with depthin the column, whereas the bound residues were mainlylocated in the upper end of the column (at the inlet),and they decreased with depth.

DISCUSSIONFlow Characteristics in Grignon Soil

Symmetrical, nonretarded ETC of chloride are ex-pected for the displacement of a nonreactive tracer inrepacked homogeneous water-saturated columns ofsieved soils, as reported, for instance, by Gamerdinger etal. (1990) and Chen and Wagenet (1997). The invariant,very low dispersivity X measured for Grignon soil issmaller than the aggregate size of the sieved soil, whichsuggests that some destruction of the aggregates oc-curred during packing of the saturated soil columns.Low dispersivities and high Peclet numbers (P) indicat-ing poor dispersion are characteristically reported inwell packed homogeneous soil columns (Lee et al., 1988;Romero et al., 1997). In undisturbed soils, where prefer-ential flow may occur, much higher dispersivities, rang-ing from 4.5 to 65.8 cm, were observed by Jaynes (1991),and a low P of 0.7 was reported by O'Dell et al. (1992).

The adequate fit of CDeq to our experimental chlo-ride data and the symmetrical shape of the ETC at thethree measured water velocities in our study is represen-tative of systems that are not influenced by transportnonequilibrium. Thus, physical nonequilibrium, whichis suggested by the small decrease in 3 that was ob-served, can reasonably be assumed to be negligible in

BEIGEL & DI PIETRO: INFLUENCE OF NONEQUILIBR1UM SORPTION ON TRANSPORT IN SOIL 1083

Measured

— 'CDnoneq / no decay

CDnoneq / first-order decay

V/V0Fig. 5. Measured and CDnoneq-simulated breakthrough curves for displacement through Grignon soil of a pulse of 0.5 F0 of 14C-triticonazole

at 5 mg L"1. Simulated curves were calculated with the model parameters v and D set to 208 cm d"1 and 12.48 cm2 d~', respectively, withoutdecay (dotted line) or with first-order decay in the liquid phase (solid line).

our columns, and nonequilibrium conditions for triticon-azole reactive tracer would mostly arise from sorptionnonequilibrium.

Nonequilibrium Transport of TriticonazoleThe asymmetrical shape of the triticonazole ETC is

indicative of nonequilibrium, and the considerable tail-ing suggests that it is primarily sorption related. Similar

sorption-related nonequilibrium transport characteris-tics have been reported for various organic chemicalsin repacked homogeneous soil columns (Lee et al., 1988;Angley et al., 1992; Gaber et al. 1995), and in fieldstudies (Jaynes, 1991). The failure of the CDeq modelis then expected, since the LEA would not be valid fortriticonazole transport in soil.

Sorption nonequilibrium with a two-stage approachto equilibrium has been evidenced for a large number

14

12

1 10to£••oen 8jf ° io>

—^Extractable residues

—• 'Bound residues

0 1 2 3 4 5 6

Soil depth (cm)Fig. 6. Measured concentration profiles of the extractable (solid line) and bound (dotted line and symbols) "C residues remaining in the soil

columns after elution of a pulse of 0.5 pore volume of "C-triticonazole at 5 mg L"1 with 16 pore volumes of water. Total recovered amountsof the extractable and bound fractions respectively averaged 2.7 and 2.5% of the initial amount applied.


of pesticides. A short, initial fast phase of sorption isgenerally reported in the first few minutes (Kookana etal., 1993; Gaston and Locke, 1995), which is followedby an extended period of much slower uptake, occuringover periods of days or months. Brusseau and Rao (1989),and Brusseau (1991) reviewed the different rate-limitingprocesses of nonequilibrium, and they attributed therate-limiting step in the sorption nonequilibrium ofHOC in soil to intrasorbent (intraorganic matter) masstransfer diffusion. Our previous results (Beigel et al.,1997 and 1999) support this hypothesis, because tritico-nazole sorption was related to nonspecific interactionmechanisms with soil OM, and because rate-limiting,slow desorption of triticonazole was evident during pro-longed incubation, which was attributed to intrasorbentdiffusion. Under these conditions, instantaneous Type1 sites would represent the sites that are directly accessi-ble, while kinetic Type 2 sites would be sites that aremore remote in the soil organic constituents. The first-order rate a for sorption kinetics of the time-dependentsites would in fact reflect a rate-limiting diffusion pro-cess. If the water velocity is low enough for all theaccessible Type 1 sites to be reached, the two-site as-sumption may be valid, and this approach has beensuccessfully used to predict the nonequilibrium trans-port of pesticides in homogeneous soil columns underlow velocities (Gamerdinger et al., 1990; Gaber et al.1995).

The CDnoneq model correctly described the earlytailing of the asymmetrical distal portion of triticonazoleelution peak, but it failed to describe the extended tail-ing of triticonazole ETC. The two-site approach is thusnot adequate for predicting the transport of triticona-zole in soil at high velocity. Chen and Wagenet (1997)observed similar failure of a two-site, first-order modelto describe atrazine [6-chloro-/V-ethyl-./V'-(l-metriyl-ethyl)-l,3,5-triazine-2,4-diamine] transport in homoge-neous soil columns at high velocities. In such cases, therate of sorption-desorption from the kinetic sites cannotbe described by a single first-order rate constant, as alsoreported by Connaughton et al. (1993). The two-stagetailing that we observed suggests that at least two typesof kinetic sites need to be considered in addition to theinstantaneous sorption sites.

The estimated KA values in the range of 1.3 to 1.7 Lkg"1 account for the instantaneous sorption of triticona-zole during transport. These estimates were much lowerthan the measured batch equilibrium KA of triticonazoleof 4.35 L kg"1, indicating that the batch equilibriumKA would considerably overpredict the retardation oftriticonazole during transport at high flow rate. Suchleftward shift of the experimental ETC compared withbatch-measured partition coefficient predictions hasbeen reported frequently at high velocities, while betteragreement is obtained at lower velocities (O'Dell et al.,1992; Gaber et al., 1995; Chen and Wagenet, 1997).This clearly shows that the batch-measured partitioncoefficient would not be appropriate for description ofthe transport of triticonazole when the residence timein soil is not long enough for some sorption sites toachieve equilibrium.

The very high value of the first-order rate a (>58d"1) estimated from the two-site model, which accountsfor the asymmetry in the elution peak, indicates a quicktransport to accessible sorption sites. In the water-satu-rated soil columns packed with finely sieved soil, triti-conazole is directly in contact with the dispersed soilorganic constituents, and it can then rapidly diffuse toaccessible sorption sites in the internal voids of organicmatter. A great portion of the kinetic sorption siteswould, thus, rapidly attain equilibrium and extensivelyincrease the sorption of triticonazole.

On the other hand, the extended slow and continuousrelease of sorbed residues and the fraction of extractableresidues remaining in the column suggest that anotherpart of the kinetic sites would be much more rate-lim-ited. Methanol-extractable residues may be subject todesorption in water, as observed by Barriuso et al.(1992), and thus are potentially available for transport.Indeed, their location at the effluent end of the soilindicates that they were subject to convective-dispersivetransport, and their release would continue after the 16pore volumes of water were applied, resulting in a longertailing than observed. These slow kinetics may be attrib-uted to rate-limiting sorption-desorption that is con-trolled by mass transfer diffusion from remote soil sites.This was not accounted for in the two-site model, andit needs to be determined independently for an adequatedescription of triticonazole transport in soil. Use of thebatch equilibrium KA would not be appropriate, as thebatch-obtained value does not represent the true sorp-tion equilibrium of triticonazole in soil (Beigel et al.,1997). Measurement of apparent desorption coefficients(Kapp) at prolonged incubation times may be more ade-quate for evaluation of the time-dependent desorption,as shown by O'Dell et al. (1992) and Beigel et al. (1997).

Kinetics of Bound Residue FormationIn our previous experiments (Beigel et al., 1997 and

1998), the formation of a significant fraction of soil-bound residues of triticonazole in both nonsterile and•y-radiated sterile soil was evident immediately aftertreatment. In such conditions, bound residue formationmay arise from physical trapping in the internal voidsof soil organic matter (Khan, 1982; Calderbank, 1989).Bound residues are resistant to desorption and wouldnot be available for transport. This would affect theelution of triticonazole in soil and might also partlyaccount for the two-site, two-region model's incapacityto describe the experimental data. Indeed, small butsignificant amounts of bound residues were detected inthe soil columns. Their localization at the inlet end ofthe soil profile shows that the removal from soil solutionand stabilization in the soil colloids occurred immedi-ately after treatment and that those residues were notsubject to convective transport. The nonreversible dis-appearence of triticonazole from soil solution as boundresidues cannot be accounted for by sorption in themodels, even in the two-site approach, which explainswhy the fraction of kinetic sites is considerably higherwhen no sink term is provided. Bound residues would


be partly accounted for in the model by the degradationterm, and the relatively high degradation rate n,|iq thatis estimated from the two-site model may be attributedto the rapid formation of triticonazole-bound residues.Hence, the estimated decay rate would depend on theresidence time in the columns (i.e., on the experimentalset up) and is, therefore, unreliable. Moreover, the rapidformation of bound residues in soil is followed by a muchslower, continuous stabilization (Beigel et al., 1999), anda single first-order rate would not be suitable to describethe dissipation of triticonazole as stabilized, bound resi-dues. The kinetics of bound residue formation need tobe further studied and correctly described to allow foran accurate modeling and prediction of transport in soil.

CONCLUSIONOur results clearly show that triticonazole convec-

tive-dispersive transport in homogeneous saturated soilcolumns is primarily influenced by its interactions withthe soil matrix through rate-limited sorption-desorptionand through the formation of bound residues. Both pro-cesses may be related to rate-limiting, intrasorbent diffu-sion into soil organic constituents and thus would bestrongly dependent upon the water flow rate. Failureof the two-site, nonequilibrium CDE model to describethe extended tailing of triticonazole experimental ETCat a high flow rate shows that nonequilibrium sorptioncannot be accounted for by a single first-order rate con-stant. Similarily, whereas the immediate formation ofbound residues may be accounted for with a high first-order rate in the decay term, use of a single rate constantwould prove erroneous for describing the much slower,continuous stabilization of triticonazole residues duringprolonged incubation time, which has been evident inprevious studies.

ACKNOWLEDGMENTSThis research has been conducted under the Bio Avenir

programme funded by Rhone-Poulenc with the participationof the French Ministere de la Recherche et de 1'Espace andMinistere de 1'Industrie et du Commerce Exterieur. All theradiochemical analyses were performed at Laboratory of SoilScience, INRA, 78850 Thiverval-Grignon, France. The au-thors wish to thank Ghislain Sevenier for performing theCl~ analysis.

..1,086 ' • . . . . . SOIL SCI. SOC, AM. J., VOL. ;63, SEPTEMBER-OCTOBER 1999

Toride, N., KJ. Leij, and M.Th. vari Genuchten. 1995. The CXTFlf Walker, A., Y.-H. Moon, and S.J. Welch. 1992. Influence of tempera-:; '• ' •• : code for estimating, transport parameters"from .laboratory or field . ture, soil moisture and soil characteristics on the persistence of

"tracer experiments. Version 2:6.•Research Rep. 137. U.S. Salinity, alachlor. Pestic. Sci. 35:109-116.' • Lab, -Riverside, CA. Weber, J.B. 1991. Fate and behaviour of herbicides in soils. Appl.Tprstensson,N.T.L: 1987. Microbial'decomposition of herbicides'iri Plant Sci. 5:28̂ 41. . •

soil. p. 249-270. In D.H. Hutson and T.R. Roberts.(ed.) Herbicides. Weber, J.B., and C.T. Miller. 1989. Organic chemical movement over•'John Witey & Sons, Chichester, UK. ., •. , and through soil. p. 305-334. In Reactions and movements of or-

Veeh, R.H., W.P. Inskeep, and A.K. Camper. 199.6. Soil depth and ganic chemicals in soil. SSSA Spec. Publ. 22. SSSA and ASA,temperature effects on microbialdegradatioh of 2,4rD. J. Environ, Madison WIQual. 25:5-12, . • . . . ' ' ' '

transport of triticonazole in homogeneous soil columns

Documents