transport of picloram in relation to soil physical conditions and pore-water velocity1

Transport of Picloram in Relation to Soil Physical Conditions and Pore-Water Velocity1

J. M. DAVIDSON AND R. K. CHANG2

ABSTRACTMiscible displacement techniques were used to study the

movement of a solution containing 4-amino-3, 5, 6-trichloro-picolinic acid (picloram) through an initially herbicide-freeNorge loam soil. Picloram mobility was reduced significantlyby decreasing the average pore-water velocity from 5.8 to 0.59cm/hr. A variation in herbicide adsorption with pore-watervelocity was observed at each bulk density (1.55 and 1.65g/cm3) and aggregate size (<2.0 and <0.42 mm) studied.For a specific bulk density, picloram adsorption was greaterwhen the largest soil aggregate size was <0.42 mm than whenthe soil contained <2.0-mm aggregates. Differences in theeffluent concentration distribution owing to variations in aggre-gate size were more evident at the lower bulk density. Theaverage pore-water velocity influenced picloram movementmore significantly than variations in bulk density or largestaggregate size at a given flow rate.

Additional Index Words; molecular diffusion, herbicide trans-port, miscible displacement, adsorption, pore-size distribution.

IN GENERAL, most organic herbicides are adsorbed by thesoil to some extent. This phenomenon is an important

factor in establishing the depth of herbicide penetrationand/or the concentration distribution with soil depth. Muchof the adsorption work to date has been for equilibriumconditions (1), with only a few studies giving considerationto the herbicide adsorption rate (7). If herbicide adsorptionis time dependent, then this factor as well as soil physicalproperties must be evaluated before we can describe a her-bicide's mobility in a transient soil-water environment.

Several mathematical models have been presented in anattempt to describe the displacement of a herbicide applica-tion through soil (5, 10, 11 and 12). These models have, ingeneral, tried to simplify the description of the transportprocess by deleting one or more of the mechanisms in-volved. Because of insufficient quantitative data, thesemodels have not been adequately tested or evaluated. Forexample, adsorption in many of the models is assumed tobe instantaneous and the adsorption and desorption processis assumed to be linear and completely reversible. David-son et al. (5) and Green et al. (6) have both shown thatless herbicide adsorption occurred when it was displacedthrough a soil than when equilibrium soil-to-herbicide solu-tion evaluations were conducted. These studies also show atailing or slow release of herbicide during desorption. Thetwo studies suggest that for high pore-water velocities, equi-librium adsorption may not exist within all pore sizes. If thisis true, it may be necessary to use the pore-size or pore-

1 Contribution from the Department of Agronomy, Okla-homa Agr. Exp. Sta., Stillwater, and as part of Project H-1324and H-1366. Journal Manuscript no. 2281. Received July 14,1971. Approved Nov. 16, 1971.

2 Associate Professor and Research Associate, Agronomy De-partment, Oklahoma State University, Stillwater 74074. Secondauthor is deceased.

water velocity distribution to describe the transport of thematerial through the soil (11).

Instantaneous equilibrium between the adsorbed and solu-tion concentration has been assumed by several investiga-tors studying herbicide movement through soils (5, 9, 10).Few studies report the rate at which equilibrium adsorptionis achieved. Hance (7), using the batch-adsorption proce-dure, in an adsorption study with four herbicides and fouradsorbents, found that 3 of the 16 systems reached equilib-rium in 1 hour and 8 reached equilibrium in 4 hours. Heconcluded that at least 24 hours were necessary for all sys-tems to reach equilibrium. The procedure used by Hance(7) assumes optimum contact between the herbicide andsoil particle surface at all times.

Owing to the natural pore-size distribution within a soilmedium, the resident time of the herbicide molecule withthe soil particle surface will vary with the average velocityof the water within each pore size. Small pores provide thelongest residence time owing to their low average pore-water velocity. However, small pores do not have a largereservoir of solute, and therefore, are dependent upon mo-lecular diffusion to reach the same adsorption equilibriumthat exists within the larger pores. This is a different ad-sorption equilibrium problem from the one discussed byHance (7).

The objective of this study was to illustrate the influenceof the average pore-water velocity on herbicide adsorption.Also, by using different pore-size distributions within thesoil medium, the importance of pore geometry to adsorp-tion was evaluated. Using the mathematical model proposedby Lindstrom et al. (10) and used by Davidson et al. (5)and adsorption isotherm data, calculated and measuredresults are compared and discussed.

ExperimentalThe procedure described by Davidson and Santelmann (4)

was used to study the movement of 4-amino-3, 5, 6-trichloro-picolinic acid (picloram) through a Norge loam soil. The her-bicide was !4C-labeled at the carboxyl position and had aspecific activity of 4.13 /tCi/mg. The herbicide displacementrate was varied by controlling the soil-water flux with a con-stant volume pump. The two soil-water fluxes used were 2.0and 0.2 cm/hr. The soil-water flux was divided by the volu-metric water content, determined after each displacement, togive average pore-water velocity. Average pore-water velocitywill be used in this manuscript when discussing soil-waterflow rate. Five-mi effluent water samples for picloram analysiswere collected sequentially with a fraction collector. A relativeconcentration (C/C0) was calculated from the picloram con-centration in the effluent (C) and the original concentration(C0) of the picloram at the inflow position. All studies wereconducted at 25 ± 0.5C.

The soil, initially calcium saturated, was air-dried and thecomplete sample forced through either a 2.0- or 0.42-mm sieve.The sieved soil was then packed into glass cylinders to aver-age bulk densities near 1.55 or 1.65 g/cm3. The cross sectionalarea and length of the soil columns were 45 cm2 and 30 cm,respectively. The soil was saturated with 0.01N CaSO4 and thedesired soil-water flux (2.0 or 0.2 cm/hr) was established.

257

258 SOIL SCI. SOC. AMER. PROC., VOL. 36, 1972

EQUILIBRIUM SOLUTION CONC. (/ig/ml)

Fig. 1—Equilibrium adsorption isotherm of picloram on Ca-saturated Norge loam.

At the completion of each herbicide displacement, the volumeof water held in the soil was determined gravimetrically. Thenumber of pore volumes (V/V0) displaced through the soilwas calculated by dividing the volume of effluent (V) by thevolumetric water capacity (V0) of the soil. The pH, cationexchange capacity, and organic matter content of the Norgeloam are 6.6, 9.2 meq/lOOg, and 1.7%, respectively. The soilhas 46% sand, 38% silt, and 16% clay.

Solutions of picloram were prepared in O.OIN CaCl2 using10.6 ,«Ci/liter (2.52 ppm). Two hundred-mi (approximately0.41 pore volume) of herbicide solution were added to thewater-saturated soil at a given soil-water flux and the herbi-cide displaced through the soil with 0.0IN CaSO,, at the sameflux. The 200-ml of picloram solution was equivalent to 1.1kg/ha. Displacements for all combinations of largest aggre-gate size, bulk density, and soil-water flux were made. Thepicloram concentration in each 5-ml effluent sample was de-termined from the 14C activity using a liquid scintillation coun-ter. The liquid scintillation counting solution contained 120 gof naphthalene, 4 g of 2, 5-diphenyloxazole (PPO), and 50mg of 1, 4-bis-2-(5-phenyloxazolyl)-benzene (POPOP) in 1liter of p-dioxane. One-half ml aliquots from each effluentsample were used in 15-ml of counting solution.

The adsorption isotherm for picloram and Norge loam wasdetermined from duplicate 1:1 mixtures of 5 g of soil and 5 mlof various herbicide concentrations in 0.01JV CaCl2. Each sam-ple was shaken for 5 hours and then centrifuged at 1,250 X gfor 15 min. Samples of the clear supernatant solution were ana-lyzed for 14C activity. Earlier experiments had shown thatpicloram reached herbicide adsorption equilibrium within the5 hour period.

RESULTS AND DISCUSSIONHerbicide adsorption can generally be described by the

Freundlich adsorption equation (1). The Freundlich equa-tion is:

5 = [1]

where S is the quantity of herbicide adsorbed per gram ofadsorbent (/ig/g), C the equilibrium herbicide concentra-tion in solution (/ig/ml), K the distribution coefficient, andn a constant. Both K and n are dependent upon the natureof the adsorbent, adsorbate, and temperature. Using equa-tion [1] to describe picloram adsorption by Norge loam,the values of K and 1/n were 0.18 and 0.97, respectively.The value of 1/n is nearly one and for this study will beassumed unity. Figure 1 shows the relationship between theadsorbed and solution concentration of picloram at equi-librium. Assuming a linear relation between S and C, aregression analysis was made and the best fit line drawn

through the data. The equation of the best fit line (Fig. 1)illustrates that only a small error was introduced by assum-ing 1/n equal to unity in equation [1],

The differential equation frequently assumed to describethe longitudinal mixing process (simultaneous mixing bymolecular diffusion and convective flow) and linear adsorp-tion in soils for one dimensional flow is:

_D n-v — -[!+—] — =0°dx 8 dt [2]

where D0 is the apparent molecular diffusion coefficient(cm2/hr), C the herbicide concentration in solution (g/cm3),x the distance from the input (cm) v0 the average pore-water velocity in the x direction (cm/hr), p the bulk den-sity (g/cm3), K the distribution coefficient (crnVg), 9 thevolumetric water content (cm3/cm3), and t the time (hr).Equation [2] has been solved by Lindstrom et al. (10) forthe same boundary conditions used in this study. The quan-tity (1 + pK/8) in equation [2] is called the retardationfactor by Hashimoto et al. (8). The pK/6 represents theapparent increase in pore volume as a result of the adsorp-tion process.

Retardation factors [1 + (pK/9)] were calculated foreach set of column parameters in Table 1 using a distribu-tion coefficient of 0.18. An increase in soil bulk densityfrom 1.55 to 1.63 g/cm3 produces an increase in the retar-dation factor (Table 1). This means that more water isrequired to move the herbicide a fixed distance through awater saturated soil at a high as compared to a low bulkdensity. Figure 2, left-hand curves, shows the concentrationdistributions of a fast displacement of picloram through the< 2.0-mm aggregate size fraction for bulk densities of1.55 and 1.63 g/cm3. Note the early arrival and greaterpeak concentration of the picloram for the high as com-pared to the low bulk density. The order of the two experi-mental curves is contradictory to that predicted by theretardation factors given in Table 1. The calculated curve(10) on the left-hand side of Fig. 2 is based on the solutionof equation [2] for a bulk density of 1.55 g/cm3, retarda-tion factor of 1.78, pore-water velocity of 5.5 cm/hr, andapparent diffusion coefficient of 0.356 cm2/hr. The appar-ent diffusion coefficients used in this "study were selectedsuch that the shape of the calculated curve was similar tothe shape of the left-hand portion of the experimental data.The tables presented by Brenner (3) were modified to in-clude adsorption and then used to obtain the appropriateapparent diffusion coefficient. A calculated curve for asimilar pore-water velocity and a bulk density of 1.63g/cm3 and retardation factor of 1.83 would be to the rightof the present calculated curve with the left-hand portionof the curve passing through C/C0 = 0.5 at V/V0 = 1.83.

The position of the two left-hand curves with respect tothe calculated line in Fig. 2 indicates that incomplete mix-ing and adsorption exist at both bulk densities at the highsoil-water flow rate. At relatively high average pore-watervelocities, the effect of diffusion on the spreading of theinvading herbicide front will be small. This fact coupledwith adsorption means that the small pores, owing to their

DAVIDSON & CHANG: TRANSPORT OF PICLORAM 259

Table 1—Physical data for picloram displacement throughNorge loam 0°

Xo

Largestaggregate size

2mm2mm2mm2mm0.420.420.420.42

Vo

Average pore-water velocity

cm/hr6.15.50.650.565.95.70.590.56

1 + 8

Retardationfactor

1.831.781.911.721.811.761.801.74

8Soil-watercontent

cm3/cms

0.3450.3630.3230.3750.3560.3670.3560.372

p

Bulk densityg/cm>1.631.5S1.671.531.651.581.631.56

smaller pore-water velocities, will not be in equilibriumwith the large pores which have a higher pore-water veloc-ity. This incomplete mixing of the herbicide solution be-tween various pore sizes results in an early arrival of thematerial in the effluent as well as less adsorption during itsdisplacement through the soil. The distribution coefficientdetermined from Fig. 1 is too large to describe the observedadsorption at the high flow rate. Distribution coefficients of0.042 and 0.066 for the high and low soil bulk density,respectively, will predict the left-hand location of the rela-tive picloram concentration distributions obtained at thehigh flow rate. A similar reduction in the distribution coef-ficient was found by Davidson et al. (5) when attemptingto predict the effluent data with the dispersion model pre-sented by Brenner (3) and modified by Lindstrom et al.(10). The average pore velocity is greater in the soil withthe higher bulk density (Table 1). Also, the pore-velocitydistribution is different for each bulk density.

Reducing the average pore-water velocity approximately10-fold increased the amount of herbicide adsorbed duringits displacement. The retardation factors for the 1.53 and1.67 g/cm3 bulk density soil columns used at the lowerflow rate are 1.72 and 1.91 (Table 1), respectively. At thelower average pore velocity, the two curves assume theorder expected based on their retardation factors.

The calculated curve on the right-hand side of Fig. 2 isfor the 1.53 g/cm3 bulk density and the correspondingretardation factor (Table 1). The apparent molecular diffu-sion coefficient used to calculate the curve is 0.078cm2/hr. This is a 5-fold reduction from that used for thehigh soil-water flow condition and illustrates the couplingthat exists between the pore-water velocity and moleculardiffusion. It also raises a question regarding the deletion ofdiffusion from the mathematical model as suggested byOddson et al. (12). A reasonable description of the locationand shape of the left-hand portion of the 1.53 g/cm3 datawas obtained. A similar fit was obtained for the 1.67 g/cm3

data. This agreement was primarily the result of the longerresident time available for equilibrium adsorption at thereduced pore-water velocity. Youngson et al. (14) show thatthe adsorption of a soil fumigant was also dependent uponsoil-water flow rate. However, Green et al. (6) in a studywith 2-chIoro-4-ethylamino-6-isopropyl-amino-.j-triazine(atrazine) and a soil-water flux range of 0.6 to 7.0 cm/hrobserved no difference in the effluent concentration distri-butions owing to soil-water flux. It is of interest to notethat in Fig. 7 of Kay and Elrick (9), an accidental increasein the average soil-water pore velocity produced a sharp

I 0.6)

' 0.4

NORGE LOAM< 2.0 mm

•CALCULATED

l»ol.67g/cmj

0.65 cm/hrf •= l.53g/cm>V0 = 0.56cm/hr

1.0 1.5 20 2.5^ 1.5 2.0 25PORE V O L U M E , V/V0

Fig. 2—Experimental and calculated relative picloram con-centration distributions from Norge loam ( <2.0 mm aggre-gates ) for two average pore-water velocities. Solid lines werecalculated with physical data (Table 1) for the 1.55 and1.53 g/cm3 bulk density soil columns.

increase in the 1, 2, 3, 4, 5, 6 hexachlorocyclohexane (lin-dane) effluent concentration. This suggests that lindane ad-sorption was dependent upon soil-water flux in a similarmanner to the results shown in Fig. 2.

Reducing the largest aggregate size in the soil from 2.0to 0.42 mm increases the amount of picloram adsorbed(Fig. 3) during a fast displacement of picloram through thesoil. The two left-hand curves in Fig. 3 are shifted to theright and the peak concentration lower than the two left-hand curves in Fig. 2 for similar bulk densities and averagepore-water velocities. A reduction in the largest aggregatesize will decrease the pore-size distribution range. Biggarand Nielsen (2) have shown that as the pore-size distribu-tion range is decreased, the mixing in the soil column be-comes more complete and the flow is no longer dominatedby the large pores. This phenomenon is particularly impor-tant when solute adsorption exists in the column. However,comparing these two curves with the calculated curve for abulk density of 1.58 g/cm3, pore-water velocity of 5.7cm/hr, and apparent diffusion coefficient of 0.354 cm2/hr,it is apparent that incomplete mixing still exists at the highflow rate.

Reducing the flow rate 10-fold increases picloram ad-sorption (Fig. 3). The two right-hand curves in Fig. 3 aresimilar to the results for the high bulk density and low flowrate column in Fig. 2. It should be noted that the high bulk

oz i .o-oo5 0.8

tx.O 0.6o"" 0.4

f f = 1.65 g/cm5

1 V0 = 5.9 cm/hrIf = l.58g/cm»IV0 = 5.7 cm/hr

NORGE LOAM< 0.42 mm f f * l.63g/«ns

1 V0 = 0.59cm/hrIf = l.56g/cm>1 vo =0.56cm/hr

ALCULATED

10 1.5 2.0 2.5 ' 1.5 2.0 2.5PORE VOLUME, V/V0

Fig. 3—Experimental and calculated relative picloram con-centration distributions from Norge loam ( < 0.42 mm aggre-gates ) for two average pore-water velocities. Solid lines werecalculated with physical data (Table 1) for 1.58 and 1.56g/cm3 bulk density soil columns.

260 SOIL SCI. SOC. AMER. PROC., VOL. 36, 1972

g 1.0

E 0.8

§3 0.6o°-o.4

K 0.23LJa °l

f»o • 5.5 cm/hrI < 2.0 mm|v0 = 5.7 cm/hrI < 0.42 mm

NORGE LOAMf " I.SSg/cm5

CALCULATED

f »o =0.56 cm/hrI < 2.0 mmfv0 =0.56 cm/hr[ < 0.42 mm

PORE VOLUME, V/V0

Fig. 4—Experimental and calculated picloram concentrationdistributions from Norge loam (1.55 g/cm3 bulk density) fortwo average pore-water velocities. Solid lines were calculatedwith physical data (Table 1) for the two < 0.42 mm soilaggregate columns.

density column at the low flow rate in Fig. 2 had the high-est retardation factor (1.91), of any column in this study.The two right-hand curves in Fig. 3 may be approachingequilibrium adsorption and therefore, would not vary withadditional changes in soil aggregate size or soil-water ve-locity. However, comparing the calculated line based onthe 1.56 g/cm3 bulk density column for the low flow ratewith the experimental data shows more adsorption occur-ring than predicted by the mathematical model. Also, thetwo experimental curves are in the reverse order to thatexpected based on their retardation factor (Table 1). Sucha situation could occur, assuming the difference to be real,if the adsorption and desorption process was not reversible.The calculated line used an apparent molecular diffusioncoefficient of 0.078 cmVhr.

Based on Fig. 2 and 3 it is readily apparent that theadsorption and mobility of picloram was significantly influ-enced by the average pore-water velocity, largest aggregatesize, and bulk density. It appears that adsorption may becoupled with pore-water velocity distribution as well asmolecular diffusion. Lindstrom and Boersma (11) present amethod of treating mass transport through a uniform soilusing the pore-size distribution. They show that with anincrease in the pore-size distribution, the concentration dis-tribution is widened at the base and the peak concentrationreduced.

The problem associated with pore-size distribution andmolecular diffusion can be better examined by replottingonly those curves from Fig. 2 and 3 with an average bulkdensity near 1.55 g/cm3. Figure 4 shows the picloram con-centration distribution for the < 2.0 mm aggregate size tothe left of the 0.42-mm size at both flow rates. The averagepore-water velocity is about the same for each aggregatesize at the two flow rates; however, this does not mean thatthe velocity distribution is the same. The < 0.42-mm aggre-gate size soil column should have a more uniform or nar-rower velocity distribution than the larger aggregate size. Asmaller pore size and narrower size range would offer moreparticle surface area for herbicide adsorption during itsdisplacement and in general, a longer average resident timewould exist in the larger pores of the < 0.42-mm samplethan in the larger pores of the < 2.0-mm aggregate sizefraction. It should also be noted that the percent of the total

porosity filled with water (Table 1) was, for most cases,greater for the < 0.42-mm size range. The calculated linesin Fig. 4 are based on measurements taken from the< 0.42-mm aggregate size columns.

The molecular diffusion of picloram from the solutionin the flowing pore to the interior of the porous aggregatemust also be considered. The pores within the aggregatecan be considered smaller than the pores between adjacentaggregates. At the lower flow rate, molecular diffusion maybe a significant transport process and a contributing factorto the large difference in the two right-hand curves in Fig.4. The difference may result from the longer time thatwould be required for the picloram to diffuse into the< 2.0-mm aggregates and reach equilibrium adsorption.However, if adsorption and desorption are not reversible,then the displacement rate of picloram through the soilwould be reduced similar to the < 0.42 mm column at thelow flow rate. At the higher bulk densities, larger pores thatexisted at lower bulk densities will not be present becauseof compaction. This change will result in more piclorammovement through the pores within the aggregate. Thus,complete adsorption equilibrium would be less dependentupon molecular diffusion under compacted conditions.Over small distances, diffusion can be an important processcontributing to herbicide movement (13).

The chloride in the 200-ml of picloram solution was alsomeasured in each effluent sample (Fig. 5). A reduction inthe average pore-water velocity caused a spreading at thebase of the concentration distribution curve owing to mo-lecular diffusion. No consistent difference in the concen-tration distribution data was noted between aggregate sizeor bulk density at the same soil-water velocity. Therefore,the differences in picloram mobility brought about bychanges in the aggregate size and bulk density must besignificantly coupled with the adsorption process. Part ofthe observed variation among the relative chloride concen-tration distributions in Fig. 5 can be attributed to the volu-metric water capacity (V0) difference that exists betweenthe two bulk densities. The area under the data from the1.55 g/cm3 soil columns is smaller than that under the1.65 g/cm3 soil column by approximately 7%.

2 1.0-OO

£0.8rri 0.6

> 0.4

• l.63g/cm3, < 2.0mm° l.64g/cm5, < 0.42mmD l.55g/cms, < 2.0mm^ l.58g/cm5, < 0.42mm

» 5.8 cm/hr

1.0

• 1.66g/cm3, < 2.0mm° 1.63 g/cm3, < 0.42mma 1.53g/cm3, < 2.0mm* 1.55g/cm3, < 0.42mm

0.59cm/hr

1.5 2.01.5 0.5 1.0PORE VOLUME, V/V0

Fig. 5—Relative chloride concentration distributions fromNorge loam for two bulk densities and aggregate sizes at twoaverage pore-water velocities.

STONE: INSTRUMENTATION EFFECTS ON ERRORS IN NUCLEAR METHODS 261

The adsorption and mobility of picloram is influencedby average pore-water velocity, bulk density, and aggregatesize. Picloram adsorption is dependent upon average pore-water velocity at relatively high flow rates. However, thecomplexity of the adsorption process appears to be in-creased by the influence of the pore-size distribution. Thephysical geometry of the soil system as well as the molecu-lar diffusion of the herbicide molecule require additionalstudy. Several models treating various aspects of the prob-lem (10, 11, and 12) are available, but a complete modelhas not been presented or tested.

ACKNOWLEDGMENTWe would like to thank Dow Chemical Company for sup-

plying the 14C-labeled picloram used in this study.