UFDC Home  |  Search all Groups  |  UF Institutional Repository  |  UF Institutional Repository |   Help

# Movement and adsorption of pesticides in sterilized soil columns

## Material Information

Title:
Movement and adsorption of pesticides in sterilized soil columns
Series Title:
Florida Water Resources Research Center Publication Number 16
Physical Description:
Book
Creator:
Mansell, R. S.
Hammond, L. C.
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

## Notes

Abstract:
Rapid transport of systemic and soil sterilant herbicides in soil during periods of net water flow may decrease the effectiveness of the chemicals to control unwanted vegetation and produce undesirable pollution of the ground water. An investigation of the influence of physical-chemical soil properties upon the transport of 2,4-D and paraquat in columns of organic and sandy soils was therefore performed. These herbicides are water soluble organic chemicals which are used extensively in agriculture. The toxicant portions of 2,4-D and paraquat behave as anion and cation, respectively. Miscible displacement of aqueous solutions of these herbicides through columns of Everglades mucky peat resulted in most of the 2,4-D and all of the paraquat being removed from solution by adsorption. Limited transport of 2,4-D was observed for the same fine sands. Very small quantities of organic matter in the fine sands effectively removed paraquat from the flowing soil solution. The presence of large concentrations of KCl in the soil solution was observed to decrease the quantity of paraquat sorbed. Mathematical transfer function theory was used in connection with statistical hydrodynamics to develop a technique for analysis and prediction of herbicide elution from soil columns during miscible displacement experiments.

## Record Information

Source Institution:
University of Florida Institutional Repository
Holding Location:
University of Florida
Rights Management:
System ID:
AA00001476:00001

Full Text

-I- -, i- I-

Publication No. 16
Pesticides in Sterilized Soil
Columns

By
R.S. Mansell and L.C. Hai,,nond
Soils Department, IF S

Water Resources Research
Univer, Ja

MOVEMENT AND ADSORPTION OF PESTICIDES IN STERILIZED
SOIL COLUMNS

By

R.S. MANSELL
(Principal Investigator)
and
L.C. HAMMOND

PUBLICATION NO. 16

of the

FLORIDA WATER RESOURCES RESEARCH CENTER

RESEARCH PROJECT TECHNICAL COMPLETION REPORT

OWRR Project Number A-013-FLA

Annual Allotment Agreement Numbers

14-01-0001-1628 (1969)
14-31-0001-3009 (1970)
'14-31-0001-3209 (1971)

Report Submitted: August 9, 1971

The work upon which this report is based was supported in part
by funds provided by the United States Department of the
Interior, Office of Water Resources Research as
Authorized under the Water Resources
Research Act of 1964.

Page

ABSTRACT . .. . . . .

SUMMARY.

INTRODUCTION AND REVIEW OF LITERATURE . .

Movement of 2,4-D and Paraquat Herbicides
in Soils . . . .

Chemical Properties of Paraquat and 2,4-D
Herbicides . . . .

Physical-Chemical Properties of Soil-Water
Systems that Influence the Movement and
Adsorption of Herbicides in Soils .. ..

EXPERIMENTAL MATERIALS, METHODS AND PROCEDURES

Herbicides . . . .. .

Soils . . . . .

Miscible Displacement Technique for Studying
Herbicide Movement in Soil Columns. ... .

Herbicide Concentration in Soil Effluent. .

INVESTIGATIONS . . . . .

Theoretical Analysis of the Movement of
Solutes in Soils: An Application of Transfer
Functions . . . . .

Miscible Displacement of 2,4-D Herbicide
Through Water-Repellant Soils . .

Miscible Displacement of 2,4-D Herbicide
During Constant Liquid Flow Velocity Through
Initially Dry Soils . . .

Sterilization of Intact Soil Columns Using
Exposure to Gamma Radiation and Methyl
Bromide Fumigation . . .. .

Miscible Displacement of Paraquat Herbicide,
Cl36, and Tritiated Water Through Sterile
and Nonsterile Soil Columns . ....

. 1

. 4

. 5

7

8

S 8

9

. 11

11

11

24

28.

35

38

Page

Displacement of Paraquat and Diquat
Herbicides by KC1 Solution from Water-
Saturated Columns of Soil . . 42

Soil Adsorption of Paraquat Herbicide ... 47

Continuous Measurement of Chloride in
Effluent Flowing from a Soil Column . .. 55

ACKNOWLEDGMENTS . . . . .. 60

LITERATURE CITED .. . . . 61

ABSTRACT

MOVEMENT AND ADSORPTION OF PESTICIDES IN
STERILIZED SOIL COLUMNS

Rapid transport of systemic and soil sterilant herbicides
in soil during periods of net water flow may decrease the
effectiveness of the chemicals to control unwanted vegetation
and produce undesirable pollution of the ground water. An
investigation of the influence of physical-chemical soil prop-
erties upon the transport of 2,4-D and paraquat in columns of
organic and sandy soils was therefore performed. These herbi-
cides are water soluble organic chemicals which are used ex-
tensively in agriculture. The toxicant portions of 2,4-D and
paraquat behave as anion and cation, respectively. Miscible
displacement of aqueous solutions of these herbicides through
columns of Everglades mucky peat resulted in most of the 2,4-D
and all of the paraquat being removed from solution by adsorp-
tion. Limited transport of 2,4-D was observed for the same
fine sands. Very small quantities of organic matter in the
fine sands effectively removed paraquat from the flowing soil
solution. The presence of large concentrations of KC1 in
the soil solution was observed to decrease the quantity of
paraquat sorbed. Mathematical transfer function theory was
used in connection with statistical hydrodynamics to develop
a technique for analysis and prediction of herbicide elution
from soil columns during miscible displacement experiments.

Mansell, R. S., and L. C. Hammond
MOVEMENT AND ADSORPTION OF PESTICIDES IN STERILIZED SOIL COLUMNS
Completion Report of the Office of Water Resources Research,
Department of Interior, August, 1971, Washington, D. C. 20240
KEYWORDS: paraquat/ pesticide movement/ herbicides/ adsorp-
tion/ water pollution/ ground water.

SUMMARY

Chemicals applied to crops or soils for the purpose of
killing weeds may be classified as contact, systemic or soil
sterilant herbicides. Contact materials require direct appli-
cation to foliage of the target plants; whereas, systemic
chemicals may be applied directly to foliage or indirectly to
the soil. Systemic herbicides may be absorbed through leaves
or roots and may be translocated through the entire plant
system. Soil sterilants prevent plant growth when present
in the soil. Thus, most herbicides used in agriculture reach
the soil, whether by direct application or indirectly as re-
sidual loss from spray applications to the plant leaves. Dur-
ing periods of rainfall or irrigation these chemicals may
move with water downward through the soil. If an herbicide
moves through the soil profile to the water-saturated zone
beneath the water table, the groundwater may become contami-
nated. The objective of this research project was to evalu-
ate the influence of specific physical-chemical properties
of the soil environment upon movement and consequent adsorp-
tion of 2,4-D and paraquat herbicides in agricultural sands.
Properties investigated were water flow velocity, initial
soil water and soil organic matter content.

Paraquat and 2,4-D chemicals were selected for this re-
search because of their extensive use as contact, systemic
and soil sterilant herbicides. Both are highly water-soluble;
therefore, they are suspects for leaching and movement through
soils. In aqueous solution, however, paraquat and 2,4-D
molecules respectively, form organic cations and anions, which
are highly toxic to plants. As these ions move in water
through the interconnected pores of soils, retention may occur
as physical and chemical sorption to the pore walls. For
soils high in clay mineral or organic matter content the di-
tion of the 2,4-D anion is much less than for paraquat.

A technique based upon statistical hydrodynamics and
Laplace transfer function theory was developed to mathemati-
cally describe the movement of herbicides or other solutes
through columns of soil. For miscible displacement of herbi-
cide solutions, through columns, experimentally determined
breakthrough curves (concentration of herbicide in the effluent
as a function of time) may be used to calculate a transfer
function for a known influent application function. The trans-
fer function is characteristic of the specific soil flow sys-
tem, and it indicates how the influent concentration function
is modified to give the effluent concentration function dur-
ing displacement through the soil. From the transfer function
the value of the coefficient of hydrodynamic dispersion (in-
cludes effects of convection, diffusion, and retention mech-
anisms such as adsorption) may easily be calculated. The

transfer function technique also provides a means to predict
breakthrough curves (or effluent concentration functions)
of a given herbicide for a given soil when the dispersion
coefficient, the flow velocity, the column length, and the
soil transfer function are known.

Movement of 2,4-D in columns of moist fine sand was found
to be influenced by water-repellancy of the sand. Displace-
ment of a volume "slug" of 2,4-D solution through columns
gave maximum recovery in the effluent from water-repellant
Blanton fine sand, less recovery in ignited Blanton, and even
less for naturally water-wettable Blanton. The larger re-
covery of 2,4-D in the effluent from the water-repellant sand
was attributed to incomplete and non-uniform water-saturation
of the packed column. When this column was dismantled at
the end of the experiment small zones of dry soil were ob-
served within the moist soil. Thus the water-filled porosity
was less in the water-repellant soil and thus adsorption of
2,4-D should be less. Ignition removed the soil organic matter;
therefore, 2,4-D adsorption should be low. Increasing the
liquid flow velocity resulted in greater herbicide recoveries
in the effluents from all the porous materials.

Miscible displacement experiments showed 2,4-D to move
readily with water through initially dry columns of glass
beads, Lakeland fine sand, and Fellowship sandy clay. Most
of the chemical was recovered within two pore volumes of ef-
fluent in these soils. Movement of 2,4-D greatly lagged water
movement in columns of Everglades mucky peat. More than five
pore volumes were required to recover only 60% of soil-applied
2,4-D. Adsorption of 2,4-D appeared to be reversible in the
mineral soils but not so in the organic peat.

Intact columns of dry Oldsmar fine sand were sterilized
by fumigation with methyl bromide gas and by irradiation with
an intense field of gamma rays. Both treatments gave complete
sterilization and provide convenient means for evaluating
the influence of specific physical-chemical soil properties
upon miscible displacementof herbicides through soil columns.

Displacement of three successive "slugs" or paraquat
and tritiated water through water-saturated columns of irradi-
ated, fumigated and non-treated Oldsmar fine sand gave unex-
pected breakthrough curves for the tritiated water. The suc-
cessive curves for the sterilized soil gave almost complete
recovery, but for the untreated soil recovery decreased with
each "slug" applied. The phenomenon was attributed to rapid
growth of microorganisms in the non-sterile soil. The break-
through curves for tritiated water behaved similarly when
Cl-36 and tritiated water were applied as three successive
"slugs" to another non-sterile column. A definite separation
was observed for Cl-36 and tritium curves. Tritium lagged the
chloride anion in the effluent, as might be expected.

The concentration of KC1 in the soil solution was found
to influence movement and sorption of the paraquat cation
in columns of soils. Application of a dilute KC1 solution
to water-saturated columns of Wabasso fine sand resulted in
elution of small quantities of adsorbed paraquat and diquat.
Desorption of both chemicals was greater from Wabasso fine
sand taken from 33-76 cm profile depth than for sand taken
from 0-10 cm depth. The organic matter content in the sur-
face soil is approximately 10 times greater than in the
material taken from 33-76 cm.

Increasing concentrations of KC1 over the range from 0
to 9000 ppm in aqueous solutions of paraquat resulted in de-
creasing adsorption of paraquat by Pomello, Wabasso and
Blanton fine sands. Initial solution concentrations of 17
and 51 ppm paraquat were used. The KC1 concentration had
peat, ignited Pomello fine sand, and ignited Blanton fine
sand. With no KC1 present in the solution adsorption iso-
therms were linear for the organic peat over the range of
0 to 500 ppm paraquat in the initial solution phase. Linear
sorption of paraquat by the fine sands was limited to narrower
solution concentration ranges. The assumption of linear ad-
sorption is therefore valid for these soils for the recommended
application rates of paraquat for herbicidal purposes. The
presence of large concentrations of KC1 or other salts in
the soil solution following fertilizer applications to agri-
cultural sands may stimulate limited profile movement of ad-
sorbed paraquat.

An inexpensive flow cell method is described for contin-
uously recording the concentration of chloride in effluents
from soil columns. The method was potential for miscible
displacement studies which involve chloride as a "tracer"
for herbicide movement in soils.

In conclusion, movement of 2,4-D and paraquat with water
in sandy soils investigated indicate that both chemicals under-
go retention due to sorption by small amounts of soil organic
matter. Paraquat was completely sorbed in most cases; whereas
2,4-D did move through soil columns under most conditions.
At the very low concentrations of 2,4-D normally applied to
these soils, significant amounts of the chemical would not
be expected to traverse field profiles to contaminate ground
water.

INTRODUCTION AND REVIEW OF LITERATURE

Movement of 2,4-D and Paraquat Herbicides in Soils

Organic pesticides applied directly or indirectly to

surface soil may move with water downward through the profile
to ultimately contaminate ground water. The pollution poten-
tial of an individual pesticide is a complex function of prop-
erties of the soil-water system which contribute to solute
transport and properties of both the pesticide and the soil-
water system which contribute to attenuation. Movement of
one of these chemicals into and through porous media occurs
primarily by mass flow (convection) coupled with molecular
diffusion. These two mechanisms also cause a chemical to
undergo dispersion (mixing) with water in the soil pores.
Mixing of a solute during flow of a liquid through porous
media is referred to as hydrodynamic dispersion (Bear, et al.,
1968). As the pesticide moves through the soil pores, adsorp-
tion, fixation, precipitation, degradation, and other attenu-
ating mechanisms tend to remove the chemical from the flowing
stream of soil solution. Those materials which interact
strongly with the soil and which also resist decomposition
are commonly called persistent pesticides (Van Middelem, 1966).
The capacity of a soil to adsorb chemical molecules has been
cited (Shaw, 1966) as one of the most important inactivating
mechanisms in man's total environment. Properties of a
specific pesticide such as solubility in water, ionic charge,
etc., also determine the extent of attenuation during move-
ment through the soil-water environment.

Van Middelem (1966) states that herbicides present a
special problem of pesticide movement in soils since many
are applied directly to the soil as selective pre-emergence
sprays and as nonselective soil sterilants. Intensive rain-
fall and sandy soils in Florida create a favorable environ-
ment for contamination of ground water with soil-applied herb-
icides. This research investigation was performed to deter-
mine the influence of specific physical-chemical properties
of soil-water environments upon the movement of 2,4-D and
paraquat herbicides through columns of sandy soil.

Chemical Properties of Paraquat and 2,4-D Herbicides

Chemical properties have been used by Goring (Kirkham,
1964) to classify pesticides into three broad categories as
far as movement in soil by water is concerned. The first
category includes chemicals with a reasonably high water solu-
bility, which are nonionic or the toxicant portion is anionic.
Movement of these materialsin soil should be somewhat similar
to that of the inorganic anions, nitrate and chloride. An
example of the latter is 2,4-D herbicide. Soluble organic
salts where the toxicant portion is an organic cation form a
second category. Movement of these chemicals in soils should
be similar to that of potassium or calcium cations except that
organic cations are usually more strongly sorbed than inorganic
cations. The quaternary bipyridylium herbicides, paraquat
and diquat, occur within that classification. The third cate-
gory is composed of highly water insoluble nonionic chemicals.

Movement of these materials in soils is particularly compli-
cated because they dissolve in both soil water and soil or-
ganic matter. Triazine herbicide is an example of this cate-
gory.

The dichloride salt of paraquat is a quaternary dipy-
ridylium herbicide which has the structural formula (Herbicide
Handbook, pages 137-141, 1967)

LCH3-N N-CH3 C 1

The toxicant portion of the molecule is an organic cation
with an electrical charge of +2 and a molecular weight of
186.2 g/mole. Paraquat is used effectively as a pre-emergence
herbicide, general contact herbicide, direct post-emergence
herbicide, crop desiccant, and as a crop defoliant. It is
normally applied in water as a spray at a rate of 0.5 to 1.0
lb of cation per acre (Calderbank, 1961) for general weed
control. The nonvolatile salt is a strong electrolyte and
largely dissociates in aqueous solution. Paraquat has the
distinction of being an almost completely water-soluble (70%
at 200C, Akhavein and Linscott, 1968) organic cation which
becomes very strongly adsorbed to materials with cation-ex-
change properties. It undergoes very rapid adsorption (Boon,
1965) with clay and organic matter constituents of soils.
Adsorption is primarily physical and is not dependent upon
pH, temperature, or exposure time (Akhavein and Linscott,
1968). Paraquat has a very long persistence (Herbicide Hand-
book, 1967) in soils which is only limited by microbial break-
down of the strongly sorbed chemical. Chemical properties
(1969), Boon (1965), Calderbank (1968), and Akhavein and
Linscott (1968).

The herbicide 2,4-dichlorophenoxyacetic acid (2,4-D)
is a phenoxyacetic acid with the structural formula (Herbi-
cide Handbook, 1967)

Cl O-C2-HCOOH

Cl
The molecular weight is 221 g/mole and the vapor pressure
at 160C is 0.4 mm Hg. Solubility in water ranges from 6
to 7% (Herbicide Handbook, 1967) over the temperature range
20 to 230C. Water, diesel oil or oil-water emulsion are com-
mon carriers for 2,4-D application. At low dosage rates of
0.25 to 1 lb/acre (Herbicide Handbook, 1967) 2,4-D is used
for pre-emergence control of weeds in crops, and at high dosage
rates of 3 to 4 lb/acre it is used in non-cropped areas as a

temporary soil sterilant against perennial weeds. It behaves
as an organic anion in aqueous solution. Adsorption occurs
with clay and organic matter constituents, but in sandy soils
it may be (Herbicide Handbook, 1967) readily leached. Low
dosage rates of 2,4-D undergo microbial breakdown in warm,
moist soil. Average persistence in soils is 1-4 weeks. Fur-
ther chemical properties of 2,4-D have been reported by Crafts
(1957 and 1961).

Physical-Chemical Properties of Soil-Water Systems that
Influence the Movement and Adsorption of Herbicides in Soils

A nonvolatile chemical applied to the surface of a soil
may move with water into the soil pore space. Because recom-
mended application rates to soils are generally low, most
organic herbicides dissolve at least partially in the soil
water. Therefore herbicides will be referred to here as
solutes. Mass flow (convection) and molecular diffusion are
physical processes (Kirkham, 1964) which move a given solute
through the soil pores. The behavior of a given herbicide
as it moves through the environment of a specific soil deter-
mines (Freed, 1966) the effectiveness of the chemical as a
weed-controlling agent and predicts the capacity of that chem-
ical to contaminate ground water. The soil environment de-
pends upon temperature, water content, pore size distribution,
ionic exchange capacity of the soil, pH of soil solution,
salt content of soil solution, organic matter content of soil,
clay content of soil, microbial activity, and many other vari-
ables. As the solute moves along tortuous pathways through
the soil, interactions such as adsorption to pore walls, chem-
ical precipitation, and biological degradation tend to remove
the herbicide from the moving solution. If the sorption is
reversible, the herbicide may be slowly released with time
into the mobile soil solution and the soil may be thought of
as a reservoir (Freed, 1966) for the chemical.

The flow velocity of the soil solution is important from
the standpoints of influencing the convection flow rate of
the solute and of determining the average detention time of
a solute molecule withinthe soil environment. Slower liquid
flow velocities give longer detention times which are favor-
able to sorption and other deactivating mechanisms. Tortu-
ous pathways of movement for a solute molecule also increase
the detention time. In water-saturated soil the pore size
distributions control the degree of tortuosity. In sandy
soils, decreasing the water content below saturation rapidly
increases tortuosity, which in turn gives increased solute
detention times and decreased liquid flow velocities (because
of decrease in percentage of pore space filled with water).
The probability for adsorption also increases with a decrease
in water content because the solute is forced to move in a
film of water in intimate contact with the electrically
charged pore wall surfaces.

Hartley (1964), Freed (1966), and LeGrand (1966) have
published on the behaviour of herbicide movement and reten-
tion in soil.

EXPERIMENTAL MATERIALS, METHODS
AND PROCEDURES

Herbicides

Carbon-14 labeled quantities of 2,4-D acid and paraquat
chloride were purchased as standard catalog items from commer-
cial suppliers of radioisotopes. The carboxyl carbon atom
of the 2,4-Dichlorophenoxy acetic acid molecule was labeled
with C-14, and the specific activity of the aqueous solution
purchased was 3.03 m Ci per mM (221 mg/mM). The methyl car-
bon atom of the paraquat chloride was labeled with C-14 and
was supplied as a freeze-dried solid under a nitrogen atmos-
phere in glass ampoules. Specific activity of the solid was
14.5 m Ci per mM (257 mg/mM). Analytical standards of non-
labeled paraquat chloride was provided free of charge from
the Ortho Division of Chevron Chemical Company, 940 Hensley
Street, Richmond, California 94804. Reagent grade non-
labeled 2,4-D acid was purchased from a chemical supplier.

Tritiated water and chloride-36 labeled NaC1 were also
used as inorganic tracers for the movement of the organic
herbicides through soil columns. Herbicides were applied
individually in aqueous solutions to the surface of columns
of soil, and for several of these investigations either H3-
labeled water or NaC136 was included with each influent
herbicide solution. Both of the radioactive tracers have
been widely used as tracers of water flow in soils. Because
of the relatively small physical-chemical reaction of H3 and
C136 during movement through moist soil, these materials
were selected as "non-reactive herbicides."

Soils

Soil materials used in this study were one organic soil,
a soil high in clay content, and several sands. Samples of
these materials were collected from various field locations
in Florida, air-dried, passed through a 2 mm sieve, and then
stored for future use. Samples of Lakeland fine sand and
Fellowship sandy clay were collected from'near Gainesville,
Everglades mucky peat was collected near Zellwood, Wabasso
and Oldsmar fine sands were collected near Ft. Pierce, and
Pomello and Blanton fine sands were collected from Orange
County. These soils are typical of large land areas in the
meter range from 105-210 pM (Cataphote Corporation, Jackson,
Mississippi) were used. Glass beads were selected as a "non-

reactive soil" because of the low adsorption properties of
the beads for many organic materials. Movement of an herbi-
cide through glass beads provides a valuable indication of
the influence of the mean pore size upon hydrodynamic dis-
persion under conditions of minimal herbicide sorption to
the pore walls. Each porous media was hand-packed into
cylindrical columns before each miscible displacement of herb-
icide.

Miscible Displacement Technique for Studying Herbicide Move-
ment in Soil Columns

Miscible displacement in porous media refers to the dis-
placement of one fluid by a second fluid when the two fluids
are mutually soluble (Collins, 1961, page 207). When the
fluids used are aqueous solutions where the displacing solu-
tion contains solute ions or molecules of interest, we may
speak of miscible displacement of solutes. During displace-
ment, the moving front undergoes spreading as a result of
mixing of the two solutions by molecular diffusion and con-
vection (mass flow). Gradients in solute concentration in
the vicinity of the advancing front or within individual pores
drive the diffusion, and convection occurs as a consequence
of the distribution of microscopic flow velocities within
the porous medium. If the solute or aqueous solution inter-
acts with the porous medium, miscible displacement can be
complicated by the processes of ionic exchange, anion repul-
total surface area exposed and the retention time for a sol-
ute in porous medium increase with increasing length of the
porous medium, the amount of mixing which occurs during dis-
placement of a solution through a column of porous medium
will be greater for longer columns.

Miscible displacement experiments in soil are generally
accomplished by applying a displacing solution of a chemical
to one end of a soil column. Flow velocities and water con-
tents through the soil are controlled. Small aliquots of
liquid effluent from the column are collected and analyzed
to determine the concentration of the displacing solute. The
solute concentration in the effluent plotted versus the vol-
ume of effluent collected give plots that have been termed
"breakthrough curves." Solute concentration is normally
expressed as a dimensionless ratio, C/C of the solute con-
centration in the effluent, and the initial solute concentra-
tion in the displacing solution. A dimensionless ratio, V/V ,
referred to as "pore volume" is sometimes used to express
the accumulative volume of effluent. Pore volume is obtained
by dividing the effluent volume by the volume of water in
the soil column under the conditions of the experiment.

A laboratory technique developed by Nielsen and Biggar
(1961, 1962, 1963) gives a convenient procedure for study of

miscible displacement of water and herbicide solutions applied
to soil columns. Movement of different chemical solutes may
be studied under controlled environmental conditions, such
as flow velocity or water content, with this procedure. In
this method, a solution containing a chemical of interest
is applied to one end of a soil column. Effluent from the
column is collected in small volume increments and each ali-
quot is then analyzed to determine the concentration of sol-
ute. If the displacing solution is applied as a pulse or
volume "slug" the curve will increase from C/C,=0 to a maximum
or peak and then decrease until C/Co=O is again reached.

In a discussion of theoretical considerations for mis-
cible displacement in porous media, Nielsen and Biggar (1962)
discussed the types of breakthrough curves predicted from
existing theories of dispersion. For the extremely ideal
situation where no mixing due to diffusion or convection occurs,
the tracer front proceeds by piston flow. For piston flow
the breakthrough curve is a vertical line passing through the
point at one pore volume of effluent. This type of transport
process would be expected for a single capillary tube of
constant radius, and cannot be expected to occur in soils.
But the model gives insight into the occurrence of mixing in
soil by establishment of the vertical line at one pore volume
as a point of reference for comparison with more realistic
types of breakthrough curves. If mixing results entirely
from convection (mass flow) in a nonreactive soil having rather
large microscopic flow velocities narrow in their distribu-
tion, a skewed sigmoid breakthrough curve will pass through
C/Co=0.5 at one pore volume. Coupling of molecular diffusion
with the convection during miscible displacement results in
translation of the curve to the left at the smaller flow
velocities as well as a slight clockwise rotation. Another
type of breakthrough curve occurs for miscible displacement
in nonreactive porous media having an extremely wide range
in velocity distribution. Assuming dispersion results from
both diffusion and convection, this type of curve is concave
to the right. For each type of breakthrough curve in non-
reactive media, the area enclosed beneath the curve and to
the left of a vertical line passing through one pore volume
is equivalent to the area enclosed above the curve but to
the right of that line.

Two types of curves are given for cases where the solute
and the soil interact. If the reaction is a chemical process,
a precipitation or an exchange, the sigmoid curve is displaced
to the right with only a small portion of the curve extend-
ing to the left of one pore volume. Exclusion of the solute
by a solute-solid interaction or a velocity distribution with
velocities near zero results in a sigmoid breakthrough curve
with a large displacement to the left of one pore volume.

A useful quantity referred to as holdbackk" has been

defined asthe area beneath the portion of the breakthrough
curve displaced to the left of the vertical line for piston
flow which occurs at one pore volume. For soils, holdback
is a measure of the volume of original soil solution not
displaced but remaining within the sample. Values for hold-
back vary from zero for piston flow to values less than 100%
for other cases. Nielsen and Biggar (1961) have shown that
the magnitude of holdback for water-unsaturated soils greatly
exceeds that for saturated soils. A slow approach to a max-
imum C/Co=l also occurs and the skewness of breakthrough curves
increase upon desaturation. They attributed part of this
to the extremely wide range in microscopic pore velocity and
a changing cross-sectional area between displacing and dis-
placed solutions for miscible displacement through unsaturated
soil. The influence of molecular diffusion upon dispersion
is greater for displacement through unsaturated soil than
for saturated soil.

Radioassay Procedure for Measurement of Herbicide Concentra-
tion in Soil Effluent

Chemical assays for 2,4-D and paraquat are frequently
time-consuming and require elaborate equipment. Radioassay
rather than chemical assay was used in this study for measure-
ment of herbicide concentrations. That method offers advan-
tages of being simple, being rapid, providing precise measure-
ment of very small quantities of herbicide, and requiring
access only to liquid scintillation counting equipment.

One ml aliquots of aqueous effluent from soil columns
were added to counting vials containing 15 ml of Bray's scin-
tillation solution (Bray, 1960). Both plastic and glass
vials were used. The vials were placed in a Packard Tri Carb
Liquid Scintillation Counter to determine the radioactivity.
Count rates (cpm) for soil effluent were corrected for back-
ground and were divided by the corrected count rate for the
initial influent herbicide solution. This relative concen-
tration, C/Co, was then plotted versus accumulative volume
of soil effluent to provide breakthrough curves for each
herbicide.

INVESTIGATIONS

Theoretical Analysis of the Movement of Solutes in Soils:
An Application of Transfer Functions

Introduction

Most recent efforts to provide an analytic description
of solute movement with water media have been directed at
solving the hydrodynamic dispersion equation as a boundary

value problem, with the boundary values determining the par-
ticular solution. An alternative approach is the method of
Laplace transfer function theory. This alternative approach
has been used with considerable success in other fields which
deal with diffusion-like problems. Basically, this theory
treats the problem of miscible displacement statistically,
and properties of the Laplace transform provide an analytic
form to the solute content of effluent from a soil column.
This treatment should provide additional understanding of
those processes which influence solute transport through
porous media.

Simple analytic examples illustrate the diverse utility
of the theory when applied to miscible displacement. In con-
trast to the boundary value solution, the transfer function
theory can be applied to cases of complex influent functions
with considerable ease. When applied to experimental data,
the transfer function theory gives results identical with
the boundary value solution; however, the transfer function
method requires considerably less effort, illustrating the
power of the method.

In this paper, the soil is treated as a continuous medi-
um in order to ascribe to it the average macroscopic proper-
ties of the soil system. The statistical nature of the
macroscopic viewpoint obscures the actions or mechanics of
individual pores on the macroscopic scale.

Probability Density

The statistical approach to the analysis of the flow of
solutes through porous media as performed by Day (1956),
Scheidegger (1957 and 1964) and Bear et al. (1968), provides
the probability density function for steady one-dimensional
flow.

This approach assumes that Darcy's Law holds for liquid
flow and that it follows from the central limit theorem, which
says that irregardless of the probability distribution of
each step in the process, after a sufficiently large number
of steps, the probability distribution tends to be Gaussian.
This leads to the probability density distribution of a sol-
ute "particle" to be

i(x,t) = (4)Dt)-1/2exp[-(x-vt)2/4Dt] 1.

where p(x,t) = probability density distribution as
a function of x and t.

x = spatial variable with its zero at the
entrance to the column and proceed-
ing in the direction of flow.

t = temporal variable with its zero at
the time that the solute "pulse"
enters the column at x = 0.

D = coefficient of hydrodynamic disper-
sion (not the coefficient of
molecular diffusion).

v = average liquid pore velocity or the
rate of advance of the center of the
"pulse".

It is convenient to transform equation 1 to a set of
coordinates traveling with the "pulse" center. The distribu-
tion function is changed only to the extent that in the new
system, the observer travels with the peak of the "pulse."
Using the center of mass coordinate, z = x-vt, equation 1
transforms to the form

i(z,t) = (4TDt)-1/2exp[-z2/4Dt] 2.

A quantity of immediate interest is the concentration C(z,t)
of solute in the soil solution, expressed as g per cm3.
Since i is the fraction of the total solute mass per unit
pore volume (total volume of solution present in soil column)
of the soil solution

C(z,t) = (z,t)Lm 3.

m = total mass of solute.

L = length of soil column.

A = cross-sectional area of soil
column.

e = volumetric water content of soil.

The concentration of a volume "slug" of solution applied to
the soil surface at time t = 0 is simply

C(o,0,, = L 4
LAB

Therefore the probability density is just the relative con-
centration of the solute or the concentration ratio c(z,t) =
C(z,t)/C(,O.). The concentration ratio is therefore a normal
Gaussian distribution

c(z,t) = *(z,t) = (4tDt)-l/2expE[-z/4Dt] 5.

with zero mean (x = vt) and variance 2Dt.

Impulse Response and Transfer Functions

It is often useful when dealing with the transmission
of aqueous solutions through porous media, to mentally visu-
alize a flow system which in effect neglects the microscopic
aspects of the medium and regards only macroscopic quantities
such as distance (in the one dimensional case), average liquid
flow velocity, concentration of solute, and some characteristic
quantity or quality of the system as a whole. The Laplace
transform, the impulse response function, and transfer func-
tion have been applied here to the problem of solute trans-
port in porous media for the purpose of giving a workable
macroscopic model. This method has been applied success-
fully in electronic circuit design and analysis, and mechan-
ics, and it is related to the Fourier transform used in ob-
taining the point spread function and the modulation transfer
function of optics, radar, sonar and other wave phenomena.
The Laplace transform method is an extension of the Laplace
transform equations used in solving boundary value problems
of diffusion and heat flow.

Consider the case of one-dimensional flow through a por-
ous medium. Assume that no information is given about the
microscopic nature of the porous medium, as this does not
aid us in determining the gross behavior. Let the solute be
applied in solution to a column of porous media with the
liquid flow velocity maintained constant with time. The
solute concentration of the influent will be taken as a func-
tion of time, Co(t) and will be referred to as the influent
function. Let Co(t) be properly behaved such as to possess
a Laplace transform which will be referred to as the influent
amplitude where

ho(s) = foo(t) e-Stdt 6.

and its inverse is defined by

o(t) = a+i ho(s)etds 7.
2iJ a-ioo

co(t) = influent function.

ho(s) = influent amplitude.

s = complex variable = c+iw where o
and w are real variables.

t = time variable.

i = v/-zi-

The constant a is chosen to be to the right of any singularity
of h (s) and the integration path is sometimes known as a

Bromwich contour. The response function of the system is
the concentration of solute in the effluent and will be re-
ferred to as the effluent function, c1(t). It will also
possess a Laplace transform which will be called the effluent
amplitude

hi(s) = c1(t) e-stdt 8.

and the inverse,
1t _a+im
c a(t)i hl(s)estds 9.
2(t) i J a_-i0

The effluent amplitude, h1(s), is related to the influent
amplitude, ho(s), by the transfer function, H(s), i.e.

h,(s) = H(s)ho(s) 10.

The transfer function has the property of a spectral
filter inasmuch as it modifies the spectrum (Laplace trans-
form) of the influent function to produce the spectrum of
the effluent function. Once the transfer function is known
for a specific flow system of a given porous medium, no other
information is necessary in order to predict the effluent
function for a known influent function. Note that so far these
are generalized functions with the only restricting condition
being that each possess a Laplace transform.

The usefulness of the transfer function will be illustrated
by showing how it can be found for select cases of influent
functions.

Consider an initially solute-free medium into which is
introduced a pulse of solute whose width may vary with time
but is small in time. The magnitude of the pulse is subject
to the condition

o 6T(t) dt = 1 11.

where 6T(t) is defined as

1/T 0 (t) = t>T 12.

6T(t) might represent a very short "slug" of solute whose
concentration is very high. The magnitude of 6 (t) is inverse-
ly proportional to its duration which is small in comparison
to some quantity which characterizes the duration of the efflu-

ent function (breakthrough curve). Thus the solute transport
may be described by

co(t) = 6T(t) 13.

and
h (s) = 1(1-e-T) 14.
sT

For a pulse of this nature, T is very small so that when the
right hand side of equation 14 is expanded in a series and
the limit taken as T approaches zero we have

ho(s) = 1 15.

Then the transfer function becomes

H(s) = lim h1(s) 16.
T + o

or is equal to the effluent amplitude when co(t) is a very
short pulse. This method provides an excellent procedure
for determining the transfer function for a specific flow
system as will be seen later. Under these specific condi-
tions the effluent function is simply the inverse Laplace
transformation of the transfer function, thus

ce(t) = L-'[H(s)] 17.

where L-1[] is the inverse Laplace transformation and L[] is
the Laplace transform. In general, the effluent function
will be

c,(t) = L- [H(s)L[co(t)]] = L[H(s)ho(s)] 18.

If for example co(t) ={0 t < o 19.
1 t > o

then from tables of Laplace transforms or by applying equa-
tion 6 directly we obtain

h (s) = 2 20.

Hence the effluent function becomes

ci(t) = L-'[H(s)L ] = g(t)dt 21.

where g(t) is the inverse Laplace transform of H(s) and is
the impulse response function. If co(t) is a long pulse,

that is, long in duration compared with the duration of g(t),
then

O (t) =/to 0 o I0 to
and by equation 14
l-e-sto
h (s) =- t 23.

Thus the effluent function is

c,(t) = L- H(s) (1-e-sto) 24.
toS
which gives i-
cw(t) =t g(t)dt t-tg(t-t )dt 25.

The last two examples suggest a second practical method of
obtaining the transfer function, H(s), or its inverse trans-
form, g(t), the impulse response function. From equation 21
we see that

g(t) = -Ec(t)] 26.
dt

when c (t) is a step function. By substitution for g(t) in
the definition of H(s) we obtain

H(s) = L[g(t)] = L[d[c(t)]] 27.

Thus one may easily apply solute in solution as a step func-
tion to the surface of a porous medium, differentiate the
resulting effluent function and perform the Laplace transform-
ation. The result then is the characteristic transfer func-
tion for the flow system.

Evaluation of inverse Laplace transforms can sometimes
be difficult with or without the use of tables of transforms,
so that an alternative method would be useful in evaluating
the effluent function from a known impulse response function
and influent function. The convolution integral provides
a convenient method. Equation 10 gives the product of two
transforms and its inverse transformation is given by

c,(t) = L-'[H(s)ho(s)] g(t-T g(t-T)c()dT = g(t) (t)28.

when is the notation for the convolution. The relationships

between the influent function, influent amplitude, effluent
function, effluent amplitude, impulse response function and
transfer function are presented in the schematic diagram in
Figure 1. Horizontal arrows indicate Laplace transforms and
are reversible when the inverse Laplace transform is taken.
Vertical arrows -represent the convolution integral on the
left (not generally reversible) and multiplication by the
transfer function (reversible) is given on the right. Thus
there are two pathways from a known influent function to a
desired effluent function when an impulse response function
is known. One way involves direct use of the convolution
and the second way involves a Laplace transformation, a mul-
tiplication and an inverse Laplace transformation.

Applications to Miscible Displacement

A practical application of the transfer function theory
to solute transport in soils is miscible displacement of
herbicides and fertilizer nutrients in soils. Potentially,
this field promises to be one of its most important uses.

Consider the effluent function, c (t), given by equation
5 when z = L-vt where L is the column length. This is the
effluent function c,(z,t) for a case when the influent func-
tion is an infinitesimal thickness pulse, so that equation
12 applies. c~(z,t) is the impulse response function of a
system whose length is L and the system characteristic is
the coefficient of hydrodynamic dispersion, D. Thus

g(z,t) = (4fVDt)-1/2exp[-(L-vt)2/4Dt] 29.

and the Laplace transform of g(z,t) is the transfer function

H(s) = H(z,s) = 2(Ds)-1/2exp [-z(s/D)1/2] 30.

In practice it will sometimes be easier to obtain the efflu-
ent functions by the convolution integral

C1(z,t) = (4TDT)-1/2exp -T 2/DT] co(t-T)dT 31.

The integral cannot be evaluated until the form of Co(t)
is known due to the convolution. However, for a specified
co(t), c,(z,t) can be evaluated by analytical or numerical
methods. If co(t) is a step function as described by equa-
tion 19, c1(z,t) is given by

c(z't) = p1/2 (L-vt) (L-vt) erfcL 32.
If (t) takes the form of equation 22, then c ,t) takes

If Co(t) takes the form of equation 22, then c,(z,t) takes

Inverse Laplace Transform

----- ---- --Laplace Transform
Laplace Transform

ho(s)

Inverse Laplace Transform
- --------------- ----------

* g(t)

Laplace Transform

H(s) x

Inverse Laplace Transform
-----------------------

Laplace Transform

co (t)

A

-H(s)

I

hi(s)

Figure 1. Relationships Between
c0(t), h (s), c,(t), h,(s), g(t), and H(s)

The symbol stands for convolution.

the form of equation 25 and using g(z,t) from above we get

(z1,t) t 1/2 (L-vt) (L-vt) erfL-vt
c(zt) =- 1 lexp (Dt 2D erfc(t
jto D 4Dt 2D ____

33.

-(t-to' /2exp (-v(t-to))2 +L-v(t-to) erfc L-v(t-to)\
T rD 4D(t-to) 2D /4D(t-to)/

If Co(t) is periodic and of the form of equation 22 with
period, 2to, it can be written as

00
co(t) = 1/to z (-1)kS(t)
k=0 kto

k = 0,1,2,3...

where

S(t) = fO
kto fl

0 t>kto

In this case, it is more convenient to determine c2(z,t)
by the transform method. By that method,

h1) (-l)ke-ks
h,(s)- e e
sto k=0

S2 Lz D x ,-1/2l o k -ks
h1(z,s) = exp -z(s/D) 2 (-1) e
to0DS I k=0

35.

36.

The resulting effluent function is

c(z,t) =- t

exp(L-vt) 2
e 4Dt '

(L- ) erfc vt) k (-1) Skt (t)
2D Dt k=0 0

37.

Note that equation 37 is equivalent to equation 33 when k =
0,1.

Results and Discussion

In order to have any practical value, soil transfer
function theory must predict the concentration of solute in
the effluent from a soil column when characteristic parameters
of the specific flow system are given and the solute concen-
tration in the liquid influent is given as a function of time.

and

Alternatively, if the impulse response is known for the flow
system then the characteristic parameters -L, v, and D- can
be recovered by a Least Squares curve fitting routine. Of
course, it is easier to work with a situation where the in-
fluent function is known and the parameters of L, v, and D
have been measured. This case may be illustrated using the
data from Davidson et al. (1968) where fluometron pesticide
and chloride were infiltrated through a 30 cm long column
of 250 pm diameter glass beads. Figure 2 shows the calculated
impulse response function using the parameters of the experi-
mental flow system. This is the theoretical effluent func-
tion which would be obtained if the influent had been a very
narrow solute "pulse" of unit area. This has been calculated
from equation 29 and normalized to unity. Using the same
parameters, the effluent function was calculated for a 200
ml "slug" from equation 33 and this is shown in Figure 3 along
with experimental curves for fluometron and chloride. The
curve which Davidson calculated by solving the diffusion equa-
tion is also given. Note that both theoretical curves fall
to the right of the data indicating perhaps an occluded vol-
ume, as Davidson has suggested.

It might be noted that this work uses time, t, as the
independent variable and that it was necessary to use the
following relation in order to calculate the abcissa, pore
volume, Vp, for Figure 3.

V = 38.
P L

Also it might be noted that when the impulse response func-
tion is used, the maximum value of the effluent function
occurs when
L
tmax v 39.

and thus for a unit area pulse, the amplitude of the peak
on the breakthrough curve is given by

c (o,tmax) = (4wDtmax)-1/2 40.

from which the value of the dispersion coefficient, D, can
be recovered.

Conclusions

Given values of length, L, and dispersion coefficient,
D, for the specific soil and solute under observation and the
average flow velocity, v, of the fluid, transfer function
theory can be used to calculate breakthrough curves (plots
of relative solute concentration versus pore volume of efflu-
ent) for elution of herbicides or fertilizer solutes from

Pore Volume
0.5

S1.0-
I D=0.123 cmr/h
C-
0 a8 L=30.0 cm
U
I v=4.81 cm/h
3
L .6-

C
SA-
0.
n.
4)
2 .2-

QO 1 2 3 4 5
E ( r
Time (hrs)

Figure 2. Impulse Response Function

Pore Volume
0.5

1.5

jv pk UIu a3 Dt=US o
Fluometuron
Chloride
Calculated by Davidson
and by this work
= 0.123 cm/h
= 30.0 cm
=4.81 cm/h
0

1 2 3 4 5 6 7 8 9 10

Time (hrs)

Figure 3. Effluent Function

0
o

-

L

v

I-) nc

columns of soil. The function describing the solute con-
centration in the effluent can be calculated with excellent
accuracy if the mathematical form of the solute concentra-
tion of the influent is known. This method provides a means
to predict the effluent without actually performing the ex-
periment in the field, when such experiments are not practi-
cal. It is based solely on statistical hydrodynamics and
transfer function theory and is not directly subject to the
boundary and initial conditions as are solutions to the dif-
ferential equations.

Miscible Displacement of 2,4-D Herbicide Through Water-
Repellant Soils

Many agricultural sands of Florida exhibit resistance
to wetting with liquid water. Soils with this property are
referred to as being water-repellant. With time water will
move through most water-repellant sands, although wetting
often proceeds along channels leaving pockets of dry soil
in between the channels. Since water is the principal car-
rier for the transport of non-volatile herbicides through
soils, water-repellancy should alter the movement and con-
sequent adsorption of a soil-applied herbicide. Aqueous
solutions of dilute 2,4-D herbicide plus tritiated water were
applied as influent slugs to water-saturated columns of
water-wettable, water-repellant, ignited (water-repellant
sand was ignited at 600C for 12 hours), and ignited plus
silicone (R-20 Silicone, Union Carbide Corporation, 270
Park Avenue, New York 17, New York) coated Blanton fine sand.
Movement and adsorption were evaluated by means of breakthrough
curves of 2,4-D and tritiated water in the column effluents.

The dry soil materials were hand-packed into a glass
column of 2.54 cm inside diameter and 25 cm length. Average
bulk densities of 1.47, 1.52, 1.57, and 1.56 g/cm3, respec-
tively, were obtained for columns of the wettable, water-
repellant, ignited and silicone coated materials. The columns
were then wet with distilled water. Long time periods were
required to wet the water-repellant and silicone coated por-
ous materials. A variable-flow pump was used to establish
constant liquid flow velocities of 2.04 and 4.08 cm/h through
the columns. The column influent was initially water, but
at a time we shall call zero a 10 ppm solution of C14
labeled 2,4-D was introduced to the column. After a 25 ml
"slug" of the 2,4-D solution had entered the column, the
influent was changed to water.

Breakthrough curves of 2,.4-D are presented only for
wettable and water-repellant Pomello fine sands (Figures 4
and 5)for liquid flow velocities of 2.04, and 4.08 cm/h.
Recoveries of the applied 2,4-D in the effluents for columns
of all four porous media are given in Table 1. Increasing
the flow velocities from 2.04 to 4.08 cm/h resulted in increased

WETTABLE POMELLO

V=4-08 CM/H

V= 204 CM/H

A- -&

/1

0 20 40 60 80
ML

100 120 140 160

Figure 4. Breakthrough Curves of 2,4-D Corresponding
to 2 Liquid Flow Velocities in Water-Wettable Pomello Fine Sand

1-0

0-8

0.6

C/C,

WATER REPELLANT POMELLO

V-4.08 CM/H

0.6

G/C4
0.4

A-
/
A

'7
I
/

A- -----A V 2.04 CM/H

\

\-

0 20 40 60 80 100 120 140 160

ML
Figure 5. Breakthrough Curves of 2,4-D Corresponding to 2 Liquid
Flow Velocities in Water-Repellant Pomello Fine Sand

Table 1. Tabulation of Recoveries of Applied 2,4-D
in the Effluent From Columns of Water-Repellant,
Water-Wettable, Ignited, and Ignited
Plus Silicone Coated Samples of Pomello Fine Sand

Recovery of 2,4-D

% 2,4-D Reed.

Soil

2.04 cm/h

Water Repellant

Wettable

Ignited

Silicone Coated

82.0

76.0

86.3

98.5

4,08 cm/h

85.5

83.2

95.0

99.8

2,4-D recoveries for all columns. At the slower velocity,
recoveries of 2,4-D were 76.0, 82.0, 86.3, and 98.5%,
respectively, for wettable, water-repellant, ignited, and
silicone-coated Blanton fine sand. Thus the recovery of
2,4-D was almost complete for the silicone-coated sand re-
gardless of the liquid flow velocity. The 2,4-D recovery
from displacement through naturally water-repellant sand was
less than that for the ignited sand but greater than for the
water repellant sand. Ignition of the sand removed the organic
matter, and thus soil adsorption of 2,4-D was therefore de-
creased during movement. The lower 2,4-D recovery in the
effluent from the wettable soil as compared to that for the
water-repellant soil was attributed to incomplete and non-
uniform water-saturation. After termination of the experi-
ments, small zones of dry soil were observed within the water-
repellant column. The incomplete wetting should give a
smaller volume of soil pores participating in transport of
water and 2,4-D. The quantity of 2,4-D adsorbed during water
flow through the wettable soil was therefore greater than
for the water-repellant soil.

Solutions of 10 ppm 2,4-D and tritiated water were also
displaced through columns of wettable, water-repellant, and
ignited plus silicone-coated Pomello fine sand. Breakthrough
curves of 2,4-D and tritiated water for the water-repellant
and silicone-coated columns are presented in Figures 6 and
7. Elution of the 2 chemicals were similar for the silicone-
coated sand, but elution of tritiated water and 2,4-D from
the water-repellant sand column gave separate breakthrough
curves. Tritiated water appeared in the effluent before
the 2,4-D and the greater spread for the 2,4-D curve indicates
adsorption by the soil. Behaviour of 2,4-D and tritiated
water curves for wettable sand (not shown) were similar to
the curves in Figure 3.

Work performed on the study indicates that water-repellancy
in water-saturated sands influences the movement of 2,4-D.
Further studies are needed to evaluate movement of herbicide
solutions into initially dry water-repellant soils.

Miscible Displacement of 2,4-D Herbicide During Constant
Liquid Flow Velocity Through Initially Dry Soils

A theoretical model was developed to predict the movement
of organic chemicals through porous materials packed in columns
of finite length. Movement due to diffusion was assumed to
be negligible. The model assumes that transport of the chem-
ical is mainly due to the liquid flow velocity and that at
equilibrium adsorption is linear. The model also considers
the rate at which equilibrium between the dissolved and the

The theory was tested by studying th.e movement of C-14
labeled 2,4-D herbicide through glass beads C105-210 .m),

WATER REPELLANT POMELLO

1.0

0-8

0-6
C/C,

0-4

0-2

0-0

. .. .. .. .

0 20 40 60 80
ML

-- 3 H

A-----A C1 2,4-D

V = 2.04 CM/HR

\

100 120 140 160

Figure 6.

Breakthrough Curves of 2,4-D and Tritiated Water
in Water-Repellant Pomello Fine Sand

U!l:; E OfCOATED I: L"

V 2.04 CM/

S1 20

,i~":

, ,
A

Figure 7. Breakthrough curves of 2,4-D and Tritiated Water
in Ignited and Silicone Coated Pomello Fine Sand

1-0

0-61

C/C,

0.2

Li .IS.
: '" '''' P~sCI

Lakeland fine sand, Fellowship sandy clay and Everglades
mucky peat, under constant flow velocities. Air dry soils
were packed uniformly in a column of 7.5 cm inside diameter
and 30 cm length which was held vertical during the studies.
A photograph of one of the soil columns is shown in Figure
8. A dilute solution (10 mg/l) of 2,4-D was introduced at
the bottom of the column with constant liquid flux main-
tained at 50 ml/h with a positive displacement pump. The
same flux was continued until 200 ml of solution had entered
the soil, at which time the solution was immediately replaced
by pure water containing 0.5% phenol. Thus, a "volume slug"
of herbicide solution was displaced upward in the column by
water at the same flow rate. Studies were also made at a
liquid flux of 100 ml/h. Effluent was collected in equal
volume aliquots for analyses. Soil solution was extracted
at the midway point (15 cm elevation) with a specially con-
structed sampling device using a porous plate and partial
vacuum. The extract was collected periodically in amounts
of 25 microliters. Effluent samples and the extracts were
analyzed for 2,4-D content with a liquid scintillation count-
ing system. Random effluent samples were subjected to
additional analyses for 2,4-D to determine if the herbicide
was degraded in the soil columns. The analyses were done
by thin layer chromatography. The thin layer chromatographic
studies revealed that the herbicide did not disintegrate dur-
ing transport through the soil.

Experimental breakthrough curves of 2,4-D applied as a
500 ml "slug" of aqueous solution to columns of glass beads
and Lakeland fine sand are shown in Figures 9 and 10. The
almost rectangular shape of the curve for elution from glass
beads indicates that the herbicide underwent only limited
retention and the herbicide movement closely paralleled the
water movement. The breakthrough curve for displacement
through Lakeland fine sand is very wide which indicates in-
teraction of 2,4-D with the porous medium. The low value of
/C o (approximately 0.2) for the first aliquot of column efflu-
ent indicates that the movement of 2,4-D lagged behind that
for the water. The herbicide also lingered in the soil efflu-
ent for longer periods than for the glass beads. These two
curves were included in this report as being representative
of data reported in the Ph.D. dissertation of V. Jyothi (1971).

Theoretical and experimental breakthrough curves of herb-
icide concentration were expressed as functions of accumulative
volume of solution introduced at the bottom of the columns.
These elution curves were plotted for effluent and soil ex-
tracts at both liquid flow velocities. It was observed that
the theoretical curves compared fairly well with experimental
results. However, some discrepancy was observed between calcu-
lated and experimental curves during the initial time periods.
Calculated concentrations were higher than the observed con-
centrations in the effluent. This indicated that there was

*m

-,r gi a
Nobr.AUea io ole- i son

GLASS

105o-200

V.o 0.22 CM/H

600 700 800

Figure 9. Experimentally Determined Breakthrough
for a 500 ml "Slug" of 2,4-D Applied as Influent to
of Initially Dry Glass Beads at a Liquid Flow Velocity

Curve
a Column
of 0.22 cm/h

C/c.

ML.

LAKELAND SAND

V,*0-22 CM/H

0 100 200 500 400 500 600 700 800

Figure 10. Experimentally Determined Breakthrough Curve
for a 500 ml "Slug" of 2,4-D Applied as Influent to a Column of
Initially Dry Lakeland Fine Sand at a Liquid Flow Velocity of 0.22 cm/h

1-0

0-8

0-6

C/C.

0-41

02

higher adsorption on the porous material during the wetting
stage since all the surface area was readily available for

Observed concentrations were slightly higher than the
calculated values near the tail end of the curve. This pro-
longed skewness was attributed to diffusion which is likely
to show in a run of longer time period but the simplified
model does not account for diffusion. In addition, hysteresis
may exist in the partition coefficient during adsorption and
desorption. Calculated and observed peaks of maximum
herbicide concentration appear at almost the same time. This
mainly depended upon the partition coefficient determined in
the spreading of the breakthrough curves.

fine sand and Fellowship sandy clay. Most of the herbicide
was recovered within two pore volumes of effluent in these
soils. Movement was very much hindered in Everglades mucky
peat. It took more than 5 pore volumes to recover only 60%
of applied 2,4-D. Adsorption of 2,4-D appeared to be irrever-
sible in the organic soils but not in the mineral soils.

Sterilization of Intact Soil Columns Using Exposure to Gamma

Sterilization of intact soil columns permits a convenient
means for eliminating microbiological effects during miscible
displacement of herbicide solutions through soil columns.
The influence of specific physical and chemical variables
upon herbicide movement may therefore be easily evaluated in
sterilized columns of soil.

Two methods commonly employed for sterilization of soil
are radiation with gamma rays and fumigation with methyl bro-
mide gas. Eno and Popenoe (1964) observed that a radiation
dose in excess of 106 rads (one rad is the dose of any
100 ergs per gram of soil) is required for sterility. Corey
et al. (1967) used gamma radiation to sterilize intact columns
of Webster-silica sand soil. Solutions of chloride and
nitrate anions were then miscibly displaced through both
sterilized and non-sterile soil columns. Recoveries of applied
nitrate were 100 and 66%, respectively, for sterile and non-
sterile columns. Steam, methyl bromide fumigation, and gamma
radiation methods were experimentally compared by Eno and
Popenoe (1964) on disturbed soil samples.

The effectiveness of methyl bromide fumigation and gamma
radiation sterilization methods were evaluated in this study
for intact columns of Oldsmar fine sand.

Experimental Methods and Procedure

Air-dry samples of Oldsmar fine sand (taken from 0.10
cm depth in the field profile) were hand packed into glass
columns of 2.54 cm internal diameter and 30 cm length. Sec-
tions of silicon rubber tubing were attached to each end of
the columns. The columns were hermetically sealed by attach-
ing a rubber septum to the end of each section of tubing.
Two of the air-tight columns were placed in the sample con-
tainer of the Research Food Irradiator, Food Science Depart-
ment. The radiation source in the facility was 105 Ci of
cesium-137 which gives a dosage rate of 2 x 105 rads per
hour in the center of the stainless steel sample container
(15 cm x 30 x 45 cm). A diagram of the irradiator is pre-
sented in Figure 11. The soil columns were exposed to the
radiation field for a period of 20 hours to give a calculated
total dose of 4 x 106 rads. After sterilization these columns
were stored for later incubation analysis.

Two other soil columns were fumigated with methyl bromide
gas in a laboratory hood. A pressurized can of the gas was
purchased from a garden store for this purpose. Thick-walled
tygon tubing was used to connect a valve which was connected
to the outflow from the methyl bromide container to a 40 liter
glass carboy and the carboy was connected by tygon tubing
to a glass Y-tube. Both forks of the tube were connected
to large syringe needles which were inserted into the rubber
septums at the ends of the two columns. Needles attached to
separate sections of tygon tubing were also inserted into
the septums at the gas outflow end of the columns. The out-
flow ends of each piece of tubing were placed at about 1 cm
depth in a beaker of water. Visually monitoring the relative
flow rate of gas bubbles in the water permitted a rapid deter-
mination of the approximate gas flow rate in the columns.
The flow system was first tested with air flow to test for
leaks; however, the test revealed the system functioned proper-
ly. The valve was then opened on the methyl bromide container
and allowed to move into the soil columns, initially flushing
air from the pores. Rubber gloves, a long-sleeved lab coat,
and protective eye goggles were worn during the fumigation
as safety precautions. Also the glass door of the hood separated
the experimenter from the experiment. The columns were left
exposed to methyl bromide for a 24-hour period, after which
the needles were removed from the septums and the sealed
columns were stored for later incubation analyses.

Samples of radiated, fumigated and control soils were
incubated in several growth media to determine the growth of
microorganisms. Seven media were specific for the growth of
aerobic organisms and one medium was specific for the growth
of anaerobic organisms. Trypic soy agar, Bacillus medium,
sporulation agar, mycological agar, peptone yeast trypticose
agar, mannitol agar, and gelatin agar were the culture media

Figure 11. Schematic Diagram of University of Florida

BACKFILL

-S.S. TANK 6FT DIA.x IIFT DEEP

for aerobes. The culture media and procedure for anaerobes
was that described by Smith (1964). Approximately one gram
of radiated, fumigated or non-treated soil was placed upon
the growth media in each petri dish. Incubation was allowed
to proceed for seven days at 280C. For a control, 1 ml of
steam sterilized distilled water was placed on the growth
media in petri dishes. The steam sterilized media were ini-
tially poured into steam sterilized petri dishes in a trans-
fer chamber equipped to provide a positive outflow of steri-
lized air. At the end of the incubation period the presence
or absence of growth was recorded for each petri dish. Ob-
servations were visually determined under a microscope at
2X magnification. The binomial data was reported as ratios,
R, of the number of petri dishes with growth to the total
number of dishes. The ratios were transformed by the rela-
tion X = arc sinVR before statistical analysis of the data.

Results and Discussion

Data for the study is shown in Table 2 and statistical
analysis is presented in Tables 3 and 4. Both the methyl
bromide fumigation and the gamma irradiation treatments of
the soil columns were shown to be highly effective means for
sterilization.

The two treatments were observed to be equally effective
methods of sterilization for both aerobes and anaerobes.

Miscible Displacement of Paraquat Herbicide, C136, and
Tritiated Water Through Sterile and Nonsterile Soil Columns

Steam sterilized aqueous solutions of 10 ppm C14 labeled
paraquat herbicide and tritiated water were displaced through
3 columns each of methyl bromide fumigated, gamma irradiated
and unsterilized Oldsmar fine sand (0-10 cm depth). Air dry
soil was packed in glass columns of 2.54 cm inside diameter
and 30 cm length to give a soil bulk density of 1.00 g per
cm3. Each column was water-saturated one day before displace-
ment was to begin and water was then pumped through the column
at a velocity of 5.2 cm per h for 24 hours. The purpose of
the study was to determine if sterilization influenced the
retention of the paraquat or tritiated water during movement
through the soil. Soil columns were sterilized by placing
each intact column of air-dry soil within a high intensity
field of gamma radiation (Cs'37) or by flushing gaseous methyl
bromide through the intact columns. Sterilization procedures
are described previously in this report. Three successive
applications of paraquat and tritiated water solution were
applied to water-saturated soil columns. As expected for
this soil, which had 8% organic matter, no paraquat was re-
covered in the liquid effluent from radiation sterilized,
methyl bromide sterilized, or unsterilized columns due to
strong adsorption by the soil particles. However, tritiated

Table 2. Growth of Microorganisms in Samples
of Fumigated, Irradiated, and Non-treated Oldsmar Fine
Sand Incubated with Seven Culture Media for Aerobes
and Anaerobes

The data are reported as values of R, ratio of number
of samples with growth observed to the total number of samples.

Replicate No. 1

Replicate No. 2

Culture Media

specific for aerobes:

1. trypic soy agar

2. Bacillus medium

3. sporulation agar

4. mycological agar

5. peptone yeast -
trypticose agar

6. mannitol agar

7. gelatin agar

specific for anaerobes:

Fumi-
gated
Soil

ated
Soil

Non-
treated
Soil

12.92 12.92 69.30

12.92 12.92 69.30

12.92 12.92 69.30

26.56 12.92 69.30

12.92 12.92 69.30

12.92 12.92 69.30

26.56 12.92 69.30

12.50 12.92 69.30

gated ated
Soil Soil

12.92 12.92

12.92 26.56

12.92 12.92

12.92 12.92

26.56 12.92

12.92 12.92

26.56 12.92

12.50 12.50

Non-
treated
Soil

69.30

69.30

69.30

69.30

69.30

91.7

Table 3. Analysis of Variance for Growth of Aerobic
Microorganisms in Sterilized and Non-treated Oldsmar Fine
Sand During Incubation with Seven Growth Media.

The data were transformed with the relation X = arc sin /R-.

Degrees of
Freedom

Mean
Squares

F-value

Replicates

Media

Treatments

4.45

16.21

13,608.00

Error

Total

0.23

0.85

721.50**

18.86

** Indicates significance at 1% level of probability.

Table 4.. Analysis of Variance for Growth of Anaerobic
Microorganisms in Sterilized and Non-treated
Oldsmar Fine Sand During Incubation with a Single
Growth Media

The data were transformed with the relation X = arc sin /R .

Degrees of
Freedom

Replicates

Treatment

Error

Total

Mean
Squares

0

3683.40

F-Value

water recovery was pronounced as three successive breakthrough
curves (one for each application) in the effluent from all
soil columns. Percentage recovery of tritiated water was
very nearly the same for each of the three successive curves
(Figures 12 and 13) for the sterilized columns, but recovery
progressively decreased with the 3 successive curves (Figure
14) from the unsterilized soil column. Recovery of tritiated
water in the effluent from the unsterilized column was 92.5,
86.1, and 81.2%, respectively. Although data reported in
Figures 1, 2 and 3 represent average values from 3 replicate
soil columns, a fourth column of unsterilized soil was pre-
pared as a check. This column was water-saturated and three
successive 5 ml "slugs" of an aqueous solution of NaC136 and
tritiated water were applied as influent to the column. A
constant liquid flow velocity was maintained through the soil,
and water was applied as influent prior to and following each
slug. Breakthrough curves of Cl36 and tritiated water in
the column effluent are given in Figure 15. Note that the
Cl36 recovery was 99.2, 98.1, and 98.0%, respectively, for
the successive "slugs;" whereas the tritiated water recovery
was 95.2, 92.2, and 86.1%, respectively. The chloride anion
recovery was constant and almost complete, but the recovery
of tritiated water decreased for the successive "slugs."
For a given "slug" the breakthrough curves indicate that the
chloride appeared in the effluent prior to the tritiated water.
The tritiated water also lingered in the effluent after elu-
tion of chloride has ceased. Corey et :al. (1963) also ob-
served separation of chloride and tritiated water breakthrough
curves for single displacements through water-saturated columns
of sandstone. They observed that separation was greater for
lower than for higher liquid flow velocities due to molecular
diffusion effects.

The decrease in recovery of tritiated water with succes-
sive "slug" applications as influent to columns of unsterilized
Oldsmar fine sand was definitely related in some way to micro-
biological activity. Since microorganisms multiply at very
rapid rates (population may double within one-half hour under
correct conditions) the population of soil microbes probably
increased considerably during the 2 to 3 days required to
displace three successive "slugs" through each unsterilized
soil column. We therefore postulate that the observed de-
crease in recovery of tritiated water was due to increased
microbial retention in the unsterilized soil with increasing
time. Skewness of breakthrough curves for tritiated water
in Figures 3 and 4 imply that retention by some mechanism did
occur. Comparison of symmetry for these curves to. those in
Figures 1 and 2 show that minimal retention of tritiated water
occurred in the fumigated or irradiated soil columns.

Displacement of Paraquat and Diquat Herbicides by KC1 Solution
from Water-Saturated Columns of Soil

The toxicant portions of paraquat and diquat herbicides

Oldsmar fine sand: sterilized with gamma
A B
recovery recovery
o paraquat: 0 0/o 0 0/o

3
H : 99.0 0/o

C
recovery
0 o/o

98.9 0/o 98.7 /o
A^A
A A AA
A
A A
A A

A A
A a
A
AA
A
A
A A

A
A AA
AA
A A

A A
A

0 1 / 2
A--1 2 3 4 5

B ----- 1 2 3 4

C BO 1 2
Pore Volume, V/

Figure 12. Breakthrough Curves for C14 Labeled Paraquat and Tritiated
Water in the Effluent From a Column (Average of Data From
Three Replicates) of Gamma Irradiated Oldsmar Fine Sand
After Three Successive Influent Applications of "Slugs"

0

UoJ
O.

L

u0
C

0
U

Oldsmar fine sand: sterilized with

A
recovery
paraquat: 0 o/o

H3: 974 /o

A A
A A

0

oo

U\
C (0
00

L
4-0
U
0
Ucj
0

A-0

B
recovery
0 0/o
97.1 /o

P A
A A

A A

A
A A
"AA A
As aa

methyl bromide
C
recovery
0 0/0

96.7 o/o

A A
A
A A

A
A

1 2 3 4 5

I
B 0
C

Pore Volume, V/
V.

Figure 13. Breakthrough Curves for C14 Labeled Paraquat and Tritiated Water
in the Effluent From a Column (Average of Data from Three Replicates)
of Methyl Bromide Fumigated Oldsmar Fine Sand After Three Successive
Influent Applications of "Slugs"

A
a
a
a"

a

~"a,

/

Oldsmar fine sand: unsterilized

A
recovery
o paraquat: 0 0/o
3925
H: 92.50/o

A
A
A
A
A

1
Pore Volume, V/V

Figure 14. Breakthrough Curves for C14 Labeled Paraquat and Tritiated Water
in the Effluent From a Column (Average Data from Three Replicates)
of Non-Sterilized Oldsmar Fine Sand After Three Successive Influent
Applications of "Slugs"

B
recovery
0 /lo

86.1 /o

0

ci)
\o
O

0

u

ud

C
recovery
0 /0o

81.2 /o

4
a
a
a
a
A a

A
aI
A
A
a
A
A

a

66A

A
A
A A
dnl^

B -0

A

a
a^

Oldsmar fine sand: unsterilized
A B
recovery recovery
Cl 99. 20/ 98.10/o
H3 95. 2%/ 92.2%/o

C
recovery
98.00/0
86.1 /o

I I
B 0 1
C F
0

1
Pore Volume, Vv

Figure 15. Breakthrough Curves for Cl36 and Tritiated Water in the Effluent
From a Column of Non-Sterilized Oldsmar Fine Sand After Three Successive
Influent Applications of "Slugs"

1.0-

are divalent organic cations which undergo rapid physical ad-
sorption with soil components. (Weber et al., 1965; Weber and
Weed, 1968; Tucker et al., 1967; Coats and Funderburk, 1966;
and Knight and Tomlinson, 1967). A short study.was performed
to determine if these chemicals could be desorbed from soil
columns by aqueous solutions of KC1.

Air-dry Wabasso fine sand taken from 0-10 and 33-76 cm
profile depths were packed into glass columns of 2.54 cm in-
side diameter and 30 cm length. The organic matter contents
of these two soil materials were determined to be 1.3 and
0.1%, respectively, both composed of greater than 97% sand.
The average bulk density for the two columns was 1.32 g per
cm3. The columns were saturated with distilled water and a
variable-flow pump was used to maintain a fluid flow velocity
of 6 cm per hour. After one day of flow, the influent was
changed from water to an aqueous solution of 10 ppm paraquat
for one column and diquat for the other. Both chemicals were
labeled with C14. After one hundred ml of herbicide solution
entered the columns, the influent was switched to water. Water
was allowed to displace the herbicides into the column until
5 pore volumes of column effluent had been collected by an
automatic fraction collector. At that time the influent was
switched to a 746 ppm aqueous solution of KC1.

Breakthrough curves for paraquat and diquat in the efflu-
ents of columns of 0-10 and 33-76 cm soil materials are
presented in Figures 16 and 17. Between 0 and 5 pore volumes
neither of the chemicals appeared in the effluent for either
soil material. The paraquat and diquat were thus assumed
to be completely adsorbed in the soil columns. After 5 pore
volumes, paraquat occurred in the effluent from the 0-10 cm
soil but the recovery was very small. None of the diquat
appeared in the effluent. For the 33-76 cm soil both herbi-
cides appeared in the effluent with a sharp breakthrough;
however, the amount of paraquat exceeded that of diquat.
Small quantities of the chemicals remained in the effluent
even after 20 pore volumes of liquid flow.

Paraquat and diquat, to a lesser extent, were observed
to be desorbed from Wabasso fine sand by KC1 solution. Thus,
it would appear that applications of fertilizer to this soil
could influence the desorption and consequent movement of
paraquat and diquat in the soil solution.

Equilibrium adsorption isotherms of paraquat were per-
formed for Everglades mucky peat, water-repellant and ignited
(600C for 12 hours) Blanton fine sand, and Wabasso fine sand.
Five grams (oven dry basis) of soil were placed with 25 ml of
C14 labeled aqueous solutions of known concentration of para-
quat into centrifuge tubes. The tubes were placed on a mechani-

Wabasso fine sand:O-10cm depth

herbicide

recovery

S104 paraquat 0.0009 0/o
diquat 0.0 0/o

C
Co

-5

0
0 5 10 15 20
Pore Volumes, V
0V

Figure 16. Displacement of Paraquat and Diquat Herbicides by KC1
From a Column of Wabasso Fine Sand, 0-10 cm Profile Depth

Wabasso fine sand:33-76 cm depth
herbicide
o paraquat
-4

C

5x10

0
0 5 10 15 20
Pore Volumes, V

Figure 17. Displacement of Paraquat and Diquat Herbicides by KC1 From
A Column of Wabasso Fine Sand, 33-76 cm Profile Depth

cal shaker for one hour and centrifuged for 15 minutes. The
concentration of paraquat in the supernatent liquid was deter-
mined by placing one ml in 15 ml of Bray's counting solution
and counting the radioactivity with a liquid scintillation
counter. The concentrations of paraquat in the initial solu-
tions ranged from zero to 500 ppm (pg/ml).

The resulting adsorption isotherms are presented in
Figures 18 and 19. For linear isotherms, the slope gives
the value R, the distribution coefficient.. Sorption was
clearly linear over the range of 0 to 500 Pg/ml of paraquat
in solution and had an R value of 6 ml/g. The water-repellant
Blanton soil was linear over the range of 0-100 pg/ml with
an R of 4.8 ml/g, but the ignited Blanton was linear over
approximately 0 to 50 .ig/ml with an R of about 3 ml/g. Over
the narrow range of 0-10 pg/ml paraquat adsorption was linear
for Wabasso fine sand with an R. of 3 ml/g. Thus adsorption
by the peat .soil was clearly greater than for either of the
sand soils used. Ignition of the Blanton fine sand removed
all soil organic matter and greatly decreased the adsorption
of paraquat. The assumption of linear adsorption for dilute
solutions of paraquat is valid for the organic soil and the
two sands used.

Adsorption experiments were also performed with solutions
of 17 and 51 yg/ml paraquat when various concentrations of
KC1 were present in the paraquat solutions. The influence
of 0-9000 ppm KC1 concentrations upon paraquat adsorption
in Everglades mucky peat, water-repellant and ignited Pomello
fine sand, Wabasso fine sand, and water-repellant and ignited
Blanton fine sand is presented by curves in Figures 20 and
21. The level of KC1 concentration in the initial solution
mucky peat at either level of paraquat concentration. This
was expected because of the tremendous affinity of soil organic
matter for paraquat. At the 51 ppm level of paraquat, ad-
sorption decreased sharply and curvilinearly with increased
level of KC1 for Wabasso fine sand. At the 17 ppm paraquat
concentration, adsorption decreased only slightly but linearly
as the KCl concentration was increased from 0 to 9000 ppm
for Wabasso soil. Paraquat adsorptions by ignited samples
of Pomello and Blanton fine sands were much less than for
the unignited counterparts. However, it is important to note
that destruction of the organic matter did not eliminate all
adsorption of paraquat. Although Pomello and Blanton soils
are classified as sands, small quantities of clay and silt
are present in the sand. An increase in KC1 concentration
in the 51 ppm paraquat solution gave a small linear decrease
crease for unignited Pomello and Blanton fine sands, respec-
tively. Thus for high levels of KC1 concentration present
in the soil solution, paraquat adsorption in sands may be less
than if no salts were in the soil water. This observation
has important implications to movement of paraquat in sandy

0
Ln

a Blanton fine sand: unignited
A Blanton fine sand:ignited

O IIl Ii ll

0 50 100 150 0 2 250 300 350 400 450
Paraquat in Initial Solution, /9 ml

Water-Repellant Blanton Fine Sand, and Ignited Blanton Fine Sand

30-
0E Wabasso sand
,-
o 25
E

20-
20
O
6- O
an
WL 15-
O

10-

S5

0 | i -- I --- i -- --- ------ -- i --- i
0 1 2 3 4 5 6 7 8 9 10

PARAQUAT CONCENTRATION (pg per ml)

Adsorption Isotherm of Paraquat for Wabasso Fine Sand

Figure 19.

solution concentrations

7m,1 paraquat

t 11

o Pomello:unignited
* /1 :ignited

o Wabasso

0(0
0)
X

0

-Qm
o

L
0
()

CT
l...
=!

^<

2 3 4 5 6 7
KCI Initially in Paraquat Solution, [mg/

8 9

Figure 20. Influence of KC1 in Solution Upon Adsorption of Paraquat in
Everglades Mucky Peat, Water-Repellant and Ignited Pomello Fine Sand,
and Wabasso Fine Sand

---51

---17

-- e s-

----- ---
I------------ --- --------~--U --- -- -- -- 6----------------- -- --*

soils

In -

F__ --~

I

2000 4000 6000 8000

KCI in Solution

(mg/ I)

Figure 21. Influence of KCI in Solution Upon Adsorption of Paraquat
in Water-Repellant and Ignited Blanton Fine Sand

0 o

4-
O

O

0

40
CT O
L

0

soils. For example, under conditions of intense fertilizer
application, paraquat may move further down the profile than
otherwise expected.

Continuous Measurement of Chloride in Effluent Flowing from
a Soil Column

Miscible displacement of chloride solutions through a
soil column normally requires the use of a fraction collector
to collect aliquots of effluent and some means for measure-
ment of the Cl" concentration in each aliquot. Yoo and Kirkham
(1971) described a liquid scintillation flow cell for contin-
uously recording Cl36 concentration in soil effluent. Their
method has advantages of eliminating the need for fraction
collectors, provides instantaneous concentration measurements
without need for sample storage and provides a continuous
recording of concentration; however, the method has the in-
active. An inexpensive flow cell method is described in
this study which does not require that the chloride be radio-
active.

Experimental Apparatus

An Orion Model 96-17 Combination reference and chloride
electrode provides the transducer for the method. The con-
centration range of the electrode is 1 to 5 x 10,5M chloride
and the minimum sample size is 10~2 ml. The outer sleeve of
the electrode is constructed of unbreakable plastic which is
resistant to most solvents and is resistant to mechanical
shock or stress. The reference portion of the combination
electrode produces a stable, drift-free reference potential
and low stirring and junction potentials.

A flow cell was constructed from a cylindrical section
of lucite plastic. A schematic of the flow cell and the com-
bination electrode is presented in Figure 22. A water-tight
seal around the sleeve of the electrode was insured by the
use of silicon rubber sealant rather than the rubber 0-ring
as shown in the schematic. The 0-ring has the advantage of
providing easy removal of the electrode from the flow cell
for cleaning purposes.

The electrical voltage from the combination electrode
was measured by a pH meter and the output from the meter was
connected to a strip chart recorder. The transducer was
calibrated by measuring the voltage for a range of chloride
solutions of known concentration. The resulting calibration
curve is shown in Figure 23. Calibration was performed in
the flow cell except that the solutions were stationary during
voltage measurements. Selected points from a strip chart
recording of the transducer voltage can thus be converted
to activities of chloride through the calibration curve.

COMBINATION REFERENCE

& CL

ELECTRODE

RING

LUCITE FLOW CELL
A

OUTPUT*--

INPUT

FLOW

CELL

FOR CL MEASUREMENT

Figure 22. Schematic Diagram
Reference and Chloride
a Lucite Flow

Showing Combination
Electrode in
Cell

Calibration Curve of Chloride

Electrode

E

--, +
e0

] -o
O0
-4 0

WIJ

Concentration of NoCI, moles per liter

Figure 23. Calibration Curve for the Combination Reference
and Chloride Electrode

To test the transient response of the flow cell and
transducer system a concentrated aqueous solution of NaCI
was displaced through a column of glass beads. A glass
chromatography column with 2.54 cm internal diameter and
30 cm internal length was packed with 105-210 pm glass
beads. The bulk density of the packed column was 1.57
g/cm3. The column was saturated with 5 ppm NaCI solution,
and a variable flow pump was used to maintain a flow velocity
of 7.0 cm/h of the dilute chloride solution through the
porous media. After one day the 5 ppm influent solution was
replaced by a 5005 ppm solution. A 2 ml "slug" of the con-
centrated NaC1 solution was allowed to enter the column be-
fore switching back to the original dilute chloride influent.
During the miscible displacement of the concentrated slug
of chloride through the column the electrode voltage was
recorded on the strip chart recorder. The chart speed was
maintained at 31.17 cm/h. Column effluent flowing from the
flow cell was collected in 2 ml aliquots in glass tubes on
a fraction collector. Samples from the aliquots were later
titrated with AgNO3 to obtain the concentration of chloride
in solution. The titration measurement provided a means for
checking the concentrations obtained by the electrode method.

Results and Discussion

The breakthrough curve (relative concentration of
chloride versus the number of pore volumes of column efflu-
ent) for concentrated chloride slug displaced through the
column of glass beads is presented in Figure 24. Zero
pore volume corresponds to the time at which the slug was
first introduced as influent to the column. Circles on
the graph correspond to concentrations determined by titra-
tion of effluent aliquots, and the smooth curve was provided
by connecting concentrations determined by the transducer.
Very close agreement between the two methods is clearly
shown. Although the breakthrough curve is skewed, the ali-
quot concentrations also indicate a similar skewness. Integ-
ration of the curve by Simpson's rule indicated 96.0% re-
covery of the chloride slug. This large recovery percentage
is only 4% less than the expected value of 100% and thus is
an indication that the electrode method gave realistic con-
centrations of chloride.

Conclusion

An inexpensive flow cell and combination chloride electrode
system is described for continuously recording the chloride
concentration in effluent flowing from a column of soil. A
breakthrough curve for chloride solution displaced through a
column of glass beads was in close agreement with a correspond-
ing breakthrough curve obtained by titrati.on of aliquots of
the same effluent.

Displacement of NaCI through a column of

C= 5000ppm CI
/= 1.57 g/cm'

ao
0
O
0_

bo

U
0

Pore Volume,
Pore Volume, V

Figure 24.

Breakthrough Curve of Chloride in Effluent from Column

V= 7.0 cm/h
recovery of Cl = 96.0%o

o titration of aliquots
- continuous measurement

1 .

OL
0

ACKNOWLEDGMENTS

The investigators were assisted by the following
authors and co-authors:

R. M. McCurdy, Graduate Student in Physics;
V. Jyothi, Graduate Student in Soil Science;
P. G. Hunt, Research Associate in Soil Science; and
A. Elzeftawy, Graduate Assistant in Soil Science.

LITERATURE CITED

1. Akhavein, A. A. and D. L. Linscott. 1968.
The dipyridylium herbicides, paraquat and diquat.
Residue Reviews, Vol. 23:97-145.

2. Bear, J., D. Zaslavsky, and S. Irmay. 1968.
Hydrodynamic dispersion. Physical Principles of
Water Percolation and Seepage, Publication 29 of
Arid Zone Research Series, UNESCO, Paris, pages 307-349.

3. Boon, W. R., 1965. Diquat and paraquat new agricultural
tools. Chemistry and Industry, May 8, 1965, pages 782-
788.

4. Bray, G. A., 1960. A simple efficient liquid scintillator
for counting aqueous solutions in a liquid scintillation
counter. Anal. Biochem. 1:279-285.

5. Calderbank, A., 1968. The bipyridylium herbicides.
Advances in Pest Control Research, Vol. 8:127-235.

6. Coats, G. E., H. H. Funderburk, Jr., J. M. Lawrence, and
D. E. Davis. 1966. Factors affecting persistence and
inactivation of diquat and paraquat. Weed Res. 6:58-66.

7. Collins, R. E., 1961. Flow of Fluids Through Porous
Materials. Reinhold Publishing Corporation, New York, N.Y.

8. Corey, J. C., D. R. Nielsen, and J. W. Biggar. 1963.
Miscible displacement in saturated and unsaturated
sandstone. Soil Sci. Soc., Amer. Proc. 27:258-262.

9. Corey, J. C., D. R. Nielsen, J. C. Picken, Jr., and Don
Kirkham. 1967. Miscible displacement through gamma
Science and Technology, Vol. 1:144-147.

10. Crafts, A. S., 1957. The chemistry and mode of action
of herbicides. Advances in Pest Control Research, Vol.
1:39-79.

11. Crafts., A. S., 1961. The chemistry and mode of action
of herbicides. Interscience Publishers, New York. 269 pages.

12. Davidson, J. M., C. E. Rieck, and P. W. Santelman. 1968.
Influence of water flux and porous material on the move-
ment of selected herbicides. Soil. Sci. Soc., Amer.
Proc. 32:629-633.

13. Davidson, J. M., and P. W. Santelman. 1968. Displacement
of Fluometuron and Diuron through saturated glass beads
and soil. Weed Science 16:544-548.

14. Day, Paul R., 1956. Dispersion of a moving salt-water
boundary advancing through saturated sand. Transactions,
American Geophysical Union, Vol. 37:595-601.

15. Eno, C. F., and Hugh Popenoe. 1964. Gamma radiation
compared with steam and methyl bromide as a soil
sterilizing agent. Soil Sci. Soc. Amer. Proc. 28:533-535.

16. Freed, V. H., 1966. Chemistry of herbicides. Pesticides
and Their Effects on Soils and Water, American Society
of Agronomy, Special Publication Number 8, pages 25-43.

17. Funderburk, H. H., Jr. 1969. Diquat and paraquat.
Degradation of Herbicides, Marcel Dekker, Inc., New York,
N.Y., pages 283-298.

18. Hartley, G. S., 1964. Herbicide behaviour in the soil.
The Physiology and Biochemistry of Herbicides.
Academic Press, New York, pages 111-161.

19. Herbicide Handbook of the Weed Society of America. 1967.
W. F. Humphrey Press, Inc., Geneva, New York.

20. Jyothi, V., 1971. Miscible displacement of 2,4-D
herbicide during constant liquid flow velocity into
initially dry soils. Unpublished Ph.D. dissertation,
Soil Science Department, University of Florida, Gainesville.

21. Kirkham, Don. 1964. Some physical processes causing
movement of ions and other matter through soil. Overdruk
Uit De Mededelingen Van De Landbouwhogeschool En De
Opzoekingstations Van De Staat Te Gent. DEEL XXIX; paper
presented at 15th Annual Phytopharmacy Symposium, May
6, 1963, Ghent, Belgium.

22. Knight, B. A. G., and T. E. Tomlinson. 1967. The inter-
action of paraquat (l:1'-dimethyl 4:4 -dipyridylium
dichloride) with mineral soils. J. Soil Sci. 18:233-243.

23. LeGrand, H. E., 1966. Movement of pesticides in the
Soil. Pesticides and Their Effects on Soils and Water,
American Society of Agronomy, Special Publication Number
8, pages 71-77.

24. Nielsen, D. R., and J. W. Biggar. 1961. Miscible dis-
placement in soils. I. Experimental information. Soil
Sci. Soc. Amer. Proc. 25:1-5.

25. Nielsen, D. R., and J. W. Biggar. 1962. Miscible dis-
placement in soils. II. Theoretical considerations.
Soil Sci. Soc. Amer. Proc. 26:216-221.

26. Nielsen, D. R., and J. W. Biggar. 1963. Miscible dis-
placement in soils. III. Mixing in glass beads. Soil
Sci. Soc. Amer. Proc. 27:10-13.

27. Scheidegger, A. E., 1957. Physics of Flow Through Porous
Media. The MacMillan Co., New York, N.Y., pages 31-35
and 197-202.

28. Scheidegger, A. E., 1964. Statistical hydrodynamics in
porous media. Advances in Hydroscience, Vol. 1:161-181.

29. Smith, P. H., 1964. Pure culture studies of methanogenic
bacteria. Proceedings, 20th Industrial Waste Conference,
Purdue University, Lafayette, Indiana.

30. Tucker, B. V., D. E. Pack, and J. N. Ospenson. 1967.
Adsorption of bipyridylium herbicides in soil.
J. Agr. Food Chem. 15:1005-1008.

31. Weber, J. B. and S. B. Weed. 1968. Adsorption and de-
sorption of diquat, paraquat, and prometone by mont-
morillonitic and kaolinitic clay minerals. Soil Sci.
Soc. Amer. Proc. 32:485-487.

32. Weber, J. B., P. W. Perry, and R. P. Upchurch. 1965.
The influence of temperature and time on the adsorption
of paraquat, diquat, 2,4-D, and prometone by clays,
charcoal, and anion exchange resin. Soil Sci. Soc.
Amer. Proc. 29:678-688.

33. Yoo, Sun-Ho and Don Kirkham. 1971. Flow Cell System for
Miscible Displacement Experiments. Water Resources
Research. 7:211-213.

Full Text

PAGE 1

Publication No. 16 Movement and Adsorption of Pesticides in Sterilized Soil Columns By R.S. Mansell and L.c. Hammond Soils Department, [FAS ( I Watpr Research Center Univeru Ja

PAGE 2

MOVEMENT AND ADSORPTION OF PESTICIDES IN STERILIZED SOIL COLUMNS By R.S. MANSELL (Princi pal Investigator) and L.c. HAMMOND PUBLICATION NO. 16 of the FLORIDA WATER RESOURCES RESEARCH CENTER RESEARCH PROJ ECT TECHNICAL COMPLETION REPORT OWRR Project Number A-FLA Annual Allotment Agreement Numbers 14 01-0001-1628 (1969) 14-31-0001-3009 (1970) 14-31-0001-3209 (1971) Report Submitted: August 9, 1971 The work upon which this report is based was supported in part by funds provided by the United States Department of the I nterior, Office of Water Resources Research as Authorized under the Water Resources Research Act of 1964

PAGE 3

TABLE OF CONTENTS ABS.TRACT SUMMARY. INTRODUCTION AND REVIEW OF LITERATURE. Movement of 2 and Paraquat Herbicides in Soils. .. ..... .. Chemical Properties of Paraquat and Herbicides. . . .. Physical-Chemical Properties of Soil-Water Systems that .Influence the Movement and Adsorption of Herbicides in Soils . EXPERIMENTAL MATERIALS, METHODS AND PROCEDURES Herbicides. Soils Miscible Displacement Technique for Studying Page 1 2 4 4 5 7 8 8 8 Herbicide Movement in Soil Columns. 9 Radioassay Procedure for Measurement Df Herbicide Concentration in Soil Effluent. 11 INVESTIGATIONS Theoretical Analysis of the Movement of Solutes in Soils: An Application of Transfer 11 Functions . . 11 Miscible Displacement of 2,4-DHerbicide Through Water-Repellant Boils . Miscible Displacement of 2,4-D Herbicide During Constant Liquid Flow Velocity Through 24 Initially Dry Soils . 28 Sterilization of Intact Soil Columns Using Exposure to Gamma Radiation and Methyl Bromide Fumigation. . . 35 Miscible Displacement of Paraquat Herbicide, C136, and Tritiated Water ThrDugh Sterile and Nonsterile Soil Columns .. . . 38

PAGE 4

Displacement of Paraquat .and Diquat Herbi.cides by.KCl Solution from Water.Saturated Columns of Soil . Soil Adsorpti.on of Paraquat Herbi.cide Continuous Measurement of Chloride in Effluent Flowing from a Soil Column ACKNOWLEDGMENTS. LITERATURE CITED Page 42 47 55 60 61

PAGE 5

ABSTRACT MOVEMENT AND ADSORPTION OF PESTICIDES IN STERILIZED SOIL COLUMNS Rapid transport of systemic and soil sterilant herbicides in soil during periods of net water flow may decrease the effectiveness of the chemicals to control unwanted vegetation and produce undesirable pollution of the ground water. An investigation of the influence of physical-chemical soil properties upon the transport of 2,4-D and paraquat in columns of organic and sandy soils was therefore performed. These herbicides are water soluble organic chemicals which are used extensively in agriculture. The toxicant portions of 2,4-D and paraquat behave as anion and cation, respectively. Miscible displacement of aqueous solutions of these herbicides through columns of Everglades mucky peat resulted in most of the 2,4-D and all of the paraquat being removed from solution by adsorption. Limited transport of 2,4-D was observed for the same fine sands. Very small quantities of organic matter in the fine sands effectively removed paraquat from the flowing soil solution. The presence of large concentrations of KCl in the soil solution was observed to decrease the quantity of paraquat sorbed. Mathematical transfer function theory was used in connection with statistical hydrodynamics to develop a technique for analysis and prediction of herbicide elution from soil columns during miscible displacement experiments. Mansell, R. S., and L. C. Hammond MOVEMENT AND ADSORPTION OF PESTICIDES IN STERILIZED SOIL COLUMNS Completi.on Report of the Office of Water Resources Research, Department of Interior, August,. 1971, Washington, D. C. 20240 KEYWORDS: paraquat/ pesticide movement/ herbicides/ adsorption/ water pollution/ ground water. 1

PAGE 6

SUMMARY Chemicals applied to crops or soils for the purpose of killing weeds may be classified as contact, systemic or soil sterilant herbicides. Contact materials require direct application to foliage of _the target plants; whereas, systemic chemicals may be applied directly to foliage or indirectly to the soil. Systemic herbicides may be absorbed through leaves or roots and may be translocated through the entire plant system. Soil sterilants prevent plant growth when present in the soil. Thus, most herbicides used in agriculture reach the soil, whether by direct application or indirectly as residual loss from spray applications to the plant leaves. During periods of rainfall or irrigation these chemicals may move with water downward through the soil. If an herbicide moves through the soil profile to the water-saturated zone beneath the water table, the groundwater may become contaminated. The objective of this research project was to evaluate the influence of specific physical-chemical properties of the soil environment upon movement and consequent adsorption of 2,4-D and paraquat herbicides in agricultural sands. Properties investigated were water flow velocity, initial soil water and so11 organic matter content. Paraquat and 2,4-D chemicals were selected for this research because of their extensive use as contact, systemic and soil sterilant herbicides. Both are highly water-soluble; therefore, they are suspects for leaching and movement through soils. In aqueous solution, however, and 2,4-D mOlecules respectively, form organic cations and anions, which are highly toxic to plants. As these ions move in water through the interconnected pores of soils, retention may occur as physical and chemical sorption to the pore walls. For soils high in clay mineral or organic matter content the divalent cation of paraquat is very strongly adsorbed. Adsorption of the 2,4-D anion is much less than for paraquat. A technique based upon statistical hydrodynamics and Laplace transfer function theory was developed to mathematically describe the movement of herbicides or other solutes through columns of soil. For miscible displacement of herbicide solutions, through columnq, experimentally determined breakthrough curves (concentration of herbicide in the effluent as a function of time) may be used to calculate a transfer function for a known influent application fUnction. The transfer function is characteristic of the specific soil flow system, and it how the influent concentration function is modified to give the effluent concentration function during displacemerit through the soil. From the transfer function the value of the hydrodynamic dispersion (includes effects of convection, diffusion, and retention mechanisms such as adsorption) may easily be calculated. The 2

PAGE 7

transfer function technique also provides a means to predict breakthrough curves (or effluent concentration functions) of a given herbicide for a given soil when the dispersion coefficient, the flow velocity, the column length, and the soil transfer function are known. Movement of 2,4-D in columns of moist fine sand was fourid to be influenced by water-repellancy of the sand. Displacement of a volume "slug" of 2,4-D solution through columns gave maximum recovery in the effluent from water-repellant Blanton fine sand, less recovery in ignited Blanton, and even less for naturally water-wettable Blanton. The larger recovery of 2,4-D in the effluent from the water-repellant sand was attributed to incomplete and non-uniform water-saturation of the packed column. When this column was dismantled at the end of the experiment small zones of dry soil were observed within the moist soil. Thus the water-filled porosity was less in the water-repellantsoil and thus adsorption of 2,4-D should be less. Ignition removed the soil organic matter; therefore, 2,4-D adsorption should be low. Increasing the liquid flow velocity resulted in greater herbicide recoveries in the effluents from all the porpus materials. Miscible displacement experiments showed 2,4-D to move readily with water through initially dry columns of glass beads, Lakeland fine sand, and Fellowship sandy clay. Most of the chemical was recovered within two pore volumes of effluent in these soils. Movement of 2,4-D greatly lagged water movement in columns of Everglades mucky peat. More than five pore volumes were required to recover only 60% of soil-applied 2,4-D. Adsorption of 2,4-D appeared to be reversible in the mineral soils but not so in the organic peat. Intact columns of dry Oldsmar fine sand were sterilized by fumigation with methyl bromide gas and by irradiation with an intense field of gamma rays. Both treatments gave complete sterilization and provide convenient means for evaluating the influence of specific physical-chemical soil properties upon miscible herbicides through soil columns. Displacement of three successive "slugs" or paraquat and tritiated water through water-saturated columns of irradiated, fumigated and non-treated Oldsmar fine sand gave unexpected breakthrough curves for the tritiated water. The successive curves for the sterilized soil gave almost complete recovery, but for the untreated soil recovery decreased with each "slug" applied. The phenomenon was attributed to rapid growth of microorganisms in the non-sterile soil. The breakthrough curves for tritiated water behaved similarly when Cl-36 and tritiated water were applied as three successive "slugs" to another non-sterile column. A definite separation was observed for Cl-36 and tritium curves. Tritium lagged the chloride anion in the effluent, as might be expected. 3

PAGE 8

The concentration of KCl in the soil solution was found to influence movement and sorption of the paraquat cation in columns of soils. Application of a dilute Kel solution to water-saturated columns of Wabasso fine sand resulted in elution of small quantities of adsorbed paraquat and diquat. Desorption of both chemicals was greater from Wabasso fine sand taken from 33-76 cm profile depth than for sand taken from 0-10 cm depth. The organic matter content in the surface soil is approximately 10 times greater than in the material taken from 33-76 cm. Increasing concentrations of Kel over the range from 0 to gOOD ppm in aqueous solutions of paraquat resulted in decreasing adsorption of paraquat by Pomello, Wabasso and Blanton fine sands. Initial solution concentrations of 17 and 51 ppm paraquat were used. The Kel concentration had little effect upon paraquat adsorption in Everglades mucky peat, ignited Pomello fine sand, and ignited Blanton fine sand. With no KCl present in the solution adsorption isotherms were linear for the organic peat over the range of o to 500 ppm paraquat in the initial solution phase. Linear sorption of paraquat by the fine sands was limited to narrower solution concentration ranges. The assumption of linear adsorption is therefore valid for these soils for the recommended application rates of paraquat for herbicidal purposes. The presence of large concentrations of KCl or other salts in the soil solution following fertilizer applications to agricultural sands may stimulate limited profile movement of adsorbed paraquat. An inexpensive flow cell method is described for continuously recording the concentration of chloride in effluents from soil columns. The method was potential for miscible displacement studies which involve chloride as a "tracer" for herbicide movement in soils. In conclusion, movement of 2,4-D and paraquat with water in sandy soils investigated indicate that both chemicals undergo retention due to sorption by small amounts of soil organic matter. Paraquat was completely sorbed in most cases; whereas 2,4-D did move through soil columns under most conditions. At the very low concentrations of 2,4-D normally applied to these soils, significant amounts of the chemical would not be expected to traverse field profiles to contaminate ground water. INTRODUCTION AND REVIEW OF LITERATURE Movementof2 ,4-DandParaquat Herbicides in: Soils Organic pesticides applied directly or indirectly to 4

PAGE 9

surface soil may move with water downward through the profile to ultimately contaminate ground water. The pollution potential of an individual pesticide is a complex function of properties of the soil-water system which contribute to solute transport and properties of both the pesticide and the soilwater system which contribute to attenuation. Movement of one of these chemicals into and through porous media occurs primarily by mass flow (convection) coupled with molecular diffusion. These two mechanisms also cause a chemical to undergo dispersion (mixing) with water in the soil pores. Mixing of a solute during flow of a liquid through porous media is referred to as hydrodynamic dispersion (Bear, et al., 1968). As the pesticide moves through the soil pores, adsorption, fixation, precipitation, degradation, and other attenuating mechanisms tend to remove the chemical from the flowing stream of soil solution. Those materials which interact strongly with the soil and which also resist decomposition ar.e commonly called persistent pesticides (Van Middelem, 1966). The capacity of a soil to adsorb chemical molecules has been cited (Shaw, 1966) as one of the most mechanisms in man's total environment. Properties of a specific pesticide such as solubility in water, ionic charge, etc., also determine the extent of attenuation during movement through the soil-water environment. Van Middelem (1966) states that herbicides present a special problem of pesticide movement in soils since many are applied directly to the soil as selective pre-emergence sprays and as nonselective soil sterilants. Intensive rainfall and sandy soils in Florida create a favorable environment for contamination of ground water with soil-applied herbicides. This research investigation was performed to determine the influence of specific physical-chemical properties of soil-water environments upon the movement of 2,4-D and paraquat herbicides through columns of sandy soil. Chemical Properties of Paraquat and 2,4-D Herbicides Chemical properties have been used by Goring (Kirkham, 1964) to classify pesticides into three broad categories as far as movement in soil by water is concerned. The first category includes chemicals with a reasonably high water solubility, which are nonionic or the toxicant portion is anionic. Movement of these materials in soil should be somewhat similar to that of th.e inorganic anions, nitrate and chloride. An example of the latter is 2,4-D herbicide. Soluble organic salts where the toxicant portion is an organic cation form a second category. Movement of these chemicals in soils should be similar to that of potassium or calcium cations except that organic cations are usually more strongly sorbed than inorganic cations. The quaternary bipyridylium herbicides, paraquat and diquat, occur within that classification. The third category is composed of highly water insoluble nonionic chemicals. 5

PAGE 10

Movement of these materials in soils is particularly complicated because they dissolve in both soil water and soil organic matter. Triazine herbicide is an example of this category. The dichloride salt of paraquat is a quaternary dipyridylium herbicide which has the structural formula (Herbicide Handbook, pages 137-141, 1967) The toxicant portion of the molecule is an organic cation with an electrical charge of +2 and a molecular weight of 186.2 g/mole. Paraquat is used effectively as a pre-emergence herbicide, general contact herbicide, direct post-emergence herbicide, crop desiccant, and as a crop defoliant. It is normally applied in water as a spray at a rate of 0.5 to 1.0 lb of cation per acre (Calderbank, 1961) for general weed control. The nonvolatile salt is a strong electrolyte and largely dissociates in aqueous solution. Paraquat has the distinction of being an almost completely water-soluble (70% at 20C, Akhavein and Linscott, 1968) organic cation which becomes very strongly adsorbed to materials with cation-exchange properties. It undergoes very rapid adsorption (Boon, 1965) with clay and organic matter constituents of soils. Adsorption is primarily physical and is not dependent upon pH, temperature, or exposure time (Akhavein and Linscott, 1968). Paraquat has a very long persistence (Herbicide Handbook, 1967) in soils which is only limited by breakdown of the strongly sorbed chemical. Chemical properties of paraquat have been published by Crafts (1961), Funderburk (1969), Boon (1965), Calderbank (1968), and Akhavein and Linscott (1968). The herbicide 2,4-dichlorophenoxyacetic acid (2,4-D) is a phenoxyacetic acid with the structural formula (Herbicide Handbook, 1967) Cl The molecular weight is 221 g/mole and the vapor pressure at 160C is 0.4 mm Hg Solubility. in water ranges from 6 to 7% (Herbicide Handbook, 1967) over the temperature range 20. to 23C. Water, diesel oil or oil-water emulsion are common for 2,4-D application. At low dosage rates of 0.25 to 1 lb/acre (Herbicide Handbook, 1967) 2,4-D is used for pre-emergence control of weeds in crops, and at high dosage rates of 3 to 4 lb/acre it is used in non-cropped areas as a 6

PAGE 11

temporary soil sterilant against perennial weeds. It behaves as an organic anion in aqueous solution. Adsorption occurs with clay and organic matter constituents, but sandy soils it may be (Herbicide Handbook, 1967) readily leached. Low dosage rates of 2,4-D undergo microbial breakdown in warm, moist soil. Average persistence in soils is 1-4 weeks. Further chemical properties of 2,4-D have been reported by Crafts (1957 and 1961). Physical-Chemical Properties of Soil-Water Systems that Influence the Movement and Adsorption of Herbicides in Soils A nonvolatile chemical applied to the surface of a soil may move with water into the soil pore space. Because recommended application rates to soils are generally low, most organic herbicides dissolve at least partially in the soil water. Therefore herbicides will be referred to here as solutes. Mass flow (convection) and molecular diffusion are physical processes (Kirkham, 1964) which move a given solute through the soil pores. The behavior of a given herbicide as it moves through the environment of a specific soil determines (Freed, 1966) the effectiveness of the chemical as a weed-controlling agent and predicts the capacity of that chemical to contaminate ground water. The soil environment depends upon temperature, water content, pore size distribution, ionic exchange capacity of the soil, pH of soil solution, salt content of soil solution, organic matter content of soil, clay content of soil, microbial activity, and many other variables. As the solute moves along tortuous pathways through the soil, interactions such as adsorption to pore walls, chemical precipitation, and biological degradation tend to remove the herbicide from the moving solution. If the sorption is reversible, the herbicide may be slowly released with time into the mobile soil solution and the soil may be thought of as a reservoir (Freed, 1966) for the chemical. The flow velocity of the soil solution is important from the standpoints of influencing the convection flow rate of the solute and of determining the average detention time of a solute molecule within the soil environment. Slower liquid flow velocities give longer detention times which are favorable to sorption and other deactivating mechanisms. Tortuous pathways of movement for a solute molecule also increase the detention time. In water-saturated soil the pore size distributions control the degree of tortuosity. In sandy soils, decreasing the water content below saturation rapidly increases tortuosity, which in turn gives increased solute detention times and decreased liquid flow velocities (because of decrease in percentage of pore space filled with water). The probability for adsorption also increases with a decrease in water content because the solute is forced to move in a film of water in intimate contact with the electrically charged pore wall surfaces. 7

PAGE 12

Hartley (1964), Freed (1966), and LeGrand (1966) have published on the behaviour of herbicide movement and retention in soil. Herbicides EXPERIMENTAL MATERIALS, METHODS AND PROCEDURES Carbon-14 labeled quantities of 2,4-D acid and paraquat chloride were purchased as standard catalog items from commercial suppliers of radioisotopes. The carboxyl carbon atom of the 2,4-Dichlorophenoxy acetic acid molecule was labeled with c-14, and the specific activity of the aqueous solution purchased was 3.03 m Ci per roM (221 mg/mM). The methyl carbon atom of the paraquat chloride was labeled with c-14 and was supplied as a freeze-dried solid under a nitrogen atmosphere in glass ampoules. Specific activity of the solid was 14.5 m C1 per roM (257 mg/mM). Analytical standards of nonlabeled paraquat chloride was provided free of charge from the Ortho Division of Chevron Chemical Company, 940 Hensley Street, Richmond, California 94804. Reagent grade nonlabeled 2,4-Dacid was purchased from a chemical supplier. Tritiated water and chloride-36 labeled NaCl were also used as inorganic tracers for the movement of the organic through soil columns. Herbicides were applied individually in aqueous to the surface of columns of soil, and for several of these H3_ labeled water or NaC1 36 was included with each influent berbicide solution. Both of the radioactive tracers have been widely used as tracers of water flow in soils. Because of the relatively small physical-chemical reaction of H3 and C136 during movement through moist soil, these materials were selected as "non-reactive herbicides." Soils Soil materials used in this study were one organic soil, a soil high in clay content, and several sands. Samples of these materials were collected from various field locations in Florida, air-dried, passed through a 2mm Sieve, and then stored for future use. Samples of Lakeland fine sand and Fellowship sandy clay were collected from-near Gainesville, Everglades mucky peat was collected near Zellwood, Wabasso and Oldsmar fine sands were collected near Ft. Pierce, and Pomello and Blanton fine sands were collected from Orange County. These soils are typical of large land areas in the state.. In addition to soil materials, glass beadswithdiameter range from 105.-210 11M (Cataphote.Corporation, Jackson, Mississippi) were used. Glass beads were selected as a "non8

PAGE 13

reactive soil" because of the low adsorption properties of the beads for many organic materials. Movement of an herbicide through glass beads provides a valuable indication of the influence of the mean pore size upon hydrodynamic dispersion under conditions of minimal herbicide sorption to the pore walls. Each porous media was hand-packed into cylindrical columns before each miscible displacement of herbicide. Miscible Displacement Technique for Studying Herbicide Movement in Soil Columns Miscible displacement in porous media refers to the displacement of one fluid by a second fluid when the two fluids are mutually soluble (Collins, 1961, When the fluids used are aqueous solutions where the displacing solution contains solute ions or molecules of interest, may speak of miscible displacement of solutes. During displacement, the moving front undergoes spreading as a result of mixing of the two solutions by molecular diffusion and convection (mass flow). Gradients in solute concentration in the vicinity of the advancing front or within individual pores drive the diffusion, and convection occurs as a consequence of the distribution of microscopic flow velocities within the porous medium. If the solute or aqueous solution interacts with the porous medium, miscible displacement can be complicated by the processes of ionic exchange, anion repul Sion, adsorption, biological degradati.on, etc. Since the total surface area exp6sed and the retention time for a solute in pDrous medium increase with increasing length of the porous medium, the amount of mixing which occurs during displacement of a solution through a column of porous medium will be greater for longer columns. Miscible displacement experiments in soil are generally accomplished by applying a displacing solution of a chemical to one end of a soil column. Flow velocities and water contents through the soil are controlled. Small aliquots of liquid effluent from the column are collected and analyzed to determine the concentration of the displacing solute. The solute concentration in the effluent plotted versus the volume of effluent collected give plots that have been termed "breakthrough curves." Solute concentration is normally expressed as a dimensionless ratio, CICo of the solute concentration in the effluent, and the initial solute concentra tion in the displacing solution. A dimensionless ratio, VIVo, referred to as "pore volume" is sometimes used to express the accumulative volume of effluent. Pore volume is obtained by dividing the effluent volume by the volume of water in the soil column under the conditions of the experiment. A laboratory technique developed by Nielsen and Biggar (1961, 1962,. 1963) gives a convenient procedure for study of 9

PAGE 14

miscible displacement of water and herbicide solutions applied to soil columns. Movement of different chemical solutes may be studied under controlled environmental conditions, such as flow velocity or water content, with this procedure. In this method, a solution containing a chemical of interest is applied to one end of a soil column. Effluent from the column is collected in small volume increments and each aliquot is then analyzed to determine the concentration of solute. If the displacing solution is applied as a pulse or volume "slug" the curve will increase from C/Co;"O to a maximum or peak and then decrease until C/Co=O is again reached. In a discussion of theoretical considerations for miscible displacement in porous media, Nielsen and Biggar (1962) discussed the types of breakthrough curves predicted from existing theories of dispersion. For the extremely ideal situation where no mixing due to diffusion or convection occurs, the tracer front proceeds by piston flow. For piston flow the breakthrough curve is a vertical line passing through the point at one pore volume of effluent. This type of transport process would be expected for a single capillary tube of constant radius, and cannot be expected to occur in soils. But the model gives insight into the occurrence of mixing in soil by establishment of the vertical line at one pore volume as a point of reference for comparison with more realistic types of breakthrough curves. If mixing results entirely from convection (mass flow) in a nonreactive soil having rather large microscopic flow velocities narrow in their distribution, a skewed sigmoid breakthrough curve will pass through C/C o=0.5 at one pore volume. Coupling of molecular diffusion with the convection during miscible displacement results in translation of the curve to the left at the smaller flow velocities as well as a slight clockwise rotation. Another type of breakthrough curve occurs for miscible displacement in nonreactive porous media having an extremely wide range in velocity distribution. Assuming dispersion results from both diffusion and convection, this type of curve is concave to the right. For each type of breakthrough curve in nonreactive media, the area enclosed beneath the curve and to the left of a vertical line passing through one pore volume is equivalent to the area enclosed above the curve but to the right of that line. Two types of curves are given for cases where the solute and the soil interact. If the reaction is a chemical process, a precipitation or an exchange, the sigmoid curve is displaced to the right with only a small portion of the curve extending to the left of one pore volume. Exclusion of the solute by a solute-solid interaction or a velocity distribution with velocities near zero results in a sigmoid breakthrough curve with a large displacement to the left of one pore A useful quantity referred to as "holdback" has been 10

PAGE 15

defined asthe area beneath the portion of the breakthrough curve displaced to the left of the vertical line for piston flow which occurs at one pore volume. For soils, holdback is a measure of the volume of original soil solution not displaced but remaining within the sample. Values for holdback vary from zero for piston flow to values less than 100% for other cases. Nielsen and Biggar (1961) have shown that the magnitude of holdback for water-unsaturated soils greatly exceeds that for saturated soils. A slow approach to a maximum CICo=l also occurs and the skewness of breakthrough curves increase upon desaturation. They attributed part of this to the extremely wide range in microscopic pore velocity and a changing cross-sectional area between displacing and displaced soluti.ons for miscible displacement through unsaturated soil. The influence of molecular diffusion upon dispersion is greater for displacement through unsaturated soil than saturated soil. RadioassayProcedure for Measurement of HerbicideConcentrati.onin Soil Effluent Chemical assays for 2,4-D and paraquat are frequentl.y time-consuming and require elaborate equipment. Radioassay rather than chemical assay was used in this study for measurement .of herbicide concentrations. That method offers advantages of being simple, being rapid, providing precise measure .of very small quantities of herbicide, and requiring access only to liquid scintillation counting equipment. One ml aliquots of aqueous effluent from soil columns were added to counting vials containing 15 ml of Brayls scintillation solution (Bray, 1960). Both plastic and glass vials were used. The vials were placed in a Packard Tri Carb Liquid Scintillation Counter to determine the radioactivity. Count rates (cpm) for soil effluent were corrected for background and were divided by the corrected count rate for the initial influent herbicide solution. This relative concentration, CICo was then plotted versus accumulative volume of soil effluent to provide breakthrough curves for each herbicide. INVESTIGATIONS Theoreti-cal Analy'sisofthe MoV'ementof SoTut'e's' in Soils: An Appli'cation of Transfer Functions Tntroduct'ion Most recent efforts to provide an analytic description of solute movement with water media have been directed at solving the hydrodynamic dispersion equati.on as a boundary 11

PAGE 16

value problem, with the boundary values determining the particular solution. An alternative approach is the method of Laplace transfer function theory. This alternative approach has been used with considerable success in other fields which deal with diffusion-like problems. Basically, this theory treats the problem of miscible displacement statistically, and properties of the Laplace transform provide an analytic form to the solute content of effluent from a soil column. This treatment should provide additional understanding of those processes which influence solute transport through porous media. Simple analytic examples illustrate the diverse utility of the theory when applied to miscible displacement. In contrast to the boundary value solution, the transfer function theory can be applied to cases of complex influent functions with considerable ease. When applied to experimental data, the transfer function theory gives results identical with the boundary value solution; however, the transfer function method requires considerably less effort, illustrating the power of the method. In this paper, the soil is treated as a continuous medium in order to ascribe to it the average macroscopic properties of the soil system. The statistical nature of the macroscopic viewpoint obscures the actions or mechanics of individual pores on the macroscopic scale. Probability Density The statistical approach to the analysis of the flow of solutes through porous media as performed by Day (1956), Scheidegger (1957 and 1964) and et (1968), provides the probability density function for steady one-dimensional flow. This approach assumes that Darcy's Law holds for liquid flow and that it follows from the central limit theorem, which says that irregardless of the probability distribution of each step in the process, after a sufficiently large number of steps, the probability distribution tends to be Gaussian. This leads to the probability density distribution of a solute 11 p ar t ic 1 e" to be where $(x,t) = (4nDt)-1/2exp[_(x-vt)2/4DtJ$(x,t) = probability density distribution as a function of x and t. x = spatial variable with its zero at the entrance to the column and proceeding in the direction of flow. 12 1.

PAGE 17

t = temporal variable with its zero at the time that the solute "pulse" enters the column at x = o. D = coefficient of hydrodynamic dispersion (not the coefficient of molecular diffusion). v = average liquid pore velocity or the rate of advance of the center of the "pulse". It convenient to transform equation 1 to a set of coordinates traveling with the "pulse" center. The distribution function is changed only to the extent that in the new system, the observer travels with the peak of the "pulse." Using the center of mass coordinate, z = x-vt, equation 1 transforms to the form 2. A quantity of immediate interest is the concentration C(z,t) of solute in the soil solution, expressed as g per cm3 Since is the fraction of the total solute mass per unit pore volume (total volume of solution present in soil column) of the soil solution C(z,t) In = m = total mass of solute. L = length of soil column. A = cross-sectional area of soil column. e = volumetric water content of soil. 3. The concentration of a volume "slug" of solution applied to the soil surface at time t = 0 is simply In C(O,O) = LAe 4. Therefore the probability density is just the relative concentration of the solute or the concentration ratio c(z,t) = C(z,t)/C(O,O). The concentration ratio is therefore a normal Gaussian 5. with zero mean ex = vt) and variance 2Dt. 13

PAGE 18

Impulse Response and Transfer Functions It is often useful when dealing with the transmission of aqueous solutions through media, to mentally visualize a flow system which in effect neglects the microscopic aspects of the medium and regards only macroscopic quantities such as distance (in the one dimensional case), average liquid flow velocity, concentration of solute, and some characteristic quantity or quality of the system as a whole. The Laplace transform, the impulse response function, and transfer function have been applied here to the problem of solute transport in porous media for the purpose of giving a workable macroscopic model. This method has been applied success-fully in electronic circuit design and analysis, and mechanics, and it is related to. the Fourier transform used in obtaining the point spread function and the modulation transfer function of optics, radar, sonar and other wave phenomena. The Laplace transform method is an extension of the Laplace transform equations used in solving boundary value problems of diffusion and heat flow. Consider the case of one-dimensional flow through a porous medium. Assume that no information is given about the microscopic nature of the porous medium, as this does not aid us in determining the gross behavior. Let the solute be applied in solution to a column of porous media with the liquid flow velocity maintained constant with time. The solute concentration of the influent will be taken as a function of time, 00 (t) and will be. referred to as the influent functi.on. Let 00 (t) be properly behaved such as to possess a Laplace transform which will be referred to as the influent amplitude where and its inverse is defined by co(t) = ho(s)e ds 1 /:a+ioo st 2'ITi a-ioo co(t) = influent function. hoCs) = influent amplitude. s = complex variable =cr+iw where cr and ware real variables. t = time variable. i = .1=1 6. 7 The constant a is chosen to be to the right of any singularity. of hoeS) and the integration path is sometimes known as a 14

PAGE 19

Bromwich contour. The response function of the system is the concentration of solute in the effluent and will be referred to. as the effluent function, 0 1 (t). It will also possess a Laplace transform which will be called the effluent amplitude and the inverse, The effluent amplitude, hI (s), is rel.ated to the influent amplitude, hoes), by the transfer function, R(s), i.e. 8. 9. 10. The transfer function has the property of a spectral filter inasmuch as it modifies the spectrum (Laplace transform) of the influent function to produce the spectrum of the effluent function. Once the transfer function is known for a specific flow system of a given porous medium, no other information is necessary in order to predict the effluent function for a known influent function. Note that so far these are generalized functions with the only restricting condition being that each possess a Laplace transform. The usefulness of the transfer function will be illustrated by showing how it can be found for select cases of influent functi.ons. Consider an initially solute-free medium into which is introduced a pulse of solute whose width may vary with time but is small in time. The magnitude of the pulse is subject to the condition 11. where 0T(t) is defined as O<,tT 0T(t) might represent a very short "slug" of solute whose concentration is very high. The magnitude of 0T(t) is inversely proportional to its duration whioh is small ln comparison to some quantity which characterizes duration of the efflu-15

PAGE 20

entfunction (breakthrough curve). Thus the solute transport may be described by and h (s) =l(l_e-ST ) sT 14. For a pulse of this nature, T is very small so that when the right hand side of equation 14 is expanded in a series and the limit taken as T approaches zero we have Then the transfer function becomes R(s) = lim h1(S) T +0 15. 16. or is equal to the effluent amplitude when co(t) is a very short pulse. This method provides an excellent procedure for determining the transfer function for a specific flow system as will be. seen later. Under these specific conditi.ons the effluent function is simply the inverse Laplace transformati.on of the transfer function, thus where L -1 [Jis the inverse Laplacetransformati.on and L[] is the Laplace transform. In general, the effluent function will be If for example Q (t) ={O o 1 t < 0 t > 0 18. 19. then from tables of Laplace transforms or by applying equation 6 directly we obtain 1 h (s) = -o s Rence the effluent function becomes Cl (t) = L -]. [R(s)l] = (t g(t}dt s Jo 20 21. where get) is the inverse Laplace transform of R(s) and is the impulse response function. If co(t) is a long pulse, 16

PAGE 21

that is, long in duration compared with the duration of get), then and by equation 14 Thus the effluent function is O<.t
PAGE 22

between the influent function, influent amplitude, effluent function, effJ.uent amplitude, impulse response function and transfer function are presented in the schematic diagram in Figure 1. Horizontal arrows indicate Laplace transforms and are reversible when the inverse Laplace transform is taken. Vertical arrows jrepresent the convolution integral on the left (not generally reversible) and multiplication by the transfer function (reversible) is given on the right. Thus there are two pathways from a known influent function to a desired effluent function when an impulse response function is known. One way involves direct use of the convolution and the second way involves a Laplace transformation, a multiplication and an inverse Laplace transformation. Applications to Miscible Displacement A practical application of the transfer function theory to solute transport in soils is miscible displacement of herbicides and fertilizer nutrients in soils. Potentially, this field promises to be one of its most important uses. Consider the effluent function, ci(t), given by equation 5 when z = L-vt where L is the column ength. This is the effluent function c1 (z ,t) for a case when the influentfunction is an infinite.simal thickness pulse, so that equation l2applies. C'1 (z ,t) is the impulse response function of a system whose length is Land the system characteristic is the coefficient of hydrodynamic dispersion, D. Thus 29. and the Laplace transform of g(z,t) is the transfer function H(s) =H(z,s) = 30. In practice it will sometimes be easier to obtain the effluent functions by the convolution integral 31. The integral cannot be evaluated until the form of Co(t) is known due to the convolution. However, for a specified co(t), c1(z,t) can be evaluated by analytical or numerical methods.; If co(t) is a step function as described by equation19, c1(z,t) is given by c (zt) = l' ,TID [ 4Dt J '(L-vt) 2D. errJ:t-vt] l?4Dt J 32. Ifco(t) takes the form of equation 22, then c1{z,t) takes 18

PAGE 23

Inverse Laplace. Transform +------------------------Laplace Transform Inverse Laplace Transform *.g(t} Laplace Transform B(s) x Inverse Laplace Transform +-----.--------------------. Laplace Transform Figure 1. RelatiDnships Between do(t), hoes), CI(t), hl(S), get), and Bes) The symbol stands for convolution. 19 A\ I -B(e)

PAGE 24

the form of equation 25. and using g (z,.t) from ab.ove we. get C1(z,t) 1 t ) 1/2. ((L-vt)}CL-vt ) '(:L-vt ) ---exp-erfc ---to' 'lTD 4Dt 2D 14Dt 33. exp(_'.(L-V(t-,to))2) +L-V(t-to) ePfc 'lTD 7 4D(t-to ) 2D 14D(t-to ) J If 0o(t) is periodic and of the form of equation 22 with period, 2to' it can be written as where 00 co(t) = lito E (-l)kS(t) k;"Q kto s (t) = kto Okto k =0,1,2,3 ... In this case, it .is more convenient .to determine C l(Z,t) by. the transform method. By that method, and .. '2': h (z s) = exp 't IDS3 o The resulting effluent function is 34. 35. 36. 37. Note that equation 37 is equivalent to equation 33 when k = 0,1. Results and Discu'ss1.'on In order to have any practical value, soil transfer function theory must predict the concentration of solute in the effluent .from a soil column when characteristic parameters of the specific flow system aregiveri and the solute concentration in the liquid influeritis given as a function of time. 20

PAGE 25

Alternatively, if the impulse response is known for the flow system then the characteristi.c parameters-L, v, and Dcan be recovered by a Least Squares curve fitting routine. Of course, it is easier to work with a situation where the in fluent function is known and the parameters of L, v, and D have been measured. This case may be illustrated using the data from Davidsonetal. (1968) where fluometron pesticide and chloride were infiltrated through a 30cm long column of diameter glass beads. Figure 2 shows the calculated impulse response function using the parameters of the experimental flow system. This is the theoretical effluent function which would be obtained if the influent had been a very narrow solute "pulse" of unit area. This has been calculated from equation 29 and normalized to unity. Using the same parameters, the effluent function was calculated for a 200 ml "slug" from equati.on 33 and this is shown in Figure 3 along with experimental curves for fluometron and chloride. The curve which Davidson calculated by solving the diffusion equation is also given. Note that both curves fall to the right .of the data indicating perhaps an occluded vol ume, as Davidson has suggested. It might be noted that this work uses time, t, as the independent variable and that it was necessary to use the following relati.on in order to calculate. the abcissa, pore volume,Vp for Figure 3. = vt L 38. Also it might be noted that when the impulse response func ti.on is used, the maximum value of the effluent function occurs when L tmax = v and thus for a unit area pulse, the amplitude on the breakthrough curve is given by c (0 t ) = (47TDt )-1/2 1 'max max 39. of the peak 40. from which the value of the dispersion coefficient, D, can be recovered. Coricl1isTohs Given values of length, L, and disperSion coefficient, D, for the specific soil and solute under observation and the average flow velocity, v, of the fluid, transfer function theory can be used to calculate breakthrough curves (plots of relative solute concentration p6re of effluent) for elution of herbicides or ferti.lizer solutes from 21

PAGE 26

Pore Volume 0.5 1.0 1.5 ,-.,. +-' 'l;-1.0r A 7D= 0.123 e ml h L=30.0 em V= 4.81 em/h ::J N tJ... .6 N OJ til C o .4 0.. til (i) i .2[ I I I I \ 0.. .0 0 1 3 4 k 6 7 8 9 E Co .J Time (hrs) Figure 2. Impulse Response Function

PAGE 27

N \.JJ Pore Volume 0.5 10r 250,u Glass Beads Fluometuron 8r Chloride 6 "'"' +-' l]-4 L o o 2 u -u by DavIdson and by this work 2 D = 0.123 em/h L =30.0 em v = 4.81 cmlh 1.0 1.5 01 I I I I I' I I I I ...... 0 1 o 1 2 3 4 5 6 7 8 9 10 Time (hrs) Figure 3. Effluent Function

PAGE 28

columns of soil. The function describing the solute con centration in the eff'luentcan be. calculated with excellent accuracy if the mathematical form of the solute concentration of the influent is known. This method provides a means to predict the effluent without actually performing the experiment in the field, when such experiments are not practical. It is based solely on statistical hydrodynamics and transfer function theory and is not directly subject to the boundary and initial conditions as are solutions to the differential equations. Miscible Displacementof2 ,4-D Herbicide 'Through 'Water Repellant Soils Many agricultural sands of Florida exhibit resistance to wetting with liquid water. SoilS with this property are to as being water-repellant. With time water will move through most .water-repellantsands, although wetting often proceeds along channels leaving pockets 6f dry soil in between the charinels. Since water is the principal carrier for the transport of non-volatile herbicides through soils,water-repellancy should alter the movement and consequent adsorption of a soil-applied herbicide. Aqueous solutions of dilute 2,4-D herbicide plus tritiated water were applied as influent slugs to water-saturated columns of water-wettable, water-repellant, ignited (water-repellant sand was ignited at6bOoC for 12 hours), and ignited plus silicone Silicone, Union Carbide Corporation, 270 Park Avenue, New York 17, New York) coated Blanton fine sand. Movement and adsorption were by means of breakthrough curves of 2,4-D and tritiated water in the column effluents. The dry soil materials were hand-packed into a glass column of 2.54 cm inside diameter and 25. cm length. Average bulk densities of 1.47, 1.52, 1.57, and 1.56 g/cm3 respec tively, were obtained for columns of the wettable waterrepellant, ignited and silicone coated materials. The columns were then wet with disti.lled water. Long time periods were required to wet the water-repellantand silicone coated porous materials. A variable-flow pump was used to establish constant liquid flow velocities of 2.04, and 4 '. 08. cm/h through the columns. The column influent was initially water, but at a time we shall call zero a 10 ppm solution of labeled 2,4-D was introduced to the column. After a 25 ml Hslug" of the 2,4-D solution had entered the column, the influent was changed to water. Breakthrough curves of 2,4-D are presented only for wettable and water-repellant .Pomello fine sands (Figures 4 and 5) for liquid flow velocities of 2.04, and 4.08, cm/h. Recoveries of the applied 2,4-D in theefflueritsfor columns of all four porous media are given in Table 1. Increasing the flow velocities from 2.04 to 4.08 cm/h resulted in iricreased 24

PAGE 29

I\) Vl WETTABLE POMEllO N I .A--A ,. A---6 \ \ \ \ V=4 CM/H V-2 CM/H .; '::. >, O ('J r< o 20 40 60 80 ML 100 120 140 Figure 4. Breakthrough Curves of 2,4-D Corresponding 160 to 2 Liquid Flow Velocities in Water-Wettable Pamella Fine Sand

PAGE 30

WATER REPEllANT POMEllO Figure 5. Breakthrough Curves of 2,4-D Corresponding to 2 Liquid Flow Velocities in Water-Repellant Pamella Fine Sand

PAGE 31

Table 1. Tabulation of Recoveriea of Applied 2,4-D in the Effluent From Columns of Water-Repellant, Water-Wettable, Ignited, and Ignited Plus Silicone Coated Samples of Pamella Fine Sand Recovery of 2,4-D % 2 4-D ,. Reed. Soil 2.04cm/h. 4.08. cm/h Water Repel.lant 82.0 85.5 Wett.able 76.0 83.2 Ignited 86.3 95.0 Silicone Coated 98.5 99.8 27

PAGE 32

2,4-D recoveries for all columns. At the slower velocity, recoveries of 2,4-D were 76.0,. and 98.5%, respectively, for wettable, water-repellant, ignited, and silicone-coated Blanton fine sand. Thus the recovery of 2,4-D was almost complete for the silicone-cOated sand regardless of the liquid flow velocity. The 2,4-D recovery from displacement through naturally water-repellant sand was less than that for the ignited sand but greater than for the water repellantsand. Ignition of the sand removed the organic matter, and thus soil adsorption of 2,4-D was therefore decreased during movement. The lower 2,4-D recovery in the effluent from the wettable soil as compared to that for the water-repellant soil was attributed to incomplete and nonuniform water-saturation. After termination of the experiments, small zones of dry soil were observed within the waterrepellant column. The incomplete should give a smaller volume of soil pores participating in transport of water and 2,4-D. The quantity of 2,4-D adsorbed during water flow through the wettable soil was therefore greater than for the water-repellant soil. Solutions of 10 ppm 2,4-D and water were also displaced through columns of wettable, water-repellant, and ignited plus silicone-coated Pomello fine sand. Breakthrough curves of 2,4-D and tritiated water for the water-repellant and silicone-coated columns are presented in Figures 6 and 7. Elution of the 2 chemicals were similar for the siliconecoated sand, but elution of water and 2,4-D from the water-repellant sand column gave separate breakthrough curves. Tritiated water appeared in the effluent before the 2,4-D and the greater spread for the 2,4-D curve indicates adsorption by the soil. Behaviour of 2,4-D and water curves for wettable sand (not shown) were similar to the curves in Figure 3. Work performed on the study indicates that water-repellancy in water-saturated sands influences the movement of 2,4-D. Further studies are needed to evaluate movement of herbicide solutions into initially dry water-repellant soils. Miscible Displacement of 2,4-D Herbicide During Constant Liquid Flow Velocity Through InitiaTly Dry Soils A theoretical model was developed to the movement of organic chemicals through porous packed in columns of finite length. Movement due to diffusion was assumed to be negligible. The model assumes that .transportof the chemical is mainly due to the liquid flow velocity and that at adsorption is linear. The model also considers the rate at which equilibrium between the dissolved and the adsorbed phases progresses with time. The theory was tested by studying the movement of C14 labeled 2,4-D herbicide through glass beads (105-210 }.lm), 28

PAGE 33

f\.) \D 1 0 C/C. WATER REPEllANT POMEllO ......................... / I I ,. ,. .,../ 3 H ____ 140 2,4-D v = 2 CM/HR \ 0.0' ..... < )P ;;e o 20 40 60 80 Ml 100 120 140 160 Figure 6. Breakthrough Curves of 2,4-D and Tritiated Water in Water-Repellant Pomella Fine Sand

PAGE 34

w o SI LlCONE COATED POM ELLO 1 3 H i4C 24-D I} C/C. V &It 2 CM/H o O.OLa hi 'I: '" I o 20 40 60 80 Ml 100 U!O 140 Figure 7. Breakthrough curves of 2,4-D and Tritiated Water in Ignited and Silicone Coated Pomello Fine Sand 160

PAGE 35

Lakeland fine sand, Fellowship sandy clay and Everglades mucky peat, under constant flow velocities. Air dry soils were packed uniformly in a column of 7.5 cm inside diameter and 30cm length which was held vertical during the studies. A photograph of one of the soil columns is shown in Figure 8. A dilute solution (10 mg/l) of 2,4-D was introduced at the bottom of the column with constant liquid flux maintained at 50 ml/h with a positive displacement pump. The same flux was continued until 200 ml of solution had entered the soil, at which time the solution was immediately replaced by pure water containing 0.5% phenol. Thus, a "volume slug" of herbicide solution was displaced upward in the column by water at the same flow rate. Studies were also made at a liquid flux of 100 ml/h. Effluent was collected in equal volume aliquots for analyses. Soil solutinn was extracted at the midway point (15 cm elevation) with a specially constructed sampling device using a porous plate and partial vacuum. The extract was collected periodically in amounts of 25 microliters. Effluent samples and the extracts were analyzed for 2,4-D content with a liquid scintillation counting system. Random effluent samples were subjected to additional analyses for 2,4-D to determine if the herbicide was degraded in the soil columns. The analyses were done by layer chromatography. The thin layer chromatographic studies revealed that the herbicide did not disintegrate during transport through the soil. Experimental breakthrough curves of 2,4-D applied as a 500 ml ttslug" of aqueous solution to columns of glass beads and Lakeland fine sand are shown in Figures 9 and 10. The almost rectangular shape of the curve for elution from glass beads indicates that the herbicide underwent only limited retention and the herbicide movement closely paralleled the water movement. The breakthrough curve for displacement through Lakeland fine sand is very wide which indicates in teractinn of with the porous medium. The low value of bk 0 (approximately 0.2) for the first aliquot of column effluent indicates that the movement of 2,4-D lagged behind that for the water. The herbicide also lingered in the soil effluent for longer periods than for the glass beads. These two curves were included in this report as being representative of data reported in the Ph.D. dissertation of V. Jyothi (1971). Theoretical and experimental breakthrough curves of herbicide concentration were expressed as functions of accumulative volume of solution introduced at the bottom of the columns. These elution curves were plotted for effluent and soil extracts at both liquid flow velocities. It was observed that the theoretical curves compared fairly well with experimental results. However, some discrepancy was observed between calculated and experimental curves during the initial time periods. Calculated concentrations were higher than the observed concentrationsin the effluent. This indicated that there was 31

PAGE 36

Figure 8. Photograph of a soil column in a constant temperature cham'Jer. A linear fracti.oD collector is also shown. 32

PAGE 37

VJ VJ GLASS BEADS I05-200.ll 1.0,. .. .... .. .... o-a C/C.' 0 V. 0.22 CM/H 200 400 500 600 700 800 ML. Figure 9. Experimentally Breakthrough Curve for a 500 ml "Slug" of 2,4-D Applied as Influent to a Column of Initially Dry Glass Beads at a Liquid Flow Velocity of 0.22 cm/h

PAGE 38

w .j::-1-0 o-a 0-6 efC. LAKELAND SAND V.-0 CM/H 0' -, o 100 200 300 400 500 600 700 800 Figure 10. Experimentally Determined Breakthrough Curve for a 500 ml "Slug" of 2,4-D Applied as Influent to a Column of Initially Dry Lakeland Fine Sand at a Liquid Flow Velocity of 0.22 cm/n

PAGE 39

PAGE 40

Experimental Methods and Procedure Air-dry samples of Oldsmar fine sand (takeri from 0.10 cm depth in the field profile) were hand packed into glass columns of 2.54 cm internal diameter and 30 length. tions of silicon rubber tubing were attached to each end of the columns. The columns were hermetically sealed by attaching a rubber septum to the end of each section of tubing. Two of the air-tight columns were placed in the sample container of the Research Food Irradiator, Food Science Department. The radiation source in the facility. was 105 Oi of cesium-137 which gives a dosage rate. of 2 x 105 rads per hour in the center of the stainless ste.el sample container (15 cm x 30 x 45 cm). A diagram of the irradiator is presented in Figure 11. The soil columns were exposed to the radiation field for a period of 20 hours to give a calculated total dose of 4 x 106 rads. After sterilization these columns were stored for later incubation analysis. Two other soil columns were fumigated with methyl bromide gas in a laboratory hood. A pressurized can of the gas was purchased from a garden store for this purpose. Thick-walled tygon tubing was used to connect a valve which was connected to the outflow from the methyl bromide container to. a 40 .li ter glass carboy and the carboy was connected by tygon tubing to a glassy-tube. Both forks of the tube were connected to large syringe needles which into the rubber septums at the ends of the two columns. Needles attached to. separate sections of tygon tubing were also inserted into the septums at the gas outf.low end of the columns The outflow ends of each piece of tubing were placed at about 1 cm depth in a beaker of water. Visually monitoring the relative flow rate of gas bubbles in the water permitte.d a rapid determination of the approximate gas flow rate. in the columns .. The flow system was first tested with air flow to test leaks; however, the test revealed the system functioned properly. The valve was then opened on the methyl bromide container and allowed to move into the soil columns, initi.ally flushing air from the pores. Rubber gloves., a long-sleeved lab coat and protective eye goggles were worn during the fumigation as safety precautions. Also the glass door of the hood separated the experimenter from the experiment. The columns were left exposed to. methyl bromide for a 24-hour period, after which the needles were removed from the septums and columns were .stored for later incubation analyses. Samples of radiated, fumigated and control soils were incubated in several growth media to. determine the growth of microorganisms. Severi. media were specific for the growth of aerobic organisms and one medium wasspeclfic for the growth of anaerobic organisms. Trypic soy agar, Bacillus medium, sporulation agar, mycological agar, peptone trypticose agar, mannitol agar, and gelatin agar were the cu.lture media 36

PAGE 41

Figure 11. ......... -f!! I :e e: :l: Ii II 'I I .... I I I : : Ii It! I I I t: I:' i i t I" ., .. '0 '1,1, ij i i 1', It , I , I :1 , !X I 'i Ii 1'111 I i i RESEARCH FOOD IRRADIATOR \IPE TRENCH $.$. TANK 6FT pIA.)( II FT PEEP .Scheniati.c Diagram of Univer.sity. of Florida Research Food Irradiator 37

PAGE 42

for aerobes. The culture media and procedure for anaerobes was that described by Smith (1964). ApproximateTy one gram of radiated, fumigated or non-treated soil was placed upon the growth media in each petri dish. Incubation was allowed to. proceed for seven days at 28c. For a control, 1 ml of steam sterilized distilled water was placed on the growth media in petri dishes. The steam sterilized media were initially poured into steam sterilized petri dishes in a trans fer chamber equipped to provide a positive outflow of sterilized air. At the end of the incubation period the presence or absence of growth was recorded for each petri dish. Ob servations were visually determined under a microscope at 2X magnification. The binomial data was reported as ratios, R, of the number of petri dishes with growth to the total number of dishes. The ratios were transformed by the relation X = arc sin/1'f before statistical analysis of the data. Results and Discussion Data for the study is shown in Table 2 and statistical analysis is presented in Tables 3 and 4. Both the methyl bromide fumigation and the gamma irradiation treatments of th.esoil columns were shown to be highly effective means for sterilization. The two treatments were observed to be equally effective methods of sterilization for both aerobes and anaerobes. Miscible Displacement of Paraquat Herbicide, Clls, and Tritiated Water Through Sterile and Nonsterile Soil Columns Steam sterilized aqueous solutions of 10 ppm C14labeled paraquat herbicide and tritiated water were displaced through 3 columns each of methyl bromide fumigated, gamma irradiated and unsterilized Oldsmar fine sand .cm depth). Air dry soil was packed in glass columns of 2.54 cm inside diameter and 30 cm length to give a soil bulk density of 1.00 g per cm3 Each column was water-saturated one day before displacement was to begin and water was then pumped through the column at a velocity of 5.2 cm per h for 24 hours. The purpose of the study was to determine if sterilization influenced the retention of the paraquat or tritiated water during movement through the soil. Soil columns were sterilized by placing each intact column of air-dry soil within a high intensity field of gamma radiation (CS137) or by flushing gaseous methyl bromide through the intact columns. Sterilization procedures are described previously in this report. Three successive applications of paraquat and tritiated water solution were applied to water-saturated soil columns. As expected for this soil, which had 8% organic matter, no paraquat was recovered in the liquid effluent from radiation sterilized, methyl bromide sterilized, or unsterilized columns due to strong adsorption by the soil particles. However, tritiated 38

PAGE 43

Table 2. Growth of Microorganisms in Samples of Fumigated, Irradiated, and Non-treated Oldsmar Fine Sand Incubated with Seven Culture Media for Aerobes and Anaerobes The data are reported as values of R, ratio of number of aamples with growth observed to the total number of samples. Replicate. No. 1 Replicate. No. 2 Culture Media Fumi-Irradi-Non-Fumi-Irradi-Non-gated ated treated gated ated treated specific for aerobes: Soil SoiT Soil Soil Soil Soil 1.. trypic soy agar 12.92 12.92 69.30 12.92 12 .. 92 69.30 2. Bacillus medium 12.92 12.92 69.30 12.92 26.56 69.30 3. sporulati.on agar 12 .. 92 12.92 69.30 12 .. 92 12.92 69.30 4. mycological agar 26.56 12.92 .30 12.92 12 .. 92 69.30 5. peptone yeast -trypticose agar 12.92 12.92 69.30 26.56 12.92 69.30 6. mannitol agar 12.92 12.92 69.30 12 .. 92 12 .. 92 69.30 7 gelatin agar 26.56 12.92 69.30 26.56 12.92 69.30 specific for anaerobes: 12.50 12.92 69.30 12.50 12.50 91.7 39

PAGE 44

Table 3. Analysis of Variance for Growth of Aerobic Microorganisms in Sterilized and Non,...treated Oldsmar Fine Sand During Incubation with Seven Growth Media. The data were transformed with the relation X = arc sin 1If. Degrees of .Mean F-value .. Freedom Squares Replicate.s 1 4.45 0.23 Media 6 16.21 0.85 Treatments 2 13,608.00 721.5** Error 32 18.86 Total 41 **Indicates significance at 1% level. of probability 40

PAGE 45

Table 4. Analysis of Variance for Growth of Anaerobic Microorganisms in Sterilized and NOn-treated Oldsmar Fine Sand During Incubation with a Single Growth Media The data were transformed with the relation X = arc sin IR. Degree.s of Mean F-Value Freedom Bqua:re s Replicate.s 1 0 Treatment 2 3683.40 00 Error 2 Total 5 41

PAGE 46

water recovery was pronounced as three successive breakthrough curves (one for each application) in theef'fluent f'rom all soil columns. Percentage recovery of' tritiated water was very nearly the samef'or each of' the three successive curves (Figures 12 and 13) f'or the sterilized columns, but recovery progressively decreased with the 3 successive curves (Figure 14) from the unsterilized soil column. Recovery of tritiated water in the eff'luent f'rom the unsterilized column was 92.5, 86.1, and 81.2%, respectively. Although data reported in Figures 1, 2 and 3 represent average values from 3 replicate soil columns, a f'ourth column of unsterilized soil was prepared as a check. This column was water-saturated and three successive 5 ml "slugs" of an aqueous solution of' NaC136 and tritiated water were applied as influent to the column. A constant liquid flow velocity was maintained through the soil, and water was applied as influent prior to and f'ollowing each slug. Breakthrough curves of C13G and water in the column effluent are given in Figure 15. Note that the C13G recovery was 99.2, 98.1, and 98.0%, respectively, for the successive "slugs;" whereas the tritiated water recovery was 95.2, 92.2, and 86.1%, respectively. The chloride anion recovery was constant and almost .complete, but the recovery of' tritiated water decreased for the successive "slugs." For a given "slug" the breakthrough curves indicate that the chloride appeared-in the effluent prior to the tritiated water. The tritiated water also lingered in the eff'luent after elution of' chloride has ceased. Coreyetal. (1963) also observed separation of chloride and tritiated water breakthrough curves for single displacements through water-saturated columns of sandstone. They observed that separation was greater for lower than for higher liquid f'low velocities due to molecular dif'fusion effects. The decrease in recovery of tritiated water with successive "slug" applications as influent to columns of' unsterilized Oldsmar fine sand was def'initely related in some way to microbiological activity. Since microorganisms multiply at very rapid rates (population may double within one-half hour under correct conditions) the population of soil microbes probably increased considerably during the 2 to 3 days required to displace three successive "slugs!! through each unsterilized soil column. We therefore postulate that the observed decrease in recovery of tritiated water was due to increased microbial retention in the unsterilized soil with increasing time. Skewness of breakthrough curves for tritiated water in Figures 3 and 4 imply that retention by some mechanism did occur. Comparison of symmetry for these curves to those in Figures 1 and 2 show that minimal retention of tritiated water occurred in the fumigated or irradiated soil columns. Displacement of Paraquat and Diquat Herbicides by KCl Solution from Water-Saturated Columns of' Soil The toxicant portions of paraquat and diquat herbicides 42

PAGE 47

.j:::" w Q ..-CQ 0 C
PAGE 48

-1= -1= o 't"o J( .QO eli L 0>0 u c o UC\I o Oldsmar fine sand: A recovery o paraq uat: 0 0/0 D. H3: 97.4 0/0 D. D. 6. 6 6 6 6 6 6 D. 6 6 steri I ized with B recovery o 0/0 97.1 0/0 If D. 6. 6 6 6 D. D. 6 D. 6 6 methyl bromide C recovery o 0/0 96.70/0 D. D. D. D. D. D. D. 6 6 D. 6 D. 6 D. D. D. D. D. 6 D. D. D. D. D. D. D., D. a '),6 o l(;/ tNI I A-O r 1 2 3 4 5 I I I I I B -0 1 2 3 4 I I 2 '------------o 1 V C Pore Volume, /\to Figure 13. Breakthrough Curves for C1 Labeled Paraquat and Tritiated Water in the Effluent From a Column (Average of Data from Three Replicates) of Methyl Bromide Fumigated Oldsmar Fine Sand After Three Successive .. Influent Applications of "Slugs"

PAGE 49

.+::\,Jl Q ..-o ;'k CJ. 6 CJ. CJ. CJ. CJ. CJ. CJ. CJ. CJ. CJ. CJ. CJ. CJ. CJ. CJ. CJ. CJ. A' CJ. CJ. u CJ. CJ. CJ. CJ. ; CJ. CJ.; OV/o"cPCJ.LS ,r I I I A-a 1 2 3 4 5 I I I I I B 1 2 3 4 I I 1 2 a v/ c-Pore VolumeJ/v. Figure 14. Breakthrough Curves for Labeled Paraquat and Tritiated Water in the Effluent From a Column (Average Data from Three Replicates) of Non-Sterilized Oldsmar Fine Sand After Three Successive Influent Applications of "Slugs"

PAGE 50

.j:::" 0\ Oldsmar fine sand: unsteri I ized A B C C136 : recovery recover,)( recover,)( 99.2% 98.1% 98.0% 1.0 H3 : 95.2% 92.2% 86.1 % 0.8 36 {}; r {}; {}; CI -{}; 3 {}; H -{}; c; 0 6 {}; {}; f {}; {}; {}; {}; +' {}; en {}; {}; {}; L {}; 1. +' 0.4 {}; {}; {}; C c; {}; {}; {}; l!. Q) {}; {}; l!. U {}; {}; l!. c 80.2 I {}; 't. l.. Ll l!. 0., A--0 1 2 3 4 :> I I I I I B it 0 1 2 3 4 I I I I C 0 12 3 Pore Figure 15. Breakthrough Curves for C136 and Tritiated Water in the Effluent From a Column of Non-Sterilized Oldsmar Fine Sand After Three Successive Influent Applications of "Slugsll

PAGE 51

are divalent .organiccations which undergo rapid physical adsorption with soil components. CWeber:et' 'al., 1965; Weber and Weed, 1968; Tucker' :et'al., 1967; Co.atS:and Funderburk, 1966; and Knight and Tomlinson, 1961). A short.study. was performed to. determine if these chemicals could be desorbed from soil columns by aqueous solutions of KC1. Air-dry Wabasso fine sand taken from and 33-76 cm profile depths were packed into glass columns of 2.54 cm inside diameter and 30 cm length.' The organic matter contents of these two soil materials were determined to be 1.3 and 0.1%, respectively, both composed of gre.ater than 97% sand. The average bulk density. for the two columns was 1.32 g per cm3 The columns were saturated with distilled water and a pump was used to maintain a fluid flow velocity of 6 cm per hour. After one day of flow, the influent was changed from water to an aqueous solution of io ppm paraquat for one column and diquat for the other. Both chemicals were labeled with .. After one hundred ml of herbicide solution entered the columns, the influent was switched to water. Water was allowed to displace the herbicides into the column until 5 pore volumes of column effluent had been collected by an automati.c fraction collector. At that time the influent was switched to a 746 ppm aqueous solution of KC1. Breakthrough curves for paraquat and diquat in the effluents. of columns of 0-10 .and 33"':76 em soil materials are presented in Figures 16 and 17. Between 0 and 5 pore volumes neither of the chemicals appeared in the effluent for either soil material. The paraquat .and diquat were thus assumed to be complete.ly adsorbed in the soil columns. After 5 pore VOlumes, paraquat occurred in the effluent from the' O-.lO.cm soil but the recovery was very small. None of the diquat appeared in the effluent. For the 33"':76 cm soil both herbicides appeared in the effluent with a sharp breakthrough; however, the amount of paraquat exceeded that of Small quantities of the chemicals remained in the effluent even after 20 pore volumes of liquid flow. Paraquat .and diquat, to a lesser extent, were observed to be desorbed from Wabasso fine sand by KCl solution. Thus,. it would appear that applications of feJJtilizer to. this soil could influence the desorption and consequent movement of paraquat and diquatin the soil solution. Soil Adsorpti.on of Paraquat Herbi.cide Equilibrium adsorption isotherms of paraquat were performed for Everglades mucky peat, water.--repellantand ignited (600C for 12 hours) Blanton fine sand, .and Waba.sso fine sand. Five grams Coven dry basis) of soil were placed with 25 ml of c1 ,labeled aqueous solutions of known concentration of paraquat into centrifuge tubes. The tubes were placed on a mechani-47

PAGE 52

..j:::" co -4 1x10 Co -s 5x10 o o Wabasso fine sand:0-10cm depth herbicide paraquat d iquat recovery 0.00090/0 0.0 % 5 10 Pore Vol umesJ o 15 20 Figure 16. Displacement of Paraquat and Diquat Herbicides by KCl From a Column of Wabasso Fine Sand, 0-10 cm Profile Depth

PAGE 53

.j::" \0 -4 1x10 Yr Co -5 5x10 o o Wabasso fine sand:33-76cm depth herbicide o paraquat t:J. diquat t:J. 5 10 15 Pore Volumes, o 20 Figure 17. Displacement of Paraquat and Diquat Herbicides by KCl From A Column of Wabasso Fine Sand, 33-76 em Profile Depth .', i

PAGE 54

PAGE 55

\.Jl I--' 0 0 0 't"-X r---"1 :::t L....-...l o Everglades mucky peat 0 lfl 6 Blanton fine sand: unignited .6' C\J Blanton fine sand: ign ited "'0 (l.) ..0 L 0 i9 +-' ro ::J't"-A 0-ro L 50 100 150 200 250 300 350 400 450 Paraquat in Initial SolutionJ tL%, Figure 18. Adsorption Isotherms of Paraquat for Everglades Mucky Peat, Water-Repellant Blanton Fine Sand, and Ignited Blanton Fine Sand

PAGE 56

0 30 Wabasso sand 0 en ..... o 25 E C) ::J 20 Z 0 t-o.. D:: 15 0 IJl (/) t\) t-10 c:( ::;:) 0' c:( D:: 5 c:( 0.. 0 0 1 i 3 4 5 6 7 8 9 10 PARAQUAT CONCENTRATION (}J9 per rill) Figure 19. Adsorption Isotherm of Paraquat for Wabasso Fine Sand

PAGE 57

Ul W solutfon concentrations soils C\J 0<0--51 }1%1 paraquat o Everglades m peat o Pomello:unignited 0 Wabasso '<"---17 It 1/ : ign ited ::::l 0 lfl .0 "'0 Q)(Y) .0 L 0 (j) "'0 -----__ H ____ 0 -------------------r--------------+-----------... 01 1 o 1 2 3 4 5 6 7 8 9. Initially in Paraquat Solution, [mo/ml] Figure 20. Influence of KCl in Solution Upon Adsorption of Paraquat in Everglades Mucky Peat, Water-Repellantand Ignited Pomello Fine Sand, and Wabasso Fine Sand

PAGE 58

Vl ..j:::Blanton Sand: r--. unignited 0 0 If) ignited l:t. --0 Paraquat initially in solution: 010 51 ppm ::t. 17 ppm --"'0 0 Q) 0 .D (") L 0 If) "'0 0 <{ 0 (\j +-' CU :J ()8 cu __ L CU D---0.----_---0__ --------0 ----. O 0 2000 4000 6000 8000 KCI in Solution (mg!l) Figure 21. Influence of KCl in Solution Upon Adsorption of Paraquat in Water-Repellant and Ignited Blanton Fine Sand

PAGE 59

soils. For under conditions of intense fertilizer paraquat may moVe. further down the profile than otherwise expected. Continuous Measurement of Chloride in Effluent Flowing from a Soil Column Miscible displacement of chloride solutions through a soil column normally requires the use of a fraction collector to collect aliquots of effluent and some means for measurement.of theCl-concentration in each aliquot. Yoo and Kirkham (1971) described a liquid scintillation flow cell for contin uously recording C136 concentration in soil effluent. Their method has advantages of eliminating the need for fraotion collectors, provides instantaneous concentration measurements without need for sample storage and provides a continuous recording of concentration; however, the method has the inherent disadvantage of requiring that the chloride be radioactive. An inexpensive flow cell method is described in this study which does not require that the chloride be radioactive. Experimental Apparatus An Orion Model 96-17 Combination reference and chloride electr.ode provides the transducer for the method. The con centration range of the electrode is 1 to 5 x lO-sM chloride and the minimum sample size is ml. The outer sleeve of the electrode is constructed of unbreakable plastic which is resistant to most solvents and is resistant to mechanical shock or stress. The reference portion of the combinati.on electrode produces a stable, drift-free reference potential and low stirring and junction potentials. A flow cell was constructed from a cylindrical section of lucite plastic. A schematic of the flow cell and the com bination electrode is presented in Figure 22. A water-tight seal around the sleeve of the electrode was by the use of silicon rubber sealant rather than the rubber O-r:tng as shown in the schematic. The O-ring has the advantage of providing easy removal of the electrode from the flow cell for cleaning purposes. The electrical voltage from the combination electrode was measured by a pH meter and the .output from the meter was connected to a strip chart recorder. The transducer was calibrated by measuring the voltage for a range of chloride solutions of known concentrati.on. The resulting calibrati.on curve is shown in Figure 23. Calibrati.on was performed in the flow cell except that the soluti.ons were stationary during voltage measurements. Selected points from a strip chart recording of the transducer voltage can thus be converted to activities of chloride through the calibration curve. 55

PAGE 60

COtv1BINATION REFERENCE & CL ELECTRODE ---c.1:;:;:;:;:;:;:;:;:;:;:;:; RING LUCI TE FLOW CELL OUTPUT-+--.-5 1 INPUT FLOW CELL FOR CL MEASUREME NT Figure 22. Schematic Diagram Showing Combination Reference and Chloride Electrode in a Luc.ite Flow Cell 56

PAGE 61

Vl -..:J o >CO E + (!)"' 0 + o > (!)o u o L 00 Wi Calibration Curve of Chloride Electrode o __ CO -if 110 163 162 161 Figure 23. Concentration of Nael; moles per liter Calibration Curve for the Combination Reference and Chloride Electrode

PAGE 62

To te'St the transient response of the flow. cell and transducersY\$tem a concentrated aqueOus solution of NaCl was displaced through a column of glass beads. A glass chromatography column with 2.54 .cm internal diameter and 30cm internal length was packed withl05-210].lm glass beads. The bulk density of the packed column was 1.57 g/cm3 The column was saturated with 5 ppm NaCl solution, and a variable flow pump was used to maintain a flow velocity of 7.0 .cm/h of the dilute chloride solution through the porous media .. After one day the 5 ppm influent solution was replaced by a 5005 ppm solution. A 2 ml "slug" of the concentrated Na.Cl solution was allowed to enter the column before switching back to the original dilute chloride influent. During the miscible displacement of the concentrated slug of chloride through the column the electrode voltage was recorded on the chart recorder. The chart speed was maintained at 31.17 cm/h. Column effluent .flowing from the flow cell was collected in 2 ml aliquots in glass tubes on a fraction collector. Samples from the aliquots were later titrated with AgN03 to. obtain the concentration of chloride in solution. The titration measurement provided a means for checking the concentrations obtained by the electrode method Results and Discussion The breakthrough curve (relative concentration of chloride versus the number of"porevo.lumes of column effluent) for concentrated chloride.slug displaced through the column of glass beads is presented in Figure 24. Zero pore volume corresponds to the time at whichthe slug was first introduced as influent to the column. Circles on the graphc:orrespond to concentrations determined by titration of effluent aliquots, and the smooth curve was provided by connecting concentrations determined by the transducer. Very close agreement between the two methods is clearly shown. Although the breakthrough curve is skewed, the aliquot .concentrations also indicate a similar skewness. Integration of the curve by Simpson f s rule indic.ated 96.0% re. covery of the chloride slug. This large recovery percentage is only 4% less than the expected value of 100.%. and thus is an indication that the electrode method gave realistic concentrations of chloride. Conclu'sion An inexpensive. flow cell and combination chloride electrode system i.s described. for continuously recording the chloride concentration in effluent .flowihg from a column of soil. A breakthrough curve for chloride solution displaced through a column of glass beads was in close agreemeritwitha corresponding breakthrough curve obtained by ti.trati.on of aliquots. of the same effluent. 58

PAGE 63

\Jl \..0 Q Displacement of NoCI through a cotumn_ of glass beads(105-210j-lm) j(0C). .. 0 c 0 Co= 5000 ppm C I 1.57 Q/cm3 CU(() b6 c V=70 c% recovery of CI = 96.0% Q) u U O ill > 'C\J cuo Q) titration of aliquots -continuous measurement 0:: Pore vb,ume,y\;o 2 Figure 24. Breakthrough Curve of Chloride in Effluent from Column of Glass Beads

PAGE 64

ACKNOWLEDGMENTS The investigators were assisted by the following authors and co-authors: R. M. McCurdy, Graduate Student in Physics; V. Graduate Student Soil Science; P. G.Hunt, Research Associate in Soil Science; and A. Elzeftawy, Graduate Assistant in Soil Science. 60

PAGE 65

LITERATURE CITED 1. Akhavein, A. A. and D. L. Linscott. The dipyridylium herbicides, paraquat and diquat. Residue Reviews, Vol. 21:97-145. 2. Bear, J., D. Zaslavsky, and S. Irmay. 1968. Hydrodynamic dispersion. Physical Principles of Water Percolation and Seepage, Publication 29 of Arid Zone Research Series, UNESCO, Paris, pages 307-349. 3. Boon, W. R., 1965. Diquat and paraquat -new agricultural tools. Chemistry and Industry, May 8, 1965, pages 782-788. 4. Bray, G. A., 1960. A simple efficient liquid scintillator for counting aqueous solutions in a liquid scintillation counter. Anal. Biochem. 1:279-285. 5. Ca1derbank, A., 1968. The bipyridylium herbicides. Advances in Pest Control Reaearch, Vol. 8:127-215. 6. Coats, G. E., H. H. Funderburk, Jr., J. M. Lawrence, and D. E. Davis. 1966. Factors affecting persistence and inactivation of diquat .and paraquat. Weed Res. 6: 58-66. 7. Collins, R. E., 1961. Flow of Fluids Through Porous Materials. Reinhold Publishing Corporation, New York, N.Y. 8. Corey, J. C., D. R. Nielsen, and J. W. Biggar. 1963. Miscible displacement in saturated and unsaturated sandstone. Soil Sci. Soc., Amer. Proc. 27:258-262 .. 9. Corey, J. C., D. R. Nielsen, J. C. Picken, Jr., and Don Kirkham. 1967. Miscible displacement through gamma radiati.on sterilized soil columns.. Environmental Science and Technology, Vol. 1:144-147. 10. Crafts, A. S., 1957. The chemistry and mode of action of herbicides. Advances in Pest Contr.ol Research, Vol. 1:39-79. 11. Crafts., A. S., 1961. The chemistry and mode of action of herbi.cides. IntersciencePublishers, New York. 269 pages. 12. Davidson, J. 00., C. E. Rieck, and P. W. Santelman. 1968. Influence of water flux and porous material on the movementof selected herbicides. Soil. Sci. Soc., Amer. Proc. 61

PAGE 66

13. Davidsan, J. M., .and P. Wo. Santelman. 1968. Displacement af Fluometuran and Diuran thraughs.aturated glass beads and soil. Weed. Science 16: 544-548. 14. Day, Paul R., 1956. Dispersian of a maving salt-water baundary advancing thraugh saturated sand. Transactians, American Geaphyslcal Unlan, Val. 37:595:..601. 15. Ena, C. F., and Hugh Papenae. 1964. Gamma radiatian camparedwith steam and methyl bramide as a sail sterilizing agent. Sail Sci. Sac. Amer. Prac. 28:533-535. 16. Freed, V. H., 1966. Chemistry af herbicides. Pesticides and Their Effects an Sails and Water, American Saciety af Agranamy, Special Publicati.an Number 8, pages 25-43. 17. Funderburk, H. H., Jr. 1969. Diquat and paraquat. Degradati.an af Herbicides, Marcel Dekker, Inc., New Yark, N.Y., pages 283-298. 18. Hartley, G. S., 1964. Herbicide behaviaur in the sail. The Physialagy and Biachemistry of Herbicides. Academic Press, New York, 111-161. 19. Herbicide Handbaak af the Weed Saciety af America. 1967. W. F. Humphrey Press, Inc., Geneva, New. York. 20 Jyathi, V., 1971. Miscible displacemetit af 2,4-D herbicide during constant .liquid flow velocity into. initially dry Unpublished Ph.D. dissertatian, Sail Science Department, University af Flarida, Gainesville. 21.: Kirkham, Dan. 1964. Same physical pracesses causing mavementaf ions and other matter thraughsail. Overdruk Uit De Mededelingen Van De Landbauwhogeschaal En De Opzoekingstations Van De Staat .Te Gent. DEEL XXIX; paper presented at 15th Annual Phytapharmacy Sympasium, May 6, 1963, Ghent, Belgium. 22. Knight, B. A. G., and T. E. Tomlinsan. 1967. The interactian af paraquat (1:1"'-dimethy14:4"'-dipyridylium dichloride) with mineral sails. J. Sail Sci. 18:213-243. 23.. LeGrand, H. E., 1966. Moverrientaf pesti.cidesin the Sail. Pesticides and Their Effects on Sails and Water, American Saciety af Agranamy, Special Publicati.on Number 8, pages 71-77. 24... Nielsen, D. R., and J. W. Biggar. 1961. Miscible displacement .in sails. 1. Experimetital infarmatian. Sail Sci. Soc. Amer. Prac. 25: 1-5. 25.. Nielsen, D. R., and J. W. Biggar. 19 Miscible dis placernetit .in sails. II. Theareticalcansiderations. Sail Sci. Sac. Amer. Prac. 26:216-221. 62

PAGE 67

26. Nielsen, D. R., and J. W. Biggar. 1963 .. Miscible displacement in soils. III. Mixing in glass beads. Soil Sci. Soc. Amer. Proc. 27:10-13. 27. Scheidegger, A. E., 1957. Physics of Flow Through Porous Media. The MacMillan Co., New York, N.Y., pages 31-35 and 197-202 .. 28.. Scheidegger, A. E., 1964. Statistical hydrodynamics in porous media. Advances in Hydroscience, Vol. 1:161-181. Smith, P. H., 1964. Pure culture studies of methanogenic bacteria. Proceedings, 20th Industrial Waste Conference Purdue UniverSity, Lafayette, Indiana. 30. Tucker, B. V., D. E. Pack, and J. N. Ospenson. 1967. Adsorption of bipyridylium herbicides in soil. J. Agr. Food Chern. 15:1005-1008. 31. Weber, J. B. and S. B. Weed. 1968. Adsorption and desorption of diquat, paraquat, and prometone by montmorillonitic and kaolinitic clay minerals. Soil Sci. Soc. Amer. Proc. 32:485-487. 32. Weber, J. B., P. W. Perry, and R. P. Upchurch. 1965. The influence of temperature and time on the adsorption of paraquat, diquat, 2,4-D, and prometone by clays, charcoal, and anion exchange resin. Soil Sci. Soc. Amer. Proc. 29:678-688. 33. Yoo, Sun-Ho and Don Kirkham. 1971. Flow Cell System for Miscible Displacement Experiments .. Water Resources Research. 7:211-213.