United States
                         Environmental Protection
                         Agency
Office of
Research and
Development
Office of Solid Waste
and Emergency
Response
EPA/540/S-92/OOI
January  1992
 &EPA      Ground  Water   Issue
                         Chemical Enhancements to Pump-and-Treat Remediation

                         Carl D. Palmer* and William Fish*
Background

Conventional pump-and-treat technologies are among the
most widely used systems for the remediation of
contaminated ground water. Within recent years it has
become recognized that these systems can require
protracted periods of time to make significant reductions in
the quantity of contaminants associated with both the liquid
and solid phases which constitute the subsurface matrix.
Evaluating  the effectiveness of pump-and-treat remediations
at Superfund sites, as well as attempting to improve this
effectiveness, are issues identified by the Regional
Superfund  Ground Water Forum as a concern of Superfund
decision makers. The Forum is a group of ground-water
scientists and engineers, representing EPA's Regional
Superfund  Offices, organized to exchange the most recent
information related to ground-water remediation at Superfund
sites.

Recent research has led to a better understanding of the
processes involved in the transport and transformation of
contaminants in the subsurface. While some of these
processes are not readily amenable to enhanced removal  by
ground-water extraction, others suggest that there are
available techniques to  increase the efficiency of these types
of remediation systems. The intent of this document is to
explore the use  of chemical enhancement to improve ground-
water remediation efficiencies using pump-and-treat
technologies, and point out arenas of contamination where
such techniques are not practical.

For further  information contact John Matthews, Chief,
Applications and Assistance Branch, RSKERL, FTS 743-
2408, or Bert Bledsoe, Project Officer, RSKERL, FTS 743-
2324. Both  may be reached at 405-332-8800.

Summary and Conclusions

Recognition that conventional pump-and-treat remediation
often requires lengthy periods of time to achieve clean-up
objectives will encourage professionals involved in site
remediation to contemplate alternative methods of aquifer
     restoration. Some form of chemical enhancement for pump-
     and-treat will likely be an alternative considered for many
     waste-site cleanups. Although chemical enhancement of
     pump-and-treat may be a means of accelerating aquifer
     remediation, there are many aspects of chemical
     enhancement that need to be known before these techniques
     can be successfully implemented.

     Not all waste sites are amenable to chemical  enhancement
     methods. In particular, if tailing in the concentration-versus-
     time curves for the extraction wells is dominated by physical
     processes, then chemical enhancement methods will have no
     advantage over conventional pump-and-treat. Knowledge of
     the relative contributions of chemical and physical processes
     limiting pump-and-treat are needed during the early stages of
     site Remedial Investigations to ascertain the general
     usefulness of chemical enhancement.

     Even when it is known that physical processes contribute little
     to the tailing, specific knowledge is needed about the
     chemical processes that contribute to tailing at a particular
     site. Only then can potential chemical agents that are likely
     to influence these processes be identified.  The reactive
     agents may be chosen to compete with the contaminants for
     adsorption sites, complex the contaminant, change the redox
     state of the contaminant, change the solvation properties of
     the ground water, act as a surfactant, ionize the contaminant,
     or substitute for the contaminant in a precipitate. If the
     reactive agents are chosen on the basis of incorrectly-
     identified limiting processes, there is  a risk  that the reactive
     agents will provide no net benefit and may  even prolong
     remediation.

     Even when a reactive agent is found that specifically
     addresses the limiting chemical process, other considerations
     must be investigated to assure successful implementation.
     The key areas of concern in any chemical enhancement
     method are 1 ) the delivery of the reactive agent to those
     'Environmental Science and Engineering, Oregon Graduate
      Institute of Science and Technology, Beaverton, OR.
                         Superfund Technology Support Center for
                         Ground Water

                         Robert S. Kerr Environmental
                         Research  Laboratory
                         Ada, Oklahoma
                          Technology innovation Office
                          Office of $o8<| Waste and Emergency
                          Reeporwev us EPA, Washington, D,C.

                          Walter W. KovaUck, Jr., Ph.D.
                          Director
                                                                                    )  Printed on Recycled Paper

-------
areas of the aquifer where it is needed, 2) the enhanced
removal of the target contaminants by the reactive agent, 3)
the removal  of the  reactive agent from subsurface, 4) the
impact of the reactive agent on the treatment of the target
contaminant and the volume of sludges to be disposed.
Additional site characterization, treatability tests, and design
studies must be conducted to  address each of these
important  aspects  of a  chemical-enhancement program. At
some sites, implementation of chemical enhancement will
require additional capital expenditures for wells and treatment
facilities. The advantages and disadvantages, including the
additional  costs,  of chemical enhancements need to be
compared with other methods  of remediation, such as
conventional pump-and-treat.

While many  individual components are associated with the
implementation of chemical enhancement to pump-and-treat,
they all should be investigated. If one aspect of this process
fails, the entire system can fail. While such failure is not
necessarily a disaster (conventional pump-and-treat can
continue),  it  is a waste of resources that could be utilized for
more beneficial uses. It is believed that these issues must  be
addressed and a reasonable probability of success
demonstrated in  all aspects of a chemical enhancement
system before it is  implemented.

Introduction

The recognition that ground water in many areas of the U.S.
is contaminated has brought about demands that the quality
of these aquifers be restored.  At Superfund sites, the initial
cleanup is accomplished in a relatively short time by
removing sources of contamination from the surface,
removing highly contaminated  shallow soil, and in some
cases installing a low-permeability cap. In contrast,
remediation  of the ground water beneath  a site is often an
inexact process requiring years to complete.

A common method for aquifer  remediation is to withdraw the
contaminated water from the aquifer and treat it on-site. The
treated water may then  be returned to the aquifer, discharged
to surface water, or transferred to a public water treatment
plant. Such "pump-and-treat" technology  is widely used for
remediating  aquifers (Palmer et al., 1988) with about 68% of
the  Records  of Decision identifying it as the method of
remediation  (Travis and  Doty,  1989). However, at many sites
pump-and-treat technology will require decades  of costly
operation to  achieve the desired levels of cleanup. Extended
periods for remediation are highly undesirable because the
operation and maintenance costs associated with the
remediation  can be large, and, in  many cases, otherwise
valuable land cannot be used for any economic purpose.

The great  costs of cleanup make it essential to investigate
technologies that may speed up remediation. One such
technology is the injection of chemical constituents, "reactive
agents", that improve the rate  of removal of contaminants
from the subsurface. The applicability of such "chemical
enhancement" technology and the specific chemicals that can
be used depend on the processes that control the slow
'tailing" of  contaminant concentrations in the extraction wells.
Not all processes leading to lengthy remediations can be
corrected by chemical enhancement. However,  certain
problematic types of contamination maybe amenable to well
conceived applications of reactive agents.
 The limitations of aquifer remediation by conventional  pump-
 and-treat will encourage engineers, scientists, and regulators
 to propose various chemical enhancement methods for the
 remediation of particular sites. While these proposed
 methods must be evaluated with regard to specific site
 conditions, there are general concepts applicable to all
 chemical enhancement  methods. This document is intended
 to 1  ) outline these general concepts, 2) pose key questions
 that should be answered before any chemical-enhancement
 scheme is initiated, 3) stimulate discussion on the merits and
 limitations of chemical enhancement methods, and 4)  focus
 research on particularly  problematic areas of chemical
 enhancement.

 Processes Affecting  Pump-and-Treat  Remediation

 A major concern in pump-and-treat operations is that
 contaminant concentrations within the extraction wells  will
 decline at a progressively slower rate as pumping continues.
 When the rate of decline becomes small and  the contaminant
 concentrations are still above the target cleanup levels, an
 extraction well is said to exhibit 'tailing" (Fig.  1). Contaminant
 concentrations may have dropped several orders of
 magnitude, but they remain above the target  clean-up  level
 despite a considerable period of pumping.  A  great
 uncertainty in pump-and-treat operations is the time required
 for these tailing concentrations to decrease below the  target
 clean-up levels. Reasonable estimates of clean-up times
 under these conditions require an understanding of the
 physical and chemical processes that can cause tailing: 1)
 the differing amounts of  time required by contaminated
 waters to flow along different streamlines from the irregular
 boundary of the plume to the extraction wells, 2) multiple
 rates of mass transport within spatially variable sediments, 3)
 limited mass transfer from reserves of nonaqueous phase
 liquids and solid phase mineral precipitates, and 4) slow
 resorption reactions (Keely et al., 1987; Keely, 1989).
    MAX
HI
O

O
O
                               WITHOUT
                               "TAILING"
                              "TAILING"
                            PHENOMENON
  "RESIDUAL"
:ONCENTRATION
       TIME  FROM  INITIATION  OF EXTRACTION
Figure 1. Concentration versus lime curve for an extraction well with
continuous pumping (after Keely et al., 1987).

-------
 Physical Causes of Tailing

Ground water entering an extraction well is a mixture of
waters that have traveled along multiple subsurface pathways
between the edge of the contaminant plume and the well.
The time required for contaminated ground water to travel
along these different flow paths is controlled by 1) placement
of the extraction wells relative to the contaminant boundaries,
2) the extraction rate, 3) the aquifer porosity, 4) the
magnitude and  direction of the natural hydraulic gradient, and
5) the location and types of hydraulic boundaries. As an
example, consider a single extraction well in an aquifer with a
natural  hydraulic gradient of 0.007 towards the east (Fig. 2).
If the edge of the contaminant plume is to the west, then only
a portion of the plume's edge is captured by the extraction
well. The flow paths along the outside of the capture zone for
the well have a  greater distance to travel and are influenced
by lower hydraulic gradients than the flow paths near the
center of the  capture zone. As a consequence of these
variable residence times within each stream tube, the
concentration-versus-time curve for the extraction well
exhibits substantial tailing even in the absence of chemical
reaction (Fig. 3).

In heterogeneous porous media, ground water in  higher
permeability layers has greater velocities than water within
the lower permeability zones. The higher permeability
pathways are not necessarily sand or gravel nor are the
lower permeability zones necessarily silts or clays; it is
sufficient if one region possesses greater hydraulic
conductivity  relative to the adjacent materials. When the
contrast in hydraulic conductivity between these zones  is
large, the advective component of transport through the
lower permeability lenses becomes small. As contaminants
are transported through such a heterogeneous aquifer, they
are advected along  the high  permeability layers and diffuse
into the lower permeability layers. Such an advection-
diffusion process  can affect the concentration  of
contaminants within higher permeability layers (Gillham et al.,
1984; Sudicky et  al., 1985).  If  aqueous contaminants have
been present over many years, their concentration in the
lower permeability layers can equal the concentrations  in the
higher permeability  zones.  During pump-and-treat
remediation, contaminants in the high permeability layers are
removed more quickly than from the lower permeability
layers. These variable rates of advective transport create
concentration gradients between zones of contrasting
permeability and cause the slow diffusion of contaminants
from the low permeability zones to the high permeability
zones where they can be pumped to the surface (Fig. 4).
Thus, the contaminant concentrations in the extracted water
are initially high as the more permeable layers are flushed.
At later times, the concentration in the extracted water  is
limited by the rate of diffusion of the contaminants into the
high permeability  zones  (Fig. 5). If pumping is discontinued,
the velocities  within  the high permeability layers decrease
and the concentration of contaminants within these zones
increase (Fig. 6) because of the greater residence time of a
parcel of water within the contaminated portion of the aquifer.

The main point is  that, in most cases, lengthy tailing-off of
contaminant concentrations in extraction wells is at least
partly due to physical attributes of the system that cannot be
ameliorated by injections of chemical agents.  Thus, chemical
enhancement cannot be expected to eliminate all unexpected
delays in pump-and-treat removal. However, when the rates
of chemical mass transfer from contaminant reserves in an
aquifer are the primary limitations on removal, then the use of
reactive agents may substantially enhance remediation.
                                    CAPTURE ZONE

                                   ğ. DIRECTION OF NATURAL
                                     GROUNDWATER FLOW
                                     STAGNATION
                                         POINT
                         EXTRACTION
                            WELL
Figure 2. Flow lines from the edge of a contaminant plume toward an
extraction well.
    0.8
in
o
o
o
UJ
LU

-------
                             ADVECTION
    ' ff~ff f * TTTTTTT" * "<"TT\

      MOtECULAR DIFFUSION
 LOW PERMEABILITY LENSES  '
                                        FTT\
                                    HIGH PERMEABILITY STRATA
                            ^      nmn rtHt/itABii
                                     MOLECULAR DIFFUSION
           ADVECTION
                                                   .ğ>*./
Figure 4. Heterogeneous porous medium with advection through the
low permeability zones and mass transfer by molecular diffusion from
the lower permeability lenses.
Z  10"
<
DC

Z
LU
    10
O
u
>  10"

3
LU
    10
       -4
                   1234

                     PORE  VOLUMES
                                                             o.
                                                             5
                                                             CL
                                                                 ON
g
<
DC

UJ
o
o
o
                                                                OFF

                                                                MAX
                                                                  o  -
                                                                     "RESIDUAL"
                                                                     CONCENTRATION
                                                                                                          CESSATION
                                                                                                          OF PUMPING
                                                                                                          (CLOSURE?)
                                                                                          *1      '2
                                                                     TIME  FROM  INITIATION  OF  EXTRACTION
                                                             Figure 6.  Concentration versus time for an extraction well that is turned
                                                             oH at time I,.
Chemical Causes of Tailing

At many sites, some or even most of the contaminant mass
will not be dissolved in the ground water but will be present
as 1) adsorbed species, 2) precipitates, or 3) nonaqueous
phase liquids (NAPLs).  " These reserves of matrix-associated
contaminants and contaminants in the immobile fraction of
the NAPLs cannot be directly extracted by pump-and-treat:
they must transfer from the solid or NAPL to the ground
water before they can be removed.  If the equilibrium
concentration in the ground water is small relative to the total
mass of contaminant in the soil or if the rate of mass transfer
is small relative to the ground-water velocity, then large
quantities of water must pass through contaminated sections
of aquifer before it is remediated.

If reactions between solutes and stationary phases are rapid
relative to the flow rate, equilibrium partitioning can be
assumed. However, rapid equilibration  does not translate to
rapid removal  rates if the equilibrium concentration in solution
is very low.  The retention of contaminants by mineral
surfaces and microbial  cell walls, ion exchange reactions with
clays, and the partitioning of organic contaminants between
soil organic  matter and the ground water can significantly
increase the time required for remediation of contaminated
aquifers. For hydrophobic, nonpolar organic compounds,
resorption can often be represented by linear isotherms
(e.g., Chiou et al., 1979). In the absence of free product
(NAPL), the number of pore volumes required to remove the
organic contaminant from a homogeneous aquifer is
approximately equal to  the retardation factor, R,
Figure 5. Concentration versus time for removal of contaminants from a
ideally layered aquifer. The layers are assumed to be 10 cm thick, the
length is 10 m, the retardation factor is unity, and the diffusion
coefficient is 10 Bcrrr7s. Calculated using the equations given by
Sudickyand Frind (1962).
                                                 (1)

-------
where pb is the dry bulk density of the soil, n is porosity, foe is
the fraction of organic carbon in the soil (mass of carbon/
mass of soil), and Koc is the partition coefficient for the
contaminant into  soil organic carbon (mass per unit mass of
carbon/equilibrium concentration in water). A compound  with
a large Koc value  can have a large retardation factor even in a
soil with a small to a moderate amount of organic carbon.
Thus, many  pore volumes of water must be flushed through
the soil to remove such hydrophobic organic contaminants.

In some cases equilibrium partitioning may not be applicable.
Laboratory tests  had shown that weeks are required to
achieve equilibrium concentrations  in laboratory experiments
with sediments (Hamaker and Thompson, 1972;  Coates and
Elzerman, 1986;  Karickhoff,  1980). Resorption of pyrene,
hexachlorobenzene, and pentachlorobenzene from river
sediments requires days to weeks (Karickhoff and Morris,
1985). If such rates of resorption are slow relative to the
rate of ground water flow, then equilibrium concentrations
may not be attained during pump-and-treat and tailing in  the
concentration-versus-time curve can result.

Although the linear adsorption model  is adequate for
describing the adsorption equilibria of many nonpolar,
hydrophobic organic contaminants (Chiou et al., 1979), it
does not adequately describe the behavior of organic or
inorganic ions over a wide range of pH and adsorbate
concentrations. The adsorption of ionic solutes is often
represented by an adsorption isotherm. An adsorption
isotherm is a plot of the contaminant concentration on the soil
versus the equilibrium solution concentration of the
contaminant. Adsorption isotherms are defined according to
their general shape and mathematical representation. For a
Langmuir isotherm, the concentration on the soil  increases
with increasing concentration in the ground water until a
maximum  concentration on the soil is reached (Fig. 7). The
isotherm can be represented by the equation:
                S = S
                             KG
                      max
                          ,1  + KC
(2)
where S (mass/mass) is the concentration on the soil, S
(mass/mass) is the maximum concentration on the soil, Kmax
((length) Ymass) is the Langmuir adsorption constant, and C
(mass/( length)3) is the concentration in the ground water. A
Freundlich (or Kuster) isotherm is given by the equation:
           LU  M
           oĞ

           o S
           o "5T
           uj  ra
               8
               i
           n
                                       1  +  KC
                    AQUEOUS  CONCENTRATION
           Figure 7. A Langmuir isotherm.
           O
           p  €
               I

o  .2
UJ  %
h-  tr
m  I
DC  Ğ
g  8

<  1:
                                                       b < 1
                                            S  =  KC
                      AQUEOUS  CONCENTRATION
                        S = KC
(3)
Figure 8. Freundlich isotherms, A linear isotherm is the special case
for which the exponent, b, is equal to unity,
where K is the Freundlich adsorption constant and b is a
positive parameter. The shape of a Freundlich isotherm
depends on the value of b.  If b is greater than  1.0, the
isotherm becomes steeper with increasing concentrations in
the ground water. If b is less than 1.0, the isotherm  becomes
steeper at lower concentrations (Fig. 8). A linear isotherm is
a special case of the Freundlich isotherm where the
parameter b is equal to unity. At  constant pH, cations tend to
follow Freundlich isotherms while anions tend to follow
Langmuir isotherms (Dzombak, 1986, Dzombak and
Morel, 1990).
           Adsorption isotherms are useful for illustrating the
           dependence of the solid phase concentration of the
           contaminant on the aqueous phase concentration of the
           contaminant at a given pH. However, adsorption  of inorganic
           ions is pH-dependent and the form of the isotherm should be
           known over the entire pH range likely to be found at a site.
           Sometimes this pH-dependence is presented as the fraction
           of the contaminant adsorbed versus pH or a "pH-edge" (Fig,
           9). For cations, the pH-edge for most minerals show little or
           no adsorption at low pH. As pH increases, the  portion of the
           contaminant that is adsorbed increases until the fraction is

-------
 unity (provided that  the  mass of contaminant does not
 exceed the available adsorption  sites). The pH-edges  for
 anions are the opposite to those for cations. There is little or
 no adsorption at higher pH but as pH decreases the fraction
 of the anion that is adsorbed increases to unity or to the ratio
 of the mass of sites to the  mass of contaminant if the amount
 of contaminant exceeds the number of available sites. For
 either cations or anions, the shape and position of the pH-
 edge depends on the specific mineral surface and ions under
 consideration.
       100
HQ 80
LLJJQ 60
OC f\ 40
s§20
    ^   0
       100

h-3 8°
LUflC 60
DC (/) 40
Q-5 20
          0
                CATIONS
                         INCREASING ADSORBENT
                                        ANIONS
                    INCREASING ADSORBENT
Figure 9. pH adsorption edges for cations and anions. The fraction of
contaminant adsorbed will not reach unity if the sites are saturated.
Adsorption processes can also be modeled using surface
complexation models (e.g., Stumm et al., 1976; Schindler,
1981; Schindler and Stumm, 1987; Dzombak and Morel,
1990). The key advantage of this type of approach is that it
has a foundation in chemical theory allowing the results to be
extended beyond the exact test conditions. The dependence
of the amount of adsorption on the pH of the solution and the
competition between several adsorbates for the adsorption
sites are, in principle, accounted for in such a model. The
disadvantages of the surface complexation model are the
lack of a consistent set of equilibrium constants and the
potential lack of linear additivity when multiple adsorbents are
present. The first limitation is being overcome through
compilations of consistent data sets for oxide surfaces. At
this time, there does exist a consistent set of adsorption
 constants for adsorption onto hydrous ferric oxide (Dzombak,
 1986; Dzombak and Morel, 1990) based on a two-layer
 surface complexation model. Such data sets need to be
 derived for other oxide surfaces as well.

 The general concepts of ion adsorption can be applied to
 anticipate some of the behavior of contaminants during
 pump-and-treat remediation. The rate of removal  of ionic
 contaminants under acidic conditions can be substantially
 different than under neutral or  alkaline conditions. Adsorption
 of anions, for example,  is more likely to be a problem at lower
 pH than at more neutral or alkaline conditions. Furthermore,
 there may be changes in the amount of adsorption during
 remediation as  acid or alkaline  waters are returned to more
 neutral  pH conditions.  In all of these adsorption models, ionic
 contaminants follow nonlinear isotherms. The partition
 coefficient equals the slope of the adsorption isotherm. As
 aqueous contaminant concentrations decrease during
 remediation, the slope of the isotherm, hence the  retardation,
 changes. In most cases, the retardation will increase with
 decreasing concentrations making it more difficult to
 decrease intermediate concentrations below the maximum
 contaminant level (MCL) than to decrease the initially high
 concentrations to intermediate levels.

 Large reserves of inorganic contaminants maybe  formed as
 the result of the precipitation of crystalline and amorphous
 materials within the soils. For example, one concern is the
 potential effect  of a reserve of  solid BaCr04within  aquifer
 systems contaminated with hexavalent chromium  (Cr(VI))
 (Palmer and Wittbrodt,  1990).  As Cr(VI)-laden waters enter
 the subsurface, natural Ba may react with aqueous
 chromate  (CrO) to precipitate a reserve of BaCr04. In many
 cases, the size  of the reserve will be limited by the availability
 of Ba'in the soil rather than the mass of chromate spilled.
 During the inital phases of a pump-and-treat remediation,
 ground water containing high concentrations of Cr(VI) in
 excess of  available Ba2*are removed and the concentrations
 in the extraction wells quickly decrease (Fig. 10). At some
 point, BaCrO becomes the principal source of Cr(VI) in the
 pore water. The Cr(VI) concentration will remain  relatively
 constant for as  long as  there is BaCrOjemaining in the soil.
 Using equilibrium concepts, Palmer and Wittbrodt  (1990)
 estimated  that 25 to 50  pore volumes are required to remove
 Cr(VI) from soils at a hard chrome plating facility.  If
 equilibrium is not obtained  and  kinetic processes control the
 amount of BaCrO4dissolved as  the ground water passes
through the soil, then more pore volumes would be required.

 Similar volubility limitations may occur for other inorganic
 contaminants. The effect of precipitates on efficacy of pump-
 and-treat remediation depends on the volubility of the  mineral
 phase. The most troublesome  mineral precipitates are those
with solubilities  low enough to create a relatively  large
 contaminant reserve, yet with solubilities large enough to
 exceed the target clean-up levels. A complicating factor is
substitution of contaminants in the crystalline structure of
other minerals.  The degree of substitution affects the
 equilibrium concentration of the contaminant. Regardless of
whether the contaminant has been precipitated in pure or
substituted mineral phases, if the rate of dissolution is slow
 relative to the velocity of the ground water, then the time
 required for the removal of the contaminant from the
subsurface will be greater than when equilibrium conditions
 have been achieved.

-------
                                                            When pump-and-treat remediation is predominantly limited by
                                                            chemical processes that restrict the transfer of mass from
                                                            these contaminant reserves to the ground water,  chemical
                                                            enhancement to pump-and-treat should be considered.
                                                            Although the choice of a reactive agent that will greatly
                                                            enhance contaminant removal is a primary concern, there are
                                                            several other factors that must be considered before
                                                            implementation of a chemical-enhancement program.
   MAX
<
QC
UJ
O
o
O
             CONTAMINANT
           CONCENTRATIONS
             CONTROLLED
             BY SOLUBILITY
SOLID PHASE
  RESERVE
 DEPLETED
             tl                    t2
TIME FROM INITIATION OF  INJECTION
Figure 10. Concentration versus time for an extraction well in a
formation that contains a solid phase precipitate.
Nonaqueous-phase liquids can also provide large reserves of
contaminants in the subsurface. For example, if a cubic
meter of soil with a 35% porosity contains trichloroethylene
(TCE) at 20% residual saturation, then approximately 270
pore volumes must pass through the soil  and reach
equilibrium with the TCE (1100 mg/L)  before the solvent is
removed from the soil by dissolution.  Sandbox experiments
with  perchloroethylene (Anderson,  1988; Anderson et al.,
1992) suggest that this equilibrium is achieved very quickly
as the water passes through fingers of residual solvent.
Longer periods of time are required to remove solvents when
they  are present in  pools (e.g., Johnson and Pankow, 1992;
Anderson  et al., 1992). Using the equations given by Hunt et
al. (1988)  and reasonable choices  of transport parameters, if
can be shown that only the water that passes very close to
the edge of the solvent pool is likely to reach equilibrium
concentrations with the solvent while the  concentrations
further above the pool are limited by the rate of mass transfer
from the pool to the bulk aquifer (Fig.  11). The average
concentration of the solvent measured in  a monitoring well
with  a two-meter length of screen placed just above the pool
will increase across the length of the pool over which the
ground water has flowed (Fig. 12). If the  103 kg of TCE used
in the previous example is distributed in a 20 cm thick pool
below the  cubic meter of soil, the average concentration of
TCE in the groundwater exiting from the block of soil is 28.6
mg/L and  10,200 pore volumes must pass through the
aquifer before the TCE is removed. Thus, it takes
approximately 39 times longer to remove  solvent from a pool
than  to remove the same mass of solvent from residual
saturation.
                                                              0.30
                                                            ^0.25
O
O
a
u
o
CD
                                                              0.20
                                                     0.15
                                                     0.10
                                                            —. 0.05
                                                              0.00
                         GROUNDWATER VELOCITY = 0.1 m/d
                         TRANSVERSE DISPERSIVITY = 0.0003 m .
                         EFFECT. DIFFUSION COEFF. = 2.3E-5 m!/d
                                                                 0.0        0.2       0.4        0.6       0.8
                                                                            RELATIVE  CONCENTRATION
                                                            Figure 11. Concentration of a contaminant at different elevations above
                                                            a DNAPL pool for different pool lengths (after Johnson and Pankow,
                                                            1992).
                     z   0.05
                     O

                     <   0.04


                     UJ
                                                  o
                                                  o
                                                   UJ
                                                   CC
    0.03


    0.02


    0.01


    0.00
                                                               Average Concentration Over
                                                               2.0 m Interval Above Pool
                                                              MONITORING
                                                                WELL
                                                                  n
NAPL — ğ
POOL Ğ„Ğ,,,
^ -
-p-
2.0m
|
                                                        ALONG POOL ~~"1
                                                           0        2        4         6        8        10
                                                                  DISTANCE ALONG POOL (m)


                                                  Figure 12. Average concentration of a contaminant over a 2-m interval
                                                  above a DNAPL pool (after Johnson and Pankow,  1992).

-------
 Chemical Enhancements for  Pump-and-Treat
 Remediations

 If chemical enhancement of pump-and-treat is to be
 successful, four key areas must be satisfactorily addressed in
 the design: 1 ) delivery of the reactive agent to where it is
 needed within  the aquifer, 2) the interaction between the
 reactive agent and the contaminant,  3) the removal of the
 contaminant and the reactive agent from the subsurface, 4)
 the treatment of the extracted water  and disposal of the
 resulting sludges (Fig. 13).
INJECTION OF
 REACTIVE
 AGENTS
                        DISPOSAL
                      RECOVERY OF
                     REACTIVE AGENT
TREATMENT
 EXTRACTION OF
REACTIVE AGENTS
& CONTAMINANTS
          I DELIVERY | •ğ• I REACTION I -•• I REMOVAL I •
 Figure 13. Schematic representation of chemical enhancement ofa
 pump-and-treat operation. Key areas of concern are shown in boxes.
 In some cases, the reactive agent will be recovered and re-used.
 Delivery

 Delivery of the reactive agents to the areas within the aquifer
 where they are needed to enhance the removal of
 contaminants can be a complex process. Reactive agent
 solutions must be injected without clogging  the aquifer near
 the injection well  with particles and chemical precipitates.
 The ground water containing the reactive agents must then
 move in some reasonable period of time to  the contaminated
 portion of the aquifer. The rate, mode,  and  scheduling of the
 injection and pumping must be designed  such that the
 reactive agent reaches those areas in a relatively short
 period of time. Many of these processes are influenced by
 the heterogeneities within the aquifer.
Clogging of Injection Wells

The clogging of injection wells is a common problem.
Clogging can be the result of the physical filtration of
suspended particles at the well interface, the formation of
inorganic precipitates, or the growth of microorganisms. As
more materials are entrapped, precipitated, or grown in the
pore space adjacent to the well face, they occupy increasing
amounts of the pore space and severely reduce the hydraulic
conductivity (Palmer and Cherry, 1984). A plot of the ratio of
the new permeability, k, to the original permeability,  k0,
versus the new porosity,6, to original porosity,90, using the
Carmen-Kozeny model (Carmen,  1937)'and'the Rumpf-
Gupte (1971) model (Fig.  14) illustrates that small changes in
porosity can result in order of magnitude reductions in
permeability. Because this reduction in  permeability is
immediately adjacent to the well screen, it is generally
manifested as a reduction in well efficiency.

The mechanisms responsible for well clogging dictate the
actions to avoid the problem. Additives  may be required to
prevent precipitation. Problems with particles in the injection
water can be overcome by flocculation and removal with filter
presses.  Microbiological activity can be inhibited by the
removal of nutrients or dissolved oxygen.  Otherwise,
periodic treatment such as surging, jetting, development, and
acid treatment of the injection well may be required. These
options need to factored into estimates of capital costs  of the
remedial design as well as the operation and maintenance
costs.
                                                                   0.1
                                                                 0.01
                                                                0.001
                                                Carmen (1937), n0 =0.47  	
                                               I Carmen (1937), nQ =0.20  	
                                                Rumpf and Gupte (1971)   	
                                                                            0.9    0.8
                                                                  0.7
                                                     0.6     0.5    0.4    0.3
                                                                    n/n,
                                      Figure 14. Ratio of permeability to initial permeability versus porosity to
                                      initial porosity using several empirical models (after Palmer and Cherry,
                                      1964).
                                      Transport of the Reactive Aaent to Contaminated Areas
                                      A fundamental problem with chemical enhancements to
                                      pump-and-treat remediation is getting the reactive agent to

-------
the contaminated portions of the aquifer so that it can interact
with the contaminants to facilitate their removal. Two major
problems should be considered: 1) differential flow paths
governed by well hydraulics, and 2) mass transfer between
heterogeneities that is governed by molecular diffusion.  The
paths the injected water follows depend on the hydraulic
properties of the medium, the aquifer thickness, the natural
hydraulic gradient, the rate of injection, the placement and
pumping (injection) rate of other nearby wells, and the
location and type of hydraulic boundaries in the vicinity of the
site. Some of these effects are illustrated for homogeneous,
isotropic aquifers with stream function calculations (Fig. 15).

The two wells in Fig. 15A are 50 feet apart. The up-gradient
well injects fluid  at a rate of 0.5 gpm while the down-gradient
extraction well removes water at the same rate. Not all of the
stream  lines from the injection well are  captured by the
extraction well. In this particular case, 20% of the injection
fluid continues to be transported through the aquifer. If the
wells are  placed 25 feet from one another (Fig. 15B), then
less of the injected fluid is lost (159'.) but a much smaller
volume of the aquifer is remediated with the chemical
extractant. If the rate of injection is reduced to 0.25 gpm, all
of the injected fluid  is captured by the extraction well (Fig.
15 C);  however, the volume of affected aquifer is diminished
even further.
    150
    100
      50
LJJ
O
52   -so
Q

    -100
   -150
                                          NATURAL  -
                                         HYDRAULIC
                                          GRADIENT
                                       EXTRACTION
                                          WELL
               INJECTION
                  WELL
       -150    -100
-50      0      50
 DISTANCE (ft)
                                             100   150
Figure 15a. Flow lines between an injection well and an extraction well
located 50 feet apart and with a natural gradient of 0.007 to the north.
Extraction rate and injection rate are both 0.5 gpm.
While, in principle, the distribution of wells and the rate of
injection and extraction could be optimized, in practice, it may
prove to be  difficult. Firstly, the factors to be optimized are
not universally agreed upon. Possibilities may include
minimizing the number of wells,  the total cost, or the time for
cleanup, or maximizing the mass of contaminant removed
per unit time. Secondly, none of the necessary parameters
                                       are known with  certainty.  For example,  concentration
                                       isopleths are known only within some range of distance, the
                                       variation in the hydraulic parameters have not been
                                       measured, and economic factors such as interest rates and
                                       costs over the next few years are based on extrapolation of
                                       current conditions.
                                          150
                                          100
                                      ^   50
                                      UJ
                                      O     0
                                          -50
                                      O
                                         -1oo
                                                                -150
                                                                                NATURAL
                                                                               HYDRAULIC
                                                                                GRADIENT
                                                                            EXTRACTION
                                                                                WELL
                                                    INJECTION
                                                       WELL
                                                                   -150    -100
                                                            -50      0      50
                                                              DISTANCE (ft)
                       100    150
                                      Figure 15b, Flow lines between an injection well and an extraction well
                                      located 25 feet apart and with a natural gradient of 0.007 to the north.
                                      Extraction rate and injection rate are both 0.5 gpm.
                                                                 150
                                          100
                                      ^   50
                                      £
                                      UJ
                                      O     0
                                                                 -50
                                                             Q
                                         -100
                                                                -150
                                                                                NATURAL  -
                                                                               HYDRAULIC
                                                                                GRADIENT
                                                                                                   EXTRACTION
                                                                                                       WELL
                                                                           INJECTION
                                                                              WELL
                                                                              I	|   |    1
                                             -150   -100
-50      0      50
 DISTANCE (ft)
                                                                                   100    150
                                      Figure 15c. Row lines between an injection well and an extraction well
                                      located 25 feet apart and with a natural gradient of 0.007 to the north
                                      Extraction rate is 0.5 gpm while the injection rate is 0.25 gpm.

-------
 Aquifer heterogeneity is an important factor in the rate of
 transport of dissolved constituents. As a reactive agent is
 injected, it is advected along the higher permeability
 pathways through the aquifer. When the hydraulic
 conductivity of the low permeability lenses is two or more
 orders of magnitude less than that of the high permeability
 zones, transport of the reactive agent into the lower
 permeability zones is controlled by molecular diffusion. The
 concentration of the reactive agent in the higher permeability
 zones should be maintained long enough to permit the
 diffusion of the reactive agent into the lower permeability
 lenses. The time it takes the reactive agent to diffuse into the
 lower permeability lenses depends on the diffusion
 coefficient, the concentration, and the soil-water interactions
 of the reactive agent.

 The process  of diffusion can be complex in a multicomponent
 system.  If a completing agent is used to enhance the pump-
 and-treat remediation, the complex initially forms in the
 higher permeability zone, creating a concentration gradient
 between the  higher permeability layers and the lower
 permeability layers. Thus, not only does the  reactive agent
 diffuse into the lower permeability zones, but so may the
 complex. The extent to which this affects the time for
 remediation depends upon the relative  rates of diffusion of
 the solutes into and out of the lower permeability lenses and
 the relative rates of reaction for the important chemical
 processes.
 Modes of Injection

 The above discussion implicitly assumes that the reactive
 agent is continuously injected into the aquifer, but there are
 several options for injecting the fluid as well as for adding the
 reactive agent. The water may be injected continuously, as a
 series  of pulses, or as a slug. With continuous fluid injection
 the reactive agent may be added to the injection fluid
 continuously or it may be added in a pulsed or slug mode.
 For pulsed injection of the fluid, the  reactive agent maybe
 injected with each pulse or as a slug into a single pulse of
 fluid. For a slug injection  of fluid, the reactive agent can be
 added  as  a single slug. Thus, there  are six general
 combinations of fluid and reactive agent injection  modes (Fig.
 16). The advantages and disadvantages of these modes of
 injection (Table 1  ) were used to determine the suitability of
 each method for chemical enhancements to pump-and-treat
 (Fig. 16).

 The modes most likely to be useful are continuous fluid
 injection with either continuous or pulsed addtition of the
 reactive agent. The use of the slug mode for the  injection of
 water is not practical; a relatively small slug of water injected
 into the aquifer does not affect a large volume of the aquifer.
 The major concern is the volume of aquifer over which the
 injected water is transported and how long it takes to
 accomplish this distribution. While continuous injection may
 provide the shortest time  period over which it takes to
 distribute the fluid over a  given volume of aquifer, there may
 be inherent economic  benefits of using a  pulsed mode.  If
 staff must be present during injection, then coordinating the
 injection pulses with personnel shifts may  be advantageous.

A key advantage of continuous addition of the reactive agent
 during  continuous fluid injection is that relatively high
 concentrations of the agent can be maintained within the
 higher permeability zones of the aquifer. This creates a large
 concentration gradient between the higher permeability
 lenses and the lower permeability  lenses, potentially reducing
 the time to remove the contaminants from those zones.
 However, this means that a larger mass of the reactive agent
 is required. A slug addition of the  reactive agent would
 require less mass; however, the concentrations may
 decrease with time because of the nonlinear relationship
 between the adsorbed and  aqueous concentrations. Pulsed
 addition  would repeatedly increase the concentration
 gradients between the higher and lower permeability zones,
 however, the gradients may be locally reversed between the
 passing  pulses. The system would have to be more carefully
 designed to ensure that the net direction  of diffusion of the
 reactive  agent is into the low permeability zones.
REACTIVE
AGENT
ADDITION
MODE
CONTINUOUS
PULSED
SLUG
FLUID INJECTION MODE
CONTINUOUS
good
good
fair
PULSED
N.A.
good
fair
SLUG
N.A.
N.A.
poor
                                   N.A. = not applicable

 Figure 16. Relative rating for different combinations of the modes of
 injection of the fluid and reactive agent.
Timing  of Injectibn 'of the  Reactive  Agent

Chemical enhancement can be initiated at any time during
the pump-and-treat remediation.  Injection of the  reactive
agent may begin 1 ) as the extraction program begins, 2) after
the concentration-versus-time curve significantly flattens, or
3) at some time intermediate to the first two.

An advantage of initiating injection of the reactive agent at
the same time that extraction  of the contaminated water
begins is that it provides the earliest start to the chemical
enhancements. The flow paths over which the reactive agent
must travel remain approximately the same and therefore
any delay in initiating the injection also delays aquifer
cleanup. However, several problems may arise suggesting
that this method may not always  be the most advantageous.
If high concentrations of contaminant are still within the
aquifer, then a greater concentration of the reactive agent
may be required to remove this material. The cost of the
additional reactive chemicals should be compared with the
cost of a longer extraction time.  Geochemical interactions
within the contaminant plume  can result in elevated
concentrations of  solutes other than the contaminant (e.g.,
                                                        10

-------
Table 1. Modes of injection for chemical enhancement to pump-and-treat
            FLUID
          INJECTION
            MODE
    ADVANTAGES
      DISADVANTAGES
         CONTINUOUS
         SLUG
         PULSED
         ACTIVE  AGENT
           INJECTION
             MODE
         CONTINUOUS
         SLUG
         PULSED
Fluid distributed over wide area.


Less maintenance of pumping schedules.

Less volume of water.


Less potential dogging of wells.

Can be developed around working
schedules.
Maintain concentration in high
permeability zones allowing for
diffusion into low permeability zones.

Requires less mass of active agent.
Less total mass of active agent.

Can be planned around work schedules.
                              Allows for sufficient time for
                              diffusion.
Greater potential for clogging of
screens.

Greater pumping costs.

Fluid distributed over very small volume
of aquifer.
Requires more design to insure injection
and off periods are balanced relative to
natural groundwater flow.
Requires more mass of active agent,
May not allow sufficient time for
diffusion into low permeability lenses.

Concentration decreases with
time/distance which can reduce
effectiveness of the active agent.

Requires greater maintenance/control.

Requires more analysis to insure that
injection and off periods are of
sufficient length.
Fe and Mn in anoxic plumes). High concentrations of these
solutes may severely impede the effectiveness of the reactive
agent, thereby requiring a greater mass of reactive agent to
be injected into the aquifer. In some cases, these elevated
levels of solutes may result in precipitation of the reactive
agent and clogging of the aquifer.

Some of these problems can be avoided by using
conventional pump-and-treat until the rate of decline in the
concentration of the contaminant is low. As the
concentrations of the contaminant decline, the high levels of
other solutes present within the contaminated area are also
likely to decline toward background levels. At lower
concentrations of interfering solutes, the reactive agent
interacts more efficiently with that portion of the contaminant
that is most difficult to remove by conventional pump-and-
treat.  However, it may require the removal of several pore
volumes over several years to reach these lower
concentration levels so added costs for operations and
maintenance may be incurred.
                                 The third possibility is to initiate chemical enhancement at
                                 some time between the initial start-up of the pump-and-treat
                                 system and the time it takes for the concentrations to level
                                 off. Deciding the optimal time requires a  more sophisticated
                                 analysis then either of the previous two choices, however,
                                 some simple criteria may serve as guides. For example,  if
                                 there is concern over the precipitation of a solid phase, the
                                 criterion may be the time at which the concentration of the
                                 interfering ion decreases below the critical concentration
                                 computed from the volubility limit and the concentration of the
                                 reactive agent. While this is not necessarily the "optimal"
                                 time,  it may serve as a practical estimate.
                                 Rptarrlatinn anrl thp Rate of Transport nf thp
                                 Agent
                                 The reactive agent must travel through the aquifer almost as
                                 quickly as the water.  If the reactive agent is significantly
                                 retarded, then it may  take longer for the  reactive agent to be
                                 transported to the target areas of the aquifer than it takes to
                                                          11

-------
 remove the contaminant from the subsurface without
 chemical enhancement. Thus, there is a possible paradox
 here: the reactive agent must react within the subsurface to
 enhance the removal of the contaminant, yet it must not be
 retarded. However, careful consideration of the chemistry of
 the system may allow this paradox to be resolved so that
 both objectives are achieved.

 If the reactive agent is chosen to compete with the
 contaminant for adsorption sites, both of these objectives can
 be  realized by controlling the concentration of the reactive
 agent. For  example, if the adsorption of the reactive agent
 follows a Langmuir-type adsorption isotherm, the amount of
 retardation is insignificant  if the concentrations are high
 enough to saturate all of the available adsorption sites in the
 soil. If a completing agent is utilized to  enhance the removal
 of the contaminant, then it should be chosen so that neither
 the agent nor the complex are significantly retarded.
 Reactive agents that are reducing or oxidizing agents maybe
 retarded as they react  with materials other than the
 contaminant. Injection  may have to simply continue until all
 of the material between the injection point and the extraction
 well that reacts faster than the contaminant is titrated from
 the aquifer. The amount of oxidant or reductant required to
 titrate the soil can be estimated from the oxidation and
 reduction capacities of the soil (Barcelona and Helm, 1991).
 In some cases, if something is known about the reaction
 rates for these  redox reactions, some control can  be obtained
 through control of the ground-water velocities  (i.e., by
 controlling the rates of  injection and extraction).

 Reactive Agent - Contaminant Interactions

 Different reactive agents can be chosen (Table 2) depending
 upon  the processes that control the tailing in the
 concentration-versus-time  curve for the extraction wells. The
 reactive agent may  compete with the contaminant for
 adsorption sites, complex the contaminant, change the redox
 state  of the contaminant, change the solvation properties of
the  groundwater, act as a  surfactant,  ionize the contaminant,
 or substitute for the contaminant in a precipitate. These
 possibilities are not necessarily exclusive of one another. For
 example, a reactive agent may change the redox state of the
 contaminant and then form a complex with the altered form.

 Completion for Adsorption  Sites

 If the tailing of the concentration-versus-time curves for the
 extraction wells are controlled by adsorption processes, the
 reactive agent can be chosen to compete for the adsorption
sites.  Such competition is  most likely to be effective for ionic
solutes and least effective  in  displacing neutral organic
 molecules partitioned into soil organic matter.

The general concepts of ion adsorption can be applied to
 anticipate some of the constraints on the use of reactive
 agents to enhance pump-and-treat remediation. The
adsorption of ions onto hydrous ferric oxide may be used as
 a model for appreciating the qualitative effects. Ultimately,
 laboratory tests utilizing the site contaminants  and geologic
 materials should be performed. In general, we expect
competition to be significant only when the adsorption sites
are  near saturation. An ionic contaminant can be easily
displaced by a reactive agent with similar adsorption
 properties if the concentration of the reactive agent is
 sufficient to saturate the adsorption sites. Ionic contaminants
 can also be displaced by a reactive agent with a lower
 adsorption affinity but only if the agent is present in great
 excess of the total number of adsorption sites in the soil.
 Table 2. Reactive agents - contaminant interactions


         • Competition for adsorption sites

         • Complexation of the contaminant

         • Cosolvent effects

         • Enhanced mobilization and solubilization by surfactants

         • Oxidation

         • Reduction

         • Precipitation/Dissolution

         • lonization



 Complexation

 The reactive agent may be effective in forming aqueous
 complexes with an ionic contaminant. The aqueous
 complexes are not expected to be adsorbed as readily as the
 noncomplexed contaminant, therefore they are more mobile
 and relatively easy to remove by pump-and-treat technology.
 For example, James and Bartlett (1 983) found that citric and
 diethylenetriaminepentaacetic (DTPA) acids  complexed
 Cr(lll) sufficiently to maintain elevated solution concentrations
 at pH 7.5 and 6.5, respectively. These organic acid anions
 can also contribute to the removal of chromium by competing
 with chromate for  adsorption sites at oxide  surfaces. Citrate
 may also contribute to the reduction of the Cr(VI) to Cr(lll)
 which can then be complexed by another citrate molecule.

 While there  are many such potential chelators, those that are
 environmentally safe enough to be used in aquifers are
 relatively weak and non-specific in their binding action. Two
 key consequences of these properties are that 1) the chelator
 must be present in great excess of the contaminant
 concentrations and 2) high concentrations of common,
 nonhazardous soil constituents such as Fe,  Mn,and Al  may
 also be removed (e.g., Grove and Ellis, 1980; Norvell, 1984) .
 The presence of these constituents in the waste stream may
 substantially increase the costs of treatment  and disposal
 over conventional  pump-and-treat. Some chelating agents,
 such as citrate, can be utilized as  substrate  by bacteria in the
 subsurface.  As the concentrations of contaminants decrease
 below toxic levels around injection wells, bacterial growth
 may increase to the level where the increased biomass clogs
the aquifer and wells.

 In some cases, the adsorption properties of the soil matrix
 may be altered by the  use of chemical extractants. The
 removal of iron and aluminum oxide surfaces should
 decrease the adsorption density of the geologic materials.
 Zachara et al, (1988b); however, found the  adsorption of
                                                        12

-------
 chromate onto kaolinite increased with treatment with
 dithionate-citrate-bicarbonate  (DCS) or hydroxylamine-
 hydrochloride (NHZOHHCI) solutions. The reasons for this
 increase in adsorption are not clear.

 Treatability studies should be  conducted to determine not
 only the efficacy with which contaminants are removed by
 such chelators but also to estimate the total load of metals
 that must be treated and  disposed, and the potential
 increases in biomass. The cost of these additional loads
 must then be compared with the costs of conventional pump-
 and-treat remediation.

 Cosolvents

 The rate of removal of hydrophobic organic contaminants is
 often limited by their relatively low volubility in water.
 However, the solubilities of many of these  contaminants are
 much greater in  other solvents. Theoretical models
 suggesting an exponential decrease in the amount of
 adsorbed organic contaminant with increasing fractions of
 water miscible solvents (Rao et al., 1985; Woodburn et al.,
 1986) have been substantiated in laboratory experiments for
 several  organic compounds (Rao et al., 1985; Nkedi-Kizza et
 al., 1985,  1987; Mahmood and Sims, 1985; Woodburn et al.,
 1986; Fu and Luthy, 1986a, 1986b; Zachara et al., 1988a).
 For example, the adsorption coefficient for anthracene in
 methanol-water mixtures  decreased by four orders of
 magnitude as the fraction of methanol was increased from O
 to 1 (Nkedi-Kizza et al., 1985). The injection of cosolvents
 may therefore be expected to  drastically increase the
 volubility and decrease the retardation factors for these
 organic compounds thereby facilitating their removal from the
 subsurface. Cosolvents that are used as substrate by
 microbes may have the added advantage of promoting co-
 metabolism of primary contaminants. Small amounts of
 biodegradable cosolvent that are difficult to remove from  the
 subsurface will be of less concern because of their eventual
transformation. Thus, cosolvents, such as  alcohols, are
 potentially effective reactive agents for chemical
 enhancement to  pump-and-treat of hydrophobic organics.
 However,  some consequences of cosolvent  injection maybe
 less desirable.

 The order-of-magnitude decreases in adsorbed contaminants
 are generally achieved with cosolvent concentrations greater
than 20%. Fluids containing this amount of cosolvent will
 have densities and viscosities that differ substantially from
the ground water. Thus, the transport behavior of these
fluids is  more complex and more difficult to predict than for
fluids with homogeneous  properties. Cosolvent interaction
with clays in the aquifer matrix may either increase or
 decrease the permeability of the soil. Cracks have appeared
 in soils treated with methanol (Brown and Anderson, 1982).
The formation of such high permeability pathways maybe
 particularly troublesome at sites where  dense nonaqueous
 phase liquids (DNAPLs) are present. Cosolvents such as
 methanol can serve as substrate for subsurface microbes
 resulting in biofouling of the aquifer. Biotransformation may
substantially alter the geochemistry of the aquifer and
 promote the reductive dissolution of Fe and Mn oxides.
These metals can create problems with well clogging and
 interfere with  surface  treatment. Also, additional treatment
facilities must be  constructed for the separation of the
 cosolvent from the water. These facilities incur capital
 expenditures as well as operation and maintenance costs.

 Surfactants

 A surfactant adsorbs to interfaces and significantly decreases
 the interracial tension (Rosen, 1978). This property of
 surfactants has made these chemicals useful in enhanced oil
 recovery and several  researchers have proposed their use in
 the remediation of NAPL-contaminated sites (e.g., Ellis et al,
 1985). In  general,  surfactants are  composed of  a
 hydrophobic moiety,  often a long chain aliphatic (Cloto C20)
 group, and a hydrophilic moiety (Fig. 17) that can be anionic,
 cationic, nonionic, or zwitterionic (i.e., possess both positive
 and negative charges).  The orientation of the surfactant
 molecules at an interface can reduce the interracial tension
 and alter the wetting properties of the soil matrix. When the
 interface is a nonaqueous phase liquid, the lowering of the
 interracial tensions decreases the capillary forces keeping the
 NAPL in place and results in greater mobility of the NAPL.
                                            HYDROPHILIC
                                               MOIETY
      ALKYLBENZENE  SULFONATE
     CH,    CH,   CH,    CH
CH3CHCH2CHCH2CHCH2CH
HYDROPHOBIC MOIETY
 Figure 17. The surfactant alkylbenzene sulfonate.
 For enhanced oil recovery, increased mobility of the NAPL
 allows a greater fraction of the available oil to be  pumped to
the surface. In the case of NAPLs that have a greater
 density than water (DNAPLs), increased mobility  is not
 necessarily desirable. Once mobilized, the DNAPL may
 migrate deeper into the aquifer.  If the DNAPL migrates into
 areas that were previously uncontaminated, additional wells
 and pumps will be required and the costs of remediation will
 increase accordingly.

 Surfactants can also promote the solubilization of
 hydrophobic organic contaminants. Above a critical
concentration known as the "critical micelle concentration,"
colloidal-size micelles can form by the aggregation of the
 monomeric surfactant molecules. In water, the micelles
form by the hydrophobic moieties grouping together in the
core of the micelle, and the hydrophilic moieties orienting
toward the surface of the micelle (Fig.  18). Hydrophobic
organic contaminants partition into the hydrophobic core
of the micelle thereby increasing the volubility of the
organic  contaminant.
                                                       13

-------
MICELLE FORMATION
                                      HYDROPHOBIC
                                           CORE
                                       HYDROPHILIC
                                         SURFACE
 Figure 18. Aggregation of surfactant molecules into a micelle.
 Although the successful application of surfactants to
 enhanced oil recovery has been demonstrated, transfer of
 this knowledge to aquifer remediation is not direct.
 Surfactants used for enhanced oil recovery are chosen on
 the basis of temperatures and salinities that are much higher
 than those at most hazardous waste sites. To achieve the
 desired behavior, the surfactant must be chosen for the
 solvent under the conditions of use (Rosen,  1978). Incorrect
 surfactant formulations may result in high-viscosity
 macroemulsions that are difficult to remove. The surfactant
 can alter the wetting properties of the soil matrix and cause
 the NAPL to become the wetting phase. The  NAPL would
 then occupy the  smaller pores of the soil matrix,  thereby
 exacerbating clean-up efforts. The toxicity and potential
 biodegradation of surfactants that will remain  in  the aquifer
 following NAPL removal is  of great concern in shallow
 aquifers.

 The use of surfactants for aquifer remediation looks
 promising; however, there  is little experience in their
 application.  Laboratory experiments have demonstrated
 enhanced removal of anthracene and biphenyls (Vignon and
 Rubin, 1989), petroleum hydrocarbons (Ellis  et al.,  1985),
 DDT and trichlorobenzene (Kile and Chiou, 1989), automatic
 transmission fluid (Abdul et al.,  1990), and PCBS (Ellis et al.,
 1985; Abdul and Gibson, 1991). Surfactant mixtures  that
 specifically address the needs for aquifer remediation need to
 be developed and tested in the field as well as in the
 laboratory. When DNAPLs are present, mixtures that
 increase solubilization more than mobilization may  be
 desired.

 Qxidants-Reductants

 The addition of a reactive agent that changes the oxidation
 state of a contaminant is potentially useful  for 1 ) decreasing
 the toxicity of the contaminant,  2) increasing its  mobility, or 3)
 increasing its susceptibility to completing  agents. For
 example,  chromium can be reduced from the more toxic
Cr(VI) to the less toxic Cr(lll). The oxidation of selenite
(Se(IV)) to selenate (Se(VI)) results in a solute that is less
toxic and more mobile. However, oxidants and reductants
are not specific and must .therefore, be in excess of the
amount of contaminant. This will locally alter redox
conditions within  the aquifer and may result in the
precipitation of solid phases that may clog the aquifer and
injection/extraction  wells or mobilization of other metals that
must be handled in the treatment train.

The rate of reaction is an important factor in considering an
oxidation or reduction  reaction to facilitate the removal of a
contaminant from the  subsurface.  Often the rates are
strongly dependent on pH. For example, rate of reduction of
Cr(VI) by ferrous iron varies with {H*}3(Wiberg, 1965).

Precipitation-Dissolirtion-lonfzatiorl

At metal-contaminated sites, it maybe  possible to add a
chemical constituent that will cause the precipitation of the
contaminant in a solid phase with very low volubility. For
example, the neutralization of acid mine waters by
carbonate-buffered solutions will cause the  precipitation of
metal-oxides,  hydroxides, and carbonates.  Pb2*can
precipitate as  a  relatively insoluble PbC03phase. While this
may reduce the risk of contaminant concentrations of
exceeding the  MCL, it does not remove the metals from the
site. The precipitates can continue to act as long-term, low-
level sources and the  contaminants may still enter the
biosphere through  plant root systems or erosion. In addition,
the precipitation of  metal oxyhydroxides and carbonates can
cause clogging of the aquifer and severe reduction in well
efficiency.

Remediation of contaminated sites by conventional pump-
and-treat may  often be limited by dissolution of sparingly
soluble mineral phases and nonaqueous phase liquids.
Reactive agents  that increase the volubility  of these phases
will release the contaminant into solution where it can be
removed via an extraction well. For example, many phenolic
compounds can be ionized at higher pH (e.g., Palmer and
Johnson, 1992).  The use of a base as a reactive agent will
enhance the volubility of the phenolic phase and decrease
the retardation factor of the dissolved  compounds. If
cadmium is being released into solution from CdCO3, the
addition of acid can dissolve the carbonate mineral phase
and bring the  Cd2*into solution. However, such treatments
are not selective and other ions including Fe, Al, and SiO2
will be added to solution.  These ions may interfere with
treatment processes and increase  the volume of sludge to
be  disposed. The natural buffering capacity  of the aquifer
will require that the concentration of injected acid or base be
in excess of the amount of contaminant in the subsurface.

At metals-contaminated sites where remediation is  limited by
the presence of a sparingly  soluble mineral  phase, it may be
possible to release the contaminant more rapidly by the
addition of ion  that  will substitute for the contaminant within
the mineral phase.  This is most likely to be applicable where
the availability  of one of the counter ions in the solid phase is
limited.  By scavenging the counter ion into  another solid
phase, the contaminant will  be released into solution where it
can be  easily  removed.  For example,  if BaCrOJimits the
remediation of chromium-contaminated sites, the injection of
high levels of  su If ate would precipitate BaS04and increase
                                                         14

-------
the volubility of the BaCr04component, thereby allowing the
removal of the Cr(VI) in fewer pore volumes than by
conventional pump-and-treat.

Such chemical enhancement methodologies require specific
knowledge about the mineral phases limiting aquifer
remediation and the geochemistry of the ground waters on
site. Detailed geochemical studies will increase the cost of
site investigation and feasibility studies but may more than
pay for themselves if efficient removal methods can be
identified and problems associated with  the implementation
and operation of the clean-up effort are  avoided.

Removal of the Contaminants and Reactive Agents
from the Subsurface

The basic concept of chemical enhancement to pump-and-
treat is to increase the mobility of the contaminants in the
subsurface so they may be more easily  removed via
extraction wells. Removal of contaminants from the
subsurface in a chemical enhancement  scheme, therefore,
requires decisions about the density, placement, and
pumping rates of these extraction wells. Several  aspects of
such an enhanced extraction system design will be similar to
those utilized in conventional pump-and-treat remediations.
For example, the contaminants must be contained within the
capture zones of the extraction wells. However, key
differences between conventional pump-and-treat and
chemical-enhancement extraction system designs need to be
explored. For example, in  conventional  pump-and-treat,
contaminants are expected to be retarded.  Therefore, the
density of the well system is typically chosen to shorten travel
times and thereby decrease the time for remediation to the
extent possible. In  chemical-enhancement  methods, the
reactive agent is used to make the contaminants behave
more like a nonreactive tracer. Therefore, it may be possible
to utilize a lower density of extraction wells to achieve an
effective removal of the mobilized contaminant. Another
important consideration in the  design of an  extraction system
for chemical enhancement is the coordination with the
injection of the reactive agent. As described above, there
can be advantages to initially removing high levels of
contaminants by  conventional  pump-and-treat before
initiating the injection of the reactive agent. Also, pumping
rates for the extraction wells may be adjusted during the
injection to aid in the distribution of the reactive agent within
the aquifer.

Use of a reactive agent in a pump-and-treat scheme
introduces one or more new chemical constituents into the
subsurface. To be effective, reactive agents generally must
be added to an aquifer at non-trace concentrations.  Even if
the reactive agent is harmless to human health, state and
federal regulations will often require that concentrations of
the reactive agent be lowered to some permissible level.
Removal of the agent then involves  all of the problems
encountered in the  removal of the original contaminant and in
some cases the agent may even be more difficult to remove.
For example, if the  reactive agent is a solute that is used to
compete with the contaminant for the adsorption sites, then
the reactive agent must have a greater affinity for the
adsorbent; but this  greater affinity also makes it more difficult
to remove from the subsurface.  There still  maybe a net
benefit if the target clean-up level for the reactive agent is
greater than for the contaminant. It can  also be argued that
the net risk is reduced because the reactive agent must, by
any reasonable choice, be less toxic than the original
contaminant.

One complication that may arise during the removal of the
reactive agent from the subsurface is clogging of the screen
and filterpack as waters are mixed at the extraction wells.
This problem is likely to be most acute when the reactive
agent changes the redox conditions in the  subsurface.  As
oxidized waters mix with reduced waters that contain iron,
precipitates may clog the screen, pipes, and treatment tanks.

Treatment and Disposal

The previous sections have outlined many of the technical
considerations that must be addressed to implement an
effective  chemical enhancement strategy. However, even a
chemical enhancement plan that is completely satisfactory in
terms of subsurface deployment and removal of solutions
may still present technical difficulties in the  treatment and
handling  of the extracted wastes. Three broad categories  of
post-extraction problems are discussed in this section: the
effects of the reactive agent on the treatment of the target
contaminants, the removal of the reactive agent from the
waste stream before discharge, and the recovery and reuse
of the reactive agent.
     Removal of the  Reactive Agent before Discharge

As described above, the use of a reactive agent in a pump-
and-treat scheme introduces one or more new chemical
constituents in non-trace levels into the water brought to the
surface. Extracted water will, therefore, contain substantial
quantities of the reactive agent, probably in excess of the
target contaminants.  State and federal regulations will often
require that concentrations of the reactive agent be lowered
to some permitted discharge level. For example, if
phosphate is used as an extractant, standards may restrict
the permissible concentration in discharges from the
treatment facility, even if the treated wastes are routed into a
municipal  sewage treatment system. Limits on  phosphate
discharges can be anticipated in localities in which
eutrophication  is a problem in waters receiving regional waste
waters.

If the levels of reactive agent in discharges from the site  are
regulated, then the treatment plan must explicitly include a
means of  removing  the agent.  In many cases, the most
efficient system will  effect the removal of the reactive agent
simultaneously with  the treatment of the targeted
contaminants.  For example, if phosphate were  used  to
enhance chromate removal, then the neutralization step in a
treatment  process could be modified to induce the
precipitation of much of the phosphate.  Phosphate and
reduced Cr would precipitate in the same step and could be
removed collectively in the sludge.

For the specific system mentioned, note that the treatment
procedure would need to be modified. Removal of Cr3*alone
can be effected by addition of NaOH to achieve  a basic pH.
The resulting sludge then contains a mixture of  Cr
hydroxides, probably coprecipitated with  by-products  of the
reductant  step. For instance, if bisulfife is the reducing
agent, sulfate and bisulfite will constitute  part of the sludge.
                                                        15

-------
 If phosphate is present in the extracted water, some
 phosphate is likely to coprecipitate with  Cr3*. However, the
 concentration of phosphate remaining in solution in such a
 complex system would be difficult to accurately predict and
 tests would be necessary to find the optimal pH  for
 phosphate precipitation.

 It is possible that pH adjustment alone would not precipitate
 sufficient phosphate. In that case, an additional treatment
 reagent would be needed; for example, substituting Ca(OH)2
 for  some or all of the NaOH in the neutralization step. Ca-
 phosphates are relatively insoluble and would strip out much
 of the  phosphate. However, pilot studies would  be needed to
 ensure that the sludges produced behaved in the desired
 fashion.  The presence of phosphate  might decrease the
 density of the sludge so that longer settling times are
 required. Furthermore, the presence of  Ca2*could lead  to
 the buildup of scale in unexpected parts of the system.  In
 general, it is wise to  bear in mind that the treatment process
 will be efficient only if it is regarded as a  coordinated
 chemical system in which the alteration of one part can cause
 dramatic changes in  the behavior of another part.

 If the reactive agent  is successfully coprecipitated with the
 target contaminants,  the total volume of  sludge sent to
 disposal will increase correspondingly. Although the reactive
 agent may be harmless, once it is commingled with a toxic
 waste, the  entire volume  could be classified as hazardous
 and the cost of disposal assessed accordingly. Thus,  while
 the removal of contaminant and reactive  agent in a single
 step may save operation  or capital costs, the increased
 volume of sludge to be landfilled over the case where  no
 reactive agent is used, will generate costs that will offset
 some of the savings. The cost of testing and design of the
 removal  system for the reactive agent must also be factored
 into the economic analysis.

 In some cases, it will be desirable or essential to remove the
 reactive  agent in a separate stage of  the treatment system.
 One reason for a separate treatment  step is to achieve a
 better removal of the reactive agent than could be
 conveniently achieved in  a single step. In the example
 above, it might be disirable to optimize Cr3*precipitation
 without regard for phosphate and then strip phosphate out of
 the supernatant with Ca(OH)2or alum treatment  in a
 subsequent step. If the Cr3*precipitation  step could be
 designed to minimize phosphate coprecipitation,  and if the
 phosphate sludge were sufficiently free of Cr to be classified
 as non-hazardous, the two-stage removal would have the
 added advantage of minimizing the volume of hazardous
 solids for disposal.
Interference  of  Reactive Agent with  Treatment
Processes
Even if regulations do not require the removal of reactive
agents, it may be necessary to remove them from the
process stream. Some reactive agents may be harmless to
humans or to the environment, but they nonetheless may
have chemical properties that alter the behavior of the
contaminants in the waste stream. For example, many
conceivable reactive agents would function by completing the
target contaminants and enhancing their volubility.
Specifically, citrate or oxalate salts might be used to bind up
and mobilize metal ions. The same solubilization of metal
 ions that is desirable in the extraction step may become a
 major headache in the treatment step.

 As discussed above, most chelators that are environmentally
 safe enough to be used in aquifers, such as malonate,
 succinate, and citrate, will be relatively weak and non-specific
 in their binding action.  A weak affinity for the target metal
 means that the chelator must be present in great excess and
 will be found in corresponding excess in the extracted water.
 The excess chelator may interfere with  one or more
 segments of the treatment process by binding to the target
 contaminant.

 The most obvious interference would be the inhibition of
 precipitation. The soluble metal-chelator complex will not
 readily precipitate out of  solution. Lowering the pH will
 dissociate most metal complexes (such as citrate or oxalate),
 but metal  ions will be soluble at the lower pH. Raising the pH
 will favor precipitation,  but the higher pH also favors the
 binding of most  chelators. Very caustic pH levels maybe
 required to induce precipitation of metals in the presence of
 excess  chelator. An alternative scheme is  to remove the
 chelator from solution before the metal precipitation step. An
 organic chelator such as  citrate could be degraded by
 biological  treatment. An inorganic chelator such as
 polyphosphate would not biodegrade and would be difficult to
 remove economically by  chemical means.  Any process that
 requires a separate treatment step  for the chelator will  have
 greater operation and capital costs than the corresponding
 process in the absence of that step.

 The nonspecificity of chelators used as  reactive agents
 creates another substantial  problem for treatment processes,
 Aquifer  materials may contain large amounts of naturally-
 occurring metals that may be solubilized by the chelator. Iron
 and manganese will be the most important in many aquifers,
 but copper, zinc, aluminum,  and other metals may be
 extracted to varying degrees. The solution brought to the
 surface will contain not only free chelator, but also substantial
 quantities of chelator bound to non-target metals. In some
 situations, the amount of  chelator actually bound to the target
 metal(s) may be only a small fraction of the total chelator in
the extracted water. The  additional  metals  would not be
 solubilized in the absence of chelator, so these metals are a
 specific feature of chemical  enhancement.

 Large quantities of iron and manganese in  the extracted
water will require special  attention in the treatment process.
 Precipitation of the target metals will be accompanied by the
 precipitation of substantial amounts of iron  and manganese
 hydroxides. The sludge volume will be correspondingly
 increased, with the concomitant elevation of disposal costs.
 Furthermore, the behavior of iron in particular is apt to be
 different from that of many target metals. Iron is an
 especially insoluble metal (under aerobic conditions) and may
therefore precipitate out of the waste stream before the
target metal precipitates.  If this precipitation can be
 anticipated and controlled, it may simplify the treatment
 process. However a more likely scenario is that iron
 precipitation will be somewhat unpredictable and will
 occur in inopportune sections of the treatment facility,
causing plugging or fouling of the equipment and interfere
with the treatment process by coprecipitating the
contaminants.
                                                        16

-------
Pornuorw anrl
                       nf tho  Poartiuo
                                       Agent
                                                             Site Characterization for Chemical  Enhancements
In some cases, it maybe beneficial to remove the reactive
agent from the waste stream so that if can be re-injected into
the aquifer and re-used for additional extraction of the
subsurface contaminants. Such re-use  may be particularly
advantageous when the reactive agent is expensive or when
it must be removed because of concerns about interference
with treatment or because of regulatry  requirements on
discharges of the treated water. However, the methods that
may be used to extract the reactive agent from the waste
stream for re-use may not be the optimal methods for
removal for other purposes.

Effects  of Rapid and Concentrated  Extraction

One of the assumptions underlying the use of chemical
enhancement in a pump-and-treat operation is that the rapid
extraction of concentrated waste solutions is  beneficial to the
clean-up operation. Of course, this will be true at many
sites because annual operation and maintenance costs are
directly reduced by a more rapid removal of contaminants.

Furthermore, more concentrated wastes may be easier to
treat than dilute wastes. However, an accurate economic
analysis of the various options available in restoring an
aquifer should consider all costs associated with chemical
enhancement. Besides the above-mentioned increases in
research and development and operation and maintenance
associated with chemical enhancement, there may be
added costs due to the rapid extraction  and concentration
of wastes.

Careful consideration should be given to the capital  costs of
designing facilities that can efficiently handle  and treat large
volumes of a target contaminant in a short time. Injection/
extraction wells, settling basins, sludge  pumps, metering
pumps, and other facilities may need to be greatly expanded
to handle the concentrated waste  load.  Although the system
will be operated for a shorter period of time there will be a
tradeoff between  increased capital costs and  lower operation
and maintence costs.  It may be cheaper to run a pump-and-
treat operation for ten years with a small facility,  rather than
build a much larger facility that will only  need to operate for
one or two years.

Furthermore, planning in the pilot stages should give careful
attention to the performance of the treatment process at
different contaminant concentrations. If the process is
especially efficient and reliable at the high concentrations
attainable only with chemical enhancement, then the
additional costs will be mitigated. If, however, the
treatment process becomes more difficult or unreliable when
contaminants are very concentrated, then the use of
chemical enhancement may be contraindicated. Special
care is needed if a biological treatment step is anticipated.
Microbes generally thrive at higher substrate  concentrations,
but if higher contaminant levels lead to toxic levels of the
target contaminant or some secondary constituent, then
the biological treatment may fail altogether. Many of the
problems of high  concentration can be circumvented by
appropriate dilution, but this is a feature that should be
anticipated and incorporated into the design.
Rational implementation of a chemically enhanced pump-
and-treat remediation will require many of the same
characterization and testing methods required for
conventional  pump-and-treat operations.  Physical
hydrogeological parameters such as hydraulic conductivity,
the potentiometric surface, and porosity (Table 3) can be
obtained using the methods outlined by Mercer and Spalding
(1992a, b), Palmer and Johnson (1989), Rehm et al. (1985),
and Ford et al. (1984). These physical parameters can be
used in modelling studies to ascertain the feasibility of getting
the reactive agent to the contaminated areas within a
reasonable period of time while maintaining a capture zone
for the contaminant and the reactive  agent. The results of
such studies should help identify the optimum injection
concentrations, the number of wells, and their location.

If chemical enhancements are to be considered, greater
effort must be placed on the chemical characterization of the
site. In particular, the  key chemical processes that limit
pump-and-treat remediation must be  identified if the  proper
type of reactive agent is to be chosen, Important chemical
processes and their characterization have been recently
addressed  by Boulding and Barcelona (1992a,b,c), Palmer
and Johnson  (1992), and Palmer and Fish (1992).


Table 3. Physical-hydrogeological and chemical parameters that
should  be identified during site characterization for chemical
enhancements to pump-and-treat remediation


Physical-Hydrogeologlcal  Parameters

        • bulk density
        • porosity
        • hydraulic conductivity
        • storativity
        • potentiometric surface
        • site boundary conditions
        • ground water-surface water interactions
        • infiltration rates
        • leakage from adjacent aquitards

Chemical Parameters

        • pH
        • redox conditions
        • contaminant concentrations and spatial distribution
        • non-contaminant concentrations
        • oxidation capacity of the aquifer
        • reduction capacity of the aquifer
        • organic contaminant partition coefficients
        • ionic adsorption parameters
Several approaches must be used to determine the chemical
processes limiting pump-and-treat remediation. For ionic
solutes, adsorption tests are important for quantifying the
fraction of the solute adsorbed onto the surfaces of the soil
as a function of pH and the aqueous concentration of the
contaminant. The potential for mineral controls can be
identified by calculating mineral saturation indices using
geochemical models such as MINTEQ (Felmy et al., 1984)
and  may be verified through x-ray diffraction or electron
                                                         17

-------
 microscopy. Oxidation and reduction tests (e.g., Barcelona
 and Helm, 1991 ) are useful for determining the amounts of
 oxidant and reductant necessary to alter the redox state of a
 contaminant in the subsurface.  Bench-scale tests to
 measure the increase in solute concentrations following the
 addition of proposed reactive agents can provide information
 about potential compositions of interfering solutes entering
 the treatment train.  Treatment studies  using water
 compositions based on these tests can be used to determine
 potential problems and test proposed solutions to the
 treatment process.

 For neutral organic contaminants,  batch-sorption tests can be
 conducted to determine the fraction of the contaminant
 partitioned into the soil organic matter. However, at most
 sites, the partitioning can be determined from published
 values of the Kocof the contaminant (e.g. Maybey et al.,
 1982; Montgomery and Welkom, 1989) and the fraction of
 organic carbon in the soil. This approach is simpler than
 batch experiments;  however,  it  does require that the f of
 the soil be measured. When nonaqueous phase liquid are
 present, they are the limiting factor in site remediation; pools
 of NAPLs are more  problematic  than NAPLs retained at
 residual saturation in the soil. Again, proposed reactive
 agents should be tested at the bench scale and the
 treatability of the extracted water tested before pilot testing
 and implementation  of the chemical enhancement operation.

 The tests discussed above are generalizations of a few that
 can be  conducted. The specific tests required  at a site
 depend on the target contaminants and the nature of the soil
 materials from which they must  be extracted. Utilizing the
 knowledge from laboratory studies and the experience from
 other hazardous waste sites will  be important in directing the
 type of tests that need to be conducted.  Unfortunately, at
 this time, there have been few field demonstrations of
 chemical enhancement methods from which to obtain such
 experience.

 References

 Abdul, A. S., and T.L. Gibson, 1991.  "Laboratory studies of
 surfactant enhanced washing of  polychlorinated biphenyl
 from sandy material."  Environmental Science and
 Technology, 25(4): 665-671.

 Abdul, A. S., T.L. Gibson, and D.N. Rai, 1990.  "Selection of
 Surfactants for the Removal of Petroleum Products from
 Shallow Sandy Aquifers." Ground Water 28(6): 920-926.

 Anderson, M. R., 1988. "The Dissolution of Residual Dense
 Non-Aqueous Phase Liquids in Saturated Porous Media."
 Ph.D dissertation, Oregon Graduate Institute of Science and
 Technology,  Beaverton,  OR. 260p. (Available from
 University Microfilms).

Anderson, M. R., R.L. Johnson, and J.F. Pankow, 1992.
 "Dissolution of Dense Chlorinated Solvents into Ground
Water.  1. Dissolution from a Well-Defined Residual Source."
 Ground Water, In press.

 Barcelona, M.J. and T.R. Helm,  1991. "Oxidation-Reduction
 Capacities of Aquifer Solids". Environmental Science and
Technology,  25(9): 1565-1572.
 Boulding, J.R. and M.J. Barcelona, 1992a. "Geochemical
 Characterization of the Subsurface: Basic Analytical and
 Statistical Concepts." IN:  Handbook on  Site Chacteri'zation
  for Subsurface Remediation.,  Chapter 7. U.S. EPA, In  press.

 Boulding, J.R. and M.J. Barcelona, 1992b. "Geochemical
 Variability of the Natural and Contaminated Subsurface
 Environment." IN: Handbook on Site  Characterization for
 Subsurface  Remediation ,  Chapter 8. U.S. EPA, In press.

 Boulding, J.R. and M.J. Barcelona, 1992c. "Geochemical
 Sampling of Subsurface Solids and Groundwater." IN:
 Handbook on Site Characterization for Subsurface
 Remediation., Chapter 9. U.S. EPA, In  press.

 Brown,  K.W. and D.C. Anderson,  1982. "Effects of Organic
 Solvents on the  Permeability of Clay  Soils." Draft Report.
 Texas A&M University, Texas Agric.  Exp.  Stn., College
 Station. (Cited by Sheets,  P.J and W.H. Fuller, 1986.
 Transport of Cadmium by Organic Solvents through Soils."
 Soil Sci. Soc. Am. J. 50: 24-28).

 Carmen, P.C., 1937,  "Fluid Flow through a Granular Bed."
 Transactions of the Institute of Chemical Engineers,   15:   150-
 156.

 Chiou, C.T., L.J. Peters, and V.H. Freed,  1979. "A Physical
 Concept of Soil-Water Equilibria for Nonionic Organic
 Compounds". Science 206:831-832.

 Coates, J.T. and A.W. Elzerman,  1986.  "Resorption Kinetics
 for  Selected PCS Congeners from River Sediments." J.
 Contaminant Hydrology, 1:191-210.

 Dzornbak, D. M., 1986. "Towards a Uniform Model for
 Sorption of Inorganic Ions on Hydrous Oxides." Ph.D.
 Dissertation, Department of Civil Engineering, Massachusetts
 Institute of Technology.

 Dzombak, D.A.  and F.M.M. Morel, 1990.  Surface
 Complexation   Modeling. Hvdrous Ferric Oxide  Wley-
 Interscience Publication, New York.

 Ellis, W.D, J.R.  Payne, and G.D.  McNabb, 1985. "Treatment
 of Contaminated Soils with Aqueous Surfactants." U.S.
 Environmental Protection Agency,  Cincinnati, OH, EPA/600/
 2-851129, 84p.

 Felmy, A. R., D.C. Girvin, and E.A. Jenne, 1984. "MINTEQ: A
 computer Program for Calculating  Aqueous Geochemical
 Equilibria." EPA/600/3-84-032 (NTIS PB84-157148).

 Ford,  P.J, P.J. Turina, and D.E. Seely, 1984.
"Characterization of Hazardous Waste Sites - A Methods
 Manual, 11, Available Sampling Methods",  2nd ed. EPA600/
4-84-076 (NTIS  PB85-521596).

 Fu,  J.K. and R.G. Luthy.  1986a. Aromatic Compound
Volubility in Solvent/Water Mixtures. J. Environ. Eng.
 112:328-345.

 Fu,  J.K. and R.G. Luthy. 1986b. Effect of Organic Solvent
on Sorption of Aromatic Solutes onto Soils. J. Environ. Eng.
 112:346-366.
                                                        18

-------
Gillham, R.W., E.A. Sudicky, J.A. Cherry, and E.O. Frind,
1984. "An Advection-Diffusion Concept for Solute Transport
in Heterogeneous Unconsolidated Geologic Deposits. " Water
Resources Research, 20(3): 369-378.

Grove, J.H. and E.G. Ellis, 1980. "Extractable Iron and
Manganese as Related to Soil pH and Applied Chromium."
Soil Science Society of America Journal, 44: 243-246.

Hamaker, J.W. and J.M. Thompson,  1972. "Adsorption",
Chapter 2 in Organir P.hpmirals in thp Snil Fnuirnnmpnt.  C.A
Going and J.W. Hamakeer, Ed., pp. 49-143.

Hunt, J. R., M. Sitar, K.S. Udell, 1988. "Nonaqueous Phase
Liquid Transport and Cleanup 1. Analysis of Mechanisms"
Water Resources  Research, 24(8):1247-1259. (see
correction, Water Resources Research 25(6): 1450.)

James, B.R. and R.J. Bartlett,  1983.  "Behavior of Chromium
in Soils: VII. Adsorption  and Reduction of Hexavalent Forms."
Journal of Environmental Quality, 12(2): 177-181.

Johnson, R.L. and J.F. Pankow,  1992. "Dissolution of Dense
Chlorinated Solvents into Groundwater. 2. Source Functions
for Pools of Solvent." Environmental Science and
Technology, In press.

Karickhoff, S.W., 1980.  "Sorption Kinetics of Hydrophobic
Pollutants in Natural Sediments." IN: Contaminants-arid
Sediments. Vol. 2, Robert Baker, ed. Ann Arbor Science, pp.
193-205.

Karickhoff, S.W. and K.R. Morris, 1985. "Sorption Dynamics
of Hydrophobic Pollutants in Sediment Suspensions."
Environmental Toxicology and Chemistry, 1: 469-479.

Keely, J.  F., C.D. Palmer, R.J. Johnson, M.D. Piwoni, C.G.
Enfield, 1987. "Optimizing Recovery of Contaminant
Residuals by Pulsed Operation  of Hydraulically Dependent
Remediations," Proceedings nf thp Petroleum Hyrlrncarhnns_
anrl Orqanlr- P.homir-alQ In f^rniinrl \A/ator Prouontinn
Restoration Conference  anrl  Fypnsitinn	
Houston, Texas, November 4-6.

Keely, J. F., 1989. "Performance Evaluations of Pump-and-
Treat Remediations." Ground Water Issue Paper, U.S.
Environmental Protection Agency, EPA/540/4-89/005,
October,  1989.

Kile, D.E. and C.T. Chiou., 1989.  "Water volubility
enhancements of DDT and trichlorobenzene by some
surfactants  below and above the critical micelle
concentration." Environmental Science and  Technology,
23(7): 832-838.

Maybey,  W.  R., J.J. Smith, R.T. Podoll, H.L Johnson, T. Mill,
T.-W. Chou, J. Gates, 1. Waight Partridge, H. Jaber, and D.
Vandenberg,  1982. "Aquatic Fate Process Data for Organic
Priority Pollutants." EPA 440/4-81-014 (NTIS PB87-169090).

Mahmood, R.J. and R.C. Sims,  1985. "Enhanced Mobility of
Polynuclear Aromatic Compounds in Soil  Systems." IN:
Environmental Engineering Specialty Annual Conference
(Boston MA), American Society of Civil Engineers: 128-135.
Mercer, J.W. and C.P. Spalding, 1992a. "Geologic Aspects
of Site Remediation." IN: Handbook on Site Characterization
 or Subsurface Remediation  Chapter 3. U.S. EPA, In  press.

Mercer, J.W. and C.P. Spalding, 1992b. "Characterization of
Water Movement in the Saturated Zone." IN: Handbook on
Site Characterization for Subsurface Remediation  Chapter
4. U.S. EPA, In press.

Montgomery, J.H.  and L.M. Welkom, 1989. Groundwater
   Chemicals  Desk Reference. Lewis Publishers,  Inc., Chelsea,
Ml, 640 pp.

Nkedi-Kizza, P.,  P.S.C. Rae, and A.G. Hornsby. 1985.
"influence of Organic Cosolvents on Sorption of Hydrophobic
Organic Chemicals by Soils." Environmental Science and
Technology,  19:975-979.

Nkedi-Kizza, P., P.S.C. Rae, and A.G. Hornsby. 1987.
"Influence of Organic Cosolvents on Leaching of Hydrophobic
Organic Chemicals through Soils." Environmental Science
and Technology,  21(1  1): 1107-1111.

Norvell, W. A., 1984. "Comparison of Chelating Agents as
Extractants for Metals in Diverse Soil Materials". Soil
Science Society of America Journal, 48: 1285-1292.

Palmer, C. D. and J. A. Cherry, 1984. "Geochemical
Reactions Associated with  Low Temperature Thermal
Storage in Aquifers,"  Canadian Geotechnical Journal,  21,
475-488.

Palmer, C. D.,  W. Fish, J.F. Keely, 1988. "Inorganic
Contaminants:  Recognizing the Problem,"  Proceedings of
the Second National Outdoor Action Conference on Aquifer
Restoration,  Ground-Water Monitoring, and Geophysical
Methods,  May 23-26, 1988, Las Vegas, NV: 555-579.

Palmer, C.D. and W.  Fish, 1992. "Physiochemical
Processes - Inorganic Contaminants." IN:  Handbook on Site
Characterization for Subsurface Remediation.  Chapter 10.
U.S.  EPA, In press.

Palmer, C.D. and R.J. Johnson, 1992. "Physiochemical
Processes - Organic Contaminants." IN: Handbook on Site
Characterization for Subsurface Remediation. Chapter 10.
U.S.  EPA, In press.

Palmer, C.D. and R.J. Johnson, 1989. "Determination  of
Physical Transport Parameters." Chapter 4,  IN: Transport
and Fate of Contaminants in the Subsurface.  EPA162514-891
019.

Palmer, C.D. and P.R. Wttbrodt, 1990. "Geochemical
Characterization of the United Chrome Products Site,"  IN:
"Stage 2 Deep Aquifer Drilling Technical" Report, CH2M-HNI
to U.S. Environmental  Protection Agency.

Rae, P. S. C., A.G. Hornsby, D.P. Kilcrease, and P. Nkedi-
Kizza.  1985. "Sorption and Transport of Toxic Organic
Substances  in Aqueous and Mixed Solvent Systems."  J.
Environmental Quality, 14:  376-383.
                                                       19
               •U.S.Government Printing Ofllce: 1992—648-003/40716

-------
   Rehm, B. W., T.R. Stolzenburg, D.G.  Nichols, R. Taylor, W.
   Kean, B. Lowery, 1985. "Field Measurements Methods for
   Hydrogeologic Investigations: A Critical Review of the
   Literature." Electrical Power Research Institute Report  EA-
   4301, Project 2485-7, Electrical Power Research Institute,
   Palo Alto, CA

   Rosen,  M.J., 1978. Surfactants and Interracial Phenomena.
   John Wiley & Sons, New York, 304p.

   Rumpf,  H. and Gupte, A.R. 1971. "Einflusse der Porositat
   und Korngroenverteilung im Widerstandsgesetz der
   Proenstromung.  Chemie Ingenieur Technik, 43: 367-375.

   Schindler, P. W., 1981. Surface Complexes at Oxide-Water-
   Interfaces, IN: Adsorption of Inorganics at Solid-Liquid
   interfaces. M.A. Anderson and A.J. Rubin (editors), Ann
   Arbor Science, Ann Arbor, Ml, p. 1-49.

   Schindler, P.W. and W.  Stumm, 1987. "The Surface
   Chemistry of Oxides, Hydroxides, and Oxide Minerals."  In:
   Aquatic Surface Chemistry. W. Stumm, editor, pp. 83-110.

   Stumm, W.,  H. Hohl, F.  Dalang, 1976. "Interaction of Metal
   Ions with Hydrous  Oxide Surfaces." Croat.  Chem. Acta.,
   48(4): 491-504.

   Sudicky, E.A. and  E.O.  Frind, 1982. "Contaminant Transport
   in Fractured Porous Media: Analytical Solutions for a System
   of Parallel Fractures," Water Resources Research, 18(6):
   1634-1642.
              Sudicky, E. A., R.W. Gillham, and E.O. Frind, 1985.
              "Experimental Investigation of Solute Transport in Stratified
              Porous Media: 1. The  Nonreactive Case." Water Resources
              Research, 21(7):  1035-1041.

              Travis, C.C. and C.B. Doty, 1989. "Superfund: A Program
              without Priorities". Environmental Science and Technology,
              23(1 1): 1333-1334.

              Vignon, B.W. and A.J.  Rubin, 1989. "Practical
              Considerations in the Surfactant-aided Mobilization of
              Contaminants in Aquifers." J. Water Pollution Control
              Federation, 61 (7): 1233-1240.

              Wberg, K. B., 1965. "Oxidation by Chromic Acid and
              Chromyl Compounds."  IN: Oxidation in Organic Chemistry
              K.B. Wiberg (Editor), Academic Press,  New York, pp. 69-184.

              Woodburn, K. B., P.S.C. Rae, M, Fukui, and P. Nkedi-Kizza,
              1986. "Solvaphobic Approach for Predicting Sorption of
              Hydrophobic Organic Chemicals on Synthetic Sorbents and
              Soils". J. Contaminant  Hydrology, 1:227-241.

              Zachara, J. M., C.C. Ainsworth, R.L. Schmidt, and C.T.
              Resch.  1988b.  "Influence  of Cosolvents on Quinoline
              Sorption by  Subsurface Materials and  Clays." J.
              Contaminant Hydrology, 2:343-364.

              Zachara, J.M, C.E. Cowan, R.L. Schmidt, and C.C.
              Ainsworth. 1988a. "Chromate Adsorption by Kaolinite." Clays
              and Clay Minerals, 36(4): 317-326.
United  States
Environmental  Protection
Agency
Center for Environmental Research
Information
Cincinnati, OH 45268
      BULK RATE
POSTAGE & FEES PAID
 EPA PERMIT NO. G-35
Official Business
Penalty for Private Use $300
EPA/540/S-92/001

-------