Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Contaminants in Groundwater Chemical Processes 2 INTROD UCTION JOHN A. CHERRY, ROBERT W. GILEHAM, an] JAMES F. BARKER University of Waterloo, Canada AB STRACT The movement of most toxic contaminants in groundwater is affected by chemical reactions that cause transfer of contaminant mass between the liquid and solid phases or conversion of dissolved species from one form to another. The chemical attenuation of inorganic contaminants occurs mainly by adsorption, precipitation, oxidation, or reduction. Organic contaminants can be adsorbed or degraded by microbiological processes, but at present little is known about their behavior, particularly under the anaerobic conditions that are common in contaminated groundwater. Field and laboratory studies have established that various toxic heavy metals, transition metals, metalloids, radionuclides, and other inorganic species can be mobile or immobile in the groundwater zone, depending on the hydrogeochemical conditions represented by the pH, the redox condition, the ionic strength, the mineralogy, the solid-phase surface area, and the complexing capacity. Although the importance of chemical reactions in the attenuation of contaminants is widely recognized, the capabilities for attenuation predictions are not well developed. This is the case because the chemical processes within dynamic groundwater systems are complex; consequently, many of the geochemical parameters in predictive models are problematic. The prediction problem is complicated by the fact that the chemical processes are continually influenced by the redistribution of dissolved species caused by molecular diffusion and mechanical dispersion. The complexities of these mixing processes contribute to the difficulties in developing reliable methods for predicting the chemical behavior of contaminants in the groundwater zone. Contaminants can enter the groundwater zone from regional sources such as agricultural fields on which fertilizer or pes- ticides have been applied or from local sources such as landfills or waste spills. Within the groundwater zone, migration of the contaminants is influenced by advection, mechanical disper- sion, molecular diffusion, and chemical mass transfer. To pre- dict the behavior of contaminants in groundwater the effects of each of these influences must be adequately represented in a model or group of models. This chapter focuses on the pro- cesses that cause chemical mass transfer and on the manner in which these processes are represented in models. Emphasis is 46 placed on inorganic contaminants and, in particular, on those elements or compounds that have concentration limits specified in drinking-water standards. These include various transition metals, heavy metals, metalloids, and nonmetallic constituents such as nitrate. Nearly all of the literature describing the chem- ical behavior of contaminants in groundwater pertains to these inorganic constituents. The number and quantity of organic chemicals that are pro- duced have increased continuously since World War II. More than 3,000,000 organic compounds are known to exist and more than 40,000 are currently manufactured. Many of these are hazardous or potentially hazardous. In recent years many com- mon organic chemicals have been recognized to be hazardous
Contaminants in Groundwater: Chemical Processes and relatively mobile in permeable groundwater systems. In some regions widespread contamination by organic compounds exists in shallow groundwater. Organic compounds in ground- water can also cause major changes in the chemical behavior of inorganic constituents because of inorganic-organic reac- tions. Because there is a paucity of information on the behavior of dissolved organic compounds in groundwater, only a brief review of this subject is presented here. The emphasis in this chapter is on the chemical behavior of dissolved contaminants in porous geologic deposits. Only minor consideration is given to contaminant behavior in fractured rock or in fractured non- indurated materials. The movement of immiscible organic liq- uids in subsurface systems is not considered in this chapter. TRANSPORT WITH CHEMICAL MASS TRANSFER A common problem in the evaluation of the existing and future quality of groundwater is the determination or prediction of the effect of local sources of contamination. Examples of local sources are municipal and industrial landfills, industrial la- goons, roadsalt storage piles, septic sewage systems, mine and mill wastes, livestock feedlots, and local spills of industrial or agricultural chemicals. When local degradation of groundwater quality occurs, it is usually the result of downward movement into the groundwater zone of leachate from wastes or of spilled chemicals or waste liquids. As the contaminant solution mi- grates through the groundwater system, it displaces the original groundwater. The chemical composition of the zone of contam- ination continually changes because of dispersion, diffusion, and chemical reactions. The chemical reactions are driven by the changes in the chemical conditions caused by dispersion and diffusion and by the contact of the contaminated water with the surfaces of minerals and amorphous solids in the medium. Chemical changes in the zone of contamination may also be fostered by bacteria in the groundwater system. There are two main approaches that are commonly used for the prediction of the behavior of most chemically reactive in- organic contaminants in groundwater. The first approach in- volves the incorporation of a simple chemical mass-transfer term representing adsorption in the advection-dispersion equa- tion. This approach has the objective of predicting the advance rate and the shape of the front of a contaminant zone emanating from a continuous or a temporary source. The second approach has the objective of predicting the contaminant concentrations that will occur in the zone of contamination after chemical mass transfer has caused equilibrium concentrations to be achieved by precipitation, dissolution, oxidation, or reduction. The first approach has its origins in applied chemical chromatography, and the second was adapted from the chemistry of electrolyte solutions by geochemists interested primarily in seawater and surface continental waters. The processes of adsorption, dis- solution, precipitation, oxidation, and reduction usually occur simultaneously in zones of contaminated groundwater; there- fore, the use of predictive methods that represent adsorption with exclusion of the effects of the other processes or the other processes with the exclusion of adsorption is usually based on 47 convenience of conceptualization and computation rather than a quest for realism. Adsorption in Advective-Dispersive Systems Most currently used models for the prediction of transport of nonreactive dissolved contaminants in the groundwater zone are based on the advection-dispersion equation. The devel- opment of this equation is described by Anderson (Chapter 2 of this volume). Adsorption is usually incorporated into this equation in a manner based on the assumption that the con- centration of the contaminant in the solution phase (C) is a function of the concentration in solid phase (C), or C = fC. (3.1) With this assumption, the advection-dispersion equation for homogeneous saturated porous media is dCl~t= Dd2C/6x2 V&Cl~x f~p/n)Ksj3ClOt, (3.2a) where x is the flow direction, D is the dispersion coefficient in the direction of flow, V is the average linear groundwater ve- locity, p is the dry bulk density of the porous medium, n is the porosity, t is time, and Ks is the slope of the functional relationship expressed in Eq. (3.1~. The last term on the right- hand side of Eq. (3.2) represents the contaminant mass that is lost from solution as a result of adsorption. When the Ks func- tion is known, mathematical solutions for Eq. (3.2a) or for its two- or three-dimensional forms (including representation for heterogeneous geologic systems) can be obtained by means of numerical methods. It is important to recognize that Eq. (3.1) is based on the assumption that equilibrium conditions exist between the solution-phase and soIid-phase concentrations. Theoretically, Eq. (3.2a) is only applicable if the reactions are instantaneous or, in practice, only if the reactions are fast rel- ative to the groundwater velocity. In some situations, the relation between C and C is linear, and thus the slope of the partitioning function (Ks) becomes a constant and is generally referred to as the distribution coef- f~cient IKE. In this situation Eq. (3.2a) can be expressed as dC/6t = D'd2C/6x2 V'~C/6x, (3.2b) where D' is the effective dispersion coefficient (D' = DIR) and V' is the rate of advance of the front of the contaminant zone (V' = VIR) in the absence of dispersion. This front is retarded relative to the rate of advection of the front of nonreactive contaminants Ad. R is the retardation factor, defined as V/V' and is represented by the retardation equation, V/V' = 1 + (pln)K~. (3.3) In the absence of dispersion the front is conceptualized as a plug-displacement front. In the presence of Gaussian disper- sion, it represents the 50th percentile concentration level of an advancing slug of contamination (i.e., the middle of the dispersed front) emanating from a continuous source. The re- tardation concept is illustrated schematically in Figure 3.1 for laboratory conditions and for a hypothetical field situation. For the field example, the groundwater velocity is assumed to have little spatial or temporal variability.
48 FIGURE 3.1 Schematic illustration of the retardation concept: the ideal laboratory case and a hypothetical field case. When K`' represents the partitioning relation in Eq. (3.3), the many available analytical and numerical mathematical so- lutions (e.g., Bear, 1972; Fried, 1975; Finder and Gray, 1977) for the advection-dispersion equation for nonreactive contam- inants in saturated homogeneous porous media can be used for predictive purposes. In these models the velocity of the ad- vancing front becomes the effective contaminant velocity, VIR, and the dispersion coefficient becomes the effective dis- persion coefficient, DIR. The retardation relation and the use of the distribution coef- ficient in the advection-dispersion equation were introduced by Higgins (1959) into the literature on contaminant migration in groundwater. The use of the distribution coefficient in stud- ies pertaining to the disposal of radioactive waste became com- mon in the 1960s and 1970s. More recently this approach has been used in studies of the behavior in groundwater of various adsorbed nonradioactive elements, such as transition metals, JOHN A. CHERRY, ROBERT W. GILEHAM, and JAMES F. BARKER LABORATORY REPRESENTATION TRA`:£R INPUT: _ NON REACTIVE CO Into _ r ADSORBED C ~ O _ TRACER ARRIVAL c/c tC ~ tCH I FIELD REPRESENTATION _ IN P UT to T I ME ~ RETARDATION FACTOR = tC /1C OUTPUT PLUG FLOW ,~ ARRIVAL ,, , _ '/NON REACTIVE '/BREAKTHROUGH 1 TRACER ~ ADSORBED TRACER I I /, /, to tc* tC TIME ~ RETARDATION FACTOR = DISTANCE AC / DISTANCE AB 1 r~ ~ , PLUG F LOW C `~ I ~ ' PC)s1TIoN /CO ~ Y~ O . , A B heavy metals, metalloids, and trace organic compounds. A1- though the convenience of the approach is beyond dispute, its validity as a means of developing reliable predictions of the behavior of inorganic contaminants in actual groundwater sys- tems is questionable in many situations. The C = fC relation is normally determined in the laboratory by means of batch tests in which a known mass of the geologic medium (in particulate form) is immersed in a solution rep- resenting the leachate or groundwater. The solution contains a specified concentration of the contaminant of interest. After agitation of the liquid-solid mixture for a period of hours or days, the contaminant concentration in solution is determined and, by difference, the concentration adsorbed on the solids is known. When this test is repeated using different concentra- tions of the contaminant in solution, the C = fC relation, which is known as the adsorption isotherm, is obtained. There are many possible functional forms of adsorption iso-
Contaminants in Groundwater: Chemical Processes therms, a large number of which are described by Smith (1970~. However, in studies of trace-level contaminants in geologic media, isotherms from batch tests usually fit closely to a func- tional relation known as the Freundlich isotherm, C = kCa, (3 4) where k and a are empirical coefficients. If a = 1, the isotherm is linear, then k = K``, and Eq. (3.3) is applicable. If a ~ 1, the concentration versus distance profile in the flow direction is narrow and the contaminant mass in solution advances less rapidly than would be the case for linear adsorption. If a < 1, the concentration profile is broad and the contaminant mass in solution advances more rapidly than in the linear case. Much of the earlier literature on distribution coefficients and the retardation equation pertains to cationic radionuclides, such as radioisotopes of Sr and Cs, present in very low concentra- tions. The isotherms for these isotopes at low concentrations are nearly always linear. In recent years batch isotherms have been determined for various nonradioactive elements such as Ag, As, Cd, Cr. Pb, and Se. When present at trace concen- trations under conditions where adsorption rather than pre- cipitation is the controlling mass-transfer process, these ele- ments commonly yield isotherms that in some cases are nonlinear but that can be described by the Freundlich relation. Davidson et al. (1976) presented the results of numerical advection-dis- persion-based simulations of the movement of slugs of contam- inants with different Freundlich isotherms having various val- ues of a. They concluded that serious errors in predictions of migration may occur when a linear isotherm is assumed for contaminants that exist at high concentration. Limitations of the Adsorption-Isotherm Approach Limitations and uncertainties are inherent in the use of the isotherm approach within the framework of the advection-dis- persion equation. The uncertainties exist because of the het- erogeneous nature of geologic deposits, because of geochemical effects, and because of effects caused by the dynamics of the transport process. The primary mechanism of adsorption is commonly ion ex- change, which, for univalent species, can be represented as X+ + AS = A+ + XS (3.5) where X+ is the cationic contaminant in solution, XS is the contaminant in the adsorbed states on the exchange medium designated as S. and A+ is the resident cation initially on the exchange medium. Application of the law of mass action pro- vides KeX tA+ 1~XS]I~X+~tAS], (3.6) where KeX is the equilibrium coefficient for the exchange re- action and the bracketed terms are thermodynamic concentra- tions or activities. Conversions between activities and concen- trations for dissolved species can be made using activity coefficients as described below. For situations when the con- taminant of interest is present in very small concentrations relative to the concentration of the exchangeable cation in the medium and when the exchange between the contaminant (X+ ~ 49 and the exchangeable cation (A+) does not cause a significant change in the activity ratio tA+~/tAS] (designated as r), it is apparent that Kit is a constant, and Kd = Kex/r= (XS)I(X+), (3. 7) where the quantities in parentheses are concentrations. The activity coefficients are neglected. This development of the distribution coefficient provides a convenient basis for identification of one of the major difficulties in the distribution-coefficient approach. Consider a situation where a leachate or spilled liquid enters a groundwater-flow system. The leachate contains contaminant X+, which exists in a zone that is retarded relative to the movement of the rest of the leachate zone. In addition, this zone contains a variety of dissolved constituents including the competitor cation A+, which exists in the leachate at concentrations different than the ambient groundwater. Therefore, each segment of the porous medium contacted by X+ has been contacted previously by the major cations (including A+) in the leachate that travels in advance of the retarded zone that contains X+. Because of this contact, the occupancy of the exchange sites evolves toward a new steady-state condition that is not represented in the normal batch tests used to determine the distribution coefficient. It is common practice in batch tests to use water that has major- ion concentrations made up to represent the ambient ground- water from the field site, or in some cases field samples of ambient groundwater are used. Even if the aqueous solution used in the determination of the distribution coefficient by the batch method has the same composition as the leachate, the condition in the field is different because the advancing zone of contamination continuously supplies cations to the exchange sites. In the normal type of batch test the solids are immersed in the test solution only once. Using exchange theory and a mixing-cell model, Reardon (1981) demonstrated that exchange involving major cations, such as Car+ and Na+, can cause a gradual change in the ratio r as the zone of contamination continuously passes through the porous medium. As can be deduced from Eq. (3.7>, this causes a progressive change in the Kit for X+ in an advancing zone of X+ contamination. In recognition of this difficulty, however, batch tests can be con- ducted in a manner that more closely represents field condi- tions. In some cases additional laboratory tests combined with the use of models such as the one described by Reardon can be used to show that the magnitude of change in r is insignif- icant relative to the accuracy required of the K,, values for the particular predictive task. In some situations, the contact between the porous medium and the less retarded contaminants can cause alteration of the exchange properties of the porous medium as a result of pre- cipitation, dissolution, oxidation, or reduction. For the move- ment of a species controlled by adsorption to be simulated in a realistic manner, it would be necessary to simulate the changes in exchange properties that occur because of all geochemical influences. This is currently beyond the scope of existing models, except for the simplest cases involving exchange of major cat- ions or of trace constituents with exchange properties known for specified or measurable conditions of pH and major dis- solved constituents.
50 Transport models that represent the exchange of major cat- ions and that are based directly on the law of mass action were described by Valocchi et al. (1981), who incorporated exchange theory into a numerical advection-dispersion model, and by Dance and Reardon (1982), who used mixing-cell models, and Schultz and Reardon (1983~. These models have been used successfully in the simulation of the transport of major cations in aquifers into which wastewater or tracer solutions were in- jected. The use of the isotherm approach rests on the premise that isotherms can be determined in the laboratory on samples that are representative of the geologic materials as they exist in the field. It is generally not feasible to collect samples of geologic materials and transfer them to the laboratory without, in some manner, altering their geochemical characteristics. Some of the more obvious processes that are difficult to control or exclude during field sampling and during batch testing include the invasion of oxygen and the degassing of CO2. These processes can alter the redox condition and pH of the samples, with the potential to cause significant changes in the concentrations of the competing cations and to cause alterations in the exchange characteristics of the solid phase. Oxygen invasion and sample drying can cause precipitation of iron and manganese hydrox- ides, which have a strong affinity for adsorption of cationic and anionic trace contaminants. Studies of batch tests (Hajek and Ames, 1968; Routson and Serne, 1972) have shown that the measured Kit can also be influenced by experimental factors such as the ratio of solution to solids used in the tests. Ralyea et al. (1980) recommended procedures for determining K,~ val- ues of radionuclides by batch tests. Although these procedures would tend to minimize the influence of the foregoing effects, the effects cannot be entirely eliminated and will generally contribute to the uncertainty in the predictions generated by models in which K`` values are used. The second inherent type of uncertainty in the isotherm approach to advection-dispersion modeling pertains to the dy- namic conditions that occur in the porous medium during trans- JOHN A. CHERRY, ROBERT W. GILEHAM, and JAMES F. BARKER port. Breakthrough curves that are simulated using the advec- tion-dispersion model with linear isotherms are symmetrical or nearly symmetrical, whereas curves obtained from labora- tory-column experiments are asymmetrical with extended tails. Reynolds (1978) compiled data from several published studies in which tracers with linear batch isotherms were used to obtain breakthrough curves from column experiments. The results were scaled and plotted on the same graph of dimensionless concentration versus dimensionless time (Figure 3.2~. The data for Figure 3.2 represent the work of several investigators and a variety of sorbates and sorbents. All the breakthrough curves are asymmetrical and fit within a single narrow band on the dimensionless graph. This suggests the presence of a common factor that is currently not accounted for in the advection- dispersion formulation. One proposed explanation for the asymmetry of the break- through curves invokes the presence of physical or chemical kinetic effects. For example, Cameron and Klute (1977) and others simulated curves of the type shown in Figure 3.2 by the addition of an empirically derived first-order kinetic term to Eq. (3. 2~. Van Genuchten et al. (1974) developed an equation having a similar form but attributed the kinetic parameters to the rate-dependent migration of the solute into zones of im- mobile water. Reynolds et al. (1982) obtained breakthrough curves for Sr2+ that exhibited the typical asymmetrical form with extended tails even though the velocities used in the tests ranged over an order of magnitude. This apparent velocity independence of the shape of the breakthrough curves casts doubt on the kinetic explanation of the observed tailing. lames and Rubin (1978) showed that the transport equation with a linear-adsorption term accurately predicted the migra- tion of cations when the flow velocity was so slow that molecular diffusion was the dominant influence in the dispersion process. At higher velocities when mechanical mixing was dominant in the dispersion process, the differences between measured and calculated breakthrough curves were similar to that repre- sented in Figure 3.2. The results of James and Rubin at very NORMALIZED BR EAKTHROUGH CURVES 1 00 080 o ' 060 040 0 20 0.00 it.. a NON- REACTIVE SOLUTE DOMAIN I I I! 11 !I i! I REACTIVE SOLUTE DOMAIN Q6 0.8 1.0 1.2 1.4 1.6 EFFECTIVE PORE VOLUMES ( EPV) 1.8 2.0 2.2 2.4 FIGURE 3.2 Breakthrough graphs for nonreactive and adsorbed tracers in column experiments with homogeneous porous media. Each graph represents numerous sets of experimental data from Gillham and Cherry (1982).
Contaminants in Groundwater: Chemical Processes low velocity are consistent with unpublished diffusion data ob- tained at the University of Waterloo, using the same materials as those used by Reynolds et al. (1982~. The evidence suggests that the discrepancy between measured and predicted curves is, in some way, the result of the advection process; however, the manner by which advection causes the discrepancy is not known. The measured curves can be closely matched with simulated curves when a kinetic term is included in advection- dispersion models, but at present the kinetic parameters are determined arbitrarily by curve fitting. An understanding of the processes may lead to a less arbitrary means of specifying values for the parameters, thus improving the predictive re- liability of the models. The asymmetrical shape and tailing of breakthrough curves for adsorbed contaminants have considerable practical impor- tance for predictions of the first arrivals of contaminants and for predictions of the rates and effectiveness of purging of zones of contamination from aquifers by means of pumping-well sys- tems. For contaminants that are hazardous at very low con- centration levels, prediction of the behavior of the front or tail of a contaminant zone (where the contaminant occurs only at extremely low concentration levels) can be more important than predictions of the arrival of the center of mass or mid-level concentrations that are represented by the retardation equation fEq. (3.3~. The fact that laboratory studies provide evidence of a lack of predictive capability for behavior of fronts and tails does not bode well for prediction under field conditions. Some of the uncertainties that are inherent in the use of laboratory-determined distribution coefficients in advection- dispersion models can be avoided if the effects of adsorption are assessed by field tests using tracers that adsorb and others that do not react with the porous medium. Values for the retardation factor (R) are computed from the differences in rates of passage through the test zone in the aquifer. Goodwin and Gillham (1982) conducted miniature field tracer tests to obtain radionuclide retardation factors using a device attached to the head of hollow-stem augers used in boreholes in sand deposits. A more common type of test involves the injection of hundreds or thousands of liters of tracer solution into aquifers. An ex- ample of a field test of this type is described by Ewing (19591. These tests are relatively expensive, and, in order to obtain results in a practical length of time, they are generally only suitable for application in zones that have moderate or high hydraulic conductivity and low retardation factors. Although values for the retardation factor can be estimated from field- tracer tests, the suitability of these values for use in predictions that pertain to spatial or temporal scales much different than those represented by the tracer test is problematic. When advection-dispersion models are used in the analysis of the tracer-test data, apparent nonlinear behavior may necessitate the use of a separate kinetic adsorption term as a means of obtaining a close match between simulations and the test data or the use of nonequilibrium physical/chemical or adsorption parameters within the dispersion term. An example of the latter approach in a field tracer investigation is provided by Pickens et al. (1981~. Whether these approaches provide a reliable basis for prediction remains to be established. Not many detailed comparisons between migrations of ad- 51 sorbed inorganic contaminants observed in the field and pre- dicted migrations based on laboratory batch tests exist in the literature. The few that we are aware of involve cationic radio- nuclides in groundwater. These comparisons show moderately good agreement between observed retardation and simulated retardation based on data from laboratory tests (Ewing, 1959; Jackson and Inch, 1980; Comer, 1981~. In some situations or- der-of-magnitude estimates of contaminant retardation are all that is required for solution of a practical problem, in which case many of the sources of uncertainty in the predictive ap- proach may be unimportant. Precipitation and Solubility Controls The behavior of many deleterious or toxic inorganic contami- nants in groundwater is influenced by chemical precipitation that occurs as the contaminant adjusts to solubility constraints. The prediction of solubility constraints is based on the law of mass action and the associated principles of equilibrium-chem- ical thermodynamics. The objective is to predict the equilib- rium concentration of the contaminant species of interest under the conditions that exist or will exist in the groundwater zone. In the conceptualization of the predictive task, the leachate or waste solution in some initial state enters the groundwater system and adjusts by means of various reactions to an equi- librium state within the groundwater system. Equilibrium is achieved as a result of one or more of the following processes: precipitation, dissolution, oxidation, re- duction, hydrolysis, or complexation. This approach is similar to that described above for contaminants that are adsorbed in that both approaches are based on the assumption of equilib- rium; however, the two approaches are very different. The first describes the advance of the front of the contaminant zone as chemical mass transfer continuously occurs in response to a changing solution-phase concentration, whereas the second de- scribes the equilibrium concentrations that occur after chemical mass transfer in a reaction zone has caused equilibrium to be achieved. According to this conceptualization, chemical mass transfer occurs in the reaction zone, and in front of the reaction zone the contaminants are transported at the equilibrium con- centrations established previously in the reaction zone. This is the case if the chemical nature of the porous medium does not change along the flow path. Although it may be possible to specify the rate of entry of the contaminant solution into the groundwater system, the chemical-equilibrium approach pro- vides no information on the rate of advance of the reaction zone because reaction rates and the availability of reactants are not included in the description of the system. Even if kinetic terms were included in the computational routines, so little is known about the reaction-rate controls in contaminated-groundwater systems that little additional predictive capability would be gained. The computational routines for representation of com- plex equilibrium hydrogeochemical systems have not yet been incorporated in a practical manner in advection-dispersion models. Excessive cost of computer time is currently one of the major limiting factors in the development of this approach as a practical means of prediction. The equilibrium relation for a contaminant species controlled
52 by precipitation or dissolution is specified as xX + bB = yY + cC + dD, (3.8) where X is the inorganic contaminant species in the solution phase; Y is a mineral or solid amorphous compound in which the contaminant species is incorporated by precipitation or from which it is released by dissolution; B. C, and D are other elements or compounds in solution; and x, y, b, c, and d are the stoichiometric mole numbers. From the law of mass action, the equilibrium expression is obtained: X = fC] · fD]/KeqfB], (3.9) where Ken is the equilibrium constant and the quantities within the brackets are chemical activities, which can be converted to concentration using well-established conversion relations (Stumm and Morgan, 19803. If X (in the leachate or spilled liquid) is initially above the equilibrium concentration when it enters the groundwater system, adjustment toward equilibrium will occur by precipitation of mineral or amorphous solids. If X is below the equilibrium concentration, minerals or amor- phous solids that contain X as part of the chemical structure will dissolve when such solids are present in the system. If they are not present, which is usually the case for most con- taminants of interest, the condition of undersaturation will per- sist. For example, if Pb is the contaminant of interest, and if PbCO3 is the solid phase that may control the solubility of Pb, the equilibrium expression would be Kit `, = ~pb2 + ~ ICON - I, (3. 10) where ~CO32-] would depend on the pH and the dissolved inorganic carbon. The total concentration of dissolved Pb would be represented by the sum of the concentrations of Pb-+ and the other species of Pb in solution, which would include com- plexes and hydrolyzed species (see next section). To determine which solid phase would be expected to exert the major sol- ubility control, equilibrium relations for many compounds of Pb would be considered. For reactions involving nearly all minerals or amorphous inorganic compounds of interest, values for the equilibrium constant at 25°C can be obtained from published listings, or they can be computed from values of the standard free energies of reaction in published compilations of thermodynamic data. When a contaminant in a leachate or waste liquid enters the subsurface domain, the chemical reactions cause adjustment of the solution to the new conditions. The reactions are influenced by the minerals that comprise the porous medium, by the degree of mixing that occurs between the contaminant solution and the ambient groundwater, by changes in temperature and pressure, and, in some cases, by microbial action. To predict the equilibrium concentration of a particular trace contaminant by means of Eq. (3.9), it is first necessary to predict the gross chemical composition of the contaminant solution after the dominant reactions in the subsurface domain have proceeded to equilibrium or to some other expected status. The gross chemical composition is represented primarily by the major ions such as Na+, Ca2+, Mg2+, C1-, SO42-, and dissolved inorganic carbon. In many cases K+, Fez+, H2S, NH4+, or JOHN A. CHERRY, ROBERT W. GILLHAM, and JAMES F. BARKER NO3- also occur as important components of the gross chemical composition of the water. The reactions that involve the above constituents normally determine the ionic strength, pH, and redox status of the groundwater. These parameters have an important influence on the solubility of most toxic inorganic contaminants, which usually constitute only a small percentage of the total dissolved solids. In practice, the use of Eq. (3.9) is the last stage in a series of predictions involving numerous chemical reactions with various assumptions regarding the choice and status of the reactions. So little is known about the iden- tities and quantities of reactive minerals in many groundwater systems that considerable speculation is often used in the rep- resentation of the suite of reactions that control the gross water chemistry. Computer codes, such as the one described by Parkhurst et al. (1980), are available for performing the computational tasks in predictions of the gross equilibrium chemistry of ground- water that reacts with various minerals and amorphous solids. The available equilibrium constants and free energy data are generally for 25°C. Extrapolations to the lower temperatures common for most contaminated groundwater are generally ac- complished without large uncertainties. Nearly all of the ther- modynamic data for solid phases is derived for chemically pure manifestations of the minerals, whereas in groundwater im- purities are the norm. When solids precipitate in the ground- water zone, amorphous or poorly crystalline forms commonly exist at first and then slowly undergo conversion to more crys- talline forms. The thermodynamic properties for the less crys- talline forms are not well known. Some reactions for precipi- tation, dissolution, oxidation, or reduction do not proceed quickly to equilibrium. These deficiencies notwithstanding, the chem- ical equilibrium approach for prediction of the concentrations of hazardous inorganic contaminants in groundwater can often provide useful estimates of the maximum concentration levels that can be expected. Estimates of equilibrium concentrations within a factor of 10 or even 100 are often relevant. Hydrolysis and Chemical Speciation The concentrations of toxic inorganic contaminants that are reported in chemical analyses of groundwater normally rep- resent the total concentrations of each element in solution. The concentration limits specified in drinking-water standards also are expressed in terms of total concentrations of each element or ion of interest. In aqueous systems, however, most inorganic contaminants exist in more than one molecular or ionic form. These forms, or species, can have different valences and, there- fore, different mobilities in groundwater owing to different affinities for adsorption and different solubility controls. Knowl- edge of the distribution of species in solution is therefore nec- essary for consideration of the behavior of most inorganic con- taminants in groundwater. The metals and metalloids are particularly prone to formation of a variety of aqueous species. These species form as a result of hydrolysis and complexation. The simple ionic species combine with ligands to form ionic or neutral-charge aqueous complexes. The major inorganic ligands in contaminated groundwater are generally C1-, HCO32-, CO2, and SO42- and in some cases NH3, NO3-, and F-. Even in
Contaminants in Groundwater: Chemical Processes highly contaminated groundwater the metallic elements and metalloids are rarely present at concentrations that exceed sev- eral milligrams per liter, whereas the major ligands are com- monly present at levels of hundreds or thousands of milligrams per liter. The complexation of a small percentage of a major ligand with a contaminant can result in the formation of a complex that represents a large percentage of the total con- taminant concentration in solution. Hydrolysis also influences the species of occurrence of a contaminant in groundwater. For example, dissolved Pb in water is represented by Pb2+ and various complexes and hy- drolyzed species, Pb~total) = Pb2+ + PbCl+ + PbCl + PbSO4 + PbCO3 + ... + Pb(OH)+ + Pb(OH)2 + Pb(OH>3- + .... (3.11) The percentage of the total Pb represented by each ligand complex depends on the equilibrium constant for each species and on the concentration of the available ligand. The hydrolysis species depend on the equilibrium constant and the pH. Com- puter codes such as that described by Ball et al. (1978) are available for speciation calculations for many metallic contam- inants of interest. For some species there is considerable un- certainty in the values that are currently used to represent the equilibrium constants. Nevertheless, attempts at evaluating the species distributions are essential in investigations of the be- havior of most inorganic contaminants in groundwater. If ad- sorption or retardation experiments are conducted in the lab- oratory under conditions where the contaminant of interest exhibits a much different speciation than would occur under field conditions, the laboratory result may have limited appli- cability to field conditions. However, if one of the major species of an element, such as Pb2+ in the example provided by Eq. (3.11), is limited to very low concentrations by adsorption or solubility constraints, the concentrations of complexes will also generally exist at very low concentrations except in situations where nearly all of the contaminant mass exists in the com- plexed form. Oxidation and Reduction Redox processes (i.e., oxidation and reduction) are important because they can cause changes in the mobility of many in- organic contaminants. Of the 16 inorganic constituents that, for regulatory purposes, have recommended or mandatory con- centration limits in drinking-water supplies, 9 of these have more than one possible oxidation state in groundwater. These are As, Cr. Fe, Hg, Mn, Se, U. N. and S and are referred to here as the redox elements. The latter 2 elements occur in various ionic or molecular forms, such as NO3- and SO42-, which are included in the drinking-water regulations. Of the remaining elements listed in drinking-water standards, 4 can be strongly influenced by redox processes even though they have only one valence state in aqueous systems; these are Ag, Cu. Cd, and Zn. The only elements in the drinking-water regulations that are relatively insensitive to the redox condi- tions are Cl, F. Ba, and Ra, although in some systems even Ba and Ra are influenced indirectly by the redox conditions 53 because of reactions with SO~2- and Fe, which are redox de- pendent. The ionic or molecular forms of the redox elements in aqueous systems are commonly deduced from simplified geochemical models of equilibrium. For illustrative purposes, the results of such computations are commonly expressed as pe-pH or Eh- pH diagrams. These diagrams are ubiquitous in the geochem- ical literature. They are used as guides to the redox status of the aqueous system. Eh or pe can be used interchangeably through a direct numerical conversion. Each has a scale that ranges from negative to positive. Positive values indicate con- ditions that are more oxidizing, and negative values indicate conditions that are more reducing. Figure 3.3 illustrates the potential for Eh and pH to control the redox state of metals and metalloids in groundwater. Dia- grams such as these can be used as a conceptual guide to some of the possibilities of element behavior in groundwater. The CFAn ( b) CADMIUM 1 ~ -0.5 1.0 0.5 - O c - I,, 0 _ - 0 5 _ -1.0 l l , , , . . . 1 3 5 7 9 pH ~ ~ , ~ :) -1.0 , 1 1, 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 (C ) CHROMIUM c, (OH) : ~ C'2O' (d ) ARSENIC 1 _ H2AS O4 l I I I 1 1 1 1 1 1 1 1 11 1 3 5 7 9 11 _. pH _ 2O 10 1O ~20 10 pe -10 FIGURE 3.3 Eh-pH diagrams for Pb, Cd, Cr, and As for 25°C and 1 atm [from Longmire et al. (1982) except for Cd]. a, Pb-C-S-O-H system. Activity of dissolved Pb and S at 10-6 m and 1O-3 m, re- spe.ctively. b, Cd-C-S-Si system. Activity of dissolved Cd at 1O-7 m, C = 10-2~ m, S = Io-3~ m, Si = 1O-3~ m. c, Cr-O-H system. Activity of dissolved Cr = 10-6 m. d, As-S-O-H system. Activity of dissolved As and S at 10-6 m and iO-3 m, respectively. pe
54 boundary lines between stability fields on the Eh-pH diagrams represent transition domains between the fields rather than abrupt discontinuities. In fields where solid phases are des- ignated, the element would be expected to be immobile be- cause of solubility constraints. In stability fields where an aqueous species is designated, this species would not be constrained by solubility if the conditions assumed for the preparation of the diagram are applicable. Each of the four elements considered in Figure 3.3 has Eh- pH domains in which they will be immobile owing to insolu- bility, if equilibrium conditions are achieved. Under extremely low redox conditions, Pb and Cd, which have only one oxidation state in groundwater, form insoluble sulfide minerals. At pH levels above about 7 or 8 and at redox conditions that are not so low, these elements form insoluble carbonate minerals. Ar- senic, which has two possible oxidation states in groundwater, has sulfide-mineral insolubility under very low redox conditions and does not form a carbonate mineral under any Eh or pH conditions. Cr. which also has two possible oxidation states in groundwater, forms a relatively insoluble oxide under all con- ditions except at very low pH or high Eh. In the absence of complexing and where solubility constraints are absent, Pb and Cd normally occur as divalent cationic species. Pb and Cd are normally relatively immobile in permeable unfractured groundwater zones. Nearly all common geologic materials have a significant capability for cation adsorption. Of the metals and metalloids that have mandated limits spec- ified in drinking-water standards, the four that may have the greatest potential for being relatively mobile in permeable geo- logic materials under Eh-pH conditions that are common in shallow groundwater are As, Cr. Se, and U. Under oxidizing conditions and normal pH conditions (see Figure 3.3) As and Cr exist as monovalent or divalent anionic species. This is also the case for Se, which has sulfide-mineral insolubility under low Eh conditions and has no severe solubility constraints under oxidizing conditions. In groundwater that has an oxidizing re- dox condition and has appreciable carbonate alkalinity, U in the + 6 oxidation state occurs predominantly as anionic com- plexes such as UO2(CO3~32- and (C03~34-. The mobility of soluble anionic species in groundwater can be limited to some degree because of adsorption. Anion adsorption is most likely to be significant when the geologic materials contain oxides of iron and aluminum. In general, there have been few field studies of the mobility or geochemical behavior of these ele- ments in groundwater. Although Eh-pH relations provide a framework for consid- eration of the behavior of redox-sensitive Elements in ground- water, they do not lead directly to predictions of contaminant mobility in groundwater. To identify the condition of redox stability applicable to a particular contaminant that enters a groundwater system, one must predict the pH, Eh, and major- ion chemistry that will exist in the contaminated zone. Com- puter models for equilibrium water chemistry such as the one described by Parkhurst et al. (1980) can provide predictions of the pH and Eh, if the initial conditions and the mineral and amorphous solids that comprise the porous medium are spec- ified. At present, however, there is little experience with which to judge the predictive capability of such models for real JOHN A. CHERRY, ROBERT W. GILEHAM, and JAMES F. BARKER groundwater systems. The usefulness of these models in many situations is also limited because of the lack of quantitative information on the influence of bacteria on redox processes in groundwater systems. It is known, however, that bacteria have an influence on the rates at which many important redox re- actions proceed. A further complication is that reliable natural or contaminated-groundwater measurements of redox condi- tions are difficult or impossible to make because of disequilib- rium or other factors. In many contaminated-groundwater systems, dispersion can be a major influence on the redox state of the groundwater. Contaminated groundwater at waste-disposal sites commonly has a much lower initial redox state than the ambient ground- water. Dispersion commonly causes a continual mixing of waters that are different in chemical composition and in redox status. As dispersion occurs, the redox and pH conditions may change, and with these and other effects of dispersion various chemical reactions involving mineral and amorphous solids take place. These reactions can cause further changes in pH, the redox condition, the gross water chemistry, and other factors. The problem of determining the influence of dispersion on the chemical mass transfer of contaminants in groundwater is a particularly difficult one at present because the manner in which dispersion influences concentration distributions at the field scale is poorly understood, as indicated by Anderson (Chapter 2 of this volume). Mineral Dissolution and Acid Consumption In some situations the dissolution of minerals has a major in- fluence on contaminant mobility in groundwater. Mineral dis- solution may cause contaminant concentrations to increase if the contaminants of interest are released from the minerals as dissolution occurs, or it may cause the contaminants to be removed from solution by adsorption or precipitation if the dissolution of minerals results in changes in the water chem- istry, which, in turn, cause adsorption or precipitation of other solid phases. Examples of this latter condition have been ob- served in aquifers contaminated by acidic leachate (pH 1.5 to 4) from mill wastes in the uranium mining districts of the west- ern United States and northern Ontario (Taylor, 1980; Morin et al., 1982~. At sites in the western United States where U- mill impoundments contain ponded water or pore water at low pH, the water has exceptionally high concentrations of tran- sition metals, heavy metals, metalloids, and radionuclides, such as 226Ra, 2~0Pb, Huh, and 238U. Except for the high radionuclide concentration, similar conditions exist at many base-metal mines, where acid-leach milling occurs or where tailings become acidic because of the oxidation of pyrite. Monitoring of plumes of contaminated groundwater in shal- low unconfined sandy aquifers at leaky acidic U-tailings im- poundments has established that considerable neutralization commonly occurs as the acidic water moves through the aqui- fers. This causes the front of the low-pH zone to advance at a rate that is retarded relative to the advance of mobile constit- uents such as C1- or SO42-. Neutralization of the acid is at- tributed primarily to the dissolution of carbonate minerals, such as calcite or dolomite, in the aquifers. Using calcite as the
Contaminants in Groundwater: Chemical Processes reactive solid phase, the neutralization process can be repre- sented as CaCO3 + 3H+Ca2+ + HCO3- + H2CO3. (3.12) The main species representing dissolved inorganic carbon have the equilibrium relation Keg = tHCO3- ASH+ ]/tH2Co34' (3.13) where Keq is the equilibrium constant for the dissociation of H2CO3. Field and laboratory studies have established that on reaction with calcite, the pH generally rises above 6 and in some cases above 7. If neutralization causes the pH to rise above 7 nearly all of the dissolved inorganic carbon occurs as HCO3-, and if the pH is below 6 the dominant species is H2CO3. Acid neutralization also occurs as a result of the dis- solution of aluminosilicate minerals, such as feldspars. It can also occur by ion exchange, when H+ competes successfully with other cations for occupancy on exchange sites. In cases where the acid neutralization occurs primarily by the dissolution of calcite and where the pH rises above 7, the rate of advance of the acid front can be represented in a manner similar to that depicted by Eq. (3.3) for equilibrium adsorption of trace contaminants: Va/V = 1/~1 + (IMcaco3IZMH+) = 1/Ra, (3.14) where Va is the rate of advance of the acid front, MCaco3 is the number of moles of calcite and dolomite per unit volume of porous medium, MH+ is the number of moles of H+ in the water per unit volume of porous medium, and Ra is the acid- front retardation factor. Equation (3.14) has the same form as Eq. (3.3), and because of this similarity the acid-front retardation factor has been in- corporated directly into the advection-dispersion equation in the same manner as the retardation factor for adsorption of trace contaminants. The acid-front retardation factor has been used in this manner by Haji-Djafari et al. (1979) and Highland et al. (1981) in numerical advection-dispersion simulations of the movement of acid fronts at two uranium tailings impound- ments in Wyoming. Inherent in this approach is the assumption that the dissolution of calcite or dolomite occurs rapidly so that local equilibrium exists within the porous medium. If acid neu- tralization occurs primarily by dissolution of aluminosilicate minerals, it is much less likely that local equilibrium will occur. Whether this use of the acid-front retardation factor in advec- tion-dispersion models provides realistic shapes for acid-front breakthrough curves cannot be determined from existing field data and has not yet been evaluated by laboratory experiments. Because the concentration of acid in solution is not dependent on the solid-phase acid neutralization capacity, it is unlikely that the shape of breakthrough curves will be closely repre- sented by the model. The arrival of the midpoint of the break- through curve, however, may be adequately represented by the model. BEHAVIOR OF ORGANIC CONTAMINANTS Spurred by the awareness of potential environmental hazards and the development of sophisticated analytical equipment, studies of the occurrence and behavior of organic compounds in contaminated groundwater have been initiated recently. Many organic compounds are of environmental concern in part per billion (ppb) or part per trillion (ppt) quantities. Faced with these problems, research has concentrated on the 120 or so organic compounds designated as priority pollutants by the U. S. Environmental Protection Agency. The major chemical or biochemical processes currently rec- ognized as having a potential to be significant with respect to the occurrence and migration of these compounds in hydro- geologic regimes include sorption, chemical reaction, and bi- ological reaction. Sorption Trace organic solutes, especially those that are nonpolar and relatively insoluble, tend to be sorbed by sediments and soils. For some polar organic solutes, these sorption processes are essentially electrostatic. For less polar organics the process has not been established but appears to involve weak hydrogen bonding and, more importantly, solvation or chemical parti- tioning similar to the distribution of solutes between two im- miscible solvents. This is similar to the partitioning of solutes between oil and aqueous phases and is controlled essentially by solubility. Sorption of organics on mineral surfaces has been docu- mented. Field and laboratory research indicates that the sorp- tion of organic compounds is dominantly onto particulate or- ganic matter in the sediments. This partitioning of the solute between the aqueous phase and solid organic matter generally appears to reach equilibrium rapidly and is reversible; thus it can be introduced into the chemical mass-transfer term of the transport equation as a K`~. The sorption of various nonionic organic solutes at trace con- centrations onto sediments and soils has been shown to follow an essentially lirtear isotherm and to be readily reversible for a broad range of organic solute concentrations (Chiou et al., 1979; Karickhoff et al., 1979; Means et al., 1980; Schwarzen- bach and Westall, 1981~. In these studies, the slopes of the linear isotherms (K,~s), which are referred to by some authors as partition coefficients (<Kps'), were relatively independent of sediment concentration and inorganic aqueous chemistry as well. In mixtures of trace levels of organic solutes, the sorption behavior of each solute was independent of the occurrence of the other solutes in solution. In all cases, increased sorption was matched by increased solid-phase organic carbon content of the solids. When the sorption is "keyed" solely to the organic carbon content of the soil/sediment, the partition coefficient is expressed as KoC' where Koc = Kp/OC, with OC being the weight percent of solid-phase organic carbon. Karickhoff et al. (1979) noted that sorption on silt/clay-sized fractions was greater than the sand-sized fraction. This suggests that surface area and/or the structure of organic matter are significant variables in sorption of organic contaminants. A promising relationship between Koc and structural chem- ical properties of organic compounds has been developed, which permits the estimation of Koc or Kp to within a factor of 10 or better for most nonpoIar organic compounds. This relationship,
56 which involves the use of partition coefficients for mixtures of water and octanol, has been established for soils in which the organic carbon content exceeds 0.1 percent. The development of this approach is reviewed by Hansch and Leo (1979) and others. A linear relationship exists for par- titioning of organic solutes between sedimentary organic matter and groundwater and for partitioning between octanol and water. The octanollwater system can be used as surrogate for the real groundwater system in describing relative partition coeffi- cients. The following relationships have been observed for var- ious organic solute/sediment systems when OC > 0.1 percent: log Koc = 1.00 log Kow0.21 (Karickhoff et al., 1979~; log K`'c = 0.72 log Kow + 0.49 (Schwarzenback and log K = oc Westall, 1981~; 2.00 log K 0.317 (Means et al., 1980~. OW Kow is the octanol/water partition coefficient, values of which can be obtained from Leo et al. (1971) or Hansch and Leo (1979~. The variation in the slope parameter may be due to differences in the sediment/soil organic matter or to differences in the organic solutes investigated. At least three areas of uncertainty must be investigated be- fore these partition coefficients can be used with confidence in predictive models for groundwater systems. One area is the effect of solute competition for sorption sites. Although Kar- ickhoff et al. (1979) found that nonpolar organic solutes sorbed independently at low total solute concentration, it is likely that at some higher organic solute/solid organic matter ratio sorption will not be independent. Another problem may occur in or- ganic-rich groundwaters such as landfill leachates. Trace or- ganic compounds can become associated with high-molecular- weight, dissolved organic matter, often humic or fulvic acids (Schnitzer and Khan, 1972~. These associated organics may not be available for sorption by matrix organic matter. Perhaps organic partition coefficients will have to be defined for a three- phase system: matrix organic matter, aqueous unassociated sol- utes, and aqueous humic associated solutes. As with inorganic sorption, there is a need for further theoretical description of the process and additional field data and field-scale testing for assessment of the applicability of laboratory results to field conditions. The low KoW~KoC relationships described above have been determined for porous geologic materials that have appreciable contents of particulate organic carbon, such as 0.1 wt. % or more. Unfortunately, many very permeable sand or gravel aquifers may have less than this amount of organic matter. It The major mechanisms of chemical transformation of organic is these aquifers in which groundwater velocities are often compounds in aqueous systems are photolysis, oxidation, hy- highest and that have the greatest potential for widespread drolysis, and reduction. In groundwater, photolysis is not sig- contamination by halogenated hydrocarbons. Whether useful nificant. Callahan (1979) compiled and assessed transformations predictive relationships can be developed for sand and gravel affecting priority pollutants in aqueous systems. Only a brief, aquifers with very low contents of organic carbon remains to generalized discussion of the chemical reactions and the state be determined. Although numerous investigations have shown that dissolved organic compounds at trace levels commonly exhibit linear isotherms, some organic compounds that have a potential to cause severe contamination of groundwater characteristically JOHN A. CHERRY, ROBERT W. GILEHAM, and JAMES F. BARKER have nonlinear isotherms. Notable in this regard are polychlo- rinated biphenyls (PCBs), which are considered to have a po- tential to cause adverse health effects if they are present in drinking-water supplies at concentrations as low as a fraction of a part per billion (ppb) level. Relative to the levels at which PCBs are considered to be undesirable in drinking water, the solubility of PCBs is very large and ranges from approximately 50 ppb (Haque et al., 1974) to about 250 ppb, depending on the isomer that is being considered. Haque et al., who con- ducted batch studies of the adsorption of PCBs (Arochlor 1254) on several soils, obtained nonlinear Freundlich isotherms with the value of K,~ ranging from 1.1 to 0.81. Griffen et al. (1978), who conducted batch adsorption studies (Arochlor 1242 and 1254) of three natural soil materials, also obtained nonlinear Freundlich isotherms with K`~ values of between 1.5 and 0.19. These investigators noted that adsorption was favored in ma- terials with higher organic matter and larger surface area. Be- cause of the nonlinearity, it would be inappropriate to use Eqs. (3.2b) or (3.3) in the development of predictions of the mobility of PCBs in groundwater. The batch adsorption studies by Haque et al. and by Griffin et al. were done with solution-phase concentrations in the range of about 5 ppb to several tens or a few hundreds of ppb. These concentration levels may be viewed as trace levels; however, they are close to the solubility limits for PCBs in water. Non- linear isotherms are most likely to occur when the solution- phase concentration approaches the solubility limit. Batch adsorption isotherms for many organic pesticides in a variety of soils are reported in the literature. At concentration levels much below the solubility limits, the isotherms are typ- ically linear, and the distribution coefficients are generally large. Organic pesticides are generally considered to be relatively immobile in soil, and their use on agricultural land is rarely regarded as a significant hazard to groundwater resources. Davidson et al. (1976, 1980), however, have shown that at high concentrations several organic pesticides have very nonlinear Freundlich isotherms and, in some cases, low Ks values. Pes- ticides at high concentrations may occur in groundwater be- cause of leakage of residual pesticide solutions from used con- tainers that are deposited in road ditches, sanitary landf~lls, or dumps. Because millions of these containers are discarded each year, high pesticide concentrations are cause for concern. Chemical Reactions of predictive capability is presented here. Oxidation often requires the presence of 02, but the reaction usually involves free radicals, especially OH., peroxy RO2, alkoxy RO., and singlet oxygen iO2 as the oxidant. The rate of oxidation of an organic compound (OC) can be expressed in
Contaminants in Groundwater: Chemical Processes terms of the concentration of oxidants (e.g., tRO2O],(~O2~) and the individual rate constants (e.g., k, RON: - dfOCJ/dt = fOCI(kRO. PRO.] + kRO2 fRO2] + kHO tHO] + . . . ). Prediction of the rate constants is possible for most organic compounds through the use of compiled, empirical structure- reactivity relationships (Mill, 1980~. Estimates of reaction rates within a factor of 3-5 are possible. However, the free-radical content of groundwaters is essentially unknown. Prediction is then limited to approximations of relative rates of chemical oxidation. Chemical structures most susceptible to oxidation include phenols, aromatic amines, and dienes. Saturated alkyl com- pounds such as alkenes, halogenated alkenes, alcohols, esters, and ketones may not be significantly oxidizable in the ground- water environment. Hydrolysis usually involves the introduction of a hydroxyl (OH) group into an organic compound, usually at a point of unbalanced charge distribution. Hydrolysis often displaces hal- ogens (X): R'COOR + H2O ~ R'COOH + ROH, RX + H2O > ROH + X. Hydrolysis may be catalyzed by acid (H+), base (OH-), or metal ions (My thus, the rate of hydrolysis is pH and metal- ion-concentration dependent. Surface effects may also influ- ence the rate of hydrolysis. Hydrolysis of atrazine and other pesticide derivatives is faster when humic material is present. Wolfe (1980) suggested that such catalytic processes are slow and reasonable predictions of hydrolysis rates at fixed pH could be obtained from structure-reactivity relationships that define kH for the hydrolysis rate relationship: -d[OC]/dt= [OC]kH. Mill (1980) felt that such a prediction could be made within a factor of 2 or 3 for specific chemicals within a family of closely related molecular structures for which rate data was available. It is difficult to generalize relative hydrolysis rates, but the data of Mabey and Mill (1978) indicate that for halogenated hydrocarbons (RX) the hydrolysis rate increases where X is F- Cl-Br, where R changes from primary to secondary to tertiary type, and when allyl or benzyl groups were added. Reductive dehalogenation involves the removal of a halogen atom via an oxidation-reduction reaction. Although this mech- anism is not usually considered in the chemical degradation of organics it may be operative in low-redox state groundwaters. The abiological reaction requires mediators, such as Fe+3 or biological products to accept electrons generated by oxidation of reduced organics and to transfer these electrons to the hal- ogenated organic to bring about dehalogenation. Esaac and Matsumuna (1980) suggested that Eh c 0.35 V is required so that electrons can be made available for dehalogenation. No studies have been made to evaluate this mechanism for organic transformation in groundwaters, but because it is operative 57 under low-Eta conditions, it is a potentially significant process that warrants evaluation. Biological Reactions Enzymatic reactions brought about by the microbes inhabiting the groundwater environment may be the most important mechanism for transformation of organic contaminants. This activity is dominantly bacterial, although yeasts, fungi, and viruses may also be present. We have measured, by direct microscopic counting, 106 bacteria per gram of dry, sandy aqui- fer material in a number of landfill-leachate-contaminated groundwater systems. Although this is less than the bacteria content of fertile agricultural soils (108 - 109 bacteria per gram), it indicates that contaminated-groundwater systems can be mi- crobially active. Biodegradation of a broad range of organic compounds has been demonstrated in laboratory studies of soils, sediments, and waters. Compounds include pesticides, halogenated hy- drocarbons, aromatic hydrocarbons, amines, and alcohols. The broad range of enzymatic activities in actual mixed populations of microbes permits a broad range of enzyme-catalyzed reac- tions (oxidation, reduction, hydrolysis, dehydration). Because the proportion of each species present at any point in space and time is environmentally dependent, predictions of actual organic transformation pathways and rates are all but impos- sible. The difficulty of prediction is illustrated by the biological transformation of many chlorinated hydrocarbons, which were originally thought to be essentially nonbiodegradable (recal- citrant). It was believed that observed biogradation was limited to aerobic bacteria because anaerobic bacteria did not have sufficient energy available for biogradation because of the types of energy-yielding reactions that they could utilize. Studies of chlorinated pesticides, trihalomethanes, and other halogenated hydrocarbons revealed biodegradation in anaerobic environ- ments. In fact, degradation of trihalomethanes was observed only in anaerobic systems (Bouwer et al., 19811. In their review of microbial degradation of organics, Kobayashi and Rittmann (1982) indicated that anaerobic reductive dehalogenation may be important in transforming certain classes of organic com- pounds. Although there is research activity in this area, few data pertinent to groundwater systems are available at present. The mechanisms of biodegradation of synthetic, often hal- ogenated, organic contaminants are not well understood. Mi- crobes that use organic compounds convert these substrates into inorganic products (e.g., CO2, H2S) and into cell constit- uents and often obtain energy for biosynthesis from these re- actions. The populations responsible for such transformations increase in numbers of biomass as a result of the introduction of the organic chemical into the system. This is direct metab- olism. Degradation of many synthetic organic compounds is unlikely by this direct utilization mechanism because the re- quired enzymatic pathways have not been developed by mi- crobes that have not been previously exposed to such organic structures. Microorganisms capable of directly utilizing DDT, 2,4,5-T, and many halogenated hydrocarbons, for example,
58 have yet to be isolated. However, these compounds have been observed to be biodegradable. The mechanism has been termed cometabolism. Microbes apparently utilize other organic sub- strate while performing the transformation of the organic con- taminant. Alexander (1981) points out two environmental consequences of cometabolism: (1) the responsible populations do not increase in number or biomass as a result of the introduction of the organic compound because it is not utilized for biosynthesis; and (2) a compound subject to cometabolism is modified slowly, and products structurally similar to the contaminant accumu- late because the organism does not possess a sufficient array of enzymes to bring about its extensive transformation. Considerable additional research is required before biode- gradation can be adequately predicted. Correlations of organic structure and reactivity need to be improved. Models such as the biofilm-based theory of Rittmann et al. (1980) need to be improved by incorporation of in situ parameters and by eval- uation at a number of field sites. It must be pointed out that microbial processes can also increase groundwater toxicity as well as reduce it. Biodegra- dation of toxic organics such as DDT can produce more toxic intermediate products such as DDE. Also, microbial methy- lation of metals such as mercury and lead has been shown to increase the metals' toxicity to other organisms. Organic Compounds as Complexing Agents Complexation of some inorganic contaminants, such as trace metals, by inorganic ligands (e.g., CO32-, OH-) has been dis- cussed with respect to chemical speciation. Cationic inorganic contaminants may also be complexed by organic ligands. Many such complexes are stable in groundwater. Although most pre- dictive models include inorganic complexes in consideration of aqueous speciation and transport, the effects of organic ligands are usually not included. Models considering the possible in- fluence of simple organic ligands such as NTA, citric acid, and EDTA on the chemical speciation in seawater or freshwater have produced conflicting indications of the importance of these organic ligands. Organic complexation should increase as the concentration of organic ligands increases and is expected to be significant in groundwaters with high dissolved organic car- bon (DOC). The transport of inorganic contaminants as organic complexes has been documented. Means et al. (1978) reported unexpected mobility of 60Co and U from liquid disposal sites at the Oak Ridge National Laboratory. Apparently, 60Co and some U were complexed by EDTA that was a constituent in wastes from decontamination facilities and by dissolved humic substances. The neutral or anionic species so formed were not subject to the expected retardation by adsorption onto the soil. Iron, toxic trace metals, and other radionuclides have been reported to be transported as organic complexes in groundwaters and sur- face waters. Prediction of the extent of organic complexation in contam- inant transport is limited by the lack of knowledge concerning the nature and content of organic ligands in groundwaters and JOHN A. CHERRY, ROBERT W. GILEHAM, and JAMES F. BARKER by the lack of a comprehensive thermochemical data base for complex ligands. A majority of organic ligands are natural, high- molecular-weight, structurally complex fulvic and humic acids. Assuming that these ligands dominate the groundwater DOC and that they have properties similar to analyzed material (Schnitzer and Khan, 1912), the complexing capacity of ground- waters can be calculated as 1 x 10-2 to 5 x 10-2 meq/L for each 1 mg/L of DOC. It is expected that most ofthis complexing capacity is taken by major cations and that only a part is avail- able to trace metals or radionuclides. The potential for signif- icant complexation of trace metals, for example, has been dem- onstrated in high-DOC groundwaters contaminated by landfill leachate (Knox and Jones, 1979~. Prediction of inorganic con- taminant complexation by complex organic ligands has been attempted through models that simplify the organic-metal in- teraction (Sposito, 1981), but the applicability of these models for application to contaminated-groundwater systems has yet to be assessed. FIELD STUDIES OF CONTAMINATED GROUNDWATER Although most of the literature on the chemical behavior of contaminants in groundwater is based on batch and column experiments and on thermodynamic models, insight pertaining to contaminant behavior is also obtained from field investiga- tions of zones of contamination. In favorable circumstances data from existing zones of contamination at waste-disposal or chem- ical spill sites can be used to assess the applicability of predic- tive models. There are 10 inorganic constituents for which maximum con- taminant levels are specified in the National Interim Primary Drinking Water Regulations of the U. S. Environmental Pro- tection Agency (1975~. These are As, Ba, Cd, Cr, F, Pb, Hg, NO3-, Se, and Ag. Except for NO3- and Cr, these constituents are rarely reported as a cause of significant deterioration of groundwater quality. Nitrate is a particularly serious cause of groundwater con- tamination. NO3-, which has a maximum permissible concen- tration of 10 mg/L as N2 in drinking water, is a common cause of groundwater-quality deterioration because nitrogen is a ma- jor component of agricultural fertilizer and of animal and human wastes and because nitrogen commonly occurs as NO3-, which is very soluble and generally unretarded by adsorption. The widespread nature of NO3- sources and the mobility of NO3- cause unconfined aquifers to be particularly susceptible to grad- ual long-term increases in NO3- concentrations. Increases will probably continue because the quantity of NO3- available for entry into the groundwater zone will probably not diminish appreciably in the next few decades. To cause a decrease would require widespread changes in agricultural practice and in the management of animal and human sewage. It is fortunate that in at least some groundwater systems there exists a biogeochemical process, namely denitrification, that tends to ameliorate the influence of NO3- inputs to the groundwater zone. The denitrification process can be expressed
Contaminants in Groundwater: Chemical Processes schematically as NO3- ~ organic carbon ~N2 + CO2 + H2O. According to thermodynamic data, denitrif~cation is expected to occur only in zones that are essentially devoid of dissolved oxygen (Lindsay 19791. However, field studies have established that it will occur in groundwater zones where labile organic matter and denitrifying bacteria exist even if the groundwater contains low but measurable concentrations of dissolved oxygen (Gillham and Cherry, 1978~. Although groundwater from wells in zones where denitrification apparently occurs contains de- tectable quantities of dissolved oxygen, it is likely that the reduction of NO3- takes place in microenvironments in small pores or on grain surfaces where the redox conditions are most suitable. Organic matter is a source of energy and of cell carbon for the bacteria that mediate the reduction process. The loss of NO3 in the groundwater zone due to denitrifi- cation cannot be predicted simply from measurements of the redox condition of groundwater samples. Based on laboratory- column experiments, Doner and McLaren (1976) developed a mathematical expression to describe steady-state NO3 loss in sandy soils due to denitrification. The expression contains sev- eral parameters, including the mass of N utilized per unit bi- omass per unit time for maintenance of the bacterial population, the mass of N utilized per unit biomass in wasted bacterial metabolism, and the mass of organic matter available to the bacteria that mediate the reaction. The work by Doner and McLaren demonstrated the importance of bacteria and organic matter in the denitrification process. It is unlikely, however, that mathematical expressions derived from laboratory exper- iments will be of much use in the development of predictions of denitrification in the groundwater zone. Whether the mi- crobiological conditions that are critical to the denitrification process in the field can be adequately represented in the lab- oratory remains to be determined. The development of field techniques for the in situ measurement of the main rate-de- termining factors is desirable. Of the other nine inorganic contaminants, Cr has probably caused the most degradation of groundwater. Cr is used in many manufacturing processes and is a constituent in sewage sludge. Cr that leaks from waste lagoons or that leaches from industrial landfills and sludge-disposal areas or from soil con- taminated by spills of Cr-rich liquids can pose a hazard to groundwater quality. Shallow aerobic groundwater zones are particularly sus- ceptable to contamination by Cr; + VI) because under oxidizing conditions the stable Cr species (HCrO3 or CrO2-) are rela- tively soluble (Figure 3.3) and undergo little retardation by adsorption in many types of permeable geologic deposits. Perl- mutter and Lieber (1970) and Pinder (1973) have described a major occurrence of Cr; +VI) contamination in a sand aquifer on Long Island, New York. The zone of contamination is ap- proximately 1400 m long, 350 m wide, and 25 m thick. Shallow zones of Cr; + VI) contamination in sand aquifers have recently been one of the main subjects of legal proceedings related to groundwater protection legislation in the state of Michigan. Cr(+ III) is not a cause of significant groundwater contamina- 59 tion because the Cr(+ III) is insoluble in water except at low pH and because it occurs as cationic species that are absorbed. Of the other eight inorganic contaminants, a few have such severe solubility limitations that they are generally immobile in groundwater under normal pH conditions. For example, the concentration levels at which Ba can occur are limited to very low levels by the solubility of BaSO4. Two of the contaminants, Hg and Ag, rarely cause groundwater contamination because they are uncommon constituents in waste materials deposited on land and because their mobility in normal groundwater conditions is probably limited by solubility constraints and ad- sorption. The remaining contaminants As, Cd, F. Pb, and Sehave been reported in only a few areas as causes of severe local groundwater degradation. Some of these contaminants are particularly prone to causing groundwater contamination in situations where the pH is much lower or much higher than the normal range for groundwater. Conditions of extreme pH in groundwater are common in waste materials at metal or uranium mines or in ash disposal areas associated with coal- fired power plants. Se and As, for example, have been reported at exceptionally high concentrations in groundwater at fly-ash disposal sites in North Dakota in zones where pH levels are above 10 (Groenewold et al., 1981~. In contrast, similar inves- tigations at fly-ash disposal sites in southern Ontario, where the pH of the groundwater is between 7 and 9.5, have estab- lished that the concentrations of these elements and the other elements with maximum permissible limits specified in drink- ing-water regulations are low (Dodd et al., 19811. Field investigations of contaminated aquifers at uranium tail- ings impoundments in Wyoming and in northcentral Ontario (Morin et al., 1982) have established that the contaminated zones at low pH (i.e., generally less than about 4.5) invariably have high concentrations of many transition metals, heavy met- als, metalloids, and radionuclides, whereas the neutral pH zones of contamination that exist in advance of the retarded low-pH fronts rarely have any of these constituents at levels above the drinking-water limits. This is the case because pH exerts a dominant influence on the solubility controls and the adsorp- tion of these contaminants. The hydrogeochemical nature of sandy aquifers that receive acidic water from tailings impound- ments can be represented as three main zones: the acidic zone, the neutralization zone, and the neutral-pH zone. In the neu- tralization zone, the hazardous contaminants are transferred from the water phase to the solid phase by precipitation and adsorption. In the prediction of the movement of metals and radionuclides in groundwater systems that are receiving acidic, metal-rich, or radionuclide-rich water, the critical task is the prediction of the advance of the front of the acidic zone. Field investigations, e.g., Morin et al. (1982), have shown that low pH fronts can be greatly retarded because of reactions with porous media, but the development of a methodology to predict neutralization-zone behavior at new sites is in the early stages. A potential cause of groundwater contamination that is an issue of concern in many communities is municipal landfills. Detailed monitoring of zones of contaminated groundwater has taken place at landfills on permeable deposits of sand and gravel on Long Island (Kimmel and Braids, 1980) in Ontario (Cherry,
60 1983), in Delaware (see Chapter 10 of this volume), and at many other locations in North America. The investigations in- dicate that, although major ions such as C1-, HCO3-, Na+, Ca2+, and Mg2+ and minor constituents such as NH4-, Fe, and Mn are mobile, the toxic inorganic constituents generally do not occur at concentration levels above the mandatory drink- ing-water limits. At the Ontario sites, no values above the limits were reported for toxic inorganic constituents. At the Long Island sites only Se occurred in some samples at levels slightly above the limit. At all of these sites, the pH values of the zone of contaminated groundwater were near neutral. The leachate from municipal landfills has high concentrations of dissolved organic compounds. However, despite the poten- tial for mobilization of toxic inorganic compounds by complex- ing with organic compounds, immobility of the heavy metals and metalloids is the rule rather than the exception. In some contaminated groundwater, mineral dissolution oc- curs because the contaminant solution that invades the porous medium causes reduction of Fe or Mn. This is particularly the case when the invasion takes place at shallow depth where Fe and Mn occur in the porous medium as oxides of Fe; + III) and Mn(+IV). When the water in the reduced state encounters the oxides, they become unstable and the Fe and Mn go into solution as Fe(+II) and Mn(+II). Their solubility in the re- duced oxidation states can be high. Oxides of Fe(+III) and Mn(+IV) are natural scavengers of metallic elements in the geochemical environment, and, therefore, their dissolution can cause the release of elements that accumulated in the oxides under natural conditions before the invasion of the contaminant solution. An example of the influence of this source of metals on the chemistry of contaminated water beneath a landfill is described by Suarez and Langmuir (19767. The importance of these oxides as metal scavengers is described by [enne (1968, 1977~. It has only been in recent years that increased concern for groundwater quality and the availability of greatly improved analytical methods for the identification of organic compounds has resulted in appreciable monitoring of toxic or potentially toxic organic contaminants in groundwater. This work is re- vealing widespread contamination of groundwater by organic chemicals, which indicates significant mobility of many of these substances through soil and in the groundwater zone. Some specific examples (Wilson et al., 1981) are occurrences of tri- chloroethane in many groundwaters in the United States and Europe as a result of spills, leaks, and disposal of wastes in soil; occurrence of 1,2-dibromo-3-chloropropane in ground- water in areas in California, where it was applied to soils as a nematocide; contamination of a large body of groundwater by phenol in the vicinity of an accidental spill of this compound in Wisconsin; and the movement into groundwater of 4,4'- methylene bis(2-chloroaniline) from a wastewater lagoon in Michigan. The processes that control the behavior of specific organic contaminants in groundwater have been evaluated at only a few field sites. One such study was conducted during injection of tertiary-treated sewage effluent in a sand aquifer near Palo Alto, California (Roberts et al., 1980~. It was observed that organic trace contaminants under anaerobic conditions were JOHN A. CHERRY, ROBERT W. GILLHAM, and JAMES F. BARKER attenuated to varying degrees during the passage of the treated effluent through the groundwater zone. Of the various low- molecular-weight halogenated organics studied, chlorobenzene was most mobile but traveled at a rate of approximately 1/36th of the rate of nonadsorbed inorganic constituents. Dichloro- benzene isomers and 1,2,4-trichlorobenzene isomers were ap- parently more strongly adsorbed than chlorobenzene. Naph- thalene showed evidence of biodegradation. Another injection study is under way in southcentral Ontario in an aerobic zone in a sand aquifer containing a very low solid-phase organic carbon content (Mackay et al., in press). At this site, 12,000 L of water containing chloride and bromide salt as nonreactive tracers and containing five toxic halogenated hydrocarbons (o- dichlorobenzene, bromoform, carbon tetrachloride, hexa- chloroethane, and tetrachloroethylene) were injected during a 14-h period as a slug into the aquifer. The behavior of the salt tracers and the organic compounds under natural flow condi- tions was monitored for more than a year following the injec- tion. Two of the compounds (carbon tetrachloride and bro- moform) were found to be quite mobile; they traveled at a rate of about two thirds of the groundwater velocity. Tetrachloro- ethylene traveled at a rate of about one third of the groundwater velocity and o-dichlorobenzene and hexachloroethane at a rate of less than a quarter of the groundwater velocity. The first three compounds above were not noticeably biodegraded. There is a possibility that the latter two compounds were biodegraded to some degree, however, at most very slowly. At the above-cited landfills on Long Island and Ontario, landfill-derived, dissolved organic carbon exists throughout the zones of contaminated groundwater. Investigations of the iden- tifiable compounds in this dissolved organic fraction indicated that many toxic or potentially toxic compounds are mobile in these groundwater systems (Reinhard et al., in press). Even with modern analytical techniques, only organic compounds that comprise as much as 5 to 10 percent of the DOC in zones of contaminated groundwater at landfill sites are identifiable. It is expected that toxic organic contaminants are mobile in groundwater at many municipal landfills situated on permeable deposits. The dissolved organic fraction in contaminated groundwater at landfills has a much greater potential to cause severe contamination of groundwater resources than dissolved inorganic contaminants. CHEMICAL REACTIONS AND PERMEABLE MEDIA The degree to which chemical reactions can cause attenuation of contaminants in groundwater can be dependent on the type of permeable media in which the contaminants occur. In this chapter the discussions pertain implicitly to nonindurated po- rous media such as gravel, sand, silt, or clay in which contam- inants are transported by groundwater flow through the pore spaces between the grains or particles that comprise the media. The advection-dispersion theory and the isotherm approach to predictive modeling of the behavior of adsorbed contaminants were developed specifically for this type of medium. In many regions of North America, the occurrence and movement of
Contaminants in Groundwater: Chemical Processes contaminants in other types of permeable media such as frac- tured rock or fractured fine-grained, nonindurated deposits are also important. The effects of chemical processes in these de- posits can be very different from those that occur in porous media. In groundwater systems in fractured crystalline rock in which the rock matrix is relatively nonporous (such as granite, marble, and many other types of igneous and metamorphic rocks), the migrating contaminants contact only the mineral surfaces ex- posed on the fracture walls and the amorphous geochemical weathering or alteration products that exist on these surfaces. Measurement of adsorption parameters for predictive modeling is particularly difficult for this type of medium because the measurements must adequately represent the conditions that occur on the surface of the fractures as they exist in the ground- water zone. The distribution coefficient expressed in the nor- mal manner relative to the mass of solids is inappropriate be- cause rock mass and fracture surface area have no known general relation. For this reason the distribution coefficient for frac- tured crystalline rocks has been defined in terms ofthe effective surface area of reaction in the fractures. There have been few attempts to determine the adsorptive properties of relatively undisturbed fracture surface, and there have been no assess- ments under field conditions of the predictive capabilities of the models using parameters determined by the existing mea- surement techniques. The flow of groundwater in fractured porous media such as weathered deposits of silt or clay or fractured rocks that have considerable intergranular porosity such as shale and porous sandstone commonly occurs through the fractures. Little or no flow takes place in very porous but relatively impermeable matrix. As contaminants are transported through the fracture network, transient chemical-concentration gradients exist be- tween the water in the fractures and the pore water in the matrix. These gradients cause the contaminants to diffuse into the matrix in the frontal part of the contaminant zone and diffuse out of the matrix in the trailing part. Models of the advection, dispersion, and diffusion of contaminants in frac- tured porous media have established that diffusion into the porous matrix can have a strong influence on contaminant be- havior (Tang et al., 1981; Grisak and Pickens, 1981~. The surface area that controls adsorption is generally the surface area con- tacted by the contaminants in the porous matrix, which is many orders of magnitude larger than the surface area of the fracture surfaces. The chemistry of the groundwater in the porous ma- trix can have a dominant effect on chemical mass transfer by precipitation or dissolution. The bacterial population that exists in the porous matrix may be more important than exists on the fracture surfaces. Research pertaining to the chemical and bio- chemical behavior of contaminants in fractured porous media Is In its Infancy. SUMMARY AND CONCLUSIONS Most of what is known about the chemical behavior of contam- 61 behavior of inorganic contaminants in groundwater involve the use of the distribution coefficient, which is incorporated in a simple retardation term into the advection-dispersion equation, and the use of thermodynamics-based chemical equilibrium models. When the contaminant of interest exhibits a linear equilibrium adsorption isotherm, the retardation relation can be used to estimate the relative rate of advance of the mid- concentration position of the front of the contaminant zone. This approach is applicable in situations where contamination emanates from a continuous source. It can also be used to estimate the center of the mass of the contaminant zone in situations where the contamination originated from a distinct temporary source. Field experiments have shown that for trace- level inorganic contaminants controlled by adsorption, such as some cationic radionuclides, the retardation relation, when ap- plied in favorable circumstances, provided estimates of relative velocity within a factor of about 5 or better. The usefulness of this approach for prediction of the relative advance rates of toxic nonradioactive inorganic or toxic organic contaminants in groundwater has not yet been subjected to a comprehensive evaluation. When the retardation relation is incorporated into one-di- mensional advection-dispersion models, simulated break- through curves for column experiments using adsorbed tracers that have linear isotherms do not agree closely with experi- mental data. The experimental breakthrough curves are dis- tinctly asymmetrical, whereas the simulated curves are sym- metrical or nearly symmetrical. Hypotheses to account for this discrepancy have not yet been subjected to comprehensive evaluation. This area of uncertainty and the uncertainties as- sociated with sample disturbance and related geochemical ef- fects do not bode well for predictions of the first arrival of fronts or of the tails of zones of adsorbed toxic contaminants. The maximum concentration levels at which many inorganic contaminants occur in groundwater is controlled by the solu- bility of minerals or other solids. Thermodynamic-based equi- librium models are available for prediction of these maximum concentrations in groundwater in which dissolved organic mat- ter is not a complicating factor. Although the thermodynamic data for many of the solid phases and complexes of relevance are questionable, and although equilibrium is probably not achieved in many situations, these models can provide useful order-of-magnitude estimates of maximum possible concentra- tion levels. The models can be used to assess the possible influences of various geochemical scenarios on the occurrence of inorganic contaminants in groundwater. Although the most advanced equilibrium geochemical models that are currently available have well-developed computational capabilities for complex inorganic aqueous systems, they are static models in that they do not include formal representations of the effects of advection or dispersion. Advanced models of this type have not yet been incorporated into advection-dispersion models, although the effects of advection and dispersion have been approximated by the use of equilibrium models in combination with cell models. The chemical behavior of most toxic inorganic contaminants inants in groundwater pertains to inorganic contaminants. The depends strongly on the redox and pH conditions of the con- two main approaches used for the prediction of the chemical tarninated groundwater. When the pH is very low or very high,
62 some heavy metals or metalloids are commonly mobile. A very low redox condition generally promotes immobility. Field in- vestigations have established that, at neutral pH, metals and metalloids are not commonly mobile, except for Cr; + VI) and Se, which occur in anionic forms and tend to be mobile when oxidizing conditions prevail. The most critical task in the pre- diction of the chemical behavior of most toxic inorganic con- taminants is the prediction of the pH and redox conditions. This necessitates prediction of the gross water chemistry of the migrating zone of contamination. In addition to the influence of geochemical factors, the gross water chemistry is affected by dispersion, which often causes mixing of waters of different pH and redox status as the zone of contamination moves through the groundwater system. The task of predicting the chemical behavior of reactive inorganic contaminants therefore cannot be isolated from the problems inherent in predictions of dis- persion in the groundwater zone. Dissolved organic contaminants in groundwater can be in- fluenced by adsorption, oxidation, hydrolysis, or microbial deg- radation. Laboratory experiments indicate that many trace or- ganic compounds exhibit linear adsorption isotherms and, therefore, may be favorable for transport simulation using ad- vection-dispersion-retardation models. Considerable success has been achieved in estimating, from solubility data, the dis- tribution coefficients for adsorption of halogenated hydrocar- bons by solid organic matter in porous geologic materials that have appreciable organic matter. The other processes that can cause attenuation of dissolved organic compounds are much less amenable to quantification, particularly for anaerobic groundwater where biological transformations are poorly understood. The current paucity of information on the chemical and biochemical behavior of organic compounds in ground- water is a particularly serious liability in the assessment of future changes in groundwater quality because entry of organic compounds to the groundwater zone is now common and be- cause many toxic organic compounds are relatively mobile. REFERENCES Alexander, M. (1981). Biodegradation of chemicals of environmental concern, Science 21 1, 128-132. Bear, J. (1972). Dynamics of Fluids in Porous Media, Elsevier, New York, 764 pp. Ball, J. W., E. A. Jenne, and D. K. Nordstrom (1978). WATEQ2- a computerized chemical model for trace and major element spe- ciation and mineral equilibria of natural waters, in Chemical Mod- eling in Aqueous Systems, E. A. Jenne, ea., ACS Symposium Series 93, American Chemical Society, Washington, D.C., pp. 815-835. Bouwer, E. J., B. E. Rittmann, and P. L. McCarty (1981). Anaerobic degradation of halogenated 1- and 2-carbon organic compounds, En- viron. Sci. Technol. 15, 596-599. Callahan, M. A. (1979). Water-Related Environmental Fate of 129 Priority Pollutants, EPA-4401 4-79-029 a and b (2 volumes), U.S. Environmental Protection Agency, Washington, D.C. Cameron, D. R., and A. Klute (1977). Convective dispersive solute transport with a combined equilibrium and kinetic adsorption model, Water Resour. Res. 13, 183-188. Cherry, J. A. (1983). Occurrence and migration of contaminants in groundwater at municipal landfills on sand aquifers, in Environment JOHN A. CHERRY, ROBERT W. GILEHAM, and JAMES F. BARKER and Solid Waste, C. W. Francis, S. I. Auerbach, and V. A. Jacobs, eds., Butterworths, Boston, Mass., pp. 127-147. 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. Dance, J. T., and E. J. Reardon (1982). Migration of contaminants in groundwater at a landfill: A case study, 4. Cation migration in the dispersion test, J. Hydrol. 63, 109-130. Davidson, J. M., L. T. Ou, and P. S. C. Rao (19761. Behavior of high pesticide concentrations in soil water systems, in Proceedings of the Hazardous Waste Research Symposium, Residual Management by Land Disposal, EPA-60019-76-015, U.S. Environmental Protection Agency, Washington, D.C., pp. 206-212. Davidson, J. M., P. S. C. Rao, and L. T. Ou (1980). Movement and biological degradation of large concentrations of selected pesticides in soils, in Proceedings Sixth Annual Research Symposium, Disposal of Hazardous Waste, EPA-600/9-80-010, U. S. Environmental Pro- tection Agency, Washington, D.C., pp. 93-107. Dodd, D. J. R., A. Golomb, H. T. Chan, and D. Chartier (1981). A comparative field and laboratory study of fly ash leaching charac- teristics, in Proceedings 1st ASTM Symposium on Hazardous Solid Waste Testing, American Society for Testing and Materials, Phila- delphia, Pa. Doner, H. E., and A. D. McLaren (1976). Soil nitrogen transformation: A modeling study, in Environmental Biogeochemistry 1, J. Nriagu, ea., Ann Arbor Science Publishers, Ann Arbor, Mich., pp. 245-258. Esaac, E. G., and F. Matsumuna (1980). Pharmacol. Ther. 9, 1-26. Ewing, B. B. (1959). Field test of the movement of radioactive cations, J. Sanitary Eng. Div., ASCE, 85(SA1), 39-59. Fried, J. J. (1975). Groundwater Pollution, Elsevier, Amsterdam, 330 PP Gillham, R. W., and J. A. Cherry (1978). Field evidence of denitri- fication in shallow groundwater flow systems, in Proceedings, Thir- teenth Canadian Symposium on Water Pollution Research, Mc- Master University, Hamilton, Ontario, Canada, pp. 53-71. Gillham, R. W., and J. A. Cherry (1982). Contaminant transport by groundwater in nonindurated deposits, in Recent Trends in Hydro- geology, T. N. Narisimhan, ea., Geol. Soc. Am. Spec. Publ. 189, Boulder, Colo., pp. 31-62. Gomer, M. D. (1981). Field evaluation of dispersivity and strontium K~ in a sandy aquifer, M.Sc. project, Dept. Earth Sci., University of Waterloo, Ontario, Canada, 86 pp. Goodwin, M. J., and R. W. Gillham (1982). Two devices for in situ measurements of geochemical retardation factors, in Proceedings of the Second International Hydrogeological Conference, G. Ozoray, ea., International Association of Hydrogeologists, Canadian National Chapter, pp. 91-98. Griffen, R., R. Clark, M. Lee, and E. Chian (1978). Disposal and removal of polychlorinated biphenyls in soil, in Proceedings Fourth Annual Research Symposium, Land Disposal of Hazardous Wastes, EPA-60019-78-016, U.S. Environmental Protection Agency, Wash- ington, D. C., pp. 169-181. Grisak, G. E., and J. F. Pickens (1981). An analytical solution for solute transport through fractured media with matrix diEusion, J. Hydrol. 52, 47-57. Groenewold, G. H., J. A. Cherry, O. E. Manz, H. A. Gullicks, D. J. Hasset, and B. Rehm (1981). Potential effects on groundwater of fly ash and FGD waste disposal in lignite surface mine pits in North Dakota, in Proceedings Symposium on Flue Gas Desulfurization, U.S. Environmental Protection Agency, Washington, D.C. Hajek, B. F., and L. L. Ames, Jr. (1968). Trace strontium and cesium equilibrium distribution coefficients: Batch and column determi- nations, Battelle Pacific Northwest Laboratories, BNWL-SA-843, Richland, Wash.
Contaminants in Groundwater: Chemical Processes Haji-Djafari, S., P. E. Antommaria, and H. L. Crouse (1979). Atten- uation of radionuclides and toxic elements by in situ soils at a uranium tailings pond in central Wyoming, in Proceedings ASTM Symposium Permeability and Groundwater Contaminant Transport, T. F. Zum- mer and C. O. Riggs, eds., American Society for Testing and Ma- terials, Philadelphia, Pa., pp. 221-242. Hansch, C., and A. Leo (1979). Substituent Constants for Correlation Analysis in Chemistry and Biology, Wiley, New York. Haque, R., D. W. Schmedding, and V. H. Freed (1974). Aqueous solubility, adsorption and vapor behavior of polychlorinated biphenyl Aroclor 1254, Environ. Sci. Technol. 8, 139-142. Higgins, G. H. (1959). Evaluation of the groundwater contamination hazard from underground nuclear explosions, J. Geophys. Res. 64, 1509-1519. Highland, W. R., L. T. Murdock, and E. Kemp (1981). Design and seepage modeling studies of below grade disposal, in Proceedings Fourth Symposium on Uranium Mill Tailings Management, Colorado State University, Fort Collins, pp. 367-388. Jackson, R. E., and K. J. Inch (1980). Hydrogeochemical processes affecting the migration of radionuclides in a fluvial sand aquifer at the Chalk River Nuclear Laboratories, Scientific Series No. 104, National Hydrology Research Institute, Environment Canada, Ot- tawa, 58 pp. James, R. V., and J. Rubin (1978~. Applicability ofthe local equilibrium assumption to transport through soils of solutes affected by ion ex- change, in Proceedings 176th American Chemical Society National Meeting, pp. 225-235. Jenne, E. A. (1968~. Control of Mn, Fe, Ni, Cu. and Zn concentration in soils and waters, significant role of hydrous Mn and Fe oxides, in Trace Inorganics in Water, ACS Adv. Chem. Ser. 73, American Chemical Society, Washington, D.C., pp. 337-387. Jenne, E. A. (1977). Trace element sorption by sediments and soils- sites and processes, in Symposium on Molybdenum in the Environ- ment 2, W. Chappel and K. Petersen, eds., DeLker, New York, pp. 425-523. Karickhoff, S. W., D. S. Brown, and T. A. Scott (1979). Sorption of hydrophobic pollutants on natural sediments, Water Res. 13, 241- 248. Kimmel, G. E., and O. C. Braids (1980). Leachate plumes in ground water from Babylon and Islip landfills, Long Island, New York, U.S. Geol. Surv. Prof. Pap. 1085, 37 pp. Knox, K., and P. H. Jones (1979). Complexation characteristics of sanitary land~ill leachates, Water Res. 13, 839-846. Kobayashi, H., and B. E. Rittmann (1982). Microbial removal of haz- ardous organic compounds, Environ. Sc~. Technol. 16, 170A-183A. Leo, A., C. Hansch, and D. Elkins (1971). Partition coefficients and their uses, Chem. Rev. 71, 525-616. Lindsay, W. L. (1979~. Chemical Equilibria in Soils, Wiley, New York, 449 pp. Longmire, P. A., R. T. Hicks, and D. T. Brookins (1982). Aqueous chemical interactions between groundwater and uranium mirestope backflling: Grants Mineral Belt, New Mexico, Application of Eh- pH diagrams, in Uranium Mill Tail~ngs Management, Organization for Economic Co-Operation and Development, Paris. Mabey, W., and T. Mill (1978). Critical review of hydrolysis of organic compounds in water under environmental conditions, J. Phys. Chem Ref. Data 7, 383-415. MacKay, D. M., J. A. Cherry, D. L. Freyberg, G. D. Hopkins, P. L. McCarty, M. Reinhard, and P. V. Roberts (in press). Imple- mentation of a field experiment on groundwater transport of organic solutes, in Proceedings National Conference on Environmental En- gineering, ASCE, U. of Colorado, Boulder. Means, J. L., D. A. Crerar, and J. O. Duguid (1978). Migration of 63 radioactive wastes: Radionuclide mobilization by complexing agents, Science 200, 1477-1481. Means, J. C., S. G. Wood, J. J. Hassett, and W. L. Banwart (1980). Sorption of polynuclear aromatic hydrocarbons by sediments and soils, Environ. Sci. Technol. 14, 1524-1528. Mill, T. (1980). Data needed to predict the environmental fate of organic chemicals, in Dynamics, Exposure and Hazard Assessment of Toxic Chemicals, R. Haque, ea., Ann Arbor Science Publishers, Ann Arbor, Mich., pp. 297-322. Morin, K. A., J. A. Cherry, T. P. Lim, A. J. Vivyurka (1982). Con- taminant migration in a sand aquifer near an inactive uranium tailings impoundment, Elliot Lake, Ont., J. Can. Geotech. 19, 49-62. Parkhurst, D. L., D. C. Thorstenson, L. N. Plummer (1980). PHREEQE a computer program for geochemical calculations, U.S. Geol. Surv. Waste-Resour. Inv. 80-96, 209 pp. Perlmutter, N. M., and M. Lieber (1970). Dispersal of plating wastes and sewage contaminants in ground water and surface water, south Farmingdale-Massapequa area, Nassau County, New York, U.S. Geol. Surv. Water Supply Pap. 1879-G, 67 pp. Pickens, J. F., R. E. Jackson, and K. J. Inch (1981). Field measurement of distribution coe~icients using a radial-injection dual-tracer test, Water Resour. Res. 17. Pinder, G. F. (1973). A gelerkin-finite element simulation of ground- water contamination on Long Island, New York, Water Resour. Res. 9, 1657-1669. Pinder, G. F., and W. G. Gray (1977). Finite Element Simulation in Surface and Subsurface Hydrology, Academic, New York, 295 pp. Reardon, E. J. (1981). K~'s Can they be used to describe reversible ion sorption reactions in contaminant migration? Ground Water 19, 270-286. Reinhard, M., J. F. Barker, and N. L. Goodman (in press). Occurrence and distribution of organic chemicals in two landfill leachate plumes, Environ. Sci. Technol. Relyea, J. F., R. J. Serne, and D. Rai (1980). Methods for determining radionuclide retardation factors: Status report, Battelle Pacific Northwest Laboratory, PNL-3349/UC-70, Richland, Wash. Reynolds, W. D. (1978). Column studies of strontium and cesium transport through a granular geologic porous medium, M. Sc. thesis, University of Waterloo, Ontario, Canada, 149 pp. Reynolds, W. D., R. W. Gillham, and J. A. Cherry (1982). Evaluation of distribution coefficients for the prediction of strontium and cesium migration in a uniform sand, J. Can. Geotech. 19(1). Rittmann, B. E., P. L. McCarty, and P. V. Roberts (1980). Trace- organics biodegradation in aquifer recharge, Ground Water 18, 236- 243. Roberts, P. V., P. L. McCarty, M. Reinhard, J. Schreiner (1980~. Organic contaminant behavior during groundwater recharge, J. Water Pollution Control, 161-172. Routson, R. C., and B. J. Serne (1972). Experimental support studies for the PERCOL and transport models, Battelle Pacific Northwest Laboratory, 1719, Richland, Wash. Schnitzer, M., and S. U. Khan (1972). Humic Substances in the En- vironment, Dekker, New York. Schultz, H. D., and E. J. Reardon (1983). A combined mixing cell/ analytical model to describe two-dimensional reactive solute trans- port for unidirectional groundwater flow, Water Resour. Res. 19, 493-502. Schwarzenbach, R. P., and J. Westall (1981). Transport of nonpolar organic compounds from surface water to groundwater. Laboratory sorption studies, Environ. Sci. Technol. 15, 1360-1367. Smith, J. W. (1970~. Chemical Engineering Kinetics, McGraw-Hill, New York. Sposito, G. (1981). Trace metals in contaminated waters, Environ. Sci. Technol. 15, 396-403.
64 Stumm, W., and J. I. Morgan (1980). Aquatic Chemistry, Wiley, New York, 583 pp. Suarez, D. L., and D. Langmuir (1976). Heavy metal relationships in a Pennsylvania soil, Geochim. Cosmochim. Acta 40, 589-598. Tang, D. H., E. O. Frind, and E. A. Sudicky (1981). Contaminant transport in fractured porous media: Analytical solution for a single fracture, Water Resour. Res. 17, 555-564. Taylor, M. J. (1980). Radionuclide movement in seepage and its con- trol, in Proceedings First International Conference on Uranium Mine Waste Disposal (Vancouver, B.C.), Soc. Mining Engineers ofAIME, New York. U.S. Environmental Protection Agency (1975). Water Programs: Na- tional Interim Primary Drinking Regulations, Federal Register 40(248). Valocchi, A. J., R. L. Street, and P. V. Roberts (1981). Transport of JOHN A. CHERRY, ROBERT W. GILEHAM, and JAMES F. BARKER ion-exchanging solutes in groundwater: Chromatographic theory and field simulation, Water Resour. Res. 17, 1517-1527. Van Genuchten, M. T., J. M. Davidson, and P. J. Wierenga (1974). An evaluation of kinetic and equilibrium equations for the prediction of pesticide movement through porous media, Soil Sci. Soc. Am. Proc. 38, 29-35. Wilson, J. T., C. G. Enfold, W. J. Dunlap, R. L. Cosby, D. A. Foster, and L. B. Baskin (1981). Transport and fate of selected organic pol- lutants in a sandy soil, Environ. Sci. Technol. 15. Wolfe, N. L. (1980). Determining the role of hydrolysis in the fate of organics in natural water, in Dynamics, Exposure and Hazard As- sessment of Toxic Chemicals, R. Haque, ea., Ann Arbor Science Publishers, Ann Arbor, Mich., pp. 162-178.