Robert W. Gillham

Sorption nonideality during organic contaminant transport in porous media

In modeling subsurface contaminant transport, sorption is often simplified by assuming instantaneous equilibrium, isotherm linearity, and sorption‐desorption singularity. Data exhibiting behavior that deviates from that predicted by this simple model have been reported, challenging the validity of these assumptions. This nonideal sorptive behavior has been attributed to several different factors, including kinetic sorption reactions, diffusive mass transfer resistances, isotherm nonlinearity, and sorption‐desorption nonsingularity. These factors are examined and their relative impact on contaminant transport is evaluated. For hydrophobic organic compounds, physical nonequilibrium (i.e., rate‐limited mass‐transfer in aggregated or layered systems) and intraorganic matter diffusion (rate‐limited diffusion within the sorbent organic matter matrix) are probably the predominant factors causing nonideality. The relative importance of these factors is scale‐dependent. For smaller scale systems, mass‐transfer lim...

Migration of contaminants in groundwater at a landfill: A case study: 1. Groundwater flow and plume delineation

Abstract A landfill-derived contaminant plume with a maximum width of ∼600 m, a length of ∼700 m and a maximum depth of 20 m in an unconfined sand aquifer was delineated by means of a monitoring network that includes standpipe piezometers, multilevel point-samplers and bundle-piezometers. The extent of detectable contamination caused by the landfill, which began operation in 1940 and which became inactive in 1976, was determined from the distributions of chloride, sulfate and electrical conductance in the sand aquifer, all of which have levels in the leachate that are greatly above those in uncontaminated groundwater. The maximum temperature of groundwater in the zone of contamination beneath the landfill is 12°C, which is 4–5°C above background. The thermal plume in the aquifer extends ∼150 m downgradient from the centre of the landfill. A slight transient water-table mound exists beneath the landfill in the late spring and summer in response to snowmelt and heavy rainfall. Beneath the landfill, the zone of leachate contamination extends to the bottom of the aquifer, apparently because of transient downward components of hydraulic gradient caused by the water-table mound and possibly because of the higher density and lower viscosity of the contaminated water. Values of hydraulic conductivity, which show variations due to local heterogeneity, were obtained from slug tests of piezometers, from pumping tests and from laboratory tests. Because of the inherent uncertainty in the aquifer parameter values, the 38-yr. frontal position of the plume calculated using the Darcy equation with the assumption of plug flow can differ from the observed frontal position by many hundreds of metres, although the use of mean parameter values produces a close agreement. The width of the plume is large relative to the width of the landfill and can be accounted for primarily by variable periods of lateral east- and westward flow caused by changes in water-table configuration due to the variable nature of recharge. Northward from the landfill, the vertical thickness of the plume decreases and the top of the plume is farther below the water table. The thickness of the zone of uncontaminated groundwater above the plume increases northward as the area of recharge of uncontaminated water downflow from the landfill increases. Because dispersion in the vertical direction is weak, there is very little mixing between the overlying zone of recharge water and the contaminant plume. Concentration profiles are irregular beneath and near the landfill and become smooth downgradient where the maximum concentrations are much less than those beneath landfill. These features are attributed to a strong influence of longitudinal dispersion. The plume passes beneath a small shallow stream near the landfill without significant influence on the stream.

The capillary fringe and its effect on water-table response

Abstract Shallow water tables are often observed to respond in a highly disproportionate manner to precipitation events. That is, the magnitude of the response is often much greater than would be predicted on the basis of the specific yield of the geologic material and the amount of rainfall. Physical arguments are presented to show that if the capillary fringe (the zone of tension saturation) extends to ground surface, then the addition of a very small amount of water can result in an immediate and large rise in the water table. The arguments are supported by a field experiment in which the addition of 0.3 cm of water caused the water table to rise 30 cm in 0.25 min., thus demonstrating the large and highly transient influence of the capillary fringe on the position of the water table. The evidence in this study, as well as recent work of others, suggests that the capillary-fringe effect could have a major influence on the process of stream-flow generation, contaminant transport to surface waters and contaminant transport in hydrogeologic regimes having shallow water tables. It is further shown that calculations of groundwater recharge and consumptive use, based on water-table response and an assumed specific yield, could be substantially in error.

AN IN SITU STUDY OF THE OCCURRENCE AND RATE OF DENITRIFICATION IN A SHALLOW UNCONFINED SAND AQUIFER

Denitrification in shallow groundwater flow systems has been inferred from the observation of declining nitrate concentrations below the water table, and corresponding decline in dissolved oxygen concentrations. To provide direct evidence of denitrification within the saturated zone, and to determine the rate of denitrification, an in-situ injection experiment was conducted using a specially designed injection-withdrawal-sampling drive point. Nitrate and a conservative tracer (bromide) were added to natural groundwater and injected at 3 m depth into a shallow, unconfined sand aquifer. The relative changes in concentration were then observed with time. After 356 h the concentration of nitrate-N in the injected water declined from the initial 13 g m−3 to less than 0.1 g m−3. The decrease in nitrate concentration was much greater than the corresponding decrease in the concentration of bromide, confirming a preferential loss of nitrate. The loss of nitrate was preceded by a decline in dissolved oxygen concentration to less than 0.1 g m−3, and coincided with an increase in bicarbonate concentration of 142 g m−3. The production of bicarbonate observed in the injection experiment, 2.59 mmole HCO−3 per mmole nitrate denitrified, agreed with that calculated using an equilibrium geochemical model of the denitrification process. An increase in the population of denitrifying organisms from 1 to 23 organisms per gram of soil was detected in core samples collected at the depth of injection 169 h after the start of the experiment. The measured rate of denitrification ranged from 0.0078 to 0.13 g m−3 NO−3/1bN h−1, and is in reasonable agreement with published rates for saturated soils. The organic carbon source required for denitrification is either dissolved organic carbon or soil organic carbon. Soil organic carbon, at 0.08–0.16% by weight, is adequate to denitrify large amounts of nitrate.

Pyrite oxidation in carbonate-buffered solution: 2. Rate control by oxide coatings

Abstract The kinetic behaviour of pyrite oxidation in the laboratory was studied over a period of about 10,000 hours in reactors through which a carbonate-buffered solution and air (20% O2) flowed continuously. Three grain size fractions were monitored. The concentration of sulphate and the mass of the effluent solution were measured periodically to calculate oxidation rates. The results indicate that the rates of reaction decreased significantly with time. The rates initially exhibited an inverse dependence on grain size (within 400 h) then became more linear with the square of the inverse grain size at later times (after 8000 h) suggesting a surface-layer control of the reaction with time. Surface analysis by X-ray Photoelectron Spectroscopy revealed the presence of ferric oxide on the pyrite surfaces, and ion boring with auger electron spectroscopy indicated a layer thickness on the order of 0.6 microns on the 215 micron grains. The data are represented by a shrinking core model which includes the effects of the surface rate constant plus the diffusive resistance to oxygen transfer through the accumulating reaction layer as oxidation proceeds. The three grain sizes (representing different specific surface areas) exhibited consistent estimates of the surface rate constant (Ks = 3.07 × 10−6mh−1 ± 46%) and the diffusion coefficient for oxygen through the oxide layer (Ds = 1.08 × 10−12m2h−1 ± 30%). The estimated thickness of the oxide layer at the end of the experiment agreed well with the measured value. Oxide accumulation on the pyrite surfaces under neutral pH conditions results in a significant reduction in oxidation rates over time. This behaviour has important implications for the reduction of the rate of release of oxidation products, including hydrogen ions, to environments where sulphide mineral wastes are exposed to the atmosphere.

Field experiments in a fractured clay till: 2. Solute and colloid transport

A field tracer experiment was conducted in a lateral flow field in the weathered and highly fractured upper 6 m of a 40-m-thick clay-rich till plain in southwestern Ontario. In the upper 3 m where fractures are closely spaced ( 5 m/d. Simulations with a discrete fracture/porous matrix flow and transport model, which used the cubic law for flow in fractures, showed that diffusion of the solutes, but not the much larger colloids, into the matrix pore water between fractures is sufficient to cause the observed difference in solute and colloid transport rates. Transport-derived and hydraulic conductivity-derived fracture aperture values were similar, within a factor of 3 and falling mainly within a range of 5–40 μm. In the upper 3 m the solute tracers were evenly distributed between pore water in the fractures and the matrix, and as a result, solute transport can be closely approximated with an equivalent porous medium (EPM) approach. Below this depth, fractures are more widely spaced (0.13 to >1 m) with concentration peaks tending to occur near visible fractures, and solute transport cannot be adequately described with an EPM approach.

Spatial Variability of Strontium Distribution Coefficients and Their Correlation With Hydraulic Conductivity in the Canadian Forces Base Borden Aquifer

Distribution coefficients (Kd), defined as the ratio of the concentration of solute associated with the solids to the concentration in solution, are widely used in the prediction of reactive solute transport. With the advent of stochastic approaches to describe solute transport, there is a need to examine the spatial distribution of Kd, and its correlation with the hydraulic conductivity (K). Distribution coefficients were measured in triplicates for strontium on 1279 subsamples of cores from Canadian Forces Base Borden for which K measurements were available. The Kd values ranged from 4.4 to 29.8 mL/g, with a mean of 9.9 and standard deviation of 2.89 mL/g. The standard error on the triplicate means was 0.95 mL/g or approximately 10% of the mean. The spatial behavior of Kd and K (expressed as In (Kd) and ln (K)) was examined in three directions: horizontally along two orthogonal transects and vertically. The two variables each behaved nearly identically in the two horizontal directions, suggesting horizontal isotropy. Horizontally, ln (Kd) appeared as “white noise” suggesting that the horizontal spacing between cores (1 m) was too large to detect any self-correlation. The distribution coefficient displayed increasing power spectral density with increasing scale in the vertical direction, while In (K) showed these trends in all directions. Depending on the model used, the, correlation lengths obtained by least squares fits of the power spectra varied from 1 to 7.5 m horizontally and from 10 to 30 cm vertically for ln (K); and from 30 cm to 2 m horizontally and from 30 to 70 cm vertically for ln (Kd). The ln (Kd) values showed a significant but very weak negative overall correlation with ln (K) at the 99.95% confidence level. The cross-spectral and coherency analysis showed that the sign and degree of correlation between ln (Kd) and ln (K) depended on the scale and direction considered. The correlations in all directions and at all scales were weak, and could not always be declared significantly different than zero at the 95% confidence level.

Field experiments in a fractured clay till: 1. Hydraulic conductivity and fracture aperture

Field values of horizontal hydraulic conductivity measured in the upper 1.5–5.5 m of a weathered and fractured clay-rich till were strongly influenced by smearing around piezometer intakes, which occurs during augering, and by the physical scale of the measuring device. Values measured in conventional augered piezometers were typically 1–2 orders of magnitude lower than those measured in piezometers designed to reduce smearing. Measurements of hydraulic conductivity in small-scale seepage collectors or piezometers, which typically intersect fewer than 10 fractures, vary over a much greater range, 10−10 to 10−6 m/s, than large-scale values based on infiltration into 5.5-m-deep trenches which intersect thousands of fractures (range 10−7 to 3×10−7 m/s). Values of hydraulic fracture aperture, 1–43 μm, and fracture porosity, 3×10−5 to 2×10−3, were calculated using the cubic law with fracture orientation/distribution measurements and the small-scale hydraulic conductivity measurements. This paper provides the first reliable determination of the magnitude and spatial distribution of hydraulically derived fracture parameters in a clay deposit. The absence of such data has, until now, severely limited the application of quantitative groundwater flow and contaminant transport models in this type of deposit.

Mechanism of oxide film formation on iron in simulating groundwater solutions : Raman spectroscopic studies

Abstract In the use of iron for reductive dehalogenation of chlorinated solvents in ground water, the formation of surface films may cause long-term problems by reducing the activity of the metal surfaces or by causing the clogging of pores. Normal (NRS) and enhanced (SERS) Raman spectroscopy was used to identify the surface film(s) formed during contact of iron particles with solvent-free simulated ground water solutions in column experiments. It was found that anaerobic corrosion of iron leads to the initial formation of ferrous hydroxide at the beginning of the reaction. Independent of the ground water composition, however, the final corrosion product is magnetite. The spontaneous (no current applied) formation of magnetite takes place by a dissolution/precipitation mechanism with the separation of anodic and cathodic sites across the surface film. The cathodic reaction, which takes place at the porous film/solution interface, requires the film to be electron conducting.

An in situ study of the role of surface films on granular iron in the permeable iron wall technology

Permeable walls of granular iron are a new technology developed for the treatment of groundwater contaminated with dissolved chlorinated solvents. Degradation ofthe chlorinated solvents involves a charge transfer process in which they are reductively dechlorinated, and the iron is oxidized. The iron used in the walls is an impure commercial material that is covered with a passive layer of Fe2O3, formed as a result of a high-temperature oxidation process used in the production of iron. Understanding the behaviour of this layer upon contact with solution is important, because Fe2O3 inhibits mechanisms involved in contaminant reduction, including electron transfer and catalytic hydrogenation. Using a glass column specially designed to allow for in situ Raman spectroscopic and open circuit potential measurements, the passive layer of Fe2O3 was observed to be largely removed from the commercial product, Connelly iron, upon contact with Millipore water and with a solution of Millipore water containing 1.5 mg/l trichloroethylene (TCE). It has been previously shown that Fe2O3 is removed from iron surfaces upon contact with solution by an autoreduction reaction; however, prior to this work, the reaction has not been shown to occur on the impure commercial iron products used in permeable granular iron walls. The rate of removal was sufficiently rapid such that the initial presence of Fe2O3 at the iron surface would have no consequence with respect to the performance of an in situ wall. Subsequent to the removal of Fe2O3 layer, magnetite and green rust formed at the iron surface as a result of corrosion in both the Millipore water and the solution containing TCE. The formation of these two species, rather than higher valency iron oxides and oxyhydroxides, is significant for the technology. The former can interfere with contaminant degradation because they inhibit electron transfer and catalytic hydrogenation. Magnetite and green rust, in contrast, will not inhibit the mechanisms involved in contaminant reduction, and hence their formation is beneficial to the long-term performance of the iron material.