In situ chemical oxidation (ISCO) is a leading-edge technology for soil and groundwater remediation, and involves injecting a chemical oxidant (e. g. , permanganate, hydrogen peroxide, or persulfate) into the subsurface to deplete contaminant mass through oxidation. Since the delivery of the chosen oxidant to the target treatment zone must occur in situ, the interaction between the injected oxidant and the aquifer material is a key controlling factor for a successful ISCO application. While many published ISCO studies have focused on the interaction between an oxidant and target contaminants, many questions still remain on the interaction between a potential oxidant and the aquifer material. Through a series of bench-scale experiments with aquifer materials collected from 10 sites throughout North America, the research presented in this thesis provides insight into the interaction between these aquifer materials and two widely used ISCO oxidants; permanganate and hydrogen peroxide. <br /><br /> The investigation into the interaction between aquifer materials and permanganate consisted of three series of bench-scale experiments: (1) long-term batch experiments which were used to investigate permanganate consumption in response to fundamental geochemical properties of the aquifer materials, (2) short-term batch experiments which were designed to yield kinetic data that describe the behavior of permanganate in the presence of various aquifer materials, and (3) column experiments which were used to investigate permanganate transport in a system that mimics the subsurface environment. The long-term experiments which involved more than 180 batch reactors monitored for ~300 days showed that the unproductive permanganate consumption by aquifer materials or natural oxidant demand (NOD) is strongly affected by the initial permanganate concentration, permanganate to solid mass ratio, and the reductive components associated with each aquifer material. This consumption cannot be represented by an instantaneous reaction process but is kinetically controlled by at least a fast and slow reactive component. Accordingly, an empirical expression for permanganate NOD in terms of aquifer material properties, and a hypothetical kinetic model consisting of two reaction components were developed. In addition, a fast and economical permanganate NOD estimation procedure based on a permanganate COD test was developed and tested. The investigation into short-term permanganate consumption (time scale of hours) was based on the theoretical derivation of the stoichiometric reaction of permanganate with bulk aquifer material reductive components, and consisted of excess permanganate mass experiments and excess aquifer material mass experiments. The results demonstrated that permanganate consumption by aquifer materials can be characterized by a very fast reaction on the order of minutes to hours, confirming the existence of the fast reaction component of the hypothetical kinetic model used to describe the long-term permanganate NOD observations. A typical experimental column trial consisted of flushing an aquifer-material packed column with the permanganate source solution until sufficient permanganate breakthrough was observed. The permanganate column results indicated the presence of a fast and slow consumption rate consistent with the long-term batch test data, and an intermediate consumption rate affecting the shape of the rising limb of the breakthrough curve. Finally, a comparison of the experimental results between batch and column systems indicated that permanganate NOD was significantly overestimated by the batch experiments; however, permanganate consumption displayed some similarity between the batch and column systems and hence an empirical expression was developed to predict permanganate consumption in physically representative column systems from batch reactor data. <br /><br /> The interaction between hydrogen peroxide and aquifer materials was also investigated with both batch and column experiments. A series of batch experiments consisting of a mixture of 2% hydrogen peroxide and 15 g of aquifer materials was used to capture the overall hydrogen peroxide behavior in the presence of various aquifer materials. The results indicated that the decomposition of hydrogen peroxide in the presence of various aquifer materials followed a first-order rate law, and was strongly affected by the content of amorphous transition metals (i. e. , Fe and Mn). Although hydrogen peroxide decomposition is related to the total organic carbon (TOC) content of natural aquifer materials, the results from a two-week long exposure to hydrogen peroxide suggests that not all forms of natural organic matter contributed to this decomposition. A multiple linear regression analysis was used to generate predictive relationships to estimate hydrogen peroxide decomposition rate coefficients based on various aquifer material properties. The enhanced stability of hydrogen peroxide was investigated under six scenarios with the addition of chelating reagents. The impact of a new green chelating reagent, S,S'-ethylenediaminedisuccinate (EDDS), on the stability of hydrogen peroxide in the presence of aquifer materials was experimentally examined and compared to that of the traditional and widely used chelating reagent, Ethylenediaminetetraacetic (EDTA). The results demonstrated that EDDS was able to significantly increase the stability of hydrogen peroxide, especially for aquifer materials with low TOC contents and/or high dissolvable Fe and Mn contents. Finally, to complement and expand the findings from the batch experiments, column experiments were conducted with aquifer materials from five representative sites. Each column was flushed with two types of source solutions (with or without EDDS addition) at two flow rates. The column experiments showed that the use of EDDS resulted in an earlier breakthrough and a higher stable concentration of hydrogen peroxide relative to the case without the addition of EDDS. The hydrogen peroxide decomposition rate coefficients generated from the column data were significantly higher than those generated from the batch test data and no correlation between hydrogen peroxide decomposition coefficients obtained from column and batch experiments was observed. Based on the column experimental results, a one-dimensional transport model was also calibrated to capture the hydrogen peroxide breakthrough process. <br /><br /> Data from bench-scale tests are routinely used to support both ISCO design and site screening, and therefore the findings from this study can be used as guidance on the utility of these tests to generate reliable and useful information. In general, the behavior of both permanganate and hydrogen peroxide in the presence of aquifer materials in batch and the column systems clearly indicates that the use of batch test data for ISCO system design is questionable since column experiments are believed to mimic in situ conditions better since column systems provide more realistic aquifer material contact. Thus the scaling relationships developed in this study provide meaningful tools to transfer information obtained from batch systems, which are widely employed in most bench-scale studies, to column systems.

In situ chemical oxidation (ISCO) is a leading-edge technology for soil and groundwater remediation, and involves injecting a chemical oxidant (e. g. , permanganate, hydrogen peroxide, or persulfate) into the subsurface to deplete contaminant mass through oxidation. Since the delivery of the chosen oxidant to the target treatment zone must occur in situ, the interaction between the injected oxidant and the aquifer material is a key controlling factor for a successful ISCO application. While many published ISCO studies have focused on the interaction between an oxidant and target contaminants, many questions still remain on the interaction between a potential oxidant and the aquifer material. Through a series of bench-scale experiments with aquifer materials collected from 10 sites throughout North America, the research presented in this thesis provides insight into the interaction between these aquifer materials and two widely used ISCO oxidants; permanganate and hydrogen peroxide. <br /><br /> The investigation into the interaction between aquifer materials and permanganate consisted of three series of bench-scale experiments: (1) long-term batch experiments which were used to investigate permanganate consumption in response to fundamental geochemical properties of the aquifer materials, (2) short-term batch experiments which were designed to yield kinetic data that describe the behavior of permanganate in the presence of various aquifer materials, and (3) column experiments which were used to investigate permanganate transport in a system that mimics the subsurface environment. The long-term experiments which involved more than 180 batch reactors monitored for ~300 days showed that the unproductive permanganate consumption by aquifer materials or natural oxidant demand (NOD) is strongly affected by the initial permanganate concentration, permanganate to solid mass ratio, and the reductive components associated with each aquifer material. This consumption cannot be represented by an instantaneous reaction process but is kinetically controlled by at least a fast and slow reactive component. Accordingly, an empirical expression for permanganate NOD in terms of aquifer material properties, and a hypothetical kinetic model consisting of two reaction components were developed. In addition, a fast and economical permanganate NOD estimation procedure based on a permanganate COD test was developed and tested. The investigation into short-term permanganate consumption (time scale of hours) was based on the theoretical derivation of the stoichiometric reaction of permanganate with bulk aquifer material reductive components, and consisted of excess permanganate mass experiments and excess aquifer material mass experiments. The results demonstrated that permanganate consumption by aquifer materials can be characterized by a very fast reaction on the order of minutes to hours, confirming the existence of the fast reaction component of the hypothetical kinetic model used to describe the long-term permanganate NOD observations. A typical experimental column trial consisted of flushing an aquifer-material packed column with the permanganate source solution until sufficient permanganate breakthrough was observed. The permanganate column results indicated the presence of a fast and slow consumption rate consistent with the long-term batch test data, and an intermediate consumption rate affecting the shape of the rising limb of the breakthrough curve. Finally, a comparison of the experimental results between batch and column systems indicated that permanganate NOD was significantly overestimated by the batch experiments; however, permanganate consumption displayed some similarity between the batch and column systems and hence an empirical expression was developed to predict permanganate consumption in physically representative column systems from batch reactor data. <br /><br /> The interaction between hydrogen peroxide and aquifer materials was also investigated with both batch and column experiments. A series of batch experiments consisting of a mixture of 2% hydrogen peroxide and 15 g of aquifer materials was used to capture the overall hydrogen peroxide behavior in the presence of various aquifer materials. The results indicated that the decomposition of hydrogen peroxide in the presence of various aquifer materials followed a first-order rate law, and was strongly affected by the content of amorphous transition metals (i. e. , Fe and Mn). Although hydrogen peroxide decomposition is related to the total organic carbon (TOC) content of natural aquifer materials, the results from a two-week long exposure to hydrogen peroxide suggests that not all forms of natural organic matter contributed to this decomposition. A multiple linear regression analysis was used to generate predictive relationships to estimate hydrogen peroxide decomposition rate coefficients based on various aquifer material properties. The enhanced stability of hydrogen peroxide was investigated under six scenarios with the addition of chelating reagents. The impact of a new green chelating reagent, S,S'-ethylenediaminedisuccinate (EDDS), on the stability of hydrogen peroxide in the presence of aquifer materials was experimentally examined and compared to that of the traditional and widely used chelating reagent, Ethylenediaminetetraacetic (EDTA). The results demonstrated that EDDS was able to significantly increase the stability of hydrogen peroxide, especially for aquifer materials with low TOC contents and/or high dissolvable Fe and Mn contents. Finally, to complement and expand the findings from the batch experiments, column experiments were conducted with aquifer materials from five representative sites. Each column was flushed with two types of source solutions (with or without EDDS addition) at two flow rates. The column experiments showed that the use of EDDS resulted in an earlier breakthrough and a higher stable concentration of hydrogen peroxide relative to the case without the addition of EDDS. The hydrogen peroxide decomposition rate coefficients generated from the column data were significantly higher than those generated from the batch test data and no correlation between hydrogen peroxide decomposition coefficients obtained from column and batch experiments was observed. Based on the column experimental results, a one-dimensional transport model was also calibrated to capture the hydrogen peroxide breakthrough process. <br /><br /> Data from bench-scale tests are routinely used to support both ISCO design and site screening, and therefore the findings from this study can be used as guidance on the utility of these tests to generate reliable and useful information. In general, the behavior of both permanganate and hydrogen peroxide in the presence of aquifer materials in batch and the column systems clearly indicates that the use of batch test data for ISCO system design is questionable since column experiments are believed to mimic in situ conditions better since column systems provide more realistic aquifer material contact. Thus the scaling relationships developed in this study provide meaningful tools to transfer information obtained from batch systems, which are widely employed in most bench-scale studies, to column systems.

Ce travail a été consacré à l'étude des sous-produits de désinfection formés lors de la chloration des eaux de piscine publiques. En effet, parallèlement à son action désinfectante, le chlore réagit sur les composés organiques et minéraux introduits dans l'eau des bassins par les baigneurs (urine, sueur, ...) pour former des sous-produits indésirables (chloramines et composés organohalogénés).Des analyses d'urée, principal composé précurseur de chloramines inorganiques, ont été effectuées dans une cinquantaine d'eaux de piscine. Les concentrations mesurées ont été comprises entre 0,14 et 3,67 mg/L (valeur moyenne : 1,08 mg/L ; écart-type : 0,70 mg/L). L'étude de la réactivité du chlore sur l'urée (cinétique et consommation de chlore) effectuée sous différentes conditions expérimentales a mis en évidence une très grande stabilité de l'urée en présence de chlore libre dans les eaux de piscines.Les analyses de sous-produits de chloration ont démontré que l'hydrate de chloral représente l'un des sous-produits de chloration majoritaire avec les acides dichloroacétique et trichloroacétique. Cette étude a aussi permis de déterminer la constante cinétique d'hydrolyse de l'hydrate de chloral dans l'eau, l'influence du pH et de la température sur la vitesse d'hydrolyse, ainsi que les potentiels de formation d'hydrate de chloral à partir de quelques constituants de l'urine.La dernière partie de ce travail a porté sur l'étude de la réactivité de l'ozone sur le chlore et sur des sous-produits de chloration ainsi que sur l'étude de l'incidence d'une préozonation des eaux de piscines (en absence et en présence de chlore libre) sur la formation des sous-produits organohalogénés lors d'une post-chloration. / The aim of this work was to study of the formation of disinfection by-products during the chlorination of public swimming pools water. In parallel to its disinfecting action, chlorine reacts with organic and inorganic compounds introduced into the swimming pool water by bathers (urine, sweat, ...) to form undesirable by-products (chloramines and organohalogenated matters).A statistical study of the presence of urea, the major component of urine and sweat, and the main precursor compound of inorganic chloramines, in public swimming poolwater has been made. Measured concentrations were between 0.14 and 3.67 mg/L (mean value: 1.08 mg/L, s.d: 0.70 mg/L). The study of the reactivity of chlorine with urea (kinetic and chlorine consumption) made under different experimental conditions, showed a very high stability of urea in the presence of free chlorine in the pools water.Analysis of chlorination by-products showed that chloral hydrate, with the dichloro and the trichloroacetic acids, is one of the major chlorination by-products found. This study also determined the kinetic rate constant of hydrolysis of chloral hydrate in water, the influence of the pH and the temperature on the rate of hydrolysis and the potential of chloral hydrate formation from some constituents of the urine.The last part of this work was focused on the study of the reactivity of ozone on chlorine and some disinfection by-products, as well as the study of the impact of preozonation of swimming pools water (in absence and presence of free chlorine) on the formation of organohalogenated by-products during a post-chlorination.