Predicting the Adsorption Capacity of Activated Carbon for Organic Contaminants from Fundamental Adsorbent and Adsorbate Properties

Show full item record

Title: Predicting the Adsorption Capacity of Activated Carbon for Organic Contaminants from Fundamental Adsorbent and Adsorbate Properties
Author: Mezzari, Isabella Anna
Advisors: Detlef R. U. Knappe, Committee Chair
Abstract: In drinking water treatment, activated carbon is an effective barrier against many organic contaminants. Approximately 87,000 chemical substances and mixtures are in use, and activated carbons with a wide range of physical and chemical characteristics are manufactured. To assess the effectiveness of activated carbons for the removal of emerging contaminants, a model that permits the prediction of adsorption isotherms from fundamental adsorbate and adsorbent properties would be useful. Therefore, the objectives of this research were (1) to identify a set of molecular descriptors that define the affinity of organic contaminants for activated carbon and (2) to incorporate the resulting quantitative structure-property relationship (QSPR) into the Polanyi-Dubinin-Manes (PDM) model to predict single-solute adsorption isotherms for emerging contaminants (e.g., endocrine disruptors, pharmaceutically active compounds), and chemical agents (e.g., nerve agents, blister agents, selected biological toxins) on activated carbons with a wide range of physico-chemical properties. For QSPR development, affinity coefficients (β[subscript l,i/benzene]) of individual adsorbates were calculated from adsorption isotherm data collected by the U.S. EPA for 62 neutral organic contaminants. Coefficients describing the affinity of water for the activated carbon surface were estimated from the adsorbent oxygen content using a previously developed correlation. Molecular descriptors of contaminants were calculated with commercially available quantum mechanics software. Among the one-parameter models, a non-linear relationship between β[subscript l,i/benzene] and molecular surface volume, V[subscript s], (β[subscript l,i/benzene] = 3.333x10⁻² V[subscript s][superscript 0.76]) best described the affinity coefficients calculated from the experimental data. The more complex poly-parameter relationship β[subscript l,i/benzene] = 6.146x10⁻¹ * V[subscript s] — 3.123 x10⁻¹ * E[subscript d] + 1.160 * ε[subscript HOMO] + 1.869x10⁻¹ * q- + 1.6163, which includes the additional molecular descriptors dielectric energy (E[subscript d]), highest occupied molecular orbital (HOMO) energy (ε[subscript HOMO]), and electrostatic hydrogen-bond basicity (q-), improved the predictive power of the model as assessed by external validation of the QSPR. External validation tests showed that trichloroethene (TCE) isotherm predictions obtained with the poly-parameter model were in close agreement with TCE isotherm data for activated carbons with a wide range of physicochemical properties. In addition, isotherm predictions for the neutral form of the antimicrobial compound sulfamethoxazole (SMX) were in close agreement with isotherm date collected for three activated carbons. Isotherm predictions were made for 33 compounds in six chemical classes including chemical warfare agents, pharmaceutically active compounds and endocrine disruptors, and pesticides. Predictions showed that the affinity of a contaminant for the activated carbon surface tends to increase with molecular size; however, this trend was not universally true. Notable exceptions included fluorine-containing contaminants such as soman or trifluoromethyl benzene, which were predicted to have a lower affinity for the activated carbon surface than what would be expected on the basis of molecular size alone. Additional predictions were made to compare the adsorption capacities of six activated carbons with a wide range of physicochemical properties for the endocrine disruptor ethinyl estradiol, which is relatively non-polar, and the nerve agent soman, which is relatively polar. Among the six activated carbons, the activated carbon with the highest oxygen content (Picazine, 15.9% oxygen) was predicted to have the lowest adsorption capacities for ethinyl estradiol and soman at low C/Cs values. This result was particularly interesting because the wood-based Picazine activated carbon had the largest BET surface area and micropore volume of the tested activated carbons. Overall, a new procedure was developed that permits the prediction of adsorption isotherms from fundamental adsorbent and adsorbate properties. With respect to adsorbent properties, the procedure requires the development of a reference curve from N₂ and CO₂ isotherm data and knowledge of the adsorbent's oxygen and ash contents. With respect to adsorbate properties, molecular descriptors are required that can be obtained quickly with molecular modeling software packages. Once the affinity coefficient of a target contaminant is calculated from the poly-parameter QSPR and the affinity coefficient of water is calculated from the adsorbent's oxygen content, a composite contaminant/water affinity coefficient can be obtained for the adsorbate/adsorbent pair of interest. This composite contaminant/water affinity coefficient can then be used to scale the reference curve such that the adsorption isotherm for the desired adsorbate/adsorbent pair is obtained. To date, only a limited external validation of the QSPRs developed in this study was conducted. To enhance the confidence in the predictive powers of the developed QSPRs, a more comprehensive external validation test should be conducted. Of particular importance would be to assess the importance of specific interactions between polar organic contaminants and oxygen-containing functional groups on the activated carbon surface.
Date: 2006-01-06
Degree: MS
Discipline: Civil Engineering

Files in this item

Files Size Format View
etd.pdf 1.101Mb PDF View/Open

This item appears in the following Collection(s)

Show full item record