Citation
Competitive multi-solute adsorption on activated carbon

Material Information

Title:
Competitive multi-solute adsorption on activated carbon
Creator:
Ryczak, Robert Stanley, 1948-
Publication Date:
Language:
English
Physical Description:
xii, 276 leaves : ill. ; 28 cm.

Subjects

Subjects / Keywords:
Activated carbon ( jstor )
Adsorption ( jstor )
Carbon ( jstor )
Isotherms ( jstor )
Modeling ( jstor )
Molecules ( jstor )
Phenols ( jstor )
Solubility ( jstor )
Solutes ( jstor )
Surface areas ( jstor )
Adsorption ( fast )
Carbon, Activated ( fast )
Dissertations, Academic -- Environmental Engineering Sciences -- UF
Environmental Engineering Sciences thesis Ph. D
Soil absorption and adsorption ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1985.
Bibliography:
Includes bibliographical references (leaves 265-275).
Additional Physical Form:
Also available online.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Robert Stanley Ryczak.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
029447652 ( ALEPH )
14399147 ( OCLC )

Downloads

This item has the following downloads:


Full Text


















COMPETITIVE MULTI-SOLUTE ADSORPTION
ON ACTIVATED CARBON









By


ROBERT STANLEY RYCZAK



















A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMLEIS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY


UNIVERSITY OF FLORIDA 1985











This work is dedicated to my wife, Sonia. Our degree would never have become a reality without you. You never doubted me or lost confidence in me, even when I lost confidence in myself. You gave unselfishly of yourself to accommodate my every whim, need, and desire. You pushed me when I needed to be pushed and patted me on the back when I needed that, too. You spent many hours washing laboratory glassware and keeping me company through the many long nights in the lab. You spent long hours learning and performing lab analyses to free me to do other work. You found the time, usually late at night, to type my drafts. You helped me with many time-consuming tedious calculations. You were always there for me. There could never be adequate recompense for you for the stresses, the frustrations, the sufferings, and sacrifices you endured over the last three years. I can only begin by saying thank you for your saintly patience, for always being there when I needed you, and for your love.















ACKNOWLEDGEMENTS


The word "acknowledgement" does not begin to repay the contributions made by those who have shared the high and low points with me during the past few years:

My advisor, Dr. John Zoltek, who took me under his wing after the untimely death of my original advisor. Thank you for your constant motivation, your sage advice, your concern for me both professionally and personally, your confidence, your support of my laboratory work, and your friendship.

Dr. Joseph Delfino, thank you for your excellence in teaching, your freely given support during my analytical struggles, and your openness to students.

Dr. Lamar Miller, thank you for your constant optimism and moral support.

Dr. Benjamin Koopman and Dr. Dinesh Shah, thank you for your encouragement and interest in me as a student.

To Ann Hill, my graphic artist, thank you for your

patient and generous support and especially for sharing your home with us.

To Don Carrara, thank you for help with my computer encounters and for your willingness to give of yourself.





iii









To Joe Angley, Lisa Drinkwater, Kevin Topp, John Earle, Sang Ii Lee, Ho Kang, Chan Won Lee, Chang Won Kim, and Rick Mestan, thanks for many good conversations and friendship.

To Barbara Smerage, who converted my drafts to the

final product. Thank you for your long and unconventional hours to accommodate my time schedule and for your confident support.

Finally, to my parents. Thank you for your many years of hard work and sacrifices to give me the tools to become what I am. Thank you for your unwavering confidence, your boundless love, and for always being there.

































iv

















TABLE OF CONTENTS


ACKNOWLEDGEMENTS..........................................iii

LIST OF TABLES ................ ...... ................ vii

LIST OF FIGURES........................................ ix

ABSTRACT................................................... xi

CHAPTERS

1 INTRODUCTION ................................... 1

2 LITERATURE REVIEW ......................... 6

2.1 Introduction......................... 6
2.2 The Nature of Adsorption.............. 7
2.3 The Nature and Characteristics of
Activated Carbon...................... 16
2.4 Effects of Adsorbate Properties on
Activated Carbon Adsorption........... 45
2.5 Effects of Aqueous Solvent System
Properties on Adsorption............. 53
2.6 Descriptive Single-solute Equilibrium
Models................................ 63
2.7 Approaches to Predictive Singlesolute Equilibrium Models ............ 73
2.8 Multi-solute Equilibrium Models...... 81

3 RESEARCH OBJECTIVES ...................... 102

4 MATERIALS AND METHODS........................ 104

4.1 Adsorbent ............................ 104
4.2 Adsorbates ........................... 106
4.3 Solutions............. .... .......... 109
4.4 Aqueous Solubilities.................. 110
4.5 Single-solute Isotherms............... ill
4.6 Single-solute Desorption............. 116
4.7 Analytical Procedures for Singlesolute Analysis...................... 118
4.8 Multi-solute Adsorption Studies...... 124



v










5 RESULTS AND DISCUSSION ..................... 135

5.1 Single-solute Adsorption Studies ..... 135 5.2 Single-solute Desorption Studies..... 152 5.3 Competitive Adsorption............... 159
5.4 Sequential Solute Addition........... 183

6 SUMMARY AND CONCLUSIONS ................... 189

7 ENGINEERING SIGNIFICANCE.................. 191

APPENDICES

A SINGLE SOLUTE DESORPTION, ADSORPTION,
AND SOLUBILITY DATA........................ 194

B SEQUENTIAL AND SIMULTANEOUS SOLUTE
ADDITION DATA FOR THE O-CRESOL/ALLYL
PHENOL SOLUTE PAIR......................... 207

C SEQUENTIAL AND SIMULTANEOUS SOLUTE
ADDITION DATA FOR 2,4-DIHYDROXYACETOPHENONE (DAP)/ALLYL PHENOL SOLUTE PAIR.... 225

D SEQUENTIAL AND SIMULTANEOUS SOLUTE
ADDITION DATA FOR O-CRESOL/2,4-DIHYDROXYACETOPHENONE (DAP) SOLUTE PAIR ............ 242

E MULTI-SOLUTE SIMULTANEOUS ADDITION DATA... 259 REFERENCES....... ......... .......................... 265

BIOGRAPHICAL SKETCH.................................. 276






















vi
















LIST OF TABLES

TABLE PAGE

2-1 Pore volumes and surface areas for
various pore size distributions for
one activated carbon (20, 28)............ 20

2-2 Raw materials that have been used to
produce activated carbons ................ 40

4-1 Characteristics of 12 x 40 mesh Calgon
Filtrasorb 400 granular activated
carbon.................................. 105

4-2 Physical and chemical properties of
organic compounds studied ............... 107

4-3 Wavelengths and concentration ranges for
single-solute UV analysis................ 123

4-4 Method for sequential solute addition.
Compound A is followed by compound B.... 132

5-1 Comparison of Freundlich and Langmuir
adsorption isotherm equations for
single-solute studies ................... 137

5-2 Experimental solubility data ........... 145

5-3 Single-solute adsorption
irreversibility.......................... 156

5-4 Simultaneous addition equilibrium data
for ALP/OC solute pair................... 160

5-5 Simultaneous addition equilibrium data
for ALP/DAP solute pair ................. 162

5-6 Simultaneous addition equilibrium data
for OC/DAP solute pair................... 164

5-7 Predicted and measured solid-phase
equilibrium values for simultaneous
addition of ALP/OC solute pair .......... 168


vii









5-8 Predicted and measured solid-phase
equilibrium values for simultaneous
addition of ALP/DAP solute pair......... 169

5-9 Predicted and measured solid-phase
equilibrium values for simultaneous
addition of OC/DAP solute pair.......... 170

5-10 Competition terms for the improved
Simplified IAS model..................... 173

5-11 Multi-solute simultaneous addition
equilibrium data and predicted values... 178

5-12 Irreversible adsorption for bisolute
systems of differing equilibrium
concentration ratios (mg/L:mg/L). All
values in mg/g......................... 185



































viii
















LIST OF FIGURES

FIGURE PAGE 2-1 Acidic oxygen surface functional groups.. 30 2-2 Basic oxygen surface functional groups... 34 4-1 UV absorbance wavelength scan for
o-cresol................................. 119

4-2 UV absorbance wavelength scan for allyl
phenol.......................................120

4-3 UV absorbance wavelength for 2,4dihydroxyacetophenone.................... 121

4-4 UV absorbance wavelength scan for
a-terpineol............................... 122

4-5 Schematic diagram of continuous flow
carbon contact column system.............. 125

5-1 Single-solute adsorsption isotherms for
o-cresol................................. 138

5-2 Single-solute adsorption isotherms for
allyl phenol............................. 139

5-3 Single-solute adsorption isotherms for
2,4-dihydroxyacetophenone................ 140

5-4 Single-solute adsorption isotherms for
a-terpineol ................................ 141

5-5 Single-solute adsorption isotherms for
all four solutes using <200 mesh carbon.. 143 5-6 Single-solute adsorption isotherms for
all four solutes using 100 x 140 mesh
carbon.................................... 144

5-7 Single-solute adsorption and desorption
isotherms for o-cresol.................... 154



ix










5-8 Single-solute adsorption and desorption
isotherms for allyl phenol ............... 155

5-9 Correlation between the solubility
factor and the competition factor, nl.... 175 5-10 Correlation between the solubility
factor and the competition factor, n2 .... 176













































x
















Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment
of the Requirements for the Degree of Doctor of Philosophy


COMPETITIVE MULTI-SOLUTE ADSORPTION ON ACTIVATED CARBON

by

Robert Stanley Ryczak

December 1985

Chairman: John Zoltek, Jr. Major Department: Environmental Engineering Sciences

Fundamental to the successful development and use of

dynamic, multi-solute carbon adsorption models is an accurate multi-solute equilibrium model. Current models often fail to adequately describe experimental data. Interest in describing and predicting the adsorption of specific organic solutes from dilute aqueous solutions has led to the need to better understand nonideal mechanisms affecting adsorption. An experimental program was conducted to examine the effects of unequal competition and irreversible adsorption in bisolute and multi-solute systems. Carbon particle size significantly affected single-solute adsorption isotherm parameters, which are the most sensitive component of any multi-solute equilibrium model.




xi










Single-solute studies indicated substantial irreversible adsorption for all four of the organic solutes studied. Unequal competition for adsorption sites due to carbon surface heterogeneity and solute affinity for carbon was observed in all bisolute combinations studied. Significant irreversible adsorption was also observed. The application of several current multi-solute equilibrium models failed to provide an accurate description of the data. A recently improved, simplified ideal adsorbed solution (IAS) theory model was superior to all other models tested when used with the unequal competition modification. The model is predictive through correlations between solute solubilities and competition factors. The model has not been expanded beyond bisolute systems. Much more work is needed to test, validate, and expand this improved model to account for unequal competition and irreversible adsorption in the ultimate design process.




















xii















CHAPTER 1
INTRODUCTION


The surface and ground water resources of today's

industrial societies were once a reliable source of supply for domestic and agricultural consumption. Ground water supplies were consistently of good quality and free of insidious constituents that would threaten man's health. The large industrial expansion and rapid growth societies have experienced in the last 50 years have seen a concomitant increase in the variety and volume of chemicals manufactured, used, and disposed of in the environment. Many of these new chemicals were refractory and also capable of producing adverse health effects in man. A large number of these chemicals have exhibited toxic, carcinogenic, mutagenic, or teratogenic properties. Lack of awareness has resulted in the casual disposal of these chemicals into soil and water resources, often through industrial and municipal wastewater streams.

As knowledge of the potential adverse health effects of the chemicals grew, concern over uncontrolled exposure through water supplies also grew. National concern, reflected in a proliferation of federal, state, and local legislation, developed when improved analytical techniques


1








2

revealed the presence of hundreds of these chemicals in water supplies throughout the country. A list of hazardous or potentially hazardous compounds was promulgated by the Environmental Protection Agency (EPA). The EPA also placed maximum acceptable limits on their concentration in drinking waters and wastewaters.

Activated carbon adsorption has evolved as one of the most effective and dependable treatment technologies suited to the task of removing the broad spectrum of dissolved organic compounds from waters and wastewaters. Empirical research has generally demonstrated the advantages and limitations of the many variations of applied adsorption technology. The broad applicability of activated carbon adsorption was most recently emphasized when serious consideration was given to mandating granular activated carbon treatment nationwide as a control strategy for organic precursors in trihalomethane formation.

The rapid development in empirical applications of

activated carbon technology was not accompanied by a similar growth in understanding the fundamental principles underlying the technology. This imbalance is readily noticed in the design approach for activated carbon systems. Laboratory experimentation followed by extensive pilot-scale testing prior to full-scale design is the norm for every activated carbon application in water and wastewater treatment. Little is known of the forces and mechanisms of








3

adsorption of organic compounds from aqueous solution by activated carbon.

Rational approaches to carbon system design have only

just begun to emerge. Simple descriptive models for singlesolute equilibrium adsorption systems have been developed. Some have a solid theoretical basis, and others are empirically formed. Elucidation of the structure and nature of the activated carbon surface has begun but is incomplete. The interactions between solute, solvent, and sorbent still remain to be clearly defined. A few studies of competitive adsorption in bisolute systems have resulted in descriptive, multi-solute equilibrium models. As basic knowledge of the adsorption process increased, descriptive equilibrium modeling has improved. Models are currently being studied which attempt to describe the fundamental thermodynamics of specific adsorption interactions. The crucial role played by equilibrium models lies in the development of predictive dynamic models for full-scale system design and analysis.

The ultimate goal of fundamental research into the

nature of activated carbon adsorptions is the development of a comprehensive, theoretically based design approach that can predict dynamic column behavior for individual solutes in the complex matrices of the water and wastewater streams to be treated. The development of predictive models has generally been limited to simple matrices of one or two compounds. Essential in the development of such dynamic








4

predictive models is the inclusion of a general equilibrium predictive model. Current multi-solute equilibrium models are limited by inadequate descriptions of adsorption mechanisms. The failure of existing models lies in the assumptions of equal competition and reversible adsorption. Such assumptions are based on a thermodynamically inert, homogeneous sorbent surface. Activated carbon is known to be heterogeneous with adsorption characteristics that vary with each commercially available carbon. Such assumptions have been made to simplify model development. However, irreversible adsorption and unequal competition can significantly influence final equilibrium conditions.

The consequences of carbon surface heterogeneity are best observed in operating granular activated carbon adsorbers where influent and effluent concentrations usually vary considerably with time. Irreversible adsorption is especially highlighted in dynamic systems. For example, current predictive models indicate that the introduction of a new solute should not compromise effluent quality. The new solute should be safely adsorbed with no detectable trace in the effluent. However, if significant irreversible adsorption had occurred with previous solutes, then the new compound could pass right through the adsorbent much quicker than predicted. Consequently, the modification of existing models to account for carbon surface heterogeneity should improve predictive accuracy. This research was directed at








5

examining competitive and irreversible adsorption in order to increase fundamental understanding of the adsorptive process, compare existing model performance, and test a model recently modified to include unequal competition.















CHAPTER 2
LITERATURE REVIEW


2.1 Introduction


The literature addressing adsorption and its application to the removal of organic solutes from aqueous systems by activated carbon is widely dispersed internationally among a variety of disciplines. Sources of the literature include many diverse and occasionally uncommon technical journals, a number of books, and many theses and dissertations. No single source of information describing the state-of-the-art of activated carbon adsorption exists. Little communication is evident in the literature among scientists and engineers studying and applying carbon adsorption. Discussion of the theory of adsorption from solution can be found in several basic references (1-7). General information on adsorption of organic compounds from aqueous solutions by activated carbon is presented in several books and review articles (8-19). More specific information can be found through the extensive lists of references in the above works.

This chapter begins with a brief summary of the nature and forces of adsorption. Activated carbon as a successful adsorbent is discussed in Section 2.3. The effects of the

6








7

nature of the adsorbate and solvent on carbon adsorption are reviewed in Sections 2.4 and 2.5. Several single-solute equilibrium models are reviewed in Section 2.6, and various multi-solute equilibrium models are discussed in Section

2.7. The development of predictive, multi-solute adsorption models is the goal of many current research efforts. The evolution of activated carbon adsorption as an effective, dependable technology in the removal of a wide spectrum of organic contaminants from aqueous systems has stimulated efforts to improve the ability to describe and predict adsorption in multi-solute systems. Such ability is necessary in order to improve process modeling and design.


2.2 The Nature of Adsorption


Adsorption has been defined as the accumulation of substances at a surface or interface, primarily as a result of forces active at or near the boundaries of the surface (8, 10, 20). The processes resulting in the adsorption of many organic solutes from aqueous solutions by activated carbon involve complex interactions of these physical and chemical forces. Much research has been conducted in an effort to understand and describe these surface chemistry phenomena. As the chemistry of carbon adsorption is better understood, the activation processes to produce activated carbon can be modified to increase carbon adsorption capacity and specificity. Subsequently, mathematical models describing








8

and predicting adsorption phenomena can be greatly improved. Ultimately, pilot and full-scale plant design procedures can then be modified to optimize the adsorption process for particular applications.

The various forces existing between a solid surface and nearby solute molecules originate in the electromagnetic interactions of nuclei and electrons. The forces of adsorption can be classified into four types: physical, chemical, ion exchange, and specific adsorption. The balancing of these surface forces minimizes the free energy of the surface and results in a state of dynamic equilibrium in which a solute is concentrated on the surface of a solid phase, such as activated carbon.

Physical adsorption results from the action of van der Waals forces and classical electrostatic forces (2, 7, 9). The van der Waals forces include London dispersion and repulsion forces and are always present. Electrostatic interactions involve polarization, dipole and quadrapole interactions and are significant only when the surface is ionic or polarizable. In physical adsorption, the electron distributions of both adsorbent and adsorbate experience some distortion, but there is no exchange or sharing of electrons (20). The net adsorption force, consequently, is weak, and adsorbate molecules are not fixed to a specific site on the surface but can move laterally from site to site or readily desorb. This type of reversible adsorption is










sometimes referred to as "ideal adsorption." The heat of adsorption is low for physical adsorption, on the order of a few hundred calories per mole of adsorbate. Physical adsorption is nonspecific, can occur in monolayer or multilayer configurations, is relatively rapid and reversible, and requires little or no activation energies (7). Specie dissociation does not occur during physical adsorption.

The London dispersion-repulsion forces are relatively weak, electrostatic forces which exert a relatively longrange influence compared to the very short-range, wave mechanical chemical bond force (2). London forces account for the observed general attraction between atoms and molecules that are not charged and do not possess permanent dipole or quadrapole moments. The attractive potential arises from the coupling of instantaneously induced dipoles and quadrapoles. These arise from the fact that even neutral atoms consist of systems of oscillating charges due to the presence of positive nuclei and negative electrons. London theorized a quantum mechanical perturbation of electron distribution caused by the continuous motion of electrons in atoms and molecules that was manifested as rapidly fluctuating temporary dipole and quadrapole moments

(20). These fluctuating moments in solute molecules near a solid surface can affect the electron distribution in surface molecules and induce temporary dipoles and quadrapoles. The surface can also induce such moments in nearby








10

solute molecules as well. The attractive potential arising from dispersion forces then is the sum of dipole-dipole, dipole-quadrapole, and quadrapole-quadrapole interactions

(7). Dipole-dipole interactions are dominant with shortrange effects inversely proportional to the sixth power of distance. Dipole-quadrapole and quadrapole-quadrapole interactions are inversely proportional to the eighth and twelfth powers of distance, respectively. The short-range repulsive force is considered to be inversely proportional to the twelfth power of distance and, consequently, becomes significant only when the electron orbitals of the surface and adsorbate molecules interpenetrate. London forces are sometimes considered long-range forces since oscillating fields are mutually propagated between atoms and molecules. These forces operate between polar and nonpolar molecules and covalent, metallic, and ionic solid surfaces (20).

Classical electrostatic interactions of physical adsorption are more specific than London forces (2, 7). Several different surface-adsorbate interactions are possible. If the solid surface possesses an electrical field, an attractive potential can be developed by the polarization of molecules within the field's influence. These attractive potentials depend on the surface electric field strength and the polarizability of the solute molecule. Solute molecules possessing permanent dipole or quadrapole moments will also interact with the surface field to an extent dependent on











the field characteristics and the magnitude of the permanent moments. The surface and adsorbate molecules can also contain hydrogen bond donor and acceptor groups (1). For example, hydrogen bonding by phenolic protons with oxygen in a surface functional group on activated carbon has been offered as one adsorption mechanism (21). However, if bonding groups on the surface or adsorbate molecule have a strong affinity for water, hydrogen bond adsorption will not be significant (1). Solvation with water will prevent close approach by solute molecules, and solvation of strongly ionized groups may screen hydrogen bonding groups. Such electrostatic interactions are not significant where the solid surface is covalent and a strong enough electrical field does not exist and cannot be induced by polar molecules. However, the surfaces of activated carbon are sufficiently heterogeneous to participate in all of the discussed interactions.

Chemical adsorption, or chemisorption, is a much more specific adsorption mechanism than physical adsorption. Chemisorption occurs when elecrons are exchanged or shared between adsorbate molecules and the surface of the adsorbent so that chemical reaction actually occurs. Consequently, chemisorption exhibits all the characteristics of a chemical bond. These characeristics include irreversibility, substantial activation energy requirements, localized reactions at specific sites on the adsorbent surface, and immobility









12

of adsorbed species. The chemisorption bond is very short and very strong, on the order of 10 to 100 kilocalories per mole compared to the few hundred calories per mole observed in physical adsorption (1). Chemisorption exhibits specificity for adsorbate molecules while physical adsorption does not. Chemisorption results in only monolayer adsorption. Subsequent adsorbed layers may form but result from physical adsorption forces. Physical adsorption, a thermodynamically exothermic process, is usually the significant adsorption type at temperatures well below 00C. Sufficient activation energies may not be available for any appreciable chemisorption to occur. As temperature is increased, physical adsorption decreases since the free energy driving forces decrease. Chemisorption, however, becomes more significant as more energy is available to overcome activation barriers, and reaction rates increase. There will be a temperature, however, beyond which chemisorption decreases, since chemisorption is also a net exothermic process. Chemisorption occurs at the energetically most favored sites with the extent of binding dependent on the energies of adhesion and activation. The covalent bond thus formed can impart an ionic character to the adsorbed molecule and the overall surface. This ionic character depends on the relative electronegativities of the bonding atoms. A dipole moment is usually formed with the comcomitant electric double layer also forming at the surface.








13

The effect of the electric double layer is to reduce the strengths of subsequently formed bonds and to impede mass transfer across the surface film to the interior sites of the carbon surface. Chemisorption alters the energetics of the surface and can affect the adsorption tendency of neighboring sites. When a surface is almost completely covered, lateral interactions among adsorbed species may become important.

Ion exchange adsorption results from charged functional groups on the adsorbent surface attracting oppositely charged adsorbate ions. An originally neutral adsorbent may also gain a surface electrical change by adsorption of an ionic solute molecule (8). This process may result in the release, or exchange, of a different ion which passes into solution. In exchange adsorption the sorbent remains electroneutral or retains the original charge. Hydrolytic adsorption, or the simultaneous adsorption of a hydated proton with adsorption of an anion, may also occur (8). Solids such as silica or carbon, which are normally negatively charged, readily adsorb cationic dyes and surfactants

(1). Sometimes ions carrying the same charge as the surface are adsorbed through van der Waals and other forces. Such ion exchange adsorption is usually independent of temperature, since surface charge varies little with temperature. Ion exchange adsorption does occur by activated carbon but








14

is not thought to contribute as much as physical or chemical adsorption to the net adsorption result.

Specific adsorption describes the result of specific interactions between adsorbate molecules and surface functional groups on the activated carbon. These specific interactions can exhibit a wide range of adsorption energies from low-energy physical adsorption to high-energy chemisorption. An example of specific adsorption is the interaction of aromatic hydroxyl and nitro-substituted compounds with activated carbon (21). Adsorption of phenols occurs through the formation of strong donor-acceptor complexes with surface carbonyl oxygen groups, in addition to hydrogen bonding. The oxygen group dipole moment determines the strength of the donor-acceptor complex formed. The carbonyl oxygens of the carbon surface act as electron donors, and the aromatic rings of the adsorbate molecule act as acceptors. Such an adsorption interaction exhibits reversibility, and the adsorbate molecule adsorbs in a planar orientation. Once the surface carbonyl sites are exhausted, adsorption continues by complexation with the rings of the basal planes of the carbon structure. Nitro group substitution enhances the donor-acceptor interaction by acting as an electron withdrawing group. Enhanced adsorption has also been observed for p-chlorophenol and p-cresol on activated carbon, further demonstrating specific adsorption interactions (22).








15

The preceding discussion described the nature of adsorption in terms of adsorbent-adsorbate affinity. However, adsorbate-solvent interactions also greatly influence the extent of adsorption. The reduction of interfacial tension through adsorption reduces the overall interfacial free energy. Surfactants have long been known to lower interfacial tension significantly (1, 2). The reduction of interfacial free energy results from the loss of solvent-solid interfacial area which lowers that interfacial free energy. The creation of a solute-solid interfacial area and a solute-solvent interfacial area through accumulation of a surfactant on the surface of the solid results in an increase in the system interfacial energy. However, the magnitude of the increase in the system's interfacial energy is less than the magnitude of the decrease, and the net result is an overall decrease in total interfacial free energy. Such an energy balance promotes the concentration of surfactant molecules on a surface or at a interface.

The extent of adsorption is also related to the degree of hydrophobicity of the solute, expressed as solubility. Increasing hydrophobicity decreases solubility and generally increases the extent of adsorption (20). Lundelius' rule states that the greater the solubility, the stronger the solute-solvent bond and the smaller the extent of adsorption. This qualitative relationship generally holds true;








16

however, there are many exceptions. A special case of Lundelius' rule is Traube's rule which contends that adsorption from an aqueous solution increases as a homologous series is ascended. As a homologous series is ascended, the increasing number of carbons and chain length makes the solute less soluble. The removal of more hydrophobic solutes allows more water-water bonds to reform. Conversely, the greater the solubility, the stronger the solute-solvent bond and the smaller the extent of adsorption.

A general outline of the nature of the adsorption

process has been presented. These general principles apply to most adsorption systems with various adsorbents, solvents, and adsorbates. Adsorption from aqueous solution by activated carbon is just one application of this separation process. To describe the nature and extent of adsorption with activated carbon, other relevant factors must be examined, including the nature of activated carbon and the effect on adsorption of varying carbon characteristics, the nature of the solute, and the effects of the solvent, water.


2.3 The Nature and Characteristics of Activated Carbon


2.3.1 Surface Area


Activated carbon refers to a large group of carbonaceous materials prepared in a manner to exhibit extremely








17

large surface areas, high degrees of porosity, and unique surface reactivities. It can be envisioned as a solid foam. Such characteristics make activated carbons an effective adsorbent for a broad spectrum of organic solutes in water and wastewater. Activated carbon has a long history of use in hundreds of separation applications which consume hundred of millions of pounds of the material annually (10, 23). Two general classes of activated carbon include gas-adsorbent carbon and liquid-phase carbon. Gas-adsorbent carbons for chemical recovery or impurity removal are characterized by a rather homogeneous system of 0.5 to 5.0 nm (5
0
to 50 A) radius micropores and surface areas exceeding 2000 m2/g (23). Liquid phase carbons, in contrast, exhibit a wide variety of surface areas, pore sizes, and pore size distributions.

The surface area of activated carbon is essentially

intraparticle. Commercially available carbons for water and wastewater applications have specific surface areas ranging from 480 to 1800 m2/g (24). When specific surface areas exceed 2000 m2/g, the pore sizes become too small (less than

1 nm) to effectively adsorb aqueous organic solutes (12). Surface areas are determined through a variety of methods. Adsorption of nitrogen gas described by the Brunauer, Emmett, and Teller (BET) isotherm equation has been the most common method (7, 10). The surface areas reported by carbon manufacturers are determined by the N2, BET method (-220C).









18

Carbon dioxide adsorption at 250C as described by the Polanyi-Dubinin isotherm equation is thought to be a better measure of the total surface area of microporous carbons, since reversible capillary condensation of N2 at very low relative vapor pressures before monolayer adsorption is complete may yield erroneously high specific surface areas

(25). Other methods to measure surface area in attempts to characterize and compare activated carbons have included adsorption of phenol, p-nitrophenol, methylene blue, rhodamine B, oxalic, and succinic acids from aqueous solutions; adsorption of stearic acid from organic solvents; retention of ethylene glycol against vacuum after wetting with the liquid; and selective adsorption from phenolbenzene and ethylene glycol-water mixtures (26). Since only that portion of surface area and pore volume which is in the pores of proper size will be available for adsorption, the use of surface area measurements or pore volumes is of little value in describing the possible effectiveness of a carbon. Equal weights of carbons prepared from different raw materials by different activation methods may have the same total N2, BET, or CO2, Polanyi-Dubinin surface areas, yet will exhibit very different adsorption behavior in the same aqueous solution of organic compounds. Other carbons with significantly different total surface areas performed equally well when treating domestic wastewaters (24). An adsorption study of dyes with molecular weights from 350 to









19

1370 Daltons indicated that the carbon with the lower surface area performed much better than another carbon with a higher surface area (24). One granular activated carbon's specific surface area as determined by the BET method with N2 was 950 to 1050 m2/g (27). Since this carbon was used in adsorption experiments with nonyl and dinonyl phenol ethoxylates, a better measurement of available surface area was desired. Methylene blue was selected to better represent large organic molecules, and the specific surface area measured was 68.3 m2/g for the same granular carbon. However, when compared to a sodium-montmorillonite clay with a methylene blue specific surface area of 711.5 m2/g, the granular carbon was much more effective in removing the phenol ethoxylates (27). Consequently, it is the nature of the carbon surface and not the magnitude of the surface area which determines its adsorption characteristics.


2.3.2 Pore Sizes and Distribution


Pore sizes and their relative distribution within a carbon particle are two of the most significant factors determining adsorption capacities and kinetics. Pores are formed during the activation process and exhibit a trimodal distribution in activated carbon (8, 20, 23, 28). Table 2-1 lists two definitions of pore size distributions and corresponding pore volumes and surface areas for one activated carbon (20, 28). Approximately 40 percent of the








20



TABLE 2-1. Pore volumes and surface areas for various pore
size distributions for one activated carbon (20, 28).



Pore Type and Diameter Pore volume Specific2Surface Area
(nm) (cm /g) (m /g) Micropore : <2 0.32 Mesospore : 2-60 0.24 Macropore : 60-1000 0.25 Total 0.81 950(N2,BET) Micropore : 3.6-4 0.15-0.50 >95% Transitional: 3.2-200 to 400 0.02-0.10 20-70 Developed
Transitional: 3.2-200 to 400 0.7 200-450 Macropore 1000-4000 0.2-0.8 0.5-2 Note: 1 nm = 10 Angstroms (A).








21

total pore volume consists of micropores. The specific surface area contained in the micropores usually exceeds 95 percent of the total N2, BET surface area (8, 21, 29). The pore pattern in a carbon particle is thought to be arranged such that very few micropores actually lead directly to the outer surface (8, 28). The macropores are thought to lead from the particle surface into the interior. Transitional pores then branch off from the macropores, and the micropores, in turn, branch off from the transitional pores. The consequences of this structural arrangement on adsorption capacities and rates are clear. Exclusion of compounds due to physical size limitations, reduced pore diffusion rates due to stearic hindrances, van der Waals forces, and electrostatic interactions in channels approaching molecular dimensions as well as pore blockage will all significantly influence the extent of site competition. Most of the adsorption occurs in the micropores, unless pore sizes or blocked transitional and micropores physically exclude a molecule. Adsorption can occur on the surfaces of transitional pores, although this size range primarily serves to connect micropores with the macropores. Transitional pore adsorption is thought to contribute to the adsorptiondesorption hysteresis effect which many carbons exhibit

(30). Macropores simply serve as bulk fluid phase conduits where solute mass transport from the bulk phase to the interior surfaces of the carbon occurs. The iodine number








22

has been used to measure micropore surface areas while the molasses number has been used for transitional pore surface area estimates (31).


2.3.3 Carbon Structure


The molecular and crystalline structure of activated

carbon is a major factor in determining the types of surface functional groups which can exist. These surface functional groups contribute significantly to the absorptive behavior of an activated carbon. Although very little work has been done on activated carbons directly, the structure of activated carbons has been described through studies of carbon black. Chemically there appears to be no difference between the two substances. The only physical difference appears to be in internal surface areas (28). Two review articles on activated carbon surface chemistry list the seminal references for the derivation of the structure of activated carbon from studies of graphite and carbon black (32, 33). Three forms of carbon have been defined: diamond, graphite, and amorphous carbon. Amorphous carbon has been renamed microcrystalline carbon to reflect a better understanding of that substance's structure. There are several forms of microcrystalline carbon including activated carbon, carbon blacks, carbon brushes, cokes, and carbon electrodes. It is the starting materials, the preparation process, the very large specific surface area, and surface








23

functional groups that distinguish activated carbon from other microcrystalline forms.

The development of the structural model begins with the structure of ideal graphite (8, 32, 33). Carbon atoms form flat hexagonal ring structures that are joined to form a two-dimensional layer. The combination of covalent and resonant bonds gives each carbon-carbon bond a one-third double bond character. The ideal graphite crystal is completed by the addition of an infinite system of parallel layers of these fused hexagons. The layers are kept parallel and approximately 0.33 nm apart by relatively weak van der Waals forces. Microcrystalline carbon differs from graphite by consisting of small, elementary crystallites instead of homogeneous, parallel, infinite two-dimensional planes. The crystallites are composed of layers of hexagonal carbon rings, but the layers are not perfectly parallel. The planes in a microcrystallite are estimated to be 2-5 nm in diameter (8, 28). About three of these planes form the microcrystallite, giving a height of about 0.9-1.21 nm, although as many as 30 planes can be present. The planes in the microcrystallite are angularly displaced in a random manner and overlap one another irregularly. This irregular geometry is accompanied by the formation of interior vacancies within the microcrystallite that, in activated carbon, are often filled with impurities, some of which are intentionally added to catalyze the carbonization








24

and oxidation steps in the activation process. In addition, the microcrystallites may contain "unorganized" carbon that participates in tetrahedral bonding between tilted layers resulting in layer cross-linkage. The edges of the microcrystallites are high energy sites where noncarbon atoms are always present in differing amounts. These foreign atoms may bond to the edge of the crystallite to form functional groups or may actually be incorporated into the fused ring structure to form "heterocyclic" ring systems.

The microcrystallites are interconnected to form the carbon matrix by tetrahedrally bonded carbon atoms, common hexagonal ring layers, cross-linked carbon hexagons, and functional groups at the edges of the crystallites. The microcrystallites are interconnected to form two types of structures: a graphitizing or a nongraphitizing structure. A graphitizing structure is normally considered to be a soft structure formed at temperatures exceeding 10000C. That is, at high temperatures the unorganized (nonplanar) carbon is consumed in expanding the microcrystallite graphite layers both in diameter and height and orienting the larger crystallites into a more ordered, or organized, graphitelike parallel arrangement. A nongraphitizing carbon does not form a three-dimensional graphitic structure, even at temperatures higher than 30000C (8). This implies considerable cross-linking between microcrystallites. Such strong cross-linking is promoted by the presence of oxygen or by a









25

lack of hydrogen in the raw material. Activated carbon is classified as a nongraphitizing material. There are some structural features of graphite formed by the burning out of unorganized carbon, but throughout the entire volume of material activated carbon is considered to be structurally heterogeneous with a much greater pore volume and surface area than graphitizing carbons. The heterogeneity in the physical structure of activated carbon greatly influences adsorptive capacities and adsorption kinetics.


2.3.4 Surface Chemistry


Adsorptive behavior in activated carbon is determined not only by its porous, physical structure but also by the chemical nature of its surface. The complexities of the carbon surface are reflected in the diversity of adsorption reaction mechanisms for organic compounds in aqueous systems. The presence of edges on the broken graphite planes shifts electron distributions in the graphite ring structures (32). As a result, unpaired electrons and permanent and temporary dipole and quadrapole moments appear, altering adsorption properties for polar or polarizable substances. The presence of noncarbon atoms in the heterocyclic ring structures also alters adsorption properties. Chemically bonded elements on the edges of the microcrystallites, such as hydrogen and oxygen, form reactive chemisorption sites. Inorganic impurities, both on









26

the surface and in microcrystallite vacancies, significantly influence the adsorption of electrolytes and nonelectrolytes from solution. All of the variations in the chemical nature of the carbon surface are, to a great degree, controllable by the choice of raw material as well as activation conditions, especially activation temperatures (32). The details of the surface chemistry of activated carbon are presented in several specific and general review articles which also contain extensive lists of references (8, 9, 17, 21, 25, 26, 28, 32-37).

Most of the carbon surface can be described by two

distinct regions (32). The first region includes the planar surfaces of the microcrystallites. Such portions of the surface appear to be relatively uniform in nature and not likely to contain any attached functional groups. This homogeneity is attributed to the involvement of carbon electrons in covalent bonding with neighboring carbon atoms. Few vacancies in the ring structures are expected, and primarily van der Waals forces would be responsible for most adsorption interactions. However, interactions between the w-bonds of the planar ring structures and certain organic solutes can also result in adsorption. Such a w-bond adsorption mechanism has been suggested for the adsorption of phenol (34). The bulk of the surface area of carbon particle is found in the micropores. Since they are formed in the activation process by the removal of








27

interplanar, unorganized carbon, most of the total particle surface area is thought to be of the planar surface type

(32).

The second major type of surface area is that provided by the edges of the microcrystallites and is quite different from the planar surface areas. A wide variety of functional groups and vacancies characterize this type of surface. Since carbon atoms at the edges of the basal planes are not completely surrounded by other carbon atoms, the electron distributions are not completely balanced by 7-bond interactions, and site reactivity is much higher. These "free valances" at the edges of the microcrystallite planes readily react with, or chemisorb, oxygen, hydrogen, and, to a lesser extent, sulfur, nitrogen, and chlorine (8). The relative amounts of noncarbonaceous material in an activated carbon varies significantly depending on the raw material selected and the activation process used. Oxygen constitutes 2 to 25 percent, by weight, of activated carbon. Hydrogen is present in amounts ranging from 8 to 19 times that of oxygen, on a molar basis, for carbons containing

0.94 to 2.25 percent oxygen by weight (32). Inorganic matter, both originating in the raw material and added during the activation process, can be present in amounts up to 4 to 5 percent, by weight, of the activated carbon (10). Although much of this matter is removed after activation by various acid elution processes, leaving less than 1 percent








28

ash in the commercial product, even small amounts of ash can significantly affect adsorption processes (28).

Surface compounds formed with oxygen are the most important determinants of the carbon surface chemistry because of the high reactivity of such compounds and their significant presence, evidenced by the oxygen weight fraction of carbon. Various surface oxide groups exhibit an acid-base chemistry that determines the classification of a carbon as acidic or basic (38). Acidic carbons are defined as carbons that lower the pH of neutral or alkaline distilled water and that are relatively hydrophilic. Basic carbons raise the pH of neutral or acidic distilled water and are relatively hydrophobic (9). Oxygen is added to carbon in four principal ways (32). Oxygen can be present in the raw material, as a carbohydrate, for example. Oxygen can be chemisorbed on the surface of carbon during the activation process or chemisorbed on the activated carbon at room temperature. Finally, treatment of activated carbon with chemical oxidizing agents can also result in oxygen chemisorption. Oxygen complexes can be removed from the carbon surface by degassing at very high temperatures. The forms of the degassed oxygen are carbon monoxide, carbon dioxide, and water. The form of the removed oxygen can indicate the type of surface oxides that were present on the carbon (35).








29

Acidic surface oxides develop when the carbon is

exposed to oxygen at activation temperatures below 400-5000C or by the action of aqueous oxidizing solutions, such as acidified potassium permanganate, nitric acid, mixtures of nitric and sulfuric acids, hypochlorous acid, sodium hypochlorite, and ammonium persulfate (20, 25, 28, 38). Some of the known acidic surface oxides include quinone carbonyl groups, cyclicperoxide, carboxylic anhydride groups, phenolic hydroxyl groups, normal lactone groups, carboxyl, and fluorescein lactone groups (8, 28, 32, 38, 39). Figure 2-1 illustrates some of these groups as they might appear on the edges of the microcrystallite basal planes. Carboxylic, lactone, and phenolic groups are thought to be the most common (35). Such surface oxides result in a negative surface potential for the carbon.

The specific adsorption mechanisms for each type of

group with various classes of organics have not been clearly defined. A few studies have examined the effect of acidic surface oxides on the adsorption of a limited number of organic solutes in single and bisolute systems (21, 33, 34, 40). The lack of information on the chemical surface characteristics of each carbon studied poses great difficulty in mechanism definition. Aromatic compounds, especially phenol and its derivatives, are the most commonly studied class of organics for the effect of surface oxides on adsorption. An early study with six different commercial











CYCLIC PEROXIDE GROUP CARBOXYLIC ANHYDRIDE GROUP QUINONE CARBONYL GROUP OO C C O--C C---O








RING STRUCTURE OF MICROCRYSTALLITE BASAL PLANES









O H C-0 H H O= O C-0 O

H O O H
H

PHENOLIC GROUP NORMAL LACTONE GROUP CARBOXYLIC GROUP FLUORESCEIN LACTONE GROUP


FIGURE 2-1. Acidic oxygen surface functional groups.








31

activated carbons (41) revealed that acidic oxygen surface groups tended to reduce the capacity of the carbon surfaces for adsorption of metanil yellow from aqueous solution but did not affect the adsorption of methylene blue. The behavior was attributed to electrostatic repulsion between the anionic metanil yellow and the surface oxide groups.

Subsequent studies investigated the effect of acidic surface oxides on the adsorption of phenol, nitrobenzene, and sodium benzenesulfonate (33, 34, 40). The alteration of the carbon surface by chemical treatment, especially with respect to the formation and removal of surface oxides, significantly changed the adsorptive capacity of the carbons studied. In all of the experiments oxidation strongly reduced the adsorption capacity for all adsorbates, reduction of the oxidized carbon slightly increased the capacity, and high temperature outgassing of the oxidized carbon restored the original capacity. Two possible explanations for the observed inhibition of adsorption were the removal of electrons from the n electron system of the carbon by the chemisorbed oxygen and the bonding of water molecules to the acidic oxide functional groups. Since phenol was thought to adsorb through interaction of its n-electron system with the n bonds of the graphitic carbon planes, removal of available electrons would reduce phenol adsorption. The adsorption of water molecules would result in the formation of complexes of associated water within the carbon pores and could









32

prevent mass transfer of organic molecules to a large portion of the active intraparticle surface area. These, and other studies (25, 36), suggest that carboxylic and lactone-type surface oxides reduce the adsorption of aromatics. These are the two principal types of acidic surface oxides formed by chemical oxidation or by activation temperatures below 4000C. Carbonyl groups in the form of quinone and hydroquinone are formed at activation temperatures about 4000C and improve the adsorption of aromatics through the formation of an electron donor-acceptor complex with the surface carbonyl group. Adsorption of aromatics can also occur, however, by 7-bond interaction with the rings of the basal planes (35). The adsorption of nonpolar aliphatic compounds is also hindered by acidic surface oxides (35). These hydrophobic compounds preferentially adsorb on carbons free of acidic surface oxides. Again, the formation of water complexes through hydrogen bonding with the surface oxides blocks the surface from these hydrophobic aliphatics.

Basic surface oxides are formed when a carbon surface is first freed from all surface compounds by heating in a vacuum or an inert atmosphere to 800-10000C and then brought into contact with oxygen after being cooled to temperatures as low as -400C (38). Activated carbons with basic surface oxides as the predominant type of oxygen functional groups are not common. Basic surface oxides may only cover 2









33

percent of the surface area, at most, whereas acidic surface oxides can cover as much as 20 percent (38). Such carbons exhibit a positive surface potential (35). The structures of basic surface oxides are not well defined. Some suggested structures and mechanisms to account for observed acid adsorption are illustrated in Figure 2-2 (28, 35). The formation of carbonium ions, when carbon with oxygen bound in chromene-like structures is oxidized in the presence of a strong acid, is the only concept able to explain all the observed acid adsorption phenomena (38).

In addition to specific adsorption mechanisms, the

presence of both acidic and basic surface oxides impart a polar nature to the carbon surface (35). As a result, the carbon will exhibit preferential adsorption for a more polar component of a binary mixture. For example, a carbon essentially free of oxygen will preferentially adsorb benzene over methanol while a carbon with acidic surface oxides will preferentially adsorb the methanol (35). With increased surface polarity pore constriction or blockage can result from the adsorption of water through hydrogen bonding. For polar solutes, the surface oxide-organic interaction is stronger than for nonpolar solutes, compensating somewhat for the extra energy needed to desorb water (34).

This brief overview of the surface chemistry of

activated carbon illustrates some of the many parameters and mechanisms involved in determining the adsorptive capacity









34

BENZOPYRYLIUM ION R R (CARBONIUM ION)



0- H 02 + H2 02
O R STRONG ACID

CHROMENE ACID REACTION



R+ R R + R










BENZOPYRYLIUM DERIVATIVES (RESONANCE HYBRIDS)









= 0 STRONG O0 ACID






PYRONE-LIKE STRUCTURES







FIGURE 2-2. Basic oxygen surface functional groups.








35

of an activated carbon for an organic compound. The specificity of mechanisms for certain classes of organics, the variety in type and relative abundance of surface functional groups among different carbons, and the lack of laboratory studies beyond single or bisolute systems point out the complexity and difficulty in modeling carbon adsorption on a micro scale.


2.3.5 Particle Size


The effect of carbon particle size on adsorption

capacity, especially equilibrium capacity, is not clear. Theoretically, grinding and sieving of an activated carbon to a particular size fraction should have no effect on adsorption capacity. Assuming that the carbon particle is a nonporous cube, the outer surface area of the largest particle in a 12x40 mesh carbon (1.68 mm x 0.420 mm) would
2
be 16.9 mm Assuming a carbon density of 0.41 g/cu cm, 1 gram of 1.68 mm carbon particle cubes would have an outer
2
surface area of approximately 0.01 m. Grinding of this carbon to pass a 200 mesh sieve (0.074 mm) would produce a theoretical outer surface area of approximately 0.2 m2 Further grinding to pass this gram of carbon through a 325 mesh sieve (0.044 mm) would result in a theoretical outer
2
surface area of approximately 0.33 m2. Compared to the 1000 m2/g surface area of a typical activated carbon, the outer surface area contributes less than 0.03 percent. Therefore,








36

grinding a larger mesh carbon into smaller-sized particle fractions should not affect total carbon surface area to any observable extent.

Several researchers have examined the effect of carbon particle size on adsorptive capacity and, for the specific carbons studied, observed little or no effect. In two separate studies a coconut-shell-based carbon exhibited no effect of carbon particle size in the adsorption of phenolic compounds for particle-size fractions ranging from 12 x 16 mesh to 230 x 270 mesh (42,43). The variation of adsorptive capacity with particle size for a coal-based carbon was examined for five fractions of mean particle diameters ranging from 0.273 to 1.011 mm (44). No effect on equilibrium capacities was observed. Other studies have produced similar results (45-47). Based on such results and theoretical considerations, a number of researchers using a particular size fraction of a ground and sieved carbon have assumed that particle size did not affect carbon adsorption capacity (48-50).

Other researchers, however, have observed an increase in adsorption capacity with a decrease in carbon particle size. In one study, a coconut-shell-based carbon was crushed and sieved to produce two narrow-range particlesize fractions (51). In a second study a coconut-shellbased carbon was only sieved to produce the desired particle-size fractions and, in both studies, a significant








37

increase in capacity with decreasing particle size was observed (52). In both studies the times required to attain equilibrium were carefully determined. A bituminous-based carbon was studied in the development of a rapid method for determining carbon adsorptive capacities (53). Carbon particle size again affected adsorption capacity for the specific carbon used. Adsorptive capacity for humic acid on a lignite-based carbon also varied with particle size (54). One recent study concluded that there was not enough information available to state with certainty that particle size will affect adsorption capacity for the carbon of interest (47).

Several phenomena have been postulated to explain the observed variations in adsorptive capacity. Available pore volume may increase as carbon particle size decreases due to the opening of previously sealed pores and channels. This increase in available pore volume may result in an increase in the adsorptive capacity for smaller, low molecular weight solutes but may not affect the adsorptive capacity for larger, high molecular weight solutes. If blockage of long channels at constrictions is a mechanism affecting adsorptive capacity, then reducing carbon particle size by fracturing along such channels may increase the total pore volume available to large organic molecules. It is also likely that the pore structure throughout a carbon particle is not uniform as a result of activation conditions. The








38

adsorptive capacity may be greater in the outer regions of a carbon particle due to more extensive burnout of pores and channels, whereas the particle interior may consist only of a small volume of micropores and sealed channels. Grinding and sieving to achieve a particle size fraction may result in the retention of the low-capacity core and the discarding of the higher capacity outer shell. Further studies are needed to define the effect of available pore volume, pore distribution, molecular size of organic solute molecules, uniformity of activation, and pore blockage on adsorptive capacity as the diameter of a carbon particle is reduced by standard laboratory grinding and sieving techniques.


2.3.6 Manufacture of Activated Carbon


The rapid development and use of activated carbon as an adsorbent was begun during World War I when a protective mask was developed to shield soldiers from chlorine gas. Since that time the applications and varieties of activated carbons have expanded into a complex industry where many different activated carbons are available. The unique characteristics of specific surface area and surface functional groups are primarily determined by the raw material and the manufacturing, or activation, process. The importance of activation temperatures and contact with oxygen in the production process has already been mentioned. Continued improvement in raw material selection,









39

carbon preparation, and carbon selection promotes greater success and cost effectiveness in the removal of organics from the aqueous phase.

The selection of raw material is an important first

step in manufacturing an activated carbon. Table 2-2 lists some of the raw materials that have been used in the production of activated carbon. Bituminous coal, lignite, and petroleum coke are the most common raw materials for water and wastewater treatment applications (55). The choice of raw materials determines the amount of tarry substance formed during carbonization, the subsequent development of pore structure, the amount of inorganic matter to catalyze surface functional group and heterocyclic ring structure formation, and the amount of oxygen available to form surface functional groups.

The actual manufacturing process has two objectives. The first is to produce a very large surface volume ratio with a specific distribution of pore sizes. The second objective is the formation of specific types and densities of surface functional groups. Two methods of manufacture are used: A previously charred carbonaceous substance is oxidized with steam and high temperatures, or the carbonaceous raw material is chemically dehydrated and oxidized at lower temperatures. Almost all of the activated carbon used in the United States for water and wastewater treatment applications is produced by the high








40






TABLE 2-2. Raw materials that have been used to produce
activated carbons.



*Coal (bituminous) Beet sugar sludges Asphaltic material Cane sugar Blood Rubber Coconut shells Coffee beans Waste tires Sawdust Corncobs and stalks Leather wastes
*Lignite Oil shale Molasses Wood Distillery wastes Pulp mill wastes Peat Fruit pits Rice hulls
*Petroleum coke Seaweed Fish Bone Petroleum heavy oil Cereals


*Most common raw materials for water and wastewater application.









41

temperature-steam process (55). The details of the process vary considerably among manufacturers with critical steps kept proprietary. The general process for manufacturing activated carbon consists of a dehydration step, a carbonization step, and an activation step. Descriptions of each step for a variety of raw materials are available in the literature (8, 10, 56).

The manufacture of granular activated carbon from a bituminous, subbituminous, or lignite coals generally is accomplished by the following operations (55). Pulverizing, compacting, and crushing followed by sizing to a range of 2 x 40 mesh U.S. Standard Sieves begins the process. Dehydration is accomplished in air at temperatures ranging from 150 to 2150C for 0.5 to 18 hours. Dehydrating agents such as zinc chloride or phosphoric acid may be added. Depending on the oxygen content of the raw material, 1 to 3 percent by weight of oxygen may be added to the coal if the subsequent activation process occurs in a controlled oxygen atmosphere. Acid washing may precede the dehydration step to reduce the inorganic content of the raw material.

The second operation in the manufacturing process, carbonization, occurs in the absence of oxygen and at temperatures ranging from 400 to 6000C. Metallic chlorides can be added to increase the effectiveness of the carbonization process. During this phase, also called pyrolysis, a char suitable for steam activation is produced. Most of the









42
noncarbon elements as well as hydrogen and oxygen are removed as gases formed during this pyrolitic decomposition. Elementary microcrystallites are formed in irregular arrangements with many free interstices that become filled or blocked by unorganized carbon (amorphous) from the decomposition of tarry substances. The resultant char has little adsorptive capacity as a consequence.

The third essential operation, activation, determines

the final characteristics of the carbon. Temperature is the most critical parameter and varies from 400 to 12000C. The intercrystallite residues are oxidized, removing pore blocking materials. The pores are enlarged and cleaned. The microcrystallites grow and form the final rigid skeletal foam structure that determines the abrasion resistance, or friability, of the carbon. Low-capacity carbons are partially activated by removing tarry products by heating them in a stream of inert gas or by extracting them with a solvent. High adsorption capacity carbons are activated under conditions where the activating agent (steam, carbon dioxide, or oxygen [air]) reacts with the carbon. Disorganized carbon is first oxidized to open the pores between the microcrystallites. Then the crystallites themselves are partially oxidized to create a much greater pore volume. Macropores and transitional pores are formed at the expense of micropores. A burn-off between 50 and 75 percent of the








43

original char results in the mixed pore structure typical of activated carbons used in water and wastewater treatment.

The activating agent reacts with the carbon at the

edges of the microcrystallites to form surface functional groups. The type and density of groups are determined by activation temperatures and presence of activating agents. Since higher temperatures promote the formation of larger microcrystallite planes, some control over surface functional group density is exerted. The type and density of oxygen-containing surface functional groups are usually controlled by the temperature of the degassing step, which completes the activation process. The higher the temperature, the less oxygen remains on the carbon surface. As an example, activation temperatures of 400 to 5000C with low degassing temperatures will produce an abundance of acidic surface oxide groups which are best suited for adsorption in alkaline solutions. Activation temperatures of 800 to 10000C and low degassing temperatures produce large quantities of basic surface oxides which readily sorb acid. Degassing temperatures around 10000C will remove almost all oxygen from the carbon surface producing a product suitable to remove trihalomethanes, such as chloroform, which do not adsorb well on carbons having a high density of surface oxide groups (26, 37).

Other treatments to increase surface area after

activation have included de-ashing the carbon with hot








44

hydrofluoric or hydrochloric acid, burn-offs in oxygen under low oxygen partial pressures at 6000C, and exposure to ozone at 300C (26). Postactivation processes can include crushing, grading, grinding, acid washing, water washing, drying, and packaging. Powdered carbon is pulverized so that 95 to 100 percent pass a 100 mesh U.S. Standard Sieve (149 micron opening) and 50 to 95 percent pass a 325 mesh sieve (44 micron opening). Granular carbon is described by two sieve sizes, one which passes the carbon and one which retains it. A typical commercial granular carbon is described as 12 x 40 mesh, where carbon would pass a 12 mesh sieve (1.68 mm) and be retained by a 40 mesh sieve (0.42 mm). Carbon is usually packaged in airtight containers to minimize oxygen adsorption and adsorption of volatile organics from the surrounding air.


2.3.7 Summary


The foregoing discussion has shown that activated

carbon is a complex, highly heterogeneous adsorbent. The surface of activated carbon can be manipulated by manufacturing processes to exhibit a wide variety of adsorptive behavior. Adsorption energies can range from the very weak van der Waals physical adsorption energies to the very strong energies of the chemical bonds typical of chemisorption. Adsorption mechanisms are also greatly affected by the physical structure of activated carbon, where pore








45

size distributions affect the rate and capacity of adsorption for solutes of various physical dimensions and charge characteristics. Consequently, activated carbon cannot be considered a relatively homogeneous material, with adsorption phenomena described by a few simple physical mechanisms.


2.4 Effects of Adsorbate Properties on Activated Carbon Adsorption


The effects of physical and chemical properties on the adsorption capacity for specific organic solutes in aqueous systems have been studied extensively in efforts to correlate such properties with observed adsorption behavior. The ultimate goal of such correlations was some degree of predictability in the adsorption process to improve the cost effectiveness of system design. Some adsorbate properties studied include molecular weight, functionality, polarity, hydrophobicity, dipole moments, aqueous solubility, molecular size, geometry, and surface area. Although these properties were studied individually, results indicated that few were independent of other properties in their effects on adsorption. However, all were shown to directly affect adsorption capacities under some set of well defined experimental conditions.

Adsorptive capacity was observed to increase with

molecular weight in homologous series (52, 57, 58). The increase in molecular weight was due to the addition of hydrophobic nonpolar groups to the basic molecular








46

structure, such as increases in alkyl chain length. As molecular weight increased in these studies, hydrophobicity increased, solubility decreased, molecular size increased, and molecular geometry and surface areas changed. However, adsorption capacity studies using a series of nonyl-phenol ethoxylates indicated that adsorption capacity decreased with increase in molecular weight (27). The increase in molecular weight was due to the addition of ethylene oxide groups into the nonyl chain. As molecular weight increased, hydrophobicity decreased while polarity, solubility, and molecular size increased. Consequently, the use of molecular weight as a predictive parameter for adsorption capacity is limited by interrelationships with other sorbate properties.

Functionality has also been examined as an indicator of adsorption behavior. The addition and/or substitution of such functional groups as NH2, OH, CH20OH, NO2, alkyl groups, and benzene rings to benzene and phenols have been the most commonly studied functional systems (42, 52, 57-59, 60-61). Adsorptive capacity for phenolic compounds increased as alkyl substitutions were made on the phenol and as the alkyl chain length decreased. Polycyclic compounds adsorbed better than corresponding monocyclic compounds. The adverse effect of OH, CH20OH, and NH2 groups on adsorption was attributed to hydrogen bonding of these groups with water molecules rendering the organic more hydrophilic. In two








47

instances the introduction of the OH group enhanced, rather than hindered, adsorption even though the addition of OH resulted in an increase in aqueous solubility. The presence of neighboring CHO and OCH3 groups was thought to result in intramolecular hydrogen bonding with the OH group. Neighboring groups, like OCH3, could also sterically hinder hydrogen bond formation between water and the OH group. Functionality appears to be a qualitative, rather than a quantitative, indicator of trends in adsorptive capacity because of the many exceptions to the expected behavior and is not suitable as a general predictive parameter.

Polarity has also been examined as an indicator of adsorption capacities and for predicting adsorption characteristics of other organics. The dipole moment, as a measure of the polarity of a compound, exhibited no correlation with the carbon capacity for 26 organic compounds studied (58). Aniline and phenol possess the same dipole moment, yet exhibited different adsorption capacities on the same carbon. Nitrobenzene had the highest dipole moment of five organics, yet was the most favorably adsorbed. One review article described seven different polarity scales that have been used to classify organics (12). Predictability of adsorptive capacities was limited to compounds of the same class as the representative compounds. Lack of knowledge concerning how each chemical in a class would react in a particular system also limits the usefulness of








48

such polarity scales. Again, the diversity of compounds as well as reactions involving more than just one type of force interaction do not allow any one classification scheme for organics and adsorptive capacities.

Aqueous solubility seems to provide the most widely applicable means of organic classification for adsorptive capacities. All the previously described adsorbate properties can be reflected in an aqueous solubility parameter. The semiquantitative Lundelius' rule describes an inverse relationship between the extent of adsorption of a solute and its solubility in the solvent from which adsorption occurs (9). One interpretation postulated that the greater the solubility, the stronger the solute-solvent bond that must be broken and the smaller the extent of adsorption. The special case of Lundelius' rule is Traube's rule which relates the aqueous solubility of various organics in a homologous series inversely to the extent of adsorption

(4). An investigation of 93 organic solutes representing 10 different functional groups commonly found in petrochemical wastes showed, without exception, that as aqueous solubility decreased, adsorptive capacity increased (57). This relationship held true not only for decreasing solubility within a functional group but also for decreasing solubility between functional groups.








49

Further development of the adsorptive capacitysolubility relationship has resulted in the application of a solubility parameter, 6, to a macroscopic single solute equilibrium model for organics in aqueous solutions (62, 63). This parameter attempts to relate the physical, chemical, and structural characteristics of neutral organic compounds to their potential to undergo a change from the liquid to a surface phase. The numerical value of the total solubility parameter consists of components which represent specific types of force interactions. The total parameter is a function of a dispersion components, the permanent dipole orientation and induced dipole components (lumped to form a polar component), and acidic and basic components (lumped to form a hydrogen bonding component). Assigning meaningful values to these components is difficult. In the study applying the total solubility parameter to the macroscopic single solute equilibrium model to calculate net adsorption energies, the dispersion, or nonpolar, component was calculated using refractive indices (62). The hydrogen bonding component was determined using aqueous solubility data. The calculation of the polar component was restricted to several homologous series and could not be calculated independently. The study reported good agreement among the four of five methods available to calculate the total solubility parameter. The calculated value was used together with values for the dispersion and hydrogen bonding








50

components to estimate the polar component values. This scheme of classification lacks a quantitative, theoretically justified basis and is limited by the lack of a single method to calculate the component solubility parameters for all applications. However, the use of total solubility parameter values in the equilibrium model has resulted in correlations between amounts of compounds adsorbed by activated carbon and the calculated net adsorption energies for several organic compounds (62, 63).

The Polanyi adorption potential theory has been applied to adsorption of organic solutes from aqueous solutions

(64). A method was developed that makes it possible, given a calibrating isotherm for a specific carbon, to estimate the adsorption isotherms with that carbon for a wide variety of organic liquids from water over the complete range from saturation to very low concentrations, given only their solubilities and densities. The discussion of the application of the Polanyi theory to adsorption from water solution onto activated carbon also provided a list of references describing the fundamental studies on liquid-phase adsorption onto activated carbon (64). The model assumes heterogeneity of adsorption energies over the adsorption space or surface and that any interactions between solute and solvent and between different solutes in solution are reflected in their effects on solubilities. However, the solubilities that enter into the calculations for competitive adsorption








51

are the solubilities of each component in the presence of the other.

Adsorption capacity, however, has not always correlated inversely with aqueous solubility. One study showed that the adsorptive capacity for phenol on a specific carbon was the same as for benzyl alcohol and higher than that of aniline, yet phenol is more soluble than either (58). Nitrophenol adsorbed just as well as cresol, yet has a solubility an order of magnitude less than cresol. The adsorptive capacity for o-methoxyphenol was greater than that for cresol or chlorophenol yet its solubility is lower. Similar observations were recorded for other organic solutes. The conclusion of the study was that no overall general tendency of increase in adsorption with decrease in solubility was observed over the range of the 26 compounds studied.

Aqueous solubilities and single solute adsorption

capacities have been used in bisolute systems to attempt to predict relative adsorbabilities (59, 61, 65-67). The adsorptive behavior of alkyl phenols in two-component systems was better described by the Ideal Adsorbed Solution (IAS) equilibrium model than by the Langmuir model for competitive adsorption (59). The IAS model uses solubilities and single solute isotherm constants. Another study used isotherm constants and solubilities to predict the preferentially adsorbed compound in 22 bisolute systems

(65). Solubility was a successful predictive parameter for








52

20 out of 22 pairs tested. The Freundlich K constant, an indicator of the adsorptive capacity of a carbon for a solute, was a successful index in all 22 pairs tested. Three other isotherm constants from the Langmuir and Freundlich models were not as successful. An improved calculation procedure for the IAS equilibrium model was used successfully to describe adsorption in binary and ternary systems containing similar and dissimilar components and mixtures containing different initial concentrations of solutes (66). However, another study showed that molecular weight was a more reliable indicator of multi-solute adsorption preference than solubility when comparing polycyclic and monocyclic compounds (67). The effect was attributed to stronger adsorptive forces due to greater surface area contact. Solubility was a reliable indicator of preference for monocyclic sorbate pairs.

There does not seem to be any one adsorbate property that is universally applicable for predicting relative single solute adsorption capacities. Under limited conditions, adsorptive capacity generally increases with increasing molecular weight and with decreasing solubility. In multi-solute systems preferential adsorption appears to be proportional to some measure of single solute adsorption capacity, such as the Freundlich K, and to molecular weight. Such adsorption also appears to be inversely








53

proportional to solubility. Single solute adsorption isotherm parameters have been used for attempting quantitative predictions of multi-solute adsorption while parameters describing the physical-chemical nature of the organic have been used qualitatively. However, the presence of exceptions to all efforts at predicting preference and quantities of adsorption imply that no single adsorbate property is applicable for general use. The development of several recent multi-solute equilibrium models attempts to combine several of the more significant properties to improve overall predictability.


2.5 Effects of Aqueous Solvent System
Properties on Adsorption


Aqueous system characteristics which exert influence on activated carbon adsorption include temperature, pH, dissolved solids, and matrix complexity. Since adsorption is normally an exothermic process, adsorptive capacity generally decreases with increasing temperature. Gas phase adsorption involves heats of adsorption on the order of several kilocalories per mole, and such systems are very temperature sensitive. However, adsorption in aqueous systems involves much smaller heats of adsorption, on the order of a few hundred calories per mole, due to the endothermic desorption of water when adsorption of a solute occurs on a carbon surface. Normal temperature variations experienced in








54

water and wastewater adsorption processes have only a minor effect on adsorption capacities (9, 21, 30, 49, 52).

Dissolved salts can significantly influence the adsorptive capacity of carbon for ionized organic molecules but exert little influence on neutral solutes. The adsorption of sodium benzenesulfonate was greatly increased when CaCl2 was added to the solution (34). This was attributed to a decrease in the mutual repulsion between surfactant head groups through interaction with the divalent calcium ions. At pH 2.0 no significant differences in adsorptive capacity for p-nitrophenol were observed with NaCl concentrations up to 1 mole/liter. At pH 2.0 the p-nitrophenol exists solely as a neutral species. When the pH was raised to 10.0 where ionic species dominate a significant salt effect was observed at the higher NaCl concentrations (49). The explanation involved an ion pairing of the cation with the anionic p-nitrophenol, thereby reducing the electrical double layer. A second possible mechanism was a partial repulsive charge reduction between adsorbed anions by the salt. A study of the effect of phosphate buffer on 2,4-dichlorophenol and 2,4-dinitrophenol indicated no influence on adsorptive capacities for undissociated solutes, but an increase in adsorption of dissociated solutes as buffer concentration increased (68).

The effect of salts on the adsorption of fulvic acids has received increasing attention as these substances have








55

been cited as precursors to trihalomethane formation. The concentration of a phosphate buffer in solution had a significant effect on the adsorption of soil fulvic acids at pH 7.0 (69). Another study showed that the adsorptive capacity for dissociated dinitrophenol nearly doubled in the presence of a 0.01 M phosphate buffer (70). However, a later study demonstrated that enhanced adsorption of fluvic acid was due to cation concentrations (calcium, magnesium, and sodium) and that anion concentrations (chloride, sulfate, bicarbonate, and phosphate) had little or no effect

(71). All salts studied showed little additional effect on adsorption capacities beyond salt concentrations of 3 to 4 millimoles per liter. The effect of salts varied significantly with pH showing increased influence on adsorption as pH increased, confirming the effect of cations on the adsorption of dissociated fulvic acid species.

Available literature indicated that inorganic salts can influence the extent and rate of adsorption of anionic organics onto negatively charged surfaces. Few studies have attempted to define the mechanisms responsible for enhanced or hindered adsorption in the presence of salts. Speculative mechanisms include solute-salt interactions to alter specie distribution, salt-adsorbed organic interactions to alter surface packing, and salt-carbon interactions to alter surface charge characteristics. Salts do not appear to significantly influence the adsorption of neutral compounds,









56

but few data are available. Available data are insufficient to draw general conclusions about the effects of co-ions, various salts, and salt concentrations on activated carbon adsorption of organics in aqueous solutions.

The effect of pH on adsorptive capacity has been widely studied. The two components of the adsorption mechanism thought to be affected by solution pH are the chemical characteristics of the carbon surface and the extent of adsorbate dissociation (12). Equilibrium capacities for sulfonated alkylbenzenes in pH regions far from the pK range for these compounds increased with increasing hydronium ion concentrations (52). Far from the pK range pH changes would not significantly affect the net ionic character of the adsorbing species. The increase in adsorptive capacity was attributed to partial neutralization of the carbon surface's negative charge character, reducing resistance to pore diffusion. A study of the adsorption of phenol at pH's from

2.0 to 10.6 showed an unexpected reduction in adsorption capacity with decreasing pH (49). The interaction of hydronium ions with surface carbonyl groups competitively with phenol was postulated to explain the decreasing phenol capacity. The effect on nitrophenols was not as great and was attributed to stronger bonding energies through the phenolic acceptor-donor complex reaction. The adsorption of the hydronium ion as a competitive solute at low pH values has been supported by other work (21, 46, 72, 73).









57

The extent of dissociation for ionizable solutes is greatly affected by pH, especially for weak organic acids like phenols. Observed changes in adsorptive capacities with pH for phenol and nitrophenols were attributed to changes in solubility (21). Adsorptive capacities for aromatic acids passed through a maximum at pH values near the pKa points of the organic acids where both ions and molecules exist in comparable amounts in the bulk solution

(72). This study also indicated that substantial amounts of ionic species were adsorbed as well as molecular species. In a study involving chloro-substituted phenols the adsorptive capacity for trichlorophenol increased substantially as pH decreased (74). Dichlorophenol adsorption was significantly less affected by pH changes. Adsorption capacity was observed to reach a maximum near pH = pKa. The neutral species of both solutes adsorbed more strongly than the anionic species. The effect of pH on adsorption from a mixture of the two substituted phenols was large but could not be described by existing competitive models. Adsorption capacities for 2-methylpyridine and o-cresol decreased as pH was lowered from 9.5 to 3.5. The effect of o-cresol was much less than that for 2-methylpyridine. In the pH range of 6.8 to 9.5 the amount of ionic species of o-cresol increased and of 2-methylpyridine decreased, yet adsorption capacities were unaffected. This indicated that both anionic and neutral species were adsorbed on the carbon.









58

The decrease in adsorptive capacity of the two compounds was attributed to competitive adsorption of hydronium ions. Substantial increases in the adsorption of fulvic acid were reported as pH values decreased in a study that attributed the enhanced adsorption to a decrease in fulvic acid solubility (69).

The effects of pH on adsorption capacity in aqueous solutions are not well understood. Observed pH effects depend on the nature of the carbon surface, especially the functional groups thereon, the nature and concentration of the organic solutes, temperature, and ionic strength. The effects of pH are not well described by existing competitive equilibrium models outside narrow and well-defined experimental conditions. Differences in solubility between neutral and anionic species contribute to observed changes in adsorptive capacities. Development of repulsive forces between the anion and the surface or between sorbed anions is also significant. Changes in the surface charge of the carbon due to hydroxyl or hydronium ion adsorption or ionization of sorbed molecules or weakly acidic surface functional groups could substantially affect adsorptive capacities. In water treatment processes the adsorption of neutral organic compounds is probably not greatly affected by solution pH changes in the ranges encountered as are the weak acids used in most of the studies reported in the literature (12). Most of the organics of health









59

significance are such neutral compounds. However, in the case of organics susceptible to ionization, pH conditions yielding the highest undissociated state seem to provide the best conditions for adsorption.

Matrix complexity is the final major characteristic of an aqueous system which influences adsorption capacities. Matrix effects are particularly significant in wastewater treatment applications. The presence of other organics in the solution usually reduces the adsorptive capacity of a particular carbon for each organic solute, although the total adsorptive capacity of the carbon is usually increased. Mutual inhibition can be predicted to occur if adsorption is restricted to one or two molecular layers. Two additional criteria for predicting mutual inhibition are that the adsorption affinities of the solutes do not differ significantly and that there are no strong interactions among solutes (9). Since the final equilibrium positions are dictated by thermodynamics, many multi-solute equilibrium models have been developed in attempts to provide some predictability in multi-solute systems.

The bulk of the data available in the literature on competitive adsorption involve bisolute studies. In a phenol-dodecylbenzene sulfate (DBS) system the presence of the DBS reduced the single solute adsorptive capacity of phenol by approximately 30 percent (51). Although the phenol was a much smaller molecule and had a higher









60

effective rate of diffusion, pore blockage by DBS was thought to reduce the surface area and pore volume available to phenol. A study of competitive adsorption between dichloro- and trichlorophenol showed that, as the equilibrium concentration of one solute increased relative to the second solute, the adsorption of the second solute decreased (74). In a study of competitive adsorption with bisolute mixtures of organic anions and neutral species, similar results were seen for some mixtures but not for others (75). Some adsorbed with substantial competitive effects while others showed little competition. In some cases, electrostatic repulsive forces were significant. Little competition was expected where there was a large difference in solute molecular sizes because of pore size distributions in the carbon. Mutual suppression of single solute adsorption capacities was also observed in a study of six fatty acids and four phenolic compounds (76). In competitive adsorption from bisolute mixtures of a fatty acid and a phenolic solute reduction in the adsorptive capacity for fatty acids was substantial but was not so extensive for phenol. Other bisolute studies are also available (77-79). Additional bisolute data will be discussed in the presentation of multi-solute equilibrium models.

Much more limited are data from multi-solute studies involving three or more compounds. In a mixture of three fatty acids the adsorptive capacity was reduced for acetic










61

acid substantially and slightly for propionic acid. However, the capacity for butyric acid actually increased somewhat (76). In a study of competitive adsorption in ternary phenolic mixtures, the equilibrium capacities were successfully predicted by the Ideal Adsorbed Solution model with a modified calculation procedure (66). A fourcomponent mixture of glucose, alkyl benzene sulfonate, an amino acid, and lactic acid was studied in a column reactor to validate a dynamic packed bed column model. Some calculations were also made for a nine-component influent (80). Adsorption equilibria of individual solutes in ternary mixtures of varying initial concentrations showed decreases in single solute adsorptive capacities (81). The most strongly adsorbed solute was described by a Freundlich isotherm, but the more weakly adsorbed solutes had curved isotherms on a log-log scale. The curved isotherms of individual solutes adsorbed from a mixture were typical if one or more of the competing solutes had higher adsorption capacities as a single solute than the remaining solutes. However, all solutes showed decreased adsorptive capacities in mixtures. These observations were also confirmed by equilibrium data in mixtures of up to six compounds.

The previously discussed research all described or

postulated factors determining interactions among competing solutes. The relative size of adsorbates seemed to primarily affect transport processes by pore blockage. The









62

relative affinities of competing solutes, if sufficiently different, can result in the displacement of previously adsorbed solutes. A very strongly adsorbed solute can even cause an otherwise adsorbable compound to appear to be nonadsorbable. In column studies of mixtures of solutes, effluent concentrations of a solute exceeding influent concentrations in a steady state system are commonly observed. The relative concentration of solutes can also play a major role in site competition. A solute can compete more successfully when its concentration relative to other competing solutes is increased. Solute-solute interactions such as electrostatic repulsion/attraction between ions, dipole, and w-electron clouds can be significant. The heterogeneity of the carbon surface can include adsorption sites so specific as to preclude any competition. Such noncompetitive adsorption is believed to be limited to ionic species and, therefore, determined by solution pH and carbon surface characteristics. The increase in combined adsorptive capacity with multi-solute systems is thought to reflect these phenomena. Site specificity can also result from limited accessibility due to pore size constrictions in addition to reaction specificity of certain surface functional groups. The exponential increase in mechanism complexity in competitive adsorption renders mathematical description extremely difficult beyond bisolute systems. Current efforts to develop predictive models consequently









63

must rely on such techniques as lumped isotherm parameters, organic homologues, total organic carbon isotherm analysis, and other approximations to reduce a multi-solute adsorption problem to a two- or three-component system.


2.6 Descriptive Single-solute Equilibrium Models


Adsorption results in the accumulation of a solute from a solution onto a surface of a solid until a state of dynamic equilibrium is reached in the system at a fixed temperature. At this equilibrium there is defined distribution of solute between solid and liquid phases. Mathematical descriptions of this equilibrium are referred to as adsorption isotherms and attempt to express the solid phase concentration of solute as a function of the liquid phase concentration at equilibrium. Single-solute adsorption systems are generally described by one or two of several isotherm equations. Some apply solely to monolayer adsorption while others can describe multilayer adsorption.


2.6.1 Henry's Law


Henry's Law is the simplest of the single-solute

equilibrium models. This linear isotherm is valid only at very low concentrations and assumes that all molecules are isolated from their nearest neighbors. This linear relationship is analogous to the limiting behavior of solutions of gases in liquids and the constant of proportionality








64

referred to as the Henry constant. This linear partitioning model is thermodynamically based and is the lower boundary condition for any adsorption equilibrium model (59, 82). It is expressed as


qe = KhCe (2-1) where qe is the equilibrium solid phase concentration, Ce is the equilibrium liquid phase concentration, and Kh is the Henry constant or the partition coefficient. The thermodynamically based Langmuir equation reduces to Henry's Law at small solute concentrations. Several other single and multi-solute equilibrium models have been modified to reduce to Henry's Law as dilute solution conditions are approached. A three-parameter isotherm equation was developed to describe the adsorption equilibria of five
-5 -i
organic compounds over a concentration range of 10 to 10 M (22). Three of the solutes (propionitrile, 2-propanol, and acetone) exhibited isotherm slopes near unity at the lower concentration ranges, and the slopes approached unity even closer as temperature rose. Such adherence to Henry's Law was typical of poorly adsorbed solutes. The remaining solutes (p-chlorophenol and p-cresol) did not exhibit the same behavior, which indicated that the phenol-carbon surface interactions were much greater than those of the first three solutes. The proposed three-parameter isotherm reduced to Henry's Law at low concentrations and was








65

successful for describing the adsorption of the solutes studied.

Linear partitioning behavior was also observed for

polychlorinated biphenyls (PCBs) when adsorbed on a river sediment (83). At higher solution concentrations the adsorption equilibrium was nonlinear. Adherence to Henry's Law was observed for the adsorption of phenol and several alkyl phenols at concentrations in the 10-4 M range (59). Isotherm data for many different types of organic compounds over a concentration range of 10-9 to 10-2 M showed compliance with Henry's Law as equilibrium concentrations decreased (62). The more poorly adsorbed compounds, such as urea, showed linearity at higher equilibrium concentration values (10-3 to 10-2 M), whereas more strongly adsorbed compounds did not exhibit linearity. For example, the isotherm for 2,4,6-trichlorophenol over a range of 10-8 to 10-3 M had a maximum slope of only 0.592 at 10-8 M. Three other three-parameter isotherms not commonly used in water and wastewater studies also reduced to Henry's Law at low concentrations (82).

The importance of Henry's Law in equilibrium modeling has been rather neglected since organic solutes of concern in adsorption studies were in the mg/L range. However, with analytical techniques available today to analyze for organics at the nanogram per liter range and with the presence of almost 200 compounds of health concern in water









66

supplies at the nanogram level, the ability of any equilibrium model to reduce to Henry's Law has become much more significant.


2.6.2 The Langmuir Equation


The Langmuir equation has been one of the most popular single-solute equilibrium models because of its theoretical basis, success, and simplicity. The model can be derived from statistical thermodynamics as well as kinetically (4, 7). The kinetic derivation is based on the following assumptions:

1. Molecules are adsorbed at a fixed number of welldefined localized sites;

2. Each site can hold only one adsorbate molecule;

3. All sites are energetically equivalent;

4. There is no interaction between adsorbed and gaseous phases;

5. Adsorption is reversible.

The adsorbed layer, therefore, can only be a monolayer if the model is to successfully describe the equilibrium. The Langmuir equation is commonly written as follows (9):


QbC
q e (2-2)
e (l+b C )
e


where qe is the equilibrium solid phase concentration, Ce is the equilibrium liquid phase concentration, Q is the solid








67

phase concentration when a complete monolayer is formed on the surface, and b is related to the net enthalpy of adsorption, H (b is proportional to e(- H/RT), where R is the universal gas constant and T is absolute temperature). The Langmuir equation will reduce to Henry's Law when bC << 1. The two Langmuir constants are determined by best
e
fitting the following linearized form of equation 2-2 to experimental data:


1 1 + ) (2-3)
q Q bQ Ce


The basic assumptions of the Langmuir equation are never met in activated carbon adsorption systems treating water or wastewater because of the heterogeneity of the carbon surface. Adsorption data have been successfully described by the Langmuir equation for very narrow concentration ranges, but it usually does not fare well when compared to other isotherm equations.


2.6.3 The Brunauer, Emmet, and Teller (BET) Equation


The BET model was an extension of the Langmuir equation to describe multilayer adsorption (4, 7). Consequently, this equation also assumes a surface composed of uniform, localized sites and that adsorption at one site does not affect adsorption at neighboring sites. The model development is based on the same kinetic picture as the Langmuir









68

equation but assumes that layers of molecules adsorb on top of the previously adsorbed monolayer. The BET model also assumes that the Langmuir equation applies to each successive layer. The first layer's heat of adsorption may have a unique value, but in all succeeding layers the heat of adsorption is equal to the heat of condensation of the liquid adsorbate. The BET equation is commonly written as follows:


BC Q
e = [C-C [l + (B-1)(C /C ) (2-4)



where B is a constant reflecting an energy of interaction with the surface, Cs is the saturation concentration of the solute in the liquid phase, and the remaining symbols are as defined for equation 2-2. Equation 2-4 is normally rearranged to a linear form for application with experimental data:


C C
e 1 B-1 C e
e 1 + 1 1 1 (2-5)
(C -C )qe BQ BQ C


The BET equation has become the standard one for surface area determinations, usually with nitrogen at 770K as the adsorbate (4). However, the equation has not enjoyed wide use in the field of environmental engineering because of difficulties in obtaining appropriate values for Cs and also because at the concentrations found in most water and








69

wastewater applications Ce << Cs and the BET equation reduces to the Langmuir equation. The saturation concentration, Cs can only be estimated in practice by an iterative procedure to fit the model to experimental data

(84). The model is also restricted by the same limitations of the Langmuir equation.


2.6.4 The Freundlich Equation


This single-solute equilibrium model is one of the most widely used in the water and wastewater treatment field. The equation was developed to empirically describe adsorption and therein lies its greatest weakness. This equation was meant to describe that portion of the isotherm beyond the dilute solution region of Henry's Law (85). The development of the equation was based on the assumption that the adsorbent had a heterogeneous surface composed of different classes of adsorption sites, with adsorption on each class of sites conforming to the Langmuir equation (86). The development of the Freundlich equation begins with the Langmuir constant, b, which is related to the net enthalpy of adsorption. On heterogeneous surfaces b will not be a constant but rather a distribution function of energies that can be related to fractions of occupied surfaces. The new adsorption isotherm is the integration of each adsorption isotherm function for each value of b in its distribution function. This integral equation is the experimentally








70

observed adsorption isotherm. One particular solution of this integral equation assumes the Langmuir model to be the isotherm function and attributes the variation in b solely to the heat of adsorption. The solution to the integral equation, in this case, has the form


1/n
qe = K Ce (2-6)



and is known as the Freundlich adsorption isotherm (4), where K and n are statistically determined, best fit constants. The equation does not become linear at low concentrations nor does it show a saturation or limiting value. Although the derivation is somewhat theoretically based, it is considered an empirical equation limited in its usefulness to its ability to fit data. The limitations of the Freundlich equation are often ignored by many investigators who incorrectly extend the model beyond the valid experimental range it describes (85). However, the model has successfully correlated experimental data for adsorption on activated carbon from aqueous systems over a wide range of concentrations, and the model has been incorporated into several multi-solute adsorption equilibrium predictive models.








71

2.6.5 Three-Parameter Equations


The lack of fit of the Langmuir equation over wide

concentration ranges and the empirical limitations of the Freundlich equation have led researchers to propose several other equations to describe experimental single-solute adsorption data. One study compared the Toth, RedlichPeterson, and Newman (three-parameter) equations with a new three-parameter equation (82). Six dilute aqueous bisolute systems were examined. The Toth and the study's equations best described the data, although all four had absolute relative percent deviations less than 10 percent and, for most of the eight compounds studied, less than 5 percent. The Toth equation is little known in the West, having been published in east European literature. Its derivation follows a similar line as the Freundlich isotherm but specifies the dependence of the integral enthalpy of adsorption as a function of the solid phase concentration of a particular solute. This dependence is specified through a dimensionless quantity that was defined for ease in obtaining experimental data. It described experimental data well in the above study. The isotherm equation developed in the study incorporated the concept that highest energy sites are filled first so that the heat of adsorption declines rapidly with increased surface coverage. The equation had the form








Kq P 72 q e e
C = (2-7)
e H


where H and K are functions of temperature and p is a constant (82). Both isotherm equations were used with the Ideal Adsorbed Solution (IAS) multi-solute equilibrium model successfully in prediction adsorption for six bisolute systems at200C.

A modified, three-parameter Freundlich model was

developed to correct for the crossover from Freundlichmodeled behavior to Henry's Law behavior (59). The equation was successfully used to predict adsorption in several bisolute mixtures and one ternary mixture using the Ideal Adsorbed Solution (IAS) multi-solute model. A similar segmented set of Freundlich equations was used to describe single solute adsorption isotherms for the prediction of multi-solute adsorption using the IAS theory in five dilute, aqueous, bisolute systems (87). The single solute isotherms were divided into five segments, each with a different set of Freundlich constants. This approach was chosen over existing three- and five-parameter isotherm equations to minimize computational time and expense. Another threeparameter model was developed to combine Henry's Law and the Freundlich equation to describe adsorption data over a wider range of concentrations (22). This equation takes the form aC
qe = c (2-8)
1 + bC
e








73

where a, b, and c are determined by statistical fit of experimental data to the equation. At low solution concentrations the equation reduces to Henry's Law. At high solution concentrations the equation is equivalent to the Freundlich model, and when c = 1, the equation becomes the Langmuir isotherm. This equation has generally described adsorption data well. Unfortunately, this eqution does not readily adapt to multi-solute equilibrium models.

There are several other single-solute descriptive

equilibrium models using a variety of parameters that treat adsorption data effectively. Some of these are listed and described elsewhere in a brief review of single-solute equilibria models (87).


2.7 Approaches to Predictive Single-solute Equilibrium Models


2.7.1 Thermodynamic Approach


Physical adsorption has been described by classical

thermodynamics applied in a macroscopic approach in two- and three-dimensional models (1, 2, 4-7, 78, 88). The fundamental expression of the thermodynamics of adsorption is the Gibbs equation which, for dilute solutions of one solute, may be approximated by



1 8 ] (2-9)
RT 8 In C
e T








74

where r is the amount of solute adsorbed per unit area of surface at equilibrium, in excess of the bulk concentration (also termed the surface excess). If y is defined as the interfacial tension, then a solute which reduces y will concentrate at the surface (1).

The development of another form of the Gibbs adsorption isotherm for the solid-liquid interface results in the form



n n di -Ad7 = 0 (2-10)




at constant temperature and where A is the molar surface area of the adsorbent, nim is an invariant adsorption term approximated by VACi in dilute solutions, pi is the chemical potential of component i, 7 is the spreading pressure, V is solution volume, and ACi is the change in concentration caused by adsorption (78).

The development of the Gibbs isotherm assumes a

thermodynamically inert solid with an available specific surface area identical for all adsorbates (78). Given the heterogeneous nature of the activated carbon surface, such assumptions are limitations to the use of the equation. However, this approach is incorporated in other methods of describing adsorptive behavior in multi-solute systems.








75

2.7.2 Energy Balance Approach


This approach takes a microscopic view of the adsorption process at the surface of a solid phase and applies an energy balance to the adsorption. Two models under investigation which take this approach include an application of the solvophobic theory and the net adsorption energy approach. These two models emphasize the importance of solute, solvent, and adsorbent interactions in the adsorption process.

The solvophobic theory was developed to estimate the effect of solvents on several types of reactions involving molecules common in biology, including isomerization, association, and conformational changes (89). The adaptation of the theory to adsorption in aqueous systems gave the net solvent effect as a summation of six free energy terms (89, 90). Although the terms could be calculated from the physicochemical properties of the system, the calculations were very complicated. If the solute molecules were much larger than the water molecules, then the equation simplified to one needing only the total solute molecular surface area to predict adsorption capacity. The solvophobic theory, while finding excellent application in reverse-phase liquid chromatography, has several limitations in its use in aqueous systems with heterogeneous carbon (91). So far, its application is restricted to un-ionized, nonpolar solutes. The model








76

assumes a uniform, nonpolar, noninteractive carbon surface and does not describe effects on adsorption of such solvent characteristics as pH and ionic strength. Such inclusions would require microscopic details not yet available, such as surface heterogeneity. It also does not provide information on the shape or type of adsorption isotherm for a particular solute. The theory also produces only a single adsorption capacity value, rather than a broad-ranged isotherm.

The net adsorption energy model was developed to

include the solvent effect in aqueous phase adsorption by relating a calculated net adsorption energy to adsorption capacity at specific solution concentrations (62, 63). The model uses the solubility parameter concept, which was discussed earlier in Section 2.4, to determine the net adsorption energy. The calculated energy is a function of a compound's aqueous solubility, molecular weight, density, a dispersion component of the solubility parameter for the organic compound, and a dispersion component of the solubility parameter for the activated carbon surface. The net adsorption energy (E T) is calculated by combining the solute's affinity for both the adsorbent (Ejs ) and the solvent (E..) and the interaction between solvent and adsorbent (Eis). The E values are in calories per mole, and the subscripts i, j, and s refer to solvent, solute, and adsorbent, respectively. The equation is of the form








77

Em = E. (E. + E..) (2-11)
S 3s is 31


Each of the energy terms is calculated using component values of the solubility parameter, 6, for the organic compounds. The application of the net adsorption energy concept to evaluate adsorption data for medium-polarity to high-polarity organic compounds showed that measured adsorption capacities increased as calculated net adsorption energies increased (62).

The application of this concept has shown several

limitations to its usefulness. The compounds that have been tested were qualitatively ranked. The model assumes a uniformly nonpolar activated carbon surface. Quantitative determinations of the components of the solubility parameter are not well developed. Application of the theory to predict the adsorption capacities of neutral organic compounds is not clearly understood. A comparison of this theory with the Polanyi theory found that the relationship between adsorptive capacity and net adsorption energy was predictive only if separate relationships were defined for branched and linear alcohols (91). If the model is to be generally predictive of adsorption capacities of organic solutes, a single relationship is needed. As adsorption energy increased, the study showed greater deviation from linearity for the alcohols. The method was not able to predict adsorption isotherms accurately, although rough








78

estimates were obtained using only the refractive index, the molar volume, and a rough estimate of solubility. The model is relatively new and untested but shows promise in elucidating the importance of solute, solvent, and adsorbent interactions and roughly estimating relative adsorption capacities for different organic compounds.


2.7.3 Adsorption Potential Approach


This approach to multilayer adsorption postulates a potential field at the surface of a solid into which adsorbate molecules "fall" (4). The adsorbed layer resembles a planetary atmosphere being most compressed at the surface and growing less dense as the distance from the surface increases. This concept of a changing adsorption potential in the adsorption process was incorporated into the Polanyi adsorption potential theory, one of the best known of the potential theories (4, 64). The development of this theory is described and well documented in several studies that applied the theory to adsorption from aqueous solution onto activated carbon (64, 92, 93).

The Polanyi theory postulates a fixed volume close to the carbon suface where the adsorption process occurs. For carbon this volume essentially consists of micropores of varying size and shape. The adsorption potential is defined as the work required to remove a molecule from its location in the micropore to infinity. The magnitude of this








79

potential depends on the nature of the surface and the distance of the solute molecule from the surface. The potential will be greatest for micropores whose volume approximates that of the solute molecules since surface area contact will be a maximum as well as the van der Waals forces of interaction.

One important consequence of this theory is that for

many single solute systems using the same carbon all of the single-solute isotherms from trace concentrations to saturation may be readily calculated from the isotherm of any one solute (64). That is, the theory postulates a single characteristic curve from a single-solute isotherm at any temperature. For a different solute, an abscissa scale correction factor will produce the same characteristic curve. To produce the unique characteristic curve of a carbon, it is possible to transform the isotherm to the characteristic curve by the conversion of qe and Ce to a volume of adsorbate adsorbed and an adsorption potential. Dividing qe by the liquid density results in the volume/mass carbon ordinate. The adsorption potential is calculated by


C
6 = RTZn ( ) (2-12)
e

where E is the adsorption potential and Cs is solute solubility. From this characteristic curve, the isotherm for any temperature can be calculated for the single solute.








80

A plot of adsorbate volume per unit weight of carbon against the adsorption potential per unit liquid volume is approximately the same for a series of similar compounds, such as a homologous series of hydrocarbons (94). An abscissa correction factor will give the same correlation curve for a wide variety of compounds. Consequently, it should be possible, given one single solute isotherm, to predict single solute isotherms for a wide variety of organic solutes across a wide concentration range. One study correlated experimentally determined Freundlich K constants for nine alcohols and plotted K against calculated net adsorption energies (91). The resultant correlation line was used to predict K values for 13 additional compounds. The net adsorption energy concept predicted K values well, better than the Polanyi with which it was being compared. Both methods deteriorated for compounds with loadings greater than one millmole per gram. The model also had difficulty with branched-to-linear alcohol relationships. The adsorbate volume also had to be empirically adjusted for variations in the maximum adsorption of all solutes even when on a volume basis.

The Polanyi model assumes heterogeneity of adsorption energies over the adsorption surface which contrasts with the Langmuir assumption of equal energies everywhere. This allows better accommodation of competitive and noncompetitive adsorption than the Langmuir semicompetitive model.








81

Any interactions between solute and solvent are reflected conveniently in solubility effects. The model has several limitations. One is the assumption that all pores are accessible to the sorbate. The model also assumes that only physical adsorption occurs and does not include chemisorption. However, this method is useful and somewhat easy to apply.


2.8 Multi-solute Equilibrium Models


Several models have been developed to describe competitive adsorption of organic solutes in aqueous solutions. Most are extensions of single-solute descriptive and predictive models. Two different approaches to modeling are currently being developed in activated carbon adsorption. Modeling the equilibrium distribution between solid and solution phases in descriptive and predictive ways has been the thrust of this review of the literature. A different modeling approach which takes into account the dynamics of adsorption has been pursued by several investigators as a basis of a rational carbon column design. The interrelationship between these two approaches lies in the thermodynamics driving a carbon adsorption system toward some equilibrium. All dynamic models must incorporate these thermodynamic equilibria considerations in order to link thermodynamics with mass transport limitations adsorption processes.








82

Although each of the current multi-solute equilibrium models adequately describes or predicts equilibria under certain specified conditions, one limitation shared by all is the failure to include the nature of the adsorbent. All the models assume that there are no specific adsorptive interactions or that chemisorption, if it occurs at all, is too small to affect overall equilibria as the models describe them. The interaction of organic substances and surface functional groups can markedly change the pattern of distribution of adsorption energies for different solutes. Molecular sieving effect due to differing pore size distributions in various adsorbents and pore blockage can be more significant than physical adsorption forces reflected in solubility effects in determining adsorption behavior

(95). Limitations of predictive and descriptive models, then, can be attributed to assumptions made during their development. Assumptions common to almost all models are a homogeneous, thermodynamically inert surface, equal competition for sites, and adsorption determined by physical forces only, that is, complete reversibility.


2.8.1 Descriptive Models


Two descriptive models include a model proposed by Fritz and Schluender and an extension of the RadkePrausnitz three-parameter model (77, 96). The Fritz and








83

Schluender model involves a general empirical equation of the form



b.
a. C 10
q9 e,i
e,i n b..2-13) c. + a.. C 1
10 j= 13 e,j


where i and j refer to different solutes; n is the total number of solutes; and aio, aij, bio, and c.io are constants determined by the best fit of multi-solute experimental adsorption data (77). The equation includes the Langmuir competitive adsorption equation, the Jaeger and Erdoes relationship, a Radke and Prausnitz isotherm, the Freundlich equation, and the Langmuir single-solute equation when the empirical constants are assigned certain values. The model is not predictive because multi-solute adsorption data are required to estimate model parameters.

An extension of a previously described single-solute three-parameter model was tested for multi-solute data predictability (96). The model is described by



(Ai.)(C i/n.)

e,i n 1 (2-14)
1 + I B. (C ./n.)
j=l 3 e,j








84

where A, B, and 8 are parameters determined from singlesolute isotherms, and the interaction terms, .i and nj, are determined from multi-solute system data.

Both these models adequately describe multi-solute

equilibria data for the concentration ranges of the specific solutes studied. The extended three-parameter model has been used in one dynamic column simulation model (96). However, because of the need for multi-solute data in order to estimate certain model parameters, these equations are valid only for the data sets to which they were applied. The model parameters are not readily correlated to any solute properties. Consequently, the models are not applicable, in a general way, to a wide variety of solute types.


2.8.2 Langmuir Competitive Model


This model extends the Langmuir single-solute equation to a competitive, multi-solute model in the form




Q. b. C
1 1 e,i
qe h (2-15)
1+ b. C
j=l 3 e,j


where i and j refer to the solutes in solution and the remaining terms are as defined previously for the Langmuir isotherm. The constants, Q and b, are determined from single-solute data. The model assumes equal competition








85

for sites, complete reversibility, and no interaction among adsorbed solutes. This model was applied to the adsorption of alkyl phenols in bisolute systems (59). Single-solute adsorpton data were described by the Langmuir equation, but the competitive model did not adequately describe the bisolute data. Other studies also showed that the Langmuir competitive model did not perform as well as several other models (30, 74-76). The thermodynamic basis of the model was valid only when the constant, Q, was identical for all solutes (equal competition).


2.8.3 Semicompetitive Langmuir Model


The Langmuir competitive model was modified to describe adsorption from a bisolute system when a portion of the adsorption occurs without competition (75). The modification was based on the hypothesis that such adsorption occurs when the constant, Q, in the competitive model was not the same for both solutes. The model was expected to be valid only when a fraction of the adsorption occurs without competition, such as when the larger of the two solutes is unable to enter the smaller pores of the carbon or when a site is unavailable to one solute due to the chemical nature of the site. The number of adsorption sites where noncompetitive adsorption would occur was assumed to be equal to Q1-Q2 in a bisolute system. The modified Langmuir competitive equation became, for Q1 > Q2'








86


(Q1-Q2) b C e b C
S(QQ2) b1 e,l + Q21 el (2-16)
e,l 1 + blC1 1 + b 1 C e,l + b2Ce,2





QbC
= 2 2 e,2 (2-17)
e,2 1 + blCe,l + b2Ce,2


where the terms are as defined for the single-solute Langmuir equation. The noncompetitive term is the first term on the right side of equation 2-16.

In the study that developed this modification (75), the semicompetitive model gave a much better description of adsorption data for four out of five bisolute pairs than the original competitive model when Q1 Q2. When Q1 Q2 in one pair, the two models were almost identical. This was taken as evidence of noncompetitive adsorption due to surface heterogeneity on the carbon. Noncompetitive adsorption has been implied by other research (74, 76). In a study of competitive adsorption between selected fatty acids and phenolic substances, the semicompetitive model was significantly better in describing adsorption data compared to the competitive model (76).

The most significant limitation to this model is its restriction to bisolute systems. The model also assumes that unequal competition occurs only when Q1 Q2. However, unequal competition has been observed when Q1 = Q2 (67).








87

The model is limited by the ability of the single-solute Langmuir equation to describe isotherm data. Other limiting assumptions of the Langmuir development also apply. The model does illustrate the occurrence of unequal competition due to the heterogeneous nature of the carbon surface.


2.8.4 Competitive Solvophobic Interaction Theory


The Solvophobic Interaction Theory discussed as a

single-solute predictive modeling approach is being studied for application in multi-solute systems (90, 97). The approach is based on the application of several physicochemical parameters (the infrared [IR] shift parameter, a polarity parameter, and a steric parameter) with the solvophobic theory to predict the adsorption of organic homologues in multi-solute systems.

The IR shift parameter is used to measure a solute's hydrogen bonding ability. It is determined by collecting acidity and basicity IR shift values (using an arbitrary scale) for various organics and then multiplying these values by the ratio of their molecular weight to percent aqueous solubility. This composite IR shift parameter is then normalized with respect to phenol. As the value of this normalized parameter increased, observed single-solute adsorbability also increased. From this it was concluded that molar adsorbability is proportional to acidity and








88

molecular weight and inversely proportion to solubility

(97).

The polarity parameter describes the polar effect of the addition of a substituent group to a solute in a class of compounds when the parameter is correlated with singlesolute, maximum adsorption capacities. The Hammet equation is used to produce a parameter that reflects polar effects of meta- and para-substituted derivatives of benzene. For reactions involving aromatic ortho-substituents and aliphalic compounds, the Taft equation provides the polarity parameter. Both parameters are tabulated in the literature for organic solutes commonly found in natural waters (97). In general, as these parameters increase, the polarity effect of the substituent decreases and adsorption increases. This agrees with solubility correlations since a decrease in polarity generally results in a decrease in solubility.

Steric effects usually become important for aromatic

substituents and a modified Taft equation is used to define a steric parameter. Values for this parameter are also available in the literature. Generally, as the steric effect of a substituent group increases, single-solute adsorbability increases, and there is a decrease in the polarity parameter. Therefore, polarity and steric parameters correlate inversely with single-solute adsorption.




Full Text

PAGE 1

COI-IPETITIVE MULTI-SOLUTE ADSORPTION ON ACTIVATED CARBON By ROBERT STANLEY RYCZAK A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMEaTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1985

PAGE 2

This work is dedicated to my wife, Sonia. Our degree would never have become a reality without you. You never doubted me or lost confidence in me, even when I lost confidence in myself. You gave unselfishly of yourself to accommodate my every whim, need, and desire. You pushed me when I needed to be pushed and patted me on the back when I needed that, too. You spent many hours washing laboratory glassware and keeping me company through the many long nights in the lab. You spent long hours learning and performing lab analyses to free me to do other work. You found the time, usually late at night, to type my drafts. You helped me with many time-consiaming tedious calculations. You were always there for me. There could never be adequate recompense for you for the stresses, the frustrations, the sufferings, and sacrifices you endured over the last three years. I can only begin by saying thank you for your saintly patience, for always being there when I needed you, and for your love.

PAGE 3

ACKNOWLEDGEMENTS The word "acknowledgement" does not begin to repay the contributions made by those who have shared the high and low points with me during the past few years: My advisor. Dr. John Zoltek, who took me under his wing after the untimely death of my original advisor. Thank you for your constant motivation, your sage advice, your concern for me both professionally and personally, your confidence, your support of my laboratory work, and your friendship. Dr. Joseph Delfino, thank you for your excellence in teaching, your freely given support during my analytical struggles, and your openness to students. Dr. Lamar Miller, thank you for your constant optimism and moral support. Dr. Benjamin Koopman and Dr. Dinesh Shah, thank you for your encouragement and interest in me as a student. To Ann Hill, my graphic artist, thank you for your patient and generous support and especially for sharing your home with us. To Don Carrara, thank you for help with my computer encounters and for your willingness to give of yourself.

PAGE 4

To Joe Angley, Lisa Drinkwater, Kevin Topp, John Earle, Sang II Lee, Ho Kang, Chan Won Lee, Chang Won Kim, and Rick Mestan, thanks for many good conversations and friendship. To Barbara Smerage, who converted my drafts to the final product. Thank you for your long and unconventional hours to accommodate my time schedule and for your confident support Finally, to my parents. Thank you for your many years of hard work and sacrifices to give me the tools to become what I am. Thank you for your unwavering confidence, your boundless love, and for always being there. iv

PAGE 5

TABLE OF CONTENTS ACKNOWLEDGEMENTS iii LIST OF TABLES vii LIST OF FIGURES ix ABSTRACT xi CHAPTERS 1 INTRODUCTION 1 2 LITERATURE REVIEW 6 2.1 Introduction 6 2.2 The Nature of Adsorption 7 2.3 The Nature and Characteristics of Activated Carbon 16 2.4 Effects of Adsorbate Properties on Activated Carbon Adsorption 45 2.5 Effects of Aqueous Solvent System Properties on Adsorption 53 2.6 Descriptive Single-solute Equilibrium Models 63 2.7 Approaches to Predictive Singlesolute Equilibrium Models 73 2.8 Multi-solute Equilibrium Models 81 3 RESEARCH OBJECTIVES 102 4 MATERIALS AND METHODS 104 4 1 Adsorbent 104 4.2 Adsorbates 106 4.3 Solutions 109 4.4 Aqueous Solubilities 110 4.5 Single-solute Isotherms Ill 4.6 Single-solute Desorption 116 4.7 Analytical Procedures for Singlesolute Analysis 118 4.8 Multi-solute Adsorption Studies 124 v

PAGE 6

5 RESULTS AND DISCUSSION 135 5.1 Single-solute Adsorption Studies 135 5.2 Single-solute Desorption Studies 152 5.3 Competitive Adsorption 159 5.4 Sequential Solute Addition 183 6 SUMMARY AND CONCLUSIONS 18 9 7 ENGINEERING SIGNIFICANCE 191 APPENDICES A SINGLE SOLUTE DESORPTION, ADSORPTION, AND SOLUBILITY DATA 194 B SEQUENTIAL AND SIMULTANEOUS SOLUTE ADDITION DATA FOR THE 0-CRESOL/ALLYL PHENOL SOLUTE PAIR 207 C SEQUENTIAL AND SIMULTANEOUS SOLUTE ADDITION DATA FOR 2 4-DIHYDROXYACETOPHENONE (DAP)/ALLYL PHENOL SOLUTE PAIR 225 D SEQUENTIAL AND SIMULTANEOUS SOLUTE ADDITION DATA FOR O-CRESOL/2 4-DIHYDROXYACETOPHENONE (DAP) SOLUTE PAIR 2 42 E MULTI-SOLUTE SIMULTANEOUS ADDITION DATA... 259 REFERENCES 265 BIOGRAPHICAL SKETCH 276 vi

PAGE 7

— "1 LIST OF TABLES TABLE PAGE 2-1 Pore volumes and surface areas for various pore size distributions for one activated carbon (20, 28 ) 20 2-2 Raw materials that have been used to produce activated carbons 4 0 4-1 Characteristics of 12 x 40 mesh Calgon Filtrasorb 400 granular activated carbon 105 4-2 Physical and chemical properties of organic compounds studied 107 4-3 Wavelengths and concentration ranges for single-solute UV analysis 123 44 Method for sequential solute addition. Compound A is followed by compound B. 132 51 Comparison of Freundlich and Langmuir adsorption isotherm equations for single-solute studies 137 5-2 Experimental solubility data 145 5-3 Single-solute adsorption irreversibility 156 5-4 Simultaneous addition equilibrium data for ALP/OC solute pair 160 5-5 Simultaneous addition equilibrium data for ALP/DAP solute pair 162 5-6 Simultaneous addition equilibrium data for OC/DAP solute pair 164 5-7 Predicted and measured solid-phase equilibrium values for simultaneous addition of ALP/OC solute pair 168 vii

PAGE 8

5-8 Predicted and measured solid-phase equilibrium values for simultaneous addition of ALP/DAP solute pair 169 5-9 Predicted and measured solid-phase equilibrium values for simultaneous addition of OC/DAP solute pair 170 5-10 Competition terms for the improved Simplified IAS model 173 5-11 Multi-solute simultaneous addition equilibrium data and predicted values... 178 5-12 Irreversible adsorption for bisolute systems of differing equilibrium concentration ratios (mg/L:mg/L). All values in mg/g 185 viii

PAGE 9

LIST OF FIGURES FIGURE PAGE 2-1 Acidic oxygen surface functional groups.. 30 2-2 Basic oxygen surface functional groups... 34 4-1 UV absorbance wavelength scan for o-cresol 119 4-2 UV absorbance wavelength scan for allyl phenol 120 4-3 UV absorbance wavelength for 2,4dihydroxyacetophenone 121 4-4 UV absorbance wavelength scan for a-terpineol 122 45 Schematic diagram of continuous flow carbon contact column system 125 51 Single-solute adsorsption isotherms for o-cresol 138 5-2 Single-solute adsorption isotherms for allyl phenol 139 5-3 Single-solute adsorption isotherms for 2 4-dihydroxyacetophenone 140 5-4 Single-solute adsorption isotherms for a-terpineol 141 5-5 Single-solute adsorption isotherms for all four solutes using <200 mesh carbon. 143 5-6 Single-solute adsorption isotherms for all four solutes using 100 x 140 mesh carbon 144 5-7 Single-solute adsorption and desorption isotherms for o-cresol 154 ix

PAGE 10

5-8 Single-solute adsorption and desorption isotherms for allyl phenol 155 5-9 Correlation between the solubility factor and the competition factor, n-j^. • • 175 5-10 Correlation between the solubility factor and the competition factor, n-,.... 176 X

PAGE 11

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy COMPETITIVE MULTI-SOLUTE ADSORPTION ON ACTIVATED CARBON by Robert Stanley Ryczak December 19 85 Chairman: John Zoltek, Jr. Major Department: Environmental Engineering Sciences Fundamental to the successful development and use of dynamic, multi-solute carbon adsorption models is an accurate multi-solute equilibrium model. Current models often fail to adequately describe experimental data. Interest in describing and predicting the adsorption of specific organic solutes from dilute aqueous solutions has led to the need to better understand nonideal mechanisms affecting adsorption. An experimental program was conducted to examine the effects of unequal competition and irreversible adsorption in bisolute and multi-solute systems. Carbon particle size significantly affected single-solute adsorption isotherm parameters, which are the most sensitive component of any multi-solute equilibrium model. xi

PAGE 12

Single-solute studies indicated substantial irreversible adsorption for all four of the organic solutes studied. Unequal competition for adsorption sites due to carbon surface heterogeneity and solute affinity for carbon was observed in all bisolute combinations studied. Significant irreversible adsorption was also observed. The application of several current multi-solute equilibrium models failed to provide an accurate description of the data. A recently improved, simplified ideal adsorbed solution (IAS) theory model was superior to all other models tested when used with the unequal competition modification. The model is predictive through correlations between solute solubilities and competition factors. The model has not been expanded beyond bisolute systems. Much more work is needed to test, validate, and expand this improved model to account for unequal competition and irreversible adsorption in the ultimate design process. xii

PAGE 13

CHAPTER 1 INTRODUCTION The surface and ground water resources of today's industrial societies were once a reliable source of supply for domestic and agricultural consumption. Ground water supplies were consistently of good quality and free of insidious constituents that would threaten man's health. The large industrial expansion and rapid growth societies have experienced in the last 50 years have seen a concomitant increase in the variety and volume of chemicals manufactured, used, and disposed of in the environment. Many of these new chemicals were refractory and also capable of producing adverse health effects in man. A large number of these chemicals have exhibited toxic, carcinogenic, mutagenic, or teratogenic properties. Lack of awareness has resulted in the casual disposal of these chemicals into soil and water resources, often through industrial and municipal wastewater streams. As knowledge of the potential adverse health effects of the chemicals grew, concern over uncontrolled exposure through water supplies also grew. National concern, reflected in a proliferation of federal, state, and local legislation, developed when improved analytical techniques 1

PAGE 14

2 revealed the presence of hundreds of these chemicals in water supplies throughout the country. A list of hazardous or potentially hazardous compounds was promulgated by the Environmental Protection Agency (EPA). The EPA also placed maximum acceptable limits on their concentration in drinking waters and wastewaters. Activated carbon adsorption has evolved as one of the most effective and dependable treatment technologies suited to the task of removing the broad spectrum of dissolved organic compounds from waters and wastewaters. Empirical research has generally demonstrated the advantages and limitations of the many variations of applied adsorption technology. The broad applicability of activated carbon adsorption was most recently emphasized when serious consideration was given to mandating granular activated carbon treatment nationwide as a control strategy for organic precursors in trihalomethane formation. The rapid development in empirical applications of activated carbon technology was not accompanied by a similar growth in understanding the fundamental principles underlying the technology. This imbalance is readily noticed in the design approach for activated carbon systems. Laboratory experimentation followed by extensive pilot-scale testing prior to full-scale design is the norm for every activated carbon application in water and wastewater treatment. Little is known of the forces and mechanisms of

PAGE 15

3 adsorption of organic compounds from aqueous solution by activated carbon. Rational approaches to carbon system design have only just begun to emerge. Simple descriptive models for singlesolute equilibrium adsorption systems have been developed. Some have a solid theoretical basis, and others are empirically formed. Elucidation of the structure and nature of the activated carbon surface has begun but is incomplete. The interactions between solute, solvent, and sorbent still remain to be clearly defined. A few studies of competitive adsorption in bisolute systems have resulted in descriptive, multi-solute equilibrium models. As basic knowledge of the adsorption process increased, descriptive equilibrium modeling has improved. Models are currently being studied which attempt to describe the fundamental thermodynamics of specific adsorption interactions. The crucial role played by equilibrium models lies in the development of predictive dynamic models for full-scale system design and analysis. The ultimate goal of fundamental research into the nature of activated carbon adsorptions is the development of a comprehensive, theoretically based design approach that can predict dynamic column behavior for individual solutes in the complex matrices of the water and wastewater streams to be treated. The development of predictive models has generally been limited to simple matrices of one or two compounds. Essential in the development of such dynamic

PAGE 16

4 predictive models is the inclusion of a general equilibrium predictive model. Current multi-solute equilibrium models are limited by inadequate descriptions of adsorption mechanisms. The failure of existing models lies in the assumptions of equal competition and reversible adsorption. Such assumptions are based on a thermodynamically inert, homogeneous sorbent surface. Activated carbon is known to be heterogeneous with adsorption characteristics that vary with each commercially available carbon. Such assumptions have been made to simplify model development. However, irreversible adsorption and unequal competition can significantly influence final equilibrium conditions. The consequences of carbon surface heterogeneity are best observed in operating granular activated carbon adsorbers where influent and effluent concentrations usually vary considerably with time. Irreversible adsorption is especially highlighted in dynamic systems. For example, current predictive models indicate that the introduction of a new solute should not compromise effluent quality. The new solute should be safely adsorbed with no detectable trace in the effluent. However, if significant irreversible adsorption had occurred with previous solutes, then the new compound could pass right through the adsorbent much quicker than predicted. Consequently, the modification of existing models to account for carbon surface heterogeneity should improve predictive accuracy. This research was directed at

PAGE 17

examining competitive and irreversible adsorption in order to increase fundamental understanding of the adsorptive process, compare existing model performance, and test a model recently modified to include unequal competition.

PAGE 18

CHAPTER 2 LITERATURE REVIEW 2 1 Introduction The literature addressing adsorption and its application to the removal of organic solutes from aqueous systems by activated carbon is widely dispersed internationally among a variety of disciplines. Sources of the literature include many diverse and occasionally uncommon technical journals, a number of books, and many theses and dissertations. No single source of information describing the state-of-the-art of activated carbon adsorption exists. Little communication is evident in the literature among scientists and engineers studying and applying carbon adsorption. Discussion of the theory of adsorption from solution can be found in several basic references (1-7). General information on adsorption of organic compounds from aqueous solutions by activated carbon is presented in several books and review articles (8-19). More specific information can be found through the extensive lists of references in the above works. This chapter begins with a brief summary of the nature and forces of adsorption. Activated carbon as a successful adsorbent is discussed in Section 2.3. The effects of the

PAGE 19

7 nature of the adsorbate and solvent on carbon adsorption are reviewed in Sections 2.4 and 2.5. Several single-solute equilibrium models are reviewed in Section 2.6, and various multi-solute equilibrium models are discussed in Section 2.7. The development of predictive, multi-solute adsorption models is the goal of many current research efforts. The evolution of activated carbon adsorption as an effective, dependable technology in the removal of a wide spectrum of organic contaminants from aqueous systems has stimulated efforts to improve the ability to describe and predict adsorption in multi-solute systems. Such ability is necessary in order to improve process modeling and design. 2 2 The Nature of Adsorption Adsorption has been defined as the accumulation of substances at a surface or interface, primarily as a result of forces active at or near the boundaries of the surface (8, 10, 20). The processes resulting in the adsorption of many organic solutes from aqueous solutions by activated carbon involve complex interactions of these physical and chemical forces. Much research has been conducted in an effort to understand and describe these surface chemistry phenomena. As the chemistry of carbon adsorption is better understood, the activation processes to produce activated carbon can be modified to increase carbon adsorption capacity and specificity. Subsequently, mathematical models describing

PAGE 20

8 and predicting adsorption phenomena can be greatly improved. Ultimately, pilot and full-scale plant design procedures can then be modified to optimize the adsorption process for particular applications. The various forces existing between a solid surface and nearby solute molecules originate in the electromagnetic interactions of nuclei and electrons. The forces of adsorption can be classified into four types: physical, chemical, ion exchange, and specific adsorption. The balancing of these surface forces minimizes the free energy of the surface and results in a state of dynamic equilibrium in which a solute is concentrated on the surface of a solid phase, such as activated carbon. Physical adsorption results from the action of van der Waals forces and classical electrostatic forces (2, 7, 9). The van der Waals forces include London dispersion and repulsion forces and are always present. Electrostatic interactions involve polarization, dipole and quadrapole interactions and are significant only when the surface is ionic or polarizable. In physical adsorption, the electron distributions of both adsorbent and adsorbate experience some distortion, but there is no exchange or sharing of electrons (20). The net adsorption force, consequently, is weak, and adsorbate molecules are not fixed to a specific site on the surface but can move laterally from site to site or readily desorb. This type of reversible adsorption is

PAGE 21

9 sometimes referred to as "ideal adsorption. The heat of adsorption is low for physical adsorption, on the order of a few hundred calories per mole of adsorbate. Physical adsorption is nonspecific, can occur in monolayer or multilayer configurations, is relatively rapid and reversible, and requires little or no activation energies (7). Specie dissociation does not occur during physical adsorption. The London dispersion-repulsion forces are relatively weak, electrostatic forces which exert a relatively longrange influence compared to the very short-range, wave mechanical chemical bond force (2). London forces account for the observed general attraction between atoms and molecules that are not charged and do not possess permanent dipole or quadrapole moments. The attractive potential arises from the coupling of instantaneously induced dipoles and quadrapoles. These arise from the fact that even neutral atoms consist of systems of oscillating charges due to the presence of positive nuclei and negative electrons. London theorized a quantum mechanical perturbation of electron distribution caused by the continuous motion of electrons in atoms and molecules that was manifested as rapidly fluctuating temporary dipole and quadrapole moments (20). These fluctuating moments in solute molecules near a solid surface can affect the electron distribution in surface molecules and induce temporary dipoles and quadrapoles. The surface can also induce such moments in nearby

PAGE 22

10 solute molecules as well. The attractive potential arising from dispersion forces then is the sum of dipole-dipole dipole-quadrapole, and quadrapole-quadrapole interactions (7). Dipole-dipole interactions are dominant with shortrange effects inversely proportional to the sixth power of distance. Dipole-quadrapole and quadrapole-quadrapole interactions are inversely proportional to the eighth and twelfth powers of distance, respectively. The short-range repulsive force is considered to be inversely proportional to the twelfth power of distance and, consequently, becomes significant only when the electron orbitals of the surface and adsorbate molecules interpenetrate. London forces are sometimes considered long-range forces since oscillating fields are mutually propagated between atoms and molecules. These forces operate between polar and nonpolar molecules and covalent, metallic, and ionic solid surfaces (20). Classical electrostatic interactions of physical adsorption are more specific than London forces (2, 7). Several different surf ace-adsorbate interactions are possible. If the solid surface possesses an electrical field, an attractive potential can be developed by the polarization of molecules within the field's influence. These attractive potentials depend on the surface electric field strength and the polarizability of the solute molecule. Solute molecules possessing permanent dipole or quadrapole moments will also interact with the surface field to an extent dependent on

PAGE 23

11 the field characteristics and the magnitude of the permanent moments. The surface and adsorbate molecules can also contain hydrogen bond donor and acceptor groups (1). For example, hydrogen bonding by phenolic protons with oxygen in a surface functional group on activated carbon has been offered as one adsorption mechanism (21). However, if bonding groups on the surface or adsorbate molecule have a strong affinity for water, hydrogen bond adsorption will not be significant (1). Solvation with water will prevent close approach by solute molecules, and solvation of strongly ionized groups may screen hydrogen bonding groups. Such electrostatic interactions are not significant where the solid surface is covalent and a strong enough electrical field does not exist and cannot be induced by polar molecules. However, the surfaces of activated carbon are sufficiently heterogeneous to participate in all of the discussed interactions. Chemical adsorption, or chemisorption, is a much more specific adsorption mechanism than physical adsorption. Chemisorption occurs when elecrons are exchanged or shared between adsorbate molecules and the surface of the adsorbent so that chemical reaction actually occurs. Consequently, chemisorption exhibits all the characteristics of a chemical bond. These characeristics include irreversibility, substantial activation energy requirements, localized reactions at specific sites on the adsorbent surface, and immobility

PAGE 24

12 of adsorbed species. The chemisorption bond is very short and very strong, on the order of 10 to 100 kilocalories per mole compared to the few hundred calories per mole observed in physical adsorption (1). Chemisorption exhibits specificity for adsorbate molecules while physical adsorption does not. Chemisorption results in only monolayer adsorption. Subsequent adsorbed layers may form but result from physical adsorption forces. Physical adsorption, a thermodynamically exothermic process, is usually the significant adsorption type at temperatures well below 0C. Sufficient activation energies may not be available for any appreciable chemisorption to occur. As temperature is increased, physical adsorption decreases since the free energy driving forces decrease. Chemisorption, however, becomes more significant as more energy is available to overcome activation barriers, and reaction rates increase. There will be a temperature, however, beyond which chemisorption decreases, since chemisorption is also a net exothermic process. Chemisorption occurs at the energetically most favored sites with the extent of binding dependent on the energies of adhesion and activation. The covalent bond thus formed can impart an ionic character to the adsorbed molecule and the overall surface. This ionic character depends on the relative electronegativities of the bonding atoms. A dipole moment is usually formed with the comcomitant electric double layer also forming at the surface.

PAGE 25

13 The effect of the electric double layer is to reduce the strengths of subsequently formed bonds and to impede mass transfer across the surface film to the interior sites of the carbon surface. Chemisorption alters the energetics of the surface and can affect the adsorption tendency of neighboring sites. When a surface is almost completely covered, lateral interactions among adsorbed species may become important. Ion exchange adsorption results from charged functional groups on the adsorbent surface attracting oppositely charged adsorbate ions. An originally neutral adsorbent may also gain a surface electrical change by adsorption of an ionic solute molecule (8). This process may result in the release, or exchange, of a different ion which passes into solution. In exchange adsorption the sorbent remains electroneutral or retains the original charge. Hydrolytic adsorption, or the simultaneous adsorption of a hydated proton with adsorption of an anion, may also occur (8). Solids such as silica or carbon, which are normally negatively charged, readily adsorb cationic dyes and surfactants (1). Sometimes ions carrying the same charge as the surface are adsorbed through van der Waals and other forces. Such ion exchange adsorption is usually independent of temperature, since surface charge varies little with temperature. Ion exchange adsorption does occur by activated carbon but

PAGE 26

14 is not thought to contribute as much as physical or chemical adsorption to the net adsorption result. Specific adsorption describes the result of specific interactions between adsorbate molecules and surface functional groups on the activated carbon. These specific interactions can exhibit a wide range of adsorption energies from low-energy physical adsorption to high-energy chemisorption. An example of specific adsorption is the interaction of aromatic hydroxyl and nitro-substituted compounds with activated carbon (21). Adsorption of phenols occurs through the formation of strong donor-acceptor complexes with surface carbonyl oxygen groups, in addition to hydrogen bonding. The oxygen group dipole moment determines the strength of the donor-acceptor complex formed. The carbonyl oxygens of the carbon surface act as electron donors, and the aromatic rings of the adsorbate molecule act as acceptors. Such an adsorption interaction exhibits reversibility, and the adsorbate molecule adsorbs in a planar orientation. Once the surface carbonyl sites are exhausted, adsorption continues by complexation with the rings of the basal planes of the carbon structure. Nitro group substitution enhances the donor-acceptor interaction by acting as an electron withdrawing group. Enhanced adsorption has also been observed for p-chlorophenol and p-cresol on activated carbon, further demonstrating specific adsorption interactions (22).

PAGE 27

15 The preceding discussion described the nature of adsorption in terms of adsorbent-adsorbate affinity. However, adsorbate-solvent interactions also greatly influence the extent of adsorption. The reduction of interfacial tension through adsorption reduces the overall interfacial free energy. Surfactants have long been known to lower interfacial tension significantly (1, 2). The reduction of interfacial free energy results from the loss of solvent-solid interfacial area which lowers that interfacial free energy. The creation of a solute-solid interfacial area and a solute-solvent interfacial area through accumulation of a surfactant on the surface of the solid results in an increase in the system interfacial energy. However, the magnitude of the increase in the system's interfacial energy is less than the magnitude of the decrease, and the net result is an overall decrease in total interfacial free energy. Such an energy balance promotes the concentration of surfactant molecules on a surface or at a interface. The extent of adsorption is also related to the degree of hydrophobicity of the solute, expressed as solubility. Increasing hydrophobicity decreases solubility and generally increases the extent of adsorption (20). Lundelius' rule states that the greater the solubility, the stronger the solute-solvent bond and the smaller the extent of adsorption. This qualitative relationship generally holds true;

PAGE 28

16 however, there are many exceptions. A special case of Lundelius' rule is Traube s rule which contends that adsorption from an aqueous solution increases as a homologous series is ascended. As a homologous series is ascended, the increasing number of carbons and chain length makes the solute less soluble. The removal of more hydrophobic solutes allows more water-water bonds to reform. Conversely, the greater the solubility, the stronger the solute-solvent bond and the smaller the extent of adsorption. A general outline of the nature of the adsorption process has been presented. These general principles apply to most adsorption systems with various adsorbents, solvents, and adsorbates. Adsorption from aqueous solution by activated carbon is just one application of this separation process. To describe the nature and extent of adsorption with activated carbon, other relevant factors must be examined, including the nature of activated carbon and the effect on adsorption of varying carbon characteristics, the nature of the solute, and the effects of the solvent, water. 2 3 The Nature and Characteristics of Activated Carbon 2.3.1 Surface Area Activated carbon refers to a large group of carbonaceous materials prepared in a manner to exhibit extremely

PAGE 29

17 large surface areas, high degrees of porosity, and unique surface reactivities. It can be envisioned as a solid foam. Such characteristics make activated carbons an effective adsorbent for a broad spectrum of organic solutes in water and wastewater. Activated carbon has a long history of use in hundreds of separation applications which consume hundred of millions of pounds of the material annually (10, 23). Two general classes of activated carbon include gas-adsorbent carbon and liquid-phase carbon. Gas-adsorbent carbons for chemical recovery or impurity removal are characterized by a rather homogeneous system of 0.5 to 5.0 nm (5 o to 50 A) radius micropores and surface areas exceeding 2000 2 m /g (23). Liquid phase carbons, in contrast, exhibit a wide variety of surface areas, pore sizes, and pore size distributions The surface area of activated carbon is essentially intraparticle. Commercially available carbons for water and wastewater applications have specific surface areas ranging from 480 to 1800 m^/g (24). When specific surface areas 2 exceed 2000 m /g, the pore sizes become too small (less than 1 nm) to effectively adsorb aqueous organic solutes (12). Surface areas are determined through a variety of methods. Adsorption of nitrogen gas described by the Brunauer, Emmett, and Teller (BET) isotherm equation has been the most common method (7, 10). The surface areas reported by carbon manufacturers are determined by the N„, BET method (-22C).

PAGE 30

18 Carbon dioxide adsorption at 25 C as described by the Polanyi-Dubinin isotherm equation is thought to be a better measure of the total surface area of microporous carbons, since reversible capillary condensation of N2 at very low relative vapor pressures before monolayer adsorption is complete may yield erroneously high specific surface areas (25). Other methods to measure surface area in attempts to characterize and compare activated carbons have included adsorption of phenol, p-nitrophenol methylene blue, rhodamine B, oxalic, and succinic acids from aqueous solutions; adsorption of stearic acid from organic solvents; retention of ethylene glycol against vacuum after wetting with the liquid; and selective adsorption from phenolbenzene and ethylene glycol-water mixtures (26). Since only that portion of surface area and pore volume which is in the pores of proper size will be available for adsorption, the use of surface area measurements or pore volumes is of little value in describing the possible effectiveness of a carbon. Equal weights of carbons prepared from different raw materials by different activation methods may have the same total BET, or CO^, Polanyi-Dubinin surface areas, yet will exhibit very different adsorption behavior in the same aqueous solution of organic compounds. Other carbons with significantly different total surface areas performed equally well when treating domestic wastewaters (24). An adsorption study of dyes with molecular weights from 350 to

PAGE 31

19 1370 Daltons indicated that the carbon with the lower surface area performed much better than another carbon with a higher surface area (24). One granular activated carbon's specific surface area as determined by the BET method with N2 was 950 to 1050 m^/g (27). Since this carbon was used in adsorption experiments with nonyl and dinonyl phenol ethoxylates, a better measurement of available surface area was desired. Methylene blue was selected to better represent large organic molecules, and the specific surface 2 area measured was 68.3 m /g for the same granular carbon. However, when compared to a sodium-montmorillonite clay with a methylene blue specific surface area of 711.5 m^/g, the granular carbon was much more effective in removing the phenol ethoxylates (27). Consequently, it is the nature of the carbon surface and not the magnitude of the surface area which determines its adsorption characteristics. 2.3.2 Pore Sizes and Distribution Pore sizes and their relative distribution within a carbon particle are two of the most significant factors determining adsorption capacities and kinetics. Pores are formed during the activation process and exhibit a trimodal distribution in activated carbon (8, 20, 23, 28). Table 2-1 lists two definitions of pore size distributions and corresponding pore volumes and surface areas for one activated carbon (20, 28). Approximately 40 percent of the

PAGE 32

20 TABLE 2-1. Pore volumes and surface areas for various pore size distributions for one activated carbon (20, 28). Pore Type and (nm) Diameter Pore Yolurne (cm /g) Specif ic, Surf ace Area (m^/g) Micropore : <2 Mesospore : 2-60 0.24 Macropore : 60-1000 0 .25 Total 0.81 950(N2,BET) Micropore : 3.6-4 0.15-0.50 >95% Transitional : 3.2-200 to 400 0.02-0.10 20-70 Developed Transitional : 3.2-200 to 400 0.7 200-450 Macropore : 1000-4000 0.2-0.8 0.5-2 o Note: 1 nm = 10 Angstroms (A).

PAGE 33

21 total pore volume consists of micropores. The specific surface area contained in the micropores usually exceeds 95 percent of the total N2 BET surface area (8, 21, 29). The pore pattern in a carbon particle is thought to be arranged such that very few micropores actually lead directly to the outer surface (8, 28). The macropores are thought to lead from the particle surface into the interior. Transitional pores then branch off from the macropores, and the micropores, in turn, branch off from the transitional pores. The consequences of this structural arrangement on adsorption capacities and rates are clear. Exclusion of compounds due to physical size limitations, reduced pore diffusion rates due to stearic hindrances, van der Waals forces, and electrostatic interactions in channels approaching molecular dimensions as well as pore blockage will all significantly influence the extent of site competition. Most of the adsorption occurs in the micropores, unless pore sizes or blocked transitional and micropores physically exclude a molecule. Adsorption can occur on the surfaces of transitional pores, although this size range primarily serves to connect micropores with the macropores. Transitional pore adsorption is thought to contribute to the adsorptiondesorption hysteresis effect which many carbons exhibit (30). Macropores simply serve as bulk fluid phase conduits where solute mass transport from the bulk phase to the interior surfaces of the carbon occurs. The iodine number

PAGE 34

22 has been used to measure micropore surface areas while the molasses number has been used for transitional pore surface area estimates (31). 2.3.3 Carbon Structure The molecular and crystalline structure of activated carbon is a major factor in determining the types of surface functional groups which can exist. These surface functional groups contribute significantly to the absorptive behavior of an activated carbon. Although very little work has been done on activated carbons directly, the structure of activated carbons has been described through studies of carbon black. Chemically there appears to be no difference between the two substances. The only physical difference appears to be in internal surface areas (28). Two review articles on activated carbon surface chemistry list the seminal references for the derivation of the structure of activated carbon from studies of graphite and carbon black (32, 33). Three forms of carbon have been defined: diamond, graphite, and amorphous carbon. Amorphous carbon has been renamed microcrystalline carbon to reflect a better understanding of that substance's structure. There are several forms of microcrystalline carbon including activated carbon, carbon blacks, carbon brushes, cokes, and carbon electrodes. It is the starting materials, the preparation process, the very large specific surface area, and surface

PAGE 35

23 functional groups that distinguish activated carbon from other microcrystalline forms. The development of the structural model begins with the structure of ideal graphite (8, 32, 33). Carbon atoms form flat hexagonal ring structures that are joined to form a two-dimensional layer. The combination of covalent and resonant bonds gives each carbon-carbon bond a one-third double bond character. The ideal graphite crystal is completed by the addition of an infinite system of parallel layers of these fused hexagons. The layers are kept parallel and approximately 0.33 nm apart by relatively weak van der Waals forces. Microcrystalline carbon differs from graphite by consisting of small, elementary crystallites instead of homogeneous, parallel, infinite two-dimensional planes. The crystallites are composed of layers of hexagonal carbon rings, but the layers are not perfectly parallel. The planes in a microcrystallite are estimated to be 2-5 nm in diameter (8, 28). About three of these planes form the microcrystallite, giving a height of about 0.9-1.21 nm, although as many as 30 planes can be present. The planes in the microcrystallite are angularly displaced in a random manner and overlap one another irregularly. This irregular geometry is accompanied by the formation of interior vacancies within the microcrystallite that, in activated carbon, are often filled with impurities, some of which are intentionally added to catalyze the carbonization

PAGE 36

24 and oxidation steps in the activation process. In addition, the microcrystallites may contain "unorganized" carbon that participates in tetrahedral bonding between tilted layers resulting in layer cross-linkage. The edges of the microcrystallites are high energy sites where noncarbon atoms are always present in differing amounts. These foreign atoms may bond to the edge of the crystallite to form functional groups or may actually be incorporated into the fused ring structure to form "heterocyclic" ring systems. The microcrystallites are interconnected to form the carbon matrix by tetrahedrally bonded carbon atoms, common hexagonal ring layers, cross-linked carbon hexagons, and functional groups at the edges of the crystallites. The microcrystallites are interconnected to form two types of structures: a graphitizing or a nongraphitizing structure. A graphitizing structure is normally considered to be a soft structure formed at temperatures exceeding 1000C. That is, at high temperatures the unorganized (nonplanar) carbon is consumed in expanding the microcrystallite graphite layers both in diameter and height and orienting the larger crystallites into a more ordered, or organized, graphitelike parallel arrangement. A nongraphitizing carbon does not form a three-dimensional graphitic structure, even at temperatures higher than 3000 C (8). This implies considerable cross-linking between microcrystallites. Such strong cross-linking is promoted by the presence of oxygen or by a

PAGE 37

lack of hydrogen in the raw material. Activated carbon is classified as a nongraphitizing material. There are some structural features of graphite formed by the burning out of unorganized carbon, but throughout the entire volume of material activated carbon is considered to be structurally heterogeneous with a much greater pore volume and surface area than graphitizing carbons. The heterogeneity in the physical structure of activated carbon greatly influences adsorptive capacities and adsorption kinetics. 2.3.4 Surface Chemistry Adsorptive behavior in activated carbon is determined not only by its porous, physical structure but also by the chemical nature of its surface. The complexities of the carbon surface are reflected in the diversity of adsorption reaction mechanisms for organic compounds in aqueous systems. The presence of edges on the broken graphite planes shifts electron distributions in the graphite ring structures (32). As a result, unpaired electrons and permanent and temporary dipole and quadrapole moments appear, altering adsorption properties for polar or polarizable substances. The presence of noncarbon atoms in the heterocyclic ring structures also alters adsorption properties. Chemically bonded elements on the edges of the microcrystallites, such as hydrogen and oxygen, form reactive chemisorption sites. Inorganic impurities, both on

PAGE 38

26 the surface and in microcrystallite vacancies, significantly influence the adsorption of electrolytes and nonelectrolytes from solution. All of the variations in the chemical nature of the carbon surface are, to a great degree, controllable by the choice of raw material as well as activation conditions, especially activation temperatures (32). The details of the surface chemistry of activated carbon are presented in several specific and general review articles which also contain extensive lists of references (8, 9, 17, 21, 25, 26, 28, 32-37). Most of the carbon surface can be described by two distinct regions (32). The first region includes the planar surfaces of the microcrystallites Such portions of the surface appear to be relatively uniform in nature and not likely to contain any attached functional groups. This homogeneity is attributed to the involvement of carbon electrons in covalent bonding with neighboring carbon atoms. Few vacancies in the ring structures are expected, and primarily van der Waals forces would be responsible for most adsorption interactions. However, interactions between the ir-bonds of the planar ring structures and certain organic solutes can also result in adsorption. Such a n-bond adsorption mechanism has been suggested for the adsorption of phenol (34). The bulk of the surface area of carbon particle is found in the micropores. Since they are formed in the activation process by the removal of

PAGE 39

27 interplanar, unorganized carbon, most of the total particle surface area is thought to be of the planar surface type (32). The second major type of surface area is that provided by the edges of the microcrystallites and is quite different from the planar surface areas. A wide variety of functional groups and vacancies characterize this type of surface. Since carbon atoms at the edges of the basal planes are not completely surrounded by other carbon atoms, the electron distributions are not completely balanced by Tt-bond interactions, and site reactivity is much higher. These "free valances" at the edges of the microcrystallite planes readily react with, or chemisorb, oxygen, hydrogen, and, to a lesser extent, sulfur, nitrogen, and chlorine (8). The relative amounts of noncarbonaceous material in an activated carbon varies significantly depending on the raw material selected and the activation process used. Oxygen constitutes 2 to 2 5 percent, by weight, of activated carbon. Hydrogen is present in amounts ranging from 8 to 19 times that of oxygen, on a molar basis, for carbons containing 0.94 to 2.25 percent oxygen by weight (32). Inorganic matter, both originating in the raw material and added during the activation process, can be present in amounts up to 4 to 5 percent, by weight, of the activated carbon (10). Although much of this matter is removed after activation by various acid elution processes, leaving less than 1 percent

PAGE 40

28 ash in the commercial product, even small amounts of ash can significantly affect adsorption processes (28). Surface compounds formed with oxygen are the most important determinants of the carbon surface chemistry because of the high reactivity of such compounds and their significant presence, evidenced by the oxygen weight fraction of carbon. Various surface oxide groups exhibit an acid-base chemistry that determines the classification of a carbon as acidic or basic (38). Acidic carbons are defined as carbons that lower the pH of neutral or alkaline distilled water and that are relatively hydrophilic. Basic carbons raise the pH of neutral or acidic distilled water and are relatively hydrophobic (9). Oxygen is added to carbon in four principal ways (32). Oxygen can be present in the raw material, as a carbohydrate, for example. Oxygen can be chemisorbed on the surface of carbon during the activation process or chemisorbed on the activated carbon at room temperature. Finally, treatment of activated carbon with chemical oxidizing agents can also result in oxygen chemisorption. Oxygen complexes can be removed from the carbon surface by degassing at very high temperatures. The forms of the degassed oxygen are carbon monoxide, carbon dioxide, and water. The form of the removed oxygen can indicate the type of surface oxides that were present on the carbon ( 35 )

PAGE 41

29 Acidic surface oxides develop when the carbon is exposed to oxygen at activation temperatures below 4 00-500C or by the action of aqueous oxidizing solutions, such as acidified potassium permanganate, nitric acid, mixtures of nitric and sulfuric acids, hypochlorous acid, sodium hypochlorite, and ammonium persulfate (20, 25, 28, 38). Some of the known acidic surface oxides include quinone carbonyl groups, cyclicperoxide, carboxylic anhydride groups, phenolic hydroxyl groups, normal lactone groups, carboxyl, and fluorescein lactone groups (8, 28, 32, 38, 39) Figure 2-1 illustrates some of these groups as they might appear on the edges of the microcrystallite basal planes. Carboxylic, lactone, and phenolic groups are thought to be the most common (35). Such surface oxides result in a negative surface potential for the carbon. The specific adsorption mechanisms for each type of group with various classes of organics have not been clearly defined. A few studies have examined the effect of acidic surface oxides on the adsorption of a limited number of organic solutes in single and bisolute systems (21, 33, 34, 40) The lack of information on the chemical surface characteristics of each carbon studied poses great difficulty in mechanism definition. Aromatic compounds, especially phenol and its derivatives, are the most commonly studied class of organics for the effect of surface oxides on adsorption. An early study with six different commercial

PAGE 42

30

PAGE 43

31 activated carbons (41) revealed that acidic oxygen surface groups tended to reduce the capacity of the carbon surfaces for adsorption of metanil yellow from aqueous solution but did not affect the adsorption of methylene blue. The behavior was attributed to electrostatic repulsion between the anionic metanil yellow and the surface oxide groups. Subsequent studies investigated the effect of acidic surface oxides on the adsorption of phenol, nitrobenzene, and sodium benzenesulf onate (33, 34, 40). The alteration of the carbon surface by chemical treatment, especially with respect to the formation and removal of surface oxides, significantly changed the adsorptive capacity of the carbons studied. In all of the experiments oxidation strongly reduced the adsorption capacity for all adsorbates, reduction of the oxidized carbon slightly increased the capacity, and high temperature outgassing of the oxidized carbon restored the original capacity. Two possible explanations for the observed inhibition of adsorption were the removal of electrons from the tt electron system of the carbon by the chemisorbed oxygen and the bonding of water molecules to the acidic oxide functional groups. Since phenol was thought to adsorb through interaction of its ir-electron system with the TT bonds of the graphitic carbon planes, removal of available electrons would reduce phenol adsorption. The adsorption of water molecules would result in the formation of complexes of associated water within the carbon pores and could

PAGE 44

32 prevent mass transfer of organic molecules to a large portion of the active intraparticle surface area. These, and other studies (25, 36), suggest that carboxylic and lactone-type surface oxides reduce the adsorption of aromatics. These are the two principal types of acidic surface oxides formed by chemical oxidation or by activation temperatures below 4 00C. Carbonyl groups in the form of quinone and hydroquinone are formed at activation temperatures about 4 00C and improve the adsorption of aromatics through the formation of an electron donor-acceptor complex with the surface carbonyl group. Adsorption of aromatics can also occur, however, by Tf-bond interaction with the rings of the basal planes (35). The adsorption of nonpolar aliphatic compounds is also hindered by acidic surface oxides (35). These hydrophobic compounds preferentially adsorb on carbons free of acidic surface oxides. Again, the formation of water complexes through hydrogen bonding with the surface oxides blocks the surface from these hydrophobic aliphatics Basic surface oxides are formed when a carbon surface is first freed from all surface compounds by heating in a vacuum or an inert atmosphere to 800-1000C and then brought into contact with oxygen after being cooled to temperatures as low as -40C (38). Activated carbons with basic surface oxides as the predominant type of oxygen functional groups are not common. Basic surface oxides may only cover 2

PAGE 45

33 percent of the surface area, at most, whereas acidic surface oxides can cover as much as 20 percent (38). Such carbons exhibit a positive surface potential (35). The structures of basic surface oxides are not well defined. Some suggested structures and mechanisms to account for observed acid adsorption are illustrated in Figure 2-2 (28, 35). The formation of carbonium ions, when carbon with oxygen bound in chromene-like structures is oxidized in the presence of a strong acid, is the only concept able to explain all the observed acid adsorption phenomena (38). In addition to specific adsorption mechanisms, the presence of both acidic and basic surface oxides impart a polar nature to the carbon surface (35). As a result, the carbon will exhibit preferential adsorption for a more polar component of a binary mixture. For example, a carbon essentially free of oxygen will preferentially adsorb benzene over methanol while a carbon with acidic surface oxides will preferentially adsorb the methanol (35). With increased surface polarity pore constriction or blockage can result from the adsorption of water through hydrogen bonding. For polar solutes, the surface oxide-organic interaction is stronger than for nonpolar solutes, compensating somewhat for the extra energy needed to desorb water (34). This brief overview of the surface chemistry of activated carbon illustrates some of the many parameters and mechanisms involved in determining the adsorptive capacity

PAGE 46

34 BENZOPYRYLIUM ION p (CARBONIUM ION) BENZOPYRYLIUM DERIVATIVES (RESONANCE HYBRIDS) PYRONE-LIKE STRUCTURES Figure 2-2. Basic oxygen surface functional groups.

PAGE 47

35 of an activated carbon for an organic compound. The specificity of mechanisms for certain classes of organics, the variety in type and relative abundance of surface functional groups among different carbons, and the lack of laboratory studies beyond single or bisolute systems point out the complexity and difficulty in modeling carbon adsorption on a micro scale. 2.3.5 Particle Size The effect of carbon particle size on adsorption capacity, especially equilibrium capacity, is not clear. Theoretically, grinding and sieving of an activated carbon to a particular size fraction should have no effect on adsorption capacity. Assuming that the carbon particle is a nonporous cube, the outer surface area of the largest particle in a 12x40 mesh carbon (1.68 mm x 0.420 mm) would 2 be 16.9 mm Assuming a carbon density of 0.41 g/cu cm, 1 gram of 1.68 mm carbon particle cubes would have an outer surface area of approximately 0.01 m Grinding of this carbon to pass a 200 mesh sieve (0.074 mm) would produce a theoretical outer surface area of approximately 0.2 m^ Further grinding to pass this gram of carbon through a 325 mesh sieve (0.044 mm) would result in a theoretical outer surface area of approximately 0.33 m^. Compared to the 1000 2 m /g surface area of a typical activated carbon, the outer surface area contributes less than 0.03 percent. Therefore,

PAGE 48

36 grinding a larger mesh carbon into smaller-sized particle fractions should not affect total carbon surface area to any observable extent. Several researchers have examined the effect of carbon particle size on adsorptive capacity and, for the specific carbons studied, observed little or no effect. In two separate studies a coconut-shell-based carbon exhibited no effect of carbon particle size in the adsorption of phenolic compounds for particle-size fractions ranging from 12 x 16 mesh to 230 x 270 mesh (42,43). The variation of adsorptive capacity with particle size for a coal-based carbon was examined for five fractions of mean particle diameters ranging from 0.273 to 1.011 mm (44). No effect on equilibrium capacities was observed. Other studies have produced similar results (45-47). Based on such results and theoretical considerations, a number of researchers using a particular size fraction of a ground and sieved carbon have assumed that particle size did not affect carbon adsorption capacity (48-50). Other researchers, however, have observed an increase in adsorption capacity with a decrease in carbon particle size. In one study, a coconut-shell-based carbon was crushed and sieved to produce two narrow-range particlesize fractions (51). In a second study a coconut-shellbased carbon was only sieved to produce the desired particle-size fractions and, in both studies, a significant

PAGE 49

37 increase in capacity with decreasing particle size was observed (52). In both studies the times required to attain equilibrium were carefully determined. A bituminous-based carbon was studied in the development of a rapid method for determining carbon adsorptive capacities (53). Carbon particle size again affected adsorption capacity for the specific carbon used. Adsorptive capacity for humic acid on a lignite-based carbon also varied with particle size (54). One recent study concluded that there was not enough information available to state with certainty that particle size will affect adsorption capacity for the carbon of interest (47). Several phenomena have been postulated to explain the observed variations in adsorptive capacity. Available pore volume may increase as carbon particle size decreases due to the opening of previously sealed pores and channels. This increase in available pore volume may result in an increase in the adsorptive capacity for smaller, low molecular weight solutes but may not affect the adsorptive capacity for larger, high molecular weight solutes. If blockage of long channels at constrictions is a mechanism affecting adsorptive capacity, then reducing carbon particle size by fracturing along such channels may increase the total pore volume available to large organic molecules. It is also likely that the pore structure throughout a carbon particle is not uniform as a result of activation conditions. The

PAGE 50

38 adsorptive capacity may be greater in the outer regions of a carbon particle due to more extensive burnout of pores and channels, whereas the particle interior may consist only of a small volume of micropores and sealed channels. Grinding and sieving to achieve a particle size fraction may result in the retention of the low-capacity core and the discarding of the higher capacity outer shell. Further studies are needed to define the effect of available pore volume, pore distribution, molecular size of organic solute molecules, uniformity of activation, and pore blockage on adsorptive capacity as the diameter of a carbon particle is reduced by standard laboratory grinding and sieving techniques. 2.3.6 Manufacture of Activated Carbon The rapid development and use of activated carbon as an adsorbent was begun during World War I when a protective mask was developed to shield soldiers from chlorine gas. Since that time the applications and varieties of activated carbons have expanded into a complex industry where many different activated carbons are available. The unique characteristics of specific surface area and surface functional groups are primarily determined by the raw material and the manufacturing, or activation, process. The importance of activation temperatures and contact with oxygen in the production process has already been mentioned. Continued improvement in raw material selection.

PAGE 51

39 carbon preparation, and carbon selection promotes greater success and cost effectiveness in the removal of organics from the aqueous phase. The selection of raw material is an important first step in manufacturing an activated carbon. Table 2-2 lists some of the raw materials that have been used in the production of activated carbon. Bituminous coal, lignite, and petroleum coke are the most common raw materials for water and wastewater treatment applications (55). The choice of raw materials determines the amount of tarry substance formed during carbonization, the subsequent development of pore structure, the amount of inorganic matter to catalyze surface functional group and heterocyclic ring structure formation, and the amount of oxygen available to form surface functional groups. The actual manufacturing process has two objectives. The first is to produce a very large surface volume ratio with a specific distribution of pore sizes. The second objective is the formation of specific types and densities of surface functional groups. Two methods of manufacture are used: A previously charred carbonaceous substance is oxidized with steam and high temperatures, or the carbonaceous raw material is chemically dehydrated and oxidized at lower temperatures. Almost all of the activated carbon used in the United States for water and wastewater treatment applications is produced by the high

PAGE 52

40 TABLE 2-2. Raw materials that have been used to produce activated carbons. *Coal (bituminous) Beet sugar sludges Asphaltic material Cane sugar Blood Rubber Coconut shells Coffee beans Waste tires Sawdust Corncobs and stalks Leather wastes *Lignite Oil shale Molasses Wood Distillery wastes Pulp mill wastes Peat Fruit pits Rice hulls *Petroleum coke Seaweed Fish Bone Petroleum heavy oil Cereals *Most common raw materials for water and wastewater application.

PAGE 53

41 temperature-steam process (55). The details of the process vary considerably among manufacturers with critical steps kept proprietary. The general process for manufacturing activated carbon consists of a dehydration step, a carbonization step, and an activation step. Descriptions of each step for a variety of raw materials are available in the literature (8, 10, 56). The manufacture of granular activated carbon from a bituminous, subbituminous or lignite coals generally is accomplished by the following operations (55). Pulverizing, compacting, and crushing followed by sizing to a range of 2 X 40 mesh U.S. Standard Sieves begins the process. Dehydration is accomplished in air at temperatures ranging from 150 to 215 C for 0.5 to 18 hours. Dehydrating agents such as zinc chloride or phosphoric acid may be added. Depending on the oxygen content of the raw material, 1 to 3 percent by weight of oxygen may be added to the coal if the subsequent activation process occurs in a controlled oxygen atmosphere. Acid washing may precede the dehydration step to reduce the inorganic content of the raw material. The second operation in the manufacturing process, carbonization, occurs in the absence of oxygen and at temperatures ranging from 400 to 600C. Metallic chlorides can be added to increase the effectiveness of the carbonization process. During this phase, also called pyrolysis, a char suitable for steam activation is produced. Most of the

PAGE 54

42 noncarbon elements as well as hydrogen and oxygen are removed as gases formed during this pyrolitic decomposition. Elementary microcrystallites are formed in irregular arrangements with many free interstices that become filled or blocked by unorganized carbon (amorphous) from the decomposition of tarry substances. The resultant char has little adsorptive capacity as a consequence. The third essential operation, activation, determines the final characteristics of the carbon. Temperature is the most critical parameter and varies from 400 to 1200 C. The intercrystallite residues are oxidized, removing pore blocking materials. The pores are enlarged and cleaned. The microcrystallites grow and form the final rigid skeletal foam structure that determines the abrasion resistance, or friability, of the carbon. Low-capacity carbons are partially activated by removing tarry products by heating them in a stream of inert gas or by extracting them with a solvent. High adsorption capacity carbons are activated under conditions where the activating agent (steam, carbon dioxide, or oxygen [air]) reacts with the carbon. Disorganized carbon is first oxidized to open the pores between the microcrystallites. Then the crystallites themselves are partially oxidized to create a much greater pore volume. Macropores and transitional pores are formed at the expense of micropores. A burn-off between 50 and 75 percent of the

PAGE 55

43 original char results in the mixed pore structure typical of activated carbons used in water and wastewater treatment. The activating agent reacts with the carbon at the edges of the microcrystallites to form surface functional groups. The type and density of groups are determined by activation temperatures and presence of activating agents. Since higher temperatures promote the formation of larger microcrystallite planes, some control over surface functional group density is exerted. The type and density of oxygen-containing surface functional groups are usually controlled by the temperature of the degassing step, which completes the activation process. The higher the temperature, the less oxygen remains on the carbon surface. As an example, activation temperatures of 400 to 500C with low degassing temperatures will produce an abundance of acidic surface oxide groups which are best suited for adsorption in alkaline solutions. Activation temperatures of 800 to 1000 C and low degassing temperatures produce large quantities of basic surface oxides which readily sorb acid. Degassing temperatures around 1000 C will remove almost all oxygen from the carbon surface producing a product suitable to remove trihalomethanes such as chloroform, which do not adsorb well on carbons having a high density of surface oxide groups (26, 37). Other treatments to increase surface area after activation have included de-ashing the carbon with hot

PAGE 56

44 hydrofluoric or hydrochloric acid, burn-offs in oxygen under low oxygen partial pressures at 600 C, and exposure to ozone at 30C (26). Postactivation processes can include crushing, grading, grinding, acid washing, water washing, drying, and packaging. Powdered carbon is pulverized so that 95 to 100 percent pass a ICQ mesh U.S. Standard Sieve (149 micron opening) and 50 to 95 percent pass a 325 mesh sieve (44 micron opening). Granular carbon is described by two sieve sizes, one which passes the carbon and one which retains it. A typical commercial granular carbon is described as 12 x 40 mesh, where carbon would pass a 12 mesh sieve (1.68 mm) and be retained by a 40 mesh sieve (0.42 mm). Carbon is usually packaged in airtight containers to minimize oxygen adsorption and adsorption of volatile organics from the surrounding air. 2.3.7 Summary The foregoing discussion has shown that activated carbon is a complex, highly heterogeneous adsorbent. The surface of activated carbon can be manipulated by manufacturing processes to exhibit a wide variety of adsorptive behavior. Adsorption energies can range from the very weak van der Waals physical adsorption energies to the very strong energies of the chemical bonds typical of chemisorption. Adsorption mechanisms are also greatly affected by the physical structure of activated carbon, where pore

PAGE 57

45 size distributions affect the rate and capacity of adsorption for solutes of various physical dimensions and charge characteristics. Consequently, activated carbon cannot be considered a relatively homogeneous material, with adsorption phenomena described by a few simple physical mechanisms. 2 4 Effects of Adsorbate Properties on Activated Carbon Adsorption The effects of physical and chemical properties on the adsorption capacity for specific organic solutes in aqueous systems have been studied extensively in efforts to correlate such properties with observed adsorption behavior. The ultimate goal of such correlations was some degree of predictability in the adsorption process to improve the cost effectiveness of system design. Some adsorbate properties studied include molecular weight, functionality, polarity, hydrophobicity dipole moments, aqueous solubility, molecular size, geometry, and surface area. Although these properties were studied individually, results indicated that few were independent of other properties in their effects on adsorption. However, all were shown to directly affect adsorption capacities under some set of well defined experimental conditions. Adsorptive capacity was observed to increase with molecular weight in homologous series (52, 57, 58). The increase in molecular weight was due to the addition of hydrophobic nonpolar groups to the basic molecular

PAGE 58

46 structure, such as increases in alkyl chain length. As molecular weight increased in these studies, hydrophobicity increased, solubility decreased, molecular size increased, and molecular geometry and surface areas changed. However, adsorption capacity studies using a series of nonyl-phenol ethoxylates indicated that adsorption capacity decreased with increase in molecular weight (27). The increase in molecular weight was due to the addition of ethylene oxide groups into the nonyl chain. As molecular weight increased, hydrophobicity decreased while polarity, solubility, and molecular size increased. Consequently, the use of molecular weight as a predictive parameter for adsorption capacity is limited by interrelationships with other sorbate properties Functionality has also been examined as an indicator of adsorption behavior. The addition and/or substitution of such functional groups as NH^ OH, CH2OH, alkyl groups, and benzene rings to benzene and phenols have been the most commonly studied functional systems (42, 52, 57-59, 60-61). Adsorptive capacity for phenolic compounds increased as alkyl substitutions were made on the phenol and as the alkyl chain length decreased. Polycyclic compounds adsorbed better than corresponding monocyclic compounds. The adverse effect of OH, CH^OH, and NH2 groups on adsorption was attributed to hydrogen bonding of these groups with water molecules rendering the organic more hydrophilic. In two

PAGE 59

47 instances the introduction of the OH group enhanced, rather than hindered, adsorption even though the addition of OH resulted in an increase in aqueous solubility. The presence of neighboring CHO and OCH^ groups was thought to result in intramolecular hydrogen bonding with the OH group. Neighboring groups, like OCH^ could also sterically hinder hydrogen bond formation between water and the OH group. Functionality appears to be a qualitative, rather than a quantitative, indicator of trends in adsorptive capacity because of the many exceptions to the expected behavior and is not suitable as a general predictive parameter. Polarity has also been examined as an indicator of adsorption capacities and for predicting adsorption characteristics of other organics. The dipole moment, as a measure of the polarity of a compound, exhibited no correlation with the carbon capacity for 26 organic compounds studied (58). Aniline and phenol possess the same dipole moment, yet exhibited different adsorption capacities on the same carbon. Nitrobenzene had the highest dipole moment of five organics, yet was the most favorably adsorbed. One review article described seven different polarity scales that have been used to classify organics (12). Predictability of adsorptive capacities was limited to compounds of the same class as the representative compounds. Lack of knowledge concerning how each chemical in a class would react in a particular system also limits the usefulness of

PAGE 60

48 such polarity scales. Again, the diversity of compounds as well as reactions involving more than just one type of force interaction do not allow any one classification scheme for organics and adsorptive capacities. Aqueous solubility seems to provide the most widely applicable means of organic classification for adsorptive capacities. All the previously described adsorbate properties can be reflected in an aqueous solubility parameter. The semiquantitative Lundelius rule describes an inverse relationship between the extent of adsorption of a solute and its solubility in the solvent from which adsorption occurs (9). One interpretation postulated that the greater the solubility, the stronger the solute-solvent bond that must be broken and the smaller the extent of adsorption. The special case of Lundelius' rule is Traube's rule which relates the aqueous solubility of various organics in a homologous series inversely to the extent of adsorption (4). An investigation of 93 organic solutes representing 10 different functional groups commonly found in petrochemical wastes showed, without exception, that as aqueous solubility decreased, adsorptive capacity increased (57). This relationship held true not only for decreasing solubility within a functional group but also for decreasing solubility between functional groups.

PAGE 61

49 Further development of the adsorptive capacitysolubility relationship has resulted in the application of a solubility parameter, 6, to a macroscopic single solute equilibrium model for organics in aqueous solutions (62, 63). This parameter attempts to relate the physical, chemical, and structural characteristics of neutral organic compounds to their potential to undergo a change from the liquid to a surface phase. The numerical value of the total solubility parameter consists of components which represent specific types of force interactions. The total parameter is a function of a dispersion components, the permanent dipole orientation and induced dipole components (lumped to form a polar component), and acidic and basic components (lumped to form a hydrogen bonding component). Assigning meaningful values to these components is difficult. In the study applying the total solubility parameter to the macroscopic single solute equilibrium model to calculate net adsorption energies, the dispersion, or nonpolar, component was calculated using refractive indices (62). The hydrogen bonding component was determined using aqueous solubility data. The calculation of the polar component was restricted to several homologous series and could not be calculated independently. The study reported good agreement among the four of five methods available to calculate the total solubility parameter. The calculated value was used together with values for the dispersion and hydrogen bonding

PAGE 62

50 components to estimate the polar component values. This scheme of classification lacks a quantitative, theoreticallyjustified basis and is limited by the lack of a single method to calculate the component solubility parameters for all applications. However, the use of total solubility parameter values in the equilibrium model has resulted in correlations between amounts of compounds adsorbed by activated carbon and the calculated net adsorption energies for several organic compounds (62, 63). The Polanyi adorption potential theory has been applied to adsorption of organic solutes from aqueous solutions (64). A method was developed that makes it possible, given a calibrating isotherm for a specific carbon, to estimate the adsorption isotherms with that carbon for a wide variety of organic liquids from water over the complete range from saturation to very low concentrations, given only their solubilities and densities. The discussion of the application of the Polanyi theory to adsorption from water solution onto activated carbon also provided a list of references describing the fundamental studies on liquid-phase adsorption onto activated carbon (64). The model assumes heterogeneity of adsorption energies over the adsorption space or surface and that any interactions between solute and solvent and between different solutes in solution are reflected in their effects on solubilities. However, the solubilities that enter into the calculations for competitive adsorption

PAGE 63

51 are the solubilities of each component in the presence of the other. Adsorption capacity, however, has not always correlated inversely with aqueous solubility. One study showed that the adsorptive capacity for phenol on a specific carbon was the same as for benzyl alcohol and higher than that of aniline, yet phenol is more soluble than either (58). Nitrophenol adsorbed just as well as cresol, yet has a solubility an order of magnitude less than cresol. The adsorptive capacity for o-methoxyphenol was greater than that for cresol or chlorophenol yet its solubility is lower. Similar observations were recorded for other organic solutes. The conclusion of the study was that no overall general tendency of increase in adsorption with decrease in solubility was observed over the range of the 2 6 compounds studied. Aqueous solubilities and single solute adsorption capacities have been used in bisolute systems to attempt to predict relative adsorbabilities (59, 61, 65-67). The adsorptive behavior of alkyl phenols in two-component systems was better described by the Ideal Adsorbed Solution (IAS) equilibrium model than by the Langmuir model for competitive adsorption (59). The IAS model uses solubilities and single solute isotherm constants. Another study used isotherm constants and solubilities to predict the preferentially adsorbed compound in 22 bisolute systems (65). Solubility was a successful predictive parameter for

PAGE 64

52 20 out of 22 pairs tested. The Freundlich K constant, an indicator of the adsorptive capacity of a carbon for a solute, was a successful index in all 22 pairs tested. Three other isotherm constants from the Langmuir and Freundlich models were not as successful. An improved calculation procedure for the IAS equilibrium model was used successfully to describe adsorption in binary and ternary systems containing similar and dissimilar components and mixtures containing different initial concentrations of solutes (66). However, another study showed that molecular weight was a more reliable indicator of multi-solute adsorption preference than solubility when comparing polycyclic and monocyclic compounds (67). The effect was attributed to stronger adsorptive forces due to greater surface area contact. Solubility was a reliable indicator of preference for monocyclic sorbate pairs. There does not seem to be any one adsorbate property that is universally applicable for predicting relative single solute adsorption capacities. Under limited conditions, adsorptive capacity generally increases with increasing molecular weight and with decreasing solubility. In multi-solute systems preferential adsorption appears to be proportional to some measure of single solute adsorption capacity, such as the Freundlich K, and to molecular weight. Such adsorption also appears to be inversely

PAGE 65

53 proportional to solubility. Single solute adsorption isotherm parameters have been used for attempting quantitative predictions of multi-solute adsorption while parameters describing the physical-chemical nature of the organic have been used qualitatively. However, the presence of exceptions to all efforts at predicting preference and quantities of adsorption imply that no single adsorbate property is applicable for general use. The development of several recent multi-solute equilibrium models attempts to combine several of the more significant properties to improve overall predictability. 2 5 Effects of Aqueous Solvent System Properties on Adsorption Aqueous system characteristics which exert influence on activated carbon adsorption include temperature, pH, dissolved solids, and matrix complexity. Since adsorption is normally an exothermic process, adsorptive capacity generally decreases with increasing temperature. Gas phase adsorption involves heats of adsorption on the order of several kilocalories per mole, and such systems are very temperature sensitive. However, adsorption in aqueous systems involves much smaller heats of adsorption, on the order of a few hundred calories per mole, due to the endothermic desorption of water when adsorption of a solute occurs on a carbon surface. Normal temperature variations experienced in

PAGE 66

54 water and wastewater adsorption processes have only a minor effect on adsorption capacities (9, 21, 30, 49, 52). Dissolved salts can significantly influence the adsorptive capacity of carbon for ionized organic molecules but exert little influence on neutral solutes. The adsorption of sodium benzenesulf onate was greatly increased when CaCl2 was added to the solution (34). This was attributed to a decrease in the mutual repulsion between surfactant head groups through interaction with the divalent calcium ions. At pH 2.0 no significant differences in adsorptive capacity for p-nitrophenol were observed with NaCl concentrations up to 1 mole/liter. At pH 2.0 the p-nitrophenol exists solely as a neutral species. When the pH was raised to 10.0 where ionic species dominate a significant salt effect was observed at the higher NaCl concentrations (49). The explanation involved an ion pairing of the cation with the anionic p-nitrophenol, thereby reducing the electrical double layer. A second possible mechanism was a partial repulsive charge reduction between adsorbed anions by the salt. A study of the effect of phosphate buffer on 2 4-dichlorophenol and 2 4-dinitrophenol indicated no influence on adsorptive capacities for undissociated solutes, but an increase in adsorption of dissociated solutes as buffer concentration increased (68). The effect of salts on the adsorption of fulvic acids has received increasing attention as these substances have

PAGE 67

55 been cited as precursors to trihalomethane formation. The concentration of a phosphate buffer in solution had a significant effect on the adsorption of soil fulvic acids at pH 7.0 (69). Another study showed that the adsorptive capacity for dissociated dinitrophenol nearly doubled in the presence of a 0.01 M phosphate buffer (70). However, a later study demonstrated that enhanced adsorption of fluvic acid was due to cation concentrations (calcium, magnesium, and sodium) and that anion concentrations (chloride, sulfate, bicarbonate, and phosphate) had little or no effect (71). All salts studied showed little additional effect on adsorption capacities beyond salt concentrations of 3 to 4 millimoles per liter. The effect of salts varied significantly with pH showing increased influence on adsorption as pH increased, confirming the effect of cations on the adsorption of dissociated fulvic acid species. Available literature indicated that inorganic salts can influence the extent and rate of adsorption of anionic organics onto negatively charged surfaces. Few studies have attempted to define the mechanisms responsible for enhanced or hindered adsorption in the presence of salts. Speculative mechanisms include solute-salt interactions to alter specie distribution, salt-adsorbed organic interactions to alter surface packing, and salt-carbon interactions to alter surface charge characteristics. Salts do not appear to significantly influence the adsorption of neutral compounds.

PAGE 68

56 but few data are available. Available data are insufficient to draw general conclusions about the effects of co-ions, various salts, and salt concentrations on activated carbon adsorption of organics in aqueous solutions. The effect of pH on adsorptive capacity has been widely studied. The two components of the adsorption mechanism thought to be affected by solution pH are the chemical characteristics of the carbon surface and the extent of adsorbate dissociation (12). Equilibrium capacities for sulfonated alkylbenzenes in pH regions far from the pK range for these compounds increased with increasing hydronium ion concentrations (52). Far from the pK range pH changes would not significantly affect the net ionic character of the adsorbing species. The increase in adsorptive capacity was attributed to partial neutralization of the carbon surface's negative charge character, reducing resistance to pore diffusion. A study of the adsorption of phenol at pH's from 2.0 to 10.6 showed an unexpected reduction in adsorption capacity with decreasing pH (49). The interaction of hydronium ions with surface carbonyl groups competitively with phenol was postulated to explain the decreasing phenol capacity. The effect on nitrophenols was not as great and was attributed to stronger bonding energies through the phenolic acceptor-donor complex reaction. The adsorption of the hydronium ion as a competitive solute at low pH values has been supported by other work (21, 46, 72, 73).

PAGE 69

57 The extent of dissociation for ionizable solutes is greatly affected by pH, especially for weak organic acids like phenols. Observed changes in adsorptive capacities with pH for phenol and nitrophenols were attributed to changes in solubility (21). Adsorptive capacities for aromatic acids passed through a maximum at pH values near the pK points of the organic acids where both ions and a molecules exist in comparable amounts in the bulk solution (72). This study also indicated that substantial amounts of ionic species were adsorbed as well as molecular species. In a study involving chloro-substituted phenols the adsorptive capacity for trichlorophenol increased substantially as pH decreased (74). Dichlorophenol adsorption was significantly less affected by pH changes. Adsorption capacity was observed to reach a maximum near pH = pK^. The neutral species of both solutes adsorbed more strongly than the anionic species. The effect of pH on adsorption from a mixture of the two substituted phenols was large but could not be described by existing competitive models. Adsorption capacities for 2-methylpyridine and o-cresol decreased as pH was lowered from 9.5 to 3.5. The effect of o-cresol was much less than that for 2-methylpyridine. In the pH range of 6.8 to 9.5 the amount of ionic species of o-cresol increased and of 2-methylpyridine decreased, yet adsorption capacities were unaffected. This indicated that both anionic and neutral species were adsorbed on the carbon.

PAGE 70

58 The decrease in adsorptive capacity of the two compounds was attributed to competitive adsorption of hydronium ions. Substantial increases in the adsorption of fulvic acid were reported as pH values decreased in a study that attributed the enhanced adsorption to a decrease in fulvic acid solubility (69). The effects of pH on adsorption capacity in aqueous solutions are not well understood. Observed pH effects depend on the nature of the carbon surface, especially the functional groups thereon, the nature and concentration of the organic solutes, temperature, and ionic strength. The effects of pH are not well described by existing competitive equilibrium models outside narrow and well-defined experimental conditions. Differences in solubility between neutral and anionic species contribute to observed changes in adsorptive capacities. Development of repulsive forces between the anion and the surface or between sorbed anions is also significant. Changes in the surface charge of the carbon due to hydroxyl or hydronium ion adsorption or ionization of sorbed molecules or weakly acidic surface functional groups could substantially affect adsorptive capacities. In water treatment processes the adsorption of neutral organic compounds is probably not greatly affected by solution pH changes in the ranges encountered as are the weak acids used in most of the studies reported in the literature (12). Most of the organics of health

PAGE 71

59 significance are such neutral compounds. However, in the case of organics susceptible to ionization, pH conditions yielding the highest undissociated state seem to provide the best conditions for adsorption. Matrix complexity is the final major characteristic of an aqueous system which influences adsorption capacities. Matrix effects are particularly significant in wastewater treatment applications. The presence of other organics in the solution usually reduces the adsorptive capacity of a particular carbon for each organic solute, although the total adsorptive capacity of the carbon is usually increased. Mutual inhibition can be predicted to occur if adsorption is restricted to one or two molecular layers. Two additional criteria for predicting mutual inhibition are that the adsorption affinities of the solutes do not differ significantly and that there are no strong interactions among solutes (9). Since the final equilibrium positions are dictated by thermodynamics, many multi-solute equilibrium models have been developed in attempts to provide some predictability in multi-solute systems. The bulk of the data available in the literature on competitive adsorption involve bisolute studies. In a phenol-dodecylbenzene sulfate (DBS) system the presence of the DBS reduced the single solute adsorptive capacity of phenol by approximately 30 percent (51). Although the phenol was a much smaller molecule and had a higher

PAGE 72

60 effective rate of diffusion, pore blockage by DBS was thought to reduce the surface area and pore volume available to phenol. A study of competitive adsorption between dichloroand trichlorophenol showed that, as the equilibrium concentration of one solute increased relative to the second solute, the adsorption of the second solute decreased (74). In a study of competitive adsorption with bisolute mixtures of organic anions and neutral species, similar results were seen for some mixtures but not for others (75). Some adsorbed with substantial competitive effects while others showed little competition. In some cases, electrostatic repulsive forces were significant. Little competition was expected where there was a large difference in solute molecular sizes because of pore size distributions in the carbon. Mutual suppression of single solute adsorption capacities was also observed in a study of six fatty acids and four phenolic compounds (76). In competitive adsorption from bisolute mixtures of a fatty acid and a phenolic solute reduction in the adsorptive capacity for fatty acids was substantial but was not so extensive for phenol. Other bisolute studies are also available (77-79). Additional bisolute data will be discussed in the presentation of multi-solute equilibrium models. Much more limited are data from multi-solute studies involving three or more compounds. In a mixture of three fatty acids the adsorptive capacity was reduced for acetic

PAGE 73

61 acid substantially and slightly for propionic acid. However, the capacity for butyric acid actually increased somewhat (76). In a study of competitive adsorption in ternary phenolic mixtures, the equilibrium capacities were successfully predicted by the Ideal Adsorbed Solution model with a modified calculation procedure (66). A fourcomponent mixture of glucose, alkyl benzene sulfonate, an amino acid, and lactic acid was studied in a column reactor to validate a dynamic packed bed column model. Some calculations were also made for a nine-component influent (80). Adsorption equilibria of individual solutes in ternary mixtures of varying initial concentrations showed decreases in single solute adsorptive capacities (81). The most strongly adsorbed solute was described by a Freundlich isotherm, but the more weakly adsorbed solutes had curved isotherms on a log-log scale. The curved isotherms of individual solutes adsorbed from a mixture were typical if one or more of the competing solutes had higher adsorption capacities as a single solute than the remaining solutes. However, all solutes showed decreased adsorptive capacities in mixtures. These observations were also confirmed by equilibrium data in mixtures of up to six compounds. The previously discussed research all described or postulated factors determining interactions among competing solutes. The relative size of adsorbates seemed to primarily affect transport processes by pore blockage. The

PAGE 74

62 relative affinities of competing solutes, if sufficiently different, can result in the displacement of previouslyadsorbed solutes. A very strongly adsorbed solute can even cause an otherwise adsorbable compound to appear to be nonadsorbable In column studies of mixtures of solutes, effluent concentrations of a solute exceeding influent concentrations in a steady state system are commonly observed. The relative concentration of solutes can also play a major role in site competition. A solute can compete more successfully when its concentration relative to other competing solutes is increased. Solute-solute interactions such as electrostatic repulsion/attraction between ions, dipole, and ir-electron clouds can be significant. The heterogeneity of the carbon surface can include adsorption sites so specific as to preclude any competition. Such noncompetitive adsorption is believed to be limited to ionic species and, therefore, determined by solution pH and carbon surface characteristics. The increase in combined adsorptive capacity with multi-solute systems is thought to reflect these phenomena. Site specificity can also result from limited accessibility due to pore size constrictions in addition to reaction specificity of certain surface functional groups. The exponential increase in mechanism complexity in competitive adsorption renders mathematical description extremely difficult beyond bisolute systems. Current efforts to develop predictive models consequently

PAGE 75

63 must rely on such techniques as lumped isotherm parameters, organic homologues, total organic carbon isotherm analysis, and other approximations to reduce a multi-solute adsorption problem to a twoor three-component system. 2 6 Descriptive Single-solute Equilibrium Models Adsorption results in the accumulation of a solute from a solution onto a surface of a solid until a state of dynamic equilibrium is reached in the system at a fixed temperature. At this equilibrium there is defined distribution of solute between solid and liquid phases. Mathematical descriptions of this equilibrium are referred to as adsorption isotherms and attempt to express the solid phase concentration of solute as a function of the liquid phase concentration at equilibriiam. Single-solute adsorption systems are generally described by one or two of several isotherm equations. Some apply solely to monolayer adsorption while others can describe multilayer adsorption. 2.6.1 Henry s Law Henry's Law is the simplest of the single-solute equilibrium models. This linear isotherm is valid only at very low concentrations and assumes that all molecules are isolated from their nearest neighbors. This linear relationship is analogous to the limiting behavior of solutions of gases in liquids and the constant of proportionality

PAGE 76

referred to as the Henry constant. This linear partitioning model is thentiodynamically based and is the lower boundary condition for any adsorption equilibrium model (59, 82). It is expressed as % = Ve (2-1) where is the equilibrium solid phase concentration, is the equilibrium liquid phase concentration, and Kj^ is the Henry constant or the partition coefficient. The thermodynamically based Langmuir equation reduces to Henry's Law at small solute concentrations. Several other single and multi-solute equilibrium models have been modified to reduce to Henry's Law as dilute solution conditions are approached. A three-parameter isotherm equation was developed to describe the adsorption equilibria of five -5 -1 organic compounds over a concentration range of 10 to 10 M (22). Three of the solutes (propionitrile, 2-propanol, and acetone) exhibited isotherm slopes near unity at the lower concentration ranges, and the slopes approached unity even closer as temperature rose. Such adherence to Henry's Law was typical of poorly adsorbed solutes. The remaining solutes (p-chlorophenol and p-cresol) did not exhibit the same behavior, which indicated that the phenol-carbon surface interactions were much greater than those of the first three solutes. The proposed three-parameter isotherm reduced to Henry's Law at low concentrations and was

PAGE 77

65 successful for describing the adsorption of the solutes studiedLinear partitioning behavior was also observed for polychlorinated biphenyls (PCBs) when adsorbed on a river sediment (83). At higher solution concentrations the adsorption equilibrium was nonlinear. Adherence to Henry's Law was observed for the adsorption of phenol and several alkyl phenols at concentrations in the lO"'^ M range (59). Isotherm data for many different types of organic compounds -9 -2 over a concentration range of 10 to 10 M showed compliance with Henry's Law as equilibrium concentrations decreased (62). The more poorly adsorbed compounds, such as urea, showed linearity at higher equilibrium concentration -3 _2 values (10 to 10 M), whereas more strongly adsorbed compounds did not exhibit linearity. For example, the isotherm for 2 4 6-trichlorophenol over a range of 10~^ to 10 ^ M had a maximum slope of only 0.592 at 10~^ M. Three other three-parameter isotherms not commonly used in water and wastewater studies also reduced to Henry's Law at low concentrations (82). The importance of Henry's Law in equilibrium modeling has been rather neglected since organic solutes of concern in adsorption studies were in the mg/L range. However, with analytical techniques available today to analyze for organics at the nanogram per liter range and with the presence of almost 200 compounds of health concern in water

PAGE 78

66 supplies at the nanogram level, the ability of any equilibrium model to reduce to Henry's Law has become much more significant. 2.6.2 The Lanqmuir Equation The Langmuir equation has been one of the most popular single-solute equilibrium models because of its theoretical basis, success, and simplicity. The model can be derived from statistical thermodynamics as well as kinetically (4, 7). The kinetic derivation is based on the following assumptions : 1. Molecules are adsorbed at a fixed number of welldefined localized sites; 2. Each site can hold only one adsorbate molecule; 3. All sites are energetically equivalent; 4. There is no interaction between adsorbed and gaseous phases; 5. Adsorption is reversible. The adsorbed layer, therefore, can only be a monolayer if the model is to successfully describe the equilibrium. The Langmuir equation is commonly written as follows (9): Q b C "^e = (1+b C^) (2-2) where q^ is the equilibrium solid phase concentration, C i e the equilibrium liquid phase concentration, Q is the solid

PAGE 79

67 phase concentration when a complete monolayer is formed on the surface, and b is related to the net enthalpy of adsorption, H (b is proportional to e^" ^/RT)^ where R is the universal gas constant and T is absolute temperature). The Langmuir equation will reduce to Henry's Law when bC^ << 1. The two Langmuir constants are determined by best fitting the following linearized form of equation 2-2 to experimental data : qe Q ^ bQ ^ ^ r ^ (2-3) e The basic assumptions of the Langmuir equation are never met in activated carbon adsorption systems treating water or wastewater because of the heterogeneity of the carbon surface. Adsorption data have been successfully described by the Langmuir equation for very narrow concentration ranges, but it usually does not fare well when compared to other isotherm equations. 2-6.3 The Brun auer, Emmet, and Teller (BET) Equation The BET model was an extension of the Langmuir equation to describe multilayer adsorption (4, 7). Consequently, this equation also assumes a surface composed of uniform, localized sites and that adsorption at one site does not affect adsorption at neighboring sites. The model development is based on the same kinetic picture as the Langmuir

PAGE 80

68 equation but assumes that layers of molecules adsorb on top of the previously adsorbed monolayer. The BET model also assumes that the Langmuir equation applies to each successive layer. The first layer's heat of adsorption may have a unique value, but in all succeeding layers the heat of adsorption is equal to the heat of condensation of the liquid adsorbate. The BET equation is commonly written as follows: BC Q e ^e [C_-C^J[1 + (B-1)(C /C )] (2-4) => e e s where B is a constant reflecting an energy of interaction with the surface, is the saturation concentration of the solute in the liquid phase, and the remaining symbols are as defined for equation 2-2. Equation 2-4 is normally rearranged to a linear form for application with experimental data: C c (C -C )qe BQ ^ BQ ^ ^ C~ ^ (2-5) St; S The BET equation has become the standard one for surface area determinations, usually with nitrogen at 77 K as the adsorbate (4). However, the equation has not enjoyed wide use in the field of environmental engineering because of difficulties in obtaining appropriate values for C and s also because at the concentrations found in most water and

PAGE 81

69 wastewater applications C << C and the BET equati on reduces to the Langmuir equation. The saturation concentration, Cg, can only be estimated in practice by an iterative procedure to fit the model to experimental data (84). The model is also restricted by the same limitations of the Langmuir equation. 2.6.4 The Freundlich Equation This single-solute equilibrium model is one of the most widely used in the water and wastewater treatment field. The equation was developed to empirically describe adsorption and therein lies its greatest weakness. This equation was meant to describe that portion of the isotherm beyond the dilute solution region of Henry's Law (85). The development of the equation was based on the assumption that the adsorbent had a heterogeneous surface composed of different classes of adsorption sites, with adsorption on each class of sites conforming to the Langmuir equation (86). The development of the Freundlich equation begins with the Langmuir constant, b, which is related to the net enthalpy of adsorption. On heterogeneous surfaces b will not be a constant but rather a distribution function of energies that can be related to fractions of occupied surfaces. The new adsorption isotherm is the integration of each adsorption isotherm function for each value of b in its distribution function. This integral equation is the experimentally

PAGE 82

70 observed adsorption isotherm. One particular solution of this integral equation assumes the Langmuir model to be the isotherm function and attributes the variation in b solely to the heat of adsorption. The solution to the integral equation, in this case, has the form % X C.'''" (2-6) and is known as the Freundlich adsorption isotherm (4), where K and n are statistically determined, best fit constants. The equation does not become linear at low concentrations nor does it show a saturation or limiting value. Although the derivation is somewhat theoretically based, it is considered an empirical equation limited in its usefulness to its ability to fit data. The limitations of the Freundlich equation are often ignored by many investigators who incorrectly extend the model beyond the valid experimental range it describes (85). However, the model has successfully correlated experimental data for adsorption on activated carbon from aqueous systems over a wide range of concentrations, and the model has been incorporated into several multi-solute adsorption equilibrium predictive models

PAGE 83

71 2.6.5 Three-Parameter Equations The lack of fit of the Langmuir equation over wide concentration ranges and the empirical limitations of the Freundlich equation have led researchers to propose several other equations to describe experimental single-solute adsorption data. One study compared the Toth, RedlichPeterson, and Newman (three-parameter) equations with a new three-parameter equation (82). Six dilute aqueous bisolute systems were examined. The Toth and the study's equations best described the data, although all four had absolute relative percent deviations less than 10 percent and, for most of the eight compounds studied, less than 5 percent. The Toth equation is little known in the West, having been published in east European literature. Its derivation follows a similar line as the Freundlich isotherm but specifies the dependence of the integral enthalpy of adsorption as a function of the solid phase concentration of a particular solute. This dependence is specified through a dimensionless quantity that was defined for ease in obtaining experimental data. It described experimental data well in the above study. The isotherm equation developed in the study incorporated the concept that highest energy sites are filled first so that the heat of adsorption declines rapidly with increased surface coverage. The equation had the form

PAGE 84

Kq P 72 e C = [ e (2-7) where H and K are functions of temperature and p is a constant (82). Both isotherm equations were used with the Ideal Adsorbed Solution (IAS) multi-solute equilibrium model successfully in prediction adsorption for six bisolute systems at 20C. A modified, three-parameter Freundlich model was developed to correct for the crossover from Freundlichmodeled behavior to Henry's Law behavior (59). The equation was successfully used to predict adsorption in several bisolute mixtures and one ternary mixture using the Ideal Adsorbed Solution (IAS) multi-solute model. A similar segmented set of Freundlich equations was used to describe single solute adsorption isotherms for the prediction of multi-solute adsorption using the IAS theory in five dilute, aqueous, bisolute systems (87). The single solute isotherms were divided into five segments, each with a different set of Freundlich constants. This approach was chosen over existing threeand five-parameter isotherm equations to minimize computational time and expense. Another threeparameter model was developed to combine Henry's Law and the Freundlich equation to describe adsorption data over a wider range of concentrations (22). This equation takes the form aC e c (2-8) 1 + bC e

PAGE 85

73 where a, b, and c are determined by statistical fit of experimental data to the equation. At low solution concentrations the equation reduces to Henry's Law. At high solution concentrations the equation is equivalent to the Freundlich model, and when c = 1, the equation becomes the Langmuir isotherm. This equation has generally described adsorption data well. Unfortunately, this eqution does not readily adapt to multi-solute equilibrium models. There are several other single-solute descriptive equilibrium models using a variety of parameters that treat adsorption data effectively. Some of these are listed and described elsewhere in a brief review of single-solute equilibria models (87). 2 7 Approaches to Predictive Single-solute Equilibrium Models 2.7.1 Thermodynamic Approach Physical adsorption has been described by classical thermodynamics applied in a macroscopic approach in twoand three-dimensional models (1, 2, 4-7, 78, 88). The fundamental expression of the thermodynamics of adsorption is the Gibbs equation which, for dilute solutions of one solute, may be approximated by (2-9)

PAGE 86

74 where r is the amount of solute adsorbed per unit area of surface at equilibrium, in excess of the bulk concentration (also termed the surface excess). If Y is defined as the interfacial tension, then a solute which reduces y will concentrate at the surface (1). The development of another form of the Gibbs adsorption isotherm for the solid-liquid interface results in the form ^ nT dy^ AdTT = 0 (2-10) at constant temperature and where A is the molar surface area of the adsorbent, n^"* is an invariant adsorption term approximated by VAC^ in dilute solutions, y^ is the chemical potential of component i, tt is the spreading pressure, V is solution voliame, and AC^ is the change in concentration caused by adsorption (78). The development of the Gibbs isotherm assumes a thermodynamically inert solid with an available specific surface area identical for all adsorbates (78). Given the heterogeneous nature of the activated carbon surface, such assumptions are limitations to the use of the equation. However, this approach is incorporated in other methods of describing adsorptive behavior in multi-solute systems.

PAGE 87

75 2.7.2 Energy Balance Approach This approach takes a microscopic view of the adsorption process at the surface of a solid phase and applies an energy balance to the adsorption. Two models under investigation which take this approach include an application of the solvophobic theory and the net adsorption energy approach. These two models emphasize the importance of solute, solvent, and adsorbent interactions in the adsorption process. The solvophobic theory was developed to estimate the effect of solvents on several types of reactions involving molecules common in biology, including isomerization, association, and conformational changes (89). The adaptation of the theory to adsorption in aqueous systems gave the net solvent effect as a summation of six free energy terms (89, 90). Although the terms could be calculated from the physicochemical properties of the system, the calculations were very complicated. If the solute molecules were much larger than the water molecules, then the equation simplified to one needing only the total solute molecular surface area to predict adsorption capacity. The solvophobic theory, while finding excellent application in reverse-phase liquid chromatography, has several limitations in its use in aqueous systems with heterogeneous carbon (91). So far, its application is restricted to un-ionized, nonpolar solutes. The model

PAGE 88

76 assumes a uniform, nonpolar, noninteractive carbon surface and does not describe effects on adsorption of such solvent characteristics as pH and ionic strength. Such inclusions would require microscopic details not yet available, such as surface heterogeneity. It also does not provide information on the shape or type of adsorption isotherm for a particular solute. The theory also produces only a single adsorption capacity value, rather than a broad-ranged isotherm. The net adsorption energy model was developed to include the solvent effect in aqueous phase adsorption by relating a calculated net adsorption energy to adsorption capacity at specific solution concentrations (62, 63). The model uses the solubility parameter concept, which was discussed earlier in Section 2.4, to determine the net adsorption energy. The calculated energy is a function of a compound's aqueous solubility, molecular weight, density, a dispersion component of the solubility parameter for the organic compound, and a dispersion component of the solubility parameter for the activated carbon surface. The net adsorption energy (E^) is calculated by combining the solute's affinity for both the adsorbent (E. ) and the solvent (Ej^) and the interaction between solvent and adsorbent (Ej^^). The E values are in calories per mole, and the subscripts i, j, and s refer to solvent, solute, and adsorbent, respectively. The equation is of the form

PAGE 89

E„=E. -(E. +E..) 11 (2-11) Each of the energy terms is calculated using component values of the solubility parameter, 6, for the organic compounds. The application of the net adsorption energy concept to evaluate adsorption data for medium-polarity to high-polarity organic compounds showed that measured adsorption capacities increased as calculated net adsorption energies increased (62). The application of this concept has shown several limitations to its usefulness. The compounds that have been tested were qualitatively ranked. The model assumes a uniformly nonpolar activated carbon surface. Quantitative determinations of the components of the solubility parameter are not well developed. Application of the theory to predict the adsorption capacities of neutral organic compounds is not clearly understood. A comparison of this theory with the Polanyi theory found that the relationship between adsorptive capacity and net adsorption energy was predictive only if separate relationships were defined for branched and linear alcohols (91). If the model is to be generally predictive of adsorption capacities of organic solutes, a single relationship is needed. As adsorption energy increased, the study showed greater deviation from linearity for the alcohols. The method was not able to predict adsorption isotherms accurately, although rough

PAGE 90

78 estimates were obtained using only the refractive index, the molar volume, and a rough estimate of solubility. The model is relatively new and untested but shows promise in elucidating the importance of solute, solvent, and adsorbent interactions and roughly estimating relative adsorption capacities for different organic compounds. 2.7.3 Adsorption Potential Approach This approach to multilayer adsorption postulates a potential field at the surface of a solid into which adsorbate molecules "fall" (4). The adsorbed layer resembles a planetary atmosphere being most compressed at the surface and growing less dense as the distance from the surface increases. This concept of a changing adsorption potential in the adsorption process was incorporated into the Polanyi adsorption potential theory, one of the best known of the potential theories (4, 64). The development of this theory is described and well documented in several studies that applied the theory to adsorption from aqueous solution onto activated carbon (64, 92, 93). The Polanyi theory postulates a fixed volume close to the carbon suface where the adsorption process occurs. For carbon this volume essentially consists of micropores of varying size and shape. The adsorption potential is defined as the work required to remove a molecule from its location in the micropore to infinity. T^ie magnitude of this

PAGE 91

79 potential depends on the nature of the surface and the distance of the solute molecule from the surface. The potential will be greatest for micropores whose volume approximates that of the solute molecules since surface area contact will be a maximum as well as the van der Waals forces of interaction. One important consequence of this theory is that for many single solute systems using the same carbon all of the single-solute isotherms from trace concentrations to saturation may be readily calculated from the isotherm of any one solute (64). That is, the theory postulates a single characteristic curve from a single-solute isotherm at any temperature. For a different solute, an abscissa scale correction factor will produce the same characteristic curve. To produce the unique characteristic curve of a carbon, it is possible to transform the isotherm to the characteristic curve by the conversion of q and C to a e e volume of adsorbate adsorbed and an adsorption potential. Dividing q^ by the liquid density results in the volume/mass carbon ordinate. The adsorption potential is calculated by C e = RT^n ( ^ ) (2-12) e where e is the adsorption potential and C is solute s solubility. From this characteristic curve, the isotherm for any temperature can be calculated for the single solute.

PAGE 92

80 A plot of adsorbate volume per unit weight of carbon against the adsorption potential per unit liquid volume is approximately the same for a series of similar compounds, such as a homologous series of hydrocarbons (94). An abscissa correction factor will give the same correlation curve for a wide variety of compounds. Consequently, it should be possible, given one single solute isotherm, to predict single solute isotherms for a wide variety of organic solutes across a wide concentration range. One study correlated experimentally determined Freundlich K constants for nine alcohols and plotted K against calculated net adsorption energies (91). The resultant correlation line was used to predict K values for 13 additional compounds. The net adsorption energy concept predicted K values well, better than the Polanyi with which it was being compared. Both methods deteriorated for compounds with loadings greater than one millmole per gram. The model also had difficulty with branched-to-linear alcohol relationships. The adsorbate volume also had to be empirically adjusted for variations in the maximum adsorption of all solutes even when on a volume basis. The Polanyi model assumes heterogeneity of adsorption energies over the adsorption surface which contrasts with the Langmuir assumption of equal energies everywhere. This allows better accommodation of competitive and noncompetitive adsorption than the Langmuir semicompetitive model.

PAGE 93

Any interactions between solute and solvent are reflected conveniently in solubility effects. The model has several limitations. One is the assumption that all pores are accessible to the sorbate. The model also assumes that onl physical adsorption occurs and does not include chemisorption. However, this method is useful and somewhat easy to apply. 2 8 Multi-solute Equilibrium Models Several models have been developed to describe competi tive adsorption of organic solutes in aqueous solutions. Most are extensions of single-solute descriptive and predic tive models. Two different approaches to modeling are currently being developed in activated carbon adsorption. Modeling the equilibrium distribution between solid and solution phases in descriptive and predictive ways has been the thrust of this review of the literature. A different modeling approach which takes into account the dynamics of adsorption has been pursued by several investigators as a basis of a rational carbon column design. The interrelationship between these two approaches lies in the thermodynamics driving a carbon adsorption system toward some equilibrium. All dynamic models must incorporate these thermodynamic equilibria considerations in order to link thermodynamics with mass transport limitations adsorption processes.

PAGE 94

82 Although each of the current multi-solute equilibrium models adequately describes or predicts equilibria under certain specified conditions, one limitation shared by all is the failure to include the nature of the adsorbent. All the models assume that there are no specific adsorptive interactions or that chemisorption, if it occurs at all, is too small to affect overall equilibria as the models describe them. The interaction of organic substances and surface functional groups can markedly change the pattern of distribution of adsorption energies for different solutes. Molecular sieving effect due to differing pore size distributions in various adsorbents and pore blockage can be more significant than physical adsorption forces reflected in solubility effects in determining adsorption behavior (95). Limitations of predictive and descriptive models, then, can be attributed to assumptions made during their development. Assumptions common to almost all models are a homogeneous, thermodynamically inert surface, equal competition for sites, and adsorption determined by physical forces only, that is, complete reversibility. 2.8.1 Descriptive Models Two descriptive models include a model proposed by Fritz and Schluender and an extension of the RadkePrausnitz three-parameter model (77, 96). The Fritz and

PAGE 95

83 Schluender model involves a general empirical equation of the form q io e, i (2-13) e, 1 c + ) a. C e, j lO where i and j refer to different solutes; n is the total number of solutes; and a. a. ., b. and c are constants lO 1] lO lO determined by the best fit of multi-solute experimental adsorption data (77). The equation includes the Langmuir competitive adsorption equation, the Jaeger and Erdoes relationship, a Radke and Prausnitz isotherm, the Freundlich equation, and the Langmuir single-solute equation when the empirical constants are assigned certain values. The model is not predictive because multi-solute adsorption data are required to estimate model parameters. An extension of a previously described single-solute three-parameter model was tested for multi-solute data predictability (96). The model is described by q, e, 1 (2-14)

PAGE 96

84 where A, B, and 3 are parameters determined from singlesolute isotherms, and the interaction terms, and n^, are determined from multi-solute system data. Both these models adequately describe multi-solute equilibria data for the concentration ranges of the specific solutes studied. The extended three-parameter model has been used in one dynamic column simulation model (96), However, because of the need for multi-solute data in order to estimate certain model parameters, these equations are valid only for the data sets to which they were applied. The model parameters are not readily correlated to any solute properties. Consequently, the models are not applicable, in a general way, to a wide variety of solute types. 2.8.2 Lanqmuir Competitive Model This model extends the Langmuir single-solute equation to a competitive, multi-solute model in the form r, 1 1 e 1 ^^e E (2-1 1 + y b. C j=l ^ ^'3 where i and j refer to the solutes in solution and the remaining terms are as defined previously for the Langmuir isotherm. The constants, Q and b, are determined from single-solute data. The model assumes equal competition

PAGE 97

85 for sites, complete reversibility, and no interaction among adsorbed solutes. This model was applied to the adsorption of alkyl phenols in bisolute systems (59). Single-solute adsorpton data were described by the Langmuir equation, but the competitive model did not adequately describe the bisolute data. Other studies also showed that the Langmuir competitive model did not perform as well as several other models (30, 74-76). The thermodynamic basis of the model was valid only when the constant, Q, was identical for all solutes (equal competition). 2.8.3 Semicompetitive Langmuir Model The Langmuir competitive model was modified to describe adsorption from a bisolute system when a portion of the adsorption occurs without competition (75). The modification was based on the hypothesis that such adsorption occurs when the constant, Q, in the competitive model was not the same for both solutes. The model was expected to be valid only when a fraction of the adsorption occurs without competition, such as when the larger of the two solutes is unable to enter the smaller pores of the carbon or when a site is unavailable to one solute due to the chemical nature of the site. The number of adsorption sites where noncompetitive adsorption would occur was assumed to be equal to Qj-Q2 in a bisolute system. The modified Langmuir competitive equation became, for Q > Q„,

PAGE 98

86 'e,l We,l 1 + 1 + b,C + b-C „ 1 e 1 2 e 2 (2-16) Q^b^C^ 2 %r2 1 + b,C '+ b„C 7 (2-17) 1 e, 1 2 e, 2 where the terms are as defined for the single-solute Langmuir equation. The noncompetitive term is the first term on the right side of equation 2-16. In the study that developed this modification (75), the semicompetitive model gave a much better description of adsorption data for four out of five bisolute pairs than the original competitive model when Q, 7^ Q_. When Q, = Q in one pair, the two models were almost identical. This was taken as evidence of noncompetitive adsorption due to surface heterogeneity on the carbon. Noncompetitive adsorption has been implied by other research (74, 76). In a study of competitive adsorption between selected fatty acids and phenolic substances, the semicompetitive model was significantly better in describing adsorption data compared to the competitive model (76). The most significant limitation to this model is its restriction to bisolute systems. The model also assxomes that unequal competition occurs only when However, unequal competition has been observed when Q, = Q (67).

PAGE 99

87 The model is limited by the ability of the single-solute Langmuir equation to describe isotherm data. Other limiting assumptions of the Langmuir development also apply. The model does illustrate the occurrence of unequal competition due to the heterogeneous nature of the carbon surface. 2.8.4 Competitive Solvophobic Interaction Theory The Solvophobic Interaction Theory discussed as a single-solute predictive modeling approach is being studied for application in multi-solute systems (90, 97). The approach is based on the application of several physicochemical parameters (the infrared [IR] shift parameter, a polarity parameter, and a steric parameter) with the solvophobic theory to predict the adsorption of organic homologues in multi-solute systems. The IR shift parameter is used to measure a solute's hydrogen bonding ability. It is determined by collecting acidity and basicity IR shift values (using an arbitrary scale) for various organics and then multiplying these values by the ratio of their molecular weight to percent aqueous solubility. This composite IR shift parameter is then normalized with respect to phenol. As the value of this normalized parameter increased, observed single-solute adsorbability also increased. From this it was concluded that molar adsorbability is proportional to acidity and

PAGE 100

88 molecular weight and inversely proportion to solubility (97) The polarity parameter describes the polar effect of the addition of a substituent group to a solute in a class of compounds when the parameter is correlated with singlesolute, maximum adsorption capacities. The Hammet equation is used to produce a parameter that reflects polar effects of metaand para-substituted derivatives of benzene. For reactions involving aromatic ortho-substituents and aliphalic compounds, the Taft equation provides the polarity parameter. Both parameters are tabulated in the literature for organic solutes commonly found in natural waters (97). In general, as these parameters increase, the polarity effect of the substituent decreases and adsorption increases. This agrees with solubility correlations since a decrease in polarity generally results in a decrease in solubility. Steric effects usually become important for aromatic substituents and a modified Taft equation is used to define a steric parameter. Values for this parameter are also available in the literature. Generally, as the steric effect of a substituent group increases, single-solute adsorbability increases, and there is a decrease in the polarity parameter. Therefore, polarity and steric parameters correlate inversely with single-solute adsorption.

PAGE 101

89 This approach to multi-solute equilibria predictability has not yet been well defined. Parameter data for a wide variety of organic solutes are limited. Comparisons of parameters are also restricted to classes of compounds. The approach also assumes reversible adsorption due to van der Waals forces only, which limit its potential use in adsorption systems where chemisorption is frequently encountered. 2.8.5 The Polanyi Competitive Model The Polanyi adsorption potential theory has been applied to adsorption from aqueous solution of single organic liquids and single and multiple organic solids (64, and the references listed therein). Given the correlation curves for two individual solutes in water, the effective molar volumes, and the solubilities of each solute in the presence of the other, the competitive correlation curves may be constructed for the more strongly adsorbing solute. From the competitive correlation curve the competitive isotherm can be constructed. Its application to multisolute systems of organic liquids partially miscible in water and completely miscible in each other led to a successful comparison with the Ideal Adsorbed Solution model (93). Binary and ternary solute systems were examined in this study. All work developing this approach has been done on one type of carbon. The limitations of the Polanyi model still

PAGE 102

90 apply (all pores accessible to the adsorbate, London-force adsorption only, effect of carbon ash content not included and limitations on predicting abscissa scale factors and adsorbate densities). Some problems may be encountered with group functionality and steric effects. The Polanyi Competitive Model requires molar volumes, refractive indices, aqueous solubilities, and the characteristic curve of the carbon as the data base. Not all of these parameters are readily available, especially aqueous solubilities. The development and application of the theory are in terms comfortable to physical chemists' point of view, and correspondence with other competitive models, such as the IAS model, are not well understood. The model promises to provide more insight into adsorption mechanisms and better predictability for multi-solute competitive adsorption equilibria. However, more testing is required using different carbons and matrices commonly encountered in water and wastewater treatment. 2.8.6 Ideal Adsorbed Solution (IAS) Theory The IAS theory was developed originally to calculate the adsorption equilibria for components in a gaseous mixture, using only data for pure component adsorption equilibria at the same temperature and with the same adsorbent (98). The theory was based on the concept of an ideal adsorbed solution, and the resultant expression was

PAGE 103

91 obtained using classical surface thermodynamics. The expansion of the theory involved the application of the thermodynamics of ideal dilute solutions to developing a method to calculate multi-solute equilibria, again using only data for single-solute adsorption (78). The central concept is that in an ideal solution, the concentration of a solute in a mixture is given by the product of the mole fraction of that solute in the adsorbed phase and the concentration of that solute which would exist in a singlesolute system at the same temperature and spreading pressure as the mixture. This equation has the form = (Z.) (C^) (2-18) where is the concentration of solute i in the mixture, is the corresponding single solute concentration, and is the mole fraction of solute i in the adsorbed phase. The thermodynamic basis for this theory assumes an inert solid with a specific surface area identical for all adsorbates. At equilibrium the chemical potentials of adsorbed and liquid phases are identical, and the application of the Gibbs-Duhem equation results in the Gibbs adsorption isotherm presented earlier (Equation 2-10). In this relationship the work of adsorption, represented as AdTT, is a function of the mass of solute adsorbed and the solute's chemical potential at constant temperature. The

PAGE 104

92 spreading pressure, ir, is defined as the difference between the interfacial tension (y) of the pure solvent-solid interface and that of the solution-solid interface at the same temperature : = Y pure solvent-solid y solution-solid (2-19) The details of the thermodynamic development are found in the original references (78, 98). The spreading pressure of a solute can be related to the single-solute adsorption equilibria according to RT o i ~ T Jq "Ji — (2-20) i where is the single-solute solid phase concentration of solute i, A is the specific surface area (surface area per unit weight), R is the universal gas constant, and T is absolute temperature. Since q. = f(C.) (2-21) through the mathematical representation of the single-solute isotherm (Freundlich, Langmuir, three-parameter models, e.g.), then spreading pressure becomes a function of C.. When solute species adsorb simultaneously from dilute solution at constant temperature and spreading pressure, the

PAGE 105

93 theory proposes that the adsorbed phase forms an ideal solution. Combining this theory with the Gibb's equation for isothermal adsorption (the thermodyanamic basis) results in (2-22) where is the total amount adsorbed from the mixture. Two other relationships needed for the IAS calculation are n Z. = 1 1 (2-23 and ^i (2-24: The calculation procedure begins with equations 2-18, 2-20, and 2-23 and a set of experimental values for the multi-solute equilibrium concentration for solute i (80). With an initial guess for Z^, an iterative procedure is used on a computer to solve this set of simultaneous equations. The iterative procedure results in Z^ and values that allow all spreading pressures to be equal, as required by the IAS theory. That is. = IT = TT = TT (2-25)

PAGE 106

94 Using the single-solute isotherm equation, q. can be determined (Equation 2-21). The total amount adsorbed from solution is then calculated From Equation 2-22. The amount adsorbed of each solute in the mixture can next be calculated using Equation 2-24. The difficulty in calculating multi-solute equilibria lies in the analytical solutions for the spreading pressure calculations (99). For accuracy in spreading pressure values, experimental data must be available over the loading range from zero to q^Consequently, although analytical solutions for equation 2-20 exist for Langmuir and Freundlich equations, unless Henry's Law is used, as the concentration approaches zero errors in tt will occur. Limitations of the Langmuir and Freundlich equations in representing wide concentration ranges have already been discussed. Some improvement in determining can be achieved by using the following form of the spreading pressure equation (100): A study of several three-parameter, single-solute models for use in the IAS model showed that use of the Toth isotherm, as well as a three-parameter model proposed in the study, in the IAS calculations represented experimental data well in several bisolute systems (82). However, when TT RT q. d log C. 1 i o ^ q^ (2-26)

PAGE 107

95 anionic species were present, large deviations were observed between predicted and experimental values. Another threeparameter, single-solute isotherm equation was used with the IAS theory to quantitatively predict breakthrough concentrations in continuous column studies (101). Single-solute data were represented by a set of Freundlich adsorption isotherm segments to simplify the application of the IAS theory (87). The method was tested with five, dilute, aqueous bisolute solutions on four different activated carbons with good success. The intent of the study was to provide an easier way to calculate spreading pressure rapidly and accurately for predicting multi-solute equilibria. A modified calculation procedure for the IAS model used solid phase solute concentrations instead of the original parameter based on an undefined weight of adsorbent. The procedure also included equation 2-2 6 for spreading pressure calculations; a modified three-parameter Freundlich model for single solute isotherm input; a new equation to allow prediction of equilibrium concentrations without using experimental data; and a numerical method to solve the simultaneous set of (5n+l) equations, where n is the number of solutes in the mixture (66). This procedure was tested with 10 sets of binary and ternay phenolic mixtures. The Langmuir competitive model in the IAS calculations was used as a comparison. The modified calculation procedure was

PAGE 108

96 successful and superior to the Langmuir isotherm model in describing the competitive adsorption of the bisolute systems studied. The IAS theory was also compared to the competitive Langmuir equation in predicting multi-solute equilibria for the calculation of multi-solute adsorption in fixed beds (102). The complexity of calculations for the two isotherm models varied with the intraparticle diffusion models used. Although The IAS theory required greater computational effort, it proved more successful in predicting breakthrough curves for the single and bisolute systems studied. The IAS model is the most theoretically sound of the multicomponent adsorption models. It can theoretically be extended to any number of competing components, and good agreement between experimental and predicted results has been shown. However, the model poses a formidable computational exercise, even using a computer, which increases enormously as the number of components increases. The iterative solution procedures contribute to the model's complexity. The model exhibited increasing deviations when low solute concentrations were studied or where low carbon doses were used (66). The model assumes an inert carbon surface with all sites equally available for competition and assumes complete reversibility. The accuracy of the model depends on the accuracy of the single-solute isotherm

PAGE 109

97 data and the isotherm model used to calculate spreading pressures 2.8.7 Simplified IAS Model A simplified calculation approach in the IAS model was suggested to reduce or remove the difficulties in determining spreading pressure curves for pure solutes in gas equilibria systems (100). The development of the simplification was described in an article reporting on the relationship of the simplified model to the original theory (99). In the model development, when all competing solutes have identical isotherms, the total loading in the mixture is the same as that in each single-solute system, providing the spreading pressure is also the same. Under these conditions, single-solute concentrations are equal to the total concentration of the mixture. For the case when solutes have different isotherms, hypothetical solutes are defined in such a way that all have the same isotherm equation and, thus, the same spreading pressure dependence. In the development of the model these hypothetical solute concentrations are defined in terms of actual concentrations. For multisolute systems the individual solid phase loadings are represented by 1/n' K. n 1/"' [I C. -) J (n'-l) = K' (n'-l) n' [K.C.] (2-27)

PAGE 110

98 where = the single solute Freundlich constant, K; n^ = the single solute Freundlich constant, n; K' = average value ; n' = average value n^. The simplified model is identical to the IAS model when single-solute isotherms are identical and for unequal isotherms when the isotherms are described by one Freundlich equation and n values are equal. When n and K values are unequal, the model can only approximate the IAS model (99, 103). The simplified model was in good agreement with the IAS model in the concentration range of 0.01-0.1 M but began to diverge at higher concentrations (99). The simplified model can also be extended to include any number of competing solutes. The model can be used in fixed adsorber beds because calculations of concentrations are just as possible as calculating loadings. However, the inability to account for chemisorption and unequal competition also limit this model, especially in columns with varying influent conditions. But, the great reduction in the number of computations while still maintaining IAS theoretical assumptions has made the simplified model very appealing.

PAGE 111

99 2.8.8 Improved Simplified IAS Model One of the criticisms of all multi-solute models is the inability to account for irreversible adsorption and unequal competition. Such phenomena arise from the heterogeneity of the carbon surface, especially varying pore sizes and specific surface functional groups. The simplified IAS model was chosen for modification because of the fundamental soundness of the model's development and the ease with which terms could be added to account for irreversibility and unequal competition for adsorption sites (104). Irreversible adsorption is most often observed in dynamic fixed-bed adsorbers where changes in influent concentrations can produce a pronounced hysteresis effect. Experiments performed to determine the degree of reversible adsorption of phenol on active carbon showed only about 50 percent reversible adsorption, even with long equilibrium times (49). Similar behavior has been observed for adsorption of gases by porous solids (4). Postulated mechanisms for the observed irreversibility of phenol included a breakdown chemical reaction after sorption on the surface and chemisorption of the phenol through '"-bond donor-acceptor complexes. Another study compared a rapid adsorption/desorption procedure to a slow equilibration technique with p-nitrophenols (21). No hysteresis was observed in the rapid method, but significant irreversibility was observed in the slower technique. A discussion of

PAGE 112

100 adsorption-desorption hysteresis in organic chemical-soil systems reviewed the many examples of hysteresis in adsorption and desorption experiments and proposed a model to account for this anomaly (105). A study of single-solute irreversibility of five low molecular weight substituted phenols showed significant irreversible adsorption (106). The irreversibility was influenced by sorbate functional group type and position. The study discussed and referenced several literature citations that showed that multi-solute adsorption equilibria models fail to accurately predict solid-phase loadings on activated carbon for certain system conditions. Unequal competition and irreversible adsorption due to a heterogeneous carbon surface were not accounted for in existing models and were postulated as the cause of much of observed multi-solute model inaccuracies. Modification of the simplified model resulted in the inclusion of a competition factor that improved model performance in several simultaneous addition studies (104). The competition factor was found to be significant only for the solute with a higher solubility. This competition factor was correlated with a solubility term for several solute pairs and gave the improved model a predictability capability. The competition factor was determined by using nonlinear SAS programming for each set of data. Modification of the simplified model to account for multi-solute irreversibility resulted in the evaluation of

PAGE 113

101 two irreversibility parameters in bisolute systems. Due to a lack of multi-solute irreversibility data in the literature, the study was limited in its evaluation of the two added parameters. Within the limited confines of the study, some improvement in describing sequential solute addition data was shown. A limited correlation of the irreversibility parameters with a solubility term also gave some degree of predictability to the improved model. The model was developed from bisolute adsorption data, and, therefore, its application to more complex solutions has yet to be tested. However, the model and its development do provide an increased understanding of the adsorption process with heterogeneous activated carbon surface as well as the causes and effects of irreversible adsorption and unequal competition.

PAGE 114

CHAPTER 3 RESEARCH OBJECTIVES The review of the literature has shown that the surface of activated carbon is heterogeneous. Carbon surfaces exhibit a wide variety in the distribution of pore sizes and surface energies. The heterogeneity in the physical structure of activated carbon greatly influences adsorption capacities and kinetics. Adsorptive behavior is also determined by intricate surface chemistries which reflect solute and solvent interactions with surface functional groups The literature has also shown that existing multisolute equilibrium models fail to account for carbon surface heterogeneity. The unequal competitive adsorption and irreversible adsorption arising from surface characteristics can be significant and cause these models to fail. When these equilibriiam models do not accurately describe, or predict, observed equilibria, then the dynamic predictive models using these equilibrium models will also fail. The overall objective of this research was to contribute to the state of the art for predicting multi-solute equilibria. Four organic solutes representative of an existing contaminated ground water aquifer at a hazardous 102

PAGE 115

103 waste site were selected to study competitive adsorption. The specific objectives of this research were to 1. Determine the extent of irreversible adsorption to improve understanding of the effects of irreversibility on existing models. 2. Determine the extent of unequal competition for adsorption sites and the effect of such competition on existing models. 3. To compare several multi-solute equilibrium adsorption models. 4. To test a modified model which attempts to include unequal competition in a predictive approach. To pursue these objectives, several phases of experimental work were conducted. Single-solute isotherms were developed, and the equilibrium model best fitting the observed data selected. Single-solute desorption studies were next begun to determine the extent of single-solute irreversible adsorption. Bisolute studies with sequential and simultaneous addition variations were conducted to evaluate competitive adsorption and compare existing models. Limited multi-solute experiments were also completed to examine model effectiveness in describing and predicting observed equilibria. Finally, the results of the bisolute equilibrium studies were used to test a model and its predictability potential.

PAGE 116

CHAPTER 4 MATERIALS AND METHODS 4 1 Adsorbent The adsorbent used in this work was Calgon Filtrasorb 400, manufactured by the Calgon Corporation, Pittsburgh, Pennsylvania. This bituminous coal-based activated carbon is commonly used for removal of organic contaminants from industrial and municipal wastewaters. The manufacturer's specifications are provided in Table 4-1. The carbon was supplied by the manufacturer in a 12 x 40 mesh size (1.68mm X 0.4 2mm ) Two different carbon particle sizes were used in this work. Batch isotherm experiments for single-solute systems were initially conducted using a carbon particle size of less than 200 mesh (sieve opening, 0.074mm). Subsequent continuous flow column experiments and other batch studies were conducted with a 100 x 140 mesh carbon size (0.149mm x 0.074mm). The powdered carbon was used to minimize equilibrium times in batch experiments. The 100 x 140 mesh carbon was chosen for use in continuous flow column studies to minimize carbon losses while allowing reasonable equilibrium times. 104

PAGE 117

TABLE 4-1. Characteristics of 12 x 40 mesh Calgon Filtrasorb 400 granular activated carbon. Parameter Value 2 Total surface area, m /g (N2/ BET procedure) 950-1050 Effective size, mm 0.55-0.65 Mean particle diameter, mm 0.9-1.1 Iodine number 1 n n n J. u u u Bulk density, Ibs/^t^ g/cm 25-26 0.40-0.42 Water soluble ash, percent 0.5 3 Total pore volume, cm /g 0. 94 Micropore volume, cm /g (pore radius < 2 nm) 0.23 Transitional pore volume, cm^/g (2 nm < pore radius < 50 nm) 0.35 Macropore volume, cm /g (pore radius > 50 nm) 0.36 Note: 1 nm = 10 Angstroms (A).

PAGE 118

106 The powdered carbon was prepared by pulverizing a quantity of the 12 x 40 mesh virgin carbon in a laboratory blender and sieving with a 200 mesh U.S. Standard Sieve. Carbon retained by the sieve was repulverized and sieved until more than 9 5 percent of the original carbon volume had passed through the 200 mesh screen. The powdered carbon was washed with distilled, deionized water that had been passed through a bed of granular activated carbon. Washed carbon was dried in porcelain dishes in a 105 C convection oven, cooled, and stored in a screw-cap glass bottle in a dessicator until use. The 100 X 140 mesh carbon was prepared by pulverizing and sieving 12 x 40 mesh carbon in a similar manner. Carbon which passed the 100 mesh sieve and was retained by the 140 mesh sieve was washed, dried, and stored. Approximately 20 percent of the original 12 x 40 mesh carbon volume was actually retained as 10 0 x 14 0 mesh carbon after pulverizing. The remainder passed the 2 00 mesh sieve. Washing, drying, and storage procedures were identical as those used for the < 200 mesh carbon. 4 2 Adsorbates The four organic compounds used in this investigation are listed in Table 4-2 and were chosen for several reasons. Three different classes of organic chemical s are represented by the selection: alcohols ( a-terpineol )

PAGE 119

107 0 (0 c a a (0 Q) -p 0) •H TD D P 1 03 >i 0) X c 01 0 0 'O !-i C C P >ix; 0 X Q4 H 0 E T3 -P 0 1 0) u •rr u (0 0 fN •H c (0 0 4-1 r-l 0 0 c 03 0) Q) x; •H a -P >-i >i Q) H a rH 0 (0 1 a (N tC 0 •H e x; u iH 73 0 C 03 10 Q) H U 10 1 0 0 •H 03 >i Si • 1 >i -P S-i M 1 -p H 03 c 0) Q \ 03 1 E — c 00 QJ H -H 0 1 D 00 in rH CT 1 a^ fH 1 -H X 1^ rH iH 4-t c •H .H o 03 H D CT t3 -H D CT •H U o Ln CN fC E O o (N P c •H o a •p c H o tji o C •H iH •H o u CQ o C71 O C -H 1^ -P rH ^ Q) U 2 o 0) E I Qjoo I >i X -H 0 >, u c >iX X a •H .H 0) 1 X o ^ 4J 4J ^ QJ OJ rM E ^ rH 0 1 H 1 > 1 (0 x: 0 c -a 0) rH -H u u U 3 0 u 0 0 0 -H C (C 0 in 0 0 Q) CN CN -P X 0 >( Dj-H X rH rH in 0 00 0 >i >, • • • 1-1 X 03 ^ in 0 T3 -P QJ CN (N ro >i 0) X E 0 1 1
PAGE 120

108 ketones ( 2 4-dihydroxyacetophenone ) and phenols (o-cresol and allyl phenol). All four are relatively small, neutral molecules with varying degrees of polar character and solubilities. Consequently, equilibrium with carbon should be reached rather quickly. All four are aromatic and, thus, are readily analyzable using a UV spectrophotometer for single-solute systems and high performance liquid chromatography (also a UV technique) for mixtures. All four are also present in a hazardous waste site currently on the Environmental Protection Agency's national priority list (107). The surficial aquifer at the site has been contaminated with leachate from wastes left in and on the ground by a no longer operating wood preserving plant. Studies are currently underway to determine a permanent site recovery scheme. The use of granular activated carbon is one of several alternative technologies being considered. These four compounds are also representative of the waste streams of several industries including wood preservation, wood pulping mills, and coal gasification plants. The organic compounds used in this research were procured from Aldrich Chemical Company, Milwaukee, Wisconsin. The 2 4-dihydroxyacetophenone, 2-allylphenol and a-terpineol were available at 98 percent purity. The o-cresol was available at 99+ percent purity. The physical and chemical properties of the solutes are listed in Table 4-2, as well as the symbols chosen to represent them

PAGE 121

109 throughout the remainder of this work. Stock solutions of 500 mg/L were prepared using a reagent grade, buffered dilution water (described below). Stock solutions were stored at 5C until needed. Solutions for analytical standard curves and solute quantitation were prepared by diluting aliquots of the stock solutions with more buffered dilution water. Feed solutions for continuous flow studies were prepared by transferring aliquots of stock solutions to volumetric flasks and diluting to volume with buffered dilution water. 4 3 Solutions A high purity, reagent grade, buffered dilution water was used for all experimental work. The pH of the buffered dilution water was adjusted to 6.0 + 0.2 for all experiments. The pH of 6.0 was chosen because it is below the pK values of the solutes studied and is more typical of contaminated groundwater and dilute industrial waste streams. At pH of 6.0 the solutes are predominantly neutral species in aqueous solutions. Consequently, adsorption resulting from force interactions between the carbon surface and dissociated, ionic species are avoided, as are solubility effects and neutral-ionic species interactions. The buffered dilution water was prepared by passing distilled water through a MILLI-Q water purification system manufactured by the Millipore Corporation, Bedford,

PAGE 122

110 Massachusetts. The treatment sequence consisted of activated carbon adsorption followed by mixed-bed deionization. The buffered dilution water was made in 15 L batches. The reagent grade water was collected in a 20 L glass bottle and 4 0 mL of a stock 1.52 M phosphate buffer solution and 12 mL of approximately 1.0 N hychochloric acid were added and mixed. The resulting solution was approximately 0.004 M with phosphate buffer and had a pH of 6.0 + 0.2 The 1.52 M phosphate buffer was made by adding 97.75 g of dibasic potassium phosphate (K2HP0^) and 126.0 g of monobasic potassium phosphate (KH^PO^) to reagent grade water and diluting to 1.0 L in a volumetric flask. All stock solutions were stored in screw cap glass bottles solely used for those reagents. A pH 10.0 + 0.2 buffered dilution water was prepared by adding 2.092 g of sodium bicarbonate (NaHCO^) and 2.64 g of sodium carbonate (Ua^CO^) to reagent grade water and diluting to volume in a 1.0 L volumetric flask. The resulting buffered dilution water was 0.025 M in bicarbonate and 0.025 M in carbonate. 4 4 Aqueous Solubilities The difficulty of obtaining aqueous solubility data from the literature resulted in the determination of approximate solubilities during this investigation. The method used was similar to the method reported in a study of the solubility of organic mixtures in water (108). To seven

PAGE 123

Ill 50 mL screwcap glass tubes were added 2.00-2.015 g of allyl phenol. Five 50 mL tubes were used for each of the remaining compounds, with approximately the same weight of organic added to the tubes. Approximately 3 0 mL of reagent grade water were added to each tube and a lined screw cap installed. Buffered dilution water was added to two tubes of allyl phenol to determine if a measurable change in solubility would occur in the presence of the 0.004 M buffer. The tightened caps were sealed with wax film and placed inside a double layer of twist lock bags. The bags, with the tubes sealed inside, were submerged in a temperature-controlled, mixing water bath and agitated for 8 7 hours at 25 + 0.5C. After removal from the bath the tubes sat for 48 hours to allow phase separation to occur. Aliquots of the saturated solution were withdrawn and diluted in volumetric flasks to a concentration range convenient for analysis. 4 5 Single-solute Isotherms 4.5.1 Glassware All isotherm studies were conducted with 125 mL Erlenmeyer flasks. All work with solutions was done with glassware only. All glassware was soaked in a chromic acid wash at the start of the experimental program. The glassware was rinsed, washed with hot, soapy water, triple rinsed with hot water followed by a distilled water triple rinse.

PAGE 124

112 methanol and methylene chloride rinses prior to air drying. The chromic acid wash was performed only once at the beginning of the study. However, all of the subsequent cleaning steps were used on all glassware between uses throughout the remainder of the study. As soon as the flasks were dry, an aluminum foil cover was placed on each for storage until next use. 4.5.2 Isotherm Preparation Single-solute adsorption studies were conducted using a bottle isotherm procedure where solute mass was constant and carbon mass was varied. Various amounts of activated carbon were weighed on weighing paper using an analytical balance and transferred to labeled 125 mL flasks with aluminum foil covers. Sorbate solutions were prepared by diluting aliquots from previously made stock solutions in volumetric flasks. To determine if different initial solute concentration affected single-solute equilibria, several concentrations were used for each solute and for both carbon particle sizes. Appendix A contains the single solute isotherm data and lists the initial concentrations used for each solutecarbon combination. Initial solute concentrations varied from approximately 50-500 mg/L with most data points collected at 50,100 and 200 mg/L levels. Aliquots of 100 mL of the sorbate solution were then transferred to each carboncontaining flask using volumetric pipets. Three additional

PAGE 125

113 flasks received 100 mL of sorbate solution, but no carbon, and were used as blanks to check for solute losses due to volatilization and/or adsorption on flask walls. Two other sets of replicates were set up for each isotherm. Each set consisted of three flasks containing the same carbon mass. The carbon masses chosen for these two replicate sets represented a low and a high solute final equilibrium concentration over the range of the experimental data. The flasks were tightly sealed by the use of a wax film over the aluminum foil cover. The isotherm flask sets, including blanks and replicates, were placed on a rotary shaker and agitated at 160 rpm for the selected equilibrium time of 10 days Isotherms for all solutes on both carbon particle sizes were determined at 16.5 + 0.5C in a constant temperature room (the room was in use for other, unrelated research as well). Isotherms for OC, ALP, and DAP using 100 x 140 mesh carbon were measured at room temperature (25 + 2C) as well to determine if temperature significantly altered the single-solute equilibria. 4.5.3 Equilibria Times One of the most common criticisms of single-solute equilibrium adsorption data is inadequate equilibrium contact time. A review of attainment of equilibrium in activated carbon isotherm studies with phenol (109) revealed

PAGE 126

114 "equilibrium" times from a few hours to several weeks. Significant differences in adsorptive capacities were reported, in excess of what was expected due to variations in carbon properties and environmental conditions. The study showed that granular-activated carbon took up to 3 weeks to reach equilibrium with phenol and up to 5 weeks for o-chlorophenol. Powdered carbon isotherms took from 3 to 5 days to reach equilibrium. Up to 80 percent of equilibrium values were reached in the first few hours, but the remaining capacity was utilized very slowly. The contact time required to reach equilibrium varies with the type and size of activated carbon used, the nature and concentration of the solute, and mixing conditions. The potential for microbial activity which can interfere with adsorption equilibrium studies must also be considered for extended contact times. Preliminary tests were run to determine the appropriate equilibrium contact time for this study. A series of 125 mL flasks were used with identical masses of carbon and initial solute concentrations. Two or three flasks were withdrawn for replicate analysis at various times into the equilibrium study. Blanks were also analyzed throughout the run. For the 100 x 140 mesh carbon DAP and AT reached equilibrium in less than 3 days, AT in less than 5 days, and OC in less than 8 days. A 10-day equilibrium period was assumed to allow a safety factor.

PAGE 127

115 4.5.4 Sample Analysis Samples for analysis were prepared by filtration using an all-glass, Millipore filtration unit and 47 mm filters. Glass fiber or 0.45 um membrane filters are commonly used. Whatman GF/A glass fiber filters (1.6 \im nominal diameter of retained particles) were selected for this study because of successful use of similar glass fiber filters in previous studies with phenolic compounds after testing for solute adsorption and carbon particle retention (66, 104). Two filters were used for each sample to prevent any carbon particles from passing into the filtrate. Whatman GF/F fibers (0.7 ym nominal diameter of retained particles) were also available but were almost four times the cost of the GF/A filters, and since the smallest carbon particle size in the 100 X 140 mesh carbon was theoretically 105 um, the additional expense was deemed unnecessary. No carbon was ever visually observed in any filtrate although some fines were captured in the second filter. Filtration was accomplished with only a slight vacuum from a faucet aspirator at a low flow of water to minimize any volatilization of solute during passage through the filters and glass frit. About half of the 100 mL suspension was passed through the filter initially to rinse the frit and glass walls. Then a 50 mL screw-cap glass tube was inserted into the filter flask under the filter down spout, and the remainder of the carbon suspension was filtered. This

PAGE 128

116 eliminated the need to clean the filter flask between every sample and flushed the frit and filters with approximately 30 to 40 frit void volumes of sample prior to any sample actually being collected. The first portion of the filtered suspension was discarded prior to insertion of the 50 mL collection tube. New filters were used for each sample. Filtrate samples were sealed by tightened, lined screw caps until analysis. 4 6 Single-solute Desorption Experiments were conducted to determine the extent of single-solute irreversible adsorption for each of the four organic compounds used in this study. A 1.08 L glass screw-cap bottle was used as the batch desorption reactor for each of the four solutes. The mass of carbon added to each reactor was 1.0000 + 0.0002 g. The corresponding single-solute isotherm was used to estimate the initial solute concentration at which to equilibrate the carbon in order to start with a solid phase loading on the upper end of the isotherm data range. The sorbate solution was made by diluting an aliquot of a stock concentrated solution to 1.0 L in a volumetric flask. The entire 1.0 L solution was added to the reactor and carbon. The reactor was closed with a lined screw cap and a wax film seal. All four reactors were placed inside a round drum which was rotated

PAGE 129

117 on a ball mill drive for a 10-day equilibrium period in the constant temperature room (16.5 + 0.5C). After the 10-day equilibrium period the bottles were removed from the rotating drive. The contents of each bottle were filtered in a similar manner to that described earlier for the adsorption isotherm flasks. After a sample of the filtrate was collected, as before, the carbon on the filters and on the funnel walls was carefully rinsed back into the reactor with a portion of the 1.0 L of solute-free buffered dilution water that would be added to begin the next desorption extraction. After the carbon was returned to the reactor, the remainder of 1.0 L was added and the reactor resealed. When all four reactors were resealed, they were placed again in the rotating drum for another 10 days, after which the filtration, filtrate sample collection, and carbon rinse procedures were repeated. This batchwise extraction, or desorption process, was continued until the liquid-phase solute concentration was less than the detectable limit of approximately 0.5 mg/L. It was assumed that the equilibration time for desorption would not be longer than that for adsorption. The mass of solute irreversibly adsorbed was calculated by, first, calculating the mass of solute adsorbed during initial equilibration, second, summing the masses desorbed in all the extraction steps, and, third, subtracting the total mass desorbed from the mass initially adsorbed.

PAGE 130

118 4 7 Analytical Procedures for Single-solute Analysis Each of the compounds studied adsorbed UV radiation; therefore, solute concentrations were analyzed by UV spectroscopy. All single-solute analyses were performed with a Perkin-Elmer Model 552 UV-Visible Spectrophotometer. All samples were measured at a 2.0 nm slit width using 1.0 cm standard fused quartz rectangular cells in a dual beam arrangement. Wavelength scans were conducted for each solute between 190 and 400 nm in order that the most useful wavelength for each solute could be selected. A representation of the wavelength scan results are shown in Figures 4-1 through 4-4. The solute concentrations for all four scans were approximately 2 0 mg/L. The wavelengths selected for the analysis were chosen by setting an arbitrary upper limit on concentrations to be measured and scanning for wavelengths at which absorbance conformed to Beer's Law throughout the desired concentration range. The concentration ranges and wavelengths used in this study are listed in Table 4-3. A series of nine standards covering the analytical concentration ranges for each solute were prepared before each set of sample analysis. The absorbance of each standard was determined and a least squares linear regression analysis performed to generate the standard curve by which sample concentrations were calculated from sample absorbance values. No standard curve regression equations

PAGE 131

119

PAGE 132

120

PAGE 133

121 D t5 30NVaH0S9V H

PAGE 134

122

PAGE 135

123 TABLE 4-3. Wavelengths and concentration ranges for singlesolute UV analysis. Concentration Compound Wavelength (nm) range (mg/L) o-cresol 211.0 0-40 allyl phenol 211.0 0-40 a-terpineol 193.0 0-30 2 4-dihydroxyacetophenone 227.0 0-40

PAGE 136

124 2 showed an R value (coefficient of determination) less than 0.995. The equilibrium solute concentrations, C^, were then used to calculate solid-phase concentrations, q^, by using the equation C -C q^ = _2_e (4-1) where C is the initial solute concentration and M is the o mass of carbon. These values, C and q were then used to e ^e calculate the adsorption isotherms. 4 8 Multi-solute Adsorption Studies 4.8.1 Continuous Flow Reactors A dual, continuous flow carbon contact system was used for all multi-solute studies. A schematic of the carbon contact system is presented in Figure 4-5. The dual system was designed to allow two independent experiments to be conducted simultaneously using only one pump drive. The feed container for each system was a 60 L glass bottle. Feed from the bottle was pumped through glass tubing to the top of the column by a dual Masterflex progressive cavity pump system. The pumping rate was adjusted by a variable voltage motor speed control. The only nonglass segment of the feed transmission lines was the tubing used in the pump heads. The tubing was Tygon F-4040-A, a tubing designed to be resistant to the adsorption of organics. Two model 7016

PAGE 137

125 GLASS COLUMN CARBON MASS GLASS WOOL SUPPORT — 1 STOPCOCKDUAL HEAD PUMP FEED BOTTLES EFFLUENT COLLECTION FLASKS FIGURE 4-5. Schematic diagram of continuous flow carbon contact column system.

PAGE 138

126 pump heads were used with this tubing. Glass tubing transmitted the feed to the top of each contacting column. Tubing life was in excess of 500 hours of operation at the chosen pump speed. The columns were 50 mL graduated glass burets (1.0 cm I.D.) with glass stopcocks. The supporting base for the carbon mass inside the columns was a Pyrex-brand glass wool. Teflon tubes carried the column effuent a short distance to glass effluent containers. Effluent was collected in a variety of containers ranging in volume from 25 mL volumetric flasks to 12 L glass bottles. All glassware used in the multi-solute studies was silanized. Nonsilanized glass contains surface OH groups which form hydrogen bonds with oxygen atoms in solute molecules and can result in high adsorption, especially when the nonsilanized glass has a large surface area, as does glass wool (110). On silanized material the adsorption of solute molecules is much weaker, since the hydroxyl groups are replaced by much less active siloxane groups. The silanization process was accomplished by soaking the glass wool in toluene containing 1 percent, by volume, of dimethyldichlorosilane (PGR Incorporated) for 8 hours followed by a methanol rinse and a methylene chloride rinse. To prepare for a column experiment the columns were first cleaned according to the standard procedure described earlier. A glass wool plug was wetted in a beaker of

PAGE 139

127 buffered dilution water. When the air bubbles had been removed, the plug was inserted into the buret, which was filled with buffered dilution water. The glass wash plug was evenly and lightly compacted in the bottom of the buret with a piece of glass tubing. In this manner all air was removed from the plug. Buffered dilution water was pumped through the system to insure that the plug was packed loosely enough to allow adequate flow with minimum head loss. The carbon mass selected weighed 0.4000 + 0.0002 g for all multi-solute experiments. Previous work found that this mass of carbon allowed reasonable equilibrium times while minimizing the effect of carbon loss and weighing errors (104). It was also found that less than 0.4 g of carbon resulted in desorbed masses of solute too low to be accurately measured. The carbon was measured on weighing paper with an analytical balance and transferred to a 50 mL beaker for wetting. The slurry was then rinsed out of the beaker and into the column using a small glass funnel. Buffered dilution water was pumped through the column to check the head loss across the carbon and glass wool and, if the head loss was small, continued to be pumped overnight to equilibrate the carbon with the solvent. Pump rates were maintained at 16.0 + 1.0 milliliters per minute. The long glass burets used for the columns allowed head losses to increase while maintaining a constant

PAGE 140

128 flow through the carbon bed. The stopcock was used to reduce head loss if the combined loss due to the partiallyopen valve, glass plug, and carbon threatened to overlow the buret. 4.8.2 Multi-solute Experiments Two different methods of solute addition to the contacting system were used. Simultaneous addition allowed the examination of competitive effects in twelve combinations of 3 solute pairs, four combinations of 3 solutes, and two combinations of 4 solutes. Sequential addition provided information regarding irreversible adsorption for solute pairs. Concentration ratios (mg/L:mg/L) of approximately 10:10, 10:100, 100:10, and 100:100 were used for both methods of addition for OC/AL, DAP/ALP, and OC/DAP solute pairs. Concentration ratios of approximately 10:10:100, 10:100:10, 100:10:10, and 10:10:10 were used in simultaneous addition of DAP/OC/ALP combinations and ratios of 10:10:10:10 and 10:100:10:10 were used in simultaneous addition of all four compounds. The observed adsorption equilibria, together with single-solute isotherms and physicochemical data, provided information regarding unequal competition and irreversible adsorption for a group of small, low molecular weight, neutral organic compounds.

PAGE 141

4.8.3 Simultaneous Solute Addition 129 A 30 L feed solution was prepared at the desired concentration ratio by weighing amounts of pure solutes in a 10 mL glass beaker and dissolving the solutes in 2.0 L volumetric flasks with buffered dilution water. After the first 2.0 L solution was added to the feed bottle, the same flask was used to measure the remaining volume of buffered dilution water. This minimized the amount of solute that may have remained on the walls of the flask initially. The buffered dilution water in the column was drawn down almost to the level of the carbon. The feed bottle was connected to the pump and the flow started. A stopwatch timed the experiment for effluent sample collection. At timed intervals effluent sample volumes were recorded and a portion of that volume stored in 15 mL screw cap tubes with teflon-lined caps until analysis. Depending on the initial concentrations of solutes and time into the experiment, sample volumes collected and time intervals ranged from 25 mL and about 3 minutes to 13 L and 12 hours. Data tables for each experiment are found in Appendices B through E and contain the time intervals of sampling and sample volumes used for that solute combination and concentration ratio. If sample analyses could not be completed concurrently with the experiment to determine when equilibrium had been achieved, then the experiment was continued well beyond previously observed equilibrium times. Upon completion of

PAGE 142

130 one experiment, the entire contacting system was emptied, rinsed with buffered dilution water, 1.0 N hydrochloric acid, and then buffered dilution water. The carbon and the glass wool support were discarded, and the buret, feed bottle, and effluent containers were cleaned with the standard washing method described earlier. 4.8.4 Sequential Solute Addition Each set of equilibrium experiments at the concentration ratios listed above was performed twice in order to reverse the order of solute addition. For example, for the solute pair OC/ALP, OC was the first solute added to the carbon, and when it had reached equilibrium, ALP was then added to the feed. The experiment was repeated at the same concentration ratios, except that ALP was the first solute added, then OC. After equilibrating the column and fresh carbon with water overnight, the carbon was exposed to the first solute of the selected pair. The feed preparation procedure was the same as for simultaneous addition. Efficient collection, volume measurements, and sample preservation were also conducted with the same procedures. After the first solute reached equilibrium, a new feed solution was prepared containing both solutes in the selected concentration ratio. The feed was initiated and sample collection began immediately. Since the carbon bed

PAGE 143

131 depth was so short, desorption of the first solute began almost immediately. Consequently, many samples were collected in the first few hours of the run. The same procedures were used for effluent collection, volume measurement, and sample storage as for the simultaneous addition experiments. The run continued until influent and effluent values were within the approximate error of the analysis. Finally, the concentration of the second solute was increased to the next desired concentration while that of the first solute remained the same. The feed preparation, sample collection, and volume measurements were repeated to generate a new set of equilibrium data at the new concentration ratio. A summary of the sequential addition process is given in Table 4-4 for compound A followed by compound B. The two experimental runs were repeated for B followed by A. 4.8.5 Multi-solute Analysis High Performance Liquid Chromatography (HPLC) was selected to analyze for individual compounds in solute mixtures. A Perkin-Elmer HPLC system, including a Series 2 high pressure pump, a constant temperature oven for the column, a Model LC-75 UV spectrophotometer, and a recorder, was used for all multi-solute sample analysis. An Alltech C18 reverse phase HPLC column provided solute separation at a constant column temperature of 3 0C. The mobile phase

PAGE 144

132 TABLE 4-4. Method for sequential solute addition. Compound A is followed by compound B. Concentration for A = 10 mg/L Concentration for A = 10 0 mg/L Carbon equilibrated with A alone Carbon equilibrated with A alone. New feed — B added at 10 mg/L and carbon equilibrated with A & B. New feed — B added at 10 mg/L and carbon equilibrated with A & B. New feed — B added at 10 0 mg/L and carbon equilibrated with A & B. New feed — B added at 10 0 mg/L and carbon equilibrated with A & B. End of experimental run End of experimental run

PAGE 145

133 selected was a 60:40 volume ratio of water and acetonitrile Several ratios were tested during preliminary work to determine the optimum ratio. All separations were performed isocratically and were generally clean and sharp. Flow rates were varied between 1.0 and 3.0 mL/min to minimize analytical time while maintaining good peak separation. The flow rates selected resulted in column pressures less than 2500 psi. Sample volume was 200 pL, determined by the loop injection system on the instrument. A series of standards of known concentrations was analyzed during each run to produce a standard curve for each solute. A linear regression analysis was used to determine the equation of the linear standard concentration/peak height relationship. Solute concentrations were calculated by substituting the sample peak heights into the appropriate standard curve equation. Wavelengths for each analysis were determined through a trial and error process and with the help of the singlesolute absorbance versus wavelength scans presented earlier. Where possible, wavelengths were selected to enable the analysis of two solutes at the same wavelength, reducing the number of times a set of multi-solute samples had to be analyzed in order to quantify all solute concentrations. The full-scale absorbance values on the spectrophotometer were also varied for each analytical run to allow the peak height of the maximum expected concentration to

PAGE 146

134 reach about 85 percent of full scale. This allowed maximum sensitivity for the expected range of concentrations.

PAGE 147

CHAPTER 5 RESULTS AND DISCUSSION 5 1 Single-solute Adsorption Studies The defined partitioning of a solute between liquid and solid phases is functionally expressed as an adsorption isotherm. The isotherm relates the solid-phase concentration of a solute to the equilibrium concentration of that solute in solution at a given temperature. Experimental isotherms have a variety of uses. They describe the capacity of an activated carbon to adsorb organic solutes from water or wastewater solutions. They provide information on the feasibility of adsorption for chosen solutes, for estimating carbon requirements, and for evaluating the performance of a variety of carbons under various loading conditions. More recently, adsorption isotherms have played an essential part in the development of predictive models and procedures for carbon system design and performance analysis. 5.1.1 Single-solute Isotherm Data Single-solute equilibrium studies in batch reactors were conducted for all four solutes as discussed earlier. The data are presented in Tables A2-A9, Appendix A. Two single-solute equilibrium models, the Freundlich and the 135

PAGE 148

136 Langmuir equations (Equations 2-6 and 2-2) were fit to the data using the least squares method in a linear regression analysis. The empirically determined constants for the two models and the coefficients of determination are presented in Table 5-1. The Freundlich model best described the data for all four solutes and both carbon particle sizes. The isotherms are presented on log-log plots in Figures 5-1 through 5-4 as linear traces of the fitted Freundlich equation. Each of the traces shown in the figures represent several initial solute concentrations. (Appendix A contains the initial solute concentrations used for each compound. ) The results indicate that the initial concentration had no detectable influence on the resulting isotherms for the solutes studied. This agrees with other researchers who evaluated the attainment of equilibrium in activated carbon isotherm studies (109). These researchers cited several examples where previous work showed lower capacities for higher initial concentrations. However, it was concluded that incomplete attainment of equilibrium would exhibit this effect and that no difference in adsorptive capacities would be observed if true equilibrium were reached. The lack of any observed effect on capacity with varying initial concentration in this study supports the assumption that equilibrium was attained in the batch isotherm experiments.

PAGE 149

<0 A +J 0 n •H 0 •H P u 0 to T3 (0 M mui ang n c 02 x; (1) 0 H •H -a iH T! -P C w 3 0) Q) >-l P fa D rH W-l 0 0 m c 1 QJ 0 rH (0 Cn -H C H to e 0 0 U U-l CM 05 to 4-1 C (0 p CP to c 0 CP o p -H 3 S (0 tJi Di E C \ (C 4-4 0 p 0) ja E to c o -H +J > P QJ 12 X! O c =1 o o u oi n n 00 r~ vo 1^ • • • o o o ^ ro Q^ 00 o -sr CN no n • • I o o o m rn rH IT) 00 o CTi • • • H O tN m rH r- CN CN ro o 0> TJ" ON iH • • o o ro O O O • • o o in o • • '3ro V£> rH fS CO CO VD • • o o CN ro o o ^ 00 00 rH o o O x: \ r~ vo o as O fO o ro o rH o o H o o rH rH rH H H • • • • • • • • • rH o o O o o o o O O T3 C =1 Q) P o cn ro rH CO o 00 r>• • • • • • • • in in in cri in o in 00 00 ro in o CN CN CN CN CN rH (N rH CM CN rH ro ro x: x: rH x; x: x; CO CO 0 CO CO rH CO (U QJ C a) 0) o QJ E E x: QJ E x; E x; QJ E x; rH CO x; CO CO c CO 0 O O QJ a O QJ O QJ •rl O QJ w ^ e to "^JE a E 0) rH rH rH rH T3 rH p rH p X X O X O 1 X O 0) X o u o o o rH O O o o +J o o 1 o o og rH O CN O (N 1 O CN 0 rH rH V to rH V CN rH V s rH V

PAGE 150

138

PAGE 151

139 o o o H (6/6uj) fa

PAGE 152

140 (6/Buj) w,x

PAGE 153

141

PAGE 154

The relative adsorptive capacities with two different mesh sizes for various equilibrium solution concentrations of the solutes are shown in Figures 5-5 and 5-6. The observed trend in capacities did not correspond to trends in relative molecular weights or solubilities. Table 5-2 contains the experimentally determined solubilities for all four compounds. From the solubility data, OC would be expected to have the lowest adsorption capacity and DAP and AT the highest. The solubilities did decrease as molecular weight increased, but neither parameter could be used to explain or anticipate the observed relative adsorptive capacities. Allyl phenol exhibited the highest capacity over the entire 3 log equilibrium concentration range for both carbon particle sizes. For the powdered carbon isotherms (Figure 5-5) DAP adsorptive capacity was least, whereas for the 100x140 mesh carbon (Figure 5-6) AT was least adsorbed. The relative adsorptive capacities also changed as equilibrium solute concentrations increased. With the powdered carbon the order of increasing adsorption was DAP, OC, AT, and ALP, with AT and OC having very similar adsorptive capacities. However, at the upper end of the equilibrium concentration range the order was OC, DAP, ALP, and AT. At the upper concentration range both OC and DAP exhibited similar capacities as did AT and ALP. For the 100x104 mesh carbon only OC and DAP reversed relative capacities over the equilibrium concentration range and

PAGE 155

143 oo Z uj (6/61U) lAi/x

PAGE 156

144

PAGE 157

145 TABLE 5-2. Experimental solubility data. Compound Solubility Standard deviation (g/L) g/L mM/L o-cresol (OC) 21.60 199.7 0.35 allyl phenol (ALP) 6.337 47.2 0.053 2 4-dihydroxyacetophenone (DAP) 2.298 15.1 0.014 a-terpineol (AT) 2.393 15.5 0.034

PAGE 158

146 showed similar capacities in the 10-4 0 mg/L equilibrium soluble concentration range. The relative adsorption behavior observed in this study illustrates the difficulty in predictive modeling efforts for neutral, low molecular weight organic studies representing different classes of compounds. The adsorption mechanisms for such compounds are a complex combination of physical and chemical force interactions that are the result of carbon surface heterogeneity, molecular physical and chemical characteristics and solvent-solute interaction. Hydrophobic compounds adsorb in a much more predictive manner and correlate fairly well with solute characteristics such as molecular weight or solubility. This is because physical forces alone drive the adsorption to equilibrium. Similar success in predictability is also expected for ionic solutes where classical electrostatics determine the extent of adsorption. But the combinations of mechanisms for neutral, low molecular weight solutes of varying polarity preclude any rational explanation of relative adsorptive capacities, such as those observed in this work. The reported isotherms describe a set of adsorption conditions that also include an organic loading factor. This carbon loading is expressed in terms of initial solute concentrations as milligrams of solute per milligram carbon. Preliminary studies indicated that organic loadings greater than 1.5, in general, resulted in the onset of

PAGE 159

147 multilayer adsorption and a rapid increase in solid phase solute concentrations over a very short range of equilibrium solution concentrations. This rapid deviation from the Freundlich model could be induced at any equilibrium concentration and was typical of Type II isotherms for the onset of multilayer adsorption in the Brunauer classification of isotherms (7). Organic loadings for all experiments were kept below this critical organic loading and provided an upper boundary for the adsorption systems studied. The Freundlich model was found to give a good fit to the data throughout the concentration ranges studied. Deviations at the lower end of the equilibrium concentration ranges corresponding to dilute solution theory and Henry's Law were not observed. However, the Freundlich model will not reduce to Henry's Law when solute concentrations become low enough. Since the deviation points for the isotherms in this study were not determined, the isotherm equations have, as a lower boundary condition, the lowest measured equilibrium solute concentration. The traces of the isotherms in Figures 5-1 through 5-6 represent the entire range of applicability of the Freundlich model parameters reported here. Extrapolations above and below these ranges may or may not conform to the model. At some point below the lowest equilibrium solution concentration conformance to Henry's Law will occur. The Freundlich model can be modified to include a Henry's Law term if lower equilibrium

PAGE 160

148 concentrations are expected. A three-parameter model, such as the Toth equation discussed earlier in Section 2.5.5, could also be used, but the constants must be determined from the data by a nonlinear regression analysis. 5.1.2 Effect of Carbon Particle Size The initial experimental approach for this work was to use powdered carbon for single-solute isotherm studies and 100x140 mesh carbon for single-solute desorption and all multi-solute studies. The use of powdered carbon results in rapid attainment of equilibriiam. The larger carbon particle size facilitated desorption studies where recovery of the carbon mass was required after each extraction. The larger particle size also minimized carbon losses through the glass wool support in column studies. However, calculated solid phase loadings from powdered carbon isotherm equations did not agree with observed loadings of the 100x140 mesh carbon in desorption experiments. Several flasks were prepared for each solute using the 100x140 mesh carbon to compare solid phase loadings with those observed for the powdered carbon. These flasks were equilibrated by the same method used to develop the powdered carbon isotherm data. In all cases the 100x40 mesh carbon exhibited lower adsorption capacities than the powdered carbon over the entire equilibrium concentration range. Consequently, all single-solute isotherm studies were repeated using the 100x140 mesh carbon.

PAGE 161

149 The results of the additional isotherm determinations indicate that carbon particle size significantly affected equilibrium adsorption capacities. The isotherm data for both particle sizes for each solute are included in Appendix A. Figures 5-1 through 5-4 illustrate the observed effect of carbon particle size. In all four figures the adsorption capacities are lower for the larger particle size; however, the isotherm slopes changed only slightly. Since isotherm slope values are related to adsorption energies, the effect of particle size is to alter the number of sites and/or the surface area available to the sorbate. The effect of carbon particle size on adsorption capacities could be a result of several factors. Failure to attain true equilibrium will show the same trend of reduced capacity with increasing particle size. However, successive isotherms for a series of particle sizes will also show increasing slopes as particle size decreases. This reflects the more rapid attainment of equilibrium for smaller particles at the larger concentration gradients at higher "equilibrium" concentrations. In this study preliminary experiments were completed to determine when equilibrium was reached for each compound in batch adsorption experiments. A two-day safety factor was added to the equilibrium time of the slowest solute. Yet, all four solutes exhibited the same behavior. The absence of any observed effect by different initial solute concentrations

PAGE 162

150 supports the assumption of equilibrium in this study. In a previous work with 100x140 mesh Calgon F400 carbon, equilibrium was attained in less than eight days for OC and four other low molecular weight neutral phenolic compounds (104). Therefore, failure to attain equilibrium is not considered to have caused the observed particle size effects A second cause of observed particle size effects on adsorption capacities involves significant differences in sorbate molecular dimensions and weight and a change in the pore size distribution caused by the grinding process. Should the micropore volume fraction increase and the transitional pore volume fraction decrease, then a powdered carbon would show increased capacities for the small, low molecular weight sorbates and a reduced capacity for high molecular weight sorbates. All four solutes used in this study were small, low molecular weight compounds very similar in dimensions and molecular weight. Consequently, some other factor caused the observed capacity reduction. The nature of the carbon particle may result in significant variation in activity as the particle is ground and sieved into size fractions. Activation conditions may have been such that the inner cores of the 12x40 mesh carbon were not completely activated and possessed a lower pore volume than the outer shells of the particles. Grinding away the outer shell and retaining only the inner core would then

PAGE 163

151 result in lower adsorption capacities for larger particle sizes containing the less active inner cores. The smaller particle sizes would contain the more active outer shell fragments and exhibit higher adsorptive capacities; thus, decreasing particle size would result in an increase in adsorptive capacity. The larger fraction of micropore and transitional pore volume would be lost as the outer fragments pass the 14 0 mesh sieve. The inner cores retained by the 140 mesh sieve would contain significantly less micropore volume, some of which may exist in sealed, unavailable channels. The effects of particle size on adsorption capacities observed in this study are probably due to such an unequal pore volume distribution with particle radius caused by activation conditions. However, further work is required before definitive conclusions can be drawn. The isotherm traces for OC in Figure 5-1 include a trace from a previous study (104). This isotherm was developed under very similar conditions to those used in this study. Carbon particle size and preparation, OC concentrations, equilibrium times, and isotherm procedures were similar. Since the referenced isotherm was conducted at 25C and this study's isotherm at 16.5C, the referenced isotherm should have shown lower capacities, not higher, if temperature affected the results. Selected points on the 16.5C isotherms in this study were also determined at 25C, and no effect of temperature was observed. The most

PAGE 164

152 reasonable explanation may involve the method of grinding used. Both carbons were prepared in high speed blenders, but differences in blender configuration, blade design, and speeds could have resulted in particle splitting in the referenced study and particle shearing in this study. If the particles were split, then outer and inner pore volume distributions would be retained. A shearing process would leave behind a lower capacity carbon particle. The effect of carbon particle size should be examined in any study where a particle size fraction does not include at least 95 percent of the original starting material. Conclusions about the commercially available carbon particle size, such as 12x40 mesh carbon, cannot be made when laboratory studies use smaller selected size fractions without determining the effect of particle size. The evaluation and composition of equilibrium models, however, can be accomplished without knowing the effect of particle size on capacities, since the carbon's unique characteristics are held constant throughout such studies. 5 2 Single-solute Desorption Studies Desorption studies were conducted to determine the extent of irreversible adsorption of each of the four organic solutes using the 100x140 mesh carbon. The carbon was first equilibrated with a solute concentration calculated from isotherm data. The chosen concentration

PAGE 165

153 corresponded to the highest equilibrium concentration measured during the isotherm experiments. The desorption/ extraction procedure should retrace the adsorption isotherm, if the adsorption were truly ideal and reversible. Two of the four desorption experiments were terminated prior to completion of the extraction because of equipment failure. The desorption data for all four solutes are listed in Table A-1, Appendix A. Figures 5-7 and 5-8 illustrate the desorption isotherm compared with the adsorption isotherm for two of the four solutes. A clear hysteresis effect is evident indicating significant irreversible adsorption. Table 5-3 lists the weight percent of each solute that was irreversibly adsorbed. The hysteresis effect is well known in the adsorption of gases where molecular and pore sizes are of similar dimensions. The resulting large surface area of contact can exhibit strong bonding energies, even though the forces are physical, not chemical, in nature. Consequently, long equilibrium times are required to allow for the slow diffusional process to remove the sorbed molecule. When the equilibrium solution concentrations in this study were below the analytical detection limit of 0.1 mg/L for o-cresol and allyl phenol, equilibration continued for 20 days. No detectable desorption was observed following this extended contact time. This indicated that failure to attain equilibrium was not a significant factor in the study. The

PAGE 166

154 (B/Buj) w/x 'ONiaVOI 3SVHd QHOS

PAGE 167

155 H (6/6ui) lAl/X 'ONiaVOI aSVHd QHOS

PAGE 168

156 TABLE 5-3. Single-solute adsorption irreversibility. Percent Compound irreversibly adsorbed o-cresol 87.8 allyl phenol 84.7 2, 4 -DAP 70.6* a-terpineol 69.5* *Desorption experiments terminated before reaching lower detection limits in liquid phase equilibrium concentrations. Completion of the extraction would have resulted in slightly lower percent irreversible adsorption.

PAGE 169

157 observation that initial solute concentration did not affect solid phase loadings during isotherm determinations also supports the assumption of equilibrium in the 10-day contact time. These results agree with other studies in which initial solute concentrations and different procedures for isotherm development were evaluated as experimental biases in explaining observed hysteresis (105, 106). The differences in the amount of irreversible adsorption among the solutes could be caused by several factors. The interaction of benzene Tr-electrons with partial positive charges on carbonyl or quinonic surface functional groups as a chemisorption mechanism has already been discussed. Altering this basic phenolic bonding mechanism are the type and position of solute molecule functional groups. These structural variations may intensify site bonding energies through inductive effects; however, much work is needed to substantiate such a hypothesis. Steric interferences may not permit sufficient solute-surface contact to allow the T-bond, charge transfer mechanisms to function. Hydrogen bonding on a molecule like DAP may form a large hydration layer and may hinder the solute from close approach to a surface site. The extent of irreversible adsorption will also vary from carbon to carbon since no two carbons have identical surface structure or chemistry. Results of previous research on desorption using single-solute batch systems have been conflicting. In a

PAGE 170

158 study of eight commercially available carbons the amount of irreversibly adsorbed phenol constituted only 3-4 percent of the total adsorbed (26). Irreversible adsorption for p-nitrophenol was less than 5 percent of the total observed adsorption. However, in another study, irreversible adsorption of phenol was approximately 5 0 percent of total adsorption (99). In studies where short equilibrium times were used for initial adsorption (on the order of hours), little or no irreversible adsorption was observed. However, in studies where equilibrium times were much longer and the second, slow, diffusion-controlled phase of adsorption allowed to reach equilibrium, there is evidence of significant irreversible adsorption. To improve predictive models by accounting for irreversible adsorption, a more complete understanding of single-solute irreversibility is needed. Currently only one model has been modified in an effort to account for the results of surface heterogeneity. The level of understanding of the mechanisms behind irreversible and competitive adsorption is rudimentary, at best. The approach used in the model modification was to add parameters and estimate their value by statistical curvefitting analysis to produce a best fit of existing multisolute data. The development of a mechanistically fundamental model or statistically relating large amounts of multi-solute data to some physicochemical properties are beyond existing capabilities.

PAGE 171

159 5 3 Competitive Adsorption 5.3.1 Bisolute Simultaneous Addition All current multi-solute equilibrium models assume a homogeneous surface and equal competition for adsorption sites on that surface. Activated carbon, however, has been shown to be heterogeneous. The nature of the surface functional groups, especially those containing oxygen, has already been discussed. The single-solute desorption results in this study imply that adsorption to such specific functional groups was the principal mechanism of adsorption for the four chosen solutes. The results of the bisolute simultaneous addition studies are found in Tables 5-4 through 5-6. The results in Table 5-4 show that ALP adsorbed to a much greater extent than OC. The relative adsorptions observed are consistent with relative molecular weights and solubility. Allyl phenol has a higher molecular weight, lower solubility, and greater hydrophobicity from the larger allyl group on the benzene ring. Single-solute isotherm data reflect the expected relative surface concentrations. However, the large differences in adsorption of each solute at similar equilibriiim solution concentrations were not expected. Adsorptive capacities for each solute in single-solute systems differed by less than 10 percent at equilibrium concentrations of 10 and 100 mg/L. In the bisolute system

PAGE 172

160 TABLE 5-4. Simultaneous addition equilibrium data for ALP/OC solute pair. ALP OC Experiment X/M X/M (mg/L) (mg/g) (mg/L) (mg/g) 1 9.5 135 9.9 39 2 10.0 63 103 146 3 97.0 234 9.8 8 4 98.5 240 96.5 52

PAGE 173

161 the capacities for ALP and OC differed by more than 75 percent. Such results could be explained by adsorption rates observed during preliminary equilibrium studies in which ALP adsorbed much faster than OC, which adsorbed at a rate slower than any of the remaining three solutes. The higher affinity for carbon exhibited by ALP with a much faster mass transfer rate would result in significant unequal competition for sites between ALP and OC. The high degree of irreversibility discussed earlier would preclude any significant desorption of ALP and adsorption of OC once the OC molecule reached the vicinity of the adsorption sites The results for the ALP/DAP solute pair are included in Table 5-5. In this bisolute system DAP outcompeted ALP for adsorption sites. The DAP molecule has a higher molecular weight and a lower solubility than does the ALP molecule. However, single-solute isotherm data show ALP to have a much greater affinity for the carbon than DAP (see Figure 5-6). Based on the isotherm data, ALP would be expected to outcompete DAP for adsorption sites. A possible explanation for this apparent anomaly may be a difference in mass transfer rates (bulk transport) that would allow the DAP molecule to reach available sites faster than AL molecules. No mass transfer rates were measured during this study, but the results from the preliminary single-solute equilibrium time determination indicated that DAP had the highest rate

PAGE 174

162 TABLE 5-5. Simultaneous addition equilibrium data for ALP/DAP solute pair. ALP DAP Experiment X/M X/M (mg/L) (mg/g) (mg/L) (mg/g) 1 9.8 61 9.7 149 2 10.1 15 94.1 288 3 91.0 187 9.1 56 4 100.5 125 96.5 190

PAGE 175

163 of adsorption of all four solutes. The combination of greater hydrophobicity higher mass transport rates, and irreversible adsorption would allow DAP to outcompete ALP. Since the chemisorption mechanisms for DAP and AL are not known, it is difficult to assess the effect of type and density of carbon surface functional groups that may be specific or preferential for DAP and ALP. The results for the OC/DAP solute pair are found in Table 5-6. The DAP molecule outcompeted the OC molecule for sites. Based on molecular weight and solubility considerations, such a result could be anticipated. The singlesolute isotherms for DAP and OC are almost identical (see Figure 5-6), especially near 10 mg/L, one of the bisolute initial concentrations. At 100 mg/L DAP showed a slightly higher capacity (less than 5 percent). Consequently, given similar mass transport rates, the two solutes should have exhibited similar capacities. However, since OC showed the slowest rate of adsorption and DAP the highest rate in equilibrium time determination, DAP successfully outcompeted OC for sites by occupying them first. Little displacement would occur since significant irreversible adsorption has been shown for both solutes on the carbon used in this research. Again, the type, density, and specificity/ preference of carbon surface functional groups are unknown, and their effect on unequal competition cannot be defined without further research.

PAGE 176

164 TABLE 5-6. Simultaneous addition equilibrium data for OC/DAP solute pair. OC DAP Experiment X/M X/M (mg/L) (mg/g) (mg/L) (mg/g) 1 10.3 35 9.8 170 2 9.4 6 90.0 278 3 101.0 143 10.1 135 4 105.0 53 94.0 94

PAGE 177

165 In all the bisolute experiments, the total adsorptive capacity for both solutes together did not exceed the single-solute adsorptive capacity of the carbon for the solute with the highest capacity of the pair. This is consistent with an earlier study of a four-component system where only about 60 percent of the anticipated carbon adsorption capacity from single-solute capacities was realized (57). However, others have reported an overall increase in the total carbon adsorptive capacity for multisolute mixtures (51). The equilibrium capacity for each solute on the activated carbon was adversely affected by the presence of another solute. One explanation of the slight decrease in total adsorptive capacity lies in the use of available surface area. If the total surface area of the activated carbon available for the adsorption of two solutes was not more than the area available for one solute, then sharing this area with a less effectively adsorbed compound would result in a less efficient use of the available area. Since all four solutes are fairly similar in molecular weight and size, such an assumption is reasonable. The above results confirm the heterogeneous nature of activated carbon with respect to adsorption sites and the physicochemical nature of the solutes. These effects, such as unequal competition, are not included in existing multisolute equilibrium models.

PAGE 178

166 The structural nature of the solutes can increase or decrease adsorption effectiveness. The addition of functional groups to the benzene ring of the phenol molecule affects polarity and stericity and, consequently, adsorbability. The decrease in polarity of OC by the addition of the allyl group in AL and the other groups for DAP and AT have obviously affected solubilities. Increases in stericity also increase hydrophobicity improving adsorption effectiveness. Polar and steric effects may partially explain the more effective adsorption of DAP over OC even though their single-solute isotherms are very similar. Assuming a perpendicular orientation with adsorption occurring at the hydroxyl end of the molecule, the hydroxyl and methyl group hydrogen atoms interfere with each other. The resulting steric arrangement is spatially expanded and relatively inflexible and would reduce intraparticle diffusion. However, the meta and para groups on the DAP molecule could form an intramolecular hydrogen bonding arrangement reducing the molecules' stericity and improving micropore transport. 5.3.2 Multi-solute Equilibrium Model Application Several existing multi-solute equilibrium models were tested using the previously discussed results. The Competitive Langmuir model and the Semicompetitive Langmuir model require single-solute isotherm determinations and the

PAGE 179

167 calculation of the Langmuir constants describing those isotherms. The Simplified IAS model also requires singlesolute isotherm data but uses constants from the Freundlich model describing the isotherms. Because empirically derived constants from single-solute data are required, the models are difficult to use in a truly predictive manner. The models also assume equal competition and reversible adsorption in the adsorption process. Since the results of this, and other recent work, indicate significant irreversibility and unequal competition for the solutes studied, the above models would not be expected to adequately describe the data. The results of the application of these three models are presented in Tables 5-7 through 5-9 for the simultaneous addition of the ALP/OC, ALP/DAP, and OC/DAP solute pairs. The experimental results are included for comparison. The Simplified IAS model was recently modified in an attempt to include the effects of unequal competition and irreversible adsorption (104). The Simplified IAS model was selected for empirical modification for several reasons. This model uses the Freundlich parameters, which often yield more accurate descriptions of single-solute data. The model, when modified by placement of empirical terms, promised to describe multi-solute equilibria even when unequal competition and irreversible adsorption are not significant.

PAGE 180

CN to Ln \ \ ?s in t— 1 00 o ro in O ^ CN CN \ iH t \ fit i-< 10 t— 1 r \ \J P • 00 Ln LO o in H 00 CO c \ i-f to ?s rn UJ O (11 n \J n, ri +J 1 vj r*^ r** i-H ^ P +-* o^ ro lO iH rH ro 10 CN LO n CN H \ rH r— ( St* r* r s \J (11 O •H rn \ UJ \ 10 m 10 r \ 1— 1 rH o ro I 1 W in 4-' H i-H 1 1 \ rrt ill /II CD ( J r\ \J r* -t r\ \J n to 0 u o M "H r—i rn O iH > -a C •H •H -H rH i-l QJ 1 0 rH 0) 0) to Ck •H a, en a U Ch ffl -d a< -u E C e E C e w a E w rtl 0 X to 0 0 to H < E -H < M s U CO O hq CO H H C/2 H

PAGE 181

169 to <0 3 rH (0 > •H U ja -H •H a -p 3 H O 0) A s \ • X M • • 0 o Pn •H -P rH < r~o ro 00 n3 CTi 00 in CM o o >-i \ H in CN X H rH c 0 -H +J (0 < 00 CN CN CTi )H Q 00 vo -P rH 2 (N CN CN c • \ X u c 0 rH u o < in o m rH s rH iH \ iH X 00 0) o Q X s \ X cri in in s rH in o CN \ X vo rH CTi o CN u rH 0 m rH cn tC 0) -p -P > > T3 H c •H •H a) Q) c d) P u P M •H ^3 -H D B h -h H -H IW 0) iP H -P 3 -P 3 •H > -H Sh 1 0) g H 0 rH
PAGE 182

(0 (0 > -H H u •H (0 a 4J •H M 3 o tr w (U C^ 0) < W Q ta \ ao I •H O f-i o c W O •H 0) 1-1 T! (0 0) 03 E 3 T3 O C (0 (0 <1) -P U I in H E-i '9' n m n < o Q CN s \ X o • u IT) o O O s in o iH in in fM x; \ Ijl E • • \ < in o o !Ti >-( Q n 00 00 ro E • O \ X CD 0 O •H • u 4-1 r-H o n o 00 (0 O o S-l rH iH 00 c 0 •H +J 03 1-1 < 00 00 00 in -P Q r-C o 2 r~ in Q) \ U X c 0 • u o o 00 00 s; 00 "J \ H X < 00 Q o o o 00 • S in \ H n n X • o u o in in o 2 m n 00 o \ X O s +J c OJ E H i-l Q) (0 Q4 -J-l X to W > •H 4-1 H P 0) E C O to U J I H E (D > H 4-1 •H 4-1 a Cn 6 C O to -H 3 'a q; •H E w H < W H O CN O VO 'aCN in 0) o 14-1 > -H O ^ a E cQ E -H < H W H

PAGE 183

171 Unequal competition for adsorption sites was included in the Simplified IAS model by adding terms as shown below : ( H'-l ^ ^ l/n^ ^ 1/n' ^"'-1) ^^Ti ) ^• K. ,r) n. qi = ^ t TT ^i ^--^ ^i ^) J (5-1) 'l N where I (K /n ) y n = and „ = i The Freundlich constant, K^, was modified to allow adjustment of K values of individual solutes such that the solute with a larger K would not outcompete a solute with a lower K. The term K is proportional to the magnitude of a single-solute isotherm. The largest K results in an isotherm that is higher than a lower K-value isotherm. All multi-solute models predict the solute with the larger K to outcompete the solute with the smaller K. In this work DAP outcompeted ALP even though ALP had a significantly higher Freundlich K value. The DAP molecule also outcompeted the OC molecule even though the K for OC was higher. A reduction in the K values for ALP and OC would result in improved model performance. The values for the competition term were determined by nonlinear Statistical Analysis System ( SAS ) programming which provided the best fit of the model to the data. The values for the competition terms for each solute pair are

PAGE 184

172 listed in Table 5-10. From the magnitude of the values it is apparent that both terms are significant in the model development. The results of the application of the improved Simplified IAS model are also included in Tables 5-7 through 5-9. In all three solute pair simultaneous addition studies the improved Simplified IAS model gave, by far, the best description of the experimental data. All models predicted ALP to outcompete OC. The two Langmuir-based models failed to predict the relative effectiveness of adsorption. This is not surprising since the single-solute Langmuir model failed to describe that data. The Simplified IAS model fared better, primarily because it uses Freundlich parameters. The Freundlich model accurately described the single-solute adsorption data in this work. It is interesting to note that the Semi-competitive Langmuir model gave predictions that are in greater error than those of the Competitive Langmuir model. The Semicompetitive Langmuir model is intended to correct for adsorption on sites that are not subject to competition. However, the model assumes that unequal competition occurs only for the solute with the larger solid-phase capacity. When solutes with equal or lower solid-phase capacities outcompete the solute with the larger solid-phase capacity, the model fails. The results in Tables 5-8 and 5-9 describe just such situations, and the Langmuir models both fail even to predict the more effectively adsorbed solute. The Simplified IAS model predicted

PAGE 185

173 TABLE 5-10. Competition terms for the improved Simplified IAS model. Solute Pair ^ (1/2) ALP/OC 1.321 1.360 ALP/DAP 1.285 1.105 OC/DAP 1.528 1.372

PAGE 186

174 the more effectively adsorbed solute in all cases except three. The improved Simplified IAS model provided a much more accurate description of the order of effectiveness and the magnitude of the equilibrium solid-phase loadings. 5.3.3 Model Predictability The modification of the Simplified IAS model required parameters from multi-solute data. This provided an improved description of the data but did not result in any model predictability. If the model is to be applied to multi-solute mixtures, some predictability is desired to avoid excessive and extensive laboratory analyses. In the work which initially improved the Simplified IAS model (104), the competition factors were correlated with a solubility term. Solubility was chosen because, of all the solute physicochemical parameters, solubility best correlated with adsorbability as discussed earlier. The form of the solubility factor is (S S ) q (5-2) ^=1 where S is solubility and S^^ > S2. This allows for discrimination between equal (Sj^-S2) values when and S2 may differ. Plots of the solubility factors are shown in Figures 5-9 and 5-10. The trend of increasing competition factor

PAGE 187

175 FIGURE 5-9. Correlation between the solubility factor and the competition factor, rii •

PAGE 188

176 1.8-1 1.61.41.21.00.80.6. 02 a ALP/ DAP 04 0.6 0.8 to '1 FIGURE 5-10, Correlation between the solubility factor and the competition factor, n,,.

PAGE 189

177 with increasing solubility term agrees with a similar trend in the original model modification work. In this work, both competition factors are significant. In the previous work only the competition factor associated with the higher solubility sorbate was found to be significant. Although much more work is needed to verify the correlation and determine the effects of different carbons on such a correlation, the results provide some measure of predictability to the model. Thus, knowing what constituents are present, what the single-solute Freundlich isotherm parameters are for those constituents, and the constituent solubilities, improved predictions of multisolute equilibria are possible. The extension of this predictive model to multi-solute systems may require the development of a lumped solubility parameter and combined Freundlich parameters to represent the combined effects of all solutes in the calculations. 5.3.4 Multi-solute Simultaneous Addition Four tri-solute and two four-solute simultaneous addition experiments were conducted to demonstrate the effect of unequal competition in more complex matrices. The adsorption data are tabulated in Appendix E, Tables E-1 through E-6. The final equilibrium solid and fluid phase concentrations are presented in Table 5-11.

PAGE 190

178 TABLE 5-11. Multi-solute simultaneous addition equilibrium data and predicted values DAP OC ALP AT c o 9 4 10.1 94.0 X/M-observed 64 2 4 274 -predicted 11 15 306 — C o 9.9 102.0 9.6 X/M— observed 99 107 4 7 -predicted 20 218 75 — C o 97.0 9.9 10.2 X/M-observed 271 6 5 15 -predicted 248 19 56 — C o 10 3 10 4 9.8 X/M-observed 146 22 71 -predicted 50 64 175 — O y 6 10.2 10.0 10.0 X/M-observed 101 19 -predicted 55 73 195 10 C o 10.3 101. 0 10.0 10.0 X/M-observed 73 92 -predicted 24 204 78 4 Note: C is the solution feed concentration in mg/L. X/M is the equilibrium solid phase loading in mg/g.

PAGE 191

179 In the first tri-solute experiment the feed concentration ratios of DAP:OC:ALP were 9.4:10.1:94.0 mg/L. The effect of the higher concentration of ALP agrees with what would be expected from the relative isotherm plots in Figure 5-6. Allyl phenol had the highest single-solute capacity of the three compounds in this experiment. The remaining solutes showed very similar isotherms at equilibrium concentrations around 10.0 mg/L. The large difference in the solid phase loadings between the two demonstrates the effect of significant unequal competition. Differences in mass transfer rates, surface specif icity /preference and irreversible adsorption all probably contributed to the observed unequal competition. The combined adsorption of all three solutes (340 mg/g) was only slightly more than the single-solute capacity (274 mg/g) of the most effectively adsorbed solute, ALP. This indicates that the available surface area for ALP was essentially the same area available for OC and DAP. This implies that the density of solutespecific surface functional groups was low, and all three solutes were competing for the same sites. In the second tri-solute experiment, the feed concentration ratios of DAP:OC:ALP were 9.9:102.0:9.6 mg/L. The total adsorption for all solutes (253 mg/g) was slightly less than the OC single-solute capacity (294 mg/g). This again implies that most sites were active adsorption sites for all three solutes, although not of equal adsorption

PAGE 192

180 energies. The DAP molecule, at less than 10 percent of the OC feed concentration, showed an almost equivalent solid phase equilibrium concentration. This demonstrates again the unequal competition between two solutes whose singlesolute isotherms nearly coincide. The success of DAP in out-competing OC and ALP for adsorption sites is evident in the results of the third tri-solute experiment. In this experiment the DAP:OC:ALP feed ratios were 97.0:9.9:10.2 mg/L. The total adsorption for all three solutes (293 mg/g) was only slightly less than the single-solute solid phase capacity for DAP (306 mg/gm) at 97.0 mg/L feed concentration. Over 90 percent of the total solid phase organic loading consisted of DAP. If equal competition occurred, the ALP fraction should have been much higher, based on relative single-solute adsorptive effectiveness (see Figure 5-6). In the fourth tri-solute experiment the feed concentration ratios of DAP:OC:ALP were 9.6:10.2:10.0 mg/L. The most effective adsorption was exhibited by DAP. If equal competition had occurred, then the DAP and OC solid phase loadings should have been much closer, based on singlesolute capacities. The ALP molecule should have been the most effectively adsorbed solute in this experiment, assiaming equal competition. The effects of unequal competition were again clearly evident. The argument that such an order in adsorption effectiveness could have been expected based

PAGE 193

181 on relative solubilities fails when confronted with the single-solute isotherm data. The two solutes with almost identical single-solute isotherm traces (OC and DAP) had very different solubilities (DAP: 0.3 percent, OC: 2.2 percent) Results from the two experiments using four solutes each confirm the previously observed trends. The solid phase equilibrium concentrations of ALP and AT were not measured because peak separation on the liquid chromatograph did not occur with the column (C8) used. Other mobile phases and mobile phase concentrations as well as another column (CIS) failed to separate the two solutes. Unequal competition was further demonstrated by the time varying concentrations of the experiments as they approached equilibrium. In all four of the tri-solute experiments significant amounts of OC were initially adsorbed, then desorbed as the slower adsorbing solutes displaced OC. This unequal competition would have been more extensive if significant irreversible adsorption had not occurred. The time-concentration profiles in Tables El through E4 imply that adsorption thermodynamics favor the adsorption of DAP over OC. This could also be viewed as functional group preference for DAP over OC due to the chemisorption mechanism. The actual chemisorption mechanisms for all four solutes are unknown but probably involved the Ti-electron system of the benzene ring and

PAGE 194

182 specific functional group interactions discussed in the literature review. Such time-concentration profiles typify dynamic column effluents when influent composition and concentration change. In order to develop dynamic models that accurately predict such observed behavior, equilibrium models must be improved to include the effects of the type of unequal competition observed here. Application of the Simplified IAS model to the tri-solute and four-solute experiments show similar effects of unequal competition. The predicted values of equilibrium solid phase concentrations for DAP all significantly underestimate the observed solid phase loadings. The model overpredicts the OC and ALP equilibrium solid phase loadings. A comparison of observed and predicted values is shown in Table 5-11. Some of the deviation must be attributed to the model development itself. The Simplified IAS model was developed to streamline calculations, and, consequently, some accuracy was sacrificed. However, the consistent trends in overand underpredicting equilibrium values illustrate the problems inherent in the assumption of equal competition and irreversible adsorption that both the original IAS and Simplified IAS models have made. The application of the improved. Simplified IAS model discussed for multi-solute studies cannot be accomplished until a satisfactory procedure can be developed to estimate competition factors for all solutes in the matrix. Several

PAGE 195

183 different approaches in this study to use the bisolute correlations to arrive at competition factors in the threesolute system were unsuccessful. The results of this research indicate that a competition factor for each solute could be significant. The work in which the improved model was developed found that only the competition factor of the solute with the lower Freundlich K value was significant (104). Since only these two studies have attempted to use the improved model, much more work is needed to determine the validity and applicability of the competition factor approach. Extension of the improved model beyond bisolute systems is essential if the model is to be of value in multi-solute dynamic column modeling approaches. 5 4 Sequential Solute Addition Sequential solute addition studies can yield information concerning irreversible adsorption in bisolute systems. If irreversible adsorption is a significant event, then current multi-solute equilibrium models will not be able to accurately describe such data. Single-solute irreversible adsorption has already been discussed. A series of sequential solute addition experiments were performed to determine whether or not irreversible adsorption was a significant mechanism in explaining deviations from expected behavior. Current multi-solute equilibrium models assume complete reversibility.

PAGE 196

184 The results of the sequential addition experiments are provided in Table 5-12. The data are arranged by solute pairs and relative concentration ratios. The calculated amount of irreversible adsorption is tabulated for the first solute of the pair fed to the contactor. The amount of irreversible adsorption can be calculated by X/M.^ = VM^^g X/M^.^ (5-3) where X/M is in mg/g and the subscripts refer to irreversibly adsorbed (ir), adsorbed during sequential addition (seq), and adsorbed during simultaneous addition (sim) (104). Adsorption during simultaneous addition includes all effects of competition. The effect of irreversible adsorption is minimized since both solutes compete simultaneously for available sites. However, in sequential addition the sites at which both solutes can adsorb are already occupied by the primary solute. The chemisorption bonding energies have a wide range, and a series of displacement, or desorption/adsorption, reactions can occur. To compare simultaneous and sequential addition, all experiments must be done at the same concentration ratios. The reversible adsorption described in Table 5-12 generally is less at lower concentrations. When the carbon is equilibrated first with a single solute, the solute begins to adsorb on the most thermodynamically favored

PAGE 197

185 TABLE 5-12. Irreversible adsorption for bisolute systems of differing equilibrium concentration ratios (mg/L:mg/L). All values in mg/g. Solute Concentrations ir reversibly adsorbed (mg/g) 10:10 10:100 100:10 100:100 o-cresol OC/ALP OC/DAP 49.6 114.5 77.4 135.5 84.4 2.7 165.0 72.9 allyl phenol ALP/OC ALP/DAP 61. 6 52.0 112.5 87.6 46.4 61.3 26.9 82.9 DAP DAP/ALP DAP/OC -33.0 24.4 6.4 4.5 -19.8 -10.4 -1.1 14.3

PAGE 198

186 sites. Consequently, the chemisorption bonding energies are greatest at these sites. When these sites become occupied, the solute adsorbs on less energetically favored sites, and so on, until physical adsorption is the final mechanism. At lower solute concentrations a larger solute mass fraction is adsorbed at high energy sites. As the solute concentration is increased, the adsorption occurs primarily on lower energy sites. These are the sites where desorption begins. The high energy sites are where irreversible adsorption occurs As the second solute is added, it, too, adsorbs by thermodynamic preference. However, the desorption energy of an already adsorbed solute must be provided. Obviously, where the primary solute is most weaky bound, there the energetics are most favorable for a desorption-adsorption interaction. If the primary solute is at a low concentration, most molecules will be bound on the high energy sites, and the greatest irreversibility is expected. As the secondary solute concentration increases, the effect on solute-solute repulsive energies improves the thermodynamics of displacement, and more desorption of the primary solute from higher energy sites is expected. When the initial concentrations of the primary solute are higher (100 mg/L versus 10 mg/L), then a larger mass fraction is adsorbed on lower energy sites. The addition of a second solute will result in greater desorption, or lower irreversibility. The

PAGE 199

187 results in Table 5-12 show, in general, higher reversibility at lower solute concentrations and less reversibility at the 10:100 mg/L ratio. The extent of irreversibility is effected by concentration ratios. The difference in irreversibilities supports the concept of a wide spectrum of chemical and physical bonding energies. If most of the primary solute is bound on high energy sites, then the percent reversible adsorption will be less. If the primary solute is bound to a significant number of low energy sites, then the percent reversibility will be greater. The least amount of reversibility was demonstrated by OC, followed by ALP. The DAP molecule exhibited very little irreversibility with either OC or ALP at any concentration ratio. Such results indicate a rather strong chemisorption bonding energy at a large number of sites for OC. The bonding energies for DAP seem weaker. In the singlesolute irreversibility studies DAP exhibited the lowest percent of irreversible adsorption of the three solutes used in this phase of study. However, in the simultaneous adsorption studies, DAP competed unexpectedly well in bisolute mixtures. This seems to indicate that competitiveness had a greater influence on final equilibrium conditions than irreversibility. The heterogeneity of the carbon surface and its functional group chemistry results in a myriad of adsorption phenomena spanning adsorsption energies from the weak, physical adsorption to the strongest of the

PAGE 200

188 chemisorption bonds. The nature of the solute also plays a role in competitiveness and irreversibility. No current multi-solute equilibrium model attempts to include the effects of irreversible adsorption. The improved version of the Simplified IAS model was also modified to attempt to empirically alter the equation by inclusion of statistically determined irreversibility parameters (104). Some improvement in model description of adsorption data was observed. Attempts to correlate the parameter with solute properties, like the previously discussed solubility term, showed some trends but lacked sufficient data to draw any conclusions. Little data are available in the literature to be used to validate the model. A successful competition parameter inclusion is a prerequisite for model modification for irreversibility as well.

PAGE 201

CHAPTER 6 SUMMARY AND CONCLUSIONS The described work was undertaken to examine the effects of activated carbon heterogeneity on the competition and irreversible adsorption of four organic solutes in multi-component systems. Single-solute studies to develop isotherms and to determine the extent of irreversible adsorption were followed by bisolute simultaneous and sequential addition studies. Several bisolute equilibrium models were compared and a recent improvement in the Simplified IAS model tested. Several conclusions are drawn from this work. 1. Carbon particle size can significantly alter isotherm equation parameters and must be considered in experimental design. 2. Good single-solute isotherm data and accurate mathematic descriptions of the data are critical for the success of equilibrium and dynamic predictive model development for the carbon adsorption process 3. The solutes studied in this research showed significant irreversibility in single-solute systems, confirming the importance of 189

PAGE 202

190 chemisorption at specific sites as an important mechanism in carbon adsorption. 4. Unequal competition for adsorption sites was a significant factor in the adsorption of the solutes used in this research. Current multisolute equilibrium models did not successfully describe the experimental data. 5. Irreversibility in the bisolute experiments was significant and is not included in existing descriptive or predictive equilibrium models. 6. An improved version of the Simplified IAS model was tested for description and prediction of unequal competition in adsorption. The improved version was superior to three other models in describing the experimental data. A correlation between competition factors and relative solubility terms indicated a predictive capability for the bisolute version of the model.

PAGE 203

CHAPTER 7 ENGINEERING SIGNIFICANCE As the understanding of the adsorption process improves, mathematical descriptions of the process find more application in design strategies. Several researchers are developing and applying dynamic adsorption models in an effort to reduce the time and costs associated with extensive preliminary and pilot scale studies. Currently, design criteria are based on empirical rules of thumb or singlesolute adsorption parameters. Lumping all solutes into total organic carbon or chemical oxygen demand parameters is no longer adequate, given the concern for, and presence of, small concentrations of toxic or hazardous organic solutes in the nation's water resources. Design strategies must change to be able to optimally separate such solutes from aqueous systems, be they industrial waste streams or contaminated rivers and aquifers. The development of predictive, dynamic, multi-solute models attempts to optimize the design process at minimum cost. Essential to the success of such dynamic models is the equilibrium portion of that model. Current multi-solute equilibrium models fail to adequately describe adsorption equilibria data where significant unequal competition and 191

PAGE 204

192 irreversible adsorption occur. Then models are based on the assumptions of equal competition and complete reversibility. Improved understanding of these phenomena can lead to improvements in the multi-solute equilibrium models and eventually in the design process. The improved. Simplified IAS model has the potential to significantly upgrade existing dynamic models. The use of solubility data, single-solute isotherm parameters, and knowledge of the constituents of a waste stream make the application of the model practical. However, the model must be modified for application to multi-solute systems beyond two-compound mixtures before its application in design or process evaluation can occur.

PAGE 205

APPENDIX A SINGLE SOLUTE DESORPTION, ADSORPTION, AND SOLUBILITY DATA

PAGE 206

o r-i 1 1 1 X rIX) in r— 1 r— 1 rH •H fo ^^ H vU ID o •H 1 fH 3 i (L) X 00 CN CO in in CN 0) X C iH 0 0 H u c (0 > x; a •H 0 (C 1 0) u (N in 00 00 in iH (C tC og (S O 00 in o in O -p X in o 00 £> in in 00 00 CN OSJ CN CN CN CN CN o c 0) x: rH o Q) M U I O X ^ IX) v£> o O in 00 CN CM in CN X 13 X X 00 00 CN in 00 00 VO iH O in iH 00 IX> 00 r-l O CTl CTi 00 00 00 00 00 CN CN CN OM cy> (Ti 00 cx) in in 00 X) 00 CN H O 00 o iH 00 o in in T (N O o VD 1X> in OM CN CM OM CM CN CN 4•H B •H H Ci 0 H 4J 0 (U • P c 0) 0 -0 •H -p u (C M 0 +J • H X c (U 0 CD -H c 4-1 p -H 0 u x: (0 0 rO -p 0) (0 X J-) Q) Q) Cn i-l -p c 0 H • M-) u c Q) 03 0 JQ in ^3 •H (0 +J e t3 u 0) (U (U +J Xi -p ns -p U C 0 T3 •H iH to E 0 QJ 4-1 (fl Xi 0 -p 0) CO +J w (0 •H -p ro CO 6 c x; g •H 0) e -H t3 -P •H 0 .H rH a 0 0 0) X m H X X K 194

PAGE 207

195 TABLE A2 Isotherm data for o-cresol, 100 X 140 mesh carbon. o v^aj-Don xoauiny e ( ma /T. ) ( a ) \ my / my ) { mrr /T \ \ mg/ X) ; img/g; f) n 9 s> 7 u z o 9 4 9 9 9 51 6 U KJ \J O n 9 s A. K 9 9 K Z Z D 51 6 KJ • \J ^ \J W u z ^ Q 9 9 Q 5 ] 6 n 0 9 0 R u • u ^ u o U Z D 4.x 9 9 C 51. 6 0.0200 0.26 5.3 232 51. 6 0.0191 0.27 6.1 238 •J J• \j U • U X O *x n 9 fi o X "3 C Z JO J J. • u U • U J. / z ATA X U 4 o /I n 2 4 0 SI fi u • u X u / X U o O /I c Z 4 D ^ X u u • U X o / U J X X U D O /I c Z 4 D ^ J. u U • U X D / U J X X U 4 ? n fi n n 0 7 01 U • U / U X n 9 Q O O 1 o ^ o Z DZ ^ \J u • u 0 0 R Q Q n 1/1 u o 4 /in o 4 U J O T "7 2.11 ^ w \J \J 0 0 ^ Q Q J y J T "7 O z 78 206.0 0.0599 0.34 37.1 282 206.0 0. 0499 0.41 60.0 293 ^ W
PAGE 208

196 TABLE A-3. Isotherm data for o-cresol, < 2 00 mesh carbon. o (mg/L) Carbon mass (g) Carbon loading ( mg/mg ) e (mg/L) X/M (mg/g) 0 0 .0 207. 0 207.0 207.0 207.0 207.0 207.0 207. 0 207.0 207. 0 207.0 207. 0 207.0 207.0 207.0 207, 207, 207, 207.0 207.0 207.0 207.0 103. 0 103. 0 103.0 103.0 103.0 103. 0 103.0 103. 0 103. 0 103. 0 103. 0 519 519 519 519 519 0. 0851 0. 0801 0.0750 0.0750 0.0750 0.0700 0. 0651 0.0602 0.0555 0.0499 0. 0450 0.0450 0.0450 0.0399 0.0349 0. 0299 0251 0200 0199 0200 0. 0100 0.0500 0.0500 0.0401 0.0350 0.0300 0.0250 0. 0200 0.0200 0. 0201 0.0149 0. 0101 0.0500 0.0500 0.0500 0.0450 0.0401 0.24 0.26 0.28 0.28 0.28 0.30 0.32 0.34 0.37 0.42 0.46 0.46 0.46 0, 0, 52 59 0.69 0. 83 1. 04 1. 04 1. 04 2.07 0.21 0.21 0.26 0.29 0.34 0.41 0.52 0.52 0.51 0. 69 1. 02 1. 04 1. 04 1.04 1. 15 1. 29 1. 1. 3, 3, 3, 6, 10.5 17.1 25.5 34.1 47.5 48.0 47.0 62 77 92 109 127 128 128.0 167.3 0 0 2 6 13 23 36.8 37.1 36.8 51. 3 66.0 310 309 306 324 340 242 256 272 271 271 287 302 315 327 346 354 353 356 363 372 383 389 396 395 395 397 205 206 251 277 299 317 332 331 330 348 368 418 420 426 433 446

PAGE 209

197 TABLE A4 Isotherm data for allyl phenol, 100 X 140 mesh carbon. o Carbon mass Carbon loading C e X/M (mg/L ; (g) (mg/mg ) ( mg/L ) (mg/g one A z U b U n ^ r\ 0.0502 0.34 29.1 292 one n Z U D U n n e n n 0 UbOO A A ^ 0.41 5 0.4 309 205.0 0.0550 0.37 42.6 295 205. 0 0. 0550 0.37 42.5 295 one n z U b U U 0550 0.37 41.5 297 one n Z U b (J 0.0400 A C 1 0.51 77.7 318 one n / (J b U 0 OzOO 1.03 138.8 331 one n A A O A A 0.0200 1. 03 140.1 325 one n z U D U A A T A A 0 UzOU T AT 1.03 138.6 332 1 n o n 1 (J Z U A A A 1 A 0 0zl9 0.47 37.0 297 T n o n 1 U Z (J A A A A '5 0 0z03 0.50 42.3 294 1 n o n i U Z U A A A A 1 0.0203 0.50 4 0.3 304 1 n o n 1 U z U n n o n o 0 Uz03 0.50 4 0.7 302 1 n o n X U z U n n 1 n o 0 UXy J 0.53 43.4 304 1 n 0 n J. U z U n n 1 o >i U UXo4 A C C 0.55 46.5 302 102. 0 0.0178 0.57 46.4 312 102. 0 0.0171 0.60 49.9 305 1 n 0 n X u z u n n 1 c U UXob A ^ A 0.62 52.8 298 1 n 0 n X U Z U n n T c n U UXoO 0.64 53.0 306 1 n 0 n X u z u n n T ^ n U (J Xb U A ^ A 0.64 52.5 309 1 n 0 n X u z u A men U UXo U A ^ A 0.64 52.6 309 1 n 0 n X u z u n n 1 c ^ U UX54 0.66 53.6 314 X u Z D n n 0 "7 e U U J / b A AT 0.27 7 3 254 X U Z D n n o c U U 0 J D 0.31 11.5 271 X U Z D n n "J "3 c n o T 0.31 13 2 266 1 n 0 e X u z b n n o "3 U U J J D A AT 0.31 12 6 268 1 n 0 e X U z b A n o o e 0 0zo5 0.36 21. 8 283 102.5 0. 0239 0.43 33.9 287 102.5 0. 0202 0.51 42.1 299 X U Z • D n n 1 c /I U U X D 4 n ^ o 0.63 54.2 295 102.5 0.0097 1. 06 72.2 312 102.5 0.0097 1. 06 71. 3 322 102.5 0.0097 1. 06 71. 9 315 102.5 0.0064 1.60 82.3 316 530 1.000 0.53 194 336

PAGE 210

198 TABLE A-4. Continued. Carbon mass Carbon loading X/M (mg/L) (g) (mg/mg) (mg/L) (mg/g) 51. 5 0,0219 0.24 1. 2 230 51. 5 0.0203 0.25 2.4 242 51. 5 0. 0203 0.25 2.3 242 51. 5 0.0203 0.25 2.5 241 51. 5 0. 0194 0.27 3.7 246 51. 5 0.0185 0.28 5.1 251 51. 5 0.0179 0.29 5.3 258 51. 5 0.0171 0.30 6.8 261 51. 5 0.0165 0.31 7.2 268 51. 5 0.0160 0.32 8.0 272 51. 5 0.0160 0.32 8.1 271 51. 5 0.0160 0.32 8.3 270

PAGE 211

199 TABLE A5 Isotherm data for allyl phenol, < 200 mesh carbon. C o Carbon mass Carbon loading c X/M (mg/L) (g) ( ma/ma ) (ma/L ) 103.4 0. 0401 0.26 0.5 257 103. 4 0.0349 0.30 1. 8 291 103.4 0.0303 0 34 6 1 *j ^ J. 103.4 0. 0199 0 52 28.2 78 103.4 0 0200 0 .52 28.2 103.4 0. 0199 0 52 2 8.5 376 103.4 0. 0100 1.03 6 2 7 4 07 207.5 0.0800 0.26 0 4 ^ .J o 207.5 0.0750 0 28 0 9 97S 207.5 0.0750 0.28 0 9 275 207.5 0.0750 0 .28 0.8 27S 207.5 0.0704 0 .29 1.5 292 207.5 0. 0651 0 .32 3 0 313 207.5 0.0603 0.34 6.5 333 207.5 0.0550 0.38 12 7 3 53 207.5 0.0502 0 41 21.9 ^ \J ^ 207.5 0. 0451 0.46 3 4.0 ^84 J O *i 207.5 0. 0451 0.46 32.7 ^ O \J 207.5 0. 0451 0.46 3 3 3 R S O O J 207.5 0.0398 0.52 47.4 401 2 07.5 0.0349 0.59 62.2 415 207.5 0.0302 0.69 7 9 8 1 ^ X 207.5 0.0249 0 83 9 9.6 4 1 J J. 207.5 0.0200 1.04 12 0 0 *4 J Q 207.5 0.0200 1. 04 J^ \j \j 4 1 0 J 207.5 0. 0200 1. 04 120. 0 435 514 0.0501 1. 03 277 473 514 0.0500 1. 03 277 474 514 0.0499 1. 03 277 475 514 0. 0402 1. 28 316 493 514 0. 0349 1. 47 340 499

PAGE 212

200 TABLE A-6. Isotherm data for 2 4-dihydroxyacetophenone, 100 X 14 0 mesh carbon. C Carbon mass Carbon loading C X/M (mg/L) (g) ( mg/mg ) (mg/L) (mg/g ) AO A 0 0z42 0.20 1. 7 193 48.4 0. 0226 0.21 2.2 204 48.4 0.0214 0.23 3.0 212 48.4 0.0214 0.23 3.1 212 48.4 0.0214 0.23 3.2 211 48.4 0. 0204 0.24 3.8 219 48.4 0. 0195 0.25 4.7 224 48.4 0.0189 0.26 6.1 224 48.4 0.0183 0.26 6.2 231 48.4 0.0177 0.27 7.1 233 48.4 0. 0177 0.27 7.2 233 48.4 0.0177 0.27 7.1 233 48.4 0.0172 0.28 8.6 231 4 fi A 'i O H U UlDD 0.29 9.3 237 96.0 0.0413 0.23 4.8 221 96.0 0.0373 0.26 7.5 237 96.0 0.0314 0.31 16.2 254 96.0 0. 0314 0.31 16.8 252 96.0 0. 0314 0.31 16.8 252 96.0 0.0265 0.36 24.7 269 96.0 0.0223 0.43 34.8 274 96.0 0. 0185 0.52 43.4 284 96.0 0.0143 0.67 54.5 290 96.0 0. 0107 0.90 65.1 289 96.0 0. 0070 1.37 75.3 296 96.0 0.0035 2.74 85.0 314 192.5 0. 0547 0.35 37.6 283 192.5 0. 0502 0.38 45.9 292 192.5 0. 0461 0.42 58.7 290 192.5 0.0461 0.42 59.0 290 192.5 0.0424 0.45 66.2 298 192.5 0. 0384 0.50 75.4 305 192. 5 0. 0344 0.56 88.4 303 192.5 0.0310 0.62 97.7 306 192.5 0. 0272 0.71 109.0 307 192.5 0. 0202 0.95 129. 5 312 192.5 0 0202 0. 95 128.0 319 192.5 0. 0202 0.95 129. 0 314 192.5 0.0133 1. 45 151. 5 308 192. 5 0. 0066 2. 92 171. 5 318

PAGE 213

201 TABLE A-7. Isotherm data for 2 4-dihydroxyacetophenone < 200 mesh carbon. Carbon mass Carbon loading X/M (mg/L) (g) (mg/mg) (mg/L) (mg/g) 99.4 0.0601 0.17 0.1 165 9 9.4 0.0499 0.20 0.3 199 9 9.4 0.0501 0 .20 0.3 198 9 9.4 0.0499 0.20 0.3 199 9 9.4 0.0400 0 .25 1. 6 245 o r\ fi c z u y 5 0.1001 0 .21 0.3 209 o n o c 2 (J y 5 0.1000 0 21 0.4 209 z u y 5 0.0999 0 21 0.4 209 209.5 0.0945 0 .22 0.5 221 2 09.5 0. 0900 0.23 0.9 232 209.5 0 0849 0.25 1. 3 245 2 0 9.5 0.0800 0.26 1. 9 260 2 u y 5 0 0749 0.28 3.1 276 z u y o n n "7 c: n U U / 3 U 0.28 2 9 275 209.5 0.0750 0.28 2.9 275 209.5 0.0700 0.30 4.8 292 209.5 0.0650 0.32 8.0 310 209.5 0.0600 0.35 13. 0 328 209.5 0. 0550 0.38 19.8 345 209.5 0.0501 0.42 28.9 360 209.5 0.0450 0.47 41. 4 374 209.5 0.0450 0.47 40.4 376 209.5 0.0450 0.47 41. 4 374 209.5 0.0399 0.53 54.4 389 209.5 0. 0351 0.60 69.2 400 209.5 0. 0301 0.70 85.4 412 209.5 0.0250 0.84 102.5 428 209.5 0. 0199 1. 05 120.0 450 209.5 0.0200 1. 05 119.5 450 209.5 0.0200 1. 05 119.5 450

PAGE 214

202 TABLE A8 Isotherm data for a-terpineol. 100 X 140 mesh carbon C o Carbon mass Carbon loading X/M (mg/L) (g) (mg/mg) (mg/L) (mg/g 194.5 0. 0900 0.22 10.7 204 194.5 0. 0801 0.24 20.6 217 194.5 0.0800 0.24 21. 6 216 194.5 0.0801 0.24 20.5 217 194. 5 0.0700 0.28 31 a ? 3 ? 194.5 0.0499 0.39 72.5 244 194.5 0.0400 0.49 93. 6 252 194.5 0.0300 0.65 118.5 253 194.5 0.0300 0.62 116.5 260 194.5 0. 0300 0.65 119. 0 252 194.5 0.0200 0. 97 139.0 278 49.5 0.0266 0.19 3.0 175 49.5 0.0253 0.20 3.6 181 49.5 0.0241 0 21 4 3 X o o 49.5 0.0232 0.21 5.0 192 49.5 0.0223 0.22 6.2 194 49.5 0. 0214 0.23 6.8 200 49.5 0.0214 0.23 6.8 200 49.5 0. 0214 0.23 6.9 199 49.5 0.0206 0.24 7.6 203 100.0 0.0509 0.20 5.5 186 100.0 0.0447 0.22 8.9 204 100.0 0. 0447 0.22 9.7 202 100.0 0.0447 0.22 11. 7 198 100.0 0 0372 0.27 ^ J. -7 100. 0 0.0311 0.32 29.7 226 100. 0 0 0258 0.39 38.6 238 100. 0 0. 0208 0.48 47.4 253 100.0 0.0122 0.82 68.4 259 100.0 0. 0122 0.82 68.0 262 100. 0 0. 0122 0.82 69.3 252 477 1. 000 0.48 193.5 284

PAGE 215

203 TABLE A-9. Isotherm data for a-terpineol, < 200 mesh carbon. o (mg/L) Carbon mass (g) Carbon loading (mg/mg) e (mg/L) X/M (mg/g) 103.0 103.0 103.0 103. 0 205.0 205.0 205. 0 205.0 205. 0 205.0 205. 0 205.0 205.0 205.0 205.0 205. 0 205.0 205.0 205. 0 205.0 205.0 205.0 205. 0 205.0 205.0 205.0 205.0 519.0 519.0 519.0 519.0 0.0503 0.0400 0.0398 0.0398 0.1000 0.0950 0.0900 0. 0853 0.0798 0.0750 0.0754 0.0749 0.0702 0650 0600 0. 0554 0.0500 0450 0450 0.0451 0. 0402 0.0350 0.0300 0.0253 0. 0203 0.0200 0.0202 0.0600 0. 0500 0.0503 0.0500 0 0, 0, 0, 0.20 0.26 0.26 0.26 0.21 0.22 0.23 0.24 0.26 0.27 0.27 0.27 0.29 0.32 0, 0, 34 37 0.41 0 .46 0.46 0.45 0.51 0.59 0.68 0. 81 1. 01 1. 03 1. 01 0. 87 1. 04 1. 03 1. 04 1.2 3.8 3.8 4.0 1.3 1.4 2.2 3.1 4 6 6 6 9 13 19.2 26.8 41. 3 51. 8 51. 6 50.9 63.3 75.0 97.0 113.0 128.0 129.5 129.5 248.0 306. 5 303. 0 299.0 202 248 249 249 204 214 225 237 252 265 264 265 278 295 310 322 327 340 341 342 352 371 360 364 379 378 374 452 425 429 440

PAGE 216

204 Table A-10. Allyl phenol solubility data (25C). Sample Concentration (g/L) Mean Standard deviation (g/L) (g/L) Al A2 A3 All A Bl B2 B3 All B CI C2 C3 Dl D2 D3 E F G 6.296 6.327 6.359 6, 6, 6, 6, 6, 6. 6 6. 6. 6, 6, 6, 6, 6. 6. 327 265 265 374 390 327 390 390 452 405 343 343 390 359 327 6.237 6.237 6.265 6.327 6.327 6.327 6.327 6.327 6.390 6.327 6.286 6.364 6.326 6.411 6.364 6.359 6.378 6.246 6.327 0.032 0.036 0.033 0.045 0.036 0.036 0.032 0.039 All Samples 6.337 6.337 0. 053

PAGE 217

205 Table A-11. Solubility data for o-cresol, 2 4-dihydroxyacetophenone and a-terpineol (25C). Sample Concentration Mean (g/L) (g/L) Standard deviation (g/L) o-cresol A B C D E 2, 4 -DAP A B C D E a-terpineol A-1 A-2 A-3 All A B C D E 21. 4 22.0 21. 1 21. 7 21. 8 2.303 2. 315 2.303 2.284 2.284 2.366 2.352 2,345 2, 2. 2, 2, 2, 2. 381 352 373 423 395 409 2.423 2.416 2.451 2.423 21. 6 0.35 2.298 0.014 2.354 2.369 2.409 2.377 0.011 0.015 0. 014 0. 027 All Samples 2.393 0.034

PAGE 218

APPENDIX B SEQUENTIAL AND SIMULTANEOUS SOLUTE ADDITION DATA FOR THE O-CRESOL/ALLYL PHENOL SOLUTE PAIR

PAGE 219

TABLE B-1. Sequential solute addition data, o-cresol followed by allyl phenol. Feed concentration: 10.0 mg/L (o-cresol) Time (hr) Effluent composite samnle vol ump (L) ^\JLL\^^ 11 L. J. a. \^ j.\^Ll \J 1. 7 0 D y 12.5 4.95 8.3 8.5 21. 5 8.17 9.2 9.2 29.6 7.53 9.6 9.7 35.0 5.07 9.7 9.7 43.75 8.34 9.8 9.9 51. 5 7.47 9.9 10. 0 207

PAGE 220

208 TABLE B-2. Sequential solute addition data, o-cresol followed by allyl phenol. Feed concentrations: 10.8 mg/L (o-cresol), 10.2 mg/L (allyl phenol) Concentration (mq/L) Time (hr ) Sample volume (L) o-cresol allyl phenol U 0 6 0.050 10.8 1. 3 0 11 0.050 11. 1 2.5 0 16 0 050 11. 4 2.9 0.22 0 050 11. 6 3.2 0.27 0 050 11. 9 3.5 0 32 0 050 11. 7 3.6 0.38 0.050 11. 9 3.8 0.43 0.050 12.2 4.1 0.48 0 050 12.1 4.2 0.53 0 050 12.0 4.2 0.59 0 050 12.3 4.5 0.64 0 050 12.3 4.6 0.69 0 050 12.2 4.7 A "in 0.74 0 050 12.4 4.9 0.85 0 100 12.3 5.0 0.95 0.100 12.1 5.1 T A ^ 1. 06 0 100 12.2 5.4 1. 16 0.100 12.1 5.5 1.27 0 100 12.2 5.8 1. 37 0.100 12.1 5.9 1.47 0.100 12.2 6.1 1.58 0 100 12.1 6.1 1.68 0.100 12.3 6.4 1.79 0. 100 12. 0 6.4 1. 89 0.100 12.2 6.6 1 Q Q X -7 17 n inn u J. u u 12 0 6.7 2.10 0.100 12.0 6.6 2.36 0.250 12.2 7.2 2.62 0.250 12.0 7.3 2.88 0.250 11. 9 7.3 3.12 0.250 11. 9 7.7 4.15 1. 000 11. 7 8.0 5.42 1. 000 11. 5 8.5 5.42 0.050 11. 4 8.6 13.85 8. 05 11. 2 9.4 13. 85 0.050 11. 0 9.7 24.32 9.90 10.9 9.9 24.32 0 050 10.8 10.2 34.69 9.73 10. 9 10.2 34.69 0. 050 10. 9 10.2

PAGE 221

209 TABLE B-3. Sequential solute addition data, o-cresol followed by allyl phenol. Feed concentrations: 10.0 mg/l (o-cresol), 8 9.0 itig/L (allyl phenol) Concentration (mg/L) Time Sample volume (hr) (L) o-cresol allyl phenol 0.07 0. 050 11. 2 32. 0 0.13 0.050 12.2 55.5 0.18 0.050 12.2 66.0 0.24 0.050 12. 0 71. 0 0.29 0.050 11. 7 74.5 0.35 0. 050 11. 6 78.0 0.40 0. 050 11. 3 79.5 0.45 0. 050 11. 2 84.0 0.51 0.050 11. 0 86.5 0.56 0.050 10.9 86.0 0.62 0.050 10. 9 87.5 0.67 0. 050 10.7 87.0 0.73 0. 050 10.6 87.0 0.78 0.050 10.5 85.5 0.83 0.050 10.4 87.0 0.89 0.050 10.4 88.0 0.94 0.050 10.4 90.5 0.99 0.050 10.4 89.0 1. 10 0.100 10.4 89.5 1. 21 0.100 10.2 87.5 1. 32 0.100 10.1 89.0 1. 42 0.100 10.2 89.5 1. 53 0.100 10.1 89.0 1. 64 0.100 10.1 90.0 1. 74 0.100 10.1 90.0 1. 85 0.100 10.1 89.0 1. 95 0.100 10. 0 89.0

PAGE 222

210 TABLE B-4. Sequential solute addition data, o-cresol followed by allyl phenol. Feed concentration: 102.0 mg/L (o-cresol) Time Sample volume Concentration, o-cresol (hr) (L) (mg/L) 0.11 0.100 0 0.22 0.100 0.5 0.33 0.100 2.9 0.44 0.100 8.6 0.55 0.100 18.2 0.66 0.100 30.4 0.77 0.100 47.7 0.87 0.100 64.3 0.98 0.100 76.5 1. 09 0.100 85.0 1. 19 0.100 90.5 1. 30 0.100 94.5 1. 41 0.100 96.0 1.51 0.100 98.0 1. 79 0.250 99.0 2. 05 0.250 100.5 2.31 0.250 101. 0 2.58 0.250 100. 0 2.85 0.250 100. 0 3.10 0.250 100.0 3. 64 0.500 101. 0 4.70 1. 000 101. 0 6.82 1. 950 100.0 13. 40 5.880 102.0 13.40 0.050 102. 0

PAGE 223

211 TABLE B-5. Sequential solute addition data, o-cresol followed by allyl phenol. Feed concentrations: 102.5 mg/L (o-cresol), 9.5 mg/L (allyl phenol) Concentration (mg/L) OdlU^Xt:^ VUXLlXUc: (h-r) ( T ) V -Li / — /"^ T"^cr^l KJ O X c o cixxyx pilcIlUX n nfi n n n inn n X u u • u 1 n X • u n n R n inn n X u u • u 1 Q n ^ 0 ^ inn n X u u • u 9 A n 0 0 Q Q n 9 7 n 9 Q n n R 0 inn n X u u • u 1 O • X 0 35 n nsn inn s ^ • J 0 41 U • L/ ^ L/ 1 nn n 7 0 46 n nsn 1 nn n 4 n D n R 0 u • u ^ u 1 n 1 n X U X • u 4 9 u • u ^ u inn n X VJ u • u 4 4 0 63 0 050 4 S 0.69 0 050 inn s 4 7 0.74 0.050 100 5 s n 0 80 n nsn 1 n 1 n X U X • u S 9 f) ft S \J m U ^ \J inn n X u u u D o 0 9] u • u J y inn n X u u u S 4 D.I 0 96 • tJ U inn n X u u u S fi n nsn 1 nn n X u u u S ft X • X n inn inn n X u u u n 1 9 4 n inn U X u u inn n X u u u 1 0 X X J D n inn u J. u u inn K X U U • D D 4 i D n inn U J. u u inn n X U U U D b n inn U X u u 1 n 1 n X u X U C "7 1 ft X o o n inn inn n X U U U 7 n 1 7 Q X / y n inn inn n X U U U 7 1 7 i 1 ft Q X o n inn U X u u Q Q K / z 2.00 0.100 100. 0 7.4 2.11 0.100 99.5 7.4 2.22 0.100 99.5 7.5 2.33 0.100 100.5 7.6 2.44 0.100 101. 0 7.7 2.55 0.100 101. 0 7.9 2.65 0.100 100.0 8.0 2.76 0.100 100.0 8.1 2 .87 0.100 100. 0 7.9 2.98 0.100 100.0 8.0 3.24 0.250 100. 5 8.3 3.51 0.250 100.5 8.4 3.78 0.250 101. 0 8.4 4.05 0.250 100.0 8.4 5.13 1. 000 101. 0 8.9

PAGE 224

212 TABLE B-5. (Continued) Time Sample volume (hr) (L) Concentration (mg/L) o-cresol allyl phenol 6.24 7.58 8.66 13.13 13.13 1. 025 1. 050 1. 000 4.360 0. 050 101. 0 102.5 102.5 102.0 102. 5 9.1 9.3 9.4 9.3 9.6

PAGE 225

213 TABLE B-6. Sequential solute addition data, o-cresol followed by allyl phenol. Feed concentration: 101.0 mg/L (o-cresol), 96.0 mg/L (allyl phenol) Concentration (mg/L) Time Sample volume (hr) (L) o-cresol allyl phenol n n t^ n u U D u 1 n Q n X u y U 4 Z J n in U J. u U U D J 1 X U U D O U n 1 n n c; fi u U D u X U o U 71. 0 n 01 u Z X n n n u U 3 u 1 u y u "7 T A 7 7 U u z o n n n u U D u X U o U OA C n s> u ^ ^ U U J u 10 7 n X u / u O O A n n ^ n u U D u X u / u OA A O 4 U n n R K U U D D 1 n "7 n XU / U o c c o 5 5 U I 0 n n R n u U D u X U / U o7 5 n R 4 U D 4 n n n u U D u X U D U 8 7.5 u U D u X U D U 8 8.5 u U D u X U D U 8 y 5 n 7 n n n R 9 u U D z XU4 U 87.5 n 7 n n n U U J u X U 4 U 8 8.5 0.81 0. 050 104.0 90.0 0.86 0 .050 104. 0 90.0 0.92 0.050 105.0 90.5 1. 02 0.100 104.0 90.5 1. 13 0.100 103.0 91. 0 1. 24 0.100 103. 0 91. 0 1. 35 0.100 101. 0 90.5 1. 47 0.115 102.0 92. 0 1. 58 0.100 101. 0 90.5 1. 69 0.100 102.0 93.0 1.79 0.100 102.0 93.0 1. 90 0.100 101. 0 93.0 2.01 0.100 101. 0 93.0 2.55 0.500 102. 0 94.0 2.55 0.015 101. 0 93.5 3. 08 0.500 101. 0 93.5 3.08 0. 015 101. 0 94.5 3.63 0.500 102. 0 94.5 3.63 0.015 102.0 95.0

PAGE 226

214 TABLE B-7. Sequential solute addition data, allyl phenol followed by o-cresol. Feed concentration: 10.0 mg/L (allyl phenol) Time Effluent composite Concentration sample volume of allyl phenol (hr) (L) (mg/L) n n Q 0.100 0.1 u Z 1 0.100 0.1 n TO 0.100 0.2 n A A U 4 ft 0.100 0.2 U 3 3 0.100 0.3 U D / U 10 0 0.2 U / 0 0.100 0.3 n Q n u y u 0.100 0.3 1 n 1 U 1 0.100 0.3 1 19 X z 0.100 0.3 X Z J 0.100 0.4 X J D 0.100 0.4 1. 46 n inn 0 5 1. 58 0.100 0.5 1. 86 0.250 0.6 2.15 0.250 0.7 2.44 0.250 0.9 2.73 0 250 1. 0 3. 01 0.250 1. 3 3.29 0.250 1. 5 3.58 0.250 1. 7 3.86 0.250 2.0 4.58 0.550 2.4 5. 96 0. 180 3.5 10.50 3.070 5.0 10.50 0.050 5.9 20.47 8.890 7.8 20.47 0.050 9.2 28.48 7.200 9.3 28.48 0. 050 10. 0

PAGE 227

215 TABLE B-8. Sequential solute addition data, allyl phenol followed by o-cresol. Feed concentrations: 9.8 mg/L (allyl phenol), 10.0 mg/L (o-cresol) Concentration (mg/L) Time Sample volume (^^) (L) allyl phenol o-cresol 0 04 0.050 -? o 1 1 O X 0. 10 0.050 in n 4.2 0 15 0.050 in A X U 4 D U 0 21 0.050 in fi X u o 0 b 0 .26 0.050 in R X u o o 1 0 32 0.050 in ft X u o D 4 0.37 0.050 in Q c o D O 0.43 0.050 11 n X X u 7.0 0.48 0.050 in Q X u y 7 3 0.53 0.050 11 n X X u 7 5 0.59 0.050 in Q X u ^ 7 5 0 64 0.050 11 1 X X X 7 8 0.70 \J \J J \J 7 9 0.75 n n R n in Q ID ^ 8 1 0.86 n inn 11 1 1. 1 8 3 0.97 0 100 11 0 ft K 1. 08 0.100 11. 2 8.7 1.19 0.100 11. 0 8.8 1. 30 0.100 11. 1 8.9 1. 41 0.100 10. 9 9.1 1. 52 0.100 11. 0 9.1 1. 63 0.100 10.9 9.2 1. 92 0.258 10.9 9.3 2.19 0.250 10.7 9.5 2.46 0.250 10.5 9.6 2.77 0.258 10.7 9.6 3.04 0.250 10.6 9.7 4.11 1. 000 10.5 9.8 4.11 0.010 10.2 9.9 5.23 1. 000 10.0 9.9 5.23 0.010 10.0 9.9 6.33 1. 000 9.9 9.9 6.33 0. 015 9.9 10.0 15.23 8. 020 9.9 10.0 15.23 0.010 9.8 10.0

PAGE 228

216 TABLE B-9. Sequential solute addition data, allyl phenol followed by o-cresol. Feed concentrations: 10.3 mg/L (allyl phenol), 97.0 mg/L (o-cresol) Concentration (mq/L) Time Sample volume (hr) (L) allyl phenol o-cresol n n K u U 0 u U D u 11 A 11. 4 o D i u n n n u u _) U 4 J 56.7 n 1 Q U X o n n c: n U U D U lb 4 71. 0 n "5 ? u Z J n n c: n U U 3 U 1 / J 7 9.5 u u o u 0 4 U n n n u U 3 u 17 ^ O 7 C n /in n n R n u U D u 17 / y u 0 U D n n n U U 0 U 1 "7 O 1 / J 91. 0 n n c: n U U D U 1 C Q lb y 92.0 U D / n n c; r> U U D U lb 7 93.0 U D J n n R n u U D u 1 b • 4 y i 5 n <^ Q o A n U U D u ib U 94.5 n fl n u U 3 u It; 0 Xo o y 4 D 0.80 0.050 15.7 95.5 0.91 0.100 15.4 95.5 1. 02 0.100 14. 9 95.5 1. 13 0.100 14.3 96.0 1. 24 0.100 14.1 95.5 1. 35 0. 100 13.7 96.0 1. 46 0.100 13.4 96.0 1. 57 0.100 13.2 96.0 1. 68 0.100 13.1 96.5 1.80 0.100 12.8 97.0 1. 91 0.100 12.4 96.0 2.20 0.258 12.0 96.0 2.48 0.254 11. 8 97.0 2.76 0.250 11. 4 96.5 3.03 0.250 11. 1 96.5 4.30 1. 000 10. 9 97.0 5.38 1. 000 10.4 97.0 11. 40 5.450 10.1 97.0 11. 40 0.015 10.2 97.0

PAGE 229

217 TABLE B-10. Sequential solute addition data, allyl phenol followed by o-cresol. Feed concentration: 9 9.5 mg/L (allyl phenol) Time Effluent composite Concentration sample volume of allyl phenol (hr) (L) (mg/L) 0.05 0.051 0.2 0.11 0. 050 1. 1 0.17 0.053 1. 2 0.22 0.050 1.6 0.28 0. 050 1.6 0.33 0.050 2.0 0.39 0. 050 2.5 0.44 0. 050 3.1 0.50 0. 050 4.1 0.55 0.050 5.5 0.61 0. 050 7.3 0.66 0.050 9.7 0.72 0.050 12.7 0.77 0.050 16.6 0.88 0.100 23.3 0.99 0.100 34.1 1. 10 0.100 44.6 1. 22 0.106 54.5 1. 33 0.100 63.7 1.44 0.100 71. 4 1. 55 0.100 78.0 1. 66 0.100 83.0 1. 77 0.100 87.0 1.88 0.100 90.5 1. 99 0.100 92.5 2.26 0.250 95.0 2.54 0.250 97.0 2.81 0.250 97.5 3.10 0.257 97.5 3.67 0.518 98.5 4.22 0.500 97.5 5.32 1. 000 98.0 5.32 0.050 98.5 6.55 1. 030 98.0 6.55 0.050 98.5 7.60 1. 000 99.5 7.60 0. 050 99.5

PAGE 230

218 TABLE B-11. Sequential solute addition data, allyl phenol followed by o-cresol. Feed concentrations: 97.5 mg/L (allyl phenol), 9.9 mg/L (o-cresol) Concentration (mg/L) Time Sample volxime (hr) (L) allyl phenol o-cresol 0.05 0.050 94.2 3.0 0.10 0.050 95.5 5.5 0.16 0. 050 97.5 6.8 0.21 0. 050 96.0 7.6 0.27 0.051 96.5 8.2 0.32 0. 050 97.5 8.5 0.37 0. 050 96.0 8.7 0.43 0.050 91. 0 8.9 0.48 0.050 96.0 9.2 0.59 0.100 97.5 9.3 0.70 0.100 97.5 9.5 0.80 0.100 97.0 9.6 0.91 0.100 97.0 9.7 1. 02 0.100 97.0 9.8 1. 30 0.252 98.0 9.8 1. 56 0.250 98.0 9.8

PAGE 231

219 TABLE B-12. Sequential solute addition data, allyl phenol followed by o-cresol. Feed concentrations: 95.0 mg/L (allyl phenol), 100.0 mg/L (o-cresol) Concentration (mq/L) Time Sample volume (hr) (L) allyl phenol o-cresol 0.05 0.050 105.0 45.6 0.10 0.050 113.0 72.5 0.16 0.050 112.0 82.5 0.21 0.051 108.0 86.5 0.26 0.050 105.0 89.5 0.32 0.050 103.0 91. 5 0.37 0.050 102. 0 94.0 0.42 0.050 102.0 95.0 0.47 0.050 99.0 95.0 0.58 0.100 98.0 96.5 0.69 0.100 97.0 97.0 0.79 0.100 96.0 97.5 0.90 0. 100 96.0 98.0 1. 00 0.100 95.0 98.0 1. 27 0.250 95.0 98.0 1.54 0.250 95.0 98.0 1. 81 0.250 95.0 99.0 2.07 0.250 95.0 100.0

PAGE 232

220 TABLE B-13. Simultaneous solute addition data, allyl phenol and o-cresol. Feed concentrations: 9.5 mg/L (allyl phenol), 9.9 mg/L (o-cresol) Time Concentration (mg/L) Sample volume (hr ) (L) allyl phenol o-cresol 0.43 0.406 0 48 0 050 0.1 0 54 0 050 0.1 0.59 0 .050 0.1 0 64 0 050 0.1 0.70 0 050 0.2 0 .75 0 050 0.2 0.80 0.050 0.2 0 91 0 100 0.3 1. 02 0.103 0.5 1. 12 0. 100 0.1 0.6 1.23 0.100 0.1 0.8 1.34 0.100 0.1 1. 0 1.44 0 100 0.1 1. 3 1. 55 0.100 0.2 1. 6 1.66 1. 100 0.3 1. 9 1.76 0 100 0.3 2.3 1.87 0.100 0.4 2 8 2.15 0.257 0.5 3.7 2.41 U / 5 1 2 .68 0.250 1. 0 6.5 2. 94 0.250 1. 3 8.1 3.01 1. 000 2.1 9.8 4.07 1. 005 3.7 10.8 4.07 0.015 4.5 11. 0 5.12 1. 000 5.4 11. 0 5.12 0.015 6.0 10.9 10.65 5.240 7.8 10.4 10.65 0.015 8.9 10.1 24.33 12. 820 9.3 9.9 24.33 0.015 9.5 9.9 *Below detection limit.

PAGE 233

221 TABLE B-14. Simultaneous solute addition data, allyl phenol and o-cresol. Feed concentrations: 10.0 mg/L (allyl phenol), 103 mg/L (o-cresol) Concentration (mg/L) Time Sample volume (hr) (L) allyl phenol o-cresol 0.05 0. 050 0.4 0.11 0.050 0.4 0.16 0. 050 0.6 0.22 0.050 1.1 0.27 0.050 1. 9 0.33 0.050 0.1 3.4 0.38 0.050 0.2 6.1 0.43 0.050 0.2 10.4 0.49 0.050 0.3 16.6 0.61 0.108 0.5 32.6 0.72 0.101 0.8 57.0 0.83 0.100 1. 0 74.5 0.94 0.100 1. 2 86.5 1. 04 0.100 1. 5 94.0 1. 31 0.250 2.0 99.0 1. 58 0.250 2.6 103.0 1. 85 0.250 3.3 105.0 1. 92 1. 000 5.0 106.0 4.02 1. 000 7.2 104.0 4.02 0. 025 7.9 103.0 5.15 1. 025 8.5 105.0 5.15 0.025 8.9 103. 0 6.27 1. 000 9.2 102.0 6.27 0.025 9.4 103.0 7.36 1. 000 9.5 102. 0 7.36 0.025 9.6 102. 0 8.14 0.025 9.7 Below detection limit.

PAGE 234

222 TABLE B-15. Simultaneous solute addition data, allyl phenol and o-cresol. Feed concentrations: 97.0 mg/L (allyl phenol), 9.8 mg/L (o-cresol) Concentration (mg/L) Time Sample volume (hr) (L) allyl phenol o-cresol 0.04 0.050 0.2 0.1 0.09 0.050 0.1 0.1 0.15 0.050 0.2 0.1 0.20 0.050 0.6 0.1 0.25 0.050 0.6 0.1 0.30 0.050 1.0 0.3 0.35 0.050 1. 5 0.4 0.41 0.050 2.4 0.8 0.46 0.050 3.6 1. 2 0.56 0.100 6.5 2.2 0.67 0.100 12. 9 4.3 0.77 0.100 21. 2 6.9 0.88 0.100 33.4 9.4 0.98 0.100 45.3 11. 3 1. 09 0. 100 56.2 12.2 1. 34 0.250 70.5 12.8 1. 63 0.277 86.5 12.1 1. 89 0.250 90.0 11.1 2.18 0.250 92.5 10.7 3.25 1. 005 100.0 10.0 3.25 0. 025 96.5 9.9 4.37 1. 045 96.5 9.8 4.37 0. 025 96.5 9.8

PAGE 235

223 TABLE B-16. Simultaneous solute addition data, allyl phenol and o-cresol. Feed concentrations: 9 8.5 mg/L (allyl phenol), 96.5 mg/L (o-cresol) Concentration (mq/L) Time Sample volume (hr) (L) allyl phenol o-cresol 0 10 0 100 0 15 0.050 0.1 0.6 0 21 0 050 0.4 2.0 0.26 0.050 1. 1 5.8 0.32 0. 050 2.3 13.6 0.37 0.050 4.5 24.5 0.42 0.050 7.9 27.9 0.48 0.050 12.2 48.3 0.58 0.100 19.1 68.1 0.69 0.100 28.1 95.5 0.80 0.100 41.2 118.0 0. 91 0. 106 52.5 122.5 1. 02 0.100 61. 5 117.5 1. 28 0.250 73.5 111. 0 1. 54 0.250 84.5 104.5 1. 81 0.250 89.0 101. 0 2.90 1. 000 93.0 98.5 2.90 0.025 95.0 98.5 4 00 1. 000 95.0 98.0 4.00 0.025 95.0 98.0 5.12 1. 032 96.0 98.5 5.12 0. 025 97.0 99.0

PAGE 236

APPENDIX C SEQUENTIAL AND SIMULTANEOUS SOLUTE ADDITION DATA FOR 2,4-DIHYDROXYACETOPHENONE (DAP)/ALLYL PHENOL SOLUTE PAIR

PAGE 237

TABLE C-1. Sequential solute addition data, 2,4dihydroxyacetophenone (DAP) followed by allyl phenol. Feed concentration: 9.6 mg/L (DAP) Time (hr) Effluent composite sample volume (L) Concentration of DAP (mq/L) Composite sample Grab sample* 1. 00 0.1 3.18 2. 000 0.2 0.5 4.28 1. 000 0.8 1.1 5.37 1. 000 1. 5 2.0 6.46 1. 000 2.6 3.2 7.56 1. 000 3.8 4.5 8.66 1. 000 5.2 5.8 9.75 1. 000 6.4 6.9 10.86 1. 010 7.4 7.8 11. 96 1. 000 8.2 8.5 13.05 1. 000 8.6 8.9 14.15 1. 000 8.9 9.1 15.25 1. 000 9.2 9.3 24. 08 8.400 9.4 9.6 *Grab sample volume was 0.050L. 225

PAGE 238

226 TABLE C-2. Sequential solute addition data, 2,4dihydroxyacetophenone (DAP) followed by allyl phenol. Feed concentrations: 9.5 itig/L (DAP), 9.5 mg/L (allyl phenol) Time Sample volume Concentration of DAP (mg/L) (hr) (L) DAP allyl phenol 0.06 0.050 9.9 0.5 0.17 0. 105 10.4 1. 1 0.28 0.100 11. 1 1. 9 0.40 0.107 11. 5 2.4 0.53 0.112 12.1 3.0 0.63 0. 100 12.3 3.5 0.91 0.250 13.0 4.2 1.22 0.288 13.0 5.0 1.49 0.250 13. 0 6.2 2.66 1. 028 12.6 6.6 2.66 0.050 12.2 7.2 3.80 1. 000 11. 9 7.6 3.80 0.050 11. 4 7.9 4.93 1. 000 11. 4 8.3 4.93 0.050 11. 0 8.3 6.09 1. 000 10.9 8.4 6.09 0, 050 10.7 8.7 7.27 1. 015 10.6 8.7 7.27 0.050 10.3 8.7 8.44 1. 000 10.5 8.9 8.44 0. 050 10.3 8.9 9.56 1. 000 10.3 9.0 9.56 0. 050 10.3 9.2 10.69 1. 000 10.0 9.4 10.69 0. 050 10.2 0.2 11. 83 1. 000 10. 0 9.5 11. 83 0.050 10. 0 9.5

PAGE 239

227 TABLE C-3. Sequential solute addition data, 2,4dihydroxyacetophenone (DAP) followed by allyl phenol. Feed concentrations: 10.0 mg/L (DAP), 99.5 mg/1 (allyl phenol) Time Sample volume Concentration of DAP (mg/L) (hr ) (L) DAP allyl phenol 0.05 0.050 12.4 38.6 0.10 0.050 14.1 51. 0 0.15 0.050 15.1 57.0 0.20 0.050 15.8 62.5 0.26 0.050 16.1 66.5 0.31 0.050 16.6 72.5 0.36 0.050 16.7 70.5 0.41 0.050 17.0 73.0 0.47 0.050 17.1 74.0 0.52 0.050 17.3 76.5 0.63 0.100 17.1 78.5 0.73 0.100 17.0 80.5 0.84 0.100 16.7 81. 5 0.94 0.100 17. 0 83.0 1. 05 0.100 16.3 84.5 1. 15 0.100 16.2 86.0 1.26 0.100 16.1 86.5 1. 36 0.100 15.6 87.5 1. 47 0.100 15.7 89.0 1. 73 0.250 15.2 91. 0 2.00 0.258 14.7 91. 5 2.28 0.256 14.2 92.5 2.54 0.250 13. 9 93.0 3.66 1. 000 12. 9 95.0 3.66 0.050 12.4 96.5 4.80 1. 020 12.0 96.5 4.80 0.050 11. 7 97.0 5. 92 1. 000 11. 4 97.5 5. 92 0.050 10.8 97.5 7.11 1. 060 10.7 98.5 7.11 0.050 10.6 99.5 8.25 1. 025 10.4 99.5 9.37 1,000 10.2 99.5 10.50 1. 032 10.2 99.5

PAGE 240

228 TABLE C-4. Sequential solute addition data, 2,4dihydroxyacetophenone (DAP) followed by allyl phenol. Feed concentration: 9 6.5 mg/L (DAP) Time Sample volume Concentration of DAP (hr) (L) (mg/L) 0.05 0 050 0.6 0 10 0 050 0.3 0 16 0. 050 0.3 0.27 0.100 0.5 0 .38 0.100 1. 0 0.48 0.100 2.2 0.59 0.100 4.5 0.69 0.100 8.6 0.80 0.100 14. 9 0 93 0.124 24.2 1. 04 0.100 35.3 1. 15 0.100 44.7 1. 25 0.100 53.2 1. 36 0.100 60.5 1. 47 0.100 66.6 1. 57 0.100 71. 9 1. 68 0.100 76.6 1. 80 0.113 80.5 1. 90 0.100 84.0 2.02 0.104 86.5 2.28 0.250 89.5 2.55 0.250 2.83 0.262 92.0 3.10 0.250 93.5 4.23 0.251 94.5 4.23 0.050 95.5 5.35 1. 000 95.5 5.35 0.050 96.0 6.45 1. 000 96.5 6.45 0.050 96.5 7.58 1. 000 96.5 7.58 0.050 96.5

PAGE 241

229 TABLE C-5. Sequential solute addition data, 2,4dihydroxyacetophenone (DAP) followed by allyl phenol. Feed concentrations: 9 7.0 mg/L (DAP), 9.9 mg/L (allyl phenol) Time Sample volume Concentration of DAP (mg/L) (hr) (L) DAP allyl phenol 0.05 0.050 101. 0 2.7 0.10 0.050 103.0 4.1 0.15 0. 050 101. 0 4.9 0.21 0.050 102.0 5.5 0.26 0.050 101. 0 6.0 0.31 0.050 101. 0 6.4 0.37 0.050 102.0 6.7 0.42 0.050 101. 0 7.0 0.48 0.050 101. 0 7.2 0.58 0. 100 99.0 7.4 0.69 0.100 98.5 7.8 0. 80 1. 102 98.5 8.0 0. 92 0.109 98.0 8.2 1. 04 0.113 99.0 8.4 1. 30 0.250 98.5 8.6 1. 61 0.287 98.0 8.9 2.73 1.000 98.0 9.2 2.73 0.050 98.0 9.6 3.87 1. 025 98.0 9.6 3.87 0.050 97.5 9.8 6. 08 2.017 97.5 10.0 6.08 0. 050 10.0 8.30 2.020 9.8 8.30 0.050 9.8

PAGE 242

230 TABLE C-6. Sequential solute addition data, 2,4dihydroxyacetophenone (DAP) followed by allyl phenol. Feed concentration: 92.5 mg/L (DAP), 94.5 mg/L (allyl phenol) Time Sample volume Concentration of DAP (mg/L) (hr) (L) DAP allyl phenol 0 05 0.050 106.0 41. 7 0 10 0.050 114. 0 56.1 0.16 0.050 112. 0 63.6 0.21 0.050 112.0 69.0 0.26 0.050 106.0 70.0 0.32 0.050 108.0 75.0 0.37 0.050 107.0 77.0 0.42 0.050 107.0 79.5 0.47 0.050 108.0 81. 5 0.58 0.100 105. 0 82.0 0.69 0.100 103.0 84.5 0.79 0.100 101. 0 85.0 0.90 0.100 103. 0 87.5 1. 00 0.100 102.0 88.5 1. 11 0.100 100.0 88.5 2.22 1. 000 97.5 92. 0 2.22 0.050 93.5 91. 0 3.34 1. 000 94.5 92.5 3.34 0.050 92.5 92.0 5.52 2. 000 93. 0 93.0 5.52 0.050 93.5 94.5

PAGE 243

231 TABLE C-7. Sequential solute addition data, allyl phenol followed by 2 4-dihydroxyacetophenone Feed concentration: 10.0 mg/L (allyl phenol) Time (hr) F!"F"Flnfin1~ c nm o q "i i~ J-idJ\^ Will (3 J. u c sample volume (L) v^vjxiueiit.ira. uxon or axxyj. pnenoi img/L ) Composite sample Grab samole* 2.12 2.000 0.1 0.2 3.21 1. 000 0.3 0.4 4.30 1. 000 0.6 0.9 5.40 1. 000 1. 8 2.5 6.49 1. 000 3.4 4.4 8.65 2.000 6.0 7.1 10.81 2.000 7.8 8.4 15.48 4.330 9.0 9.4 24.45 8.510 9.6 9.7 *Grab sample volume was 0.050L.

PAGE 244

232 TABLE C-8. Sequential solute addition data, allyl phenol followed by 2 4-dihydroxyacetophenone (DAP). Feed concentrations: 10.1 mg/L (allyl phenol), 9.4 mg/L (DAP) Time Sample volume Concentration (mg/L) (hr) (L) allyl phenol DAP 0. 05 0.050 10.1 0.1 0.15 0.100 10.5 0 1 0.26 0.100 11. 0 0 1 0.36 0.100 11. 6 0 1 0.47 0.100 12.2 0 2 0.57 0.102 12. 8 0.2 0.67 0.100 13. 0 0.3 0. 93 0.250 13.8 0.5 1. 19 0.250 15.0 0.7 1. 45 0.250 15.2 1. 4 1. 71 0.250 15.5 1. 6 2.85 1. 057 15.2 2.2 2.85 0.050 14.6 3.2 3. 94 1. 000 14.0 4.1 3.94 0.050 13.4 4.9 5.01 1. 000 12.8 5.6 5.01 0.050 12.4 6.3 10.07 4.840 11. 2 7.9 10.07 0.050 10.7 8.8 14. 65 4.375 10.5 9.2 14.65 0. 050 10.4 9.4

PAGE 245

233 TABLE C-9. Sequential solute addition data, allyl phenol followed by 2 4-dihydroxyacetophenone (DAP). Feed concentrations: 9.9 mg/L (allyl phenol), 93.0 mg/L (DAP) Time Sample volume Concentration (mg/L) (hr) (L) allyl phenol DAP 0 04 0.050 •3 c c J b • D 0 10 0.050 1 A -L D *i 0 Q 0 J 0 • 0 0.15 0.050 17 4 R n 1 D \J m X 0.20 0.050 17.4 59.3 0.25 0.050 17.3 66.5 0.31 0.050 16.6 71. 5 0.36 0. 050 16.3 77.0 0.46 0.100 15.7 80.0 0.57 0.100 14. 9 83.5 0.67 0.100 14.2 85.5 0.78 0.100 13.7 87.0 0.88 0.100 13.0 88.5 0.99 0.100 12.7 89.0 1. 55 0.500 11. 8 91. 0 1.55 0.050 11. 3 92. 0 2.13 0.505 11. 2 92.5 2.13 0. 050 10.8 92.5 3.24 1. 010 10.5 92.5 3.24 0.050 10.3 92.5 4.34 1. 000 93.0 4.34 0 050 93.0

PAGE 246

234 TABLE C-10. Sequential solute addition data, allyl phenol followed by 2 4-dihydroxyacetophenone (DAP). Feed concentration: 9 8.0 mg/L (allyl phenol) Time Effluent composite Concentration of sample volume allyl phenol (hr) (L) (mg/L) U Z 7 0.250 7 4 U 3 0 0.100 17 8 A An u 4 y 0.100 24.0 0.100 29.3 U / U 0.100 33.7 n on 0.100 37.2 u y 1 0.100 44.6 X U X n inn 4 9.2 1. 12 0 100 53.5 1. 22 0.100 57.4 1. 33 0.100 62.0 1. 43 0.100 66.0 1. 54 0.100 70.0 1. 64 0.100 73.5 1.75 0.100 76.5 1. 86 0.100 79.0 1. 96 0.100 81. 0 2.54 0.500 86.5 2.54 0.050 90.5 3.65 1. 000 94.5 3.65 0. 050 97.0 4.76 1. 000 96.5 4.76 0 050 97.0 5.88 1.000 98.0 5.88 0.050 98.0

PAGE 247

235 TABLE C-11. Sequential solute addition data, allyl phenol followed by 2 4-dihydroxyacetophenone (DAP). Feed concentrations: 97.0 mg/L (allyl phenol), 9.6 mg/L (DAP) Time Effluent composite Concentration (mg/L) sample volume (hr) (L) allyl phenol DAP 0 .05 0.050 96.0 0.5 0.10 0.050 97.5 0.8 0.15 0.050 98.5 1. 1 0.20 0.050 99.5 1. 2 0.25 0.050 99.5 1. 4 0.30 0.050 101. 0 1. 6 0 .41 0. 050 100.5 1. 8 0 52 0 100 100. 5 2.0 0.62 0.100 101. 5 2.2 0.73 0.100 101. 5 2.5 0.84 0.100 101. 5 2.7 0.95 0.100 102.0 3.0 1. 21 0.250 101. 5 3.3 1. 48 0.250 100.0 3.9 1. 75 0.250 100.0 4.5 2.01 0.250 100.0 5.0 2.60 0.500 5.8 2.60 0.050 99.0 6.2 3.23 0.540 6.7 3.23 0.050 99.0 7.1 4.35 1. 000 7.8 4.35 0.050 100.0 8.1 6. 55 2.015 8.7 6.55 0. 050 98.0 9.1 8.75 2.010 9.3 8.75 0.050 98.0 9.4

PAGE 248

236 TABLE C-12. Sequential solute addition data, allyl phenol followed by 2 4-dihydroxyacetophenone (DAP). Feed concentrations: 9 9.0 mg/L (allyl phenol), 101. 0 mg/L (DAP) Time Effluent composite Concentration (mg/L) sample volume (hr) (L) allyl phenol DAP 0.05 0.050 104. 0 28.7 0.11 0.050 111. 5 40.7 0.17 0. 050 115.5 50.0 0.22 0.050 115.5 57.2 0.27 0.050 113.0 62.5 0.33 0.050 112.0 65.5 0.38 0.050 111. 5 71. 0 0.44 0.050 111. 0 73.5 0.49 0.050 107.5 76.5 0.60 0.100 106.0 80.0 0.71 0.100 104.5 84.5 0.82 0.100 103.0 88.0 0.92 0.100 103.0 89.0 1. 03 0.100 102.5 90.0 2.16 1. 000 103.5 94.5 2.16 0.050 99.0 99.0 3.28 1. 000 100.5 98.0 3.28 0.050 99.5 99.5 5.51 2.040 100.0 99.5 5.51 0.050 99.0 100.5

PAGE 249

237 TABLE C-13. Simultaneous solute addition data, allyl phenol and 2 4-dihydroxyacetophenone (DAP). Feed concentrations: 9.8 mg/L (allyl phenol), 9.7 mg/L (DAP) Time Sample volume Concentration (mg/L) (hr) (L) allyl phenol DAP 1. 15 1.000 0 1 0 1 1. 15 0 .050 0 1 0 1 2 .31 1. 010 0 3 0 1 2 31 0.050 0 6 0 2 3 .46 1.000 1. 4 0 4 3.46 0.050 2 6 0 7 4.05 0.500 3 7 0 9 4 05 0 050 4 8 1. 3 4.64 0.500 6 2 1. 7 4.64 0.050 7 4 2 2 5.30 0.565 8.5 2.7 5.30 0.050 9.5 3.3 6.44 1. 000 10.5 4.3 6.44 0.050 11. 0 5.2 7. 03 0.505 11. 0 5.6 7. 03 0.050 11. 1 6.1 7.62 0.505 11.1 6.5 7.62 0.050 11. 0 6.9 8.22 0.500 11. 1 7.1 8.22 0.050 11. 1 7.3 9.37 1. 000 10.8 7.7 9.37 0.050 10.9 8.2 11. 58 2.000 10.6 8.5 11. 58 0.050 10.6 9.2 13.78 2. 000 10.5 9.5 13.78 0. 050 10.1 9.7 14.91 1. 000 10.2 9.7 14.91 0.050 10.0 9.7

PAGE 250

238 TABLE C-14. Simultaneous solute addition data, allyl phenol and 2 4-dihydroxyacetophenone (DAP). Feed concentrations: 10.1 mg/L (allyl phenol), 9 4.1 mg/L (DAP) Time Sample volume Concentration (mg/L) (hr) (L) allyl phenol DAP 0. 06 0.050 0.1 0.5 0.12 0.050 0.2 • 1.2 0.18 0. 053 0.3 1. 9 0.24 0.050 0.5 2.8 0.29 0.050 0.7 3.8 0.35 0 050 0.9 4.9 0.41 0.050 1. 1 6.5 0.46 0.050 1.4 7.8 0.52 0.050 1. 7 9.8 0.63 0.100 2.1 12.8 0.74 0.100 3.0 15.8 0 .85 0.100 3.8 22.5 0. 96 0.100 4.9 29.6 1. 08 0.100 6.1 34.6 1. 36 0.255 8.4 46.0 1. 63 0.250 10.5 66.0 1. 91 0 250 11. 1 75.0 2.18 0.250 11. 1 82.0 2.78 0.500 10.7 87.0 2.78 0.050 10.5 87.5 3.38 0.500 10.4 88.0 3.38 0. 050 10.3 89.5 4.53 1. 000 10.5 91. 5 4.53 0.050 10.2 91. 0 5.68 1. 000 10.2 91. 5 5.68 0.050 10.1 90.5 6.87 1. 040 10.1 94.5 6.87 0.050 10.1 95.0 8.01 1. 000 10.1 94.5 8. 01 0.050 10.1 94.0

PAGE 251

239 TABLE C-15. Simultaneous solute addition data, allyl phenol and 2 4-dihydroxyacetophenone (DAP). Feed concentrations: 91.0 mg/L (allyl phenol), 9.1 mg/L (DAP) Time Sample volume Concentration (mg/L) (hr) (L) allyl phenol DAP 0.05 0 .050 0.2 0.1 0.11 0.050 0.6 0.1 0 16 0 050 1. 0 0.1 0 21 0.050 1. 5 0.1 0.26 0. 050 2.2 0.1 0 32 0 .050 3.2 0.1 0 .37 0.050 4.6 0.2 0.42 0.050 6.5 0.3 0.48 0.050 9.0 0.3 0 .58 0.100 14.5 0.5 0.69 0.100 24.1 0.7 0.79 0 100 35.4 1. 0 0 90 0.100 45.9 1. 4 1.00 0 100 55.5 1. 6 1. 26 0.250 67.0 2.1 1. 54 0.258 77.0 2.8 1. 80 0.250 84.5 3.3 2. 06 0.250 89.0 4.1 2.64 0.500 91. 5 4.8 2.64 0.050 93.0 5.4 3.22 0.500 92. 0 6.1 3.22 0.050 93.0 6.4 4.31 1. 000 92.0 7.0 4.31 0.050 94.0 7.5 5.42 1. 000 93.0 8.1 5.42 0.050 8.4 6.55 1.030 91. 5 8.5 6.55 0. 050 90.0 8.8 7.65 1. 000 90.5 8.8 7.65 0.050 89.0 8.9 9.83 2.000 91. 0 9.1 9.83 0.050 91. 0 8.9

PAGE 252

240 TABLE C-16. Simultaneous solute addition data, allyl phenol and 2 4-dihydroxyacetophenone (DAP). Feed concentrations: 100.5 mg/L (allyl phenol), 9 6.5 mg/L (DAP) Time Sample volume Concentration (mg/L) (hr) (L) allyl phenol DAP 0.12 0.114 0.2 0.1 0.23 0.100 1. 8 0.8 0.34 0.100 8.7 3.8 0.44 0.100 24.9 12.1 0.55 0.100 47.5 24.6 0.66 0.106 70.5 40.0 0.77 0.100 84.5 54.0 0.88 0.100 96.0 64.0 0.98 0.101 99.5 69.5 1. 09 0.102 100.0 76.5 1. 36 0.250 103.0 82.5 1. 62 0.250 103.5 87.5 1. 89 0.250 100.0 90.5 2.15 0.250 102. 0 92.5 3.27 1. 000 101. 5 94.5 3.27 0.050 102.0 96.0 5.48 2.025 98.5 95.5 5.48 0.050 101. 0 95.0

PAGE 253

APPENDIX D SEQUENTIAL AND SIMULTANEOUS SOLUTE ADDITION DATA FOR O-CRESOL/2 4-DIHYDROXYACETOPHENONE (DAP) SOLUTE PAIR

PAGE 254

TABLE D-1. Sequential solute addition data, o-cresol followed by 2 4-dihydroxyacetophenone (DAP). Feed concentration: 10.2 mg/L (o-cresol). Time Sample volume Concentration of o-cresol (hr) (L) (mg/L) 2.13 2.000 0.1 2.13 0.050 0.4 4.30 2.015 1. 3 4 .30 0.050 3.7 5.38 1. 000 4.9 5.38 0. 050 6.2 6.48 1. 000 7.1 6.48 0. 050 7.9 7.55 1. 000 8.1 7.55 0.050 8.3 9.77 2 050 8.7 9.77 0. 050 9.0 14.10 4. 000 9.1 14.10 0.050 9.5 24.97 10.250 9.7 24.97 0. 050 9.7 35.79 10.400 9.7 35.79 0.050 10.1 46.66 10.200 10.0 46.66 0.050 10.4 242

PAGE 255

243 TABLE D-2. Sequential solute addition data, o-cresol followed by 2 4-dihydroxyacetophenone (DAP). Feed concentrations: 10.3 mg/L (o-cresol), 9.7 mg/L (DAP) Time Sample volume Concentration (mg/L) (hr) (L) o-cresol DAP u z y U Z O U 11. 0 3 8 U Db f\ OCA 11. 5 4 1 Ci o o U o2 0.250 11. 7 4 5 1. (Jo 0.250 11.7 4 6 1.34 0.250 11.7 4 9 1. 62 0.260 11. 7 5.2 1. 88 0.250 11. 7 5.4 2.14 0.250 11. 6 5.7 2.41 0.250 11. 5 5.8 2.67 0.250 11. 5 6.0 2.93 0.250 11. 4 6.1 3.20 0.250 11.4 6.3 4.31 1. 000 11. 3 6.7 4.31 0.050 11.1 6.9 5.47 1. 000 11.1 7.1 5.47 0. 050 11. 0 7.4 7.59 2.000 10. 9 7.7 7.59 0.050 10.8 8.0 11. 87 4.020 11. 0 8.4 11. 87 0. 050 10.5 8.7 14.53 2.450 10.4 8.9 14.53 0.050 10.3 8.9 23. 93 7.060 10.3 9.5 23. 93 0.050 10.3 9.7

PAGE 256

244 TABLE D-3. Sequential solute addition data, o-cresol followed by 2 4-dihydroxyacetophenone (DAP). Feed concentrations: 9.8 mg/L (o-cresol), 94.5 mg/L (DAP) Time (hr) Sample volume (L) Concentration (mq/L) o-cresol DAP 0.021 0.025 0.048 0.025 11. 1 62.5 0.102 0.050 11.1 67.5 0.15 0.050 11.2 72.0 0.21 0.050 11. 0 75.0 0.26 0.050 11. 0 78.0 0.31 0.050 11. 0 80.0 0.36 0.050 11. 0 82.0 0.47 0.100 11. 0 82.0 0.57 0. 100 10. 9 85.0 0.69 0.115 10.7 85.0 0.79 0.100 10.7 82.5 0.89 0,100 10.7 87.5 1. 00 0.100 10.6 87.0 1. 57 0.500 10.6 88.0 1. 57 0.050 10.5 87.5 2.13 0.500 10.3 88.5 2.13 0.050 10.3 91. 5 3.25 1. 000 10.3 92.0 3.25 0. 070 10.2 92.5 4.33 1.000 10.1 92.0 4.33 0. 050 10.1 91. 5 5.46 1. 030 10.0 92. 0 5.46 0. 050 10.0 92.5 6.54 1. 000 10.0 94.5 6.54 0.050 9.9 95.0 7.63 1. 000 9.9 96.0 7.63 0. 050 9.9 95.5 8.72 1. 000 9.8 95.5 8.72 0.050 9.9 95.5 9.81 1. 000 9.9 94.0 9.81 0.050 9.9 93.5 *Sample lost.

PAGE 257

245 TABLE D-4. Sequential solute addition data, o-cresol followed by 2 4-dihydroxyacetophenone (DAP). Feed concentration: 101.0 mg/L (o-cresol) Time Sample volume Concentration of o-cresol (hr) (L) (mg/L) 0.11 0 100 0 3 0.21 0 100 1. 9 0 31 0 100 6 1 0.42 0.100 15.4 0.52 0.100 27.8 0.63 0.100 41. 4 0.73 0.100 54. 0 0.83 0.100 62.0 0. 94 0.100 69.5 1. 04 0.100 77.0 1. 14 0.100 83.0 1. 27 0.123 87.0 1. 37 0.100 90.5 1.47 0.100 93. 0 1.58 0.100 94.0 2.68 1. 000 98.0 2.68 0.050 99.5 4.81 2.000 100.0 4 81 0.050 100.5 6.95 2.000 100.5 6. 95 0.050 100.5

PAGE 258

246 TABLE D-5. Sequential solute addition data, o-cresol followed by 2 4-dihydroxyacetophenone (DAP). Feed concentrations: 100.5 mg/L (o-cresol), 9.7 mg/L (DAP) Time Sample volume Concentration (mg/L) (hr) (L) o-cresol DAP 0.05 0.050 100.0 0.1 0.11 0.050 101. 5 0.1 0.16 0.050 103.0 0.2 0.21 0.050 103.0 0.3 0.27 0. 050 103.5 0.3 0.32 0. 050 105.5 0.4 0.42 0.100 104.0 0.5 0.53 0.100 105.5 0.6 0.79 0.250 105. 0 0.7 1. 05 0.250 105.0 1. 0 1. 31 0.250 105.0 1.4 1. 57 0.250 105.5 1. 8 1. 82 0.250 105.0 2.2 2.08 0.250 105. 0 2.7 2.65 0.500 104.0 3.4 2.65 0. 050 104.0 3.7 3.22 0.500 104. 0 4.3 3.22 0.050 104. 0 4.9 4.29 1. 000 103.5 5.7 4.29 0.050 103.0 6.5 6.39 2.000 102.0 7.6 6.39 0.050 101. 5 8.5 8.49 2.000 101. 0 9.1 8.49 0.050 101. 0 9.4 14.32 5.550 100.5 9.4 14.32 0.050 100.5 9.6

PAGE 259

247 TABLE D-6. Sequential solute addition data, o-cresol followed by 2 4-dihydroxyacetophenone (DAP). Feed concentrations: 100.5 mg/L (o-cresol), 94.0 mg/L (DAP) Time Sample volume Concentration (mg/L) (hr) (L) o-cresol DAP 0.023 0.025 105.5 15. 7 0.050 0.025 115.5 25.0 0.10 0.050 118.0 35.5 0.15 0. 050 117.0 49.0 0.21 0.050 114.0 59.0 0.26 0. 050 112.5 66.0 0.31 0 050 109.5 71. 5 0.36 0.050 108. 0 75.5 0.47 0. 100 106.0 79.5 0.57 0.100 106.0 84.5 0.67 0.100 103.5 86.0 1. 82 1. 040 102.5 91. 0 1. 82 0.050 100.5 93.0 2.89 1. 000 100.5 94.0 2.89 0. 050 100.5 94.0

PAGE 260

248 TABLE D-7. Sequential solute addition data, 2 4-dyhydroxyacetophenone followed by o-cresol. Feed concentration: 9.7 mg/L (DAP) Time Sample volume Concentration of DAP (hr) (L) (mg/L) 2.11 2.000 2.11 0.050 4.22 2.000 0.2 4.22 0.050 0.5 5.31 1. 000 0.8 5.31 0.050 1. 2 6.40 1. 000 1. 8 6.40 0.050 2.6 7.48 1. 000 3.3 7.48 0.050 4.2 8.56 1. 000 4.7 8.56 0. 050 5.4 9.65 1. 000 5.8 9.65 0.050 6.2 10.75 1. 000 6.6 10.75 0.050 7.1 12.89 2. 000 7.6 12.89 0.050 8.2 15.05 2.000 8.5 15.05 0.050 8.7 24.77 4.330 9.7 24.77 0.050 9.7 *Below detection limit.

PAGE 261

249 TABLE D-8. Sequential solute addition data, 2 4-dihydroxyacetophenone followed by o-cresol. Feed concentrations: 9.6 mg/L (DAP), 10.1 mg/L ( o-cresol ) Time Sample volume Concentration (mg/L) (hr) (L) DAP o-cresol 0.05 0.050 10.0 2.0 0 15 0.100 10.4 4.0 0.25 0.100 10.7 5.6 0.36 0.100 11. 0 6.6 0.46 0.100 11.1 7.3 0.56 0.100 11. 2 7.8 0.67 0.102 11.2 8.2 0.77 0.100 11. 2 8.5 1. 03 0.250 11. 1 8.8 1.29 0.250 11. 1 9.2 1. 55 0.250 10.9 9.4 1. 81 0.250 10.7 9.5 2.88 1. 000 10.5 9.7 2.88 0.050 10.3 9.8 3.97 1. 000 10.2 9.9 3. 97 0.050 9.5 9.5 6.09 2.000 10.1 10.1 6.09 0.050 10.0 10.1 8.21 2.000 9.8 10.1 8.21 0.050 9.7 10.1

PAGE 262

250 TABLE D-9. Sequential solute addition data, 2 4-dihydroxyacetophenone followed by o-cresol. Feed concentrations: 9.5 mg/L (DAP), 10 3.0 mg/L ( o-cresol ) Time Sample volume Concentration (mg/L) (hr) (L) DAP o-cresol 0.05 0.050 12.5 40.0 0.10 0.050 14. 9 60.0 0.15 0.050 16.2 70.5 0.20 0.050 16.8 77.5 0.25 0.050 17.3 81. 5 0.31 0.050 17.7 85.0 0.41 0.100 17.7 88.5 0.51 0.100 17.7 91. 0 0.62 0.100 17.6 94.0 0.72 0.100 17.2 95.0 0.82 0. 100 17.0 96.5 1. 39 0.500 14. 9 97.5 1. 39 0.050 15.0 98.5 2.49 1. 015 13.8 99.5 2.49 0.050 12. 8 100.0 3.57 1. 000 12.3 100.5 3.57 0.050 11. 8 101. 0 5.70 2.000 11. 0 101. 5 5.70 0.050 10.5 103.0 7.82 2.000 10.3 103.0 7.82 0.050 10.2 103.0 9. 96 2. 000 10.0 103. 0 9.96 0.050 9.9 103. 0

PAGE 263

251 TABLE D-10. Sequential solute addition data, 2 4-dihydroxyacetophenone followed by o-cresol. Feed concentration: 95.5 mg/L (DAP) Time Sample volume Concentration of DAP (hr) (L) (mg/L) 0.06 0.050 0.11 0.050 0.2 0.16 0.050 0.3 0.27 0.100 0.7 0.37 0.100 1. 6 0.47 0.100 3.6 0.58 0.100 7.3 0.68 0.100 13.2 0.78 0.100 21. 8 0.89 0.100 32.1 0.99 0.100 43.1 1. 09 0.100 51. 6 1. 20 0.100 58.6 1. 30 0.100 64.6 1.42 0.112 68.5 1. 52 0.100 72.5 1. 64 0.117 76.0 1.74 0.100 79.0 1. 84 0.100 81. 5 1. 95 0.100 83.5 3.05 1. 000 89.5 3.05 0.050 93.0 4.15 1. 000 94.0 4.15 0.050 95.0 5.25 1. 000 95.0 5.25 0.050 95.5 6.38 1. 025 95.5 6.38 0.050 95.5 *Below detection limit.

PAGE 264

252 TABLE D-11. Sequential solute addition data, 2 4-dihydroxyacetophenone followed by o-cresol. Feed concentrations: 9 8.5 mg/L (DAP), 9.7 mg/L ( o-cresol ) Time Sample volume Concentration (mg/L) (hr) (L) DAP o-cresol 0. 026 0.025 100.0 2.8 0.08 0.050 100. 0 6.2 0.13 0.050 100.5 7.5 0.19 0.050 100.5 8.2 0.24 0.050 101. 0 8.5 0.29 0.050 100.0 8.7 0.40 0.100 100.5 9.0 0.51 0.100 100.5 9.1 0.77 0.250 100.0 9.3 1. 04 0.250 99.5 9.5 2.15 1. 000 99.0 9.6 2.15 0.050 99.0 9.7 3.25 1. 000 99.0 9.7 3.25 0.050 98.5 9.7 5.36 2.000 98.5 9.7 5.36 0.050 98.5 9.7

PAGE 265

253 TABLE D-12. Sequential solute addition data, 2 4-dihydroxyacetophenone followed by o-cresol. Feed concentrations: 96 mg/L (DAP), 101.0 mg/L (o-cresol ) Time Sample volume Concentration (mg/L) (hr) (L) DAP o-cresol 0.021 0.025 103.5 46.5 0.044 0.025 107.0 74.5 0.10 0.050 106. 0 82.0 0.15 0.050 106.5 87.5 0.20 0.050 105.5 90.5 0.25 0.050 104.5 93.0 0.30 0.050 103.5 94.5 0.35 0. 050 102.0 94.5 0.46 0.100 102. 0 97.0 0.56 0.100 101. 0 97.5 0.67 0.100 100.0 98.5 1. 75 1. 000 98.5 100.0 1. 75 0.050 97.0 100.5 2.33 0.510 97.0 100.5 2.33 0.050 96.0 100.5

PAGE 266

254 TABLE D-13. Simultaneous solute addition data, o-cresol and 2 4-dihydroxyacetophenone. Feed concentrations: 10.3 mg/L (o-cresol), 9.8 mg/L (DAP) Time (hr) Sample volume (L) Concentration (mq/L) o-cresol DAP 1. 07 1. 000 1. 07 0.050 0.2 2.14 1. 000 0.9 2.14 0.050 2.1 3.22 1. 000 5.1 0.3 3.22 0.050 8.1 0.4 3.79 0.505 9.7 0.5 3.79 0.050 10.7 0.6 4 .35 0.500 11. 4 0.7 4.35 0.050 12. 1 0 8 4.92 0.500 12.5 1. 1 4.92 0.050 12.9 1. 3 5.48 0.500 13.0 1. 6 5.48 0.050 13.2 1. 9 6.05 0.500 13. 1 2.4 6.05 0.050 13.1 2.8 6.60 0.500 12. 9 3.2 6.60 0.050 12. 8 3.8 7.18 0.500 12.5 4.3 7.18 0.050 12.4 4.8 8.27 1. 000 12.1 5.6 8.27 0.050 11.7 6.4 10.38 2.000 11. 3 7.6 10.38 0.050 11, 0 8.5 12.48 2. 000 10.5 8.9 12.48 0. 050 10.5 9.4 14.61 2.000 10.4 9.5 14.61 0.050 10.3 9.7 24.65 0.800 10.3 9.8 24.65 0.050 10.3 9.8 *Below detection limit.

PAGE 267

255 TABLE D-14. Simultaneous solute addition data, o-cresol and 2 4-dihydroxyacetophenone. Feed concentrations: 9.4 mg/L (o-cresol), 9 0.0 mg/L (DAP) Time (hr ) Sample volume (L) Concentration (mq/L) o-cresol DAP 0 10 0 100 0.22 0.124 0.2 0.32 0 100 0.2 0.42 0 100 0.3 0.5 0 52 0. 100 0.7 1. 1 0.63 0 103 1.7 2.6 0.73 0.100 3.3 5.7 0 83 0. 102 5.7 10.9 0 93 0 100 8.5 19.2 1. 03 0 100 10.7 29.7 1. 30 0.275 12.4 48.3 1. 54 0.250 12.4 66.0 J. o u u Z D D 11. 7 76.5 2.05 0.250 11. 1 78.5 3.10 1. 000 10.3 83.5 3.10 0.050 9.9 86.0 4.16 1. 000 9.7 86.5 4.16 0.050 9.6 86.0 6.22 2.000 9.5 91. 0 6.22 0. 050 9.4 91. 5 8.27 2.000 9.4 91. 5 8.27 0.050 9.4 91. 5 10.33 2.000 9.4 91. 5 10.33 0.050 9.4 91. 5 *Below detection limit.

PAGE 268

256 TABLE D-15. Simultaneous solute addition data, o-cresol and 2 4-dihydroxyacetophenone Feed concentrations: 101.0 mg/L (o-cresol), 10.1 mg/L (DAP) Time Sample volume Concentration (mq/L) (hr) (L) o-cresol DAP 0.10 0.100 0.21 0.100 0.2 0.31 0.100 1. 5 0.42 0.100 5.0 0. 52 0.100 13.2 0.63 0 100 29.6 0.73 0.100 50.3 0.84 0.100 72.5 0.95 0.100 86.5 0.1 1. 05 0.100 95.0 0.1 1.31 0.250 100.5 0.1 1. 58 0.250 104.0 0.3 1. 84 0.250 104.5 0.3 2.10 0.250 104.5 0.5 3.2 1. 000 104.5 0.9 3.2 0.050 104.5 1. 6 4.3 1. 000 104.5 2.4 4.3 0.050 104.0 3.3 5.46 2.000 102.5 5.6 5.46 0.050 102.5 7.5 7.64 2.000 101. 0 8.7 7.64 0.050 101. 0 9.3 13. 92 5. 990 101. 0 9.4 13.92 0.050 101. 0 9.7 *Below detection limit.

PAGE 269

257 TABLE D-16. Simultaneous solute addition data, o-cresol and 2 4-dihydroxyacetophenone Feed concentrations: 105.0 mg/L (o-cresol), 9 4.0 mg/L (DAP) Time Sample volume Concentration (mg/L) (hr) (L) o-cresol DAP 0.11 0.100 0.22 0.100 2.8 0.5 0.32 0.100 14.4 1. 6 0.43 0.100 43.0 4.9 0.53 0.100 78.5 11.2 0.64 0.100 103.5 20.0 0.74 0.100 115.5 28.0 0.85 0.100 121. 0 35.0 0.95 0.100 122.5 45.5 1. 21 0.250 122.0 58.0 1.47 0.250 118.0 70.5 1. 73 0.250 113.0 79.5 1. 99 0.250 111. 0 85.0 2.54 0.500 108. 0 89.5 2.54 0.025 106.0 91. 0 3.13 0.545 106.0 92.5 3.13 0.025 105.0 93.0 5.28 2.055 105.0 94.5 *Below detection limit.

PAGE 270

APPENDIX E MULTI -SOLUTE SIMULTANEOUS ADDITION DATA

PAGE 271

TABLE E-1. Simultaneous solute addition data, 2,4dihydroxyacetophenone, o-cresol, and allyl phenol Feed concentrations: 9.4 mg/L (DAP), 10.1 mg/L (o-cresol), 94.0 mg/L (allyl phenol) Time Sample volume Concentration (mq/L) (hr) (L) DAP o-cresol allyl phenol 0.11 0.100 0.1 0.21 0.100 0.3 0.4 0.33 0.100 0.6 1. 5 0.43 0.100 0.2 1. 5 4.2 0.54 0.100 0.3 3.3 9.0 0.65 0.100 0.5 5.1 15.7 0.76 0.100 0.7 6.6 23.7 0.87 0.100 0.9 8.0 30.9 0. 97 0.100 0.9 9.6 39.1 1. 08 0.100 1.2 11.2 46.3 i 3 5 0.250 1. 6 13. 3 59.5 1. 62 0.250 2.2 13.3 73.0 1. 89 0.250 2.9 12.3 84.0 2.16 0.250 3.4 11.7 89.0 3.26 1. 025 4.9 10.8 91. 5 3.26 0.025 6.0 10.5 90.0 5.44 2 000 7.4 10.2 90.5 5.44 0.025 8.4 10.1 91. 0 7.70 2.075 8.9 10.1 90.5 7.70 0.025 9.2 10.1 93.5 8.88 2.000 9.4 10.1 95.0 8.88 0. 025 9.5 10.1 94.0 *Below detection limit. 259

PAGE 272

260 TABLE E-2. Simultaneous solute addition data, 2,4dihydroxyacetophenone o-cresol, and allyl phenol. Feed concentrations: 9.9 mg/L (DAP), 102.0 mg/L (o-cresol), 9.6 mg/L (allyl phenol) Time (hr) Sample volume (L) Concentration (mq/L) DAP o-cresol allyl phenol 0 10 0 100 0 21 0 100 0 31 0 100 0.7 0.42 0 100 4.5 0.52 0.100 0 2 15.5 0.1 0.63 0 105 0 2 40.0 0.2 0.74 0 100 0.2 67.0 0.4 0 84 0 100 0.2 90.0 0.6 0.95 0.100 0.2 100.0 0.8 U 0 n 1 n "7 u xu / U z iUD 5 1. 0 1. 32 0.250 0.3 108.5 1.6 1. 58 0.250 0.4 109.5 2.6 1. 84 0.250 0.5 109.5 3.4 2.10 0.250 0.8 108.5 4.5 3.16 1. 000 1. 7 107.0 6.9 3.16 0.025 2.7 105.5 8.5 5.33 2. 000 5.1 104. 0 9.5 5.33 0.025 7.2 103. 0 9.7 7.50 2.060 8.5 102.5 9.7 7.50 0.025 9.2 102.0 9.6 14.78 7.000 10.0 102.0 9.6 14.78 0.025 10.1 101. 5 9.6 *Below detection limit.

PAGE 273

261 TABLE E-3. Simultaneous solute addition data, 2,4dihydroxyacetophenone, o-cresol, and allyl phenol. Feed concentrations: 97.0 mg/L (DAP), 9.9 mg/L (o-cresol), 10.2 mg/L (allyl phenol) Time Sample volume Concentration (mq/L) (hr) (L) DAP o-cresol allyl phenol 0.10 0.100 0.21 0.100 0.1 0.32 0.100 0.9 0.5 0.2 0.42 0.100 2.4 1. 3 0.8 0.53 0.100 5.8 3.0 1.7 0.64 0.100 12.3 5.1 3.2 0.74 0.100 21. 0 7.1 5.0 0.86 0.104 29.9 8 6 6.0 0. 96 0.100 40.0 9.6 7.8 1. 07 0.100 47.6 10.2 8.6 1. 34 0.250 59.1 11.2 9.6 1.60 0.250 73.3 12.0 10.8 1. 87 0.250 83.0 11. 8 11. 3 2.24 0.345 85.5 11.7 11. 1 3.33 1. 000 93.0 10.5 10.6 3.33 0.025 94.0 10.1 10.3 5.49 2.000 97.5 10.0 10.1 5.49 0.025 97.5 9.9 10.1 *Below detection limit.

PAGE 274

262 TABLE E-4. Simultaneous solute addition data, 2,4dihydroxyacetophenone o-cresol, and allyl phenol Feed concentrations: 10.3 mg/L (DAP), 10.4 mg/L (o-cresol), 9.8 mg/L (allyl phenol) Time Sample volume Concentration (mg/L) (hr) (L) DAP o-cresol allyl phenol 0 .26 0.250 4r ie 0 51 0.254 4r 0 3 "k 0.76 0.251 ie 0 5 ie 1. 02 0.250 1. 1 0.1 1.27 0.250 2 2 0 3 1.53 0.250 0 3 4 0 0 6 1.78 0.250 0 4 5 9 1. 6 2 04 0.253 0 6 8 6 1. 8 2.30 0.250 0 8 10.3 2 6 2.55 0.250 1. 1 11.5 3.5 2 92 0.360 1. 6 12 1 4.5 3 17 0.335 1. 8 12 3 5.8 3.71 0 500 2 5 12 3 6.9 3 .71 0 025 3 2 12.3 7.5 4.25 0 500 3 5 12.3 8.1 4.25 0. 025 3 9 12 1 8.6 4.79 0.500 4 4 11. 9 9.0 4.79 0.025 4.9 11.7 9.2 5.33 0.500 5.3 11.7 9.4 5.33 0. 025 5.7 11.4 9.6 5.85 0.500 6.0 11.2 9.5 5.85 0. 025 6.4 11. 3 9.8 6.39 0.500 6.7 11.1 9.8 6.39 0 025 7.1 11. 0 9.9 7.44 1.000 7.5 10.9 9.9 7.44 0.025 7.9 10.7 9.9 8. 54 1. 030 8.3 10.6 9.9 8.54 0. 025 8.7 10.6 9.9 10.60 2.000 9.1 10.4 9.8 10.60 0.025 9.6 10.4 9.8 12.68 2. 000 9.7 10.4 9.7 12.68 0.025 9.9 10.3 9.6 14.74 2.000 10.1 10.4 9.7 14.74 0.025 10.2 10.4 9.7 24. 00 9.100 10.2 10.4 9.7 24.00 0. 025 10.2 10.4 9.7 *Below detection limit.

PAGE 275

263 Table E-5. Simultaneous solute addition data, 2,4dihydroxyacetophenone o-cresol, allyl phenol, and a-terpineol. Feed concentrations: 9.6 mg/L (DAP), 10.2 mg/L (o-cresol), 10.0 mg/L (allyl phenol), 10.0 Mg/L (a-terpineol ) Time Sample volume Concentration (mg/L) (hr) (L) DAP o-cresol 0.26 0.250 0.53 0.250 0.2 0.80 0.250 0.3 1. 06 0.250 0.8 1.33 0.250 0.1 1.8 1.60 0.253 0.3 3.5 1.86 0.250 0.5 5.9 2.13 0.250 0.8 8.4 2.40 0.250 1.1 10.5 2.65 0.250 1.7 11. 7 2.92 0.250 2.2 12.5 3.19 0.250 2.8 12. 9 3.74 0.500 3.7 12.8 3.74 0.025 4.4 12.8 4.30 0.500 5.0 12.5 4.30 0. 025 5.4 12.1 4.86 0.500 5.9 11. 9 4.86 0.025 6.4 11. 7 5. 96 1. 000 7.0 11. 4 5.96 0.025 7.5 11. 0 7.06 1. 010 8.0 10.7 7.06 0.025 8.4 10.5 9.27 2.080 8.8 10.5 9.27 0.025 9.1 10.3 14.81 5.220 9.5 10.2 14.81 0.025 9.7 10.2 *Below detection limit.

PAGE 276

264 Table E-6. Simultaneous solute addition data, 2,4dihydroxyacetophenone, o-cresol, allyl phenol, and a-terpineol. Feed concentrations: 10.3 mg/L (DAP), 101.0 mg/L (o-cresol), 10.0 mg/L (allyl phenol), 10.0 mg/L (a-terpineol ) Time Sample volume Concentration (mg/L) (hr) (L) DAP o-cresol 0.25 0.255 0.4 0.36 0.100 4.3 0.46 0.100 0.1 14.4 0.56 0.100 0.1 38.2 0.67 0. 100 0.2 66.5 0.78 0.105 0.2 90.5 0.88 0.100 0.3 102.5 0.98 0.100 0.3 108.0 1. 08 0.100 0.4 110.0 1. 18 0.100 0.5 111. 0 1. 44 0.250 0.7 110.5 1. 70 0.260 1. 1 109.5 1. 96 0.250 1.7 109. 0 2.21 0.250 2.5 107. 0 3.26 1.000 4.6 105.0 3.26 0.025 6.2 103.0 5.35 1. 020 8.1 102.5 5.35 0.025 9.4 101. 0 7.45 1. 040 9.9 99.5 7.45 0. 025 10.3 100.5 9.64 2.060 10.3 101. 0 9.64 0.025 10.3 100.5 *Below detection limit.

PAGE 277

REFERENCES Osipow, L.I. 1962. Surface Chemistry Robert E. Krieger Publishing Company, Huntington, New York. Clark, A. 1970. The Theory of Adsorption and Catalysis Academic Press, New York. Rosen, M.J. 1978. Surfactants and Interfacial Phenomena. Wiley-Interscience New York. Adamson, A.W. 1982. Physical Chemistry of Surfaces Wiley-Interscience, New York. Parfitt, G.D., and Rochester, C.H. Eds. 1983. Adsorption from Solution at the Solid/Liquid Interface Academic Press, London. Ottewill, R.H., Rochester, C.H., and Smith, A.L. Eds. 19 83. Adsorption from Solution Academic Press, London. Ruthven, D.M. 19 84. Principles of Adsorption and Adsorption Processes Wiley-Interscience, New York. Smisek, M. and Cerny, S. 1970. Active Carbon; Manufacture, Properties and Applications Elsevier Publishing Company, Amsterdam. Weber, W.J., Jr. 1972. Physicochemical Processes for Water Quality Control Wiley-Interscience, New York. Hassler, J.W. 1974. Purification with Activated Carbon. Chemical Publishing Company, New York. McCreary, J.J., and Snoeyink, V.L. 1977. Granular Activated Carbon in Water Treatment. Journal American Water Works Association 6_9 ( 8 ): 437-444 (August). McGuire, M.J., and Suffet, I.H. 1978. Adsorption of Organics from Domestic Water Supplies. Journal American Water Works Associatio n 70(11) : 621-636 ( November )

PAGE 278

266 13. Cherimisinof f P.N., and Ellerbusch, F. Eds. 1978. Carbon Adsorption Handbook Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan. 14. Suffet, I.H. 1980. An Evaluation of Activated Carbon for Drinking Water Treatment: A National Academy of Science Report. Journal American Water Works Association 7^(l):44-50 (January). 15. Miller, S. 1980a. Adsorption on Carbon: Theoretical Considerations. Environmental Science and Technology 14(8) :910-914 (August). 16. Miller, S. 1980b. Adsorption on Carbon: Solvent Effects on Adsorption. Environmenal Science and Technology 14 ( 9) :1037-1049 (September). 17. Suffet, I.H., and McGuire, M.M. Eds. 1980. Activated Carbon Adsorption of Organics from the Aqueous Phase — Volume 1 Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan. 18. Suffet, I.H., and McGuire, M.J. Eds. 1981. Activated Carbon Adsorption of Organics from the Aqueous Phase — Volume 2 Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan. 19. Suffet, I.H., and McGuire, M.J. Eds. 1983. Treatment of Water by Granular Activated Carbon American Chemical Society, Washington, D.C. 20. Weber, W.J., Jr., and Van Vliet, B.M. 1980. Fundamental Concepts for Application of Activated Carbon in Water and Wastewater Treatment. In: Activated Carbon Adsorption of Organics from the Aqueous Phase, Volume 1 Edited by Suffet, I.H., and McGuire, M.J. Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan. pp. 15-41. 21. Mattson, J.S., Mark, H.B., Malbin, M.D., Weber, W.J., Jr., and Crittenden, J.C. 1969. Surface Chemistry of Active Carbon: Specific Adsorption of Phenols. Journal of Colloid and Interface Science 31(1): 116-130 (September). 22. Radke, C.J., and Prausnitz, J.M. 1972. Adsorption of Organic Solutes from Dilute Aqueous Solution on Activated Carbon. Industrial Engineering and Chemistry Fundamentals, 11 ( 4 ): 445-451.

PAGE 279

267 23. Fornwalt, H.J., and Hutchins, R.A. 1966. Purifying Liquids with Activated Carbon. Chemical Engineering 73:179-184 (April). 24. Hutchins, R.A, 1981. Activated Carbon. In: Activated Carbon Adsorption for Wastewater Treatment Edited by Perrich, J.R. CRC Press, Inc., Boca Raton, Florida, pp. 30-40. 25. Mahajan, O.P., Moreno-Castilla, C. and Walker, P.L., Jr. 1980. Surface-treated Activated Carbon for Removal of Phenol from Water. Separation Science and Technology 15 ( 10 ): 1733-1752 26. Puri, B.R. 1980. Carbon Adsorption of Pure Compounds and Mixtures from Solution Phase. In: Activated Carbon Adsorption of Organics from the Aqueous Phase Volume 1 Edited by Suffet, I.H., and McGuire, M.J. Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan. pp. 353-378. 27. Narkis, N. and Ben-David, B. 1985. Adsorption of Non-ionic Surfactants on Activated Carbon and Mineral Clay. Water Research 19 ( 7 ) : 815-824 28. Cookson, J.T., Jr. 1978. Adsorption Mechanism: The Chemistry of Organic Adsorption on Activated Carbon. In: Carbon Adsorption Handbook Edited by Cheremisinof f P.N., and Ellerbusch, F. Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan. pp. 241-279. 29. Bell, B.A-, and Molof, A.H. 1975. A New Model of Granular Activated Carbon Adsorption Kinetics. Water Research, 9:857-860. 30. Keinath, T.M. and Carnahan, R. 1973. Mathematical Modeling of Heterogeneous Sorption in Continuous Contactors for Wastewater Decontamination Clemson University, Clemson, South Carolina. 31. Wagner, N.J., and Jula, R.J. 1981. Activated Carbon Adsorption. In: Activated Carbon Adsorption for Wastewater Treatment Edited by Perrich, J.R. CRC Press, Inc., Boca Raton, Florida. pp. 41-60. 32. Snoeyink, V.L., and Weber, W.J., Jr. 1967. The Surface Chemistry of Active Carbon: Discussion of Structure and Surface Functional Groups. Environmental Science and Technology l{3):228-234 (March).

PAGE 280

268 33. Coughlin, R.W., and Ezra, F.S. 1968. Role of Surface Acidity in the Adsorption of Organic Pollutants on the Surface of Carbon. Environmental Science and Technology 2(4):291-297 (April). 34. Coughlin, R.W. Ezra, F.S., and Tan, R.N. 1968. Influence of Chemisorbed Oxygen in Adsorption onto Carbon from Aqueous Solution. Journal of Colloid and Interface Science 28^( 3/4 ): 386-396 (November-December). 35. Cookson, J.T., Jr. 1975. The Adsorption Chemistry of Organics from Waters on Activated Carbon. AIChE Symposium Series, Water-1975 71 ( 151 ): 376-386 36. Mahajan, O.P., Youssef, A., and Walker, P.L., Jr. 1978. Surface-treated Activated Carbon for Removal of Ammonia from Water. Separation Science and Technology 13 (6 ) :487-499. 37. Ishizaki, C, Marti, I., and Ruiz, M. 1983. Effect of Surface Characteristics of Activated Carbon on the Adsorption of Chloroform from Aqueous Solution. In: Treatment of Water by Granular Activated Carbon Edited by McGuire, M.J., and Suffet, I.H. American Chemical Society, Washington, D.C. pp. 95-107. 38. Boehm, H.P. 1966. Chemical Identification of Surface Groups. In: Advances in Catalysis and Related Subjects, Volume 16 Edited by Eley, D.D., Pines, H. and Weisz, P.B. Academic Press, New York. pp. 179-269. 39. Cookson, J.T. 1980. Physicochemical Changes of Substances By or Within Carbon Adsorption Beds. In: Activated Carbon Adsorption of Organics from the Aqueous Phase, Volume 1 Edited by Suffet, I.H., and McGuire, M.J. Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan. pp. 379-396. 40. Coughlin, R.W. and Tan, R.N. 1968. Role of Functional Groups in Adsorption of Organic Pollutants on Carbon. Chemical Engineering Progress Symposium Series. 64(90 :207-214. 41. Graham, D.J. 1955. Characterization of Physical Adsorption Systems. III. The Separate Effects of Pore Size and Surface Acidity upon the Adsorbent Capacities of Activated Carbons. Journal of Physical Chemistry 59:896-900. 42. Zogorski, J.S., and Faust, S.D. 1978. Operational Parameters for Optimum Removal of Phenolic Compounds

PAGE 281

269 from Polluted Waters by Columns of Activated Carbon. In: Carbon Adsorption Handbook Edited by Cheremisinof f P.N., and Ellerbusch, F. Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan, pp. 753-777. 43. Letterman, R.D., Quon, J.E., and Gemmell, R.S. 1974. Film Transport Coefficient in Agitated Suspensions of Activated Carbon. Journal Water Pollution Control Federation, 46 ( 11 ): 2536-2546 44. Keinath, T.M., and Weber, W.J., Jr. 1968. A Predictive Model for the Design of Fluid-Bed Adsorbers. Journal Water Pollution Control Federation 40^(4) :741-765. 45. Dedrick, R.L., and Beckman, R.B. 1967. Kinetics of Adsorption by Activated Carbon from Dilute Aqueous Solution. Chemical Engineering Progress Symposium Series 63(74) ;68-78. 46. Martin, R.J., and Al-Bahrani, K.S. 1978. Adsorption Studies Using Gas-Liquid Chromatography — III. Experimental Factors Influencing Adsorption. Water Research 12 ; 879-888. 47. Randtke, S.J., and Snoeyink, V.L. 1983. Evaluating GAC Adsorptive Capacity. Journal American Water Works Association 75(8) :406-413. 48. Martin, R.J. 1980. Activated Carbon Product Selection for Water and Wastewater Treatment. Industrial Engineering and Chemistry Product Research and Development 19 ( 3 ): 435-441. 49. Snoeyink, V.L., Weber, W.J., Jr., and Mark, H.B. 1969. Sorption of Phenol and Nitrophenol by Activated C a rb on Environmental Science and Technology 3 (10 ) :918-926. 50. McCreary, J.J., and Snoeyink, V.L. 1980. Characterization and Activated Carbon Adsorption of Several Humic Substances. Water Research 14 ; 151-160 51. Weber, W.J., Jr., and Morris, J.C. 1964. Adsorption in Heterogeneous Aqueous Systems. Journal American Water Works Association 56(4); 447-457 52. Weber, W.J., Jr., and Morris, J.C. 1964. Equilibria and Capacities for Adsorption on Carbon. Journal of the Sanitary Engineering Division, American Society of Civil Engineers 90 (SA3 ): 79-107

PAGE 282

270 53. Bilello, L.J., and Beaudet, B.A. 1983. Evaluation of Activated Carbon by the Dynamic Minicolumn Adsorption Technique. In: Treatment of Water by Granular Activated Carbon Edited by McGuire, M.J., and Suffet, I.H. American Chemical Society, Washington, D.C. pp. 213-246. 54. Weber, W.J., Jr., Voice, T.C. and Jodellah, A. 1983. Adsorption of Humic Substances: The Effects of Heterogeneity and System Characteristics. Journal American Water Works Association 75(12) : 612-619 55. Perrich, J.R. Ed. 1981. Activated Carbon Adsorption for Wastewater Treatment CRC Press, Inc., Boca Raton, Florida. pp. 30-31. 56. Yehaskel, A. 1978. Activated Carbon: Manufacture and Regeneration Noyes Data Corporation, Park Ridge, New Jersey. 57. Giusti, D.M., Conway, R.A. and Lawson, C.T. 1974. Activated Carbon Adsorption of Petrochemicals. Journal Water Pollution Control Federation 46 ; 947-965. 58. Al-Bahrani, K.S., and Martin, R.J. 1976. Adsorption Studies Using Gas-Liquid Chromatography — 1. Effect of Molecular Structure. Water Research 10 ; 731-736 59. Singer, P.C., and Yen, C.Y. 1980. Adsorption of Alkyl Phenols by Activated Carbon. In: Activated Carbon Adsorption of Organics from the Aqueous Phase, Volume 1. Edited by Suffet, I.H., and McGuire, M.J. Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan, pp. 167-191. 60. Martin, R.J., and Al-Bahrani, K.S. 1977. Adsorption Studies Using Gas-Liquid Chromatography — II. Competitive Adsorption. Water Research 44 : 994-999 61. El-Dib, M.A., and Badawy, M.I. 1979. Adsorption of Soluble Aromatic Hydrocarbons on Granular Activated Carbon. Water Research 43 : 255-258 62. McGuire, M.J., and Suffet, I.H. 1980. The Calculated Net Adsorption Energy Concept. In: Activated Carbon Adsorption of Organics from the Aqueous Phase, Volume 1. Edited by Suffet, I.H., and McGuire, M.J. Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan, pp. 91-115. 63. McGuire, M.J., Suffet, I.H., and Radziul, J.V. 1978. Assessment of Unit Processes for the Removal of Trace

PAGE 283

271 Organic Compounds from Drinking Water. Journal American Water Works Association ^ 70(10) ; 565-572 64. Manes, M. 1980. The Polanyi Adsorption Potential Theory and Its Applications to Adsorption from Water Solution onto Activated Carbon. In: Activated Carbon Adsorption of Orqanics from the Aqueous Phase, Volume 1. Edited by Suffet, I.H., and McGuire, M.J. Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan, pp. 43-64. 65. Arbuckle, W.B., and Romagnoli, R.J. 1980. Prediction of the Preferentially Adsorbed Compound in Bisolute Column Studies. American Institute of Chemical Engineers Symposium Series 76 : 77-85 66. Yen, C, and Singer, P.C. 1984. Competitive Adsorption of Phenols on Activated Carbon. Journal of Environmental Engineering 110 ( 5 ) ; 976-989 67. Martin, R.J., and Al-Bahrani, K.S. 1979. Adsorption Studies Using Gas-Liquid Chromatography: IV. Adsorption from Bi-Solute Systems. Water Research 13:1301-1304. 68. Zogorski, J.S., and Faust, S.D. 1976. The Effect of Phosphate Buffer on the Adsorption of 2 4-Dichlorphenol and 2 4-Dinitrophenol. Journal of Environmental Science and Health, All(8 & 9):501-515. 69. McCreary, J.J., and Snoeyink, V.L. 1980. Characterization and Activated Carbon Adsorption of Several Humic Substances. Water Research 14^:151-160. 70. Digiano, F.A. and Weber, W.J., Jr. 1973. Sorption Kinetics in Infinite Bath Experiments. Journal Water Pollution Control Federation £5 { 4 ) : 713-725 71. Randtke, S.J., and Jepsen, CP. 1982. Effects of Salts on Activated Carbon Adsorption of Fulvic Acids. Journal American Water Works Association 72 (2 ) : 84-93 72. Getzen, F.W. and Ward, T.M. 1969. A Model for the Adsorption of Weak Electrolytes on Solids as a Function of pH. Journal of Colloid and Interface Science 3_4 (4 ) :444-453. 73. Snoeyink, V.L., and Weber, W.J., Jr. 1968. Reaction of the Hydrated Proton with Active Carbon. In: Advances in Chemistry Series. 79 Edited by Gould, R.F. American Chemical Society Publications, Washington, D.C. pp. 112-134.

PAGE 284

272 74. Murin, C.J., and Snoeyink, V.L. 1979. Competitive Adsorption of 2 4-Dichlorophenol and 2 4 6-Trichlorophenol in the Nanomolar to Micromolar Concentration Range. Environmental Science and Technology 13 (3 ) :305-311. 75. Jain, J.S., and Snoeyink, V.L. 1973. Adsorption from Bisolute Systems in Active Carbon. Journal Water Pollution Control Federation 45 (12 ) :2643-2479. 76. Huang, J.C., and Steffens, C.T. 1976. Competitive Adsorption of Organic Materials by Activated Carbon. Proceedings of the 31st Purdue University Industrial Waste Conference Edited by Bell, J. Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan. pp. 107-121. 77. Fritz, W. and Schluender, E.U. 1974. Simultaneous Adsorption Equilibria of Organic Solutes in Dilute Aqueous Solution on Activated Carbon. Chemical Engineering Science 29 ; 1279-1282 78. Radke, C.J., and Prausnitz, J.M. 1972. Thermodynamics of Multi-Solute Adsorption from Dilute Liquid Solutions. American Institute of Chemical Engineering Journal, 1^( 4 ): 761-768 79. Weber, W.J., Jr., and Pirbazari, M. 1983. Removal of Carbon Tetrachloride from Water by Activated Carbon. In: Treatment of Water by Granular Activated Carabon Edited by McGuire, M.J., and Suffet, I.H., Advances in Chemistry Series No. 202. American Chemical Society, Washington, D.C. pp. 121-166. 80. Wilde, K.A. 1980. Milticomponent Adsorption Column Parameter Studies. In: Activated Carbon Adsorption of Organics from the Aqueous Phase, Volume 1 Edited by Suffet, I.H., and McGuire, M.J. Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan. pp. 251-272. 81. Frick, B.R. and Sontheimer, H. 1983. Adsorption Equilibria in Multisolute Mixtures of Known and Unknown Composition. In: Treatment of Water by Granular Activated Carbon Edited by McGuire, M.J., and Suffet, I.H., Advances in Chemistry Series No. 202. American Chemical Society, Washington, D.C. pp. 247-268. 82. Jossens, L. Prausnitz, J.M., Fritz, W., Schluender, E.U., and Myers, A.L. 1978. Thermodynamics of MultiSolute Adsorption from Dilute Aqueous Solutions. Chemical Engineering Science 33 : 1097-1106

PAGE 285

273 83. Weber, W.J., Jr., Voice, T.C., Pirbazari, M. Hunt, G.E., and Ulanoff, D.M. 1983. Sorption of Hydrophobic Compounds by Sediments, Soils and Suspended Solids — II: Sorbent Evaluation Studies. Water Research 17 (10) :1433-1452. 84. Schroeder, E.D. 1977. Physical-Chemical Removal of Dissolved Materials. In: Water and Wastewater Treatment McGraw-Hill Book Company, New York. pp. 53-72. 85. Miller, S. 1980. Adsorption on Carbon: Theoretical Considerations. Environmental Science and Technology 14 (8) :910-914. 86. Benefield, L.D. Judkins, J.F. and Weand, B.L. 1982. Process Chemistry for Water and Wastewater Treatment Prentice-Hall Publishing, Inc., New York. pp. 191-210. 87. Fritz, W. and Schluender, E.U. 1981. Competitive Adsorption of Two Dissolved Organics onto Activated Carbon — I: Adsorption Equilibria. Chemical Engineering Science 36 : 721-730 88. Myers, A.L., and Prausnitz, J.M. 1965. Thermodynamics of Mixed-Gas Adsorption. American Institute of Chemical Engineering Journal 11(1) ; 121-126 89. Altshuler, G. and Belfort, G. 1983. Selective Adsorption of Organic Homologues onto Activated Carbon from Dilute Aqueous Solutions. In: Treatment of Water by Granular Activated Carbon Edited by McGuire, M.J., and Suffet, I.H. Advances in Chemistry Series No. 202. American Chemical Society, Washington, D.C. pp. 29-61. 90. Miller, S. 1980. Adsorption on Carbon: Solvent Effects on Adsorption. Environmental Science and Technology 14 ( 9 ): 1037-1049 91. Arbuckle, W.B. 1981. Estimating Equilibrium Adsorption of Organic Compounds on Activated Carbon from Aqueous Solution. Environmental Science and Technology 15 ( 7 ): 812-819 92. Manes, M. and Hofer, L.J.E. 1969. Application of the Polanyi Adsorption Potential Theory to Adsorption from Solution on Activated Carbon. Journal of Physical Chemistry 73 ( 3 ): 584-590 93. Manes, M. and Greenbank, M. 1983. Adsorption of Multicomponent Liquids from Water onto Activated

PAGE 286

274 Carbon: Convenient Estimation Methods. In: Treatment of Water by Granular Activated Carbon Edited by McGuire, M.J., and Suffet, I.H. American Chemical Society, Washington, D.C. p. 9-27. 94. Wohleber, D.A. and Manes, M. 1971. Application of the Polanyi Adsorption Potential Theory to Adsorption from Solution on Activated Carbon. II. Adsorption of Partially Miscible Organic Liquids from Water from Solution. Journal of Physical Chemistry 75(1); 61-64 95. Weber, W.J., Jr. 1980. Discussion I: Adsorption Models. In: Activated Carbon Adsorption of Organics from the Aqueous Phase, Volume 1 Edited by Suffet, I.H., and McGuire, M.J. Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan. pp. 117-129. 96. Crittenden, J.C., and Weber, W.J., Jr. 1978. Model for Design of Multi-Component Adsorption Systems. Journal of the Environmental Engineering Division, American Society of Civil Engineers 104 ( EE6 ) : 1175-1195. 97. Belfort, G. 1979. Selective Adsorption of Organic Homologues onto Activated Carbon from Dilute Aqueous Solutions. Solvophobic Interaction Approach and Correlations of Molar Adsorptivity with Physicochemical Parameter. Environmental Science and Technology 13 (8) :939-946. 98. Myers, A.L., and Prausnitz, J.M. 1965. Thermodynamics of Mixed-Gas Adsorption. American Institute of Chemical Engineering Journal 11(1): 121-126 99. Digiano, F.A. Baldauf, G. Frick, B. and Sontheimer, H. 1978. A Simplified Competitive Equilibrium Adsorption Model. Chemical Engineering Science 32(12) :1667-1673. 100. Kidnay, A.J., and Myers, A.L. 1966. A Simplified Method for the Prediction of Multicomponent Adsorption Equilibria from Single Gas Isotherms. American Institute of Chemical Engineering Journal 12(5): 981-986. 101. Famularo, J., Mueller, J.A. and Pannu, A.S. 1980. Prediction of Carbon Column Performance from PureSolute Data. Journal Water Pollution Control Federation 52 ( 7 ): 2019-2032 102. Tien, C. 1983. Recent Advances in the Calculation of Multicomponent Adsorption in Fixed Beds. In:

PAGE 287

275 Treatment of Water by Granular Activated Carbon Edited by McGuire, M.J., and Suffet, I.H. American Chemical Society, Washington, D.C. pp. 167-199. 103. DiGiano, F.A. Baldauf, G. Frick, B. and Sontheimer, H. 1980. Simplifying the Description of Competitive Adsorption for Practical Application in Water Treatment. In: Activated Carbon Adsorption of Organics from the Aqueous Phase, Volume 1 Edited by Suffet, I.H., and McGuire, M.J. Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan. pp. 213-228. 104. Yonge, D.A. 1982. The Effects of Unequal Competition and Irreversible Adsorption on Multi-Component Adsorption Equilibria. Ph.D. Dissertation, Clemson University, Clemson, South Carolina. 105. Curl, R.L., and Keolelan, G.A. 1984. ImplicitAdsorbate Model for Apparent Anomalies with Organic Adsorption on Natural Adsorbents. Environmental Science and Technology 18 (12 ) ; 916-922. 106. Yonge, D.R., Keinath, T.M. Poznanska, K. and Jiang, Z.P. 1985. Single-Solute Irreversible Adsorption on Granular Activated Carbon. Environmental Science and Technology 19 ( 8 ): 690-694 107. Zoltek, J., Jr., McCreary, J.J., Spangler, D.P., Eades, J.L., Jackson, J.A.G., Touchett, R.S., and Sprague, C.W. 1982. Toxic Wastes and Phenolic Groundwater in the Upper Hogtown Creek Region. A Report to the Cabot Corporation Foundation, Inc. Department of Environmental Engineering Sciences, University of Florida, Gainesville. 108. Banerjee, S. 1984. Solubility of Organic Mixtures in Water. Environmental Science and Technology 18(8); 587. 591. 109. Peel, R.G., and Benedek, A. 1980. Attainment of Equilibrium in Activated Carbon Isotherm Studies. Environmental Science and Technology 14 ( 1 ) ; 66-71 110. Davidowicz, A., Waksmundzki, A., and Sokolowski, S. 1977. Gel Permeation Chromatography Using Porous Glass. Investigations of the Influence of Silanization on Exclusion and Adsorption. Separation Science 12(6 ) :573-586.

PAGE 288

BIOGRAPHICAL SKETCH The author was born on December 14, 1948, in Richmond, Virginia. He received a Bachelor of Science degree in June, 1971, from the Department of Chemical Engineering at the Masschusetts Institute of Technology, Cambridge, Massachusetts. He completed the Master of Science in Engineering degree in sanitary engineering in May, 1973, at West Virginia University, Morgantown, West Virginia. He was commissioned a Second Lieutenant in the United States Army Medical Service Corps in August, 1973, with a military occupational specialty of sanitary engineer. He served tours of duty in Maryland, West Germany, and Texas prior to beginning studies at the University of Florida in August, 1982. He currently holds the rank of Major. He is married to the former Sonia Evalina Collazo. Upon degree completion he will commence a three-year tour of duty with the United States Army Pacific Environmental Health and Engineering Agency in Tokyo, Japan. 276

PAGE 289

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Jol^ Prof 1 Zolteki essor oi i \Jr cH^irman r-Environmental Engineering Sciences I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. fseph J. '^elf ino Professor of Environmental Engineering Sciences I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. W. Lamar Miller Professor of Environmental Engineering Sciences

PAGE 290

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. This dissertation was submitted to the Graduate Faculty of the College of Engineering and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. December 1985 Dinesh 0. Shah Professor of Chemical Engineering Benjamin L. Koopman Associate Professor of/^ Environmental Engin^^ering Sciences Dean, College of Engineering Dean, Graduate School


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E1Z1M1ZPC_CUF5SM INGEST_TIME 2015-01-16T17:49:14Z PACKAGE AA00026600_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES