EPA 600/R-07/008 I November 2007 I www.epa.gov/ord
United States
Environmental Protection
Agency
In-situ Regeneration of Granular
Activated Carbon (GAC) Using
Fenton's Reagents
FINAL PROJECT REPORT
Cs
Office of Research and Development
National Risk Management Research Laboratory I Ground Water and Ecosystems Restoration Division
-------
-------
EPA/600/R-07/008
November 2007
In-situ Regeneration of Granular Activated Carbon
(GAC) Using Fenton's Reagents
Final Project Report
Cooperative Agreement with the U.S. Environmental Protection Agency
Principal Investigator: Robert G. Arnold, Ph.D.
Co-Principal Investigator: Wendell P. Ela, Ph.D.
A. Eduardo Saez, Ph.D.
Carla L. De Las Casas, M.S.
Chemical and Environmental Engineering Department
University of Arizona
Tucson, AZ 85721
rga@engr. arizona.edu
520-621 -2410 520-621 -6048 (fax)
Project Officer: Scott Ruling
U.S. Environmental Protection Agency
Office of Research and Development
National Risk Management Research Laboratory
Ground Water and Ecosystems Restoration Division
Ada, OK 74820
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
NATIONAL RISK MANAGEMENT RESEARCH LABORATORY
CINCINNATI, OH 45268
-------
Notice
The U.S. Environmental Protection Agency through its Office of Research and Development funded and managed the
research described herein under Cooperative Agreement No. CR-829505 to the University of Arizona. It has been
subjected to an administrative review but does not necessarily reflect the views of the Agency. No official
endorsement should be inferred. EPA does not endorse the purchase or sale of any commercial products or services.
-------
Foreword
The U.S. Environmental Protection Agency is charged by Congress with protecting the Nation's land, air, and water
resources. Under a mandate of national environmental laws, the Agency strives to formulate and implement actions
leading to a compatible balance between human activities and the ability of natural systems to support and nurture
life. To meet these mandates, EPA's research program is providing data and technical support for solving
environmental problems today and building a science knowledge base necessary to manage our ecological resources
wisely, understand how pollutants affect our health, and prevent or reduce environmental risks in the future.
The National Risk Management Research Laboratory is the Agency's center for investigation of technological and
management approaches for reducing risks from threats to human health and the environment. The focus of the
Laboratory's research program is on methods for the prevention and control of pollution to air, land, water, and
subsurface resources; protection of water quality in public water systems; remediation of contaminated sites and
ground water; and prevention and control of indoor air pollution. The goal of this research effort is to catalyze
development and implementation of innovative, cost-effective environmental technologies; develop scientific and
engineering information needed by EPA to support regulatory and policy decisions; and provide technical support and
information transfer to ensure effective implementation of environmental regulations and strategies.
Stephen G. Schmelling, Director/
Ground Water and Ecosystems pestpfation Division
National Risk Management Research Laboratory
in
-------
-------
Table of Contents
LIST OF FIGURES x
LIST OF TABLES xiii
EXECUTIVE SUMMARY El
BACKGROUND 1
Fenton's Mechanism 3
Copper as a Fenton Metal 5
Fenton-Driven GAC Regeneration 6
Contaminant Removal/Destruction Mechanism 6
Theoretical Considerations- Fenton's Treatment for GAC Regeneration 10
PROJECT OBJECTIVES 12
MATERIALS AND METHODS 13
Chemicals 13
Analytical 13
Target Organic Compounds 13
Hydrogen Peroxide 14
pH 14
Iron 14
Quality Assurance 14
Experimental 15
General 15
Batch Kinetic Experiments 15
-------
Adsorption Isotherms 15
Column Experiments 16
GAC Selection and Preparation 16
Iron-amended GAC Preparation 17
Extraction of VOCs from GAC 18
Sample Analysis 19
Desorption Rate Experiments 20
Regeneration Rate Experiments 20
Fixed-Bed Desorption Experiments 21
Field-Scale Column Experiments 22
RESULTS AND DISCUSSION 24
Homogeneous Experiments 24
Dependence of PCE Reduction Kinetics on Total Iron 24
Rate Enhancement via Hydroxylamine Addition 25
Quinone Addition 26
Copper Effects 31
Mechanism of Rate Enhancement by Copper 34
Temperature Effects 34
Homogeneous Model Formulation 35
Simplified Fenton Mechanism 35
Complications in the Actual Fenton Reactions 35
Complexation with H2O2 and Complex Disproportionation - Implication for pH Effects
on Rate of Reaction 36
Model Limitations 37
Inorganic Radical Formation 38
Chloride, Perchlorate, Sulfate and Nitrate 38
Carbonate (CCv) and Carboxyl (CCv) Radicals 40
Radical Formation Kinetics and Reactivity 41
VI
-------
Model Applications 42
Simulation of HXO2 Decomposition 42
Effect of pH on the Observed Rate Constant for Decomposition of H2O2 by Fe(III) 43
Non-Halogenated Organics Degradation in Homogeneous Systems 43
PCE Destruction Kinetics Considering Cl Effects 44
Simulation of Fenton's Reaction for PCE Degradation 45
pH Effect 45
Cl Effect 47
Complexation of Cl with Fe(III) 48
Ionic Strength 49
Temperature Effects 49
Role of Superoxide Radical (O2-) in Fenton's Reaction 50
Background and Reactions Involving O2- 50
Solvent Effects on O2- Reactivity in Aqueous Solutions 51
Fenton-Driven Transformation of PCE 52
Inhibition of PCE Oxidation by Cl 55
CT Transformation in the Presence of 2-propanol 55
Homogeneous, Bench-Scale Experiements - Summary and Conclusions 59
Bench-scale, Heterogeneous, Column Experiments 61
Adsorption Isotherms 61
Bench-Scale Column Experiments 62
Recovery Using Fenton's Reagents 62
Clean Water and Fenton-Driven Recovery Experiments 65
Fixed-Bed Recovery Trials - Effect of Particle Size 68
Theoretical Considerations - Modeling 69
Pore Diffusion Modeling 73
Pore and Surface Diffusion Modeling 76
vn
-------
Role of Iron Phase (Precipitated vs. Dissolved) 78
Heterogeneous, Bench-Scale Experiments - Conclusions 80
Field-Scale Regeneration Trials 82
Equipment Testing - Methylene Chloride and Chloroform Recovery Tests 82
Sequential Adsorption/Regeneration Experiments 83
Loading Carbon with SVE Gases 85
Economic Analysis 87
Cost Estimation Based on Iron-Amended GAC Regeneration 88
GAC Economic Analysis 89
Cost of H2O2 Consumption in Fenton's System 91
Engineering Considerations 92
Hydrogen Peroxide Stability 92
The Heat of H2O2 Decomposition 93
Oxygen Formation via Fenton's Reactions 95
Field-Scale Regeneration Trials Summary 97
SUMMARY AND RECOMMENDATIONS FOR ADDITIONAL STUDY 99
General Observations 99
Homogeneous, Bench-Scale Experiments 100
Heterogeneous, Bench-Scale Experiments 101
Field-Scale Regeneration Trials 102
Data Quality and Limitations 103
Research Recommendations 104
REFERENCES 106
APPENDICES 113
Appendix A - Computer Simulation of Homogeneous System 114
Appendix A.I - Variable's Identificaction and Significance 114
Appendix A.2 - Fortran Code for Non-Chlorinated Compounds 116
Vlll
-------
Appendix A.3 - Fortran Code for Chlorinated Compounds 122
Appendix B - GAC Cost Economic Analysis 129
IX
-------
List of Figures
Figure 1. Cross-section of the Park-Euclid Arizona state Superfund site 3
Figure 2. Potential sources of rate limitation for adsorbate desorption 7
Figure 3. GAC particle cross section 8
Figure 4. Intraparticle porous transport of hydroxyl radical to GAC particle surface 10
Figure 5. Pore and surface diffusion, and surface desorption of intraparticle contaminant
within a GAC particle pore 11
Figure 6. Expanded bed desorber set up 18
Figure 7. Fixed bed desorber set up with mechanical vacuum pump 19
Figure 8. SVE-GAC system at field site 21
Figure 9. Field-site diagram 22
Figure 10. Effect of iron concentration on PCE disappearance 23
Figure 11. PCE degradation with hydroxylamine as a reducing agent 24
Figure 12. Simplified quinone mechanism 25
Figure 13. Structures of quinones investigated 26
Figure 14. PCE degradation as a function of 1,4-hydroquinone added 26
Figure 15. First order rate constants for different hydroquinones concentrations 27
Figure 16. Two-phase hydroquinone addition experiment 28
Figure 17. Effect of benzoquinone on the PCE degradation 29
Figure 18. Benzoquinone effect on the kobs for PCE disappearance 30
Figure 19. Effect of 0.3 mM BQ addition on the (log transformed) PCE concentration 30
-------
Figure 20. Effect of 0.5 mM AQDS addition to PCE transformation 30
Figure 21. Effect of Copper addition on the rate of PCE destruction 31
Figure 22 (a)(b). Copper on the PCE-pseudo-first-order rate constant 32
Figure 23 (a)(b)(c). Arrhenius plots for copper-free and fixed Cu:Fe ratio (8:1) cases 33
Figure 24. Simulation of the effect of pH on the k0bs 35
Figure 25. Simulation for the oxidation of formic acid 42
Figure 26. PCE degradation and model fits using various kCi,oH 43
Figure 27. Effect of pH on the PCE degradation rate in Fenton's reaction 44
Figure 28. Effect of chloride ion concentration on the PCE degradation 45
Figure 29. Effect of light and isopropanol on carbon tetrachloride transformation 46
Figure 30. Effect of isopropanol on the degradation of CT 47
Figure 31. Effect of benzoquinone on CT degradation 53
Figure 32. Effect of Cl and IP in PCE transformation 54
Figure 33. Effect of Cl on PCE transformation with IP in solution 56
Figure 34. Isotherm data for methylene chloride on URV-MOD 1 57
Figure 35. PCE extraction efficiencies in four different solvents 58
Figure 36. Removal of MC, CF and TCE from GAC 61
Figure 37. Semi-log plot of the data in Figure 36 62
Figure 38. Correlation between kobs and compound-specific (l/K)n 63
Figure 39. Recovery using eluant solutions with and without Fenton's reagents 65
Figure 40. Liquid phase reservoir concentrations for CF-loaded GAC recovery 66
Figure 41. 14-hour carbon recovery profiles for CH2C12 and TCE 66
Figure 42. 14-hour carbon recovery profile for CH2C12 and TCA 67
Figure 43. Fixed-bed carbon recovery kinetics for three discrete size distributions 68
Figure 44. GAC particle cross section 69
Figure 45. Bulk aqueous-phase concentration profiles for CH2C12 71
XI
-------
Figure 46. Bulk aqueous phase contaminant concentrations between experimental TCA data
and a pore diffusion model 74
Figure 47. Log-transformed isotherm data for CH2C12 (a) and TCA (b) 75
Figure 48. Bulk aqueous-phase contaminant concentration comparison between experimental CH2C12 data
and a pore and surface diffusion model 75
Figure 49. A comparison of bulk aqueous-phase contaminant concentrations between experimental TCA data
and a pore and surface diffusion model 76
Figure 50(a)(b). TCE Recovery with background GAC, iron amended GAC and iron in solution 77
Figure 51. pH effect on iron-amended GAC loaded with PCE 79
Figure 52. Carbon regeneration for MC and CF in the field trials 80
Figure 53. TCE carbon recovery during three sequential regeneration phases 82
Figure 54. Ratio of AQ (Ceq - Ciiq) to Ceq, where Ceq is the aqueous-phase TCE concentration in equilibrium
with the residual adsorbed concentration (q) and Ciiq is the measured, liquid phase concentration 83
Figure 55. Carbon regeneration for SVE-loaded GAC 84
Figure 56. Fractional saturation of GAC before regeneration 85
Figure 57. Field regeneration for TCE and PCE 86
xn
-------
List of Tables
Table I. Hydroxyl Radical (»OH) Reactivity with Organic Compounds 2
Table II. Chemical Properties of the Organic Compounds Studied 16
Table III. Physical Properties of Calgon URV-MOD 1 17
Table IV. Extraction Efficiencies 18
Table V. First-Order Rate Constants for PCE Disappearance as a Function of the Initial
Hydroquinone Concentration 28
Table VI. Reaction Mechanism for Fe(III)-Catalyzed Decomposition of H2O2 (25° C; /= 0.1M) 36
Table VII. Apparent First-Order Rate Constant for the Decomposition of H2O2 (k0bs) 39
Table VIII. Measured Pseudo-First-Order Kinetic Constants (kobs) for the Initial Rate
of Decomposition of H2O2- Sulfate effect 39
Table IX. Measured Pseudo-First-Order Kinetic Constants (kobs) for the Initial Rate
of Decomposition of H2O2- Chloride Effect 39
Table X. Carboxyl Radical Anion, CO2-, Properties 40
Table XI. Rate Constants for Potential Hydroxyl Radical Sinks 42
Table XII. Additional Second Order Reaction Rate Constants for Organic Targets with -OH 44
Table XIII. Calculated Effect of Chloride and Perchlorate Ion Concentrations on Rate of Hydrogen Peroxide
Degradation in Fenton's Reaction 48
Table XIV. Equilibrium Constants as a Function of Ionic Strength 49
Table XV. Equilibrium Constants as a Function of Temperature 50
Table XVI. Standard Redox Potentials 50
Table XVII. Carbon-Chlorine Bond Dissociation Energies for Chlorinated Compounds 51
Xlll
-------
Table XVIII. Summary Table for Rates Observed (k0bs) at 24° C and 32° C, and Cost Estimates
at the Bench and Field Scales 64
Table XIX. Expanded-Bed Recovery Data 67
Table XX. Characteristics of GAC and Results after 14 hours of Iron-Amended, Regeneration Trials 79
Table XXI. Observed Degradation Rate Constant and Average Hydrogen Peroxide Use
for PCE-Laden GAC Regeneration With and Without Iron-Amended to the Carbon 80
Table XXII. Comparison of GAC Replacement vs. Regeneration Costs 88
Table XXIII. Cost Evaluation Based on Bench-Scale Results Using TCE-Loaded and Iron-Amended GAC 89
Table XXIV. Cost Estimates Comparing Hazardous Waste Disposal of Spent CAG (#1) and Virgin Carbon
Replacement Versus Fenton's Reagent Regeneration of GAC (#2) 91
Table XXV. Summary of Estimated H2O2 Cost Contribution to Total Cost of Fenton's Regeneration 91
Table XXVI. Summary of Efficiency Results for Fenton's Reagent Regeneration of GAC
in Bench and Field Trials 97
xiv
-------
Executive Summary
Fenton-dependent recovery of carbon initially saturated with one of several chlorinated aliphatic contaminants
was studied in batch and continuous-flow reactors. A specialty carbon, URV-MOD 1 (Calgon) was employed to
minimize non-productive H2O2 demand - that which does not yield hydroxyl or superoxide radicals. Because the
reductive reaction of Fe(III) to Fe(II) limits the overall rate of radical generation via Fenton's mechanism, it was
hypothesized that steps designed to increase the rate of ferrous iron generation would accelerate Fenton-dependent
contaminant destruction, enhancing carbon recovery kinetics. Homogeneous-phase experiments were designed to
establish the effects on PCE destruction kinetics of total iron concentration and additions of NH2OH, various
quinones or copper to the Fenton mixtures.
Up to almost the solubility limit of Fe(III), the pseudo-first-order rate constant for PCE disappearance was nearly
proportional to the mass of iron added. At Fe(III)T = 2.0 mM, the half time for PCE disappearance was 19 minutes.
Unfortunately, the limited solubility of ferric iron, even at pH 2.0, prevents further rate enhancement via this method.
As expected, the addition of hydroxylamine initially accelerated the destruction of PCE many fold, so that the half
time for PCE disappearance was a few minutes. The effect of NH2OH addition was rapidly lost, however, suggesting
that hydroxylamine, initially provided at 0.01 M was approaching exhaustion after about 20 minutes. A second
addition of 0.01 NH2OH did not promote PCE transformation to the same degree. The observation is consistent with
the hypothesis that time-dependent reduction in the specific rate of PCE loss was due to temporary accumulation of
partial oxidation products derived from the initial reactions involving PCE.
Quinones are known to be electron shuttles in many environmental systems and, as such, were investigated as
possible means to accelerate the iron reduction reaction in Fenton's system. Quinone addition increased the pseudo-
first-order rate constant for PCE loss approximately three-fold. Above a 1,4-hydroquinone concentration of 0.2 mM,
further addition of the compound had little effect on PCE loss. Unfortunately hydroquinone appears to have been
quickly destroyed in mixtures containing iron and H2O2. After an hour, the initial, hydroquinone-dependent increase
in the specific rate of PCE disappearance was completely lost. Reactors amended with 1,4-benzoquinone or 9,10-
anthraquinone-2,6-disulfonic acid as electron shuttles provided similar results.
Only copper addition to the Fenton mixtures sustainably enhanced the specific rate of PCE loss. At a copper-to-
iron ratio of 2 moles per mole, copper addition increased the pseudo-first-order rate constant for PCE transformation
El
-------
by a factor of 4.3. It is apparent that the effect of copper addition on Fenton-dependent reaction rates is complex and
involves a shift in chemical mechanism. The slopes of Arrhenius plots in copper-free and copper-amended solutions
were significantly different suggesting that the overall rate limitation for PCE disappearance is derived from different
steps in these two cases.
The effects of pH, chloride ion and other hydroxyl radical scavengers on the rate of Fenton-dependent compound
degradation were evaluated. The PCE decomposition rate increased with pH in the range 0.9-3.0. PCE mineralization
yields chloride ions, which accumulate in solution and inhibit the PCE degradation rate by scavenging «OH radicals.
The PCE degradation rate decreased with increasing Cl" concentration in the range of 0 to 0.058 M. Isopropanol, a
known «OH radical scavenger, enhanced CT degradation in Fenton's reaction but inhibited PCE decomposition. The
nature of the pathway for CT destruction was indicated by the presence of phosgene, CO2 and chloride, which
suggests that the superoxide radical (O2*~), not *OH, is the species responsible for CT degradation (Smith et al., 2004).
A parallel experimental program was carried out to examine the sources of rate limitation when halogenated
contaminants were initially adsorbed to activated carbon. Possibilities included rates of reaction between hydroxyl
radicals and either the adsorbed or (bulk) aqueous-phase contaminant, the rate of contaminant desorption, the rate of
diffusive transport of contaminants from the carbon interior, or some combination of these.
The feasibility of using Fenton's reagents for in-place recovery of spent granular activated carbon (GAC) was
investigated in continuous-flow experiments. Fenton's reagents were cycled through spent GAC to degrade sorbed
chlorinated hydrocarbons. Little carbon adsorption capacity was lost in the process. Seven chlorinated compounds
were tested to determine compound-specific effectiveness for GAC regeneration. The contaminant with the weakest
adsorption characteristics, methylene chloride, was 99% lost from the carbon surface during a 14-hour regeneration
period. Results suggest that intraparticle mass transport generally limits carbon recovery kinetics, as opposed to the
rate of oxidation of the target contaminants.
Mathematical models were developed to optimize Fenton-driven degradation of organic compounds in solution or
adsorbed to GAC. Models can evaluate the effect of operational parameters ([Fe(III)]T:[H2O2]0 ratio, pH) on
degradation kinetics. Computer modeling efforts were divided into homogeneous and heterogeneous simulations. The
first model simulated experimental degradation of the organic target in a homogeneous Fenton-reaction system. This
model was based on the system of reactions proposed by De Laat and Gallard (1999). The model was tested with
experimental data for H2O2 disappearance and PCE degradation (pH 2), and verified against published data using non-
halogenated organic targets (i.e. HCOOH). Although the model simulated the results well for simple Fenton's
systems, it requires further refinement to closely simulate Fenton's systems in which reaction by-products (e.g.,
chloride, partially oxidized organics) play a significant role.
The second model simulated experimental GAC recovery, testing the hypothesis that intraparticle diffusion
governs overall recovery kinetics. Analytical solutions were developed for mass transport-limited GAC recovery rates
when contaminant adsorption is governed by a linear isotherm. A single fitting parameter (tortuosity) brought
E2
-------
simulations and data into reasonable agreement. However, a common tortuosity could not be obtained for all
compounds, suggesting that desorption effects can also limit GAC recovery kinetics. More work is recommended in
this area.
A pilot-scale reactor was designed for use in a field demonstration. The field site is among those on the state of
Arizona Superfund list - the Park-Euclid site. Primary contaminants at that site include trichloroethylene,
perchloroethylene and a mixture of volatile and semi-volatile hydrocarbons. Gases brought to land surface via soil
vapor extraction (SVE) are treated via carbon adsorption and returned to the SVE collection system. In our study,
carbon was regenerated periodically via Fenton's reaction and returned to service in order to establish process
feasibility under field conditions. In the field, up to 95% of the sorbed TCE was removed from GAC during
regeneration periods of 50-60 hours. Recovery of PCE was significantly slower. Although the process, as employed,
was not cost-effective relative to thermal regeneration or carbon replacement, straightforward design and operational
changes are likely to lower process costs significantly.
The field trials supported the bench-scale trial conclusions that the rate of GAC regeneration is compound
specific. For the most soluble VOCs, with modest to low solid partitioning, the bulk radical reaction rate with the
target compound can control the GAC recovery rate. For these cases, Fenton's based regeneration is a very attractive
treatment option, from both economic and ease-of implementation perspectives. Less soluble, more reactive
compounds like TCE are limited by desorption or intraparticle transport. The least soluble and most strongly
partitioning compounds (e.g., PCE) are likely limited by the desorption reaction rate.
The field trials indicate that there is minimal loss of carbon adsorption capacity after Fenton-driven regeneration.
The largest driver of Fenton system (operation and maintenance) costs is hydrogen peroxide (H2O2) usage. Its
utilization rate can be optimized by: (i) using an optimal H2O2/iron concentration ratio, that which generates an
optimum [«OH]SS without scavenging most of the radicals by reaction with H2O2 itself, (ii) reducing the size of the
reservoir, (iii) pulsed addition based on bulk VOC levels or injection of H2O2 before the GAC column for compounds
that desorb rapidly, and (iv) employing iron-amended GAC, which can generate radicals near the surface of the
carbon, thus reducing the use of H2O2. Activated carbons with more macroporous structures (and more frequent,
shorter regeneration periods) may provide a means to reduce mass transfer limitations and increase overall system
efficiency. Scoping-level cost estimates indicated that field use of Fenton regeneration is not cost effective without
optimization and/or iron surface amendments, except in the case of the most soluble VOCs.
E3
-------
-------
Background
Ten of the 25 most frequently detected hazardous contaminants at National Priority List (Superfund) sites are
chlorinated volatile organic compounds (VOCs) (NRC, 1994). VOCs are commonly recovered from contaminated
groundwater or soils using pump-and-treat, air stripping, and soil vapor extraction (SVE) methods. After the
contaminated fluid is extracted from the subsurface, granular activated carbon (GAC) adsorption is often used to
separate VOCs from liquid and gas streams derived from these recovery techniques. When the carbon is loaded to
capacity, it must be regenerated or replaced. Advanced oxidation processes (AOPs) also can be employed for the
remediation of the effluent of these fluid extraction processes (Kommineni et al., 2003; Prousek, 1995; Zoh and
Stenstrom, 2002). AOPs can destroy VOC contaminants, leaving only mineralized products, but AOPs are relatively
expensive for treatment of low-concentration pollutants.
In this project we explore the feasibility of using Fenton's reaction for regeneration of spent GAC that has been
used to collect and concentrate VOCs. Fenton's reaction is an AOP process in which reaction of hydrogen peroxide
(H2O2) with iron (Fe), generates two radical species (*OH, and HO2«/O2«) A very broad range of organics, including
a variety of prominent contaminants, are oxidized by hydroxyl radicals, which are among the strongest and least
specific oxidants known (Table I; Gallard and De Laat, 2000; Ruling et al., 2000a; Duesterberg et al., 2005; Chen et
al., 2001). Fenton-dependent processes can mineralize even heavily halogenated targets such as PCE and TCE (Teel
et al., 2001). Rate limitations, potential rate acceleration strategies and process feasibility of Fenton-dependent
regeneration of VOC-loaded GAC were investigated in a series of bench-scale experiments. Field-scale process
feasibility was also examined at an Arizona State Superfund site, where GAC is being used to separate chlorinated
VOCs and volatile hydrocarbons from an SVE gas stream. Carbon recoveries at bench and field scales were compared
and evaluated.
The field site selected was the Park-Euclid (Tucson, Arizona) State Superfund site, where the primary vadose
zone contaminants are PCE, TCE, dichloroethene isomers, and volatile components of diesel fuel. A SVE system with
GAC treatment of the off-gas was installed at the site as an interim remediation scheme while the state Remedial
Investigation at the Park-Euclid site was underway. There are four distinct zones of contamination ~ the upper
vadose zone, perched aquifer, lower vadose zone, and regional aquifer (Figure 1). Both the upper and lower vadose
zones contain dry-cleaning-related contaminants (i.e. PCE, TCE, DCE isomers). The Park-Euclid SVE system draws
-------
Table 1. Hydroxyl Radical ('OH) Reactivity with Organic Compounds
Target Compound
Trichloroethylene
Tetrachloroethylene
1,1 -Dichloroethane
1,1,1 -Trichloroethane
1,1,2-Trichloroethane
Chloroform
Methylene Chloride
Carbon Tetrachloride
1,4-Benzoquinone
Isopropanol
Atrazine
Formic Acid
H202
2.90E+09
(3.3-4.3)E+09b
2.00E+09
7.90E+08
l.OOE+08
3.00E+08C
5.00E+07C
9.00E+07C
-------
Ground Surface '" SVE system
Upper Vadose Zone
Perched Aquifer
Upper Aquitard
-115'
Lower Vadose Zone
-200'
Regional Aquifer
Figure 1. Cross-section of the Park-Euclid Arizona state Superfund site. PCE and TCE contamination is observed in the regional and perched
aquifers and in both the lower and upper vadose zones. SVE gases are extracted from the upper vadose zone.
Fenton's Mechanism
In Fenton's mechanism, iron cycles between the Fe(II) and Fe(III) oxidation states due to reaction with H2O2.
The oxidation of Fe(II) produces a highly aggressive hydroxyl radical (*OH) that is relied upon to attack organic
environmental contaminants (Ruling et al., 2000a; Teel et al., 2000). Reduction of Fe(III) to Fe(II) limits the overall
rate of radical production under most circumstances (Chen and Pignatello, 1997; De Laat and Gallard, 1999; Teruya,
2000). Thus, iron speciation is a strong determinant of Fenton's kinetics (Gallard et al., 1999; De Laat and Gallard,
1999). The distribution of Fe(III) among free ferric ion and hydroxylated forms (predominantly Fe3+, FeOH2+,
Fe(OH)2+, Fe2(OH)24+) depends primarily on solution pH. Furthermore, the insolubility of Fe(OH)3 (s) (pKso for
Fe(OH)3 (s, amorph) = 38.7; Stumm and Morgan, 1981) is expected to limit total iron solubility at all but very low pH
values. At pH < 2, the free ferric ion is the predominant Fe(III) species. Changes in iron speciation in the low pH
range, however, can account for the dependence of Fenton reaction kinetics on pH. Fe(III)-hydroperoxyl
complexation reactions are fast, and equilibrium conditions are generally satisfied on the timescale of Fenton
applications. Complexation with peroxide ion precedes Fe(III) reductions.
The following mechanism for organic contaminant destruction is simplified from a more detailed Fenton's
scheme proposed by De Laat and Gallard (1999):
Initiation reactions:
Fe(III) + H2O2 -> Fe(II) + HO2- + FT (1)
Fe(II) + H2O2 -» Fe(III) + -OH + OH (2)
•OH + R -» OH + -R (3)
-------
Chain propagation:
•R + Fe(III) -» R+ + Fe(II) (4)
(where Fe (II) reacts again with H2O2 to yield another hydroxyl radical)
The organic cation is consumed through a highly exothermic reaction as follows:
R+ + H2O -> ROH + FT (5)
The dominant chain termination steps are:
•R + -R -> R-R (6)
•R + Fe(II) -» Fe(III) + R (7)
The overall rate is limited normally by the rate of reduction of Fe(III) to Fe(II) by H2O2 (reaction 1). The
hydroxyl radical produced from reaction 2 can oxidize the target organic compound (R), producing an organic radical
(•R). These organic radicals can reduce Fe(III) to Fe(II), propagating the chain reaction. The chain reaction is
terminated by radical dimerization, as in reaction 6, or by reaction of the organic radical with Fe(II). The same
radical terminating reactions have been used to explain various inhibition effects (Walling and Kato, 1971).
In homogeneous systems, contaminant reaction kinetics follow the second-order rate equation,
at
where k.0H,R is the compound-specific reaction rate constant for reaction of the target compound with «OH. It is
assumed that upon initiation of the Fenton's reaction, the concentration of »OH rises quickly and stabilizes at a near-
steady concentration within a very short period of time (seconds or less). A steady-state approximation for «OH
concentration is frequently assumed for kinetic analysis so that pseudo-first-order decay kinetics are often the basis of
kinetic analyses of contaminant destruction:
dt
where k ' is defined as
(10)
Pseudo-first-order kinetics are only expected, however, if the hydroxyl radical concentration is essentially
constant. Furthermore, it is expected that [»OH]SS will depend on total iron, H2O2 concentration, pH, and temperature.
At a constant total iron concentration, the near-steady «OH concentration may be fairly insensitive to H2O2
concentration (Ruling et al., 2000a). If H2O2 concentration is low, for example, the Fenton's mechanism slowed,
which limits the rate of production of radicals. However, the overall rate of *OH scavenging may also be relatively
low. If, on the other hand, excess H2O2 is present, it acts as both a source of radicals and a scavenger for «OH.
-------
•OH + H2O2 -> HO2- + H2O (11)
However, it has been shown that because H2O2 both generates and consumes «OH, the steady concentration of
•OH, and thus the pseudo-first-order rate constant for target destruction, are not strongly influenced by H2O2
concentration over a broad range of conditions. Therefore, small fluctuations in H2O2 have little effect on the
observed pseudo-first-order rate constant for contaminant destruction via oxidation with «OH in homogeneous Fenton
systems. Although not explicitly studied in this or previously published work, the near-steady superoxide radical
concentration is expected to be insensitive to small fluctuations in H2O2 concentration (analogous to *OH radical as
discussed previously).
Copper as a Fenton Metal
There is considerable debate regarding the status of copper as a Fenton metal. Although copper cycles between
the Cu(I) and Cu(II) oxidation states via reaction with H2O2, hydroxyl radicals are not generally produced (Masarwa
et al., 1988). Nevertheless, the reaction of the cuprous ion and hydrogen peroxide (reaction 12) results in the
formation of a copper complex, (H2O)m Cu *O2H, that may react with organics present in solution. In acidic solution
and in the absence of organics, the copper complex decomposes into free cupric ion and hydroxyl radicals (reaction
13) (Masarwa etal., 1988).
Cu+ + H2O2 -> Cu+-O2H + H+ (12)
(H2O)Cu+'O2H -» Cu2+ + -OH + 2OH (13)
Cupric copper can also react with organic radicals (reaction 14). The Cu(I) product of reaction 14 can either react
with Fe(III) to regenerate Fe(II) or with H2O2 to produce cuprous ions (Walling and Kato, 1971). Since an array of
unidentified organic radicals may be produced during Fenton-driven decomposition of target contaminants,
particularly in the presence of non-target organics, this pathway may be of practical importance in environmental
applications. The reaction of Cu(II) with R« is comparable to Fe(III) reduction (reaction 4). The reduction of Fe(III)
by Cu(I) (reaction 15) and subsequent reaction of Fe(II) with H2O2 would increase the overall rate of «OH generation
(Walling and Kato, 1971).
R- + Cu(II) -> Cu (I) + products (14)
Cu(I) + Fe(III) <-> Fe(II) + Cu(II) (15)
Walling and Kato (1971) indicated that iron undergoes electron-transfer during reaction with R«, whereas the
copper reaction can involve either complexation (reaction 16) or reduction via formation of an organo-copper
intermediate (reactions 16 and 17).
R- +Cu2+(H2O)n -> RCu2+(H2O)n (16)
RCu2+(H2O)n -> ROH + Cu+(H2O)n.! + H+ (17)
-------
Furthermore, they found evidence that organic radicals with strong electron withdrawing groups (e.g. Cl") are
preferentially reduced by ferrous iron. Trichloroethene radical, for example, would satisfy requirements for such a
reaction.
Cu(II) is a weaker oxidizing agent than Fe(III), and therefore less capable of oxidizing R« (as in reaction 4) by
outer-sphere electron transfer. Instead, the reaction involves formation of an organo-copper intermediate (Walling,
1975). Comparison of oxidation rates attributable to Cu(II) and Fe(III) indicated that Cu(II) oxidations were slower,
but that ligand-Cu(II) exchanges were fast (Walling, 1975).
Cu(II) and Cu(s) are thermodynamically favored over Cu(I) as shown by the following reactions (Holleman and
Wiberg, 2001; Cotton et al., 1999):
Cu+ + e- = Cu(s) E° = 0.52V (18)
Cu+ = Cu2+ + e" EH° =-0.16V (19)
2Cu+ = Cu (s) + Cu2+ AE° = 0.36 V (20)
At equilibrium, only low concentrations of Cu(I) (<10~2 M) can exist in aqueous solutions (Cotton et al., 1999).
Stability differences arise in part because the energy of hydration of Cu2+ is much higher than that of Cu+ (2100 vs.
582 KJ/mol), all of which helps to explain why Cu(I) does not normally exist in aqueous solution, but rather
disproportionates to Cu and Cu(II) (Heslop and Robinson, 1967):
6
106 (21)
Cu*
However, Cu2+ is favored with anions unable to make covalent bonds or bridging groups (e.g. SO42") (Cotton et
al., 1999). Copper participates in outer sphere electron transfers as opposed to inner sphere electron transfers in which
the transfer of electrons occurs through a chemical bridge. The ligand covalently links the two metal redox centers
and typically has more than one lone electron pair, serving as an electron donor to both the reductant and the oxidant
(www.wikipedia.com). The coordination chemistry of the Cu2+ ion is dominated by nitrogen- and oxygen-donating
ligands followed by chloride and sulfur-containing groups (King, RB. 2006).
Fenton-Driven GAC Regeneration
Contaminant Removal and Destruction Mechanisms
Removal of adsorbates from GAC may be kinetically limited by any of four distinct steps: desorption from the
GAC surface, "1"; pore and surface diffusion, "2"; film transport, "3", in which the film thickness (8) is inversely
related to the local bulk velocity; and removal of reactant from bulk aqueous phase, "4", due to reaction and advective
transport. In most cases, intraparticle effects (pore diffusion-"2" and/or desorption-" 1") control the observed rate of
transfer from the particle to the bulk aqueous phase (Crittenden et al., 1987).
-------
Figure 2 illustrates the possible sources of rate limitation experienced by a desorbing species. Partitioning
between the solid and liquid within a pore is governed by the surface desorption rate or equilibrium condition "1".
Once the contaminant is in the liquid within a pore, intraparticle transport "2" is largely governed by molecular
diffusion. At the surface of the particle, transport into the bulk aqueous phase can be limited by molecular diffusion
across a mass transfer boundary layer. Removal from the bulk aqueous phase relies on a mixture of convection and (in
the presence of Fenton's reagents) reaction.
Because hydroxyl radicals are extremely reactive and short-lived, they do not persist any significant distance
beyond the point where they are generated. Thus, if the Fenton's reagents are primarily present in the bulk phase (as
opposed to in the pore volume of the GAC), it is reasonable to hypothesize that the reaction of the target compound
with hydroxyl radicals will occur mainly in the bulk aqueous phase. This physical model represents a hypothesis to be
tested using experimental carbon recovery data.
1 2 3
Solid Pore Film
/&&
•
Figure 2. Potential sources of rate limitation for desorption of adsorbate from the GAC (solid) to the bulk fluid. 1 .Desorption from solid to
liquid phase. 2.Diffusive transport within the pores (pore or surface diffusion). 3.Diffusive transport through a quiescent film surrounding the
particle. 4.Convective transport or reaction in bulk fluid.
In the general case, it is possible that carbon recovery kinetics are limited by either the rate of reaction of target
contaminants with Fenton-dependent radicals, by the rate of desorption of contaminants from the particle surface or
by a combination of pore and surface diffusion. This physical description provides a starting point for model
development and data analysis.
To provide a foundation for transport characterization, the physical processes of chemical adsorption will be
briefly covered. Convection along the column's axial direction and axial dispersion are the mass transport
mechanisms in the bulk aqueous phase (Ma et al., 1996). If these processes are rapid compared to other transport
steps, the bulk aqueous phase can be modeled as well mixed. Molecules from the bulk are transported across a
boundary layer. The thickness of this hydrodynamic boundary layer is reduced when fast convective mixing occurs in
-------
the bulk phase; the resistance to overall diffusive transport is often neglected in fast flowing, expanded bed columns
(Noll et al., 1992). Upon reaching the adsorbent surface, molecules diffuse into or out of the interior of the particle
through an intricate porous network (Figure 3).
Solid
Pore
Bulk
contaminant
Figure 3. GAC particle cross-section. A tortuous path is experienced by desorbing particles on the particle surface and in the intra-particle
aqueous phase.
Intraparticle diffusion describes the mass transport of adsorbate molecules within GAC, and consists of pore-
volume and surface diffusion. Pore -volume diffusion, which characterizes contaminant transport through the porous,
fluid-filled void, is normally expressed in terms of an effective diffusion coefficient (Defip) that is lower than the
molecular diffusivity (Dmoi). This arises because the diffusive path experienced by the adsorbate during radial
transport can be exceptionally tortuous (Figure 3). In the absence of other diffusive mechanisms, an inverse
relationship exists between Defp and pore space tortuosity, T,
(22)
where Dmot is the contaminant molecular diffusivity (cm2/s), and Defp is the effective pore-volume diffusivity (cm2/s).
Conversely, a direct relationship exists between particle porosity and the Defip. The effective pore-volume diffusion
coefficient approaches the liquid-phase molecular diffusion coefficient as porosity approaches 1.0 (Furuya et al.,
1996).
Surface diffusion describes concentration-driven contaminant transport on the particle surface. The mechanism is
analogous to pore-volume diffusion but occurs on the internal surface of the particle. Due to an aqueous concentration
gradient within the pores, a surface gradient will also exist on the pore surface in the same direction as the aqueous-
phase gradient (Yang, 1987). Such a (surface) gradient is assured if adsorption/desorption reactions on the
-------
particle/pore wall are relatively fast. Surface diffusion occurs by molecular hopping between adsorption sites. Thus,
surface diffusion depends on the ability of molecules to overcome an energy barrier. Molecules can also desorb to the
intra-particle fluid by overcoming a similar desorptive energy barrier. A simplifying assumption that is often made for
modeling purposes is that adsorption and desorption processes are much faster than diffusive transport (Crittenden et
al., 1986). Under these circumstances, local equilibrium is assumed to exist between the local sorbed and intraparticle
aqueous contaminant concentrations, and the interfacial flux can be characterized by an adsorption isotherm.
As for pore-volume diffusion, calculating surface diffusion flux in GAC requires estimation of a tortuosity factor.
Although the tortuosity factor for surface diffusion is not necessarily equal to that of porous contaminant transport
within the same structure, the same T is usually assumed for both diffusion modes (Yang, 1987). The relative
contributions of surface and pore diffusion to the overall rate of intraparticle transport are a function of adsorbate
affinity for the surface as well as other factors, making empirical determination of an overall diffusion constant a
necessity. Several researchers have established that surface diffusion is also related to the solid phase (adsorbed)
concentration and temperature (Furuya et al., 1987; Sudo et al., 1978; Suzuki and Fujii, 1982). Consequently,
determination of accurate kinetic and equilibrium data across the gamut of experimental conditions is generally
required for adequate surface diffusion characterization.
For design, surface and pore diffusion are often combined into a single term, the apparent diffusivity (D), and an
overall mass transfer rate is established from recovery data. General conclusions can be deduced from such an
analysis - mainly the relative importance of individual transport mechanisms. However, mechanism-specific transport
rates cannot be established. To verify the importance of diffusive transport to overall recovery kinetics in the Fenton-
based treatment scheme considered here, the effect of particle size on recovery rate can be examined in fixed-bed
columns. That is, a specific pore-diffusion-limited process would yield faster particle recovery kinetics (transport to
the bulk liquid phase) as the size of the particles decreases. The importance of this transport mechanism can be
determined by measuring the rate of carbon recovery from several GAC populations that differ in size.
A Fenton-dependent GAC recovery (kinetic) model was developed based on assumptions that (i) particle surface
and proximate pore water concentrations of the contaminant are at equilibrium and (ii) overall recovery kinetics are
limited by pore diffusion or a combination of surface and pore diffusion. An analytical solution governing
contaminant removal kinetics from carbon was developed and compared to experimental recovery profiles.
Calibration was based on a single fitting parameter, T. To validate the pore diffusion approximation, a single T must fit
the desorption profiles of compounds with differing sorption characteristics. Similarly, a single, unique T must fit
these same desorption profiles for a surface and pore diffusion approximation to be validated.
-------
Theoretical Considerations - Fenton's Treatment for GAC Regeneration
This section discusses possible scenarios and theoretical considerations that may take place in heterogeneous
systems. It should be clear to the reader that the discussion encompasses theoretical speculations as opposed to a
supportable set of hypotheses related to the physical-chemical events in these experiments.
In heterogeneous systems, there exist a variety of scenarios in which pseudo-first-order recovery kinetics could be
expected. Similar to homogeneous reaction kinetics, Fenton-dependent recovery rates in expanded-bed heterogeneous
trials may be limited by the bulk-phase reaction of the target with the *OH. Since the overall rate of contaminant loss
is related to the «OH concentration present in the bulk, a steady state approximation for «OH would lead to pseudo-
first-order kinetics. Under these circumstances, the bulk liquid-phase concentration of contaminant must be in
approximate equilibrium with its adsorbed concentration throughout the recovery period. Therefore, an approximately
linear adsorption isotherm would be an additional, necessary condition for first-order kinetics. In such a case, the
observed recovery rate would be directly related to the second-order rate constant for the aqueous-phase reaction.
Alternatively, it is possible that the reaction rate of the *OH with the adsorbed contaminant limits overall recovery
kinetics. In these circumstances, the bulk aqueous-phase concentrations of H2O2, total iron and (therefore) *OH would
seemingly extend to the carbon surface, where the reaction rate would depend on the abundance of adsorbed
contaminant (Figure 4). The bulk aqueous-phase contaminant concentration would be essentially zero, or at least
much less than the equilibrium aqueous-phase concentration. If the concentration of the aqueous-phase radical is
steady, the rate of target destruction will be proportional to the adsorbed concentration under surface-reaction limited
conditions.
Solid Pore Bulk
contaminant
Figure 4. Intraparticle porous transport of hydroxyl radical to GAC particle surface.
Intraparticle transport mechanisms, such as pore and surface diffusion and surface desorption may also limit
carbon recovery in heterogeneous systems (Figure 5). If the overall rate of contaminant destruction were limited by
the rate of chemical desorption from the carbon surface, first-order kinetics would be expected. That is, the overall
10
-------
desorption rate of compound disappearance would be proportional to the adsorbed concentration. There are, however,
some underlying assumptions here, such as the equivalence of adsorption sites on the carbon surface. Thermodynamic
equivalence must exist among the carbon adsorption sites, so that rapid desorption from less energetically favorable
sites does not result in recovery rates that are initially faster, and subsequently slower, than predicted from a single
first-order kinetic relationship.
Finally, first-order recovery kinetics might be expected if the overall rate of compound disappearance were
controlled by intraparticle diffusive transport to the bulk solution (Figure 5). No fraction of the carbon surface,
however, could be kinetically inaccessible relative to the remainder of the surface. This restriction seems unlikely
considering that relatively small pores within the particle interior typically dominate carbon surface area. Therefore,
kinetic equivalence of adsorption sites in the particle interior and those at the surface, in terms of accessibility to the
bulk solution, seems improbable (Knudsen effects). The loss of contaminant from the carbon surface would, under
those circumstances, be proportional to the aqueous-phase concentration in equilibrium with the sorbed concentration
of contaminant. The diffusion coefficient for intraparticle transport is related to a geometric tortuosity factor, T, and
reflects, at least to a degree, the physical characteristics of the adsorbent (22).
Solid
Pore
Bulk
contaminant •
Figure 5. Pore and surface diffusion, and surface desorption of contaminant within a GAC particle pore.
In summary, the rate of disappearance of initially adsorbed contaminants can be proportional to the mass of target
in the heterogeneous system under a variety of circumstances. Chemical contaminants that react slowly with «OH may
accumulate in the bulk liquid-phase reflecting a reaction-limited rate of chemical destruction. Chemicals that are
insoluble or show a very high affinity for solid surfaces (e.g. carbon) may experience overall rates of Fenton-driven
transformation that are limited by pore and surface diffusion or by the rate of desorption from the surface. Each
situation could yield pseudo-first-order recovery kinetics.
11
-------
Project Objectives
The project seeks to increase the knowledge and applicability of an innovative means of destroying GAC-sorbed
VOCs without the need to remove the sorbent from the adsorption reactor. The project focuses on laboratory bench-
scale experimentation, evaluation and modeling combined with limited pilot-scale field testing.
The specific laboratory objectives were:
• Investigate the use of reductants and electron shuttles, other than hydrogen peroxide to accelerate the
Fenton-driven process for organics degradation.
• Evaluate the effect of chloride build-up in the Fenton reagent regenerant solution during chlorinated
organic degradation.
• Establish the relative merits of regenerant liquid amended versus surface precipitated iron for catalyzing
Fenton's reaction.
• Identify the mechanism and rate limiting step(s) for Fenton reagent destruction for a range of chlorinated
organics.
• Evaluate the dependence of reaction kinetics and efficiency on solution pH.
• Provide a scoping level evaluation of the relative economics of Fenton reagent regeneration of VOC-
bearing GAC versus conventional off-site thermal regeneration or hazardous waste disposal.
The specific field testing objectives were:
• Provide proof of concept of Fenton's reagent, in-place regeneration of VOC-laden GAC.
• Establish basic performance characteristics and challenges for field use of Fenton's reagents for GAC
regeneration.
• Establish the effect on GAC adsorption capacity of multiple regenerations by Fenton's reagents.
• Establish a scoping level estimate of field scale economics of the process.
12
-------
Materials and Methods
Chemicals
Purified water (Milli-Q™ Water System by Millipore) was used in all experiments. The following chemicals
were obtained from Sigma-Aldrich: carbon tetrachloride, CT (99.9% HPLC grade), 1,1,1-trichloroethane, TCA
(>99%), hexachloroethane, HCA, ferric sulfate [Fe2(SO4)3'5.6H2O], cupric sulfate pentahydrate (CuSO4«5H2O),
hydroxylamine hydrochloride (H3NOHC1; minimum 99%), copper (I) chloride (CuCl; 97%), 1,4-benzoquinone (1,4-
BQ; 98%), hydroquinone (HQ; 99%), 9,10-anthraquinone 2,6-disulfonic acid disodium salt (>98%). Hydrogen
peroxide (30%, reagent grade) and methanol (HPLC grade) were from Fischer Scientific. Ultra resi-analyzed
tetrachloroethylene (PCE), methylene chloride (CH2C12), trichloroethylene (TCE), chloroform (CF), «-heptane, 1,10-
phenanthroline monohydrate, ferrous sulfate hepta hydrate (FeSO4«7H2O), and hydrochloric acid were from J.T.
Baker. Titanium sulfate solution was obtained from Pfaltz and Bauer, and sulfuric acid, ethyl acetate, ammonium
acetate, and 1,2-dichloroethane (DCA) were from EM Science. Potassium permanganate and isopropyl alcohol were
obtained from EMD Pharmaceuticals, and ferrous ammonium sulfate [FeSO4(NH4)2SO4*6H2O] from Spectrum
Chemical Mfg. Corporation. All chemicals were reagent grade or better and were used as obtained.
Analytical
Target Organic Compounds
The target VOCs (methylene chloride, chloroform, carbon tetrachloride, 1,2-dichloroethane, 1,1,1-
trichloroethane, trichloroethylene, tetrachloroethylene) were analyzed using a modified version of the EPA method
551.1, "Determination of Chlorinated Solvents by Liquid-Liquid Extraction and Gas Chromatography with Electron-
Capture Detection." Samples were prepared for analysis by placing 20-jaL (15-jaL in some experiments) in a 2-mL
glass crimp-top vial containing 1 mL of heptane. Using an auto sampler 1 jaL of the extract was injected into a
Hewlett Packard 5890 Gas Chromatograph (GC, Palo Alto, CA) equipped with a DB-624 fused silica capillary
column (J & W Scientific, Fulsom, CA; 0.53 mm ID, 30 m in length). The GC used an electron-capture detector
(ECD) for quantification of chlorinated compounds. Nitrogen and helium were used as the make-up and carrier gases.
The gas flow rate was 26 mL/min. The temperatures of the detector and inlet were 275°C, and 150°C, respectively.
The oven temperatures ranged from 35°C to 100°C and the sample run times were 5-20 minutes, depending on the
13
-------
compound analyzed. A chlorinated compound (e.g. carbon tetrachloride) was added to each sample as an internal
standard to adjust for instrument error and sample size inaccuracy. A response factor was obtained from the
calibration curve relating the response of the analyte to that of the internal standard. This factor was then used to
determine the analyte concentrations. A calibration curve was run prior to all analysis; samples were diluted as
needed. To ensure proper results and monitor instrument performance, a check standard was run every tenth sample.
The percent deviation between check standards was no larger than 5%, indicating proper operation of the GC. A more
complete description of procedures for adjustment of analyte concentrations using the internal standard is in the
Quality Assurance Project Plan (QAPP) submitted to EPA project officers prior to beginning project trials
(November, 2003).
Hydrogen Peroxide
Hydrogen peroxide was analyzed using a peroxytitanic acid colorimetric method (Boltz and Holwell, 1978), as
modified by Teruya (2000). The procedure was as follows. The sample (50 jaL) and 50 jaL of titanium sulfate (Pfaltz
and Bauer, Inc., Waterbury, CT) solution were added to 4.9 mL of deionized water. Titanium sulfate was provided in
stoichiometric excess to react with H2O2 leading to color development and quenching the Fenton reaction. After 1
hour, color development was measured at a wavelength of 407 nm using a Hitachi U-2000 doubled-beamed
spectrophotometer (Hitachi Corporation, Schaumburg, IL). Samples were diluted with deionized water as necessary to
fall within the range of the standards.
pH
A Hach One pH/ISE meter (Hach Company, Loveland, CO) was used to monitor the pH of the solution
containing Fenton's reagents. The meter was calibrated using standard pH calibrating buffers (pH 2 and 4) from VWR
(Aurora, CO).
Iron
Total iron content of all the samples was analyzed using the phenanthroline method (Standard Methods for the
Examination of Water and Wastewater, 1995).
Quality Assurance
The utmost care was taken to assure the quality of laboratory work. In accordance with the Quality Assurance
Project Plan (QAPP), the lab maintained logbooks, internal standards, proper storage of samples, and check standards.
The Quality Assurance Officer specified all analytical methods, evaluated analysts for competency to perform the
analyses, and monitored all phases from sample collection to disposal. The project was subject to audit by EPA and
14
-------
quarterly audits by the Quality Assurance Officer. These audits consisted of reviews of analytical methods, analyst
familiarity with methods, reviews of standard solution quality, reviews of calculations, and instrument performance.
Experimental
General
Experiments were conducted in either batch or column reactors. In general, batch reactors were used to
investigate homogeneous Fenton reaction kinetics. Experiments in bench-scale column reactors were designed to
examine the role of mass transfer limitations to Fenton-driven carbon recovery rates and to expose key operational
characteristics of the Fenton-dependent carbon regeneration mechanism. It is not known with certainty, for example,
whether compound desorption must precede reaction with Fenton-derived free radicals and (consequently) whether
desorption rates control overall process kinetics. Column reactors were also designed and tested at the field-scale.
Isotherms for target compounds were developed by varying the mass of carbon in a series of batch reactors that
contained identical masses of the target contaminant. Masses of carbon and contaminant were estimated in advance
to provide a broad range of aqueous-phase concentrations following attainment of equilibrium.
Batch Kinetic Experiments
Reaction vessels consisted of 65-mL glass vials, capped with mini-inert valves. Chemicals in the reaction
mixtures always included PCE or CT (the target compounds in these experiments), ferric iron (added from a pH-
adjusted solution of ferric sulfate), and H2O2 (added to achieve target initial concentrations from a 30% stock
solution). Depending on experimental objectives, one of several quinones, copper, hydroxylamine or radical
(•OH/O2*) scavengers were at times part of the reacting mixture. Reactors were filled to near capacity and placed in
an Orbit Shaker Bath for temperature control. In a subset of the experiments, H2O2 and/or PCE were periodically
replenished. Both PCE and CT degradation was studied in the presence of radical scavengers in Fenton's reaction. In
these experiments, the same 65-mL vials were utilized but with 5 mL headspace.
Experiments with 0.5 mM total iron were run at room temperature (22-24°C). When total iron was 0.1 mM, the
temperature was maintained at 30-31°C. Experiments involving quinones were conducted at room temperature (22-
24°C). Experiments studying the effect of pH, Cl" accumulation and radical scavengers in Fenton's reaction were at
32°C. Temperatures in the range 8.8-54.4°C were also explored in the experiments designed to establish the effect of
copper addition on Fenton's (iron based) reaction for PCE decomposition.
Adsorption Isotherms
Isotherms were obtained for each of seven chlorinated target compounds (Table II). For each compound, five
160-mL glass serum bottles containing varying masses of carbon (1-6 grams) were filled with water containing the
15
-------
respective target chemical and crimp sealed. A blank (compound solution with no carbon) was utilized when running
all isotherms. The measurement of this blank was used as the C0 (initial concentration).
Water at near-saturation levels with contaminant was used to initially load the target VOC into the serum bottle to
avoid co-solvent effects and/or adsorption of neat-phase contaminant directly onto the carbon. The serum bottles
were placed in a temperature-controlled water bath (32°C) equipped with a shaker element. The bottles were allowed
to equilibrate for at least 36 hours. The liquid phase was then sampled and analyzed for the contaminant by GC-ECD.
The mass adsorbed on the carbon was determined from the difference between the initial (saturated water)
concentration and the liquid concentration following adsorption. Isotherm data were fitted using a Freundlich model.
Model parameters were determined via simple linear regression analysis of the log-transformed data.
Column Experiments
GAC Selection and Preparation
Granular activated carbon (URV-MOD 1, Calgon Corporation, Pittsburgh, PA) was used in all trials. This
experimental-type carbon was selected because of its relatively high iron and low manganese contents. The GAC was
steam-activated to minimize reactivity with H2O2 (Huling et al., 2005a). It is a bituminous coal, 8x30 mesh (effective
size 0.6-2.4 mm), with a specific surface area of 1290 m2/g and pore volume of 0.64 mL/g (Huling et al., 2005a).
Physical properties of this experimental carbon provided by the Calgon Corporation are summarized in Table III. The
8x30 mesh size fraction was used in expanded-bed column and isotherm adsorption experiments. In fixed-bed studies,
sieve-sorted particle distributions of 1.0-1.18 mm, 1.4-1.7 mm, and 2.0-2.38 mm were used. Calgon URV-MOD 1
carbon was sieve sorted using USA Standard test sieves with ASTM-11 specification.
Table II. Chemical Properties of the Organic Compounds Studied
Name
Methylene Chloride (MC)
1,2-DCA
1,1,1-TCA
Chloroform (CF)
Carbon Tetrachloride (CT)
TCE
PCE
Formula
CH2C12
C2H4C12
C2H3C13
CHC13
CC14
C2HC13
C2C14
Log Kowa
1.15
1.47
2.48
1.93
2.73
2.42
2.88
i T-.-CC • -t. Adsorbed
KM OH Diitusivity , ,.
/-M-i -Kb / 2; NC concentration
(M s ) (cm /s) , . . d
^ ' ' ' (mg/g)d
9.00E+07
7.90E+08
l.OOE+08
5.00E+07
2.00E+06
2.90E+09
2.00E+09
1.21E-05
1.01E-05
9.24E-06
1.04E-05
9.27E-06
9.45E-06
8.54E-06
275
146
20
188
N/A
103
11
Freundlich Parameters6
K (mg/g) 1/n
(L/mg)1/n
0.07
0.04
0.65
1.48
12.30
5.82
45.66
1.06
1.33
0.87
0.77
0.59
0.70
0.56
Note: kM OH is the second-order rate constant for the reaction of hydro xyl radical with the target organic compound.
Source: "Swarzenbach et al., 1993; bwww.rcdc.nd.edu, except carbon tetrachloride (Haag and Yao, 1992); 'calculated from Wilke-Chang equation (Logan, 1999);
dFrom analysis of initial carbon concentration for carbon recovery experiments (data at 32°C); 'From isotherm data obtained in this lab at 32°C.
16
-------
Table III. Physical Properties of Calgon URV-MOD 1
Property Value
BET Surface Area (95% CI) 1290 m2/g ± (1260-1330 m2/g)
Total Pore Volume (95% CI) 0.643 mL/g ± (.613-.673 mL/g )
Micropore 0.386 mL/g
Meso & Macropore 0.257 mL/g
Porosity 0.592
Source: Huling et al., 2005a.
Prior to use in experiments, GAC was dried overnight at 103°C to obtain the dry weight. Dried carbon was cooled
in a vacuum dessicator, and then wetted with de-ionized water from a Milli-Q-water system. To guarantee a fully
hydrated surface, the suspended carbon was shaken at room temperature for 24 hours prior to contaminant loading.
Loading of a target contaminant onto the carbon was done in an aqueous-phase suspension to avoid co-solvent effects,
and the adsorbed mass was determined by GC-ECD analysis, and verified by determining the difference between pre-
and post-adsorption liquid-phase concentrations. Pre-loading of carbon for bench-scale experiments was
accomplished by placing 10-16 g of GAC into 1.00 L of water containing the experiment-specific target contaminant.
Normally, the water was nearly saturated with the target at room temperature prior to the introduction of carbon. The
1-L batch reactors with negligible headspace were then sealed and tumbled for approximately 60 hours (room
temperature) for attainment of equilibrium between the dissolved and adsorbed chemical. Final (measured) adsorbed
concentrations are provided in Table II.
Iron-Amended GAC Preparation
In one set of trials, iron was precipitated on the GAC surface to evaluate the effectiveness of using iron-amended
GAC versus providing iron in the bulk solution. These trials were conducted in bench-scale column reactors using
GAC loaded with either TCE or PCE as the target VOCs.
To precipitate iron on the surface of the carbon, an iron solution was equilibrated with the carbon for
approximately 4 days. The iron loading was based on a critical iron loading on GAC particles that was established by
Huling et al., (2007). The critical loading was established as the GAC-bound level that maximized the rate of H2O2
consumption for the iron-amended GAC in water plus H2O2. To prepare the iron solution, FeSO4«7H2O was dissolved
in water to obtain 2.2 g/L Fe (0.039 M Fe). Ten grams of dry GAC were placed in each vial with 30 mL of the iron
solution. Sulfuric acid was added as necessary to maintain the pH of the iron-GAC solution near 2.5 during the
equilibrium process (~4 days). After 4 days, the pH was increased to 3.0 using a solution of NaOH. The liquid was
analyzed for total iron using the phenanthroline method. Subsequently, the iron-amended GAC from each vial was
combined in a beaker, dried, weighed and utilized in the column experiments. For analysis of the iron content on the
carbon, 5 g of iron-amended and background (clean) GAC were crushed to homogenize the samples. Replicates of the
17
-------
crushed GAC samples were analyzed by Shaw Environmental, Inc., under EPA direction. Samples were prepared by
microwave extraction and filtration and analyzed by ICP-OES (Perkin Elmer Optima 3300DV ICP). A standard
operating procedure for determination of total nitric acid extractable metals from solids and sludges was used.
Extraction of VOCs from GAC
Four organic solvents were evaluated in terms of their ability to extract PCE from activated carbon. The
extracting solvents analyzed were heptane, pentane, ethanol, and ethyl acetate, and the target organic compound was
tetrachloroethylene (PCE). Results are summarized in Figure 6. Five identical samples of carbon were loaded with
equal masses of PCE and subsequently extracted with one of the four solvents. Extraction mixtures were sampled at
6, 12, and 24 hours and analyzed by GC-ECD.
Of the four solvents, ethyl acetate provided the highest extraction efficiency (>93%) after 12 hours. All other
solvents took considerably longer. Table IV contains the percent recovery of PCE in the same experiment.
Extractions
n 6 hours
• 12 hours
n 24 hours
• Initial Mass Added
Ethyl Acetate (1) BhyI Acetate (2) Bhanol
Solvents
Heptane
Pentane
Figure 6. PCE extraction efficiencies in four different solvents.
Table IV. Extraction Efficiencies
Solvent
Ethyl Acetate (1)
Ethyl Acetate (2)f
Ethanol
Heptane
Pentane
6 hours
92%
95%
20%
71%
55%
12 hours
100 %
93%
22%
76%
68%
24 hours
94%
95%
21 %
77%
68%
Note: f A second trial using ethyl acetate was conducted and verified performance.
18
-------
A period of 12 hours was considered adequate for extraction of PCE from the activated carbon using ethyl
acetate. Based on these results, all subsequent extractions of chlorinated solvents from GAC used ethyl acetate
solvent and a 12-hour extraction period.
Sample Analysis
In all column experiments, representative volumes of carbon (0.5-0.7 grams) were withdrawn from the reactor,
weighed, and extracted in ethyl acetate in crimp sealed vials. After the 12-hour extraction period, samples were
prepared for analysis by withdrawing a 20-(iL sample of ethyl acetate from the crimped carbon vials, and injecting it
into a 2-mL glass crimp top vial containing 1 mL of heptane. The mass of VOC in the heptane was subsequently
quantified by GC-ECD (see earlier description). Liquid samples from the reactor bulk fluid were collected at the same
time as the GAC samples to measure contaminant desorption rates and to complete a mass balance. Liquid-phase
contaminant levels were monitored in the bulk aqueous-phase by placing samples from the reservoir (Figure 7) in 1-
mL crimp sealed vials. Dilutions were performed in heptane as needed, and 1.00 (iL was then injected in the GC-
ECD. An internal standard, appropriate for each compound, was added to each sample to analyze and adjust for
instrumentation inconsistencies. Analyte-specific internal standards were selected based on GC peak retention times,
to produce sufficient separation between the target and the internal standard. The oven temperature and run time were
also selected based on chromatographic peak analysis. Oven temperatures ranged from 60°C-100°C (inlet
temperature: 200°C, detector temperature: 275°C) with run times of 4-9 minutes, depending on the compound being
analyzed. A calibration curve was established prior to all analyses.
Pump
Figure 7. Expanded bed desorber setup.
19
-------
Desorption Rate Experiments
Expanded-bed column experiments were performed at a temperature of 32°C (± 2°C) using a Chromaflex®
borosilicate glass chromatography column with a 2.5-cm inner diameter and 15-cm length (obtained from VWR). In
these experiments, the temperature was controlled by means of a water bath.
The column was fitted with 2.5-cm inner diameter PTFE frits tapped with 5/8-inch Swagelok stainless steel
fittings that fed to 5/8-inch Teflon tubing. A 6-600 rpm Masterflex® peristaltic pump transported the bulk liquid in an
up-flow mode through the column at a flow rate of 950 mL per minute, a volume flow-rate sufficient to expand the
carbon bed and ensure proper mixing. To improve hydrodynamic mixing, a 5-cm Chromaflex® extension packed with
3-mm glass beads was added to the column influent (Figure 7).
Desorption rates were measured by circulating clean (contaminant free) water through the column. That is, elution
water did not contain Fenton's reagents. Two methods were used in this type of experiment. In the first, the water was
recirculated through the reservoir (T = 32°C) and the contaminant was stripped from the reservoir using a bubbler.
MC, CF and TCE were used in these trials. Later, similar experiments were run by wasting the column effluent rather
than stripping and recirculating it. In this case, tap water was employed (T = 29°C). It should be noted that even
thought no efforts were made to analyze the tap water in these experiments, the typical ion concentrations in Tucson
groundwater are all below the levels that are known to have an effect on reaction rates for Fenton's reaction (see
section entitled, Effect of Cl" on PCE Degradation by Fenton's Reaction).
In both trials, water was passed through the pre-loaded columns at rates designed to minimize aqueous-phase
contaminant concentrations at the column exit, so that rates of mass transport from the carbon surface to bulk solution
were not affected by the bulk liquid-phase concentration. In this manner, contaminant transport kinetics from surface
to bulk could be established directly. Desorption kinetics were established by periodically measuring both residual
contaminants on carbon samples and the contaminant concentration in the liquid exiting the column. In these trials
MC, TCA and TCE were used as the VOC targets, and the results were used in the modeling efforts described later.
Regeneration Rate Experiments
Regeneration kinetics were first studied by circulating Fenton's reagents through a fluidized bed of pre-loaded
GAC as described for the desorption rate experiments above. Column recovery was monitored by periodically
withdrawing carbon samples and extracting residual contaminants for analysis (see earlier descriptions). Preliminary
experiments of this type were conducted at room temperature. When it became apparent that room temperature was
subject to significant uncontrolled changes, temperature was controlled at 32°C in subsequent experiments. In the
initial bench-scale column experiments, VOC-loaded carbon was packed into a Chromaflex® borosilicate glass
chromatography column, I.D.=2.5 cm, L=15 cm, V=85 mL (Ace Glass, Inc., Louisville, KY). Fenton's reagents were
prepared on the day of use. During regeneration, a 10 mM Fe (ferric sulfate) solution was recirculated in up-flow
mode through the column at a rate sufficient to expand the carbon bed by approximately 50%. To initiate recovery,
20
-------
0.2 M H2O2 was added to the recirculating fluid. At 10-60 minute intervals, sufficient H2O2 was added to restore the
original concentration. This generally produced an H2O2 concentration that differed from the original concentration by
less than 50% (data not shown). Periodically the reagent circulation was stopped, while carbon samples were
withdrawn from the top and bottom of the column for extraction and analysis of the target compound. Extraction
periods were 12 hours in ethyl acetate on a shaker table. Extracts were analyzed using GC-ECD with the methods
described below. Carbon samples were then dried at 103°C and weighed. Data are reported as the mean results of the
values for samples from the top and bottom (one each) of the fluidized bed. Aqueous-phase samples were taken from
the recirculation reservoir for analysis of the target compound, reaction by-products and residual hydrogen peroxide.
In iron-amended GAC trials, the reservoir contained water (no iron solution). A smaller reservoir size was utilized
in these experiments (approx. 400 mL). Hydrogen peroxide was added using a peristaltic pump (as described
previously), but the rate of addition necessary to maintain a constant H2O2 concentration was relatively low (see
discussion). In these experiments, the pH was uncontrolled, but monitored.
Fixed-Bed Desorption Experiments
Fixed bed (as opposed to the previously described fluidized bed) experiments were conducted using an Emerson
mechanical vacuum pump rated at 1725 rpm to draw water through a contaminant-loaded carbon bed in a down-flow
direction. The pump was connected to an enclosed reservoir by 1/4-inch Teflon tubing, which fed a 1 L Pyrex® filter
flask that acted as a support base for the reactor assembly. The reactor consisted of a 350-mL borosilicate graduated
funnel with removable support screen (Chemglass®). Prior to experiments, carbon was sieve-sorted and 8-9 grams of
each sample fraction (1.0-1.18 mm, 1.4-1.7 mm, and 2.0-2.38 mm) were equilibrated for 24 hours with an initially
saturated water solution. The aqueous-phase was then decanted off, and the loaded GAC was transferred to the funnel
assembly. Deionized water was drawn through the carbon bed at a rate of 200 mL per hour. A constant volume of 250
mL of deionized water was maintained above the carbon bed using a 1-100 rpm Masterflex® peristaltic pump, which
pumped deionized water from a reservoir at room temperature (32°C ± 2°C). The fixed-bed adsorber design is
presented in Figure 8.
Pump
Vacuum
Pump
Figure 8. Fixed bed desorber setup with mechanical vacuum pump.
21
-------
Representative samples of carbon were removed from the funnel over the course of each 250-minute experiment,
extracted with ethyl acetate on a shaker table for 24 hours, and analyzed for residual contaminant, per above.
Field-Scale Column Experiments
Fenton-driven carbon regeneration was applied to soil vapor extraction (SVE) gas at the Park-Euclid (Arizona)
State Superfund site, in which the primary contaminants are perchloroethene, trichloroethene, dichloroethene isomers,
and the volatile hydrocarbon components of diesel fuel. Local groundwater and soil gases have been studied
intensively so that the extent and severity of pollution are well characterized. A side stream was taken off the full-
scale SVE system at the field site to provide a source of SVE gases (containing mainly TCE and PCE) to the project's
GAC column. The carbon was packed into a borosilicate glass chromatography column, I.D.=5.00 cm, L=30.0 cm,
V=600.0 mL. The gas flow rate passing through the column was 4.0 cfm. Column effluent was returned to the SVE
system. The carbon was typically loaded for approximately 72 hours. After loading, the carbon was regenerated in
place via Fenton's reaction, reloaded with contaminant, and re-regenerated to study process feasibility (see Figure 9).
During regeneration, solid and aqueous-phase samples were withdrawn and extracted as in the bench-scale
experiments (see Figure 10). Initially, hydrogen peroxide was added to the 7-liter reservoir to maintain a near-
constant concentration (0.2 M) throughout the regeneration period. Less frequent pulse additions of H2O2 were later
used as a strategy to reduce H2O2 utilization during carbon recovery.
GAC-Adsorption process
(gas phase)
GAC-Regeneration process
(liquid phase)
Packed
Bed
Adsorption
spent GAC
Fluidized
Bed
Fenton s
Reagents
Ground
Soil Vapor Extraction (SVE)
Figure 9. SVE-GAC system at field site.
22
-------
Rotameter
Shut-off
Valve
Tee
Downflow
for gas
Shut-off Valve
(liquid)
Sampling port
Thermocouple for
temperature control
Upflow
for liquid
Sampling port
SVE return
Shut-off Valve
(Outlet gas)
Tee
Shut-off Valve
(liquid)
Pump
7-L reservoir
Figure 10. Diagram of the field experimental set-up for GAC-regeneration experiments. The carbon was loaded in a down flow mode and
regenerated in an up flow mode. Column ID = 50 mm, L = 300 mm, 0.60 L, HRT (column) = 2 sec., HRT (reservoir) = 0.9 min, pH = 2.0,
[Fe(III)]T = 10 mM, and [H2O2]0 =0.38M.
23
-------
Results and Discussion
Homogeneous Experiments
Dependence ofPCE Reduction Kinetics on Total Iron
Preliminary experiments were carried out at initial concentrations of 0.10 M H2O2, pH 2.0, and either 0.1, 0.5 or
2.0mM FeT (added as Fe(III)). Starting PCE concentrations were 50 jjJVI. PCE disappearance obeyed first order
kinetics (Figure 11) even though there was no attempt to maintain H2O2 at a constant level over the 2-hour
experiments. The magnitude of the first order rate constant varied directly with total iron concentration. At the
highest total iron concentration (2.0 mM), the half time for PCE disappearance was 19 minutes. The reaction
proceeded with essentially no lag following the addition of H2O2, indicating that near steady concentrations of iron
species and hydroxyl radical were established quickly.
• [Fe] = 2.0 mM
A[Fe] = 0.5mM
0
50 . , . ,100 150
time(mm)
Figure 11. Effect of iron concentration on PCE disappearance at a fixed initial hydrogen peroxide concentration of 0.10 M. Total iron
concentrations were 0.1 mM(*), 0.5 mM(A), and2.0 mM(«). The data yield pseudo-first-order rate constants ki of 0.0018 min"1 (*), 0.0047
min"1 (A), and 0.0366 min"1 (•), respectively.
The pseudo-first-order rate constant for PCE disappearance in the experiment with a total iron concentration of
2.0 mM was about 20 times that of the experiment with [Fe]T = 10"4 M. However, the rate constant for PCE
destruction at [Fe]T = 5 x 10"4 M was unexpectedly low. There is no convincing explanation for the seeming
inconsistency.
24
-------
Rate Enhancement via Hydroxylamine (NH2OH) Addition
Many investigators have shown that Fenton-driven reaction rates are initially very fast but subsequently
decelerate rapidly when iron is provided as Fe(II) (Chen and Pignatello, 1997; Gallard and De Laat, 2000; Poppe,
2001). The observation directly supports assertions that Fe(III) reduction limits the overall rate of radical generation
and, in these cases, contaminant destruction. On this basis, it was hypothesized that chemical reductants that convert
Fe(III) to Fe(II) faster than the reaction of Fe(III) with H2O2 would increase the overall rate of hydroxyl radical
formation and the rate of PCE disappearance. Several chemical additives were tested as potential means to circumvent
or accelerate this rate limiting step in the simple Fenton's reagent system.
In the first experiments, identical reaction mixtures were established in two parallel reactors with [Fe(III)]T = 0.5
mM, [H2O2]0 = 0.1 M, pH = 2.0, and [PCE]0 = 70 |jM. Eighteen minutes after the start of the experiment, 0.1 M
NH2OH (a common reductant used to convert Fe(III) to Fe(II)) was added to one of the two reactors. After an
additional 21 minutes, a second, identical dose of hydroxylamine was provided to the same reactor. The aqueous-
phase PCE concentration was measured as a function of time.
NH2OH addition dramatically increased the specific rate of PCE disappearance, decreasing the half time for PCE
disappearance from 148 minutes to, at least initially, less than five minutes (Figure 12). It is also apparent that the
hydroxylamine-dependent (specific) rate of PCE conversion was not sustained over the course of the experiment,
probably due to hydroxylamine consumption. Because hydroxylamine concentration was not monitored, this
explanation cannot be established directly. The second hydroxylamine addition at 39 minutes also increased the
specific rate of PCE disappearance, although more modestly. It is hypothesized that reaction products from the first
stages of the experiment consumed a significant fraction of the radicals generated at that point. Accumulation of
reaction intermediates could also account for the observed decrease in the specific rate of PCE disappearance during
minutes 20-39 of the experiment. It seems likely, then, that although NH2OH addition initially greatly enhanced
Fenton-dependent contaminant transformations, consumption of NH2OH and the consequent need for semi-
continuous chemical addition limits the utility of such strategies for enhancing PCE degradation rates.
With hydroxylamine
Without hydroxylamine
0
20
40
80
100
120
60
Time (min)
Figure 12. PCE degradation with hydroxylamine as a reducing agent. T=22-24°C. Initial concentrations: [Fe(III)]T = 0.50 mM, [H2O2] = 0. 10
M, pH= 2.0, and [PCE]= 70E-4 M. Hydroxylamine doses (0.01 M) were added to the system at 18 and 39 min.
25
-------
Quinone Addition
Quinone addition was investigated as a means for accelerating the rate limiting step, Fe(III) reduction, in the
Fenton's mechanism. Quinones are effective agents for facilitating electron transfer to Fe(III) (Fredrickson et al.,
2000). The mechanism involves a series of 1-electron transfers that yield a semiquinone radical intermediate (Chen
and Pignatello, 1997). The initial reductant in this reaction series is thought to be the superoxide radical produced via
Fenton's mechanism (Figure 13). These experiments do not distinguish between superoxide or hydroxyl radicals as
participants in PCE transformation. Later experiments using IP as an «OH radical scavenger, however, suggest that
•OH is the primary reactant with PCE (see later section on superoxide radical effects). Three different quinones were
tested in these experiments (Figure 14). It was hypothesized that the more complex structure of 9,10-anthraquinone-
2,6-disulfonic acid (AQDS) might protect the molecule against radical attack, preserving or extending its ability to
shuttle electrons to the Fe(III) target.
•OH +OH
H2O2
Figure 13. Simplified quinone mechanism (adapted from Chen and Pignatello, 1997). 'SQ stands for semiquinone radical, and O2'" for
superoxide radical.
Quinones were tested by adding them at concentrations up to 5xlO"4 M to Fenton reagents consisting initially of
5xlO"4 M FeT (initially as Fe(III)), 0.10 M H2O2, pH 2.0 and 70 |jM PCE. All experiments were at room temperature
(22-24°C). Initial PCE conversion rates increased monotonically with increasing initial quinone concentration (Figure
15). Addition of 5xlO"5 M hydroquinone, for example, doubled the initial rate of PCE disappearance. An additional
10-fold increase in the hydroquinone concentration, to 5xlO"4 M, increased the initial rate constant of PCE
disappearance by another factor of two. It is apparent that for periods approaching an hour hydroquinone addition
increased the overall rate of Fe(III) reduction and, consequently, the concentration of hydroxyl radicals in these
experiments.
1,4-Hydroquinone (HQ)
1,4-Benzoquinone (BQ)
9,10-Anthraquinone-2,6-Disulfonic Acid
HO
N3SQ3
'SO3N3
Figure 14. Structures of quinones investigated as agents for enhancing Fe(III) reduction rates in Fenton's mechanism.
26
-------
A summary of quinone-dependent first-order rate constants is provided as Table V. First-order rate constants
represent reaction kinetics at the outset of the experiment, before the specific rates of PCE disappearance decreased as
a consequence of quinone destruction or accumulation of reaction intermediates. Initial rates of PCE disappearance
are plotted as a function of the hydroquinone addition in Figure 16. A linear relationship between the specific rate of
PCE loss and the initial hydroquinone concentration is evident.
-control
-0.05 mM
0 mM
0.2 mM
-0.01 mM
-0.5 mM
50
100
time (min)
150
200
0 •
-0.5 -
o
-i -
-1.5 -
-2 -
-2.5
X
• OmMHQ
A O.OSmMHQ
D 0.5 mMHQ
• 0.01 mMHQ
X 0.2 mMHQ
^—Linear (0 mM HQ)
50 100 150
time (min)
200
Figure 15. PCE degradation as a function of 1,4-hydroquinone added.
Initial conditions: [Fe(III)]T = 0.5 mM @ pH = 2.06, [H2O2]0 = 0.10 M (0.25%), [PCE]0 = 4E-5M.
27
-------
Table V. First-Order Rate Constants for PCE Disappearance as a Function of the Initial Hydroquinone Concentration
HQ (mM)
0.5
0.4
0.3
0.2
0.05
0.01
0.005
0
ki(min~!)
1.89E-02
1.59E-02
1.41E-02
1.28E-02
8.30E-03
6.90E-03
6.60E-03
4.45E-03
R2
0.946
0.947
0.972
0.979
0.996
0.998
0.989
0.990
Note: Values were derived via linear regression using the (semi-log) transformed data from Figure 15. In general, only the first 4-5 data
points (first 30-min) from each experiment were used. R2 values are included to illustrate goodness of fit.
0.025
0.000
0.1 0.2 0.3 0.4
[HQ] (mM)
0.5
0.6
Figure 16. Initial first order rate constants for different hydroquinones concentrations. When two or more experiments were run at a single HQ
concentration, symbols indicate average values. Error bars represent ± 1 standard deviation (n=3).
Inspection of the Figure 15 data indicates that, in the presence of hydroquinone, the specific rate of PCE
conversion decreased continuously during the course of each three-hour experiment. It was therefore posited that
Fenton-derived radicals gradually destroy HQ. To test this hypothesis, the reaction time for PCE destruction was
extended by renewing the PCE and H2O2 concentrations in both the HQ-free control and the reactor initially amended
with 0.5 mM HQ (Figure 17). The kinetic advantage initially afforded by HQ addition was completely absent in the
second half of the experiment. Results strongly suggest that HQ no longer served as an electron shuttle after the first
three-hour period of the experiment. Others have noted that quinone function is rapidly lost in the presence of
reactions that yield hydroxyl radicals (Chen and Pignatello, 1997). The practical implication of these results is that
28
-------
hydroquinone addition can accelerate Fenton-driven conversions only briefly unless the quinone concentration is
periodically renewed. Practical and economic considerations probably dictate against this.
0.000
-1.200
100
200
time(min)
300
Figure 17. Two-period hydroquinone addition experiment comparing the performances of Fenton's mechanism for PCE destruction, with (4/0)
and without (B/D) 5xlO"4 M hydroquinone addition. The second period of the experiment involved reestablishment of initial conditions via PCE
and H2O2 addition after the first three hours of the experiment. Initial conditions: [Fe(III)]T = 0.5 mM, [H2O2] = 0.10 M, [PCE] = 7.26xlO"5 M,
pH=2.0, room temperature.
Experiments with 1,4-benzoquinone (BQ) were parallel in design and similar in result to those involving HQ.
Reactors, reaction mixtures and physical conditions of the experiments were identical to those described for HQ. The
initial BQ concentration was varied from 0.0 to 0.5 mM. Initially, the specific rates of PCE disappearance were
significantly enhanced by addition of BQ (Figure 18), although a saturation effect was apparent: BQ concentrations >
0.3 mM all produced approximately the same effect on reaction kinetics (Figures 18 and 19). PCE disappearance was
initially 7 times faster in the reactor amended with 0.3 mM BQ than in the BQ-free reactor. However, prior to the end
of the first 5-hour phase of the experiment, the kinetic advantage offered by BQ addition was entirely lost and likely
reversed. After 5 hours, H2O2 and PCE concentrations were restored to their original levels in the reactors that
initially contained 0.0 or 0.3 mM BQ. The BQ-free control clearly outperformed the BQ-amended reactor at that
point (Figure 20). It is hypothesized, but not verified, that organic residuals derived from the destruction of quinones
served as radical scavengers, thus depleting the *OH radical pool available for PCE depletion.
29
-------
-0.5 mM
-0.4 mM
-0.3 mM
-0.2 mM
-0 mM
-control
50 100
time (min)
150
Figure 18. Effect of initial benzoquinone concentration on the Fenton-dependent rate of degradation of PCE.
Initial conditions: [Fe(III)]T = 0.5 mM, pH = 2.0, [H2O2] = 0.10 M, [PCE] = 8E-5 M. The control had no H2O2.
^
^e
J,
£
0.03
0.03 -
0.02 -
0.02
•
• *
•
•
A.
0.01 - w
0.01 f
f\ f\f\ 1
0 0.1 0.2 0.3 0.4 0.5 0.6
[BQHmM)
Figure 19. Effect of initial benzoquinone concentration on the observed first-order rate constant for Fenton-dependent PCE disappearance.
Initial conditions are given in Figure 18.
0.10
0.00
-0.70
100
200 300
Time(min)
400
500
Figure 20. Effect of 0.3 mM BQ addition on the (log transformed) PCE concentration. Initial conditions: [Fe(III)]T= 0.5 mM, [PCE]0 =
1E-4 M, [H2O2]0 = 0.10 M, pH = 2.0. PCE and H2O2 concentrations were restored to original levels after 300 minutes. The slopes for the
segments of straight lines are 0.0018 min1^) and 0.0020 min^o) for the BQ-free reactors and 0.0055 min1 (4) and 0.0008 min^O) for
BQ-amended reactors. Closed symbols indicate the original reactors.
30
-------
Experiments with 9,10-anthraquinone-2,6-disulfonic acid (AQDS) produced similar results to the BQ and HQ
trials. That is, the initial catalytic effect was quickly lost, and the AQDS-free control outperformed the amended
reactor in the second phase of the experiment (Figure 21). It is apparent that addition of functional groups to the
quinone did not enhance quinone longevity. Neither did the added functional groups make organic by-products from
quinone destruction less effective radical scavengers. Again, the experiment highlights the practical limitations of
quinone addition for enhancement of Fenton-driven destruction of hazardous organic compounds.
-0.5 -
o
-i.o -
-1.5 -
50 100 150 200
tiiiie(niin)
250 300
Figure 21. Effect of 0.5 mM AQDS addition to Fenton's reagents for PCE transformation. Log-transformed PCE data are on the ordinate.
Initial conditions: [Fe(III)]T=0.05 mM, [H2O2]=0.10 M, [PCE]=9.9xlO'5 M, pH= 2.1. Initial PCE andH2O2 levels were restored after 180
minutes. The slopes for the segments of straight lines are 0.0126 min"^*) and 0.0186 min'^o) for the AQDS-free reactors and 0.0492 min"1^)
and 0.0024 min'^O) for AQDS-amended reactors. Closed symbols indicate the original reactors.
Copper Effects
Based on the findings of Walling and Kato (1971), it was hypothesized that the presence of copper in solution can
propagate hydroxyl radical chain reactions, possibly enhancing the rate of PCE destruction in homogeneous
experiments. To test this hypothesis, initial experiments were conducted in which 0.0, 5xlO"4 M (Figure 22), 2.0xlO"3
M, 4.0xlO"3 M and 2.0xlO"2 M Cu(II) were added to the aqueous-phase system that consisted of 5xlO"4 M Fe(III),
0.10 M H2O2 and 10"4 M PCE. The initial pH was 2.1, and the experiment was conducted at room temperature.
Negative controls contained 5x10~4 M Cu(II), but lacked either iron or H2O2. In both controls, the loss of PCE was
negligible over the 5-hour experiments. Rate dependence on copper concentration proved to be complex.
At Cu/Fe molar ratios from 4-80, however, rates of PCE degradation were enhanced by copper addition. Process
kinetics remained first-order in PCE, but the apparent first-order rate constant was increased more than two-fold at
[Cu(II)] = 4xlO"3 M (Figure 23). Unlike the quinone and hydroxylamine trials, PCE conversion kinetics remained first
31
-------
order throughout each experiment since copper was conserved during the radical generation and reaction process.
Presumably the concentrations of copper species changed little over the course of the experiment, following
establishment of near-steady conditions in the first minutes of each experiment.
o
U
U
0.6 -
0.4 -
0.2 -
0.0
-0.5 mMCu
-no Cu
-noFe
- Control
50 100 150 200
time (min)
250
300
0.1
-0.1 -
o -0.2 H
u
- -0.3 H
-0.4 -
-0.5 -
-0.6
0.5 mM Cu, cycle 1
0 mM Cu, cycle 1
no Fe
0.5 mMCu, cycle 2
0 mMCu, cycle 2
50 100 150 200
time (min)
250
300
Figure 22(a)(b). Effect of SxlCT4 M Cu(II) addition on the Fenton-driven rate of PCE destruction. Initial conditions: [Fe(III)]T = 0.5 mM,
[H2O2]0 = 0.10 M, [PCE] = 10'4 M, pH = 2.1, room temperature. At 180 minutes, PCE and H2O2 were added to the system to reestablish their
initial conditions.
32
-------
3.0 -
2.5 -
)2'° "
jl.5 -
) <
<1.0 -
0.5 -
0.0 f-
0
20
40 60
Cu:Fe
80
100
3.0 -,
2.5 -
2.0 -
1.5 -
1.0 -
0.5 -
0 0
b
.111
* •
0 20 40 60 80 100
Cu:Fe
a
5.0
4.5-
4.0-
3.5-
3.0-
2.5-
2.0-
1.5-
0.5-
n n
U.U T 1 1 1 1 1 1
0 0.5 1 1.5 2 2.5 3
Cu:Fe
3.5 4
Figure 23. Effect of Cu(II) addition on the pseudo-first-order rate constant for Fenton-driven PCE transformation. Initial conditions:
(a)[Fe(III)]T = 0.5 mM, [H2O2]0 = 0.10 M, [PCE]0 = 5xlO'5, [Cu(II)] from 0 to 20 mM, pH = 2.1, room temperature; (b) [Fe(III)]T = 0.1 mM,
[H2O2]0=0.10 M, [PCE]0=10~4 M, [Cu(II)] from 0 to 8 mM, pH = 2.1, T = 30°C. (c) [Fe(III)]T = 0.1 mM, [H2O2]0=0.10 M,[PCE]0=6xlO'5 M,
[Cu(II)] from 0 to 0.4 mM, pH = 2.0, T = 30°C. Rate constants obtained at other temperatures were corrected to 30°C using the Arrehenius plot
(see Figure 24). Ordinate values are the ratio of rate constants measured in the presence and absence of copper at the molar Cu:Fe ratio
indicated. Symbol indicates the average value. Error bars were calculated for some data points (n>3) using the standard deviations.
33
-------
The same series of copper-addition experiments were repeated using a total ferric iron concentration of l.OxlO"4
M, with Cu/Fe (molar ratios) from zero to 80 (T = 30°C). Results (Figure 23b) were similar to those obtained with
0.5 mM total iron. Cu:Fe ratios from 20-40 increased the pseudo-first-order rate constant for PCE disappearance by a
factor > 2.0.
Mechanism of Rate Enhancement by Copper
The acceleration of PCE degradation by copper addition at Cu/Fe ratios near 10:1 can be explained in the
following way. The reaction of Cu(I) with H2O2 may be slow or proceed without the production of hydroxyl radicals.
This is supported by the lack of PCE degradation in mixtures that initially contained Cu(II), H2O2, and PCE but with
no iron present. For the case with both Cu and Fe present, if the reaction of Cu(II) with H2O2 is fast, the Cu(I) formed
may reduce Fe(III) to yield Fe(II) and regenerate Cu(II). In the Fenton's system without copper amendment, iron
reduction limits the rate of hydroxyl radical generation, thus, acceleration of ferrous iron production by copper
addition would increase the overall rate of hydroxyl radical generation, and thus the pseudo-first-order rate constant
for PCE disappearance. That is, the following sequence of reactions would yield additional free radicals and
accelerate PCE oxidation:
Cu(II) + H2O2 <-» Cu(I) + O2' + 2FT (23)
Cu(I) + Fe(III) <-> Fe(II) + Cu(II) (15)
Fe(II) + H2O2 -> Fe(III) + -OH + OH (as above) (2)
Superoxide radicals produced in reaction 23, may reduce another metal ion yielding molecular oxygen. The
cuprous ion may also terminate the chain reaction that propagates the PCE conversion reaction. That is,
Cu(I) + R- -> Cu (II) + R- (24)
The ability of copper addition to accelerate PCE degradation is fairly modest, at least at room temperature.
Nevertheless, Cu(II) solubility is greater than that of iron so that copper addition may provide attractive benefits in the
pH range where the rate is limited by iron solubility. Furthermore, copper-dependent effects on contaminant treatment
kinetics may be greater at higher temperatures.
Temperature Effects
Dependence of the first-order rate constant for PCE disappearance on temperature was established for both
copper-free and copper-amended reaction mixtures. It is apparent that the copper-dependent mechanism of PCE
destruction is more sensitive to temperature than is the conventional (Cu-free) Fenton-driven mechanism. Arrhenius
plots corresponding to the Cu-free and Cu/Fe = 8:1 cases (Figure 24) provided different activation energies and pre-
exponential factors, indicating that the addition of copper to the Fenton's reactants alters the mechanism of
destruction, including the rate-limiting step. In the copper-free solution, at least, rate limitation is thought to arise
34
-------
from reduction of Fe(III)-peroxo complexes by H2O2. The greatest benefit of copper addition, in terms of accelerated
PCE destruction was achieved in the highest temperature range investigated (40 - 54°C). At 54°C, the expected rate
constant for PCE transformation was increased almost four-fold by copper addition at an 8:1 molar (Cu/Fe) ratio. At
30°C, rates with and without copper addition were essentially indistinguishable. Because Fenton's reaction is
exothermic, reaction heat could enhance the benefits of copper addition to hydroxyl radical generation and PCE
degradation rates. Iron was provided at nearly its solubility limit in these experiments (at pH 2.0 Fe(III) solubility is
about 2xlO"3 M). Therefore, acceleration of Fenton's reaction via further iron addition is infeasible. Thus, copper
addition offers an attractive method for further enhancing reaction rates, particularly at higher temperatures.
= -13200x + 38
R2 = 0.94
y = -9770x + 26
R2 = 0.99
-2.0 -
-6.0 -
-10.0
3.0E- 3.IE- 3.2E- 3.3E- 3.4E- 3.5E- 3.6E-
03 03 03 03 03 03 03
1/T (1/K)
Figure 24. Arrhenius plots for copper-free and fixed Cu:Fe ratio (8:1) cases. Initial concentrations: [Fe(III)]T = 0.1 mM, [H2O2]0 = 0.10 M,
[PCE]0= 9.0x10" M. Temperature range: 8.8-54.4°C. Symbols used: * Fenton's system without copper and • Fenton's system with copper.
Homogeneous Model Formulation
A. Simplified Fenton Mechanism
In Fenton's mechanism, successive reactions of hydrogen peroxide cycle iron between the plus two and plus three
oxidation states, generating two radical species (*OH, and HO2«/O2«) in each cycle. The radicals produced are capable
of oxidizing/reducing many organic targets per the following simplified mechanism:
Fe(III) + H2O2 -> Fe(II) + HO2- + FT (1)
Fe(II) + H2O2 -» Fe(III) + -OH + OH (2)
•OH + R -> OH + -R (3)
B. Complications in the Actual Fenton Reactions
A mathematical model based on the kinetic model of De Laat and Gallard (1999) is presented in this section. The
model describes the decomposition of hydrogen peroxide by iron in a homogeneous aqueous solution, taking into
account the rapid formation and the slower decomposition of Fe(III)-hydroperoxo complexes (Fem(HO2)2+ and
Fem(OH)(HO2)+). The objective of this modeling effort is to achieve a better understanding of the effect that the
35
-------
operational parameters in Fenton's reaction, such as pH and initial concentrations of H2O2 and iron, have on the utility
of the system for degradation of organic toxins.
L Complexation with H2O2 and Complex Disproportionation - Implication for pH Effects on Rate of Reaction
Gallard et al. (1999) identified two ferric hydroperoxo complexes by spectrophotometric analysis. These two
complexes are formed rapidly after mixing Fe(III) and H2O2 in solution. The species are formulated as [Fem]3+, (HO2~)
(or Fem(HO2)]2+; Ij) and [Fem](OH)]2+, (HO2) (or [Fem](OH)(HO2)]+; I2), which are in acid base equilibrium (KI1/I2 =
1.8 xlO"4 M) (Table VI). Once generated, these species are assumed to decompose to Fe2+ and HO2« (Table VI).
The reaction rate constants and the equilibrium constants for the species involved in the Fenton's reaction and
included in the model are listed in Table VI. Reactions (1) to (5) are sufficiently fast to be at equilibrium in this
application.
Table VI. Reaction Mechanism for Fe(III)-Catalyzed Decomposition of H2O2 (25 °C; 7 = 0.1 M)
No
(1)
(2)
(3)
(4)
(5)
(6a)
(6b)
(7)
(8)
(9)
(lOa)
(lOb)
(lla)
(lib)
(12a)
(12b)
(13a)
(13b)
(14a)
(14b)
(15)
Reaction
Fe3+ + H20 <-> FeOH2+ + H+
Fe3+ + 2H2O <-> Fe(OH)2+ + 2H+
2Fe3+ + 2H20 <-> Fe2(OH)24+ + 2Ff
Fe3+ + H2O2 <-> Feln (HO2)2+ + Ff
FeOH2+ + H202 <-> Fem (OH)(HO2)+ + Ff
Fem (H02)2+ -> Fe2+ + HO2-
Feln (OH)(HO2)+ -> Fe2+ + HO2- + OH"
Fe2+ + H202 -> Fe3+ + -OH + OFF
Fe2+ + -OH -> Fe3+ + OH"
•OH + H2O2 -> HO2- + H2O
Fe2+ + HO2- -> Feln (HO2)2+
Fe2+ + 02-- + Ff -+ Feln (HO2)2+
Fe(III) + HO2- -^ Fe2+ + O2 + H+
Fe(III) + O2-- -> Fe2+ + O2
H02- -> 02-- + H+
O2-- + H+ -^ HO2-
H02-+ H02- -+ H202 + 02
HO2- + O2-- + H2O -> H2O2 + O2+ OH"
•OH + H02.^H20 + 02
•OH + O2-~ -^ OH'+ O2
•OH+.OH^H202
Constants
K,= 2.9xlO-3M
K2=7.62xl(r7M2
K2)2 = 0.8xlQ-3M
KI^S.lxlO'3
KI2 = 2.0xlQ-4
k6!ll = 2.7xlO-3s-1
k6.I2 = 2.7xlO-3s-1
k^es.o^M-'s-1
k8 = 3.2xl08M'1s-1
k9 = 3.3x!07M-1s-1
ki0a= 1.2 x It/M^s"1
klob=1.0x!07M-1s-1
klla<2xl03M'1s-1
kllb=5x!07M-1s-1
k12a= l.SSxlO's'1
k12b= IxlO^M-1 s'1
k13a=8.3xl05M-V
k13b = 9.7xl07M'1s-1
k14a = 0.71xl010M-1s-1
k14b= 1.01xl010M'1s-1
k15 = 5.2xl09M-V
Note: a Apparent second-order rate constant for the reaction of H2O2 with Fe(II) at pH < 3.5.
Source: De Faat and Gallard, 1999. De Faat and Gallard (1999) fit kg to the experimental data (pH <3.5),
where l{ is the predominant Fe(III)-hydroperoxy complex.
36
-------
In a constant volume batch system, the reactions listed in Table VI lead to the following mass balances:
£6([Ii]+ M) - £7[Fe2+][H202] - £8[Fe2+][-OH] - £10a[Fe2+][HO2-] -
£10b[Fe2+][02--] + klla[Fe(III)][H02.] + kllb[Fe(III)][O2.-] (25)
d[Fe(III)]T/d^ = -d[Fe2+]/d^ (26)
2]/d^ = - £7[Fe2+][H2O2] - £9[H2O2]['OH] + £13a[HO2'][HO2']
+ £i3b[O2'-][HO2'] + £15['OH]2 (27)
= £7[Fe2+][H2O2] - £8[Fe2+]['OH] -£9[H2O2]['OH] - £14a['OH][HO2'] -
£14b[-OH][02--] -2£15['OH]2 (28)
= MIiMt]) + ^[H202][-OH] -£10a[Fe2+][H02-] - £lla[Fe(III)][HO2-] -
+£12b[02'-][H+] - 2£13a[H02-]2 - £i3b[02--][H02-] -£14a[-OH][H02-] (29)
= -£lob[Fe2+][02--] -£llb[Fe(III)][02--] +^2a[HO2-] -^2b[O2.-][lT]
-£i3b[02--][H02-] -kub['OH][02'l (30)
The total concentration of Fe(III) and the concentration of non-hydroxy Fe(III) species are given by:
[Fe(III)]T = [Fe(III)] + [I1] + [I2] (31)
[Fe(III)] = [Fe3+] + [FeOH2+] + [Fe(OH)2+] +2[Fe2(OH)24+] (32)
The system of nonlinear ordinary differential equations was solved numerically using the fourth-order Runge-
Kutta method. Quasi-steady-state conditions were assumed for radical species (d[«OH]/dt = d[HO2«]/dt = d[O2«"]/dt =
0) and their concentrations were obtained using Newton's method. The reaction parameters (pH, [H2O2]0, [Fe(III)]0),
rate constants and equilibrium constants were specified as inputs to the program (for nomenclature of variables - see
Appendix A.I). Model details are provided in Appendix A. The concentration - time profiles for H2O2, and Fe2+ were
calculated by the program, and the concentration of H2O2 predicted by the model was compared to experimental
measurements and/or published data when available.
ii. Model Limitations
Precipitation reactions of Fe(III) have not been incorporated in the model, hence the model should not be applied
in situations where Fe(III) precipitation is expected or observed (typically pH> 3.0).
Differences between the experimental conditions in our study and those employed in the investigation by De Laat
and Gallard (1999) are worth mentioning. De Laat and Gallard conducted their kinetic study at 25.0 °C and I = 0.1 M
(HClO4/NaClO4), while our experiments were performed at 31±1°C (water bath), and the ionic strength varied from
trial to trial but was typically about 0.02 M. In contrast to De Laat and Gallard's use of ferric perchlorate salt, ferric
sulfate salt was used in all experiments and the pH was adjusted using sulfuric acid (H2SO4), to avoid adding
background chloride in the system. In Fenton's reaction, destruction of chlorinated compounds results in
37
-------
accumulation of chloride ions in solution. Therefore, ferric sulfate was chosen over ferric chloride for iron addition, to
avoid the complication of a high initial chloride ion concentration when attempting to sort out chloride effects. Ferric
sulfate was preferable to ferric perchlorate since a rate constant for the reaction of the hydroxyl radical and sulfate
was found in the literature (3.5 x 105 Lmol'V1, Radiation Chemistry Data Center, 2003). No value was found for the
hydroxyl radical and perchlorate reactions.
Hi. Inorganic Radical Formation
a. Chloride, Perchlorate, Sulfate and Nitrate
De Laat et a/., (2004) studied the effect of chloride, perchlorate, sulfate and nitrate ions on the rate of
decomposition of H2O2 and transformations of organic compounds (atrazine, 4-nitrophenol, and acetic acid) by
Fenton's reaction. They observed that the rates of reaction between iron (added as Fe(II)) and H2O2 were
increased by specific anions. Relative effects were in the order of SO42>ClO4"=NO3"=Cr. When iron was
provided as Fe(III), sulfate and chloride decreased both the decomposition rate of H2O2 and the rate of
disappearance of specific organic targets. It was suggested that the latter observation was due to decreased
formation of Fe(III) - hydroperoxy complexes, and that H2O2 does not complex with Fe(III)-sulfato complexes
(De Laat and Giang Le, 2005). The pseudo-first-order kinetic constant (kobs) for the initial rate of decomposition
of H2O2 by Fe(III) decreased by 30%, and almost 20% of the iron was complexed as sulfato-Fe(III) at pH 2.0,1 =
0.2 M, [SO42~] = 2 mM (lowest concentration studied in the range 0-200 mM). Almost complete inhibition of the
rate was observed at [SO42~] = 67 mM, where 93% of iron was complexed with sulfate (De Laat and Giang Le,
2005). Nevertheless, under our experimental conditions, sulfate ([SO42~] = 1.5 mM) effects on H2O2
decomposition were not evident.
In the study by De Laat et al., (2004), inhibitory effects observed in the presence of chloride (100 mM) or
sulfate (33.33 mM) were explained by a decrease in the hydroxyl radical production rate due to the formation of
non-reactive (with H2O2) Fe(III) complexes and the formation of inorganic radicals from «OH (SO4* and C12*~)
that are less reactive than «OH. The same study suggested that kinetic models validated in NaClO4/HClO4
solutions should not be used to predict Fenton reaction rates in the presence of chloride and sulfate. Moreover, De
Laat and Giang Le (2006) recently showed that in the presence of 50 mM chloride, 20% of the Fe(III) was
complexed by the Cl", decreasing the H2O2 degradation rate by 23%.
Tables VII - IX provide reaction rates for the decomposition of hydrogen peroxide and the fraction of Fe(III)
that is complexed with the different anions at the concentrations given.
38
-------
Table VII. Apparent First-Order Rate Constant for the Decomposition of H2O2
Anion
[Fe(III)]
Perchlorate 200
Nitrate 200
Chloride 200
Sulfate 200
[H202]
(mM)
10
10
10
10
[Anion
(mM)
100
100
100
33.33
] Complexes Fe(III) k0bs
(%) (min1)
0 4.67xlO'3
0 4.66xlO'3
21.0 3.95xlO'3
83.7 7.37xlO'4
Source: De Laat et al, (2004)
Table VIII. Measured Pseudo-First-Order Kinetic
I(M)
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.6
Note: T
Source:
[S042-]0
mM
0
2
5
10
20
40
66.67
200
[H+]0 [F
(mM)
10
10
11
11.5
13.5
19
25
40
Constants (k^bs) for the Initial Rate of Decomposition of H2O2 -Sulfate Effect
e(IH)],
mM
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
, [H202]0
(mM)
49.9
50.0
50.1
49.7
49.4
49.3
48.9
48.5
pH
2.00
2.01
2.00
2.01
2.01
2.00
2.01
2.02
Complexes
Fe(III) (%)
0
19.6
38.8
57.3
74.6
87.0
92.8
97.7
^obs /r.rvi
(ID'5 s-1) ([ 01
10.46
8.04
5.78
4.02
2.41
1.25
0.79
0.53
3]/[S04--])
4418
1688
855
419
193
114
51
= 25.0±0.5°C; ionic strength adjusted with NaClO4.
De Laat and Giang Le (2005).
Table IX. Measured Pseudo-First-Order Kinetic
I(M)
0.2
0.2
0.2
0.2
0.2
0.4
0.5
0.8
[cr]T I
mM
0
50
100
200
400
500
800
1000
;Fe(III)]0 [C1O4-]
mM (mM)
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
200
150
100
0
0
0
0
0
Constants (kobs)
[H202]0
(mM)
49.9
49.5
49.7
50.2
50.5
48.2
47.5
48.6
for the Initial Rate of Decomposition of H2O
PH
2.00
2.00
2.00
2.02
2.02
2.00
2.00
1.92
Complexes
Fe(III) (%)
0
17.2
30.6
49.6
49.6
65.5
71.2
84.5
(lO'V) ([C
10.46
8.07
6.80
5.12
5.16
3.81
3.56
2.67
2 -Chloride Effect
l2'-]/[-OH])
0
6x1 02
1.6xl02
3.8xl03
3.8xl03
8.4xl03
l.lxlO4
l.SxlO4
Note: [[HC1O4]0 = 10 mM in all experiments.
Source: De Laat and Giang Le (2006).
39
-------
b. Carbonate (CO3-) and Carboxyl (CO2-") Radicals
Carbonate system effects may be important in reactions involving oxygen containing free radicals.
Michelson and Maral (1983) reported an increase in the oxidation of luminal by hydrogen peroxide in the
presence of bicarbonate and carbonate anions. The following mechanism was proposed:
If + O2 + HCO3'-> CO3 + H2O2 (33)
2H+ +2O2- -»H2O2 + O2 (34)
H2O2 + CO32 ^ HO-+ HO + CO3 (35)
The increase in luminal destruction was attributed to reaction with the carbonate radical. The carbonate
radical is much less reactive than the hydroxyl radical, but enjoys a relatively long lifetime (Michelson and Maral,
1983).
Both the superoxide and hydroxyl radicals (produced in the Fenton's reaction) are able to react with carbonate
ions (CO32~) to produce carbonate radicals (CO3-). In addition, reduction of carbonate by superoxide radical can
produce formate radicals:
FT + CO32 + O2 -> CO2 + O2 + HO (36)
However, the formate radical so formed further reacts to reform carbonate radical via:
H+ + CO32 + CO2 -> CO3 + HCOO (37)
Thus, each pathway yields carbonate radicals from superoxide radicals.
The carboxyl radical anion (CO2-) is widely used in free-radical studies due to its ease of formation in
aqueous solutions by ionizing radiation, and its excellent reductive properties [E(CO2/CO2-")=-1.9V] (Flyunt et
al., 2001) (Table X). The product of the bimolecular decay of CO2- depends strongly on the pH of the solution
with an inflection point at pH 3.8. In acidic solution, CO2 is the main product, while in neutral to basic solution
oxalate is dominant. To explain the dependence of product formation on pH, Flyunt et al., (2001) proposed that
CO2'~ radicals react mainly (>90%) by head-to-tail recombination to form an intermediate. In this mechanism,
rearrangement of the intermediate competes with the proton-assisted disproportionation reactions.
Table X. Carboxyl Radical Anion, CO2-~, Properties
E(CO2/CO2--) -1.9V
pKaCCOjH)1 2.3
Bimolecular decay rate constant2 2k«l .4xl09 dnAnol'V1
Products CO2 and oxalate anion
Note: 'determined by pulse radio lysis with conductometric detection,
2independent of pH in the range 3-8 at constant ionic strength, 3depend on pH.
Source: Flyunt et al., 2001.
40
-------
The carboxyl radical has also been reported to enhance HCOOH oxidation kinetics by assisting in the redox
cycling of iron (Duesterberg et al., 2005).
CO2- + Fe(III) ^ CO2 + Fe(II) (38)
In the absence of oxygen, Fe(II) concentrations are higher due to the absence of autoxidation, which leads to
higher steady-state hydroxyl radical concentrations in Fenton-driven systems. However, in the presence of oxygen
(air-saturated solutions) rate enhancement is diminished by competition for the organic intermediates. With O2
present, the carboxyl radical reacts at diffusion-limited rates to yield CO2 and superoxide radicals
CO2 + O2 + lT-> CO2 + HO2VO2- (39)
Oxygen is the primary carboxyl radical scavenger. The superoxide radicals produced can then reduce Fe(III)
or oxidize Fe(II) in the Fenton system. In the presence of oxygen, organic radicals are converted into species that
can both oxidize and reduce Fe. Under anoxic conditions, only reducing species are available.
The large difference in species profiles between air-saturated and deaerated experiments demonstrates that
oxygen reactions should be included in the kinetic models to accurately simulate system behavior. However,
under this project's experimental conditions, oxygen is always present in the aqueous solutions (see oxygen
generation section). Thus, the effect of the carboxyl radical is not likely to be significant in our system.
c. Radical Formation Kinetics and Reactivity
Hydroxyl radical scavenging by bicarbonate (HCO3~) or carbonate (CO32~) should not have been an issue in
project's trials due to the low pH. At a pH <3, both ions are present at low concentrations.
Hydroxyl radical reaction rates with perchlorate are not available in the literature. If a perchlorate radical is
formed by reaction, it does not affect the rate of hydrogen peroxide decomposition by either Fe(II) or Fe(III), as
discussed above. Even though nitrate radicals can be formed by reaction of hydroxyl radicals with nitric acid
(koH,HN03 = 5.3xl07 M"1 s"1, www.rcdc.nd.edu). their formation does not affect the rate of reaction in the Fenton's
system.
Hydroxyl radical reaction rates with sulfate, hydrogen peroxide, chloride and PCE under conditions expected
here are compared in Table XI. The data indicate that hydroxyl radicals are consumed mainly by reacting with
hydrogen peroxide, PCE, and chloride (produced from PCE dechlorination) and to a much lesser extent, sulfate.
Hence, the computer modeling efforts neglected sulfate effects on «OH.
41
-------
Table XI. Rate Constants for Potential Hydroxyl Radical Sinks
*kOH,
Concentration, M
Rate, s'1
H202
cr
PCE
3.5xl05
3.3xl07
S.OxlO9
2.0xl09
ISxlO'5
0.1
200x10'6
50x10'6
52.5
3.3 xlO6
6.0xl05
1.0 xlO5
Note: [Fe(III)]T = O.lmM, [SO42']=(3/2)[Fe(III)] T since the source of sulfate in our experiments is the Fe2(SO4)3,
[PCE]0 = SOjiM, [Cf]0=0 but [Cf]f = 4x[PCE]0 when mineralization is complete.
Source: *www.rcdc.nd.edu.
Moreover, typical groundwater concentrations of chloride and sulfate in Tucson, AZ where the field
studies were conducted are 0.45 mM and 0.50 mM, respectively. At these concentrations, the effect of
the anions on H2O2 consumption is insignificant, but buildup of chloride from the dechlorination of
compounds like PCE/TCE may eventually interfere with »OH generation and degradation kinetics.
Nitrate and perchlorate do not seem to affect the rate of H2O2 reactions in Fenton systems (De Laat et
al., 2004). Typical groundwater concentrations of nitrate and perchlorate in Tucson, AZ are low - less
than 1 mg/L and < 0.01 mg/L, respectively.
C. Model Applications
L Simulation ofH2O2 Decomposition.
The rate limiting step in Fenton-dependent destruction of organic compounds is typically the
disproportionation of the ferric hydroperoxy species (k6n and k6J2 of Table VI). In the De Laat and Gallard (1999)
formulation, these rate constants were assumed to be equal, and their value was obtained by fitting experimental
data. However, experimental trials in that study were conducted at much higher ionic strengths and lower
hydrogen peroxide concentrations than those used in the present work. Efforts to validate use of the
De Laat/Gallard constant for simulations relevant to our chemical conditions are recommended as follow-on
research.
10 12 14 16 18 20 22 24
time (hrs)
Figure 25. Simulation of H2O2 decomposition at initial concentrations of 0.01 and 0.955M at a constant Fe(III) concentration. Initial conditions:
pH= 3.0(25°C,I=0.1M;R=[H2O2]0/[Fe(III)]0, [Fe(III)]0=2xlO'4M. Data from De Laat and Gallard, 1999.
42
-------
H. Effect ofpHon the Observed Rate Constant for Decomposition ofH2O2 by Fe(III)
The effect of pH on Fenton kinetics has been studied extensively (Burbano et al., 2005; Kremer, 2003;
De Laat and Gallard, 1999). The kobs for the decomposition of H2O2 by catalytic iron increases with pH in the
range 1-3.2, but decreases above pH 3.2. This decrease can be attributed to the precipitation of Fe(III), as
confirmed by measurements of dissolved Fe(III) concentrations (De Laat and Gallard, 1999).
In an effort to validate our mathematical model, the effect of pH on the observed first-order rate constant
(kobs) for the decomposition of H2O2 in Fenton's mixtures was simulated (Figure 26) and compared to data from
De Laat and Gallard (1999). This study's model produced similar results prior to pH 3, after which Fe(III)
precipitation began. However, constant initial rates were not obtained at each pH indicating that the simulation
efforts are not satisfactory. Future research is recommended to resolve these discrepancies.
120
100
in
o 40
20 -
0
0
2 3
PH
Figure 26. Computer simulation showing the effect of pH on the simulated first order rate constant for the decomposition of H2O2 by Fe(III)
(solid line). Initial conditions: [Fe(III)]0=200)o,M, [H2O2]0=10mM. Data points (observations) were obtained from De Laat and Gallard, 1999
and the continuous line indicates the computer generated simulation.
HI Non-Halogenated Organics Degradation in Homogeneous Systems
Table XII shows hydroxyl radical reaction rate constants for non-halogenated organics (i.e. formic acid) that
were studied previously. These constants were incorporated into the mathematical model in order to analyze the
transformation rates of the target compounds. Duesterberg et al. (2005) studied the effect of oxygen and by-
product formation on the oxidation rate of HCOOH in a Fenton's system with initial Fe(II) concentrations of 100-
200 jjJVI and H2O2 concentrations from 0.2 - 2.2 mM. It was reported that the intermediate, carboxyl radical,
enhances oxidation efficiency by assisting in the redox cycling of iron. However, in the presence of molecular
oxygen the improvement was attenuated.
In our experiments, we made no effort to remove oxygen as it was generated. The effect of the carboxyl
radical was neglected at initial formic acid concentrations that were low relative to the dissolved oxygen
concentration (DOsat = 2.7xlO"4 M at 25°C) (Duesterberg, et al. 2005). That is, under the experimental conditions
43
-------
employed in the simulation, the intermediate generated by reaction 1 (Table XII) is assumed to be oxidized
immediately to CO2. Therefore, a simplified version of the oxidation pathway of formic acid (Table XII) was
included in our mathematical model to simulate formic acid degradation in Fenton's reaction (Figure 27). The
mathematical model (see Appendix A.2 for coding) simulates satisfactorily the oxidation of HCOOH (Figure 27)
as reported by Duesterberg et al (2005) in an Fe(II)/H2O2 system.
Table XII. Additional Second Order Reaction Rate Constants for Organic Targets with 'OH
No.
Reactions
constants
(1) -OH + HCOOH -> C02 + H02V02--
(2) PCE + -OH -> 'CC^CC^OH -> (CO2 + HC1)
(3) Cr + -OH -> Cran-
k17 = 6.5xl08MVa
k21 = 2.0xl09MVb
k22 = 3.0x!09M-1s-lb
k22 = 4.3xl09MVc
Source: aDuesterberg et al., 2005; www.rcdc.nd.edu; ° Buxton et al., 1988.
0 100 200 300 400 500 600 700
Time (sec)
Figure 27. Computer simulation (solid lines) for the simplified oxidation of formic acid (HCOOH) in a H2O2/Fe(II) system (pH = 3).
Experimental data (symbols) were obtained from Duesterberg et al., 2005. [HCOOH]0=200 nM, [H2O2]0=1.1 mM (•) or 2.2 mM
(A,«),[Fe(II)]0=200nM(«»orlOOnM(«).
iv. PCE Destruction Kinetics Considering Cl' Effects
Published second-order rate constants for the chloride ion reaction with the hydroxyl radical (reaction 3,
Table XII) indicate chloride should be a significant, unavoidable radical scavenger in Fenton's destruction of
chlorinated organics. In our model, the chloride ion is considered a conservative species. It reacts with the
hydroxyl radical and generates a radical that does not participate in the organic degradation pathway, and then is
regenerated by radical-radical reaction, resulting in reestablishment of the chloride ion.
44
-------
Figure 28 shows the effect of different second-order reaction rate constants for the reaction of hydroxyl
radical with the chloride ion. The literature value (k = 3.0 x 109 M'V1) employed in this simulation was obtained
in a system with [Cl~] = 1.2 M (Grigor'ev et a/., 1987), which exceeds the range of concentrations studied here.
The chloride rate constant was reported to decrease as [Cl~] increased in the range 3.5-12.2 M (Grigor'ev et a/.,
1987). The lack of a reasonable model fit suggests that the literature value for the reaction constant may not be
accurate at the much lower chloride concentrations of our experiments. Using k22 (rate constant for reaction of
•OH and Cl", Table XII) as a fitting parameter, the model (see Appendix A.3) was applied to our experimental
data to find reasonable empirical values for k22. The data fall in between the simulation trials with kflt of 0 - 3x
109 MV. A kflt value of l.OxlO9 A/TV1 is near optimal.
a
u
a
u
CLi
Experimental data
kCl/OH = OM-ls-l
kCl/OH (lit)= 3.0E09 M-ls-1
kCl/OH (fit) = 1.0E09 M-ls-1
20
40 60
Time (min)
100
120
Figure 28. PCE degradation and model fits using various second-order 'OH reaction rate constants (ka OH) f°r the reaction with Cl" ion. The
Cl" is produced from PCE dechlorination and no Cl" was added initially to the solution. Initial conditions: [Fe(III)]0=0.1 mM, [H2O2]0= 0.11 M,
[PCE]0=6.52E-5 M , pH = 2.0, T= 31±1°C. A literature value (k^on = 3.0 x 109 M"1^1), one kfit and one neglecting the reaction of-OH with Cl"
were used for the simulation trials.
D. Simulation of Fenton's Reaction for PCE Degradation
/. pHEffect
Numerous studies have shown that the Fenton-dependent rates of degradation of H2O2 and organic
compounds are optimum near pH 3 (Gallard and DeLaat, 2000; Arnold etal., 1995; Sedlak and Andren, 1991). In
this section, experimental data for the degradation of PCE in the pH range 0.9-3.0 are reported and compared to
simulations using the mathematical model (see Appendix A.3) described previously.
The PCE transformation rate increases with increasing pH (0.9-3.0), with a maximum near pH 3 (Figure 29).
Most of the rate acceleration was evident in the pH range 2.1-3.0. PCE degradation experiments at pH>3.0 are not
presented because precipitation (as indicated by solution turbidity) was observed. The effect of the precipitated
45
-------
Fe(III)-species on the PCE degradation rate is not completely understood. Precipitated Fe(III)-species are not
likely to form complexes with hydrogen peroxide, which changes the rate of decomposition of H2O2 and that of
any organic target in solution. The concentration of the Fe(III)-hydroperoxy species is greater at pH 3 than 2,
which explains why Fenton's reaction is faster at pH 3.
Time dependent PCE concentrations are compared to model predictions in Figure 29. The simulations are not
in good agreement with the experimental data, but model and experiment agree on the general effect of increasing
pH. That is, increasing pH is expected to accelerate PCE degradation. At every pH except pH = 0.9, however, the
model over-predicts the PCE transformation rate. Furthermore, the model predicts greater sensitivity to pH values
from 0.9 to 2.1 than was observed.
W
o
o
£1
pH=0.9
pH=1.5
pH=2.1
pH=2.7
pH=3.0
• 0.9 Simulation
•1.5 Simulation
•2.7 Simulation
3.0 Simulation
12.1 Simulation
0
50
100
150
Time (min)
200
250
300
Figure 29. Effect of pH on the PCE degradation rate in Fenton's reaction. Initial conditions: [Fe(III)]0 =0.103 mM, [H2O2]0=0.11 M, [PCE]0=
62 )o,M, T = 31±1°C. A literature value (kci.on = 3.0 x 109 M'V1) was utilized in all simulations trials.
Degradation of PCE and other chlorinated solvents yields chloride ions that can accumulate in solution.
Chloride ion is a hydroxyl radical scavenger. Reported values for the rate constant for reaction of Cl" with «OH
are near that of PCE (Table II and XII). If mineralization is complete and the rate of production of chloride ion is
four times the rate of PCE destruction, then the accumulation of chloride in the regenerant solution should be
considered in terms of its effect on Fenton-dependent reactions.
To further investigate chloride ion effects, PCE degradation experiments were conducted in which the initial
chloride ion concentrations were varied from 0 - 0.058 M. Figure 30 shows the effect of chloride on the
decomposition rate of PCE in Fenton's reaction. PCE loss decreases with increasing Cl" concentration in the range
0 - 0.058 M Cl". At [Cl"]=0.058 M, the observed first-order rate constant for PCE transformation was about one-
fourth that of the chloride-free solution. The results show that the accumulation of chloride ion in the Fenton's
46
-------
solution can markedly decrease the rate of organic degradation. The accumulation of chloride in the regenerant
due to the destruction of the target is a key aspect of the application of this process for the regeneration of
chlorinated solvents adsorbed to the GAC. It is important to evaluate how much chloride would be added to a
regenerating solution during one carbon recovery event and at what stage chloride accumulation will matter. A
simplistic, yet illustrative, approach is taken below to quantify the chloride accumulation from PCE destruction at
the field-scale.
W
u
u
£1
[Cl]=0 M
[Cl]=0.00198 M
[Cl]=0.02 M
[Cl]=0.034 M
[Cl]=0.058 M
OM
0.00198 M
0.02 M
0.034 M
0.058 M
0 M w/out Cl rxn
0
100 200 300
Time (min)
400
Figure 30. Effect of chloride ion concentration on the PCE degradation rate in Fenton's reaction. Initial conditions: [Fe(III)]0 =0.09 mM,
[H2O2]o=0.11 M, [PCE]0= 87 nM, pH = 2.0, T = 31±1°C. Chloride concentrations: 0 M - 0.058 M. Model simulations are shown by lines
without data points.
H. Cl' Effect
The reaction rate of chloride with the hydroxyl radical alone cannot fully explain the Cl" effect observed
(Figure 30). The rate constant (kCi,oH = 3.0 x 109 M'V1) used in the mathematical model may not be appropriate
for our experimental conditions, as discussed in the earlier section. Data and fit of the mathematical model for
PCE degradation (accounting for reaction of Cl" with «OH) were presented previously (Figure 28). The data fall
between the simulation trials using kflt of 0 - 3x 109 M'V1. A value to 1 x 109 M'V1 provided the best fit.
The problem is to determine the [Cl~] in the regeneration water produced from PCE dechlorination during
Fenton's reaction and then predict its impact. The solution strategy was to calculate the AC1" concentration
attributable to one regeneration of the spent carbon and compare that concentration to Cl" levels shown to impair
Fenton-dependent process performance. Conditions were selected to mimic observations in field-scale
47
-------
experiments. Those conditions included mass of PCE on the carbon at breakthrough, regenerant volume and
fractional recovery during the regeneration event.
Assumptions:
(i) PCE concentration in the solid, Cs = 250 mg/g
(ii) 50% PCE degradation/regeneration
(iii) 4 moles of Cl" produced/mole PCE degraded
(iv) Bulk density of GAC = 0.5 kg/L
(v) 8 L of regenerant for recovery of 1 L of (unexpanded) carbon
The mass of PCE on the carbon at saturation is
O.SkgGAC 0.250kgPCE ILbed n e, „ . J.SgPCE
x x x 0.5 (recovery _ efficiency) - —
Lbed kgGAC SLwater Lwater
and the resultant fractional Cl" mass is
l42gCr/mol J.SgPCE _ 6.68gC/' mold' _ QMr/-
\65.S3gPCE/mol Lwater ~ L 35.5gCr ~
a. Complexation of Cl" with Fe(III)
Iron(III)-chlorocomplexes (FeCl2+ and FeCl2+) are the predominant Fe(III) species at [Cl"] above 200 mM.
The rate of decomposition of hydrogen peroxide decreases 49% at 200 mM Cl", and almost 84% at 1 M Cl"
(De Laat and Giang Le, 2006). In our analysis, 190 mM Cl" is produced per regeneration, which could have a
significant impact on the rate of decomposition of H2O2 (Table XIII).
Based on this analysis, use of regenerant solution for multiple regeneration events would not be
recommended due to the rate-suppressing effects of chloride ion accumulations.
Table XIII. Calculated Effect of Chloride and Perchlorate Ion Concentrations on Rate of Hydrogen Peroxide Degradation in Fenton's Reaction
I,
M
0.2
0.2
0.4
1
[cr],
mM
0
200
400
1000
[ClOJo,
mM
200
0
0
0
pH0
2.00
2.02
2.00
1.76
[H202]0,
mM
49.9
50.2
48.2
46.1
a
0
49.6
65.5
90.7
a (Ia + Ib)
1.29
0.65
0.43
0.09
(iob§)
10.46
5.12
3.81
1.64
Note: [HClO4]0=10mM except 15mM for lOOOmM Cl", [Fe(III)]0=lmM. a represents the molar fraction of Fe(III) present as
iron(III)-chlorocomplexes, a (Ia + Ib) represents the fraction of Fe(III)-hydroperoxy species, and kobs the rate of H2O2 decomposition.
Source: De Laat, J. and Giang Le, T. (2006)
48
-------
b. Ionic Strength Effects
Although not incorporated into the model formulation, Table XIV illustrates the expected dependence of the
Fe(III) equilibrium constants on the ionic strength of the solution. Because the equilibrium constants for the
Fe(III) species dominating in the sub-neutral pH range where the project trials were conducted are affected by
factors of up to five, it is recommended that ionic strength corrections be incorporated in future modeling work.
Table XIV. Equilibrium Constants as a Function of Ionic Strength
Ionic Strength (M)
Fe3+
Fe3+
+ H20<->
Reaction
FeOH2+ + FT
+ 2H2O [Fe(OH)2]+ + 2Ff
2Fe3+ + 2H20
o [Fe2(OH)2]4+
0
logKi
logK2
log K2)2
-2
-5
-2
.19
.67
.95
0
-2,
-6,
-2,
.1
.63
.33
.95
0.2
-2.72
-6.47
-2.95
0.5
-2.79
-6.57
-2.95
0.6
-2.79
-6.57
-2.95
1
-2.72
-6.47
-2.95
Source: DeLaat and Giang Le, 2005.
c. Temperature Effects
The effect of temperature on reaction rate constants is analogous to its effect on equilibrium constants (Morel
and Hering, 1993). In general, rate constants increase with increasing temperature. As a rule of thumb, rate
constants double for every 10°C increase in temperature. The exponential effect of temperature arises from the
exponential temperature dependence of viscosity (r\). Since the rate constant is inversely proportional to viscosity,
then
k oc exp(-Ea/RT)
This dependence corresponds to the empirically derived Arrhenius equation:
k = Aex.p(-Ea/RT)
where Ea is the activation energy of the reaction and A is the preexponential factor.
From the Arrhenius plot (Figure 24),
lnk = -9770(l/T) + 26
we can adjust the rate constants from De Laat and Gallard's model rate constant for a temperature of 25°C (Table VI)
to our experimental conditions (T = 30°C). Table XV shows the temperature corrections for some of the reactions in
the model.
49
-------
Table XV. Equilibrium Constants as a Function of Temperature
Reaction
Constants
Temperature (°C)
25 30
FeJ+ + H2O <-» FeOFT
Fe3+ + 2H2O <-» [Fe(OH)2]+ + 2Ff
2Fe3+ + 2H2O <-» [Fe2(OH)2]4+ + 2F1+
Kj 2.9x10'3M 5.0xlO'3M
K2 7.62xlO'7M2 1.31xlO-6M2
K,, 0.8xlQ-3M 1.4xlQ-3M
Source: DeLaat and Gallard, 1999.
d. Role of Superoxide Radical (O2*) in Fenton's Reaction
/'. Background and Reactions Involving O2'~
A number of studies suggest that a subset of the environmental contaminants of interest, including carbon
tetrachloride (CT) and N-nitrosodimethylamine (NDMA), are transformed by a non-hydroxyl radical mechanism
that probably involves direct attack by the superoxide radical anion (Roberts and Sawyer, 1981; Stark and Rabani,
1999; Watts et al., 1999; Teel and Watts, 2002; Smith etal, 2004; Kommineni etal, 2003).
Although the superoxide radical is a relatively weak reductant (Table XVI), O2* is an effective nucleophile in
aprotic solvents (Roberts and Sawyer, 1981). As such, it may react readily with heavily halogenated targets such
as PCE and carbon tetrachloride. The reductive transformation of CT in dimethyl sulfoxide by O2* follows
second order kinetics with a rate constant, k = 3800 M"1 s"1 (Teel and Watts, 2002). However, the role of
perhydroxyl radical, HO2% in CT reduction was held to be negligible. Since the pKa for HO2* in water is 4.8, the
utility of superoxide radical as a reductant in the destruction of CT and similar (heavily halogenated) targets may
be largely confined to waters with pH > 4.
Table XVI. Standard Redox Potentials
Equilibrium
Li (aq/Li(s)
K+(aq)/K(s)
Ca (aq)/Ca(s)
Na+(aq)/Na(s)
Mg2+(aq)/Mg(s)
Al3+(aq)/Al(s)
Zn (aq)/Zn(s)
Fe2+(aq)/Fe(s)
Pb2+(aq)/Pb(s)
E°(V)
-3.03
-2.92
-2.87
-2.71
-2.37
-1.66
-0.76
-0.44
-0.13
Equilibrium
H+(aq)/!/2H2(g)
02(a(1)/02'-(g)
-pulse radiolysis (duroquinone-superoxide )
- irradiated solutions
-DMF
-DMSO
Cu2+(aq)/Cu(s)
Ag+(aq)/Ag(s)
Au3+(aq)/Au(s)
E°(V)
0
0.137a
0.13a
0.33"
0.86"
0.78"
0.34
0.8
1.5
Note: The standard reduction potential is that of the hydrogen electrode under standard conditions (1 M or 1 atm and pH 0.0). At pH 7.0, the potential of the
hydrogen electrode is -. 42 V. Solvents: dimethyl sulfoxide (DMSO), dimethylformamide (DMF).
"The standard reduction potential for the couple O2(aq)/O2'~(g) = 0.137 V was calculated from reversible electrochemical cell measurements without a liquid
junction.
Source: URL http://www.chemguide.co.uk/phvsical/redoxeqia/ecs.htmlftop (Dec 2005), except O2'" ("Petlicki and van de Yen, 1998; bSmith et al., 2004).
50
-------
/'/'. Solvent Effects on O2'~ Reactivity in Aqueous Solutions
CT degradation in solutions containing KO2 was reported to be insignificant compared to the loss in control
reactors (deionized water). Only 25% of CT was lost over 2 hrs under the experimental conditions (1 mM CT, 2
M KO2, 33 mM purified NaOH, 1 mM DTPA, pH = 14, T= 4±1°C) (Smith et al, 2004). The kinetics of this
reaction, however, are substantially improved via the addition of specific organic co-solvents (Smith etal., 2004).
In alkaline solutions (pH =14) with H2O2 (as HO2"), cosolvent enhancement of CT transformation by O2* was
reported in the following order: acetone > 2-propanol > ethanol > H2O2 (as HO2") > methanol> ethylene glycol. In
Fenton's reaction, the kinetics of CT degradation increased as a function of acetone concentration in the range
0.01 M < [Acetone] < 1 M (Smith et al, 2004). At sufficiently high concentrations (> 0.1 M), HO2" increased the
observed rate of CT transformation. It was concluded that the superoxide radical, and not HO2", initiated CT
transformation in this reaction mixture, and that HO2" increased the reactivity of O2« with CT through a Co-
solvent effect (Smith et al., 2004). Others (Peyton et al., 1995) have noted enhancement of degradation kinetics
due to ethanol addition during H2O2/UV treatment of water containing CT.
The products of CT reaction with O2* included primarily carbon dioxide and phosgene (Smith et al., 2004).
Both products are consistent with a reductive mechanism initiated via nucleophilic attack by O2* on CT.
A number of independent investigators have noted that the reaction of CT with «OH is slow (< 2xl06 M"1 s"1
Haag and Yao, 1992; <6xl05 M"1 s"1 Buxton et al., 1988). These observations support the existence of a reductive
pathway involving superoxide radicals in Fenton-based systems. Moreover, inverse relationships between rates of
hydrogenolysis of chlorinated targets and carbon-chlorine bond energies and/or the energy of formation of
chloromethyl radicals from their chlorinated parents have been repeatedly shown for a number of relevant
reductive systems (Liu et al., 2000). Carbon tetrachloride and other heavily halogenated targets are particularly
well suited to transformation via hydrogenolysis due to their low carbon-chlorine bond energies (Table XVII).
Table XVIL Carbon-Chlorine Bond Dissociation Energies for Chlorinated Compounds
Species
CC14
CHC13
CH2C12
CH3C1
CC12=CC12
CC12=CHC1
CC12=CH2
trans-CHCl=CHCl
cis-CHCl=CHCl
Abbreviation Experimental enthalpies (C-C1)
(kcal/mol)
CT
CF
DCM
CM
PCE
TCE
1,1-DCE
trans-DCE
cis-DCE
72.0 ±2.1
77.8 ±1.4
82.1 ±1.3
83. 5 ±0.9
91.0
93.3
93.8
88.7
88.2
Theoretically calculated D(C-C1)
(kcal/mol)
72.65
77.54
81.85
NA
94.52
93.56
93.67
97.33
98.98
51
-------
Table XVII (continued). Carbon-Chlorine Bond Dissociation Energies for Chlorinated Compounds
Species
CHC1=CH2
C2C16
C2HC15
CHC12CHC12
CC13CH2C1
CH2C1CHC12
CH3CC13
CH2C1CH2C1
CH3CHC12
C2H5C1
.,, ... Expenmental enthalpies (C-C1)
Abbreviation ^ ,. .. ,f
(kcal/mol)
VC
HCA
PCA
1,1,2,2-TeCA
1,1,1,2-TeCA
1,1,2-TCA
1,1,1-TCA
1,2-DCA
1,1 -DCA
CA
107.6 ±2. 3
71.2 ±3.3
68.4 ±3.6
NA
NA
NA
NA
82.8 ±2.8
79.2 ±2.7
84.4 ±0.8
Theoretically calculated D(C-C1)
(kcal/mol)
NA
68.83
68.95
74.65
70.19
76.04
73.6
82.23
79.12
84.13
Note: Experimental enthalpies for C-C1 bond dissociation were derived from enthalpies of formation. Theoretical calculations
of D(C-C1) values were performed at G2MP2 level using Gaussian 94.
Source: Liu et al., 2000.
The existence of a reductive transformation pathway for CT in Fenton's solutions was supported by
observation of CT reactivity in the presence of «OH-scavenging reactants at concentrations that should have
quenched *OH reaction with CT (Smith et al., 2004). A reductive mechanism for CT degradation was also
suggested by the results of experiments in which excess CHC13 was added to react with superoxide radicals (Teel
and Watts, 2002). Reduction pathways have been proposed for other advanced oxidation processes (Stark and
Rabani, 1999; Glaze etal, 1993).
Hi. Fenton-Driven Transformation ofPCE
Advanced oxidation processes are known to promote both oxidative and reductive contaminant
transformations. In TiO2-mediated photocatalysis, for example, PCE degradation occurs via both oxidative and
reductive pathways (Glaze et al., 1993). It was suggested that the reductive pathway involved electrons that were
elevated to the conduction band by photons and produced di-chlorinated byproducts. The oxidative pathway
involved semiconductor holes and produced both mineralized products and tri-chlorinated intermediates. Peyton
et al. (1995) observed products from reductive transformations of chlorinated target compounds in UV/ozone
reactors, and Watts et al. (1999) indicated that in Fenton systems, PCE is degraded exclusively by hydroxyl
radicals only when reducing species are consumed by a suitable radical scavenger. In soil systems, PCE
desorption was enhanced via co-solvent addition (Watts et al., 1999), suggesting that similar methods could be
used to circumvent transport/desorption limitations to PCE recovery in heterogeneous systems such as the GAC
adsorption/destruction system described elsewhere in this report.
52
-------
Our own (previous) investigation and work by others showed that PCE destruction by Fenton's reagents in
homogeneous, aqueous solutions is fast (Figure 31). Here PCE reached the method detection limit after about 100
minutes. Again, halogenated intermediates derived from PCE conversion were not detected via GC/ECD analysis
of reactor contents at any time during the experiments. The published second-order rate constant for PCE reaction
with hydroxyl radical is 2.0xl09 M"1 s"1 (www.rcdc.nd.edu). This is near the diffusion limitation. Addition of 1.0
M IP stabilized the PCE concentration in the same two-hour experiments. Results suggest that under the
conditions used here (low pH), reaction with hydroxyl radical accounts for observable PCE degradation. In this
system, under these conditions, it seems that a reductive pathway for initiation of PCE destruction can be
neglected. Reduction reactions may still play a role, however, in subsequent transformations involving reaction
intermediates.
No Cl or IP
[IP]=1M
[Cl]=9.58E-4 M
[Cl]=0.0288 M
-PCE SIMULATION
-IP SIMULATION
-[C1]=9.58E-4M
• [C1]=2.88E-2M
0
50
Time (min)
100
150
Figure 31. Effect of Cl" and IP as -OH radical scavengers on PCE transformation rate in Fenton's system. Initial conditions: [Fe(III)]T = 0.389
mM, [H2O2]0 =0.3 M, pH = 2.8, [PCE]0 = 61 ^M, [IP]0= 1 M, T = 31± 1 °C. Data points represent the experimental data and the continuous
lines the simulation using the mathematical model described previously in this report.
Consistent inhibition of PCE transformation was observed in a series of experiments with Fenton's reagents
at various concentrations of chloride ion. The experiments were originally designed to investigate the inhibitory
effect of Cl" on reactions involving «OH. The second-order rate constant for the reaction between Cl" and «OH is
S.OxlO9 M"1 s"1 (http://www.rcdc.nd.edu/) making the chloride ion a strong candidate for inhibition of Fenton-
driven oxidations. Here it seems, however, that PCE oxidation by «OH was essentially eliminated by IP addition,
leaving chloride ion no role to play. Figure 32 is included only to show the consistency of experiments illustrating
the reaction antagonism produced through IP addition.
53
-------
1.1
1
0.9 t
^ 0.8
ij °-7
Si 0.6
W 0.5
£i 0.4
0.3
0.2
0.1
0
0
• [Cl]=0 M
•[C1]=0.00198M
• [Cl]=0.02 M
• [Cl]=0.034 M
•[C1]=0.058M
50
100
150
200
250
300
Time (min)
Figure 32. Effect of Cl" concentration on PCE transformation in Fenton's reaction with IP as-OH radical scavenger. Initial conditions: [Fe(III)]T
= 0.238 mM, [H2O2]0 =0.3 M, pH = 2.2, [PCE]0 = 53^M, [IP]0 = 1 M, [Cr]0 = 0 M - 0.058 M,, T= 31±1°C.
Elimination of hydroxyl radical-dependent reaction pathways by IP addition and competition for Fenton-derived
hydroxyl radicals might have been expected. The initial concentration of IP (1 M) was more than 16,000 times that of
PCE (61 jjJVI) (Figure 31). Since reported second-order rate constants for reaction of IP and PCE are similar (Table
VII), the very high concentration of IP employed should have virtually eliminated the direct reaction of «OH with
PCE (or Cl") in these experiments. A comparison of reaction rates for PCE, H2O2 and IP with hydroxyl radical
follows:
d[PCE]
dt
d[H202]
dt
d[IP]
• = k
PCE,.OH
[PCE][-OH]
HO ,OH
dt
= k
IP,-OH
(39)
(40)
(41)
If [-OH] ss=1.0xlO-12 M and [PCE]0 = 61 nM, [H2O2]0 =0.3 M, [IP]0 = 1 M, then it is clear that the IP rate exceeds
by nearly four orders of magnitude that of PCE.
d[PCE]
dt
dt
d[IP]
dt
-i
(42)
(43)
(44)
54
-------
Since IP, PCE and H2O2 all compete for the same hydroxyl radical pool and the IP is in excess, it is expected that
the direct PCE reaction with OH will be essentially eliminated through the addition of IP.
a. Inhibition of PCE Oxidation by Cl"
The oxidation of chloroalkenes such as PCE by *OH is expected to yield primarily mineralized products (CO2 and
HC1). No chlorinated intermediates were detectable via GC/ECD during the course of these experiments. There was
no attempt to measure CO2 evolution, however. Under these conditions, the accumulation of Cl" due to target
compound mineralization may eventually inhibit «OH-dependent pathways, as radical consumption by Cl" begins to
rival H2O2 as a sink for hydroxyl radicals.
Here PCE degradation was modestly impeded at the highest concentration of chloride addition (0.0288 M).
However, at a concentration near 1 mM, Cl" had essentially no effect on PCE transformation kinetics (Figure 31).
iv. CT Transformation in the Presence of2-propanol
Here we measured Fenton-dependent CT degradation kinetics in the presence and absence of 2-propanol (IP),
hydroxyl radical scavenger (Smith et al., 2004; Watts et al., 1999). Data were compared to previously published
results and to predictions obtained using a mathematical model (described subsequently). The importance of light to
observed transformations was investigated by comparing CT reaction rates in the presence and absence of room light.
Carbonyl-photosensitized destruction of CT in the simultaneous presence of acetone and isopropanol was reported
previously (Betterton et al., 2000).
All experiments were run at 31±1°C in 65-mL (batch) threaded Pyrex tubes that were sealed with mini-inert caps
to permit sampling for residual CT without opening the reactors. Reactors were sampled 5-8 times over the course of
each experiment. The 15-jaL samples were transferred to 1 mL of heptane in GC vials for analysis via gas
chromatography using an electron-capture detector. Duplicates of each reaction mixture/treatment were provided.
When duplicate reactors provided conflicting results, experimental results were discarded and the trial was repeated.
Initial reactor conditions are provided in figure captions.
Room lighting had no effect on observed reaction kinetics (Figure 33). That is, CT transformation kinetics were
essentially independent of light/dark conditions for paired reactors in room lighting and minimal light. Thirty percent
of the CT originally present was lost over 2.5 hours in reactors containing Fenton's reagents plus IP in both the
presence and absence of room light. The disappearance of CT was clearly dependent on Fenton's reaction, as
indicated by control reactors that lacked either hydrogen peroxide or iron. In no case did light induce CT
transformation in the absence of Fenton's reagents.
55
-------
-IP+LightNoH202(l)
-IP+LightNoH2O2(2)
-IP+LightNoFe(l)
-IP+LightNoFe(2)
- IP + Light (1)
- IP + Light (2)
- IP + minimum Light (1)
- IP + minimum Light (2)
50
100
Time (min)
150
200
Figure 33. Effect of light on transformation rates of carbon tetrachloride (CT) in the presence of isopropanol (IP) as an OH radical scavenger.
Initial conditions: [Fe(III)]T= 0.26 mM , [H2O2]0=0.21 M, [CT]0= 0.81 mM, pH =2.1, T=31±1°C
Acceleration of CT transformation kinetics in the presence of 1.0 M IP was clearly evident (Figure 34). Again,
room light had no effect on reaction kinetics. In the presence of IP, CT was completely consumed in 10-15 hours. In
the absence of IP, Fenton's reaction produced -50% reduction in CT concentration during the 28-hour experiment.
With IP present, reaction kinetics were zero-order in CT for the first 10 hours of the experiment, during which time
CT removals approached 90%. No chlorinated intermediates were observed in the GC/ECD analysis.
The role of IP in Fenton-driven CT transformations remains obscure despite the careful work of Smith et al.
(2004), in which co-solvent effects were firmly established. That work was conducted at the opposite end of the pH
spectrum (pH =14) and lacked a detailed mechanistic explanation for co-solvent effects. It is possible, for example,
that IP serves primarily as a hydroxyl radical scavenger (k = 1.9xl09 M"1 s"1, Buxton, 1988; k = 1.6xl09 M"1 s"1,
www.rcdc.nd.edu), eliminating heterogeneous radical-radical extinctions involving hydroxyl and superoxide radicals
per
•OH + HO2'/O2' -» H2O/OH + O2 k = (0.71-1.01) xlO10 M'1 s'1 (33)
This explanation seems unlikely, however, in light of the low, steady concentrations of radicals in these reactors
and the presence of other effective radical scavengers, including H2O2 itself.
That is, reaction (33) would tend to lower quasi-steady levels of superoxide radicals in IP-free reactors. In this
way, IP addition might increase the observed rates of superoxide radical-dependent pathways, including CT
transformation. However, this mechanism is thought to play a minimal role in IP-dependent kinetic effects. At the
concentration provided, H2O2 is already a capable scavenger for hydroxyl radicals. The rate constant for «OH
56
-------
reactions with H2O2 is 3.3xl07 M"1 s"1 (De Laat and Giang Le, 2005). Calculations show the overwhelming majority
of hydroxyl radicals produced in these experiments were consumed by that reaction. The importance of «OH
scavenging by H2O2 on Fenton-dependent contaminant transformations was the subject of extensive previous
commentary (Ruling et a/., 1998). Thus, the contribution of the hypothesized radical-radical extinction pathway to
regulation of *OH levels in Fenton reactions is probably small compared to those of alternative scavenging reactions.
-No IP, No light
-No IP, Light
-IP, No light (1)
-IP, No Light (2)
-IP, Light (1)
-IP, Light (2)
• CT \M32noIP
Simulation
C1XOH&O2) no
IP Simulation
CT v Acetone + CT- + HC1 (46)
where CT« is the trichloromethyl radical. Betterton et al. (2000) proposed a chain reaction for CT destruction in
which reaction (46) was among the initiation steps. A reaction of this form and antecedent chain reaction could
account for the observed IP-dependent acceleration of the low-pH, Fenton-dependent destruction of CT.
Nevertheless, a series of additional experiments was performed to clarify the role of IP in promoting Fenton-
dependent CT transformations. Because superoxide radical rapidly reduced quinones (Watts et al., 1999),
benzoquinone (p-isomer k = ~109 M'V1, Chen and Pignatello, 1997) was added to a series of reaction mixtures (with
and without IP) in an effort to scavenge O2«, thus preventing its reaction with other reactor components. This line of
reasoning was, admittedly, compromised from the start, inasmuch as the reactivity of reduced benzoquinone species
with CT and CT transformation intermediates was not known. Nevertheless, experiments with benzoquinone were
conducted, primarily because a more suitable superoxide radical scavenger could not be identified.
57
-------
Benzoquinone addition decreased the rate of CT degradation relative to the Fenton system with IP and no
benzoquinone (Figure 35 versus Figure 34). Even at the relatively high concentrations used, however, benzoquinone
addition did not entirely quench CT transformation. Although results of these experiments tend to confirm the
importance of IP oxidation to the kinetics of CT transformation, they provide little of the anticipated mechanistic
insight. If benzoquinone does in fact react with superoxide radical to produce the corresponding semiquinone, that
species must be as reactive or nearly as reactive with CT as is the superoxide radical. The mechanism of IP
participation, although only modestly affected by benzoquinone, is no less uncertain as a consequence of these
experiments.
Results of these experiments with benzoquinone are in stark contrast with those in previously reported work
involving benzoquinone with PCE as the target compound. That work indicated clearly that benzoquinone addition
initially accelerated PCE transformation rates, an effect that was subsequently reversed due to quinone destruction,
presumably by hydroxyl radicals (see previous section, this report). Seemingly, Fenton-driven CT and PCE
transformations proceed via different radical mediated pathways.
-NoIP/BQ(l)
•NoIP/BQ(2)
•[IP]=1M,[BQ]=0.01M(1)
-[IP]=1M,[BQ]=0.01M(2)
•[BQ]=0.01M(1)
-[BQ]=0.01M(2)
100 200 300 400
Time (min)
500
600
700
Figure 35. Effect of benzoquinone (an O2'~ radical scavenger) onCT degradation rates in Fenton's reaction. Initial conditions: [Fe(III)]T = 0.11
mM, [H2O2]0 = 0.21 M, [CT]0=0.33 mM, pH= 2.1, T=31±1°C.
One additional observation is perhaps worth noting. Reactors containing both IP and benzoquinone experienced
no color change during the course of the experiments. Those that contained only benzoquinone became dark, perhaps
because the quinone was maintained in an oxidized state.
58
-------
Homogeneous, Bench-Scale Experiments-Summary and Conclusions
• Fenton-dependent tetrachloroethylene (PCE) degradation followed first order kinetics with a rate constant that
was about proportional to total soluble iron. The reaction proceeded with essentially no lag following the
addition of H2O2, indicating that near steady concentrations of iron species and hydroxyl radical were
established quickly.
• The initial rate of PCE degradation was increased by over an order of magnitude by the addition of
hydroxylamine hydrochloride, a strong reductant. The rate enhancement could not be sustained, however,
indicating that hydroxylamine was consumed in the reaction. The result supports the research consensus that
Fe(III) reduction to Fe(II) via H2O2 consumption is the rate limiting step in the Fenton system as
hydroxylamine can rapidly, directly reduce Fe(III) to Fe(II).
• Quinones are known electron shuttles that may facilitate iron reduction. 1,4-Hydroquinone (HQ), 1,4-
benzoquinone (BQ) and 9,10-anthraquinone-2,6-disulfonic acid all initially increased PCE degradation in
Fenton's system. The increase was proportional to the quinone concentration. However, as with
hydroxylamine addition, the rate enhancement was not sustained, suggesting that the quinones were gradually
destroyed. The PCE degradation rate stabilized at a rate that was slower than that of the unamended Fenton's
system suggesting that the by-products of quinone degradation may themselves retard the contaminant
degradation rate.
• PCE degradation was negligible when Fe(III) was replaced with Cu(II) in the Fenton system. However, when
both copper and iron were present at a Cu:Fe ratio of 2, the first order rate constant for PCE degradation
increased by a factor of 4.3. Analysis of potential chemical mechanisms for this outcome suggest that Cu(II)
is reduced to Cu(I) by H2O2 after which Cu(I) reduces Fe(III) to Fe(II). This sequence of reactions may
provide a more rapid pathway for iron reduction than direct reaction with H2O2. Although the ability of
copper to accelerate PCE degradation is modest, Cu(II) is more soluble than Fe(III), so copper may provide
kinetic benefits in the pH range where the iron concentration in Fenton's system is limited by solubility
considerations.
• The pseudo-first-order rate constant for PCE degradation increases more rapidly with increasing temperature
in the copperiron system than when iron alone is provided. Thus, the kinetic benefit of copper addition is
increased in Fenton systems operated above ambient temperature.
• A homogeneous phase, kinetic model was formulated based on earlier work by De Laat and Gallard (1999) in
which the rate constant for disproportionation of the Fe(III)-hydroperoxy complex (the rate limitation for
radical production kinetics) was fitter to H2O2 utilization data.
• As chlorinated VOCs are degraded in Fenton systems, chloride anions will build up in the regenerant
solution. The rate constants for the reaction of hydroxyl radical with chloride ion indicates that chloride
59
-------
accumulation through repeated carbon bed regenerations will retard VOC degradation rates. However, model
simulations using the literature rate constant for the «OH/C1 reaction overestimated chloride ion effects in
experimental trials. A revised *OH/Ci rate constant was fitted on the basis of these data. Our second-order
rate constant was lower than the previously reported value by a factor of 3.
The rate of PCE degradation by Fenton's reaction increases with increasing pH in the range 1< pH <3. Above
pH 3, iron solubility limits free iron concentration availability lowering the rate of PCE degradation.
Observed PCE degradation rates at 0.9 < pH <3 were compared to the results of mathematical simulations.
There was considerable difference between experiment and simulation although general effects of pH were in
agreement. At every pH (except pH = 0.9), the model over-predicted the PCE transformation rate.
The rate of Fenton-dependent carbon tetrachloride (CT) degradation was increased in the presence of
isopropanol (IP), a *OH scavenger. The work suggests that the mechanism of CT degradation involves direct
reaction with superoxide radical (O2*) A more complete mechanism is under investigation. The increase in
rate may be due to an IP co-solvent effect that increases O2* activity.
PCE degradation by Fenton's reagents is negligible in the presence of IP, an «OH scavenger, indicating that
PCE destruction is initiated by hydroxyl radical (*OH) attack. No chlorinated intermediates were detected in
this experiment. PCE degradation diminished modestly at the highest concentration of chloride addition
(0.0288 M). A concentration near 1 mM, Cl" had a negligible effect on PCE transformation kinetics.
60
-------
Bench-scale, Heterogeneous, Column Experiments
Adsorption Isotherms
Freundlich isotherm parameters for GAC (URV-MOD 1) adsorption of all compounds tested are summarized in
Table II (repeated below). The data for adsorption of methylene chloride, a representative contaminant, is provided as
Figure 36.
Table IL (Repeated from earlier section.) Chemical Properties of the Organic Compounds Studied
Name
Methylene Chloride (MC)
1,2-DCA
1,1,1-TCA
Chloroform (CF)
Carbon Tetrachloride (CT)
TCE
PCE
Formula
CH2C12
C2H4C12
C2H3C13
CHC13
CC14
C2HC13
C2C14
Log Kowa
1.15
1.47
2.48
1.93
2.73
2.42
2.88
i r^-ff • -i. Adsorbed
KM OH Diffusivity , ,.
,, ,:r "-b / 2/ xc concentration
(Ms) (cm /s) , . xd
(mg/g)
9.00E+07
7.90E+08
l.OOE+08
5.00E+07
2.00E+06
2.90E+09
2.00E+09
1.21E-05
1.01E-05
9.24E-06
1.04E-05
9.27E-06
9.45E-06
8.54E-06
275
146
20
188
N/A
103
11
Freundlich Parameters6
K (mg/g) 1/n
(L/mg)1/n
0.07
0.04
0.65
1.48
12.30
5.82
45.66
1.06
1.33
0.87
0.77
0.59
0.70
0.56
y = 1.06x-1.16
R2 = 0.97
Note: kM OH is the second-order rate constant for the reaction of hydro xyl radical with the target organic compound.
Source: "Swarzenbach et al., 1993; bwww.rcdc.nd.edu, except carbon tetrachloride (Haag and Yao, 1992); 'calculated from Wilke-Chang equation (Logan, 1999);
''From analysis of initial carbon concentration for carbon recovery experiments (data at 32°C). 'From isotherm data obtained in this lab at 32°C.
3.5 -,
3.0 -
2.5 -
2.0 -
1.5 -
1.0 -
0.5 -
0.0 -
-0.5 -
-1.0 -
0.0
0.5
1.0
1.5 2.0 2.5
Log Ceq (mg/L)
Figure 36. Isotherm data for methylene chloride on URV-MOD 1 (32°C).
3.0
3.5
4.0
61
-------
Bench-scale Column Experiments
Recovery using Fenton's Reagents
Contaminants with a range of compound hydrophobicities and reactivities with «OH (Table I) were selected for
Fenton-driven regeneration experiments in which spent GAC was regenerated in columns. Preliminary carbon
recovery experiments were run at 24 ± 2°C for seven chlorinated VOCs: (methylene chloride (MC), 1,2-
dichloroethane (DCA), chloroform (CF), 1,1,1-trichloroethane (TCA), carbon tetrachloride (CT), TCE and PCE). The
same experiments were repeated at 32° C. Carbon recovery data for MC, CF and TCE at 32°C are summarized in
Figure 37. Recovery kinetics followed the order MC>CF>TCE. The transformed data are shown on a semi-log plot in
Figure 38. After an initial period of rapid recovery that lasted 1-3 hours, further reductions in the sorbed concentration
conformed to first-order kinetics. TCE removal from GAC was only 50% complete after 14 hours. This was
unexpected since the second-order rate constant (kM,oH, Table II) for the reaction of hydroxyl radical with TCE
(2.9E+09 M'V1) is near the molecular-collision diffusion limit (www.rcdc.nd.edu) and is higher than the rate
constants for reaction of *OH with MC and CF. Lack of dependence of recovery kinetics on reaction rate with «OH
suggests that the kinetics of Fenton-driven recovery of GAC is controlled by mass transport, as opposed to the rates of
hydroxyl radical generation or radical reaction with contaminant targets.
5 10
Time (hr)
Figure 37. Fractional removal of adsorbate from GAC for MC (•), CF (4) and TCE (•). Fractional q/q0 represents the mass of contaminant
remaining in the carbon. The regenerant solution contained 10 mM iron, pH = 2.0, and 0.15 M H2O2 average concentration throughout the
experiment. Temperature was controlled in the reservoir at 32°C. The lines are a smoothed fit to the data. Average error bars are indicated for
each curve. n=2 for each data point (average between top and bottom samples).
The shapes of the recovery curves (Figure 37) are consistent with an intraparticle diffusion limitation: at short
times, target concentrations in most of the GAC pores were nearly uniform, until concentration profiles developed
along particle radii. With time the concentration gradient penetrated further into the particle increasing the apparent
diffusion length and leading to slower, nearly first-order contaminant removal kinetics. The apparent first-order rate
constants derived from the latter portion of each experiment are summarized in Table XVIII.
62
-------
Time (hr)
Figure 38. Semi-log plot of the data in Figure 37forMC (4), CF (A) andTCE (•). The slopes of lines from about t= 2 h to the end of the
experiment were used to calculate the observed rates (kobs).
Experimental results (Figure 38 and related discussion) suggest that mass transfer rates limit the slow recovery
stage in GAC regeneration under the conditions employed in this work. If intraparticle diffusion controls the overall
recovery rate, the concentration of the target compound in the bulk liquid should be low due to rapid consumption via
Fenton's reaction in the bulk liquid phase. In that case, it can be shown that the flux of contaminant from the particle
surface (J) follows a relation like
T = tr- (47)
where km is a hybrid, conditional coefficient and Cp is the average concentration of the target compound in the liquid
that fills the pores of the particles. If adsorption and desorption rates are fast, the local aqueous- and solid-phase
concentrations of the adsorbate are in equilibrium so that:
C = -2-
K
(48)
When the Freundlich isotherm is linear (n=l), equation 48 can be expressed in terms of average concentrations so
that:
K
(49)
without error. For the analysis that follows, it was assumed that average concentrations could be inserted in the non-
linear isotherm (equation 48) without generating excessive error, so that
(50)
63
-------
Table XVIH. Summary Table for Rates Observed (kobs) at 24°C and 32°C, and Cost Estimates at the Bench and Field Scales
Compound
MC
1,2-DCA
1,1,1-TCA
CF
CT
TCE
PCE
Bench-scale3
T=24°C
0.17
0.10
0.075
0.066
0.052,
0.060
0.045
0.037, 0.062
kobs (hr
Bench-scale15
T= 32°C
0.25
0.24
0.058
0.11
N/A
0.027,
0.060
0.042
')
Bench-scale
Desorption0
T= 32°C
0.19
N/A
N/A
0.11
N/A
0.043
N/A
Field-scale"1
0.12
N/A
N/A
0.059
N/A
0.015
0.033
0.068
0.011
Cost ($/kg
Bench- Scale
T= 32°C
4.6
N/A
N/A
3.50 (93%)
N/A
5.14
(52%)
2.63(35%)
GAC)e
Field-scale
6.63 (100%)
3.57(97%)
N/A
N/A
6.63 (93%)
N/A
2.55(73%)
4.53(82%)
6.54(95%)
6.54(50%)
Note: aBench-scale regeneration column experiments at 24°C (Figure 38), and bT= 32°C (Figure 37).
°Bench-scale desorption experiments (clean water, no Fenton reagents as eluant - Figure 39).
dField-scale regeneration column experiments, tost is based on hydrogen peroxide consumption
for spent GAC regeneration. The number in parenthesis indicates the percentage GAC recovery for each trial.
We recognize, however that the magnitude of error introduced by this approximation increases with the degree of
non-linearity in respective isotherms.
A mass balance on the adsorbate in the carbon particles, assuming that aqueous-phase mass can be neglected, then
yields:
s
dt
where M and As are the total carbon mass and total external surface area of GAC, respectively, and q is the average
concentration of the adsorbate on the carbon surface.
The preceding equation can be solved with the initial condition
(52)
Substituting equations (47) and (50), equation (51) becomes
dt s m\ K
(53)
64
-------
This equation implies that the initial rate of decrease of the average target concentration in the solid,
dq
dt i
at t = 0, should be directly proportional to (l/Kn) for all target compounds, assuming that km does not vary appreciably
among the various targets. A rigorous analysis would show that km is equal to the observed pseudo-first order rate
constant for target removal (kobs).
The preceding analysis motivated us to plot the observed pseudo-first order rate constant for target removal (kobs)
vs. l/Kn for all target compounds (Figure 39). The fact that most compounds follow the expected trend (positive
correlation between l/Kn and initial rate of adsorbate loss) is consistent with the hypothesis that intraparticle mass
transfer controls the overall carbon recovery process.
1 -,
0.1 -
0.01
PCE
0.001
0.010
0.100
1.000
10.000 100.000
(g/mg)n(mg/L)
Figure 39. Correlation between first-order observed rate constant (k^) and compound-specific (l/K)n. Rates for a 14-hour regeneration period
with 10 mMiron, 0.1-0.15 MH2O2, pH = 2.0. Experiments were conducted at room temperature (24°C) - closed symbols and 32°C -open
symbols. The line is a linear fit of the log- transformed data.
Clean Water and Fenton-Driven Recovery Experiments
Comparison of clean water (no Fenton's reagents present) and Fenton-driven recovery experiments (Figure 40)
indicate that carbon recovery trajectories for the Fenton-driven and no-reaction (clean-water elution) cases matched
for MC and TCE. For those contaminants, it seems evident that contaminant reaction did not limit recovery kinetics.
For CF, however, degradation in the presence of Fenton's reagents was slower than in the clean water circulation
experiments. This occurred because the bulk aqueous-phase concentration was maintained near zero in the latter
experiments by continuously feeding clean influent to the column reactor. Since the reaction of *OH with CF is
relatively slow (Table II), CF accumulated to some extent in the bulk liquid when Fenton's reaction was relied upon
to eliminate CF from the recirculated fluid. This diminished the driving force for CF transport from the GAC and
protracted the recovery period. This interpretation is supported by liquid-phase CF measurements during GAC
65
-------
recovery using Fenton's reagents (Figure 41). It is apparent that recovery kinetics were limited, at least in part, by the
rate of reaction of bulk liquid phase CF.
1.2 -,
Eluant used
Water, open symbols
Fenton's reagents, solid symbols
0 5 10
Time (hr)
Figure 40. Comparison of recovery rates using eluant solutions with and without Fenton's reagents. When eluant consisted of water, bulk
aqueous phase contaminants were near zero. Target compounds were MC (•), CF (»), and TCE (•). Regenerant solutions contained 10 mM
iron, 0.15 M average H2O2, pH = 2.0, T = 32°C. Concentrations are normalized by the initial contaminant concentration on the carbon (q/q0).
Data points represent the average value between carbon extractions of GAC samples from the top and bottom of the GAC bed. An average error
bar is indicated for one data point for each curve.
As indicated previously, the Fenton solution was replaced with a pure water eluant for selected trials. That is, the
external (bulk aqueous-phase) concentration of adsorbate was maintained near zero by passing water through the
expanded carbon bed rapidly and discarding the once-used eluant. In these trials, 10-gram samples of carbon were
preloaded with MC (CH2C12) or TCE. Initial contaminant loadings were 291 and 69 mg VOC/g GAC, respectively.
Carbon was transferred into the regeneration column, where tap water was flushed through at 950 mL/min to expand
the carbon bed. Figure 42 shows the solid's time-dependent concentration profiles for MC and TCE.
Figure 41. Liquid phase reservoir concentrations for CF-loaded GAC recovery. Times of data points correspond to those in Figure 40.
66
-------
• TCE -*- CH2CI2
o
in"
O
t/)
O
\.Z. -
1.0 i
0.8-
0.6-
0.4-
0.2-
o.o -
(
i
\ \
\ 'i i T
| »
\ *
1 •. H
1 ||.* ~' " •
1 * * * T
1 T"*"""'--- i
\ X X """*""••.
\ T""""1---
'
i
\
, , ^ , ~~m — — , ii
) 2 4 6 8 10 12 1
Time (hrs)
Figure 42. 14-hour carbon recovery profiles for CH2C12 and TCE in expanded-bed columns using fresh water (29°C) flushing to remove the
VOC from the bulk liquid. n=2 for each data point (average between top and bottom samples).
Again, pseudo-first-order kinetics were observed. Respective first-order recovery constants for CH2C12 and TCE
removals were 0.219 and 0.043 hr1 (Table XIX).
As in the Fenton-dependent trials, a correlation between carbon recovery and compound solubility (see Log Kow,
Table II) was evident. Recovery of the more soluble compound (MC) was 98% over a 14-hour period, versus 51%
recovery of TCE.
Table XIX. Expanded-Bed Recovery Data
Compound
Methylene Chloride (MC)
1,1,1 -Trichloroethane (TCA)
Trichloroethylene (TCE)
Reaction
Whr1)
0.242
0.291
0.041
% Mass
Destroyed*
97
67
52
Flushing
k^flif1)
0.219
0.223
0.043
% Mass
Removed
98
48
51
k.QH,R
(M'V1^
9.0xl07
l.OxlO8
2.9xl09
* - For 14-hour exposure to 10 mM iron and 7500 mg/L H2O2
#-From Buxton( 1988)
Recovery profiles for the Fenton-dependent trials and those in which fresh water was flushed through (without
recirculation) are superimposed in Figure 43. Recovery kinetics were similar (independent of eluent composition) for
both compounds. This suggests that desorption or mass transfer, as opposed to aqueous-phase reaction kinetics,
limited overall GAC recovery rates for both MC and TCE.
67
-------
- — - CH2CI2 flushing
- CH2CI2 - — - TCE flushing
-TCE
6 8
Time (hrs)
10
12
14
Figure 43. 14-hour carbon recovery profile comparison for CH2C12 and TCA in expanded bed columns. Fenton-dependent and water flushing
trials are presented.
Fixed-Bed Recovery Trials—Effect of Particle Size
To verify the importance of diffusive transport to overall recovery kinetics, the effect of particle size on recovery
rate was examined. Recovery data were collected over a 250-minute flushing period (with clean water) for GAC
preloaded with chloroform. Chloroform is relatively soluble in water with a moderately non-linear adsorption
isotherm (Table II). Despite operating at a significantly lower flow rate than the expanded-bed design (200 mL/hr
versus 950 mL/hr) the assumption of fast convective mixing in the bulk was maintained, and diffusion limitations
through the hydrodynamic boundary layer surrounding carbon particles were neglected. Dependence of carbon
recovery rate on particle size (other being factors equivalent) was taken as an indication that pore-volume or surface
diffusion limits the carbon recovery process. That is, decreasing pore length/particle diameter would result in faster
contaminant removal only if the recovery process were limited by intra-particle mass transfer.
Time dependent residual mass loadings on sieved GAC are represented for each particle size range tested (Figure
44). The smallest particle range (1.0-1.18 mm diameter) yielded the fastest carbon recovery rate (apparent k: 0.226
hr"1). Recovery rates decreased as particle size increased (1.4-1.7 mm: 0.204 hr"1; 2.0-2.38 mm: 0.129 hr"1). Results
suggest that pore or surface diffusion has an effect on chloroform removal kinetics from Calgon URV-MOD 1 carbon
in Fenton driven recovery systems.
68
-------
— 1-1.18 mm —1.4-1.7 mm —2.0-2.38 mm
1.0
Time (hrs)
Figure 44. Fixed-bed carbon recovery kinetics for three discrete size distributions of GAC, preloaded with chloroform and normalized to the
respective loading concentration.
Theoretical Considerations - Modeling
In aqueous-phase GAC applications, effects due to Knudsen diffusion are generally considered negligible as it is
assumed that surface tension limits the availability of micro-pores that approach the molecular size of the
contaminant. Pore-volume diffusion is an intrinsic property of the GAC-adsorbate system and is often the rate-
limiting step in the sorption mechanism (Furuya et al., 1996). However, it has been established by several researchers
(Snoeyink, 1990; Weber 1972; DiGiano and Weber, 1973) that a parallel transport mechanism (surface diffusion) is
often needed to characterize the complicated adsorption process (Furuya et al, 1996). The relative contribution of
each mechanism to the overall effective diffusivity is a function of several factors. For example, unlike pore-volume
diffusion, which is always present, the importance of surface diffusion depends on the affinity of the adsorbate to the
adsorbent (Furuya et al., 1996). It has been shown (Suzuki, 1990) that although surface diffusivities are often two
orders of magnitude smaller than pore volume diffusivities, a large surface concentration gradient, such as those
experienced with high-affinity solutes, can result in a surface flux that is much greater than the contribution of pore
diffusion to overall mass transfer (Ma et al., 1996). Therefore, a general transport function in which pore and surface
diffusion are individually characterized is generally employed.
In both aqueous-phase and surface diffusion, transient contaminant fluxes obey Pick's second law:
— = DV2C
dt
(54)
69
-------
where C is the concentration of diffusing substance, and D is the diffusion coefficient (Crank, 1975). Here we
approximate a GAC particle to a sphere. The following equation results from a mole balance of contaminant in the
pore space of the GAC particle:
8Cp Def^p 8 . 2dC
~^T = ^^^(r ^~^~a*'J*
ot r or or
where Cp is the local concentration of contaminant in the liquid within the pores, Defp is its effective (aqueous-phase)
diffusivity in the pores, at is the surface area of solid per unit volume of pore space, and Js is the interfacial flux of
contaminant, expressed as moles of contaminant transferred to the solid phase per unit time and per unit solid surface
area.
Analogous to pore-volume diffusion, a spherical GAC particle is assumed for the characterization of surface
diffusion. For adsorbed contaminant a similar mass balance results in
(56)
dt r dr dr
where Cs represents the concentration of contaminant adsorbed on the solid surface in moles per unit solid surface
area, Defs is the effective surface diffusivity, and the interfacial flux of contaminant is again represented as Js.
Here, we will make the assumption that adsorption and desorption processes are much faster than diffusive
transport. Under these conditions, we can assume that the solid and the liquid in the pore space are at equilibrium
(Figure 45). Therefore, the interfacial flux of contaminant, Js, is governed by the Freundlich isotherm, and represented
as follows:
C,=K-Cp-» (57)
where K is often interpreted as the adsorbent capacity ((mg/g)(L/mg)1/n), and 1/n the relative strength of adsorption. A
smaller value of 1/n, suggests a stronger adsorption bond (Letterman, 1999). The Freundlich parameters K and 1/n
have no theoretically derived physical significance, however, and must be determined empirically. Compound-
specific Freundlich parameters were experimentally determined on URV-MOD 1 carbon, as presented earlier.
70
-------
"""j^ pores
pores •
n
&.3&SSKf'*%
:-Wv-ft;Xft
-------
individually based on the fitting parameter, D. For example, assuming pore diffusion is the dominant transport
mechanism, (60) may be interpreted as,
£> = - ^^ (61)
However, when surface diffusion cannot be ignored, equation (60) can also be simplified to the following
equation by assuming a similar tortuosity, T, for both pore and surface diffusion,
D = Ba2L (62)
The initial and boundary conditions are derived in part from the experimental setup, which typically provides fast
convective mixing in the bulk liquid phase (negligible film resistance), a negligible bulk liquid-phase concentration
(rapid consumption of desorbing contaminant in the bulk liquid-phase) and negligible contaminant concentration on
the GAC particle exterior surface (for maintenance of equilibrium with the bulk liquid-phase concentration). That is,
r=e[0,/Z] t = 0 (63)
Cp(r,t) = 0 r = R f=e[0,oo] (64)
dC,
8r
. = 0 r = Q ^=e[0,ooj (65)
where Cs 0 is the initial solid contaminant concentration (mg/m2), Cp the aqueous-phase intraparticle contaminant
concentration (mg/L), and K and 1/n the empirical Freundlich parameters.
An analytical solution is available for equations (59-65, for n=l only), which yields the aqueous bulk
concentration. The intraparticle liquid-phase concentration profile (generated by (55)) can be translated to a solid
loading by assuming that equilibrium exists throughout the particle and integrating over the particle volume. The
loading so obtained can be represented in terms of a volume-averaged effective liquid-phase concentration in a single
GAC particle. Following a contaminant mass balance on the system, solution for the bulk aqueous-phase contaminant
concentration yields,
n 'W • D't t-\ j.
(66)
7
V,
C -V-LLV/4
U' ~ L, r> 2 L,An,}
-e
where Q is the bulk aqueous-phase concentration (mg/L), /*,- is the particle population fraction, R} the population of
particles with radius R, Q the aqueous flow rate (m3/min), Vt the reactor liquid volume (m3), and An the consolidated
constant represented as
72
-------
6-C^-D-a,^-M
"'J ~ O n2-n2-D
/v
The consolidated term includes the previously determined average solid concentration Cs 0, Mthe mass of GAC
(g), the particle porosity 0, pap the particle density (g/m3), and D the fitting parameter. It was determined from
experimental data that it was advantageous to periodically sample the liquid effluent from the reactor instead of the
residual contaminant mass on the solids, as a greater sampling frequency could be achieved while introducing less
experimental error due to depletion of carbon mass in the bench-scale reactor. It is important to note that although the
liquid-phase contaminant was measurable, the bulk liquid-phase concentration remained very small compared to the
contemporary intra-particle concentrations, validating (approximately) the selection of the bulk liquid-phase boundary
condition (equation (62)).
Model calibration was based on fitting T using experimental data for CH2C12 and TCA desorption. That is, T was
selected to minimize the sum of squared error between model output and experimental results (time-dependent liquid-
phase concentration of contaminant in the reactor effluent). If the effective diffusivity is dominated by pore and
surface diffusion, then T can be deduced from the effective diffusivity by (22) (assuming a similar T for pore and
surface diffusion). T should remain constant (independent of sorbate identity) if pore-volume and surface diffusion are
the dominant transport mechanisms for all sorbates. Furthermore, a single T should be experienced independent of the
residual contaminant concentration on the solid if an energetically homogeneous (no range of binding energies)
surface exists.
Pore Diffusion Modeling
For compounds with a near-linear isotherm on URV-MOD 1 carbon (i.e., CH2C12, l/n=1.0; TCA l/n=0.93; Table
II), an analytical solution (64) to the apparent diffusion model (57) is available. Tortuosity (i)-dependent curves are
compared to experimental recovery data by partial differential analysis using MathCad software. The physical
properties of Calgon URV-MOD 1 carbon are considered uniform with respect to porosity and internal surface area
(Table III) throughout the diameter range 0.6-2.38 mm (the range of particle sizes used in experiments). The actual
pore volume is approximated as the sum of all pore volumes, 0.643 mL/gGAC, and includes the volume of
micropores (Table III). The result is a constant surface area to pore-volume ratio, at (2*109 m"1), across all particle
sizes. As a result, an error in D is anticipated, as a potentially significant fraction of the micro-sized pores may be
unoccupied because of surface tension effects in liquid-phase application (Knudsen effects).
73
-------
A series of contaminant recovery simulations was developed for desorbed CH2C12 by varying the single fitting
parameter, T (Figure 46).
2.5 ft—r-
10 12 14 16 18 20 24 26 28-*- Trial 1 - «- Trial 2
1.5 -
O
O
1 -
0.5 -
YV\
1
'
0.5 2.5 4.5 6.5 8.5 10.5 12.5 14.5
Time (min)
Figure 46. A comparison of experimental and modeled, bulk aqueous-phase concentration profiles for CH2C12 where the single fitting
parameter, T, was varied.
For CH2C12, model predictions are compared to the two experimental data sets obtained using modestly different
loading concentrations. Although initial surface loading concentrations differ slightly, transport effects associated
with initial concentration differences are considered negligible. Therefore, the difference in recovery rates between
the data sets is probably due to experimental error, and a minimized sum-of-squares method was applied to select an
appropriate simulation curve (T = 26) using the experimental average. Notice that selection of this value depends on
an assumption that surface diffusion can be neglected (equation 61). Any appreciable role for surface diffusion would
result in a larger fitted tortuosity value. Column washout due to liquid loading and contaminant flux by non-diffusive
mechanisms was accounted for by fitting modeled profiles to data obtained after a few liquid retention periods had
passed. Based on a column residence time of approximately 4 seconds, it is assumed that a uniformly distributed flow
and complete mixing will occur within five bed volumes. However, a more accurate determination of the physical
characteristics of column washout, and the hydraulic detention time is desirable. Figure 47 shows the normalized
bulk-phase concentration profiles for the two CH2C12 data sets against the model prediction over a 15-minute flushing
period.
74
-------
2.5
2
O 1.5
in"
O
O !
0.5 -
- Trial 1
-Tortuosity of 26
- Trial 2
1 -»-.
6 8
minutes
10
12 14
Figure 47. The normalized, bulk aqueous-phase contaminant concentration versus time-comparison for experimental CH2C12 data and results
of the pore diffusion model.
A T of 26 minimized the squared error between experimental and modeled recovery profiles, although the pore
diffusion model predicted contaminant removal slightly after approximately seven minutes.
For TCA, the isotherm was approximated as linear (l/n=0.93), and only a single experimental data set was
available. Carbon properties and a column flow rate consistent with the CH2C12 simulations were used in the model. A
value of 0.65 L/g is used for the isotherm fitting parameter, K, (Table II) and a series of i-dependent recovery curves
was again developed for the pore diffusion model. Here, a T of 129 minimized the relative error between experimental
data and the pore diffusion model (Figure 48). Again, this value is based on an assumption that surface diffusion does
not contribute to intra-particle mass transfer (equation 61). A similar washout period (30 seconds) was used to justify
rejection of the first few data points based on residence time, and the recovery process was modeled over a 250-
minute flushing period.
- Experimental
Tortuosity of 129
0 50 100 150 200
minutes
Figure 48. A comparison of bulk aqueous phase contaminant concentrations between experimental TCA data and a pore diffusion model.
75
-------
Initially, a large discrepancy exists between the measured and simulated TCA profiles with the model over-
predicting contaminant removal rate over the first 30 minutes, and under-predicting thereafter. Like the profile
experienced with CH2C12, the fitted curve for TCA does not provide an adequate physical representation of
contaminant loss over the entire removal period, suggesting that contaminant removal cannot be accurately predicted
with a pore diffusion model of the form used here. Also, it is evident that a single T cannot be utilized to account for
transport-limited recovery of both compounds (T0pt=26 for CH2C12; Topt=129 for TCA).
A remarkably broad range of T has been reported for activated carbon macropores, ranging from 5-65 (Yang,
1987). From equation (58) it is apparent that the accuracy of fitted T values can be no greater than the accuracy with
which K (the capacity parameter in the Freundlich isotherm) is estimated from equilibrium data. Based on a linear
regression for the CH2C12 isotherm data, a 95% confidence interval corresponds to a range of adsorbent capacities of
0.036
-------
2.5
2 -
9. 1-5
O
O !
0.5 -
- Trial 1
-Tortuosity of 1920
- Trial 2
10 12
14
minutes
Figure 50(a). A normalized, bulk aqueous-phase contaminant concentration comparison between experimental CH2C12 data and a pore and
surface diffusion model.
A similar analysis applied to the minimized sum-of-squares recovery profile for TCA resulted in a T of 129,530
(Figure 5Ob).
- Experimental
Tortuosity of 129,530
50
100
minutes
150
200
Figure 50(b). A comparison of bulk aqueous-phase contaminant concentrations between experimental TCA data and a pore and surface
diffusion model.
Like the pore diffusion model, it is evident that a single T cannot be utilized to account for transport-limited
recovery of both compounds (T0pt=1920 for CH2C12; T0pt=129,530 for TCA). On this basis, it seems likely that an
intraparticle diffusion model cannot adequately account for observed transport kinetics during carbon recovery in the
desorption experiments. This conclusion is based on the inability to model linear isothermal contaminants for CH2C12
and TCA with a single T using the pore and surface diffusion model described. Transport limitations other than those
arising from intraparticle diffusion, such as adsorption/desorption kinetics, may produce the deviations between the
77
-------
experimental and predicted curves. Although an assumption of local equilibrium (fast sorption kinetics) is often
employed in intraparticle transport mechanisms, the validity of this assumption is seldom validated. For example,
comparison of surface diffusion and desorption energy barriers has not been adequately considered for systems such
as these. Furthermore, it seems unlikely that an energetically homogenous surface exists with a uniform heat of
adsorption. A reasonable view is that variation in Def:Sis an outcome of variation in heats of adsorption, and that D^ is
in fact a sensitive indicator of energetic surface heterogeneity (Carman, 1956).
In addition, although accounting for both pore and surface diffusion, an apparent diffusion model may be
insufficient for predicting concentration effects over a broad range of experimental conditions. The apparent
diffusivity determined at a particular temperature and concentration has very limited applications for other conditions
(Furuya et a/., 1996). This may result from the fact that the pore and surface diffusion mechanisms require different
values of T due to an energetically heterogeneous surface. When surface flux is relatively high (high affinity solutes),
a diffusion model would then be inadequate for characterizing removal kinetics. For compounds with limited
solubility, for example, intraparticle aqueous concentrations of the sorbate may be much lower than their equilibrium
levels. Under these circumstances, surface diffusion may be the predominant mechanism of sorbate transport to the
particle exterior. Surface effects resulting from pore volume distribution may prove to be significant as well. As pore
dimensions approach the molecular size of the contaminant, it is anticipated that surface diffusion mechanisms limit
transport kinetics. It is known from literature that in micropores found in zeolites (Ruthven, 1984), surface diffusion is
the dominant transport mechanism, while pore diffusion dominates in macropores (Ma et al., 1996).
Results presented in this section lead the investigators to conclude that mass transport mechanisms limit the
effectiveness of Fenton's reaction for carbon recovery, at least for slightly soluble compounds that are reactive with
Fenton's reagents. Therefore, optimal design for this type of treatment would maximize contaminant flux from the
sorbent while minimizing the use of H2O2, the primary contributor to process cost. Just how this is done will probably
be compound specific.
Role of Iron Phase (Precipitated vs. Dissolved)
The following section evaluates the effectiveness of loading iron onto the surface of GAC prior to the Fenton
treatment. Localizing the reaction on the carbon surface was considered as a means to increase contaminant
destruction efficiency and minimize reagent use. That is, it was hypothesized that radical production in the immediate
vicinity of adsorbed contaminants would allow a higher proportion of radicals to react with the target compounds as
opposed to H2O2 and other competitors. In addition, the eluant reservoir size was reduced to decrease non-productive
(outside the column) H2O2 consumption.
To test this hypothesis, native and iron-amended GAC were used in parallel column regeneration trials. Both
types of GAC (15 g each) were loaded with TCE and regenerated. Results were compared with previous column
experiments with iron in solution. After 14 hours, TCE recovery was essentially the same in all three situations
78
-------
(Figure 51). Table XX summarizes the results, as well as column iron content, and volume of H2O2 used in each trial.
Precipitating the iron on the surface of the GAC lowers the use of H2O2, which is the primary cost associated with the
GAC regeneration process (see later section on process cost estimation). Iron addition to the carbon surface,
however, provided little advantage in terms of carbon recovery kinetics.
0
10
Time (hrs)
Figure 51. TCE Recovery with background GAC, iron amended GAC and iron in solution.
15
Bench-scale PCE regeneration column experiments (Table XXI) were conducted using a lower iron loading than
in the TCE trials (Table XX). In this set of experiments, the effect of pH for PCE recovery was established. Iron-
amended GAC was loaded with PCE and regenerated at pH 2 and 3 (Figure 52). As in the TCE trials, PCE destruction
efficiencies with iron-amended GAC and with iron in solution (pH 2) were very similar, however H2O2 use was
reduced by almost a factor of three (Table XXI). Results also suggest that pH is not an important process parameter in
the range 2.0
-------
Unfortunately, enhancement of degradation rates for compounds that are diffusion or desorption limited has not yet
been achieved. Alternative methods, such as use of alternative solvents and GAC with a greater proportion of surface
area in macropores, should be explored to overcome the kinetic limitations arising from mass transport.
o
0
0
1.2 n
1.0 <
0.8
0.6
0.4
0.2
0.0
(
^**^*^*-_ ___.
4 ^^ +
A Iron in solution, pH =2
• Iron on GAC, pH = 2
V Iron on GAC, pH = 3
I I
D 5 10 15
Time (hrs)
Figure 52. pH effect on iron-amended GAC loaded with PCE in column regeneration trials.
Table XXI. Observed Degradation Rate Constant and Average Hydrogen Peroxide Use for PCE-Laden GAC Regeneration With and Without
Iron-Amended to the Carbon
pH/Iron
2/Iron in solution
2/Iron-amended GAC
3/Iron-amended GAC
kobsChr-1)
0.024
0.025
0.034
H2O2 added (mls/hr)
16
5.6
5.6
Note: *Uncertainty about Iron content (not accurately measured)
Heterogeneous, Bench-Scale Experiments - Conclusions
• Ethyl acetate was the best solvent tested for solid: liquid extraction of PCE from GAC. An extraction period
of 12 hours provided nearly complete recovery of adsorbed PCE.
• Degradation of adsorbed chlorinated VOCs in heterogeneous Fenton's systems indicated that carbon recovery
kinetics was bi-phasic. A fast initial degradation phase was followed (after 1-3 hours) by a slower second
phase. The fraction of contaminant degraded in the initial rapid phase increased as the aqueous-phase
solubility and (l/K)n value of the contaminants increased. Results suggest that intraparticle diffusion of the
80
-------
contaminant limits recovery kinetics during the slow phase treatment, for at least a subset of the compounds
tested.
When GAC was loaded with CF and recovery was accomplished by rapidly flushing the carbon with clean
water, the rate of CF recovery was inversely related to the size of carbon particles. Results suggest that pore
and/or surface diffusion, and thus particle size, affect the overall removal kinetics for chloroform from
Calgon URV-MOD 1 carbon in Fenton driven systems.
A mass transfer model in which local solid: liquid equilibrium is assumed between the carbon surface and
proximate pore was formulated to simulate mass transfer and recovery of carbon adsorption capacity. An
analytical solution was developed for the special case in which contaminant adsorption was governed by a
linear Freundlich isotherm (n « 1). A single fitting parameter (tortuosity) brought recovery simulations and
data into reasonable agreement. However, a common tortuosity could not be obtained for all compounds,
suggesting that surface diffusion and/or desorption effects can also limit GAC recovery kinetics.
Based on the combined experimental and model results, it is concluded that mass transport mechanisms can
limit the effectiveness of Fenton's reaction for carbon recovery, at least for slightly soluble compounds that
are reactive with Fenton's reagents. Therefore, optimal design for this type of treatment would maximize
contaminant flux from the sorbent while minimizing the use of H2O2, the primary contributor to process cost.
Trials were conducted using Calgon URV-MOD 1 carbon on which iron had been precipitated onto the pore
and outer surfaces. No iron was added to the bulk regenerant (Fenton's solution). It was hypothesized that this
would localize the Fenton-driven radical generation near the GAC surface, in the vicinity of target
compounds, and potentially avoid rate limitations due to pore and surface diffusion and/or compound
desorption from the carbon surface. Improvement in the rate of carbon recovery due to the iron amendment
was negligible, however. However, iron amendment to the carbon surface did decrease by about 3-fold the
rate of H2O2 usage, which is the primary driver in operating cost of the system.
81
-------
Field-Scale Regeneration Trials
Equipment Testing - Methylene Chloride and Chloroform Recovery Tests
Initial field regeneration trials were carried out using a larger column (I.D. = 5 cm, L = 30 cm, V = 600 mL,
residence time = 2 s) containing 100 g URV-MOD 1 GAC that was pre-loaded with MC or CF (under lab conditions).
Contaminant selection was based on hydrophobicity and reactivity with hydroxyl radicals (Table I).
Regenerant solution was recirculated continuously during each 30-hour experiment. Hydrogen peroxide was
added at 1-hour intervals during hours 0-6 and 23-28 (Figure 53). At each point of addition, 150 mL of a 50% H2O2
stock solution was added to the 7 L recirculation reservoir. Since H2O2 was essentially exhausted at each addition
point, the H2O2 concentration immediately after addition was about 0.38 M.
Carbon was periodically withdrawn from the top and bottom of the reactor for extraction and measurement of
residual contaminants. Methylene chloride was essentially gone (full recovery) after 6-7 hours of operation. After 30
hours, just 6% of the original CF loading (125 mg CF/g carbon) remained on the GAC. The cost of recovery ranged
from $2.5/kg to $6.6/kg GAC treated for the target contaminants studied. When multiple contaminants are
simultaneously adsorbed to GAC, the compound with slowest recovery would determine the overall cost. Little was
done in these experiments to limit the non-productive consumption of H2O2. That is, neither the configuration of the
recovery system nor the schedule of H2O2 additions was designed to reduce H2O2 consumption/radical production that
did not result in MC or CF destruction. Further discussion of this point is provided below.
10
15
Time (hr)
20
25
30
Figure 53. Carbon regeneration for (4) MC and (•) CF in field experiments. Units in the y-axis are "mg VOC/g GAC". Degradation to below
detection limit and 93% for MC and CF, respectively, was achieved after a 30-hour regeneration period. Reservoir concentrations: 10 mM
iron, 0.15 M average H2O2, pH = 2.0. Dotted lines indicate times of 150 mL hydrogen peroxide additions. Error bars indicate difference in
concentrations measured for samples taken from the top and bottom of the carbon bed.
82
-------
Sequential Adsorption/Regeneration Experiments
The feasibility of carbon regeneration depends on both the acceptability of regeneration costs (primarily H2O2
consumption) and maintenance of carbon adsorption capacity through multiple degradation steps. Here, carbon
adsorption capacity was tested before and after each of three surface regenerations. The work discussed in this section
is part of a paper accepted for publication (De Las Casas et al., 2006).
Carbon was loaded with 100-110 mg TCE/g GAC in a batch reactor in the lab, then transferred to the field site for
regeneration in the field column. GAC (100 g, dry weight) was suspended in 1 L of pure water that was pre-saturated
with TCE at room temperature (initial TCE concentration « 1100 mg/L). After 3 days, the distribution of TCE
between carbon and liquid was near equilibrium with more than 99% of the contaminant on the carbon surface. The
process was repeated twice, after field regeneration, using the same carbon sample to determine whether TCE
adsorption was adversely affected by Fenton-driven regeneration. During regeneration periods, 0.7±0.2 g carbon
samples were periodically withdrawn from the top and bottom of the column and extracted in ethyl acetate for
determination of residual TCE. The regenerant solution containing 10 mM total Fe (pH 2) was recirculated at a rate
that produced 50% GAC bed expansion. To initiate regeneration, 150 mL of 50% H2O2 was added to the 7 L
regenerant volume to produce an initial H2O2 concentration of 0.38 M. Thereafter, the schedule of H2O2 additions was
as indicated in Figure 54. At each point, an additional 150 mL of the 50% H2O2 stock solution was added to the
regenerant reservoir.
50
100
Time (hr)
150
Figure 54. TCE carbon recovery during three sequential regeneration cycles. Vertical dotted lines indicate points of hydrogen peroxide
addition. The horizontal dashed line represents the TCE load (107 mg/g GAC) at the start of the first carbon recovery procedure. TCE
degradation of 73% (•), 82% (+), and 95% (A) was obtained for the three, consecutive, regeneration cycles. Error bars indicate the difference
between the carbon extraction values from top and bottom of the column.
In each phase of the experiment, carbon recovery was initially fast with 50% TCE loss from the carbon surface in
4 hours or less. Thereafter, recovery was much slower so that final (60-hour) TCE recoveries were 73, 82 and 95%
during the sequential regenerations. Improvement in the later regeneration cycles was probably a consequence of
83
-------
more frequent H2O2 addition rather than a treatment-derived change in the physical characteristics of the URV-MOD1
carbon. Most important was the maintenance of TCE adsorption capacity after the 180-hour experiment (Figure 54).
This finding is supported by previous investigations involving jV-nitrosodimethylamine (Kommineni et al, 2003) and
methyl fert-butyl ether (Ruling et al., 2005a) adsorption/regeneration on GAC.
Ruling et al. (2005a) discussed two mechanisms that could adversely affect the performance of the activated
carbon regenerated via an aggressive oxidative treatment such as the Fenton's reaction. In the study, the authors
discussed possible chemical and physical alterations to GAC due to oxidative treatment. In a previous study, Ruling
et al. (2005b) reported reduction in surface area, microporosity, total porosity and sorptive capacity as a result of
repeated (10-15) regeneration treatments to the GAC. In addition, incomplete transformation of the target
compound(s) may affect the performance of the GAC by accumulating reaction intermediates on sorption sites and in
this way diminishing the availability of sites for the target compounds. Other studies have also reported no loss of
carbon sorption capacity under aggressive oxidative conditions (Toledo et al., 2003).
TCE was measured periodically in the regenerant reservoir to gauge the adequacy of the H2O2 addition schedule.
Comparison of regenerant TCE concentrations with (calculated) aqueous-phase concentrations in equilibrium with
residual sorbed TCE concentrations (Figure 55) suggests that less frequent H2O2 could have produced similar
recovery kinetics while reducing H2O2 consumption.
5
1°
' — '
^
;s.
&
O
1
O"
4J
n ?
0.9 T
0.8 -
0.7 -
0.6 -
n r
'
0.4 -
0 "3
\J . J
0.2 -
0.1 -
n n
. n
A
A o
m A
° O
n d
n n
p.
DD
0 1st Regeneration
D 2nd Regeneration
n A 3rd Regeneration
10
20
30
Time (hr)
40
50
60
Figure 55. Ratio of AQ (Ceq - Cliq) to Ceq, where Ceq is the aqueous-phase TCE concentration in equilibrium with the residual adsorbed
concentration (q) and Cliq is the measured, liquid phase concentration. Results for three consecutive regeneration periods are superimposed.
Equilibrium concentrations were calculated using measurements of residual adsorbed TCE (Figure 53) and TCE isotherm parameters (Table II).
Note: t = 0 marks the beginning of each recovery cycle.
Were liquid levels to approach equilibrium with residual adsorbed TCE between peroxide additions, reactor
performance could be improved significantly by increasing the frequency of H2O2 additions. Conversely, if aqueous-
phase concentrations remain low relative to equilibrium levels calculated on the basis of adsorbed mass, then the
84
-------
period of H2O2 addition could be extended to lower operational costs. The data (Figure 56) suggest that dissolved
TCE was quickly destroyed following each H2O2 addition to the regenerant. However, aqueous-phase TCE
concentrations also recovered quickly after H2O2 was exhausted. There was apparently little to gain by decreasing the
frequency of H2O2 addition in this experiment.
Temperature was measured in the regenerant solution during the experiment. Overall, temperature increased
from ambient (~30°C) to 55-60°C during the 180-hour procedure. This is consistent with the exothermic nature of the
Fenton's reaction. Temperature decreased slowly following H2O2 exhaustion (l-2°C/hr). Greater temperature
increases might be expected in larger reactors although more judicious application of H2O2 or reduction in regenerant
iron levels would tend to mitigate temperature rise. Because TCE mass transport and reaction kinetics are favorably
affected by higher temperature, the exothermic decomposition of H2O2 via reaction with iron might, if handled
carefully, increase carbon recovery rates and lower overall costs for carbon surface regeneration.
O)
o
50 -,
45 II
40 -
35 -
30 -
25 -
20 -
15 -
10 -
5 -/
0 I
—•—Cliquid (mg/L)
• Ceq (mg/L)
o
10
20
30
Time (hr)
40
50
60
Figure 56. Comparison between Ceq (aqueous-phase TCE concentrations in equilibrium with the residual adsorbed concentration (q)) and Cliq
(measured, liquid phase concentration). Original data from Figure 54. Equilibrium concentrations were calculated using measurements of
residual adsorbed TCE (Figure 54) and TCE isotherm parameters (Table II). Note: t = 0 marks the beginning of the recovery cycle.
Loading Carbon with SVE Gases
Vadose zone gases from the soil vapor extraction system at the Park-Euclid (Arizona) state Superfund site
containing primarily PCE, TCE and light diesel components as contaminants were used to load URV-MOD 1 GAC in
a final set of field experiments.
It is likely that gas-phase contaminants experience more facile access to adsorption sites in carbon micropores, so
that the contaminant mass removed prior to apparent breakthrough may be greater in gas-phase applications.
Nevertheless, a water film is almost always present on the carbon surface in such situations, so that, ultimately, the
adsorption capacity is dictated by the heterogeneous equilibrium between the adsorbed and liquid-phase chemical. If
85
-------
Henry's Law limits the liquid-phase chemical concentration, the carbon load at breakthrough should be indifferent to
the form in which the contaminant is applied. That is, gas-phase treatment should yield the same loading as treatment
of a liquid that is in Henry's Law equilibrium with that gas. Crittenden et al. (1988) suggested that 45% relative
humidity represents a critical cutoff point, in that a liquid film is fully developed on the carbon at sustained relative
humidity greater than 45%. Gas removed from surface solids should be near saturation levels with water vapor (at the
soil temperature). For this reason, a moisture knockout box is usually included in SVE designs.
The GAC was loaded for approximately 72 hours using a 4-cfm SVE side-stream. Effluent gases were pumped
back into the system of extraction wells. To determine the carbon loading, GAC samples were taken from the top and
bottom of the column, extracted in ethyl acetate and analyzed with GC-ECD. Initial, 6-hour regeneration trials
produced 80% reduction in the adsorbed TCE concentrations, but only a 30% loss of adsorbed PCE (Figure 57). As in
the previous field trials, degradation was initiated by adding 150 mL of the 50% H2O2 stock to the regenerant solution
(total volume 7 L). Subsequently, 50 mL of the stock H2O2 solution was added every 15-30 minutes to replenish the
initial H2O2 concentrations throughout the regeneration period. This procedure led to excessive H2O2 utilization and
attendant cost.
0123456
Time (hr)
Figure 57. Carbon regeneration for SVE-loaded GAC. The two primary pollutants at the site are PCE and TCE. Overall, 30% and 80%
degradation for PCE (•) and TCE (4) were achieved during the 6-hour regeneration period. Reagent concentrations were 10 mMiron, 0.15 M
H2O2 (average), at pH = 2.0. Error bars indicate the difference between the carbon samples from top and bottom of the column.
A total of 1.1 L of 50% H2O2 was consumed to destroy 7.0 g of PCE and 4.0 g TCE. This gives a molar yield of
2.2xlO"3 and 1.6xlO"3 for PCE and TCE, respectively. Pulsed addition of H2O2 with intervening periods in which H2O2
was exhausted (for contaminant transport to the bulk regenerant phase) might have produced comparable recoveries
using a fraction of the oxidant. Peroxide costs could also be lowered significantly by reducing the volume of
regenerant in the system. In the presence of Fenton's reagents, aqueous-phase concentrations of contaminants were
generally near zero. Because the rate of H2O2 consumption is independent of the contaminant concentration, however,
86
-------
H2O2 use continued during such periods without affecting contaminant transport out of the carbon particles. Under all
circumstances, H2O2 consumption was proportional to the total regenerant volume, including the volume in the
recirculation tank, where little or no contaminant was consumed under conditions of the field test. Consequently,
future field experiments using similar conditions could further reduce the recirculation tank volume to lower reagent
costs.
Economic Analysis
The cost of an in-place GAC recovery system based on Fenton's mechanism would include reagent (H2O2)
consumption, power requirements to circulate regenerant, replacement of carbon that is lost due to abrasion and
chemical (treatment) inactivation/destruction, capital expenses associated with the regeneration system itself (pumps,
pipes and tanks) and the additional capacity necessary for temporary column retirement for regeneration. Here, it is
assumed that H2O2 costs dominate the additional cost for an in-place Fenton carbon recovery system (see subsequent,
detailed GAC Economic Analysis for a more complete cost estimate supporting this assumption). The primary sinks
for «OH are H2O2 itself, due to concentration effects (Table II) and possibly (as regeneration cycles progress) free
chloride ion. Thus, free radical concentrations (and H2O2 consumption rates necessary to maintain those levels) are
essentially independent of the identity or concentration of the target compounds. This is not to say, however, that
H2O2 costs are independent of contaminant identity. Compounds that desorb slowly from GAC require significantly
greater time to achieve comparable degrees of recovery and, hence, greater recovery costs.
When multiple contaminants are present simultaneously, the compound most resistant to Fenton-driven recovery,
in this case PCE, is likely to dominate recovery costs. Consequently, PCE degradation was used to estimate overall
carbon regeneration costs. Other assumptions and economic or operational factors were:
Unit cost of H2O2
(50% solution, 1.18g/mL)
H2O2 utilization for carbon recovery
Carbon in experimental columns
Carbon purchase cost (EPA, 2000)
Carbon change out/disposal
S0.34/L (Kommineni etal, 2003)
(transportation cost not included)
95-232 mL (bench-scale column)
1-2 L (field column)
10-16 g (bench-scale column)
78-100 g (field column)
$1.54-2.64/kg virgin coal carbon
$1.10-1.72/kg regenerated carbon
$0.66/kg (soiltherm.com, 2006)
Costs for carbon recovery/replacement alternatives are compared in Table XXII. The comparison is necessarily
simplistic. The mechanism, rate limitation and kinetics of PCE recovery on GAC are poorly known, certainly not well
enough to produce a most refined economic analysis. In-place oxidations of all other contaminants tested were
87
-------
significantly faster than that of PCE and, therefore, more economically attractive. Nevertheless, PCE was among the
important contaminants in this application and should be considered when assessing the utility of the technology
locally.
Table XXII. Comparison of GAC Replacement vs. Regeneration Costs
Option
Cost/kg Critical assumptions/parameters
1. Replacement $3.30
2. Thermal Regeneration $2.64
3. Fenton-based, in-place recovery
a. Lab column $2.69
b. Field column $6.54
$2.64/kg purchase cost
$0.66/kg disposal of spent carbon
$1.65/kg regeneration cost
$0.66/kg transportation
$0.33/kg carbon replacement
Peroxide costs dominate (0.34/L)
PCE recovery dictates treatment time
95 mL H2O2/12 g GAC
1.5LH2O/78gGAC
Methods for increasing the efficiency of H2O2 use have been discussed to some extent. The mixing reservoir
allowed the Fenton reaction to take place and H2O2 to be consumed without oxidation of the target contaminants.
Based on the dimensions of the pilot-scale column, and assuming a porosity of 0.5 of the GAC, the pore volume
within the column is approximately 0.3 L. The total liquid volume of the reactor system was 7 L. This suggests that
more than 95% of the H2O2 applied to this system was consumed outside of the column (i.e. 0.3 L/7 L). Although we
assume that the Fenton reaction (in our system) occurred predominantly in the reservoir, limiting the H2O2 reaction
outside the mixing reservoir would economize H2O2. It may be possible to immobilize iron on the carbon surface and
run regenerations in a pH range to avoid iron dissolution during recovery operations. The feasibility of such a scheme
depends on selection of iron loadings that allow degradation reactions to proceed without blocking the carbon surface
or interfering with contaminant access to carbon pores. Preliminary tests of Fenton-driven recovery in such iron-
mounted systems have been discussed in a previous section. The potential advantage lies in localization of H2O2-
consuming reactions and radical generation in the vicinity of the carbon surface. Bulk-aqueous-phase Fenton
reactions can be minimized (i.e. sizing down reservoir) so that non-productive H2O2 consumption (that which destroys
no contaminants) is greatly diminished.
Cost Estimation Based on Iron-Amended GAC Regeneration
The cost of carbon recovery in the field-scale reactors using iron-mounted on the surface of the GAC is discussed
here by considering the scale-up factors and issues for going from bench to field scale. The field-scale set-up is
approximately seven times bigger than the bench-scale, in which experiments with iron-amended carbon were
-------
actually run (based on g GAC and reservoir size). This factor was used to roughly scale costs to the field-scale. Table
XXIII summarizes the results from the bench-scale trials, discussed previously (see bench-scale section). The last
column shows the estimated demand for H2O2, scaled-up to field dimensions. If the iron-amended GAC were
employed in field trials, attendant costs will be significantly lower than those actually encountered. For example, if
we regenerate 100 g GAC, it is estimated that 181 mL H2O2 will be consumed. Actual field trials utilized 1-2 L H2O2
when iron was provided. That is at least 8 times higher than the amount to be used if an iron-amended GAC system
were implemented. In the field site studied, the SVE stream contains mainly TCE and PCE. However, the presence of
other organic compounds can increase the demand for H2O2. Nevertheless, for ease of calculations it is assumed that
the same amount of H2O2 will be applied in the recovery of the spent GAC when both TCE and PCE are present.
Consequently, the cost of GAC regeneration for SVE-loaded GAC will be the same for both compounds, and
estimations based on TCE regeneration will be applied to estimate the field-scale cost. If the iron-mounted GAC is 10
times more efficient than the soluble iron system, then the peroxide cost would be $0.62/kg GAC ($0.28/lb GAC).
Under these circumstances, the iron-amended GAC regeneration is probably less expensive than replacement of spent
GAC (Table XXIII). A more complete economic analysis follows.
Table XXIII. Cost Evaluation Based on Bench-Scale Results Using TCE-Loaded and Iron-Amended GAC.
„.„, 7^** C/Co H2O2 H2O2 (mL) scaled-up to field dimensions
GAC type Iron Content ,TY^ , -, ( ,+•+->
_ (TCE) (mL) (estimate)
Iron-amended GAC 7.4mgFe/gGAC (5 mM) 47% 25.8 181 ($0.62/kg GAC)+
Iron in Solution, pH 2 10 mM 50% 160-224 1.6*
* Actual field scale trials employed 1-2 L H2O2 for experiments using iron in solution. Field-scale is approximately
7x bigger than bench-scale experiments. Costs can be easily calculated using the cost of H2O2 (S0.34/L). Field
scale reservoir can be reduced in size as it was done with the bench-scale trials, reducing costs by half.
+ Costs calculated for 181 mis H2O2 to regenerate 100 g GAC. Both background GAC and iron-amended GAC
trials employed a 400 mi-size reservoir (less than half the size of the iron in solution experiment (1 L reservoir).
GAC Economic Analysis
The economic analysis conducted here was designed to compare alternative carbon replacement/regeneration
strategies in processes using activated carbon for contaminant adsorption. In scenario #1, spent carbon is replaced
with new activated carbon, and the waste carbon is disposed of as a hazardous waste. Scenario #2 differs in that
carbon is periodically regenerated or, at least, partially regenerated using Fenton's reagents to destroy the adsorbed
contaminants. The economic analysis was carried out by comparing costs that are unique to each scenario on both a
present worth and an annual cost basis. Most of the costs from activated carbon adsorption for treatment of gas-phase
streams derived from SVE (the basic scheme for both scenarios) are common to both alternatives. As such they are
omitted from the analysis. These include energy costs for the SVE system, capital costs for the carbon adsorption unit,
initial carbon costs and some maintenance and other labor activities. A description of the costs that are unique to each
alternative follows:
89
-------
Scenario # 1 Scenario #2
GAC replacement cost 5% loss/regeneration cycle
Carbon costs GAC cost = $ 1.60/lb GAC cost = $ 1.60/lb
Disposal cost = $ 1.00/lb No disposal costs
$0.345/lb H2O2
H2O2 consumption Not Applicable plus transportation from Houston
$3.50/mi, 1067 miles
Incremental capital for chemical AT * A i • ui ™ u/^^ ^ i
, ., Not Applicable Dosing pumps, H2O2 storage tanks
The scenarios were built in part from field experience in this project (H2O2 dosing and time to recovery) and in
part based on engineering rules of thumb or professional judgment. Cost comparisons were carried out for carbon
recovery when (i) PCE and (ii) methylene chloride were the target contaminants. The behaviors of these compounds
during carbon regeneration were diametrically opposed. PCE provides a challenging recovery problem, presumably
because of its affinity for the carbon surface and consequent slow desorption rate. Carbon recovery following
methylene chloride breakthrough is remarkably fast, probably due to its low affinity for the carbon surface.
Other assumptions or data used to support the economic comparison were:
• bulk GAC density = 0.5 kg/L
• gas-flow rates during SVE = 10, 100 cfm (2 distinct analyses)
• equilibrium is assumed to exist among gas, liquid and solid phases at breakthrough during SVE
• T = 25°C
• The concentration of contaminant (PCE or MC) in the gas treated by SVE is 100 ppmv.
• GAC column diameter = 1 m
• The mass of carbon in the column is irrelevant since carbon wastage rates are calculated on the same basis for
each scenario investigated
• The pump efficiency during recirculation of Fenton's reagents is 0.70.
• The economic discount operator of 0.08 was assumed
• Equipment for H2O2 dosing and storage has a service life of 20 years
• GAC disposal cost ($1.00/lb) is an engineering estimate
• All GAC costs are in year 2000 dollars and all other cost are in 2006 dollars
Details for the analysis, including calculations, are provided in Appendix B. Additional assumptions are exposed
in the appendix. The following summary represents the annualized costs that are unique to each alternative (Table
XXIV). Annual costs are provided for hypothetical SVE systems treating gas flows of 10 and 100 cfm for removal of
90
-------
PCE or methylene chloride. Alternative #2 consists of in-place recovery using the Fenton-driven method investigated
here.
The analysis shows that the chemical costs are in fact the dominant expense in Scenario #2 (Appendix B). The
following analysis was added to this section and to Appendix B to highlight the importance of H2O2 purchase to the
total cost in Scenario #2.
Table XXIV. Cost Estimates Comparing Hazardous Waste Disposal of Spent GAC (# 1) and Virgin Carbon Replacement versus Fenton' s
Reagent Regeneration of GAC (#2)
Target Contaminant
SVE Flow
lOcfm
100 cfm
PCE
#1: $2,545
#2: $25,212
#1: $25,416
#2: $239,745
MC
#!:$!.1M
#2: $0.83M
#1: $11.1M
#2: $8.2M
Note: Option 1: GAC cost = $2.60/lb. Option 2: GAC cost = $1.60/lb (no disposal), H2O2 cost (purchase and transportation), additional process
energy, labor, and capital (recirculation pumps, chemical storage tanks).
Cost ofH2O2 Consumption in Fenton's System
The following table (XXV) contains both total costs for each treatment alternative and the associated H2O2 cost in
order to illustrate the importance of reagent costs to overall process economics.
Table XXV. Summary of Estimated H2O2 Cost Contribution to Total Cost of Fenton's Regeneration
Compound
PCE
MC
Total H2O2 Cost
for GAC regeneration
($/yr)
0^=10 cfm 0^=100 cfm
17,966 179,495
0.75M 7.4M
Total Annual Cost
($/yr)
Q^lOcfm 0^=100 cfm
25,212 239,745
0.83M 8.2M
Fraction of H2O2 Cost
for GAC regeneration
($/yr)
0^=10 cfm 0^=100 cfm
71% 75%
90% 90%
Note: All values were obtained from Appendix F.
91
-------
Engineering Considerations
A number of practical considerations that may affect operation of a Fenton-driven carbon recovery system have
been neglected to this point. Chief among these is (i) the generation of molecular oxygen from the decomposition
reactions of H2O2 in the presence of iron and (ii) the exothermic character of the same reactions. In this section, each
of these is analyzed using engineering tools in order to gain perspective on the magnitude of related operational
difficulties.
In the case of gas-phase O2 generation, field data on H2O2 consumption are combined with operational flow rates
and hydrogen peroxide decomposition stoichiometry to determine the probable oxygen volume rate of flow during
regeneration. For perspective, this is compared to the volume rate of flow of the regenerant stream that is necessary to
expand the carbon bed during regeneration. The analysis can go only so far, however, since the effects of oxygen
generation in carbon pores cannot be adequately addressed without much more study.
The engineering analysis of heat generation/transport is again based on the H2O2 consumption data and regenerant
flow using the (known) heat of hydrogen peroxide decomposition. In this case, the regenerant flow and specific heat
of water allow us to calculate the regenerant temperature rise necessary to balance the rate of heat liberation due to
reaction during one pass of regenerant through the reactor during recovery. Then, based on the volume of regenerant
in the storage tank and regenerant flow rate it is possible to predict the rate and extent of fluid temperature rise during
recovery operations.
The methods and calculations are fully explained below.
Hydrogen Peroxide Stability
The primary factors contributing to H2O2 decomposition during storage include: increasing temperature (2.2
factor increase in first-order reaction rate constant for each 10°C); increasing pH (especially at pH > 6-8); increasing
contamination (especially transition metals such as copper, manganese or iron); and to a lesser degree, exposure to
ultraviolet light. In most cases, pH and contamination work in tandem as the dominant factors (US Peroxide, 2006).
rhttp://www.h2o2.com/intro/faq.html1. Generally, 50% hydrogen peroxide loses less than 1% per year when stored
properly (according to manufacturer's specifications.) [http://www.h2o2.com/h2o2update/volume2/hvpochlorite.html1
During Fenton based reactions, of course, H2O2 decomposition is exactly what is to be promoted.
Hydrogen peroxide often decomposes exothermically into water and oxygen gas spontaneously:
2H2O2 ->• 2H2O + O2 + Energy (68)
This process is very energetically favorable; it has a AH° of-98.2 KJ/mol, a AG° of -19.2 KJ/mol and a AS of 70.5
J/mol °K. The rate of decomposition is dependent on the temperature, concentration of hydrogen peroxide, pH and the
presence of impurities (transition metals) and stabilizers.
[Wikipedia, http://en.wikipedia.org/wiki/Hydrogen_peroxide#Decomposition1.
92
-------
Although pure hydrogen peroxide is fairly stable, it decomposes into water and oxygen when heated above about
80°C. rhttp://www.infoplease.com/ce6/sci/A0824724.html1
The Heat of H2O2 Decomposition
Accumulation of heat in the regenerant and, as a consequence, in the carbon bed from the exothermic
decomposition of H2O2 could be beneficial within manageable limits, but would be a problem if excessive. However,
moderate heat generation and consequent temperature increases could aid the recovery process by increasing radical
generation rates and the rates of temperature dependent mass transfer processes. An energy balance was used to
estimate the temperature increase in the regenerant. In the balance, the average rate of heat generation by hydrogen
peroxide decomposition was exactly balanced by the rate of heat loss in advective flow through the column. That is
2 = pQCpAT where
•
AH rxn rriH2o2 = average rate of heat generation from the H2O2 decomposition (J/min)
p = density of the solution (g/mL)
Q= flowrate through the column (mL/min)
Cp = specific heat (cal/g °C)
AT = temperature change = Tf - T (°C)
•
It should be noted that mn2o2 is taken here as the average rate of H2O2 use during experiments that comprised the
field demonstration project. H2O2 was unevenly applied and consumed during those experiments, however.
Consequently, there will be periods in which heat is generated more rapidly and more slowly than estimated here. The
average rate of heat generation provides an adequate estimate of temperature rise as long as that rate is not so rapid as
to become dangerous during particularly fast reaction periods (immediately following addition of H2O2 to the
regenerant). In fact, predicted and (field) observed rates of regenerant temperature rise proved to be modest (see
below).
Furthermore, it was assumed that:
The heat capacity of the carbon and column materials could be ignored;
Radiative heat losses were comparatively small;
1 . There was no cumulative heat energy in the regenerant or carbon bed (one-pass analysis);
2. Excursions from the average rate of H2O2 consumption and heat generation could be ignored for convenience;
and
3. All reactions involving H2O2 other than its breakdown to water and molecular oxygen could be ignored.
93
-------
Heat production:
Rate of heat
generation
pQCpAT
GAC
Column
AH rxnis the heat of reaction at constant pressure and standard conditions
Cp=1.00 cal/g °C and psoiution = 1 kg/L
lcal=4.184J
For the reaction: 2H2O2 0) -> 2H2O 0) + O2 (g) AHrxn° (KJ/mol) = 2(-286)+0-2(-187.8)= -196.4
AH/ (KJ/mol)
H202 (1)
H2O (1)
02(g)
-187.8
-286
0
Source: Fundamentals of Chemistry. Brady, J. and Holum, J. 3rd ed. John Wiley & Sons. NY: 1988.
Since 2L of 50% wt/wt H2O2 in water (p = 1.2 g/ml) were used to regenerate the pilot column over a 54-hour
period, the mass of H2O2 consumed was
\mol
S solution
'solution
H2O2
L
17.65 — x2L = 35.30molH2O2
94
-------
Thus, the average rate of H2O2 consumption was
3530molesH2O2 \hr
-x
54hrs 60 min
and the average rate of heat generation from the reaction was:
\.09xW-2molesH2O2 196AKJ 1000.7
min 2molH,O, \KJ
= 1.07jd03J7min
The rate of regenerant flow through the column was estimated at 8 L/min (field data), so that the average increase
in regenerant temperature to balance the rate of heat generation was:
1.07x103./ Imin leal ImL /C .
min 8000ml 4.184J \g leal
This is the temperature increment in regenerant fluid per single pass through the carbon column due to the
regeneration reaction. The average detention time in the regenerant tank was about 1 minute, so that there were
opportunities for heat energy to accumulate in the regenerant, immediately after one of the H2O2 pulses, when the
decomposition reaction was most rapid. In fact, however, the reservoir temperature never exceeded 52°C, after an
overnight low of 36 C, suggesting that the regeneration system can be conveniently regulated through management of
the H2O2 addition rate.
In an hour, without any regenerant cooling, the temperature rise would be
60min 8Z/min „„,-„,, - -,„
A^r= — x—-—x0.032 C = 2.2 C
hr 1L
The field reaction was operated for 7-hour periods with consequent temperature changes of about 16°C, or
2.2°C/hr, suggesting that the analysis is valid. However, the rate of H2O2 use during the 7-hr daylight periods was
about 3x the average daily rate, so that perhaps two-thirds of the heat generated is unaccounted for in the regenerant
fluid.
Oxygen Formation via Fenton's Reactions:
The mechanism proposed by De Laat and Gallard (1999) describes in detail the reactions that result in oxygen
formation. Depending on the pH of the solution, HO2*/ O2* will react with iron to generate oxygen.
Fe(III) + HO2- -» Fe2+ + O2 + FT klla < 2 x 103 M'1 s'1
Fe(III) + O2' -»Fe2+ + O2 kllb = 5 x 107 M'1 s'1
95
-------
HO2'+ HO2- -> H2O2 + O2 k13a = 8.3 x 105 M'1 s'1
HO2- + O2- + H2O -> H2O2 + O2+ OH k13b = 9.7 x 107 M'1 s'1
•OH + HO2- -> H2O + O2 k14a = 0.71 x 1010 M'1 s'1
•OH + O2- -»OH+O2 k14b= 1.01xl010M-V
Given the typical conditions under which these reactions occur, however, radical-radical recombination reactions
are not a significant source of oxygen. Consequently, oxygen production is estimated according to the overall
hydrogen peroxide decomposition reaction (above) resulting in liquid water and oxygen gas formation. This analysis
ignores
1. Compressibility effects
2. Dissolved oxygen transport in the regenerant
3. Temperature effects
Stoichiometrically, 2H2O2 molecules generate one molecule of oxygen during decomposition:
2H202(1)^2H20(1) + 02(g)
From previous calculations based on the average rate of H2O2 decomposition, 2L of 50% wt/wt H2O2 in water were
used to regenerate the pilot column over each 54-hour period. The average rate of O2 production was therefore
35.30molesH7O7 lmolO7 \hr c AC 1rt3 , _ . .
2—2-x 2_x = 5.45jdO molesO7/mm
54hrs 2molH2O2 60mm
Normalized to the rate of flow of water:
5A5xW~3molesO7 32000mgO7 min „, 0 ^ IT
-x ^-^x = 2l.8mgO2/L
min molO2 8Z
Now, the solubility of oxygen in water can be calculated as follows:
P02 - latm
Using Henry's law, the concentration of oxygen dissolved in water is
[02} = KHP02
where KH = Henry's law constant =0.0012630 M-atm (From Table 2.4 at T=25°C. Introduction to Environmental
Engineering and Science. 2nd Ed. Masters, G. Prentice Hall. New Jersey, 1998.)
96
-------
[O2]= O.OOU630M-atmxlatm = 0.0012630M
mol
[O2 \ = 40.4mg IL (at equilibrium with pure O2 gas at a total pressure of 1 atm.).
The concentration of oxygen for a pure oxygen saturated solution at 25 °C is about 40 mg/L, which means that the
oxygen concentration produced from the decomposition of hydrogen peroxide does not exceed (21.8mg/L < 40 mg/L)
the solubility of oxygen in water.
The oxygen concentration was calculated for the 54-hour period of regeneration. In general, the hydrogen
peroxide was added over short periods (30 min) followed by longer periods (6 hours) in which hydrogen peroxide was
exhausted and the target compound was allowed to desorb from the GAC. Thus, oxygen formation rates were
considerably higher during those short periods. The likely formation of O2 gas bubbles, which was actually observed
in the lab and field experiments, indicates that course media should be used and that the GAC bed should be expanded
during regeneration to aid in the release of gas bubbles from the column.
Field-Scale Regeneration Trials-Summary
• The data and analysis suggest that intraparticle diffusion or desorption from the solid phase of the
contaminant is a limiting factor in the second, slow phase of degradation.
• The following table contains bench- and field-scale recovery data that reflect the ease or speed of column
recovery when respective compounds have initially saturated carbon adsorption sites. The consensus of the
study group regarding the source of rate limitation during the recovery process is as indicated (Table XXVI).
Table XXVI. Summary of Efficiency Results for Fenton's Reagent Regeneration of GAC in Bench and Field Trials
Compound
Methylene Chloride (MC)
1,2-DCA
1,1,1-TCA
Chloroform (CF)
Carbon Tetrachloride (CT)
Trichloroethylene (TCE)
Perchloroethylene (PCE)
Percentage Removed from GAC (%)
Bench-scale
(14 hrs @ 32C)
99
98
67
93
73 @25C
52
35
Field scale
(7-54 hours)
98 (7 hrs)
99 (30 hrs)
N/A
N/A
82 (7 hrs)
93 (29 hrs)
N/A
73-95 (50-54 hrs)*
50 (52 hrs)
Controlling Mechanism
Radical Reaction rate
Radical Reaction rate
Desorption or pore diffusion
Radical Reaction rate
Desorption or pore diffusion
Desorption or pore diffusion
Desorption or pore diffusion
*depending on H2C>2 application frequency (see Field-scale Results Section).
97
-------
• GAC was preloaded with methylene chloride (MC) and chloroform (CF) for field-scale regeneration
experiments. H2O2 was added hourly between hours 0-6 and 23-28. The reduction of solid-phase CF was 93% after
30 hours. Residual MC could not be detected.
• GAC was loaded with trichloroethylene (TCE) in the laboratory and regenerated in the field in three
successive loading/regeneration cycles. In each cycle, the solid-phase TCE concentration decreased by >50% in the
first 4 hours. After 60 hours, however, carbon recoveries ranged from 73-95%. No loss of adsorption capacity was
observed during the overall period of the experiment. H2O2 was added intermittently during regeneration. By allowing
the bulk aqueous-phase contaminant concentration to approach equilibrium levels prior to H2O2 addition, it may be
possible to minimize H2O2 costs. Since recovery appears to be limited by compound desorption at least in some cases,
accumulation of the target in the aqueous phase prior to H2O2 addition is likely to minimize chemical costs.
• Temperature in the field reactor was observed to increase from an ambient value of about 30°C to 55-60°C
during a 60-hour regeneration period. Because VOC mass transport and reaction kinetics are favorably affected by
higher temperatures, the exothermic decomposition of H2O2 could, if handled carefully, increase carbon recovery
rates and lower overall costs for carbon surface regeneration.
• PCE, TCE and light diesel contaminants from the field site soil vapor extraction system were used to load
URV-MOD 1 GAC in a final set of field experiments. A 6-hour regeneration trial reduced the adsorbed TCE
concentration by 80%, but PCE by only 30%. H2O2 was added periodically but without any attempt to minimize
chemical consumption.
• A scoping-level economic analysis was conducted, based on regeneration of PCE-loaded GAC. PCE presents
the most challenging recovery situation in terms of recovery kinetics. The cost for Fenton-based regeneration,
determined from results of bench-scale studies (~$2.70/lb), was comparable to that of conventional thermal
regeneration (~$2.60/lb) and new carbon replacement (~$3.30/lb). The cost of Fenton regeneration based on the field
trials was higher (~$6.50/lb), but this cost may not represent operational cost following optimization of chemical
addition frequency. There was no attempt to minimize H2O2 consumption, the primary cost-driver, in the field trial.
Further study in this are is recommended.
• It may be possible to both minimize unproductive H2O2 consumption and to shift the operational pH range for
carbon recovery by precipitating iron on the carbon surface prior to use in adsorption/recovery operations. The
feasibility of such a scheme depends on selection of iron loadings that allow degradation reactions to proceed without
blocking carbon pores, interfering with contaminant access. Bench-scale trials were carried out. Although field-scale
trials using iron-amended GAC were not conducted, a first-cut, estimate of H2O2 cost ($0.34/L) was undertaken. The
analysis suggests that iron-amended GAC regeneration costs would be ~$0.28/lb GAC, or about 10-fold lower than
the cost of thermal carbon regeneration or GAC replacement.
98
-------
Summary and Recommendations for Additional Study
General Observations
Data suggest that a single mechanism does not control the rate of VOC-loaded carbon regeneration by Fenton's
mechanism for all contaminants. Weakly adsorbed compounds with relatively low reactivity with «OH, like
chloroform, can be limited by reaction in the bulk aqueous phase. Less soluble, more reactive compounds like TCE
are limited by intraparticle transport, and the desorption reaction rate may also play a role in the most strongly bound
compounds (e.g., PCE). For compounds with limited solubility, for example, intraparticle aqueous concentrations of
the sorbate may be much lower than their equilibrium levels. Under these circumstances, surface diffusion may be the
predominant mechanism of sorbate transport to the particle exterior. Surface effects resulting from pore volume
distribution may prove to be significant as well. As pore dimensions approach the molecular size of the contaminant,
it is anticipated that surface diffusion mechanisms limit transport kinetics. It is known from literature that in
micropores found in zeolites (Ruthven, 1984), surface diffusion is the dominant transport mechanism, while pore
diffusion dominates in macropores (Ma et al., 1996). Results lead the investigators to conclude that mass transport
mechanisms limit the effectiveness of Fenton's reaction for carbon recovery, at least for slightly soluble compounds
that are reactive with Fenton's reagents. Therefore, optimal design of this type of treatment would maximize
contaminant flux from the sorbent while minimizing the use of H2O2, the primary contributor to process cost. Just
how this is done will probably be compound specific. Pulsed addition of H2O2 probably offers advantages over
continuous maintenance of a target H2O2 concentration when mass transport governs the carbon recovery rate.
Temperature management may be an essential issue inasmuch as the kinetics of physico-chemical processes that
determine recovery rate are temperature dependent and the Fenton reactions are exothermic. Fenton-based carbon
regeneration was shown to be cost competitive for VOCs of modest binding strength on GAC but was not clearly cost
effective for strongly binding VOCs. However, the cost efficiency can be improved substantially by implementing the
design/operational changes discussed. Additional cost saving may be possible by directing radical-generating
reactants to the carbon surface. Selection of carbon (or other sorbents) with a pore size distribution that minimizes
mass transfer limitations should also be considered.
The specific project findings on which this summary discussion is based follow.
99
-------
Homogeneous, Bench-Scale Experiments
• Tetrachloroethylene (PCE) degradation followed first order kinetics in which the rate constant was a function
of total iron in the Fenton system. The reaction proceeds with essentially no lag following the addition of
H2O2, indicating that near steady concentrations of iron species and hydroxyl radical are established quickly.
• The initial rate of PCE degradation was increased by more than an order of magnitude by the addition of
hydroxylamine hydrochloride, a strong reductant. However, the rate enhancement could not be sustained,
indicating that hydroxylamine was consumed in the reaction. The result supports the research consensus that
Fe(III) reduction to Fe(II) via H2O2 consumption limits the radical production rate in the Fenton system.
• Quinones are known electron shuttles that can facilitate iron reduction. 1,4-hydroquinone (HQ), 1,4-
benzoquinone (BQ) and 9,10-anthraquinone-2,6-disulfonic acid all initially increased PCE degradation in
Fenton's system. The increase was proportional to quinone concentration. However, as with hydroxylamine
addition, the rate enhancement was not sustained, probably reflecting the gradual destruction of the quinone.
After quinine addition, the PCE degradation rate eventually stabilized at a rate that was slower than that of the
unamended Fenton's system, suggesting that the by-products of quinone degradation retarded the
contaminant degradation rate.
• PCE degradation ceases if Cu(II) replaces Fe(III) in the Fenton's system. However, if both copper and iron
are present in a Cu:Fe ratio of 2, the rate of PCE degradation increases by a factor of 4.3. This accelerated rate
was steady over the course of the experiments. Among the possible mechanistic explanations, Cu(II) may be
reduced to Cu(I) via H2O2 consumption after which the conversion of Cu(I) to Cu(II), Cu(I) reduces Fe(III) to
Fe(II). The hypothesized mechanism may provide a more rapid pathway for iron reduction than reaction of
Fe(III) with H2O2 in the Fenton system. Although the ability of copper to accelerate PCE degradation is
modest, Cu(II) solubility is greater than that of iron, and copper may provide greater benefit in the pH range
where the iron concentration solubility in Fenton's system is limited.
• The first-order rate constant for PCE degradation increased more rapidly with temperature in the copper: iron
system than in the Fe-only Fenton system. Thus, the benefit of copper addition will be increased for Fenton's
reactor systems operating above ambient temperature.
• A homogeneous-phase kinetic model was formulated based on earlier work by De Laat and Gallard (1999), in
which the rate constant for Fe(III)-hydroperoxy complex reduction is the sole fitted parameter. Although
model simulations were in qualitative agreement with results generated here, there were several noteworthy
quantitative departures between model and experiment. It is suspected that the fitted rate constant requires
revision based on the substantially lower ionic strength in the present work and the insensitivity of the fitted
rate constant to pH in the De Laat and Gallard case.
• As chlorinated VOCs are degraded in a Fenton's system, chloride anions build up in the regenerant solution.
Literature rate constants for the reaction of the hydroxyl radical and chloride indicate that chloride
100
-------
accumulation should significantly retard VOC degradation rates. However, model simulations using the
literature rate constant overestimate the chloride impact observed in the experimental trials. Additional
modeling efforts using a fitted rate constant for the "OH/Cl reaction are in progress.
• The rate of PCE degradation by Fenton's system increases with increasing pH in the range 1 to 3. Above pH
3, decreased iron solubility limits free iron availability and, consequently, limits the rate of degradation.
Measured PCE degradation rates at 0.9 < pH < 3.0 were not in good agreement with the model results
although predicted and observed trends in pH-dependent data are similar. At every pH (except pH = 0.9), the
model over-predicts the PCE transformation rate.
• The rate of carbon tetrachloride (CT) degradation was increased by the addition of isopropanol (IP), a *OH
scavenger. The work strongly suggests CT degradation occurs via superoxide radical (O2*) attack in Fenton's
system. Literature studies suggest the increase in rate may be due to a co-solvency effect due to IP addition
that increases O2* activity.
• PCE degradation by Fenton's reagents was negligible in the presence of IP, which indicates PCE destruction
occurs via hydroxyl radical (*OH) attack. No chlorinated intermediate or final products were detected. PCE
degradation diminished modestly at the highest concentration of chloride added (0.0288 M). However, at a
concentration near 1 mM, Cl" had essentially no effect on PCE transformation kinetics.
Heterogeneous, Bench-Scale Experiments
• Ethyl acetate was the best solvent tested for solid:liquid extraction of PCE from GAC. An extraction duration
of 12 hours provided practically complete extraction.
• Contaminant degradation kinetics in a heterogeneous Fenton's system (VOC loaded on GAC) follow bi-
phasic, first-order kinetics. A fast initial phase was followed (after 1-3 hours) by a slower second phase. The
fraction of total contaminant degraded in the rapid initial phase increased with the aqueous-phase solubility of
the contaminants. The data and analysis suggest intraparticle diffusion or desorption of the contaminant is a
limiting factor in the second, slow phase of degradation.
• When chloroform (CF) -loaded GAC was rapidly flushed with fresh water, the rate of CF loss increased as
the size of the GAC particles on which it was adsorbed decreased. These results suggest that pore and/or
surface diffusion, affect the overall removal kinetics.
• A mass transfer model in which it was assumed that solid:liquid equilibrium exists throughout the porous
carbon particles (kinetics are not limited by the desorption reaction rate) was formulated to simulate mass
transfer of contaminants in the heterogeneous system. Analytical solutions were developed when contaminant
adsorption was governed by a linear isotherm. A single fitting parameter (tortuosity) brought simulations and
data into reasonable agreement. However, a common tortuosity could not be obtained for all compounds,
suggesting that desorption effects limit GAC recovery kinetics for some contaminants.
101
-------
• Based on experimental and model results, it is concluded that mass transport mechanisms limit the
effectiveness of Fenton's reaction for carbon recovery, at least for slightly soluble compounds (e.g., PCE) that
are reactive with Fenton's reagents. Therefore, optimal design for this type of treatment would maximize
contaminant flux from the sorbent while minimizing the use of H2O2, the primary contributor to process cost.
• Trials were conducted using Calgon URV-MOD 1 carbon on which iron had been precipitated onto the pore
and outer surfaces. No iron was added to the bulk regenerant (Fenton's solution). This was to localize the
Fenton's-driven radical generation near the GAC surface and the desorbing VOC targets, to minimize the rate
limitation due to desorption or pore and surface diffusion. However, negligible improvement in the rate of
carbon recovery was observed compared to the rate observed using non-iron-amended carbon. The rate of
H2O2 usage, the primary driver in operating cost of the system, was decreased by about three-fold in the iron-
amended system.
Field-Scale Regeneration Trials
• GAC was preloaded with methylene chloride (MC) and chloroform (CF) for field-scale regeneration
experiments. H2O2 was added hourly between hours 0-6 and 23-28. The reduction of solid-phase CF was 93%
after 30 hours. Residual MC could not be detected.
• GAC was loaded with trichloroethylene (TCE) in the laboratory and regenerated in the field using Fenton's
reagents through three loading/regeneration cycles. In each regeneration cycle, an initial loss of over 50% of the
TCE occurred in the first 4 hours, and after 60 hours TCE recovery was 73-95%. No loss of adsorption capacity
was observed after three GAC regenerations. H2O2 was added intermittently during regeneration. Analysis
indicates H2O2 additions timed to allow the bulk aqueous contaminant concentration to reach near-equilibrium
levels would minimize H2O2 cost.
• Temperature in the field reactor increased from an ambient value of about 30°C to 55-60°C during a 60-hour
regeneration period. Because VOC mass transport and reaction kinetics are favorably affected by higher
temperatures, the exothermic decomposition of H2O2 could increase carbon recovery rates and lower overall costs
for carbon surface regeneration.
• PCE, TCE and light diesel contaminants from the field site soil vapor extraction system were used to load
URV-MOD 1 GAC in a final set of field experiments. A 6-hour regeneration trial reduced adsorbed TCE by 80%,
but PCE recovery was only 30%.
• A scoping-level economic analysis was conducted, based on regeneration of PCE-loaded GAC (the most
challenging case investigated). The cost for Fenton Reagent regeneration based on the bench-scale studies
(~$2.70/lb) was comparable to that for conventional thermal regeneration (~$2.60/lb) and new carbon
replacement (~$3.30/lb). The cost of Fenton regeneration based on the field trials was higher (~$6.50/lb), but this
102
-------
may be a considerable overestimate since the primary cost driver, H2O2 consumption, was not optimized to any
extent in the field trial. Further tests in this regard are needed.
• It may be possible to immobilize iron on the carbon surface and regenerate in a pH range to avoid iron
dissolution during recovery operations. The feasibility of such a scheme depends on selection of iron loadings that
allow degradation reactions to proceed without blocking the carbon surface or interfering with contaminant access
to carbon pores. Field-scale trials using iron-amended GAC were not conducted. The results of bench-scale
experiments were used to estimate recovery costs in the iron-amended carbon systems. Projected regeneration
costs, ~$0.28/lb GAC, were approximately 10-fold lower than the cost of thermal regeneration or GAC
replacement.
Data Quality and Limitations
The data presented in this report was gathered using the procedures and tools detailed in the Quality
Assurance Project Plan (QAPP) submitted prior to project initiation. None of the procedures detailed in the QAPP
were required to be modified during the subsequent execution of the project. Consequently, there are no limitations on
the use of the data related to its original intended application beyond the following specific instances considerations.
• Using isopropanol as a hydroxyl radical scavenger, the experimental work was able to establish that
tetrachloroethylene was degraded in the Fenton's system by hydroxyl radical attack. However, because
no specific superoxide radical scavenger could be identified that itself was not subject to hydroxyl radical
attack, the work was not able to conclusively indicate that other halogenated VOCs, such as carbon
tetrachloride, are degraded by superoxide attack. The experimental results suggest that this is the case, but
other possible mechanisms cannot be completely ruled out.
• Considerable work was undertaken in modeling both the homogeneous and heterogeneous system
degradation of VOCs by Fenton's reagents. This work is largely beyond the original scope of the project
and was undertaken as a no cost extension of benefit to the project. Quality control and assurance for the
modeling approach was not detailed in the QAPP. Details of the approach, computer code, mathematical
algorithms, model assumptions and sources of input data are provided in Appendix A. In an effort to
validate the homogeneous model functioning, the project model was used to simulate the data and model
fit from the study by De Laat and Gallard (1999). The De Laat and Gallard study was conducted in a
much higher ionic strength and lower iron environment than the current work. The project model was
successful in simulating the literature study data, yet when the model was used to simulate data generated
in the conditions of interest in this project, the agreement was not nearly as good. Several efforts that
were made to reconcile the issue are detailed earlier, including temperature corrections, ionic strength
corrections and chloride concentration adjustments. Although the fit improved with these revisions, the
model is still not considered ready for use as a predictive tool and should not be applied unless first
103
-------
validated for a data set with the appropriate water composition conditions as will be used in the model.
Recommendations for addressing this shortcoming are given later in this section.
• The economic analysis reported was based (unless otherwise specified) on the results from the pilot-scale
treatment unit deployed in the field. Only very limited efforts were made to optimize this system,
particularly with respect to H2O2 use, which is the primary driver of the technology's cost. This fact
coupled with the observation that H2O2 use in the field was greater than 5-fold higher (per mass of carbon
regenerated) than in the lab trials, suggests that the cost estimates for the Fenton's based system over-
estimate the real cost of the technology deployment.
Research Recommendations
Based on this work and the perspective provided by other investigators it is possible to recommend additional,
follow-on research in a number of areas. These are:
1. Despite the contributions of De Laat and co-workers, uncertainties remain in details of both the Fenton
mechanism itself and its application for destruction of organic contaminants. The primary uncertainties
from the perspective of advanced oxidation processes include (i) the respective roles of hydroxyl and
superoxide radicals in transforming specific halogenated contaminants (e.g. CC14 and PCE) and (ii) matrix
effects on process kinetics and efficiency - that is, those imposed by non-target organics, partial
degradation products and reactive anions like Cl" and SO42". Extension of the De Laat kinetic model to
more chemically complex waters depends on reasonable representation of radical reaction kinetics with a
variety of chemical species that are usually omitted for simplicity. Difficulty in matching chloride-
dependent reaction kinetics with model predictions using published rate constants is apparent in this work.
This is unfortunate in light of the obvious importance of chloride ion concentrations in the context of our
investigation.
2. A reinforced De Laat chemical model can be used to help understand the accelerating effects of copper
and isopropanol additions that were observed in this work. In neither case are we able to offer a
convincing mechanism for compound-dependent changes in reaction kinetics, and explanations remain
speculative. Without additional work, the potential benefits of copper and cosolvent addition in Fenton-
based advanced oxidation systems are unlikely to be realized.
3. Certain other minor additions to the homogeneous Fenton model would be helpful. In particular, an easy
way to incorporate effects due to solution ionic strength and temperature could be important.
4. A great deal remains to be learned in the heterogeneous application of Fenton chemistry for carbon
recovery. That is, the roles of transport and reaction on overall recovery kinetics have not been fully sorted
out, as indicated by the inability to model recovery kinetics using the simplified approach adopted to date.
Pore diffusion is an unlikely controlling process for tightly held contaminants like PCE and the relative
104
-------
roles of pore diffusion and surface diffusion on intraparticle transport have not been sorted out. Desorption
kinetics need to be considered for such compounds (PCE, TCE) in the next modeling attempts.
5. Field operations provide yet another level of uncertainty that has not yet been fully evaluated. It is evident
that process economic feasibility depends on the efficient use of H2O2 for contaminant destruction, but
additional work is necessary to find a target-dependent H2O2 feed rate to promote such efficiency. A
systems approach to H2O2 scheduling and additional verifying experiments seem warranted.
6. Provision of Fe on the carbon surface provides operational possibilities (pH>3.0, soluble Cu(II) provision
to enhance kinetics, cosolvent effects) that have not yet been investigated. Much more effort here is
warranted.
7. The full breadth of compounds that can be destroyed in place following carbon adsorption has not yet
been identified. Reactivity with «OH seems not to be the most important issue, as indicated by the relative
success in previous NDMA experiments and the apparent stubbornness of PCE to Fenton-based recovery.
A through examination of compound parameters that affect the recovery kinetics for specific compounds
would be useful.
8. The presence of reactive co-contaminants on carbon to be regenerated using the propanol Fenton-based
technology may have consequences that have not yet been investigated.
105
-------
References
Adams, J.Q. and R.M. Clark. (1989) Cost estimates for GAC treatment systems. J. AWWA, 35-41.
Arnold, S., Hickey, W. and Harris, R. (1995). Degradation of Atrazine by Fenton's Reagent: Condition Optimization
and Product Quantification. Environ. Sci. Techno!., 29, 2083-2089.
ATSDR Information Center (last updated on June 22, 2001), ATSDR-Public Health Statements: Tetrachloroethylene,
URL http://atsdrl .atsdr.cdc.gov/ToxProfiles/phs8822.html.
Bansal, R., Bonnet J. and Stoeckli F. (1988). Active Carbon, New York: Marcel Dekker, Inc.
Bercik, L., (2003). "Heterogeneous Degradation of Chlorinated Solvents via Fenton's Reaction". Masters Thesis,
University of Arizona.
Betterton, E., Hollan, N., Arnold, R., Gogosha, S., Mckim, K. and Liu, Z. (2000). Acetone-Photosensitized Reduction
of Carbon Tetrachloride by 2-Propanol in Aqueous Solution. Environ. Sci. Technol, 34, 1229-1233.
Boltz, D. and Holwell, J. (1978). Colorimetric Determination ofNonmetals, 2nd. Edition, Chemical Analysis 8; John
Wiley and Sons, Inc.: New York, p 543.
Burbano, A., Dionysiou, D., Suidan, M., and Richardson, T. (2005). Oxidation kinetics and effect of pH on the
degradation of MTBE with Fenton reagent. Water Research, 39, 107-118.
Buxton, G., Greenstock, C., Helman, P. and Ross, A. (1988). Critical Review of Rate Constants for Reactions of
Hydrated Electrons, Hydrogen Atoms and Hydroxyl Radicals (OH/O). J. Phys. Chem. Ref. Data, Vol. 17, No. 2, p
513.
Carman, P., (1956). Flow of Gases through Porous Media, Academic Press, New York, N.Y.
Casero, I., Sicilia, D., Rubio, S. and Perez-Bendito, D. (1997). Chemical Degradation of Aromatic Amines by
Fenton's Reagent. Wat. Res. Vol. 31, No. 8, pp. 1985-1995.
Chen R. and Pignatello JJ. (1997). Role of quinone intermediate as electron shuttles in Fenton and Photoassisted
Fenton Oxidations of Aromatic compounds. Environ. Sci. Technol., 31, 2399-2406.
Chen, G., Hoagb, G., Chedda, P., Nadim, F., Woody, B. and Dobbs, G.(2001). The mechanism and applicability of in
situ oxidation of trichloroethylene with Fenton's reagent. J. Haz. Mat. 171-186.
106
-------
Clark, J. (2002) (December, 2005). The Electrochemical Series, URL
http: //www .chemguide. co .uk/physical/redoxeqia/ecs .html#top.
Clean Harbors Environmental (2004). Clean Harbors Environmental Services, Inc. 1 Hill Avenue, PO Box 859048,
Braintree, MA 02185-9048, ph: 781-792-5000. Personal (email) communications to procure hazardous waste disposal
costs.
Cotton, F., Wilkinson, G., Murillo, C. and Bochmann, M. (1999). Advanced Inorganic Chemistry, 6th Edition, John
Wiley & Sons, Inc. New York.
Crank, J. (1975). The Mathematics of Diffusion, Oxford University Press, New York, N.Y.
Crittenden, J., Hutzler, N., Geyer, D., Oravitz, J. and Friedman, G. (March 1986). Transport of Organic Compounds
with Saturated Groundwater Flow: Model Development and Parameter Sensitivity, J. Water Resources Research,
Vol. 22, No. 3, pp. 271-284.
Crittenden, J.C., Hand, D.W., Arora, H. and Lykins, B.W. (1987). Design Considerations for GAC Treatment of
Organic Chemicals. J. AWWA. 79, No. 1, 74-82.
Crittenden, J.C., Cortright, R.D., Rick, B., Tang, S.R. and Perram, D. (1988). Using GAC to Remove VOCs From Air
Stripper Off-Gas. J. AWWA. 80, 73-84.
De Laat, J. and Gallard, H. (1999). Catalytic Decomposition of Hydrogen Peroxide by Fe(III) in Homogeneous
Aqueous Solution: Mechanism and Kinetic Modeling, Environ. Sci. Technol. 33, 2726-2732.
De Laat, J. and Giang Le, T. (2005). Kinetics and Modeling of the Fe(III)/H2O2 System in the Presence of Sulfate in
Acidic Aqueous Solutions. Environ. Sci. Technol., 39, 1811-1818.
De Laat, J. and Giang Le, T (2006). Effects of Chloride Ions on the Iron(III)-Catalyzed Decomposition of Hydrogen
Peroxide and on the Efficiency of the Fenton-Like Oxidation Process. Appl. Cat. B: Environ., 66, 137-146.
De Laat, J., G.T. Le and B. Legube. (2004) A comparative study of the effects of chloride, sulfate and nitrate ions on
the rates of decomposition of H2O2 and organic compounds by Fe(II)/H2O2 and Fe(III)/H2O2. Chemosphere, 55:
715-723.
De Las Casas, C., Bishop, K., Bercik, L., Johnson, M., Potzler, M., Ela, W., Saez, A.E., Huling, S. and Arnold, R. In-
Place Regeneration of GAC using Fenton's Reagents. In Clark, C. and Lindner, A. (eds), "Innovative Approaches for
the Remediation of Subsurface-Contaminated Hazardous Waste Sites: Bridging Flask and Field Scales." ACS
Symposium Series, accepted for publication, Feb. 2006.
DiGiano, F.A. and Weber, W. J., Jr. (1973). Sorption Kinetics in Infinite-Batch Experiments. JWPCF, 45, 713-725.
Duesterberg, C., Cooper, W. and Waite, T. (2005). Fenton-Mediated Oxidation in the Presence and Absence of
Oxygen. Environ. Sci. Technol., 39, 5052-5058.
Eaton, A., Clesceri, L. and Greenberg, A. (1995). Standard Methods for the Examination of Water and Wastewater;
19th Edition, Publication Office, American Public Health Association: Maryland.
Environmental and Remediation Equipment (August, 2005). URL
http://www.soiltherm.com/GAC Comparison/gac comparison.html.
107
-------
EPA (2000). Granular Activated Carbon Adsorption and Regeneration,EPA-832-f'-00-017.
Fredrickson, J., Kostandarithes H., Li S., Plymale A. and Daly, M. (2000). Reduction of Fe(III), Cr(VI), U(VI), and
Tc(VII) by Deinococcus radiodurans Rl., Appl. Environ. Microbiol, 66, 2006-2011.
Furuya, E., Ikeda, H., and Takeuchi, Y. (1987). Fundamentals of Adsorption. Engineering Foundation, New York,
N.Y., 235.
Furuya, E., Chang, H., Miura, Y., Yokomura, H., Tajima, S., Yamashita, S. and Noll, K. (1996). Intraparticle Mass
Transport Mechanism in Activated Carbon Adsorption of Phenols. J. Environ. Eng., Vol. 122, No. 10, pp. 909-916.
Gallard, H., De Laat, J. and Legube, B. (1999). Spectrophotometric Study of the Formation of Iron (III)-Hydroperoxy
Complexes in Homogeneous Aqueous Solutions, Wat. Res., Vol. 33, No. 13, pp. 2929-2936.
Gallard, H. and De Laat, J. (2000). Kinetic Modeling of Fe(III)/H2O2 Oxidation Reactions in Dilute Aqueous Solution
Using Atrazine as a Model Organic Compound, Wat. Res. Vol. 34, No. 12, pp. 3107-3116.
Glaze, W., Kenneke, J. and Ferry, J. (1993). Chlorinated byproducts from the titanium oxide-mediated
photodegradation of trichloroethylene and tetrachloroethylene in water. Environ. Sci. Technol., 27, 177-184.
Grigor'ev, A., Makarov, I. and Pikaev, A. (1987). Formation of C12" in the Bulk Solution During the Radiolysis of
Concentrated Aqueous Solutions of Chlorides. High Energy Chem., 21, 99-102.
Haag, W.R. and Yao, C.C.D. (1992). Rate constants for reaction of hydroxyl radicals with several drinking water
contaminants. Environ. Sci. Technol., 2(5,1005-1013.
Heslop, R. and Robinson, P. (1967). Inorganic Chemistry. A Guide to Advanced Study, 3rd Ed., Elsevier Publishing
Company, Amsterdam.
Holleman, A. and Wiberg, E. (2001). Inorganic Chemistry, Academic Press, San Diego.
Ruling, S., Arnold, R., Sierka, R. and Miller, M. (1998). Measurement of Hydroxyl Radical Activity in a Soil Slurry
Using the Spin Trap -(4-Pyridyl-1 -oxide)- TV-fert-butylnitrone. Environ. Sci. Technol., 32, 3436-3441.
Ruling, S., Arnold, R., Jones, P. and Sierka, R. (2000a). Predicting Fenton-driven degradation using contaminant
analog. J. Environ. Eng., pp. 348-353.
Ruling, S., Arnold, R., Sierka, R., Jones, P. and Fine, D. (2000b). Contaminant Adsorption and Oxidation via Fenton
Reaction, J. Environ. Eng., pp. 595-600.
Ruling, S., Jones, P., Ela, W. and Arnold, R. (2005a). Fenton-driven chemical regeneration of MTBE-spent GAC,
Wat. Res. Vol. 39, No. 10, pp. 2145-2153.
Ruling, S.G., Jones, P.K., Ela, W.P., Arnold, R.G. (2005b). Repeated reductive and oxidative treatments on granular
activated carbon. J. Environ. Eng. Vol. 131, No. 2, pp. 287-297.
Ruling, S., Jones, P. and Lee, T. (In Press 2007). Iron Optimization for Fenton-Driven Oxidation of MTBE-Spent
Granular Activated Carbon. Environ. Sci. Technol.
King, RB. (2006). Encyclopedia of Inorganic Chemistry, 2nd. Edition, Copper: Inorganic and Coordination
Chemistry; John Wiley and Sons, Inc.(online service); Chichester; Hoboken: New Jersey, p. 14.
108
-------
Kommineni, S., Ela, W., Arnold, R., Ruling, S., Hester, B. and Betterton, E. (2003). NDMA Treatment by Sequential
GAC Adsorption and Fenton-Driven Destruction. Environ. Eng. Sci. Vol. 20, pp. 361-373.
Kremer, M.L. (2003) The Fenton reaction. Dependence of the rate on pH. J. Phys. Chem. A. 107(11): 1734-1741.
LaGrega, M., Buckingham, P., Evans, J. and Environmental Resources Management (2001). Hazardous Waste
Management, 2nd edition. San Francisco: McGraw Hill.
Letterman, RD. (1999). Water Quality and Treatment. A Handbook of Community Water Supplies. Fifth Edition,
Adsorption of Organic Compounds; McGraw-Hill, Inc.:New York, p. 13.3.
Liu, Z., Betterton, E. and Arnold, R. (2000). Electrolytic Reduction of Low Molecular Weight Chlorinated Aliphatic
Compounds: Structural and Thermodynamic Effects on Process Kinetics. Environ. Sci. Technol., 34, 804-811.
Logan, B. (1999). Environmental Transport Processes; John Wiley & Sons, Inc.: New York, p 654.
Lowry, G. and Reinhard, M. (2000). Pd-Catalyzed TCE Dechlorination in Groundwater: Solute Effects, Biological
Control, and Oxidative Catalyst Regeneration. Environ. Sci. Technol., Vol. 34, pp. 3217-3223.
Ma Z., Whitley, R. D. and Wang, N.-H. L. (May 1996). Pore and surface diffusion in multicomponent adsorption and
liquid chromatography systems. AIChEJ., Vol. 42, No. 5, pp.1244-1262.
Masarwa, M., Cohen, H., Meyerstein, D., Hickman, DL., Bakac, A. and Espenson, JH (1988). Reactions of Low-
Valent Transition-Metal Complexes with Hydrogen Peroxide, Are They "Fenton-like" or Not? 1. The Case of Cu+aq
and Cr2+aq., J. Am. Chem. Soc.110, 2493-4297.
Miller Brooks Environmental Inc. (2004). Final Draft Park-Euclid Remedial Investigation Report, Arizona
Department of Environmental Quality, Project No. 365-0011-01, Phoenix, AZ, 707.
Morel, F. and Hering, J. (1993) Principles and Applications of Aquatic Chemistry, Wiley-Interscience, New York.
National Research Council (1994). Alternatives for Ground Water Cleanup, National Academy Press, Washington,
DC, 315.
Noll, K. E, Gounaris, V. and Hou, W. (1992). Adsorption Technology for Air and Water Pollution Control, Lewis
Publishers, Michigan.
Petlicki, J. and van de Ven, T. (1998). The equilibrium between the oxidation of hydrogen peroxide by oxygen and
the dismutation of peroxyl or superoxide radicals in aqueous solutions in contact with oxygen. Faraday Trans., 94,
2763-2767.
Peyton, G., Bell, O., Girin, E. and Lefaivre, M. (1995). Reductive Destruction of Water Contaminants during
Treatment with Hydroxyl Radical Processes. Environ. Sci. Technol., 29, 1710-1712.
Poppe, MA. (2001). Analysis of MTBE and MTBE By-product Degradation Using Fenton-Type Reactions in a
Homogeneous System. Master's Thesis, Unpublished. University of Arizona.
Prousek, J. (1995). Fenton Reaction after a Century, Chem. Lisy, Vol. 89, pp. 11-21.
Radiation Chemistry Data Center of the Notre Dame Radiation Laboratory (May, 2003). URL
http://www.rcdc.nd.edu/.
109
-------
Radiation Chemistry Data Center of the Notre Dame Radiation Laboratory (August, 2005). URL
http://www.rcdc.nd.edu/.
Roberts, J. and Sawyer, D. (1981). Facile Degradation by Superoxide Ion of Carbon Tetrachloride, Chloroform,
Methylene Chloride, and p,p'-DDT in Aprotic Media. J. Am. Chem. Soc., 103, 712-714.
Ruthven, D. M. (1984). Principles of Adsorption and Adsorption Processes, Wiley-Interscience, New York.
Sedlak, D. and Andren, A. (1991). Aqueous-phase oxidation of poly chlorinated biphenyls by hydroxyl radicals.
Environ. Set. Technol, 25, 1419-1427.
Smith, B., Teel, A. and Watts, J. (2004). Identification of the Reactive Oxygen Species Responsible for Carbon
Tetrachloride Degradation in Modified Fenton's Systems. Environ. Sci. Technol., 38, 5465-5469.
Snoeyink, V. L. (1990). Adsorption of Organic Compounds, Water Quality and Treatment, 4th Ed., McGraw Hill,
New York, N.Y., 781-875.
Stark, J. and Rabani, J. (1999). Photocatalytic Dechlorination of Aqueous Carbon Tetrachloride Solutions in TiO2
Layer Systems: A Chain Reaction Mechanism. J. Phys. Chem. B, 103, 8524-8531.
Stumm, W. and Morgan, J. (1981). Aquatic Chemistry: An Introduction Emphasizing Chemical Equilibria in Natural
Waters. 2nd ed. Wiley-Interscience, New York.
Sudo, Y., Misic, M., and Suzuki, M. (1978). Concentration dependence of effective surface diffusion coefficient in
aqueous phase adsorption on activated carbon. Chem. Eng. Sci., 33, 1287-1290.
Suzuki, M., and Fujii, T. (1982). Concentration dependence of surface diffusion coefficient of propionic acid in
activated carbon particles. AIChEJ., 28, 380-385.
Suzuki, M., (1990). Adsorption Engineering, Elsevier, Amsterdam.
Swarzenbach, R., Gschwend, P. and Imboden, D. (1993). Environmental Organic Chemistry, John Wiley & Sons,
Inc., New York, p 681.
Teel, A., Warberg, C., Atkinson, D. and Watts, R. (2001). Comparison of Mineral and Soluble Iron Fenton's Catalysts
for the Treatment of Trichloroethylene. Wat. Res., Vol. 35, No. 4, pp. 977-984.
Teel, A. and Watts, R. (2002). Degradation of carbon tetrachloride by modified Fenton's reagent. J. Haz. Mat. 179-
189.
Toledo, L.C., Silva, A.C.B., Augusti, R., Lago, R.M., 2003. Application of Fenton reagent to regenerate activated
carbon saturated with organochloro compounds. Chemosphere, Vol. 50, pp. 1049-1054.
Teruya, E. (2000). Potential Regeneration of Activated Carbon Using Fenton's Reaction after MTBE Adsorption,
Master's Thesis, Unpublished, University of Arizona.
U.S.EPA. Wastewater Technology Fact Sheet (2000). Granular Activated Carbon Absorption and Regeneration,
U.S.EPA-Office of Water: Washington, DC, EPA 832-F-00-017.
Walling, C. (1975). Fenton's Reagent Revisited, Ace. Chem. Res., Vol. 8.
110
-------
Walling C. and Kato S. (1971). The Oxidation of Alcohols, by Fenton's Reagent, The Effect of Copper Ion., J. Am.
Chem. Soc., 93, pp. 4275-4280.
Watts, R., Bottenberg, B., Hess, T., Jensen, M. and Teel, A. (1999). Role of Reductants in the Enhanced Desorption
and Transformation of Chloroaliphatic Compounds by Modified Fenton's Reactions. Environ. Sci. Technol., 33, 3432-
3437.
Weber, W. J. Jr., (1972). Physiochemical Processes, John Wiley, New York, N.Y., 199-259.
Yang, T. Ralph (1987). Gas Separation by Adsorption Processes, Massachusetts, Butterworth Publishers.
Yoshida, M., Lee, B-D. and Hosomi, M. (2000). Decomposition of aqueous tetrachloroethylene by Fenton oxidation
treatment, Wat. Sci. Technol. Vol. 42, pp. 203-208.
Zoh, K. and Stenstrom, M. (2002). Fenton oxidation of hexahydro-l,3,5-trinitro-l,3,5-triazine (RDX) and octahydro-
l,3,5,7-tetranitro-l,3,5,7-tetrazocoine (HMX), Wat. Res., Vol. 36, pp. 1331-1341.
Ill
-------
-------
Appendices
113
-------
Appendix A. Computer Simulation of Homogeneous System
Appendix A. 1 Variable's Identification and Significance
Variables and their significance in the homogeneous model
Q1=Q(1)= [-OH]
Q2=Q(2)= [H02-]
Q3=Q(3)= [02- ]
xl = Wl = [Fe2+]
x2 = W2 = [H2O2]
W3 = [PCE]
W4 = [cr]
W5 = [CT]
x6 = FCTOTAL
x7 = [Fe3+]
x9 = [FeOH2+]
xlO = [Fe(OH)2+]
xll = [Fe2(OH)24+]
x!2 = [Ij] = [Fem(H02)2+]
x!3 = [I2] = [Fem(OH)(HO2)+]
x!4 = [FT]
x!5 = [SO42-]
Fl=d[Fe2+]/dt
F2 = d[H2O2]/dt
F3 = d[PCE]/dt
F4 = d[Cl']/dt
F5 = d[CT]/dt
Equilibrium constants are used to calculate the Fe(III) species. The constants were obtained directly from De Laat and
Gallard (1999). The value of the equilibrium constants was employed directly without substitution of the variables.
Rate constants named as XK6 through XK15 correspond to k6 to k!5 as defined by De Laat and Gallard (1999) and
included in this report (Table IV).
Additional rate constants XK20 - XK25 are included to simulate reactions of organic targets with the radicals as
follows:
XK20 = kOH,pcE PCE + OH -»
XK21=kOH,ci Cr
114
-------
XK22 = kOH,so4 SO42- + OH ->
XK23 = k02,cT CT + O2- ->
XK24 = k02,ci Cr + (V->
XK25 = k02,so4 SO42- + (V->
Grouping of variables to solve the Newton's method for the steady state concentration of the radicals was employed.
The terms ALPHA, BETA, GAMMA, DELTA, EPS and their equivalence are included in the coding. For example,
ALPHA = W1*W2*XK7.
In this model, the concentration of Fe3+ was calculated by solving a quadratic equation of the total iron. AFE, BFE,
CFE correspond to the a,b,c coefficients in the quadratic equation. The term DFE groups the terms inside the square
root.
The terms DISC, DI1, DI2, DI3 are used in solving the matrices in the Newton's method and Zl 1 - Z 45 in the Runge
Kutta method.
115
-------
Appendix A.2 Fortran Code for Non-Chlorinated Compounds
*
c
•*•
C RUNGE-KUTTA FOR SYSTEMS OF DIFFERENTIAL EQUATIONS ALGORITHM 5.7
*
c
C
C
c
C TO APPROXIMATE THE SOLUTION OF THE MTH-ORDER SYSTEM OF FIRST-
C ORDER INITIAL-VALUE PROBLEMS
C UJ' = FJ(T,U1,U2, . . . ,UM) , J=1,2,...,M
C A <= T <= B, UJ(A)=ALPHAJ, J=1,2,...,M
C AT (N+l) EQUALLY SPACED NUMBERS IN THE INTERVAL [A, B].
C
C INPUT ENDPOINTS A, B; NUMBER OF EQUATIONS M; INTIAL
C CONDITIONS ALPHA1, . . . , ALPHAM; INTEGER N.
C
C OUTPUT APPROXIMATIONS WJ TO UJ(T) AT THE (N+l) VALUES OF T.
C
DOUBLE PRECISION, DIMENSION ( 3, 4 ) :: AA
DOUBLE PRECISION, DIMENSIONS) :: Y
DOUBLE PRECISION, DIMENSION{3) :: Q
CHARACTER NAME 1*30, BB*1
INTEGER OOP, FLAG
LOGICAL OK
DOUBLE PRECISION XI, X2, X6, X7, X9, X10, Xll, X12, X13, X14 , X15
DOUBLE PRECISION XK6, XK7, XK8, XK9, XK1QA, XK10B, XK11A, XKiiB, XK12A
DOUBLE PRECISION XK12B, XK13A, XK13B, XK14A, XK14B, XK15, XK20
DOUBLE PRECISION Wl, W2, W3, TOL, ALPHA, BETA, GAMMA, DELTA, EPS, R, U
DOUBLE PRECISION AFE, BFE, CFE, DFE, DISC, DI1, DI2, DI3
DOUBLE PRECISION Zll, Z12, Z13, Z21 , Z22, Z23, Z31, Z32, 233, Z4 1, Z42, Z43
C CHANGE FUNCTION Fl THROUGH F5 FOR A NEW PROBLEM
C NOTE: THIS VERSION IS TO BE USED FOR NON-CHLORINATED COMPOUNDS.
C HCOOH REACTS ONLY WITH OH AND NO BY-PRODUCT REACTIONS ARE
CONSIDERED.
C Fl=Fe2+, F2=H202, F3=HCOOH OR ANY OTHER ORGANIC COMPOUND
Fl(T,Xi,X2,Ql,Q2,Q3,X7,X9,X10,Xll,X12,X13) = 2.7d-03*(X12+
*X13)-63.dO*Xl*X2-3.2d08*Xl*Ql-1.2d06*Xl*Q2-1.0d07*Xl*Q3+
*1.0d03* (X7+X9+X1Q+2*X11) *Q2 + 5. Od07* (X7+X9+X10+2*X11) *Q3
F2{T,Xi,X2,Ql,Q2,Q3) = -63.dO*Xl*X2-3.3d7*X2*Ql+
*8.3d5*Q2*Q2+9.7d7*Q3*Q2+5.2d9*Ql*Ql
F3{T,W3,Q1)= -XK20*W3*Q1
OPEN (UNIT=5, FILE= ' CON ' , ACCESS= ' SEQUENTIAL,' )
OPEN(UNIT=6,FILE='CON',ACCESS='SEQUENTIAL')
C DEFINE FUNCTIONS Fl,...,FM
WRITE(6,*) 'This is the Runge-Kutta Method for systems with m=2.'
WRITE(6,*) 'Remember to modify the function according to the
rxns'
116
-------
WRITE{6,*} 'Enter Y or N '
WRIT£{6,*} ' '
READ(5,*) BB
IF{( BB .EQ. 'Y' ) .OR. ( BB .EQ. 'y1 }} THEN
OK = .FALSE,
10 IF (OK) GOTO 11
WRITE(6,*) 'Input left and right endpoints separated by1
WRITE{6,*| 'blank'
WRITE(6,*) ' '
READ(5,*) A, B
IF (A.GE.B) THEN
WRITE(6,*) 'Left endpoint must be less1
WRITE(6,*) 'than right endpoint'
ELSE
OK = .TRUE.
ENDIF
GOTO 10
11 OK = .FALSE.
c WRITE(6,*) 'Input the TWO initial conditions.'
c WRITE{6,*} ' '
c READ(5,*) Wl, W2
12 IF (OK) GOTO 13
WRITE(6,*) 'Input a positive integer for the number'
WRITE(6,*) 'of subintervals '
WRITE(6,*) ' '
READ{5,*} N
IF ( N ,LE. 0 ! THEN
WRITE(6,*) 'Must be positive integer '
ELSE
OK = .TRUE.
ENDIF
GOTO 12
13 CONTINUE
ELSE
WRITE(6,*) 'The program will end so that the functions'
WRITE{6,*} 'F1-F5 can be created '
OK = .FALSE.
ENDIF
IF(.NOT.OK) GOTO 400
WRITE{6,*} 'Select output destination: '
WRITE(6.*) '1. Screen '
WRITE{6,*} '2. Text file '
WRITE(6,*) 'Enter 1 or 2 '
WRITE(6,*S ' '
READ(5,*} FLAG
IF ( FLAG .EQ. 2 ) THEN
WRITE(6,*) 'Input the file name in the form - '
WRITE(6,*) 'drive:name.ext'
WRITE(6,*5 'with the name contained within quotes'
WRITE(6,*5 'as example: "A:OUTPUT.DTA" '
WRITE(6,*5 ' '
READ(5,*) NAME1
OOP = 3
OPEN{UNIT=OUP,FILE=NAME1,STATUS='NEW')
ELSE
OOP = 6
ENDIF
WRITE(OOP,*) 'RUNGE-KUTTA AND NEWTON METHODS FOR SYSTEMS'
WRITE(OUP,6)
117
-------
6 FORMAT(5X,'T(i)',9X,'Wl(i)',8X,'W2(i)',8X,'W3(i)',8X,
"fa * f*\ 1 /-i1! * G V * lO""? f "i \ ' Q V * f°"\ "^! / 4 \ * QV "Vd/-i^ ' QV ' V "7 / i \ * QV * ' V G f -i ^ *
*fc/X \ X j / Q A ^ y^ \ X / f Q A f ^,3 | X / ^ O A f AO\X/ / O A ^ A / \ 1 / / o A / A:?l,iJ ,f
*8X,'XlO(i)',8X,'Xll(i)',8X,'X12(i)',8X,'X13(i)',8X,'XI4(i)')
C STEP 1
H=(B-A)/N
T=A
C STEP 2 INITIAL CONDITIONS
TOL=l.d-5
NN=10000
Ql=l.d-40
Q2=l.d-40
Q3=l.d-40
Q(1)=Q1
Q(2)=Q2
Q(3)=Q3
C W1=FE2+,W2=H202,W3=HCOOH,X15=S04
Wl=2.d-4
W2=2.2d-03
W3=2.d-07
X15=0.dO
C REACTION RATE CONSTANTS FROM GALLARD AND DE LAAT 1999
XK6=2.7d-3
XK7=63.dO
XK8=3.2d08
XK9=3.3d07
XK10A=1.2d06
XK10B=l.d07
XKllA=l.d3
XKllB=5,d07
XK12A=1.58d05
XK12B=l.dlO
XK13A=8.3d05
XK13B=9.7d07
XK14A=0.71dlO
XK14B-1.01dlO
XK15=5.2d09
C OH RXNS W/HCCOH
XK20=6.5d08
C IRON SPECIES. INITIAL CONDITIONS CALCULATED WITH EXCELL SOLVER
X6=2.d-G4
X7=l.d-40
X9=2,9000d-40
X10=7.6200d-41
Xll=8.0000d-78
X12=0.dO
X13=0.dO
X14=1.0000d-03
NJUMP=10000
XM=1
U=l
C STEP 3
WRITE(OOP,1) T,W1,W2,W3,Q1,Q2,Q3,X6,X7,X9,X10,X11,X12,X13,X14
C STEP 4
NJ1=0
C WRITE(OUP,*} 'l',Ql,Q2,Q3,Q(l),Q(2),Q(3)
DO 110 1=1,N
C STEP 5
118
-------
ALPHA=W1*W2*XK7
BETA=XK8*W1+XK9*W2+XK20*W3
GAMMA=XK6*(X12+X13)
DELTA=XK10A*W1+XK11A*(X7+X9+X10+2*X11)+XK12A
EPS=XK10B*W1+XK11B*(X7+X9+X10+2*X11)+XK12B*X14
AFE=2*.8d-3/{X14**2)
BFE=l+2.9d-3/X14+7.62d-7/(X14**2)+3.id-3*W2/X14+
*2.0d-4*2.9d-3*W2/(Xi4**2)
CFE=W1-X6
DFE=BFE**2-4*AFE*CFE
2200 IF(DFE.GT.O) GOTO 2100
WRITE(oup,*) 'WARNING: quadratic eqn is negative'
2100 X7=(-BFE+SQRT(DFE))/(2*AFE)
IF(X7.LT.Q) X7=0
X9=2.9d-03*X7/X14
IF(X9.LT.O) X9=0
X10=7.62d-07*X7/(X14**2)
IF(XIO.LT.O) X10=0
Xll=8.0d-04*(X7**2)/(X14**2)
IF(Xll.LT.O) Xll=0
X12=3.1d-03*X7*W2/X14
IF(X12.LT.O) X12=0
X13=2.0d-04*2.9d-G3*W2*X7/(X14**2)
1F{X13.LT.O) X13=0
X6=W1+X7+X9+X1Q+2*X11+X12+X13
IF(X6.LT.O) X6=0
C STEP 11
K=l
C STEP 12
100 IF (K.GT.NN) GOTO 200
C STEP13
C COMPUTE J(Q)
AA ( 1 , 1 ) =-BETA-XK14A*Q (2 ) -XK14B*Q { 3 ) -4 *XK15*Q ( 1 )
AA(1,2)=-XK14A*Q{1)
AA(1,3)=-XK14B*Q(1)
AA(2,1)=XK9*W2-XK14A*Q(2S
AA ( 2 , 2 ) =-DELTA-4 *XK1 3A*Q { 2 ) -XK13B*Q { 3 ) -XK14A*Q { 1 )
AA(2,3)=XK12B*X14-XK13B*Q(2)
AA(3,1)=-XK14B*Q(3)
AA ( 3 , 2 ) =XK12A-XK1 3B*Q { 3 )
AA(3,3)=-EPS-XK13B*Q(2)-XK14B*Q{1)
C COMPUTE -F(Q)
AA { 1 , 4 ) =-ALPHA+BETA*Q { 1 } +XK14A*Q ( 1 ) *Q ( 2 ) *XK14B*Q { 1 ) *Q { 3 )
*+2*XK15*Q(l) *Q(1)
AA ( 2, 4 ) =-GAMMA-XK9*W2*Q { 1 ) +DELTA*Q (2 ) -XK12B*X14 *Q { 3 )
*+2*XK13A*Q(2)*Q(2) +XK13B*Q{2) *Q ( 3) +XK14A*Q ( 1 ) *Q(2)
AA(3,4)=EPS*Q(3)-XK12A*Q(2)+XK13B*Q{2) *Q ( 3) +XK14B*Q ( 1 } *Q (3)
C STEP 14
C SOLVES THE N X N LINEAR SYSTEM J(Q) Y = -F(Q)
DI SC=AA (1,1) *AA (2,2) *AA (3,3) +AA (1,3) * AA (2,1) * AA (3,2)
* +AA(1,2) *AA(2,3)*AA(3,1)-AA(1,3)*AA(2,2)*AA{3,1)
* -AA(1,2)*AA{2,1S*AA(3,3)-AA(3,2)*AA(2,3)*AA(1,1)
DI1=AA(1,4) *AA(2,2S *AA(3, 3) +AA(1, 3} *AA(2, 4 ) *AA(3,25
+AA(1,2] *AA(2,3) *AA (3, 4 ) -AA ( 1 , 3) *AA(2, 2) *AA (3, 4 )
-AA(1,2)*AA{2,4)*AA{3,3)-AA(3,2}*AA(2,3)*AA{1,4}
119
-------
DI2=AA(1,1)*AA{2,4)*AA(3,3}+AA(1,3)*AA(2,1}*AA{3,4]
* +AA(1,4)*AA(2,3)*AA(3,1)-AA(1,3)*&A{2,4}*AA(3,1)
* -AA(1,4)*AA(2,1)*AA(3,3}-AA(3,4)*AA(2,3j *AA{1,1}
DI3=AA(1,1)*AA(2,2)*AA(3,4)+AA(1,4)*AA(2,1)*AA{3,2]
* +AA(1,2)*AA(2,4)*AA(3,1)-AA(1,4)*AA{2,2}*AA(3,1)
* -AA(1,2)*AA(2,1)*AA(3,4)-AA(3,2)*AA(2,4)*AA(1,1}
Y(1)=DI1/DISC
Y(2)=DI2/DISC
Y(3)=DI3/DISC
C STEPS 15 AND 16
R=0.
DO 20 L=l,3
IF (ABS{Y{L)/Q(L)).GT.R) R=ABS(Y(L)/Q(L))
20 CONTINUE
Q(1)=Q(1)+Y(1)*XM
Q(2)=Qf2)+Y{2)*XM
Q(3}=Q(3)+Y{3)*XM
C WRITE(OUP,*) t2',Ql,Q2,Q3,Q(l),Q(2),Q{3)
IF {Q(L).LT.O) Q(L)=ld-40
Ql = Qd!
Q2 = Q(2)
Q3 = Q(3)
C WRITE(OOP,*) I4',Q1,Q2,Q3,Q{1),Q(2},Q{3)
IF (R.LT.TOL) THEN
C PROCESS IS COMPLETE
C WRITE(OOP,4)
GOTO 70
END IF
C STEP 17
K=K+1
C WRITE{OUP,*} 'OUTPUT NEWTON'
C WRITEjOUP,*) K,X7,R,Y(1),Y(2),Y(3),Q(1},Q(2),Q{3}
C WRITE(OUP,*) AA(1,1),AA(1,2),AA(1,3),AA(1,4)
C WRITE(OOP,*} AA(2,1),AA{2,2),AA(2,3),AA{2,4)
C WRITEfOUP,*) AA(3,1),AA(3,2),AA(3,3),AA{3,4)
C WRITE(OOF,*) 0150,011,012,013
GOTO 100
70 CONTINUE
Z11=H*F1{T,W1,W2,Q1,Q2,Q3,X7,X9,X10,X11,X12,X13S
Z12=H*F2{T,W1,W2,Q1,Q2,Q3)
Z13=H*F3(T,W3,Q1)
C STEP 6
Z21=H*F1(T+H/2,W1+Z11/2,W2+Z12/2,Q1,Q2,Q3,X7,X9,X10,X11,
*X12,X13)
Z22=H*F2{T+H/2,W1+Z11/2,W2+Z12/2,Q1,Q2,Q3)
Z23=H*F3(T+H/2,W3+Z13/2,Q1)
C STEP 7
Z31=H*F1(T+H/2,W1+Z21/2,W2+Z22/2,Q1,Q2,Q3,X7,X9,X10,X11,
*X12,X13!
Z32=H*F2(T+H/2,W1+Z21/2,W2+Z22/2,Q1,Q2,Q3J
Z33=H*F3(T+H/2,W3+Z23/2,Q1)
120
-------
STEP 8
Z41=H*F1(T+H,W1+Z31,W2+Z32,Q1,Q2,Q3,X7,X9,X1Q,X11,X12,
*X13)
Z42=H*F2(T+H,W1+Z31,W2+Z32,Q1,Q2,Q3)
Z43=H*F3(T+H,W3+Z33,Q1)
STEP 9
Wl=Wl+(Zll+2*Z21+2*Z31+Z41)/6
W2=W2+«Z12+2*Z22+2«Z32+Z42)/6
W3=W3+{Z13+2*Z23+2'Z33+Z43)/6
WRITE(OOP,*3 '3',Q1,Q2,Q3,Q(1),Q(2),Q[3)
C STEP 10
T=A+I*H
C STEP 19
NJ1=NJ1+1
IF(NJl.LT.NJUMP) GOTO 110
NJ1=0
WRITE {OOP, 1) T,W1,W2,W3,Q(1) ,Q{2) ,Q(3) , X6, X7 , X.9, X10, Xll, X12,
*X13,X14
c WRITECQUP,*) Fl(T,X1,X2,Q1,Q2,Q3,X7,X9,X10,X11,X12,X13) ,
c *F2(T,X1,X2,Q1,Q2,Q3)
110 CONTINUE
GOTO 400
C STEP 18
C DIVERGENCE
200 WRITE{OOP,5) NN
C STEP 20
400 CLOSE(UNIT=5)
CLOSE(UNIT=OUP)
IF(OUP.NE.6) CLOSE(UNIT=6)
STOP
1 FORMAT(16(1X,E12.6)}
2 FORMAT(5{IX,E12.6})
3 FORMAT(IX,'ITER.=',IX,12,IX,'APPROX. SOL.
IS1,1X,3(1X,E15.8),/,lX,
*'APPROX. ERROR IS',IX,E15.8)
4 FORMAT(IX,'SUCCESS WITHIN TOLERANCE l.OE-4')
5 FORMAT(IX,'DIVERGENCE - STOPPED AFTER ITER.',IX,12)
END
121
-------
Appendix A.3 Fortran Code for Chlorinated Compounds
c
*
C RUNGE-KUTTA FOR SYSTEMS OF DIFFERENTIAL EQUATIONS ALGORITHM 5.7
*
C
*
Q*******^*************#****^**^************
*
C
C
C
C TO APPROXIMATE THE SOLUTION OF THE MTH-ORDER SYSTEM OF FIRST-
C ORDER INITIAL-VALUE PROBLEMS
C UJ' = FJ(T,U1,U2,...,UM) , J=1,2,...,M
C A <= T <= B, UJ(A)=ALPHAJ, J=1,2,...,M
C AT (N+15 EQUALLY SPACED NUMBERS IN THE INTERVAL [A,B].
C
C INPUT ENDPOINTS A,B; NUMBER OF EQUATIONS M; INTIAL
C CONDITIONS ALPHA1,..,,ALPHAM; INTEGER N.
C
C OUTPUT APPROXIMATIONS WJ TO UJ{T) AT THE (N+l) VALUES OF T.
C
DOUBLE PRECISION, DIMENSION(3,4) :: AA
DOUBLE PRECISION, DIMENSION(3) :: Y
DOUBLE PRECISION, DIMENSION(3} :: Q
CHARACTER NAME1*30,BB*1
INTEGER OUP,FLAG
LOGICAL OK
DOUBLE PRECISION XI,X2,X6,X7,X9,X10,Xll,X12,X13, X14,X15
DOUBLE PRECISION XK6,XK7,XK8,XK9,XK10A,XK10B,XK11A,XK11B,XK12A
DOUBLE PRECISION
XK12B,XK13A,XK13B,XK14A,XK14B,XK15,XK20,XK21, XK22
DOUBLE PRECISION
Wl,W2, W3,W4,W5,TOL,ALPHA,BETA,GAMMA,DELTA,EPS,R,U
DOUBLE PRECISION XK23,XK24,XK25,AFE,BFE,CFE,DFE,DISC,DI1, DI2,DI3
DOUBLE PRECISION Zil,Z12,213,Z14,215,221,Z22,Z23,Z24,225,Z31,Z32,
*Z33,234,235,241,242,Z43,Z44,Z45
C CHANGE FUNCTION Fl THROUGH F5 FOR A NEW PROBLEM
C NOTE: THIS VERSION TO BE USED FOR CHLORINATED COMPOUNDS.
C IN THIS MODEL, CT REACTS ONLY WITH 02 AND PCE ONLY WITH OH.
C Chloride is a conservative specie.
C Fi=Fe2+, F2=H202, F3=PCE, F4=C1-, F5=CT
F1(T,X1,X2,Q1,Q2,Q3,X7,X9,X10,X11,X12,X13) = 2.7d-03*(X12+
*X13)-63.dO*Xl*X2-3.2d08-Xl*Ql-1.2d06*Xl*Q2-1.0d07*Xl*Q3+
*1.0d03*{X7+X9+XIO+2*X11}*Q2+5.0d07*{X7+X9+X10+2*X11)*Q3
F2(T,X1,X2,Q1,Q2,Q3} = -63dQ*Xl*X2-3 . 3d7*X2*Qlt-
*8 . 3d5*Q2*Q2-r9 . 7d7*Q3*Q2T-5 . 2d9*Ql*Ql
F3{T,W3,Q1)= -XK20*W3*Q1
F4{T,W3,Q1,W5,Q3)= 4*XK20*W3*Q1+4*XK23*W5*Q3
F5{T,W5,Q3)= -XK23*W5*Q3
122
-------
OPEN(UNIT=5, FILE='CON',ACCESS='SEQUENTIAL')
OPEN(UNIT=6,FILE='CON',ACCESS='SEQUENTIAL')
C DEFINE FUNCTIONS Fl,..,,FM
WRITE(6,*) 'This is the Runge-Kutta Method for systems with m=2.'
WRITE(6,*) "Remember to modify the function according to the
rxns1
WRITE(6,*) 'Enter Y or N '
WRITE(6,*5 ' '
READ(5,*) BB
IF({ BB .EQ. 'Y' ) .OR. ( BB .EQ. 'y' )} THEN
OK = .FALSE.
10 IF (OK) GOTO 11
WRITE(6,*) 'Input left and right endpoints separated by"
WRITE(6,*) 'blank1
WRITE(6,*) ' '
READ(5,*) A, B
IF (A.GE.B) THEN
WRITE(6,*) 'Left endpoint must be less'
WRITE(6,*) 'than right endpoint1
ELSE
OK = .TRUE.
ENDIF
GOTO 10
11 OK = .FALSE.
c WRITEJ6,*) 'Input the TWO initial conditions.'
c WRITE(6,*) ' '
c READ(5,*} Wl, W2
12 IF (OK) GOTO 13
WRITE(6,*) 'Input a positive integer for the number'
WRITE(6,*) 'of subintervals '
WRITE(6,*) ' '
READ(5,*) N
IF ( N .LE. 0 } THEN
WRITE(6,*) 'Must be positive integer '
ELSE
OK = .TRUE.
ENDIF
GOTO 12
13 CONTINUE
ELSE
WRITE(6,*) 'The program will end so that the functions'
WRITE{6,*} 'F1-F5 can be created '
OK = .FALSE.
ENDIF
IF{.NOT.OK) GOTO 400
WRITE{6,*) 'Select output destination: '
WRITE(6,*3 '1. Screen '
WRITE{6,*) '2. Text file '
WRITE(6,*) 'Enter 1 or 2 '
WRITE(6,*) ' '
READ(5,*) FLAG
IF ( FLAG .EQ. 2 } THEN
WRITE(6,*) 'Input the file name in the form - '
WRITE(6,*) 'drive:name.ext'
WRITE(6,*) 'with the name contained within quotes'
WRITE(6,*) 'as example: ''A:OUTPUT.DTA'' '
WRITE(6,*) ' '
READ(5,*) NAME1
OUP = 3
OPEN{UNIT=OUP,FILE=NAME1,STATUS='NEW')
123
-------
ELSE
OOP = 6
ENDIF
WRITE(OUP,*) 'RUNGE-KUTTA AND NEWTON METHODS FOR SYSTEMS'
WRITE(OUP,6)
FORMAT(5X,'T(i)',9X,'Wl(i)',8X,'W2(i)',8X,'W3(i)',8X,'W5',8X,
*'Ql{i)',8X,'Q2(i)',8X,'Q3{i)',8X,'X6(i>',SX,'X7(i)',8X,'X9(i)',
*8X, 'XlO(i) ',8X, 'XH (i) ',8X, 'X12 (i) ', 8X, 'X13(i) ', 8X, 'XI4 (i) ' }
STEP 1
H=(B-A)/N
T=A
STEP 2 INITIAL CONDITIONS
TOL=l.d-5
NN=10000
Ql=l.d-40
Q2=l.d-40
Q3=l.d-40
Q(D=Q1
Q(2)=Q2
Q{3)=Q3
W1=FE2+,W2=H202,W3=PCE,W4=C1-,W5=CT,X15=S04
Wl=l.d-40
W2=l.ld-01
W3=8.9d-05
W4=0.dO
W5=0.dO
X15=0.dO
REACTION RATE CONSTANTS FROM GALLARD AND DE LAAT 1999
XK6=2.7d-3
XK7=63.dO
XK8=3.2d08
XK9=3.3d07
XKiOA=1.2d06
XK10B=l.d07
XKllA=l.d3
XKllB=5.d07
XK12A=1.58d05
XK12B=l.dlO
XK13A=8.3d05
XK13B=9.7d07
XK14A=0.71dlO
XK14B=1.01dlO
XK15=5.2d09
OH RXNS W/PCE,C1,S04
XK20=2.d09
XK21=3.d09
XK22=3.5d05
02 RNXS W/CT,C1,SO4
XK23=3.8d03
XK24=140.dO
XK25=0.dO
IRON SPECIES. INITIAL CONDITIONS CALCULATED WITH EXCELL SOLVER
X6=9.00d-05
X7=6.75d-05
X9=1.9573d-05
X10=5.1431d-07
Xll=3.6444d-08
X12=2.30157d-06
X13=4.30616d-08
X14=1.0000d-02
NJUMP=10000
124
-------
XM=1
U=l
C STEP 3
WRITE(OOP,1) T,W1,W2,W3,W5,Q1,Q2,Q3,X6,X7,X9,X10,X11,X12,X13,X14
C STEP 4
NJ1=0
C WRITECOUP,*) 1l',Ql,Q2,Q3,Q(l),Q(2),Q(3)
DO 110 1=1,N
C STEP 5
ALPHA=W1*W2*XK7
BETA=XK8*W1+XK9*W2+XK20*W3+XK21*W4+XK22*X15
GAMMA=XK6*(X12+X13)
DELTA=XK10A*W1+XK11A*(X7+X9+X10+2*XX1)+XK12A
EPS=XK10B*W1+XK11B*(X7+X9+X10+2*X11!+XK12B*X14+XK23*W5+XK24*W4+
*XK25*X1S
C QUADRATIC EQN TO SOLVE FOR FE3+
AFE=2*.8d-3/(X14**2)
BFE=l+2.9d-3/X14+7.62d-7/(X14**2)+3.ld-3*W2/X14+
*2.Qd-4*2.9d-3*W2/{X14**2)
CFE=W1-X6
DFE=BFE* *2-4 *AFE*CFE
2200 IF(DFE.GT.O) GOTO 2100
WRITE(oup,*) 'WARNING: quadratic eqn is negative'
2100 X7=(-BFE+SQRT(DFE))/(2*AFE)
IF(X7.LT.O) X7=0
X9=2.9d-03*X7/X14
IF(X9,LT.O) X9=0
X10=7.62d-07*X7/(X14**2)
IF(XIO.LT.O) X10=0
Xll=8.0d-04*(X7**2)/(X14**2}
IF(Xll.LT.O) X-ll=0
X12=3.ld-03*X7*W2/X14
IF(X12.LT.O) X12=0
X13=2.0d-04*2.9d-03*W2*X7/!X14**2)
IF(X13.LT.O) X13=0
X6=Wl+X7-fX9+X10+2*Xll+X12-i-X13
IF(X6.LT.O) X6=0
C STEP 11
K=l
C STEP 12
100 IF (K.GT.NN) GOTO 200
C STEP13
C COMPUTE J|Q5
AA(1,1)=-BETA-XK14A*Q(2)-XK14B*Q(3)-4*XK15*Q(1)
AA(1,2)=-XK14A*Q(l)
AA(1,3)=-XK14B*Q(l)
AA(2,1)=XK9*W2-XK14A*Q(2)
AA(2,2)=-DELTA-4*XKl3A*Q{2)-XK13B*Q(3)-XK14A*Q(i;
AA(2,3)=XK12B*X14-XK13B*Q(2)
AA(3,1)=-XK14B*Q(3)
AA{3,2)=XK12A-XK13B*Q{3)
AA{3/3)=-EPS-XK13B*Q(2)-XK14B*Q(l}
125
-------
C COMPUTE -F(Q)
AA(i, 4}=-ALPHA+B£TA*Q{l)-i-XK14A*Q (1) *Q(2) +XK14B*Q (1) *Q(3)
*+2*XK15*Q(l)*Q(l)
AA{2,4}=-GAMMA-XK9*W2*Q(l)+DELTA*Q(2)-XK12B*X14*Q{3)
*+2*XK13A*Q(2)*Q(2}+XK13B*Q(2}*Q(3}+XK14A*Q(1}*Q(2}
AA(3,4)=EPS*Q(3)-XK12A*Q(2)+XK13B*Q{2)*Q(3)+XK14B*Q(1)*Q(3)
C STEP 14
C SOLVES THE N X N LINEAR SYSTEM J(Q) Y = -F{Q)
DISC=AA(1,1}*AA(2,2)*AA(3,3)+AA(1,3)*AA(2,1}*AA(3,2)
* +AA(1,2)*AA(2,3}*AA(3/1}-AA(1,3)*AA(2,2)*AA{3,1}
* -AA(1,2)*AA(2,1)*AA(3,3)~AA(3,2)*AA(2,3)*AA(1,1)
DI1=AA(1,4}*AA(2,2)*AA(3,3)+AA(1,3}*AA(2,4)*AA(3,2}
* +AA(1,2)*AA{2,3)*AA(3,4)-AA(1,3}*AA(2,2}*AA(3,4}
* -AA(1,2)*AA(2,4)*AA{3,3)-AA(3,2)*AA(2,3)*AA(1,4)
DI2=AA(1,1)*AA(2,4)*AA(3,3)+AA(1,3)*AA(2,1)*AA(3,4)
* +AA(1,4)*AA(2,3)*AA (3,1)-AA(1,3)*AA(2,4)*AA(3,1)
* -AA(1,4)*AA(2,1)*AA{3,3)-AA{3,4)*AA{2, 3}*AA(1, 1}
DI3=AA(1,1)*AA{2,2)*AA{3,4}+AA{1,4)*AA{2,1)*AA(3,2)
* +AA(1,2)*AA(2,4)*AA(3,1)-AA(1,4)*AA(2,2)*AA(3,1)
* -AA(1,2)*AA(2,1)*AA(3,4)-AA(3,2)*AA(2,4)*AA(1,1)
Y(1)=DI1/DISC
Y(2)=DI2/DISC
Y(3)=DI3/DISC
C STEPS 15 AND 16
R=0.
DO 20 L=l,3
IF (ABS(YfL)/Q(L)).GT.R) R=ABS(Y(L)/Q(L)}
2,0 CONTINUE
Q(1)=Q(1)+Y{1)*XM
Q{2)=Q(2}+Y(2)*XM
Q(3)=Q(3)+Y(3)*XM
C WRITECOUP,*} '2',Q1,Q2,Q3,Q{1),Q(2),Q(3)
IF (Q(L).LT.O) Q(L}=ld-40
Ql = Q(D
Q2 = Q{2)
Q3 = 0(3}
C WRITECOUP,*} '4',Ql,Q2,Q3,Q(l),Qt2),Q(3)
IF (R.LT.TOL) THEN
C PROCESS IS COMPLETE
C WRITE(OUP,4}
GOTO 70
END IF
C STEP 17
K=K+1
C WRITE(OOP,*) 'OUTPUT NEWTON'
C WRITEfOUP,*) K,X7,R,Y{1),Y(2),Y(3},Q(1),Q(2),Q{35
C WRITE(OUP,*) AA(1,1),AA(1,2),AA(1,3),AA{1,4)
C WRITE(OUP,*) AA(2,1),AA(2,2),AA{2,3],AA(2, 4}
C WRITE(OOP,*) AA(3,1),AA(3,2),AA{3,3),AA(3,4}
C WRITE(OUP,*) DISC,DI1,DI2,DI3
126
-------
GOTO 100
70 CONTINUE
Z11=H*F1(T,W1,W2,Q1,Q2,Q3,X7,X9,X10,X11,X12,X13)
Z12=H*F2(T,Wl,W2,Ql,Q2,Q3)
Z13=H*F3(T,W3,Q1)
Z14=H*F4(T,W3,Q1,W5,Q3)
Z15=H*F5(T,W5,Q3)
C STEP 6
Z21=H*Fl{T+H/2,Wl+Zil/2,W2+212/2,Ql,Q2,Q3,X?,X9,X10,Xil,
*X12,X13)
Z22=H*F2(T+H/2,Wl+Zll/2,W2+Z12/2,Qi,Q2,Q3)
Z23=H*F3(T+H/2,W3+Zi3/2,Ql)
Z24=H*F4(T+H/2,W3+Z13/2,Q1,W5+Z15/2,Q3)
Z25=H*F5(T+H/2,W5+Z15/2,Q3)
C STEP 7
Z31=H*F1(T+H/2,W1+Z21/2,W2+Z22/2,Q1,Q2,Q3,X7,X9,X1Q,X11,
*X12,X13)
Z32=H*F2(T+H/2,W1+Z21/2,W2+Z22/2,Q1,Q2,Q3)
Z33=H*F3(T+H/2,W3+Z23/2,Q1)
Z34=H*F4(T+H/2,W3+Z23/2,Q1,W5+Z25/2,Q3)
Z35=H*F5(T+H/2,W5+Z25/2,Q3)
C STEP 8
Z41=H*F1 {T+H,W1 + Z31,W2+Z32,Q1,Q2,Q3,X7,X9,X10,X11,X12,
*X13)
Z42=H*F2(T+H,W1+Z31,W2+Z32,Q1,Q2,Q3)
Z43=H*F3(T+H,W3+Z33,Q1)
Z44=H*F4(T+H,W3+Z33,Q1,W5+Z35,Q3)
Z45=H*F5(T+H,W5+Z35,Q3)
C STEP 9
W1=W1+(Z11+2*Z21+2*Z31+Z41)/6
W2=W2+ (Z12'+2*Z22+2*Z32 + Z42) /6
W3=W3+(Z13+2*Z23+2*Z33+Z43)/6
W4=W4+(214+2*224+2*234+244)/6
W5=W5+(Z15+2*Z25+2*Z35+Z45]/6
C WRITE{OUP,*} '3',Q1,Q2,Q3,Q(1),Q(2},Q(3)
C STEP 10
T=A+I*H
C STEP 19
NJ1=NJ1+1
IF(NJl.LT.NJUMP) GOTO 110
NJ1=0
WRITE(OOP,1)
T,W1,W2,W3,W5,Q{1),Q{2),Q(3),X6,X7,X9,X10,Xll,X12,
*X13,X14
c WRITE(OUP,*) F1{T,X1,X2,Q1,Q2,Q3,X7,X9,X10,X11,X12,X13),
c *F2(T,X1,X2,Q1,Q2,Q3)
110 CONTINUE
GOTO 400
C .STEP 18
C DIVERGENCE
200 WRITE(OUP, 5} NN
C STEP 20
127
-------
400 CLOSE(UNIT=5)
CLOSE(UNIT=OUP)
IF{OUP.NE.6) CLOSE(UNIT=6)
STOP
1 FORMAT(16(IX,E12.6))
2 FORMAT(5(1X,E12.6])
3 FORMAT(IX,'ITER.=',IX,12,IX,'APPROX. SOL,
IS',1X,3(1X,E15.8),/,IX,
*'APPROX. ERROR IS',IX,E15.8)
4 FORMAT(IX,'SUCCESS WITHIN TOLERANCE 1.OE-4')
5 FORMAT(IX, 'DIVERGENCE - STOPPED AFTER ITER. ',IX, 12;
END
128
-------
Appendix B. Comparing Costs of Carbon Disposal vs. (Fenton) Carbon Regeneration
The economic analysis conducted here was designed to compare alternative carbon replacement/regeneration
strategies in processes using activated carbon for contaminant adsorption. In scenario #1, spent carbon is replaced
with new activated carbon, and the waste carbon is disposed of as a hazardous waste. Scenario #2 differs in that
carbon is periodically regenerated, or at least partially regenerated using Fenton's reagents to destroy the adsorbed
contaminants. The economic analysis was carried out by comparing costs that are unique to each scenario on both a
present worth and an annual cost basis. Most of the costs for activated carbon adsorption for a gas-phase streams
derived from SVE (the basic scheme for both scenarios) are common to the alternatives compared. As such they are
omitted from the analysis. These include energy costs for the SVE system, capital costs for the carbon adsorption unit,
initial carbon costs and some maintenance and other labor activities. A description of the costs that are unique to each
alternative follows:
Scenario # 1 Scenario #2
GAC replacement cost
GACcost = $1.60/lb -0/1 . .. ,
„.. ,
-------
• The concentration of contaminant (PCE or MC) in the gas treated by SVE is 100 ppmv
• GAC column diameter = 1 m
• The mass of carbon in the column is irrelevant since we will calculate carbon wastage rates for each scenario
investigated
• The pump efficiency during recirculation of Fenton's reagents is 0.70
• The discount generator is 0.08
• Equipment for H2O2 dosing and storage has a service life of 20 years
• All GAC costs are in year 2000 dollars and all other costs are current
Calculations:
Estimating time to breakthrough:
At a MC-gas phase concentration of 100 ppm, the concentration in the liquid can be calculated using Henry's law.
From La Grega et al., (2001),
H=exp(A-B/T) where A = 6.65 and B=3.82xl03
then the Henry's law constant can be calculated at a T = 25°C (298 K)
3 82 T! o3 I
6.65 — : - \ = 2.09xW-3atm-m3 I mol = 2.\atml M
298
=xx
L,ML. -r~r r\ -\ j I ~k f 7 *-*
H 2.\atmlM mol g
At saturation, the concentration of MC in equilibrium with the solid is
CSMC = KCeql'"=0.069(mg/g)(L/mg)lx(4.05mg/L)l=0.280mgMC/gGAC
where K and 1/n are the Freundlich parameters for MC and the URV-MOD1 GAC. The concentration of PCE in the
solid is 250 mg/g (from the field data).
The carbon wastage rate at Qair =10 cfm,
• Wft3 l.Sgal 3.78Z Imol 1 11Ain-4 ,/ •
m°C,=QairCair=^^x—%-x - -x— — — x— T = 11.6*10 mollmm
mm ft gal 24.4651 10
\65.S3xW3mgPCE
mc = - mn - _ - mo - = 0.769gO4C/min
250mgPCE / gGAC
mc - 0.769^G^C / min - 439lbsGAC 1 6months - 878/fc / yr
130
-------
At a gas flowrate of 10 cfm, the carbon wastage rate for PCE-loaded GAC is nearly 9001bs/yr and for MC is
mc -•
\\.6xWmol 85x\tfmgMC
x —
min mol
0.2SOmgMC / gGAC
= 356gGAC/mm
The carbon wastage rate at Qair =100 cfm,
= 7.69 gGAC I'mm = 4385lbsGAC / 6months = SJJOlbs / yr
mc,MC = 35 ISgGAC I mm = 2,005,979lbsGAC 16months = 4,011,958/fe Iyr
Up to this point, all calculations are common to both option 1 and 2 (See the following table).
Properties and GAC Wastage Rate for Methylene Chloride and PCE
Compound
Methylene
Chloride (MC)
Perchloroethylene
(PCE)
Form.
CH2C12
C2C14
Initial
Cone.
(ppm)
100
H
(atm/M)
2.1
cs
(mg/g)
KCeq1/n
0.280
250
GAC wastage rate
(Ibs /yr)
Qair=10
cfm
405,961
878
Qair=100
cfm
4,011,958
8770
Note: MC Henry's law constant calculated with the parameters A = 6.65, B=3.82xl03 and T = 25°C.
The PCE concentration in the solid is measured from field data.
For option 2, it is necessary to perform additional calculations to find out the cost of regeneration via Fenton's
reaction.
131
-------
Annual Cost for H2O2 - Option 2
SVE+TCE (aqueous phase) and MC (aqueous phase) loaded GAC-Field data
Target
Comp.
PCE
MC
CM
(mg/g)
270
79
%Target
degraded
50
86
GAC
(g)
78
(0.171bs)
100
(0.221bs)
mass
degraded
(g)
10.5
6.8
H2O2
added
(L)
1.5
0.30
Regeneration
period
(hrs)
7L
reservoir
48
2
Note: For PCE, H2O2 additions (50%) to 7-L reservoir: 150 mLs at time 0, 1,2,4,6,23,25,27,29,48 hrs. Annual hydrogen peroxide cost is for 2
regenerations/yr (assumes 50% regeneration for PCE). For MC, H2O2 additions (50%) to 7-L reservoir: 150 mLs at time 0,1,2,3,4,5,6 hrs.
Analysis uses (2)-150 mL H2O2 additions as they are sufficient to degrade 86% of MC in 2 hrs.
Source: Field data (see section Field-Scale Regeneration Trials in final report).
The H2O2 Cost is $0.34/lb (S0.90/L)- US. Peroxide
$0.34 2.2Ibs 1.2kg
Ib
\kg
L
= $0.90/1
In the field experiments, 1.5 L H2O2 were added to a 7-L reservoir to regenerate 78 g GAC-loaded with PCE
(50% in 48 hrs). For MC, 0.3 L H2O2 were added to regenerate 100 g GAC (86% in 2 hrs).
IZgGAC x
lOOOg- 1kg
- 0. 1 llbsGAC
for PCE
1000 g 1kg
= 0.22lbsGAC
forMC
If 1000 Ibs of GAC will be regenerated at a time, the H2O2 utilization (Ibs) per pound of carbon for PCE-loaded
GAC is
lOOOlbs
(O.nibs
n \.2kgH ,O2
9O9 x — z x
LH202 kgH202
23,295 Ibs H2O2to regenerate 1000 Ibs GAC
132
-------
23,295lbsH202 = 23.3lbsH202 llbsGAC = 46.6lbsH2O2 for PCE-loaded GAC
WOOlbsGAC 0.5 IbsGAC
and for MC,
202 2.2lbsH202
77
LH2O2 kgH2O2
3600 Ibs H2O2to regenerate 1000 Ibs GAC,
3600lbsH202 = 3. 6lbsH202/ IbsGAC = 4.2lbsH2O2 for MC.loaded GAC
WOOlbsGAC 0.85 IbsGAC
Given the carbon wastage rate for PCE and MC (calculated above), the annual cost of H2O2 can be calculated for
PCE at 10 and 100 cfm respectively,
46.6lbsH7O7 SISlbsGAC 40,9\5lbsH7O7 $0.345 ^1,11^/
^-^-x = — *_^_x = $14,116/yr
IbsGAC yr yr lbH2O2
46.6lbsH7O7 SllOlbsGAC 40S,6S2lbsH7O7 $0.345 „,
2—2-x *—*-x $140,9957 yr
IbsGAC yr yr lbH2O2
The annual cost of H2O2 for MC at 10 and 100 cfm
4.2lbsH7O7 405,96llbsGAC \,705,036lbsH7O7 $0.345 «rnn ^^,
±-^-x = — ^-^-x $588,2371 yr
IbsGAC yr yr lbH2O2
4.2lbsH2O2 4,01 l,958lbsGAC _ 16£50,224lbsH2O2 $0.345 _ *, 01Q Q0- ,
x — x — $5,o 13,327 / yr
IbsGAC yr yr lbH2O2
The H2O2 transportation cost is $3.50/mile -US Peroxide (1067 miles from Houston, TX to Tucson, AZ)
The cost of H2O2 depends on the requirements (e.g., H2O2 strength and grade, volume delivered per year,
packaging, and distance to the production plant, etc.). Within the U.S., the list price for 50% Technical Grade,
delivered in full tank trucks with a 40,000 Ibs capacity, and shipment from the nearest production plant, is as follows:
Product: $0.345 per lb-50% (FOB Houston, TX)
Freight: $3.50 per mile (regardless of delivery volume)
133
-------
Total H2O2 cost summary
Comp.
PCE
MC
Annual H2O2 Cost ($/yr)
2m3 reservoir
($0.345/lb H2O2)
Qair=10cfm
14,116
(40,9151bs
H2O2 /yr)
588,237
(1.7Mlbs
H2O2/yr)
Qair=100cfm
140,995
(408,682 Ibs
H2O2/yr)
5,813,327
(16.8Mlbs
H2O2/yr)
Transportation Cost ($/yr)
($3850/trip)
40,000 Ibs truck load
Qair=10cfm
3,850
163,625
Qair=100cfm
38,500
1,617,000
Total H2O2 Cost
($/yr)
Qair=10cfm
17,966
751,862
Qair=100cfm
179,495
7,430,327
GAC Costs ($/yr) - Comparison
GAC Purchase Cost = $1.60/lb based on annual usage rate (Adams and Clark, 1989)
Cost index 1989 = 355
Cost index 2000 = 395.7
GAC disposal cost ($/yr) = $0.48/lb
Plus disposal transportation cost ($/yr) = $3000/8316 Ibs GAC
Plus UTtax ($/yr) = $28/1000 kg = $12.73/10001bs
Calculation of GAC cost - Option 1
Compound
MC
PCE
GAC Purchase Cost
($/yr)-asof2000
Qair=10cfm
0.72M
1,566
Qair=100cfm
7.2M
15,641
GAC Disposal Cost
($/yr)
Qair=10cfm
0.35M
749
Qair=100cfm
3.4M
7,485
Total GAC Cost
($/yr)
Qair=10cfm
1.1M
2,315
Qair=100cfm
11M
23,126
Note: Costs of GAC include purchase ($ 1.60/lbs) and disposal (disposal + transportation + tax).
134
-------
Calculation of GAC Cost - Option 2
Compound
PCE
MC
GAC Cost ($/yr),
($1.60/lb)- 1989 (2000)
Qair=10cfm
140
(156)
38,208
(42,588)
Qair=100cfin
1,403
(1,564)
376,471
(419,633)
Total H2O2 Cost for GAC
regeneration ($/yr)
Qair=10cfm
17,966
0.75M
Qair=100cfm
179,495
7.4M
Total GAC Cost
($/yr)asof2000
Qair=10cfm
18,122
792,588
Qair=100cfm
180,898
7,819,633
Note: Option 2 assumes 5% GAC loss/regeneration event. GAC Disposal Cost is $0/yr.
Additional GAC Cost for Option 2
Compound
PCE
MC
Option 1
GAC Cost ($/yr),
($2.60/lb)
Qair=10cfm
2,545
1.1M
Qair=100cfm
25,416
11. 1M
Option 2
GAC Cost ($/yr)
(H2O2 + GAC Costs)
Qair=10 cfm
18,122
792,588
Qair=100cfm
180,898
7,819,633
Option 2
Additional GAC Cost
($/yr)
Qair=10cfm
15,577
-307,412
Qair=100cfm
155,482
-3,280,367
Note: Negative numbers indicate savings by using option 2.
Labor Costs ($/yr) - Comparison
Labor costs are comparable for both systems assuming breakthrough time and regeneration occur at the same
time. Since, methylene chloride can be degraded almost completely (86%) in-place, then the capacity of the GAC is a
100% for both systems. Hence, regeneration and breakthrough times coincide for both systems. For PCE on the other
hand, the GAC capacity is only 50% after regeneration. Thus, GAC regeneration events occur twice as often as the
breakthrough times. Consequently, the labor costs are double for option 2 (In-place GAC regeneration).
Labor Cost - Option 1 (Assumes 8 hr/GAC replacement event)
Compound
PCE, 100% GAC capacity
MC, 0.28 mg/g
GAC Replacement Events/yr
Qair=10cfm
1
406
Qair=100cfm
9
4000
Labor Cost $/yr, ($40.00/hr)
Qair=10cfm
320
129,920
Qair=100cfm
2880
1.3M
-------
Labor Cost -Option 2 (Assumes 8 hr/GAC regeneration event)
Compound
PCE, 50% GAC capacity
MC, 0.28 mg/g
Regeneration Events/yr
Qair=10cfm
2
406
Qair=100cfm
18
4000
Labor Cost $/yr, ($40.00/hr)
Qair=10cfm
640
129,920
Qair=100cfm
5,760
1.3M
Additional Labor Cost - Option 2
Compound
PCE, 50% GAC capacity
Regeneration Events/yr
Qair=10 cfm
1
Qair=100cfm
9
Labor Cost $/yr, ($40.00/hr)
Qair=10cfm
320
Qair=100cfm
2,880
Pump Selection and Process Energy Cost Calculations
Assuming that 1000 Ibs of GAC will be regenerated at a time,
PGAC = 0.5 g/cm3
Then, Volume/regeneration is 0.9 m3 « 1 m3
Storage tank estimation is 2 m3
Detention time estimation is 10 min
Q = 0.1 nvVmin = 100 L/min = 26.5 gpm
Assuming a GAC column diameter of 1 m, the pump pressure head is:
VGAC bed = 0.9m3
AGAcbed= Hr2 = n(0.5m)2= 0.79 m2
hoACbed = V/A = 0.9 m3 70.79 m2 = 1.15m
doAcbed = lm(h:d = 2.5:1)
Then, hcoiumn = 2.5 m
hcoiumn = 2.5 m (8.2 ft) = 3.55 psi (1 foot head = 0.433527502 psi)
Based on the pump requirements of flowrate (26.5 gpm) and pressure head (3.55 psi),
Air-Powered Double Diaphragm Pump (Positive Displacement Pump)
High-Flow Double Diaphragm Pump with Air Filter Regulator
Max GPM
44
Max psi
120
Wetted parts
PP/Teflon PTFE
Price, $
935
Source: Cole Farmer Catalog n(p. 1562).
136
-------
Pump Power requirements,
P theoretical = QlPlg^ where
Qi = 30gpm(1.89xlO-3gps),
pi =998 kg/m3,
g = 9.8 m/s2 and
h(m) = height + 50% due to head loss = 2.5 m x 1.5 = 3.75 m
y»3| (998kg} (9.&
theoreticai ii * - ^ x —
{ mm ) \ m3 ) \ s2
p -p in
actual theoretical
where n is the efficiency of the pump (assume n=0.7)
x3.75m = 69.3W
Pactual=69.3W/0.7 =
PowerCoSt($) = Pactual (kW) x
kw — nr year
The pump usage (hr/year) depends on the regeneration events (see table below). For example, for the PCE-loaded
GAC regeneration at 10 cfm air flowrate:
regenerationtime (hrs)
year
= 4&hrs x x (2regenerations I yr) = 564.706hrs / yr
( 0.17/fo J
PowerCost($) =
$0.11 564,7067?™ „,
x x '- = $6,149 7 jr
kw — hr year
Additional Process Energy Cost - Option 2
Pump requirements for regeneration events
Target
PCE
0.171bsGAC
MC
0.221bsGAC
Time/regen
(hrs) - 0.2 Ibs
GAC
48
2
Time/regen
(hrs)- 1000
Ibs GAC
282,353
9,091
Regen/year
lOcfin
2
406
lOOcfm
18
4000
Hrs/year
lOcfin
564,706
3,690,946
lOOcfm
5,082,354
36,364,000
Power Cost, $
lOcfin
6,150
40,194
lOOcfm
55,347
396,004
Note: the regeneration time (hrs) and the number of regenerations required per year were calculated above.
137
-------
Additional Capital Cost - Option 2
2 Chemical storage tanks1, 2m3 1,014
2 Positive Displacement Pumps2 2,000
Incremental Labor (10% capital cost) 3,077
Additional Capital Cost, $ 6,091
Note: Capital Cost for 2 GAC columns is $30,772.
Source: 'www.watertanks.com, 2Eng. estimate.
Summary Annual Costs - Option 2 (In-place GAC Regeneration)
PCE, Qair = 10 cfm, Cs=250 mg/g
Energy
Labor
GAC
O & M, $/yr
Present Worth (n=20 years, i=8%), $
Total Cost, $
Annual Cost, $
6,150
320
18,122
24,592
241,448
247,539
25,212
Note: Additional capital is $6,091 .
PCE, Qair = 100 cfm, Cs=250 mg/g
Energy
Labor
GAC
O & M, $/yr
Present Worth (n=20 years, i=8%), $
Total Cost, $
Annual Cost, $
55,347
2,880
180,898
239,125
2,347,764
2,353,855
239,745
Note: Additional capital is $6,091 .
138
-------
MC, Qair = 10 cfm, Cs=0.28mg/g
Energy 40,194
Labor 0
GAC 792,588
O & M, $/yr 832,782
Present Worth (n=20 years, i=8%), $ 8,176,376
Total Cost, $ 8,182,467
Annual Cost, $ 833,402
Note: Additional capital is $6,091.
MC, Qmr = 100 cfm, Cs=0.28mg/g
Energy 396,004
Labor 0
GAC 7,819,633
O & M, $/yr 8,215,637
Present Worth (n=20 years, i=8%), $ 80,662,335
Total Cost, $ 80,668,426
Annual Cost, $ 8,216,257
Note: Additional capital is $6,091.
Summary annual costs - Option 1 (GAC replacement/disposal)
Total GAC Cost ($/yr) as of 2000
Compound
Qair=10cfm Qair=100cfm
PCE 2,315 23,126
MC 1.1M 11M
Note: Costs of GAC include purchase ($1.60/lbs) and disposal costs.
139
-------
The annual cost analysis provided the following summary results (incremental cost only, or costs unique to the
alternative).
SVE Flow
lOcfm
100 cfm
Target Contaminants
PCE
#1: $2,315
#2: $25,212
#1: $23,126
#2: $239,745
MC
#!:$!.1M
#2: $0.83M
#1:$11M
#2: $8.2M
Note: Option 1: GAC cost = $1.60/lb + disposal costs (disposal, transportation and UT tax). Option 2: GAC cost = $1.60/lb (no disposal),
H2O2 cost (purchase and transportation), additional process energy, labor, and capital (dosing pumps, chemical storage tanks).
Fraction ofH2O2 of the Total Cost for Option 2
The following table shows the estimate of the fraction of total cost that arises from H2O2 in our scenario #2:
Compound
PCE
MC
Total H2O2 Cost for GAC
regeneration
($/yr)
Qair=10cfm
17,966
0.75M
Qair=100cfm
179,495
7.4M
Total Annual Cost
($/yr)
Qair=10 cfm
25,212
0.83M
Qair=100cfm
239,745
8.2M
Fraction of H2O2 Cost
for GAC regeneration
($/yr)
Qair=10 cfm
71%
90%
Qair=100cfm
75%
90%
Note: All values were obtained from previous tables.
Some references for GAC costs:
Clark, Robert M. 1989. Granular Activated Carbon: Design, Operation and Cost. Lewis Publishers.
Activated carbon costs range from $1.40 to $2 per pound ($2,800 to $4,000 per ton) (U.S. EPA 1998). U.S. EPA.
1998. "Technical Bulletin: Zeolite, A Versatile Air Pollutant Adsorber." EPA 456/F-98-004. Office of Air Quality,
Clean Air Technology Center. July.
Carbon cost is $2 to $3 per pound, http://www.frtr.gov/matrix2/section4/4-61.html
DOE, 1994. Technology Catalogue, First Edition. February.
EPA, 1991. Granular Activated Carbon Treatment, Engineering Bulletin, EPA, OERR, Washington, DC, EPA/540/2-
91/024.
LaGrega, M., Buckingham, P., Evans, J. and Environmental Resources Management (2001). Hazardous Waste
Management, 2nd edition. San Francisco: McGraw Hill.
140
-------
-------
xvEPA
United States
Environmental Protection
Agency
PRESORTED STANDARD
POSTAGES FEES PAID
EPA
PERMIT NO. G-35
Office of Research and Development (8101R)
Washington, DC 20460
Official Business
Penalty for Private Use
$300
Recycled/Recyclable
Printed with vegetable-based ink on
paper that contains a minimum of
50% post-consumer fiber content
processed chlorine free
------- |