<%BANNER%>

Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2011-08-31.

Permanent Link: http://ufdc.ufl.edu/UFE0024331/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2011-08-31.
Physical Description: Book
Language: english
Creator: Miyittah-Kporgbe, Michael
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: Soil and Water Science -- Dissertations, Academic -- UF
Genre: Soil and Water Science thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Statement of Responsibility: by Michael Miyittah-Kporgbe.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Rechcigl, John E.
Local: Co-adviser: Stanley, Craig D.
Electronic Access: INACCESSIBLE UNTIL 2011-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024331:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024331/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2011-08-31.
Physical Description: Book
Language: english
Creator: Miyittah-Kporgbe, Michael
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: Soil and Water Science -- Dissertations, Academic -- UF
Genre: Soil and Water Science thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Statement of Responsibility: by Michael Miyittah-Kporgbe.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Rechcigl, John E.
Local: Co-adviser: Stanley, Craig D.
Electronic Access: INACCESSIBLE UNTIL 2011-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024331:00001


This item has the following downloads:


Full Text

PAGE 1

TAILORING/ADAPTING APPLICATION OF SORBENTS FOR PHOSPHATE IMMOBILIZATION: ENERGETICS AND SIMULATION MODELING By MICHAEL MIYITTAH-KPORGBE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009 1

PAGE 2

2009 Michael Miyittah-Kporgbe 2

PAGE 3

Dedicated as a memorial to the nature of the Rhema 3

PAGE 4

ACKNOWLEDGMENTS My sincere gratitude and thanks to Drs. J. Rechcigl and C. Stanley for the financial supports and the serene atmosphere at Gulf Co ast Research & Education Center, Wimauma. I would also like to thank other members of my committee Dr. R.D Rhue, who introduced me to the world of calorimetries. I learnt some valu able lessons-patience. My thanks go out to Dr. Cheryl Mackowiak for the encouragement; I still have your hand written notes, Calculus will be handy if you are interested in simulation modeling. That small note has become a self-fulfilling prophecy. To, Dr. Jim Jones who introduced me to simulation modeling, it was fun. Furthermore, my thanks go out to Dr. J-C Bonzongo for the en couragement and allowing me to use the lab, whenever in Gainesville. To the late Dr. M.B Adje i, whose efforts where not in vain at the time I was planning to move to another University. My regr et is that he did not li ve to see the fruits. I also owe special thanks to the technical sta ff at Major Analytical Instrumentation Center, Department of Material Science and Engineerin g for SEM-EDS analysis, and to Gill Brubakar and Gary Sheiffele of the Particle Science a nd Engineering Research Center, for the surface analysis of the sorbents, University of Florida. Above all, I thank God for given me that strong inner peace that surpasses every kind of knowle dge and understanding; it keeps me above every storm of life. Indeed, I dedicate this dissertation as a memorial to the nature of the Rhema. 4

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4 LIST OF TABLES ...........................................................................................................................8 LIST OF FIGURES .........................................................................................................................9 LIST OF ABBREVIATIONS ........................................................................................................1 3 ABSTRACT ...................................................................................................................... .............14 CHAPTER 1 PHOSPHORUS POLLUTION, SORPTION MECHANISMS AND SPECIATION IN SOILS/AQUEOUS SYSTEMS ..............................................................................................17 Introduction .................................................................................................................. ...........17 Mechanisms of P sorption and Speciation in soils .................................................................18 The Rationale of the Study and Remedi ation of P in Contaminated Soils .............................20 2 EVALUATION OF SORBENTS/CHEMICAL AMENDMENTS FOR P IMMOBILIZATION ..............................................................................................................23 Introduction .................................................................................................................. ...........23 Materials and Methods ...........................................................................................................25 General Sorbents Sources and Descriptions ....................................................................25 Initial Equilibrium Sorption ............................................................................................31 Results and Discussion ........................................................................................................ ...32 Preliminary Sorbent Selections ......................................................................................... ......32 Conclusion .................................................................................................................... ..........36 3 SURFACE ANALYSIS OF SO RBENTS: PHYSISORPTION .............................................37 Introduction .................................................................................................................. ...........37 Surface Area and Pore Size Distribution .........................................................................38 Material and Methods .............................................................................................................40 N2-Physisorption (BET, DR, BJH) Pore Size Distribution.....................................................40 Results and Discussion ........................................................................................................ ...42 Conclusions .............................................................................................................................51 4 SORBENTS-P INTERACTIONS AND IM PACTED SOIL: EVIDENCE WITH INCUBATION AND SEQUENT IAL EXTRACTION .........................................................54 Introduction .................................................................................................................. ...........54 Materials and Methods ...........................................................................................................55 5

PAGE 6

Soil Incubation with Sorbents/chemical amendments .....................................................55 Sequential Extraction of So il/Sorbent Amended Soils ....................................................56 Results and Discussion ........................................................................................................ ...57 General Properties of the Amendments and Soils ...........................................................57 Sequential Extractions of Soil and Soil Incubated with Amendments ............................58 Conclusions .............................................................................................................................62 5 SOLID/LIQUID REACTIONS: EQUILIBRI UM VS. KINETICS SORPTION OF P .........63 Introduction .................................................................................................................. ...........63 Materials and Methods ...........................................................................................................64 Equilibrium vs. Kinetics Sorption of P ................................................................................. ..64 Results and Discussion ........................................................................................................ ...65 Modeling of P Sorption Kinetics .....................................................................................68 Reaction Order of the Model ....................................................................................70 Model Evaluation .....................................................................................................72 P Density on Sorbent ................................................................................................75 Adsorption Isotherm ........................................................................................................78 Langmuir and Freundlich Adsorption Model ..................................................................78 Conclusions .............................................................................................................................81 6 CONCEPTUAL FRAMEWORK OF CO-BLENDING: EFFECTS OF CO-BLENDING ON LEACHED SOIL AND SOIL SURFACE MORPHOLOGY .........................................83 Introduction .................................................................................................................. ...........83 Reactions Equations for P Minerals Formation ......................................................................86 Materials and Methods ...........................................................................................................87 Experimental Design/Set up ............................................................................................87 Column Leaching Study ..................................................................................................88 Scanning Electron Microscopy/Ener gy Dispersive Spectroscopy ..................................89 Results and Discussion ........................................................................................................ ...90 Chemical Analysis of Leachates .....................................................................................90 Effects of Co-blending on Leached Soils/Sequential Extractions ...................................92 Effects of Co-blending on Soil Morphology (SEM-EDS) ..............................................96 Control Soil and Sorbents Amended Images ...........................................................98 Sorbents Images .....................................................................................................100 Morphological Images of Co-blended Soils ...........................................................102 Conclusions ...........................................................................................................................107 7 GEOCHEMICAL MODELING ...........................................................................................108 Introduction .................................................................................................................. .........108 Materials and Methods .........................................................................................................110 Results and Discussion ........................................................................................................ .111 Visual-Minteq and Phreeqc ...........................................................................................111 Polymath Model ............................................................................................................114 Comparison of Polymath M odel with Visual-Minteq ...................................................115 6

PAGE 7

Contributions of Solid phases from Polymath Model ...................................................116 Struvite: Slow Release of P and N ................................................................................117 Conclusions ...........................................................................................................................118 8 FLOW CALORIMETRY AN D SURFACE REACTIONS .................................................125 Introduction .................................................................................................................. .........125 Materials and Methods .........................................................................................................131 Instrumentation ..............................................................................................................1 31 Operation ..................................................................................................................... ..133 Ion Exchange .................................................................................................................134 Surface Charge ..............................................................................................................135 Results and Discussion ........................................................................................................ .135 Ion Exchange/Surface Charge .......................................................................................135 Microcalorimetry Analysis ............................................................................................137 Conclusion .................................................................................................................... ........141 9 SIMULATION MODELING OF P DYNAMICS IN FLOW ..............................................142 Introduction .................................................................................................................. .........142 The Role of Simulation Modeling ........................................................................................144 Materials and Methods .........................................................................................................145 System Description and Operation ....................................................................................... 145 Model Development ............................................................................................................. 146 Numerical Equations .................................................................................................... .........148 Results and Discussion ........................................................................................................ .148 Sensitivity An alysis .......................................................................................................152 Conclusions ...................................................................................................................154 Validation of Simulation Model with Column Studies ........................................................154 Conclusions ...........................................................................................................................155 Developing Decision Support for Contaminan t Removal in Soil-Aqueous Systems ...........157 Defining Decision Support Systems ..............................................................................158 Excel Applications and Illustrations for DSS ................................................................159 Conclusions ...........................................................................................................................162 10 SUMMARY AND CONCLUSIONS ...................................................................................164 LIST OF REFERENCES .............................................................................................................168 BIOGRAPHICAL SKETCH .......................................................................................................182 7

PAGE 8

LIST OF TABLES Table page 3-1 BET-N2 of specific surface area, micropore volume and mesopores volumes of sorbents, obtained at P/Po= 0.99 of total pore volume. .....................................................44 4-1 General properties of amendments and sequential extraction of soil P. ............................59 5-1 Kinetic parameters of P sorption on so rbents using pseudo-first-order model ..................75 5-2 Kinetic parameters of P sorption on so rbents using pseudo-second-order model .............75 5-3 Langmuir and Freundlich constants for P sorption on sorbents ........................................81 6-1 Elemental composition (%) of contro l soils with or w ithout leaching. .............................99 7-1 Phosphate species distribution (Visual-Minteq) in control soil without amendments showing potential negatively charged P solid phases. .....................................................120 7-2 Saturation indices calculat ed using Visual-Minteq for treatments with and without co-blending. .................................................................................................................. ...120 7-3 Saturation indices calculated using Phr eeqc for treatments with and without coblending............................................................................................................................121 7-4 Data taken from literature to simulate model predictions of Polymath and VisualMinteq. ....................................................................................................................... ......122 7-5 Calculated values of solid (mg) found per litter of residual leac hate as determined by Polymath model ...............................................................................................................1 23 A-1 Weekly mean pH data for sorbents amended and unamended soils. ...............................165 A-2 Weekly mean Eh (mV) data for sorbents amended and unamended soils. ......................165 A-3 Cumulative P mass of column leaching study. ................................................................166 A-4 Sequential extraction for so luble and extractable P. Soils extracted after 12 wks of leaching. ..................................................................................................................... ......166 A-5 Sequential extraction of Fe-Al bound P. Soils extracted after 12 wks of leaching. ........167 A-6 Sequential extraction of Ca-Mg bound P. So ils extracted after 12 wks of leaching. ......167 8

PAGE 9

LIST OF FIGURES Figure page 2-1 Phosphorus sorbed from soil solution extr act as a function of sorbents after 8 h of equilibration. Values are means of triplicate. ....................................................................35 2-2 Phosphorus sorbed (%) from soil solution ex tracts as a function of sorbents after 24 h equilibration. Values are means of triplicate. ....................................................................35 3-1 Types of physisorption isotherms (IUPAC, 1985) ............................................................44 3-2 Types of hysteresi s loops (IUPAC, 1985) .........................................................................45 3-3 Nitrogen adsorption and desorption isothe rm for Al-WTR as a function of relative partial pressures conditions n ear to saturation pressure P/Po = 0.99. ...............................45 3-4 Nitrogen adsorption and desorption isotherm for Slag as a functi on of relative partial pressures condition near to saturation pressure P/Po = 0.99. ............................................48 3-5 Nitrogen adsorption and desorption isot herm for MgO as a function of relative partial pressures condition n ear to saturation pressure P/Po = 0.99. .................................49 3-6 Nitrogen adsorption and desorption isothe rm for Gypsum as a function of relative partial pressures condition n ear to saturation pressure P/Po = 0.99. ..................................50 4-1 Sequential fractionation of manure-impact ed soils (Ona and Okeechobee), showing distribution of P.............................................................................................................. ....60 4-2 Soluble and extractable P (KCl-extraction) for soil inc ubated with Al-WTR, Slag and MgO. Means (n = 3) of the same letters are not significantly ...........................................60 4-3 Iron-Al bound P (NaOH-extraction) for so il incubated with Al-W TR, Slag and MgO. Means (n = 3) of the same ................................................................................................61 4-4 Calcium-Mg bound P (NaOH-extraction) for soil incubated with Al-WTR, Slag and MgO. Means (n = 3) .........................................................................................................61 5-1 Phosphorus sorption isotherm with in itial P concentrations of 40, 60, 100, 300, and 500 mg L-1 on sorbents. .....................................................................................................68 5-2 Effect of reaction time on P so rption of sorbents at 500 mg L-1 initial P concentrations. Values ar e means of tr iplicate. .................................................................69 5-4 Residuals plot of pseudo-first-order P sorption on Al-WTR, MgO, Slag, Gypsum, and LimeKD showing non-scatter points for each sorbents. .............................................73 5-3 Pseudo-second-order fitting for P sorpti on to sorbents of Al-WTR, MgO, Slag, Gypsum, and LimeKD. ......................................................................................................74 9

PAGE 10

5-5 Residuals plot of pseudo-second-order P sorption on Al-WTR, MgO, Slag, Gypsum, and LimeKD showing scatter points. .................................................................................74 5-6 Phosphorus density on sorbents (Al-WTR, Slag, MgO, Gypsum, LimeKD and MgO) as a function of equilibrium concentration at initial P concentration ..............................77 5-7 Kinetics of P surface coverage on sorbents of Al-WTR, MgO, Slag, Gypsum and LimeKD at initial P concentration ...................................................................................78 6-1 A framework of complexation reactions due to co-blending of sorbents for rapid P immobilization. ............................................................................................................... ...85 6-2 The conceptual transport of P species with respect to surface chemistry of sorbents as influenced by speciation and pH. .......................................................................................85 6-3 Effects of co-blending 10 g kg-1 Al-WTR with 10 g kg-1 each of slag and MgO respectively on soluble P from leachates.. .........................................................................92 6-4 Soluble and extractable P. Means (n = 3) of the same letters are not significantly different at = 0.05, determined by LSD.. ........................................................................94 6-5 Iron-Al bound phosphorus. Mean (n = 3) of the same letters are not significantly different at = 0.05, determined by LSD.. ........................................................................95 6-6 Calcium-Mg associated P. Means (n = 3) of the same letters are not significantly different at = 0.05 determined by LSD. .........................................................................96 6-7 SEM image of Control (Time zero) soil without co-blending/no leaching and the corresponding EDS spectra. ...............................................................................................99 6-8 SEM image of control (Time zero) soil w ithout co-blending and with leaching (after 12 wks) and the corresponding EDS spectra. ....................................................................99 6-9 SEM images of Al-WTR and the co rresponding EDS. The EDS indicates high amount of elemental Al content. .....................................................................................100 6-10 SEM image of gypsum, a nd the corresponding EDS. The EDS showing presence of Ca and S as major elemental composition. Traces of impure minute Mg particles were also observed.. .........................................................................................................101 6-11 SEM images of slag and the corresp onding EDS spectra. The EDS showing presence of major elements as Si, Ca Mg and Al. Traces of minute levels of S and K were also observed. Scale ................................................................................................................101 6-12 SEM images of LimeKD and the co rresponding EDS spectra. The EDS showing major elements present as Ca and Si, with traces of Al, Mg, S and K. Scale bar = 200 m. ........................................................................................................................... ........101 10

PAGE 11

6-13 SEM images of MgO and the corresponding EDS spectra. The EDS reveals major element as Mg. Scale bar = 200 m................................................................................102 6-14 SEM and EDS spectra of (MgO + Al-WTR ) sorbents co-blended with P impacted soil. The marked arrow indicates the mo rphological observation due to co-blending. EDS spectra ......................................................................................................................103 6-15 SEM and EDS spectra of (Slag + Al-WTR ) sorbents co-blended with P impacted soil. The marked arrow indicates the mo rphological observation due to co-blending. EDS spectra. .....................................................................................................................104 6-16 SEM and EDS spectra of (Gypsum + Al-W TR) sorbents co-blended with P impacted soil. The marked arrow indicates the mo rphological observation due to co-blending. EDS s ...............................................................................................................................104 6-17 SEM and EDS spectra of (LimeKD + Al-WTR) sorbents co-blended with P impacted soil. The marked arrow indicates the mo rphological observation due to co-blending. EDS spectra ......................................................................................................................105 6-18 Electron dot maps of (LimeKD +Al-WTR) co-blended soil. Bright spots indicate the location of elements within the new solid phase. ............................................................105 6-19 Electron dot maps of (Slag + Al-WTR) co -blended soil. Bright spot indicates the location of elements within the new solid phase. ............................................................106 6-20 Electron dot maps of (Al-WTR+ Gyps um) co-blended soil, showing location of elements within the new solid phase Circled spots showing likely PO4 minerals. .........106 7-1 Experimental values plotted vs. predicted values of struvite from a Visual-Minteq. ......124 7-2 Experimental values plotted vs. predicted va lues of struvite from a Polymath model. ...124 8-1 Schematic diagram of column, thermistors, and calibrating resistors as used in flow calorimetry ................................................................................................................... ....132 8-2 A linear curve depicting the relationsh ip between peak area and flow rate ....................134 8-3 Calorimetric response for nitrate replacing chloride (NO3 -/Cl-) and chloride replacing nitrate (Cl-/NO3 -) on Al-WTR. .........................................................................................136 8-4 Exothermic calorimetric response for nitrate replacing chloride (NO3 -/Cl-) before and after PO4 treatment on Al-WTR. .....................................................................................137 8-5 Microcalorimetric response of nitrate replacing chloride (NO3/Cl) and (Cl/NO3) for Al-WTR. ..........................................................................................................................140 8-6 Microcalorimetric response of nitrate replacing chloride (NO3/Cl) and (Cl/NO3) for Boehmite. ..................................................................................................................... ....141 11

PAGE 12

9-1 An image of Flow Calorimetry set up, indi cating sections of P in jection (syringe), a monitor recording signal effects and leachate collection area. ........................................146 9-2 Forrester diagram of P tr ansport through compartmented ce lls in Flow calorimetry. .....147 9-3 Simulating amount of P sorbed in the fi rst cell, depicting maximum sorption capacity as a function of time. ........................................................................................................ 149 9-4 Simulation of maximum sorption for segmented cells with of ~2.0 mgP cell-1 maximum vs time. ............................................................................................................14 9 9-5 Simulation of concentration as it flows through segmented cells to final cell vs time. ...150 9-6 Simulation of maximum sorption for segmented cells with of ~ 0.5mgP cell-1 maximum vs time .............................................................................................................15 2 9-7 Simulation of concentration as it flows through segmented cells to a final cell vs time. ......................................................................................................................... ........152 9-8 The measured P values vs predicted values by the model as used in column leachates. .................................................................................................................... ......156 9-9 Plots of residual error vs predicted va lues on the model for column leachate. ...............156 9-10 A screen shot of initial start and data input base for sorbents. ........................................159 9-11 A screen shot of the parameters (sor bents, sorption capacity, and rate) columns attached to sorbent data input. .........................................................................................160 9-12 A user-interface on start button showing sorbent selection, precision, initial concentration, number of steps, computes step/concentration and show results buttons. .............................................................................................................................161 9-13 User-interface showing the results of init ial concentration and tim e steps and types of sorbent used and amount of contaminant removed. .........................................................162 12

PAGE 13

LIST OF ABBREVIATIONS WTR Water treatment residual Al-WTR WTR generated using alum Fe-WTR WTR generated using iron salts HCl-P Ca-Mg associated inorganic P forms KCl-P Soluble or labile P NaOH-P Fe-Al associated inorganic P forms AEC Anion exchange capacity CEC Cation exchange capacity SEM Scanning electron microscopy EDS Energy dispersive X-ray spectroscopy SRP Soluble reactive phosphorus SE Standard error SD standard deviation TP Total phosphorus LimeKD Lime kiln dust BET Brunauer, Emmett and Teller IUPAC International Union of Pure and Applied Chemistry PSD Pore size distribution SA Surface area SSA Specific surface area 13

PAGE 14

Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy TAILORING/ADAPTING APPLICATION OF SORBENTS FOR PHOSPHATE IMMOBILIZATION: ENERGETICS AND SIMULATION MODELING By Michael Miyittah-Kporgbe August 2009 Chair: John E. Rechcigl Co-chair: Craig D. Stanley Major: Soil and Water Science The goal of this research is to develop a remediation technology for a rapid P uptake and complete immobilization of P in soils, and evaluate the potential for re-use of such immobilized P. Many studies exist in the peer-reviewed literature on the use of sorbents for P immobilization. However, there is little to no published informa tion on the development of remediation strategies that focus on the tailoring and use of sorbents to target various P species concurrently for a rapid and efficient P immobilization. To gain knowledge in tailoring an application of different sorbents in P immobilization, the following tasks were performed: 1. Selection and evaluation of various sorbents (Al drinking water treatment residuals (AlWTR), Slag, Gypsum, Lime kiln dust (LimeKD) and MgO for P immobilization. 2. Characterization of the surf ace properties of all selected sorbents by physisorption analysis. 3. Evaluations of the sorbents-P interactions in phosphorous contaminated water and soil matrices through multiple techniques of incubation, operationally defined speciation (sequential extraction), and geochemical modeling. 4. Development of a coblending technique and assessment of its effi ciency on immobilized P through leaching, geochemical modeling and scanning electron microscopy (SEM), equipped with energy 14

PAGE 15

dispersive spectroscopy (EDS). 5. Flow calorimetry which provides a controlled environment for quantifying parametric inputs and ou tflows of P in this case was used to help in developing a simple mathematical model for simulation of P sorption. The major findings from these experimental and modeling studies can be summarized as follows: Results from the evaluation of sorbents suggest Al-WTR and Gypsum had limitation in immobilizing P at near natural condition (soil solution extracts contai ning P), whereas, Slag, LimeKD and MgO removed P greater than 90% from solution. On the contrary, using lab reagents suggest, all the sorbents were excellent candida tes for sorption. The specific surface areas of the sorbents were 21, 3.1, 3.3, 2.0, and 5.0, m2 g-1 for Al-WTR, Gypsum, Slag, LimeKD and MgO respectively. Physisorption isotherm furthe r revealed that the sorbents were non-porous with limited microporosity, suggesting that adsorption of P ma y be governed by other secondary mechanisms of chemisorption and or precipita tion reactions. Sorbents-P interactions revealed that Al-WTR removed P associated with Al-Fe, significantly gr eater than sorbents containing Ca-Mg i.e. Slag, and MgO ( P < 0.001). On the other hand, sorbents cont aining Ca-Mg immobilized P associated with Ca-Mg than that of Al-Fe (P < 0.001) Thus showing that co -blending the sorbents, could immobilize all P forms liable to cause environmental pollution. The results from the co-blending technique of using Al-WTR with Ca-Mg based so rbents suggest the method can effectively and rapidly immobilized P. This was assessed w ith soil leaching experiments. Data of 10g kg-1 each of Al-WTR and Slag reduced P to almost zero in the first week of leaching. In addition, the results showed that, the co-blending of (Al-WTR+ Slag) is statistically and significantly greater than depending on 20 g kg-1 of solely Al-WTR for P immobi lization. Further, co-blended AlWTR+Slag had a pH that is about the same as th at of the control (unamended soil). Furthermore, 15

PAGE 16

with the aid of scanning electron microscopy (S EM) and energy dispersive spectroscopy (EDS) revealed presence of new solid phases. Geochemical models also revealed the natu re of these solid phases. Results suggest presence of various PO4 minerals including struvite. Although not identified experimentally, the occurrence of struvite would be pa ramount as it is not only a sink fo r P but can also be used as a source of controlled P release. Accordingly, the po tential formation of this mineral could lead to the development of a remediation strategy th at may help in sustainable re-use of PO4. Phosphorus sorption on Al-WTR and its dynamics were modeled using flow calorimetry, and a new and simple mathematical model was proposed. A sensitivity analysis on parameters influencing P sorption showed clearly that for P sorption, the most important single parameter is the rate and not the ad sorption capacity. The P model valida tion studies resulted in very good predictions with R2 ~ 90%. A decision support tool was developed based on the mathematical model, with the goal to provide easy access to decision makers on th e types of sorbent to use for water quality considerations. Finally, future laboratory studies are needed to identify and quantify struvite formation, if formed under the above described experimental conditions and as predicted by the geochemical models. This could then lead to optimization studies focusing on slow release of previously sorbed P and its potential applic ations. Overall, the results s uggest that the novel co-blending technique can lead to rapid and complete immobilization of P. 16

PAGE 17

CHAPTER 1 PHOSPHORUS POLLUTION, SORPTION MECHANISMS AND SPECIATION IN SOILS/AQUEOUS SYSTEMS Introduction Phosphorus (P) along with other nutrient load ings from non-point sources threatens the water quality of many surface and ground water bodi es in many parts of the world, including Florida (Pant et al., 2003, 2004). Phosphorous loss to surface/ground waters depends on various factors, including soil physicochemical characteristics (Pant and Reddy, 2003), as well as the P mineral phase formation. Retention of P in the so il is therefore, important for limiting P loss to water bodies. Phosphorus retention in soil/aqueous systems is controlled by Al-Fe hydroxides and Camineral phases. Further, these oxides/hydroxides or minerals exist as mixed/blended forms rather than as pure Al or Fe or Ca-m inerals. Depending on the pH of the ambient environment, mineral phases of Al-Fe or Ca may be dominant in the soil. Manure deposition often provides additional complexities including increased soil pH (Eghball, 2002). However, over time, due to leac hing, the initial soil pH which is typically neutral has been observed to become alkaline. This alkaline pH of a bout 8.3 or higher may be due to continuous release of carbona tes associated with Ca and M g, and or release of ammonia. In alkaline environments, Ca-Mg likely controls P due to activities of Ca2+/Mg2+ controlling P in the solid phase, while Al-Fe control P under acidic conditions (Bohn et al., 2001). In manureimpacted soils, dominant P forms have been observed to be loosely associated with Ca-Mg (60% 70%) than that of Al-Fe (20% 30%), (Sharpley et al., 2004, S ilveira et al., 2006). These large pools of alkaline forms of Ca-Mg components, wh ich are loosely associated with P, can be responsible for major P release. Thus, the need to control this greater Ca-Mg P forms to limit P release. 17

PAGE 18

Additionally, it has been shown that different orthophosphate species can exist in solution as a function of pH (Lindsay, 2001). A decr ease of one unit in pH can increase the H2PO4 -/HPO4 2ratio by 10-fold. Conversely, an increase of one unit in pH can decrease the H2PO4 -/HPO4 2ratio by 10-fold. Thus, using a sorbent suitable for an acidic condition may not completely remove certain phosphorous species as pH fluctuates from approximately neutral to values greater than eight pH units. On the other hand, using a sorbent suitable for an alkaline environment may not completely remove a ll P species. Under practical conditions, a good sorbing material should be able to effectively i mmobilize P from moderately acidic to alkaline pH range (about 6 to about 8.5). To immobilize P, two separate appr oaches have been applied to PO4 solubility. One is based on formation of various PO4 minerals by precipitation, while the second is based upon the theories of adsorption onto the surfaces of hydrated Al-Fe oxides and clay minerals (Haynes, 1982). To exploit these two approaches (mechanis ms), there is a need to characterize surface properties of the sorbents. In addi tion, sorbent application must also consider the sp eciation of P, especially in manure-impacted soils for e ffective immobilization. A look at P sorption mechanisms would shed more lights on ways to u tilize the mechanisms of precipitation and that based on surface adsorption for effective P immobilization. Mechanisms of P sorption and Speciation in soils The source of P has great influence on the ava ilable species, as well as the process for P immobilization and removal. Three main sources of P to soils have been recognized: rainfall, fertilizers, (inorganic), and orga nic (feeds/manure). These P sources can lead to incidental loss to surface waters resulting in eutrophicati on. The soluble P from organic sources may accumulate in, or be released by soils depending on the capacities of the soils to retain P through adsorption and precipitation reactions. 18

PAGE 19

Two P adsorption reactions mechanisms have been described taking place in soils. 1. Nonspecific/ion exchange and 2. Specific/ligand exch ange (Rhue and Harris, 1999). Ion exchange is rapid and reversible, and typica lly accounts for small fraction of adsorbed P in Florida soils. Ligand exchange occurs when a PO4 anion replaces a surface hydroxyl coordinated with a metal cation, usually with reactive su rfaces of Al or Fe oxyhydroxides. Further, depending on the availability of the coordina ted position of OH groups and H2O molecules on the surfaces of hydroxides of Fe, Al, Mn or layer silicates, chemis orption may occur, leading to inner or outer sphere complex formation (McBride, 1994). Precipitation reaction may also occur, and can be strongly a ffected by pH. Under alkaline conditions, Ca generally controls P solubility, and orthophosphate readily forms less soluble di and tricalcium phosphates. Under acidic condition however, Al and or Fe controls P solubility and orthophosphate read ily precipitates primarily as Al or Fe PO4. The rendering of P as insoluble is either through ligand exchange with the addition of Al oxides or through precipitation reactions. Hydroxides and oxyhydroxides provide major source of P sorption capacity in soil and form insoluble surface complexes/solid phases when they react with P from solution (Bohn et al., 2001; Mc Bride, 1994). Phosphate is adsorbed onto hydrous metal oxides through the following liga nd exchange reaction: >M-OH( s ) +H+( aq) >M-OH2 + >M-OH2 +( s ) +H2PO4 ( aq) >M-H2PO4 -(s) +H2O( l) where, M is a metal, usually Al or Fe. The covalent bond formed between the hydrous oxide and PO4 is very stable (Sposito, 1989). Speciation of surface sorbents affects P sorp tion by influencing the reactivity of the surfaces. Speciation of metal hydr(oxides) is la rgely pH dependent and can be observed through 19

PAGE 20

the behavior of metal in soluti on (McBride, 1994). For example, Al and Fe exist as hydrolyzed ions in solution when the pH is > 4, thus, rese mbling an Al-OH surface site. The extent of the hydrolysis determines the number of ligands present. Metals that easily hydr olyzed tend to have more ligands associated with them than other metals. Other ligands such as SO4, F, SiO4 and organic acids also can be sorbed to the surfaces of Al or Fe crystals and can exchange or compete with, P for sorption sites (Violante and Gianfreda, 1993; Rhue and Harris, 1999). Ligands coordinated to two metals atoms (bi-dente liga nds) tend to have less dissociation than those coordinated to one metal atom (mono-dentate ligands) and are therefor e, less reactive. In P speciation/chemistry, reactions are great ly dependent on pH. Phosphorus can exist as PO4 3or as protonated species of H2PO4 and HPO4 2-. As pH increases, th e change in dominance of each species coincides with the pKa1, pKa2, and pKa3 of H3PO4, representing pH between, 2-7 (mono-valent), 7-12 (di-valent), and 12-14 (tri-valent) of H2PO4 and HPO4 2and PO4 3respectively (Lindsay, 2001). Anions are readily attracted to protons at surfaces of hydroxides when protonated at low pH values. Phosphate is therefore, expected to adsorb more at low pH and less readily at pH values above the pKa of the given surface sites. Further, P adsorption would be inhibited at higher pH due to competition from OH groups in solution (Lijklema, 1980). Consequently, it is unlikely to have significant qua ntities of tri-valent species under normal soils conditions. However, H2PO4 and HPO4 2-species exist as ratios relative to the pH of the soils (Lindsay, 2001). The Rationale of the Study and Remediation of P in Contaminated Soils Immobilization of P is a technique devised to capture P species within the contaminated soil mass. Thus, it reduces the tendency of P en tering the ground/surface water, to ultimately cause eutrophication. To achieve this with manure-impacted soils, several researchers use 20

PAGE 21

sorbents containing Al/Fe and Ca to imm obilize P (e.g. Agyin-Birikorang, et al., 2007, Kordlaghari and Rowell, 2006). Howe ver, P exists in manure-impacted soils as different species with respect to the prevailing pH Thus, there is a limitation of retaining some of the species depending on sorbents containing solely Al/Fe or Ca. The limitations with the use of sorbents with ma inly Al/Fe and Ca appear to be high pH for alkaline-based materials, low pH (acidic based materials), together with slow sorption process if the reaction is heavily co ntrolled by kinetics. In addition, th e effectiveness of the sorbent is relative to the P species existing at that prevailing pH environmen t. However, as discussed above regarding manure-impacted soils, the pH is never static. Thus, suggesti ng appearance of new P species as the pH changes, ther eby reducing the efficiency of th e sorbent in question. A case in point is research by Novak and Watts, 2005, where Al-WTR is used to reduce the mobility of labile P or extractable P. However, the paper di d not address how the use of this sorbent would remove other species of P such as P associated with loosely bound Ca-Mg. Further, a paper from Agyin-Birikorang et al ., (2007), used Al-WTR to immobilize P from manure-impacted soil where Al-WTR was a pplied only once. After 7.5yrs only ~ 50% of bioavailable P was removed. In addition, majority of the P appeared to be immobilized within the first yr. Thus, supporting spectro scopic studies on P reactions that initial P reac tion is a fast process. However, the subsequent reactions are slow and take years to complete (Sparks, 2002). Currently, no study exists in th e peer-reviewed literature on the slow immobilization of P in manure-impacted soils. Further, many reported studies in the peer-reviewed literature have failed to take into consideration the methods n ecessary to immobilize all P species concurrently under fluctuating soil pH. Therefor e, there exists a gap in kn owledge, to fully address the concerns of P immobilization in manure-impact ed soils. Furthermore, no study in the peer21

PAGE 22

reviewed literature, addressed the fixing of exces s P in the soil, while making sure that such fixed P is released for later re-u se. This is important because, PO4 is an exhaustible natural resource, and care must be taken to maintain its su stainability for future re-use. It is against this background; this project was designed to: (i). ra pidly immobilize P. ( ii). immobilize of all P species regardless of pH fluctuati ons. (iii). address the re-use of immobilized P for future re-use. To fully address the P immobilization issu es, experiments were designed to gain knowledge on: The surface physical properties of the sorbents. The chemistry of P, and P speciation pert aining to P contaminated environment. The chemistry of sorbents interactions with the P species. The remediation of P contaminated soils is based on data obtained in experiment 1, 2, and 3 above. Developing a strategy for sustaina ble re-use of the fixed P. Developing a mathematical model that may he lp in predicting the time required for complete P immobilization. Finally, a unique and novel a pproach of presenting research information to policymakers was also addressed through user defined decision support tools. It is hoped, that its application will meet the needs of decision makers addressing water quality issues. 22

PAGE 23

CHAPTER 2 EVALUATION OF SORBENTS/CHEMICAL AMENDMENTS FOR P IMMOBILIZATION Introduction Phosphorus has been implicated in water impe rilment leading to eutr ophication of streams, lakes and groundwater pollution. The cost of wate r quality degradation in the US has been estimated to be ~ $2.2 billion annually (Dodds et al., 2008). This has necessitated efforts to reducing P loadings from non-point sources. Non-point sources include from over fertilized fields, manure-impact ed soils, and from PO4 mining. Several Florida soil types (Entisols and Spodosols) have low P retenti on capacities, which allow sign ificant P leaching and runoff (OConnor et al., 2001). Various management practices have been suggested to reduce potential P losses and subsequent impacts to water bodies. One such pr actice involves the use of soil amendments or sorbents to improve soil retention of P and to reduce P solubility in soil-aqueous system. Chemical amendments, such as metal salts cont aining Fe, Al and Ca have been suggested to reduce potential P losses and s ubsequent impacts on water bo dies (Moore and Miller, 1994, Moore et al., 1999). Some exampl es may include dolomite (CaMg(CO3)2), Gypsum (CaSO4.2H2O), Alum (Al2(SO4)3), Ferric chloride (FeCl3), and Al-based water treatment residuals (Al-WTR). Synthetic iron oxide-gypsum (OX) was used to remove P from water in different parts of Lake Butgenbach (Bastin et al., 1999), where eu trophication was suspect. Further, OX was also tested with inorganic/organic PO4 solution. The results showed that P removal was pH dependent, where rates were consta nt between pH 4 and 8, but increas ed significantly at pH >8. Gypsum was used by Kordlaghari and Rowell, (2006), to immobilize PO4 in soils. Results from the study using scanning electron micros copy (SEM) and equilibrium modeling showed 23

PAGE 24

small crystals of gypsum disappeared and roughl y spherical particles of dicalcium phosphate (DCPD) formed. Water treatment residuals (WTR), a waste prod uct of drinking water treatment, behave like amorphous Al oxides and have been shown to i mmobilize P in manure-impacted soils (Silveira et al., 2006). Reduction in extractable P has been observed with addition of Al-WTR to manure/manure-impacted soils (Elliot et al., 2002 Dayton et al., 2003). Ho wever, both Dayton et al., 2003 and Novak and Watts (2004), have reported that Al-WTRs can diffe r substantially in P binding maxima. Such differences in sorption were attributed to variations in their oxalate extractable Al and Fe concentra tions. (Novak and Watts, 2005) Limitations to WTRs are slow P sorption kinetics requiring prolonged contact times (Makris et al., 2004). Secondly, la rge amounts of WTRs are require d in heavily impacted soils due to competition of soluble organics with the ac tive sorption sites, thus reducing the efficiency of WTRs ability to retain P (Lane, 2002; Silveira, et al., 20 06). Additionally, at high WTR application rates presence of Al toxicity ha s been observed (Novak and Watts, 2004). However, using moderate rates of 114 dry Mg ha-1 Al-WTR one time applicati ons, suggest that after 7.5 yrs, bioavailable P and dissolved was reduced by~50% (Agyin-Birikorang et al., 2007). This implies that, P continuous to be released with passage of time. Further, under alkaline conditi on of manureimpacted soil, Al-based WTR is limited in controlling P solubility. This is because in alkaline condition Ca-Mg tended to control P in the solid phase, whereas Al-Fe contro ls P in acidic soils (Bohn et al ., 2001). Thus, soil pH and the nature of the chemical amendments must be taken into account when trying to immobilize P in manure-impacted soil. 24

PAGE 25

Phosphorus speciation is pH depe ndent. Phosphorus can exist as PO4 3or the protonated species of H2PO4 and HPO4 2-. As pH increases, the change in dominance of each species coincides with the pKa1, pKa2, and pKa3 of H3PO4, at pH ranges between, 2-7, 7-12, and 12-14 of H2PO4 and HPO4 2and PO4 3-, respectively (Lindsay, 2001). In Florida soils, H2PO4 and HPO4 2predominate as P species at the pH of soil solutions (Lu and OConnor, 1999). Due to the acid and sandy nature of Florida soils, low P sorption capacity often leads to P losses, thereby aff ecting water bodies. The P sorption (retention) capacity of Florida upland acid so ils (Freese et al., 1995) as well as wetlands systems (Reddy et al., 1995) have been strongly correlated with Al and Fe hydr(oxide) content. of sediment and soils P revealed that P is associated with Ca -Mg, Al-Fe, labile and organic (70, 20, 10, and < 1)%, respectively fractions (Chang et al., 1983). Sequential extraction suggests the role played by metal cations (Fe, Al, Ca, and Mg) in bind ing different P species. Thus, there is a need to evaluate the effects of different chemical amendments on P using soil solution extracts from manure-impact ed soil, as it might occur under a natural system. This is to imitate the likely sorptio n process that would have occurred, to a close natural system e.g. of soil containing P. It is against this background, in this study, the following hypotheses were set up: (i) Sorption of P to chemical amendments using so il solution extracts wo uld indicate different sorption rates of the tested amendments. (ii) Ca-Mg amendments would remove greater P from solution extracts. The specific objectives are: (i) to determine sorption characteristics of selected amendments. (ii) to select suitable sorpti on materials for the subsequent experiments. Materials and Methods General Sorbents Sources and Descriptions Sorbents used in the study included, aluminum water treatment residuals (Al-WTR), MgO, Lime Kiln dust (LimeKD), Gypsum (CaSO4), dolomite, and superMag (MgSO4). A preliminary 25

PAGE 26

study was carried out, which lead to the selection of some of the sorbents for a succeeding column study. Al-WTR is a byproduct of drinking water trea tment having hydr(oxides) like properties and potential to use as P fixing sorbents. Alum (aluminum sulfate) or polymers (polyaluminum chloride) are coagulants used, in conjunction with lime to form an amorphous Al hydroxide (AlOH3) gel during drinking water treatment. Coagulation is used to remove turbidity, color, taste, and odor from raw water and speed sedi mentation. WTRs contain sediments from the raw water and the reaction products of coagulati on, amorphous Al oxides, which account for 50 to 150 g kg-1 of the total residuals (ASCE and AWWA, 1996). Drinking water facilitie s throughout the USA generate vari ous types of residuals from water purification processes. Two major types ar e generated (Al-based a nd Fe-based). Residuals are produced from the process of sedimentation and flocculation, where the primary coagulant is either Al salts (e.g. alum) or Fe salts (e.g. ferric chloride). These residuals are hitherto referred to as Alor Fe-WTR, respectively. Another major type is calcium-WTR, also produced in water treatment facilities where lime is used to remove hardness in water. The Al-WTR used in this study was taken from Bradenton drinking water treatment facility, Florida. Blast furnace slag is a byproduct of pig iron production. Limestone, iron-ore, and coke are heated to about 1900oC in a blast furnace. This causes the ir on to separate out while the silicates and alumina in the core and coke combine with Ca and Mg from the limestone. The molten slag is tapped from the furnaces and is either allowed to air-cool into a crystalline, cellular substance or cooled rapidly in cold wate r, which causes it to form light weight amorphous, porous granules. Slag is mainly Ca and Mg alumino-silicates, but it contains smaller quantities of other elements 26

PAGE 27

such as Fe, Mn, or S (Barber, 1967). The granul ated blast furnace slag used was collected from Holcim Inc., (USA) Alabama. Magnesium oxide (MgO) is produced from natura lly occurring minerals such as magnesite (magnesium carbonate), MgCl2 rich brine, or seawater. When magnesium carbonate is heated at 700-1000oC, MgO and CO2 are produced (i.e. MgCO3 MgO + CO2). Brine, which contains MgCl2 and CaCl2 it is reacted with calcined dolomite (CaO. MgO). The react ion generates the following. CaCl2 + MgCl2 + H2O + (CaO. MgO) + H2O 2Mg(OH)2 + CaCl2 +H2O The slurry Mg(OH)2 is thermally decomposed (calcined) to produce MgO. 2Mg(OH)2 2MgO + 2H2O (steam). The calcinations of magnesium hydroxide generate different grades of MgO depending on the temperature of cal cinations. Thermal altera tions affect surface area, porosity and reactivity of MgO. Thr ee grades of MgO are produced, dead burned, hard burned and light burned MgO. Dead burned Mg O produced when the temperature used in calcinations is >1800oC. This high temperature virtually e liminates the reactivity, hence dead burned. Surface area of the MgO is < 0.1m2 g-1. However when the temperature is between 12001800oC, with surface area between 0.1-1.0 m2 g-1, such MgO is referred to as hard burn due to narrow range of reactivity (Van de Walle et al., 1993). Further, when the temperature is between 350-1000oC is referred to as light burned. This product has a wide ra nge of surface areas between 1.0-250 m2 g-1. The MgO used in this study is a re sidual mixture of different grades discussed above (Personal Communication, Mart in Marietta, Magnesia Specialties Inc). Lime kiln dust (LimeKD) is a waste residual dust or particles from decomposition of magnesium carbonate and or calci um carbonate. It was collected from O-N Minerals, Luttrell Operation (TN). It is a white crystalline soli d. Lime KD is produced from calcinations at 27

PAGE 28

temperature of > 1000oC. The high temperature volatilizes the CO2 as indicated. CaCO3 + Heat CaO +CO2 The term LimeKD is used to describe both CaO or Ca(OH)2. Calcium oxide/hydroxide has a strong exothermic reaction with water, and has an additional drying effect on the soil forming cementitious hydrates. A simplified qualitative repr esentation of the reaction with the soil is shown below: CaO +H2O Ca(OH)2 + Heat H KJ mol-1 Ca(OH)2 Ca2+ + 2OHCa2+ + 2OH+ SiO2 (Clay Silica) CSH Ca2+ + 2OH+ Al2O3 (Clay Alumina) CAH Where: C= CaO, S= SiO2, A= Al2O3, and H = H2O. A wide variety of hydrate forms can be obtained, depending on reactions e.g. quantity and type of calcium, soil characteristics, curing time and temperature (Dermatas and Meng, 2003). The reaction of LimeKD with water is also called slaking yieldi ng hydrated LimeKD. The heat liberated by hydration is dependent upon the content of CaO. The heat liberated by the reaction of (CaO) with water is about 1140 kJ kg-1 and in the case of dol omite (CaO. MgO) with water, it is ~880 KJ kg-1 (Boynton, 1980). LimeKD is less expens ive to ship because it weighs less. However, it can be dangerous to handle because of the high energy released when it is mixed with water. LimeKD will react with atmospheric moisture and CO2 in the air to reform calcium carbonate. This carbonation is a reversal of the calcining reaction. It is a relatively slow reaction, but, once carbonated, LimeKD is rendered as an ineffective so rbent (Little, 1995). When LimeKD is added to a clayey soil, it has an immediate effect on th e properties of the soil chemistry. Ion exchange takes place between the meta llic ions associated with the surfaces of the 28

PAGE 29

clay particles and the calcium ions of the li me. Clay particles are surrounded by a diffuse hydrous double layer, which is modified by the ion exchange of calcium. This alters the density of the electrical charge around th e clay particles leading to fl occulation-agglomeration (Bell, 1996). Raising the pH of the mixt ure also increases the cation ex change capacity; encouraging further replacement of cations by calcium. The reduction in the diffuse hydrous double layer allows the particles to align in a more edge-t oface manner. This new configuration improves the workability of the soil by improving aggregation. (TRB, 1987). SuperMag is a MgO waste product from coal processing. It is converted to magnesium SO4 by a fertilizer manufacturing company (LC Supe rMag Bradley, Florida). It is a mixture of magnesium SO4 and other micronutrients. Currently, little or no information is available on this magnesium SO4 residue. It has a tremendous capacity to sorb P due to the presence of magnesium and or MnO or SO4. Among some of its advantages are having minimal environmental impact, has a low solubility and its essential nutrient for plant, animal and human growth. It has a high alkalinity, which helps to neutralize acids and precipitate metals. The Mg/Mn contents will form solid phases with P. Land application could function as a means of SuperMag disposal and immobilization of P in poor ly P sorbing soils. Further, manure impacted soils tend to have alkaline pH. Heavy rainfall in Florida has contributed to leaching of nutrients and to Mg deficiencies. The dissolution of magnesium SO4 will lead to SO4 forming complexes with Al (thus it will minimize the toxic effect of Al). This may allay the fear in using Al-WTR. In addition, Mg would also complexes with or thophosphoric species rende ring the bioavailable P insoluble through precipitation. The addition of SuperMag may likely change the pH to a desirable level, and provide th e right amount of nutrients for specific forage species. E.g. Bahiagrass (Paspalum notatum ) production requires pH to be s lightly acidic, pH about 5.5 (Adjei 29

PAGE 30

and Rechcigl, 2004). In addition, Mg -P bearing minerals have been implicated to control the solubility of P due its undersat uration in manure-impacted soils (Cooperband and Good, 2002; Josan et al., 2005; Silveira, 2006). Gypsum (CaSO4.2H2O) is a white crystalline naturall y occurring mineral. It is also byproduct of PO4 refining and electric power industries, as well as from other processes. The most significant producer of by-product gypsum is the PO4 fertilizer industry. Gypsum used was obtained from Ben Franklin, Agri cultural Gypsum, (United States Gypsum Co., Chicago IL). The rock-containing mineral fluroa patite is treated with sulfuric acid to produce phosphoric acid; gypsum is a product of the reaction as follows: Ca10(PO4)6F2 + 10H2SO4 +20H2O = 6H3PO4 + 10CaSO4.2H2O + 2HF Another major source of gypsum is from the removal of SO2 from exhaust gases of coalfired power plants. The reaction is as follows: SO2 (g) + H2O = H2SO3 H2SO3 + CaCO3 = CaSO3 +CO2(g) + H2O By wetting and forcing oxidation with air pumpe d through the slurry, gypsum precipitates and hydrates as CaSO4.2H2O. Phosphorus uptake by gypsum is due to adsorption on reactive surfaces, which include calcite, and precipitation of calcium phosphate minerals (Lindsay et al., 1989). Precipitation causes dicalcium phosphate crystal formation then slowly changing to octacalcium PO4 and eventual formation of a stable hydroxyapatite (Arambarri and Talibudeen, 1959). The study will address P sorption with sorb ents, SuperMag, Al-WTR, Lime KD, Slag, Gypsum (CaSO4) and MgO in P-impacted soil. This is to ensure that the various forms of P associated with either Fe-Al or Ca-Mg ions, are stabilized. 30

PAGE 31

Initial Equilibrium Sorption Mass of sorbents used was based on oven-dried weight (at 105oC), sieved through < 850m mesh to minimize slaking and improve reactivity (Dayton and Basta 2005b). Phosphorus sorption was carried out using 2 g Al-WTR MgO, Slag, Gypsum, LimeKD, and MgSO4. Two experiments were conducted. The first study used MgSO4, Al-WTR, MgO, Slag based on availability at the time of the experiment s. The second study used Al-WTR, MgO, Slag, dolomite, Gypsum, LimeKD. The control is soil so lution extracts with no sorbent added. The soil solution extracts were derived from the inherent P (manure-impacted soil). The soil used in the study was taken from Lake Okeechobee, since that is the region, largely affected by P due to manure loadings. The extracted soil solution P was used in order to mimic the natural P sorption that would occur when the sorbent was used. This is because matrix effects appear to influence P sorption. The extraction was done with a pH water of 5.0, having 0.01M KCl as background electrolyte, soil solution ratio of 1: 500. This soil solution is used in order to release all the labile P as possible. The soil solution ra tio was equilibrated us ing a horizontal shaker for 96 h, to make sure all P is in soluble form. Aliquots of the solu tion extracts at 1: 20 solid solution were used to equilibrate with the sorbents. In the first study, equilibration of 8h was allowed for the sorbent/soil solution mixture. The second study used 24 h equilibration time. Data from this P sorption were used to select promising sorbents for use in the subsequent studies. Further, a composite soil was also equilibrated with the sorb ents at 2% rate by weight to mass of the soil used. The solid solution ratio wa s 1: 20. The sorption study was conducted at room temperature (23 2 oC). The suspensions were equilibrated on a reciprocal shaker, (Eberbach Corp., Michigan), centrifuge at 3200xg and filtered through a 0.45 m membrane following the reaction, at a frequency of ~ 120 rpm. To minimize matrix interference during an alysis, a good dilution of 31

PAGE 32

aliquots of the samples were obtained. Solu tion P was analyzed us ing inductively coupled plasma-optical emission spectroscopy (ICP-OES, Perkin-Elmer Plasma 2100DV). Quality Assurance and Control: Quality assurance and quality control (QA/QC) protocol were followed, including the use of 5% repeats, 5% spikes and blanks for each procedure during ICP analysis. Standard calibration curves, as well as quality control check standards were, prepared for each procedure. Repeats were within 10% relative standard deviation. In situation were samples standard deviation fell outside the 10% range, repeats were conducted to cross check any anomaly. Phosphorus retained by the sorbents was calculated as follows: s eqolm CCV q )( (Eq. 2-1), where ms is the mass of adsorbent/sorbent (g), in contact with a volume (mL) of the solution Vl. The Co is the initial concentration of the sorbate (ug mL-1), and Ceq is the equilibrium concentration after reaction of the solute (PO4) and the adsorbent (ug mL-1). The q is the mass of the solute per mass of the sorbent (mg g-1). Equation 2-1 was used to calculate P sorbed in relation to initial P concentration. Results and Discussion Preliminary Sorbent Selections Sorption on Composite Soil with Sorbents Prior to soil solution extraction, a prelim inary laboratory equilibration study was conducted on composite soil (Okeechobee) amended with the sorbents. Application rates of 2% (by mass) as taken from previous study, Silveira et al., 2006 were chosen. The main objective for the preliminary sorption study was to observe the rate of sorption pertaini ng to the natural situation of the soil matrix. Mass of soil used was 2g with 20 mL 0.01MKCl as background electrolyte. Sorbents us ed were a granular MgSO4 and a powder type, Al-WTR and MgO. A 24 h 32

PAGE 33

equilibration was allowed. The control so il without sorbents, released 4.7 mg L-1 P (equivalent to 47g P g-1) for equilibration. Soil amended with MgSO4 sorbed the least P. A granular form of MgSO4 removed 17.9% P (8.4 g P g-1), while the powder form removed 22.5% P (10.6g P g-1) from solution. The low P immobilization from MgSO4 might indicate strong competition between the exchange sites of the sorbents and other ions such as soluble organic acids with P. On the other hand, a slow dissolution of MgSO4 takes place over time, si nce P precipitation by Mg is dependent upon Mg2+ activity. In addition, it also gave an indication of inability of MgSO4 to effectively remove P in a strong matrix enviro nment from manure-impacted soil. However, the MgO material reduced soluble P effectively, 79% P (37.1g P g-1) and Al-WTR material immobilized 50% of P in solution (23.5g P g-1). Sorption with Soil Solution Extracts To further explore the sorption differences, so il solution extracts were tested. This is because the composite soil sorption soil solution ratio (1:20) used above might be too high, hence the need for dilution. This approach was to create the environment for immobilization of P from a solution phase as is on to a solid, pertaining to normal field condition. Data are presented as in Figure 2-1. After 8 h of reacti on using soil solution extract at 1: 500, the P remaining in solution was analyzed. The results showed that after equilibration of 1:500 soil solution ratios; P in solution was 2.5 mg L-1 P. This 2.5 mg L-1 P, (equivalent to 25g P g-1) was then used in the sorption reactions with the sorbents for 8h. The Mg O material immobilizes 88% P (22g P g-1) in solution. This was followed by Slag, immobilizing 84 %, (21g P g-1) Thirdly, Al-WTR removed from solution (17.5 g P g-1) representing, ~70%. It was observed that MgSO4 removed 33

PAGE 34

the least P, at 14% (3.5g P g-1). From these data, MgSO4 was deselected from subsequent studies. The greater significant sorption observed (P < 0.001) of P by MgO and Slag to Al-WTR support the idea that mixed sorbents to include Ca-Mg may improve P sorption in manureimpacted soils. Although ~70% reduction of P was obtained from Al-WTR, time of equilibration was too short to make any signifi cant conclusion. However, it appear s that the matrix effects had a strong influence on the sorption of P to Al-WTR in alkaline environment of pH 8.3. This work is unique in that the soil extr acts were used to better repr esent real-world response to soil solution P. This study, therefore, showed the diffe rence between data available in the literature which depended heavily on lab reag ents for sorption experiments. The second study which had the sorbents, dolom ite, LimeKD, and Gypsum equilibrated at 24 h had soil solution extract of 15 mg L-1 P (150g P g-1). The time difference of 8h equilibration and that of 24 h was to allow more time in the sorption of P. The results of the second full study are presented in Figure 2-1. Dolo mite and gypsum had low P sorption of 5 and 6 % (equivalent to 7.5 g P g-1, and 9 g P g-1, respectively). Al-WTR sorbed 67 % P (102 g P g-1) after equilibration, which was close to previ ous observation by percentage (68%) measured in the first study. On the other hand, MgO, Slag and LimeKD had 99.9, 98, 99.9% P removed from solution, (equivalent to 149.85 g P g-1, 147 g P g-1, and 149.85 g P g-1, respectively). These results led to the selection of Al-WTR, MgO, Slag, Gypsum and LimeKD for further study. Gypsum sorbed poorly, but was selected because it is an inexpens ive option and can be readily obtained. Since it is sparingly soluble it is projected that with time, ions might be released into solution to react with soil solution P. 34

PAGE 35

0.0 0.5 1.0 1.5 2.0 2.5 3.0 MgSO4Al-WTRMgOSlagControl Sorbents Solution P (mg L -1) Figure 2-1. P sorbed from soil solution extract as a function of sorbents after 8 h of equilibration. Values are means of triplicate. Error bars are SE of the mean. Error bars for Al-WTR, MgO, Slag, and Soil solution are very sm all and overlap with the top bar. 0 5 10 15 20 Al-WTRMgOSlagDolomiteGypsumLimeKDControl SorbentsSolution P (mg L -1)67% 99.9%98% 5%6% 99.9%Figure 2-2. Phosphorus sorbed (%) from soil solution extracts as a function of sorbents after 24 h equilibration. Values are means of triplicate. Error bars are SE of the mean. Error bars are very small for MgO, Slag and Lime KD, and overlap with the horizontal line. 35

PAGE 36

Conclusion Seven sorbents/chemical amendments were evaluated. Equilibrium sorption at 8 and 24 h equilibration time were tested. The amendments were equilibrated with soil solution extracts containing P from manure-impacted soils. Al-W R sorbed 67%, MgO 99.9%, Slag 98% and Lime KD 99.9%. Poor sorption was observed for dolomite and MgSO4 (granular and powder forms) and therefore were not studied further. Gypsum also showed weak/ poor sorption 6% (9 g P g-1) relative to the initial concentration of P (soil solution extracts). Howe ver, gypsum was kept for further study. Sorbents selected for subsequent studies were Al-WTR, MgO, LimeKD, Slag and Gypsum. Al-WTR was selected based on the assumptions that it be haves like hydr(oxides) containing Al. Whereas, MgO, LimeKD, Slag and Gypsum were selected based on the Ca and Mg contents of the materials. The next issue to address is the surface properties of the sorbents. As mentioned in the introduction, the nature of surface properties has strong influence on the mechanisms of P immobilization. One mechanism is based on the formation of PO4 minerals by chemisorption or precipitation, while the second one is by adsorption onto surfaces of metal oxides. Chapter 3 would address the surfaces properties to gain in sight on how porosity of the sorbents may have effects on adsorption of P thr ough physisorption analyses. 36

PAGE 37

CHAPTER 3 SURFACE ANALYSIS OF SORBENTS: PHYSISORPTION Introduction It is generally accepted that majority of chemi cal interactions on solid surfaces takes place at the solid/liquid, solid/gas, and solid/solid in terfaces (Bolt et al., 1991). Thus, the nature and properties of the solid phase ha s a controlling influence on impor tant chemical interactions. Physisorption is a method used to probe the nature of pore structure of solids. It provides insight on the amount of adsorbate molecules that can be sorbed, based on the pore network and or structure. The surface properties of sorbents play a major role in controlling phenomena such as adsorption or desorption of inor ganic or organic pollutants. Typically, isotherms are plotted as an amount of adsorbate versus the adsorptive pressure. These isotherms reveal hidden behavior of the surfaces porosity and the types of likely mechanisms governing the surface interactions with the incoming adsorbate molecules (Condon, 2006). There are 6 specified classe s of adsorption isotherms based on experimental physisorption adsorption isotherms (IUPAC, 1985). Type 1 isot herm reveals that, th e sorbent is porous (microporous with relative small external surf aces), of the solid/sorbent having a homogenous monolayer coverage, as the partial pressure incr eases and reaches saturation. It also, implies that adsorption of the adsorbate onto a solid surface is controlled by the microporosity. Examples of materials having type 1 isotherm include, molecu lar sieve zeolites, certain porous oxides, and activated carbon. The other classes further reveal that mechanis ms other than adsorpti on are at play, notably chemisorption and/or precipitation. Normally, it reveals two types of interactions of adsorbate/adsorbate i.e. direct interaction be tween adjacent adsorbed molecules and indirect interactions where the adsorbat e changes the surfaces, which in turn affect the adsorption of 37

PAGE 38

other adsorbate molecules. Furthermore, direct inte ractions are similar to those in liquid, leading to clustering of adsorbate. The indirect interact ions are more complex, leading to bond formation on the surface or electrons transfer between the surface and the adsorbate molecules (IUPAC, 1985; Masel, 1996; Condon, 2006). Physisorption was used on selected sorbents to better characterize surface structures to indicate the likely nature of the adsorption phe nomena on the surface. This information would enable us tailor the sorbents through utilization of its most probable mechanisms and adapting the mechanisms towards P speciation in manure-impacted soils. Surface Area and Pore Size Distribution The surface properties of soil pa rticles or sorbents play a major role in controlling phenomena such as adsorption or desorption of inor ganic or organic polluta nts. At the molecular scale, adsorption is defined in terms of relative surface excess, when solute accumulates at the mineral-water interface, or mineral-gaseous interface (Sing et al., 1985). Langmuir, 1918 contributions suggested solutes uptake by adso rbent surface in reaching saturation coverage called the monolayer. Thus, when the average area occupied by each adsorbed solute is known, it would be possible to estimate the surface area of the adsorbent (Sing, 1998). Analysis of surface area is therefore, important in revealing first hand inform ation on possible areas on which adsorption takes place by the interacting solutes on a given adsorbent. The information to draw from these experiments is that, a porous material with high surface area has the potential to sorb more solutes than a material with a less surface area. Classical work done by Bruna uer et al., (1938) and Emme tt et al., (1938) led to quantification of surface ar ea, including and pores that contribu te to internal surface area. Highly porous materials may have both the internal and ex ternal surface areas. However, the porosity is 38

PAGE 39

mainly contributed from the internal surfaces of composite porous materials (Gregg and Sing, 1982). Solid materials have different types, size, and shape of pores Due to the complexities of determining pores width, the international uni on of pure and applied chemistry (IUPAC) has provided guidelines for adsorption studies (Sing et al., 1985). Pores are considered to be micropores, mesopores, and macropores with diameter < 2 nm, between 2 nm to 50 nm and > 50 nm, respectively. For most porous materials, the single most important parameter responsible for adsorption is pore structure. Macropores are the conduit pathways th rough which adsorptive molecules travel to the mesopores, where they eventually enter the micropores. The micropores usually constitute the largest proportion of the internal surface of porous material e.g. activated carbon, clays, and gels contribute most to the total pore volume Thus, the total pore volume and pore size distribution (PSD) determines how much solute (adsorbate) can coat the adsorbent. The use of a N2 gas adsorption isotherm developed by Brunauer, Emmett and Teller (BET) is by far the most commonly used method for determ ine the surface area of por ous materials. It is an extension of Langmuirs monolayer covera ge to multi-layer adsorption. The BET theory provided mathematical support for describing solu te uptake at point B (i.e. point at which monolayer coverage is complete, and multiplayer begins), and it was found to be in good agreement with the BET monolayer capacity ( nm). The main reason for popular usage of N2 rather than other inert gases for physisorption are that it is far, th e least expensive and it is the method recommended by IUPAC, (1985). In classification of pore size distribution of materials, methods of Dubinin and Radushkevich (DR) a nd Barret-Joyner-Halenda (BJH), 1951 are among the most useful for determining pore sizes usi ng a nitrogen isotherm. (Lowell, et al., 2004; Rouquerol, et al., 1999; Condon, 2006). 39

PAGE 40

Numerous research papers have used the BET, BJH and/or DR to characterize the surface area, and pore size distribution of soils, clay s, metal oxides, and activ ated carbon. Goldberg, et al. 2001 used the BET to characterize the surfac e area of amorphous aluminium oxides. Mikutta and Mikutta, 2006, used BET, BJH and DR techni ques to probe the microporosity of organic matter. Further, Sing, (1989) used the BET to char acterize activated carbon. However, little work has been conducted using the BET, BJH and DR to characterize the surface properties of chemical amendments such as Al-WTR, LimeKD Gypsum, MgO and Slag. The Al-WTR, as an exception, (Makris et al., 2005) since the BET, and DR were to characterize surface area and micropores volume of the materials, but they did not employ the use of BJH approach for graphical interpretation of the isotherms. Th e objective of this study was to determine the specific surface area (SSA) of the sorbents and to characterize the pore si ze distribution through physisorption. Material and Methods N2-Physisorption (BET, DR, BJ H) Pore Size Distribution Sorbent Al-WTR, LimeKD, Gypsum, MgO an d Slag were sieved through a < 250m screen to obtain enough colloidal particles according to (Emmett et al., 1938). Specific surface area (SSA) and the pore size distribution ( PSD) were measured with NOVA 1200 sorption analyzer (Quantachrome Corp.) The samples were initially out-gassed at 100 to 150oC for 4-6 h under vacuum to remove any available moisture. This was followed by N2 adsorption-desorption isotherms of the sample measured at 77.4K. The SSA was estimated using multi-point adsorption from the linear segment of the N2 adsorption isotherm (Sing et al., 1985) in the relative pressure range of 0.05-0.3 using BET method (Brunauer et al., 1938): 40

PAGE 41

om m o oP P Cn C CnPPn PP 11 )1( (Eq. 3-1), where, Po is the saturation vapor pressure at the measurement temperature, P/Po the relative gas pressure, n the amount of gas adsorbed per mass sample, nm is the monolayer adsorption capacity and C is a constant related to the enthalpy of the gas adsorpti on. The parameters nm and C were determined from equation 3-1. The surface coverage ( ) of adsorbate for both mono and multiplayer adsorption is defined as the ratio of the amount substance adsorbed to the monolayer capacity. The surface area (SAs) of the adsorbent was then ca lculated from the monolayer capacity (nm in moles), provided that the area (am) effectively occupied by an adsorbed molecule in the complete monolayer is known. Thus, SAs = nm.NA.am (Eq. 3-2), where NA is the Avogadro number and the specific su rface area (SSA) is calculated as: SSA = SAs/mass of adsorbent (Sing et al., 1985). Pore size distributions were estimated for micro-pore volume (MIV) by the DubininRadushkevich (DR) method (Gregg and Sing, 1982), and for meso-pore volume (MEV) by Barrett-Joyner-Halenda (BJH ) method (Barrett, Joyner and Halenda 1951) using NOVA 1200 sorption analyzer. The BJH method utilizes the Kelvin equation with the assumptions of cylindrical pore geometry. Th e Kelvin equation reduces to: )log( 15.4 )(0PP Ark (Eq. 3-3), where, rk is the radius of the pore in which condensation occurs at re lative pressure of P/Po. The BJH values were obtained from the software, once the information needed is set from the start of 41

PAGE 42

the experiment. The software is equipped to do the calculations. The Dubinin-Radushkevich (DR) is given by: )(log)log()log(2P P DVVo o (Eq. 3-4), where V is the volume of gas adsorbed per mass of sample (cm3 g-1) that is calculated from V = np-1, where n is the mass of gas adsorbed (g g-1) and is the liquid density of the gas used (g cm3); Vo is the total micro-pore volume (cm3 g-1), and D is a constant related to the structure of the adsorbent (D = 2.303 k R2T2/ 2, where, k and are constants) and the adsorbate-adsorbent affinity. The MIV was obtained as th e intercept in a plot of log ( V ) vs log2 ( Po/P ) after extrapolation from the linear region in the Dubinin-Radushkevich plot (Dubinin, 1960; Tsunoda, 1977; Mikutta and Mikutta, 2006). The software is equipped to pe rform numerical analyses of the pore volumes. Results and Discussion Data in Table 3-1, show the differences be tween the SSA, DR and BJH analyses using BET-N2 measurements for the sorbents. The SSA of Al-WTR was the greatest, at 20.7m2 g-1. This was followed by MgO of 5.0 m2 g-1. Gypsum, Slag, LimeKD and Dolomite had 3.1, 3.3, 2.0, 1.8 m2 g-1, respectively. All SSA sorbent values we re relatively low, although Al-WTR had the greatest. Comparing the sorben ts data with other sorbents su ch as Al oxides and activated carbon suggest, possible limited area for physisorption (Goldberg, et al ., 2001; Kruk et al., 1999, IUPAC, 1985). In addition, the PSD showed that all the sorbents had greater mesopores volumes in comparison to micropores. According to (Si ng, 1982), micropores are the seat for adsorption for porous materials. True microporous materials have a ratio of micropor es to total pore volume to close to 1 (Kruk et al., 1999). In other words, the volume of the mesopores is negligible. AlWTR had ratio of micropores to mesopores to be 0.397, a value far less than 1. Further, if we assume the sum of the micropores and mesopores to constitute the total pore volume, micropores 42

PAGE 43

volume amounted to 28% of the total. G ypsum, Slag, LimeKD, MgO and Dolomite had micropore volumes of 26, 26, 23, 29 and 25%, resp ectively to total por e volume. Since the micropores were found to be limited, (i.e. less th an 1), the adsorbate area for physisorption is also likely low. The implication is that, the te sted adsorbents may be more limited in removing contaminants than pure metal oxides. However, li quid-solid isotherm showed that sorbents could remove P (Chapter 2 and 5). T hus, suggesting that P removal is likely through other mechanisms aside from adsorption, and mo st notably by precipitation. To validate the adsorbate surface coverage resu lts, pore size distribution was carried out. The graphical physisorption data of the chemical amendments were compared with standard graphical isotherms from IUPAC (Sing, 1985). Figur e 3-1, illustrates the experimental standards isotherms from the IUPAC. In (I) under Figure 31, the isotherm represents sorbents that are highly porous with the pores represented by almo st 100 % micropores. The rest of the isotherms (II to VI) are said to be non-por ous. That is the porosity is not governed by micropores, but by mesopores or macropores. For Al-WTR, the BET-N2 graphical isotherm is presented in Figure 33. The data clearly shows Al-WTR was a type (IV) isotherm. Additionally, the graph shows a multilayer adsorption on a non-porous surface (Hodson, 1999). The non-porosity does not mean there are no micropores. This is a term used relati ve to true internal mi croporous nature for type 1 isotherm that, normally show microporosity to govern the adsorption process with little external surface area ( Lowell, et al., 2004). In ot her words, the majority of the pores for true porous material are the micropores. In principle, a solid sorbent is regarded as porous only if the surface irregularities are deeper than wider at the surface (Rouquerol, et al., 1999). 43

PAGE 44

Table 3-1. BET-N2 of specific surface area, micropore volume and mesopores volumes of sorbents, obtained at P/Po= 0.99 of total pore volume. Types of Sorbents Specific surface area SSA (m2 g-1) a DR (cm3g-1 x 10-3) (Micropores volume) b BJH (cm3g-1 x 10-3) (Mesopores volume) Al-WTR 20.7 17.3 43.53 Gypsum 3.1 2.73 7.76 Slag 3.3 1.83 5.20 LimeKD 2.0 1.68 5.52 MgO 5.0 4.30 10.52 Dolomite 1.8 1.06 3.25 a Dubinin-Radushkevish method for micropores volume determination. b Barrett-JoynerHalenda method for mesopor es volume using BET-N2 isotherm. Figure 3-1. Types of physisorption isotherms (IUPAC, 1985) 44

PAGE 45

Figure 3-2. Types of hyste resis loops (IUPAC, 1985) 0 5 10 15 20 25 30 0.00.10.20.30.40.50.60.70.80.91.0 P/PoVolume Adsorbed @STP [cm3/g] Adsorption Desorption Figure 3-3. N2 adsorption and desorption isotherm for Al -WTR as a function of relative partial pressures conditions near to saturation pressure P/Po = 0.99. Further, the porosity is regarded as an intrinsic property of the solid sorbent Type (IV) is observed for Al-WTR, with hysteresis loops, im plying the filling and emptying of capillary condensation in the mesopores (Sing, 2001) The condensation could mean direct adsorbate/adsorbate interactions taking place, thus leading to clustering of the adsorbate 45

PAGE 46

molecules in the mesopores or adsorbate react ing with the surfaces of the adsorbent for adsorption. With this phenomenon, adsorption with in the micropores is thus limited. However, comparing the curvature of point B as in Figure 3-1 to Figure 3-3, thus suggest the presence of limited micropores, which may contribute to wards adsorption and was supported by PSD calculations data of the materials (Table 3-1). T ype (IV) also means that the adsorbents possess mesopores structures. This was clearly shown with BJH calculations in Table 3-1 with majority of the pores as mesopores. Furthermore, the hysteresis loop observed in Figure 3.3, suggest a type of slit-shaped like pores i.e. a ty pe H4 loop as in Figure 3-2 (IUPAC, 1985). The low proportion of micropores observed in Al-WTR, but showing good P sorption in liquid-solid interface, enabled other researchers to account for low microporosity. Makris et al., 2004, Lazano-Catello, et al., 2004 used CO2 to probe into the struct ure of carbonaceous materials because; it is believed that CO2 can penetrate inner st ructure better than N2. However, the use of CO2 has generated debate in the li terature. The ionic radius of CO2 and N2 are similar, 2.8 and 3.0, respectively. Therefore, there may not be a ny significant penetrati on power attributed to CO2. Secondly, Al-WTR has about 20% carbon. The surface chemistry of carbon showed moieties of different functional groups. In addition, carbon has acidic and alkaline surfaces. The use of CO2 may interfere with the inner structure of pores due to interactions with the adsorbent. Also, classical work from the original developers of BET-N2, published in 1938, also use CO2 but did not encouraged it for soil-like, or soil colloidal material s (Emmett et al., 1938). In addition, gasification in CO2 is a procedure done to bring to life, dead activated carbon whose pores are blocked by other gases, and adsorbates (Para, et al., 1995). The IUPAC had recommendations and procedures for in PSD dete rmination and measurements. The standard gas used is N2. Further, recent works from Ozdemir et al., (2003);and Dutta et al.,( 2007) suggests 46

PAGE 47

CO2 can be stored on carbonaceous materials such as coal, in an attempt to mitigate the CO2 responsible for global warming. Furthermore, De Jonge and Mittelmeijer-Hazeleger, (1996) used CO2 in micropores determination for soil organi c matter. However, the authors cautioned the interpretation of CO2 data. They expressed the difficulties in reaching true equilibrium. Also, they argued, on the use of surface area concept for polymer-like structure with pores having the same scale as the penetrating molecules may have influence on the results, since there is no clear pore-surface interface as there is in the larger pores. The SSA of Al-WTR de termined was ~20.7 m2 g-1, at a partial pressure range of (0.05-0.3) whiles data from Makris, et al., (2004) was~36 m2 g-1, at a partial pressure range of (0.03-0.3). The difference in the SSA measurement is likel y due to differences in the methodology. Makris, et al., (2004) used, helium gas to outgass the samples at 70oC for 4 hrs, and selected the partial pressure range of 0.03-0.3, instead of the nor mal vacuum outgassing before the use of N2 sorption. A full physisorption isotherm determina tion as in Figure 3.3 thus, helps to throw more lights on the nature of the sorben ts. The isotherms revealed the sh ape of the graph and help to suggest the possible sorbent beha vior with incoming adsorbate. Physisorption data for Slag and MgO and G ypsum are presented in Figures 3-4, 3-5 and 3-6 respectively. All data showed the non-porosit y of the materials. Specifically, Slag and Gypsum showed type (IV) isotherm, whereas Mg O indicates type (II) isotherm. Description for type (IV) was explained for Al-WTR above. Howeve r, as observed in Table 3-1, major pores for both Slag and MgO are mesopores. According to Condon, 2006, type (IV) or type (V) generate mesopores. Mesoporosity was clearly observed in Sl ag and MgO. Graphical data for LimeKD is not shown. Data however, are close in values to that of Slag as obs erved in Table 3-1. 47

PAGE 48

Magnesium oxide on the other hand, showed a type (II) isotherm. Type (II) isotherm is the normal isotherm usually obtained for non-porous or flat surface adsorbents It also represents unrestricted monolayer-multilayer adsorption at higher P/Po (Sing. 1985; Rouquerol et al. 1999). This also indicates the heterogeneity of su rface. Surface heterogeneity complemented from observed data in pore size tabulati on (Table 3-1). Further, lack of a hysteresis loop, i.e. complete reversibility of adsorption-desorption suggests an open and stable surface 0 1 2 3 4 0.00.10.20.30.40.50.60.70.80.91.0 P/PoVolume Adsorbed @STP[cm3/g] Adsorption Desorption Figure 3-4. N2 adsorption and desorption isotherm for Slag as a function of relative partial pressures condition near to saturation pressure P/Po = 0.99. 48

PAGE 49

0 2 4 6 8 10 0.00.10.20.30.40.50.60.70.80.91.0 P/PoVolume Adsorbed @ STP [cm3/g] Adsorption Desorption Figure 3-5. N2 adsorption and desorption isotherm for MgO as a function of relative partial pressures condition near to saturation pressure P/Po = 0.99. The stability of the surface may support simila r mechanisms notably that of precipitation reactions. The shape of the isotherm observed appear s to be similar to graphs from Ribeiro et al., (1991), by calcinations of Mg(OH)2 to produce MgO at various temperatures. For the rest of the data (sorbents), there appears to be no literature available on phys isorption isotherms to compare with the present. This may be due to the time consuming nature of PSD analysis. 49

PAGE 50

0 1 2 3 4 5 60.00.10.20.30.40.50.60.70.80.91.0P/PoVolume Adsorbed @ STP[cm3/g] Adsorption Desorption Figure 3-6. N2 adsorption and desorption isotherm for G ypsum as a function of relative partial pressures condition near to saturation pressure P/Po = 0.99. The result of the physisorption isotherm analysis has implication for solid-liquid adsorption. Type (I) physisorption is analogous to L-type isotherm for solid-liquid adsorption. Kinetic data for PO4 reacting with all the sorbents showed L-type (Data not shown). This implies that the adsorbate (PO4) has a relatively high affinity for the adsorbents surface at low surface coverage (Essington, 2004). However, as cove rage increases, the affinity of the PO4 for the sorbent surface decreases. McBride, (1994) cautioned that isotherm features could not prove the actual reactions mechanisms occurring. However, the information from the isotherm can point reasonably to mechanisms, which must be confirmed with othe r molecular techniques. Other techniques such as SEM-EDS and geochemical models would be employed to elucidate the likely mechanisms occurring with phosphate reactions on the sorb ents. The main information revealed by the physisorption data is that the su rfaces of the sorbents were highly non-porous and heterogeneous, with limited microporosity. Microporosity is the seat for adsorption, which is limited by the 50

PAGE 51

sorbents. This is the usual characteristics of a type (I) isotherm, associated with a micropore filling process (Sing, 2004). This inform ation points towards the degree of PO4 adsorption that might occur on the surfaces of the adsorbents most probably reactions may be that of a precipitation reaction an d or chemisorption. Chemisorption may be defined as the adsorpti on in which valence forces results in the formation of chemical compounds (IUPAC, 1972). However, no absolute sharp distinction can be made between physisorption and chemisorpt ion. A chemisorption reaction may be observed when the adsorptive reaction occurs in such a wa y that desorption of the original species cannot be recovered. Thus, suggesting that the chemisor ption may not be reversible. A typical example is as in Figure 8-4 using flow calorimetry. In a ddition, the energy of chemisorption is of the same order of magnitude as the energy change in a chemical reaction betwee n a solid and a fluid. Thus, chemisorption like chemical reactions in general may be exothermic or endothermic and the magnitude of energy changes may range from very small to very large (IUPAC, 1972). On the other hand, in physisorption, th e energy of interaction between the molecules of adsorbate and the adsorbent is of the same order of magnitude as, but usually greater than the energy of condensation of the adsorptive. Finally, a physisorbed molecule keeps its identity and on desorption, returns to the fluid phase in its orig inal form. Whereas, if a chemisorbed molecule undergoes a reaction or dissociati on, it losses its identity and ca nnot be recovered by desorption (Rouquerol et al., 1999). A cl ose observation with the isotherms for the sorbents suggests that the materials exhibit attributes of both physisorption and chemisorption. Conclusions Physisorption data analysis reveals the pore size distribution (micr opores, and mesopores) of the sorbents, as well as the specific surface areas of the sorbents. Th e use of specific surface 51

PAGE 52

area (SSA) data was not enough to show the pore structures of the materials. Comparing the physisorption data to that of cl assified graph from (IUPAC, 1985), shows that, the sorbents were non-porous. The non-porosity is an indication of de viation from a true microporous nature of type 1 isotherm. All sorbents exam ined showed non-type 1 isotherm. The presence of limited micropores on the sorb ents does not necessary mean P would be adsorbed to the available sites at the interfaces of the solid. The amorphous nature of the AlWTR should be taken into consideration as well as the possible Al hydrolysis that may occur when the material is in solution. However, Figur e 3-3 suggests, clearly adsorption was controlled by other secondary mechanisms, such as chemis orption and/or precipitation other than by reaction driven by microporosity. This study, therefore, sugges t that knowing the amount of micropores and that of the mesopores, as well as physisorption isotherm are needed together to draw some reasonably conclusions on the nature of possible reactions that may occur on sorbents used for soil contaminant immobilization. Usi ng microporous volume alone, without the specific surface area of the micropores may not provide enough evidence on the capacity of the micropores for adsorption. Physisorption data pinpoint a direction for possible mechanisms. Perhaps, due to the long time used in determin ing the isotherm graph (physisorption) prevent other researchers in its determination. In addition, isotherm data for other sorben ts suggested non-porosity, with Slag and Gypsum behaving similarly like Al-WTR, type (IV) isotherm. The physical surface properties of the sorbents discussed above, s uggested that microporosity of the sorbents are not the main driving factor for adsorption. However, secondary mechanisms such as chemisorption and or precipitation could be possible reactions that might occur on the surface e.g. if P reacts with sorbents. A close observation with the isotherms of the sorbents suggests that, attributes of both 52

PAGE 53

physisorption and chemisorption can occur on the surface .The next chapter discusses the effects of the sorbents on manure-impacted soil with respect to the P speciation available through incubation sequential extraction. 53

PAGE 54

CHAPTER 4 SORBENTS-P INTERACTIONS AND IM PACTED SOIL: EVIDENCE WITH INCUBATION AND SEQUENTIAL EXTRACTION Introduction Soil incubation study is a general accepted ap proach adapted to mimic soil conditions under natural situation. Parameters of temperature, humidity and moisture content can be controlled to suit the condition desired. The differences in the e xperimental procedure set up, is the location, which is that of lab (Dou et al., 1996, Liu, et al., 2008). This is to allow reactions of the sorbents to the soil such as mineralization or adsorbed P species to be monitored. Thus, enables quantification of the desired chemical species to be achieved. Sequential extraction procedure is a tool that allows me tals, oxyanions, and organic compounds that form complexes in soils and sediments to be quantified. The procedure can be used to track the mobility of species of meta ls, oxyanions and organic complexes (Chang et al., 1983, Sulkowski and Hirner, 2006). Soil incubation and sequential extraction procedures were used to investigate the mobility of P species through reactions with sorbents. Addition of manure can cause an increase in soil pH (Eghball, 2002; Iyamuremye, 1996), due to inputs of large amounts of carbonate an d hydroxyl associated with Ca. (up to 60 g kg-1 Ca) has been observed in the manure sample (S harpley, et al., 2004). It has been shown that fractionation of soil-P has significa nt differences in th e distribution of orga nic and inorganic P forms. Soils that are manure-impacted revealed th at, P is associated with Ca-Mg, Al-Fe, labile, and organic complexes (Nair, 1995; Silveira, et al., 2006). It is against the background of P differences in distribution led to the followi ng hypothesis: Reaction of sorbents on manureimpacted soils may reveal differences in P distribution. The objective of the study is to characterize soil P with sorbents through sequential extraction. 54

PAGE 55

Materials and Methods Soil Incubation with Sorb ents/chemical amendments The soil used for the study was a manure-impacted sample of Immokalee fine sand (sandy, siliceous, hyperthermic Arenic Haplaquods) was obt ained from a field site on a dairy cattle ranch (Butler Oak dairy farm) located in the Lake Ok eechobee County water shed. This soil was used because it received long term manure deposition, it had low P sorbing capacity, and the soil is geographically abundance in South Florida. Multiple random samples were collected from the A horizons (0-15 cm), and were thoroughly mixed to yield a composite sample. The impacted surface soil was air dried and sieved through a (< 2 mm). A second soil taken from the Cattle Range Research Center, Ona (Pomona fine sand, sandy siliceous hyperthermic Ultic Haplaquod) was also included for fractiona tion (P distribution) only. The mass of Al-WTR samples was calculated on a dry mass basis, and sieved through < 850m to minimize slaking and increase reactiv ity (Dayton and Basta, 2005b). Low activity magnesium oxide (MgO) and Slag were treated in a similar manner as Al-WTR. Amendments were applied at ~ 2% rate to the soil. The trea tments were in triplicat es and placed in 50 mL glass centrifuge and were laboratory incubated fo r 4 wks. During the incu bation, all tubes were maintained at 10% (g g-1) moisture content, which represente d typical soil moisture content at field capacity for sandy soils of Fl orida (Silveira et al., 2006). Periodically, every 2 to 3 days, the caps were removed to allow for air exchange, an d sufficient water (DI) was added to account for moisture loses before resealing. Temp erature of incubation was at 23 2oC. Total recoverable Fe, P, Al, Ca and Mg were determined using inductively coupled plasma optimal-emission spectrometry (ICP-OE S, Perkin-Elmer Plasma 2100DV), following digestion according to EPA Method 3050B (USEPA, 1996). Soil and amendments incubated at 4 wks were sequentially extracted for P as determined according to Chang et al., (1983), using a 55

PAGE 56

1:20 soil/solution ratio, with modification as per Silveira et al., 2006. Howeve r, in this analysis the organic P was not analyzed since organic P is usually very small. Sequential Extraction of Soil/Sorbent Amended Soils Sequential extraction procedures allow soil P to be separated and char acterized as P into different forms. A modified se quential fractionation scheme of Chang et al., (1983) was adopted to distinguish between the va rious inorganic and organically bound soil P pools. Approximately 1.5 g of ( 2 mm sieved) soil was weighed into a pr e-weighed 50 mL centrifuge tube. To each tube, 30 mL of 1M KCl was added and shaken fo r 2 hr for soluble P extraction. The tubes were then centrifuged (Sorvall legend ( RT)) at 3200 X g for 20 min. The supernatant was vacuum filtered (0.45 m) into 20 mL scintillati on vials and stored at 4-7 oC until analysis for inorganic Pi using inductively coupled plasma optimal-emission spectrometry (ICP-OES, Perkin-Elmer Plasma 2100DV). Phosphorus extracted by KCl is operationally defined as soluble P, and is regarded as readily labile P (plant available and leacheable). The second extraction step involved shaking th e residual from step 1 with 30 mL of 0.1M NaOH at 250 rpm for 17 hr to extract Fe and Al bound P. The suspensions were centrifuged and filtered as described above. The solution was usually darkly colored; hence, an aliquot of the solution was acidified with one drop of concentrated H2SO4 per mL of supernatant to precipitate soluble organics. The solution was then centr ifuged at 3200 X g for 10 min before analysis (NaOH Pi). The original NaOH supernatant was digested with a sulfuric acid and potassium persulfate (USAEPA, 1993) for determination of Feand Alassociated total P (NaOH TP). Organic P sorbed by Feand Al (NaOH Po) was estimated by the difference between NaOH Pi and NaOH TP values. 56

PAGE 57

The final step in the extraction sequence wa s a 24-h reaction with 0.5 M HCl (1:20 solid: solution) to extract Caand Mg-bound P. The suspension was centrifuged and filtered as above and P analyzed for Caand Mg-bound P. The Ca-and Mg P forms are typically minor constituents in soils, but are significant in manure-amended soils (Nair et al., 1995). The sum of P in the three extractants is generally considered as total in organic P in a material (OConnor et al., 2002). The unextracted P (residual P) is usually considered to be recalcitrant organic P. In most cases, this quantity is negl igible and was not quantified here The sum of all fractions (Seq. Sum) approximated total P, but in this case minus the residual P. Results and Discussion General Properties of the Amendments and Soils The physicochemical properties of Al-WTR take n from Bradenton, Florida, showed ~ 86 g kg-1 Al content, and ~2.3 g kg Fe content (Table 4-1). The high Al content suggests there should be reasonably good P sorption. Slag showed good representation of Ca, Mg and Al (59.8, 10.6, 15.7 g kg-1) respectively. Slag represents a type of co-blended material. In comparison, MgO had greater content of Mg ~ 90g kg-1. Total P in Okeechobee soil was > 2800 mg kg1indicating the P load was very high P. Soil taken from Ona with pH ~7.7, had ~ 1000 mg P kg-1. Sequential extractions suggested a greater amount of P was associ ated with Ca-Mg, representing ~ 72% of total P for Okeechobee soil. Further, so il taken from Ona also confirmed Ca-Mg had greater P content, ~59% of total P (Figure 4-1). The above data are consistent with results from Nair et al., 1995 and Sharpley et al., 2004 showing that Ca-M g bound P has greater proportion of P distribution in alkaline and heavily manure-impacted soils. 57

PAGE 58

Sequential Extractions of Soil and Soil Incubated with Amendments A sequential extraction of soil and amended in cubated soil with Al-WTR, MgO and Slag is shown in Figure 4-2 to 4-4. Labile P, (1 M KCl extraction) revealed that soil amended with AlWTR had soluble P (~310 mg kg-1) indicating, susceptibility for eas ily leacheable P (Figure 4-2). Comparing to the control (no amendment), Al-WTR sorbed 22% labile P. On the other hand, the amount of P removed from solution by MgO and Slag amended soil was (~ 70%) to the control. The implication of the data suggested that Mg -Ca based materials sorbed more P in manure impacted soil than Al-based residual. As i ndicated above, ~ 70% of the soil P is loosely associated with Ca-Mg components. Addition of Ca-Mg based materials may simply reinforce formation of solid phase by preci pitation. Further, for amendments application (Al-WTR, Slag, MgO), each is statistically different at ( = 0.05, determined by Tukeys HSD test). Extraction with (0.1M NaOH) method removes P that is strongly chemisorbed and bound to Fe-Al, in organic and inorganic forms (Figure 4-3). The data showed P bound to Fe-Al were effectively removed by Al-WTR, ~750 mg kg-1 than MgO and Slag (440 and 395 mg kg-1) respectively. Statistically, soils ame nded with Al-WTR were significant ( P < 0.05 ) whereas, that of Slag and MgO were not significantly different from each other (Tukeys HSD test). Thus suggesting, P species (H2PO4 -/HPO4 2-) in solution were affected by the types of amendment incorporated (i.e. metal cations availability). Further, sequential extraction with (0.5 M HCl), aimed at dissolving P bound to calciummagnesium components (Figure 4-4). The data re vealed greater amount of P bound to Slag and MgO (1980 and 1900 mg kg-1) respectively. Al-WTR sorption of P bound to Ca-Mg was rather very low ~ 49% in comparison to Slag and Mg O suggesting its limitation in sorbing the large portion (70%) of P asso ciated to Ca-Mg. 58

PAGE 59

The individual applications of Al, Fe, Ca and Mg, containing amendments to the soil suggest, that P is differentially sorbed by different the amendments with respect to P species. Each P species sorbed corresponded to the major species existing at the pH of the sorbents. It is against the background of P species attacking different metal cations led to the novel idea of co-utilizing the amendments together in speeding up P immobilization. The application of the amendments would attack th e existing major P species concu rrently. Further, it would limit the amount of Al-residual appl ication. Thus the process of co-blending may be tending to reinforce dual pathways as mechanisms for P species immobilization by Mg-Ca and Fe-Al based materials concurrently. Table 4-1. General properties of amendmen ts and sequential extraction of soil P. Sorbents Total elements (g kg-1) aProperties P Fe Ca Mg Al pH Al-WTR 0.02 2.3 4.6 0.5 86.1 6.3 Slag 0.03 1.4 59.8 10.6 15.7 11.5 MgO b BDL BDL BDL 87.9 BDL 10.9 #Soil (mg kg-1) cKCl-P 205 d NaOH-P 362 f HCl-P 1400 pH 7.1 a Values are means of triplicates. b BDL (below detection limits). c,d,f Soil fractionations determined using the method of Chang et al. 1983. #Okeechobee soil. 59

PAGE 60

0 400 800 1200 1600 LabileFe-AlCa-Mg Forms of phosphorus(mg P kg-1) Ona soil Okeechobee soil Figure 4-1. Sequential fractiona tion of manure-impacted soils (Ona and Okeechobee), showing distribution of P. Errors bars represent stan dard errors of mean of triplicate. Errors bars for Fe-Al (both soils) are very sma ll and overlap with the horizontal line. 0100200300400500 Control (soil) Soil+Al-WTR Soil+Slag Soil +MgOSoil + amendment s ( mg P kg-1)a b c dFigure 4-2. Soluble and extractabl e P (KCl-extraction) for soil in cubated with Al-WTR, Slag and MgO. Means (n = 3) of the same lett ers are not significantly different at = 0.05, determined by Tukeys HSD test. 60

PAGE 61

0100200300400500600700800 Control (soil) Soil+Al-WTR Soil+Slag Soil+MgOSoil + amendment s mg P kg-1a c c b Figure 4-3. Fe-Al bound P (NaOH-extraction) for so il incubated with Al-WTR, Slag and MgO. Means (n = 3) of the same letters are not significantly different at = 0.05, determined by Tukeys HSD test. 05001000150020002500 Control (soil) Soil+Al-WTR Soil+Slag Soil+MgOSoil + amendment s mg P kg-1a c a bFigure 4-4. Ca-Mg bound P (NaOH-extraction) for soil incubated with Al-WTR, Slag and MgO. Means (n = 3) of the same letters are not significantly different at = 0.05, determined by Tukeys HSD test. 61

PAGE 62

Conclusions Sequential extractions of manur e-impacted soils suggest that a large proportion of P is associated with Ca-Mg bound complex. Sorbents were incubated with manure-impacted soils. Each tested sorbents targeted different soil P forms. Al-WTR serves as a sink for P forms associated with Fe-Al. The result of the effect of Al-WTR on Fe-Al associated P is statistically significantly greater than that of MgO and Slag amendments using Tukeys HSD. A Ca-Mg based materials also served as a sink for P a ssociated with Ca-Mg bound P fractions. There is therefore, a differential sorption of P forms by the sorbents with respect to availability of Fe-Al and Ca-Mg. The effects of Ca-Mg based material s, i.e. Slag and MgO, were significantly different than that of Al-WTR for P forms a ssociated with Ca-Mg using Tukeys HSD. Each sorbent removed a specific P forms as influe nced by the soil pH derived from the sorbent amendmended. To completely sequester P from manure-impacted soil, the results suggest the need to tie each P fractions available in the soil. Thus, the use of two different sorbents amended together by means of the socalled co-blending technique. Th e methods aimed at targeting different P species for rapid removal. The co-blending technique will be described in the subsequent section. 62

PAGE 63

CHAPTER 5 SOLID/LIQUID REACTIONS: EQUILIBRI UM VS. KINETICS SORPTION OF P Introduction The equilibrium sorption approach has been us ed extensively in the literature to study P uptake by sorbents. The drawback to equilibrium studies is that, most often is not applied to field condition, since soils are nearly always at dis-equilibrium with respec t to transformation and interactions with molecules (S parks, 1989). However, equilibrium studies are normally used to predict the maximum sorption capacity of sorbents, which requires relatively simple laboratory experiment. The experimental set up may be si mple but may be a time consuming process to complete the reactions. Batch experiments are mostly used to study sorption reactions of solid (sorbent)/liquid (adsorptive) interactions. The pr ocedure utilizes a known mass of sorbent (adsorbent), which is placed in a container with a known volume of th e reacting solutions (adsorptive). The sorbent and the reacting solution or suspensions are allo wed to equilibrate at constant temperature (isothermic) and pressure for different time intervals. The equilibration process allows the suspension to be mixed well, in order to allow maximum sorbent-adsorpti ve interactions. After definite time intervals, the suspension is filtere d, centrifuge and aliquots of the supernatant is analyzed for the concentration of the adsorpti ve remaining in solution. The difference between the initial concentration of th e adsorptive use in the reactions and that of the suspension remaining in solution is take n as sorbed concentration. Mathematical equations that relate the amount retained by the solid phase/sorbents(Sads), to the concentration remaining in solution (Ceq) at equilibrium can be used to quantify the adsorptive concentration. Plots of the quantity of P adsorbed versus equilibrium concentration of the adsorptive concentration are referred to as adsorption isotherms (Pierzynski et al., 2000, 63

PAGE 64

Sparks, 2002). Mathematical equations such as Langmuir, Freundlich, Temkin, and DubininRadushkevish can be used to express the isotherm that fits the sorption process (Ho et al., 2001, Pierzynski et al., 2000). Batch e xperiment are used in this sect ion to investigate the sorption behavior of the sorbents during the solid (sorbe nt)/liquid (P containing so lutions) interactions. Materials and Methods Equilibrium vs. Kinetics Sorption of P Calculations of sorbents are based on dry mass, and sieving to < 850m to avoid slaking (Dayton and Basta, 2005b). Phosphorus sorption was carried out using sorbents of Al-WTR, MgO, Slag, Gypsum, and LimeKD of 2 g each. Equilibrium and sorption kinetics studies were performed on the sorbents at initial concentrations of ~ 40, 60, 100, 300 and 500 mg L-1 P solution, with 0.01M KCl as a background electrol yte at pH ~ 4.8. A solid solution ratio of ~ 1:30 were used and equilibration achieved at 24 hrs (the time required for equilibrium to be reached between P adsorbed and P in solution for Al-WTR). The 24 hr period was used for all sorbents, although some of the sorbents would re ach equilibrium sooner. At time intervals of 5, 30, 60, and 240 min., suspensions were removed from the shaker for sorption kinetics determination (reactions with time). Three replic ate each of the sorbents were used in the experiments. The total number of samples was 60 i.e. [5 sorbents x 3 replicate x 4 sampling times (t)]. The sorption experiment was conducted at room temperature (23 2 oC). The suspensions were equilibrated in a reciprocal shaker, at a frequency of 120 rpm (Eberbach Corporation, Michigan), centrifuged at 3200 X g and filtered through a 0.45 m membrane after reacting on a reciprocal shak er. Phosphorus remaining in solution was analyzed using inductively coupled plasma-op tical emission spectroscopy (ICP-OES, Perkin-Elmer Plasma 2100DV). Phosphorus retained was calculated as in (Eq. 2-1). So rbed P concentrations were plotted against time, and equilibrium concentrati on at respective initial concentrations used. 64

PAGE 65

Results and Discussion The amount of P sorbed vs equilibrium concentr ation is presented in Figure 5-1 using the sorbents of Al-WTR, MgO, LimeKD, Gypsum and Sl ag. The isotherm rises sharply in the initial stages for low equilibrium concentration (Ceq) and P sorbed for Slag, Al-WTR, and Gypsum. However, for LimeKD sorption was very fast, almost instantaneously at each Ceq. The instantaneous reaction might appa rently be due to pr ecipitation of P onto LimeKD. This rapid P sorption continues to the last Ceq reaction at 500 mg L-1 P initial P concentrations. Sorption by MgO was similar like LimeKD, except that at high er initial P concentrat ions of 300 and 500 mg L-1 P, MgO was deviated lower as compared to Li meKD on the curve. A true sorption optimum was not reached for Gypsum, Slag and Al-WTR, t hus, the sorbents showed potential for further adsorption. However, the graph generated (Gypsum, Slag and Al-WTR) showed a curvature trend leading to the maximum level, at wh ich no sorption can occu r. Perhaps additional increment of initial P to ~1000 mg L-1 P may exhaust the sorbents. Thus, due to fast P removal and large sorption capacities of some of the sorbents especially, LimeKD, and MgO, it was difficult to arrive at true optimal sorption capacity. Surface physisorption data suggest limited micropores for adsorption thus, implied that most of the mechanisms for the sorption processes are more likely that of precipita tion or chemisorption reactions. Kinetic study was used togeth er with the equilibrium data to indicate possible maximum sorption capacities of the sorbents (Figure 52). Estimated maximum sorption for MgO, Slag, Gypsum, LimeKD and Al-WTR were 96.1( 6. 6), 83.2( 4.7), 26.0( 7.4), 98.0( 1.5), and 90.7( 10)% respectively at initial concentration of 500 mg L-1 P. Aside from Gypsum, which showed a low sorption capacity for P, the rest of the sorbents indicated high sorption capacities. The high sorption observed may be due to precipitation reaction for MgO, Slag and LimeKD. In the case of Al-WTR, sorption might be attributed to hydrolysis of Al ions and eventually leading 65

PAGE 66

to precipitation reactions or chemisorption. Further lab shaking process normally involves constant agitation of the suspensions to achieve equilibrium reaction. This constant agitation may lead to abrasion via vigorous shak ing for rapid reaction to occur may not be the same as solid water interactions under natu ral environmental conditions. This may account for such high sorption capacities values observed. The constant agitation by shaking of the solid solution ratio may change some of the surface chemistry of the sorbents. However, needless to say that equilibrium sorption experiments have provided some benefits and understanding to reactions involving solid-liquid interactions. However, under natural conditions, such reactions may take a longer time than recorded in the study. Furthermore, as noted in Figure 5-1, several initial concentrations and time was involved and yet equilibrium was not fully reached for G ypsum and Al-WTR. Gypsums lack in reaching full equilibrium may be due to the sparingly soluble nature and Al-WTR may be due to continuous hydrolysis of the amorphous nature of the residual during shaking. Surface physisorption data suggest that th e sorbents used in this study are not true porous materials (i.e. materials with highly defined exchange sites/micr opores for sorption), hence, it is suggested that both equilibrium and kinetic data should be used t ogether to draw some useful inferences. This is because pseudo equilibrium appears to be observed in this study. Kinetic data (Figure 5-2) how ever, appears to suggest possible sorption optimum with respect to time, with some of the sorbents. Phosphorus uptake increased rapidly up to 5 min of adsorption and then proceeded more slowly through 240 min with mo st of the sorbents. However, LimeKD sorption was complete at each sampling time of P removal. Phosphorus reaction and uptake was instantaneous and complete in ~30 min. This material may be useful in removing P from areas loaded with high and exce ss P content. However in contrast, Gypsum is 66

PAGE 67

the least material to remove P. Maximum sorption was found to be 26%, about ~ 2.6 g kg-1. Sorption by MgO follows closely that of Lime KD. Al-WTR sorption af ter 60 min was found to be greater than that of Slag. Initial pH of 40, 60, 100, 300, and 500 mgL-1 P at the beginning of the experiment were acidic i.e. 5.25, 4.96, 4.87, 4.69, 4.54 respectively. No attempts were made to control pH during the equilibration experiment. This was done to show how individual sorbents affect the solu tion pH for P sorption. Final pH for each sorbents at the end of the equilibration experiments showed that the sorben ts had impact on the pH For instance,. Al-WTR after 40 mg L-1 P equilibration had pH of 6.0. MgO, Lime KD, Gypsum and Slag had pH of 10.6, 12.6, 6.3 and 11.2, respectively, after equilibration with 40 mg L-1 P solution. The importance of the kinetic experiment is th e application of the time the materials used in reaching the optimal P uptake. Information gathered from this experiment was utilized in the co-blending technique. This means that, for Li meKD, after about ~30 min, the material has almost completed the binding of P to form a solid phase. On the other hand, Gypsum may require a much longer time for the bi nding of P to form a solid phase. The important of the model is to predict the qe values as in measured values. However, this appeared not to be the case for the sorbents. A pparently, this might be due to the physical and chemical properties of the sorbents have not re mained constant during the sorption process. In addition, due to kinetic effects the amount adso rbed appeared to be greater for measured qe as in the case of Al-WTR. For example Gypsum in cont act with water solubilizes to release Ca and SO4. On the other hand sorbent such as boehmite, and calcined alunite that are not soluble during the sorption process have the qe measured equivalent to that predicted by the model. (Ozacar, 2003). This further support the fact that adsorption behaviour agre e well with the model but it is 67

PAGE 68

not the case with precipitation governing reactions which is the likely mechanisms for sorption with most of the sorbents in this study. Modeling of P Sorption Kinetics Models for describing sorption kinetics of P on sorbents are numerous. Equilibrium models (isotherms) of sorption have been widely applied for data approaching near equilibrium (Tang, et al., 1996, Siemens et al., 2004, Shin et al., 2004, Cheung et al., 2006). However, time dependence of the sorption process appears to be important. It helps in predicting the rate at which a particular pollutant is removed from an aqueous system, and to aid in designing treatment solutions. In addition, the kinetics of sorption is one of the fundamental studies necessary for better understand ing the mechanisms associated with sorption (Azizian, 2006; Sparks, 1989). Figure 5-1. Phosphorus sorption isotherm with initial P concentrations of 40, 60, 100, 300, and 500 mg L-1 on sorbents. Conditions: 600-900 m particle size of sorbents, pH of ~ 4.8, 24 hrs equilibration at 298K. Plotted valu es are the means of triplicate samples. Initial 1st and 2nd points for LimeKD and MgO coincided with each other and overlapped with Slag and Al-WTR points. 68

PAGE 69

0 20 40 60 80 100 120 050100150200250300 Time (min)P removed (%) MgO Slag Gypsum LimeKD Al-WTR Figure 5-2. Effect of reaction time on P sorption of sorbents at 500 mg L-1 initial P concentrations. Values are means of triplicate samples. Among the kinetics equations for describing P so rption are: first orde r (Griffin and Jurinak, 1974), second order (Griffin and Jurinak, 1974), pseudo-first-order (Makris, et al., 2004), pseudo-second-order (Makris, et al., 2004), diffusion (Cooke, 1966) modified Freundlich (Kuo and Lotse, 1975; Barrow and Shaw, 1975) and Elovich (Chien and Clayton, 1980). However, in order to investigate the mechan isms of sorption, and the characteristic rate constant of sorption, pseudo-first-order and ps eudo-second-order have proven to be quite successful for several sorbents (Ho and McKay, 1999b, Ho and McKay, 2003). Reaction rate orders such as pseudo-first a nd second-orders are important in expressing the relationship occurring between the rate of the reaction and the initial concentrations of the reactants. The order of the reaction can be determined from a simplified relationship: n oXCk dt dx )( (Eq. 5-1), where, n is the order of the reaction. Co is the initial concentr ation of the reactant, X is the 69

PAGE 70

concentration units that have reacted at time t and k is the rate constant (Duffey, 2000). Depending on the initial concentrations used for sorption, a linear or complex function could be observed in the model. For the pseudo-first-o rder model (Azizian a nd Yahyaei, 2006), a linear relationship is derived be tween the rate and the in itial concentration, i.e. k1= kaCo +kd, where k1 is pseudo-order-rate constant, ka, and kd are adsorption and desorption rate constant respectively, and Co is the initial concentration. Whereas for a pseudo-second order, the rate constant k2 and the initial concentration Co relationships have been found to be a complex one and not a linear function (Azizian, 2004). Reaction Order of the Model Pseudo-First-Order Model: The pseudo-first-order equation of Lagergren (1898) has been used for characterizing the sorption of a solid/ liquid system based, based on the solid carrying capacity. It has been used for a variety of solutes such as metals, organics and anions on different sorbents (Ho and McKay, 1999b). The diffe rential form is represented as: )(1 te tqqk dt dq (Eq. 5-2) Integrating eq. 5-2 with resp ect to boundary conditions of t = 0 to t = t and qt = 0 to qt = qt to a linear form: t k qqqe te303.2 )log()log(1 (Eq. 5-3), where, qe is the amount of solute sorbed at near equilibrium (mg g-1) and calculated as in (Eq. 21), qt is the amount of solute sorbed on the surf ace of the sorbent at any given time t (mg g-1), k1 is the rate constant for ps eudo first order sorption (min-1). A plot of log ( qe-qt) versus t is a straight line with slope ( k1/2.303) and intercept log ( qe). Equation 5-3 was applied to P sorbed on 70

PAGE 71

all sorbents (Al-WTR, LimeKD, Gy psum, MgO, and Slag). The resu lt is presented as indicated in Table 5-1. The r2 of the linear plots showed correlations were poor (Table 5-1). The sorbents MgO, Slag, Gypsum, and LimeKD had r2 = 0.27, 0.41, 0.05, 0.14 respectively. The exception was with Al-WTR which had an r2 = 0.81. The Al-WTR results was in close agreement with data from Makris et al., 2005, showing WTR from Bradenton to be (r2 = 0.84). To ascertain the sufficiency of the pseudo-first-order model, it was evaluated by a residual plot as in dicated in Figure 5-4. Azizian and Yahyaei (2006) have shown that th e initial concentration of the solute used influences the sorption. Thus, r2 may not a reliable index to draw conclusions on the order of the reactions. The data was further plotted to a pseudo-second-order equa tion for verification. Pseudo-Second-Order Model: This model was first used by Blanchard et al., 1984 to characterize the removal of heavy metals onto zeo lites. Since then the equation has been applied to sorption kinetics from liquid solutions in a linear form by (Ho and McKay, 1999a, Ho and McKay, 1999b, Ho and McKay, 2000, Ho et al., 2001, Ho and McKay, 2002, Ho et al., 2006). In addition, other researchers have also found the equation useful in predicting anions onto sorbents (Azizian and Yahyaei, 2006, Makr is et al., 2005, Janos and Smidova, 2005). The rate of pseudo-second-order reaction ma y be dependent on the amount of solute sorbed on to the sorbent, and the amount sorbed at near equilibrium. The sorption equilibrium, qe is a function of the nature of solute-sorbent interaction. The rate law is described as: 2 2)(teqqk dt dq (Eq. 5-4) Integrating eq. 5-4 with the boundary conditions t = 0 to t and qt = 0 to qt = qt, gives a linear format as: 71

PAGE 72

t q qk q te e t112 2 (Eq. 5-5), where, k2 is the pseudo-second-order rate constant for sorption, qe is the amount of solute sorbed per gram of sorbent at near equilibrium, qt is the amount of solute sorbed on the surface of the sorbent at any time t (mg g-1). A plot of t/qt versus t gives a straight line with slope of 1/qe and intercept of 1/k2q2 e. Sorption rate constant k2 is evaluated from the slope and intercept respectively (Azizian, 2004, Ho and McKay, 2003) Rudzinski et al., (2006), found that the pseudo-second order-equation is appropriate for surfaces that are energetically heterogeneous such as Al-WTR, Slag, MgO, LimeKD and Gyps um. Data fitted to the pseudo-second-order model is presented in Table 5-2. The results show good correlations for Al-WTR, Slag, MgO, LimeKD and Gypsum with r2 to be 0.981, 0.996, 0.994, 1.00, and 0.999, respectively. The parameters of k2 and q2 e determined from the slope and the intercept are presented in Table 5-2. Based on a greater fit as compared to the pseudo-first-order equation it suggests that adsorption of PO4 to the sorbents can be described with a pseudo-second-order model. An exception, however, is that of LimeKD whose r eaction time was quite rapid, making it difficult to quantify the order of the reaction. Model Evaluation Correlations were poor for pseudo-first-order model, and good correlations for pseudosecond-order, suggesting acceptance of the second order model. However, according to (Azizian and Yahyaei 2006), the initial con centration of solute i.e. PO4 used in the sorption process had significant influence on the nature of model fittings. Thus, r2 values are not the best criteria to depend on. The model was evaluated using plots of residuals. Graphical plots in which the residuals do not exhibit any systematic struct ure indicate the model fits the data well (Montgomery, 2005). 72

PAGE 73

On the other hand, plots in which residual show systematic structure indicates that the form of the model may be improved in some ma nner. Residual plot of pseudo-first-order is presented in Figure 5-4. The resu lts show non-random structure (i .e. points show a significant trend or pattern), thus supporting the poor correlation of the model as seen in r2 values of pseudofirst-order. In addition, the ps eudo-second-order residual plots showed random residual plots (Figure 5-5). The combination of the residual plot and r2 values therefore s uggest that P sorption to the sorbent follows that of pseudo-second-order function. -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0 100 200 300 Time (min)Residuals error MgO Slag Gypsum LimeKD Al-WTR Figure 5-4. Residuals plot of pseudo-first-order P sorpti on on Al-WTR, MgO, Slag, Gypsum, and LimeKD showing a trend or pattern of points for each sorbents. 73

PAGE 74

Figure 5-3. Pseudo-second-order fitting for P so rption to sorbents of Al-WTR, MgO, Slag, Gypsum, and LimeKD. Figure 5-5. Residuals plot of pseudo-second-order P sorption on Al-WTR, MgO, Slag, Gypsum, and LimeKD showing scatter points. 74

PAGE 75

Table 5-1. Kinetic parameters of P sorption on sorbents using pseudo-first-order model Pseudo-first-order parameters Sorbents ak1 (min-1) b qe (mg g-1) R2 Slag 0.0029 1.78 0.408 Gypsum 0.0007 1.23 0.055 LimeKD 0.0025 2.94 0.141 MgO 0.0035 2.41 0.274 Al-WTR 0.0053 1.29 0.805 a The rate constant for pseudo-first-order. b The amount of P sorbed onto sorbents at near equilibrium. Table 5-2. Kinetic parameters of P sorption on sorbents using pseudo-second-order model Pseudo-second-order parameters Sorbents ck2 (g(mg min)-1) d q2 e (mg g-1) R2 Slag 0.1856 0.83 0.995 Gypsum 0.0877 3.68 0.999 LimeKD f N/A N/A 1.000 MgO 0.5109 0.19 0.990 Al-WTR 0.1545 0.45 0.980 c The rate constant for pseudo-second-order. d The amount of P sorbed onto sorbents at near equilibrium. f Not available. P Density on Sorbent The amount of P sorbed or covered on the surf ace of sorbents is expressed as P density (moles m-2) and is plotted as a functi on of the near equilibrium c oncentration. Surface area data used is discussed in (Chapter 3, Table 3-1). Th e result is shown in Figure 5-6. Mean P moles m-2 i.e. surface coverage or dens ity for the Al-WTR was ~2.05x 10-6, Lime KD ~1.65x 10-6, Gypsum ~4.4x10-6, Slag was ~1.13x10-5, and MgO ~3.16x10-6. In addition, the average amount of P 75

PAGE 76

remaining in solution after equilibration with the sorbents Al-WTR, LimeKD, Gypsum, Slag, and MgO were 200, 10, 450, 150, 100 mg L-1 P, respectively. Gypsum data showed equilibrated P remain ing in solution to be greater than in comparison to the rest of the sorbents. Greater P (moles m-2) may be due to the saturation of single monolayer as observed under a Langmuir mode l. Further, Gypsum is sparingly soluble hence has less surface area for reac tivity. On the other hand, Lime KD had low P density and less amounts of P remaining in solution. Further, Al -WTR has more P remaining in solution in comparison to LimeKD, MgO and Slag. In additi on, kinetics of P coverage on the surface is presented in Figure 5-7. Result showed PO4 coverage on the surface increase rapidly up to ~ 5 min, and then slowly decrease to 240 min in all so rbents used. This therefore suggests that, the area for P reactivity decreases w ith time. This may support the f act that the likely mechanisms may be precipitation/chemisorption, since the su rface to react has been covered with products resulting from precipitation. LimeKD a nd Al-WTR had maximum coverage (1.6x10-6, 3.0x10-6 moles m-2) of, respectively. On the other hand, Gypsum, Slag and MgO had maximum coverage of (5.0x10-5, 2.0x105, 6.0x10-6 moles m-2, respectively), showing much P accumulation on the surface than Lime and Al-WTR. Mean average of MgO, Sla g, Gypsum, LimeKD and Al-WTR are 3.15x10-6, 1.1x10-5, 4.4x10-5, 1.6x10-6, 2.05x10-6 (moles m-2) of P, respectively. Other researchers had mean P adsorption density of P on goethite to be ~ 2.6x10-6 (moles m-2) of P after 24 h sorption (Ippolito et al., 2003), which is consistent to value of ~ 2.05x10-6 found for Al-WTR. Al-WTR is amorphous in nature and behaves like hydr(oxides). Experimental data from the peer-reviewed data for other sorbents appears to be non-existence, for compar ison. It is, however, unclear why Slag and Gypsum (1.1x10-5, 4.4x10-5, moles m-2, respectively) appears to have greater P density. 76

PAGE 77

It may be due to different mechanisms other than adsorption. Presumably, the mechanism may be precipitation leading to new solid phase form ation due to an absence of desired micropores for adsorption. Thus suggesting that, P is reacting on the surface of the sorb ent leading to greater P density observed on such sorbents. Figure 5-6. Phosphorus density on sorbents (AlWTR, Slag, MgO, Gypsum, LimeKD and MgO) as a function of equilibrium concentration (Ceq) at initial P c oncentration of 500 mg L-1. 77

PAGE 78

Figure 5-7. Kinetics of P surface coverage on so rbents of Al-WTR, MgO, Slag, Gypsum and LimeKD at initial P c oncentration of 500 mg L-1. Adsorption Isotherm Adsorption by definition is the process by wh ich molecules from a solution bind in a condensed layer on a solid surface (Masel, 1996). By way of quantifying or modeling the process, an isotherm plot is usually used. Se veral models are availabl e. However, a commonly and extensively used model for describing sorp tion behavior is the Langmuir model (Azizian, 2004; Bolster and Hornberger, 2007; Wang and Harrell, 2005; Kleinman and Sharpley, 2002). Langmuir and Freundlich Adsorption Model Langmuir 1918, created a model to describe reversible monolayer adsorption where adsorbing molecules reversibly attach themselves to the surface of the solid. The equation is as follows: o e o e eQ C bQq C 1 (Eq. 5-6), where, Ce is the equilibrium concentration (mg L-1) in solution, qe the amount of P sorbed at 78

PAGE 79

equilibrium (mg g-1), Qo denotes Langmuir monolayer capacity (mg g-1), and b represents the Langmuir bonding term related to interaction energies (dm3 mg-1). The values of Qo, and b are determined form the slopes and intercep ts respectively for the sorbents. The Freundlich equation on the other hand, wa s developed for fitting data on rough or heterogeneous (multi-site) surfaces (Masel, 1996). It is represented in a linear format as: e eC N Kfq log 1 loglog (Eq. 5-7), where, qe is the amount of P adsorbed per mass of the sorbent (mg g-1) and Ce is the equilibrium concentration (mg L-1), Kf and N are adjustable parameters that measures adsorption capacity and N adsorption intensity resp ectively. A plot of log qe vs. log Ce is a straight line with slope and intercept as 1 /N and Kf respectively (Essington, 2004). Table 5-3 shows comparative values of Langmuir and Freundlich constants, for the test ed sorbent treatments, and their corresponding regression coefficients. A Langmuir model is based on the assumption that maximum adsorption corresponds to a saturated monolayer of adsorbate molecule on the adsorbent surface. Secondly, the Langmuir model is based on kinetic theory, and at the sa me time assumes equilibrium concentrations and conditions. From the comparative Ta ble 5-3, it is clear from the r2 values that the Freundlich model predicts best characterizes the adsorpti on of P on sorbents than Langmuir, except for Gypsum. Gypsum had r2 = 0.9974 for the Langmuir fitting, which implies that the material was homogeneous and likely had monolayer coverage The homogeneity is confirmed using the Freundlich constant N. Accord ing to Essington (2004), N is a measure of heterogeneity of adsorption sites of the sorbent. As N approaches zero, surface site heterogeneity increases, indicating there is a broad dist ribution of adsorption sites t ypes. On the other hand, as N approaches unity, surface site homogeneity incr eases, thus implying that there is a narrow 79

PAGE 80

distribution of adsorption site types. N of Gypsum was 1.0 showing clearly the surface homogeneity. This likely explains why Langmuir model fits better expl ained Gypsum sorption (assuming surface homogeneity and single monolayer coverage of the material). Conversely, the N values for Al-WTR, MgO, and Slag, were closer to zero (Table 5-3). Thus, indicating surface heterogeneity of the sorbents. Their corresponding values of r2 values for Freundlich were greater than for Langmuir; hence the model is well pr esented by Freundlich. Harter and Smith, (1981) suggested the lim itation of universally depending on the Langmuir model. This was explained based on the fact that Langmuir model was developed for simple-physico-chemical retention of solutes onto surface. However, other reactions such as ion exchange, precipitation-crystallization, and structural substitution may also be occurring other than mere retention of solute on the surface. In addition, a recent paper from Cucarella a nd Renman, (2009) has observed significant and systematic differences in sorption capacities of similar sorbents by different authors. The differences have been attributed to the origin of the sorbents, the particle size, chemical composition, initial P concentration, material-tosolution ratio and pH. The authors lamented that there is no single standard procedure governing batch experi ments. Thus results from batch experiments shows gross discrepancies. These di fferences render it impossible to compare the capacity of materials and to quest ion the validity of results. The authors suggest that procedures should be standardized for materi als within the same particle si ze range, using similar materialtosolution ratios and applied P concentrations, contact times that allow equilibration to be reached, and use of proper agitation to ensure mixing but not aggregate breakage of the sorbents. Furthermore, work from Del Bubba et al., (2003) using Langmuir isotherm accurately described the P sorption by sands, if no precipitation reactions were taking place. Other authors 80

PAGE 81

have observed the effects of precipitations on th e shape and applicability of the commonly used isotherms (Zhou and Li, 2001; Svik and Klve, 2005). Thus the fit of experimental data to a Langmuir (or other) adsorption isotherm does not always adequately predicts sorption responses in the natural environment. Most commonly, ad sorption to a surface is followed by additional interactions or reactions at the surface or within the matrix of the sorbents. It is against this background, as shown in Table 5-3, using both Langmuir and Freundlich is suggested. By using the two models together, parameters may complement each other, in revealing features/meanings of parameters, wh ich, may be unclear from a single isotherm. Table 5-3. Langmuir and Freundlich c onstants for P sorption on sorbents Langmuir parameters Freundlich parameters R2 Sorbents aQo (mgg-1) b b(dm3mg-1) cKf (mgg-1) dN Langmuir Freundlich Al-WTR 0.1729 -0.0034 9.554 0.2967 0.9094 0.9437 MgO 0.0405 -0.0025 23.25 0.1122 0.8283 0.9445 Slag 0.2312 -0.0032 9.268 0.2804 0.9782 0.9898 LimeKD eNA NA NA NA NA NA Gypsum 3.3378 -0.0723 0.9891 1.0 0.9974 0.9683 a Langmuir constant of monolayer capacity. b Langmuir constant related to energy. c Freundlich distribution parameter rela ted to sorption capacity. d is a constant value between 1 and 0. eData not available. Conclusions Sorbents were reacted with P-containing soluti ons at different initial P concentrations. A pseudo equilibrium sorption optimum was obtained for the sorbents. The results suggest the sorbents have sufficient sorption capacities to immobilize P. The mechanism for sorption appears to be that of precipitation/chemisorption for solid/liquid reactions. Sorption kinetics of the sorbents showed LimeKD having rapid sorpti on followed by MgO, Slag, Al-WTR and Gypsum. 81

PAGE 82

The sorbents appeared to reach equilibrium after one hour sorption, except for Al-WTR and Gypsum. Pseudo-second order kinetics best fit th e reaction order for the sorbents, except for LimeKD. LimeKD reaction order was difficult to very rapid sorption process. The pseudo second order kinetics model was validated using residual error plots, wh ich showed scattered points indicating that there are no interacti ng factors controlling the error variable. Phosphorus coverage on the surface of the sorb ents suggests the sorbents can accumulate P. Al-WTR had ~2.05x10-6 moles m-2 which was consistent with other values reported in the literature. However, no data existed for the othe r tested sorbents for comparison. In addition, an amount of P remaining in solution after equili brating initial P con centration of 500 mg L-1 for the sorbents is: Al-WTR, LimeKD, Gypsum Slag, and MgO, ~200, 10, 450, 150, 100 mg L-1 P, respectively, with respect to P cove rage on the surface of the sorbents. Langmuir and Freundlich models were used to fit the sorbents. Freundlich model best fitted the sorbents sorption process with the exce ption is that of Gypsum. It had a Langmuir model prediction of r2 = 0.9974, compared with the Freundlich model, r2 = 0.9683. 82

PAGE 83

CHAPTER 6 CONCEPTUAL FRAMEWORK OF CO-BLENDING: EFFECTS OF CO-BLENDING ON LEACHED SOIL AND SOIL SURFACE MORPHOLOGY Introduction Previous chapters (1 and 4) have addressed the types of P forms and species existing in manure-impacted soils. The nature of the sorb ents examined through P incubation studies, suggests that each sorbent sequestered P relative to the soil system pH. In an attempt to retain greater P, sorbent co-blending wa s considered as a nove l idea to capture all the P forms and species. This is because, under natural condition, an ideal sorbent should effectively immobilize P as pH fluctuates from 6 to 8.5. The ideal sorben t is difficult to obtain, but may be engineered. Further, pH fluctuation is a common problem in manure-impacted soils, due to large proportions of loosely bound P fractions associated with Ca -Mg as shown in chapter 2. The fluctuation of soil pH leads to rel ease of P, causing pollution to water bodies. The basic principle of co-blending is to capture all P species with re spect to a sorbent and in relation to the species existing at that particular pH. Phosphorus species are functions of pH. A single sorbent works within a given pH range an d has the limitation as the ambient pH condition changes. Thus, there is the need to tailor the available sorbent to immobilize P, as soil pH condition changes. It is proposed that the co-ble nding technique will create the enabling sorbents and the environment necessary to immobilize P, as soil pH environment fluctuates. As noted in Figure 6-1, the initial pH may be high; P species existing at that situation may be captured by the sorbents containing Ca-Mg suitable for that en vironment. Further, as the pH eventually decreases, which is typical for the tested soils under constant dynamic environmental conditions, Al-Fe based sorbents would be activated to sorb that P species. Thus, there is a narrow range for P release, which may be of important to plants if that P is of significance use. The importance of this approach is that, the P is no more in excess to cause significant environmental damage. 83

PAGE 84

Furthermore, the transport of P species with re spect to surface chemistries of the sorbent is illustrated in Figure 6-2. The schematic notation a ssumed the sorbent to be a porous medium for sorption to occur. P species may either adsorbed on the interface of the sorbent surfaces or precipitated depending on the actual mechanism occurring. Many types of PO4 minerals may be formed through co-blending with the so il amendments (Eq. 6-1 to 6-11). According to McBride (1994), depending on the availability of coor dinated position of OH groups and H2O molecules on the surfaces of hydr(oxides) of Fe-Al, Mn and layer silicates clays, chemisorption may occur, leading to an inner or outer sphere complex formation as indicated in Figure 6-2. It is therefore, postulated that a co mbination of ligand exchange (formation of inner-sphere complexes), and surface adsorption, together with surface precipitation by abundant Fe, Al, Ca and Mg compounds may provide the likely PO4 removal mechanisms (Figure 6-2). However, the actual mechanism is that may govern co-blending is unknown, although chemisorption may be an important process. The hypothesis of the study therefore is: coblending Al-Fe based materials and that of Ca-M g materials may lead to rapid P immobilization. The objective of this chapter is: 1. To determine, the effects of co-blending various sorbents on soil P leaching studies. 2. To determine, the e ffects of the co-blending products on soil surface morphology using scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDS). 84

PAGE 85

Figure 6-1. A framework of complexation reactions due to co-blend ing of sorbents for rapid P immobilization. Figure 6-2. The conceptual transport of P species w ith respect to surface chemistry of sorbents as influenced by speciation and pH. 85

PAGE 86

Reactions Equations for P Minerals Formation Many different PO4 minerals may be formed through co-blending techniques with the soil amendments. For example, the formation of a PO4 mineral from reactions of MgO and soluble PO4 in a soil-aqueous system can be described by (Eq.6-1 and Eq. 6-2). MgO + H2O Mg(OH)2 (Eq. 6-1) 3Mg(OH)2 + 2H3PO4 Mg3((PO4)4)2 + 6H2O (Eq. 6-2) If ammonium is present in the soil syst em, a reaction between MgO and di-hydrogen ammonium PO4 can occur, leading to the production of other insoluble PO4, such as magnesium ammonium phosphate hexahydrate (NH4MgPO4 .6H2O). This reaction can be described by equation 3. MgO + NH4H2PO4 + 5H2O NH4MgPO4 .6H2O (Eq. 6-3) Additionally, the formation of another PO4 mineral from reactions of calcium-based compounds and soluble PO4 in a soil-aqueous system can be described by (Eq. 6-4 through 6-7). CaO + H2O Ca(OH)2 Ca2+ + 2OH(Eq. 6-4) H2PO4 2H+ + PO4 3(Eq. 6-5) HPO4 2H+ + PO4 3(Eq. 6-6) 3Ca2+ + 2PO4 3Ca3(PO4)2 (Eq. 6-7) If soluble organic carbon compounds are present in the soil system, solid hydroxyapatite (Ca10(PO4)6(OH)2) may be formed, as described by (Eq. 6-8 through 6-11). 2R-COOH 2R-COO+ 2H+ (Eq. 6-8) Ca2+ + 2R-COOR-COO-Ca-OOC-R (Eq. 6-9) R-COO-Ca-OOC-R + Ca3(PO4)2 + H2O Ca4(PO4)2O + 2R-COOH (Eq. 6-10) Ca4(PO4)2O +2Ca3(PO4)2 + H2O Ca10(PO4)6(OH)2 (Eq. 6-11) 86

PAGE 87

In (Eq. 6-8 through 6-11), R- can be any suitable organic group, including, but not limited to, an alkyl group, an aryl group, an amino group, an amido group, a cyano group, an ester group, or a hydrogen atom. Also, while the PO4 mineral formations described in (Eq. 6-1 through 6-11) are provided for illust rative purposes, many other PO4 minerals may be formed by the soil amendments. Materials and Methods Experimental Design/Set up The soil (Immokalee fine sand-sandy, silice ous, hyperthermic Arenic Haplaquod) was collected from a dairy farm (Butler Oak), near Lake Okeechobee as described in chapter 2 and 4. The dry soils were wetted to moisture content between 10-15% by weight. The pH and Eh (redox potential) of the soil samples was me asured with a pH meter (Accument (R) x L60) on 1:2 v/v bases. Sorbents were amended at the rates of 0, (10+10) g kg-1 and 20 g kg-1, representing, 0, 1+1% and 2% by mass to the weight of the soil used, respectively. Sorbents treatments consisted of 5[Al-WTR, Slag, MgO, Gypsum, LimeKD at 2% rate]; 4 co-blended samples [MgO, Slag, MgO, Gypsum, LimeKD to Al-WTR]; 1 control (no sorbent), were replicated three times. [Total number of samples = 30]. Soils were packed in a revolving container (vol. ~150 cm3) that allowed the soil to mix well, and to allow for air exchange if necessary. Before the application of the first amendments, soils were put under anaerobi c condition to enable the release of P fixed to any Fe to desorb for seven days. This is b ecause under reducing condition (anoxic), and pH > 7, it has been observed that greater P may be rel eased (Patrick and Khalid, 1974; Ortuno et al., 2000). The soil was allowed to dry by allowing air to the container and the moisture content was reduced to that of air-dry soil. The sorbent cont aining Al, i.e. Al-WTR wa s applied, and then put through another anaerobic cycle. After the reduc ing process (anoxic condition), a second sorbent 87

PAGE 88

containing Ca-Mg was applied while maintaining the moisture content of 10-15%. The anaerobic cycle (wetting and reducing conditi on) was repeated. The soil was air dried and then loaded into column and leached over a 12-week period. Le ached P was measured up through week 5. Since no more P had leached over by that time, additional times were not analyzed for P. After 12 weeks, the leached soil was prepared for SEM-EDS analysis. The co-blended and unamended samples were examined under SEM-EDS for any structured surface morphology due to the effects of co-blending afte r the leaching process. Column Leaching Study Columns were made from PVC pipe (5cm inne r diameter x 20 cm height). Each column was equipped with a 2 cm drainage hole at the bottom. Screens of wire gauze were used to glue the holes at the bottom at each column to prevent soil loss. The columns were hanged on a wooden support customed made to hold the column in a vertical position. Prior to placing the soil in the columns, it was co-blended with the so rbents as described above, i.e. (Al-WTR, coblended to Slag, MgO, Gypsum, LimeKD) at ra tes of (1+1 % each) re spectively. The columns were packed with ~ 300 g of amended soil. All treatments were replicat ed three times in a completely randomized design. One hundred milliliters (mL) of DDI water (adjusted to pH 5 by using 6M HCl, to mimic the pH of rainfall in So uth Florida) were added to each column weekly. Each leachate volume corresponded to ~ 1 pore volum e. Leachates were analyzed for soluble P for over five weeks with ICP-OE S. Leachate pH, Eh (redox potential) was also analyzed. Soils used in leaching studies were air-dried and P was sequential extracted, as per Chang et al., 1983 (Chapter 4). Statistical analyses were performed using th e Proc GLM and REG model in SAS software version 9.1.3 (SAS Institute, 2002 ). Differences within each P fraction were examined using ANOVA at = 0.05, with mean separation determined using Tukeys Studentized Range Test. 88

PAGE 89

Significant differences in cumulative P mass was determined using Fishers Protected LSD procedure at = 0.05. Analytical and instrument quality assurance and quality control (QA/QC) was evaluated for all lab analyses by including 5% repeats, 5% spikes, and blanks. Standard calibration curves, as well as a quality control ch eck standard were prepared for each procedure. Background signal drift was consistently < 1% for al l instruments. Spike recovery all fell within an acceptable range of 90-110%. Scanning Electron Microscopy/Energy Dispersive Spectroscopy Soils used after the leaching study were ai r-dried, sieved < 600m and mounted on Al SEM stubs with doubled-sided con ductive carbon tabs (SPI Supplies, West Chester, PA), carbon coated (Ion Equipment, Santa Clara, CA), and stored in a desiccator until analyzed. Samples were imaged using JEOL JSM-6335F SEM (JEO L USA, Peabody, MA) operated at a 5keV for imaging; EDS analysis was performed w ith a JEOL JSM-6400 (JEOL USA, Peabody, MA) operated at 15keV. Back-scattering and secondary electron images were also acquired using scanning electron microscope. The SEM procedure consisted of dividing the sample on the SE M stub into four quadrants. Three parts of the quadrants were analyzed by EDS for elemental composition analysis and images were taken at locations showing specific surface structures different from the normal soil. Scatter plots were obtained for three EDS analyses taken within the quadrants with r2 close to ~ 1.0 and SD < 10% indicating samp le variances were within the acceptable range (data not shown). The locatio n with surface structures (sur face complexation) were also analyzed for elemental composition with EDS (Prochnow et al. 2001). Morphological comparison where made with the original soil without co-blendi ng to co-blended samples. This 89

PAGE 90

type of examination was meant to discriminate or recognize morphological features that where not present at the star t of the experiment. Results and Discussion Chemical Analysis of Leachates Leachates were analyzed for P for 5 weeks. Cations of Ca, Mg, Na, and K were also analyzed (data not shown). During leaching, the initial slightly neutra l pH became alkaline. Original soil samples without co-blending had pH of ~7.1 and Eh of (+100 mV). The pH of the leachate samples without any sorbents amendment fluctuated, ranging from 7.7 to 8.26 over the first 5 weeks of leaching (Table A-1). Further, the Eh (redox potential) of the leachate showed clearly, the soil were anoxic due to co-blending (Table A-2). All soil, sorbent unamended and co-blended samples had Eh values below the reported critical redox potential for the reduction of Fe (Eh = +300 to mV at pH 6-8; Gotoh and Patrick, 1974). Thus, suggesting that reduction of Fe3+ to Fe2+ should have taken place. This further support that P sequestered to Fe3+ might have been released, and recaptured with the coblending products which aimed by first releasing and then tying the P. Leachate pH for soil amended with Al-WTR (20 g kg-1) ranged from 7.7 to 8.3 ( 0.17). The fluctuating pH environment tended to make Al-WTR to preferentially sorbed P species associated to Fe-Al (Chapter 4, Figure 4-3). Thus, maybe accounting for the longtime it takes AlWTR to completely sorb P (Agyi n-Birikorang, et al., 2007 ). Leachates pH of soil amended with MgO + Al-WTR, 10 g kg-1 each ranged from 9.0 to 9.2 ( 0.14). Further, the pH of leachate from of soil amended with 10 g kg-1 each of Al-WTR + Slag ranged from 8.0 to 8.4 ( 0.13). The pH of the sorbents amended with MgO and Slag alone were alkaline, pH of 11.5 and 10.9 respectively. However, leachate data suggested th at the amount of sorbent incorporated could 90

PAGE 91

reduce the pH to a workable pH range while sorb ing P due to the buffering capacity of the soil (Table A-1). One of the reasons for using various sorbents is to allow options for selecting a particular sorbent to fit a speci fic purpose. For example, data for Gypsum co-blended with AlWTR (1+1%), had excellent pH, ~7.5 far below that of the control soil (~8.0). Such a pH would be conducive for pH sensitive plants requiring su ch a neutral pH. However in the case of Gypsum, the binding of P was more gradual as comp ared to other sorbents (Table A-1 and Table A-3). A cumulative soluble P mass (mg) showed the effects of co-blending of the sorbents for comparison. To avoid overlapping data points on each other, representative sorbents (control soil, Al-WTR, Al-WTR+MgO and Al-WTR+Slag) were presented for simplicity (Figure 6-3). The remainder of the data can be found in (Table A-3). All plots had a re gression coefficients r2 >0.9. The plots suggested a linear reduction of 97% of slope from 6.3 (control) to 0.2 as sorbents were co-blended and was significant. The regressi on equation for the control (without sorbents amendments) was f(x) = 6.28x-3.28, r2 = 0.99, ( P < 0.0001 ). The equation also showed a reduction of 88% of slope from 1.73 (Al-WTR) to 0.2 when soil was amended with half of AlWTR to 10 g kg-1 Slag and MgO each. The regression equa tion for Al-WTR amended at 20 g kg1 rate was f(x) = 1.731x-1.04 r2 = 0.99, (P < 0.0001), while the equations for soils amended with Al-WTR+ Slag, MgO + Al-WTR, 10 g kg-1 each were f(x) = 0.283-0.187 r2 = 0.98, (P < 0.001), and f(x) = 0.203x+0.045, r2 = 0.93 (P < 0.007) respectively. The findings showed that co-blending Al-WTR with Ca-Mg based sorbents, significantly reduced leachate P and that immobilization was rapid. In further support, (Bayley, et al., 2008; and Ippolito et al., 1999) have observed reduction of P when Al-WTR was co-applied with biosolids containing some form of Ca. Thus, co-blending has been suggested as the sharing of 91

PAGE 92

different P species from available P pools for immobilization with respective to sorbent amendment. The application of Al residual was also reduced c onsiderably in the study, which may allay any fears of aluminiu m toxicity in using Al-WTR. igure 6-3. Effects of co-blending 10 g kg Al-WTR with 10 g kg each of slag and MgO respectively on soluble P from leachates. Errors bars indicate SE of triplicate samples. Effects ential Extractions t hod of Chang et al., (1983) 0 10 20 3012345Leaching events Al-WTR WTR+MgO WTR+Slag Soil (Control) Soluble P Cumulative Mass (mg) b a c c F-1 -1Means sharing the same letter ar e not significantly different at = 0.05, as determined by Fishers Protected LSD. of Co-blending on Leac hed Soils/Sequ Soils in columns used for leaching studies we re air-dried. The me was used for the extraction with modifica tion, and as described in chapter 4. Results of the analysis are presented in Figure 6-4 to Figur e 6-6. The results show the effects of sorbents amendments on labile P. That is, there is a diffe rential reduction of solubl e P with respect to the sorbents. The effects of the sorbents are therefor e consistent with those obtained for incubation study in chapter 4. 92

PAGE 93

Comparing co-blended samples (1+1%) to 2% single sorbent application, there is a significant difference among the diffe rent treatments for labile soil P. Among the (1+1%), there was no significant difference among co-blend ed (Soil+Al-WTR+LimeKD), (Soil+AlWTR+Slag), (Soil+Al-WTR+MgO) with one excep tion of (Soil+Al-WTR+Gypsum). This thus suggested that any sorbents can be co-blended to gether for P removal. Gypsum co-blended with Al-WTR, however had large pool of labile P afte r reaction. The difference may be due to kinetics effects, as Gypsum has to release calciu m for any reaction to occur (Figure 6-4). Comparing between the 2% sorbents amendmen ts, it suggests that using the sorbents will remove soil P. However, as discussed previously, a single sorbent may have the slow P sorption in some sorbents reaction as well as impacting so il pH. This thus suggests that the use of coblending technique material may offer a great er advantage to only one use of a sorbent application. Figure 6-5, showed significant P differenc es between P that is bound to Fe-Al. Observation among the 2% application rates showed clearly, that Al-WTR removed greatly P bound to Fe-Al. Those of LimeKD, Slag, Gypsum and MgO lag behind that of Al-WTR at 2% and the result is significant. Thus, a second observation from soils us ed in the leachate experiments again confirmed that Al-Fe materials serve as a sink for P associated to Fe-Al. In addition, this observation is thus consistent with a recent paper (Malecki-Brown and White, 2009), which showed clearly that P bound to Al Fe were the dominant fractions in the Al treated samples. Further, additi on of Al did lead to an increas e in Al-bound P pools. This thus showed that P fractions removal is subject to basic cations available for P sequestration. Consequently, supporting the use of co-blendi ng technique to immobilize all P forms as discussed in previous chapters (2 and 4). 93

PAGE 94

Phosphorus associated to Ca-Mg was examined in Figure 6-6. The result is consistent with previous data in Chapter 4. The use of Ca-Mg based materials addition leads to sequestration of P bound to Ca-Mg. Thus there is a differential sorption of P with respect to metal cations for remediation strategy. The Ca -Mg materials removed a greater amount of P significantly, as compared to that of Al-based materials (Figure 6-6). Therefore, data from the sequential extraction procedure after leaching studies further add impetus to the novel idea of coblending of using both Al-Fe and Ca-Mg materials to immobilized P. Results from sequential extraction of soil after leaching study provided useful evidence to the use of co-blended technique in P immobilization. Soil + amendment s 020406080100120140160180 Control Soil Soil +MgO2% Soil +Slag2% Soil +Gypsum2% Soil +LimeKD2% Soil +Al-WTR2% Soil +WTR+MgO(1+1%) Soil +WTR+Slag(1+1%) Soil +WTR+Gypsum(1+1%) Soil +WTR+LimeKD(1+1%) mg P kg-1b f c d f g f e a f Figure 6-4. Labile P. Means (n = 3) of the sa me letters are not significantly different at = 0.05, determined by LSD. 94

PAGE 95

0100200300400500600700 Control Soil Soil +MgO2% Soil +Slag2% Soil +Gypsum2% Soil +LimeKD2% Soil +Al-WTR2% Soil +WTR+MgO(1+1%) Soil +WTR+Slag(1+1%) Soil +WTR+Gypsum(1+1%) Soil +WTR+LimeKD(1+1%)Soil+ amendment s mg P kg-1d b c d a f d e d b Figure 6-5. Iron-Al bound P. Mean (n = 3) of the same letters are not sign ificantly different at = 0.05, determined by LSD. 95

PAGE 96

020406080100120 Control Soil Soil +MgO2% Soil +Slag2% Soil +Gypsum2% Soil +LimeKD2% Soil +Al-WTR2% Soil +WTR+MgO(1+1%) Soil +WTR+Slag(1+1%) Soil +WTR+Gypsum(1+1%) Soil +WTR+LimeKD(1+1%)Soil+amendment s mg P kg-1f d bac d ba ef dc bc ed a Figure 6-6. Calcium-Mg associated P. Means (n = 3) of the same letters are not significantly different at = 0.05 determined by LSD. Effects of Co-blending on Soil Morphology (SEM-EDS) To evaluate the effects of co-blending on soil morphology, scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDS) was carried out. One great advantage of SEMEDS is that is a non-destructive approach, as well as examination of morp hological features and the elemental composition in-situ. Several researchers have used SEM for soil characterizations as well as in forensic identifications and classi fications of materials (Z adora and Brozek-Mucha, 2003, Pye and Croft, 2007). Analysis from SEM-EDS could be qualitative as well as quantitative depending on the approach use to evaluate results. SEM has been used by (Haapala, 1998), Steven s et al., 1993 for quantitative analysis of distribution of air pollutants in a surrounding location. Effects of calcium distribution on pine 96

PAGE 97

trees were investigated. SEM-EDS was used to identify calcium as well quantify the amount of calcium distribution across loca tions and with distance. Zadora and Brozek-Mucha (2003) illustrated th e use of SEM-EDS for quantitative forensic analysis of gunshot, glass and car paint residue s when a reference sample was compared for forensic analysis. The identific ation of residual samples matche d exactly the morphological as well as elemental content to reference samples. They concluded on the use of SEM-EDS for such studies samples spread in the soil. Soil weathering has been inve stigated by (Ewing and Nater, 2003). The investigators combine palaeoecological study as well as grain mi cromorphological approach of soil science by using SEM to reveal directly the degree of degradation of individual grains preserved in lake sediment. The studies confirm SEM-EDS as a useful complement to chemical analysis. Differences in weathering rates of two sites were supported by results from geochemical analysis of sediment fractions. The presence of black carbon in soils was invested using SEM-EDS (Brodowski et al., 2005). The results suggest that black carbon exhibit a great variety in shape and surface properties from SEM-EDS probing. The results furt her showed that black carbon interacts with minerals and various forms of C were attached to mineral phases in soil. Thus, using O/C ratios, C origin could be traced and classified. Co-blending was an attempt to rapidly P im mobilizes P from soil-aqueous systems using sorbents. Currently, no study is av ailable on the effects of co-blending and the impact it has on soil morphology as well as any elemental compos ition that may result leading to new solid phases or changes on soil morphology. It is ag ainst this background, SEM-EDS were employed 97

PAGE 98

to probe the surface morphological changes that may occur due to co-blending. More so, if structure changes occur, what are the likely features? Control Soil and Sorbents Amended Images Images of soil without co-blending with sorbents and with no leaching vs. leaching (control soil) are presented as photomicrographs in (Figure 6-7 and Figure 6-8), respectively. It should be noted that the soil used, already had P loading through manure additions. Morphologically, one can observe that the leached soil was composed of smooth sand grains, while without leaching had roughed colloidal particles surrounding it. The leaching allows washing away of the colloidal sand grains. This was observed by further reducing the scale bar from 500 to 200 m for leached soil. The EDS show ed P in no leached soil and with no visible observable P in leached soil (Figure 6-8). Similarly, percentage elemental compositions were also analyzed using EDS data of Figure 6-7. Mean values (%) are presented in Ta ble 6-1. The result shows reduction of basic cations of Mg, Al, Si, and Ca. After 12 wks, the cation decreases in order of Al, Ca, Mg, and Si as 76.9, 67.4, 45.9, and 18.2% respectively. Thus, EDS data could be useful to indicate chemical changes likely occurring on surf aces of materials such as soil or sorbents amendments. In addition, the result re vealed Al, Ca and Mg as the dominant cations likely to be loss during leaching studies. 98

PAGE 99

Figure 6-7. SEM image of Control (Time zero) soil without co-blending/no leaching and the corresponding EDS spectra. Figure 6-8. SEM image of control (Time zero) soil without co-blend ing and with leaching (after 12 wks) and the corresponding EDS spectra. Table 6-1. Elemental composition (%) of cont rol soils with or without leaching. a Control (without leaching) b Control (with leaching) Elements Mean (%) Elements Mean (%) C cNA C 33.11 ( 0.42) O 67.33 ( 9.1) O 44.65 ( 1.29) Mg 0.37 ( 0.06) Na 0.05 ( 0.08) Al 3.51 ( 0.82) Mg 0.2 ( 0.10) Si 24.79 ( 6.43) Al 0.82 ( 0.81) P 1.28 ( 0.417) Si 20.29 ( 2.70) Ca 2.72 ( 1.35) Ca 0.88 ( 0.43) a Values are means of duplicate. b Values are means of triplicate. c Not available 99

PAGE 100

Sorbents Images The tested sorbents were characterized using SEM. This was to make it possible for morphological comparison to be made between the control samples in Figure 6-7 and with sorbent amendments images in (Figure 6-9 to 6-13) at 200m. The EDS spectra of the SEM images provided a list of the major elemental constituents of the sorbents used in the co-blending techniques. Any significant observations with respect to the control and the original sorbents might indicate the effects of the co-b lending techniques on surface morphology. As expected, the EDS confirmed what the ch emical analyses provided but it is also revealed additional impurities in some of the sorben ts. A point of note is that some of the images below were not too sharp due to the sorbents pa rticles, especially Al-W TR and MgO, absorbing the conductive carbon fluid. Figure 6-9. SEM images of Al-WTR and the corresponding EDS. The EDS indicates high amount of elemental Al content. Images were poor with swollen particles due to absorption of carbon fluid. Scale bar = 200m. 100

PAGE 101

Figure 6-10. SEM image of gypsum and the corresponding EDS. The EDS showing presence of Ca and S as major elemental composition. Traces of impure minute Mg particles were also observed. Scale bar = 200 m. Figure 6-11. SEM images of slag and the corresponding EDS spectra. The EDS showing presence of major elements as Si, Ca Mg and Al. Traces of minute levels of S and K were also observed. Scale bar = 200 m. Figure 6-12. SEM images of LimeKD and th e corresponding EDS spect ra. The EDS showing major elements present as Ca and Si, with traces of Al, Mg, S and K. Scale bar = 200 m. 101

PAGE 102

Figure 6-13. SEM images of MgO and the co rresponding EDS spectra. The EDS reveals major element as Mg. Scale bar = 200 m. Morphological Images of Co-blended Soils SEM images of soil impacted with P and co-blended with sorbents (2% rate) are presented in Figure 6-14 to 6-17. The result suggested that new solid phases were formed on the surface of the soil particles. Examples of th e new structure were found scattered on the soil surface. The EDS targeted specific spots to s how the elemental composition. The EDS spectra analysis in each of the co-blended samples i. e. MgO, Slag and LimeKD co-blended with AlWTR revealed PO4 imbedded in the new morphological struct ure formed on the surface. Thus, it might suggest a type of PO4 mineral. However there exists uncerta inty to prove the real nature of the mineral phases with the use of SEM and EDS analyses alone. Electron dot maps were also created of some of the co-b lended samples as shown in Figure 6-18 to 6-20. Electron dot ma ps are used to provide information on the elemental content imaged by SEM. The dot maps show presence of elements, and the association of the elements infers presence of minerals. The SEM image may appear white (tiny brig ht) spot or black. The white spot of say, Ca indicates presence of some calcium mineral related to Ca element. For full bright spot to be clearly seen, several resolutions of electron b eam are needed to pass through the samples (~ 50 minutes per sample). However, due to the cost, only ~15 minute resolutions were achieved. In Figure 6-18, one can observe a white spot under P. This suggests the presence of 102

PAGE 103

some PO4 minerals when (Lime + Al-WTR) were co-b lended together. Likewise, the white spot under P of (Gypsum + Al-WTR) and (Slag + WT R) also indicate the presence of some PO4 minerals. Thus, co-blended samples after 12 wks of leaching suggest the possibility of PO4 minerals being formed. Other advanced spectroscopic analytical tools such as FTIR, XPS, XAS and XANES may be used to differentiate, types of bonding, mol ecular structure and oxida tion states of P in sorbents added to the soil. However, due to th e cost and availability of the spectroscopic techniques, the use of geochemical model was also used to predict the likely mineral phases from leaching studies. Geochemical models woul d be discussed in proceeding sections. Figure 6-14. SEM and EDS spectra of (MgO + Al -WTR) sorbents co-blended with P impacted soil. The marked arrow indicates the mo rphological observation due to co-blending. EDS spectra revealing the el emental components with PO4 as part of the new solid phase. White spot indicate charging of the samples. Scale bar = 200 m. 103

PAGE 104

Figure 6-15. SEM and EDS spectra of (Slag + Al -WTR) sorbents co-blended with P impacted soil. The marked arrow indicates the mo rphological observation due to co-blending. EDS spectra revealing the el emental components with PO4 as part of the new solid phase. Scale bar = 200 m. Figure 6-16. SEM and EDS spectra of (Gypsum + Al-WTR) sorbents co-blended with P impacted soil. The marked arrow indicates the morphological observation due to coblending. EDS spectra revealing th e elemental components with PO4 as part of the new solid phase. Scale bar = 200 m. 104

PAGE 105

Figure 6-17. SEM and EDS spectra of (LimeKD + Al-WTR) sorbents co-blended with P impacted soil. The marked arrow indicates the morphological observation due to coblending. EDS spectra revealing th e elemental components with PO4 as part of the new solid phase. Scale bar = 200 m. Figure 6-18. Electron dot maps of (LimeKD +AlWTR) co-blended soil. Br ight spots indicate the location of elements within the new solid phase. 105

PAGE 106

Figure 6-19. Electron dot maps of (Slag + Al-WTR ) co-blended soil. Bright spot indicates the location of elements within the new solid phase. Figure 6-20. Electron dot maps of (Al-WTR+ Gypsum) co-blended soil, showing location of elements within the new solid phase Circled spots showing likely PO4 minerals. 106

PAGE 107

Conclusions The concept or the framework of co-blending was defined. The effects of co-blending technique were validated using a column leach ing experiment. The c-blending provided several options as remediation strategy for P immobiliz ation while at the same time it enables the potential to use the land to suit the desire purpose. For instance, if P is to be removed at a gradual pace, without any significant impact on the so il pH, (Al-WTR+Gypsum) co-blended appears to meet such a condition. Conversely, if the P is to be removed rapidly at pH close to the existing soil condition, (AlWTR+Slag) demonstrated success under such co nditions. Likewise, if P is to be removed immediately, but with raised pH, LimeKD a pplication met such a condition. Co-blending, therefore, leads to immobilizat ion of P species at a rapid ra te, depending on the choice of sorbents used in the application process. The effect of co-blending on soil surface morphology was also examined. This was achieved using dried soil samples after 12 weeks of leaching. The results show that co-blending had observable effects on soil morphology, which might be attributable to precipitation or solid phase formations. Thus, confirmed clearly that co-b lending resulted in struct ured alteration of the soil surface morphology. The nature of such solid phases was suggested to be mineral phases notably PO4 minerals. The exact types of the minerals could not be ascert ained from the EDS and electron dot maps. Geochemical models may however, leads to the lik ely nature of such minerals phases through equi librium speciation modeling. 107

PAGE 108

CHAPTER 7 GEOCHEMICAL MODELING Introduction Geochemical modeling is a tool for characterizing environmental site contaminations and predicting any environmental impacts. Geochemical models quantify reactions, reaction rates, and phase transfer between water and mineral or solid phase in diverse hydrogeochemical settings. The overall objective in using the geoche mical model is to use chemical analyses of leachate to determine what chemical reactions ar e occurring between water, the sorbents and the soil, as water moves through the soil column. Th e geochemical models used in this study are Visual MINTEQ version 2.60 (Gustafsson, 2009) PHREEQC (Parkhurst and Appelo, 1999), and Polymath model (Gadekar et al ., 2009a). The geo-chemical mode ls involve anal ysis of major anions and cations plus pH and carbonate alkalini ty to minimally satisfy conditions of reactivity. Phreeqc has several advantages and capabilities. It is one of the most comprehensive geochemical models available (Zhu and Anderson, 2002) It can be used to simulate forward and inverse geochemical problems. Forward mode ling involves taking a solution composition and determining what minerals are in or near equilib rium with the solution. It also includes reaction path modeling, which tracks the evolution of water in response to chemical reaction with minerals, surfaces or during mixing. On the other hand, inverse modeling involve s at least two solution compositions and calculates geochemical reactions that account fo r the observed changes in chemical composition of water along a flow path (Toran, and Gra ndstaff, 2002). Forward modeling would be employed in this analysis. In addition, Phreeqc has kinetic capabilities, which is lacking in most other equilibrium geochemical models such as MINTEQ, MINEQL+, and WATEQ. 108

PAGE 109

Phreeqc has been used by Horner et al., 2007 to model reactive transport in column experiments for a riverbank infiltration. The outputs from Phreeqc together with MATLAB tool verified the breakthrough of oxygen, DOC and inorganic hydrochem istry. Lipson et al., (2007) used Phreeqc to simulate solute transport in fractured bedrock aquifers. The dual-porosity transport model in Phreeqc could accurately simula te an advective, diffusive, and reactive solute transport process in fract ured bedrock aquifers. On the other hand, Visual-MINTEQ also has application in water qu ality assessments and is used widely by USEPA. In addition, it has incorporated wide thermodynamic base borrowed from MINTEQA2. Furthermore, complexation r eactions with dissolve organic carbon were solved using Gaussain model. The two geochemi cal models calculate saturation index (SI) of different mineral phases. The SI for each solid is defined as, log of an ion activity product (IAP) minus log of solubility constant (Ks). The Ks is drawn from the thermodynamic database, whereas log (IAP) is calculated using stoichiometry of the ions. For a particular mineral solid, if SI > 0, it implies that solution/l eachate is saturated with respect to that mineral. On the other hand, if SI < 0, it implies the leachate is under sa turated with the mineral phase. Both models can have inputs of pH, pE, and ionic strength defined as 0.013x EC (e lectrical conductivity), together with the concentrations of anions and cations. Charge balance error is ~20% tolerance limits. A new model, Polymath model developed for predicting magnesium related minerals is also employed (Gadekar et al., 2009a). It is a mathematical model of precipitation, which uses physicochemical equilibrium expressions, mass ba lance equation for nitrogen, P, magnesium and charge balance. The derived equation-model is solved using Polymath education version 6.1. The outcome suggests possible magnesium based minerals availability. 109

PAGE 110

The objective of this section is to use concen trations data from leaching experiment as inputs to be processed by Phr eeqc, VisualMINTEQ, and Poly math model in predicting the likely solid phases influencing PO4 solubility with respect to the co-blended samples. Materials and Methods Leachates were collected as in column studies described in chapter 6. Major anions (PO4, SO4, NO3, and Cl) and cations (Al, Fe, Ca, Mg, K, Na, and NH4), pH plus EC were analyzed. Cations with exception of ammonium were analy zed using ICP. Ammonium was measured with ion selective electrode, Fisher scientific, USA. PO4, SO4, Cl and NO3 were analyzed using ion chromatograph, Metrohm USA, Inc. However, the e xperimental design used is the same as that of chapter 6 with sorbents amended due to co-blending. Phosphorus chemical speciation was calculated for the leachates from the control, 2% application rate of Al-WTR, Slag, Gypsum, MgO and LimeKD, in addition to co-blende d samples of (1+1%) rates each of AlWTR+Gypsum, Al-WTR+LimeKD, Al-WTR+MgO, and Al-WTR+Slag. Previous work from Silveira et al., (2006) us ing several selected w eeks of leachates data, suggest that number of weeks of selected leachate used, did not have any significance difference in chemical speciation. Based on that, the 1st w eek of data (pore volume leachates was allowed to equilibrate for one week in the column befo re leaching began), was used in the chemical speciation equilibrium modeling. Further, af ter 2nd to 4th week, due to co-blending, PO4 concentrations were almost zero, thus difficult to predict any PO4 mineral. Leachates anions analyzed were NO3, Cl, SO4, ammonium, and PO4. Cations analyzed were Al, Mg, Ca, K, and Na. In addition, pH, Ec and alkalinity data we re also analyzed as inputs for the chemical speciation. 110

PAGE 111

Results and Discussion Visual-Minteq and Phreeqc Phosphate speciation distributions were obt ained from visual-minteq process for the control soil without any sorbent applicat ion (Table 7-1). The major identified orthophosphate ions in solution were HPO4 2and H2PO4 corresponding to ~30% and ~4% respectively of total soluble P. This was consistent with (Lindsay, 2001) observation on distribution of orthophosphate in soil solution. Orthophosphate species of H2PO4 and HPO4 2have been shown to occur in the pH range of (2-7) and (7-12) respectively. Further, the greater percentage of HPO4 2available shows clearly, the environment is slightly alkaline as supported by pH of 7.7. Thus, supporting the fact that if Al-WTR is applied alone, there may be lim itations in removing HPO4 2species in alkaline environment, since Alhydr(oxides) is noted to remove species in ac idic environment. In addition to orthophosphate ions present, Mg-P [MgHPO4(aq)] and Ca-P [CaHPO4or CaPO4-] complexes of ~48% and ~18% respectively of the potential negatively ch arged P solid phases. Thus representing, ~66% total soluble P. These unsaturated complexes are the ones that would have to be stabilized to prevent P losses. The 66% total P observation from Ca-Mgco mplexes is consistent with sequential extraction data in chapter 6. Due to loosely bound P associated with Ca-Mg, any change in pH results in P replenishment from th e Ca-Mg sources into the main labile pool. This is further supported by data from Table 7-2, i ndicating saturation indices (SI) of Ca-Mg solid phases were negative or unsaturated. For all treatments (Table 7-2), hydroxyapatite was predicted to be stable from visualminteq with SI > 1. Further, in all treatments Mg -P complexes were also shown to be unsaturated with the exception of 2% rate application of (Soil+MgO) and (Soil+ WTR+MgO) at (1+1 %) application showing positive SI indices. Thus, unde rscoring the fact that extra Mg addition to the 111

PAGE 112

soil aid in strengthening the Mg rela ted solid phases. In addition, CaHPO4, and CaHPO4.2H2O exhibit undersaturation in all trea tments except gypsum and the c ontrol. This may be explained due to incongruent solubility and phase transfor mation when Gypsum is added to the soil, which may lead to other form of calcium-P solid pha ses. Such heterogeneous nucleation complexes, may lead to other forms other than hydroxyapa tite, such as dicalcium phosphate dehydrate (DCPD), Octacalcium phosphate (OCP) and amorphous calcium phosphate (ACP) leading to positive stabilization of the comple xes (Pan and Darvell, 2009). The control had a strong SI of hydroxyapatite compared to co-blended samples, and may help in contributing to the stability index of other Ca-P complexes and Mg-P complexes having a negative SI (Table 7-2). This is because according to Ostwald step rule, it is no t the phase that is thermodynamically stable that nucleates first, but rather the meta stable phase that is closest in free energy to the present phase. Furthermore, Ca3(PO4)2 ( ), and Ca4H(PO4)3.3H2O have been found to be saturated (Table 7-2), with the exception of Soil+ WTR+LimeKD (1+1%) and Soil+WTR+ Slag (1+1%) showing negative SI indices. Such negative SI may be due to other forms of calcium solid phases that may be occurring but were not identified by visual-minteq. This is due to the fact that the SI of the hydroxyapatite was reduced fro m 13.57 of the control soil to 5.73 for Soil+WTR+LimeKD (1+1%) co-ble nded soil, and to 9.58 for Soil+ WTR+ Slag (1+1%). The results therefore support the fact that co-blendi ng may be creating new additional solid phases as observed under SEM-EDS. To ascertain these othe r forms of solid phases available but were not identified by Visual-Minteq led to the use of polymath model. Polymath model would be explained in the next sections. The use of Phreeqc was also employed to characterize the chemical speciation distributions of treatments with and without co-blending. Results are presented in Table 7-3. 112

PAGE 113

Comparing species distribution of SIs, Phreeqc appears to lack th e broad distributions of diverse solid phases as revealed by visual-minteq. Howeve r, general trends of SI with Phreeqc were similar to that of visual-minte q, although Phreeqc has a lower value. For instance, SI of control hydroxyapatite for visual-minteq was 13.57 whereas that of Phreeqc indicates 9.40. It was also noted that SI of hydroxyapatite decreases c onsiderably when soil was co-blended with WTR+LimeKD and WTR+Slag at (1+1 %) rate application of 3.07 and 5.26, respectively. Such SI reduction was also revealed using visual-m inteq. The decrease in SI may be due to other metastable solid phases that may be occurring as intermediate phase to a stable phase. The use of Phreeqc indicates species of CaSO4, CaCO3, CaMg(CO3)2, CaSO4.2H2O which as mostly associated with geological formation, although they are of environmental significance (Table 7-3). In addi tion, the species lack the interm ediate phase transformation of calcium related P solid phases as revealed by vi sual-minteq. The distribu tion observed also lack species associated with magnesium-P solid pha ses. Perhaps additional and new thermodynamic data inputs may indicate the phase transfor mation of such species indicated above. In conclusion, stability indices as reveal ed by the use of Phreeqc and visual-minteq showed the likely mineral phase occurring as a result of co-blending. Co-blending used less amount of Al-WTR application to the soil and further showed other so lid phases available through the process which otherwis e is not available in single sorbent applications. The coblending resulted in rapid P imm obilization. No single geochemical model is sufficient to reveal all forms/species of P distribution. Each mode l has strength and limitatio ns. A new model called a Polymath model developed by (Gadekar et al., 2009a ), would be used to detect any deficiency that may be lacking in the use of Phreeqc a nd Visual-Minteq for magnesium related mineral phases, such as struvite over visual-minteq. This would be discussed belo w. As discussed above, 113

PAGE 114

SEM-EDS have shown other types of PO4 mineral were present, but were difficult to fully identify. Polymath Model A polymath model is a novel mathematical model for predicting struvite formation (Gadekar and Pratap, 2009a). The model satisfactory predicted struvi te fractions in precipitates, ranging from 27% to 100% (Gadekar and Pratap, 2009b). Struvite is a PO4 mineral (MgNH4PO4) and considered as a slow releas e fertilizer. Detection of struvi te in soils has proved difficult. However, recent data from Gungor et al., ( 2008) with the aid of XANES spectroscopy has identified struvite in raw and anaerobically digested dairy manure. The use of XANES spectroscopy is an expensive tool to use. Polymath was built in the lab and cheap and easy to use. The objective of this section is to us e Polymath model to predict the various PO4 minerals especially struvite, likely to o ccur in soil leaching study as a re flection of the presence of the mineral availability. The model considers overall mass balance for magnesium, nitrogen and P, electroneutrality, and physico-chemical and solubi lity equilibrium equations, pH and total concentrations to describe the system. Total in organic carbon and calciu m are input along with equilibrium constants. Ionic specie s modeled are concentrations of NH4, PO4, Na, K, Ca, Mg, CH3COO, CO3, Cl, H, OH, HPO4, H2PO4, MgOH, MgH2PO4 and MgPO4. In addition, concentrations of the following unionized or dissolved species are also modeled: NH3, H3PO4, CH3COOH and MgHPO4 (dissolved). Possible formations of five different precipitates are considered, namely, MgNH4PO4.6H2O (struvite), MgPO4.8H2O, MgPO4. 22H2O, Mg(OH)2 and MgHPO4. What distinguish Polymath from other equi librium models like Scott et al., (1991), is that, Polymath model incorporates concentrations of all species (dissolved, ionic, and solid). 114

PAGE 115

Further the model maximizes the use of mass and charge balance as well as the physicochemical equilibrium equations. Polymath e ducational version 6.1 was used to solve the nonlinear equations of the concen trations in solution. This was used to furnish the overall concentrations of dissolved and ionic species as well as concentr ations of total solid components in the model calculations (Gadekar et al., 2009a). Comparison of Polymath Model with Visual-Minteq Random data taken from the literature usi ng experimental, synthetic as well as, actual wastewater experimental data were simulated using Visual-Minteq and Polymath model. A summary of part of the data was taken from Gade kar et al., (2009b) is pres ented below. The data utilize the concentrations of wastewater involving eighteen species that included dissolved (three), ionic (ten), and solid (five) species for initial concentrations of ammonium-nitrogen, magnesium, and PO4-P and pH. The result of the model si mulation of Polymath and that of Visual-Minteq is presented for comparison (Table 7-4). By way of observation, Polymath model predicts greater amounts of struvi te over that of Visual-Minteq. A graphical representati on of the amount of struvite predic ted by each model versus that of the experimental values is also plotted. (Figure 7-1 and Figure 72). The results showed clearly, the superior strength of Polymath (R2 = 0.94) over that of the Visual-Minteq (R2 = 0.22) in predicting struvite. The inability of Visual-Minteq to predict str uvite well was suggested to be due presence of calcium ions which tend to ca use interference (personal communications from Gadekar et al., 2009b). A polymath model, however, incorporated calcium and carbonate ions in its formulation. This might have led to lowe r reduction of errors observed in the model predictions. 115

PAGE 116

Contributions of Solid phases from Polymath Model Tabulated data of Polymath simulation using leachate samples is presented in (Table 75). Data inputs are the same as those used fo r Phreeqc and Visual-minteq. The data showed clearly, the amount of struvite pr edicted, as well as the percentage purity and the likely total solids in the leachates. Struvite is predicted in all treatments. However, their relative proportions differ with respect to a particular treatment. SEM-EDS images of Figure 6-14-to-2-19, well as electron dot maps of Figure 6-18-toFi gure 6-19 clearly sugg ested a type of PO4 mineral. However, to ascertain its va lidity demands different types of instrumentation such as synchrotron-based spectroscopy, whic h is not available at University of Florida. Polymath model however, helps to identify one such type of phosphate mineral, i.e. struvite. Struvite predicted by the Polymath model to decrease in the following order MgO > Slag > Al-WTR > LimeKD > Gypsum (120, 90, 37, and 12) mg L-1 at 2% application rate in the soil. It clearly shows that Mg application is the ke y to struvite formation in the soil. The order indicated above corresponds to the level of Mg contents in each tr eatment. As noted earlier, Mg is loosely associated in manure-impacted soil, so a little addition serves as a catalyst to precipitate struvite in the soil. In addition, at industrial forma tion of struvite, MgO is commonly used as a seed to struvite nucleation (Shu iling and Andrade, 1999; Chimenos et al., 2003). Application rate at (1+1) %, of Mg containing materials (Soil+WTR+Slag), (Soil+WTR+MgO), had ~350 and 200 mg L-1, respectively of struvite predicted from the Polymath model. The (Soil+WTR+Slag), had grea ter struvite formation apparently due to the right condition pH of ~8.2. The pH of 8.2 is al so conducive for plant growth. Thus, co-blending may not affect plant conditions provided the right a pplication rate is used. The amount of struvite formed under (1+1)% is greater than 2% appli cation of either Slag and MgO combined. This further confirmed that co-blending enhanced PO4 mineral formation especially struvite. On the 116

PAGE 117

order hand, (Soil+WTR+LimeKD) did not produce any significant struvite in the soil, ~ 0.001 mg L-1. Apparently due to the fact that addition of LimeKD is an exothermic reaction evolving heat, thus may have caused the loss of ammonium via volatilization, which would inhibit struvite formation. Further, Visual-Minteq predicted hydr oxyapatite, as the single dominant mineral under the treatment with LimeKD. Gypsum app lication as (Soil+WTR+Gypsum) 1+1%, had ~ 80 mg L-1 struvite. Gypsum is sparingly soluble, its solubility and the gradual formation of hydroxyapatite as suggested by Visual-Minteq may have allowed and led to the formation of struvite formation in moderate amount. Struvite: Slow Release of P and N The conventional school of thought in addressing soil P pollution problem is to permanently fix the P with a sorbent. This is to pr event future release of P into the soil. However, the unconventional thinking should be fixing the P in the soil and releasin g it later, for plant uptake. This school of thought is beneficial in cost terms, as well as for sustainable environmental management. Struvite is produced at industrial scale and us ed as a slow release fertilizer. It can release the P and N. The released N and P can be beneficial to plant growth. However, struvite is not use al one but rather used by mixing to other forms of inorganic and organic fertilizers (Ueno and Fujii, 2001). With the presence of struvite in the soil, as suggested by the geochemical model, would save additiona l cost from synthetic struvite manufacturing. Further research efforts are needed in this direction to save cost and for environmental sustainability of natural resources. The kinetics of P release is so slow, and may leads to saving cost to ranchers whose soils are heavil y impacted with manure. The dwindling PO4 reserves in the world and for that matter in Florida should be a concern. This, therefore, suggest that the reuse and recycling of excess P in the soil should be considered. In Florida anecdotal evidence suggest the PO4 industry has between 10-20 yrs to be exhausted. 117

PAGE 118

Ways, to recycle the P includes the removal of the P from animal waste before application, e.g. removal of P from wastewater before discha rge to land or surface waters. This industrial approach is expensive. The recycling of na tural resource has become ways of facilitating environmental sustainability, a nd protection of the environmen t. The use of co-blending has suggested that, the process could yi eld useful struvite in the soil. Thus, suggesting that P can be fixed and released later, for plant uptake. However, the release of P from struvite is very slow and trickles gradually, implying that it may not be sufficient for adequate plant growth. In addition, the release of P may not be in significant amount to cause environmental problems if supplemented with adequate plant uptake Further research is need in th at direction to cause sufficient release of P by struvite in the soil, to be used by plant. The mechanisms suggested from co-blending application including th e use of Al-based materials. Should the pH fall into the acidic ra nge would activate the sorption of P by the Al. However, it is expected that before the pH reach es the acidic level, plant may have utilized the excess P available in the alkaline form of HPO4 2-and Ca-Mg-P solid phases. Research efforts are also needed to optimize a given soil system for optimal release of P (from struvite) in a high Pimpacted environment through co-blending applications. Conclusions This chapter utilizes geochemical models to predict possible solid phases occurring in leachate samples. Geochemical models are not a means to an end, in the identification of solid phases. The models rather point towards the existence of solid phases under thermodynamic assumptions and considerations. Three geoche mical models (Visual-Minteq, Phreeqc and Polymath) were examined. Hydroxyapatite is we ll predicted by both Visual-Minteq and Phreeqc. The SI of Hydroxyapatite were positive under Vi sual-Minteq is ~13.6 and 9.4 in the control soil 118

PAGE 119

119 leachate. On the other hand, the Polymath mode l predicted a likely hood of a small amount of ~1.35E-9 mg L-1 for hydroxyapatite. Application of co-blended sorbents as amendments resulted in reduction of SI indices, thus supporting other minerals formations. However, observation from the Polymath model revealed the new solid phases, which were lacking in Vi sual-Minteq and Phreeqc as possibly a struvite mineral. A comparison of the Polymath model to Visual-Minteq was made using data from the literature. The Polymath model showed s uperior coefficient of determination (R2 = 0.94) in identification of struvite, while Visual-Minteq had (R2 = 0.22). Struvite formation was predicted to decrease in the following order, MgO > Slag > AlWTR > LimeKD at 2% application (120, 90, 37, and 12) mg L-1. Thus, suggesting that moderate Mg application is the key to struvite preci pitation in the soil. In co-blended soil, (Soil+WTR+Slag) had greater predicted struvite formation (~350 mg L-1), apparently due to the right pH of 8.2. On the other hand, (Soil+WTR+ MgO), with an average pH of ~9.6, had a predicted amount of ~200 mg L-1. Thus, suggesting that over Mg application can lead to higher pH but low amount of struvite precipitation. Optimization of Mg in relation to the pH is therefore needed to obtain the desired struvite precipitation. The prediction of struvite in th e leachate through th e use of Polymath model, suggests that, the PO4 mineral (struvite) can be prec ipitated and released later as nutrients for plant uptake. Further research efforts are needed in this direction.

PAGE 120

Table 7-1. Phosphate species distribution (Vis ual-Minteq) in control soil without sorbents showing potential negatively charged P complexes. Phosphate species Percentage (%) HPO4 230 H2PO4 4 Ca-P complexes 18 Mg-P complexes 48 Total 100 Table 7-2. Saturation indices calculat ed using Visual-Minteq for treatm ents with and without co-blending. Treatments Ca3(PO4)2 ( ) Ca4H(PO4)3.3 H2O CaH PO4 CaHPO4.2H2O Hydroxyapa tite Mg3(P O4)2 MgHPO4.3 H2O Soil (Control) 4.17 3.99 0.40 0.10 13.57 -0.48 -0.39 Soil+ LimeKD (2%) 2.77 1.71 -0.48 -0.77 11.65 -5.59 -9.76 Soil+MgO (2%) 3.13 1.03 -1.52 -1.81 13.42 1.89 -1.18 Soil+Gypsum (2%) 3.36 3.08 0.31 0.01 12.05 -1.44 -0.54 Soil+Slag (2%) 2.05 0.19 -1.28 -1.57 11.01 -2.33 -1.98 Soil+Al-WTR (2%) 3.22 2.59 -0.05 -0.35 12.13 -1.81 -0.97 Soil+WTR+Gypsum (1+1%) 3.08 2.93 0.10 -0.18 11.82 -1.65 -0.70 Soil+WTR+LimeKD (1+1%) -1.33 -4.67 -2.76 -3.05 5.73 -7.48 -4.05 Soil+WTR+MgO (1+1%) 2.37 0.14 -1.64 -1.94 12.01 0.16 -1.62 Soil+WTR+Slag (1+1%) 1.15 -0.62 -1.52 -1.80 9.58 -3.46 -2.28 120 Saturation index = log ion activity produc t (IAP)log solubility product (Ks).

PAGE 121

Table 7-3. Saturation indices calculated using Phreeq c for treatments with a nd without co-blending. Treatments Anhydrite (CaSO4) Aragonite (CaCO3) Calcite (CaCO3) Dolomite (CaMg(CO3)2) Gypsum (CaSO4.2H2O) Halite (NaCl) Hydroxyapatite (Ca5(PO4)3OH Control soil -0.48 0.57 0.71 1.87 -0.25 -5.75 9.40 Soil+Al-WTR-2% -0.34 -0.03 0.12 0.55 -0.12 -5.80 7.83 Soil+Slag-2 % -0.52 -0.31 -5.93 6.75 Soil+MgO-2% -1.43 0.86 1 3.59 -1.20 -5.61 6.35 Soil+Gypsum-2% -0.02 -0.97 -0.83 -1.26 0.19 -4.79 8.36 Soil+LimeKD-2% -0.3 0.95 1.09 2.08 -0.08 -5.85 7.83 Soil+WTR+Gypsum(1+1%) 0 0.09 0.23 0.83 0.21 -5.73 8.05 Soil+WTR+LimeKD(1+1%) -2.38 0.78 0.93 1.81 -2.16 -5.65 3.07 Soil+WTR+Slag-(1+1%) -0.39 0.76 0.9 2.21 -0.17 -5.79 5.26 Soil+WTR+MgO(1+1%) -1.05 1.31 1.46 4.16 -0.82 -5.7 6.24 121

PAGE 122

122 Table 7-4. Dataa taken from literature to simulate model predictions of Polymath and Visual-Minteq. References Type of wastewater pH Initial concentrations (mM) Exptl Struvite (mg L-1) Model Predictions Struvite (mg L-1) MgT PT NT VisualMinteq Polymath Loewenthal et al., 1994. Solutions prepared by adding NH4Cl, KHPO4, MgCl, carbonate and acetate 6.8 8.33 12.9 21.43 601 0 223.61 Harada et al., 2006. Synthetic urine containing PO4, NH4, Na, Mg, K, Ca, Cl, citrate, carbonate 8 20.0 0 13.45 20.18 1685 0 1253.26 Wilsenach et al., 2007. Synthetic urine containing PO4, NH4, Na, Mg, K, Ca, Cl, citrate, carbonate 9.4 7.42 14.83 18.70 1045 1006 987.07 Wilsenach et al., 2007. Synthetic urine containing PO4, NH4, Na, Mg, K, Ca, Cl, citrate, carbonate 9.4 14.8 3 14.83 18.70 2011 1087 1845.03 Celen et al., 2007. Liquid manure 8.5 2.39 5.51 80.00 338 319 322.2 Munch and Barr, 2001. Supernatant from anaerobically digested sludge dewatering centrifuge 8.5 1.51 1.97 43.88 195 171.05 200.86 Yoshino et al., 2003. Anaerobic digester effluent supernatant 8.5 7.02 5 6.387 24.5 805 825 818.23 Tunay et al., 1997. Synthetic samples prepared by using MgCl2, NaH2PO4, NH4Cl 9 14.2 6 14.26 14.26 1714 1581 452.39 Altinbas et al., 2002. Domestic wastewater + 2% landfill leachate 9.2 7.79 7.79 7.79 1420 681.7 207.44 Battistoni et al., 1998. Supernatant from sludge centrifuges in a biological nutrient removal plant 8.1 1.54 2.00 44.5 210.98 143.64 198.21 Burns et al., 2001. Swine waste 9 9.74 6.09 12 758.6 713.5 705.47 aData taken from Gadekar et al., 2009b.

PAGE 123

Table 7-5. Calculated values of solid (mg) found per litter of residual leach ate as determined by Polymath model Treatments Ca3PO4 Ca5(PO4)3OH CaHPO4 5H2O CaHPO4 Mg(OH)2 MgHPO4 MgPO4 22H2O MgPO4 8H2O Struvite (%Purity) Total Solids Control 26.85 1.35E-9 622.32 280.88 22.49 23.18 0.15 1.135 12.24 (1.2) 989.3 Soil +Al-WTR (2%) 48.13 4.96E-9 544.38 245.70 95.97 36.92 1.61 12.28 37.18 (3.6) 1022.2 Soil+Slag (2%) 14.34 4.95E-10 484.38 218.62 96.69 87.94 9.2 70.21 90.29 (8.4) 1071.7 Soil+MgO (2%) 377.64 9.60E-7 173.43 78.27 2295.93 3.64 0.37 2.86 119.0 (3.9) 3051.2 Soil+Gypsum (2%) 0.41 2.25E-12 88.99 40.16 14.29 2.78 0.001 0.01 1.56 (1.1) 148.24 Soil+LimeKD (2%) 0.71 6.08E-12 97.43 43.97 29.04 4.32 0.0069 0.051 12.26 (6.5) 187.8 Soil+WTR+Gypsu m (1+1%) 19.43 2.0E-10 2164.84 977.08 6.14 368.19 10.25 78.20 80.20 (2.2) 3704.4 Soil+WTR+LimeK D (1+1%) 1.1E-6 3.45E-20 0.04 0.018 54.72 0.0004 9.6E-11 6.98E-10 0.001 (0.002) 54.94 Soil+WTR+Slag (1+1%) 8.77 8.23E-11 1092.6 493.00 19.00 324.06 24.57 187.36 353.1 (14.10) 2502.6 Soil+WTR+MgO (1+1%) 92.08 6.58E-08 150.26 67.81 2080.67 8.79 1.98 15.1 199(7.4) 2610.9 123

PAGE 124

Figure 7-1. Experimental values plotted vs. predicted values of struvite from Visual-Minteq. Figure 7-2. Experimental values plotted vs. predicted values of st ruvite from a Polymath model. 124

PAGE 125

CHAPTER 8 FLOW CALORIMETRY AN D SURFACE REACTIONS Introduction Flow calorimetry (FC) is a tool that ha s been recognized to study surface reactions (adsorption/desorption) at the solid/liquid inte rface (Groszek, 1998). Other investigators used flow calorimetry to determine heat of adsorp tion and kinetics of reactions at solid/liquid interfaces (Steinberg, 1981; Rhue, et al., 2002; Harvey and Rhue, 2008). The fundamental principle is that calorimetry as an analytical method measures the change in enthalpy. Enthalpy changes occur in all physical, chemical and biological reactions and can be quantified with calorimetry (Steinberg, 1981). The role of calorim etry is to derive a numerical value to the changes associated in thermodynamic variables through measurement. The driving force of such changes on surfaces is due to thermal effects, and provides insights into physical and chemi cal interactions in a unique ma nner. Quantitative data of intensity, extent, rate, an d reversibility of interactions can become available from measurement of these heats effects. Most importantly, calo rimetry can uniquely qua ntify interactions as adsorption, desorption, and competition for a surface, as well as immersion, solubilization or solvation, mixing, chelation, and othe r transformations Steinberg, (1981). Some researchers use the name isothermal micr ocalorimeter instead of flow calorimetry. The term isothermal microcalorimeter is used to refer to experiments where the temperature remains constant and measurements are in micr owatt range. Different measurement principles are employed depending on what type of experi ments the instruments ar e intended for. Three main groups of measurement are identified: ad iabatic, heat conduction and power compensation calorimeters. 125

PAGE 126

In adiabatic calorimeter, no heat exchange ta kes place between the calo rimetric vessels and its surroundings. The amount of heat evolved or adsorbed is equal to the product of the measured temperature change and the heat capacity of the vessel. In the sec ond category, heat conduction calorimeter, heat released or absorbed in a re action vessel is allowed to flow to a heat sink (surrounding). A thermopile is us ually positioned between the vesse l and heat sink as sensors for heat flow. This allows a temperature gradient be tween the vessel and the h eat sink as heat flows, which is measured as thermopile potential (Wadso, 2001). Thirdly, in a power compensation calorimeter, the thermal power from an exothermic process is balanced by a known cooling effect. Alternatively, an e ndothermic process is balanced by a known thermal power released in heater. In spite of the differences, isothermal microcalorimeter possesses thermal power, heat production rate ( Q/ t), which is a function of thermodynamic as well as kinetic properties. In principle, all isothermal microcalorimeters can therefore be used as kinetic and thermodynamic instruments and can be used for the determina tion of enthalpy changes. Further, kinetics reactions and thermodynamic parameters can eas ily be obtained from the response/power-time curve. In principle, isothermal microcalorimeter is the same as that of flow calorimeter; it can be applied to all types of reactions or processes that are accomp anied by thermal effects (Rong et al., 2007a). The application of the techniques has been applied to wide range of studies involving life science, material science, and pharmaceutical development over the past yrs (Phipps and Mackin, 2000). The main advantage of flow/microcalorimeter is that, it permits the continuous monitoring of the reactions occurring in-situ, and without disturbing the system. This is especially useful for reactant species of contaminant transport to a sorben t, as in wastewater treatment, or interactions 126

PAGE 127

of soil minerals with contaminants. In addition, ca lorimetry is a non-specific analytical tool, and can, in principle, be applied to a ll types of systems and processes. This property of calorimetry is a powerful tool for the discovery of unexpected or unknown processes or reaction steps to be discovered in complex systems. Flow/microcalorim eter methods of investigation, therefore, can provide new insights into the su rface reactions and m echanisms involved in adsorption kinetics, ion exchange, chemisorption and other unknown mechanisms, which otherwise were unexplainable. Briggner et al., (1994) used microcalorimeter to study changes in crystallinity, induced during the processing of powders. The study illustra tes the transitions processes that solid-state material undergoes due to changes of inte rnal environment leading to amorphous and crystallinity of the material. Mi crocalorimeter can therefore, be used to ascertain the degree of crystallinity of a solid state of desire feat ures and properties. Lock and Ford, (1983) used flow microcalorimeter to measure heat output from attached or sedimentary microbial co mmunity over which flow of water allowed is to pass over. The results suggest that substantial heat outputs were obtained from the biological materials. The heat output ranges from 32-to 284W flow cell-1, and with lowest heat output found to be ten-fold higher than the lower detection lim it of the instrument (3 W). Joslin and Fowkes, (1985) used flow microcalorimetry to measure surface acidity of ferric oxides. They concluded that FeOH surface sites are acidic and adsorb nitrogen and oxygen bases with appreciable heats of adsorption. The heat of adsorption was quantitatively correlated and predicted with the Drago E and C equation. Bakri et al., (1988a and 1988b) utilize the sensitivity of flow microcalorimetry to determine the chemical stability of organic com pounds as well as the reaction rate parameter. 127

PAGE 128

The outcome of the experiment showed that fl ow microcalorimetry could be used to study stability problems. Further, due to heat involv ed in the chemical processes, thermodynamic and kinetic parameters of wide variety of chemical reactions such as the rate can be estimated. The study used theoretical derivations as well as experimental data to support the use of flow calorimetry in kinetic studies. Rong et al., (2007b), investigate the metabolic activity of Bacillus thuringiensis as influenced by kaolinite, montmorillite and goethit e. The study revealed that clay minerals and iron oxide stimulate exponential growth of B. thuringiensis, whereas the sporulation of B. thuringiensis was inhibited by the presence of kaolinite and montmorillonite. The study concludes that surface interaction between microb ial cells and soil mineral account for the effect of these minerals on th e metabolic activity of B. thuringiensis. Lantenois et al., (2006) investigate adsorpti on of Pb (II) and Cd (II) on silica used flow microcalorimetry. The study observed the adsorpti on of metals on silica is controlled by pH, chemical potential of cations and hydrations sh ells. The heat of ca dmium adsorption was found to be low, endothermic and quantitatively equivale nt to that of desorption. In the case of lead, adsorption had no thermal effects. Thus fl ow microcalorimetry provides explanatory information on the mechanisms of adsorption of these two metals on silica. Groszek, (1970) used flow calorimetry to study adsorption of long chain n-paraffins on graphite basal planes from n-heptane. The result suggests that adsorbed n-paraffins molecules consist of monolayer and are in close packing. These findings we re confirmed after 20 yrs using scanning tunneling microscope (STM) and XR D studies, McGonigal et al. (1990). Their findings demonstrated the effec tiveness of flow calorimetry. 128

PAGE 129

Groszek and Partyka, (1993) used heats of adsorption quantification from flow calorimetry, to determine specific surface areas of metal oxides. The heat produce was found to be proportional to specific surface areas of larg e polar solids and correlates well with BET (N2) surface areas. Further, Fowkes et al., (1988); Groszek, (1990) showed that heat of adsorption and amounts of adsorption measured by the flow met hod correlates well with adsorption determined by conventional batch methods. Rhue et al., (2002) used flow calorimetry to measure surface reactions of soil i.e. cation exchange and PO4 sorption. The reaction of PO4 on Al(OH)3 and Ultisol was exothermic. A repeated cycle of PO4 reaction on soil was found to saturate the adsorption capacity, due to loss of detectable heat signal. Further, the study observed that pr ecipitation was not the primary mechanism for PO4 sorption. Therefore, flow calorimetry can provide additional information about surface chemical reactions, which ca nnot be obtained readily by other methods. Kabengi et al., (2006a) use flow calorimetry to determine point of net zero charge (PZNC) on amorphous Al hydroxide (AH). The heat of ex change determined calorimetrically was found to be directly proportional to su rface charge and that PZNC can be equated to pH at which heats of cation and anion exchange are equal. The study found out that PZNC of AH determined calorimetrically was ~9.5, and is consistent with PZNC values reported in the literature. Harvey and Rhue, (2008) used flow calorimetry (FC) to study PO4 sorption on multicomponent Al-Fe hydroxides sorbents systems. The results of FC we re compared with traditional batch experiment. The results suggest that using FC, PO4 adsorption resulted in reduction of peaks of NO3 and Clcalorimetric responses, which were consistent with anion exchange capacity (AEC) losses. Further with FC, on average of ~1.9 moles of AEC were lost per mole of 129

PAGE 130

PO4 adsorbed. These results were also consistent with a bi-dentate interaction. In addition, the AEC loss was found to be irreversible with respect to NO3 and Cldue to inability to generate pre-P AEC with repeated cycles. The above selected literature reviewed showed the effectiveness of FC in revealing intrinsic pr operties of sorbents as well as some reaction mechanisms which otherwise were unexplainable and unrevealed. Furthermore, FC has been reported in terms of time resolution of detecting changes on surfaces to about one millisecond (Wadso, 2001). Al so, in adsorption science, when a metal oxide is brought into contact with an electrolyte, the outermost surface oxygens adsorb one or two protons, a cation, or an aggregate composed of two protons and an anion. This process leads to formation of complexes, electrically charged interface and geometric distortion of the oxide surfaces. This oxide surface defects and distortion has been exploited and used in catalysis. This is because; it is believed that surface defects create catalytic cent ers for catalytic reactions. This phenomenal change on the surface has been noted in adsorption science that calorimetric effects of adsorption are much more sensitive to surface energetic heterogeneity of the oxide/electrolyte interfaces than adsorption is otherm (Rudzinski et al., 1998). Al-WTR has recently attracted lots of attention in the literature on its ability to immobilize large amount of anions such as PO4, arsenic and perchlorate (Dayt on and Basta, 2005a; Makris et al., 2004). To best describe Al-WTR, it has been characterize as amorpho us Al-hydr(oxides). The complexities associated w ith Al-WTR are enormous due to large soluble organic carbon, and presence of several particulat e residuals. It is against this b ackground; FC is used to probe the surfaces of Al-WTR with in coming reactants such as Cl, NO3 and P containing solutions. Thus, may provide new insights on sorption behavior of Al-WTR. Th e hypotheses of this section were as follows: (i). Flow calorimetry may help in assessing the surface chemistry of Al-WTR. 130

PAGE 131

(ii). The mechanisms of P sorption may be reve aled as P reacts with amorphous Al-WTR. Two main objectives were set up as: (i). To determine ion/anion-exchange capacity of amorphous AlWTR, and (ii). To provide foundation for modeli ng P sorption using the parameters associated with FC. Materials and Methods Instrumentation Flow calorimetry used in this study was de signed and built as in (Rhue, et al., 2002; Kabengi, et al., 2006; Harvey and Rhue, 2008). It consists of a small column assembly containing a sorbent placed between two thermistors, one as inlet and the ot her as outlet with a calibrating resistor at one end (Figure 8-1). The column assembly is sealed and placed in a 500 mL polyethylene bottle. The bottle is placed in a 50 L water bath at room temperature. The water bath provides a good thermal stability against am bient temperature variations responsible for baseline drifts during the cause of analysis. The functions of the thermistors are temperatur e sensing in solution as changes occurs. The sensed temperature in solution produces a diffe rential output voltage, which is fed into an instrumentation amplifier. The amplified signa l is subsequently fed into a computer for processing. The main advantages of this flow calorimetry are the high sensitivity and low signal to noise ratio it exhibits (Kabengi et al., 2006; Rhue et al., 2002). A sorbent (Al-WTR) of ~100 mg is placed insi de the column and reactive solutions are allowed to pass through the column using a total pr essure drop of ~100 cm of water. Flow rates in the range of 0.30 to 0.38 ml min-1 are used. During analysis flow rate is constantly measured to ensure uniform collection of data. Run time of analysis ta kes about ~20 min to ~60 min, depending on when the signal returns to baseline. 131

PAGE 132

Peaks areas are either endothermic or exot hermic, depending on the reactive solutions and the nature of sorbent. The peak area is calcu lated by integrating the signal (in millivolts) numerically over time. This time-averaged peak area (V min) was converted to a flow rateaveraged peak area (V ml) by multiplying by the average flow rate. This was measured for each peak by collecting the effluent volume and di viding by the time over which the volume was collected. Peak areas were converted to energy (Joules) units by comparison with peaks generated with a calibrated resist or located within the flow str eam, i.e. near the inlet flow. Voltage and current for the heat pulses were measured and the heat input calculated from the relation Q (Joules) = V.A.t where V is voltage, A is amperage, and t is the time, in seconds, that the resistor was energized. The resulting graph was in tegrated as flow rate-averaged peak area and compared with 15 seconds heat pulse calibration p eaks. Differences in peak characteristics for Al-WTR (sorbent) were used to make inferenc es about the surface prope rties of the sorbent. Figure 8-1. Schematic diagram of column, thermistor s, and calibrating resistors as used in flow calorimetry 132

PAGE 133

Operation Initial amount of 15 mg of Al -WTR (dry mass) was packed into the column; however, the calorimetric signal response, repr esenting the peak area was very small. An increased amount to about ~100 mg was later packed into the column. The Al-WTR is located in the column as in Figure 3-1 situated between the two thermistors. Prior to each experiment, 100 mg of Al-WTR wa s placed in a 5mL vial and soaked in pH 5.8 solution of NaOAc/HOAc, unbuffered salt solu tions equilibrium with atmospheric CO2. This was done to expedite equilibration, since Al-WTR original pH is above 6.5. Solutions used in flow calorim etry consisted of 50 mM NaNO3, 50 mM NaCl. Phosphate solution was prepared as 1.0 mM PO4 in 49 mM NaNO3. It was prepared from a 50 mM PO4 stock solution that consisted of NaH2PO4 and Na2HPO4 salts in the proportions needed to give a final solution pH of 5.8. One milliliter of the PO4 stock solution was tr ansferred to 50 mL volumetric flask, and was brought to volume with 50 mM NaNO3. Solutions containing the known reactant species react with the sorbent are allowed to pass through the sample by a 100 cm water pressu re head. Flow rate of ~ 0.36 mL min-1 is maintained to ensure uniform peak areas. This is done by measuring amount of leachate per unit time intervals. Figure 8-2 shows peak area generated by 30 mJ heat pulses as a function of flow rate. According to Rhue et al., (2002), peak area is affected by the flow rate. This dependency was taken into account when comparing heat data obtained at different fl ow rates by applying a correction factor that was based on linear relationship. As noted by Kabengi et al., (2004), heat pulses for calibrating the instrument generally rang ed from about 5 mJ to more than 100 mJ in size, corresponding to about 2 to 45 s for energizing the calibrating resistor. Precision for 133

PAGE 134

replicated heat pulses was acceptabl e within 5% coefficient of va riation. In addition, flow rate precisions obtained were also within 1-10 %. Figure 8-2. A linear curve depicting the relati onship between peak area and flow rate Ion Exchange The flow calorimeter used in this experiment is the same as used by Harvey and Rhue, 2008. Heats of exchange for Cl/NO3 were obtained using the step mode. In this method, a baseline was first obtained using NaNO3. The solution was then switched to NaCl, which resulted in an endothermic heat associated with Cldisplacing exchangeable NO3 -. When the signal returned to baseline, the so lution was switched back to NaNO3, which resulted in an exothermic heat associated with NO3 displacing exchangeable Cl-. The heats of Cl/NO3 exchange were calculated, through integration of the Cland NO3 peaks and converting the areas of the peaks to heat units using a calibration curve. Several cycles of Cl/NO3 exchange were recorded to obtain a replicated data. The Cl/NO3 exchange was measured before a nd after P treatment. The P treatment consisted of injecting 1 mM PO4 in 49mM (NaNO3/NaCl) as background electrolyte solution. 134

PAGE 135

The pH of the reacting solution was 5.8. The result of exchange prior to P treatment is called the pre-P treatment. Whereas, the Cl/NO3 exchange after pre-P treatment is the post-P-treatment. Surface Charge A sample of 30 to 50 mg Al-WTR was placed in a cell and Cl-saturated by passing 50 mM NaCl through the cell. Excess solution was blow n out of the cell and th e cell was weighed to determine the amount of entrained NaCl solution. The entrained NaCl and exchangeable Cl were displaced using 50 mM NaNO3. A Cl electrode placed in the effluent stream was used to quantify the total amount of Cl displaced from the cell. The el ectrode response was calibrated using known amounts of 50 mM NaCl placed in a cell and displacing the NaCl with NaNO3. Exchangeable Cl was calculated by subtracting en trained Cl from the total and this value was taken as equivalent to surface positive charge. Th e Boehmite was re-saturated with NaCl and the process was repeated to obtain replic ated measurements of surface charge. Results and Discussion Ion Exchange/Surface Charge Figure 8-3 shows the results for anion exchange, NO3 replacing chloride (NO3 -/Cl-) and chloride replacing nitrate (Cl-/NO3 -) on Al-WTR. The two peaks (exothermic and endothermic) are expected to be about equal in area as known for reversible ion-exchan ge reaction. Further, the reactions observed were rapid and took about ~ 10-12 min for the signal to return to baseline, indication the end of th e reaction. Several runs were obtaine d to obtained reproducible data over time. In addition, the quantity of heat associat ed with anion exchange remains approximately constant. 135

PAGE 136

-0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25 0510152025Time (min)Calorimetric response (Volts) Cl/NO3 NO3/Cl Figure 8-3. Calorimetric response fo r nitrate replacing chloride (NO3 -/Cl-) and chloride replacing nitrate (Cl-/NO3 -) on Al-WTR. Figure 8-4 shows the effects of PO4 sorption on the calorimetric response of (NO3 -/Cl-) exchange at pH of 5.8. The data suggest that, P tr eatment reduced the number of exchange sites. This is attributed to the fact that peaks for anion-exchange, prior to PO4 treatment were higher than that of the post-PO4 treatment. Thus, indica ting that sorption of PO4 to Al-WTR resulted in a change in anion-exchange site. Further, no change in post PO4 anion exchange areas was observed during multiple (NO3/Cl-) exchange cycles, indicating th at the change made to the anion exchange sites by PO4 sorption was irreversible. These obs ervations were consistent with those of Harvey and Rhue, (2008) when PO4 was sorbed by Al-hydr(oxides). In addition, Kabengi et al., 2006b also reported similar obser vations of arsenate on Al-hydr(oxides). 136

PAGE 137

0.00 0.05 0.10 0.15 0.20 0.25 0.30 01 02 0 Time (min)Calorimetric response (Volts)3 0 Cl-/NO3-before P Cl-/NO3after P Figure 8-4. Exothermic calorimetric respons e for nitrate replacing chloride (NO3 -/Cl-) before and after PO4 treatment on Al-WTR. Surface charge method was used to calculate anion exchange capacity (AEC) on Al-WTR. The values was found to be ~ 1.5 cmol(+) kg-1 ( 0.7 Std.) after 12 replications were run. The results indicate low number of exchange si tes for anion sorption on Al-WTR. Thus, flow calorimetry data may also shed information to support data of low mi cropores (physisorption analysis), in revealing low exchange sites for anion sorption such as PO4 on Al-WTR. Although, quantitatively the two values are not the same, theoretically, the values are all pinpointing sites for sorption. Pores size analysis in previous chapter showed micropores constitute < ~28%. Microporosity is the seat for ad sorption. It is against this back ground that co-blending techniques was designed to optimize the reaction rate and to catalyze rapid PO4 sorption by making use of different reaction rates of other so rbents together with Al-WTR. Microcalorimetry Analysis Microscal limited, USA is a company that manufactures flow microcalorimeters. The company has several types of fl ow microcalorimeters. Out of curiosity and in terest, and in advancing future work in surface chemistry, and to compare previous energetic data. Samples of 137

PAGE 138

Boehmite (~ 200 m2 g-1), obtained from Sasol North America Inc, Houston, TX and Al-WTR as used in previous chapters were analysed by fl ow adsorption microcalorimetry by Microscal LTD (Groszek, 1999). A similar mass of the sorben ts was used, Al-WTR, ~ 0.167g and Boehmite, ~ 0.147g. A 0.05mM NaCl and NaNO3 cycles were passed through th e samples to determine the corresponding ion exchange and en ergetics of these sorbents. Flow rate was set to 6 mL hr-1 which is less than a third of the flow rate used in the previous study. Exothermic and endothermic reactions obtained in NO3/Cl and Cl/NO3 cycle in Figure 8-3 for Al-WTR were consistent with data from (Harvey and Rhue, 2008). Microscal results are presented in Figure 8-5 for Al-WTR. For exothermic reactions, energy re leased from Al-WTR for first cycling was 209.3mJ g-1, whereas that of Boehmite (Figure 8-6) was 616.8mJ g-1. Based on the measured anion exchange capacities i.e. 1.5 cmol(+) kg-1 for Al-WTR, and 30 cmol(+) kg-1 for Boehmite, one would have expected these two heats of excha nge to differ by a factor of 20 instead of the observed factor of 3. Subsequent second a nd third cycles (NO3/Cl) exchange shows that exchange on Boehmite rapidly reached equilibrium. Acco rding to McBride, (1994), ion exchange is a reversible process. Thus, suggesting that Boehmite can fully exchange anions of NO3 and Cl (Figure 8-6). The reversibility of anions exchange on Boehmite, an Al hydr(oxide), was consistent with that of Kabengi et al., 2005. This was suggested, to occur on protonated surface hydroxyls of Al hydroxides in aqueous solution. Thus, the reaction l eads to the non-specific adsorbed anions such as NO3 and Cl to interact directly with the surface hydroxyl groups. Conversely, for Al-WTR subsequent second and third cycles of NO3/Cl exchange show that exchange equilibrium was not achieved af ter three cycles. For example, the first NO3 exotherm registered 209mJ g-1, the Cl, 72mJ g-1. In the second cycle, the NO3 exotherm was 138

PAGE 139

59mJ g-1 the Cl, 42 mJ g-1. And in the third cycle, NO3 was 50mJ g-1 and Cl was only 38mJ g-1. Microscal Ltd attributed these changes to irreversible sorption of NO3 and Cl. However, an equal likely explanation is that the Al-WTR undergoes some type of physical/chemical changes when immersed in aqueous solution and that more than three NO3/Cl exchange cycles are required before exchange equilibrium is achieved. The Al-WTR is not considered as a pure hydroxide, but has other impure substances such as polymers, carbon, and other minute trace elements (Makris, 2004). These impurities may have hindered the free formation of non-specific, positive charges by the surface hydroxyl groups on Al-W TR during the aging process. Thus, with three different sets of analys es conducted on Al-WTR i.e. using BET-N2, with microporosity ~28%, flow calorimetry (Rhue et al., 2002) and flow adsorption microcalorimetry (Microscal Ltd) suggested clearly, Al -WTR has limited ion exchange (1.5cmol(+) kg-1).and low number of protonated surface hydroxyl for ligand ex change with respect to P sorption. If ion exchange is limited, it suggests that PO4 sorption by ligand exchange may be limited. However, as observed from equilibrium reaction of PO4 on Al-WTR, P was reasonably sorbed. Thus, the proposed reaction mechanisms of PO4 to an Al-WTR may be dom inated by precipitation or chemisorption reactions at sites othe r than the protonated surface hydroxyls. One main advantage of using flow calorimetry in sorption experiments is that, the sorbent is a stationary phase while the co ntaminant is in the mobile phase. Such a scenario is what takes place in the soil. The soil is stationary a nd contaminant moves through. Thus, the kinetics process can easily be modeled to mimic closely sorption as it pertains to field conditions. Conditions/parameters such as flow rates, ionic st rength, mass of sorbents the sorbents carrying capacity, as well the hydraulic prope rties of the sorbent, can be simulated to reflect the dynamics. This dynamic simulation would lead to prediction of the removal of the contaminant, once a 139

PAGE 140

suitable mathematical equation can represent the gi ven scenarios. Therefore, analysis of data from flow calorimetry can provide useful mean s by which sorption characteristics of materials can be probed, as well as modeled. A mathematical relationship, based on the parameters of flow calorimetry would be discusse d in subsequent chapter. Figure 8-5. Microcalorimetric response of nitrate replacing chloride (NO3/Cl) and (Cl/NO3) for Al-WTR. 140

PAGE 141

Figure 8-6. Microcalorimetric response of nitrate replacing chloride (NO3/Cl) and (Cl/NO3) for Boehmite. Conclusion In conclusion, flow calorimetry provided a dditional supporting information on the nature of sorption by Al-WTR. It can be inferred from the above experimentatio n that Al-WTR showed little anion exchange capacity. Th e AEC was measured at 1.5cmol (+) kg-1. Al-WTR had ~28% of micropores for adsorption. The microporosity da ta also suggest low sites for adsorption. Micropores are the seat for adsorp tion for highly porous sorbents. N eedless to say that, little ionexchange sites do not mean there are no exchange sites. There are exchange sites, but they are limited in number. Comparatively, Boehmite a pure Al hydro(oxide) exhibits greater ion exchange than Al-WTR, about 20 times greater. 141

PAGE 142

CHAPTER 9 SIMULATION MODELING OF P DYNAMICS IN FLOW CALORIMETRY Introduction The practice of simulating contaminant rem oval or uptake by solid-liquid (soil-water) interface is complex to illustrate. To study such a system in its entire ly in a traditionally designed experiment is a difficult task. Howeve r, it is theoretically possible to build a mathematical model that describes the system. Th e parameters in the model can be manipulated to predict the outcome of changing one or more of the input variables of a system. The practice of system analysis and manipul ation requires a comprehensive a nd quantitative description of the system before a system can be built (Smith, 1982) However, with the advent of computers and associated softwares, renew the interest in ma thematical techniques that previously had been impossible to use. Further, development of general system theory by Forrester, 1961; Van Dyne and Abramsky, 1975 have had great influence on biological simulations and also due to development of dynamic and stocha stic simulation techniques. Flow calorimetry used in the preceding se ction was designed to provide information on heat of reactions on sorbents, and assessing the chemistry of surfaces with incoming interacting molecules. It is ideally suitable for measuring of reactions occurring at liquid/solid interface. The process of sorbents interacting with liquid interface as in flow calorimetry is analogous to many natural occurring systems e.g. wastewater treatments, and in leaching of contaminants in soils. This implies that simulating and modeling P tr ansport as it reacts with sorbent in flow calorimetry would provide useful information on the dynamics of P as well as the state of the sorbents with respect to time. The nature of such simulations in flow calorimetry could be extrapolated to contaminants transport in soil systems. 142

PAGE 143

Flow calorimetry is a continuous flow proce ss, where reactions occu rring at the surface can be tracked with time. It consis ts of a column that is placed inside a water bath that provides a good thermal stability agai nst temperature changes. Within th e column are two thermistors that sense changes in the solution. A change in solu tion temperature produced a differential output voltage that is fed into an instrumentation amplif ier, and the amplified signal is transmitted into a computer for processing (Rhue, et al., 2002; Kabengi et al., 2006). P over applications has been a concern causing eutrophication problems as well as poses a threat to human and ecological health. To at tenuate the transport of P, several chemical amendments containing Fe, Al and Ca have been investigated. Sorbents such as Goethite, Boehmite, Fe-humate, and Al-WTR can be used for P immobilization. This is because these sorbents have high P sorption capacities. Batch experiments have been used extensively to determine sorption maximum or sorption capacity of sorbents. This entails determination of multipoint isotherms for P sorption capacity (Bmax), and is a laborious procedure. Data from Bmax studies provide information on, an overall amount of P the material can sorb. On the other hand, information regarding the time to reach maximum sorption is generally lacking. Another separate study involving kinetics or rates has to be conducted to gain such informati on. Despite batch limitations, its simplicity and repeatability makes it the mo st common approach in sorp tion studies (Sparks, 1989). A sorbent containing metal oxides or hydroxi des of Fe or Al (e.g. Al-WTR) can be placed within the column. A known P containing solution is allowed to pass through the column by injection/continuous flow. Any reacti on occurring between the sorbents and PO4 is recorded as peaks. An integral of the area under the peaks can be used to calculate the sorption capacity of adsorbent. Flow rates are controlled and leachat es are collected for phosphorous analysis. In 143

PAGE 144

addition, flow calorimeters are well suited for measur ing interactions at the liquid/solid interface. Data on surface reactions can be tracked and mode led. However, currently little work has been done on modeling PO4 reactions on adsorbent, e.g. Al-WTR/Boehmite under flow calorimetry. Further, little work is done on modeling and simulation of PO4 sorption reactions with FC, given a specific sorbent. It is against this background the study formulates a mathematical description to mimic the dynamics of PO4 reaction utilizing a continuous flow reactor description. Furthermore, there appears to be no study on comparing two parameters simultaneously that affects sorption of P at the same time. Hypotheses i Formulating a mathematical model would quantitatively describe the P transport dynamics under flow calorimetry. ii Simulation of the model would generate an output response. To test these hypotheses, the objectives of the study were: Objectives i To simulate the dynamics of PO4 sorption on a sorbent (Al-WTR) under flow calorimetry, leading to prediction of P removal. ii To determine the sensitivities of parameters influencing sorption reaction (rate and sorption capacity concurrently). The Role of Simulation Modeling Models are basically hypotheses and simply representation of reality and no model is 100% perfect. However, models are useful tools not because they reproduce reality, but because they simplify reality and enable the most important processes to be identified, studied, and simulated, so that outcomes can be predicted in advance (Addiscott, 1993). Further, a model when used appropriately allows extrapolation of data to reduc e repetitive and time-consuming processes. In this study, the use of computer is employed in simu lation the model with the aid of 144

PAGE 145

numerical techniques. Numerical approach wa s used because of the nonlinear differential equations involved. Further, due to simulation of a model, assessm ent of uncertainties in the models response can be evaluated. The response can be evaluated based on the parameters available. Verification and validation of the true parameters, can lead to real assessment of the model. Materials and Methods System Description and Operation Flow calorimetry consists of a column that is placed inside a water bath to provide a good thermal stability against temperature changes. With in the column are two thermistors that sensed changes in the solution. An adsorbent material is placed in between the thermistors. A change in solution temperature produces a differential output voltage that is fed into an instrumentation amplifier, and the amplified signal is transmitted into a computer for processing (Chapter 8). Approximately 100mg of adsorbent material was plac ed inside the column and solutions containing reactive species (P) we re injected through the column. Flow rates were controlled with a precision needle valve. Run time vari es between 15 to 25 minutes depending on the time required for the signal to return to the baseline (Figure 9-1). P eaks were obtained by integrating the signal (volts) numeri cally over time. This time-average pe ak area (V min) was converted to a flow rate-average peak area (V ml) by multiplying by the average flow rate. This was measured for each peak by collecting effluent volume and dividing by the time over which the volume was collected. Further, peaks areas can be converted into energy units (Joules) by comparison with peaks generated with a calibrating resist or located within the flow stream. The input is P solution concentration of 1mM ( 32 mg P), with Flow rate set to 0.36 ml min-1. Output concentration of phosphorous depends on the adsorbents carrying capacity, the rate of sorption, and the mass of the adsorbent. As more P is loaded onto the surface exchange 145

PAGE 146

site, the adsorption density/the capacity to sorb d ecreases. This system behavior can be observed as the area of the peaks decreases, until it fina lly reached the base line. Also, the rate of P sorption affects P loading unto the surface or intern al micropores, depending on the nature of the materials. Flow Calorimetry Figure 9-1. An image of Flow Calorimetry set up, i ndicating sections of P injection (syringe), a monitor recording signal effects and leachate collection area. A solution containing P is allowed to pass through a packed column. The differential change and transport of P (rate of change w ith time) through the column can be described mathematically. Output fluid passing through the column is collected and analyzed. Model Development A Forrester diagram of the set up of the m odel is represented in (Figure 9-2.). The objective of using the Forrester diagram is to translate all ideas and concepts into mathematical equations. The law of mass transfer and mass balanced were assumed. Auxiliary variables ( Bmax and r ) were also monitored at each compartmen t and attached to each state variable. A mathematical model in the form of differential equation was derived: 146

PAGE 147

rC B B CC xAH F dt dCti ti titi w ti 1, max 1, 1,1, ,]1[] [: (Eq. 9-1), where; i =1, Ci-1 = Cin (i.e. input) for initial first cell: C = concentration (mM): Bi,t-1 = sorption with time: F = flow rate (ml min-1): Hw = hydraulic property: A = area (cm3): x = total distance per cell (cm): r = rates (min-1): Bmax = sorption capacity (g kg-1). The column/sorbent was subdivided into ten compartment or cells as state variables i.e. C1 through C10. The flow of P as it passes through each cell is modeled. The time ( t ) it took each cell to reach saturation with P is noted with respect to the parameters in question. Figure 9-2. Forrester diagram of P transport th rough compartmented cells in Flow calorimetry. Parameters used in first simulation are as follows: Initial input concentration = 1mM, sorption capacity of the sorbent = 2g kg-1, mass of Al-WTR = 100 mg, rate r = 8 min-1, t = 0.025 min, flow rate = 0.36 ml min-1, volume estimated = 0.157 cm3 and hydraulic property = 0.58. The parameters in the model can be changed, especially Bmax, r of the sorbent and initial concentration of the species to be removed. The simulation-m odel output changes accordingly reflecting the robustness of the model to stim ulus (Figure 9-3, Figur e 9-4, and Figure 9-5). 147

PAGE 148

The basic idea is, as P is transported th rough cell #1, portion of the concentration is adsorbed within that compartment. The remaini ng unadsorbed P moves as output to the next cell #2. The output of cell #1 becomes the input for cell #2. The scenario go es on till P solution reaches the final compartment, cell #10. The above description is mathematically formulated in differential equations, which is solved numerically. Numerical Equations Let C1, C2.,..and C10 to represent concentrations of input and outputs, t as time step via discretization process. Hence Ci,t represents the t-th time step for Ci ( i-th concentration out). Let h be the actual time interval. Using Eulers implementation, the following derivations were obtained. )1(maxB B HxArC dt dBi w i i; With i= 1,t 0 (Eq. 9-2), )1( (max 1 1 ,11,1B B HxArChBBw t t (Eq. 9-3), rC B B CC xAH F dt dCti ti titi w ti 1, max 1, 1,1, ,]1[] [ (Eq. 9-4) Given i= 1, t 0, Ci-1= Cin ])1()( [,1 max ,1 ,1 ,11,1rC B B CC xAH F hCCt t tin w t t (Eq. 9-5) The above notation and calculation is for i =1, however for subsequent calculations of Ci,t and Bi,t for all i 1 and t 0. Results and Discussion The numerical approach of Eulers method was us ed to generate the simulated results as in Figure 4-3, Figure 4-4, and Figure 4-5. The results showed P sorb ed with respect to time. Maximum P sorbed was ~2.0 mg P cell-1 which corresponded to exactly 2.0 mg P cell-1 as input sorption capacity per cell. 148

PAGE 149

0 0.5 1 1.5 2 01020304050Time(t) minP Sorbed mgP/cell Figure 9-3. Simulating amount of P sorbed in the first cell, depicting maximum sorption capacity as a function of time. 0 0.5 1 1.5 2 020406080Time (t) minAmount Sorbed (mgP/cell ) B-1 B-2 B-3 B-4 B-5 B-6 B-7 B-8 B-9 B-10 Figure 9-4. Simulation of maximum sorption fo r segmented cells with of ~2.0 mg P cell-1 maximum vs time. 149

PAGE 150

0 0.2 0.4 0.6 0.8 1 01020304050607080Time (t) minConcentration (mg P/cm3 C2 C3 C4 C5 C6 C7 C8 C9 C10 C1 Figure 9-5. Simulation of concentr ation as it flows through segmente d cells to final cell vs time. Figure 9-3 showed simulation of sorbent in the first segmented cell. It shows clearly, the sorbent sorption capacity of 2.0 mg P has been r eached in the first cell in about ~20 minutes. A subsequent cell reaching sorption capacity is depi cted in Figure 9-4. Thus, the time to reach saturation of the sorbent can be quantified. In this simulation, with the specific parameters mentioned above, time of saturation of the sorbent is ~70 minutes. In Figure 9-5, concentration (1 mM) of the reac ting species is at unifo rm flow rate of 0.36 ml min-1 through each cell in the sorbent. At time zero, ~0.3mgP cm-3 was removed instantly. This might be attributable to diffusion and due to different interface of wetting fronts of a liquid meeting a porous solid material instantly. Afterw ards, instant uptake decreases and begins to build up. Subsequent cell #2, had ~0.1mgP cm-3 instantly removed, leading to a drop of 0.2 in comparison to the first cell. Further, a cell #3 had about ~0.5mgP cm-3 instantly, whereas, the cell# 4 upwards start from zero, indicating the wetting front has been eliminated and smooth passage takes place in the column without resi stance, up to total concentration of 1mM. 150

PAGE 151

At ~30 min, it is observed from the simu lation, no solution was flowing in cell#10. Solution begins to reach and accumulate in the ce ll #10 after ~35 min, and then build up till the cell is fully saturated. The scenario of the simula tion is in accordance with what is expected to occur. However, measured concentrations in solution should be matched with predicted simulated values to calculate th e uncertainties in the model. A second simulation is run with changes in parameter values as follows: Initial input concentration = 1mM, Bmax (sorption capacity of the sorbent) = 0.5g kg-1, mass of Al-WTR = 100 mg, rate r = 8 min-1, t = 0.025 min, flow rate = 0.36 ml min-1, volume estimated = 0.157 cm3 and hydraulic property = 0.58. The parameters in the model can be changed, especially B max, r of the sorbent and initial concentration of the species to be removed. The simulationmodel output changes accordingly reflecting the robustness of the model to stimulus (Figure 9-6, and Figure 9-7). The output of the second simulation showed similar trend as indicated above (Figure 9-6). The difference lies in the maximum inputs give n. The amount sorbed was 0.5 mg P, which was reflected in the Figure 9-6. Further, the time of reaching sorption saturation was ~20 min, a reduction ~70 min as in Figure 94. The simulation showed that with the required given inputs, the output can be predicted. The simulation of concentration of the reacting species flowing through the column/sorbent is shown in Figure 9-7. Ma ximum concentration given was ~1 mM. This implies that each cell can only remove up to 1mM, which was depicted in each cell taking up to 1mM. Further, the sorption capacity of each ce ll and the concentratio ns simulation occurs simultaneously. Thus, it was observed that the time of saturation was ba sically the same ~20 min. 151

PAGE 152

0 0.5 1 020406080Time (t) minAmount Sorbed (mgP/cell ) B-1 B-2 B-3 B-4 B-5 B-6 B-7 B-8 B-9 B-10 Figure 9-6. Simulation of maximum sorption fo r segmented cells with of ~ 0.5mgP cell-1 maximum vs time 0 0.2 0.4 0.6 0.8 1 020406080 Time (t) minConcentration (mg P/cm3) C2 C3 C4 C5 C6 C7 C8 C9 C10 C1 Figure 9-7. Simulation of concentr ation as it flows through segmen ted cells to a final cell vs time. Sensitivity Analysis Sensitivity analysis refers to techniques or processes by which the impacts of parameters or inputs are evaluated with regards to their effects on changes on the model or simulated results. The basic reasons in performing sensitivity analysis are: (i). Testing which parameters dominate a certain response in order to elim inate insensitive parameter. (ii) Testing where additional effort should be placed to reduce uncertainty. 152

PAGE 153

A parameter at a time perturbation in whic h individual parameters are varied using a certain step size and the impact of this varia tion is measured based on the objective function. This approach, although is simple, is unreliable for high-dimensional and non-linear models as in environmental or hydrological models (Wagener a nd Kollat, 2006). It is against this background that Monte Carlo analysis and re lative sensitivity was used, sin ce the model is a non-linear one. A Monte Carlo is a type of global sensitivity analysis, in which sampling of N points are drawn from multivariate uniform distribution. Th e procedure of Monte Carlo utilizes random variables and probabilities to determ ine the parameters of significant in fluence. It is often used in methodological investigation of the performan ce of statistical estim ators under various conditions. The simulation utilizes pseudo num bers generated in Excel function =RAND(). When this function is entered in spreadsheet cells, it generate s a uniformly distributed pseudo random numbers between 0 and 1. Results from the Monte Carlo analys is (data not shown) suggest that the parameter r is more sensitive than Bmax. A second sensitivity test was run using relative sensitivity, r( y | k ) which is often used to provide a normalized measure for comparing the sensitivity of a model to several variables (Jones and Luyten, 1998). It is defined as: r( y |k ) = kdk ydy= y k ky )|(, where output = y and k is a variable. (Eq. 9-6) The parameters compared were Bmax and r. Over several simu lations (> 1000) were run generating values of r outputs at fixed Bmax was 182.92 by calculation, whereas values of Bmax outputs at fixed r was .81. From relative sensitivity analys is calculation, it was observed that parameter r was most significant in in fluencing sorption through mathematical model. The implication is that the parameter r should have more attentio n during P sorption reaction. 153

PAGE 154

Practical intuitive meaning is that kinetic reac tions are perhaps one of the most important processes controlling sorption due to their dynamic nature as depi cted in the model equations. Conclusions The simulation outputs suggest clearly the model is responding to input stimulus. Simulated data of the model showed the dynamics of P sorption with respect to a particular sorbent e.g. (Al-WTR). The results show the maximum time required for a sorbent to reach saturation. Further, the model can be used to co mpare the behavior of two different sorbents. Common application of simulati on-model derived above is, for wastewater treatment using activated carbon, removal of contaminant in so ils using porous materials, and as well for drinking water treatment industries. The simulatio n model would be of pa rticular interest to researchers as well to regulators in knowing the time frame for a particular contaminant. Sensitivity analysis performed using both Mont e Carlo and relative sensitivity methods suggest that, the parameter most influential in sorption is the rate. This further buttresses the fact that in dynamic systems, kinetic is perhaps th e most important processes governing sorption reactions and should not be neglected. Validation of Simulation Mo del with Column Studies An application of the model is flow of leachates as usually determined by column studies. Parameter of initial concentrations ( C i) of P, from the control w ithout any amendment was used. Control concentration changes with time on weekly basis. Bmax was estimated to ~2-13.6 g kg-1 taking into consideration the volume of the soil used, the mass of the soil, the mass of Al-WTR added, and some sorption by the soil itself. An average value of 5.7 g kg-1 was used. Rate ( r ) used ~0.03 min-1 since the soil has lots of organics whic h retards the sorption rate. Flow rate was ~0.6 ml min-1. For easy computations, segmented cel ls were assumed as one. Time was evaluated for 7d. Result of the measured values vs predicted values are plotted as in Figure 9-8. 154

PAGE 155

Regression coefficient of determination (R2) for the measured vs the predicted was 90% for six weeks of leaching data. Also, the ratio of measured values to estimated values was closed to 1, which suggests that P coming out from the columns can be predicted by the model using the parameters mentioned above. Model efficiency (M E) were calculated to be ~ 0.86, a value close to 1 suggesting the predictive nature of the m odel. The model efficiency equation used is: N i i N i iiYY YY ME1 2 1 2)( ) ( 1 (Eq. 9-7), where Yi is observed values, the predicted values, and iY Y is the mean of Yi values. To ascertain the effects of other parameters, whic h might be having an in teractive effect on each other, residual plots vs predicted values were obtained (Figure 9-9). The plots showed a scatter points, indicating no single parameter is having an influence on the model. Conclusions It reality, the Flow calorimeter should be used to validate th e model. However, due to the long duration of time it takes to reach full saturation of the sorbents, column a study was used instead. Other independent data from the literatur e could also be used to validate the model. However, searching the literature revealed that experimental da ta available were poorly designed and method of recording was without derivatives of important parameters. Future efforts are needed in validation and calibrati on of the model on field scale with respect to a particular soil type and a named contaminant to be removed. Fu rther, the simple spreadsheet mathematical model provides useful tools for predicting contaminan ts, in which time is critical and need to be known, and also, the model can predict the time to remove the contaminant with a given sorbent. 155

PAGE 156

156 0 5 10 15 20 25 0510152025Predicted Measured R2 =0.9 Figure 9-8. The measured P values vs predicted valu es by the model as used in column leachates. -2 -1 0 1 2 3 0 5 10 15 20Residual ErrorPredictedFigure 9-9. Plots of residual erro r vs predicted values on the model for column leachate.

PAGE 157

Developing Decision Support for Contamin ant Removal in Soil-Aqueous Systems Most researches are funded from agencies that want end produc t of the research to help address real life problems. The main traditiona l way of presenting information to decision makers has been through reports or journal articles. The main pr oblems however is, the scientific language use creates problems and secondly info rmation presented are static and not in a dynamic way for funding agencies or policy agency to update or upgrade the information. Consequently, there is a dichot omy between research outputs a nd end users/decision makers. Many of the problems the decision makers hope to address are not the understanding of the mathematical models, but rather a simple solution without the details of complexities of science and mathematics. Most probably what decision makers are yearni ng for is a simple graphical user interface (GUI) that is well packaged in a friendly manner to use. Decision makers want to quickly see the results in graphs or chart or other means, and also m odify inputs to analyze different scenarios (Hanna, 2004). The desire end of scientific re search should be well presented in a simple format to help pull data from large databases and when appropr iately manipulated should yield results. This way of presenting scientific environmental data is a widely problem that is not addressed in most science-based curriculum. The writer has some basic background in educational curriculum development and rudimentary information on progra mming. The main objective in this chapter is to point the way forward for trai ning new scientists in informati on presenting, to be in a manner for decision makers. Secondly to use the model developed in chapter 4, to prove a simple decision support tools for model users. 157

PAGE 158

Defining Decision Support Systems Decision support system (DSS) is a model-base d or knowledge-based system intended to support policy decision-making in a semi-structu red or unstructured situations (Turban and Aronson, 2001). A DSS uses data, provides a clea r user interface, and can incorporate the decision makers own insights (Seref et al., 2007). The main characteristic of a DSS is that it supports but not replaces decision process. It must be capable of analyzing complicated issues and relationship in a manner that can be a pplied directly by the decision maker. A DSS application exhibits certain pr operties and characteristics as: The structure and environment of the problem is not rigid but flexible to change. The exchange between the user and the computer is interactive in nature. The system provides the user with the capabilit y to examine different situations, evaluate various scenarios, and answer a variety of what if questions. The user is afforded the flexibility to adap t the system to his pr eferences (Davis, 1988). Microsoft excel (MSE), is by far the most practical and widely used software, among scientists, engineers and decisi on makers. MSE allows for data storage and model building. MSE also has many in built programs as well as ma ny add-on programs available that allows for simulation of various models. Excel has a m acro programming language, Visual Basic for applications (VBA). VBA allows building of fr iendly GUIs and manipulations of excel objects. Thus, MSE provides a suitable platform for fair ly easy applications of DSS (Chapra, 2003; Hanna, 2004). The main question however is, assembli ng these tools together in a format that is friendly to transmit information and at the same tim e retrieve feedbacks. By this way the user and the data can communicate information that may be latent but provide good feedback to decision makers. 158

PAGE 159

Excel Applications and Illustrations for DSS DSS application is provided for the model in Chapter 4. The DSS indicated below is an attempt to show how easily information can be presented to decision makers rather than the conventional reports and journal articles used by research scientists. The model was coded in VBA with macro and creations of user control form s. Labels were attached to the user control forms. Possible questions that would likely be raised by users of the models are 1. Which sorbents is more effective in removing a particul ar contaminant? 2. What time is required to remove all contaminants? 3. What level of contam inant would remain; say after 2 yrs application of sorbent A? Questions of this nature would not suffice in scien tific publications in relation to the model. Based on the above questions, a DSS is developed using Microsoft excel. Screen shot of the segmented tools of the support is presented below. Figure 9-10. A screen shot of initial st art and data input ba se for sorbents. Data input for sorbents is a storage house for naming the sorbents and the corresponding parameters (i.e. sorption capacity (mg kg-1)) and the rate (min-1). These parameters are linked to 159

PAGE 160

the model equations for manipulations, by clicki ng the button. Figure 9-11 indicates the screen view of the database for the parameters. Figure 9-11. A screen shot of the parameters (sorbents, sorption capaci ty, and rate) columns attached to sorbent data input. In Figure 9-11, after in-putting the data for a pa rticular sorbents, there is a button created on the right to return the screen to the front view to i ndicate start. The start button leads to a user control form with label as sorb ent. A named sorbent must be selected. A label with precision indicates the decimal units you want your final result. The smalle r the decimal unit, the quicker the computational result would be calculated. The next button from the start toolbar is the column for the initial concentration of the contaminant (P), input values in mg L-1 is required. Number of steps indicates the time components in minutes. Compute time steps indi cate the time the user re quired the contaminant to be removed. And compute concentration indi cates the amount of the contaminant remove within the time steps indicated previously (Figure 9-12). 160

PAGE 161

Figure 9-12. A user-interface on start button showing sorbent selection, precision, initial concentration, number of steps, computes step/concentration and show results buttons. At each of the computational levels for step s/concentration, the user can click on show results. It would show clearly the types of sorbents inputs and the time and concentration level the user desires the contaminant to be removed. Final data output is pr esented in Figure 9-13. The screen shot shows examples of computation with respect to particular contaminants. The computation generated different time steps and levels of contaminant removal. Thus, the user/decision maker can make useful comparison, which is ot herwise difficult to see in 2dimensional charts as in reports or journal articles. Besides, th e user can update and request for different user interface tailored to suit his/her particular needs. In the final analysis, DSS provides information that is usually missing from reports and journal articles. In conclusion, excel through basic programmi ng can provide additional and easy to use GUI to serve as decision support tools for decision makers. It is hoped that, this approach would bridge the gap between policy institutions, funding ag encies and research scientists. 161

PAGE 162

Figure 9-13. User-interface showing the results of initial concentration and time steps and types of sorbent used and amount of contaminant removed. Conclusions DSS is a friendly user interface that can be bui lt to address real li fe questions faced by decision makers. It can be tailor ed to the desire of the user based on their needs. The DSS discussed above is to show what can be done to bridge the existing ga p between decision/policy makers or funding agencies and researchers. Data published in reports or journals have been shown to have little impacts unless, it is transl ated to public domain, otherwise it has become a mere academic exercise. The model developed for P removal in solution was linked to a deci sion support tool based on proposed questions. Which types of sorbents would likely remove the P at a faster rate? These are issues that decision makers or e ngineers may face when say arsenic/P has been contaminated with drinking water. How can it be removed quickly to save the community? DSS 162

PAGE 163

163 can help to arrive at decisions to complement other intermediates action for safe contaminant removal.

PAGE 164

CHAPTER 10 SUMMARY AND CONCLUSIONS A simple approach of remediation technol ogy of P immobilization, the so-called coblending technique has been found to rapidly and completely remove P in manure impacted soil. There is no study in the development of this technique for P immobilization. The technique utilized the properties of the sorb ents as well as the most major P species to be immobilized. The main advantage is that the technique can be tailore d for application of sorbents to remove what is most desired. Further, depending on the tailoring of the sorbent, pH of the environment may not be far different from the control. The use of geochemical models revealed major PO4 minerals. One of such mineral is struvite, which is of strategic importance in sust ainable re-use of the immobilized P. Future study is needed in this direction, to quantify how much struvite can be formed and when it would be released. A simple mathematical model is propose d. The model is the first of its kind in using simple spreadsheets for simulation of P dynamics. The model can predict time for P removal in a given system. Future research is, however, need ed in the application of the model under field condition. Based on the model, a decision support tool was also developed. The support tool would be of great help to policy/decision makers in selecting types of so rbents for water quality considerations. 164

PAGE 165

APPENDIX SUPPORTING DOCUMENTS Table A-1. Weekly meana pH data for sorbents amended and unamended soils. Sorbents amended Week 1 Week 2 Week 3 Week 4 Week 5 Control soil0% 7.71 8.11 7.91 7.94 8.26 Soil +MgO-2% 9.52 9.79 9.67 9.76 9.42 Soil +Slag-2% 8.45 8.73 8.56 8.73 8.35 Soil +Gypsum-2% 7.52 7.94 7.54 7.68 7.89 Soil +LimeKD-2% 8.08 8.01 7.95 8.21 8.14 Soil +Al-WTR-2% 7.74 8.12 7.95 8.04 8.49 Soil +WTR+MgO-(1+1%) 9.13 9.48 9.13 9.20 9.35 Soil +WTR+Slag-(1+1%) 8.18 8.48 8.35 8.46 8.53 Soil +WTR+Gypsum-(1+1%) 7.54 7.69 7.36 7.56 7.50 Soil+WTR+LimeKD-(1+1%) 8.03 8.22 7.88 8.25 7.93 aValues are means of triplicate. Table A-2. Weekly meana Eh (mV) data for sorbents amended and unamended soils. Sorbents amended Week 1 Week 2 Week 3 Week 4 Week 5 Control soil0% -46.33 -69.10 -64.67 -61.47 -64.07 Soil +MgO-2% -158.2 -166.23 -169.93 -167.7 -164.0 Soil +Slag-2% -103.9 -104.77 -103.63 -104.8 -102.6 Soil +Gypsum-2% -54.90 -59.20 -54.80 -45.87 -46.53 Soil +LimeKD-2% -66.60 -63.03 -67.60 -76.17 -72.87 Soil +Al-WTR-2% -48.73 -69.87 -67.43 -66.87 -70.13 Soil +WTR+MgO-(1+1%) N/A -148.73 -137.53 -133.3 -131.5 Soil +WTR+Slag-(1+1)% N/A -90.93 -91.27 -90.50 -89.07 Soil +WTR+Gypsum-(1+1%) N/A -45.67 -45.37 -38.23 -45.13 Soil +WTR+LimeKD-(1+1%) N/A -76.63 -72.07 -78.10 -71.07 aValues are means of triplicate. N/A not available. 165

PAGE 166

Table A-3. Cumulative P mass (mg) of column leaching study. Sorbents amended Week 1 Week 2 Week 3 Week 4 Week 5 Control Soil 2.64 7.79 13.51 18.87 23.71 Soil +MgO-2% 0.39 0.82 0.98 1.13 1.28 Soil +Slag-2% 0.07 0.41 0.51 0.62 0.74 Soil +Gypsum-2% 2.68 4.44 6.87 9.07 11.23 Soil +LimeKD-2% 0.36 0.70 0.77 0.88 1.03 Soil +Al-WTR-2% 0.79 2.31 4.11 5.88 7.66 Soil+WTR+MgO-(1+1%) 0.14 0.56 0.71 0.85 1.01 Soil +WTR+Slag-(1+1%) 0.02 0.47 0.68 0.94 1.20 Soil +WTR+Gypsum (1+1%) 1.50 3.22 4.79 6.41 7.88 Soil +WTR+LimeKD (1+1%) 0.00 0.36 0.47 0.66 0.88 Table A-4. Sequential extraction for soluble and extractable P (mg P kg-1). Soils extracted after 12 wks of leaching. Treatments Grouping Meana Control soil0% B 144.40 Soil+MgO2% G 45.00 Soil+Slag2% F 73.08 Soil+Gypsum2% A 167.84 Soil+LimeKD2% E 90.89 Soil+Al-WTR2% D 114.68 Al-WTR+MgO(1+1%) F 76.44 Al-WTR+Slag(1+1%) F 74.08 Al-WTR+Gypsum(1+1%) C 128.05 Al-WTR+LimeKD(1+1%) F 75.02 aValues are means of triplicate; means of th e same letter are not significantly different (LSD, = 0.05) 166

PAGE 167

Table A-5. Sequential extrac tion of Fe-Al bound P (mg P kg-1). Soils extracted after 12 wks of leaching. Treatments Grouping Meana Control soil0% B 501.72 Soil+MgO2% D 323.99 Soil+Slag2% E 241.56 Soil+Gypsum2% D 317.51 Soil+LimeKD2% F 80.93 Soil+Al-WTR2% A 662.06 Al-WTR+MgO(1+1%) D 323.38 Al-WTR+Slag(1+1%) C 363.54 Al-WTR+Gypsum(1+1%) B 522.60 Al-WTR+LimeKD(1+1%) D 312.79 aValues are means of triplicate; means of th e same letter are not significantly different (LSD, = 0.05) Table A-6. Sequential extrac tion of Ca-Mg bound P (mg P kg-1). Soils extracted after 12 wks of leaching. Treatments Grouping Meana Control soil0% F 61.72 Soil+MgO2% D 81.26 Soil+Slag2% BAC 101.01 Soil+Gypsum2% D 80.81 Soil+LimeKD2% BA 110.32 Soil+Al-WTR2% EF 65.63 Al-WTR+MgO(1+1%) DC 87.84 Al-WTR+Slag(1+1%) BC 98.09 Al-WTR+Gypsum(1+1%) ED 78.87 Al-WTR+LimeKD(1+1%) A 113.72 aValues are means of triplicate; means of th e same letter are not significantly different (LSD, = 0.05) 167

PAGE 168

LIST OF REFERENCES Addiscott, T.M. 1993. Simulation modelling and soil behaviour. Geoderma, 60:15-40 Adjei, M.B., and J.E. Rechcigl. 2004. Interact ive effect of lime and nitrogen on Bahiagrass pasture. Soil Crop Sci. Soc. Fla. Proc. 63:52-56. Agyin-Birikorang, S., G.A OConnor, L.W Jac obs, K.C Makris, and S.R Brinton, 2007. Longterm P immobilization by a drinking water tr eatment residual. J. Environ. Qual 36:316323. Altinbas, M., C. Yangin, and I. Ozturk. 2002. Struvite precipitation from anaerobically treated municipal and landfill wastewater s. Water Sci Technol. 46:271-278. Arambarri, P., O. Talibudeen. 1959. Factors influencing the isotopically exchangeable phosphate in soil. III: The effect of temperature in some calcareous soils. Plant and Soil II, 364 376. ASCE and AWWA. 1996. Management of water treatment plan t residuals. Washington, DC. ASCE and AWWA. Azizian, S. 2004. Kinetic models of sorption: a th eoretical analysis. J. Colloid Interface Sci. 276:47-52. Azizian, S. 2006. A novel and simple method for fi nding the heterogeneity of adsorbents on the basis of adsorption kinetic data. J. Colloid Interface Sci. 302: 76-81. Azizian, S., and B. Yahyaei. 2006. Adsorption of 18-crown-6 from aqueous solution on granular activated carbon: A kinetic modeling study. J. Colloid Interface Sci. 299:112-115 Bakri, A., L.H.M. Janssen and J. Wilting. 1988a. Fl ow microcalorimetry a pplied to the study of chemical stability of organic compounds. J. Thermal Anal. 33:1193-1199. Bakri, A., L.H.M. Janssen and J. Wilting. 1988b. Determination of reaction rate parameters using heat conduction microcalorimetry. J. Thermal Anal. 33:185-190. Barber, S.A. 1967. Limning materials and practices. P 125-160. In R.W. Pearson et al. (ed.) Soil acidity and liming. Agronomy monograph series no.12. Ameri can society of agronomy. Madison, Wisconsin. USA. Barret E. P., L.G. Joyner, and P.P. Halenda. 1951. The determination of pore volume and area distributions in porous substances. 1. Comput ations from nitrogen isotherms. J. Amer. Chem. Soc. 73:373-380 Barrett, E. P., L. G. Joyner and P. P. Halenda. 1951. The determination of pore volume and area distributions in porous substances .1. Computat ions from nitrogen isotherms. J. Am. Chem. Soc. 73: 373-380. 168

PAGE 169

Barrow, N. J., and T. C. Shaw. 1975. Slow reactions between soil and anions. 2. Effect of time and temperature on decrease in phosphate con centration in soil so lution. Soil Sci. 119.167177. Barrow, N. J., and T. C. Shaw.1975. Slow reactions between soil and anions 2. Effect of time temperature on decrease in phosphate concentration in soil soluti on. Soil Sci. 119:167-177. Bastin, O., F. Janssens, J. Dufey, and A. Peet ers. 1999. Phosphorus removal by a synthetic iron oxide-gypsum compound. Ecol. Eng.. 12:339-351. Battistoni, P., P. Pavan, F. Cecchi, and J. Mata-Alvarez. 1998. Phosphate removal in real anaerobic supernants: Modeling and performance of a flui dized bed reactor. Water Sci. Technol.38:275-283. Bayley, R.M, J.A. Ippolito, M.E. Stromberger, K.A. Barbarick, and M.W. Paschke. 2008. Water treatment residuals and bioso lids coapplications affect semiarid rangeland phosphorus cycling. Soil Sci. Soc. Am. J. 72:711-719. Bell, F.G. 1996. Lime stabilization of clay minerals and soils. Eng. Geol. 42: 223-237. Blanchard, G., M. Maunaye, and G. Martin. 198 4. Removal of heavy metals from waters by means of natural zeolites. Water Res.18: 1501-1507 Bohn, H.L., McNeal, B.L., and OConnor, G.A., 2001. Soil Chemistry, third ed. John Wiley and Sons, New York. Bolster, C. H, Hornberger, and G. M. 2007. On the use of linearized Langmuir equations. Soil Sci. Soc. Am. J. 71:1796-1806 Boynton, R.S. 1980. Chemistry and technology of lime and limestone. John Wiley and Sons. New York. Briggner, L. E., G. Buckton, K. Bystrom, and P. Darcy. 1994. The use of isothermal microcalorimetry in the study of changes in cr ystallinity induced duri ng the processing of powders. Int. J. Pharm. 105:125-135. Brodowski, S., W. Amelung, L. Haumaier, C. Abetz and W. Zech. 2005. Morphological and chemical properties of black carbon in physical soil fracti ons as revealed by scanning electron microscopy and energydispersive X-ray spectroscopy. 116-129. Elsevier Science Brunauer S., P.H. Emmett, and E. Teller (1938). Ad sorption of gases in multimolecular layer. J. Amer. Chem. Soc. 60:309-319. Brunauer, S., P. H. Emmett and E. Teller 1938. Ad sorption of gases in multimolecular layers. J. Am. Chem. Soc. 60:309-319. Burns, R.T, L.B. Moody, F.R. Walker, and D. R. Raman. 2001. Laboratory and in-site reductions of soluble phosphorus in swine waste slurries. Environ Technol. 22:1273-1278. 169

PAGE 170

Celen, I., J.R. Buchanan, R.T. Burns, R. B. Robinson, and D.R. Raman. 2007. Using a chemical equilibrium model to predict amendments require d to precipitate phosphorus as struvite in liquid swine manure. Water Res. 41:1689-1696. Chang, A.C., Page, A.L., Sutherland, F.H., and Grgurevic, E., 1983. Fractionation of phosphorus in sludge affected soils. J. Environ. Qual. 12, 286-290. Chapra, S.C. 2003. Power programming with vba/e xcel. Pearson Education, Inc. Upper Saddle River, New Jersey. 07458. Cheung, K.C. and T.H. Venkitachalam. 2006. Kine tic studies on phosphorus sorption by selected soil amendments for septic tank effluent re novation. Environ. Geoc hem. Health: 28:121131. Chien, S.H., and W.R. Clayton. 1980. Applicatio n of Elovich equation to the kinetics of phosphate release and sorption in soils Soil Sci. Soc. Am. J. 44: 265-268. Chimenos, J.M, A.I. Fernandez, G. Villalba, M. Segarra, A. Urruticoechea, B. Artaza, and F. Espiell. 2003. Removal of ammonium and phospha tes from wastewater resulting from the process of cochineal extraction using MgOcontaining by-product. Water Res. 37:16011607. Condon, J.B. 2006. Surface area and porosity determ inations by physisorption measurement and theory. Elsevier, Amsterdam, The Netherlands. Cooke, I. J. 1966. A kinetic approach to descripti on of soil phosphate status. J. Soil Sci. 17: 56 Cooperband, L.R, and L.W. Good. 2002. Biogenic phos phate minerals in manure: implications for phosphorus loss to surface waters. Env. Sci. Technol. 36:5075-5082. Cucarella, V., and G. Renman 2009. Phosphorus sorp tion capacity of filter materials used for on-site wastewater treatment determined in batch experiments-a comparative study. J. Environ. Qual.38:381-392. Dayton, E.A., N.T. Basta. C.A. Jakober, and J. A. Hattey. 2003. Using treatment residuals to reduce phosphorus in agricultu ral runoff. J. Am. Water Works Assoc. 95:151-158. Dayton, E.A., and N.T. Basta. 2005a. Use of dri nking water treatment residuals as a potential best management practice to reduce phosphorus risk index scores. J. Environ. Qual. 34:2112-2117. Dayton, E.A., Basta, N.T., 2005b. A method for determining the phosphorus sorption capacity and amorphous aluminum of aluminum-based drinking water treatment residuals. J. Environ. Qual 34, 1112-1118. Del Bubba, M, C.A. Arias, and H. Brix. 2003. Phosphorus adsorption maximum of sands for use as media in subsurface flow constructed reed beds as measured by the Langmuir isotherm. Water. Res. 37:3390-3400. 170

PAGE 171

De Jonge, H., and M.C. Mittelmeijer-Hazelger. 1996. Adsorption of CO2 and N2 on soil organic matter: Nature of porosity, surface area and diffusion mechanisms. Env. Sci. Technol. 30:408-413. Dodds, W.K., W.W. Bouska, J.L. Eitzmann, T.J. P ilger, K.L.Pitts, A.J. Riley, J.T. Schloesser, and D.J. Thornbrugh. 2008. Eutrophication of U. S. Freshwaters: Analysis of potential economic damages. Environ. Sci. Technol. 43: 12-19. Dubinn, M.M. 1960. The potential theory of adsorption of gases and vapors for the adsorbents with energetically nonuniform su rfaces. Chem. Rev. 60:235-241. Duffey, G.H. 2000. Modern physical chemistry: a molecular approach. Kluwer Academic Publishers, New York. Pg: 487-508. Dou, Z., J.D. Toth, J.D. Jabro, R.H. Fox, and D.D. Fritton. 1996. Soil nitrogen mineralization during laboratory incubation: dynamics and m odel fitting. Soil Biol. Biochem. 28:625-632. Eghball, B., 2002. Soil properties as influenced by phosphorus-and nitrogen-based manure and compost applications. Agron. J. 94:128-135. Elliot, H.A., G.A. OConnor. P. Lu, and S. Brin ton. 2002. Influence of water treatment residuals on phosphorus solubility and leaching. J. Environ. Qual. 31:1362-1369. Emmett, P. H., S. Brunauer and K. S. Love. 1938. The measurement of surface areas of soils and soil colloids by the use of low temperature va n der Waals adsorption isotherms. Soil Sci. 45: 57-65. Essington, M. 2004. Soil and water chemistry. An in tegrative approach. CRC LLC. Press. Boca Raton. Florida. Ewing, H. A. and E. A. Nater. 2003. Use of sca nning electron microscopy to investigate records of soil weathering preserved in la ke sediment. Holocene, 13:51-60. Freese, D., R. Lookman, R. Merckx and W.H. van Riemsdijk. 1995. New method for assessment of long-term phosphate desorption from soils. Soil. Sci. Soc. Am. J. 59:1295-1300. Forrester, J.W. 1961. Industrial dynamic s. MIT Press, Cambridge, Mass. Fowkes, F. M., Y. C. Huang, B. A. Shah, M. J. Kulp, and T. B. Lloyd. 1988. Surface and colloid chemical studies of gamma-iron oxides for magnetic memory media. Colloids and Surfaces. 29:243-261. Gadekar, S., P Pratap, and V. Amir. 2009a Validation of a comprehensive chemical equilibration model for pred icting struvite precipitation, in : Abstracts, International conference on nutrient Recovery from wast ewater streams. May 10-13, 2009. Westin Bayshore, Vancouver, BC. Canada. 171

PAGE 172

Gadekar, S., and P Pratap. 2009b. Validation and applications of a chemi cal equilibrium model for struvite precipitation. Envi ron. Model Assess. (Available at http://dx.doi.org/10.1007/s10666-009-9193-7 (verified 23rd March 2009). Goldberg, S., I. Lebron, D.L. Suarez, and H.R. Hinedi. 2001. Surface characterization of amorphous aluminum oxides. Soil Sci. Soc. Am. J. 65:78-86. Gotoh, S., and W.H. Patrick. 1974. Transformation of iron in a waterlogged soil as influenced by redox potential and pH. Soil Sci. Sc. Am. Proc 38:66-70. Gregg, S.H., and S.K.W. Sing. 1982. Adso rption, surface area and porosity. 2nd edition, Academic Press, New York. Gregg, S.J., and K.S.W. Sing. 1982. Adsorption, su rface area and porosity. Academic Press, London. Gregg, S.J., and K.S.W. Sing. 1982. Adsorption, su rface area and porosity. Academic Press, Griffin, R. A., and J. J. Jurinak. 1974. Kinetics of phosphate interaction with calcite. Soil Sci. Soc. Am. J. 38: 75-79. Groszek, A. J. 1966. Determination of surface areas of powders by flow microcalorimetry. Chem. Ind. 42: 1754-& Groszek, A. J. 1998. Flow adsorption microcal orimetry. Thermochim. Acta. 312: 133-143. Groszek, A. J., and S. Partyka. 1993. Measurem ent of hydrophobic and hydrophilic surface sites by flow microcalorimetry. Langmuir. 9:2721-2725 Groszek, A.J. 1990. 11th IUPAC conference on chemical th ermodynamics, Como, Italy, August 26-31. 534-535. Groszek, A.J. 1998. Flow adsorption microcal orimetry. Thermochim. Acta. 312:133-143. Groszek, A.J. 1999. Advances in characteri zation of adsorbents by flow adsorption microcalorimetry. p. 143-175. In A. Dabrowski (ed.) Adsorption and its applications in industry and environmental prot ection. Studies in Surface Scie nce and Catalysis. Vol. 120. Elsevier. Gungor, K., A. Jurgensen, and K.G. Karthikeyan. 2008. Determination of phosphorus speciation in dairy manure using XRD and XANES spectroscopy. J. Environ. Qual. 36:1856-1863. Gustafsson, J.P. 2009. Visual-MINTEQ. Version 2.60. Department of land and water resources engineering. Stockholm, Sweden. Haapala, H. 1998. The use of SEM/EDX for studyi ng the distribution of air pollutants in the surroundings of the emission sour ce. Environ. Pollution, 99, 361-363. 172

PAGE 173

Hanna M.M. 2004. Principles of designing and de veloping spreadsheet-based decision support systems. MSc Thesis, University of Fl. Gainesville, Fl. 32611. Hansen, I. N. C., and D. G Strawn. 2003. Kine tics of phosphorus release from manure-amended alkaline soil. Soil Sci.168: 869-879. Harada, H, Y. Shimizu, Y. Miyagoshi, S. Mitsui T. Matsuda, and T. Nagasaka. 2006. Predicting struvite formation for phosphorus recovery fr om human urine using an equilibrium model. Water Sci. Technol.54:247-255. Harris, W.G., H.G, and K.R. Reddy. 1994. Dair y manure influence on soil and sediment composition: implication for phosphorus re tention. J. Environ. Qual. 23:1071-1081. Harter, R.D. and G. Smith. 1981. Langmuir e quation and alternate methods of studying adsorption reactions in soils. p. 167-182. In R.H Dowdy et al (ed) Chemistry in the soil environment. ASA Spec. Publ. 40. ASA and SSSA Madison WI. Harvey, O. R., and Rhue, R. D. 2008. Kinetics and energetics of phospha te sorption in a multicomponent Al(III)-Fe(III) hydr(oxide) sorbent system. J. Colloid Interface Sci.322:384393. Haynes, R.J. 1982. Effects of liming on phosphate avai lability of acid soils. A critical review. Plant and Soil. 68:298-308. Ho, Y. S. and G. McKay (2000) The kinetics of sorption of divalent metal ions onto sphagnum moss flat. Water Res.34:735-742. Ho, Y. S. and G. McKay. 1999a. A kinetic study of dye sorption by biosorbent waste product pith. Resources Conserv. Recyc 25:171-193. Ho, Y. S. and G. McKay. 1999b. Pseudo-second or der model for sorption processes. Process Biochem. 34:451-465. Ho, Y. S. and G. McKay. 2002. Application of kine tic models to the sorption of copper(II) on to peat. Adsorption Sci. Technol. 20:797-815. Ho, Y. S. and G. McKay. 2003. Sorption of dyes and copper ions onto biosorbents. Process Biochem. 38:1047-1061. Ho, Y. S. and McKay, G. 1999. Pseudo-second or der model for sorption processes. Process Biochem. 34: 451-465 Ho, Y. S., 2006. Second-order kinetic model fo r the sorption of cadmium onto tree fern: A comparison of linear and non-linear methods. Water Res. 40:119-125. Ho, Y. S., and McKay, G. 2003. Sorption of dyes and copper ions onto biosorbents. Process Biochem. 38:1047-1061. 173

PAGE 174

Ho, Y. S., J. C. Y. Ng and G. McKay.2001. Rem oval of lead(II) from effluents by sorption on peat using second-order kinetics Separation Sci. Technol. 36:241-261. Hodson, M. E., 1999. Micropore surface area variati on with grain size in unweathered alkali feldspars: implications for surface r oughness and dissolution studies. Geochim. Cosmochim. Acta. 62:3429-3435. Horner, C.E., E. Holzbecher, and G. Nutzmann. A coupled transport and reaction model for long column experiments simulating bank filtration. Hydrol. Process. 21: 1015-1025 Ippolito, J.A. K.A. Barbarick, and E.F. Redente. 1999. Co-application effects of water treatment residuals and biosolids on two range grasses. J. Environ. Qual. 28:1644-1650. Ippolito, J.A. K.A. Barbarick, D.M. Heil, J.P. Chandler, and E.F. Redente. 2003. Phosphorus retention mechanisms of water treatm ent residual. J. Environ. 32:1857-1864. IUPAC (1972). Manual of symbols and termi nology for physicochemical quantities and units, Appendix 2, Definitions, Terminology, and Symb ols in Colloid and Surface chemistry Pt. 1. Pure Appl. Chem. 31:578-638. IUPAC (1985). Reporting physisorpt ion data for gas/solid systems with special reference to the determination of surface area and porosity. Pure Appl. Chem. 57:603-619 Iyamuremye, F., R.P. Dick and J. Baham. 1996. Organic amendments and soil phosphorus dynamics: I. Phosphorus chemistry a nd sorption. Soil Sci. 161: 426-435. Janos, P. and V. Smidova. 2005. Effects of surfactants on the adsorptive removal of basic dyes from water using an organomineral sorbent-ir on humate. J. Colloid Interface Sci. 291:1927. Jones J.W. and J.C. Luyten, 1998. Simulation of biological processes: In. R. Peart and R.B. Curry (ed.) Agricultural systems modeling a nd simulation. pp. 19-62. Marcel Dekker, Inc. New York. Josan, M.S., V.D. Nair, W.G. Harris, and D. Herrera. 2005. Associated release of Magnesium and phosphorus from active and abandoned dairy soils. J. Environ. Qual. 34, 184-191. Joslin, S. T., F. M. Fowkes. 1985. Surface acidity of ferric oxides studied by flow microcalorimetry. Ind. Eng. Ch em. Prod. Res. Dev. 24: 369-375. Kabengi, N. J., S. H. Daroub, and R. D. Rhue 2006b. Energetics of arsenate sorption on amorphous aluminum hydroxides studied using flow adsorption calorimetry. J. Colloid Interface Sci. 297:86-94. Kabengi, N.J., R.D. Rhue and S.H. Daroub. 2006a Using flow calorimetry to determine the molar heats of cation and an ion exchange and point of zero net charge on amorphous aluminum hydroxides. Soil Sci. 171:13-20. 174

PAGE 175

Kleinman, P. J. A., and A. N. Sharpley. 2002. Estimating soil phosphorus sorption saturation data from Mehlich-3 data. Commun. Soil Sci. Plant Anal. 33:1825-1839. Koopmans, G.F., W. J. Chardon, P. de Willig en, and W.H. van Riemsdijk, 2004. Phosphorus desorption dynamics desorption dynamics in soil and the link to a dynamic concept of bioavailability. J. Environ. Qual 33:1393-1402. Kordlaghari, M. P., and D. L. Rowell. 2006. The role of gypsum in the reactions of phosphate with soils. Geoderma. 132:105-115. Kruk, M., M Jaroniec, and K. P. Gadkaree. 1999. Determination of specific surface area and the pore size of microporous carbons from adso rption potential distributions. Langmuir. 15:1442-1448. Kuo, S., and E.G. Lotse. 1974. Kinetics of phosphate adsorption and desorption by lake sediments. Soil Sci. Soc. Amer. Proc. 38:50-54. Lane, C. T. 2002. Water treatment residuals eff ects on phosphorus in soils amended with dairy manure. M.Sc. Thesis, Univ. Fl ., Gainesville, Fl. 32611-0510. Lagergren, S. 1898. Zur theorie der sogennten ad sorption geloster stoffe. Kungliga Svenska Vetenskapsakademiens. Handlingar. 24:1-39. Laidler, K. J. 1987. Chemical kinetics. Third Edition. Pearson Educational Company. Singapore, Pte, Ltd. India. Langmuir, I. 1918. The adsorption of gases on plane surfaces of glass, mica, and platinum. J. Am. Chem. Soc. 40:1361-1382. Lantenois, S., B. Prelot, J. M. Douillard, K. Szczodrowski, and M.C. Charbonnel. 2007. Flow microcalorimetry: Experimental development a nd application to adso rption of heavy metal cations on silica. Appl. Surf. Sci. 253:5807-5813. Lijklema, L. 1980. Interaction of orthophosphate with iron (III) and aluminum hydroxides. Am. Chem. Soc. 14: 537-541. Lindsay, W.L., 2001. Chemical equilibria in soils The Blackburn Press. Caldwell, New Jersey. USA. Lindsay, W.L., P.C.G. Velk, and H.C. Chien. 1989. Phosphate Minerals. In: Dixon, J.B., Weed, S.B. (Eds.), Minerals in Soil Environments 2nd edn. Soil Science Society of America, Madison, pp. 1089 1130. Lipson, D.S., J.E. McCray, and G.D. Thyne. 2007. Usi ng Phreeqc to simulate solute transport in fractured bedrock. Grou nd Water. 45:468-472. 175

PAGE 176

Lock, M.A., and T.M. Ford. 1983. Inexpensive fl ow microcalorimeter for measuring heat of production of attached and sedimentary aquatic microorganism s. Appl. Environ. Microbiol. 46:463-467. Loewenthal, R.E., U. R.C. Kornmuller, a nd E.P. van Heerden. 1994. Modeling struvite precipitation in anaerobic treatment sy stems. Water Sci Technol. 30:107-116. Lowell, S., J.E. Shields, M.A.Thomas, and M. Thommes. 2004. Characterization of porous solids and powders: surface area, pore size and density. Kluwer Academic Publishers, Dordrecht, The Netherlands. Liu, M., D.E. Kissel, L.S. Sonon, M.L. Cabrera, and P.F. Vendrell. 2008. Effects of biological nitrogen reactions on soil lime requirement de termined by incubation. Soil Sci. Soc. Am. J. 72:720-726. Lu, P., and G.A. OConnor, 1999. Factors affec ting phosphorus reactions in soils: Potential sewage sludge effects. Soil Crop Sci Soc. Fla. 58:66-71. Makris, K. C., W. G. Harris, G. A. O'C onnor, T. A. Obreza and H. A. Elliott 2005. Physicochemical properties related to long-term phosphorus retention by drinking-water treatment residuals. Environ. Sci. Technol. 39:4280-4289. Makris, K.C., W.G. Harris, G.A. OConnor, T. A Obreza. 2004. Phosphorus immobilization in micropores of drinking-water treatment residua ls: Implications for long-term stability. Environ. Sci. Technol. 38:6590-6596. Malecki-Brown, L.M., and J.R. White. 2009. Eff ects of aluminum-containing amendments on phosphorus sequestration of wast ewater treatment wetland soil. Soil Sci. Soc. Am. J. 73:852-861. Masel, R. 1996. Principles of adsorption a nd recreation on solid surf aces. New York Wiley. McBride, M.B. 1994. Environmental chemistry of so ils. Oxford University Press, Inc. New York McGonigal, G.C., R.H. Bernhard t, and D.J. Thompson. 1990. Imaging alkane layers at the liquid graphite interface with the s canning tunneling microscope. Appl. Phys. Letters 57: 28-30. Mikutta, R., and C. Mikutta. 2006. Stabilization of organic matter at micropores (< 2nm) in acid forest subsoils. Soil Sci. Soc. Am. J. 70:2049-2056 Montgomery, D.C. 2005. Design and analysis of experiments. 6th edition. John Wiley, $Sons, Inc. New Jersey. USA. Moore, P.A., T.C. Daniel, and D.R. Edwards. 1999. Reducing phosphorus runoff and improving poultry production with alum Poultry Sci. 78:692-698. Moore, P.A., and D.M. Miller. 1994. Decreased phosphorus solubility in poultry litter with aluminum, calcium and iron amendments. J. Environ. Qual. 23:325-330. 176

PAGE 177

Munch, E.V. and K. Barr. 2001. Controlled stru vite crystallization for removing phosphorus from anaerobic digester sidest reams. Water Res. 35:151-159. Nair, V.D., D.A. Graetz, and K.M. Portier. 1995. Forms of phosphorus in soil profiles from dairies of south Florida. So il Sci. Soc. Am. J. 59:1244-1249. Novak, J.M., and D.W. Watts. 2004. Increasing the phosphorus sorption capacity of southeastern coastal plain soils using water trea tment residuals. So il Sci. 169:206-214. Novak, J.M., and D.W. Watts. 2005. An alum-b ased water treatment residual can reduce extractable phosphorus concentrations in three phosphorus-enriched coas tal plain soils. J. Environ. Qual. 34:1820-1827. OConnor, G.A., H.A. Elliot, and P. Lu. 2001. Characterizing water treatment residuals phosphorus retention. Soil and Crop Sci. Soc. Florida Proc. 61:67-73. OConnor, G.A., H.A. Elliot, D. Sarkar, a nd D.A. Graetz. 2002. Characterizing forms, solubilities, bioavailabilities and minera lization rates of phosphorus in biosolids, commercial fertilizers and manures. Fi nal Report, Water Env. Res. Foundation, Alexandria, VA. Ortuno, J.F., J. Saetz, M. Llorens, and A. So ler. 2000. Phosphorus release from sediments of a deep wastewater stabilization pond. Water Sci. Tech. 42: 265-272. Ozacar, M. 2003. Equilibrium and kinetic modell ing of adsorption of phosphorus on calcined alunite. Adsorption. 9: 125-132. Ozdemir, E., B.I. Morsi, and K. Schroeder. 2003. Importance of volume effects to adsorption isotherms of carbon dioxide on co als. Langmuir. 19: 9764-9773. Pan, H-B, and B.W. Darvell. 2009. Calcium phospha te solubility: the ne ed for re-evaluation. Crystal Growth & Design. DOI:10.1021/cg801118v. Pant, H.K., P. Mislevy and J.E. Rechcigl. 2004. Effects of phosphorus and potassium on forage nutritive value and quantity: environmen tal implications. Agron. J. 96:1299-1305. Pant, H.K., and K.R. Reddy. 2003. Potential inte rnal loading of phosphorus in a wetland constructed in agricultural land. Water Res. 37:965-972. Parkhurst, D. L., and C.A. Appelo. 1999. Users guide to PHREEQC (version 2)A computer program for speciation batch-r eaction, one-dimensional transport, and inverse geochemical calculations. U.S. Geological Survey Water-R esources Investigations. Report. 99-4259. Patrick, W.H and R.A. Khalid. 1974. Phosphate release and sorption by soils and sediments: effects of aerobic and anaerobic conditi ons. Science, New Series, 186: 53-55. Phipps, M.A., and L.A. Mackin. 2000. Application of isothermal microcalorimetry in solid state drug development. Pharm. Sc i. & Tech. Today. 3:9-17. 177

PAGE 178

Pierzynski, G.M., T.J. Logan, S. J. Traina, and J.M. Bigham. 1990. Phosphorus chemistry and mineralogy in excessively fertilized soils : quantitative analysis of phosphorus rich particles. Soil Sci. Soc. Am. Proc. 54:1576-1583. Prochnow, L. I., C. A. Clemente, E. F. D illard, A. Melfi and S. Kauwenbergh (2001) Identification of: Compounds present in single superphosphates produced from Brazilian phosphate rocks using SEM, EDX, and X-ray techniques. Soil Sci. 166:336-344 Pye, K., and D. Croft. 2007. Forensic analysis of soil and sediment traces by scanning electron microscopy and energy-dispersiv e X-ray analysis: An experime ntal investigation. Forensic Sci. Int. 165:52-63. Reddy, K.R., O.A. Diaz, L.J. Scinto, and M. Ag ami. 1995: Phospshorus dynamics in selected wetlands and streams of the lake Okeechobee basin. Ecol. Engin. 5:183-207. Rhue, R.D., C. Appel, and N. Kabengi. 2002. Me asuring surface chemical properties of soil using flow calorimetry. J. Soil Sci. 167:782-790. Rhue, R.D., and W.G. Harris. 1999. Phosphorus sorption/desorption reactions in soils and sediments. p. 187-206. In Reddy, K.R., G.A.OConnor, a nd C.L. Schelske (ed.) Phosphorus biogeochemistry in subtropical ecosystems. Lewis publishers, CRC Press LLC, Boca Raton, Florida. Ribeiro, C.M., P. Carrott, M.M. Brotas de Carvalho and K.S.W. Sing. 1991. Ex-hydroxide magnesium oxide as a model adsorbent for i nvestigation of micropore filling mechanisms. J. Chem. Soc., Faraday Trans. 87:185-191. Rong, X. M., Huang, Q. Y, and W.L. Chen. 2007a. Microcalorimetric investigation on the metabolic activity of Bacillus thuringiensis as influenced by kaolinite, montmorillonite and goethite. Appl. Clay Sci.38:97-103. Rong, X. M., Q. Y. Huang, D.H. Jiang, P. Cai, and W. Liang. 2007b. Isothermal microcalorimetry: A review of applicatio ns in soil and environmental sciences. Pedosphere. 17:137-147. Rouquerol, F., J. Rouquerol, and K. Sing. 1999. Adsorption by powders and porous solids. Principles, Methodology and Applicatio ns. Academic Press. London, UK. Rudzinski, W. and W. Plazinski. 2006. p.320. Heter ogeneity effects in adsorption and catalysis on solids (ISSHAC-6). In Proc. Int. Symp. Zakopane, Poland. Rudzinski, W., R. Charmas, W. Piasecki, F. Th omas, F. Villieras, B. Prelot and J. M. Cases .1998. Calorimetric effects accompanying ion adsorption at the charged metal oxide/electrolyte interfaces: E ffects of oxide surface energe tic heterogeneity. Langmuir, 14:5210-5225. SAS Institute. 2003. Statistical analysis software, version 9.1.3. SAS Institute, Inc, Cary, NC. USA. 178

PAGE 179

Schuiling, R.D., and A. Andrade. 1999. Recovery of struvite from calf manure. Environ. Technol. 20:765-768. Scott, W.D., Wrigley, T. J., and K.M. Webb. 1991. A computer model of struvite solution chemistry. Talanta. 38:889-895. doi.10.1016/0039-9140(91)80268-5. Seref, M.M., R.K. Ahuja, and W.L. Winst on. 2007. Developing spread sheet-based decision support systems. Dynamic idea s. Belmont, Mass. USA. Sharpley, A.N., McDowell, R.W, and P.J.A. Kl einman. 2004. Amounts, forms and solubility of phosphorus in soils receiving manure. So il Sci. Soc. Am. J. 68: 2048-2057. Shin, E.W., J.S. Han, M. Jang, S.H. Min, J. K. Park, and R.M. Rowell. 2004. Phosphate adsorption on aluminum-impregnated mesoporous silicates: surface structure and behavior of adsorbents. Environ. Sci. Technol. 38:912-917. Siemens, J., K. Ilg, F. Lang, and M. Kaupenj ohann. 2004. Adsorption controls mobilization of colloids and leaching dissolved phosphor us. Eur. J. Soil Sci. 55:253-263. Silveira, M.L., M.K. Miyittah, and G.A. O Connor. 2006. Phosphorus release from a manureimpacted spodosol: effects of a water treatm ent residual. J. Environ. Qual. 35:529-541. Sing, K. S. W 2004. Characterization of porous mate rials: past, present a nd future. 3-7. Elsevier Science Sing, K. S. W. 1989.The use of physisorption for the characterization of microporous carbons. Carbon. 27:5-11. Sing, K. S. W., D. H. Everett, R. A. W. Haul L. Moscou, R. A. Pierotti, J. Rouquerol & T. Siemieniewska. 1985. Reporting physisorption da ta for gas solid systems with special reference to the determination of surface -area and porosity (recommendations 1994) Pure Appl. Chem. 57:603-619. Sing, K., 2001. The use of nitrogen adsorption for th e characterisation of porous materials. 3-9. Elsevier Science Sing, K.S.W. 1998. Adsorption methods for the characterization of por ous materials. Adv. Colloids Interface Sc. 76-77:3-11 Smith, E.M. 1982. Systems research provi des new knowledge about agriculture. In M.G. Russel, R.J. Sauer and J.M. Barnes (eds.) Enabling Interdisciplinary research: Perspectives from agriculture, forestry and home economics pp. 155-159. Miscellaneous publication 19-1982, University of Minnesota. Sparks, D.L. 1989. Kinetics of soil chemical processes. Academic Press, Inc. London Sparks, D.L. 2002. Environmental soil chem istry. Academic Press, San Diego 179

PAGE 180

Sposito, G. 1989. The chemistry of soils. Oxford University Press. New York. Steinberg, G. 1981. What you can do with surface calorimetry. ChemTech. 12:730-737. Stevens, R, J. Pinto, Y. Mamane, J. Ondov, M. Abdulraheem, N. Ali-Majed, M. Sadek, M. Cofer, W. Ellenson, R. Kellogg. 1993. Chemical and physical properties of emissions from Kuwaiti oil fires. Water Sci. Technol. 27:223-233. Svik, A. K., and B. Klve. 2005. Phosphorus rete ntion process in she ll sand filter systems treating municipal wastewat er. Ecol. Engin.25:168-182. Sulkowski, M. and A.V. Hirner. 2006. Elemental fractionation by sequential extraction in a soil with high carbonate content. Applied Geochem. 21:16-28. Tang, W.P., O. Shima, A. Ookubo and K. Ooi. 1996. A kinetic study of phosphate adsorption by boehmite. J. Pharm Sci. 86:230-235 Toran, L, and D. Grandstaff. 2002. Phreeqc and ph reeqci: geochemical mode ling with interactive interface. Ground water. 40:462-464. TRB, 1987. Stabilization reactions, properties, de sign and construction: St ate of the art report no: 5, TRB. National Research Council, Washington. Tsunoda, R. 1977. Determination of micropore volumes of active carbon using DubininRadushkevish plots. Bulletin Chem. Soc. Jpn. 50:2058-2062. Tunay, O., I. Kabdasli, D. Orhon, and S Kolcak. 1997. Ammonia removal by magnesium ammonium phosphate precipitati on in industrial wastewaters. Water Sci. Technol. 36:225228. Turban, E. and J. Aronson. 2001. Decision support system and intelligent systems, Prentice-Hall. Ueno, Y., and M. Fujii. 2001. Three years of experience of operating and selling recovered struvite from full-scale plant. Environ. Technol. 22:1373-1381. USEPA. 1996. Acid digestion of sediments, sludges, and soils [Online]. Available at http://www.epa.gov/epaoswer/hazwaste/test/pdfs/3050b.pdf (verified 12 June 2007). USEPA, Cincinnati, OH. Van de Walle, R.H., M. T. Wajer, and D.M. Sm ith. 1993. Process for removal of metal ions from water. U. S. Patent 5 211 852. Date issued: 18 may. Van Dyne, G.M., and Z. Abramsky. 1975. Agricu ltural systems models and modelling: An overview. In. G.E. Dalton (ed.) Study of Ag ricultural systems. pp. 23-106. Applied science publishers, London. Violante, A., and L. Gianfreda. 1993. Competitio n in adsorption between phosphate and oxalate on an aluminum hydroxide montmorillonite complex. Soil Sci. Soc. Am. J. 57:1235-1241. 180

PAGE 181

Wadso, I. 2001. Isothermal microcalorimetry. Current problems and prospects. J. Therm. Anal. 64:75-84 Wagner, T. and J. Kollat. 2007. Numerical and visual evaluati on of hydrological and environmental models using the Monte Carl o analysis toolbox. Environ. Modelling & Software 22:1021-1033. Wang, J. J., and D.L. Harrell. 2005. Effect of ammonium, potassium, and sodium cations and phosphate, nitrate, and chloride anions on zinc sorption and lability in selected acid and calcareous soils. Soil Sci. Soc. Am. J. 69:1036-1046. Wilsenach, J.A., C.A.H Schuurbiers, and M. C.H. van Loosdrecht. 2007. Phosphate and potassium recovery from source separated urine through precipitati on. Water Res. 41:458466. Yoshino, M., M. Yao, H. Tsuno, and I. Somiya. 2003. Removal and recovery of phosphate and ammonium as struvite from supernatant in anaerobic digestion. Wa ter Sci. Techol.48:171178. Zadora, G., and Z. Brozek-Mucha. 2003. SEM-EDX a useful tool for forensic examinations. Mat. Chem. Phys. 81:345-348. Zhu, C., and G. Anderson. 2002. Environmental applications of geochemical modeling. Cambridge University Press. United Kingdom. Zhu, M., and Y. Li. 2001. Phosphorus-sorption charac teristics of calcareous soils and limestone from the southern Everglades and adjacent farmlands. Soil Sci. Soc. Am. J.65:1404-1412. 181

PAGE 182

182 BIOGRAPHICAL SKETCH Michael Miyittah is a native of Ghana. He gr aduated from the University of Ghana, Legon, with Bsc (Hons). After graduation, he taught briefl y and later went back to the University of Ghana for postgraduate study, and also at Univer sity of Cape Coast, Ghana (post graduate diploma in education). He later won a Japanese Government Scholarship to Chiba University, Japan, where he graduated with M.Phil (Envir onmental Science, 2002). Michael came to University of Florida, Soil and Water Science Department in 2003. In 2004, he completed MSc (Soil and Water Science). He started his Ph.D in August 2005.