INVESTIGATION INTO STRUVITE SOLUBILITY, GROWTH AND
DISSOLUTION KINETICS IN THE CONTEXT OF PHOSPHORUS
RECOVERY FROM WASTEWATER
by
MD. IQBAL HOSSAIN BHUIYAN
M.Sc, UENSCO-IHE Institute for Water Education, 2002
A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
in
THE FACULTY OF GRADUATE STUDIES
(Civil Engineering)
THE UNIVERSITY OF BRITISH COLUMBIA
October 2007
Md. Iqbal Hossain Bhuiyan, 2007 Abstract
The present research was conducted to investigate the mechanisms controlling formation,
dissolution and decomposition of the mineral, struvite (MgNPLiPO^HiO) in the context of phosphorus recovery from wastewater. Solubility, thermodynamics, kinetics and thermal
decomposition of struvite were studied in laboratory and wastewater treatment environments to
gain knowledge to optimize the phosphorus recovery process from wastewater through struvite
crystallization.
The thermodynamic solubility products (Ksp) of struvite were determined by
extrapolating measured solubility product values to zero ionic strength, with -log Ksp of
13.36(+0.07) at 25°C, using an appropriate activity coefficient model. A representative
temperature compensation factor (a = 0.0198 °CI) has been derived for electrical conductivity
(EC) correction, and a relationship between ionic strength (I) and EC has been developed for
anaerobic digester supernatant/centrate samples from five different wastewater treatment plants
in western Canada. The metastable region, where nucleation is negligible, for struvite
precipitation was explored in this study. This region was used in a kinetics study to suppress
nucleation of struvite during growth experiments in a bench-scale fluidized bed reactor (FBR). A
linear growth rate model has been tested and proposed, which was found to be effective for
struvite growth determination in FBRs. The dissolution processes of struvite were investigated in
a batch reactor system using two different theoretical models. The experimental values of
struvite dissolution were found to fit well with both models. In a mixed flow-through reactor
system, the dissolution rates for struvite pellets were found to increase with the hydrogen ion
concentration in the acidic pH, while the rate of dissolution in the alkaline pH was found to
increase due to hydroxyl-promoted dissolution. The thermal decomposition study of struvite
ii showed that the simultaneous loss of both ammonia and water molecules from the struvite structure occurred gradually as a function of temperature, rather than as a distinct step.
A pilot-scale struvite recovery FBR developed at The University of British Columbia
(UBC) was operated, using the knowledge gained from the thermodynamics and kinetics experiments. The pilot-scale FBR was found to be effective in recovering phosphate from anaerobic digester centrate in the form of a nearly pure struvite.
in Table of Contents Abstract ii
Table of Contents iv
List of Tables viii
List of Figures x
List of Abbreviations xiv
Preface xv
Acknowledgements xvii
Contribution of others xix Chapter 1 Introduction 1 1.1 Preface 1
1.2 Literature review 3
1.2.1 Why recover phosphorus? 3
1.2.2 Struvite Solubility and Thermodynamics 8
1.2.3 Estimation of Ionic Strength from Electrical Conductivity 10
1.2.4 Precipitation kinetics of struvite 11
1.2.5 Pilot-scale fluidized reactor operation 13
1.2.6 Dissolution kinetics and slow release property of struvite 14
1.2.7 Thermal decomposition of struvite and its phase transition 14
1.3 Research objectives 15 1.4 Thesis outline 17 Chapter 2 A solubility and thermodynamic study of struvite 25 2.1 Introduction 25
2.2 Materials and methods 28
2.2.1 Formation of struvite 28
2.2.2 Equilibration 28
2.2.3 Thermodynamic solubility product, Ksp 30
2.2.4 Speciation and ionic strength calculation 31
2.3 Results and discussion 33
2.3.1 Thermodynamic solubility product 33
2.3.2 Solubility product of struvite at various temperatures 34
2.3.3 Effect of pHon struvite solubility 35
2.3.4 Solubility product (Ksp) value and solubility of struvite 37
iv 2.3.5 Temperature effect on struvite solubility 39
2.3.6 Enthalpy 40
2.4 Conclusions 42 Chapter 3 Determination of temperature dependence of electrical conductivity and its relationship with ionic strength of anaerobic digester supernatant, for struvite formation 55 3.1 Introduction 55
3.2 Materials and methods 58
3.2.1 Temperature dependence ofEC 58
3.2.2 EC-I relationship 59
3.2.3 Analyses 59
3.3 Results and discussion 60
3.3.1 Temperature dependence of EC 60
3.3.2 EC-1 relationship 62
3.4 Conclusions 65 Chapter 4 Nucleation and growth kinetics of struvite in a fluidized bed reactor 77
4.1 Introduction 77
4.2 Materials and methods 81
4.2.1 Determination of induction time 81
4.2.2 Determination of relative supersaturation 83
4.2.3 Identification of metastable region 83
4.2.4 Crystallization system 84
4.2.5 Determination of crystal growth rate 85
4.2.6 Analysis 86
4.3 Results and discussion 87
4.3.1 Nucleation and induction time 87
4.3.2 Metastable region 89
4.3.3 Growth 89
4.3.4 Growth rate expressions 91
4.4 Conclusions 93 Chapter 5 Assessing struvite precipitation in a pilot-scale fluidized bed crystallizer 109 5.1 Introduction and background 109 5.2 Material and methods 112
v 5.2.1 Reactor design and operation.. 112
5.2.2 Chemicals, storage tanks and pumps 113
5.2.3 Sampling and analysis 113
5.2.4 Product Identification 114
5.3 Results and discussion 114
5.3.1 Centrate characteristics during the study 115
5.3.2 Reactor operation 115
5.3.3 Performance of the crystallization process 116
5.3.4 Supersaturation level 117
5.3.5 Induction time and mixing 118
5.3.6 Apparent upflow velocity 119
5.3.7 Mg:P and N:P molar ratio 119
5.3.8 Effect of Organic ligands 120
5.3.9 Influence of calcium and carbonate.ions 121
5.3.10 Crystal morphology 122 5.4 Conclusions 123 Chapter 6 Dissolution kinetics of struvite grown in a pilot-scale crystallizer 148 6.1 Introduction 148
6.2 Materials and Methods 150
6.2.1 Identification 150
6.2.2 Batch reactor system 151
6.2.3 Treatment of data 151
6.2.4 Mixed flow-through reactor system 153
6.2.5 Treatment of data 154
6.2.6 Analysis 155
6.3 Results and discussion 155
6.3.1 Batch reactor system 155
6.3.2 Mixed flow-through reactor system 158
6.4 Conclusions 160 Chapter 7 Thermal decomposition of struvite and its phase transition 176 7.1 Introduction 176 7.2 Materials and methods 178
vi 7.2.1 Formation of struvite 178
7.2.2 Identification of struvite and transformation compounds 179
7.2.3 Phase transition of struvite in excess water 180
7.2.4 Analytical and thermogravimetric methods 180
7.3 Results and discussion 181
7.3.1 Identification of struvite 181
7.3.2 Thermogravimetric Analysis 181
7.3.3 Evaluation of activation energy 183
7.3.4 Phase transition with heating in excess water 184
7.3.5 Phase transition with boiling in excess water 186
7.4 Conclusions 187 Chapter 8 General conclusions and direction for future research 203 8.1 Introduction 203
8.2 Overall conclusions 204
8.3 Engineering significance 207
8.4 Recommendations for future research 209
vii List of Tables
Table 2.1 Published ksp values for Struvite at 25 C from Literature 44
Table 2.2 Major equilibria involved in the computation of the solution species at 25 °C. 45
Table 2.3 Solubility products of struvite determined at various temperatures. (Values in
parenthesis are the 95% confidence intervals) 46
Table 3.1 Expressions for activity coefficients 66
Table 3.2 Temperature compensation factor (a) and RMS percentage (e) 67
Table 3.3 Chemical composition of the anaerobic digester supernatant/centrate samples. All values in mg 1"' 68 Table 3.4 pH, temperature, electrical conductivity and calculated ionic strength 69
Table 3.5 Correlation matrix for anaerobic digester centrate supernatant/centarte. All values shown are Pearson's correlation coefficients of struvite constituting ion concentrations [ ] , activities { }, EC (pS cm"1) and I (mol 1"') 70
Table 4.1 Concentrations and conditions used in different run during induction time
study 95
Table 4.2 Determination of metastable region for struvite at 25°C 96
Table 4.3 Mass transfer coefficient and surface reaction coefficient of struvite at pH=8.07 97 Table 5.1 Centrate characteristics of the Lulu Island Wastewater Treatment Plant
during the pilot-scale operation of the struvite crystallizer (n = 23) 125
Table 5.2 Operational conditions during the pilot-scale operation at LIWWTP (n=23). 126
Table 5.3 Calculated Reynolds Number at three sections of the reactor 127
Table 5.4 Major equilibria involved among acetate and struvite constituents (Ball and Nordstorm, 1991) and corresponding solubility product constants at 25 °C... 128
3 Table 5.5 pKsp values and solubility equilibria of possible precipitates in Ca-Mg-P04 "-
2 C03 "system at 25°C 129
Table 6.1 Solubility of struvite (Cs) and equilibrium pH calculated by PHREEQC for different initial pH values 162
viii Table 6.2 Dissolution rate constants with the R2 values of the regressions for different amount of struvite pellets and pH values 163
Table 6.3 Experimental conditions and corresponding rate of dissolutions rates of struvite pellets at 25°C 164
Table 7.1 Solubility product values (Ksp) available in the literature for the precipitates
2+ 3 + + in Mg - P04 " - NH4 - H system 189
Table 7.2 Activation energy of the struvite decomposition reactions at different heating rates 190
ix List of Figures
Figure 1.1 Metastable widths for different possible mechanisms of nucleation 20
Figure 2.1 Identification of crystalline solid as struvite using (a) powder X-ray diffraction. The peaks of the X-ray diffraction pattern of struvite sample matches perfectly with the struvite standard peaks (| )(b) IR spectrum where vibrational mode peak for P-0 bond of the PO4 " groups are observed at 1006 and 570 cm"1 47
Figure 2.2 Example of extrapolation for thermodynamic solubility product for struvite at 25 °C (n=5). Error bars: 95% confidence interval 48
Figure 2.3 Solubility product of struvite at different temperatures (This research and other studies) 49
Figure 2.4 Solubility (mg l"1) at different temperatures in the solutions of NaCl with different normality 50
Figure 2.5 Variation of Solubility (mg 1"') and Ksp of struvite with temperature (n=5). Error bars: 95% confidence interval 51
Figure 3.1 Electrical Conductivity vs. Temperature of the supernatant/centrate
samples from five WWTPs 71
Figure 3.2 Estimated EC25 vs. Measured EC25 of supernatant/centrate samples 72
Figure 3.3 Measured and estimated EC10 using different standard temperatures 73
Figure 3.4 Ionic Strength vs. Electrical Conductivity 74
Figure 4.1 Bench-scale fluidized bed crystallization system for struvite growth experiment... 98 Figure 4.2 Concentration profile of Cpo4 and their average in three different sections of the reactor used for struvite growth experiment 99
Figure 4.3 Induction time versus solution supersaturation at 25°C for 120 rpm (G value = 140 sec"'). Driving force for precipitation as a function of induction time 100
Figure 4.4 Struvite induction time at selected supersaturation levels at a constant mixing speed of 120 rpm 101
Figure 4.5 Variation of induction time with G value (s1) at a selected saturation level (SI - 1.83). Error bars: 95 % confidence interval 102 Figure 4.6 Determination of metastable region for struvite with (a) concentrations (b) activities of the ions 103
Figure 4.7 Variations of supersaturation in three different sections of the struvite reactor 104
Figure 4.8 Box-plots of electrical conductivity variation in three different sections of the struvite reactor (n=15). Error bars: 95% confidence interval. Percentiles shown: 10th, 25th, 75th, and 90th. The horizontal line inside the box represents the median 105
Figure 4.9 log G' vs. log a plot for pH = 8.07 106
Figure 5.1 Pilot-scale struvite crystallization system at LIWWTP 130
Figure 5.2 Constiuent concentrations in the centrate during the operation 131
Figure 5.3 Influent and effluent concentrations of three constituents of struvite in the crystallizer (n=23). Error bars: 95 % confidence interval 132 Figure 5.4 Variation in supersaturation ratio (Q) and phosphate removal efficiency at
different pHs (n = 23). Error bars: 95% confidence interval 133
Figure 5.5 Variation in supersaturation ratio (Q) in influent and effluent with time.... 134
Figure 5.6 Upflow velocity in the bottom section of the pilot-scale struvite
crystallizer vs. phosphate removal 135
Figure 5.7 Mg:P molar ratio vs. (a) phosphate removal (b) model calculated SI 136
Figure 5.8 N:P molar ratio vs. (a) phosphate removal (b) model calculated SI for struvite in the crystallizer 137 Figure 5.9 Model calculated saturation index (SI) value of struvite vs. acetate concentration in the struvite crystallizer 138
Figure 5.10 pH vs. phosphate removal in the crystallizer and inreactor filtered TOC (n=23). Error bars: 95% confidence interval 139
Figure 5.11 Identification of the product as struvite using powder X-ray diffraction. X-ray diffraction pattern matches very well with pattern of struvite standard (•) 140
Figure 5.12 Ca:P molar ratio vs. model calculated Saturation Index (SI) values of the possible precipitates 141
xi Figure 5.13 Influent and effluent HCO3" and CO32" concentrations 142
Figure 5.14 pH vs. model calculated Saturation Index (SI) values of the possible precipitates that could interfere with the struvite formation in a crystallizer (solid phases marked by *) 143
Figure 5.15 SEM analysis of the harvested product (a) full pellet (b) x-section of a pellet 144
Figure 6.1 Mixed flow-through reactor set-up 165
Figure 6.2 Concentrations in solution during dissolution of different amounts of struvite pellets for initial pH value of 5.05: (a) total ortho-phosphate (Cpo4), (b) total ammonia (CNH4), (c)mean (C); and (d) for 1000 mg struvite pellets for different initial pH values 166
Figure 6.3 Percent struvite dissolved in 1 h against initial amount of struvite for different initia pH values 167
Figure 6.4 Concentrations in solution during dissolution of 1000 mg of struvite pellets with an initial pH of 5.15: (a) 10 h, and (b) 5 d 168
Figure 6.5 Model curves of the concentrations of struvite with time for an initial amount of 1000 mg, for different initial pH values 169
1 Figure 6.6 Plot of dC/dt as a function of A V (Cs-C) drawn for 500 mg initial amount of struvite pellets and pH value of 4.38 170
2/s Figure 6.7 Plots of dC/dt as a function of ns V' (Cs - C) for an initial mass of 1000 mg, and pH 4.38 171
Figure 6.8 Struvite constituent dissolutions with time for input solution pH (a) 6.05 and (b) 8.55 172
Figure 6.9 Change of rate of dissolution with pH for struvite pellets in mixed flow through reactor showing the dissolution rates determined using river(s)/creek(s) waters (boxed values S-l to S-3 173
Figure 7.1 X-ray diffraction patterns of (a) synthetic struvite (b) struvite pellets and struvite standard (•) 191
Figure 7.2 TGA and DTGA curves of Synthetic struvite for heating rate 1, 2, 5 and 20°Cmin"' 192
Figure 7.3 TGA and DTGA curve for heating rate 1 and 5°C min"1 for struvite pellets 193
xii Figure 7.4 X-ray diffraction pattern (3-D) of struvite pellets with increasing temperature 194
Figure 7.5 NH4-N/ PO4-P molar ratio in the solution after dissolution of struvite heated at different temperatures and excess water. Error bars: 95% confidence interval 195
Figure 7.6 X-ray diffraction patterns of the heated (a) synthetic struvite (b) struvite pellets with struvite (•) and bobierrite (•) standards after heating at 50°C in excess water. Bobierrite peaks are identified by arrow 196
Figure 7.7 Powder X-ray diffraction patterns of the heated (a) synthetic struvite (b) struvite pellets with struvite (•) and bobierrite (•) standards after heating at 60°C in excess water. Bobierrite peaks are identified by arrow 197
Figure 7.8 X-ray diffraction patterns of the heated (a) synthetic struvite (b) struvite pellets with struvite (•) and bobierrite (•) standards after heating at 80°C in excess water. Bobierrite peaks are identified by arrow 198
Figure 7.9 X-ray pattern of the product after boiling struvite pellets in excess water for 1 day with the pattern of the dittmarite standard (•) 199
Figure 7.10 Schematic of the possible transformation mechanisms of various phases associated with struvite 200
xiii List of Abbreviations
BET Brunaer Emmet and Teller BNR Biological Nutrient Removal DOC Dissolved Organic Carbon DTGA Derivative of Thermogavimetric Analysis EBNR Enhanced Biological Nutrient Removal EBPR Enhanced Biological Phosphorus Removal EC Electrical Conductivity FBR Fluidized Bed Reactor HRT Hydraulic Retention Time I Ionic Strength IC Inorganic Carbon ID Internal Diameter IR InfraRed MAP Magnesium Ammonium Phosphate MEBPR Membrane Enhanced Biological Phosphorus LIWWTP Lulu Island Wastewater Treatment Plant N Nitrogen P Phosphorus RMS Root Mean Square SEM Scan Electron Microscope SD Standard Deviation SRT Sludge Retention Time SSA Specific Surface Area SI Saturation Index TC Total Carbon TGA Thermogravimetric Analysis TOC Total Organic Carbon UBC The University of British Columbia VFA Volatile Fatty Acids WWTP Wastewater Ttreatment Plant XRD X-ray Difraction Preface
The present Ph.D. thesis has been prepared in manuscript-based format. A manuscript- based thesis, as described by the Faculty of Graduate Studies at The University of British
Columbia, is a collection of published, in-press, accepted, submitted or draft manuscripts. The body of this thesis has been separated into eight main chapters. Chapter 1 is an introductory
chapter presenting the background to the engineering problem and the main objectives of the
thesis. The results of the research program are presented in Chapter 2 through 7. Chapter 8
relates the manuscript chapters to each other, outlines the engineering significance of the
research work, and provides direction for future research.
The following is the list of the manuscripts accepted, submitted, and in preparation
pertaining to this thesis.
1 Bhuiyan, M.I.H, Mavinic, D.S., and Beckie, R.D. (2007) A solubility and thermodynamic
study of struvite. Environmental Technology 28, 1015-1026.
2 Bhuiyan, M.I.H, Mavinic, D.S., and Beckie, R.D. (2007) Determination of temperature
dependence of electrical conductivity and its relationship with ionic strength of anaerobic
digester supernatant, for struvite formation. Submitted to Journal of Environmental
Engineering- ASCE (August 01, 2007).
3 Bhuiyan, M.I.H, Mavinic, D.S., and Beckie, R.D. (2007) Nucleation and growth kinetics
of struvite in a fluidized bed reactor. Submitted to Journal of Crystal Growth (May 10,
2007).
4 Bhuiyan, M.I.H, Mavinic, D.S. (2007) Assessing struvite precipitation in a pilot-scale
fluidized bed crystallizer. Submitted to Environmental Technology (April 16, 2007).
xv 5 Bhuiyan, M.I.H, Mavinic, D.S., and Beckie, R.D. (2007) Dissolution kinetics of struvite
grown in a pilot-scale crystallizer. Submitted to Journal of Environmental Engineering
and Science (May 01, 2007).
6 Bhuiyan, M.I.H, Mavinic, D.S., and Koch, F.A. (2007) Thermal decomposition of
struvite and its phase transition. Chemosphere, in press.
xvi Acknowledgements
I feel immense pleasure in expressing my gratitude to Professor Donald Stephen Mavinic who was the main supervisor of my Ph.D. study. My heartfelt thanks go to him for his guidance, patience, and invaluable support throughout this research. He was always very particular and professional in extending his support as a supervisor; timely disposal of everything from his end always inspired me and raised my spirits on many occasions. His intelligent and prudent direction and excellent behavior were the sources of encouragement and vigor to move from one step to another. I owe him a lot for what I have learnt from him to go forward in my professional life.
My special thanks go to my committee members for their invaluable support for this research. I would like to thank Dr. Roger Daniel Beckie, Department of Earth and Ocean
Science, UBC, for his support in exploring the world of equilibrium chemistry. It was so pleasing for me that he could easily understand any problem that I approached with, and gave a very quick and reasonable direction to find a solution. I really appreciate his critical thinking and spirit to carry out an innovative research. I am very much grateful to Dr. Ken. J. Hall, Dr. Victor
Lo and Dr. Bernard Laval, Department of Civil Engineering, UBC, for their invaluable support and fruitful discussions as committee members.
I would like to thank the Natural Sciences and Engineering Research Council (NSERC) of Canada, for providing financial support to my supervisor, from which he funded my research.
The present research would not have been possible without the special initiative taken by my supervisor.
The Environmental Engineering Group at UBC has a very homey atmosphere, where everyone is helpful and encouraging. I am especially grateful to Fred A. Koch, Research
xvii Associate and Pilot Plant Manager, UBC, for his valuable support and discussion in different spheres of this research. Special thanks are also due to Susan Harper and Paula Parkinson, UBC
Environmental Engineering Laboratory. Their cooperation and assistance with the analytical work are highly appreciated. I really enjoyed the company and time I spent with some former
and present students. I also enjoyed the regular discussions held among the research students presently working in different aspects of the struvite crystallization. I thank everyone in the
"struvite" discussion group for providing their valuable comments and encouragements. I would
like to thank Alessandro Monti, Zahid Mahmood, Umberto Preciado, Annette Muttray, Rob
Simm, Dean Shiskowski, Ali Adnan, Ahren Britton, Alexander Forrest, Daniel Potts, Wendong
Tao, Saifur Rahman, Kazi Parvez Fattah, Melissa Zhang, Raphael Fugere, and Ellizabeth Tilly,
for making my time at UBC so enjoyable.
I would like to thank Dr. Mark Mclakhlan, Anita Lam, and Dr. John R. Scheffer in the
Chemistry Department at UBC for their support in conducting thermogramimetric analysis, X-
ray Crystallography (XRD) and InfraRed (IR) Analyses respectively. I would also like to thank
Sally Finora in the Mining Engineering Department, UBC, for helping me with surface area
determination by the BET method. I am also grateful to James Thorn, PhD student in Earth and
Ocean Science Department at UBC, for his valuable input and discussion in understanding
PHREEQC.
My profound gratitude goes to my father, who is no more in this world, and to my
mother, whose constant wishes remain with me all the time. Finally, I would like to thank my
wife Lina from the core of my heart, for extending her outstanding support, and for her patience
in raising our children Ibtida and Ishmam while I was busy in the lab. Thanks Ismam and Ibtida
for your patience.
xviii Contribution of others
Collaborations and linkage with several parties have provided valuable support for this thesis research. The submitted papers and manuscripts in preparation of this thesis have been strengthened by input given by my supervisor Dr. D.S. Mavinic, coauthors, scientific colleagues and anonymous reviewers. Below is a summary of the contributions of coauthors to each chapter.
Chapter 2, 3, 4 and 6
As a very important and integral part of my thesis research, I had several important discussions on equilibrium chemistry with Dr. Roger Daniel Beckie of Earth and Ocean Science
Department at UBC. For speciation and to find a solution to other relevant problems related to precipitation, Dr. Beckie helped me a lot. He also extended his all possible support in understanding the chemical equilibrium model, PHREEQC. His valuable suggestions provided useful insights on the research activities and manuscript content. He also contributed to a critical review of the manuscripts.
Chapter 7
The present research was a part of a broader study in optimizing the struvite crystallization system developed at UBC. A group discussion was held regularly among all the research students in the presence of Fred A. Koch, Manager, UBC Pilot plant and my supervisor
Dr. D.S. Mavinic. Critical inputs and discussions from Fred A. Koch were always very helpful for coordination among the different components of the research. He was very helpful and resourceful in setting up bench-scale and pilot-scale reactors, as well as other instrumentation for this study. He also provided helpful suggestions and contributed with a critical review of the manuscript.
xix Chapter 1 Introduction
1.1 Preface
The limited extent and the gradual depletion of phosphate reserves in the world has long been a matter of grave concern; this has resulted in a growing demand for sustainable phosphorus (P) resources in the industrialized world. Research is currently underway in many developed countries, to recover phosphorus from wastewater, since domestic sewage offers a great potential for recovery and recycle of phosphorus (Driver et al. 1999).
The release of phosphorus to surface waters, and its subsequent contribution to eutrophication, has led to increasing concerns about water quality. Policies are, therefore, being implemented globally to reduce the level of phosphorus entering surface waters from domestic, industrial and agricultural wastewater. Phosphorus from wastewater is traditionally removed via chemical treatment or through enhanced biological phosphorus removal (EBPR) processes.
Struvite, magnesium ammonium phosphate hexahydrate, or MAP (MgNF^PCV 6H2O) is a crystalline phosphate mineral that often accumulates on equipment surfaces in anaerobic digestion and post-digestion processes in the wastewater treatment industry. This causes some particular concerns for biological nutrient removal (BNR) processes. This problem plagues the industry commercially through major downtime, loss of hydraulic capacity, and increased pumping and maintenance costs (Doyle et al, 2002). A novel solution to this problem is to recover phosphate from wastewater as struvite, before it forms and accumulates on wastewater treatment equipment. This solves a wastewater treatment problem and provides an environmentally sound and renewable nutrient source for the agriculture sector. Besides its potential use as a fertilizer, another advantage of this process is the reduced volume of generated
1 sludge, compared to the chemical process. Through the assorted benefits of pollution control, waste management, and resource recovery, the P-recovery process, through struvite formation, has already ensured its contribution to environmental sustainability. Due to all these promising
aspects, the recovery and recycling of phosphorus may soon become, economically and
ecologically, an attractive option for a new generation of wastewater treatment plants.
The P-recovery process and characteristics of the recovered material could be further
optimized, through extensive research (Battistoni et al. 2001). Predicting struvite precipitation potential is critical to designers and operators for anticipating potential struvite problems and for
removing existing problems, due to struvite deposition (Ohlinger, 1999). Before recovery can be
truly optimized, a clear understanding of the solubility, nucleation, growth and dissolution
through experimental investigation of struvite thermodynamics and its kinetics (especially in real
supernatant/centrate) is important.
The phosphorus recovery research project has been continuing in the Department of Civil
Engineering at the University of British Columbia (UBC) since 1999. The basic design of the
pilot-scale UBC MAP Crystalliser, based on fluidized bed systems, was found effective in
recovering phosphate from both real and synthetic supernatant (Adnan et al, 2003a; Britton et
al, 2005). A first large-scale struvite recovery demonstration reactor has been built by Ostara
Nutrient Recovery Technologies Inc, at the Goldbar Wastewater Treatment Facility, Edmonton,
Canada, based on the fluidized bed reactor technology developed by UBC. Once this pilot scale
is successful, a full scale facility, involving five commercial scale reactors will be built to cover
the plant's full capacity of sludge lagoon supernatant (Scope Newsletter, 2006). In the
commercialization process of UBC MAP crystallizer, a knowledge of struvite thermodynamics
2 and its growth and dissolution kinetics thus attained, and their application in fluidized bed reactors (FBR), would help enhance its feasibility and control total production costs.
1.2 Literature review
1.2.1 Why recover phosphorus?
Phosphorus is of fundamental importance to all living things, including humans, to lead healthy and productive lives; it is also an essential nutrient for crop production. It is predicted that P demand will increase by at least 1.5% each year and the resource could be exhausted in as early as 90 years (Shu et al, 2006). Global initiative to increase corn production for the eventual production of ethanol, as a biofuel, may reduce this lifespan considerably. Even if these reserves
could be processed economically in the future, they are not a renewable resource and therefore, recovering phosphorus from waste streams is a significant breakthrough. Reducing phosphorus released in wastewater has important environmental implications, as the phosphorus released into
water bodies from wastewater treatment plants can cause eutrophication in rivers and lakes.
Struvite precipitation is a recognized problem in sludge handling at many wastewater
treatment plants. It is likely to increase with the current trend towards biological nutrient
removal. A number of treatment plants have reported the occurrence of unintentional struvite
formation in plant piping, and other equipment (e.g. pumps, valves, filter belts, etc.) (Borgerding,
1972; Booram, 1975; Snoeyink and Jenkins, 1980; Stumm and Morgan, 1981; Ohlinger et al,
1999). This is due to the fact that magnesium, ammonium and phosphate are released as the
result of solids degradation in the subsequent digestion process. Struvite precipitation occurs in
2+ + 3 solution when the product of the effective concentration (activities) of Mg , NH4 and P04 ~
exceeds the struvite solubility limit. Availability of the three components can be controlled by
system pH and the total dissolved concentrations of magnesium, ammonium and the phosphate
3 species. The problem of unintentional struvite formation is more severe in anaerobic digestion since the supernatant from anaerobic digestion is rich in two important constituents of struvite, namely ortho-phosphate and ammonium, and the pH of anaerobic digestion and post-digestion processes is generally higher (Ohlinger et al, 1999).
To date, several remedial measures have been suggested for alleviating the problem of unintentional struvite formation in wastewater treatment plants. Solutions so far have included installation of water softening devices before and after sludge digestion, precipitating phosphorus by the addition of ferric chloride, diluting digester sludge with secondary effluent, adding meta- phosphates or other scale inhibitors, acidifying the waste stream and redesigning certain areas of the plants (Williams et al, 1999; Mamais et al, 1994). All of the above remedial measures are costly and, at times, only alleviate the problem, without eliminating it completely.
Intentional struvite formation appears to represent a practical solution, which, in turn, may solve many of the above-mentioned problems. If P is recovered in the form of struvite, the concentrations of soluble phosphate, ammonia and magnesium can be reduced significantly and hence, the struvite build-up problem faced by the wastewater treatment industry can be significantly reduced, or even eliminated entirely.
1.2.1.1 Phosphorus removal technologies
Recognized phosphorus removal technologies include chemical precipitation, biological phosphorus removal, crystallization, tertiary filtration, and ion exchange (Morse et al, 1998).
Most of these processes produce wastes which need to be landfilled or incinerated. The ethics of sustainability, however, makes these options unattractive (Durrant et al, 1999). Biological P- removal is mainly used for low P concentrated municipal wastewater. Crystallization processes stand out because they not only achieve high P-removal, but also recover P from wastewater as
4 useful products, including struvite, calcium phosphate, and hydroxyapatite. The recovery of
phosphorus by crystallization has been reported to reduce sludge volume, under specific
conditions, by up to 49%, compared to chemical phosphorus removal (Woods et al, 1999).
Struvite and hydroxyapatite can be used in agriculture as fertilizers; struvite, however, is
preferred for numerous reasons. First, nutrients are released as a slower rate compared to other
fertilizers. Plants can take up the nutrients before being rapidly leached, and less frequent
application is therefore required (Manch and Barr, 2001). Second, the impurities caused by
heavy metals in the recovered struvite are two or three orders of magnitude lower than that of
commercial phosphate fertilizers (Brett et al, 1997). Lastly, the essential nutrients P, N, and Mg
are applied simultaneously, without any unnecessary components in the fertilizer. The process
also has physical advantages in addition to the chemical advantages. The footprint of the struvite
crystallization reactor is considerably smaller than that of biological P-removal infrastructure,
and the process has fewer problematic operational concerns (Wang et al, 2005). Thus, the major
driving forces linked with P-recovery through struvite formation are: the potential to reduce the
diminishment of global phosphate reserves; the potential for cost savings (chemical and sludge
handling cost) through improved sewage sludge management; the potential to reduce the
phosphorus and nitrogen load of side stream and sludge liquors recirculated to the head of the
wastewater treatment works; the potential to enhance phosphorus removal through improvement
of the biological nutrient removal efficency in sewage treatment plants; the potential to control
struvite encrustation in enhanced biological nutrient removal (EBNR) plants; the potential for
cost recovery through the sale of recovered phosphate; increasing the use of struvite as a slow
release fertilizer; and the support of the important principle of sustainability.
5 Hence, the conditions for its formation that can be found naturally within the environment of wastewater treatment works, can be exploited for extraction of struvite as a commercial product and close the phosphorus loop in the soil-crop-animal-human-soil cycle, thus contributing to an ecologically sustainable future.
1.2.1.2 Phosphorus crystallization processes
Considerable world-wide research has been undertaken on phosphate crystallization techniques. A number of different techniques are used to extract phosphorus from wastewater through crystallization, in a dedicated reactor. The DHV Crystalactor ™ process is based on a fluidized bed reactor, in which calcium phosphate crystallizes on a seeding grain, typically sand.
The phosphate containing wastewater is pumped in an upward direction, maintaining the pellet bed in a fluidised state. In order to crystallize the phosphate on the pellet bed, a driving force is created by a reagent dosage or sometimes pH adjustment. Due to high rate of crystallization, a short retention time is required. During the operation, the pellets grow and move towards the reactor bottom. At regular intervals, a quantity of the largest fluidized pellets is discharged at full operation from the reactor and fresh seed material is added (Giesen, 1999). In the Netherlands, three full-scale P-recovery plants had been installed at municipal wastewater treatment plants in the past. Now, only one plant is in operation due to the decreased P concentration (3-4 mg l"1) in the biological treatment unit (Giesen, 1999).
In the Unitika Phosnix Process, wastewater is fed into the base of the reactor, where it is mixed with magnesium chloride, to achieve a desired Mg:P molar ratio. A blower forces air into the base of the column, providing the agitation required for complete mixing and suspension of the growing particles. The crystals grow in size until they sink to the base of the tower, where they are periodically removed (Stratful et al. 1999). There are a number of full-scale and pilot-
6 scale P-recovery operations documented, based on the Unitika Phosnix Process (CEEP, 2001). In
Japan, three years experience of operating and selling recovered struvite from full-scale plants has been well documented (Ueno and Fujii, 2001).
The Kurita Process uses phosphate rock as seed grains. Wastewater is introduced from the base of the column and travels upward through it. Unlike the Unitika process, this process does not employ air agitation in the reactor (Stratful et al, 1999).
The UBC MAP crystallizer design also utilized a fluidized bed. Unlike the above three processes, it has four different areas of cross section, increasing from bottom to the top, with a settling zone at the top. The diameter changes caused turbulent eddies above each transition, ensuring that sufficient mixing existed in the reactor from top to bottom (Britton et al, 2005).
The larger crystals accumulated at the bottom and were harvested periodically. The bigger size pellets are harvested through a port in the bottom section. Wastewater is fed into the bottom of the reactor, along with the recycle stream. The injection port at the bottom facilitates high energy mixing. Based on previous experience at a pilot scale reactor, designed and tested at the UBC
Environmental Engineering Pilot Plant using a synthetic feed (Adnan et al. 2003b), two scaled- up pilot-scale reactors were installed at the Lulu Island Waste Water Treatment Plant (WWTP).
One of these two reactors was used in this study, which was operated for two months, from mid-
May to mid-July, 2006.
Research suggests 'the ideal location' for recovery of struvite requires that the flow should have a high concentration of soluble ortho-phosphate and ammonium, a low concentration of suspended solids and a relatively high P load (Williams, 1999). In general, these requirements mean recovery is not possible from the mainstream of standard activated sludge plants, trickling filters and anaerobic treatment processes, as the concentrations of phosphate and
7 ammonium are too low. However, anaerobic digestion releases the elevated levels of P into the supernatant, which can be used as a potential source. EBNR plants tend to eliminate ammonium and phosphate, while the anaerobic digestion of EBNR sludge releases even higher phosphate and ammonium (Gaterell et al, 2000). Thus, the most appropriate place for struvite formation
and recovery is from the supernatant of anaerobic sludge digester of an EBNR plant, or from the
centrate from sludge filter presses or centrifuge.
A Pilot-scale, UBC crystallizer operation was successfully operated, both in an EBNR
Plant (City of Penticton WWTP) (Britton et al, 2005), and a Secondary Treatment Plant with a
Trickling Filter (Lulu Island WWTP) (Fattah, 2004). The struvite pellets produced are more than
97% pure (containing no foreign seed material, such as sand) and with diameters three times
larger than those produced by commercially operational reactors in Japan (Scope newsletter,
2006).
1.2.2 Struvite Solubility and Thermodynamics
Struvite is a white crystalline substance, consisting of magnesium, ammonium and
phosphate in equal molar concentrations (MgNH4P04-6H20). The chemistry of struvite, with
regard to the wastewater industry, is inexorably linked with solubility; for this reason struvite
chemistry is discussed with solubility being the key issue (Doyle et al, 2002). Struvite
precipitation can be separated into two stages: nucleation and growth. Nucleation occurs when
constituent ions combine to form crystal embryos; the crystal growth continues until equilibrium
is reached (Ohlinger et al, 1999). In systems continuously replenished with struvite constituents;
e.g. wastewater treatment plants, crystal growth may continue indefinitely, unless it is controlled.
Hence, a comprehensive understanding of the struvite solubility is crucial for the efficient
management of the system.
8 Struvite precipitation is controlled by pH, degree of supersaturation, temperature, presence of other ions and ionic strength of the aqueous system (Ohlinger et al., 1999;
2+ Bouropoulos et al., 2000) and can occur when combined activities of Mg , NFL* and PO43"
exceed the solubility product for struvite. The equilibrium constant for a reaction involving a precipitate and its constituent ions is known as solubility product (Tchobanoglous, 1985). A poorly soluble salt will dissolve in water until there is a dynamic equilibrium between ions,
leaving the solid to go into the liquid and ions passing from the liquid to the solids. In some
simple cases, the equilibrium conditions of a salt AxBy can be described in terms of following
expressions:
z+ z xA +yB ~ <=> AxBy
2 z {A Y {B T = Ksp
where x and y are stoichiometric coefficients or numbers of cations and anions respectively and
z+ and z~ are their valencies. In case of struvite, the product of the activities of each of the struvite
ions is referred to as ion activity product or IAP. When the solution is in equilibrium with the
solid phase, the IAP is called the solubility product, Ksp (Buchanan et al, 1994).
The solubility product published by a number of authors display a range of values.
Published values of pKsp, i.e. -log Ksp for struvite range from 12.6 to 13.36. The most commonly
used value in engineering texts is 12.6 (Snoeyink and Jenkins, 1980; Stumm and Morgan, 1996).
Thermodynamically, there should be a single value of the solubility product (Ksp) at a certain
temperature, that should apply to all solutions as long as it is possible to determine the activity of
each chemical species accurately. The thermodynamic solubility product, at the zero-ionic
strength standard state, can be determined by extrapolation of non-zero experimental data to zero
ionic strength. As a simple means of determining the saturation state of the supernatant/ centrate
9 being treated, the concept of conditional solubility is also used (Snoeyink and Jenkins, 1980).
The conditional solubility product of struvite, derived from the total analytical concentrations of
the constituents, illustrates a situation where more than one of the dissolving species is affected
+ 3 by solution pH. These species are the ammonium ion (NH4 ) and phosphate ion (P04 ~). Because
an increase in pH will decrease the ammonium ion concentration and increase the phosphate ion
concentration, it follows that there should be a pH value where the solubility of struvite is a
minimum. A number of pH values have been suggested as the pH of minimum struvite solubility
ranging from 8-10.7 (Doyle et al, 2002).
Chemical equilibria are governed by the rules of thermodynamics, so it should not come
as a surprise that temperature affects them. A change in temperature results in a shift in the
reaction equilibrium towards either products or reactants, depending on whether the reaction is
endothermic or exothermic (or has a positive or negative enthalpy). Thus, the equilibrium
constants maintain a relationship with temperature and the enthalpy of the reaction.
1.2.3 Estimation of Ionic Strength from Electrical Conductivity
Research that has taken into account the ionic strength of solutions observed and
suggested greater solubility of struvite (Doyle et al, 2002). This can be explained by the impact
that ionic strength has upon activity of individual ions. The ionic strength of a solution directly
affects the activity of each ion. The interaction of ions with other ions in solution often, but not
always, leads to an increase in solubility. Essentially, struvite solubility requires an appreciation
of ionic strength, with the associated effect upon activity and the relevant species that form in the
solution. One issue, with respect to accurate prediction and modeling of struvite, is which ions in
wastewater must be considered and to what concentration. Since all ions and associated ion-pairs
will contribute to the ionic strength of a solution, where should the cut-off point be for specific
10 ions and what concentrations should be included, when making ionic strength and ion activity calculation (Doyle et ai, 2002)? Due to the complexity in the determination of ionic strength for complex solutions, such as wastewater or supernatant, it is sometimes desirable to use an approximation of ionic strength derived from a correlation with electrical conductivity (EC).
This, in turn, will help predict the struvite solubility during intentional struvite formation.
Besides the amount and composition of ionic species, EC is strongly dependent on temperature. It is important to minimize unnecessary errors resulting from inaccurate temperature correction, especially when correction has to be made over a large temperature range. The proposed study has the objective to recommend a representative compensation factor for the supernatant/centrate used as feed, for struvite crystallization.
1.2.4 Precipitation kinetics of struvite
Precipitation kinetics can be separated into two phases, nucleation and growth.
Nucleation occurs when ions combine to form crystal embryo that can act as the foundation for growth into detectable crystals. Growth results from the assimilation of ions in the lattice structure established by a crystal embryo foundation (Ohlinger, 1999).
The nucleation corresponds to the apparition of new particles. Several mechanisms are possible for nucleation, depending on the supersaturation level in the crystallizer (Fig. 1.1). Each of these nucleation types has a metastable zone of its own.
The time that elapses between the establishment of supersaturation and the first changes in the physical properties, due to the formation of solid phases, is called the crystallization induction period. Published information about struvite precipitation kinetics is limited to only a few studies. Gunn (1976) developed expressions for struvite induction time in homogeneously nucleating solutions. It was concluded in that study that, once nucleation had occurred, crystal
11 growth rate exceeds the nucleation rate, because of greater activation energy required for nucleation of non-associated ions compared to the activation energy for deposition of ions on a crystal surface. Ohlinger (1999) found that a high degree of supersaturation would maximize the nucleation rate, while thorough mixing would maximize crystal growth.
1.2.4.1 Metastable limit for struvite
From the viewpoint of thermodynamics, supersaturated solutions are always unstable. Up to a certain concentration, however, they appear to be stable in that their properties do not change for a comparatively long period of time. This metastable state is found for supersaturated solutions for both readily and sparingly soluble substances. Isothermal and polythermal methods of metastable limit determination have been reported in the literature (Sohnel and Garside,
1992). No such method has been reported to be used in literature in the case of struvite. A previous researcher used the difference of the experimental pH's for spontaneous precipitation and pH's of minimum solubility, to determine the metastable limit of struvite (Ali and Schneider,
2005). A suitable metastable limit determined for struvite system would help in estimating growth and optimizing the process, by avoiding nucleation. Also, the growth rate, determined independent of nucleation, can be used to develop a growth model in fluidized bed reactors.
1.2.4.2 Crystal growth model
The systems previously studied for determination of growth models are mostly soluble salts, using seeded techniques. Similar studies of sparingly soluble salts are limited because large, single-seed, crystals of the system are difficult to prepare and the application of the technique was limited in the past (Tai et al., 1999). The growth of crystals from solution is influenced either by mass transport from solution to the crystal surface or by incorporation of material into a crystal lattice through the surface integration process (Sohnel and Garside, 1992).
12 Several mechanisms regarding crystal growth have been proposed in the literature (Mullin,
2003), and a number of researchers have conducted studies on crystal growth of different substances and several models have been proposed. With the determination of the mass-transfer coefficient and surface-reaction co-efficient for a specified condition, a two-step linear growth rate model was found to be effective in a fluidized bed reactor (Tai et al, 1999).
1.2.5 Pilot-scale fluidized reactor operation
Controlled struvite precipitation reduces downstream scaling potential in wastewater treatment processes. Reduction of growth rate limitations has been experimentally demonstrated with the input of mixing energy. Energy input disrupts concentration gradients in boundary layers surrounding growing crystals and increases the struvite crystal growth rate. The fluidized bed reactors process, coupled with pH control, can provide the high supersaturation and mixing energy environment necessary for rapid struvite precipitation (Ohlinger, 1999).
Based on the studies completed so far, phosphate recovery, through the struvite crystallization process, has been found to be affected by several operating parameters such as: supersaturation (Ohlinger, 1999; Bouropoulos et al., 2002;); pH (Munch et al, 2001; Doyle et al.,
2002); magnesium to phosphorus molar ratio (Mg:P) (Adnan et al., 2003a; Munch et al, 2001); crystal retention time (Adnan et al., 2003a); recycle ratio (Adnan et al, 2003a); reactor seeding
(Ohlinger, 1999; Wu and Bishop, 2004); temperature (Andrade et al, 2001; Doyle et al., 2001); turbulence and mixing (Ohlinger, 1999; Regy et al, 2002).
Using the experimentally-determined thermodynamic solubility product and the knowledge of precipitation kinetics, the factors that affect the crystallization of struvite in
fluidized bed reactor can be optimized through pilot scale studies.
13 1.2.6 Dissolution kinetics and slow release property of struvite
If both crystallization and dissolution processes were purely diffusional in nature, they should exhibit a true reciprocity; the rate of crystallization should equal the rate of dissolution at a given temperature and under equal concentration driving forces, i.e. at equal displacements
away from the equilibrium saturated conditions. These conditions are rarely obtained in practice
(Mullin, 2001). In reality, dissolution of a mineral depends on different factors.
Whilst struvite can be viewed as a problem in wastewater treatment plants, it has potential use as a fertilizer, once it is intentionally crystallized. Compared to other fertilizers, the benefits of using struvite are low leach rates and prolonged release of nutrients throughout the
growing season of plants, with the possibility of only one single application (Gaterell et al,
2000). The slow release property of struvite is well documented (Bridger et al, 1962).
Some studies have been undertaken, using batch reactors and mixed flow-through
reactors, on the dissolution kinetics of different minerals (Golubev et al, 2006; Kraemer et al,
1998). A very limited number of studies have been carried out on dissolution of struvite. A study
was conducted in a batch reactor to determine the dissolution kinetics of synthetic struvite at
three different temperatures (Babic-Ivancic et al, 2002). Dissolution experiments, both in batch
reactors and mixed flow-through reactors, under different conditions, would help gain a better
understanding of the dissolution kinetics of struvite formed in fluidized bed reactor. Thus, its
slow release property can be established as a fertilizer.
1.2.7 Thermal decomposition of struvite and its phase transition
The decomposition behavior of struvite at various temperatures and the mechanism of its
transformation to various other associated forms, are essential to obtain a good grasp on the
struvite crystallization system, as well as its end use. The conditions responsible for such
14 transformations are diverse. Apparently, the fragile equilibrium of struvite in solution leads to the presence of other crystal phases (Andrade and Schuiling, 2001). The formation of
magnesium phosphates such as MgHPCVSFbO (newberyite), Mg3(P04)2-8H20 (bobierrite) and
Mg3(P04)2-22H20 (cattiite), during struvite crystallization or dissolution, is reported in the literature. Struvite also faces different conditions depending on its uses and applications. It is
suggested that the availability of P in the soil is different for different phases of struvite, due to variable dilution with water of crystallization (Bridger et al, 1962).
A limited number of studies have been conducted on the decomposition behavior of
synthetic struvite, at different heating rates (Sarker, 1991; Frost et al, 2004). A systematic thermogravimetric study of both synthetic struvite and struvite pellets grown in the fluidized bed reactors would ascertain its fate at different temperatures and different heating rates.
1.3 Research objectives
From the literature review presented in the previous section (1.2), it appears that there is a wide
scope for conducting research into different aspects of P-recovery, through struvite
crystallization. Struvite crystallization is a chemical process, involving numerous equilibria and
myriads of different ions. To accurately estimate their contribution, a chemical equilibrium
model selection is important. A number of chemical equilibrium-based models have been
developed and used by previous researchers. Equilibrium based computer models such as
STRUVITE, MINTEQ+, MINTEQA2 and MINEQL+ have been used by previous researchers
(Loewenthal et al, 1994; Ohlinger, 1999; Miles and Ellis, 2001; Josan et al, 2005). A chemical-
equilibrium based modelling package, "PHREEQC", has also been used by several researchers
(Bourpoulos et al, 2004; Ali, 2005). In order to strengthen the scope of other objectives,
selection of a reliable chemical-equilibrium model was a key objective of this study.
15 Solubility, being the key issue in the crystallization process, needs to be addressed properly. Although there had been several studies on solubility of struvite, a true thermodynamic
solubility product, at zero ionic strength in a standard state, is not available in the literature. The
solubility product and solubility conundrum has also been an important issue found in the
literature (Clark and Judith, 1998). There are only a few cases in which solubility and published
solubility product (Ksp) values are related in a simple way (Clark and Judith, 1998; Harris, 2000).
However, the solubility data commission of International Union of Pure and Applied Chemistry
(IUPAC) has taken the initiative to publish critically select solubility data (Letcher and Battino,
2001). Determination of a thermodynamic solubility product of struvite, for zero ionic strength
standard state conditions, with a relationship among the values at different temperatures, and to
investigate how it is related to solubility, was another objective of this study.
Due to the complexity in the determination of ionic strength for complex solutions such
as wastewater or supernatant/centrate, it is sometimes desirable to use an approximation of ionic
strength derived from a correlation with electrical conductivity. No such relationship has been
found yet in the literature for wastewater and the solution matrix that we deal with for P-
recovery, through struvite crystallization. Considering the dependence of electrical conductivity
on temperature in real supernatant/centrate, developing a relationship between ionic strength and
electrical conductivity was another objective of this study.
Information on the induction time, metastable limit, nucleation and growth of struvite is
also limited in the literature. To accurately estimate the struvite precipitation in the fluidized bed
reactor, a properly determined induction time of struvite, at different conditions, has a wide
application potential. Also, the knowledge of growth kinetics of struvite would certainly help in
16 estimating struvite growth in a fluidized bed reactor. In order to achieve that objective, a growth rate model development was also deemed important.
In this study, a pilot-scale reactor, developed at The University of British Columbia
(UBC), was operated at the Lulu Island Wastewater Treatment Plant (LIWWTP), in Richmond,
BC, Canada. The specific research objective of this part of the study was to assess the precipitation potential of struvite, by calculating the saturation index with the help of the selected model PHREEQC (version 2.12) (Parkhurst and Appelo, 1999), using the experimentally- determined thermodynamic solubility product and its temperature dependence. In the commercialization process of UBC MAP crystallizer, knowledge of struvite thermodynamics and solubility and their application in fluidized bed reactors (FBR) are important. An attempt has also been made to identify the most important parameters and their impacts in P-removal, through struvite crystallization in the fluidized bed reactors.
In connection with the potential use of struvite as a fertilizer and its slow release property, the investigation of dissolution kinetics of struvite, formed at Lulu Island Wastewater
Treatment Plant in a batch and mixed flow-through set up, was set as another objective of the study.
Since temperature has a profound effect on struvite solubility, both during formation and end use, the determination of the fate of struvite at different temperatures and heating rates was deemed important and was also included as one of the objectives.
1.4 Thesis outline
As indicated in the Preface, this thesis is presented in manuscript-based format. The individual papers (manuscript-chapters) are presented in a logical manner that directs the readers towards the development of a conceptual model, while addressing the objectives.
17 A systematic solubility study at temperatures in regular intervals is absent in the literature. To determine a true thermodynamic solubility product at different temperatures, experiments were conducted at a range of ionic strengths and extrapolated to determine solubility products at zero ionic strength standard sate. This rigorous study on solubility and thermodynamics is presented in the manuscript attached as Chapter 2.
To this author's knowledge, no one has yet determined the temperature dependence of electrical conductivity, and a relationship between electrical conductivity and ionic strength for wastewater, especially for an aqueous solution matrix related to struvite formation. Since anaerobic digester supernatant/centrate has been identified as the most appropriate place for struvite formation and recovery (Gaterall et al., 2000), a relationship was determined for samples collected from five different Wastewater Treatment Plants in Canada. This relationship would facilitate the use of electrical conductivity in the operation of a crystallization system, for struvite
formation. The results of this research are presented in the manuscript attached as Chapter 3.
A kinetic study of struvite is also limited in the literature. Struvite crystallization in
fluidized bed reactors is controlled by mass-transfer resistance, as well as surface-reaction resistance (Tai, et al, 1999). Mass-transfer resistance is dependent on hydrodynamic conditions
inside the reactor. A detailed study on the effects of hydrodynamic conditions on crystallization
(which involves another student in this research group) was not within the scope of this study.
However with the determination of a metastable limit of struvite, different mechanisms of
struvite precipitation were investigated and concluded with a proposed growth model in this
study. This research has been presented in the manuscript attached as Chapter 4.
Experimentally-investigated, struvite thermodynamics and solubility results were applied
in a pilot-scale FBR at the Lulu Island Wastewater Treatment Plant to augment the
18 commercialization process of the UBC MAP crystallizer. Identification of other parameters, to optimize the struvite crystallization process in an FBR, was also made. The results of this research have been presented in the manuscript attached as Chapter 5.
The potential use of struvite pellets as fertilizer has widened the scope of this technology
as a sustainable one. Its slow release property has made it more acceptable as a fertilizer. A
study on dissolution kinetics of struvite is also very rare in the literature. A detailed investigation
into the dissolution kinetics of struvite has made in batch and mixed, flow-through conditions
and is presented in the manuscript attached as Chapter 6.
The thermogravimetric analysis of struvite pellets grown in a fluidized bed reactor is also
not available in the literature. To this end, a detail study was conducted to know the temperature, heating rate and other conditions that cause transformation of struvite to its associated phases.
The results of the study have been presented in the manuscript attached as Chapter 7.
This chapter (Chapter 1) has provided an introduction to the theme and objectives of the
study with literature review, while Chapter 8 concludes the research findings with identification
of overall significance of the thesis research to the field of study and provides specific
recommendations for future work.
19 / / /_/ /^r~y-™j>~y— / / / Metastable zone limit of / / / homogeneous primary nucleation I / / / / f *"f •/ • Metastable zone limit oi / / / heterogeneous primaiv nucleation
yy / y* y Metastable zone limit of surface secondary nucleation
Solubility
lempsratuft
Figure 1.1 Metastable widths for different possible mechanisms of nucleation.
20 References
Adnan, A., Fredric A.K. and Donald S.M. (2003a) Pilot-scale study of phosphorus recovery through struvite crystallization — examining the process feasibility, Journal of 'Environmental Engineering and Science 2, 315-324.
Adnan, A., Koch, F.A. and Mavinic, D.S. (2003b) Pilot-scale study of phosphorus recovery through struvite crystallization- II. Applying in-reactor supersatuartion ratio as a process control parameter. Journal of Environmental. Engineering and Science 2, 473-483.
Ali, M.I. and Schneider, P.A.. (2005) Crystallization of struvite from metastable region with different types of seed crystals. Journal of Non-Equilibrium Thermodynamics 30, 95-111.
Andrade, A. and Schuiling, R. D. (2001) The chemistry of struvite crystallization. Mineralogical Journal (Ukraine) 23, 37-46.
Battistoni, P., De Angelis, A., Pavan, P., Prisciandro, M., and Cecchi, F. (2001) Phosphorus removal from a real anaerobic supernatant by struvite crystallization. Water Research 35(9), 2161-2178.
Bavic-Ivancic, V., Kontrec, J., Kralj, D. and Brecevic, L. (2002) Precipitation diagram of struvite and dissolution kinetics of different struvite morphologies. Croatica Chemica et Acta 75, 89-106.
Booram, C, Smith, R. and Hazen, T. (1975) Crystalline phosphate precipitation from anaerobic animal waste treatment lagoons liquors. Transactions of theASAE. 18, 340-343.
Borgerding, J. (1972) Phosphate deposits in digestion systems. Journal of the Wastewater Pollution Control Federation 44, 813-819.
Bouropoulos, N. and Koutsoukos, P. (2000) Spontaneous precipitation of struvite from aqueous solutions. Journal of Crystal Growth. 213, 381-388.
Bouropoulos, C, Vagenas, N., Klepetsanis, P., Stravropoulos, N., and Bouropoulos, N. (2004) Growth of calcium oxalate monohydrate on uric acid crystals at sustained supersaturation. Crystal Research Technology 39, 699-704.
Brett, S., Guy, J., Morse, G.K., and Lester, J.N. (1998) Phosphorus Removal and Recovery Technologies, Selper, London.
Bridger, G.L., Murrel, L.S. and Starostka, R.W. (1962) Metal ammonium phosphates as fertilizers. Journal of Agriculture and Food Chemistry 10 (3), 181-188.
Britton, A., Koch, F.A., Mavinic, D.S., Adnan, A. Oldham, W.K., and Udala, B. (2005) Pilot- scale struvite recovery from anaerobic digester supernatant at an enhanced biological phosphorus
21 removal wastewater treatment plant. Journal of Environmental Engineering and Science 4, 265- 277.
Buchanan, J., Mote, C, and Robinson, R. (1994) Thermodynamics of struvite formation.
Transactions of the ASAE. 37, 617-621.
CEEP. (2001) Phosphate recovery: where do we stand today? Special issue of the scope newsletter, published in preparation to 2nd international conference on P-recovery from human and animal wastes, 12-14 March, 2001, Noordwijkkerhout, The Netherlands.
Clark, R.W. and Judith, M.B. (1998) The Ksp-Solubility Conundrum, Journal of Chemical Education 75, 1182-1185.
Doyle, J.D. and Simon, A.P. (2002) Struvite formation, control and recovery. Water Research 36, 3925-3940.
Driver, J., Lijmbach, D., and Steen, I. (1999) Prospects for recovery of phosphorus from animal manures: a review. Environmental Technology 2, 697-708.
Durrant, A.E., Scrimshaw, M.D., Stratful, I., and Lester, J.N. (1999) Review of the feasibility of recovering phosphate from wastewater for use as a raw material by the phosphate industry. Environmental Technology 26(5-6), 987-996.
Fattah, K.P. (2004) Pilot Scale Struvite Recovery Potential from Centrate at Lulu Island Wastewater Treatment Plant. Masters thesis, Department of Civil Engineering, The University of British Columbia, Canada.
Frost, R.L, Weier, M.L. and Erickson, K.L., (2004). Thermal decomposition of struvite- Implication for the decomposition of kidney stone. Journal of Thermal Analysis and Calorimetry 76, 1025-1033.
Gaterell, M.R., Gay, R., Wilson, R., Gochin, R.J. and Lester, J.N. (2000) An economic and environmental evaluation of the opportunities for substituting phosphorus recovered from wastewater treatment works in existing UK fertilizer markets. Environmental Technology 21, 1067-1084.
Giesen, A. (1999) Crystallization process enables environmental friendly phosphate removal at low costs. Environmental Technology 20, 769-775.
Golubev, S.V., Bauer, A., and Pokrovsky, O.S. (2006) Effect of pH and organic ligands on the kinetics of smectite dissolution at 25°C. Geochimica et Cosmochimica Acta 70, 4436-4451.
•22 Gunn, D.J. (1976) Mechanisms for the formation and growth of ionic precipitates from aqueous solution. Faraday Discussion Chemistry 61, 133-140.
Harris, C. D. (2000) Quantitative Chemical Analysis, 5th ed. W.H Freeman and Company, New York. USA.
Josan, M.S., Nair, V.D., Harris, W.G., and Herrera, D. (2005) Associated release of magnesium and phosphorus from active and abandoned dairy soils, Journal of Environmental Quality 34, 184-191.
Kraemer, S.M., Chiu, V. Q., and Hering, J.G. (1998) Influence of pH and competitive adsorption on the kinetics of ligand-promoted dissolution of aluminum oxide. Environmental Science and Technology 32: 2876-2882.
Letcher, T.M and Battino, R. (2001). An introduction to the understanding of solubility, Journal of Chemical Education, 78(1), 103-111.
Loewenthal, R.E., Kornmuller, U.R.C. and Heerden, E.P. (1994) Modelling struvite precipitation in anaerobic treatment systems, Water Science and Technology 30, 107-116.
Mamais, D., Pitt, P.A., Cheng, Y.W., Loiacono, J. and Jenkins, D. (1994) Determination of ferric chloride dose to control struvite precipitation in anaerobic sludge digesters, Water Environment Research 66,912-918.
Miles, A. and Ellis, T.G. (2001) Struvite precipitation potential for nutrient recovery from anaerobically treated wastes, Water Science and Technology 43(11), 259-266.
Morse, G.K., Brett, S., Guy, J., and Lester, J.N. (1998) Review: Phosphorus removal and recovery technologies. Science of the Total Environment 221(1), 69-81.
Mullin, J.W. (2001) Crystallization, 4th edn, Butterworth-Heinemann, Jordan Hill, Oxford, UK.
Munch, E. and Barr, K. (2001) Controlled struvite crystallization for removing phosphorus from anaerobic digester side streams. Water Research 35(17), 4191-4199.
Ohlinger, K. (1999) Struvite Controls in Anaerobic Digestion and Post-Digestion Wastewater Treatment Processes. Ph.D. Thesis, University of California Davis, U.S.A.
Parkhurst, D.L., Appelo, C.A.J. (1999) User's guide to PHREEQC (Version 2)- a computer program for speciation, reaction- path, advective-transport, and inverse geochemical calculations. USGS water-Resources Investigation Report 99-4259.
Regy, S., Mangin, D., Klein, J.P., and Lieto, J. (2002) Phophate Recovery by Struvite Precipitation in a Stirred Reactor, Lagep, Centre Europeen d' Etudes des polyphosphates.
23 Sarker, A.K. (1991) Hydration/ dehydration characteristics of struvite and dittmarite pertaining to magnesium ammonium phosphate cement systems. Journal of Material Science 26, 2514- 2518.
Scope Newsletter (2006) Phosphorus recycling, September, No. 65, pp 2.
Shu, L., Schneider, P., Jegatheesan, V., and Johnson, J. (2006) An economic evaluation of phosphorus recovery as struvite from digester supernatant. Bioresource Technology 97, 2211 - 2216
Snoeyink, V. and Jenkins, D. (1980) Water Chemistry. John Wiley & Sons, New York.
Sohnel, O. and Garside, J. (1992) Precipitation: Basic Principles and Industrial Applications. Butterworth Heinmann, Oxford, England, p.78-79.
Steen, I. (1998) Phosphorus availability in the 21st century management of a non-renewable resource. Phosphorus and Potassium 217, 25-31.
Stratful, I., Brett, S., Scrimshaw, M. and Lester, J. (1999) Biological phosphorus removal, its role in phosphorus recycling. Environmental Technology 20, 681 - 695.
Tai, C.Y, Chien, W.C., Chen, CY. (1999) Crystal growth kinetics of calcite in a dense fluidized- bed crystallizer. AIChE Journal 45(8), 1605-1614.
Tchobanoglous, G. and Schroeder, E.D. (1985) Water Quality. Addison Wesley Publishing Company, USA.
Ueno, Y. and Fujii, M. (2001) Three years experience of operating and selling recovered struvite from full-scale plant. Environmental Technology 22, 1373-1381.
Wang, J. Burken, J.G., Zhang, X., and Surampalli, R. (2005) Engineered Struvite Precipitation: Impacts of component-ion molar ratios. Journal of Environmental Engineering, 131(10), 1433- 1440.
Williams, S., (1999) Struvite precipitation in the sludge stream at Slough wastewater treatment plant and opportunities for phosphorus recovery. Environmental Technology 20, 743-747.
Woods, N.C., Sock, S.M, and Daigger, G.T. (1999) Phosphorus recovery technology modeling and feasibility evaluation for municipal wastewater treatment paints. Environmental Technology 20 (7), 663-679.
Wu, Q. and Bishop, P.L. (2004) Enhancing struvite crystallization from anaerobic supernatant, Journal of Environmental Engineering and Science 3,21 -29
24 Chapter 2 A solubility and thermodynamic study of struvite
2.1 Introduction
The accurate prediction of the solubility of struvite is important in many waste-water applications. Indeed, anaerobic sludge digestion results in the release of ammonium, magnesium, and phosphate; hence, it may cause struvite formation. Thermodynamics predicts that struvite will tend to form when the activity product of the constituting ions exceeds the thermodynamic
solubility product (Ksp), The objective of this study is to experimentally determine the solubility product of struvite and associated thermodynamic properties at temperatures between 10 and
60°C.
For a system that is at equilibrium with pure and solid-phase struvite, the thermodynamic
2+ + 3 solubility product is Ksp= {Mg } {NH4 } {PO4 "} and apparent or concentration solubility
+2 + 3 product is Kc - [Mg ] [NH4 ] [PO4 ] where {} indicates activity, [] indicates concentration. The
Ksp value of struvite has been examined by numerous researchers. A summary of Ksp values at 25
°C reported in the literature is shown in Table 2.1. The reported values range from 4.37x10"14 to
10 3.89xl0" (pKsp from 9.41 to 13.36). However, in engineering the most commonly used pKsp
value is 12.6 (Snoeyink and Jenkins, 1980). The first reported 12.6 value (Bube, 1980) was an
apparent pKsp, or concentration product, as ionic strength corrections were not made. In
experiments, concentrations not activities are typically measured in solution, whereas activities
are determined by thermodynamic modeling of activity effects. A pKsp value of 13.15
*A version of this chapter has been published:
Bhuiyan, M.I.H, Mavinic, D.S., and Beckie, R.D. (2007) A solubility and thermodynamic study of struvite. Environmental Technology 28, 1015-1026.
25 was also reported, where ion activity corrections were made using the Guntelburg approximation of the Debye-Hiickel equation (Taylor et al, 1963). Other researcher made activity corrections using the Davies approximation of the Debye-Hiickel equation and reported a value 13.12 (Burn and Finlayson, 1982).
The conventional techniques of solubility studies and determination of the Ksp value involve either formation of precipitate, or dissolution of previously formed salt in deionized
water, or any other solvent. A Ksp value has also been determined by considering equilibration by both formation and precipitation (Ohlinger et al, 1998). In that study, the solubility product was developed using measured and theoretical equilibrium total concentration of constituting ions and statistical methods. Theoretical values of the ions for each equilibrium experiment were
determined using MINEQL+ for any selected value of Ksp. The final Ksp value was determined by minimizing an objective function defined in terms of the measured and theoretical total concentration of the constituting ions and variance of their analytical results from the lab.
However, other techniques have also been employed. For instance, the Ksp values of struvite have been determined using the radioisotope 32P as tracer over a range of temperatures (Age et al,
1997). The solubility of struvite, formed in the pilot scale fluidized bed reactor, was also investigated in different water and wastewater solutions in a previous study (Rahman et al.,
2006).
2 3 + The equilibrium relationships of the ionic species H2P04", HP04 ", P04 ", MgH2P04 ,
0 2+ + + + MgHPO4 , MgP04", Mg , MgOH , NH4 , H , OH", and NH3 (aq) are considered, in most cases, for solubility studies of struvite (Buchanan et al, 1994; Ohlinger et al, 1998). However, relationships involving reaction(s) between magnesium and ammonium (Abbona et al, 1982;
Bouropoulos et al, 2004; Michalowski and Pietrzyk, 2006), the reaction between magnesium
26 and sulphate (Abbona et al, 1982), and the reaction between magnesium and carbonate
(Michalowski and Pietrzyk, 2006) were also considered in previous studies. Magnesium
2+ 3 phosphate complex formation reduces the concentration of Mg and P04 " ions available for
struvite formation and increases the solubility. A pKsp value of 13.27 was reported, where three
complexes of magnesium phosphate (MgHPCU, MgHiP04+ and MgPCV) were considered
(Ohlinger et al, 1998). When only one magnesium complex (MgHPCM) was considered, the
value was reported to be 13.15 (Taylor et al, 1963). There are many reasons that explain the
wide discrepancies in the reported solubility product values (Andrade and Schuiling, 2006):
• the solubility product may be derived using approximate solution equilibria
• the effect of ionic strength is often neglected
• mass balance and electro neutrality equations are not always used
• different chemical species are selected for calculations
• variation in the presence of both organic and inorganic complexes, as well as dissolved
species formed between the principal constituents of struvite.
• the solubility product value determined for a certain temperature is not extrapolated for
zero ionic strength.
The thermodynamic solubility product at the zero-ionic strength standard state can be
determined by extrapolation of non-zero experimental data to zero ionic strength. It is
reasonable to assume that the extrapolation to zero ionic strength is more accurate from low ionic
strengths (1=0.01 to 0.1M) than from high ionic strengths (Grenthe and Wanner, 2000). Standard
thermodynamic solubility products can be obtained from experimental data by: i) determining
the activity of species in the precipitation - dissolution reaction at infinite dilution or, ii)
extrapolating the non-zero ionic strength solubility product to the standard state of zero ionic
27 strength (Silbereman, 1996; Hefter and Tomkins, 2003). The principal objective of this study
was to determine thermodynamic Ksp values of struvite at different temperatures.
2.2 Materials and methods
2.2.1 Formation of struvite
Struvite was prepared by mixing equal volumes of equimolar quantities of magnesium chloride and diammonium phosphate. The solution was then made mildly alkaline (pH=7.4) by slow addition, with stirring, of filtered ammonium hydroxide (Johnson, 1959; Babic-Ivancic,
2002). After initial mixing of reactant solutions, the solution was sealed in a tightly closed container, leaving a minimum space above the solution. The solution was then maintained at 25
°C without agitation for 24 hours. The suspensions, after precipitation, were filtered though a
0.45 pm membrane filter, and the precipitates were washed thoroughly with distilled water and allowed to dry at room temperature overnight. Dried precipitate was examined and identified as struvite by x-ray diffraction using Bruker D8 Advance X-ray diffractometer and CuKa radiation, with an average scanning rate of 2.0° 29 min"1 (see Figure 2.1). Identification was also made by the room-temperature infrared (IR) spectra in the wave number range of 400 to 4000 cm"1, on a
Perkin Elmer 1710 Fourier Transformed Infrared Spectrophotometer. Analysis of the magnesium, ammonium and phosphate contents of weighed samples, after dissolution in 0.5% nitric acid solution, indicated an average purity of 99.7 percent.
2.2.2 Equilibration
For the solubility study, known amounts of powdered struvite, sufficient to provide an excess of the solid phase, was added to 125 ml glass jars full of deionized water and NaCl solutions of molarity 0.01, 0.02, 0.04 and 0.06. The jars were sealed and placed in an Innova
28 4230 refrigerated incubator shaker. The solubility determination requires intimate contact between the solid and the liquid phase. This can be achieved by thermostating a suspension of the solid in the solvent agitated by mechanical stirring, or by shaking or rotating the vessels.
When prolonged agitation is needed, the solid may become physically degraded or even colloidal. This not only renders phase separation more difficult, but also affects the results, since solubility depends on particle size (Hefter and Tomkins, 2003). The instrument was set at 150 rpm for shaking. The solubility was measured over a temperature range of 10°C to 60°C, with an interval of 5°C. All the mixtures were aged for 3 days to attain equilibrium (Taylor et al, 1963).
Constant pH readings of the mixtures would also indicate that equilibrium has been reached
(Hefter and Tomkins, 2003). After 3 days, the pH and conductivity of each jar were both measured quickly, to minimize temperature changes before sampling. pH was measured using an
Orion 420A bench top pH meter, equipped with a VWR Symphony temperature compensated probe which was calibrated using pH 7 and 10 buffers. Conductivity and temperature were measured using an Oakton pH/CON 300 Deluxe Waterproof pH/Conductivity meter. Next, the samples of the equilibrated solutions were filtered through 0.45 um size membrane filter paper.
The filtrates were washed three times with deionized water to remove extraneous ions. The filtered samples were then analyzed for magnesium, ammonia, and ortho-phosphates. Analyses for ortho-phosphate and ammonia were made, using the flow injection method on a LaChat
QuickChem 8000 instrument, as described in the method number 4500-P G and 4500-NH3 H of the Standard Methods for the Examination of Water and Wastewater (APHA et al, 1998).
Magnesium analysis was performed by flame atomic absorption spectrophotometry, using a
Varian Inc. SpectrAA220 Fast Sequential Atomic Absorption Spectrophotometer.
29 2.2.3 Thermodynamic solubility product, Ksp
The thermodynamic solubility product was determined by extrapolating experimental
data to zero ionic strength. In order to extrapolate, the following method was adopted.
2+ 3 +2 + 3 Ksp= {Mg } {NHV} {P04 -} = ^2+ [Mg ] ym: [NH4 ] y^. [PO, '],
+2 + 3 and Kc = [Mg ] [NH4 ] [P04 ].
Therefore, f\ y. = Kspl Kc (2-1) /=i
where, n is the number of ions in the solubility product expression and where
Ksp= Solubility product in terms of activity
Kc= Solubility product in terms of concentration
/(= activity co-efficient of the ion i
At low ionic strength, activity coefficients approach unity. One way to determine a
thermodynamic equilibrium constant is to measure the equilibrium constant at successively
lower ionic strengths and then extrapolate to zero ionic strength. Very commonly, the tabulated
equilibrium constants are not true thermodynamic constants, but just the concentration or
activity product measured under a particular set of conditions (Clark and Judith, 1998).
The Debye-Hiickel theory of interaction of ions in a solution incorporates both the
electrostatic interaction between ions and the thermal motion of the ions. The basic equation,
called the Debye- Hiickel limiting law, was developed for an ionic strength of less than
approximately IO"2 3 and can be stated as
\ogyr -0.5z,2/2 (2-2)
where z; is the valence of ion i and I is the ionic strength. Considering all the alternative models
(different approximations) of the Debye- Hiickel limiting law, to find the activity coefficients
30 from ionic strength, and their corresponding applicability, the Guntelberg approximation (I < 0.1
M) was chosen.
According to the Guntelberg approximation, logy- —Az. •= (2-3) 1 + V7
Combining equations 2-1 and 2-3,
\oZKc=-\o%Ksp-A-^ (2-4) 1 + -\l I
" 2 V7 where, A'(=A^jzj ) is a positive constant. A plot of -\ogKc against -== is linear. The i=l l + yll
intersection of this line with the ordinate axis (1=0) gives the value of -log Ksp, from which Ksp is calculated.
2.2.4 Speciation and ionic strength calculation
Essentially, to determine struvite solubility requires an appreciation of ionic strength with the associated effect upon activity and the relevant species that form in solution. Since all ions present in the solution will contribute to the ionic strength of a solution, it becomes difficult to ascertain the cut-off point for specific ions and what concentrations to include in making ionic strength and activity calculations.
A number of chemical equilibrium-based models have been developed and used by previous researchers. Computer models such as STRUVITE, MINTEQ+, MINTEQA2 and
MINEQL+ are based on equilibrium equations; however, a kinetic-based model has also been described (Musvoto et al, 2000). A chemical-equilibrium based modelling package, PHREEQC, has also been used by several researchers (Bouropoulos et al., 2004; Ali and Schneider, 1991).
PHREEQC can access thermodynamic data from large, well-established databases or rely on user-provided data, depending on user specifications. It can solve a wide range of problems,
31 having been based upon a robust numerical engine that rarely crashes. PHREEQC version 2.12, with the Lawrence Livermore National Laboratory Database (llnl.dat), was used in this study for
speciation and other thermodynamic equilibria. Unless otherwise specified in the database file or the input data set, the Davies approximation of the Debye-Huckel equation is used for charged
species. For uncharged species, the extended Debye-Hiickel equation (ln^ = bjT), where the first term of the equation becomes zero, is used. Unless otherwise specified, the coefficient b; is
assumed to be 0.1 for all uncharged species. Log K values of the equilibrium equations at
different temperatures are calculated either by the Van't Hoff equation (equation 2-5) using the
enthalpy value of the reaction or by an analytical expression (equation 2-6).
d\nKsp AH„° (2-5) dT RT 2
where, R is the gas constant (0.008314 kJmoF'deg"1); and
log10 Ksp = A, + A2T + -f + A, log10 T + -f (2-6) T2
where T is in Kelvin and Ai, A2, A3, A4, and A5 are coefficients. If defined, the analytical
expression takes precedence over the Van't Hoff equation to determine the temperature
dependence of the equilibrium constant. Major equilibria involved in the computation of solution
species are shown in Table 2.2. The equilibrium equations and the respective thermodynamic
constants at 25°C, collected from other sources, were included in the input file, if they were not
available in the llnl database. Collected enthalpy values of different equilibrium reactions from
different sources were also verified by equation 2-7. The values, which have good agreement
with this equation, have been finally selected.
A/ 0 0f 0 2 ^; = ZZ^ foducl -Ml /react™,s ( "7)
32 Enthalpy values for the reactions involving magnesium and ammonia were not available.
The selected model demonstrated that a 5-fold increase or decrease in the solubility product of either magnesium or ammonia or all of three species caused only a 0.2 to 0.4 percent change in the calculated solubility product of struvite. To establish the activities of species in solution as required for the solubility product determination, the published solubility product value for 25°C was used in calculations for all temperatures. In other words, the value of AH,0 was considered to be zero for these three equations. A similar approach was also used in previous research (Burn and Finlayson, 1982).
2.3 Results and discussion
2.3.1 Thermodynamic solubility product
Figure 2.2 shows an example (at 25°C) of how the thermodynamic solubility product of struvite was determined by extrapolation. The error bars represent the 95% confidence intervals.
The trend line has been drawn with the mean values. The intersection of the line with the
ordinate axis (1=0) gives the value of -log Ksp (equation 2-4) from which Ksp is calculated. The
experimentally determined pKsp (=-log Ksp) value at 25°C from this work is 13.36 (±0.07). The
pKsp value at 25°C (standard state) was also determined, using available formation enthalpies values of the reactant struvite of -731.24 kcal mof1 and products magnesium, phosphate, ammonium and water of -108.7, -18.97, -245.18, and -56.69 kcal mol"1 respectively (Faure,
1991) in equation 2-8.
0 AG,. =-RT \nKsp (2-8)
AG 0 AG The K where, AG,. =X °//>™ta ~ °freac^us • P sP value of 13.37, thus found, is in good agreement with the experimentally determined value in this study. The value found from this
33 work compares favorably with the values reported in previous works (Taylor et ai, 1963; Burn
and Finlaysons, 1982; Ohlinger et ai, 1998; Babic-Ivancic et al, 2002) (see Table 2.1).
2.3.2 Solubility product of struvite at various temperatures
The thermodynamic solubility product values of struvite determined with the method
presented here, in a temperature range from 10 to 60°C, with 5°C intervals, are shown in Table
2.3. The values, determined in this study at different temperatures using solutions of different
ionic strengths and then extrapolated for zero ionic strength, are lower than published Ksp values
(see Fig. 2.3). Solubility studies of struvite at different temperatures have been conducted by
previous researchers (Burn and Finlaysons, 1982; Age et ai, 1997; Babic-Ivancic et al, 2002);
however, none of the previous studies was conducted over an evenly distributed range of
temperatures. Figure 2.3 shows a comparison among the values determined in this study and
those of the previous studies. The consideration of ionic strength effect, by extrapolating to zero
ionic strength, and the use of most of the equilibrium equations are probable reasons for
differences of results with the previous studies. The Ksp value in this study increases with
temperature, until around 30°C, and then it decreases. Extrapolation to zero ionic strength was
done, as it is not practically possible to use any solution of zero ionic strength or infinite dilution.
The thermodynamic solubility products (1=0), thus determined and if used as the reference, can
accurately predict the precipitation potential of any solution. The supersaturation, calculated with
respect to these thermodynamic solubility product values (1=0), for solutions at different ionic
strengths, for a particular temperature, should then be comparable.
34 2.3.3 Effect of pH on struvite solubility
For any precipitation to take place, the ion activity product (IAP) of the solid phase must
be greater than the thermodynamic solubility product Ksp. When the solution is in equilibrium
2+ + 3 with the solid phase, the struvite Ksp is equal to the IAP of Mg , NH4 , and P04 ". The availability of these three species is controlled by system pH, the total dissolved concentration of magnesium, ammonium and phosphate species, the ionic strength, and temperature of the aqueous system. However, a fundamental requirement for an equilibrium characterization is the definition of the actual species present in the solution. In the case of struvite, the difficulty comes from the fact that all three of the reacting ions exhibit complex equilibria in aqueous solution.
Slight variation in the pH results in a change in the speciation of the struvite constituents, leading to more or less favourable conditions of struvite precipitation. The proportion of ammonium ion
+ (NH4 ) present in solution depends on its equilibrium with ammonia (NH3) and varies markedly with pH. Due to the triprotic nature of orthophosphoric acid, several orthophosphate species exist
3 in aqueous solution, resulting in a variable proportion of P04 " with pH of the solution.
Hydrolysis of the magnesium ion leading to the formation of MgOH+, is significant only at pH higher than 9.5. In the presence of phosphate and ammonia, magnesium forms a number of complexes depending on pH and concentration of the species in solution. However, as speciation of the components is pH dependent, the solubility of struvite also varies with pH. The pH of the anaerobic sludge digestion and post digestion processes is generally higher than the pH of preceding waste treatment processes. Moreover, concentrations of dissolved component ions increase in the anaerobic digestion process. Ammonia concentration in the bulk fluid increases as protein is degraded, and dissolved magnesium and phosphate concentrations increase due to cell lysis. Thus, struvite precipitation potential is greater in the anaerobic digestion and post digestion
35 process, since only the dissolved fraction of the components is free to combine to form precipitates. However, the conditional solubility product of struvite illustrates a situation where more than one of the dissolving species is affected by solution pH. A conditional solubility product can be defined in terms of total analytic concentrations as follows (Snoeyink and
Jenkins, 1980):
P = C C C = - (2-9)
a A a 2 Mg^ NH; Pot ? Mg * y NH; y PO?-
2+ + 3 a +2 The ionization fractions for Mg , NH4 , and P04 " can be defined as Mgi> ~ [Mg ] /Cx„Mg ,
3 CC + = [NH4 ] /CT, NH3 and apQ,_ = [P04" ]/CT, PO4 , where CT, ,Mg, CT, NH3, and CT, PO4 are the total analytical concentrations of magnesium , ammonia, and orthophosphates, respectively.
The conditional solubility product definition is selected for practical purposes. Plotting
Ps, calculated from the right side of equation 2-9 versus pH, establishes the struvite solubility
limit curve for a specific ionic strength, temperature and solution composition. Unlike a
thermodynamic approach which requires activities of all species, a solubility limit curve, drawn
for a certain ionic strength, can be used to determine the struvite saturation condition of water
from measured magnesium, ammonia, and orthophosphate concentrations. Extensive research
has been carried out to model struvite precipitation and to construct a solubility limit curve. The
computer program MINTEQA2 has been used in previous studies to draw the solubility limit
curve. It has the option to specify a fixed value for ionic strength, required for modeling an open
system, where struvite precipitation does not significantly influence the solution ionic strength
(Ohlinger et al, 1998). Struvite solubility limit curves, although intended to be used as examples
in textbooks (Snoeyink and Jenkins, 1980; Stumm and Morgan, 1981), have also been
extensively used to evaluate struvite precipitation potential in practice (Booram et al, 1975; Web
36 and Ho, 1992). The solubility limit curve has also been used by several researchers (Adnan et al,
2003; Britton et al, 2005). In such cases, a variation of ionic strength of the solution was avoided. However, the capability of the model to accurately predict struvite precipitation potential is dependent upon accurate determination of ionization fractions, activity coefficients, and the solubility product.
Generally, struvite solubility decreases with increasing pH within the operating pH range
(7-8.5). Because an increase in pH will decrease the ammonium ion concentration and increase the phosphate ion concentration, it follows that there should be a pH value where the solubility
of struvite is a minimum. Within the operating ionic strength range, struvite solubility increases
as ionic strength increases due to the resultant decrease in the effective concentration of the
component ions of struvite (Snoeyink an Jenkins, 1980). Including all three magnesium-
phosphate complexes and using a pKsp value of 13.27 (Ohlinger et al, 1998), the struvite
solubility was found to increase and the pH of minimum solubility was found to shift from 10.7
(Snoeyink an Jenkins, 1980) to 10.3 (Ohlinger et al, 1998). It's worth mentioning that the
experimentally derived thermodynamic solubility product value of pKsp=\3.?>6, found in this
study, includes through the PHREEQC calculations, most of the possible equilibrium reactions.
2.3.4 Solubility product (Ksp) value and solubility of struvite
Due to ion pair formation, hydrolysis, complex ion formation, activity coefficient
variations, common ion effect, temperature effect, and ionic strength effect, there are only a few
cases in which solubility and Ksp are related in a simple way (Clark and Judith, 1998). The
l/3 calculation of solubility from Ksp value (S = Ksp ), takes no account of pH, ion activity and
ionic strength and the use of this equation should only be applied with very soluble salts (Doyle
and Simon, 2002). In this solubility study, dissolution of the constituent ions from struvite was
37 found to take place in equimolar concentrations, with only minor variation. However, the molar solubility of struvite has been calculated at different temperatures from the geometric mean of the total molar concentrations of magnesium, ammonium and phosphate of the equilibrium solution (Sohnel and Garside, 1992). Figure 2.4 shows how solubility (mg l"1) of struvite changes with ionic strengths, represented by the different normality of NaCl, of the solutions at different temperatures. In deionized water, struvite solubility at 25°C is 169.2 (± 4.3) mg 1"'. An average value for the solubility of struvite, calculated from the published data, and reported by the previous researchers, is 160 mg l"1 at 25°C (Andrade and Schuiling, 2001; Salutsky et al, 1972).
Thus, the value found in this work compares favourably. Previous studies, conducted to investigate the effect of ionic strength on the solution chemistry of struvite at 25°C have suggested that ammonium salts exert a greater influence in this respect than the corresponding salts of Na and K, in equivalent concentrations (Uncle and Smith, 1946). Indeed, ammonium salts not only increase the ionic strength, but the activity of the ammonium ion in solution. The ionic strength of the medium is particularly important in the case of struvite crystallization from wastewater. This important solution parameter can be estimated indirectly by the measurement of the electrical conductivity of the solution. In a previous study, authors compared the solubility product of struvite under in situ waste digester conditions, with its solubility in pure water, and concluded that struvite was far more soluble under prevailing digester conditions (Bridger et al,
1962).
1 Figure 2.5 shows the variation of solubility of struvite in mg l" and Ksp values with
temperature. The maximum Ksp value was found at 30°C , while the solubility of struvite in deionized water reaches its maximum value of 212.7 (± 3.8) mg l"1 at 35°C. Nonideal behaviour of the solutions creates such discrepancy in the relationship between solubility and solubility
38 product. In the majority of cases, especially for sparingly soluble salts, solubilities are not simply
related to Ksp (Clark and Judith, 1998).
2.3.5 Temperature effect on struvite solubility
The effect that temperature has on the solubility of a substance is determined by the
quantity of heat released or absorbed, as the solute dissolves. Whether heat is liberated or
absorbed depends on the energy required to disrupt the crystal structure of the solid and on the
energy liberated when the solid particles interact with the solvent. If more energy is required to
break particles away from the solid than is gained by virtue of the interaction of these particles
with solvent, heat will be absorbed (endothermic), and vice versa.
The dissolution of most solids is accompanied by absorption of heat (endothermic). This
means that if the temperature of a saturated solution is raised, the process whereby solute
particles enter the solution is favored over the precipitation of solid from solution. Thus, the
solubility will increase with temperature. If the dissolution of the solid releases heat
(exothermic), the precipitation process is favored over the dissolution process and the solubility
decreases as the temperature increases (Nebergall et ai, 1980).
The effect of temperature on struvite solubility has been investigated by others for
different temperatures. A study on the anaerobic digestion system of the wastewater treatment
plant concluded that, as the temperature increased from 0 to 20°C, struvite solubility also
increased to a maximum (Borgerding, 1972). Another study suggested 30°C to be the
temperature of maximum struvite solubility, from experiments designed to define
thermodynamic parameters for its crystallization (Webb and Ho, 1992). Struvite was also found
to be more soluble at 38°C than at 25°C, as concluded from the solubility products of the
experimental solutions (Burn and Finlayson, 1982). A different study, employing radioisotope
39 32P as tracer, observed a steady increase in solubility with increasing temperature; followed by a marked decline in solubility over the range 10-65°C, with the maximum solubility at 50°C (Age et al, 1997). The solubility of struvite in mg 1"' was found to be maximum at 35°C, while the solubility products determined by extrapolating to zero ionic strength reach a maximum value at
30°C in this study. The sudden transition at around 55°C of the solubility curve of struvite in pure water shown in Figure 2.5 implies a possible phase change of struvite to magnesium ammonium
monohydrate (MgNFUPCV H20) (Bridger, 1962). The phase change of struvite is accompanied by a change in structure, which is likely to affect the solubility of the compound. The presence of higher water of crystallization (44%) in the hexahydrate makes it more soluble than the monohydrate (11.6%)(Bridger et al, 1962).
2.3.6 Enthalpy
While variations in pressure have little effect on the values of the solubility products, temperature variations are important. Variation of the solubility products with temperature are usually calculated with the van't Hoff equation (equation 2-5)
If we integrate equation (2-5) from the standard temperature (7^) to some other temperature (T), the result of integration is,
^ ^A^ 1_J_
Kvr> R T T«
At standard temperature (25°C), the value of the reaction enthalpy, AH?, is calculated from the
formation enthalpies, AHf, using equation 2-7. AHr° varies only slightly with temperature and, within a range of few tens of degrees, it can be considered as constant. In that case, equation 2-
10 can be re-written as equation 2-11, to determine solubility product at any temperature from a known solubility product and temperature.
40 KSp\ ln( ) (2-11) Kspl
where, Ksp/ and Ksp2 are the solubility products at temperature T] and T2 ( Kelvin);
However, for the temperature range of 10 to 30 °C, using equation (2-11) and the
1 measured Ksp values, the average AHr° value was found in this study to be 23.62 kcal mol" , with
a standard deviation of 0.42. The value of AHr°, derived from the slope of the curve drawn with
1 log Ksp vs. 1/r, is 24.11 kcal mol" . Thus, the value AHr°, agrees favourably with the value (20 kcal mol"1) derived from text book values (Faure, 1991) of formation enthalpies, using equation
2-5.
Based on the experimental Ksp values, the dissolution of struvite is accompanied by
absorption of heat (endothermic) up to 30°C. Thus, the solubility increases with temperature.
After that, the dissolution of struvite releases heat (exothermic), and the precipitation process is
favored over the dissolution process. This results in a decrease in solubility, as temperature
increases. As a result, the value of enthalpy of reaction changes both in magnitude and sign, after
the maximum solubility is attained. As such, considering a single value of enthalpy of reaction
for the entire temperature range (10-60°C) possible in this work, can lead to erroneous results.
An alternative way of expressing the variation of Ksp values, with temperature, is to fit
the experimental data for different temperatures with a polynomial function (Appelo and Postma,
1999). The analytical expression in equation 2-6 is recommended in PHREEQC, where the
coefficients of the expression are mentioned either in the database or can be included in the input
file. Coefficients of the analytical expression have been calculated by minimizing the sum of
squared differences between calculated and the experimental values found in this work. The
analytical expression can be written using the calculated coefficients as:
41 T 12 logl0Ksp =-1157.45-0.784r —+ 556.831og10 + ^J~ O )
Using this analytical expression, along with other required information about struvite in the input file, struvite precipitation potential of any solution can be measured from the calculated
saturation index, SI (=log IAP - log Ksp), as demonstrated elsewhere (Bhuiyan et al., 2007,
Chapter 5).
2.4 Conclusions
The solubility and thermodynamics of struvite were studied at different temperatures and under strictly controlled conditions. Based on the experimental data, the following conclusions can be made:
• The true thermodynamic solubility products of struvite were determined by extrapolating
measured Ksp values to zero ionic strength, using an appropriate activity coefficient
model. ThepKsp value of struvite, thus found, at 25°C, is 13.36(±0.07). This value agrees
favorably with the value (13.37) calculated from the standard free energies of formation
available in the literature. The pKsp values for a temperatures between 10-60 °C range
from 14.36(±0.05) to 14.01 (±0.03), with the minimum value (Ksp value maximum) of
13.17(±0.05)at30°C
• The solubility and solubility product value of struvite are not directly related, as it is
found in case of highly soluble salts. The solubility of struvite determined in deionized
water was found to be 169.2 (±4.3) mg l"1 at 25°C, with a maximum value of 212.7 (±3.8)
mgl"1 at35°C
• Standard enthalpy of reaction, AHr°, calculated from the average Ksp values, was found to
be more or less constant within the temperature range of 10-30 °C with an average value
42 of 23.62 Kcal mol"1 and standard deviation 0.42. Since the thermodynamics of struvite
changes after the temperature of maximum solubility, use of the same value of AHr° in
calculations to derive Ksp values for other temperatures, may lead to erroneous results. In such cases, the analytical expression, as proposed, can be used.
A thorough analysis of struvite thermodynamics, resulting from a consideration of more complete speciation using PHREEQC, has produced a new understanding of this subject, allowing us to predict the precipitation potential of struvite. Using the thermodynamic solubility products in PHREEQC, it is possible to readily determine the precipitation potential of struvite, at any temperature.
43 Table 2.1 Published kSD values for struvite at 25 C from literature.
pKsp Ksp Origin Reference
9.41 3.89xl(r10 Wastewater (Borgerding, 1972) 9.94 1.15x10"'° Aqueous solution (Abbona etal, 1982) 11.84 1.14xl0"'2 The same (Booram et al., 1975)
12.36 4.36xl0"13 Simulation (Buchanan etal, 1994)
12.6 2.51xl0~13 Water (Bube, 1910)
12.6 2.51xl0~13 Simulation (Loewenthal et al, 1994)
12.76 1.74X10"'3 Water (Webb and Ho, 1992)
12.93 1.17xl0-13 Aqueous solution (Age etal, 1997)
13.12 7.58xl0~14 Water (Burn and Finlayson, 1982)
13.15 7.08xl0_14 Aqueous solution (Taylor etal, 1963)
13.27 5.37xl0"14 Synthetic (Ohlinger et al, 1998) supernatant 13.36 4.37x10"14 Aqueous solution (Babic-Ivancic et al, 2002
44 Table 2.2 Major equilibria involved in the computation of the solution species at 25 C.
Equilibrium LogK Ref AHr° Ref (KJ mol"1)
+ 2 9.37 llnl database 3.744 llnl database 2H + HP04 "<=> H3PO4
+ 2 7.20 llnl database -4.205 llnl database H +HP04 » H2P04"
+ 3 2 12.32 llnl database -14.769 llnl database H +P04 "»HP04 "
2+ + + -11.44 Ball and Nordstrom, 1991 66.743 Ball and Nordstrom, 1991 Mg +H20<» MgOH + H
2+ + 0.45 Kofina and Koutsoukos, 2005 14.225 Ball and Nordstrom, 1991 Mg +H2P04-<=>MgH2P04
2+ 2 2.87 Ball and Nordstrom, 1991 13.807 Ball and Nordstrom, 1991 Mg +HP04 MgHP04°
2+ 3 4.80 Kofina and Koutsoukos, 2005 12.970 Ball and Nordstrom, 1991 Mg +P04 "»MgP04"
+ + 9.24 llnl database -51.920 llnl database NH3+H<=> NH4
+ H+OFT<=> H20 14.00 Ball and Nordstrom, 1991 -55.906 Ball and Nordstrom, 1991
2+ 2+ 0.24" Kofina and Koutsoukos, 2005 0 Burn and Finlayson, 1982 Mg + NH3 <=> MgNH3
2+ 2+ 0.20 Kofina and Koutsoukos, 2005 0 Burn and Finlayson, 1982 Mg +2NH3<=> (MgNH3)2
2+ 2+ -0.3" Kofina and Koutsoukos, 2005 0 Burn and Finlayson, 1982 Mg +3NH3<=> (MgNH3)3
2+ 2 Mg +S04 "<=> MgS04 2.37 Ball and Nordstrom, 1991 19.037 Ball and Nordstrom, 1991
2+ 2 Mg +C03 "<=> MgC03 2.98 Ball and Nordstrom, 1991 11.351 Ball and Nordstrom, 1991
2+ + Mg +HC03"<=> MgHC03 1.07 Ball and Nordstrom, 1991 3.305 Ball and Nordstrom, 1991
2+ + Mg +cr<=> MgCl -0.135 llnl database -0.586 llnl database
** 1=0.2 M in NH4N03
45 Table 2.3 Solubility products of struvite determined at various temperatures in this research (values in parenthesis are the 95% confidence intervals).
14 Temperature (°C) pKsp Ksp(xl0- ) 10 14.36 (±0.05) 0.436
15 14.04 ((+0.03) 0.916
20 13.69 (±0.02) 2.050
25 13.36 (±0.07) 4.330
30 13.17 (± 0.05) 6.840
35 13.23 (±0.03) 5.920
40 13.40 (±0.02) 4.000
45 13.60 (±0.06) 2.530
50 13.68 (± 0.08) 2.110
55 13.84 (±0.05) 1.460
60 14.01 (±0.03 ) 0.973
46 (a)
05 CL u,
2-Theta - Scale
(b)
%T
\ / /
Wave number (cm" )
Figure 2.1 Identification of crystalline solid as struvite using (a) powder X-ray diffraction. The peaks of the X-ray diffraction pattern of struvite sample matches perfectly with the struvite standard peaks (»)(b) IR spectrum where 3 vibrational mode peak for P-O bond of the P04 " groups are observed at 1006 and 570 cm I
47 13.4
R = 0.99 13.2 Y=-7.015 X+13.364
13.0
12.8
as 12.6
12.4 -I
12.2
12.0
11.8 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.5 0.5 I /(1 + I )
Figure 2.2 Example of extrapolation for thermodynamic solubility product for struvite
at 25°C (n=5). Error bars: 95% confidence interval.
48 40 • This research 35 • Age etal.(1997) 30 A Burns and Finlayson (1982) o Babic-lvanbic et al. (2002) • 10 5 0 0 10 20 30 40 50 60 70 Temperature (°C) Figure 2.3 Solubility product of struvite at different temperatures (This research and other studies). 49 350 300 • o • _ 250 • o 200 '•q 150 "o w 100 20degC • 25degC A 30 degC • 35 degC o 45 degC 50 00.0 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 Normality of NaCl Figure 2.4 Solubility (mg l"1) at different temperatures in the solutions of NaCl with different normality. 50 10 240 K x 10"" sp 220 Solubility (mg I"1) 200 i I I 180 ~ i 160 S 140 120 w 100 I 80 60 10 20 30 40 50 60 70 Temperature( c) Figure 2.5 Variation of Solubility (mg I") and Ksp of struvite with temperature (n 5). Error bars: 95% confidence interval. 51 References Abbona, F., Lundager, M.H. and Boistell, R. (1982) Crystallization of two magnesium phosphates, struvite and newberyite: Effects of pH and concentration. Journal of Crystal Growth 57,6-4. Adnan, A., Koch, F. A., and Mavinic, D.S. (2003) Pilot-scale study of phosphorus recovery through struvite crystallization- II: Applying in-reactor super saturation ratio as a process control parameter. Journal of Environmental Engineering and Science 2, 472-483. Age, H.K., Andersen, B.L., Blom, A. and Jensen, I. (1997) The solubility of struvite. Journal of Radioanalytical and Nuclear Chemistry 223, 213-215. Ali, M.I., and Schneider, P.A. (2005) Crystallization of struvite from metastable region with different types of seed crystals. Journal of Non-Equilibrium Thermodynamics 30, 95-111. American Public Health Association (APHA), American Water Works Association, and Water Pollution Control Federation. (1998) Standard Methods for the Examination of Water and Wastewater, 20th ed., Washington, DC. Andrade, A. and Schuiling, R.D. (2001) The chemistry of struvite crystallization. Mineralogy Journal ( Ukraine ) 23, 37-46. Appelo, C.A.J, and Postma, D. (1999) Geochemistry, Groundwater and Pollution, Rotterdam, The Netherlands. Ball, J.W. and Nordstrom, D.K. (1991) User's Manual for WATEQ4, with revised thermodynamic database and test cases for calculating speciation of major, trace and redox elements in natural waters. U.S.G.S Open File Report 91-183, Menlo Park, CA. Bavic-Ivancic, V., Jasminka, K., Damir, K. and Ljerka, B. (2002) Precipitation diagram of struvite and dissolution kinetics of different struvite morphologies. Croatica Chemica Acta 75, 89-106. Bhuiyan, M.I.H, Mavinic, D.S. (2007d) Assessing struvite precipitation in a pilot-scale fluidized bed crystallizer. Environmental Technology (Submitted). Booram, C, Smith, R. and Hazen, T. (1975) Crystalline phosphate precipitation from anaerobic animal waste treatment lagoons liquors. Transactions of the ASAE 18, 340 - 343. Borgerding, J. (1972) Phosphate deposits in digestion systems. Journal of the Wastewater Pollution Control Federation 44, 813-819. Bouropoulos, C, Vagenas, N., Klepetsanis, P., Stravropoulos, N., and Bouropoulos, N. (2004) Growth of calcium oxalate monohydrate on uric acid crystals at sustained supersaturation. Crystal Research Technology 39, 699-704. 52 Bridger, G.L., Murrel, L.S. and Starostka, R.W. (1962) Metal ammonium phosphates as fertilizers, Journal of Agriculture and Food Chemistry 10, 181-188. Britton, A. Koch, F., Mavinic, D., Adnan, A., Oldham, W., Udala, B. (2005) Small scale pilots at penticton sewage works: Pilot scale struvite recovery from anaerobic digester supernatant at an enhanced biological phosphorus wastewater treatment plant. Journal of Environmental Engineering and Science 4, 265-277'. Bube, K., (1910) Uber magnesiumammoniumphosphate, Zeitschrift fiir Analytische Chemie 49, 525-596. Buchanan J., Mote, C. and Robinson, R. (1994) Thermodynamics of struvite formation. Transactions of the ASAE, 37, 617-621. Burn, J.R. and Finlayson, B. (1982) Solubility product of magnesium ammonium phosphate Hexahydrate at various temperatures. Journal of Urology 128, 426-428. Clark, R. W. and Judith, M.B. (1998) The Ksp-solubility conundrum, Journal of Chemical Education 15, 1182-1185. Doyle, J.D. and Simon, A.P. (2002) Struvite formation, control and recovery. Water Research 36, 3925-3940. Faure, G. (1991) Principles and Applications of Inorganic Geochemistry. Macmillan Publishing Company, New York. Grenthe, I. and Wanner, H. (2000) Guideline for the extrapolation to zero ionic strength, Thermo dynamical database project, OECD Nuclear Agency, Paris, France. Harris, C. D. (2000) Quantitative Chemical Analysis, 5th ed. W.H Freeman and Company, New York. USA. Hefter, G.T, and Tomkins, R.P.T. (2003) The Experimental Determination of Solubilities. John Wiley & Sons, Ltd., New York. Johnson, R.G. (1959) The solubility of magnesium ammonium phosphate hexa hydrate at 38°C with considerations pertaining to the urine and the formation of urinary calculi. The Journal of Urology 81, 681-690. Kofina, A.N. and Koutsoukos, P.G. (2005) Spontaneous precipitation of struvite from synthetic wastewater solutions. Crystal Growth & Design 5, 489-496 Lawrence Livermore National Laboratory database (llnl.dat) derived from thermo.com.V8.r6t.dat 53 Loewenthal, R.E., Kornmuller, U.R.C. and Heerden, E.P. (1994) Modelling struvite precipitation in anaerobic treatment systems, Water Science and Technology 30, 107-116. Michalowski, T. and Pietrzyk, A. (2006) A thermodynamic study of struvite+water system. Talanta 68, 594-601. Musvoto, E.V. Wentzel, M.C. and Ekama, G.A. (2000) Integrated chemical-physical process modeling II. Development of kinetic based model for weak acid/base systems. Water Research 34, 1868-1880. Nebergall, W.H., Holtzclaw, H.F. and Robinson, W.R. (1980) General Chemistry. D.C. Heath and Company, Lexington, Massachusetts, pp. 309-310. Ohlinger, K. N, Young, T.M., and Schroeder, E.D. (1998) Predicting struvite formation in digestion. Water Research 32, 3607-3614. Rahaman, M.S., Mavinic, D.S., Bhuiyan, M.I.H, and Koch, F.A. (2006) Exploring the determination of struvite solubility product from analytical results. Environmental Technology 27, 951-961. Salutsky, M.L., Dunseth, M.G., Ries, K.M. and Shaphiro, J.J. (1972) Ultimate disposal of phosphate from wastewater by recovery as fertilizer. Effluent and Water Treatment Journal October, 509-519. Silbereman, R.S. (1996) Solubility and Thermodynamics: An introductory Experiment, Journal of Chemical Education 73, 426. Snoeyink, V. and Jenkins, D. (1980) Water Chemistry. John Wiley & Sons, New York, pp.306-309, 449. Sohnel, O. and Garside, J. (1992) Precipitation- Basic Principles and Industrial Applications. Butterwort-Heinemann Ltd., Oxford, UK. Stumm, W. and Morgan, J. (1981) Aquatic Chemistry. Wiley-Interscience, New York, pp. 277-278. Taylor, A., Frazier, W. and Gurney, E. (1963) Solubility products of magnesium ammonium and magnesium potassium phosphates. Transactions of the Faraday Society 59, 1580 - 1584. Uncles, R. F. and Smith, B.L. (1946) Solubility of magnesium ammonium phosphate hexahydrate. Industrial and Engineering Chemistry 30, 699-702. Webb, K. and Ho, G. (1992) Struvite solubility and its application to a piggery effluent problem. Water Science and Technology 26, 2229 - 2232. 54 Chapter 3 Determination of temperature dependence of electrical conductivity and its relationship with ionic strength of anaerobic digester supernatant, for struvite formation* 3.1 Introduction The electrical conductivity (EC) of a solution, at a constant temperature, depends upon concentration of ions present in it and their mobility. The more ions that are in solution, the greater is the conductivity of the solution. Therefore, EC measurements can be used to assess total ion concentration in a solution, as well as the solution ionic strength (I), an important thermodynamic property. The mobility of ions, and therefore the conductivity of the solution, also changes as a function of temperature. In this chapter, the temperature dependence of EC and its relationship to ionic strength for waste water solutions that are typical of conditions under which struvite precipitation occurs was examined. These results will enable more reliable application of EC measurements to predict struvite saturation in wastewater solutions. Intentional precipitation of struvite from sludge digester supernatant/centrate in fluidized bed reactors has been described elsewhere (Bhuiyan et al, 2007, chapter 5). Previous studies have identified the mechanisms by which the ions required for struvite formation (Mg2+, NliV, and PO4 "), become supersaturated in digester effluents (Jardin and Popel, 1994; Banister et al, 1998). The supernatant from an anaerobic digester may contain numerous ions, in addition to those that constitute struvite. EC variation, alone, cannot explain the change of any ions, depending on solution pH and temperature. However, struvite solubility calculations *A version of this chapter has been submitted for publication: Bhuiyan, M.I.H, Mavinic, D.S., and Beckie, R.D. (2007) Determination of temperature dependence of electrical conductivity and its relationship with ionic strength of anaerobic digester supernatant, for struvite formation. Journal of Environmental Engineering- ASCE, in review. 55 require a knowledge of ionic strength, with the associated effect upon activity and the relevant species that form in solution. Ionic strength is a measure of the total concentration of ions in solution and is important for the determination of the activities of ions in solution (Snoyeink and Jenkins, 1980). Indeed, chemical reactions and equilibria are functions of ionic activities rather than ionic or analytical concentrations. The ratio of an ion's activity to its concentration in a solution is called the activity coefficient, y. The activity coefficient of ions in sufficiently dilute solutions (I <0.5) can be accurately predicted using ionic strength and expressions based on the DeBye-Huckel limiting law (Stumm and Morgan, 1981). The DeBye-Huckel theory of interaction of ions in aqueous solution incorporates both the electrostatic interactions between ions and the thermal motion of the ions. The two most widely used expressions, based on two different approximations, are presented in Table 3.1. The relationship between EC and I is complex, depending on the chemical composition and ionic strength. However, there are many instances where the relative composition of the water is reasonably constant for a particular use and, hence, the electrical conductivity-ionic strength (EC-I) relationship can be established with a reasonable degree of certainity over a wide concentration range (Ponnamperuma et al, 1966; Griffin and Jurinak, 1973; Russel, 1976). Due to the difficulty in the determination of ionic strength for complex wastewater solutions, an approximation of ionic strength from an EC-I relationship would be faster and inexpensive, especially in monitoring the performance of a system. In addition to the amount and composition of ionic species, EC is strongly dependent on temperature. Since EC measurements cannot always be at a standard temperature, the EC of 56 water samples measured at various temperatures must be corrected to values corresponding to a standard temperature, for meaningful data interpretation and comparison. The electrical conductivity-temperature (EC-T) relation of natural waters is generally nonlinear (Millero, 2001). However, the degree of nonlinearity is relatively small in practical temperature range (0- 45°C), and a linear equation (equation 3-1)) is commonly used to represent the relation (Sorensen and Glass, 1987; Hayashi, 2004): ECT = EC 25[\ + a(T-25)] (3-1) where, ECj is electrical conductivity at temperatures T (°C), EC25 is electrical conductivity at a standard temperature of 25°C, and a ( °C~1) is a temperature compensation factor. Several values of a, ranging from 0.0191 (for 0.01 M KC1 solution) to 0.025 (commonly used by geophysicists), are commonly cited in the literature (Hyashi, 2004). The use of an arbitrary chosen value of a may be justified, considering a large degree of uncertainty in the EC-I relationship. However, it is still important to minimize unnecessary errors resulting from inaccurate temperature correction, especially when correction has to be made over a large temperature range. None of the relationships between ionic strength and conductivity found in the literature have been derived for wastewater or digester supernatant/centrate (Ponnamperuma et al, 1966; Griffin and Jurinak, 1973; Russell, 1976). A relationship between the conductivity data obtained during the struvite solubility tests in different kinds of solutions (including supernatant/centrate) and the ionic strength (manually calculated using major equilibrium equations) developed by this research group at UBC, has been proposed and reported elsewhere (Rahaman et al, 2006). This study aims to determine a representative temperature compensation factor for conductivity correction in a system associated with struvite formation from anaerobic digester 57 supernatant/centrate; in addition, an attempt has been made to develop a relationship between electrical conductivity and ionic strength, which includes a correction for ion-pairing, with the use of a chemical equilibrium model. 3.2 Materials and methods 3.2.1 Temperature dependence of EC Digester supernatant/centrate samples were collected from the Annacis Island Wastewater Treatment plant (secondary), Lion's Gate Wastewater Treatment Plant (primary), City of Penticton Wastewater Treatment Plant (BNR), and Lulu Island Wastewater Treatment Plant (secondary) in British Columbia, Canada, and the Edmonton Goldbar Wastewater Treatment Plant (secondary) in Alberta, Canada. The samples were transported in a cooler and stored at 4°C in the laboratory, until the analyses were made. Before the day of analysis, the samples were transferred to 250 ml high density polyethylene bottles, and stored at 4°C overnight. In the morning, a sample bottle was removed and placed on the laboratory bench at room temperature (21-23°C), and the measurement of temperature and EC started immediately. EC and temperature were recorded at approximately every 2°C, while the samples slowly warmed to room temperature, using a Radiometer conductivity meter (CDM3) equipped with a Radiometer CDC 304 conductivity probe and a mercury thermometer. The samples were well mixed in the bottles when measurements were taken. The conductivity meter was calibrated with 0.01 M KC1 and the probe was cleaned after each immersion in the supernatant/centrate sample. After the samples reached room temperature, they were immediately placed in a warm water bath, where EC and temperature were recorded at every 2°C until 40°C 58 3.2.2 EC-I relationship Samples collected from the same treatment plants were used to develop an EC-I relationship. To minimize the effects of ion-ion interactions and to appropriately account for the contribution of each separate ion to the measured EC, it is recommended that the solutions be diluted (APHA et al, 1998). With that view in mind, and to obtain a range of ionic strength and conductivity, samples were diluted with different dilution factors. After the required dilution with deionized water, conductivity and temperature were both recorded, using the same conductivity meter and thermometer as mentioned in the above section. pH was measured using an Orion 420A bench top pH meter, equipped with a VWR Symphony temperature compensated probe, which was calibrated using pH 7 and 10 buffers. 3.2.3 Analyses Analyses for ortho-phosphate ammonia, nitrate and chloride were made, using the flow injection method on a LaChat QuickChem 8000 instrument, as described in the method number 4500-P G, 4500-NH3 H, 4500-NO3" I, and 4500-C1" of the Standard Methods for the Examination of Water and Wastewater respectively (APHA et al, 1998). Calcium, magnesium, potassium, sodium, aluminum, and iron analyses determined by flame atomic absorption spectrophotometry, using a Varian Inc. SpectrAA220 Fast Sequential Atomic Absorption Spectrophotometer. Sulphate was quantified by the turbidimetric method. TC and IC of the filtered (0.45 pm) samples were analyzed by Shimadzu Total Carbon Analyzer TOC-500. Filtered TOC or dissolved organic carbon (DOC) was calculated from their difference. Knowing the pH of the 2 solutions, concentrations of HCO3" and CO3 " were calculated from IC (IC = H2C03* + HC03" + CO3 " ) using equilibrium equations 3-2 and 3-3, as described in the method number 4500 C02. D of the Standard Methods for the Examination of Water and Wastewater (APHA et al., 1998): 59 + 636 H2C03* <=> HC03" + H ; K,= 10" @25°C (3-2) + 10 33 2 HCO3" «• CO3 " + H ; K2= IO" ' @25°C (3-3) where, [H2C03* ] <=> [H2C03] + [C02 (aq) ] and the activity coefficients are assumed equal to unity. Volatile fatty acids (acetic, propionic and butyric) were analyzed by injecting 1 pi of sample into a Hewlett Packard 5890 series II Gas Chromatograph, equipped with a FID detector and a 25 m, 0.25 mm dia HPFFAP column. Operating temperature was 130°C for 2 minutes and increased 8°C per min. to 160°C. Since the ionic strength is a measure of the intensity of the electrical field in an electrolyte solution, correction of the analytical concentrations used to compute ionic strength for neutral ion pair species, and ion-pairs of the reduced charge, are necessary to provide an accurate measure of the EC-I relation (Griffin and Jurinak, 1973). PHREEQC version 2.12 (Parkhurst and Appello, 1999) was used for speciation and ionic strength calculation. 3.3 Results and discussion 3.3.1 Temperature dependence of EC Figure 3.1 shows the EC-T relations of the anaerobic digester supernatant/ centrate samples of Annacis Island WWTP, Goldbar WWTP, Lion's Gate WWTP, City of Penticton WWTP, and Lulu Island WWTP. Within the temperature range tested, the relationships, in all five cases, were well represented by a linear equation (R2>0.99). The solid lines in Figure 3.1 indicate equation 3-1 with a value of a ranging from 0.0197- 0.0205 °Cl with an average of 0.0198 °C"1 (Table 3.2), determined by the least-squares method. The root-mean squared (RMS) percentage error of the equation is defined by (Hayashi, 2004) 60 where e is the RMS percentage error, m is the number of data points, ECmeas is measured EC, and ECeq„ is the predicted EC. The value of e of equation 3-4 was found to range from 0.38-2.01 % (Table 3.2). Figure 3.2 shows the comparison between measured EC25 (EC at 25°C) and EC25 estimated from measured EC10 (EC at 10°C), using equation 3-1 with a =0.0198 °C' for all samples. The estimated EC25 values were found to be reasonably accurate, with a maximum error of 2.01% (Table 3.2). The 45° straight line in Figure 3.2 indicates a perfect match between measured and estimated values. Equation 3-1 is useful for calculating EC25, but it may be necessary to calculate EC at a standard temperature different from 25°C. As modified from equation 3-1, ECT = ^ (3-5) T " [\ + c(T-T0)) Where ECT is the electrical conductivity at a standard temperature 7b, and c is a constant defined by a c = (3-6) [i+fl(r0-25)] Equation 3-6 indicates that c varies with To. For example, c = a =0.0198 at 25°C, and c =0.0229 when a =0.0198 and J'o=180C. Therefore, the temperature compensation factor is dependent on the standard temperature used, even though electrical conductivity and temperature maintain a linear relationship, with a certain a value for a specific standard temperature. However, the EC value can be reasonably estimated based on any standard temperature. Figure 3.3 shows estimated EC values at 10°C for T0 =25°C (ECl0est@25) and for 7b=18°C (EC ioest@i8), and they were found to be statistically similar to each other (p=0.51) by the paired-t test. 61 3.3.2 EC- I relationship Table 3.3 shows compositions of 20 samples, of different dilutions, from the 5 WWTPs with 4 from each. PHREEQC version 2.12 with the Lawrence Livermore National Laboratory Database (llnl.dat), was used in this study for speciation and other thermodynamic equilibria. The charge balance of the selected samples was found be below 10%, using PHREEQC. An accurate assessment of all the ions present in the solution may only give an accurate charge balance. It is difficult, if not impossible, to accurately consider all the ions of the complex solution of anaerobic digester supernatant /centrate. Since the DOC level of the samples varied from 103 to 1458 mg l"1, low molecular fatty acids were suspected to make a contribution to the charge balance. However, volatile fatty acid analysis indicated that the contribution from acetic, propionic and buteric acids was insignificant. Table 3.4 shows pH, electrical conductivity and corresponding temperature readings of the samples. All ECT values have been converted to EC25 (Table 3.4), using a temperature compensation factor of 0.0198 in equation 3-1. The ionic strength of the samples was calculated, using PHREEQC where most of the ion-pairs were included in the calculation. Table 3.4 also shows the calculated I values. From a regression analysis using least squares method, the following linear EC-I relation, with r =0.94, was found: _6 / = 7.22x10 £C25 -0.0016 (3-7) where the ionic strength, I, is in mol F1 and EC25 is in pS cm"1 at 25°C The value of the intercept was found statistically insignificant (j?=0.60). Thus, the relationship can be written as 6 / = 7.22x10~ EC25 (3-8) 62 Relationships between EC25 and I for different types of water samples, have been 6 1 developed by others. A relationship of / (mol 1"') = 16xlO~ EC25 (uS cm" ) was developed for 13 waters of varying composition (Russell, 1976), while for extracts of flood soils and electrolyte 1 1 6 1 solutions of I less than 0.06 mol l" , the same relationship of/(mol l" ) = \6xlO' EC25 (uS cm" ) (Punnamperuma et al., 1966), and for arid-zone soil extracts and river waters, a relationship of 1 6 1 /(mol l" ) = 13x10' EC25 (uS cm" ) were developed (Griffin and Jurinak, 1973). In all three cases, the relationship was developed for a temperature of 25°C. The proposed relationship between the conductivity data obtained during the struvite solubility tests in different kinds of solutions, including supernatant/centrate, and the ionic strength manually calculated using major equilibia equations (Rahaman et al., 2006), was found 1 6 1 to be / (mol l" ) = 5xlO' EC25 (uS cm" ). Temperature dependence of conductivity was not considered in developing that relationship. The relationship developed in this study (equation 3- 8) was developed exclusively from anaerobic digester supernatant/ centrate, using the computer model PHREEQC. It involved most of the equilibrium equations and appropriately considered the temperature dependence of EC Figure 3.4 shows the observed EC25-I relationship in this study, as described in equation 3-8. It can be seen from Figure 3.4 that the EC25 values correlate better with / in the lower range of EC25. The higher variability of I, at a higher range of EC2s values, is due to the higher ion-ion interaction (Hall and Northcote, 1986). However, to minimize this effect, diluted samples of varying dilution factors were used to develop the relationship. Also, correction for ion-pairing was done by using the computer model PHREEQC. Table 3.5 shows a correlation matrix of the Pearson's correlation coefficients of three struvite constituting ions, namely magnesium, ammonium, and phosphate concentrations denoted 63 by [ ], their activities denoted by { }, EC25 and /. Neither the concentration nor activity of magnesium maintains a good correlation with EC25 or /. Both the concentration and activity of ammonium ion were found to have a good correlation with EC25 and I, with the maximum between the activity of ammonium and I. The correlation of phosphate ion was found to be less, in case of its effective concentration (activity) for both EC25 and I. The availability of these three species in a solution is controlled by the system pH, the total dissolved concentration of magnesium, ammonium, and phosphorus species, the ionic strength and the temperature of the system. In the case of struvite, the difficulty comes from the fact that all three of the reacting ions exhibit complex equilibria, in aqueous solution. The proportion of ammonium ion (NH4+) present in solution depends on its equilibrium with ammonia (NH3) and varies markedly with pH. Due to the triprotic nature of the orthophosphoric acid, several orthophosphate species exist in aqueous solution, resulting in a variable proportion of PO43" with pH of the solution. Examination of PHREEQC predictions for the various wastewaters used in this study showed that magnesium formed strong complexes with HCO3", 2+ 2 2 3 HP04 , CO3 ", NH3, S04 ", PO4 ", CI", and OH", depending on pH and concentration of the species in solution. Thus, the contribution of concentration or activity of free magnesium to EC, or I, decreases with the formation of such complexes. Struvite precipitation potential is greater in the anaerobic digestion and post digestion process, where comparatively higher dissolved fractions of the components are free to combine to form precipitates. Despite good correlation between PO4 " and EC25 or I, it is difficult to predict phosphorus removal from the EC variation alone, due to the presence of numerous ions and their interaction in the supernatant solution. However, the relationship developed between EC and ionic strength would certainly help predict 64 struvite precipitation as the solubility of struvite is largely affected by the ionic strength of the solution. 3.4 Conclusions • A representative temperature compensation factor for conductivity has been derived in a system associated with struvite formation from anaerobic digester supernatant/centrate samples of five different wastewater treatment plants. Using a temperature compensation factor, a =0.0198 °C~' for all samples, the estimated electrical conductivity values were found to fairly accurately match the measured values (R2~l). Although the value of the temperature compensation factor varies with the standard temperature used, the estimated EC values, based on any standard temperature, were found to be statistically similar to each other by paired t-test. 6 • Considering the temperature dependence of EC, a relationship/ = 7.22x10~ EC25, including correction for ion-pairing, was developed from anaerobic digester supernatant/ centrate. This relationship can be used to estimate the ionic strength of the solution in a system associated with struvite formation, from anaerobic digester supernatant/centrate. An in situ estimation of ionic strength, from electrical conductivity, would help predict struvite precipitation potential and effectively monitor the system performance. • Among the three constituent ions of struvite, the concentration and activity of Mg2+ was found to have a poor correlation with both EC25 and ionic strength. This is due to the fact that magnesium forms a number of complexes in situ, depending on pH and the concentration of the ions, and contributes less to EC and ionic strength, than the other two constituent ions. 65 Table 3.1 Expressions for activity coefficients. Approximation Expressions Approx. Applicability Guntelberg 7<0.1 M 1 1 A 2 J \ogy = Azt Davies K 0.5 M logy,. = -Az? {-JL--Q2I) 1 + V/ Where 7=1/2 Zm;.Zj2, mi is the molar concentration of ion i, z\ is the charge of ion i, A is a parameter associated with the absolute temperature and dielectric constant. Table 3.2 Temperature Compensation factor (a) and RMS percentage (e). Annacis Island Goldbar Lion's Gate City of Penticton Lulu Island a(°C-') 0.0198 0.0197 0.0199 0.0191 0.0205 e 1.02 0.99 2.01 0.38 1.71 ^0.0198 1.02 0.99 2.01 0.87 1.82 Table 3.3 Chemical composition of the anaerobic digester supernatant/centrate samples. All values in mg 1"'. 2 + + 3+ 2+ 2+ 2 2+ HCO3 Sample PO4-P NH4-N CI so4 NO3-N Mg Na K Al Fe Ca CO3 DOC Annacis I. 108 1290 38 10 0.7 1.0 24 150 1.0 1.4 11.5 4606 16.4 570 55 578 17 5 0.5 0.8 12 75 0.5 0.7 6.0 2534 6.6 900 30 266 11 3 0.3 1.1 6 37 0.3 0.2 3.9 1136 2.7 541 23 289 8 2 0.0 0.8 5 26 0.2 0.2 3.0 1269 3.4 299 Goldbar 118 744 101 12 0.2 2.3 93 259 0.5 0.7 11.6 3408 8.0 589 58 285 46 6 0.2 1.5 46 129 0.3 0.3 6.2 1263 3.1 385 28 148 21 3 0.1 0.9 23 65 0.1 0.2 3.5 627 1.3 214 23 118 20 2 0.1 0.9 19 52 0.1 0.1 3.2 500 1.0 154 17 Gate 6 286 78 11 0.5 6.1 45 128 1.0 1.2 20.6 1675 34.4 910 3 156 33 6 0.3 2.5 22 64 0.5 1.0 13.4 903 17.3 937 3 117 32 3 0.2 11.5 11 33 0.3 0.7 6.8 646 ' 10.3 257 7 181 36 2 0.1 1.8 9 27 0.2 0.6 5.4 775 12.4 274 C/Penticton 3 175 30 7 0.3 12.5 14 34 4.7 0.5 28.6 1035 7.5 653 3 110 15 4 0.3 5.4 7 16 2.4 0.2 14.5 517 3.5 240 2 56 8 2 0.3 3.2 4 9 1.2 0.2 7.7 257 1.1 163 2 43 7 1 0.1 2.7 3 7 1.0 0.2 6.3 193 1.0 103 Lulu I. 67 699 48 7 0.3 3.5 25 123 2.4 0.4 19.1 3190 10.4 1458 37 422 25 3 0.4 2.2 12 62 1.2 0.3 9.7 1914 6.2 600 18 194 12 2 0.1 1.3 6 31 0.6 0.1 5.0 890 2.5 291 14 181 10 1 0.3 1.2 5 24 0.5 0.1 3.9 761 2.1 274 68 Table 3.4 pH, Temperature, Electrical conductivity and calculated ionic strength. Sample Source pH ECT T EC25 I (mS cm"1) (°C) (uS cm"1) (mol 1"') Annacis Island 7.89 10790 21.1 11693 0.094 7.75 5790 21.4 6234 0.046 7.71 3230 21.5 3471 0.021 7.77 2770 21.4 2983 0.023 Goldbar 7.71 6810 18.4 7834 0.069 7.72 3610 20.7 3946 0.027 7.65 1710 21.2 1849 0.014 7.62 1430 21.2 1546 0.011 Lion's Gate 8.65 6800 19.3 7665 0.030 8.62 3670 20.6 4020 0.016 8.54 1700 21.0 1846 0.012 8.54 1430 21.1 1550 0.014 City of Penticton 8.2 2520 20.3 2779 0.018 8.17 1260 20.7 1377 0.010 7.98 690 20.9 751 0.005 8.04 590 20.9 642 0.004 Lulu Island 7.85 8120 20.1 8992 0.059 7.85 4380 20.8 4777 0.035 7.79 2540 21.1 2753 0.016 7.77 1860 21.0 2020 0.014 69 Table 3.5 Correlation matrix for anaerobic digester centrate supernatant/centrate. All values shown are Pearson's correlation coefficients of struvite constituting ion concentrations [ ], activities {}, EC (pS cm"1) and I (mol l"1). 2+ 2+ + + 3 I 3 EC25 [Mg ] {Mg } [NH4 ] {NH4 } [P04 ] {PO4 -} EC25 1.00 I 0.95 1.00 [Mg2+] -0.21 -0.27 1.00 {Mg2+} -0.28 -0.32 1.00 1.00 + [NH4 ] 0.71 0.81 -0.30 -0.34 1.00 + {NH4 } 0.91 0.99 -0.32 -0.36 0.83 1.00 3 [P04 "] 0.90 0.97 -0.28 -0.32 0.85 0.97 1.00 3 {P04 "} 0.57 0.60 -0.36 -0.39 0.54 0.61 0.50 1.00 70 Figure 3.1 Electrical Conductivity vs. Temperature of the supernatant/centrate samples from five WWTPs. 71 Figure 3.2 Estimated EC25 vs. Measured EC25 of supernatant/centrate samples. 10000 • EC10est@25 Annacis Goldbar Lion's City of Lulu Island Island Gate Penticton WWTPs Figure 3.3 Measured and estimated ECio using different standard temperatures. 73 0.12 0.10 O 0.08 R2 = 0.89 O 0.06 0.04 O 0.02 jQ&^v^ O 0.00 2000 4000 6000 8000 10000 12000 14000 1 EC25 (nS cm" ) Figure 3.4 Ionic Strength vs. Electrical Conductivity. 74 References American Public Health Association (APHA), American Water Works Association, and Water Pollution Control Federation. (1998) Standard Methods for the Examination of Water and Wastewater, 20th ed., Washington, DC. Banister, S. S., Pitman, A.R. and Pretorious, W.A. (1998) The solubilization of N and P during primary sludge acid fermentation and precipitation of the resultant P. Water SA 24(4), 337-342 Bhuiyan, M.I.H, Mavinic, D.S. (2007) Assessing struvite precipitation in a pilot-scale fluidized bed crystallizer. Environmental Technology (Submitted). Griffin, R.A. and Jurinak, J.J. (1973) Estimation of activity coefficients from the electrical conductivity of natural aquatic systems and soil extracts. Soil Science 115(1), 26-30. Hall, K.J. and Northcote, T.G. (1986) Conductivity-temperature standardization and dissolved solids estimation in a meromictic lake. Canadian Journal of Fish and Aquatic Science 43, 2450- 2454. Hayashi, M. (2004) Temperature-electrical conductivity relation of water for environmental monitoring and geophysical data inversion. Environmental Monitoring and Assessment 96(1-3), 119-128. Jardin, N. and Popel, H.J. (1994) Phosphate release of sludge from enhanced biological P- removal during digestion, Water Science and Technology 30(6), 281-292. Lawrence Livermore National Laboratory database (llnl.dat) derived from thermo.com.V8.r6t.dat Millero, F.J. (2001) The Physical Chemistry of Natural Waters, Wiley-Interscience, New York, USA. Parkhurst, D.L., Appelo, C.A.J. (1999) User's guide to PHREEQC (Version 2) - a computer program for speciation, reaction- path, advective-transport, and inverse geochemical calculations. USGS water-Resources Investigation Report 99-4259. Ponnamperuma, F.N., Tianco, E.M., and Loy, T.A. (1966) Ionic Strengths of the Solutions of Flooded Soils and Other Natural Aqueous Solutions from Specific Conductance. Soil Science 102,408-413. Rahman, M.S., Mavinic, D.S., Bhuiyan, M.I.H, and Koch, F.A. 2006. Exploring the determination of struvite solubility product from analytical results. Environmental Technology 27,951-961 (2006). 75 Russell, L.L. (1976) Chemical Aspects of Groundwater Recharge with Wastewaters, Ph.D. Thesis, University of California, Barkley, USA. Snoeyink, V.L. and Jenkins, D. (1980) Water Chemistry, John Wiley & Sons, New York, USA. PP 74. Sorensen, J.A. and Glass, J.E. (1987) Ion and temperature dependence of electrical conductance for natural waters, Analytical Chemistry 59, 1594-1597. Stumm, W. and Morgan, J. (1981) Aquatic Chemistry, Wiley-Interscience, New York, USA, pp 135,407. 76 Chapter 4 Nucleation and growth kinetics of struvite in a fluidized bed reactor 4.1 Introduction Struvite precipitation from wastewaters has gained importance as a means of nitrogen and phosphorus removal and recovery (Kabdasli et al, 2006; Doyle and Simon, 2002). While a substantial number of studies have been conducted on the thermodynamics of struvite crystallization, much less work has been carried out on the kinetics of struvite precipitation. Crystallization kinetics can be separated into two phases: nucleation and growth. Nucleation occurs when ions combine to form a crystal embryo that can act as the foundation for growth into detectable crystals. Growth results from the assimilation of ions in the lattice structure established by the crystal embryo foundation (Ohlinger, 1999). Nucleation results in the apparition of new particles. Several mechanisms are possible, depending on the supersaturation level in the crystallizer. Homogeneous primary nucleation corresponds to nuclei apparition directly in the supersaturated solution. Heterogeneous primary nucleation occurs on foreign surfaces, which can be dust in suspension or parts of the crystallizer itself. Surface secondary nucleation, also called true secondary nucleation, needs to have particles of the same species as the solid which is crystallized, already in suspension. Indeed, surface secondary nucleation corresponds to the nuclei formation on the surface of these particles. The nucleation rate of the mechanism under consideration is negligible within its own metastable zone and increases rapidly if its metastable zone limit is exceeded. Among the *A version of this chapter has been submitted for publication: Bhuiyan, M.I.H, Mavinic, D.S., and Beckie, R.D. (2007) Nucleation and growth kinetics of struvite in a fluidized bed reactor. Journal of Crystal Growth, in review. 77 different nucleation mechanisms, homogeneous primary nucleation requires the highest supersaturation level to spontaneously develop. It is followed by heterogeneous primary nucleation and surface secondary nucleation, respectively (Regy et al. 2002). Since homogenous nucleation occurs only in highly purified and/or highly supersaturated solutions, heterogeneous nucleation is expected to be the predominant nucleation mechanism in wastewater treatment environments. Growth is the process of incorporating constituent ions into the crystal lattice of the embryos, to form detectable crystals. Nucleation of crystal embryos and growth of the embryos into detectable crystals is usually combined in the study of crystal formation. The time that elapses between the establishment of supersaturation and the first changes in the physical properties (due to the formation of solid phases) is called the "crystallization induction period" (sometimes also the 'waiting time' or 'incubation period'). Several different proxies have been used for induction time: conductivity (Kabdasli et al, 2006), observation of light scintillations (Ohlinger, 1999; Gunn and Murthy, 1972), pH (Bouropoulos and Koutsoukos, 2000), and absorbance (Kabdasli et al, 2006). Mixing energy, pH, supersaturation and the presence of foreign ions are the important factors affecting induction time (Kabdasli et al, 2006). However, following the induction period, crystal growth continues until equilibrium is attained. In systems continuously replenished with struvite constituents, crystal growth continues indefinitely. Crystallization kinetics and the role of induction time are of particular importance for intentional struvite formation in the fluidized bed reactors. Two processes exert a decisive influence during the growth of crystals from solution: mass transport from the solution to the crystal surface by mass diffusion and convection, and incorporation of material into a crystal lattice through the surface integration process, sometimes 78 referred to as the surface reaction process (Sohnel and Garside, 1992). A number of researchers have conducted studies on crystal growth of different substances and several models have been proposed. Transport limitation models are typically based on Fick's law of diffusion. However, the kinetic equation for diffusion-controlled growth is more complex, as the diffusion of each ion is characterized by a different diffusion coefficient. Surface integration models take several forms and reported results vary significantly (Sohnel and Garside, 1992). This variability probably results from the difficulty in evaluating a number of the parameters appearing in the kinetic expressions; and also, multiple mechanisms can influence the growth rate (Ohlinger, 1999). The effects of operating variables, such as ionic strength, pH, superficial velocity, crystal size, seed type, and temperature have already been identified (Tai et al, 1999; Sohnel and Garside, 1992). Small crystals grow faster than large ones at high supersaturations because of diffusion mechanism. At low supersaturations, the small crystals grow more slowly due to the Gibbs- Thomson effect. This effect is also known as Ostwald ripening, in which concentration gradients cause small precipitates to dissolve and larger ones, having smaller surface-to-volume ratios, to grow. At an intermediate supersaturation, the growth rate is independent of size, since the two size-dependent effects are in balance. Temperature usually has a significant influence on crystal growth rate. It can change the relative rates of the diffusion and surface integration steps to such an extent that, at high temperatures, diffusion controls the growth rate. At low temperatures, under otherwise identical conditions, growth is controlled by the surface integration step (Sohnel and Garside, 1992). In some cases, pH influences the crystal growth. A small pH variation may cause significant changes in the zeta potential of the particles, leading to the eventual destabilization of the suspension by aggregation. Such phenomena are expected to influence the 79 kinetics of crystal growth by reducing the respective rates, since a smaller surface area is available to the growth units (Bouropoulos and Koutsoukos, 2000). It is also suggested that the acceleration of the growth rate can be achieved by controlling the pH of the solution (Omar and Ulrich, 2003). Among all the available models for growth of crystals, the two-step growth model is considered the most useful, especially from the chemical engineering point of view (Tai et al., 1999). The two-step growth model is based on the fact that growth, by mass deposition, consists of two steps, viz. a diffusion process, whereby solute molecules are transported from the bulk fluid phase to the solid surface, followed by a reaction process, when the solute molecules arrange themselves into the crystal lattice. Under steady-state conditions, the simplified scheme for two-step crystal growth can be described mathematically by equations 4-1 and 4-2: G' = Kd (cr-(Ji) mass transfer (4-1) = Kr(7i surface reaction (4-2) where, G'is the crystal growth rate, cr and cr, are the over all relative supersaturation and the interfacial relative supersaturation, Kd and Kr are mass transfer coefficient and surface reaction coefficient respectively, and r is the surface reaction order. Fluidized beds have been adopted as separators for many operations in the chemical industry, especially for the crystallization operation. The most efficient, modern industrial crystallizers are of suspended-type bed, in which the crystals are grown in a liquid fluidized zone. One such, pilot-scale, liquid-fluid, fluidized bed reactor (FBR), developed at The University of British Columbia, Vancouver, Canada (UBC), has been utilized in this research (Bhuiyan et al, 2007a, chapter 5). 80 The prime objective of this study was to investigate the nucleation and growth kinetics of struvite. An induction-time study of struvite was carried out in the laboratory, for the concentration range encountered in the pilot-scale operation. The nucleation kinetics, thus obtained, helped determine a metastable region for struvite. Subsequently, the growth experiments were performed in the metastable region, using in a bench - scale FBR with a pH controller, and using seed crystals from the pilot-scale FBR. 4.2 Materials and methods 4.2.1 Determination of induction time The driving force for nucleation of struvite in aqueous supersaturated solutions is the difference between the chemical potential of struvite in the supersaturated solution ps and the corresponding value at equilibrium, \JLX (Regy et al, 2002): Au= Uoc - ps (4-3) Combining this expression with the definition for chemical potential yields: -Au= kT ln Q. (4-4) where, k = the Boltzman's constant, T = absolute temperature and D. = activity based super saturation ratio, 2+ + 3 {Mg }{/v7/4 }{PQ4 -} For struvite, Q = , where Ksp is the thermodynamic solubility product K sP of struvite. The homogenous nucleation rate, J, is given by: py\2 J = A"exp (4-5) kT(-Ap)2 V where, A" = pre- exponential factor 81 (3 = geometric factor (167i/3 for a sphere and 32 for a cube) y = surface energy or the interfacial tension between crystals and solution vm = molecular volume Based on the assumption that nucleation time is much greater than growth time during the induction period (t„»tg), and using the statistical concept of nucleation (Ohlinger, 1999): ',«/ = -j (4-6) Combining equations 4-4, 4-5, and 4-6 and taking the log transformation yields: log^ = —- B (4-7) (logQ) .3.. 2 where, A = —^—and B = log A" (2.303kTY Experiments were conducted to measure the induction period by blending solutions in a 2-litre square beaker, using a Phipps and Bird stirrer. The time lapsed between pH adjustment and the first pH changes (which could be measured in a working solution) was defined as the induction time, tinci (Bouropoulos and Koutsoukos, 2000). The stirring speed was set and pH change was monitored closely. All experiments were conducted in a temperature-controlled o o (25 C± 0.5 C) room. Solutions were prepared and stored in the same room. The solution pH was continuously monitored using an OAKTAN Waterproof pH tester BNC 10, equipped with a VWR symphony probe. The beaker was sealed with the stirrer connected to a motor mounted on top, following the specification of Phipps and Birds for a square beaker (Ebling et al, 2003). The pH probe was also inserted through a threaded hole, made especially for the probe, and sealed with a rubber gasket. Solutions were prepared using dry reagents and distilled water. In order to make solutions with concentrations encountered during the pilot scale operation, di-ammonium 82 phosphate ((NH^HPC^), ammonium chloride (NH4CI), and magnesium chloride (MgCl2-6H20) were added, as required. NaHC03 was used for alkalinity addition. Table 4.1 shows the concentrations and conditions of different experiments in the induction time study. 4.2.2 Determination of relative supersaturation Relative supersaturation, the driving force for crystal growth, is defined as: rj = Q1/3-l (4-8) The activities of the struvite constituents were computed for the measured pH, using a computer model PHREEQC. The detailed use of this model is described elsewhere (Bhuiyan et al, 2007a). In the operation of the fluidized bed reactor, supersaturation was maintained in a suitable range. A low supersaturation results in a low growth rate and therefore a low removal rate of species, and a high supersaturation causes unwanted nucleation. Phosphate and ammonia species change with pH, with the subsequent effect on the relative supersaturation. As such, the control of pH is important in a study of crystallization kinetics. 4.2.3 Identification of metastable region From the viewpoint of thermodynamics, supersaturated solutions are always unstable. Up to a certain concentration, however, they appear to be stable in that their properties do not change for a comparatively long period of time. This metastable state can be determined for supersaturated solutions of both readily and sparingly soluble substances. To suppress nucleation of struvite in growth experiments, the supersaturation was kept in the metastable region, which is the region between super solubility and solubility curves (Tai et al, 1999). Since a supersolubility curve for struvite is not available, experiments were first performed to identify the metastable region. Under constant temperature (25°C) and constant pH (8 and 8.5), 100 ml of (NH4)2HP04 and 83 NH4C1 solutions of known molarities (to yield a range of constituent concentrations) were mixed and stirred at a constant speed of 200 rpm, while a MgCl2-6H20 solution of known molarity (0.1 M) was added from a burette. Nucleation was identified by change in pH, as mentioned in the induction time determination, as well as visually. The pH drop and detectable precipitation took place simultaneously. Table 4.2 shows the concentrations of the constituents that caused homogenous primary nucleation at pH 8 and 8.5. Subsequently, the concentrations and activities of the constituents causing nucleation were calculated, using the same computer model as was used for calculating supersaturation. The saturation index values (SI = log Q) of the solutions, as shown in Table 4.1, were also calculated by the PHREEQC model. Finally, the metastable region was constructed and growth experiments were performed in this region. 4.2.4 Crystallization system The crystallization system, containing a bench-scale fluidized-bed crystallizer, a storage tank, and a pH control system (Fig. 4.1) was used to measure the growth rate of struvite pellets. The bench-scale crystallization system was similar to the pilot-scale system (Chapter 5), except that the feed (N and P) and magnesium were injected from a storage tank. Immediately above the upper section is an enlarged section to prevent seed crystals from carrying over to the storage tank. To maintain a better control of growth within the metastable zone, all of the ingredients were mixed in one solution. The supersaturated solution (39 L), thus prepared with a desired pH, was poured into the storage tank, and another 5.65 L of the same solution was fed into the fluidized bed. The solution in the storage tank was then pumped into the fluidized bed through an injection port, using a peristaltic pump. When the flow rate and pH became steady, 40 g of seed (struvite pellets) were introduced into the crystallizer from the top, through a funnel. The funnel was inserted beyond the bottom of the enlarged section, so that the struvite pellets would remain 84 suspended within the reactor. The growth experiment began immediately afterwards. The pH of the solution was maintained constant (8.07) during the operation, by adding NaOH in combination with a pH controller. An aliquot sample of approximately 20 ml was withdrawn from the middle of each section, using a long tube connected to a plastic syringe. The sampling was undertaken every 30 min. The sample was filtered through a 0.45 pm filter. A typical run was lasted for 420 min. 4.2.5 Determination of crystal growth rate The total mass (W) and surface area (A) of crystals are given by W = NiVVL3 (4-9) A = N£f L2 (4-10) where, pp is the crystal density; fy and fA are the volume and surface area shape factor, respectively; W and L are the total mass and average diameter of the crystals; A is the surface area and N is the number of crystals. The crystal growth rate expressed in kg m"2 s"1 is (Tai et ai, 1999) R--¥T (4-N) A dt or, g fA ^ Combining equations 4-11 and 4-12, the expression for the linear growth rate (G') can be written as: dt 3fvp Adt 85 During the growth experiment, the change of crystal mass in a crystallizer is related to the change of the limiting ion (ortho-phosphate) concentration. Ortho-phosphate concentration was kept limiting by adding magnesium in a molar ratio (Mg/P) of 1.3. However, the change of mass can be expressed as: (4-14) dt I dt Where, M is the molecular weight of struvite (MgNH4P04. 6H2O), and V is the solution volume. Substituting equation 4-14 into equation 4-13 yields: f MV G' = A (4-15) dt Combining equations 4-9 and 4-10: L /A PP (4-16) fYA W Substituting equations 4-16 into equation 4-15, we obtain the linear growth rate expression: LMVf d[CP04]\ G' = 1 3W I dt' J Once the concentration profile (Fig. 4.2) of Cpo4 is determined experimentally, the linear crystal growth rate can be evaluated with the assumption that changes in size and weight for the seed crystals are negligible (Tai et al. 1999). 4.2.6 Analysis Analyses for ortho-phosphate and ammonia were made, using the flow injection method on a LaChat QuickChem 8000 instrument, as described in the method number 4500-P G and 45OO-NH3 H of the Standard Methods for the Examination of Water and Wastewater (APHA et 86 al, 1998). Magnesium analysis was performed by flame atomic absorption spectrophotometry, using a Varian Inc. SpectrAA220 Fast Sequential Atomic Absorption Spectrophotometer. 4.3 Results and discussion 4.3.1 Nucleation and induction time Measured induction times were defined operationally as the duration from initial mixing of precipitant solutions to the on set of pH change. Among various indicators conductivity, absorbance and pH were tested. pH was selected because of its consistent response as an indicator and its simultaneous response with visual observations. The measured durations from any of these indication methods were consistent within a set of experiments. Therefore, induction time measurements should not be considered absolute measurements that can be compared between methods; rather, they should be treated as relative measurements, useful for evaluating a system (Ohlinger, 1999). The results of the induction time study of this work were analyzed using equation 4-7, which has been previously used by several researchers, to model nucleation kinetics of struvite (Bouropoulos and Koutsoukos, 2000; Ohlinger, 1999; Abbona and Boistelle, 1985). The induction time preceding the onset of crystallization was found to be inversely proportional to the solution supersaturation, as seen in Figure 4.3. Results from the induction time experiments, conducted at a constant mixing speed, are presented in Figure 4.4. A trend line was fitted to the experimental data, and the slope (A= 7.16) and intercept (-B=-1.0) were indicated. Previously, it was suggested that the experimental slope is predictable using the induction time model, if all parameter values for nucleation sites are known. Conversely, the experimental intercept is a function of the selected induction period endpoint indication method, and the experimental values will vary based on the selected experimental procedures (Ohlinger, 1999). From only one 87 linear part of Figure 4.4, it can be seen that heterogeneous nucleation did not occur in the supersaturation range studied (Mullin, 2001). However, fundamental nucleation theory can be applied to heterogeneous nucleation by correcting for the reduced activation energy required, due to the catalytic effect of impurity particles (Sohnel and Garside, 1992) The rate of homogenous nucleation, J, is related to the supersaturation ratio, as can be found by combining equations 4-4 and 4-5: J = A"exp (4-18) 3 2 (£r) ln Qy When the supersaturation is such that J = 1 nucleus cm"3 s"1 (corresponding to the breakdown metastable state) it is possible to obtain a numerical value of the surface energy, y (Abbona and Boistelle, 1985). A" can be given a reasonable value of 1017 nuclei cm"3 s"1. For the sake of simplicity and in order to compare the results found in this work to those by others, it was assumed that the nucleus was cubic ((3 = 32) and that the molecular volume is given by vm= M/Npsn, where M is the molecular weight (245.44 g); N, Avogadro's number; ps, crystal density (1.711 g cm"3); and n, the number ions in the formula unit. From the experimental results, a value of y = 43.0 mJm"2 was calculated for struvite, which is in the order of magnitude expected for sparingly soluble salts and has also been reported in the literature (Abbona and Boistelle, 1985; Bouropoulos and Koutsoukos, 2000). The induction time model in equation 4-7 is based on assumptions of reaction-controlled nucleation and diffusion -controlled growth, when tn » tg. The linearity of the experimental data with a good fit (Fig. 4.4) suggests that supersaturation is the primary influence on the struvite precipitation induction time, when mixing energy and, thus, transport influences are held constant. To determine the extent of the influence of the mixing energy on the induction period, 88 experiments were conducted at a constant supersaturation (SI =1.83) and selected stir speeds. The mixing speeds were converted to G values (s1) with the help of G-curves (Ebling et al. 2003). It can be seen from Fig. 4.5 that induction time is also influenced by mixing energy, as indicated by the slope of the line. It has been previously suggested that the influence of mixing energy on the induction time is relatively small, indicating that the induction period is dominated by reaction-controlled processes (Ohlinger, 1999). 4.3.2 Metastable region 3 2+ From the concentration data in Table 4.2, concentrations [P04 ~], [NFL/], and [Mg ] and 3 + 2+ activities {P04 "}, {NH4 }, and {Mg } of the free ions were calculated using PHREEQC. The 3 Ksp value of struvite was considered to be 13.36 (Bhuiyan et al, 2007b, chapter 2). -log [P04 ~ + 2+ ][NH4 ] against -log[Mg ] has been plotted in Figure 4.6a, to mark the supersolubility curves. The solubility curves were also plotted using equilibrium values, calculated using PHREEQC. It can be seen from the curves that the solubility and supersolubility curves are straight and almost parallel lines in the concentration range studied, and the metastable region is independent of the 3 + pH values used in this study. The curve drawn with the activities, i.e. -log {P04 "} {NH4 } 2"1" against -log{Mg } (see Fig. 4.6b) also showed the same pattern. A similar observation has been suggested from a study with CaC03 except that the solubility and supersolubility curves were bent at both ends, for curves drawn with concentrations (Tai et al, 1999). This may happen when the formation of complex ions is significant (Sohnel and Garside, 1992). 4.3.3 Growth The growth of struvite was studied in a bench-scale fluidized bed reactor, which was designed following the same principle as the pilot-scale unit, having three sections (Bottom, 89 Middle and Top) of increasing diameter. The diameter changes cause turbulent eddies above each transition, ensuring that sufficient mixing existed in the reactor from top to bottom (Britton et a/., 2005). Due to the deposition of ions at the growth sites, the supersaturation level of the solution generally decreased as it moved upward (Fig. 4.7). However, the change of Q, between each level, was not found to be statistically significant by a paired-t test. The conductivity of the solution, measured in three sections, maintained a more or less constant median value (Fig. 4.8). Superficial velocity was kept constant, 380 cm min"1 in the bottom section, with a gradual decrease in the upper sections due to increasing diameter. The average diameter of the seeds, grown in the pilot-scale reactor with anaerobic digester centrate, and later identified as struvite by X-ray crystallography, was 1.25 mm (0.5-2.00 mm in size). Once dropped from top of the reactor, they rearranged themselves by gravity, with bigger ones at the bottom. The growth rates were calculated for the average concentrations in three sections. Figure 4.9 shows a log G' - log a plot for pH = 8.07 in the specified conditions. A previous study, involving calcite growth rate in a dense fluidized bed reactor, identified supersaturation, pH, ionic strength, and particle size and type of seeds as important factors affecting growth rates (Tai et al., 1999). However, growth rates can be calculated in each section separately and compared by log G' - log a plots, if the mass of crystals distributed in each section were known. The determination of mass of the seeds in each section was beyond the scope of this study. However, during the experimental runs, not as much crystal mass was visible in the top section as in the bottom and middle section, due to the size distribution of the seeds used and hydrodynamic conditions inside each section. The bottom section, having higher supersaturation, is favorable for the faster growth of the small crystals. However, the large crystals have higher 90 settling velocities and need larger upward flow velocities to be fluidized. As a result, the relative velocities between the crystals and the solution will increase with increasing crystal size. This leads to higher mass transfer in the case of larger crystals, resulting in faster growth. It has been suggested that the increase in growth rate is not a result of crystal size, rather a result of higher relative velocities of the fluid phase (Omar and Ulrich, 2003). Thus, the dependence of growth rate on particle size is apparent, and the real effect is caused by fluid dynamics. In other words, the growth is a transport-controlled process in an FBR. In a pilot-scale operation, where supersaturation cannot be strictly controlled in the metastable region, nucleation may be triggered in the bottom section. Those newly formed crystals would tend to move upward, with the flow, to the top section, where they would need a lower relative velocity to be fluidized. With a slower growth in the top section, they would soon gain a higher growth potential, as they grow bigger and gradually descend. During the pilot-scale crystallization process, agglomeration, in the presence of supersaturation or aggregation without supersaturation, is essentially observed according to these three successive steps: the collision of two particles; a sufficient period during which the two particles stay together; and adherence of the two particles with the help of supersaturation. A detailed study is thus recommended, to properly model the agglomeration phenomenon in the main UBC crystallizer. 4.3.4 Growth rate expressions One of the disadvantages of using the two-step crystal growth model is the anonymity of the surface reaction order. As summarized in a previous study, second-order kinetics of the surface reaction step was appropriate for most systems (Tai, 1997). A study involving'struvite, suggested the order to be >1 (Kofina and Koutsoukos, 2004). The validity of the assumption of a first-order surface reaction is questionable. Many inorganic salts, crystallizing from aqueous 91 solution, show rates slightly greater than first-order, while others indicate a second order reaction (Mullin, 2001). Taking r = 2, equations 4-1 and 4-2 can be combined to give, 1 1 (4-19) IG' K Thus, the mass transfer coefficient, kj, and surface reaction coefficient, kr were evaluated from the slope and intercept of the plot, o/fG' vs. fG' and shown in Table 4.3. A mass-transfer coefficient for calcite, thus obtained, has been suggested to be independent of crystal size and superficial velocity, while surface-reaction coefficient for calcite is size dependent (Tai et al., 1999). The values of both the coefficients for struvite found in this study were higher than those for calcite. A detailed study, considering different conditions, would help decide the variability of these coefficients. However, for a surface-reaction order r=2, the following expression of crystal growth rate can be used (Mullin, 2001): G' = K 1+- d -1 cr (4-20) 2kr(j j 2k,a An alternative equation (equation 4-21) has also been suggested to calculate the linear growth rate (G') (Omor and Ulrich, 2003): 1/3 c \N ' mf •1 (4-21) At \moJ where, Lo is the average size of seed crystals at the start of the run, m0 and mt- are the masses of seed crystals at the start and the end of the run, respectively, and At is the residence time. From the linear growth rates found from equations 4-17 and 4-20, the increase in crystal size for a growth time of 420 min was estimated to be 5-6% of average size (1.25 mm). When equation 4- 21 was used to estimate the increase in crystal size, it was found to be 2-3% of the average size 92 1.25 mm. The results, for the concentrations tested, were of the same order of magnitude. One of the main reasons for variation may be difficulties in recovering all the grown crystals from the reactor, in the later case. However, the linear growth rate, G' in m s"1, can be easily converted to mass growth rate or molar growth rate, using equations 4-13, 4-14 and 4-16. In a pilot scale operation for struvite crystallization, where constituents are being continuously replenished, growth, as a whole, in kg d"' can also be calculated on the basis of a decrease in the limiting ion concentration and the feed flow in the reactor (Bhuiyan et al, 2007a, Chapter 5). 4.4 Conclusions • Nucleation was found to be the controlling process for struvite crystal formation during the induction period. pH monitoring proved to be an effective method of induction time determination. The induction period for struvite, determined in the laboratory, was found to be primarily reaction controlled, with minor transport influence. • The metastable region for struvite was explored in this study. The nucleation of struvite was suppressed when the growth experiments were conducted in the metastable region. The solubility and supersolubility curves, which are the boundaries of the metastable region, are almost parallel straight lines in the concentration range studied. • The growth-rate of struvite, in a fluidized bed reactor, was found to be mainly transport-controlled. The struvite growth-rate data, at constant pH and ionic strength, were analyzed by a two-step growth model. With the determination of the mass-transfer coefficient and surface-reaction coefficient for a specified 93 condition, a linear growth rate model for struvite growth determination in a fluidized bed reactor has been proposed. A detailed study, considering different operating parameters in a fluidized bed reactor has been recommended. This would help determine the variability of the mass-transfer coefficient and surface-reaction coefficient with various factors. Agglomeration should also be addressed in detail to come up with a model involving this phenomenon. 94 Table 4.1 Concentrations and conditions used in different experiments during induction time study. + Run rpm pH HC03 PO4-P Mg NH4-N CI Na lind SI mgl"1 mgl'1 mgr' mgl' mgr1 mg r' sec 1 85 8.25 4575 105 107 1050 312 1724 50 1.83 2 100 8.25 4575 105 107 1050 312 1724 24 1.83 3 60 8.25 4575 105 107 1050 312 1724 100 1.83 4 130 8.25 4575 105 107 1050 312 1724 10 1.83 5 50 8.25 4575 105 107 1050 312 1724 134 1.83 6 120 8.25 4575 105 107 1050 312 1724 12 1.83 7 120 8.25 3050 70 71 700 208 1150 210 1.48 8 120 8.51 3050 70 71 700 208 1150 35 1.7 9 120 8.35 3050 70 71 700 208 1150 85 1.56 10 120 8.2 3050 70 71 700 208 1150 320 1.43 11 120 8.3 3050 70 71 700 208 1150 125 1.52 12 120 8.4 3050 70 71 700 208 1150 60 1.61 13 120 8.45 3050 70 71 700 208 1150 40 1.65 14 120 8.15 3050 70 71 700 208 1150 500 1.38 15 120 8.46 2440 56 57 560 167 920 240 1.46 a SI = Saturation Index, tjnd = Induction Time. 95 Table 4.2 Determination of metastable region for struvite at 25 C. pH P04-P NH4-N Mg SI nigl1 nigf1 nigl1 8.0 102 490 379 1.87 202 669 201 2.14 81 709 261 1.86 163 814 339 2.23 1286 1206 101 2.4 323 908 171 2.33 645 1008 161 2.48 8.5 101 486 181 2.23 201 667 121 2.47 41 607 241 1.98 81 706 161 2.24 161 807 181 2.57 321 904 81 2.53 643 1005 101 2.76 1303 1222 359 3.4 aSI = Saturation Index 96 Table 4.3 Mass transfer coefficient and surface reaction coefficient of struvite at pH=8.07. Particle size Superficial velocity Electrical kd K Conductivity (mm) (cm min"1) (mS cm"') (10"8ms"') (10"'ms"') 1.25 380 10.82 (+0.24) 1.11 7.99 97 106 mm m 65 nan ID Down Pipe Figure 4.1 Bench-scale fluidized bed crystallization system for struvite growth experiment. 98 Time (min) Figure 4.2 Concentration profile of Cpo4 and their average in three different sections of the reactor used for struvite growth experiment. 99 600 500 o _ 400 o o a> o 3, 300 o o ~ 200 ° ° o o 100 0 10 20 30 40 50 60 70 80 0 n Figure 4.3 Induction time versus solution supersaturation for struvite at 25°C for 120 rpm (G value = 140 sec"1). Driving force for precipitation as a function of induction time. 100 3.5 3 ^ y = 7.1622x-1.0011 2.5 R2 = 0.99 o CD 2 1.5 o 1 0.5 0 0 0.2 0.4 0.6 0.8 1/log2 Q Figure 4.4 Struvite induction time at selected supersaturation levels at a constant mixing speed of 120 rpm. 20 40 60 80 100 120 140 160 180 G Value (sec"1) Figure 4.5 Variation of induction time with G value (s"1) at a selected saturation level (SI = 1.83). Error bars: 95% confidence interval. 102 o Q_ O 0 1 -log[Mg2+] o pH=8 • pH=8.5 A pH=8 • pH=8.5 - Solubility curve Supersolubility curve (b) O CL D) O -log{Mg2+} ° pH=8 • pH=8.5 A pH=8 A pH=8.5 Supersolubility curve Solubility curve Figure 4.6 Determination of metastable region for struvite with (a) concentrations (b) activities of the ions. 103 6.00 o Bottom 5.00 A Middle • Top 4.00 C5 3.00 o o go o • a o * B B • 2.00 A 1.00 0.00 ~i r ~ r 0 50 100 150 200 250 300 350 400 450 Time (min) Figure 4.7 Variations of supersaturation in three different sections of the struvite reactor. 104 o GO > o T3 C o O o -t—I o _0B LU Bottom Middle Figure 4.8 Box-plots of electrical conductivity variation in three different sections of the struvite reactor (n=15). Error bars: 95% confidence interval. Percentiles shown: 10th, 25th, 75th, and 90th. The horizontal line inside the box represents the median. 105 References Abbona, F. and Boistelle, R. (1985) Nucleation of struvite (MgNH4P04. 6H20) single crystals and aggregates. Crystal Research & Technology 20(2), 133-140. American Public Health Association (APHA), American Water Works Association, and Water Pollution Control Federation. (1998) Standard Methods for the Examination of Water and Wastewater, 20th ed., Washington, DC. Bhuiyan, M.I.H, Mavinic, D.S. (2007a) Assessing struvite precipitation in a pilot-scale fluidized bed crystallizer. Environmental Technology (Submitted). Bhuiyan, M.I.H., Mavinic, D.S., and Beckie, R.D. (2007b) A solubility and thermodynamic study of struvite. Environmental Technology. In press. Bouropoulos, N.C and Koutsoukos, P.G. (2000) Spontaneous precipitation of struvite from aqueous solutions. Journal of Crystal Growth 213, 381-388. Britton, A., Koch, F.A., Mavinic, D.S., Adnan, A. Oldham, W.K., and Udala, B. (2005) Pilot- scale struvite recovery from anaerobic digester supernatant at an enhanced biological phosphorus removal wastewater treatment plant. Journal of Environmental Engineering and Science 4, 265- 277. Doyle, J.D. and Simon, A.P. (2002) Struvite formation, control and recovery. Water Research 36, 3925-3940. Ebling, M.J., Sibrell, P.L. Ogden, S.R., and Summerfelt, S.T. (2003) Evaluation of chemical coagulation-flocculation aids for the removal of suspended solids and phosphorus from intensive recalculating aquaculture effluent discharge. Aquaculture Engineering 29, 23-42. Gunn, D.J. and Murthy, M.S. (1972) Kinetics and mechanisms of precipitations, Chemical Engineering Science 27(6), 1293-1313. Kabdasli, I., Parsons, S.A., and Tunay, O. (2006) Effect of major ions on induction time of struvite precipitation. Croatia Chemica et Acta 79(2), 243-251. Kofina, A.N. and Koutsoukos, P.G. (2005) Spontaneous precipitation of struvite from synthetic wastewater solutions. Crystal Growth & Design 5(2), 489-496. Mullin, J.W.(2001) Crystallization. 4th ed, Butterworth -Heinemann, Jordan Hill, Oxford, UK. Ohlinger, K., Young, T.M. and Schroeder, E.D. (1999) Kinetics effects on preferential struvite accumulation in wastewater. Journal of Environmental Engineering 125(8), 730-737. Omar, W. and Ulrich, J. (2003) Influence of crystallization conditions on the mechanism and rate of crystal growth of potassium sulphate. Crystal Research Technology 38(1), 34-41. 107 Regy, S., Mangin, D., Klein, J.P., Lieto, J. (2002) Phosphate Recovery by Struvite Precipitation in a Stirred Reactor, Lagep, Centre Europeen d' Etudes des polyphosphates. Sohnel, O. and J. Garside. (1992) Precipitation: Basic Principles and Industrial Applications. Butterworth Heinmann, Oxford, England, p.78-79. Tai, CY, Chien, W.C., Chen, CY. (1999) Crystal growth kinetics of calcite in a dense fluidized- bed crystallizer. AIChE Journal 45(8), 1605-1614. Tai, CY. (1997) Crystallization kinetics revealed from experimental data analyzed by the two- step growth model. Journal of Chemical Engineering of Japan 30(3), 373-381. 108 Chapter 5 Assessing struvite precipitation in a pilot-scale fluidized bed crystallizer* 5.1 Introduction and background The recovery of phosphates from biological wastewater treatment plants, through struvite crystallization, offers an innovative and novel approach for the wastewater industry. It not only reduces unwanted struvite deposits in anaerobic digestion and post digestion processes and return lines, but also ensures environmental sustainability. To optimize the formation of struvite so as to minimize operational downtime, and offer the potential for cost effective recovery, the necessary conditions need to be established within wastewater treatment plants in a reactor dedicated to the purpose. Research suggests that the ideal location for the recovery of struvite requires that the flow should have a high concentration of soluble PO4-P and NH4-N, a low concentration of suspended solids and a relatively high phosphorus load (Williams, 1999). Since anaerobic digestion results in the formation of elevated levels of PO4-P and NH4-N, the most appropriate place for struvite formation and recovery is from supernatant/centrate of an anaerobic sludge (Gaterell et al, 2000). The only limiting factors then are Mg and pH levels, which can be corrected relatively easily by dosing with chemicals. Struvite usually precipitates as stable white orthorhombic crystals in a 1:1:1 molar ratio according to equation 5-1 (with n=0,l, and 2 as a function of pH) (Abbona and Boistele, 1979): 2+ + n 3 + Mg + NH4 + HnP04 " + 6H20 -> MgNH4P04 6H2Q + nH (5-1) A version of this chapter has been submitted for publication: Bhuiyan, M.I.H, Mavinic, D.S. (2007) Assessing struvite precipitation in a pilot-scale fluidized bed crystallizer. Environmental Technology, in review. 109 Predicting the crystallization of struvite is complex, as it is controlled by a combination of factors such as the thermodynamics of liquid-solid equilibrium, phenomena of mass transfer between solid and liquid phases, kinetics of reactions, and several physiochemical parameters (Corre et al, 2005). Supersaturation, pH, mixing energy, temperature, and presence of foreign ions are widely reported physiochemical parameters that affect struvite crystallization (Bouropoulos and Koutsoukos, 2000; Adnan et al., 2003). However, precipitation of struvite occurs when activities of magnesium, ammonium and phosphate exceeds the thermodynamic solubility product (Ksp) of struvite. As such, precipitation is probable when the supersaturation ratio (O), as defined by equation 5-2, is greater than 1. nJM^}{NH;}{PO>-} 2+ where, {Mg }, {NH/}, and {PO/~} are the activities of magnesium, ammonium and phosphate ions respectively. The determination of a thermodynamic solubility product of struvite has been described elsewhere (Bhuiyan et al, 2007a, Chapter 2). To avoid the complexities associated with the calculation of solubility products of struvite, a simple concept of conditional solubility has been used by several researchers (Adnan et al.,2003; Britton et al, 2005). The struvite conditional solubility product (Ps), is defined as the product of the analytical results for soluble magnesium, ammonia nitrogen, and ortho- phosphates: Ps — CT,M CT,mi Cr,P04 = (5-3) 2 Mg * NH,* P04 ' Mg ' NH4 ' P04 2+ + where, Y is the activity coefficient of ion i. The ionization fractions for Mg , NH4 , and PO/~ +2 + 3 can be defined as aMgl+ = [Mg J /CT,Mg , a + = [NH4 ] /CT,NH3 and a 3_ = [P04" ]/CT,po4 , 110 where CT,MG, CT,NH3, and CT,PO4 are the total analytical concentrations of magnesium, ammonia, and orthophosphates, respectively. Plotting conditional solubility product at equilibrium, Pseq, calculated from the right side of equation 5-3 versus pH establishes the struvite solubility limit curve for a particular ionic strength. The curve can be used to determine the struvite saturation condition of a process fluid by calculating Ps for that process fluid from analytical concentrations of the constituents (Ohlinger, 1998; Snoeyink and Jenkins, 1980). A simple approach was taken by the previous researchers to draw such curves, using the experimentally determined equilibrium total concentrations of the constituents in a particular solution for a range of pH values (Adnan et al, 2003; Britton et al, 2005). The main disadvantage of using the conditional solubility values, thus determined, for describing a struvite crystallization system is that comparison between studies becomes more difficult, since any changes in wastewater composition would change its inherent Pseq value. Any variation in water chemistry will result in difference in ionic strength, thus changing the precipitation potential of the wastewater (Adnan et al, 2003). In this study, a pilot-scale reactor, developed at The University of British Columbia (UBC), was demonstrated at the Lulu Island Wastewater Treatment Plant (LIWWTP), in Richmond, BC, Canada. The specific research objective of this part of the study was to predict the precipitation potential of struvite by calculating saturation index (SI = log Q) with the help of PHREEQC version 2.12 (Parkhurst and Appello, 1999), using the experimentally determined 13 36 thermodynamic solubility product Ksp = 10~ ' , and its temperature dependence (Bhuiyan et al, 2007a, chapter 2). The objectives also include the identification of the factors that effect struvite crystallization, with special attention to the use of magnesium, presence of calcium, and presence of filtered total organic carbon (TOC) and inorganic carbon (IC), especially CO32". During this 111 study, a careful observation was always made to use the knowledge of equilibrium and kinetics gained in other sections of the research (Bhuiyan et al., 2007a, Chapter 2; Bhuiyan et al, 2007b, Chapter 4). 5.2 Material and methods 5.2.1 Reactor design and operation Based on previous experience at a pilot scale reactor, designed and tested at the UBC Environmental Engineering Pilot Plant using a synthetic feed (Adnan et al. 2003), two scaled-up pilot-scale reactors were installed at the LIWWTP. One of the reactors was connected with a CO2 stripper, specially designed by the research group to investigate how CO2 stripping can effectively replace/reduce the use of caustic (NaOH), which was not considered in this study. The reactor for this study was operated for two months from mid-May to mid-July, 2006. Figure 5.1 shows the total crystallization system, including the reactor and the accessories. The reactor was designed as a fluidized bed with increasing diameter and a settling zone at the top. The diameter changes caused turbulent eddies above each transition, ensuring that sufficient mixing existed in the reactor, and also helped to classify the fluidized particles by size (Britton et al, 2005). Larger crystals accumulated at the bottom and were harvested periodically. Of the 90 L total volume, 79 L is utilized as liquid volume, as the top clarifier has some free-board. A hydraulic retention time (HRT) of 4.43 (±0.26) min was maintained during the operation. The reactor was equipped with two Oakton gel-filled epoxy body pH probes, one in the top of the bottom section above the harvest zone, and another in the external clarifier. The principal function of the external clarifier was to recycle the effluent back into the reactor. It also helped to trap the washed-out fine crystals, if any, from the reactor. The pH in the external 112 clarifier was monitored using an Oakton continuous pH monitor. The pH probe in the bottom section was connected to a pH controller having a digital monitor to maintain the desired pH in the reactor. The pH meters were regularly calibrated by a two point method, using standard buffer solutions of pH 7 and 10. 5.2.2 Chemicals, storage tanks and pumps The centrate was stored in two 5600 litre capacity holding tanks, connected to each other at the bottom. The tanks facilitated the removal of suspended solids (SS) to some extent by settling. The centrate was pumped from the holding tanks to the reactor using a Moyno™ Model 500 332 progressive cavity pump, with a V% HP motor, equipped with a digital speed controller. A 5.1 cm (2 inch) valve was used to drain off the settled sludge periodically. Magnesium was added in the form of MgCi2-6H20 solution from a tank, using a peristaltic pump.Throughout the operation, the target was to maintain a Mg:P molar ratio of about 1.3:1 within the reactor. The higher magnesium concentration caused the limiting reagent to be phosphate, and thus resulted in lower effluent phosphate concentrations (Britton et al., 2005). A Mg:P ratio of 1.3:1 was found to be effective in phosphorus (P) removal in the pilot- scale in the previous studies (Adnan et al., 2003; Britton et al., 2005). 5.2.3 Sampling and analysis Samples of the centrate, effluent, and magnesium feed were collected on a daily basis. Filtration was done immediately after sampling with a 0.45 pm filter and preserved with 5% H2S04 for N and P, and concentrated HN03 for metals. Conductivity and temperature were measured using an Oakton pH/CON 300 Deluxe Waterproof pH/Conductivity meter. Analyses for ortho-phosphate and ammonia were made, using the flow injection method on a LaChat 113 QuickChem 8000 instrument, as described in the method number 4500-P G and 4500-NH3 H of the Standard Methods for the Examination of Water and Wastewater (APHA et al, 1998). Magnesium and calcium analyses were performed by flame atomic absorption spectrophotometry, using a Varian Inc. SpectrAA220 Fast Sequential Atomic Absorption Spectrophotometer. TC and IC were analyzed by Shimadzu Total Carbon Analyzer TOC-500. Filtered TOC was calculated from their difference. Knowing pH of the solution, HCO3" and 2 2 CO3 " were calculated from IC (IC = H2C03* + HCO3" + CO3 " ) using equilibrium equations 5- 4 and 5-5 as described in method number 4500-CO2D in the Standard Methods for the Examination of water and Wastewater (APHA et al, 1998): + 636 H2C03* ^ HC03" + H ; K,= 10" @25°C (5-4) 2 + 10 33 HC03" <^C03 " + H ; K2= 10" ' @25°C (5-5) where, [H2C03* ] <=> [H2C03] + [C02 (aq) ] and the activity coefficients are assumed equal to unity. 5.2.4 Product Identification The pellets formed in the pilot-scale crystallizer were examined and identified by x-ray diffraction (XRD), using a Bruker D8 Advance X-ray diffractometer and CuKa radiation, with an average scanning rate of 2.0° 20 min"1. Analyses of the magnesium, ammonium and phosphate contents of weighed samples were done, after dissolution in 0.5% nitric acid solution. 5.3 Results and discussion The crystallization system was found to successfully recover phosphate as struvite as in the previous studies (Adnan et al, 2003; Britton et al, 2005). The diameter of the pellets formed in this study ranged from 0.5 to 3.5 mm. The crystals were found to be nearly pure struvite 114 (98.0+01.0%), which agrees with findings of a previous study with the same centrate and reactor system (Fattah, 2004). 5.3.1 Centrate characteristics during the study The characteristics of the centrate of LIWWTP, during the study, is reported in Table 5.1 The concentrations of the constituents in the centrate were found to be less stable in the month of May and July than in June (see Fig. 5.2). Supersaturation ratio (Q), of the centrate was more than 1 in some instances because of the variable magnesium concentrations; however, this did not result in any noticeable precipitation in the centrate tank. Rather, the suction side of the feed pump was clogged by struvite deposition quite a few times that reduced the number of effective days of operation. Similar clogging problems have been reported at treatment facilities elsewhere (Borgerding, 1972; Ohlinger et ai, 2000). Usually, acidic hot water was pumped to remove the clogging in the pump, except in one instance when the pump had to be opened and cleaned thoroughly. However, from the nucleation study, it was evident that a certain saturation level has to be attained to initiate precipitation, which again can be reduced to some extent by introducing mixing or seed by reducing the activation energy (Bhuiyan et al, 2007b, Chapter 4). The saturation level of the process fluid was increased by magnesium dosing and/or increasing the pH. 5.3.2 Reactor operation Table 5.2 shows the range of the operational conditions during the two-month operation. Continuous growth within the reactor, coupled with variable constituent concentrations, resulted in a shift of the steady sate condition inside the reactor. This was minimized, to some extent, by a periodic harvesting from the reactor and adjusting pH level, magnesium dosing and flow. The 115 reactor was operated in a pH range of 8.0-8.2, which yielded satisfactory results with that particular process fluid in a previous study (Fattah, 2004). 5.3.3 Performance of the crystallization process One of the main objectives of this research was to remove phosphate from the digester centrate and recover that phosphate as struvite. As such, the performance of the system was analyzed both from its phosphate removal efficiency, and struvite recovery efficiency. Since the system was monitored on a daily basis, its efficiency was also calculated for a 24 h operation. Within that period, the system was considered to reach a steady state. The phosphate removal efficiency (%) was calculated from the difference of the PO4-P concentrations in the influent and effluent of the fluidized bed reactor. Figure 5.3 shows influent and effluent concentrations of three constituents of struvite with error bars representing 95 % confidence interval. The paired t-test between the influent of and effluent concentrations of P04-P (p=0.000), NH4-N (p=0.001) and Mg (p=0.000) indicated that the removals were significant, with 40-95% P04-P removal. Since, phosphate was made limiting adding magnesium in excess, theoretically, 100% of the removed phosphate should be recovered as phosphate according to equation 5-5 1 Struvite mass, kg d" = Q.ACP04-MWS (5-5) where, Q = total flow in the reactor (1 nf1), ACpo4 = decrease in phosphate concentrations ( mol 1 1 f ), and MWS = Molecular weight of struvite (0.2454 kg mol" ). Based on mass of seed added, mass of total struvite harvested, and mass remaining in the reactor at the end, 83.0+2.0% of the 116 theoretical mass was recovered during operation. This difference may have occurred due to loss of some fines with the effluent, which was eventually deposited in the external clarifier. 5.3.4 Supersaturation level Saturation condition or supersaturation level is the key to struvite crystallization. It was measured, as defined in equation 5-2, and was calculated from SI determined by PHREEQC. The supersaturation level, that leads to struvite crystallization inside the fluidized bed reactor, was the Q inside the reactor for a particular operational pH. The saturation level, determined with respect to the thermodynamic solubility product (Bhuiyan et al, 2007a, Chapter 2), gives a precipitation potential of struvite which can be compared with another situation, regardless of pH's and ionic strengths. However, because of the wide variation of the composition of the centrate being fed, the saturation level changed within the reactor even for a particular pH. The operational parameters such as recycle ratio, magnesium dosing, and temperature also contributed to the change in the supersaturation levels inside the reactor. Recycle ratio was maintained within a range of 5.0-9.0 throughout the operation. Although, the value of Q inside the reactor appeared to be the most reliable control parameter for operating the reactor in a previous study (Adnan et al, 2003), the relationship between Q and P-removal (%) was found to be affected by several other variables in this study. Figure 5.4 shows that with an increasing pH and Q, the phosphate removal efficiency dropped at a pH of 8.1. The reason for this drop maybe attributed to the contribution of several operating parameters. For a system, performing with 100% efficiency, the effluent should reach equilibrium. Figure 5.5 shows that the effluent's supersaturation level seldom reached equilibrium during this study. This maybe partly due to the fact that effluent was collected through a down pipe to the external clarifier, where CO2 level 117 may have dropped due to some striping. This could be due to the inaccuracies in the measurement of the constituent concentrations, and pH. 5.3.5 Induction time and mixing The induction time, which is dependent on the supersaturation level, can be lowered by manipulating the pH in the system. In a previous study, it was found that the induction time for struvite precipitation from an anaerobic digester supernatant fell from days at pH 7 to less than 1 hour at pH 8.5 (Momberg and Oellermann, 1992). Induction time can also be lowered by introducing mixing or generating turbulence (Bhuiyan et ai, 2007b, Chapter 4; Ohlinger et al, 1999). The injection port of the reactor was designed in such way that it ensured high energy mixing at the inlet. Thus, a turbulent condition was maintained in each section of the reactor before the flow reached the top clarifier (see Table 5.3). Thereby, a favorable condition for crystal growth, which is a transport limited process (Bhuiyan et ai, 2007b, Chapter 4), was ensured. Although, the in-reactor supersaturation level was below the metastable limit of spontaneous homogeneous primary nucleation of struvite (Bhuiyan et al., 2007b, Chapter 4), the nature of the process fluid and the high energy mixing at the inlet resulted in heterogeneous nucleation and lowered the metastable limit. The induction time in the reactor was, thus, lowered to a level where nucleation was triggered immediately after the flow passed the injection port. This was observed by a pH drop due to the proton release as shown in equation 5-1. Once the pH dropped below the operational pH, the controller started to pump caustic. The nucleation and/or growth inside the reactor took place well within the HRT of 4.44 +0.26 min. This was also supported by the presence of a very little amount of fines that accumulated in the external clarifier. 118 5.3.6 Apparent upflow velocity Apparent upflow velocity, also called superficial velocity, calculated from the total flow and the cross sectional area of the reactor section, in the bottom section was found to be another parameter of importance. From Figure 5.6, it can be seen that for a very narrow window of the superficial velocity (400-410 cm min"1), 75-85 % phosphate removal was achieved. 5.3.7 Mg:P and N:P molar ratio Previous researches suggested that a stoichiometric excess of ammonium will help drive the reaction to form relatively pure struvite (Stratful et al. 2001), while excess magnesium decreases purity (Demeestere et al, 2001). Table 5.2 shows Mg:P and N:P ratio maintained inside the reactor during the study. Figure 5.7a shows that phosphate removal efficiency maintains a good correlation (r = 0.83, /?=0.000) with Mg:P ratio inside the reactor, and the efficiency does not change significantly after a Mg:P ratio around 2. The mean value of the inreactor concentrations of different constituents (Table 5.1) were used for calculation. PHREEQC model was used to determine the variation of SI with Mg:P ratio. Figure 5.7b shows that the SI values for measured data follow the curve of calculated model values. The scatter in the SI values from the measured data may be due to the effects of other variables and partly due to measurement errors. N:P ratio inside the reactor was also found to maintain a very good correlation (r = 0.90, p = 0.000) with phosphate removal efficiency (see Fig. 5.8a) for the pH range used. For the same mean in-reactor concentrations with a magnesium concentration of 25.5 mg l"1 (Mg:P=1.3:l), SI values were determined using PHREEQC for a range of N:P molar ratio. The model calculated curve in Figure 5.8b shows that the precipitation of struvite becomes probable for a N:P molar ratio of above 10:1. For the measured data, a decreasing trend in the SI 119 value can be noticed after an N:P molar ratio of 60:1. This may be due to the variation in the operational parameters during the operation. 5.3.8 Effect of Organic ligands The effect of organic ligands has been the subject of a few recent studies. The influence of organic ligands on the precipitation of struvite was investigated using a small molecular weight organic ligand, acetate, as it was found to be the only available short chain fatty acid during the operation. PHREEQC used the equilibria and thermodynamic parameters mentioned in Table 5.4. A range of acetate concentration 1-1000 mg F1 was used to determine its effect in struvite precipitation potential for the same in-reactor concentrations, temperature, and pH as mentioned above. Figure 5.9 shows that the SI value of struvite does not change below a concentration of acetate 103 mg F1. After that, SI value decreases slowly with an increase in acetate concentration. Thus, for the average in-reactor concentration of acetate (~ 20 mg F1) during operation, any affect in struvite precipitation potential, in terms of SI value, can be ignored. However, at neutral to alkaline pH, carboxyl groups present in most of the polysaccharides are available to interact with positively charged metal ions, with highest affinity for Mg2+ in most cases, thus affecting the growth by inhibition (Henry and Carole, 1990). Figure 5.10 shows that the phosphate removal efficiency was lower at pH 8.1, where filtered TOC level was higher. This was perhaps due to the inhibition effect on growth by organic ligands in phosphate removal. Inhibition of the crystal growth by blocking the active sites of the newly formed nuclei of hydroxylapatite by the organic ligands has been previously reported (Koutsopoulos & Dalas, 2000; Van Der Houwen, and Valsami-Jones, 2001). However, X-ray crystallography of the products from the reactor did not show any detectable impurity from struvite (see Fig. 5.11). A detailed study on the effects of different types of organic ligands 120 including fulvic and humic acids in this pilot-system, operated with anaerobic digester centrate, is recommended. 5.3.9 Influence of calcium and carbonate ions 2+ 2 It has been shown previously that the presence of Calcium (Ca ) and carbonate (C03 ~) ions can lengthen the induction time and negatively affect the growth rate (Bouropoulosand Koutsoukos, 2000). Presence of higher Ca2+ions in solution was found to have significant impact on struvite crystallization in terms of size, shape and purity of the product recovered in a previous study. At a Ca:Mg ratio of 1:2 and above, struvite crystals were found to be covered with an amorphous substance, probably a calcium phosphate compound (Corre et al, 2005). In this study, average in reactor Ca: Mg molar ratio was found to be 1:3.6, which did not affect in struvite crystallization. This was confirmed by the purity of struvite crystal by XRD method. A similar observation was made in a previous study for a molar ratio of Ca:Mg:PC>43" - 0.5:2:1, where the deposits formed at a pH of 7.8-9.2 had nearly a 1 to 1 ratios of N and P (Wang et al, 2005). However, to determine the effect of Ca2+ in struvite crystallization for a range of Ca:Mg ratio at pH =8.0, PHREEQC was used to calculate the precipitation potential of possible precipitates at 25 °C. The mean value of the inreactor concentrations of different constituents (Table 5.1) were used for calculation. pKsp values and the solubility equilibria used in calculations are shown in Table 5.5. As shown in Figure 5.12, the solution becomes supersaturated for dolomite and HAP for a very low Ca:P ratio. However, previous studies have suggested that precipitation of hydroxylapatite (HAP) occurs at pH value above 9.5, where effective struvite precipitation occurs at pH=8.0 and above (Wang et al, 2005). The range of supersaturation degree for calcium phosphate to precipitate in the presence of seeding material was found to be between 7 and 11 in a previous study (Nancollas and Tomazic, 1974). The 121 precipitation is mainly controlled by kinetic factors, and less by thermodynamic factors (Abbona et al. 1986). Hence, the solutions, although supersaturated with other calcium and magnesium phosphate, do not cause precipitation of those compounds along with struvite. 2 From Figure 5.13 it can be seen that the effluent C03 " concentration seems to be higher in the effluent, which in fact is not statistically significant (p=0.43). This apparent increase, may be, due to some CO2 release by stripping when the effluent falls in the external clarifier. HCO3" change was also not statistically significant (p=0.707). The presence of CO32" ion, however, can cause the solution supersaturated for some of the solid phases (marked with * in Fig. 5.14) for the same concentrations , temperature, Mg:P molar ratio 1.3, and an average in-reactor calcium concentration of 15 mg l"1, as calculated by PHREEQC using the pKsp values in Table 5.5. The precipitation in carbonate compounds were, perhaps, also controlled by the kinetic factors, rather than thermodynamic solubility alone. Among other possible precipitates associated with the struvite system (see Table 5.5), only bobierrite was found to be supersaturated, but above pH 8.5, which was beyond the operational pH range of the crystallizer. As such, the solid precipitates harvested were pure struvite with undetectable impurities. 5.3.10 Crystal morphology During the study, crystals of various sizes and shapes were harvested. The SEM analysis of the product shows that all the pellets are, in fact, aggregates of small crystals (Fig. 5.15a). Close examination of the images (Fig. 5.15b) reveals that the pellets were a fragmented aggregation of very fine and fused crystals. During the crystallization process, agglomeration, in presence of supersaturation or aggregation without supersaturation, was essentially observed. Agglomeration occurs through three successive steps: the collision of two or more particles; a sufficient period during which the particles stay together; and adherence of the particles with the 122 help of supersaturation. However, the edges of the pellets are bordered by aggregation of very fine crystals as a result of collision during the agglomeration process. The structural strength of the pellets seems to come from the tightly packed inner core and thicker outside coating of fine aggregates. Thus, the quality and the size and shape of the struvite pellets grown in UBC crystallizer are comparable with any such products. 5.4 Conclusions • Based on the results obtained from this study of struvite recovery from an anaerobic digester centrate, using the results of a thermodynamic study, the pilot-scale struvite recovery reactor developed at UBC was found to be effective in recovering phosphate as pure struvite. • The desired degree of phosphate removal was achieved by maintaining operating pH 8.0- 8.2, and recycle ratio 5-9, to control the supersaturation conditions inside the reactor. The performance of the system was found to be optimal when in-reactor supersaturation ratio was 2-6. • Among several other operating parameters, superficial velocity and magnesium to phosphate molar ratio were also found important to maintain system performance. In- reactor ammonium to phosphate molar ratio was also found to maintain a good correlation with phosphate removal. • Superficial velocity of 400-410 cm min"1 in the bottom section of the reactor was found to be effective in removing 75-85 % phosphate. Acetate ~ 100 mg 1"' did not affect the precipitation potential of struvite. However, filtered TOC levels inside the reactor affected the performance to some extent, perhaps by inhibiting the crystal growth. 123 Calcium and carbonate ion did not have any noticeable effect on struvite crystallization over the range studied. As the precipitation in calcium and carbonate compounds were also controlled by the kinetic factors, rather than thermodynamic solubility alone, the solid precipitates harvested were pure struvite with undetectable impurities. 124 Table 5.1 Centrate characteristics of the Lulu Island Wastewater Treatment Plant during the pilot-scale operation of the struvite crystallizer (n = 23). Parameters Mean Maximum Minimum SD pH 7.79 8.27 7.28 0.29 Temperature (°C) 27.3 31.3 23.4 2.1 Conductivity (uS cm"1) 6467 6860 5970 310 PO4-P (mg l"1) 76.1 106.8 59.1 12.1 NH4-N (mg l"1) 757.4 842.0 647.0 47.8 Mg (mg l"1) 12.3 21.4 2.0 4.4 Ca(mgl"') 19.1 28.2 10.3 5.9 Filtered TIC ( mg 1"' as C) 707.6 1417.4 285.3 328.6 Filtered TOC (mg l"1) 562.4 1340.6 37.4 325.3 125 Table 5.2 Operational conditions during pilot-scale operation of the struvite crystallizer at LIWWTP (n = 23). Parameters Mean Maximum Minimum SD Operational pH 8.0 8.2 8.2 0.09 Feed (N, P and Mg) flow, (lm"') 2.2 3.0 1.4 0.34 Magnesium flow (1 m"1) 0.07 0.08 0.05 0.01 Total flow (lm"1) 17.8 18.6 14.8 0.93 Recycle ratio 7.3 9.6 5.0 1.15 Up-flow velocity (cm min"1) 392.8 409.0 325.4 20.5 HRT (min.) 4.44 4.25 5.34 0.26 In-reactor SSR 6.4 13.5 2.7 2.65 In-reactor Mg:P molar ratio 2.76 13.05 0.45 3.47 In-reactor N:P molar ratio 71.73 140.2 33.96 28.75 In-reactor filtered TIC (mg l"1 as C) 640.9 1372.0 255.5 295.6 In-reactor filtered TOC (mg l"1) 520.5 1214.0 35.7 308.6 126 Table 5.3 Calculated Reynolds Number at three sections of the reactor. ID X-sectional Area Mean superficial velocity Reynolds Number 2 1 Section (cm) (cm ) (cm min" ) (Re*) Top clarifier 38.1 1140.1 16 1111 Top section 15.24 182.4 98 2778 Middle section 10.16 82.0 220 4167 Bottom section 7.62 45.5 392 5571 Reynolds Number, Re - DVp/ju where, D - ID of the pipe section, V = superficial velocity, and p=density of the fluid (997 kg m"3 @25°C), and u=Dynamic viscosity of the fluid (8.9xl0"4 Ns 127 5.4 Major equilibria involved among acetate and struvite constituents (Ball and Nordstrom, 1991) with corresponding solubility product constants at 25 °C. Equilibrium pKsp Acetate" + H+ = HAcetate -4.76 Mg2++ HAcetate = MgAcetate+ + H+ 3.48 2+ + Mg + 2HAcetate = Mg(Acetate)2 + 2H 7.47 NH3 + HAcetate = NH4Acetate -4.69 + NH3 + 2HAcetate = NH4(Acetate)2" + H 0.19 128 3 Table 5.5 pKsp Values and solubility equilibria of possible precipitates in Ca-Mg-PC>4 — CO32 system at 25°C. Minerals Reaction pKsp Ref. 2+ 2 Calcite CaC03<=> Ca + C03 ~ 8.42 a 2+ 2 Magnesite MgC03<=>Mg + C03 " 2.98 b Nesquehonite 2+ 2 5.19 a MgC03-3H2O^Mg + C03 " + 3H20 Dolomite 2+ 2+ 2 16.7 a CaMg(C03)2-3H20^ Ca + Mg + 2C03 " 2+ 3 HAP Ca5(P04)3(OH) « 5Ca + 3P04 ~ + OH~ 57.8 c Brushite 2+ 6.55 d CaHP04-2H2O^Ca + HP04" + 2H20 Struvite 2+ + 3 13.36 e MgNH4P04-6H20 «Mg + NH4 + P04 " + 6H20 Newberyite 2+ 3 5.8 f MgHP04-3H20 «Mg + HP04 " + 3H20 Bobieryite 2+ 3 25.2 f Mg3(P04)2-8H20 Bobieryite(am.) 2+ 3 23.1 f Mg3(P04)2-22H20 «3Mg + 2P04 " +22H20 aStumm and Morgan (1.981); bAppelo and Postma (1999); 'Ferguson and McCarty (1971); dBall and Nordstrom(1991) eBhuyian et al. (2007a) (Chapter 2); f Taylor et al. (1963) 129 Figure 5.1 Pilot-scale struvite crystallization system at LIWWTP. 120 100 A NH4-N _ 80 u> o - PQ4-P E |> 60 CL O 0. 40 300 200 20 100 0 0 26-May31-May 5-Jun 10-Jun 15-Jun 20-Jun 25-Jun 30-Jun 5-Jul 10-Jul Date Figure 5.2 Constituent concentrations in the centrate during the operation. 100 800 a Influent • Effluent 80 \ I 600 E 60 E c o ' 400 -B CO ro 1_ i_ -4—I -t—< c 40 c CD CD O o c c o o O r 200 o 20 I P04-P Mg NH4-N Figure 5.3 Influent and effluent concentrations of three constituents of struvite in the crystallizer (n = 23). Error bars: 95 % confidence interval. 132 G Figure 5.4 Variation in supersaturation ratio (Q) and phosphate removal efficiency at different pHs (n = 23). Error bars: 95% confidence interval. 133 15 o Inreactor A Effluent 10 C A A A A A A*A* A A A A A A / 0 23-May 2-Jun 12-Jun 22-Jun 2-Jul 12-Jul 22-Jul Date Figure 5.5 Variation in supersaturation ratio (Q) in influent and effluent with time. 1 Figure 5.6 Superficial Velocity in the bottom section of the pilot-scale struvite crystallizer vs. phosphate removal. 6 8 10 12 14 Mg:P (molar ratio) (b) 1.4 1.2 1 0.8 CO 0.6 0.4 o model 0.2 • measured 0 0 2 4 6 8 Mg:P (molar ratio) Figure 5.7 Mg:P molar ratio vs. (a) Phosphate removal (b) model calculated Saturation Index (SI). 136 (a) Figure 5.8 N:P molar ratio vs. (a)Phosphate removal (b) model calculated Saturation Index (SI) for struvite in the crystallizer. 137 0.65 0.6 y = -8E-05x + 0.6026 R2 = 0.98 0.55 CO 0.5 0.45 0.4 0 200 400 600 800 1000 1200 Acetate (mg I"1) Figure 5.9 Model calculated saturation index (SI) value of struvite vs. acetate concentration in the struvite crystallizer. 138 100 1400 1200 80 •{ 1000 co > 60 800 - E CD 0 E CD •4—' o CD 600 JZ o CL 40 w o 400 20 200 0 0 Figure 5.10 pH vs. phosphate removal in the crystallizer and inreactor filtered TOC (n=23). Error bars: 95% confidence interval. 139 Figure 5.11 Identification of the crystal pellets as struvite using powder X-ray diffraction. X-ray diffraction pattern matches very well with the pattern of struvite standard (•). 140 12 o Dolomite A Calcite 10 • Struvite • HAP A Brushite c/) 4 pDD • O D 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Ca:Mg (molar ratio) Figure 5.12 Ca:P molar ratio vs. model calculated Saturation Index (SI) values of the possible precipitates. 141 • Influent • Effluent HCO3- co3< _ Figure 5.13 Influent and effluent HCO3" and C03 concentrations. 15 -o- Bobierrite* -•- Cattiite —+— Brushite Calcite* 10 -Magnesite Nesquehonite & Newberyite —o— Struvite* -«- Dolomite* •HAP* (0 -5 -10 9.5 10 Figure 5.14 pH vs. model calculated Saturation Index (SI) values of the possible precipitates that could interfere with struvite formation in a crystallizer (Solid phases marked by *). 143 (a) References Abbona, F. and Boistelle, R. (1979). Growth morphology and crystal habit of struvite crystals (MgNH4P04 6H20). Journal of Crystal Growth 46, 339-354. Abbona, F., Madsen, H.E.L., and Boistelle, R. (1986) The initial phases of calcium and magnesium phosphates precipitated from the solutions of high to medium concentrations. Journal of Crystal Growth 74, 581-590. Adnan, A., Koch, F.A. and Mavinic, D.S., (2003) Pilot-scale study of phosphorus recovery through struvite crystallization- II. Applying in-reactor supersatuartion ratio as a process control parameter. Journal of Environmental Engineering and Science 2, 473-483. American Public Health Association (APHA), American Water Works Association, and Water Pollution Control Federation, (1998) 20th edn. Standard Methods for Examination of Water and Wastewater, Washington , D.C. Appelo, C.A.J, and Postma, D. (1999) Geochemistry, Groundwater and Pollution, A.A. Balkema, Rotterdam, The Netherlands. Ball, J.W. & Nordstrom, D.K. (1981) User's Manual for WATEQ4, with revised thermodynamic database and test cases for calculating speciation of major, trace and redox elements in natural waters. U.S. Geological Survey Open File Report 91-183, Menlo Park, CA. Bhuiyan, M.I.H, Mavinic, D.S., and Beckie, R.D. (2007a) A solubility and thermodynamic study of struvite. Environmental Technology 28, 1015-1026. Bhuiyan, M.I.H, Mavinic, D.S., and Beckie, R.D. (2007b) Nucleation and growth kinetics of struvite in a fluidized bed reactor. Journal of Crystal Growth (submitted). Borgerding, J. (1972) Phosphate deposits in digestion systems. Journal of Water Pollution Control Federation, 44(5), 813-819. Bouropoulos, N.C. and Koutsoukos, P.G. (2000) Spontaneous precipitation of struvite from aqueous solutions. Journal of Crystal Growth 213, 381-388. Britton, A., Koch, F.A., Mavinic, D.S., Adnan, A. Oldham, W.K., and Udala, B. (2005) Pilot- scale struvite recovery from anaerobic digester supernatant at an enhanced biological phosphorus removal wastewater treatment plant. Journal of Environmental Engineering and Science 4, 265- 277. Corre, K.S.L., Valsami-Jones, E., Hobbs, P., Parsons, S.A. (2005) Impact of calcium on struvite crystal size, shape and purity. Journal of Crystal Growth 283, 514-522. 145 Demeestere, K., Smet, E., Van Langenhove, H., and Galbacs, Z. (2001) Optimization of magnesium ammonium phosphate precipitation and its applicability to the removal of ammonia. Environmental Technology 22 (12), 14-1428. Fattah, K.P. (2004) Pilot Scale Struvite Recovery Potential from Centrate at Lulu Island Wastewater Treatment Plant. Masters thesis, Department of Civil Engineering, The University of British Columbia, Canada. Ferguson, J.F., and McCarty, P. (1971) Effects of carbonate and magnesium on calcium phosphate precipitation. Environmental Science & Technology 5(6), 534-540. Gaterell, M.R., Gay, R., Wilson, R., Gochin, R.J. and Lester, J.N. (2000) An economic and environmental evaluation of the opportunities for substituting phosphorus recovered from wastewater treatment works in existing UK fertilizer markets. Environmental Technology 21,1067-184. Henry, L.E., Carole, L.B. (1990) Microbial Mineral Recovery, McGraw-Hill Publishing company, NY, USA. Koutsopoulos, S., and Dalas, E. (2000) The crystallization of hydroxyapatite in the presence of lysine. Journal of Colloid Interface Science 231, 207-221. Momberg, G.A. and Oellermann, R.A. (1992) The removal of phosphate by hydroxyapatite and struvite crystallization in South Africa. Water Science and Technology 26, 987-996. Nacollas, G.H. and Tomazic, B. 1974. Growth of calcium phosphate on hydroxyapatite crystals. Effect of supersaturation and ionic medium. The Journal of Physical Chemistry 78 (22), 2218- 2225. Ohlinger, K. N, Young, T.M., and Schroeder, E.D. (1998) Predicting struvite formation in Digestion. Water Research 32, 3607-3614. Ohlinger, K., Young, T.M. and Schroeder, E.D. (1999) Kinetics effects on preferential struvite accumulation in wastewater. Journal of Environmental Engineering 125(8), 730-737. Ohlinger, K.N, Young, T.M., and Schroeder, E.D. (2000) Postdigestion struvite precipitation using a fluidized bed reactor. Journal of Environmental Engineering, 126(4), 361-368. Parkhurst, D.L., Appelo, C.A.J. (1999) User's guide to PHREEQC (Version 2)- a computer program for speciation, reaction- path, advective-transport, and inverse geochemical calculations. USGS water-Resources Investigation Report 99-4259. Snoeyink, V. and Jenkins, D. (1980) Water Chemistry. John Wiley & Sons, New York, pp.306- 309, 449. 146 Stratful, I., Scrimshaw, M.D., and Lester, J.N. (2001) Conditions influencing precipitation of magnesium ammonium phosphate. Water Research 35 (17), 4191-4199. Stumm, W. and Morgan, J. (1981) Aquatic Chemistry. Wiley-Interscience, New York, pp.241- 247. Taylor, A.W., Frazier, A.W., Gurney, E.L., and Smith, J.P. (1963) Solubility products of di- and trimagnesium phosphates and the dissociation of magnesium phosphate solutions. Transactions of the Faraday Society 59 1580 - 1584. Van Der Houwen, J.A.M., and Valsami-Jones, E. (2001) The application of calcium phosphate precipitation chemistry to phosphorus recovery: the influence of organic ligands. Environmental Technology 22, 1325-1335. Wang, J. Burken, J.G., Zhang, X., and Surampalli, R. (2005) Engineered struvite precipitation : impacts of component-ion molar ratios. Journal of Environmental Engineering 131(10), 1433- 1440. Williams, S., (1999) Struvite precipitation in the sludge stream at Slough wastewater treatment plant and opportunities for phosphorus recovery. Environmental Technology 20, 743-747. 147 Chapter 6 Dissolution kinetics of struvite grown in a pilot-scale crystallizer 6.1 Introduction Although struvite (MgNH4P04 6H20) can cause problems in wastewater treatment plants by clogging pipes and equipements, it has potential use as a fertilizer once it is intentionally precipitated. This realizes the benefit of reducing phosphorus and nitrogen load of side stream and sludge liquors recirculated to the head of the wastewater treatment works (Doyle and Simon, 2002; Bridger et ai, 1962). Struvite has been studied from different points of view and its chemical and physical properties have been clearly described (Babic-Ivancic et al, 2002). The dissolution and precipitation of struvite occurs simultaneously in an aqueous system, as the chemical reaction approaches equilibrium. Further knowledge on the kinetics and mechanisms of dissolution of struvite, would help optimize the intentional formation of struvite as well as its efficacy, when used as a fertilizer. It is important for a fertilizer to supply plants with nutrients efficiently and to maintain equilibrium of nutrients in crops. Slow or controlled release of nutrients can effectively achieve that goal, as well as minimize the loss of nutrients from the fertilizers and reduce environmental pollution (Guo et al., 2005). Previous research suggests that struvite can be used as slow-release fertilizer at high application rates, without the danger of damaging plant roots (Bridger et al, 1962). Suggested fertilizer applications for struvite are diverse and include ornamentals, forest outplantings, turf, orchards and potted plants. Granular forms of struvite are *A version of this chapter has been submitted for publication: Bhuiyan, M.I.H, Mavinic, D.S., and Beckie, R.D. (2007) Dissolution kinetics of struvite grown in a pilot-scale crystallizer. Journal of Environmental Engineering and Science, in review. 148 one of the most suitable slow release fertilizers, as the rate of availaibility can be controlled by granulation (Gaterell et al, 2000; Bridger et al, 1962). Research has also been undertaken with briquettes of magnesium ammonium phosphate, to investigate its slow release property in a small stream in British Columbia (BC), Canada, as a nutrient replacement source for decaying salmon carcasses (Sterling, 1997). Compared to other fertilizers, the benefits of using struvite are low leaching rates and prolonged release of nutrients, with the possibility of only one single application (Gaterell et al, 2000). When crystals dissolve in an aqueous medium, two consecutive reactions take place, a surface process and a transport process (diffusion or convective diffusion). The surface process involves transfer of substance from the crystalline phase to the solution adjacent to the crystal. This region is, for simplicity, called the "interface region". The transport process involves the transfer of dissolved substance from the interface region to the bulk solution. The rate may be controlled by either of these two processes or in combination (Christoffersen and Christoffersen, 1984). The dissolution of sparingly soluble minerals is generally controlled by a surface reaction, whereas the dissolution of soluble minerals is predominantly controlled by transport processes. Thus, the dissolution mechanism of minerals, that occurs whenever undersaturation of the bulk solution is present, can be predicted to some extent, from its solubility. Caution is warranted, since both the presence of inhibitors, the composition of the solution and other factors may affect the dissolution mechanism (Appelo and Postma, 1999). Dissolution may occur as a congruent or as an in-congruent reaction. In a congruent dissolution, a mineral dissolves into its constituent ions, while in an incongruent dissolution, secondary minerals form with the dissolution (Berner and Bemer, 1987). The equilibrium 149 relationship between struvite and its constituents can be described by equation 6-1 (Ohlinger et al, 1998). 2+ + 3 MgNH4P04 6H20<» Mg + NH4 + P04 " + 6H20 (6-1) The transformation of struvite to other polymorphic phases under different conditions has also been investigated (Bhuiyan et al 2007b, Chapter 7). A previous study was conducted in a batch reactor to determine the dissolution kinetics of synthetic struvite at three different temperatures (Babic-Ivancic et al, 2002). Some research has been undertaken, using batch reactors and mixed flow-through reactors, to study the dissolution kinetics of minerals (Golubev et al, 2006; Kraemer et al, 1998). This particular study was conducted at 25°C, both in batch reactors and mixed flow-through reactors, to gain a better understanding of the dissolution of struvite formed in a pilot-scale fluidized bed reactor, using real anaerobic digester centrate. 6.2 Materials and Methods 6.2.1 Identification Struvite pellets used in these experiments were formed in a pilot-scale, fluidized bed reactor, operating at the Lulu Island Wastewater Treatment Plant, Richmond, British Columbia, Canada (Bhuyian et al, 2007a). The composition of the pellets was examined and identified as pure struvite by x-ray diffraction (XRD), using a Bruker D8 Advance X-ray diffractometer and CuKa radiation, with an average scanning rate of 2.0° 29 min"1 . Identification was also made by the room-temperature infrared (IR) spectra in the wave number range of 400 to 4000 cm"1, on a Perkin Elmer 1710 Infrared Fourier Transform Spectrometer. Analysis of the magnesium, ammonium and phosphate contents of weighed samples, after dissolution in hydrochloric acid, indicated an average purity of 99.7 percent. 150 6.2.2 Batch reactor system Dissolution experiments in the batch reactor system were performed in a temperature - controlled room. A known amount of struvite pellets (0.5 - 2.0 mm) was suspended in a 1 litre glass jar, filled with 1100 ml deionized water. System pH was adjusted slowly, using HC1 or NH4OH to the target value. Immediately after suspending the struvite pellets, the jar was tightly closed with a rubber stopper, leaving no space for air. The experiments were carried out at 25 C. During the experiments, the system was continuously stirred at a constant rate by means of a Teflon-coated magnetic stirring bar. System pH was measured using an Orion 420A bench top pH meter, equipped with a VWR Symphony temperature compensated probe, calibrated using pH 4, 7 and 10 buffers. The pH probe was inserted through the rubber stopper, and was sealed to avoid any leakage. An aliquot of sample was collected using a syringe at 5 minute intervals, for 1 hour. The experiment was carried out for suspensions of 1000, 500, 200, and 100 mg of struvite pellets at three different pH values, namely 4.38, 5.05 and 6.05. These pH values were arbitrarily chosen within the acidic range. To see the dissolution pattern for a longer period of time, a similar set-up was also used to carry out the experiment for 5 days with an initial pH of 5.15 and 1000 mg of struvite pellets. Electrical conductivity (EC) and temperature were measured using an Oaktan pH/CON 300 Deluxe Waterproof pH/Conductivity meter. 6.2.3 Treatment of data By subtracting the amount of pellets dissolved in the solvent from the initial amount of pellets added, the amount still undissolved was determined. To simplify this procedure, the number and the shape of the pellets were assumed to be constant during the dissolution kinetics experiments. In determining dissolution rates, the rate of change of concentration, dC/dt, was determined by numerical differentiation of the concentration of struvite dissolved, as a function 151 of time. The driving force for the dissolution process, given as an absolute undersaturation, Cs - C, can be calculated by subtracting the concentration of the struvite pellets dissolved from the solubility or equilibrium concentration, Cs, The geometric mean of the concentrations of three struvite constituents was used as the dissolved concentration of the struvite pellets, C. Struvite pellet dissolution, under the given experimental conditions, was found to follow the following kinetic equation (equation 6-2) (Babic-Ivancic et al, 2002): ^ = kd.-(Cs-C) (6-2) dt y y ' where, A, is the surface area of the suspended struvite pellets, V is the volume of solution, kd is the dissolution rate constant. A can be expressed as a function of the number of crystals, N, and the mass of struvite in the crystal phase suspended in the solution, ns (equation 6-5) using equations 6-3 and 6-4: 2 A=NfAl (6-3) 3 Vmns=Nfv.l (6-4) 2/3 fV n ^ U2 A = N f/ III s (6-5) where, fA and/v are the dimensionless numerical surface and volume factors and dependent on the shape; Vm is the molar volume with units of volume per mole, and / is the characteristic linear dimension of the suspended crystals. Equation 6-2 can be further expressed as: at V r \2/3 1 m kdiss k where, kdiss=k(lN l and and d are the dissolution rate constants (cm s" ). 152 Following Babic-Ivancic et al. (2002), the dissolution rate, dC/dt, calculated from the 2/3 n s concentration versus time dependence, was plotted as a function of (Cs -C) (see equation 6-6). The slope of the line represented the dissolution rate constant, kcuss. of the struvite pellets. 6.2.4 Mixed flow-through reactor system Mixed flow-through reactors are advantageous, compared to batch reactors, because rates can be measured directly; this eliminates the need to differentiate the concentration versus time data produced by the batch reactors (Weissbart and Rimstidt, 2000). A 500 ml mixed flow- through reaction vessel was used in a temperature controlled room (25 C) (see Fig. 6.1), to obtain the steady-state dissolution rate. 500 mg of struvite pellets (0.5 - 2.0 mm) of known specific surface area (23.87 m2 g"1), measured by the B.E.T method, were put into the reaction vessel. The solution inside the vessel was continuously stirred with floating Teflon - coated magnetic stirrer controlled by a stirplate placed beneath the vessel. The input solution was stored in a polyethylene container protected against C02 uptake from the atmosphere. The chemistry and pH of the input solution varied with each experiment. The solution was injected into the reaction vessel using a peristaltic pump at a flow rate of 1.5 ml min"1. The reactive solution exited the reactor at the same flow rate as the inflow, maintained by the same pump with two heads. This was confirmed by the steady level of the solution in the reaction vessel. The outflow accumulated in the syringe, where the pH probe was inserted upto the bottom. When steady state was attained, the pH inside the reactor was identical to that of the effluent. The pH was recorded continuously by a VWR Symphony temperature compensated probe, connected to a computer through a data logger. pH probe tip was located where the solution entered the syringe, to avoid any effect of atmospheric CO2 (Fig. 6.1). The pH probe was periodically calibrated using pH 4, 7 153 and 10 buffers. The reaction vessel was sealed with a rubber stopper, having sealed inflow and outflow tubes passing through it. After a steady-state condition was reached, the dissolution rate was evaluated from the total ammonia (CNH4), total ortho-phosphates (CPOJ), and total magnesium (CMG) concentrations of the output solutions. 6.2.5 Treatment of data The accumulation of dissolved constituent ions within the reactor is equal to input minus output, plus the contribution from the dissolution. Thus, the non-steady-state material balance equation for the mixed flow-through reactor can be written as: fd\CX ^ V lout = q[C] -q[C} +VR (6-7) dt i il> i mt diss \ Ul, J 1 where, [C,],., and [C,.]0H, are the input and output concentrations of the ion of interest (mol l" ) respectively; Vis the volume of the reaction solution (1 ); q is the flow through the reactor (m3 1 1 1 h" ); Rdiss is the dissolution rate (mol f' h" ). Since [C(.]/(1~ 0 mol l" (i.e. there is no detectable constituent ion in the input solution), equation 6-7 can be rewritten as: lout V = R*.-fa}.« • (6-8) Udtl V J Solving the differential equation in equation 6-8: [C,L,=r^(l-e^) (6-9) where, r = hydraulic retention time = V/q At infinite time, normalized to surface area, the familiar equation for steady state dissolution within a flow-through reactor is (Samson et al., 2000): R = g[c,]au VA 154 where, is the dissolution rate (mol m"2 h"1) and A is the mineral surface area per unit solution volume (m2 l"1). 6.2.6 Analysis Analyses for ortho-phosphate and ammonia were made, using the flow injection method on a LaChat QuickChem 8000 instrument, as described in the method number 4500-P G and 4500-NH3 H of the Standard Methods for the Examination of Water and Wastewater (APHA et al., 1998). Magnesium was analyzed by flame atomic absorption spectrophotometry, using a Varian Inc. SpectrAA220 Fast Sequential Atomic Absorption Spectrophotometer. The specific surface area of the struvite pellets was determined by the B.E.T method, using nitrogen in an Autosorb 1-MP by Quantachrome. Before dissolution experiments, solids were examined and identified as pure struvite by x-ray diffraction (XRD), using a Bruker D8 Advance X-ray diffractometer and CuKa radiation, with an average scanning rate of 2.0° 20 min"1. The same method was adopted to determine purity of the remaining solids after each dissolution experiment. 6.3 Results and discussion 6.3.1 Batch reactor system Struvite dissolution was found to be congruent in the experimental conditions maintained in this study. Remaining solids, after each dissolution experiment, were identified as pure struvite. However, to evaluate the dissolution rate constants of struvite pellets, the geometric mean (C) of the total concentrations of three different constituents orthophosphate (Cpo4), ammonia (CNH4), and magnesium (Cjvig) was used. During the 1 hour experiments, equilibrium was not reached for any case. It can be observed from Figure 6.2a-6.2c, as an example for 155 pH=5.05, that there is consistently higher dissolution for larger initial masses of struvite pellets. Figure 6.2d, drawn for 1000 mg and different pH values, shows that dissolution increased with decreasing pH. Figure 6.3 shows that the percentage of initial struvite dissolved, is higher for a lower initial amount, during the 1 hour period. The percent of struvite dissolved does not vary markedly for the higher (500 and 1000 mg) reactor loadings. Table 6.1 shows for each initial pH the equilibrium pH, and solubility of struvite (Cs), as calculated using PHREEQC on the basis of the solubility product value, .£^=13.36 (Bhuiyan et al, 2007c, chapter 2). The Ksp value of struvite is referred to as the ion activity product (IAP) of 2+ + 3 Mg , NH4 , and P04 ", when solution is in equilibrium with the solid phase. The experiment was also carried out with an initial pH value of 5.15 and 1000 mg of struvite pellets, for 5 days. It was observed that equilibrium was not reached before 24 h (see Fig. 6.4). From the slopes of the curves, it can be noticed that the rate of dissolution was more or less the same until it reached equilibrium. The dashed line indicates the PHREEQC calculated solubility. Figure 6.5 shows best fit curves of the concentrations of struvite, with time, for an initial amount of 1000 mg, for different initial pH values. Similar curves were drawn for other amounts, but not presented in this paper. The curves fit well the experimental data points, except in the case where the initial amount (100 mg) was less than the solubility of struvite. However, the dissolution rate constant was also calculated for this amount. To determine the mechanism for the dissolution process, different theoretical models were tested. The plots, according to equation 6-2, showed straight lines. Figure 6.6 shows an example for the 500 mg initial amount of struvite pellets and a pH value of 4.38. The Rvalues and the corresponding R2 values of the regression, are shown in Table 6.2. As noted, the general trend, within the pH range tested (4.83-6.05), is that the dissolution rate constant kd decreases as 156 the pH increases, and increases as the initial amount decreases. The value of A, used in equation 6-2, was taken to be initial surface area of the struvite pellets dissolved. Since/4 was unknown, it was not possible to evaluate the change of A during the dissolution process. However, in a simpler form, the rate constant can be defined as kd' = kdA/V(Mullin, 2001). The values, as provided in Table 6.2, were also determined by substituting the relation in equation 6-2. The trend of values, thus determined, is identical to that of the rate constants, kd. Equation 6-6, which is independent of A, was also tested to determine the values of dissolution rate constant, kdiss. The dissolution rates, dC/dt, calculated from the concentration versus time dependence, were plotted as a function of the product of undersaturation, Cs - C, and 2/s the mass of struvite present in suspension at the given time, ns . The value of ns was determined from the difference of initial mass and dissolved mass of struvite. Figure 6-7 shows an example of such a plot, for an initial mass of 1000 mg, and pH 4.38. The plots were straight lines and fit well with the experimental values. From the slopes of these straight lines, the values of the rate constant, kdiss, for the dissolution processes in the batch reactor system were calculated. The results are also given in Table 6.2, with the R2 values of the regressions. The fit of the experimental curves to equation 6-6 demonstrates that the kinetics of struvite pellet dissolution can be described by this equation. In all three cases, a low R2 value for the initial amount below solubility, suggests that the initial amount should be higher than the solubility to use this model. The higher the initial amount, the better was the fit of the model to the data, as can be seen from the R2 values. The dissolution rate constants were found to decrease with an increase in pH. They also increase as the amount of initial struvite pellets decreases. During dissolution of a crystal, the rate is expected to be influenced by the amount of crystals present, the change in their shape during dissolution, and the composition of the solution (Appelo and Postma, 1999, pp. 73-74). 157 However, the dissolution process mechanism cannot be determined only by means of a kinetic model; the rate determining mechanism is also important. Previous researchers calculated an activation energy from the dissolution rates of struvite crystals, at three different temperatures, and concluded that the most probable rate determining mechanism is the process of desorption of the integrating ions (Babic- Ivancic et ai, 2002). The dissolution of sparingly soluble minerals is generally controlled by surface reaction, whereas the dissolution of soluble minerals is predominantly controlled by transport processes (Appelo and Postma, 1997, p. 76). However, the possibility of changing the rate control mechanism from one to another, with a change in concentration, has also been viewed as a realistic explanation (Babic- Ivancic et al, 2002). 6.3.2 Mixed flow-through reactor system At equilibrium, the rate of dissolution and the rate of precipitation are equal. Dissolution predominantly takes place if the system is far from equilibrium conditions. The experiments to determine dissolution rates of struvite pellets in mixed flow-through conditions were carried out far from equilibrium conditions. This was ensured by calculating the saturation index (SI) of the reaction solutions at steady-state, with respect to struvite, in each case. In all the cases, SI was « 0 (SI=0, for equilibrium). Steady-state dissolution rates were calculated from the concentrations of the constituent ions in the output solutions. The overall dissolution rates were derived from the geometric mean ( C ) of the constituent concentrations of the output solutions. At a flow rate of 3x10" min" (normalized by reactor volume), steady-sate was found to be attained at around 1500 min, while all of the experiments were carried out for 2880 minute duration. This demonstrated that the times necessary to reach mechanical (~ four fluid residence times) and chemical steady- states were very close (Pokrovsky and Schott, 2000). Figure 6.8 shows examples of constituent dissolutions, with time, for input solutions with pH 6.05 and 8.55. It can be observed from the 158 curves that the release of ammonia is higher than other two constituents in the case of lower pH ( stronger (-1:1:1) with increasing pH of the input solution. This was also verified by paired-t tests among the constituent concentrations. Table 6.3 shows the experimental conditions and corresponding calculated rates of dissolution, with respect to the constituent ions and their geometric mean (C), using equation 6-10. Due to proton promoted dissolution, the rate of dissolution of minerals increases with increasing hydrogen ion concentrations in the acidic region. Above some transition pH, dissolution rates are generally independent of pH. In the alkaline region, at pH values above 8, the rate increases with increasing pH, due to hydroxyl-promoted dissolution. Experimental dissolution rates of several minerals have been reported by different researchers that followed this pattern (Appelo and Postma, 1999, p. 222,; Golubev et ai, 2006; Kraemer et ai, 1998). Figure 6-9 shows that, for struvite pellets, transition seems to occur at a pH of 7.5 and the rate of dissolution remains independent of pH until around 8. This can be further verified in a future study with more observations in that range. However, the rate of dissolution was found to increase with the increase in pH after a value of 8. At low pH, phosphate species and their + 2 complexes such as HPO4 ", H2P(V , MgHP04°, MgH2P04 are forming. At high pH, complexes such as MgOH form. These complexes may reduce the solubility product and promote dissolution. However, ligand (inorganic and organic) promoted dissolution has also been described in previous studies (Kraemer et al, 1998), which was beyond the scope of this research. Laboratory determined mineral dissolution rates need to be normalized to allow their extrapolation to systems exposed to the natural environment. The principal normalization terms 159 used in the literature are mass, and geometric and B.E.T specific surface area (SSA). In a previous study, with biotite minerals, the normalizing term which gave the least variation in dissolution rates between the grain sizes, was the initial B.E.T SSA (Hodson, 2006). In this study, the dissolution rates were normalized to the initial B.E.T SSA. However, extrapolation of laboratory-derived dissolution rates to field dissolution rates show deviations. Possible explanations for the deviations are: (i) difference in the dissolution mechanism in the laboratory and field; (ii) lack of steady-state attainment; (iii) differences in solution chemistry; iv) improper estimation of the reactive surface area; (v) differences or variability in pH and temperature; (vi) effects of the biological parameters; (v) hydrological parameters (Brantley, 2003). In order to determine the effects of solution chemistry in the overall trend of rate of dissolution of struvite pellets, water samples were collected from three different regions of the province of British Columbia (BC), Canada. Due to the similar geological provenances, the pH of the waters were found to not vary significantly between the rivers or creeks located within a short distance. As such, to carry out the experiments, waters were collected for use as the input solutions from three different salmon bearing river(s)/creek(s) in BC, namely the Mamquam river in Squamish, BC, S-l (EC =90 pS cm"'); Deer Creek, Pavilion Lake in Kootenay, BC, S-2 (EC = 100 pS cm"1); and the Nicola River near Merritt, BC, S-3 (EC = 245 uS cm"1). The dissolution rates found from these experiments are also shown in Figure 6.9. The results indicate that the dissolution rates were most strongly controlled by the initial pH and were similar to the experimental trends observed with laboratory solutions described earlier. 6.4 Conclusions • The dissolution of struvite occurs whenever the bulk solution is undersaturated. To determine the mechanism of the dissolution processes in a batch reactor system, two 160 different theoretical models were tested. The experimental values were found to fit well with both models. As the surface area changes with dissolution, the model which is independent of the surface area was found to be more easily applicable to determine the dissolution kinetics. The dissolution of struvite pellets was found to decrease with increasing pH within the pH range tested (4.38-6.05). In general, the dissolution rate constants (Kd, kd', and KdiSS) were found to be lower for higher pH values. They were found to decrease with an increase in initial mass of struvite per volume of solution for any tested pH value. When the initial mass was less than the solubility at that particular pH, the results were not consistent, since the main driving force for dissolution is (Cs - C). In a mixed flow-through reactor system, the constituents release became stoichiometric in case of alkaline pH of the input solution. The dissolution rates for struvite pellets were found to increase with the hydrogen ion concentration, due to proton-promoted dissolution. The transition occurs at a pH around 7.5, and the rate of dissolution remains independent of pH until around 8. After that, the rate of dissolution increases with an increase in pH because of the hydroxyl- promoted dissolution. As such, supersaturation is more important than pH alone during intentional struvite formation. When used as a fertilizer, not only acidic pH will cause dissolution when the system is far from equilibrium. 161 Table 6.1 Solubility of struvite (Cs) and equilibrium pH calculated by PHREEQC for different initial pH values. Initial pH Cs Equilibrium pH (mol l"1) (mgl1) 4.38 5.596 x 10"4 137.32 9.77 5.05 5.545 x 10"4 136.07 9.84 5.51 5.542 x 10"4 136.00 9.85 6.05 5.535 x 10"4 135.83 9.86 Table 6.2 Dissolution rate constants and corresponding R2 values of the regressions for different amount of struvite pellets and pH values. 8 2 2 4 2 2 pH Initial kd xlO" R R kdiss x 10 kd' xlO" R amount of struvite pellets (mg) (cm s"1) (cm s"1) (mol"2'3 1 s1) 4.38 1000 1.47 0.99 3.0 0.99 0.77 0.98 500 6.94 0.94 8.0 0.94 1.81 0.94 200 30.1 0.71 13.1 0.71 3.11 0.71 100 47.4 0.1.3 10.29 0.13 2.39 0.20 5.05 1000 0.04 0.97 0.08 0.97 0.02 0.97 500 3.93 0.88 4.0 0.88 1.02 0.88 200 27.4 0.88 10.0 0.88 2.84 0.87 100 38.5 0.62 8.0 0.62 1.98 0.60 6.05 1000 0.05 0.97 0.1 0.97 0.03 0.97 500 3.51 0.90 4.0 0.90 0.91 0.90 200 40.2 0.81 17.5 0.81 4.16 0.80 100 84.9 0.38 18.4 0.38 4.31 0.37 163 Table 6.3 Experimental conditions and corresponding dissolution rates of struvite pellets at 25°C. pH Steady-state lOg R 164 Peristaltic pump" (with two heads) pH probe ~ (Inserted in a piston-less plastic syringe) Input Solution 20 [im filter .Reactor solution (500 ml Struvite pellets (0.5-2 mm) Floating stirring b ar Stirplate Figure 6.1 Mixed flow-through reactor set-up. (a) (b) 0.80 - 0.90 - -O-1000 mg -o-1000 mg 0.70 - 0.80 - -a— 500 mg -a— 500 mg 0.70 - 0.60 -" -a- 200 mg -ft- 200 mg . -•-100 mg 0.60 -" -•- 100 mg o 0.50 - o e 0.50 - o 0.40 - o X x 0.40 - 0.30 - O I 0.30 - Oz 6 0.20 - 0.20 - 0.10 - . • • • --. 0.10 - 0.00 ' 0.00 i 20 40 60 80 20 40 60 80 Time (min) Time (min) (C) (d) - -o-pH=4.38 - o-pH=5.05 / i . -^pH=6.05 40 20 40 60 80 Time (min) Time (min) Figure 6.2 Concentrations in solution during dissolution of different amounts of struvite pellets for initial pH value of 5.05: (a) total orthophosphate (CPo4), (b) total ammonia (CNH4), (C) Mean ( C ); and (d) for 1000 mg struvite pellets for different initial pH values. 166 4.50 4.00 • g 3.50 o g> 3.00 o 2.50 A • 2.00 o o > 1.50 -\ 0 A co 1.00 0.50 0.00 0 200 400 600 800 1000 1200 Initial weight (mg) Figure 6.3 Percent struvite dissolved in 1 h against initial amount of struvite for different initial pH values. (a) 0.0 * 0 20 40 60 80 100 120 140 Time (h) Figure 6.4 Concentrations in solution during dissolution of 1000 mg of struvite pellets with an initial pH of 5.15: (a) 10 h, and (b) 5 d. 168 10 o pH=4.38 8 • pH=5.05 A pH=6.06 o 1000 2000 3000 4000 Time (s) Figure 6.5 Model curves of the concentrations of struvite with time for an initial amount of 1000 mg, for different initial pH values. 169 2.02 2.01 Rz = 0.94/ 2.01 / / ~ 2.00 / o T o/ ^_ 2.00 o9 | 1.99 ^ 1.99 o/ y 1.98 1.98 1.97 1.97 5.5 5.7 5.9 6.1 6.3 6.5 1 1 A(Cs-C)/V(mol I" m" ) 1 Figure 6.6 Plot of dC/dt as a function of A V (Cs-C) drawn for 500 mg initial amount of struvite pellets and pH value of 4.38. 170 3.5 3.0 FT = 0.98 o C/3 | 2.5 00 o or * 2.0 O o 1.5 1.0 10.5 11.5 12.5 13.5 2/ 1 6 5/3 6 ns V (Cs-C) x10" (mol 1" ) Figure 6.7 Plots of dC/dt as a function of ns V (Cs - C) for an initial struvite mass of 1000 mg, and pH 4.38. 171 (a) 3 • c o CNH4 2.5 H o • CP04 o o 0 A CMg 2 o o o D • X A o c 0 1.5 A o © • • A s O -f—' A c 1 o • CD O • T A o O 0.5 A • 0— 0 1000 2000 3000 4000 Time (min) (b) 3 • c 2.5 o CNH4 • CP04 2 O A A A CMg • • 1.5 o o A A c 1 CD O o 0.5 O 0 •* 0 1000 2000 3000 4000 Time (min) Figure 6.8 Struvite constituent dissolutions with time for input solution pH (a) 6.05 and (b) 8.55. 172 -13.1 o CP04 • CNH4 -13.2 A CMg X S-1 o C © -13.3 o E -13.4 o o o 0 „ A . E O A -13.5 og o -13.6 S-3 9 o -13.7 XLSQ2J -13.8 10 pH Figure 6.9 Change of rate of dissolution with pH for struvite pellets in mixed flow through reactor showing the dissolution rates determined using river(s)/creek(s) waters (boxed values S-1 to S-3). 173 References American Public Health Association (APHA), American Water Works Association, and Water Pollution Control Federation, (1998) 20th edn. Standard Methods for Examination of Water and Wastewater, Washington , D.C. Appelo, C.A.J., and Postma, D. (1999) Geochemistry, Groundwater and Pollution, A.A. Balkema, Rotterdam, The Netherlands, pp. 73-76. Bavic-Ivancic, V., Kontrec, J., Kralj,D. and Brecevic, L. (2002) Precipitation diagram of struvite and dissolution kinetics of different struvite morphologies. Croatica Chemica etActa 75, 89-106 Berner, E.K. and Berner, R.A. (1987) The Global Water Cycle: Geochemistry and Environment, Prentice-Hall, New Jersey. Bhuiyan, M.I.H, Mavinic, D.S. (2007a) Assessing struvite precipitation in a pilot-scale fluidized bed crystallizer. Environmental Technology (submitted). Bhuiyan, M.I.H, Mavinic, D.S., and Koch, F.A. (2007b) Thermal decomposition of struvite and its phase transition. Chemosphere, in press. Bhuiyan, M.I.H., Mavinic, D.S., and Beckie, R.D. (2007c) A solubility and thermodynamic study of struvite. Environmental Technology 28, 1015-1026. Brantley, S.L. (2003) Reaction kinetics of primary rock forming minerals under ambient conditions. Treatise Geochemica 5,1-44. Bridger, G.L., Murrel, L.S. and Starostka, R.W. (1962) Metal ammonium phosphates as fertilizers. Journal of Agriculture and Food Chemistry 10 (3), 181-188. Christoffersen, J. and Christoffersen, M.R. (1984) Kinetics of Dissolution of Calcium Hydroxyapatite. Farady Discussion Chemica Society 77, 235-242. Doyle, J.D. and Simon, A.P. (2002) Struvite formation, control and recovery. Water Research 36, 3925-3940. Gaterell, M.R., Gay, R., Wilson, R., Gochin, R.J. and Lester, J.N. (2000) An economic and environmental evaluation of the opportunities for substituting phosphorus recovered from wastewater treatment works in existing UK fertilizer markets. Environmental Technology 21, 1067-1084. Golubev, S.V., Bauer, A., and Pokrovsky, O.S. (2006) Effect of pH and organic ligands on the kinetics of smectite dissolution at 25°C. Geochimica et Cosmochimica Acta 70, 4436-4451. 174 Guo, M., Liu, M., Hu, Z., Zhan, F., and Wu, L. (2005) Preparation and properties of a slow release NP compound fertilizer with superabsorbent and moisture preservation. Journal of Applied Polymer Science 96, 2132-2138. Hodson, M.E. (2006) Does reactive surface area depend on grain size? Results from pH 3, 25°C far-from-equilibrium flow-through dissolution experiments on anorthite and biotite. Geochimica et Cosmochimica Acta 70,1655-1667. Kraemer, S.M., Chiu, V. Q., and Hering, J.G. (1998) Influence of pH and competitive adsorption on the kinetics of ligand-promoted dissolution of aluminum oxide. Environmental Science and Technology 32, 2876-2882. Mullin, J.W. (2001) Crystallization. 4th edn, Butterworth -Heinemann, Jordan Hill, Oxford, UK. Ohlinger, K.N., Young, T.M., and Schroeder, E.D. (1998) Predicting struvite formation in digestion. Water Research 32(12): 3607-3614. Provosky, S.O. and Schott, J. (2000) Kinetics and mechanism of forsterite dissolution at 25°C andpH from 1 to 12. Geochimica et Cosmaochimica Acta 64(19), 3313-3325. Samson, S.D., Stillings, L.L., and Eggleston, C.M. (2000) The depletion and regeneration of dissolution-active sites at the mineral-water interface: I. Fe, Al, and In sesquioxides. Geochimica et Cosmochimica Acta 64 (20), 3471-3484. Sterling, M.S. (1997) Phosphorus Release From a Slow-release Fertilizer Under Simulated Stream Conditions. Masters Thesis, Department of Civil Engineering, University of British Columbia, Vancouver, BC, Canada. Weissbart, E.J. and Rimstidt, J.D. (2000) Wallastonite: incongruent dissolution and leached layer formation. Geochimica et Cosmochimica Acta 64(23), 4007-4016. 175 Chapter 7 Thermal decomposition of struvite and its phase transition 7.1 Introduction The problems associated with struvite deposition on equipment surfaces of anaerobic digestion and post-digestion processes in the wastewater treatment industry (especially Biological Nutrient Removal (BNR) processes) affects it commercially through major downtime, loss of hydraulic capacity, and increased pumping and maintenance costs (Doyle and Simon, 2002). Although struvite can be a problem in wastewater treatment plants, the conditions for its formation found within the environment of wastewater treatment works can be exploited for extraction of struvite as a commercial product. Struvite can be used as slow release fertilizer at high application rates, without the danger of damaging plant roots (Bridger et al, 1962). Fertilizer applications for struvite are diverse and include ornamentals, forest outplantings, turf, orchards and potted plants. Research has also been undertaken with briquettes of magnesium ammonium phosphate, to investigate its slow release property in a small stream in British Columbia (BC), Canada because of its high phosphorus (P) content and potential for nutrient replacement source for salmon carcasses (Sterling, 1997). Compared to other fertilizers, the benefits of using struvite are low leaching rates and prolonged release of nutrients throughout the growing season of plants, with the possibility of only one single application (Gaterell et al, 2000). A version of this chapter has been accepted for publication: Bhuiyan, M.I.H, Mavinic, D.S., and Koch, F.A. (2007) Thermal decomposition of struvite and its phase transition. Chemosphere, in press. 176 Granular forms of struvite are one of the best, slow-release P fertilizers (Gaterell, 2000; Bridger et al, 1962). In the context of competition in the marketplace as a fertilizer with diammonium phosphate (DAP) and triple superphosphate, a modification of struvite has been proposed, whereby struvite should be treated with phosphoric acid (Gaterell et al, 2000). This modified form is called enhanced struvite, containing two parts of slowly-soluble, mono hydrogen magnesium phosphate (MgHP04), to one part of highly soluble DAP [(NH4)2HP04]. This product may prove to be suitable where an initial high dose of P is required, followed by a sustainable slow release of P. It is suggested that, the availability of phosphorus to the soil is higher for hexahydrate struvite than that of the monohydrate dittmarite (MgNH4P04 H20) because of greater dilution with water of crystallization. When dittmarite contacts soil moisture at ambient temperatures, it gradually hydrates to hexahydrate (Bridger et al, 1962). Thermophilic digestion of sludge is normally carried out in a temperature range of 50- 60°C. The use of an optional pasteurization system, to treat the sludge at 70°C for over 30 min prior to mesophilic (36-38°C) digestion, is also used in Europe (Oleszkiewicz and Mavinic, 2001). Thus, a study on the thermal stability, phase transition and decomposition of the products of the struvite system would provide a better understanding of the fate of struvite in anaerobic digestion and post digestion processes, as well as off-site, agricultural use. The apparently fragile equilibrium of struvite in solution leads to the presence of other crystal phases as well (Andrade and Schuiling, 2001). The formation of magnesium phosphates - such as MgHP04 • 3H20 (newberyite), Mg3(P04)2 ' 8Ff20 (bobierrite) and Mg3(P04)2 22H20 (cattiite), during struvite crystallization or dissolution process, is reported in the literature (Michalowski and Pietrzyk, 2006; Taylor et al, 1963b; Johnson, 1959). When the pH is 177 increased from slightly acidic to slightly basic values, there is a change in predominant solid species from newberyite to struvite to cattiite (Dempsy, 1997; Taylor et al, 1963b). It is suggested that, for a pH between 6.4 and 7.7, both newberrite and struvite are thermodynamically stable (Dempsy, 1997), while both struvite and bobierrite are stable phases under alkaline pH conditions (Taylor et al, 1963b). It has also been reported that at room temperature, cattiite is stable in air but unstable in water, in which it reverts to bobierrite. The difference in the solubility products of these two hydrates of trimagnesium phosphates (see Table 7.1) indicates that they have distinctly different solubilities and that the reversion must involve dissolution of the higher hydrate and crystallization of the lower hydrate from the solution (Taylor et al, 1963b ). Table 7.1 shows the solubility product values (Ksp) available in the 2+ + + literature for the precipitates in Mg - PO43" - NH4 - H systems. In this research project, a study was carried out in the laboratory to investigate the phase relationships of various reaction products of both synthetic and real magnesium ammonium phosphate systems at different temperatures and different heating rates. In this paper, the decomposition behaviour of struvite at various temperatures, and the mechanism of its transformation to other forms, including bobierrite and dittmarite, are reported. 7.2 Materials and methods 7.2.1 Formation of struvite Synthetic struvite was prepared by mixing equal volumes of equimolar quantities of magnesium chloride and diammonium phosphate. The solution was then made mildly alkaline (pH = 7.4) by slow addition, with stirring, of filtered ammonium hydroxide (Babic-Ivancic et al, 2002; Ohlinger, 1999; Johnson, 1959). After initial mixing of reactant solutions, the solution was sealed in a tightly closed container, leaving a minimum space above the solution. The solution 178 was then maintained at 25 C, without agitation, for 24 hours. The suspensions were filtered though a 0.45 pm membrane filter, and the precipitates were washed thoroughly with distilled water and allowed to dry at room temperature overnight. Other struvite samples, used in this study, were obtained from pilot-scale struvite crystallizer demonstration studies at the Lulu Island Wastewater Treatment Pant, Richmond, BC (Bhuiyan et al, 2007a, chapter 5). The reactor design follows the concept of a fluidized bed (see Figure 5.1, Chapter 5). It has four different areas of cross section, increasing from the bottom to the top. A harvest zone is used primarily for harvesting of the struvite pellets and is located immediately above the injection ports. The process consists of a fluidized bed reactor; an external clarifier; storage tanks for centrate, magnesium chloride and sodium hydroxide solutions; pumps for the centrate as feed, magnesium and recycle flows; and a pH controller. Magnesium chloride and sodium hydroxide are added to the reactor through the injection ports, just above the feed and recycle flows. Struvite pellets, harvested from the harvesting section every alternate day, were dried in air for at least 24 h at room temperature. The size of the pellets ranged from 0.5 to 3.5 mm. 7.2.2 Identification of struvite and transformation compounds Both dried precipitate of the synthetic solution and struvite pellets formed in the pilot scale crystallizer were examined and identified as pure struvite by x-ray diffraction (XRD), using a Bruker D8 Advance X-ray diffractometer and CuKa radiation, with an average scanning rate of 2.0° 20 min" . Identification was also made by the room-temperature infrared (IR) spectra in the wave number range of 400 to 4000 cm"1, on a Perkin Elmer 1710 Infrared Fourier Transform Spectrometer. 179 7.2.3 Phase transition of struvite in excess water For investigation of the phase changes of struvite with temperature in excess water, 125 mg of struvite (sufficient to provide an excess of the solid phase) were added to 125 ml glass jars full of deionized water. The jars were sealed and placed in an Innova 4230 refrigerated incubator shaker. The instrument was set at 150 rpm for shaking. The experiments were conducted for both synthetic powdered struvite and struvite pellets at three different temperatures: 50°C, 60°C and 80°C. A known amount of struvite pellets (0.5-2.0 mm in size) was also boiled in a round flask connected to a condenser. The refrigerated incubator shaker could not be used for boiling, since it cannot be run above 80 °C. All the mixtures were heated for 24 hours. Next, the samples of the solutions were filtered through 0.45 um size membrane filter paper. The filtrates were washed three times with deionized water, to remove extraneous ions. 7.2.4 Analytical and thermogravimetric methods Analyses for ortho-phosphate and ammonia were made, using the flow injection method on a LaChat QuickChem 8000, as described in the method number 4500-P G and 4500-NH3 H of the Standard Methods for the Examination of Water and Wastewater (APHA et al, 1998). Magnesium analysis was performed by flame atomic absorption spectrophotometry, using a Varian Inc. SpectrAA220 Fast Sequential Atomic Absorption Spectrophotometer The thermogravimetric analysis (TGA) of struvite pellets and synthetic struvite was performed on a Pyris 6 TGA Perkin Elmer Thermogravimetric analyzer under nitrogen gas and at different heating rates. The conversion of the TGA curve to its derivative (DTGA) was undertaken from the rate of mass loss curve as a function of temperature. 180 7.3 Results and discussion 7.3.1 Identification of struvite The powdered XRD pattern of synthetic and struvite pellets matched very well with the pattern for struvite standarad (see Figure 7.1). IR spectra for both synthetic struvite and struvite pellets were also consistent with the published IR spectrum of struvite. As such, the struvite samples were considered to be a single-phase material. Analysis of the magnesium, ammonium and phosphate contents of weighed samples, after dissolution in 0.5% nitric acid solution, indicated purity of 98.0( ± 1.0) %. 7.3.2 Thermogravimetric Analysis The TGA and DTGA curves for synthetic struvite and struvite pellets, at different heating rates are shown in Figure 7.2 and 7.3 respectively. These data indicate that mass loss begins at a temperature around 55°C and is essentially complete when the temperature reaches above 250°C. At this point, ~ 51% of the original mass loss occurred. This mass loss corresponds to the following decomposition reaction for struvite (equation 7-1): MgNH4P04 6H20 -> MgHP04 + NH3T + 6H20t (7-1) DTGA curves of the synthetic struvite for heating rates 20, 5, 2 and 1°C min"1 show a single peak for every case, which is attained at higher temperatures for a faster heating rate (Fig. 7.2). The peaks of the DTGA curves of synthetic struvite are attained at around 190, 125, 113 and 104 °C, for heating rates of 20, 5, 2 and 1°C min"1, respectively. These peaks correspond to major mass loss during the course of heating in each case. DTGA curves of the struvite pellets in Figure 7.3 show the major mass loss peaks at around 150°C, for a heating rate of 5°C min' and at 115°C for 1°C min" . The simultaneous loss of both ammonia and water molecules from the struvite 181 structure presumably occurred gradually as a function of temperature, rather than as a distinct step; otherwise, double peaks would have appeared in the corresponding DTGA curve. In this study, decomposition of struvite was found to be dependent upon the rate of heating. The decomposition is different in nitrogen or in moist atmosphere; it also depends upon the partial pressure of water (Frost et al, 2004). However, variation of the steepness of the slopes of the TGA curves, and the sharpness of the peaks of the DTGA curves show that the decomposition takes place faster in the case of slower heating rates. Struvite is often found in the urinary tract and kidneys (Ramachandran and Nataranjan, 2004; Johnson, 1959). It was suggested that the thermal decomposition of 'kidney stone' (struvite), at very low heating rates, show that the ammonia is lost before the water of crystallization. At a thermal decomposition rate of 1°C min"1, three mass loss steps, namely at 39.5, 57.8 and 82.6 °C have been suggested. Using a heating rate of 2°C min"1, the same researcher found that the ammonia and water molecules were lost simultaneously with thermal decomposition at 85°C (Frost et al, 2004). In this study, both for synthetic struvite and struvite pellets, the temperature of the thermal decomposition was always below 250°C. For a slower rate of heating, the decomposition took place at a lower temperature. This was also verified by X-ray crystallography. After identification of struvite samples, the temperature was raised from room temperature to 150°C, at a heating rate of 0.5°C min"1. XRD was carried out with increasing temperature, at every 10°C interval. Struvite was found to lose its structure and become amorphous at around 110°C (see Fig. 7.4). However, in thermogravimatric analysis, which was carried out at heating rates of 1°C min"1 and above, the major mass losses (presumably with simultaneous loss of ammonia and/or water molecules) were found to take place above 100°C, for every case. 182 Struvite starts to become X-ray non-crystalline, as soon water molecules are expelled from the crystal structure, resulting in a mass loss. The XRD on the final products of the thermal decomposition showed that they were X-ray amorphous. It has been suggested that the struvite decomposition temperature is affected by the degree of crystallinity or the crystal size (Sarker, 1991). Upon rehydration of the dehydrated struvite, at room-temperature, it was found to return to its original phase along with some unknown phases, and mostly as newberyite, but only if some ammonia molecules were still present in the structure. However, dittmarite could not be detected in any of his rehydrated samples by the XRD technique (Sarker, 1991). 7.3.3 Evaluation of activation energy Activation energy is the threshold energy, or minimum energy that is to be overcome in order for a chemical reaction to occur. This kinetic parameter for the decomposition reactions of struvite, for different heating rates, was determined by the following method (Gyore and Ecet, 1973), using the Arrhenius equation (equation 7-2) k = Ae-Ea/RT (?_2) 1 where, Ea = activation energy (kcal mol" ), k = rate constant of the decomposition reaction (min"1), R = gas constant ( = 1.987 x 10"3 kcal mol"1 deg"1), and T = absolute temperature in °K. The rate of mass loss and reaction rate constant, k at various temperatures are calculated from Equations 7-3 and 7-4. dw__dw_dr_ ^ dt dx dt 183 ^ = K (7-4) at dw where, — = mass loss per minute, dt dw 0 — = mass loss per C, dr — = heating rate, and dt wr = residual mass of the sample at temperature r The value of Ea was calculated from the slope (m) of the plot of log k versus reciprocal absolute temperature, by means of the relation, Ea = 2.3 R. m. Table 7.2 shows the values of the calculated activation energy at different heating rates, for both synthetic struvite and struvite pellets. The activation energy for both synthetic struvite and struvite pellets was found to be higher (35.13 kcal mol"1) in a slower heating rate (heating rate 1°C min"1), while for synthetic struvite at a heating rate of 20°C min"1, it was 19.13 kcal mol"1. The kinetic parameters from experimental data, obtained under dynamic temperature conditions, provide an idea of the non-isothermal decomposition behaviour of the material. The decrease of the activation energy, the minimum energy needed to initiate the decomposition reaction, with increasing rate has been also reported previously for other compounds (Zasko and Arz, 1974). 7.3.4 Phase transition with heating in excess water The precipitation of struvite seems to be dependent, not only on the supersaturation of the solution with respect to this mineral, but also on the ammonia activity in the solution. Struvite is also reported to be metastable at low ammonia activity (Boistelle et al, 1983; Johnson, 1959). It is suggested from previous research that the precipitation of struvite in the first phase, is 184 accompanied by a sharp decrease in pH, which subsequently remains constant. If any precipitation of newberyite crystals occurs in the first phase, the solution still remains supersaturated with respect to newberyite; thus, the precipitation of newbryite then occurs, resulting in another drop in pH. This results in undersaturation with respect to struvite, which starts dissolving to restore saturation. During this process, the pH may remain nearly constant as a result of newberyite growth and struvite dissolution (Boistelle et al, 1983). Newberyite is reported to be stable at high concentrations of magnesium and at low pH values (Babic-Ivancic etal, 2003; Johnson, 1959). Formation of newberyite at higher temperatures has also been reported (Babic-Ivancic et al, 2003; Hirasawa et al, 1997). In another study, involving an investigation of the kinetics of transformation of struvite to newberyite, solid phase transformation was found to occur, through a solution-mediated process, especially below a neutral pH (Babic-Ivancic et al, 2006). However, no newberyite crystals were observed during formation of struvite in this study, most likely because the solution conditions and the system used were most favorable for pure struvite formation. When struvite was heated in excess water at temperatures 50, 60 and 80°C, transformation was found to take place partially from struvite to bobierrite, through a gradual loss of ammonium ions from the struvite structure. If ammonia is allowed to escape, the hydrolysis of the magnesium ammonium salt would continue, with precipitation of newberyite, and finally, as the more stable bobierrite (Taylor et al, 1963a). Although heating of the mixtures were undertaken in sealed conditions, most of the ammonia, if not all, was transformed into NH3 species at these high temperatures and high pH, resulting from dissolution. When a known amount of heated struvite was dissolved in 100 ml 0.235N HC1 solution at 25°C for 24 h, the 185 NH4-N to PO4-P molar ratio in the solution was found to decrease with the temperature at which the struvite was heated (see Fig. 7.5). The XRD of the struvite crystal heated at 50, 60 and 80°C, in excess water, also indicated that the bobierrite peaks became more visible with the increase in temperature (see Fig. 7.6 to 7.8). 7.3.5 Phase transition with boiling in excess water When a known amount of struvite pellets was boiled in excess water for 24 hours, the final product was found to be completely converted to dittmarite. Figure 7.9 shows that the XRD of the product matches very well with that of dittmarite. It is suggested that single-phase dittmarite could be synthesized in the laboratory in the same way as struvite, except that the two solutions have to be mixed at boiling temperatures (Sarker, 1991). The decomposition temperature of dittmarite was found to be around 220 °C in the same study under static air and a heating rate of 5 °C min"1. The percentage mass loss (around 80%) in the TGA curves of his study with dittmarite corresponds to the decomposition of dittmarite through the reaction: Mg NH4PO4' H20 -» MgHP04 + NH3T +H2OT (7-5) Since only one large mass loss peak appears in the DTGA curve for dittmarite, its dehydration and decomposition behaviour appears to be very similar to that of struvite, i.e, through simultaneous release of both NH3 and H20 molecules. The two compounds are structurally related, and the difference might be in the way the extra molecules are linked. These extra molecules make the struvite structure thermally less stable than dittmarite. However, if dittmarite is slowly hydrated at room temperature, it is transformed to the more stable hexahydrate, struvite with time (Sarker, 1991; Bridger et al, 1962). A schematic diagram is shown in Figure 7.10, for the possible transformation mechanisms of various phases associated with struvite (Bridger et al, 1962; Sarker, 1991; Babic-Ivancic et al, 2003). 186 The decomposition of struvite, under dynamic temperature conditions, was found to depend on the rate of heating; also, the decomposition occurred faster in the case of a slower heating rate. The simultaneous loss of both ammonia and water molecules from the struvite structure was found to occur gradually as a function of temperature, rather than as a distinct step. The resultant decomposition product of struvite is X-ray amorphous and chemically corresponds to MgHP04, as determined by the mass loss measurement. For the heating rates used in this study, the decomposition of both synthetic struvite and real struvite pellets was found to occur in a single stage. The activation energy of the decomposition reactions of this stage, for both synthetic struvite and struvite pellets, was found to be higher for a slower heating rate. Struvite shows different behaviours at higher temperatures in excess water. When struvite was heated at higher temperatures (50-80°C), in the presence of excess water, for 24 h, it gradually lost some ammonia and was partially transformed into bobierrite. Boiling struvite in excess water resulted in the loss of five water molecules from its structure and transformed into the monohydrate, dittmarite. From a fertilizer point of view, dittmarite has a higher P205 (45.69%) than struvite (28.92%). It has other commercial uses as well. The conditions for the transformation of struvite to dittmaritte could be exploited for intentional dittmarite formation that would extend the field of P recovery from wastewater. However, the economics associated with this transformation have yet to be proven 187 as cost-effective. Further research, along these lines, is currently under way. With quantification of water in each compound and quantification of each element in a compound, derivation of a phase diagram of the relationship between struvite, bobirite, and dittmarite is recommended for future study. 188 Table 7.1 Solubility product values (Ksp) available in the literature for the precipitates 2+ + + 3 in Mg - PO4 " - NH4 - H system. Species Chemical formula Ksp pKsp Ref. 14 Stravite MgNH4P04 6H20 (s) 4.37xl0" 13.36 (Bhuiyan et al, 2007b) 5.37xl0~14 13.27 (Ohlinger, 1999) 7.08xl0"14 13.15 (Taylor et al, 1963a) 6 Newberyite MgHP04 3H20 (s) 1.58xl0" 5.80 (Taylor et al, 1963b) 26 Bobierrite Mg3(P04)2 8H20 (s) 6.31xl0" 25.20 (Taylor et al, 1963b) 24 Cattiite Mg3(P04)2 22H20 (s) 7.94x10" 23.10 (Taylor etal, 1963b) 12 Brucite Mg (OH)2 (s) 6.92xl0" 11.16 (Stumm and Morgan, 1981) 189 Table 7.2 Activation energy of the struvite decomposition reactions at different heating rates. Heating rate Activation energy Sample (°C min"1) (kcal mol"1) Synthetic struvite 20 19.1 5 28.4 2 34.9 1 35.1 Struvite pellets 5 19.9 1 35.1 190 (a) 3000 t/) Q. O 2000 JAu—ft, 20 2-Theta - Scale (b) 4000 CL O 2000 1000 J L 2-Theta - Scale Figure 7.1 X-ray diffraction patterns of (a) synthetic struvite (b) struvite pellets, and struvite standard (•). 191 0 0 50 100 150 200 250 300 350 400 450 500 550 Temperature (°C) Figure 7.2 TGA and DTGA curves of synthetic struvite for heating rate 1, 2, 5 and 20°C min"1. 192 120 5 •©—TGA 5 degC/min 0 50 100 150 200 250 300 350 400 450 500 Temperature (° C) Figure 7.3 TGA and DTGA curves for heating rate 1 and 5°C min"1 for struvite pellets. 193 X-axis: 2-Theta, Y-axis: Cps Lin, and Z-axis: scan order from 30 to 150 °C at 10 °C interval. Figure 7.4 X-ray diffraction pattern (3-D) of struvite pellets with increasing temperature. 194 1.2 1.0 •2 0.8 CO I— ra o E a- 0.6 O CL X z 0.4 A 0.2 0.0 45 50 55 60 65 70 75 80 85 Temperature (°C) Figure 7.5 NH4-N/ PO4-P moiar ratio in the solution after dissolution of struvite heated at different temperatures and excess water. Error bars: 95% confidence interval. 195 (a) 2000 1900 1800 1700 1600 1500 1400 1300 1200 Q. 1100 1000 • S 900 800 700 600 500 400 300 200 \ 100 0 2-Theta - Scale (b) 2-Theta - Scale Figure 7.6 X-ray diffraction patterns of the heated (a) synthetic struvite (b) struvite pellets with struvite (•) and bobierrite (•) standards after heating at 50°C in excess water. Bobierrite peaks are identified by arrow. 196 (a) 2000 1900 1800 1700 1600 1500 1400 1300 1200 w Q. 1100 c 1000 Zi 900 800 700 600 500 400 300 200 100 0 20 30 2-Theta - Scale (b) 2000 1900 1800 1700 1600 1500 1400 1300 1200 Q. 1100 1000 c Zi 900 800 700 600 500 400 300 200 100 2-Theta - Scale Figure 7.7 Powder X-ray diffraction patterns of the heated (a) synthetic struvite (b) struvite pellets with struvite (•) and bobierrite (•) standards after heating at 60°C in excess water. Bobierrite peaks are identified by arrow. 197 (a) 2000 1900 1800 1700 1600 1500 1400 1300 ^-v 1200 Q. 1100 O 1000 c 900 800 700 600 500 400 300 200 100 0 30 40 2-Theta - Scale (b) 2000 1900 1800 1700 1600 1500 1400 1300 1200 2.1100 y- 1000 700 600 500 400 300 1 J 200 100 0 2-Theta - Scale Figure 7.8 X-ray diffraction patterns of the heated (a) synthetic struvite (b) struvite pellets with struvite (•) and bobierrite (•) standards after heating at 80°C in excess water. Bobierrite peaks are identified by arrow. 198 40000 -1 30000—| 3 10 20 30 <0 50 60 70 2-Theta - Scale Figure 7.9 X-ray pattern of the product after boiling struvite pellets in excess water for 1 d with the pattern of the dittmarite standard (•). 199 Room temperature . Water.. . Heating excess water Dittmarite Struvite Bobierrite Boiling excess water Room Temperature- Water Excess magnesium and < Neutral pH) (SO S Amorphous Newbervite MgHP04 temperature water Figure 7.10 Schematic of the possible transformation mechanisms of various phases associated with struvite. 200 References American Public Health Association (APHA), American Water Works Association, and Water Pollution Control Federation, (1998) 20th edn. Standard Methods for Examination of Water and Wastewater, Washington, D.C. Andrade, A. and Schuiling, R.D., (2001) The chemistry of struvite crystallization. Mineralogical Journal (Ukraine) 23, 37.-46. Babic-Ivancic, V., Kontrec, J., Brecevic, L., and Kralj, D. (2006) Kinetics of struvite to newberyite transformation in the precipitation system MgCl2-NH4H2P04-NaOH-H20. Water Research 40,3447-3455. Babic-Ivancic, V., Kontrec, J., Kralj, D., K. and Brecevic, L. (2002) Precipitation diagram of struvite and dissolution kinetics of different struvite morphologies. Croatica Chemica et Acta 75, 89-106. Bhuiyan, M.I.H, Mavinic, D.S. (2007a) Assessing struvite precipitation in a pilot-scale fluidized bed crystallizer. Environmental Technology (submitted). Bhuiyan, M.I.H., Mavinic, D.S., and Beckie, R.D. (2007b) A solubility and thermodynamic study of struvite. Environmental Technology 28, 1015-1026. Boistelle, R., Abbona, F. and Madsen, H.E.L. (1983) On the transformation of struvite into newberyite in aqueous systems. Physics and Chemistry of Minerals 9, 216-222. Bridger, G.L., Salutsky, M.L., and Starostka, R.W. (1962) Metal ammonium phosphates as fertilizers. Journal of Agriculture and Food Chemistry 10,181-188. Dempsy, B.A. (1997) Removal and re-use of ammonia and phosphate by precipitation of struvite. Proceedings of the 52nd Industrial Waste Conference, Purdue University, West Lafayette, Ind., 5-7 May, Ann Arbor Press, Chelsea, Mich., pp. 1-782. Doyle, J.D. and Simon, A.P. (2002) Struvite formation, control and recovery. Water Research 36, 3925-3940. Frost, R.L, Weier, M.L. and Erickson, K.L., (2004). Thermal decomposition of struvite- implication for the decomposition of kidney stone. Journal of Thermal Analysis and Calorimetry 76, 1025-1033. Gaterell, M.R., Gay, R., Wilson, R., Gochin, R.J. and Lester, J.N. (2000) An economic and environmental evaluation of the opportunities for substituting phosphorus recovered from wastewater treatment works in existing UK fertilizer markets. Environmental Technology 21, 1067-1084. 201 Gyore, J. and Ecet, M., (1973) Calculation method based on DTG curve for evaluation of activation energy. Journal of Thermal Analysis 5, 299-305. Hirasawa, I., Nakagawa, H., Yosikawa, O. and Itoh, M. (1997) Phosphate recovery and reactive crystallization of magnesium ammonium phosphate: application of wastewater. American Chemical Society Symposium Series 667, 267-276. Johnson, R.G. (1959) The solubility of magnesium ammonium phosphate hexahydrate at 38°C with considerations pertaining to the urine and the formation of urinary calculi. The Journal of Urology 81, 681-690. Michalowski, T., and Pietrzyk, A., (2006) A thermodynamic study of struvite + water system. Talanta 68, 594-601. Ohlinger, K., (1999) Struvite Controls in Anaerobic Digestion and Post-digestion Wastewater Treatment Processes. Ph.D. Thesis, University of California Davis, U.S. Oleszkiewicz, J.A. and Mavinic, D.S. (2001) Wastewater biosolids: an overview of processing, treatment, and management. Canadian Journal of Civil Engineering 28,102-114. Ramachandran, E. and Natarajan, S. (2004) Crystal growth of some urinary stone constituents: III. In-vitro crystallization of L-cystine and its characterization, Crystal Resarch Technology 39, 308-312. Sarker, A.K. (1991) Hydration/ dehydration characteristics of struvite and dittmarite pertaining to magnesium ammonium phosphate cement systems. Journal of Material Science 26, 2514- 2518. Sterling, M.S. (1997) Phosphorus Release From a Slow-Release Fertilizer Under Simulated Stream Conditions. Masters Thesis. University of British Columbia, Vancouver, BC, Canada. Stumm, W. and Morgan, J. (1981) Aquatic chemistry. Wiley-Interscience, New York, pp. 277- 278. Taylor, A.W., Frazier, A.W., and Gurney, E.L. (1963a) Solubility products of ammonium and magnesium potassium phosphates. Transactions of the Faraday Society, 59, 1580-584. Taylor, A.W., Frazier, A.W., Gurney, E.L., and Smith, J.P., (1963b) Solubility products of di- and trimagnesium ammonium and dissociation of magnesium phosphate solutions. Transactions of the Faraday Society 59, 1585-1589. Zasko, J. and Arz, H.E., (1974) Kinetic Analysis of thermogravimetric data VII. Thermal decomposition of calcium carbonate. Journal of Thermal Analysis 6, 651-656. 202 Chapter 8 General conclusions and direction for future research 8.1 Introduction The ability of phosphorus to have significant recovery potential gives full meaning to what should be a core principle of true sustainable development: managing today's resources for the people of tomorrow. Among all the recognized phosphorous removal technologies, crystallization processes stand out because they not only achieve high P removal, but also recover P from wastewater as useful products. The ethics of sustainability makes this option most attractive since all other processes produce wastes which need to be landfilled or incinerated. Struvite and hydroxyapatite can be used in agriculture as fertilizers; struvite, however, is preferred for numerous reasons such as slow-rrelease rate, less impurities caused by heavy metals, and simultaneous release of essential nutrients (Durant et al, 1999). The struvite crystallization process also has physical advantages, in addition to the chemical advantages. To date, several remedial measures have been suggested for alleviating the problem of unintentional struvite formation in wastewater treatment plants. Intentional struvite formation, before it forms in the equipment, appears to be the most practical solution. The conditions for struvite formation that can be found naturally within the environment of wastewater treatment works are exploited for extraction of struvite as a commercial product. It also closes the phosphorus loop in the soil-crop-animal-human-soil cycle and paves the way to an ecologically sustainable future, as far as phosphorus extraction and usage are concerned. The basic design of the pilot-scale UBC (University of British Columbia) MAP (Magnesium ammonium phosphate) Crystalliser, based on fluidized bed systems, was found effective in recovering phosphate from both real and synthetic supernatant (Adnan et al. 2003; Britton et ai, 2005). The first, large-scale, struvite recovery demonstration reactor has been built 203 by Ostara Nutrient Recovery Technologies Inc, at the Goldbar Wastewater Treatment Plant in Edmonton, Canada, based on the fluidized bed reactor technology developed by UBC (Scope Newsletter, 2006). The products of the UBC crystallizer have also been found to be comparable or even better than those of other technologies. Results from this study would obviously help move forward the UBC MAP Crystallizer technology. To address the objectives of this research (Chapter 1), different experimental set ups were used in the UBC Environmental Engineering Laboratory. A bench-scale fluidized bed reactor (FBR) was used in the laboratory, while a pilot-scale FBR was operated at the Lulu Island Wastewater Treatment Plant. 8.2 Overall conclusions The thermodynamic solubility products of struvite were determined by extrapolating measured Ksp values to zero ionic strength, using an appropriate activity coefficient model. The pKsp values for temperatures between 10-60 °C range from 13.17(±0.05) to 14.36(±0.03), with 13.36(±0.07) at 25°C. The solubility and solubility product value of struvite are not directly related, as is found in the case of highly soluble salts. The solubility of struvite determined in deionized water was found to be 169.2 (±4.3) mg l"1 at 25°C, with the maximum value of 212.7 (±3.8) mg 1"' at 35°C. Since the thermodynamics of struvite change above the temperature of maximum solubility, use of the same value of enthalpy of reaction, in calculations to derive Ksp values for other temperatures, may lead to erroneous results. For use in such cases, an analytical expression has been proposed (Chapter 2). A representative temperature compensation factor for conductivity has been derived, in a system associated with struvite formation from anaerobic digester supernatant/centrate samples of five different wastewater treatment plants. Using a temperature compensation factor, a 204 =0.0198 °C~' for all samples, the estimated electrical conductivity (EC) values were found to fairly accurately (R2~l) match the measured values. Considering the temperature dependence of EC, a relationship between EC and inonic strength (I), including correction for ion-pairing, was developed in this study exclusively from anaerobic digester supernatant/centrate. This relationship can be used to estimate the ionic strength of the solution in a system associated with struvite formation, from anaerobic digester supernatant/centrate (Chapter 3). The induction period for struvite formation, determined in the laboratory, was found to be primarily reaction controlled, with minor transport influence. The metastable region, where nucleation is negligible, was explored for struvite in this study. The nucleation of struvite was suppressed when the growth experiments were conducted in the metastable region. The solubility and supersolubility curves, which are the boundaries of the metastable region, were found to be almost parallel straight lines in the concentration range studied (PO4-P = 41-1303 mg 1"', NH4-N = 486-1222 mg l"1 and Mg = 101-379 mg l"1). With the determination of the mass-transfer coefficient and surface-reaction coefficient for a specified condition, a linear growth rate model for struvite growth determination in a fluidized bed reactor has been proposed (Chapter 4). Based on the results obtained from this study of struvite recovery from an anaerobic digester centrate, and using the results of a thermodynamic study, the pilot-scale struvite recovery reactor developed at UBC was found to be effective in recovering phosphate in the form of nearly pure struvite product. The desired degree of phosphate removal was achieved by maintaining the operating conditions between pH 8.0-8.2, and a recycle ratio (ratio between feed flow and recycle flow) 5.0-9.0, to control the supersaturation conditions inside the reactor. The performance of the system was found to be optimal when in-reactor supersaturation ratio was 2.0-6.0. Among several other operating parameters, apparent upflow velocity and magnesium to 205 phosphate molar ratio were also found important to maintain optimum system performance. Since the precipitation of other related compounds in the system were also controlled by kinetic factors, rather than thermodynamic solubility alone, the solid precipitates harvested were pure struvite, with undetectable impurities (Chapter 5). To determine the mechanisms of dissolution processes in a batch reactor system, two different theoretical models were tested. The experimental values were found to fit well with both models. The dissolution of struvite pellets was found to decrease with increasing pH within the pH range tested (4.38-6.05). In general, the dissolution rate constants were found to be lower for higher pH values. They were found to decrease with an increase in initial mass of struvite per volume of solution, for any tested pH value. In a mixed flow-through reactor system, the dissolution rates for struvite pellets were found to increase with the hydrogen ion concentration in the acidic pH due to proton-promoted dissolution. After a transition, the rate of dissolution increased again with an increase in pH, because of the hydroxyl- promoted dissolution (Chapter 6). The decomposition of struvite, under dynamic temperature conditions, was found to depend on the rate of heating. The decomposition occurred faster in the case of a slower heating rate. The simultaneous loss of both ammonia and water molecules from the struvite structure was found to occur gradually as a function of temperature, rather than as a distinct step. Boiling struvite in excess water resulted in the loss of five water molecules from its structure and transformed it into the monohydrate, dittmarite. The conditions for the transformation of struvite to dittmaritte could be exploited for intentional dittmarite formation; this would extend the field of P recovery from wastewater. However, the "economics" associated with this transformation 206 have yet to be proven as cost-effective. Further research, along these lines, is currently under way (Chapter 7). 8.3 Engineering significance The recovery of phosphorus has the potential to reduce operating costs by reducing reliance on chemicals, sludge disposal and downtime for cleaning struvite encrustations that occurs in the recycled in-plant streams, rich in phosphorus. The struvite produced annually from a wastewater treatment plant that processes 100 m3 d"1 of wastewater with a PO4-P concentration of only 7 mg l"1 and 55 % phosphorus recovery rate, would be sufficient to apply on 2.6 ha of arable land at an application rate 40 kg phosphorus as PaOs/ha/yr, as fertilizer. If struvite were to be recovered from wastewater treatment plants worldwide, 0.63 million tons of phosphorus (as P2O5) could be harvested annually, reducing phosphate rock mining by 1.6% (Shu et al., 2006). Therefore, this technology could provide opportunities to reduce wastewater treatment problems and recover phosphorus, sustainably, from waste streams, as well as preserve dwindling natural phosphorus reserves. In the commercialization process of the UBC MAP Crystallizer, Ostara Nutrient Recovery Technologies Inc. is paving the way with success in a number of experimental sites. In initial trials with the products from the UBC crystallizer on various types of turf, certain distinct advantages over currently available commercial fertilizers were ascertained (Scope Newsletter, 2006). Results from this study support this technology to be more widely accepted for sustainable development. The primary contributions of this research program to engineering practice and research include: • The determination of a thermodynamic solubility product of struvite for the standard state (1=0) and derivation of an analytical expression for the solubility product of struvite at 207 any temperature; this would help accurately predict struvite precipitation in fluidized bed reactors. The determination of a relationship between ionic strength and EC for the most potential source for phosphorus recover through struvite formation (anaerobic digester supernatant/centrate), with due consideration of temperature dependence of EC. It is suggested that this relationship would help in quickly estimating ionic strength of the solution, and thus in operating the reactor efficiently. It would also shed light on an ongoing research at UBC for automation and online control of the struvite crystallization system. Investigation of the precipitation and dissolution kinetics of struvite suggested that the kinetic experiments were supported by growth and dissolution models tested. The proposed growth models would help estimate growth under fluidized bed conditions, while the dissolution of struvite pellets grown, when used as a slow release fertilizer, can be predicted with the dissolution models tested. The successful application of the solubility and kinetic parameters, determined in the laboratory, in pilot-scale struvite recovery operation, has developed a basis for optimization of the system. The results found from the thermal decomposition of struvite, and its phase change with temperature will give insight for further research in the system dealing with higher temperatures, such as thermophilic digestion; the use of struvite and its fate under higher temperature conditions, such as composting is essential, especially, for nitrogen preservation. 208 8.4 Recommendations for future research As is the nature of research, the present study generated some exciting research questions that could not be addressed within the time allocated or were beyond the scope of this research work. Recommendations for future research and include: • Ammonia removal. Phosphorus, being the key parameter, is made limiting by adding MgCl2.6H20, based on the available phosphate in struvite crystallization processes. However, ammonia, which is also high in the anaerobic digester supernatant/centrate, is also removed to some extent, in the struvite crystallization process. The crystallization process removes and recovers phosphorus and lowers the precipitation potential of the effluent, leaving the ammonia level above the discharge limit in most cases. Limited research has been conducted on struvite crystallization from the perspective of ammonia removal (Siegrist, 1996; Tunay et al, 1997; Pace et al, 2000). Siegrist (1996) suggested that ammonia stripping and MAP precipitation are both feasible processes for eliminating ammonium from digester supernatant, while the cost of eliminating ammonium from the supernatant is higher than eliminating nitrogen, with denitrification. As such, there is a significant need for further research to couple an economically feasible process with the struvite crystallization system, that would remove additional ammonia from the solution. • Use of electrical conductivity. The relationship developed between EC and ionic strength will help estimate ionic strength on site. However, research should be conducted to see if EC can be used as an operational tool in struvite crystallization, especially investigating its effects on P-removal and other important parameters. • Optimization of process configuration. The results found in this study would focus on the optimization of the UBC MAP crystallization system. The pilot-study of struvite 209 crystallization from anaerobic digester supernatant/centrate resulted in nearly pure struvite, with undetectable impurities. However, a detailed study on the effect of different kinds of organic ligands (such as humic acids, fulvic acids and polysaccharides) in the struvite crystallization is highly recommended. Growth and agglomeration model. A detail study is recommended to verify the proposed growth model in FBR under different Conditions. The determination of the metastable limit for the struvite system is useful in process optimization. Since the metastable limit is a wastewater-specific phenomenon, determination of a metastable limit is recommended for each system, especially where the system is highly variable in terms of concentrations of the different contributing parameters. Agglomeration, being an important issue in pelletization of struvite inside the reactor, needs to be properly addressed. The ongoing research at UBC on the hydrodynamic condition inside the reactor can focus on the agglomeration phenomenon. A model, incorporating the agglomeration aspect, is recommended to optimize the system. P-recovery process coupled with Membrane Enhanced Biological Phosphorus Removal (MEBPR) process. The scope of merging P-recovery process with the MEBPR process to exploit the advantages of both technologies has been envisioned. Research is under way at UBC, to examine the process feasibility. The implementation of such technology in the MEBPR bioreactor has the potential to change substantially the layout and the operation of existing BNR Plants. Typically, municipal wastewaters do not contain sufficient carbon and, for a successful implementation of the enhanced biological phosphorus removal (EBPR) mechanism, the addition of extra volatile fatty acids (VFA), through a pre-fermented sludge stream, is often necessary. Even for those isolated cases of carbon- 210 rich wstewaters, the operation of EBPR process at long sludge retention time (SRT), entails the accumulation of a large mass of P in the bioreactor; this may result in high and prolonged peak effluent P concentrations in the event of process failure (Monti, 2005). A combined process, operated in the EBPR mode, with a side-stream P-recovery unit for struvite crystallization, is thought to be an innovative solution and warrants further study. Solubility of dittmarite. The conditions for the transformation of struvite to dittmaritte could be exploited for intentional dittmarite formation that would extend the field of P recovery from wastewater. A solubility and thermodynamic study of dittmarite is also recommended. A study to investigate the cost effectiveness associated with the struvite- dittmarite transformation is recommended. In addition to that, an investigation into the thermal decomposition of struvite, at a slower rate than the rates used in this study and at various constant temperatures, is recommended. Struvite in composting. The concept of struvite crystallization has a beneficial effect on the conservation of nitrogen in composting. Loss of nitrogen during composting may lead to the reduction of the agronomic value of the compost product, and it can also cause severe odor problems in full-scale composting facilities. The formation of struvite crystals, by adding Mg and P salts, results in a significant reduction of ammonia loss and increase in total ammonia nitrogen content of the compost (Bhuiyan et al, 2006). Further research on the application of struvite crystallization concept in composting is strongly recommended. 211 References Adnan, A., Koch, F.A. and Mavinic, D.S. (2003) Pilot-scale study of phosphorus recovery through struvite crystallization- II. Applying in-reactor supersatuartion ratio as a process control parameter. Journal of Environmental. Engineering and Science 2, 473-483. Bhuiyan, I.H., Mavinic, D.S. and Koch, F.A. (2006) Phosphorus recovery through struvite crystallization and its impact on biosolids. Proceedings of SWANA 21st Pacific Northwest Regional Symposium, April 5-7, Richmond, BC, Canada. Britton, A., Koch, F.A., Mavinic, D.S., Adnan, A. Oldham, W.K., and Udala, B. (2005) Pilot- scale struvite recovery from anaerobic digester supernatant at an enhanced biological phosphorus removal wastewater treatment plant. Journal of Environmental. Engineering and Science 4, 265- 277. Durrant, A.E., Scrimshaw, M.D., Stratful, I., and Lester, J.N. 1999. Review of the feasibility of recovering phosphate from wastewater for use as a raw material by the phosphate industry. Environmental Technology 26(5-6), 987-996. Monti, A. (2005) A Comparative Study of Biological Nutrient Removal Process with Gravity and Membrane-Liquid Separation, PhD Thesis, The University of British Columbia, Vancouver, BC, Canada. Pace, G., Berton, A., and Mantovani, A.(2000) Ammonia removal from aqueous solutions by MgNH4P04 6H20 precipitation. Annali di Chimica 90, 443-453. Scope Newsletter (2006) Phosphorus recycling, September, No.65, pp 2. Shu, L., Schneider, P., Jegatheesan, V., and Johnson, J. (2006) An economic evaluation of phosphorus recovery as struvite from digester supernatant. Bioresource Technology 97, 2211- 2216. Siegrist, H. (1996) Nitrogen removal from digester supernatant-comparison of chemical and biological methods. Water Science and Technology 34(1-2), 399-406. Tunay, O., Kabdasli, I., Orhon, D., and Kolcak, S. (1997) Ammonia removal by magnesium ammonium phosphate precipitation in industrial wastewaters. Water Science and Technology 36(2-3), 225-228. 212