Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Optimization of struvite pellet formation in a fluidized bed reactor : investigation of agglomeration… Fromberg, Marcia 2019

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2019_may_fromberg_marcia.pdf [ 8.79MB ]
Metadata
JSON: 24-1.0378385.json
JSON-LD: 24-1.0378385-ld.json
RDF/XML (Pretty): 24-1.0378385-rdf.xml
RDF/JSON: 24-1.0378385-rdf.json
Turtle: 24-1.0378385-turtle.txt
N-Triples: 24-1.0378385-rdf-ntriples.txt
Original Record: 24-1.0378385-source.json
Full Text
24-1.0378385-fulltext.txt
Citation
24-1.0378385.ris

Full Text

  OPTIMIZATION OF STRUVITE PELLET FORMATION  IN A FLUIDIZED BED REACTOR:  Investigation of Agglomeration and Growth Processes   by  Marcia Fromberg   B.A.Sc., The University of British Columbia, 2013   A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF   THE REQUIREMENTS FOR THE DEGREE OF    DOCTOR OF PHILOSOPHY  in  THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES  (Civil Engineering)   THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)    April 2019   © Marcia Fromberg, 2019      ii The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled: OPTIMIZATION OF STRUVITE PELLET FORMATION IN A FLUIDIZED BED REACTOR: Investigation of Agglomeration and Growth Processes  submitted by Marcia Fromberg in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Civil Engineering  Examining Committee: Donald S. Mavinic, Civil Engineering Supervisor  Victor Lo, Civil Engineering Supervisory Committee Member   Supervisory Committee Member Eric Hall, Civil Engineering University Examiner Lee Groat, Earth, Ocean, and Atmospheric Sciences University Examiner  Additional Supervisory Committee Members:  Supervisory Committee Member  Supervisory Committee Member    iii Abstract The precipitation of struvite (MgNH4PO46H2O) from phosphorus-rich wastewaters is well studied, but the agglomeration and growth mechanisms to form struvite pellets in fluidized-bed reactor (FBR) technologies are not well understood. Of primary concern is the unwanted production of fine struvite crystals that do not agglomerate and are generally lost to recovery. A pellet form is desirable as a final product; this can be used directly as a slow-release fertilizer and is important for the recovery process as it is easily separated from any colloidal material present in the wastewater.  The purpose of this research was to increase the fundamental knowledge of struvite pellet formation and mechanisms contributing to it within the UBC FBR system, with the goal to maximize growth rates while preventing fines losses. New methods were developed to understand struvite-pellet growth and morphologies under simulated FBR conditions. Precipitation experiments were used to determine induction times and zeta potential values of nucleating and growing crystals. A pilot UBC-FBR and lab-scale flow cells were used for separating pellet growth and agglomeration processes, and performing individual crystal and pellet growth experiments, respectively. Growth and morphology data was obtained from scanning electron microscope images. Fundamental knowledge gained from this work includes: crystal growth was the dominant process found in pellet formation; a preferential, pellet branching structure was identified; repulsive forces of precipitating and growing struvite changes as solution conditions change; agglomeration can be independently controlled in an FBR; radial growth rates for individual crystals increase with an increase in concentrations, supersaturation ratio (SSR) and fluid velocity; and individual pellet growth shows an increase with higher relative fluid velocities.  iv An overall FBR SSR between 2-6 is recommended to optimize the pelletization process. This will reduce crystal protrusions that could turn into fines while desirable pellet morphologies form, and also reduce repulsive forces that could prevent agglomeration. Higher relative fluid velocities can maximize growth rates and recover phosphorus at a faster rate, potentially reducing FBR footprints and/or changing technology designs. An alternate FBR operational process was proposed to maximize pellet growth and phosphorus recovery, while preventing fines losses by independently controlling growth and agglomeration.       v Lay Summary Recovery of phosphorus and nitrogen from wastewater into a fertilizer called struvite has been studied since the early 1990s. Many technologies have been developed; however, the basic science behind how the technologies work is not well understood. The patented UBC fluidized-bed reactor (FBR) recovers struvite as a pellet. The objective of this dissertation was to determine how these pellets form with the intention of optimizing the process and reducing process losses. New fundamental knowledge of struvite growth and crystal formations was gained. Processes within the FBR were separately controlled so optimal operational conditions could be determined. An alternate FBR operational process was designed to maximize pellet growth and phosphorus recovery while reducing losses.        vi Preface This dissertation has been presented in a manuscript-based format, and includes a collection of submitted, and draft manuscripts. All of the research-program design, experimental work, analysis of data, and writing pertaining to all chapters and/or manuscripts was carried out, and is the intellectual product of the author, Marcia Fromberg.  The Lawrence Livermore National Laboratory PHREEQC model database (file: thermo.com.V8.R6.230), modified by Dr. Sergey Lobanov, was used for modeling struvite equilibrium in this dissertation.  A version of Chapter 2 was submitted February 6, 2019 for publication as Fromberg, M., Pawlik, M., Mavinic, D.S., Induction time and zeta potential study of nucleating and growing struvite crystals for phosphorus recovery improvements within fluidized bed reactors. First author Marcia Fromberg prepared the first draft of the manuscript, and Professors Marek Pawlik (committee member) and Donald Mavinic (supervisor) advised on subsequent drafts and final editing.  Copyright Acknowledgments Images previously published by others were used in a visual literature review in Chapter 4, and throughout the chapter. Permission was granted for all previously published images in this work. The following table relates all the superscript reference numbers to specific figure headings within the text.       vii Superscript Number Figure Number 1-16 Figure 31 17 Figure 32 18 Figure 33 19 Figure 34 20 Figure 35 21 Figure 62  The following are the specific attributions relating to each copyright image used in the thesis: 1 Image republished with permission of John Wiley and Sons, from Abbona, F., & Boistelle, R. (1985). Nucleation of struvite (MgNH4PO46H2O) single crystals and aggregates. Crystal Res. Technol., 20, 133–40; permission conveyed through Copyright Clearance Center, Inc.  (Abbona & Boistelle, 1985)  2 Image republished with permission of PERGAMON, from Ali, M. I., & Schneider, P. A. (2006). A fed-batch design approach of struvite system in controlled supersaturation. Chemical Engineering Science, 61(12), 3951–3961; permission conveyed through Copyright Clearance Center, Inc.  (Ali & Schneider, 2006)  3 Image republished with permission of Canadian Science Publishing, from Britton, A., Koch, F. A., Mavinic, D. S., Adnan, A., Oldham, W. K., & Udala, B. (2005). Pilot-scale struvite recovery from anaerobic digester supernatant at an enhance biological phosphorus removal wastewater treatment plant. Journal of Environmental Engineering and Science, 4(4), 265–277; permission conveyed through Copyright Clearance Center, Inc.   4 Image republished with permission of Elsevier, from Chauhan, C. K., & Joshi, M. J. (2013). In vitro crystallization, characterization and growth-inhibition study of urinary type struvite crystals. Journal of Crystal Growth, 362(1), 330–337; permission conveyed through Copyright Clearance Center, Inc.  5 Image republished with permission of Elsevier, from Cusick, R. D., Ullery, M. L., Dempsey, B. A., & Logan, B. E. (2014). Electrochemical struvite precipitation from digestate with a fluidized bed cathode microbial electrolysis cell. Water Research, 54, 297–306; permission conveyed through Copyright Clearance Center, Inc. (Cusick, Ullery, Dempsey, & Logan, 2014) 6 Image republished with permission of Canadian Science Publishing, from Fattah, K. P., Mavinic, D. S., & Koch, F. a. (2012). Influence of Process Parameters on the Characteristics of Struvite Pellets. Journal of Environmental Engineering, 138(12), 1200–1209; permission conveyed through Copyright Clearance Center, Inc.     viii  7 Image republished with permission of Taylor & Francis, from Fattah, K. P., Mavinic, D. S., Koch, F. A., & Jacob, C. (2008). Determining the feasibility of phosphorus recovery as struvite from filter press centrate in a secondary wastewater treatment plant. Journal of Environmental Science and Health - Part A Toxic/Hazardous Substances and Environmental Engineering, 43(7), 756–764; permission conveyed through Copyright Clearance Center, Inc.   8 Image republished with permission of American Society of Civil Engineers, from Forrest, A. L., Fattah, K. P., Mavinic, D. S., & Koch, F. A. (2008). Optimizing Struvite Production for Phosphate Recovery in WWTP. Journal of Environmental Engineering, 134(5), 395–402; permission conveyed through Copyright Clearance Center, Inc.   9 Image republished with permission of Mineralogical Society of America, from Li, H., Yao, Q. Z., Yu, S. H., Huang, Y. R., Chen, X. D., Fu, S. Q., & Zhou, G. T. (2017). Bacterially mediated morphogenesis of struvite and its implication for phosphorus recovery. American Mineralogist, 102(2), 381–390; permission conveyed through MSA (Li et al., 2017) 10 Image republished with permission of American Society of Civil Engineers, from Ohlinger, K. N., Young, T. M., & Schroeder, E. D. (1999). KINETICS EFFECTS ON PREFERENTAIL STRUVITE ACCUMULATION IN WASTEWATER. Journal of Environmental Engineering, 125(8), 730–737; permission conveyed through Copyright Clearance Center, Inc.   11 Image republished with permission of American Chemical Society, from Prywer, J., & Torzewska, A. (2009). Bacterially induced struvite growth from synthetic urine: Experimental and theoretical characterization of crystal morphology. Crystal Growth and Design, 9(8), 3538–3543; permission conveyed through Copyright Clearance Center, Inc.   12 Image republished with permission of American Chemical Society, from Romanowski, Z., Kempisty, P., Prywer, J., Krukowski, S., & Torzewska, A. (2010). Density Functional Theory Determination of Structural and Electronic Properties of Struvite. Journal of Physical Chemistry, 114, 7800–7808; permission conveyed through Copyright Clearance Center, Inc.   13 Image republished with permission of Elsevier, from Ronteltap, M., Maurer, M., Hausherr, R., & Gujer, W. (2010). Struvite precipitation from urine - Influencing factors on particle size. Water Research, 44(6), 2038–2046; permission conveyed through Copyright Clearance Center, Inc. (Ronteltap, Maurer, Hausherr, & Gujer, 2010)  14 Image republished with permission of Elsevier, from Sun, W. D., Zhang, K. C., Wang, J. Y., & Wang, X. L. (2010). The chemical composition and ultrastructure of uroliths in Boer goats. Veterinary Journal, 186(1), 70–75; permission conveyed through Copyright Clearance Center, Inc. (Sun, Zhang, Wang, & Wang, 2010)  15 Image republished with permission of Elsevier, from Ye, X., Ye, Z. L., Lou, Y., Pan, S., Wang, X., Wang, M. K., & Chen, S. (2016). A comprehensive understanding of saturation index and upflow velocity in a pilot-scale fluidized bed reactor for struvite recovery from swine wastewater. Powder Technology, 295, 16–26; permission conveyed through Copyright Clearance Center, Inc.    ix 16 Image republished with permission of Elsevier, from Ye, Z., Shen, Y., Ye, X., Zhang, Z., Chen, S., & Shi, J. (2014). Phosphorus recovery from wastewater by struvite crystallization: Property of aggregates. Journal of Environmental Sciences (China), 26(5), 991–1000; permission conveyed through Copyright Clearance Center, Inc.   17 Image republished with permission of American Chemical Society, from Prywer, J., & Torzewska, A. (2009). Bacterially induced struvite growth from synthetic urine: Experimental and theoretical characterization of crystal morphology. Crystal Growth and Design, 9(8), 3538–3543; permission conveyed through Copyright Clearance Center, Inc.  18 Image republished with permission of Springer-Verlag, from Prywer, J., Torzewska, A., & Plociński, T. (2012). Unique surface and internal structure of struvite crystals formed by Proteus mirabilis. Urological Research, 40(6), 699–707; permission conveyed through Copyright Clearance Center, Inc.   19 Image republished with permission of Mark Holtkamp, Holtkamp, M. (2014). smorf crystal models. Retrieved December 11, 2018, from http://www.smorf.nl.  20 Image republished with permission of Mark Holtkamp, Holtkamp, M. (2014). smorf crystal models. Retrieved December 11, 2018, from http://www.smorf.nl.   21 Image republished with permission of Mark Holtkamp, Holtkamp, M. (2014). smorf crystal models. Retrieved December 11, 2018, from http://www.smorf.nl.            x  Table of Contents Abstract .......................................................................................................................... iii Lay Summary .................................................................................................................. v Preface........................................................................................................................... vi Table of Contents ............................................................................................................ x List of Tables ................................................................................................................. xv List of Figures............................................................................................................... xvi List of Abbreviations and Symbols ................................................................................ xxi Acknowledgments ........................................................................................................ xxii Dedication ................................................................................................................... xxiii Chapter 1 Introduction .................................................................................................. 1 1.1 Literature Review ................................................................................................ 1 1.1.1 Phosphorus and its Requirements for Life.................................................... 1 1.1.2 Uses of Phosphorus ..................................................................................... 2 1.1.3 Sources of Phosphorus ................................................................................ 2 1.1.4 Phosphorus Recovery Technologies ............................................................ 4 1.1.5 Benefits and Challenges with Struvite Recovery Technologies .................... 4 1.1.6 Chemistry of Struvite .................................................................................... 5 1.1.7 Mechanisms of Pellet Formation .................................................................. 7 1.2 Project Development & Research Gaps............................................................ 10 1.3 Research Objectives......................................................................................... 12 1.4 Dissertation Outline .......................................................................................... 13 Chapter 2 Induction Time and Zeta Potential Study of Nucleating and Growing Struvite Crystals for Phosphorus Recovery Improvements Within Fluidized Bed Reactors ......... 14  xi 2.1 Background ...................................................................................................... 15 2.2 Materials and Methods ..................................................................................... 17 2.2.1 Experimental Set-up .................................................................................. 17 2.2.2 Calculations ............................................................................................... 22 2.2.3 Analytical Methods ..................................................................................... 26 2.3 Results and Discussion .................................................................................... 28 2.3.1 Induction Time Study ................................................................................. 28 2.3.2 Aged Struvite Crystal Zeta Potential Experiments ...................................... 39 2.3.3 Struvite Nucleation Zeta Potential Experiments ......................................... 43 2.4 Chapter 2 Summary and Conclusions .............................................................. 52 Chapter 3 Decoupling Struvite Crystal Growth and Agglomeration Processes within an FBR System .................................................................................................................. 54 3.1 Background ...................................................................................................... 54 3.1.1 General Agglomeration .............................................................................. 54 3.1.2 FBR Agglomeration .................................................................................... 57 3.1.3 Observed Pellet Morphologies Produced in FBRs...................................... 58 3.1.4 Parameters Influencing Struvite Pellet Quality During Formation and Growth    .................................................................................................................. 60 3.2 Materials and Methods ..................................................................................... 63 3.2.1 Experimental FBR Setup and Operation .................................................... 63 3.2.2 Procedure for Crystal Addition ................................................................... 68 3.2.3 Analytical Methods ..................................................................................... 69 3.2.4 Calculations ............................................................................................... 70 3.3 Results & Discussion ........................................................................................ 71 3.3.1 Control Run Summary ............................................................................... 71 3.3.2 Crystal Addition Summary .......................................................................... 74  xii 3.3.3 Comparison of Control and Crystal Addition ............................................... 91 3.3.4 Clarifier Particle Size Distribution ............................................................... 96 3.3.5 Pellet Development .................................................................................... 98 3.3.6 Hard Pellet Surfaces ................................................................................ 100 3.4 Chapter 3 Summary and Conclusions ............................................................ 101 Chapter 4 Struvite Crystal Growth and Morphologies: The Influence of Concentrations, SSR and Velocity ........................................................................................................ 104 4.1 Background .................................................................................................... 105 4.1.1 Struvite Morphology ................................................................................. 105 4.1.2 Struvite Growth ........................................................................................ 111 4.2 Materials and Methods ................................................................................... 115 4.2.1 Stationary Crystal Growth ........................................................................ 115 4.2.2 Rotational Crystal Growth ........................................................................ 119 4.2.3 Bulk FBR Pellet Growth and Internal Structures ....................................... 120 4.2.4 Image Analysis ........................................................................................ 121 4.2.5 Calculations ............................................................................................. 121 4.2.6 Analytical Methods ................................................................................... 123 4.3 Results and Discussion .................................................................................. 124 4.3.1 Stationary Crystal Growth ........................................................................ 124 4.3.2 Rotational Crystal Growth ........................................................................ 147 4.3.3 FBR Bulk Pellet Growth Rate and Internal Structures .............................. 153 4.3.4 Comparison of Growth Rates for Limited Conditions ................................ 157 4.4 Chapter 4 Summary and Conclusions ............................................................ 159 Chapter 5 Engineering Application and FBR Operation ............................................ 160 Chapter 6 Conclusions and Recommendations ........................................................ 165 6.1 Conclusions .................................................................................................... 165  xiii 6.1.1 General .................................................................................................... 165 6.1.2 Nucleation ................................................................................................ 165 6.1.3 Surface Charge ........................................................................................ 165 6.1.4 Crystal Growth ......................................................................................... 166 6.1.5 Struvite Pellet Agglomeration and Growth ................................................ 167 6.2 Recommendations for Future Work ................................................................ 167 6.2.1 General .................................................................................................... 167 6.2.2 Nucleation ................................................................................................ 168 6.2.3 Surface Charge ........................................................................................ 169 6.2.4 Crystal Growth ......................................................................................... 169 6.2.5 FBR Operation ......................................................................................... 170 References .................................................................................................................. 171 Appendices ................................................................................................................. 183 Appendix A Canadian Fertilizer Imports for 2017 .................................................. 183 Appendix B XRD Results for Synthetic Struvite Crystals Used in Aged Zeta Potential Experiments  .......................................................................................................... 184 Appendix C Graph of Mixing Speed Influence on Induction Time ......................... 185 Appendix D XRD Results from Various Nucleation Experiments .......................... 186 Appendix E Zeta Potential Data ............................................................................ 187 Appendix F FBR Control Data .............................................................................. 202 Appendix G FBR Crystal Addition Data ................................................................ 206 Appendix H Struvite Mass Removed from Control FBR ........................................ 210 Appendix I Struvite Mass Removed from Crystal Addition FBR ............................ 211 Appendix J FBR Velocities Based on Configuration .............................................. 212 Appendix K Stationary Crystal Growth Experimental Data .................................... 213 Appendix L Stationary Growth Rate Data and Calculations .................................. 215  xiv Appendix M Growth Coefficient Graphs ................................................................ 219 Appendix N Published Struvite Growth Rates and Coefficients............................. 220 Appendix O Stationary Crystal Growth Morphology Data ...................................... 221 Appendix P Rotational Pellet Growth Experimental Data ...................................... 225         xv  List of Tables Table 1 Average synthetic wastewater compositions and targeted SSR values for the nucleation zeta potential experiments .................................................................... 20 Table 2 Laser & pH probe induction time analysis of differences: The Bland-Altman Method (units in seconds) ...................................................................................... 36 Table 3 Regression line intercepts - the influence of RPM on induction time .............. 38 Table 4 Dissolution of struvite crystals into NaCl and NH4Cl solutions ........................ 41 Table 5 Experimental data grouping and statistical analysis ....................................... 45 Table 6 Crystal addition amounts ............................................................................... 75 Table 7 Pellet Weights................................................................................................ 90 Table 8 Difference in harvested struvite masses between control and crystal addition FBRs (g) ................................................................................................................ 94 Table 9 Growth coefficients for varying conditions .................................................... 131 Table 10 Single crystal unit cell measurements/XRD results .................................... 144 Table 11 Comparable stationary and rotational growth rates .................................... 158         xvi  List of Figures Figure 1. FBR function in each section ....................................................................... 12 Figure 2. Laser induction time determination example: Normalized laser intensity of transmitted light for SSR 9, RPM 400 Run 1 a) entire experimental data set b) detail of the laser induction time determination ............................................................... 29 Figure 3. pH induction time determination example: SSR 9, RPM 400, Run 1 a) entire experimental data set b) detail of pH induction time determination ........................ 31 Figure 4. Struvite induction times a) raw data; b) data fit to nucleation model ............. 32 Figure 5. Analysis for homogeneous versus heterogeneous nucleation for pH probe induction time data ................................................................................................ 33 Figure 6. Laser and pH probe induction time analysis of differences .......................... 35 Figure 7. Crystal size variation between: a) Run 5, SSR 6; b) Run 4, SSR 9; c) Run 3, SSR 12 .................................................................................................................. 39 Figure 8. Comparison of zeta potential measurements in relation to pH from (Le Corre et al., 2007) final solution ionic concentrations unknown at room temperature; (Bouropoulos & Koutsoukos, 2000) solution ionic concentrations of 0.01 M NaCl saturated wrt struvite at 25°C; (Z. Ye et al., 2014) final solution ionic concentrations unknown at 23-25°C; (Prywer et al., 2015) ionic strength calculated from conductivity as 0.31-0.29 M of unknown ion concentrations at 37°C; and current study averages in solution ionic concentrations of NaCl and dissolved struvite as 0.0133-0.0134M at 25°C.  .................................................................................................................... 43 Figure 9. Zeta potential nucleation experiment 102 (induction time as dashed line) ... 44 Figure 10. Wastewater composition group regression line comparisons ..................... 47 Figure 11. Initial SSR and time of zero charge for wastewater composition 1 ............. 50  xvii Figure 12. Wastewater 1, high SSR zeta potential data (average induction time as vertical dotted line) ................................................................................................ 51 Figure 13. FBR configuration ...................................................................................... 65 Figure 14. Detail of FBR lower section ....................................................................... 66 Figure 15. (a) [Sample3_bse_01] Outer surface of pellet from control run 40x magnification; (b) [Sample14_bse_01] 2mm pellet cut-away from control run 25x magnification; (c) [Sample14_lvse_03] Edges of the cut pellet 250x magnification; (d) [Sample4_lvse_02] Seed hopper junction fines from control 150x magnification.       ................................................................................................................ 72 Figure 16. Experimental FBR operational parameter timeframes ............................... 75 Figure 17. (a) [Sample49_bse_03] 2.0 mm pellet surface after 1.5 hours of crystal addition 100x magnification; (b) [Sample49_bse_04] Agglomerates from pH probe sampling port 100x magnification; (c) [Sample51_bse_02] 2.0mm pellet surface from day 3 before start of crystal addition 100x magnification; (d) [Sample51_bse_05] 2.0mm pellet cut open from day 3 before start of crystal addition 100x magnification.  ................................................................................................................. 78 Figure 18. (a) [Sample53_bse-03] 2.0mm cut pellet from day 4 before crystal addition 100x magnification; (b) [Sample55_bse_02] agglomerates at seed hopper junction from day 4 before crystal addition 100x magnification; (c) [Sample56_bse_02] agglomerates at seed hopper junction after 4 hours of crystal addition from day 4 100x magnification; (d) [Sample56_bse_03] inset of (c) at 500x magnification. ..... 80 Figure 19. Outer surface of pellets from (a) [Sample57_bse_01] day 5 pH probe sample of a pellet after two hours of crystal addition 50x magnification; (b) [Sample69_bse_03] day 5 500x magnification; (c) [Sample60_bse_04] day 9 before crystal addition and after 2 days on recycle flow 50x magnification; (d) [Sample7_bse_01] 2.0mm pellet from day 12 25x magnification; (e)  xviii [Sample16_bse_01] day 12 harvested 2mm pellet from crystal addition ultrasonic in methanol 25x magnification; (f) [Sample7_bse_03] 1000x magnification of d. ....... 82 Figure 20. Seed hopper junction samples (a) [Sample58_bse_02] Day 5 100x magnification; (b) [Sample58_bse_03] inset in a 250x magnification; (c) [Sample64_bse_03] day 10 at 100x magnification; (d) [Sample64_bse_04] inset in c 250x magnification; (e) [Sample8_bse_02] day 12 150x magnification; (f) [Sample_8_bse03] inset in e 400x magnification. .................................................. 84 Figure 21. Day 2 hourly liquid samples ....................................................................... 86 Figure 22. Day 3 hourly liquid samples ....................................................................... 86 Figure 23. Day 4 hourly liquid samples ....................................................................... 87 Figure 24. Day 5 hourly liquid samples ....................................................................... 87 Figure 25. Difference in effluent phosphorus concentration from the control ............... 89 Figure 26. Comparison of the control and crystal addition FBR SSRs ........................ 92 Figure 27. Particle size distribution of clarifier fines compared to the injected small crystals  ................................................................................................................. 97 Figure 28. [Sample13_bse_01] Resin mounted pellet at 24x magnification ................ 98 Figure 29. Pellet development in FBR systems with primary nucleation occurring ...... 99 Figure 30. Hard Ostara pellet surface at (a) [Sample47_bse_01] 42x magnification; (b) [Sample47_bse_03] 600x magnification; (c) [Sample12_bse_02] cut and polished pellet at 1000x magnification. .............................................................................. 101 Figure 31. Review of struvite morphologies from literature; laboratory observations, and process or in-situ observations ............................................................................ 106 Figure 32. Coffin shaped struvite crystal morphology along axis: a) b-a, b) b-c, c) a-c. Image from (Prywer & Torzewska, 2009)  ............................................................ 108 Figure 33. Plate structures identified by Prywer et al (2012)  .................................... 109 Figure 34. Pyramidal 3D Model 36 of struvite by (Holtkamp, 2014)  ......................... 110  xix Figure 35. Pyramidal 3D Model 47 of struvite by (Holtkamp, 2014)  ......................... 110 Figure 36. Varying diffusion fields around a precipitating and growing crystal .......... 114 Figure 37. Stationary crystal growth: crystal holder in flow cell ................................. 116 Figure 38. Rotational pellet growth experimental setup ............................................ 120 Figure 39. Full-mix concentration growth examples .................................................. 125 Figure 40. No-mix concentration growth examples ................................................... 126 Figure 41. Full-mix radial growth rates ...................................................................... 127 Figure 42. No-mix radial growth rates ....................................................................... 128 Figure 43. Full-mix interpolated surface plot of growth rates ..................................... 129 Figure 44. No-mix interpolated surface plot of growth rates ...................................... 130 Figure 45. a) Original SEM pellet image [Sample_1_bse_1] and b) converted to binary image  ............................................................................................................... 133 Figure 46. Binary image of branch sides .................................................................. 134 Figure 47. Full-mix morphologies and experimental times in minutes ....................... 136 Figure 48. No-mix morphologies and experimental times in minutes ........................ 137 Figure 49. Example of multiple morphologies at low SSR and low velocity: (a) before (b) after growth i) pyramidal crystals ii) edge growth iii) smoothening of rough surface and forming pyramidal habit [FMLF-15] ............................................................... 139 Figure 50. Example of no growth on large flat face a) before b) after growth [NMHF-27]    ................................................................................................................ 140 Figure 51. Growth influence from seed crystal experiment 17; a) before growth b) after growth [FMHF-17] ................................................................................................ 140 Figure 52. Development of pyramidal 6-sided habit with the (00 1 ) face extending outwards, Scale bar 100 μm across both images [FMLF-16] ............................... 141  xx Figure 53. Protrusion growth at high SSRs a) experiment 14 [image Protrusions FMLF-14] b) experiment 28 [image Protrusions NMHF-28] c) experiment 3 [image Protrusions NMHF-3] ........................................................................................... 143 Figure 54. Pellet radial growth rates ......................................................................... 148 Figure 55. Interpolated surface plot of pellet radial growth rates ............................... 149 Figure 56. Pellets glued to needle ............................................................................ 150 Figure 57. New morphology with 6-sided top a) side view [Needle2_afterExp3_lsve_007] b) top view [Needle3_afterExp6_lsve_013] ......... 151 Figure 58. Clusters of small crystals [Needle1_afterExp5_lsve_005] ........................ 152 Figure 59. Comparison of pellet growth from a) experiment 2 [PelletGrowth_Exp2_after_lsve_002] b) experiment 7 [Needle4_afterExp7_lsve_002] ............................................................................ 153 Figure 60. Cut pellet with branching structures [Sample 75-03] ................................ 155 Figure 61. a) Multi-branch formation [Sample 77-03]; b) Side of branch [Sample 77-05]    ................................................................................................................ 155 Figure 62. Typical crystals in branch formation [Sample 76-02], comparable to pyramidal model inset from (Holtkamp, 2014)  ..................................................................... 156 Figure 63. Branch development from kink site or damage areas a) before growth b) after growth [NMHF-27] ............................................................................................... 157 Figure 64. Dimensions and retention time comparison of a 2.54 and 7.60 cm diameter FBR  ............................................................................................................... 161 Figure 65. Proposed FBR system ............................................................................. 164     xxi  List of Abbreviations and Symbols CV  Coefficient of Variation EF  Effluent filtered sample (mg/L) EU  Effluent unfiltered sample (mg/L) FBR  Fluidized bed reactor HRT  Hydraulic retention time (minutes) IAP  Ion activity product KSP  Solubility product PHREEQC Aqueous equilibrium geochemical modeling software package created by the US Geological Survey RPM Revolutions per minute SEM  Scanning Electron Microscope SHF Seed hopper filtered sample (mg/L) SHU Seed hopper unfiltered sample (mg/L) SSR  Supersaturation ratio TSS  Total suspended solids WWTP  Wastewater treatment plant ζ  Zeta potential (mV)     xxii Acknowledgments I would like to thank, from the bottom of my heart, my wonderful husband Sean Ellickson who supported me emotionally and financially through this endeavor. I would never have accomplished so much in life without him.  Thank you to my supervisor Dr. Donald S. Mavinic, for his support and guidance while at UBC, and my PhD committee members Dr. Marek Pawlik, Dr. Bernard Laval, and Dr. Victor Lo who provided valuable assistance and feedback.  I would like to thank Dr. Sergey Lobanov for all his teachings about struvite and PHREEQC equilibrium modeling, and all the inspiration and great times experimenting with new ideas and processes.  Thank you to Dr. Noboru Yonemitsu, for his teachings of fluid dynamics, sensors and microprocessors, and all the help and advice building my apparatuses.  Thank you to the Environmental Engineering Lab personnel Paula Parkinson, Tim Ma, and Otman Abida who were always eager to offer advice and to help with equipment, sample analysis, and discuss results.  I would like to thank Laís Mazullo M. Pereira from Universidade Federal de Pernambuco in Brazil who volunteered to assist with the laboratory nucleation experiments, and Sean Larson , MASc student, for assisting with the laboratory aged zeta potential experiments.   xxiii Dedication This work is dedicated to the late UBC Research Associate Frederic A. Koch. He inspired me to do bigger and better things, and to help make the world a better place with our environmental engineering work. Without Fred’s enthusiasm, guidance, and mentorship in the early stages of my work, this dissertation would never have been accomplished. I am forever in his debt.        1 Chapter 1 Introduction Phosphorus is a required component to all life on the planet. It is extracted from the ground and turned into various commodities. One major product it is turned into is fertilizer, which is required for mass production of food to feed our ever-growing global population. Phosphorus is a non-renewable resource, and it is suggested we are on the way to running out of it (Ashley et al., 2011). Technologies to recover phosphorus from wastewater, in the form of magnesium-ammonium-phosphate, called struvite, are proven, but require optimization to reduce process losses. The following review will discuss the importance of phosphorus, present the various aspects of current recovery-technologies, and summarize our knowledge of struvite formation. Research objectives are then outlined from the highlighted research gaps. 1.1 Literature Review 1.1.1 Phosphorus and its Requirements for Life Whether it is a plant or animal, organisms would not survive without phosphorus. It is in cell membranes, deoxyribonucleic acid (DNA), ribonucleic acid (RNA), and in the energy-carrying compound adenosine triphosphate (ATP) of living organisms. The 0.7 kg of phosphorus in an average adult human body (Ashley et al., 2011), or approximately 1% by weight (Corbridge, 2013), is integrated into our bones, tissues, and fluids from the food we eat. Once adulthood is reached, very little phosphorus from our food is utilized in our bodies, and it is estimated 0.40-0.55 kg of phosphorus/person per year is excreted (Corbridge, 2013; Jönsson et al. 2004). Sanitary sewer and treatment systems collect this excreted phosphorus, and if not removed, deposit it into water bodies where it causes eutrophication problems. Once the phosphorus is discharged into surface waters it is removed and virtually lost from the natural nutrient cycle of the land.   2 1.1.2 Uses of Phosphorus Not only is phosphorus in all living organisms, but it is also used for many other purposes in our society. Phosphoric acid is produced from acid leaching of phosphorus rich ore, and is the main starting ingredient for producing most phosphorus-containing compounds. It is used directly in the production of fertilizer, chemicals and food products, for metal treatment, and in fuel cells. Phosphorus is also widely used in products such as food and beverages, detergents, metal surface-treatments, pigments, glass and glass coatings, cements, flame-retardants, pesticides, medicines, materials used in dental repairs, pyrotechnics, catalysts for many chemical processes, metals, electronic equipment, and many other items too numerous to list (Corbridge, 2013).  Fertilizer is the largest use of phosphorus today. Since World War II, crop yield has increase by 50% by acre in part due to fertilization (Corbridge, 2013). Commercial fertilizer has allowed human populations to flourish with this increase in food production. The problem with most applied fertilizers is that plants only take up about 20-25% of the phosphorus fertilizer added to the soil; the remainder becomes fixed in the soils as insoluble compounds, unavailable for uptake, or it runs off into nearby water bodies (Corbridge, 2013). This fixing or loss of the phosphorus from soils then requires continual replenishment in the form of fertilizers to grow more food. 1.1.3 Sources of Phosphorus Sources of phosphorus used to produce fertilizers are typically ore deposits of sedimentary marine phosphorites, which take millions of years to form. The largest reserves are found in northern Africa, China, the Middle East, and the United States. Many countries are reliant on imports of phosphorus as P2O5 for fertilizers and industrial uses. In 2017, 45.7 million tonnes of it was used worldwide and an estimated 48.8 million tonnes will be used in 2021 (US Geological Survey, 2018). World resources of phosphate rock  3 are estimated to be over 300 billion metric tonnes; reserves are 70 billion tonnes; mine production (excluding China) for 2017 was estimated at 147 million tonnes; and there is the expectation that 168 million tonnes will be mined in 2021 (US Geological Survey, 2018). There are some questions about the validity of the reserve estimates, quality of the ore, and how long the finite supply will last (Ashley et al., 2011). There are no phosphorus deposits found in Canada, and the country is dependent on imports of phosphorus each year for fertilizer requirements. In 2017, approximately 1.86 million tonnes of various fertilizers containing phosphorus were imported into Canada at a value of $848 million, the breakdown of fertilizer types and values can be seen in Appendix A (Government of Canada, 2018). With human population expected to increase from the current 7.6 billion people to 9.8 billion people in 2050 (United Nations, 2017), there will be a greater need for high production farming and phosphorus fertilizers.  Phosphorus recovery from wastes has become increasingly more viable, and could reduce the dependency on phosphate ores. It is estimated that human urine and feces, if collected, could supply 22% of global phosphorus demand (Mihelcic et al., 2011); this percentage would significantly increase if livestock wastes were included. Harvesting phosphorus in the form of struvite (magnesium-ammonium-phosphate hexahydrate – MgNH4PO46H2O) from municipal or agricultural wastewater has proven a successful way to recover waste phosphorus. Struvite is a slow-release fertilizer due to its low solubility, and when mixed into soil the roots of the plants activate the release of nutrients when they require it; therefore, minimizing harmful runoff. The nitrogen-to-phosphorus-to-potassium ratio for struvite (in percentage) is approximately 5-29-0 + 10% magnesium. One application of struvite can provide the phosphorus requirement for a crops entire growing season (Ostara Nutrient Recovery Technologies Inc., 2017). Recovering phosphorus from wastewater not only reduces eutrophication of water bodies, it also mitigates buildup of  4 struvite in wastewater treatment piping and equipment; therefore, recovering it turns a problem into a product. 1.1.4 Phosphorus Recovery Technologies The technologies for recovering struvite from wastewater consist of various shapes and sizes of stirred reactors and fluidized bed reactors (FBR). The main technologies in use today for recovering struvite from wastewater are the AirPrex, Crystalactor, Multiform HarvestTM, OstaraTM, PHOSNIX, and PhospaqTM processes. All of these technologies are a form of FBR although differ in of shape, size, air addition, mixing regime, multiple or single pass, seed utilization, type of wastewater being treated, or upstream process variations (Oleszkiewicz et al., 2015).  The Phosphorus Recovery Group at the University of British Columbia developed, patented, and commercialized a novel FBR in the early 2000s (Koch et al., 2008), and created the spinoff company OstaraTM. As of December 2018, this technology has been installed at 18 wastewater-treatment facilities throughout North America and Europe (Ostara Nutrient Recovery Technologies Inc., 2017), and is the technology discussed and utilized in this work.  1.1.5 Benefits and Challenges with Struvite Recovery Technologies The main benefit of using an FBR in recovering phosphorus is its ability to produce a struvite pellet of sufficient size as to separate it from colloidal material found in wastewater. Success with the UBC FBR has produced high-quality pellets, ranging between 0.5 mm up to 5 mm in diameter, or even larger if required, valued for specialized fertilizer applications (Fattah et al., 2012). Struvite pellet formation is currently accepted to be an agglomeration, or aggregation, of smaller crystals, but one problem that many FBR technologies face is the unwanted production of fine struvite crystals that do not agglomerate. These crystals get washed out of the reactors, where they are lost to  5 recovery, and can account for 10-30% of struvite losses depending on FBR operational conditions (Shimamura et al., 2007; Ye et al., 2016).  Before phosphorus recovery can take place from municipal wastewaters, a biological process is used to concentrate the phosphorus. This process, called biological phosphorus removal, utilizes phosphorus-accumulating organisms to remove the phosphorus from the wastewater and store it in cell biomass or biosolids. Specific conditions are required to grow this biomass, which is then removed from the treatment system, and digested to release the concentrated phosphorus stored within it. This phosphorus and nitrogen-rich liquid expelled from the biomass, called centrate, is then used for struvite recovery. Approximately 55-65% of the phosphorus is released in biosolids digestion, based on volatile solids destruction (Tchobanoglous et al., 2003). If an average of 15% of fines is lost from FBRs, it is estimated that only 47-55% of the total phosphorus in the biosolids is recovered. Newer technologies are being developed to extract more phosphorus from biosolids with the use of microwave-oxidation processes (Lo et al., 2017), but with the inefficiencies of phosphorus release from digested biomass and the loss of fines from FBRs, a large portion of the available phosphorus in wastewater is not being recovered; therefore, optimization of the struvite recovery process is essential. 1.1.6 Chemistry of Struvite Struvite is a transparent to semi-transparent crystalline material made up of equal molar concentrations of magnesium, ammonium, and phosphate with six waters of hydration. Struvite precipitation is controlled by the saturation of solution with respect to ions that form the crystals lattice, and precipitates as per Equation 1  𝑀𝑔2+ +  𝑁𝐻4+ +  𝐻𝑛𝑃𝑂4𝑛−3 + 6𝐻2𝑂 →  𝑀𝑔𝑁𝐻4𝑃𝑂4 ∙  6𝐻2𝑂 ↓ +𝑛𝐻+ (1) where n = 1, 2, or 3 H+ ions depending on the pH of the solution.  6 1.1.6.1 Solubility Product and Supersaturation Ratio The solubility product (Ksp) is used in the determination of how saturated the solution is, and is defined as the equilibrium constant of a reaction when precipitate ions dissolve or form to equalize within a solution as per Equation 2. It is often expressed as pKsp for ease of expression as per Equation 3, and is calculated by the ion activities at equilibrium. Temperature is the main condition that affects the equilibrium constant value (Snoeyink & Jenkins, 1980).   𝐾𝑠𝑝(𝑠𝑡𝑟𝑢𝑣𝑖𝑡𝑒) =  {𝑀𝑔2+}{𝑁𝐻4+}{𝑃𝑂43−}{𝐻2𝑂}6{𝑀𝑔𝑁𝐻4𝑃𝑂4 ∙ 𝐻2𝑂(𝑠)}                        =  {𝑀𝑔2+}{𝑁𝐻4+}{𝑃𝑂43−} (2)   𝑝𝐾𝑠𝑝 =  −𝑙𝑜𝑔10(𝐾𝑠𝑝) (3) The ion activity product (IAP), the product of the ion activities in solution at a specific time, is used in the calculation of supersaturation ratio (SSR), and is compared with the solubility product to determine the solution saturation according to Equations 4 and 5 (Ali & Schneider, 2008; Stumm & Morgan, 1981). The SSR represents the extent of crystallization that must occur in order for the system to reach equilibrium, and can be described as the driving force of the precipitation process.  𝐼𝐴𝑃(𝑠𝑡𝑟𝑢𝑣𝑖𝑡𝑒) =  {𝑀𝑔2+}1{𝑁𝐻4+}1{𝑃𝑂43−}1 =  {𝑀𝑔2+}{𝑁𝐻4+}{𝑃𝑂43−}  (4)   𝑆𝑆𝑅(𝑠𝑡𝑟𝑢𝑣𝑖𝑡𝑒) =  {𝑀𝑔2+}{𝑁𝐻4+}{𝑃𝑂43−}𝐾𝑠𝑝(𝑠𝑡𝑟𝑢𝑣𝑖𝑡𝑒)=𝐼𝐴𝑃𝐾𝑠𝑝 (5)   IAP > Ksp (supersaturated) IAP = Ksp (equilibrium, saturation) IAP < Ksp (undersaturated)   7 There are many reported equilibrium constants in the literature, but recent work has determined a new struvite pKsp value of 13.47, which will be used in this project (Lobanov et al., 2013). All struvite SSR values have been modeled using PHREEQC software. PHREEQC is a free aqueous geochemical modeling software package created by the US Geological Survey, which can be used for a wide range of modeling of chemical reactions within aqueous solutions. 1.1.7 Mechanisms of Pellet Formation  Struvite pellet formation is currently accepted to be an agglomeration or aggregation of smaller crystals. The precipitation of struvite from phosphorus-rich wastewaters is well studied, but the agglomeration and growth mechanisms for struvite to form into a pellet are not well understood. Nucleation, crystal growth, and agglomeration are suggested as main components of pellet formation, so a better understanding of each of their roles is required to understand the processes and mechanisms that occur in FBRs. These processes occur simultaneously in FBRs; therefore, a methodical approach is required to separate and analyze them individually. 1.1.7.1 Struvite Nucleation Understanding nucleation is important for the formation of new crystals, as well as knowing the induction time or how long it takes to achieve new crystal formations. Struvite nucleation studies typically measure induction times with a drop in pH of the solution (Bhuiyan et al., 2008; Bouropoulos & Koutsoukos, 2000; Koutsoukos & Kofina, 2007; Mehta & Batstone, 2013), which does not take into account the size of the nuclei at observation. The induction time and visual observations of crystals within the solution have been reported to be at the same time for the pH method in one study (Bhuiyan et al., 2008). This raises concern with the appropriateness of the method because if the crystals are visible they are fairly large and may not be classified as nuclei. It also suggests that  8 reported values may be incorporating crystal growth into induction time values. No method to date for struvite has reported a size corresponding to an induction time. The size of forming crystals is important for agglomeration potential, not only for fitting within the pellet matrix, but also for measuring any crystal repulsive forces preventing agglomeration. Additional induction time information could enable the design of growth or agglomeration-zone locations within an FBR, or where nuclei would be located. It is important to know the struvite induction time to reduce excess nucleation and fines production within FBR systems. The FBR promotes nucleation and growth in a plug-flow regime, similar to desupersaturation in a constant volume; therefore, understanding nucleation is essential for any FBR process design. 1.1.7.2 Struvite Surface Charge Surface charge of nucleating and growing struvite crystals could have a great effect on pellet formation. As particles increase in size, attractive forces diminish (Pietsch, 1991), so smaller crystals potentially have a greater chance of overcoming repulsive forces if their solution environment does not influence them.  The measurement of zeta potential (ζ) is a means of estimating repulsive forces between particles in solution. Struvite ζ has been previously investigated by a number of researchers, and reported values for aged struvite in suspension ranges between +32 to -35 mV depending on solution composition, but most reported values are negative (Bouropoulos & Koutsoukos, 2000; Forrest, 2004; Koutsoukos & Kofina, 2007; Le Corre, Valsami-Jones et al., 2007; Liu, 2009; Prywer et al., 2015; Vol’khin et al., 2015; Z. Ye et al., 2014). These extreme values indicate the solution systems require investigation to determine how various conditions affect struvite repulsive forces with respect to agglomeration in FBRs.  9 1.1.7.3 Struvite Agglomeration in FBRs Little is known about the mechanisms behind struvite agglomeration. Problems with agglomeration of fine struvite crystals has been associated with the large negative zeta potential values, especially at higher pH values (Le Corre et al., 2007). General agglomeration literature states, the tendency to agglomerate increases if the particles have a large size distribution where voids can be filled with smaller particles, and then solid bridges can develop (Pietsch, 1991). The main factors that control agglomeration are reported to be: level of supersaturation; suspension density; particle size; degree of agitation; ionic strength; and presence of impurities (Jones, 2002).  Nucleation and growth of agglomerates in the same reactor are not ideal, whereas traditional FBRs tend to limit the particle size distribution. This can be better controlled if a series of FBRs feed each other as they will keep in the smaller particles that have a higher chance to agglomerate (Pietsch, 1991). This idea is seen in the UBC FBR design where the internal diameter increases in stages as the height in the FBR increases, resulting in classifying the pellets, but also allowing different sizes of pellets and crystals to mix at the stepped zones (Koch et al., 2008).  Four main parameters are found in literature that contribute to the pellet quality during formation: the FBR SSR; fluid velocities; magnesium concentration; and impurities or solids in the wastewater. The pellet morphologies change depending on these parameters. Loose aggregated pellets have been reported in a recent study analyzing up-flow velocity and saturation index (X. Ye et al., 2016). Harder, smooth pellets have also been produced (Fattah et al., 2012; Ostara Nutrient Recovery Technologies Inc., 2017; X. Ye et al., 2016; Z. Ye et al., 2018), but the reasons for these differences are still not fully understood. Cut pellets reveal a variety of internal characteristics: orthorhombic shapes growing from the center outwards, worn away tops of the crystals (Britton et al., 2005);  10 inner cores of the pellets not as compact as the outer surfaces (Fattah et al., 2012; X. Ye et al., 2016; Z. Ye et al., 2018); a jumbled arrangement of various crystal sizes and shapes agglomerated together (Huang et al., 2006); and rings of varying growth and compactness (Fattah et al., 2012; X. Ye et al., 2016). It is still unclear as to how to control the formations. 1.1.7.4 Crystal Growth Crystal growth encompasses both growth rates and the habits or morphologies that can develop with varying conditions. There are numerous studies for bulk growth of struvite crystals, but none on single crystals. Some examples of struvite growth-rates in literature have been done in bulk in a stirred reactor (Ariyanto et al., 2014; Galbraith et al., 2014; Harrison et al., 2011; Mehta & Batstone, 2013), an FBR (Bhuiyan et al., 2008), or in-situ with coupons in a wastewater treatment plant (Ohlinger et al., 1999). Typical bulk estimates in stirred reactors utilize SSR decay, or desupersaturation at constant volume. This is an inappropriate method to determine growth for any continuous industrial application, like an FBR, and does not provide much detail into the actual morphologies of the crystals. However, understanding how long growth occurs using desupersaturation will provide information for sizing reactors for residence time calculations. There are many forms that struvite can take: pellets; agglomerates; single crystals; or multiple crystal formations. These forms change depending on the system and solution conditions. As described above, FBR SSR and velocities affect the quality of the pellets in different ways. Precipitation of struvite has been well studied with respect to SSR, but not in conjunction with velocity. If a higher fluid velocity changes the morphology of struvite pellets, it must also change the morphology of the struvite crystals making up the pellets.  1.2 Project Development & Research Gaps Initial research for this project came from a group brainstorming-event, led by the author of this dissertation. The event included the late Research Associate (and co- 11 inventor of the UBC FBR technology), Mr. Frederic Koch, visiting scholar Dr. Sergey Lobanov, and MASc student Mr. Connor Wilson. The UBC FBR system was analyzed and broken-down into sections to determine what was understood, and what was unknown about the processes occurring in each section of the FBR. Through this brainstorming and further investigation common struvite knowledge and knowledge gaps were defined as follows: i. Struvite thermodynamic properties were well documented and understood; ii. Pellets consist of agglomerated crystals; iii. Little was known about struvite agglomeration processes in FBRs and how to control them; iv. Turbulence within the FBR was thought to play a major role in agglomeration; v. Surface charge was thought to influence agglomeration; vi. It was unknown what process was producing the fine crystals that are being lost in the FBR effluent; vii. Physical properties of individual struvite crystals were not well understood; viii. Dye tests confirmed complete mixing in the injector port of the FBR, the dye travelled upwards in a plug-flow regime, so plug-flow is assumed throughout; and ix. No systematic process breakdown was ever carried out. All the struvite growth studies were carried out in bulk, stirred-reactors, or in an FBR system, which is too complicated to properly control and study. Some of the gaps defined above led to further investigation into FBR processes, and review of the pertinent literature. Preliminary research operating pilot FBRs determined what occurs to the velocity, size of pellet, and growth rate in each sequential section of the FBR. Figure 1 visually illustrates how all variables decrease with an increase in height and diameter in the FBR.    12                                         Figure 1. FBR function in each section 1.3 Research Objectives The purpose of this dissertation was to increase the fundamental knowledge of struvite pellet formation and mechanisms contributing to it within the UBC FBR system, with the goal to maximize growth rates while preventing fines losses. A series of objectives were defined as follows: i. To develop a method to relate struvite crystal size to induction times; ii. To assess various ionic strengths/wastewater types and SSR values on the surface charge of nucleating and growing struvite crystals; iii. To separate crystal growth and agglomeration processes in the UBC FBR; iv. To develop a method for, and assess the effects of SSR, ionic strength, and fluid velocity on individual crystal growth rates and crystal morphologies; and v. To propose a new process design, or FBR operational procedure in which growth rates are maximized and fines losses are prevented.  13 1.4 Dissertation Outline As mentioned in the Preface, this dissertation is written in a manuscript format with a total of 6 chapters. The individual manuscript-chapters are in an order that highlights the progression of the research performed to determine struvite pellet formations in an FBR system.  Chapters 2 through 5 contain the technical aspects to this dissertation. Chapter 1 has provided the background including, a short literature review common to all chapters, which highlights the gaps in the research and provides the framework for the objectives. Each chapter has a more detailed review pertaining to specific aspects of the research performed. Chapter 2 investigates struvite nucleation while developing a size-dependent method to indicate struvite induction times. Similar experiments were then conducted with 5 wastewater compositions to assess the surface charge of nucleating and growing struvite. The results from Chapter 2 were used in Chapter 3 to separate growth and agglomeration processes within an FBR system, and to suggest optimal agglomeration conditions. Chapter 4 delves into the area of individual crystal growth rates and morphologies. This chapter investigates how velocity, SSR, and varying concentrations influence stationary and rotational crystal growth. Chapter 5 applies the findings from Chapter 2 to 4 to two pilot FBRs, to illustrate how the data can be used for FBR design. An FBR operational procedure is then recommended. Chapter 6 summarizes the main findings from the individual manuscript-chapters and relates each together. Overall conclusions and recommendations are presented for future work on understanding struvite pellet formation.     14 Chapter 2 Induction Time and Zeta Potential Study of Nucleating and Growing Struvite Crystals for Phosphorus Recovery Improvements Within Fluidized Bed Reactors  *A version of this Chapter has been submitted on February 6, 2019 for publication: Fromberg, M., Pawlik, M., Mavinic, D.S., Induction time and zeta potential study of nucleating and growing struvite crystals for phosphorus recovery improvements within fluidized bed reactors.  The precipitation of struvite from phosphorus-rich wastewaters has been well studied, but the agglomeration and growth mechanisms for struvite to form into a pellet are not well understood. Success with fluidized-bed reactors (FBR) has produced agglomerates large enough to separate from colloidal material found in wastewater, and has produced high-quality pellets valued for specialized fertilizer applications (Fattah et al., 2012). One problem that many technology developers face is the unwanted production of fine struvite crystals that do not agglomerate. These crystals get washed out of the reactors where they are lost to recovery, and can account for 10-30% of struvite production depending on FBR operational conditions (Shimamura et al., 2007; X. Ye et al., 2016). Dissolving the outgoing fines back into solution and returning them back into the process is one answer to the problem (Britton et al. , 2015; Shimamura et al., 2007). Attempts have been made to utilize coagulants to flocculate struvite fines (Le Corre, 2006) and binders to pelletize fines (Latifian et al., 2012). Others put harvested or specific sized fines back into the reaction column as seed, which continue to grow into pellets of larger size (Shimamura et al., 2007; Ueno & Fujii, 2001). This may work during research with synthetic wastewater where there are no colloids to interfere, but often is not feasible with biological material that is in municipal or agricultural wastewaters.  15 With the inefficiencies of phosphorus release from digested biosolids, and the loss of fines from phosphorus recovery technologies, a good portion of the available phosphorus in wastewater is not being recovered. If it is assumed that 55-65% of the phosphorus is released in biosolids digestion, based on volatile solids destruction (Tchobanoglous et al., 2003); and, if an average of 15% of this phosphorus is lost as fines from FBRs, it is estimated that only 47-55% of the total phosphorus in the biosolids is recovered. Newer technologies are being developed to extract more phosphorus out of the biosolids (Lo et al., 2017), but a better solution to fines control is required and this was the motivation for studying agglomeration and growth mechanisms within FBR systems.  2.1 Background Struvite pellet formation is currently accepted to be an agglomeration of smaller crystals; therefore, studying the development of these crystals is important in the understanding of pellet formation. The first step in crystal formation is nucleation. Nuclei are defined as very small, new crystals formed from a supersaturated solution, and the time it takes to form nuclei is called the induction time. The main focus in the present study was primary homogeneous nucleation, which occurs when molecules or groups of molecules come together on their own in solution. This is different from secondary or heterogeneous nucleation, which forms onto crystals or particles in solution, and primary nucleation generally requires higher supersaturation than secondary nucleation. Struvite induction times have typically been measured by a drop in pH readings of the solution (Bhuiyan et al., 2008; Bouropoulos & Koutsoukos, 2000; Koutsoukos & Kofina, 2007; Mehta & Batstone, 2013); others have used turbidity measurements (Triger et al., 2012), or visual observation with light scintillations (Ohlinger et al., 1999).  Other nucleation studies for different calcium compounds have used transmitted and scattered light (Lancia & Musmarra, 1999; Mazziotti di Celso et al., 2017). There is no  16 standardization between these methods or any reference to crystal size at observation. For struvite, as one molecule forms it releases hydrogen ions causing a decrease in pH; hence, the use of pH as an indicator of solid phase formation. It is important to know the struvite induction time to reduce excess nucleation within FBR systems. The pH indication of induction time and visual observations of crystals within the solution have been reported to occur at the same time (Bhuiyan et al., 2008); this raises concern with the appropriateness of the method because if the crystals are visible, they are fairly large and may not be classified as nuclei. It also suggests that reported values may be incorporating crystal growth into induction time values. The higher the SSR, the greater number of nuclei are produced and the smaller the crystal size, but it is unknown whether all of the crystals will agglomerate. Lower recovery rates have been reported when high SSR conditions are used (Shimamura et al., 2007), and high SSRs and large concentrations of crystals also seem to prevent agglomeration and growth (Britton et al., 2015). A consideration for agglomeration is the crystal’s surface charge and repulsive forces between crystals. For agglomeration to take place, any repulsive force acting between crystals must be smaller than attractive forces. The measurement of zeta potential (ζ) is a means of estimating repulsive forces between particles in solution. Struvite ζ has been previously investigated by a number of researchers, and reported values for aged struvite in suspension range anywhere between +32 to -35 mV depending on solution composition, but most reported values are negative (Bouropoulos & Koutsoukos, 2000; Forrest, 2004; Koutsoukos & Kofina, 2007; Le Corre et al., 2007; Liu, 2009; Prywer et al., 2015; Vol’khin et al., 2015; Z. Ye et al., 2014). Large ζ values, either positive or negative, indicate a stable suspension according to the Derjaguin-Landau-Verwey-Overbeek (DLVO) theory of colloidal systems (Derjaguin & Landau, 1941; Verwey & Overbeek, 1948), and could be the explanation for difficulties with struvite agglomeration processes. No research to date has been able to define ζ process parameters for FBR  17 operations to improve agglomeration processes. There are also no known studies in which the ζ has been measured for nucleating and growing struvite. Since the in-situ FBR operating solution conditions change with respect to column height, the ζ is also likely to change throughout the column creating areas with preferential agglomeration conditions.  The objectives of this work was to relate a specific size of struvite crystal to induction time, and to determine the ζ values of precipitating and growing struvite from solution. It was hypothesized that the ζ values would change as the precipitation reaction proceeded, and these values could be utilized for FBR process design and optimal agglomeration conditions within FBRs.  2.2 Materials and Methods The first step of this work was to develop a size-dependent laser sensor for struvite nucleation studies. The new laser method was compared to pH induction time measurements. The relationship between mixing speed and induction time was investigated. Nucleation experiments were then performed to include ζ measurements of nucleating and growing struvite crystals to determine SSR influence for reducing inter-crystal repulsive forces. Aged struvite ζ values were measured and compared to reported literature values. Results were compared to FBR conditions and operational parameters are suggested for greater agglomeration to occur. 2.2.1 Experimental Set-up 2.2.1.1 Induction Time Experiments Experiments were carried out within an 11.5 cm square beaker made of clear Plexiglas with a magnetic stirrer, including a digital RPM readout. Two chemical solutions were prepared using distilled water and analytical grade chemicals; one contained ammonium dihydrogen phosphate and ammonium chloride (NH4H2PO4 and NH4Cl), and the other contained magnesium chloride (MgCl26H2O). The pH of the two solutions was  18 adjusted to the required SSR value with a 1.0 M or 0.1 M sodium hydroxide solution, and 500 mL of each solution were mixed together to achieve concentrations of 800 mg/L of N-NH4, 100 mg/L of P-PO4, and 80 mg/L of Mg in the final 1000 mL solution. This mixing method reduces any high concentration gradients, which would otherwise trigger local nucleation at higher SSR values, and assured only homogeneous nucleation was initially taking place. An alternative method to estimate struvite induction time was utilized along with the traditional method of a decrease in pH. The first method used a laser of approximately 625-700 nm, a 650 nm narrow bandpass filter (638-672 nm), and sensor to detect the onset of struvite nucleation by a decrease in the solution transmissivity. The first indications by the laser are from crystals approximately the same size as the laser wavelength reflecting the light away from the sensor. A 650 nm struvite crystal would be approximately 1000 unit-cells in size. The laser and sensor with the filter were positioned on either sides of the square beaker with the beam path length through the solution equal to 11.5 cm. This long path length ensured high sensitivity. A potentiometer was used to standardize experiments and to establish a baseline for the initial mV reading through the ammonium and phosphate solution, prior to the addition of the magnesium solution and initial mixing. The baseline was set at approximately 900 mV at the beginning of each experiment. The laser mV output, pH, and temperature were logged at 0.5-second intervals using a simple PC and an Arduino microprocessor, and conductivity was measured manually with a hand-held Horiba pH/conductivity D-54 meter, in conjunction with sample collection.  Solution samples of 50 mL volume were collected after complete mixing of the stock solution, and at the end of the experiment to calculate SSR values. Final solution samples were filtered with 0.45 μm filter. The 50 mL samples were acidified with one or two drops of a 1:1 HCl:H2O solution. Conductivity was constant at 7.92 mS/cm. Solid  19 samples were collected at the end of the experiment by filtering the remaining suspension with a 2 μm glass filter paper.  Five to seven experimental runs for RPM values of 200, 300, 400, 500, and 600 for each of the initially estimated SSR values of 1.5, 3, 6, 9, and 12 were performed. Experiments were run for 30-40 minutes each. Temperature was not specifically controlled, but on average was 23.0°C ± 2.0°C with individual SSR values calculated independently, and similar SSR runs performed on the same day with temperatures within ± 1.0°C of each other.  2.2.1.2 Nucleation Zeta Potential Experiments Experiments were performed with five different synthetic wastewater compositions corresponding to different municipal wastewater treatment processes. The compositions assumed anaerobic digestion to produce centrate, and the nitrogen and phosphorus concentrations represent typical values from: 1) Non-biological phosphorus removal process (typical Annacis Island WWTP centrate);  2) Biological removal process with thermophillic digestion (highest centrate concentrations);  3) Digested conventional activated sludge (lowest centrate concentrations); 4) Nitrogen-limited centrate with high phosphorus content (mid-stage process concentrations from ammonia recovery process using thermal decomposed struvite as source of P and Mg); and 5) Composition with low overall concentrations.   The experimental procedure used for nucleating struvite, and measuring and logging the data was exactly the same as in Section 2.2.1.1 with the exception of the different chemical compositions outlined in Table 1. Analytical grade 85% phosphoric acid  20 solution was used to obtain higher phosphorus concentrations in composition 2 and 4. Targeted SSR values of 3, 6, 9, and 12 were used to represent a range of FBR conditions. Two SSR values were used for wastewater compositions 1, 2, and 4 to examine how the ζ differed by changing only the pH; composition 3 and 5 were compared to each other to investigate the influence of ammonium. A minimum of two experiments for each composition and SSR combination was attempted. All successful experimental runs can be seen in Table 5 in the results section 2.3.3. Table 1 Average synthetic wastewater compositions and targeted SSR values for the nucleation zeta potential experiments Composition # N (mg/L) P (mg/L) Mg (mg/L) N:P Ratio Mg:P Ratio Target SSR 1 770 90 80 18.9 1.1 6 & 12 2 980 520 365 4.2 0.9 6 & 9 3 450 40 40 24.9 1.3 6 only 4 430 520 580 1.8 1.4 3 & 9 5 345 40 40 19.1 1.3 9 only  Temperature was not precisely controlled, but averaged 24.0°C ± 1.0°C. Solution samples of 50 mL volume were collected and filtered with a 0.45 μm filter after complete mixing of the stock solutions and between each ζ measurement to calculate SSR values. The 50 mL samples were acidified with one or two drops of a 1:1 HCl:H2O solution. A 10-15 mL sample was injected into a Malvern Zetasizer 2000 as fast as it could process the samples; this resulted in no set time period between measurements. The Zetasizer for each injected sample performed three ζ measurements; the cell was rinsed with distilled water between each sample injection and with a 5% HCl solution between experiments. Experiment labeling consisted of the wastewater composition number 1 through 5 as the  21 hundreds digit, and the experiment run number is in the tens and one digit. For example experiment 127 has composition 1, and experiment run number 27. 2.2.1.3 Aged Crystal Zeta Potential Experiments Zeta potential measurements for aged struvite crystals were undertaken in an indifferent electrolyte of 0.01 M NaCl between a target pH of 9.4 to 10.0.  A solution of NH4Cl was also used at molar concentrations of 0.0036 M, 0.0134 M, and 0.0226 M, relating to target values of 200, 700, 1200 mg/L of NH4Cl at a pH of 8.0. Solutions were prepared with distilled water and analytical grade NaCl and NH4Cl salts. A minimum of three experimental runs per solution-pH combination was carried out. Each test consisted of 200 mL of solution in a beaker initially adjusted to the corresponding pH value, using analytical grade sodium hydroxide (NaOH) in distilled water; then 0.25 grams of synthetic struvite crystals were mixed into the solution, and the beaker was placed into an ultrasonic bath for five minutes. The crystal suspension was then allowed to settle for 15 minutes. The pH of the solution was measured without stirring right before injecting a 10-15 mL sample into a Malvern Zetasizer 2000. Total time for solution aging was 20 minutes, with 15 minutes for settling. The temperature was maintained at 25°C. Initial samples were taken from stock solutions for ionic strength confirmation; a 50 mL sample from each test was collected and filtered with a 0.45 μm filter and acidified with one or two drops of a 1:1 HCl:H2O solution for struvite dissolution analysis. Chloride concentrations were measured at the end of each experiment with the remaining solution, for ionic strength confirmation.  A particle size distribution analysis of the aged solutions was performed as a separate test, but utilized the same procedure as the aged ζ tests. A Malvern Mastersizer Hydro 2000S was used to measure the particle distribution. Triplicate runs were performed for the 0.01 M NaCl solutions at a pH of 9.6, and for each NH4Cl solution concentration at a pH of 8.0. Initial particle size distribution d(0.1), d(0.5), d(0.9) of the synthetic struvite  22 crystals used in the experiments was measured at 4.1, 29.8, 298.5 μm to determine settling time requirements. The composition of the crystals was analyzed for purity and molar ratios, and found not to be 100% pure struvite. Molar ratios of Mg:N:P were 1.02:1.0:1.12, and XRD analysis (see Appendix B) indicated a mixture of struvite and another magnesium phosphate, dittmarite (NH4MgPO4H2O).  2.2.2 Calculations 2.2.2.1 Solubility Product and Supersaturation Ratio Struvite precipitation and growth depends on how saturated the solution is with ions that form the crystal lattice. The solubility product (Ksp) is used in the determination and is defined as the equilibrium constant of a reaction when precipitate ions dissolve, or form to equalize within a solution. It is often expressed as pKsp for ease of expression, and is calculated by the ion activities at equilibrium. Temperature is the main condition that affects the equilibrium constant value (Snoeyink & Jenkins, 1980). The pertinent equations are outlined in Equations 6-8.   𝑀𝑔2+ +  𝑁𝐻4+ +  𝑃𝑂43− + 6𝐻2𝑂 =  𝑀𝑔𝑁𝐻4𝑃𝑂4 ∙ 6𝐻2𝑂 (6)   𝐾𝑠𝑝(𝑠𝑡𝑟𝑢𝑣𝑖𝑡𝑒) =  {𝑀𝑔2+}{𝑁𝐻4+}{𝑃𝑂43−}{𝐻2𝑂}6{𝑀𝑔𝑁𝐻4𝑃𝑂4 ∙ 6𝐻2𝑂(𝑠)}                        =  ({𝑀𝑔2+}{𝑁𝐻4+}{𝑃𝑂43−})𝑒𝑞𝑖𝑙𝑖𝑏𝑟𝑖𝑢𝑚 (7)   𝑝𝐾𝑠𝑝 =  −𝑙𝑜𝑔10(𝐾𝑠𝑝) (8) The ion activity product (IAP), the product of the ion activities in solution at a specific time, is used in the calculation of supersaturation ratio (SSR), and can be compared with the solubility product (Ksp) to determine the solution saturation (Ali & Schneider, 2008; Stumm & Morgan, 1981), according to Equations 9 and 10.  𝐼𝐴𝑃(𝑠𝑡𝑟𝑢𝑣𝑖𝑡𝑒) =  {𝑀𝑔2+}1{𝑁𝐻4+}1{𝑃𝑂43−}1 =  {𝑀𝑔2+}{𝑁𝐻4+}{𝑃𝑂43−}  (9)  23   𝑆𝑆𝑅(𝑠𝑡𝑟𝑢𝑣𝑖𝑡𝑒) =  {𝑀𝑔2+}{𝑁𝐻4+}{𝑃𝑂43−}𝐾𝑠𝑝(𝑠𝑡𝑟𝑢𝑣𝑖𝑡𝑒)=𝐼𝐴𝑃𝐾𝑠𝑝 (10) IAP > Ksp (supersaturated) IAP = Ksp (equilibrium, saturation) IAP < Ksp (undersaturated)  There are many reported equilibrium constants in the literature, but recent work has determined a new struvite pKsp value of 13.47, which will be used in this project (Lobanov et al., 2013). All struvite SSR values have been modeled using PHREEQC software. 2.2.2.2 Nucleation and Induction Time According to classical nucleation theory, the homogeneous nucleation rate (J) in units (nuclei/cm3 s) for three-dimensional nuclei can be described according to Equation 11 (Abbona & Boistelle, 1985; Bouropoulos & Koutsoukos, 2000; Jones, 2002; Ohlinger et al., 1999)   J =  Ω 𝑒𝑥𝑝 (−𝛽𝛾𝑆3𝑉𝑚2(𝑘𝑇)3(𝐿𝑜𝑔𝑆𝑆𝑅)2 ) (11) It is assumed the nucleation time is much longer than the growth time during the nucleation period (tn>>tg), and using the statistical concept of nucleation, the nucleation rate is related to induction time by Equation 12 (Ohlinger et al., 1999).  τ = 1𝐽 (12) Where Ω is a kinetic factor described by Equation 13   Ω =  (𝐷𝑑5𝑁∗) (4∆𝐺∗3𝜋𝑘𝑇)12 (13) Where D is the diffusion coefficient; d is the interplanar distance in the crystal lattice; N* is the number of molecules comprising a critical size nucleus; ∆𝐺∗ is the Gibb’s free energy  24 change to form a critical nucleus; β is a shape factor (=32 for cubic shape of the nucleus); γs is the surface energy; Vm is the molar volume of the nucleating solid [molecular weight/(Avogadro's Number x Density x Number of Ions in a Formula Unit) = 7.99 x 10-23 cm3)]; k is the Boltzmann’s constant; T is the absolute temperature. Combining Equations 11 and 12, the homogeneous nucleation induction time can then be calculated with Equation 14.   τ = 1Ωexp (𝛽𝛾𝑆3𝑉𝑚2(2.3𝑘𝑇)3 𝜈2(𝐿𝑜𝑔𝑆𝑆𝑅)2 ) (14) Taking the log base-10 transformation and simplifying will then yield Equation 15 (Ohlinger et al., 1999).  log 𝜏 =𝐴(Log 𝑆𝑆𝑅)2− 𝐵  (15) The heterogeneous nucleation induction time can be computed by adding a correction factor 𝑓 to the homogeneous induction time Equation 14. Taking the log base-10 transformation results in 𝐴 = 𝛽𝛾𝑆3𝑉𝑚2𝑓/(2.3𝑘𝑇)3 𝜈2 , and 𝐵 = log Ω if 𝑓 = 1, and 𝐵 =log Ωℎ𝑒𝑡  if 𝑓 < 1 . Heterogeneous nucleation requires lower activation energy than for homogeneous nucleation, and this correction factor changes the slope of the induction time Equation 15.  2.2.2.3 Electrophoresis Electrophoresis has been the only reported technique used for measuring struvite ζ, where it measures the particle velocity. Dividing the velocity by the applied electric field results in the electrophoretic mobility. This mobility is then used directly to calculate the ζ i.e. the net charge on the particle according to Equations 16-22:  𝑉 =  𝑈𝐸𝐸 (16)   𝑉𝐸⁄ =   𝑈𝐸 = (2𝜀𝜁 3𝜇⁄ ) ∙ 𝑓(𝜅a) (17)  25 Assuming a low surface charge, the Debye- Hückel parameter κ is defined as (Hunter, 1981; Kosmulski, 2009):  𝜅 =  √𝑒2𝑁𝐴 ∑ 𝑐𝑖𝑖 𝑧𝑖2𝜀𝑘𝑇 (18)   𝜅 =  3.2881 ∙ 109 ∙ √𝐼  (𝑢𝑛𝑖𝑡𝑠 𝑜𝑓 𝑚−1 𝑎𝑡 25°C) (19) where 𝑉  is the measured particle velocity; UE is the electrophoretic mobility; 𝐸  is the applied electric field; ε is the permittivity of the solution; ζ is the zeta potential; μ is the viscosity of the solution; 𝑓(𝜅𝑎)  is the Henry equation; a is the particle radius; e the electronic charge; NA the Avogadro constant; k the Boltzmann constant; T is the temperature of 298K (25°C); I is the ionic strength in units of mol/dm3  (mol/L). For 𝜅a << 1, 𝑓(𝜅a) ≈ 1, for very low ionic concentrations, the Hückel equation applies according to Equation 20  𝑈𝐸 = (2𝜀𝜁 3𝜇⁄ ) (20) For 𝜅a >> 1, 𝑓(𝜅a) ≈ 1.5, for high ionic strength that has compressed the double layer, the Smoluchowski equation applies according to Equation 21  𝑈𝐸 = (𝜀𝜁 𝜇⁄ ) (21)   𝐼 =  12∑ 𝑧𝑖2𝑐𝑖𝑖 (22) where zi is the ionic valence of each ion in solution; and ci is the concentration of each ion in solution. 2.2.2.4 Attractive and Repulsive Forces Between Crystals The attractive and repulsive forces between struvite crystals were analyzed to determine if one was more dominant than the other, and to provide some insight into struvite solution stability criteria. The total force (VT) between interacting particles is the  26 sum of the van der Waals attractive forces (VA) and the repulsive electrostatic forces (VR). These forces are described by the DLVO theory of colloidal stability, and are the basic and main forces experienced by particle systems given by Equation 23 (Derjaguin & Landau, 1941; Verwey & Overbeek, 1948).   𝑉𝑇 =  𝑉𝐴 +  𝑉𝑅 (23) The van der Waals attractive force between particles in a solution can be calculated by Equation 24  𝑉𝐴 =  −𝐴 ∙ 𝑅12𝑑 (24) where A is the Hamaker constant; R is the radius of identical spherical particles; and d is the separation distance of particles (for very short distances).  Repulsive force calculations differ depending on the solution ionic strength compressing the electrostatic double-layer. Equation 25 is used when the double-layer is substantially large (𝜅𝑑 < 5)  (Verwey & Overbeek, 1948); this criteria was confirmed during the analysis.   𝑉𝑅 =  2𝜋𝜀𝑅𝜓𝑂2  𝑒𝑥𝑝(−𝜅𝑑) (25) For low potentials (<25mV), where the electrical energy is small compared to the thermal energy, Equation 26 can be used to determine the surface potential value (ψO) for the above equation.  𝜓 =  𝜓𝑂  𝑒𝑥𝑝(−𝜅𝑥) (26) where ε is the dielectric constant of the medium; ψO is the surface potential; ψ is the potential at x distance away from the surface (if x is assumed to be the shear plane distance, then ψ can estimate the zeta potential); and κ is the Debye-Hückel parameter. 2.2.3 Analytical Methods Liquid samples were analyzed for orthophosphate and ammonia by using a LACHAT QuickChem FIA+ 8000 Series with a ASX 500 Series Autosampler, according to  27 UBC method EE-SOP# A.04.01 ‘Ammonia, Nitrate and Phosphate Determination in Water by Continuous Flow Analysis’ with reference to Standard Method 4500-P G and 4500-NH3 H (Eaton et al., 1995). Sodium and magnesium were analyzed by atomic absorption spectroscopy with a 220 Fast Sequential Varian Spectra AA with a Varian Autosampler model SPS-5, in accordance with Standard Method 3111 B (Eaton et al., 1995). Chloride concentrations for the aged struvite crystal ζ experiments were measured with a chloride-specific ion electrode and millivolt meter. pH and conductivity were measured with a hand-held Horiba pH/conductivity D-54 meter. Solid struvite samples were analyzed to confirm crystalline phases by powder X-ray defraction with a Bruker D8-Advance X-ray Deffractometer using copper (Cu) Kα radiation, and for molar ratio confirmation by dissolving in hydrochloric acid solution. Images were captured with an Environmental Scanning Electron Microscope (SEM) FEI Quanta 650 VP using the secondary electron detector (indicated by LFD on the image) and the backscatter detector (indicated by CBS on the image) in low vacuum mode. The Malvern Zetasizer 2000 program was set to standard settings for water at 25°C, the pH value was set to 7.0 and not adjusted for each experiment. The Smoluchowski equation for high ionic strengths was validated (κa>>1) for both the nucleation and aged struvite crystal experiments. This was confirmed by calculating κ using two ionic strengths representing the lowest (0.0061 mol/L) and highest (0.1744 mol/L) ionic strength from all of the ζ experiments. The smallest crystal size from any experiment (a=650 nm) was then used to calculate κa. These values resulted in the lowest possible κa of 166 and 893, compared to larger crystal sizes, which would only increase the value of κa.    28 2.3 Results and Discussion 2.3.1 Induction Time Study 2.3.1.1 Induction Time Determination Laser induction times were determined from the time of initial mixing of the two solutions until an indication of a change in the solution with a reduction in the transmitted laser beam. The laser induction times were determined for each experiment using a simple algorithm from the logged laser mV data; this consisted of smoothing out the data by a five-point running average, and then the induction time was determined from a decrease in three consecutive mV values from the initial baseline reading. The experiments for SSR values of 1.5 did not nucleate in the 30-40 minute timeframe; therefore, they are not included in the analysis. Figure 2a and b illustrate the smoothed curve of the laser-intensity (Itrans) normalized by the initial laser baseline (I0) for an SSR value of 9 and RPM of 400 for one experiment. The vertical line, after the initial mixing at t=0, indicates the induction time of 140 seconds for the particular experiment. It is apparent from the significant slope that the tangent line is signaling a change in the laser intensity. This change is the time at which the laser beam is beginning to be reflected by the growing struvite crystals, around 650nm in size. There were no visual observations of precipitates at the time when the laser indicated the initial reduction in transmitted light.  During initial data analysis, a non-biased way of determining the laser induction times was explored. The laser output data was fit with a tangent to the main slope of the data, and the baseline was drawn out to intersect with it. The point of intersection would be the induction time from this analysis, but it revealed a large discrepancy from when the laser mV value started to change. For the same experiment example given in Figure 2,  29 the laser induction time would be more than 100 seconds after the laser first indicated a change in tranmissivity, and was deemed unacceptable, and therefore, not used.     Figure 2. Laser induction time determination example: Normalized laser intensity of transmitted light for SSR 9, RPM 400 Run 1 a) entire experimental data set b) detail of the laser induction time determination  0.60.650.70.750.80.850.90.9511.051.1-50 450 950 1450 1950I trans/I 0Time (s)a)0.940.960.9811.021.041.06-50 50 150 250 350I trans/I 0Time (s)BaselineInitial      MixingInduction Timeb) 30 Measured induction times for the pH probe method were estimated from the time of initial mixing at t=0 of the two solutions until an initial decrease from the peak pH value was observed. Figure 3a and b show the time series data for the pH method of sensing the induction time for same experiment as show in Figure 2 for the laser method.  Figure 4a is a combination of all of the experimental induction times measured by the laser and the pH probe methods, and it can be seen that the induction time is inversely proportional to the SSR. Figure 4b illustrates a fairly good fit to the logarithmic nucleation model, Equation 15, with the coefficients of determination (R2) being more than 0.76 for both the pH probe and laser data.        31    Figure 3. pH induction time determination example: SSR 9, RPM 400, Run 1 a) entire experimental data set b) detail of pH induction time determination   5.76.26.77.27.78.2-50 450 950 1450pHTime (s)Induction Timea)5.76.26.77.27.7-50 150 350 550pHTime (s)Induction Timeb) 32   Figure 4. Struvite induction times a) raw data; b) data fit to nucleation model  The pH probe data was further analyzed for the possible occurrence of heterogeneous nucleation as others have suggested could occur (Bouropoulos & Koutsoukos, 2000). The data was split at 1/(LogSSR)2 = 2 arbitrarily where two groupings 02004006008001000120014002.00 7.00 12.00 17.00Induction Time (s)SSRLaser pH Probea)y = 0.5822x + 1.3275R² = 0.78088y = 0.3153x + 2.1236R² = 0.7661311.522.533.540.5 1 1.5 2 2.5 3Log(Induciton Time)1/(LogSSR)^2Laser pH Probeb) 33 of data appeared to change slope. Trendlines were fit to each group. Figure 5 illustrates the poor fit of the linear regression lines with 0.39 and 0.16 R2 values for the split data compared to the combined data trendline with an R2 value 0.76. From this analysis there is no distinction between homogeneous and heterogeneous nucleation, but based on the method used in this work where high concentration gradients were limited it is unlikely that heterogeneous nucleation occurred. Other researchers with fewer data for struvite, but over a slightly larger range have reported two linear parts that suggest a separation between homogeneous and heterogeneous nucleation (Bouropoulos & Koutsoukos, 2000). This difference from literature observations may be a result of this large data set in which small variations in the regression line slope could be concealed. However, a larger data set usually provides a better representation of average values and avoids errors from small sample sizes.   Figure 5. Analysis for homogeneous versus heterogeneous nucleation for pH probe induction time data  y = 0.3902x + 2.0383R² = 0.3594y = 0.3153x + 2.1236R² = 0.7661y = 0.1966x + 2.4043R² = 0.157622.22.42.62.833.23.40 0.5 1 1.5 2 2.5 3 3.5 4Log (Induction Time) (s)1/(Log SSR)^2 34 2.3.1.2 Laser and pH Probe Induction Time Method Comparison A comparison of the time difference between the nucleation indications from the laser and the pH probe methods was carried out. It can be seen in Figure 4a that there is a significant difference in the two methods used for sensing induction time. The pH probe regression line is shifted up from the laser regression line, indicating the pH probe takes longer to sense nucleation.  The slopes of the linearized regression lines in Figure 4b were found to be statistically different, so a comparison of intercepts is not applicable to determine the differences. Therefore, an analysis of differences between the pH-probe and laser sensing methods for all of the nucleation experiments was carried out to determine if one method was superior. The Bland-Altman method (Giavarina, 2015) for comparing two methods was used in this study. The method comprised of calculating the difference between the pH and laser method, and the average of the pH and laser method for each experiment, which are then plotted. The mean difference between the two methods was then found, which is known as the bias, to be +191 seconds with a standard deviation of 155 seconds. This bias line corresponds to a difference between the methods of 191 seconds, and since it is positive, it indicates the pH method takes longer to indicate the induction time than the laser method. Figure 6 illustrates the significance over the entire range of experiments, where data points above the bias indicate the pH method takes longer than the average difference between the two methods. If the data points are below the bias and closer to zero, there is a smaller difference between the methods. Any negative data points indicate the laser takes longer than the pH method to indicate the induction time.  35  Figure 6. Laser and pH probe induction time analysis of differences  Table 2 summarizes the means and standard deviations for estimates of induction times from the individual laser and pH-probe methods, and the mean difference and mean difference standard deviations between the methods for specific SSR ranges. As SSR increased the mean induction times decreased. This was expected, as induction time is dependent on SSR. The standard deviations also decreased highlighting improved accuracy in both methods. At lower SSR values, there was a large spread in induction time values for both the laser sensor and the pH probe. This spread is due to difficulties with both methods to accurately sense minor changes in the solution transmissivity and pH; however, as the SSR increased, there was better precision for both methods. The regression line R2 values are low due to this spread at the lower SSR values.    -600-400-20002004006008000 200 400 600 800 1000 1200Difference of pH-Laser Induction Times (s)Mean of Laser and pH Probe Induction Times (s)BiasMean (+191) 36 Table 2 Laser & pH probe induction time analysis of differences: The Bland-Altman Method (units in seconds)  Laser pH Probe Difference Analysis SSR Range Mean SD CV Mean SD CV Mean Difference (pH-L) SD of Mean Difference 4.0-5.0 570 275 0.48 763 178 0.23 193 245 6.2-8.5 207 88 0.42 418 107 0.26 211 60 8.4-10.3 82 32 0.39 262 70 0.27 180 41 10.5-12.6 83 23 0.28 254 56 0.22 171 42 13.3-15.8 53 21 0.40 - -  - - 16.1-19.9 45 8 0.18 - -  - -  The large variability in estimated induction times for low SSR values for both methods indicate a necessity for method improvement, or additional method development. Table 2 lists the calculated Coefficient of Variations (CV) for each SSR range for each method. The pH method was found to have the lowest CV values in all the SSR ranges, indicating it has a higher precision than the laser method.  The accuracy of each method is a challenge to compare, as the true value of induction time is dependent on how it is measured, and the size of struvite nuclei at the induction time has not been defined before. The mean differences for all of the SSR ranges show that the laser consistently estimated shorter struvite induction times. The laser method’s fast response time and ability to indicate an approximate size is an advantage for determining other process parameters of struvite nucleation and crystallization, like the location of nucleation in an FBR, and is important in ensuring crystals are present for the ζ nucleation experiments.  37 2.3.1.3 Influence of Mixing Speed on Induction Time To rule out any influences from the magnetic stirrer input on the nucleation and induction time of struvite, experiments were conducted at various mixer RPMs. This was to determine and reduce any additional errors that could be introduced into the data. It has been reported that mixing can influence growth (Ohlinger et al., 1999), so nucleation could be affected by mixer speeds.  Five to seven experiments were conducted at mixer RPM values 200, 300, 400, 500, and 600 as detailed in the Methods Section 2.2.1.1 for all the corresponding SSR values for the induction time experiments. The laser method subset of RPM 200, 400, and 600 was used to determine the influence of mixing speed on induction time. Each RPM data set regression line was first linearized with Equation 15. A two-tail t-test, to compare the slopes of the regression lines, was carried out to determine if the vertical shift in the regression lines represent the effect of mixing speed on the induction time. Combinations of the 200/600, 200/400, and 400/600 RPM data sets, and a significance level (alpha value) of 0.05 were used in the evaluation. The slopes of the regression lines were restricted to be identical at +0.52 across all the RPM values, and each new RPM regression line was compared to its original slope (see Appendix C for graph). The tests failed to reject the null hypothesis of equal slopes between the original and slope restricted regression lines; therefore, the parallel assumption was acceptable as there is no strong evidence from the data to suggest otherwise. The vertical shift in the regression lines was then used to find the difference between the regression line intercepts for 200, 400, and 600-RPM data sets. Table 3 summarizes the obtained values. This comparison determined that there was not a great significance to a vertical shift in RPM values, but between RPM 200 and 600 there was a difference of 12 seconds, which indicates a small influence from mixing speed.    38 Table 3 Regression line intercepts - the influence of RPM on induction time RPM Log of Intercept Intercept Actual time (s) R2 Values 200 1.505 32 0.80 400 1.376 24 0.80 600 1.310 20 0.79  The SSR and mixing speeds used in this research were lower than other published data (SSR 3-12 compared to 9-19, and RPM of 200-600 compared to 360-1060) (Ohlinger et al., 1999); however, the induction time between the lowest and highest RPM values from the two studies both show a 10-12 second difference at a constant SSR value. With optimal FBR operations usually around an SSR of 5, (discussed in the Chapter Summary) a 12 second difference from mixing will not have a great impact on the mean homogeneous induction time estimation of 570 seconds for SSR 4.0-5.0 (SSR range generally used in the FBR) reported here. The magnetic mixer was not found to have a great influence on induction time between 200-600 RPM, so any mixing speed in this range would provide adequate results for induction time measurements. It can be concluded that the stirrer speeds from a magnetic stir bar in the current study did not have a significant effect on induction time of struvite crystals; however, higher mixing energies, or different types of mixing like what is found in an FBR, could potentially have more of an influence. 2.3.1.4 Crystal Analysis XRD examination confirmed the solid samples as crystalline struvite structure from SSR 6, 9, and 12 from various induction time experiments, when compared to the XRD reference database (see Appendix D for XRD results). It is assumed no amorphous phases were present. Particle sizes ranged depending on SSR. Figure 7 illustrates the  39 size variation between the runs with SSR values of 6, 9, and 12, and crystal sizes measured to be < 20 μm, < 30 μm, and < 50 μm, respectively.   Figure 7. Crystal size variation between: a) Run 5, SSR 6; b) Run 4, SSR 9; c) Run 3, SSR 12  The laser method was found to estimate struvite induction times sooner than the pH method, but was not as precise as the pH method. Relating a specific size of crystal to the induction time with the laser ensures that crystals are present for the nucleation zeta potential experiments in Section 2.3.3. 2.3.2 Aged Struvite Crystal Zeta Potential Experiments Aged struvite crystals, in a 0.01 M NaCl solution at an average equilibrium pH value of 9.7 ± 0.2, were found to have an average ζ of -9.6 mV, and varied between -8.2 to -16.2 mV with standard deviations of individual readings of ±0.8 mV. Struvite crystals in b) a) c)  40 0.0036 M, 0.0134 M, and 0.0226 M NH4Cl solutions were found to have an average ζ of -8.2, -6.8, and -2.4 mV ±0.8 mV respectively. This showed a similar trend, with a less negative ζ with an increase in NH4Cl concentrations, as reported in the literature (Koutsoukos & Kofina, 2007).  Soluble substances, like struvite, are difficult to test in solution because the substance can dissolve, and cause changes in the particle surface and ionic strength of the solution system (Healy & Jellett, 1967). The targeted pH values and ionic strength for both the NaCl and the NH4Cl solutions were a challenge to maintain due to struvite dissolution (see Table 4). Initial solution pH values were adjusted to the target value before addition of the struvite crystals, but during the aging and settling process the pH changed. The NaCl solutions reached equilibrium between pH 9.6 and 9.7, and the NH4Cl solution pH change depended on the ionic strength, or buffering capacity of the solution. Equilibrium was obtained between solid and liquid phases for the NaCl solution during the aging and settling time. Approximately 185 mg/L of struvite dissolved in the 0.01 M NaCl solution, similar to reported struvite crystal solubility data in 0.01 M NaCl solution of 190 mg/L (from graph) and 169.2 mg/L in deionized water (Bhuiyan et al., 2007).  The lowest molarity NH4Cl solution had the largest amount of struvite dissolve (135 mg/L) for all the NH4Cl experiments. The solubility was slightly lower at higher ionic strengths for the NH4Cl solution. This is opposite to PHREEQC model predictions of an increase in solubility with a higher ionic strength, so solution equilibrium is not confirmed for the NH4Cl experiments. This contradiction is most likely due to the lattice ion NH4+ controlling and reducing the struvite dissolution at the higher concentrations. Over the 20 minutes of solution aging, an average of 14-15% of the 0.25 g of struvite dissolved in the NaCl solution, and 9-11% dissolved in the NH4Cl solution. This dissolution increased the ionic strength of the NaCl solutions by 34%, and the NH4Cl solutions by 69.4%, 16.4%, and 9.7% from lowest to highest molarity. The change in pH and ionic strength due to  41 dissolving struvite restricted the range that could be tested, and also highlights the problem with published methods and results for measuring struvite ζ. Table 4 Dissolution of struvite crystals into NaCl and NH4Cl solutions Solution Type Stock Solution Ionic Strength (mol/L) Target pH Experimental Values PHREEQC Modeling pH after aging Mass Struvite dissolved (g) Ionic Strength Increase due to Dissolved Struvite (mol/L) Total Ionic Strength (mol/L) Final pH Theoretical Mass of Struvite Dissolved (g) NaCl 0.008 9.4 9.6 0.037 0.0034 0.0134 10.0 0.039 NaCl 0.008 9.6 9.7 0.037 0.0034 0.0134 10.0 0.039 NaCl 0.008 9.8 9.7 0.036 0.0033 0.0133 10.1 0.039 NaCl 0.008 10.0 9.6 0.036 0.0033 0.0133 10.1 0.039 NH4Cl 0.0036 8.0 8.3 0.027 0.0025 0.0061 9.9 0.026 NH4Cl 0.0134 8.0 8.1 0.024 0.0022 0.0157 9.9 0.027 NH4Cl 0.0226 8.0 8.0 0.024 0.0022 0.0247 9.9 0.029  All struvite ζ data reported in the literature were determined using different ionic concentrations where no two solutions were alike, and often the solution composition was not reported. Exact methods are unclear in many studies, but the general practice is to precipitate struvite and then measure the ζ in the precipitation solution.  Not knowing the solution conditions makes it very hard to compare results, but some trends are noticeable from the reported data. pH has been found to be a controlling factor in struvite ζ. All reported data sets show a decreasing trend in the ζ values with an increase in pH (Bouropoulos & Koutsoukos, 2000; Koutsoukos & Kofina, 2007; Le Corre et al., 2007; Vol’khin et al., 2015). Figure 8 illustrates some of the comparable data. One exception to  42 this trend is from a study of semiconductor wastewater, that had many additional ions in solution, there was a decrease in the magnitude of the measured ζ between pH 8 and 10, and then an increase in magnitude as pH increased from 10 to 12 (Liu, 2009). There is also a general trend of a lower magnitude, neutralization, or even a charge reversal in ζ that corresponds with higher concentrations of magnesium ions reported in the literature (Bouropoulos & Koutsoukos, 2000; Koutsoukos & Kofina, 2007; Liu, 2009; Vol’khin et al., 2015). This trend is typically observed with divalent cations, where for struvite magnesium behaves as a specifically adsorbed ion. Bouropoulos & Koutsoukos (2000) report that the isoelectric point occurs at a magnesium concentration of pMg of 1.75 (432 mg/L) at a pH of 9.7, and Vol’khin et al. (2015) report a value of pMg of 1.82 (368 mg/L) at a pH 9.5.    43  Figure 8. Comparison of zeta potential measurements in relation to pH from (Le Corre et al., 2007) final solution ionic concentrations unknown at room temperature; (Bouropoulos & Koutsoukos, 2000) solution ionic concentrations of 0.01 M NaCl saturated wrt struvite at 25°C; (Z. Ye et al., 2014) final solution ionic concentrations unknown at 23-25°C; (Prywer et al., 2015) ionic strength calculated from conductivity as 0.31-0.29 M of unknown ion concentrations at 37°C; and current study averages in solution ionic concentrations of NaCl and dissolved struvite as 0.0133-0.0134M at 25°C.  2.3.3 Struvite Nucleation Zeta Potential Experiments Twenty-one successful experiments, similar to the induction time experiments, were carried out. Induction times were measured along with ζ measurements of the nucleating and growing crystals. Multiple liquid samples were collected for these experiments to measure SSR throughout the experiment timeframe. Figure 9 is an example of one experiment illustrating the relationship between ζ and SSR.  -30-25-20-15-10-507.5 8 8.5 9 9.5 10 10.5 11Zeta Potential (mV)pHLe Corre et al. 2007 Bouropoulos & Koutsoukos 2000Ye et al. 2014 Prywer et al. 2015Current Study 44 The data sets were combined based on wastewater composition and SSR target value, and a second order polynomial regression line was fit to each combined group. Table 5 outlines each group’s regression line and the corresponding coefficient of determination. Data from these experiments provides mostly qualitative information about the tendencies for different wastewater compositions, which can be used to further understand FBR struvite pellet formation.  Figure 9. Zeta potential nucleation experiment 102 (induction time as dashed line)  y = 0.0129x2 - 1.3006x + 14.179R² = 0.8719y = 0.0129x2 - 0.769x + 13.325R² = 0.98160.002.004.006.008.0010.0012.0014.00-40-30-20-10010200 5 10 15 20 25 30 35 41SSR Zeta Potential (mV)Time (mins)Zeta Potential SSR 45 Table 5 Experimental data grouping and statistical analysis Group Exp# Initial SSR Final SSR Final pH Data Crossing Bootstrap Crossing Time (mins) Bootstrap SSR Regression Line R2 Time (mins) SSR Min Max Min Max 1-A 102 12.59 2.34 7.0 10.8 5.06 2.35 7.59 5.5 42.5 y = 0.0099x2–0.9333x+12.651 0.61 103 11.75 2.04 7.0 8.6 5.01 3.31 7.27 4.5 14.6 104 12.59 2.34 7.0 14.7 3.72 3.08 4.58 11.1 18.9 124 13.8 2.57 7.1 23.0 3.19 2.53 9.23 4.5 39.8 125 12.59 2.51 7.1 26.3 2.71 2.46 3.43 18.5 40.9 126 12.88 2.88 7.1 27.5 3.12 2.83 3.9 19.4 41.1 127 14.45 2.95 7.1 37.6 3.00 2.94 10.13 4.1 41.5 1-B 105 5.89 2.82 7.0 - - 3.46 5.02 6.5 28.4 y = -0.0066x2+0.15051x–5.3019 0.33 106 4.79 2.14 6.9 - - 3.08 3.96 7.1 19.7 2-A 210 8.32 4.07 6.2 29.4 4.19 3.99 6.98 5.7 35.2 y = -0.0061x2+0.1327x+1.5308 0.40 2-B 209 6.03 2.82 6.2 23.3 3.36 2.66 3.59 19.1 39.5 y = 0.0099x2–0.8528x+14.077 0.54 3-A 312 6.03 2.24 7.5 22.1 2.93 2.3 4.99 4.5 35.2 y = -0.017x2+0.5649x–0.7792 0.48 313 6.31 2.51 7.5 29.9 2.86 2.58 5.21 4.4 36.1 315 6.46 2.51 7.5 39.9 2.46 2.38 5.55 4.2 42 4-A 422 10.72 3.16 6.3 34.0 3.08 3.08 3.41 22.9 40.1 y = 0.0019x2–0.4119x+11.898 0.63 423 10.96 3.09 6.3 33.2 2.91 2.91 7.11 5.1 41.1 4-B 419 3.98 - 6.2 - 1.75 1.69 1.92 30.1 41.2 y = -0.0077x2+0.3226x–0.883 0.10 421 4.17 3.89 6.3 - - - - 4.4 41.9 5-A 528 9.33 2.19 7.7 15.4 4.04 2.09 7.2 4.5 40.7 y = -0.0228x2+0.7045x–1.1189 0.66 529 9.33 3.16 7.8 29.3 3.73 3.00 7.61 3.6 40.2 530 8.91 6.46 7.9 27.2 6.28 6.05 7.95 3.4 33.8 (-) missing data or model calculations not valid  46 Figure 10a-d show general downward trends in ζ for all nucleation experimental groups. Steeper slopes and positive initial ζs correspond to higher initial SSR values for wastewater compositions 1 and 4 (groups 1-A, and 4-A in Figure 10a and b). This can be attributed directly to higher initial pH values within the same wastewater composition tested; conversely, the lower initial SSR values start at lower ζs as seen in 1-B and 4-B. The slope of the ζ regression line would be expected to become even steeper with an increase in initial pH and have a corresponding higher positive initial ζ for composition 1 and 4. This is counter-intuitive and opposite of literature results for aged crystals, where a higher pH corresponds to a more negative ζ. However, if the pH was held constant throughout the experiments, it is anticipated to achieve more negative ζ results at the end of the experiment, similar as to what is reported in literature for aged crystals (due to a higher ending pH).  Composition 1 experiments did not exceed an initial pH value of 7.6, and composition 4 did not exceed an initial pH of 6.6, compared to pH values above 8 or 9 reported in literature for the precipitation of struvite. The pH was not kept constant throughout the experiments, but dropped to a pH of 6.9-7.1 for composition 1, and to 6.2-6.3 for composition 4 within the 42 minutes of experimentation. Table 5 illustrates the distinct convergence of the final pH values for each wastewater composition. Similar final pH values for all the similar wastewater composition experiments exhibit similar final ζ values, which support previous evidence that pH is a controlling factor for struvite ζ. Group 2-A, 2-B, 4-A, and 4-B are all within a final pH of 6.2-6.3 and have average final ζ values ranging from -2.7 to -5.5mV; whereas, group 1-A, 1-B, 3-A, and 5-A have higher final pH values ranging from 7.0-7.8 and average final ζ values between -9.3 and -17.6 mV. Final ζs for composition 4 are the lowest in magnitude of all the values at -2.7 and -3.3 mV; this is likely due to the high initial Mg:P ratio of 1.4 and final magnesium concentrations of 540 mg/L, well over the published isoelectric point concentrations.  47           Figure 10. Wastewater composition group regression line comparisons -20.0-15.0-10.0-5.00.05.010.015.00 10 20 30 40 50Zeta Potential (mV)Time (mins)1-A 1-B-20.0-15.0-10.0-5.00.05.010.015.00 10 20 30 40 50Zeta Potential (mV)Time (mins)4-A 4-B-20.0-15.0-10.0-5.00.05.010.015.00 10 20 30 40 50Zeta Potential (mV)Time (mins)2-A 2-B-20.0-15.0-10.0-5.00.05.010.015.00 10 20 30 40 50Zeta Potential (mV)Time (mins)3-A 5-A 48 The trend for wastewater composition 2 (Figure 10c) is opposite from what is noted for composition 1 and 4. The lower initial SSR corresponds to a more positive initial ζ and a greater slope of the regression line (2-B); the higher initial SSR corresponds to a fairly flat regression line (2-A), with the initial ζ around +1.5 mV. This opposite trend is most likely due to lower concentrations of magnesium to phosphorus (Mg:P = 0.9) in the initial solution. Composition 2, therefore, has an excess of negative phosphate ions and the addition of caustic would then increase the concentration of anions in solution, and potentially increase the concentration of di or trivalent phosphate anions depending on pH values. Since they are the building blocks for the crystal lattice, ions of Mg2+, NH4+, and PO4-3 should all interact with struvite in predicable ways according to their ionic charge. Phosphate would be expected to continuously make struvite ζ more negative with an increase in pH when the Mg:P is less than one.  Compositions 3 and 5 produced similar regression lines, except they diverge around 30 minutes ending 10 mV apart from one another (Figure 10d). Wastewater composition 3 has approximately 100 mg/L more nitrogen than composition 5, but both have the same phosphorus and magnesium concentrations. As seen previously, magnesium can influence struvite ζ, so ammonium cations should also make the ζ less negative, but not as much as its divalent partner. A less negative ζ with an increase in ammonium concentrations is reported in the literature (Koutsoukos & Kofina, 2007) and was confirmed in the current study; this could explain why the graphs are so similar even though the initial SSR and pH were higher for composition 5. The regression line for composition 5 would be expected to shift upward if additional ammonium were added to the solution. 2.3.3.1 Time and SSR When the Potential Changes Sign Another analysis was carried out with two, second-order polynomial fixed-effects models in RStudio fitted to each ζ experiment; this was intended to determine the time at  49 which the ζ turned from positive to negative. This time represents a specific location within FBR systems, assuming plug-flow, and is important for determining where optimal agglomeration conditions are occurring. This crossing time was then correlated to the SSR value for each experiment. In addition, a parametric-bootstrap procedure was utilized to estimate the variability in the crossing time due to the high standard deviations associated with the ζ measurements. One thousand pseudo measurements were randomly created from Gaussian distributions for each data point, with the mean at the actual measurement and the Gaussian standard deviation equivalent to the measured standard deviation, for each experiment. The bootstrap procedure estimated the variability in the crossing time due to the high standard deviations associated with the ζ measurements, and it highlighted the fact that minimum and maximum-crossing times could occur within the first few minutes, right up until the end of the experiment.  Within the larger data set group of 1-A the higher SSRs correlated with longer time frames to reach a zero ζ as seen in Table 5, or Figure 11. This is a potential disadvantage to struvite agglomeration processes within FBR systems, where the agglomeration process needs to occur within a specific area of the reactor for optimum pellet formation. Wastewater composition 4-A also shows long time frames to reach zero for the higher SSR values. The lower SSR values of 1-B start at, or below, and do not cross the zero ζ, whereas the 4-B regression line is an inverted parabola and values hover around zero with two potential crossing times; hence the invalid model calculations. The low correlation coefficients for groups 1-B and 4-B are a result of variations throughout the ζ experiments that do not fit well with the model.   50  Figure 11. Initial SSR and time of zero charge for wastewater composition 1  The SSR value at the minimum and maximum crossing time is between 2-8 for 90% of the bootstrap procedure, and 2-6 for 95% of the actual data. This suggests that the solution composition between these low SSR values contributes to the low magnitude of the ζ values for all of the wastewater compositions. Other researchers have also suggested low SSR values to be optimal as a control parameter for struvite pellet formation, and for limiting fines losses in the recovery process (Fattah et al., 2012; Fattah et al., 2008; Forrest et al., 2008). Reported phosphorus removal efficiencies of 70-95% are associated with SSR conditions between an SSR of 2 and 4.5 (Mavinic et al., 2007). This would suggest that struvite agglomeration conditions could be SSR driven when trying to limit repulsive forces, and confirms the importance of SSR control for nutrient recovery systems. Table 5 compares the bootstrap minimum and maximum crossing times and corresponding SSR to the actual data set from all of the experiments.  Changing solution conditions (ionic strength) and crystals larger than the Zetasizer measuring limits are most likely the cause of the high variability in the ζ measurements, and the high standard deviations associated with them. The increase in amplitude in the y = 0.0707x + 11.435R² = 0.6421011121314150 10 20 30 40Initial SSRCrossing Time (mins) 51 induction time laser signal between 550 and 850 seconds, and the increase in laser intensity from 850 seconds to the end of the experiment (as seen in Figure 2a), is indicative of growing crystals, which were noticed to fall out of suspension. Figure 12 shows the individual ζ data points for group 1-A with standard deviations, and the average induction time as a vertical dotted line. Large standard deviations are seen in most of the experiments, but at random times, and do not correlate with only one point in the experiments. Since there are no other known studies measuring ζ for nucleating and growing struvite crystals to compare results and methodology, the variability and high standard deviations are beyond control at this time.   Figure 12. Wastewater 1, high SSR zeta potential data (average induction time as vertical dotted line)  -40-35-30-25-20-15-10-505101520253035400 5 10 15 20 25 30 35 40 45Zeta Potential (mV)Time (mins)102 (SSR 12.59) 103 (SSR 11.75) 104 (SSR 10) 124 (SSR 13.80)125 (SSR 12.59) 126 (SSR 12.88) 127 (SSR 14.45) 52 2.3.3.2 Attractive and Repulsive Force Comparison Attractive and repulsive forces between the struvite crystals during the nucleation experiments were examined. Comparison calculations using Equation 23-26, were made using a shear plane distance from the surface of 1.0 nm; based on an average between the size of a hydrated magnesium ion of 0.60-0.86 nm (Kiriukhin & Collins, 2002) and the longest length of a struvite unit cell c=1.12 nm (Prywer & Torzewska, 2009), which would be the largest ion or complex adsorbed at the struvite solution interface. The struvite Hamaker constant of 1.54x10-20Joules (Skuce, 2015), a 1.0 and 0.1 nm separation between particles, and a particle diameter of 1 μm were also used in the comparison. The calculations are very sensitive to ionic strength, since it is used to obtain the Debye-Hückel parameter and this parameter is used in both the ζ and repulsive force calculations. The ζ at the point where the forces would be balanced was calculated for wastewater compositions 1 through 5; for the separation distance of 1 nm, the results are ±9.7, ±8.6, ±11.0, ±9.1, and ±11.3 mV respectively, and for a separation distance of 0.1 nm the ζs are much higher at ±18.2, ±14.7, ±23.0, ±16.0, and ±24.6 mV. The experimental ζ measurements are generally within these ranges, except at the beginning and ending of the experiments at the higher SSR values. These results also suggest greater instability of the system for improved agglomeration is at lower ionic strengths (wastewater compositions 3 and 5). 2.4 Chapter 2 Summary and Conclusions The first objective of developing a size-dependent laser sensor for measuring struvite induction time was demonstrated to be significantly faster at sensing changes in solution characteristics than the typical pH-based approach. Struvite induction time is greatly dependent on solution supersaturation, with insignificant effect from mixing speed at low SSRs, where FBR conditions are optimal. The ability to indicate an approximate  53 crystal size at induction time ensured the presence of crystals in solution for ζ measurements during nucleation and growth of struvite crystals. Struvite ζ of nucleating and growing crystals in solution were investigated for five synthetic wastewaters. ζ values changed over time as the solution conditions changed, and became more negative as crystal precipitation and growth progressed. The charge of these crystals is important for inter-particle interactions within FBR systems, which impacts the likelihood for them to agglomerate into a pellet. Lower initial SSRs can produce low ζs within +/- 5mV for an extended period throughout the precipitation process, and would be desirable for agglomeration within an FBR. Using a high pH for nucleating struvite not only creates large amounts of unwanted crystals, but also creates conditions of high positive initial ζ; this can take longer to reach the point of zero charge and optimal agglomeration conditions. pH and magnesium concentrations had the greatest influence on struvite ζ. Magnesium molar concentrations should exceed phosphate concentrations to promote lower magnitude ζs. SSR control is important for ζ and struvite agglomeration control. SSR values between 2 and 6 are recommended to improve FBR agglomeration processes.     54 Chapter 3 Decoupling Struvite Crystal Growth and Agglomeration Processes within an FBR System There still exists a significant knowledge gap in the fundamental understanding of struvite pellet formation in FBR systems, in particular the mechanisms of agglomeration. Operational parameters such as up-flow velocities, SSR, and Mg:P ratio contribute to better quality struvite pellets, but the reasons why are still not fully understood. The main objective of this study was to de-couple struvite crystal growth from agglomeration in the UBC-designed FBR. The separation of these processes will prove the mechanisms behind struvite pellet formation, and allow for new process design to maximize pellet growth and reduce struvite fines losses. 3.1 Background Struvite pellet formation is currently accepted to be an agglomeration, or aggregation, of smaller crystals; therefore, studying the agglomeration phenomenon is crucial in reducing phosphorus losses in FBR systems. Since the agglomeration process of struvite in FBRs is not well understood or discussed anywhere in any great detail, a review of general and FBR agglomeration processes are discussed. Struvite morphologies and parameters that influence struvite pellet formation produced in FBRs, from literature, are also examined. 3.1.1 General Agglomeration Agglomeration can be defined as primary or secondary. Primary agglomeration is the growth process in which an individual crystal produces a twin or dendrite formations due to the crystal structure of the compound (Jones, 2002). Struvite commonly forms twins, x-shapes, star-shapes, and branching dendrite primary agglomerates at higher SSRs (Z. Ye et al., 2014). The bonding of twins is on an edge or crystal face, and can be caused by  55 impurities, high SSR, excess seeding, and poor mixing. Orthorhombic crystals, like struvite, often produce twins (Jones, 2002).  Secondary agglomeration comprises of the “intergrowth of aggregates formed by particle collisions, through a cementation process that forms an agglomerative bond.” (Brunsteiner et al., 2005). It is a process of enlargement by particles coming together, and in some instances can be a very fast process. Agglomerates are firmly bonded together, usually by crystalline bridges. An aggregate is the structure formed from individual crystals, but is not fully cemented together (Jones, 2002). In this work, struvite pellets will be referred to as agglomerates, not crystals. Secondary agglomeration will simply be called agglomeration and is the process of pellet formation that will be discussed.  Fine crystals can be produced through either primary or secondary nucleation (also known as homogeneous or heterogeneous nucleation), and are the main agglomerating particles in FBRs. Primary nucleation has been discussed in detail in Ch. 2, so will not be described further. Secondary nucleation is related to attrition and is dependent on SSR (Jones, 2002). Secondary nuclei are important for agglomeration processes as the crystals created could be incorporated into pellets, if the size is appropriate. A supersaturated solution is required for secondary nucleation to occur, and some form of breakage of protruding needles, or surface damage of the parent crystal creates the nuclei (Jones, 2002). This could result in fine individual crystals, which exit FBRs, if not properly controlled. Nuclei that do not get bumped off the parent crystal would grow and may develop into larger crystals, or clusters of crystals as part of a pellet.  Most of the secondary nucleation studies utilize mixers with the greatest influence being from impeller contact, which has much higher contact energy than the fluid dynamics of FBRs. Higher stirrer speeds have been reported to increase secondary nucleation in struvite-seeded, stirred-reactor experiments (Ariyanto et al., 2014). The struvite studies measure particle size and, unfortunately, not the physical appearance of the crystals to  56 see what is occurring on the crystal surface. It is assumed secondary nuclei have a different morphology than primary nuclei and are probably broken pieces of crystals. For continuous crystallization, it is advised to avoid primary nucleation by staying in the metastable or growth zone, and then only secondary nucleation needs to be controlled. If the secondary nucleation can be limited, individual crystals will grow much larger if they have long enough retention times for growth (Beckmann, 2013); hence, the struvite pellets will grow larger. Many conditions influence agglomeration processes. For agglomeration of any type of particle to occur, it is well known that repulsive forces or any destructive forces have to be smaller than attractive or bonding forces. Problems with agglomeration of fine struvite crystals has been associated with large negative zeta potential values, especially at higher pH values (Le Corre et al., 2007). Attractive forces generally decrease with an increase in particle size and surface roughness, and larger particles often require some form of bonding agent to make them adhere together (Pietsch, 1991). Ch. 2 investigated repulsive forces and how they changed as struvite precipitated from solution. Lower SSRs were found to induce lower zeta potentials. These results influenced the choice of SSR used for this investigation of agglomeration.  The tendency to agglomerate increases if the particles have a large size-distribution where voids can be filled with smaller particles, and then solid bridges can develop (Pietsch, 1991). Jones (2002) outlines the main factors that control agglomeration as: level of supersaturation; suspension density; particle size; degree of agitation; ionic strength; and presence of impurities. Increased agglomeration rates are attributed to higher SSRs, suspension densities, and particle sizes. Particles greater than 1 μm can be controlled by the fluid flow, where too high a flow can break agglomerates apart. Speed of crystal collisions could alter the probability of collision, where local velocities affect forces involved in aggregation (Brunsteiner et al., 2005).  57 After crystals come together, or aggregate, the creation of a bond is dependent on the characteristics of the solution between the two crystals. Higher SSRs have been found to create stronger agglomerate bonds for calcium oxalate dehydrate, and also create stickier particles (Brunsteiner et al., 2005). There may be a limit to the SSR used for struvite formation, with trade-offs between stickiness and excess nuclei formation. Brunsteiner et al (2005) found the agglomerate bonds differ in strength depending on which face of the potash alum was investigated. Small particles of 100 μm diameter or less, have a greater tendency to stay bonded, and not have their bonds destroyed by separation forces (Pietsch, 1991). Intergrowth of crystals by nuclei bridging requires a supersaturated solution, and requires the crystals to stay together long enough for solid bonds to form, and not only inter-particle forces like Van der Waals forces to hold them together (Linnikov, 2008; Linnikov et al., 2011). 3.1.2 FBR Agglomeration There are some limitations in using FBRs for agglomeration purposes. Typical fluidized beds, with one diameter, create higher porosity agglomerates because the particle size distribution is narrow. The occurrence of nucleation and growth of agglomerates in the same reactor is not ideal, again due to the FBR limiting the particle size distribution. This can be better controlled if a series of FBRs feed each other, since they will retain the smaller particles that have a higher chance to agglomerate (Pietsch, 1991). This idea is seen in the UBC FBR design, where the diameter increases in stages as the height increases, resulting in classifying the pellets, but also allowing different sizes to mix at the stepped zones (Koch et al., 2008). The conical shape FBR of Multiform Harvest (Bowers, 2013) would also allow various size fractions to mix and agglomerate. The low intensity of interaction in FBRs can reduce abrasion transfer and crushing, resulting in less effective agglomeration processes (Pietsch, 1991).   58 Shape, size, and number of particles is very important for agglomeration and it has been found that, within specific FBR systems, there is an optimum particle seed size and bed surface area or volume for the best recovery rates for struvite (Shimamura et al., 2007). Suspension densities (% by mass) specific to individual reactor configurations have a range, where below or above it will increase the risk of primary nucleation or cause higher attrition, respectively (Beckmann, 2013). This would mean that below this density range, there would not be enough crystal growth in the FBR to drop the SSR faster than the induction time, which is needed to prevent primary nucleation if high SSRs are used. This also means the available amount of growth in the bottom of the FBR, related to crystal surface area, potentially limits the inlet SSR when primary nucleation needs to be controlled. Surface area, reaction area, (Shimamura et al., 2007) and agglomeration sites decrease as pellets get larger. It was shown in a recent study for lower SSRs, higher struvite crystal growth and lower fines production occurred with higher seed loading within a stirred reactor (Agrawal et al., 2018). Optimization of suspension densities, or bed loading could be a means of controlling unwanted struvite fines in FBRs. 3.1.3 Observed Pellet Morphologies Produced in FBRs There are only a few studies in which the structures of pellets produced in FBRs were analyzed in any great detail. There are a number of case studies in the literature with images of the pellets they produced, but no operational data related to pellet morphologies were given; thus, they have not been included in this review. The studies referenced in this section utilize some configurations and operations similar to those of the UBC FBR. Pellet surfaces, internal structures, and varying size are discussed.  Agglomerate shapes and outer surfaces vary depending on FBR configuration and operation. Early work with the UBC FBR showed pellet morphology to be fragile “loose aggregates of plate-like crystals”, which changed during the studies into more compact- 59 round pellets (Britton et al., 2005). Other descriptions of the pellets and pellet surfaces include “tightly-agglomerated, brick-like and rod-like crystals” (Huang et al., 2006), as well as round, amorphous agglomerates of small and needle-like crystals (Crutchik et al., 2017). Higher magnification images of the pellet surfaces appear as agglomerates, with cracked crystal surfaces (Britton et al., 2005). Loose aggregated pellets have also been reported in a recent study analyzing up-flow velocity and saturation index (X. Ye et al., 2016). Harder smooth pellets have also been produced (Fattah et al., 2012; Ostara Nutrient Recovery Technologies Inc., 2017; X. Ye et al., 2016; Z. Ye et al., 2018). The round smooth surface of the pellet is described as being abraded from being in the FBR. They become smoother with longer retention times (Huang et al., 2006) and higher up-flow velocities (Z. Ye et al., 2018), however, more breakage can occur (Fattah et al., 2012). Elongated pellets were formed during a clogging event of the UBC FBR (Fattah et al., 2012), which would have increased velocities and changed flow patterns to more of a jet or spouted bed flow regime. Fattah et al. (2012) also reported that struvite pellets grown in the commercial reactors tended to be less spherical in nature. Cut pellets reveal a variety of internal characteristics. Britton et al. (2005) noted that the crystals within the pellets are of orthorhombic shape; growth of the orthorhombic shape is from the center outwards; and the outer surface of the pellet is the worn away tops of the crystals. Some have found that the inner cores of the pellets are not as compact as the outer surfaces (Fattah et al., 2012; X. Ye et al., 2016; Z. Ye et al., 2018). Some of the cutaways show a jumbled arrangement of various crystal sizes and shapes agglomerated together (Huang et al., 2006). Rings of varying growth and compactness can be seen in a few cut-open pellets (Fattah et al., 2012; X. Ye et al., 2016). This changing of the crystal habit seen in rings can be attributed to changing growth conditions (Chauhan & Joshi, 2013), and different internal versus external morphologies are due to  60 changes in driving force (Sunagawa, 2005). Morphology images from literature can be seen in detail in Ch. 4. Struvite pellet size is a difficult characteristic to compare unless the size fraction in specific sections, areas, or in the entire FBR is examined as Ye et al. (2018) have done. Larger particle sizes in the bottom section of the FBR are typical where higher up-flow velocities prevail. Higher velocities will also restrict the smaller size obtained in the top section of the FBR. The smaller size fractions from FBRs are generally loose agglomerates (Z.-L. Ye et al., 2018) or what appear to be broken pieces from pellets (Britton et al., 2005; Huang et al., 2006). However, sieving the pellets when dry could contribute to breakage and incorrect conclusions may be drawn about these size fractions and morphologies within the FBR. 3.1.4 Parameters Influencing Struvite Pellet Quality During Formation and Growth The intended usage is what dictates the required quality of the ready-to-use struvite fertilizer produced in the UBC FBR, so its shape, size, and in particular, the hardness, are important factors. The pellets have to be able to withstand handling, shipping, and machine application. The pellets need to be small enough, but not too small, so when mixed into soil the roots of the plants can activate the dissolution of the compound. Four main parameters are found in literature that contributes to the pellet quality during formation: the FBR SSR, liquid velocities, magnesium concentration, and impurities or suspended solids in the wastewater.  Higher SSRs can form stronger agglomerate bonds for other compounds, so it may also apply to struvite. Optimal UBC FBR performance (>80% phosphorus removal) occurs at SSRs between 2-6 (Bhuiyan et al., 2008). This range has not been related directly to pellet quality, but it does limit nucleation, and ensures only crystal growth conditions. SSR also controls repulsive forces as seen in Ch. 2, where recommended values are in the  61 same optimal range recommended by Bhuiyan et al. (2008). FBR pH control has been reported to be problematic at times (Fattah et al., 2012), resulting in unstable SSR control and potentially changing pellet morphology. One of the main influences on the size of a pellet is the length of time it is given to grow, although pellet size has been reported to be smaller for FBRs operating at higher SSRs (X. Ye et al., 2016). Low recovery rates of pellets, or reduced pellet formations are also reported at higher SSRs where large amounts of fines are produced (Ghosh et al., 2018). High rates of nucleation restrict pellet-surface growth rates because all the crystal surfaces contained within the FBR, including the nuclei, would be growing at the same time.  FBR velocities appear to influence the hardness of the pellets. As pellets increase in size, the velocities required to fluidize them must also increase. Pellet morphologies change when FBR up-flow velocities increase. Loose aggregates turn into smooth, harder pellets, where recently a “coating-growth” process has been observed with velocities exceeding 40.8 mm/s (245 cm/min) (Z. Ye et al., 2018). A higher velocity could fill in the voids by crystal growth, where liquid is forced inside, or contribute to small crystals forced into the agglomerates. Larger FBRs also tend to produce harder pellets with a smooth outer surface. This phenomenon could be due to the higher centerline inlet velocities in larger FBRs, which models have shown (X. Ye et al., 2017) to affect the growth or morphology of the outer surface of pellets. An optimal upper up-flow velocity of 400 cm/min was recommended for good quality agglomerates; above 500 cm/min, agglomerates can break and more fines losses could occur (Fattah et al., 2012). However, Ye et al. (2016) found no occurrence of broken pellets above this optimal velocity. Higher Mg:P was reported to create more compact and harder pellets, and an increased crushing strength was correlated with higher Mg:P values (Fattah et al., 2012). Excess magnesium concentrations could also reduce the repulsive forces of the system, as seen in Ch.2, where greater agglomeration could take place and more compact and  62 harder pellets could form. Fattah et al. (2012) also mention the fertilizer industry uses magnesium sulphate to make stronger pellets during granulation. Phosphorus removal rates are higher with higher Mg:P ratios because phosphorus is then the limiting ion, not magnesium. Higher magnesium concentrations increase the SSR of the system; therefore, a reduced pH is required for the same SSR value (Mavinic et al., 2007). Impurities such as phosphorus-competing ions like calcium or iron have not been reported to cause any reduction in strength or potential size of struvite pellets. The ions will get incorporated into the pellets so the composition or purity is compromised. Calcium has been shown to slow the growth rate of struvite crystals (Le Corre et al., 2005), so pellets might take longer to grow to a comparable size. Suspended solids concentrations within the wastewater can, however, affect the formation of struvite pellets. Struvite pellet size and purity decreases with an increase in total suspended solids (TSS) content. The suspended solids block growth-sites and are incorporated into pellets (Ping et al.2016). These solids can also provide secondary nucleation sites instead of the struvite pellet itself. This competition decreases the potential for pellet growth. It has been reported that the orthorhombic crystals inside pellets become smaller and shorter with a higher TSS. It is beneficial to have lower TSS, since a wider crystal size distribution develops in the FBR (Ping et al., 2016).  There still exists a significant knowledge gap in understanding struvite pellet formation, in particular the mechanisms of agglomeration. Pellet morphologies vary depending on FBR configuration and operational parameters, but how, and why are not fully understood. The objectives of this study were to determine the main mechanism of struvite pellet formation in an FBR, and to determine if agglomeration and growth processes can be separated and controlled.   63 3.2 Materials and Methods Two FBRs, one as a control and one for the experiment, of the same dimensions were used to separate growth and agglomeration processes. Both FBRs were operated in the crystal growth phase, to restrict primary nucleation, while small pre-formed struvite crystals were injected into the experimental FBR to promote agglomeration processes. Struvite fines have been grown separately from main reactors and then used as seed (Shimamura et al., 2007), or collected from FBR clarifiers (Adnan et al., 2003), and sometimes re-injected into the FBR if they wash out (Ueno & Fujii, 2001). Struvite fines have never been used before to try to promote or study agglomeration and pellet growth processes.  3.2.1 Experimental FBR Setup and Operation The UBC FBR design (Koch et al., 2008) was used for the pellet growth and agglomeration experiments. This FBR is a stepped FBR, which retains smaller crystals in the upper zones. The agglomerates classify by size in the different zones, but can interact at the steps, due to turbulent fluid dynamics at these areas.  3.2.1.1 FBR Design The FBR consisted of an injector port with an internal diameter of 0.95 cm at the bottom made out of stainless steel with inlet ports for the feed line, a clarified recycle line, a sodium hydroxide solution port, and a magnesium chloride solution port. Connected in sequential order to the injector port was clear PVC piping designated in ascending order as the harvest zone, active zone, fines zone, and the seed hopper. Each zone increased in internal dimension with a diameter of 2.54, 3.81, 7.62, and 19.05 cm and lengths of 53.34, 60.96, 60.96, and 38.10 cm, respectively. Below the harvest zone the length of the injector port was 19.05 cm.   64 The total volume of the FBR was approximately 14 L. An isolation valve was located at the junction between the harvest zone and the active zone. The recycle inlet contained a shutoff and bypass line above the isolation valve for continuous fluidization during the harvesting of struvite pellets. A wye-connector installed in the bottom of the harvest zone with another valve enabled the harvesting of pellets. A pH probe was installed in the bottom of the active zone, and was connected to an Omega CNi1653 controller and a MasterflexTM L/S variable speed pump for continuous addition of sodium hydroxide solution for reactor pH control. An external clarifier with a capacity of 46.5 L was connected to the outlet of the FBR seed hopper, and was the reservoir with a standpipe for the recycle flow back into the FBR. An overflow from the clarifier was used as the effluent outlet. A MoynoTM 500 Series, Model 331 progressive cavity pump with a RelianceTM variable frequency drive was used for the recycle flow. Magnesium chloride solution and feed were each pumped into the FBR using MasterflexTM L/S pumps with standard pump heads, specific for each flow requirement. Figure 13 and 14 illustrate the FBR setup.     65  Figure 13. FBR configuration   66  Figure 14. Detail of FBR lower section  3.2.1.2 Chemicals Used All chemicals were mixed with tap water. The synthetic feed comprised of 800 mg/L of nitrogen and 100 mg/L of phosphorus prepared from a combination of commercial grade ammonium chloride (NH4Cl) and diammonium phosphate ((NH4)2HPO4), and food grade phosphoric acid (H3PO4). The magnesium solution was prepared at a concentration of 1500 mg/L from commercial grade magnesium chloride hexahydrate (MgCl2.6H2O) and was diluted to operating conditions, once pumped into the FBR. Sodium hydroxide (NaOH), for pH adjustment, was mixed to approximately 1-1.5 molar. 3.2.1.3 FBR Operations  Two FBRs were operated concurrently; one as a control, which had been operating for 12 days prior to starting the second FBR with the crystal addition. The targeted operational values for initial process setup consisted of the SSR=4.0, feed flow of 0.25  67 L/min, the recycle flow at 7.0 times the feed flow, up-flow velocity of 400 cm/min within the harvest zone, temperature of 22°C, and pH of 8.16. Initial start-up for both FBRs consisted of filling up the FBR and clarifier with a feed and water mixture and turning on all the chemical pumps and pH controller. The system was allowed to stabilize to the required SSR value for a few hours. Each FBR was then seeded with 86 g of 0.5 mm pre-formed struvite pellets and 206 g of struvite crystals; this reduced the reactor bed loading time. The struvite seed used was produced from previous student research from real centrate, and was a brown colour, enabling a visual reference to new white synthetic crystal formations.  3.2.1.4 pH Control pH was controlled with a pH probe inserted in the FBR just above the harvest zone, and the use of a separate pump and a variable speed controller connected to the pH controller. This enabled the pump speed to be controlled separately from the pH controller. This is a different setup than that traditionally used in past UBC research, where proportional flow pH controllers have been used with varying success of control, and could be the reason behind excess nucleation by creating localized high SSRs. It has been found that the induction time of struvite is affected by the addition of NaOH to solutions, and a localized high SSR created fines (Agrawal et al., 2018). For this reason, the setup was changed and a slow, constant NaOH addition was established; hence, primary nucleation was prevented in this study. 3.2.1.5 Measurements & Sampling Procedures Measurements and sampling were carried out twice a day for the control, and varied from once to twice a day for the crystal addition experiment to confirm SSR operational values. The sampling varied for the crystal addition experiment, due to the extra crystals interfering with measuring the background or equilibrium condition. On days  68 when the chemicals were turned off, samples were taken at least an hour after turning them back on to make sure FBR equilibrium conditions were reached. The hydraulic retention time (HRT) in the FBR was calculated to be 5.6 minutes and in the clarifier was approximately 23 minutes, with the up-flow velocity of 400 cm/min, so an hour would allow for the system to stabilize and reach equilibrium. The recycle, feed, and magnesium flows were measured. Filtered (1.5 μm filter) and unfiltered liquid samples were collected from the seed hopper, the clarifier, and the incoming feed line. The conductivity, pH, and temperature were measured in the seed hopper and clarifier. Solid samples were collected from the pH probe sampling port and the seed hopper. Pellet harvests only occurred when the pellets were approximately 2 mm in size, and they filled the harvest zone of the FBR. Pellets were sieved in the mother liquor solution, before being left to air dry, to prevent excessive breakage. Sieve sizes were approximately 2.0, 1.0, 0.5 mm in size, and fines of less than 0.5 mm were filtered with a 2.0 μm filter.  3.2.2 Procedure for Crystal Addition Synthetic struvite crystals were mixed at a concentration of 5 g/L or 10 g/L with filtered FBR clarifier effluent. The effluent was collected from the outflow of the FBR clarifier, and was aged for at least 12 hours before it was filtered with a 2 μm filter. This ensured the SSR was close to equilibrium so the crystals would not dissolve or grow. Four liters of the crystal suspension was mixed fresh every couple of hours as required during the crystal addition. For the addition of crystals into the FBR system, a simple round container with a capacity to hold 4 liters fitted with a mixer and paddle, and a MasterflexTM L/S pump, all controlled by an analog timer, were utilized. The crystal suspension was pumped from the round container into the standpipe inlet of the FBR recycle pump within the clarifier. The concentration of crystals diluted into the total flow of 2.0 L/min depended on the crystal pumping flow, and averaged between 0.25 and 0.5 g/L for the above  69 concentrations. The crystal suspension was pumped for approximately 5 minutes, and then turned off for approximately 10 minutes per cycle. At the end of each day, the remaining crystals in the round container, and any that settled in the tubing, were collected, filtered, dried, weighed, and then subtracted from the total mass added. 3.2.2.1 Hourly Liquid Sample Collection  Additional liquid samples were collected to determine if the injected crystals stayed in or washed out of the FBR. Filtered and unfiltered liquid samples were collected from the seed hopper and the clarifier effluent 2-4 minutes before the crystal injection pump turned on, and then again approximately 3-4 minutes after the pump turned on. This was carried out for days 2 to 5.  3.2.3 Analytical Methods Liquid samples were analyzed for orthophosphate and ammonia by using a LACHAT QuickChem FIA+ 8000 Series with a ASX 500 Series Autosampler, according to UBC method EE-SOP# A.04.01 ‘Ammonia, Nitrate and Phosphate Determination in Water by Continuous Flow Analysis’ with reference to Standard Method 4500-P G and 4500-NH3 H (Eaton et al., 1995). Magnesium was analyzed by atomic absorption spectroscopy with a 220 Fast Sequential Varian Spectra AA with a Varian Autosampler model SPS-5, in accordance with Standard Method 3111 B (Eaton et al., 1995). pH and conductivity were measured with a hand-held Horiba pH/conductivity D-54 meter. Solid struvite samples, from the control FBR, were analyzed for molar ratio confirmation by dissolving in a hydrochloric acid solution. Scanning electron microscope (SEM) images were taken of the solid samples with an Environmental Scanning Electron Microscope FEI Quanta 650 VP using the secondary electron detector (indicated by LFD in the images) and the backscatter detector (indicated by CBS in the images) in low vacuum mode.   70 3.2.4 Calculations 3.2.4.1 Supersaturation Ratio Struvite is a transparent to semi-transparent crystalline material made up of equal molar concentrations of magnesium, ammonium, and phosphate with six waters of hydration. Struvite precipitation is controlled by the saturation of solution with respect to ions that form the crystals lattice, and precipitates as per Equation 27  𝑀𝑔2+ +  𝑁𝐻4+ +  𝑃𝑂43− + 6𝐻2𝑂 =  𝑀𝑔𝑁𝐻4𝑃𝑂4 ∙  6𝐻2𝑂 (27) The supersaturation ratio (SSR) represents the extent of crystallization that must occur in order for the system to reach equilibrium, and can be described as the driving force of the precipitation process. The SSR is the ratio between the ion activity product (IAP) and the solubility product (Ksp), as seen in Equation 28  𝑆𝑆𝑅(𝑠𝑡𝑟𝑢𝑣𝑖𝑡𝑒) =  {𝑀𝑔2+}{𝑁𝐻4+}{𝑃𝑂43−}𝐾𝑠𝑝(𝑠𝑡𝑟𝑢𝑣𝑖𝑡𝑒)=𝐼𝐴𝑃𝐾𝑠𝑝 (28) A pKsp value [𝑝𝐾𝑠𝑝 =  −𝑙𝑜𝑔10(𝐾𝑠𝑝)] of 13.47 was used in this project (Lobanov et al., 2013). All struvite SSR values were modeled using PHREEQC software. 3.2.4.2 Phosphorus Removal  Phosphorus removal, in the FBR, was calculated using the incoming concentrations and outgoing concentrations of phosphorus as per Equation 29.  %𝑃 𝑟𝑒𝑚𝑜𝑣𝑎𝑙 =  ([𝐶𝑓](𝑄𝑓) −  [𝐶𝑒](𝑄𝑒)/ ([𝐶𝑓](𝑄𝑓))) ∗ 100 (29) where [Cf] is the concentration of PO4-P in the feed, [Ce] is the concentration of PO4-P in the effluent, Qf is the feed flow, and Qe is the effluent flow (which in this study is the combination of the feed flow and magnesium flow). 3.2.4.3 In-Reactor Concentrations The in-reactor, mixed conditions was used to calculate the exact SSR within the FBR from collected samples, and can be calculated by Equation 30.  71  [𝐶𝑇 𝑖𝑛−𝑟𝑒𝑎𝑐𝑡𝑜𝑟] =  ([𝐶𝑓](𝑄𝑓) +  [𝐶𝑅](𝑄𝑅)/((𝑄𝑇))) (30) where Cf and CR are the feed and recycle concentration for a particular ion; Qf and QR are the feed and recycle flow; and QT is the total flow of the feed and recycle, including the magnesium flow. 3.3 Results & Discussion The two FBRs were operated using the same chemical feed, magnesium, and caustic solutions. Pumps were maintained at the same pumping rates, and pH was closely controlled to have all operational conditions as close to each other as possible. The experiment lasted for 12 days in total. Days 6 to 8 fell on a long weekend, so both of the FBRs were fluidized with only the recycle flow during that time. The following two sections provide a summary of each FBR run, followed by a comparison of the two; then a section on the crystal size distribution found in the clarifiers, a section on how pellets form, and lastly a comparison of the morphology of a hard pellet. 3.3.1 Control Run Summary The purpose of the control run was to have only crystal growth occurring within the FBR, and to compare the pellets produced in the control FBR to the crystal addition experiment pellets. Growth was confirmed as the main pellet formation with the filtered and unfiltered effluent samples, and the low amount of fines accumulation within the clarifier. The average difference between the filtered and unfiltered effluent samples was ±1 mg/L of phosphorus. The cumulative accumulation of struvite crystals collected from the clarifier for the entire run of 19 days (ten days longer than the crystal addition experiment) was 71.3 g, which is less than 3.8 g/day. Most of the accumulation would have occurred during startup of the FBR, when it was not fully loaded with pellets. It can be concluded that with this very low crystal accumulation in the clarifier, and no significant difference between effluent samples, the main pellet formation was from crystal growth  72 and not agglomeration. SEM photographs also confirm this. Figure 15 (a) and (b) show no signs of small crystals within the pellets compared to Figure 18 and 19 for the crystal addition pellets. A well-defined, branching structure can be seen growing from the center of the pellet outwards from the cut pellet (b) without any agglomerated crystals visible.     Figure 15. (a) [Sample3_bse_01] Outer surface of pellet from control run 40x magnification; (b) [Sample14_bse_01] 2mm pellet cut-away from control run 25x magnification; (c) [Sample14_lvse_03] Edges of the cut pellet 250x magnification; (d) [Sample4_lvse_02] Seed hopper junction fines from control 150x magnification.  This branching structure within the struvite pellet has not been clearly identified until now. Britton et al. (2005) noticed orthorhombic crystal shapes growing from the center outwards, and in most other cut pellets the typical shape is there, but hidden by other (a) (b) (c) (d)  73 agglomerated crystals. It can be seen in Figure 15 (a), (b), and (c) that the tops of the branches are orthorhombic crystals, which previously have been identified as ground-down crystals. The ground-down appearance, noticed in the work by Britton et al. (2005), is more than likely due to the sieving or handling of the pellets; however, cracks or defects on the tops and sides of the orthorhombic crystals are apparent in the control run. It is hypothesized that these defects create locations of higher growth, and/or secondary nucleation, and are sites where the branch changes direction. It can be concluded that this branching structure is the preferred morphology for struvite growth in a FBR at low SSRs. The smallest crystals fluidized in the FBR were collected at the junction between the fines zone and the seed hopper. They were approximately 200-300 μm in size, and shared similar formations to the pellets. Figure 15 (d) illustrates a handful of these growing crystals, which are thought to be the source of pellet cores, and are more than likely produced by attrition. Figure 15 (d) also shows no sign of smaller crystals, again confirming only growth was occurring without any primary nucleation in the FBR. Some of these smaller agglomerates show localized ruptured areas, where they would have broken off from a larger crystal structure (Brunsteiner et al., 2005), and illustrates the attrition process (Section 3.3.2 Figure 17 (b) and Figure 20 (a) white arrows).  Molar ratio comparison of the control FBR pellets and crystals, from the last day of FBR operation, indicates a Mg:N:P ratio of 1:1.1:1. This measured ratio is within the limits of measurement error to confirm struvite is the main crystal produced. The growth within the crystal addition FBR is assumed to precipitate a similar molar ratio, but pellets were not analyzed due to the influence from the added crystals.  74 3.3.2 Crystal Addition Summary Conditions varied throughout the experiment in regards to: the length of time and concentration of crystals added; the FBR operational time when all the chemicals were turned on, designated as FBR operational time; and the time when the pellets were fluidized with only the recycle flow, designated as FBR recycle time. The experiment time includes the entire cycle of crystal addition time as well as when crystal pump turned off. Figure 16 illustrates the times of all of these operations for the entire experimental run. Table 6 lists the concentrated crystal suspensions, and actual amounts of crystals added to the FBR over the 12 days.  Two distinct sets of conditions were simulated: continuous growth with limited crystal addition (days 2 to 4); and continuous growth during crystal addition (days 5, and 9 to 12). During the experiment, the amount of crystals added to the FBR was progressively increased until it was overloaded with fine crystals on day 4, during continuous growth with limited crystal addition conditions. This approach estimated the amount of crystals to use for the continuous growth during crystal addition conditions, the second half of the experiment.  75  Figure 16. Experimental FBR operational parameter timeframes   Table 6 Crystal addition amounts Day Concentrated Crystal Suspension Used (g/L) Actual Amount of Crystals Added to FBR (g) – (Added minus collected in tubing) 1 - - 2 10 13.0 3 10 41.9 4 10 101.1 5 5 52.7 9 5 62.7 10 5 67.3 11 5 67.9 12 5 59.4  05101520251 2 3 4 5 6 7 8 9 10 11 12HoursDayFBR Operational TimeFBR Recycle TimeExperiment TimeCrystal Addition Time 76 The initial start-up of the FBR was as outlined in Section 3.2.1. During the first day, the seed grew unhindered without any additional crystals added to the FBR. The brown pellets turned white, indicating synthetic struvite had grown and coated the seed struvite, and the reactor had filled up to an approximate equilibrium operating solids content. Crystal addition started on the second day.  For the continuous growth with limited crystal addition experiment, the amount of crystals added increased each day, with an increase in pumping rate, from 3.5 g/hour, to 8.2 g/hour, to a maximum of 17.9 g/hour, based on the actual amount added to the FBR during the experimental time. The FBR was overloaded with 17.9 g/hour of crystals, which could be seen in the cloudy seed hopper and clarifier on day 4. There were 76 hours of continuous FBR operation, or crystal growth conditions from the beginning of day one to the end of day four. Of those 76 hours, there were 14.5 hours (average of 4.8 hours for each of the three days) of experimental time when crystals were added to the FBR. The crystal addition pump cycled on for approximately 5 minutes, and off for 10 minutes. SEM images of the solid samples show a progression in agglomeration from day 2 to 4. Figure 17 (a) and (b) show samples taken 1.5 hours after starting the crystal addition on day 2 from the pH probe-sampling port. The 2.0 mm pellet surface shows very few small crystals attached within the pellet branches, and the cut pellets did not reveal any significant agglomeration. At this sample collection time, approximately 5.3 grams of crystals had been added to the FBR; and it would be expected that this small amount would not be apparent in the pellets. The smaller-size fractions of the sample are agglomerates and crystals of approximately 200-400μm. Figure 17 (b) shows some of these smaller agglomerates with a coating of the small crystals on their surfaces, and others without any small crystals. This lack of agglomeration on some crystals is likely a result of breakage of the pellets during sample collection. This observation is significant because it validates that agglomeration and bonding processes are taking place in the  77 FBR, and the accumulation of the small crystals is not through static attraction during handling of the dry samples.  After the crystal addition on day 2, the pellets were left to grow without crystal addition for approximately 20.3 hours. One harvest, and samples from the pH probe sample-port, were collected after this growth phase and before the crystal addition started on Day 3. These samples in Figure 17 (c) show a 2.0 mm pellet surface with few small crystals on the surface, and (d) a cut pellet with many small crystals within it. The 20.3 hours of growth enveloped the added crystals from day 2 within the branching structure. The harvested fines showed no small crystals attached to them, so they were again most likely created through breakage of the outer branches of the pellets when they were collected and sieved. Day 3 continued with a higher crystal addition rate of 8.2 g/hour.     78  Figure 17.  (a) [Sample49_bse_03] 2.0 mm pellet surface after 1.5 hours of crystal addition 100x magnification; (b) [Sample49_bse_04] Agglomerates from pH probe sampling port 100x magnification; (c) [Sample51_bse_02] 2.0mm pellet surface from day 3 before start of crystal addition 100x magnification; (d) [Sample51_bse_05] 2.0mm pellet cut open from day 3 before start of crystal addition 100x magnification.  The pellets were left to grow for approximately 19 hours after the crystal addition of day 3. One harvest, samples from the pH probe, and samples from the seed hopper junction were taken before the crystal addition started for day 4. Similar to the previous day samples, the outer surface of the pellets did not contain many small crystals; however, once cut open the small crystals are embedded throughout the internal structure of the pellet. Figure 18 (a) shows clusters of the small crystals that have been enveloped inside the branch structures during the growth phase. Figure 18 (b) shows the seed hopper (a) (b) (c) (d)  79 junction agglomerates with small crystals bonded onto them from day 3, similar to the agglomerates from the previous day in Figure 17 (b). Crystal growth is not as apparent with the small agglomerates from the seed hopper junction, as it is with the pellets. This is likely due to the SSR being the lowest in the upper section of the FBR, and as these agglomerates grow they sink down to lower sections within the FBR, so it is difficult to track their evolution. Day 4 continued with the crystal addition at 17.9 g/hour. The seed hopper and clarifier were cloudy throughout the day from this amount of crystals being added to the FBR. More than likely, the crystal addition was greater than 17.9 g/hour due to the cloudy clarifier recycling the crystals. Figure 18 (c) shows the development and different ages of seed hopper agglomerates with different amounts of small crystals attached to their surfaces. Some of the agglomerates have few small crystals attached to them, while others are almost completely covered as the magnified Figure 18 (d) shows. The FBR chemical addition was stopped after crystal addition on day 4, and the next phase of the experiment commenced.   80  Figure 18.  (a) [Sample53_bse-03] 2.0mm cut pellet from day 4 before crystal addition 100x magnification; (b) [Sample55_bse_02] agglomerates at seed hopper junction from day 4 before crystal addition 100x magnification; (c) [Sample56_bse_02] agglomerates at seed hopper junction after 4 hours of crystal addition from day 4 100x magnification; (d) [Sample56_bse_03] inset of (c) at 500x magnification.  The continuous growth during crystal addition condition was performed on days 5, and 9 to 12. The chemicals (feed, NaOH, and magnesium) were turned on each morning, and no crystals were added for approximately an hour to allow the FBR to reach steady state. After the system stabilized and samples were collected, the crystal injection commenced for the day. The average rate of crystal addition was 9.3 g/hour with the same pump cycle as the previous days. On average, the experimental timeframe was 6-8 hours each day; then, the FBR chemicals were turned off and only the recycle flow fluidized the pellets for (a) (b) (c) (d)  81 the remaining 14-17 hours, instead of having the chemicals on during the night, like the previous experiments. There was a three-day period (days 6-8) where only the recycle flow fluidized the pellets and no crystals were added for 72 hours. A total of 33.7 hours of experiment time was carried out over the five days. Solid samples were taken each day of crystal addition and photographed by SEM. No harvests were carried out until the end of the experiment, due to little pellet growth occurring during these conditions. Figure 19 (a) through (f) shows various days and magnifications of samples for the continuous growth during crystal addition conditions. The pellets all show a similar trend, with the small crystals aggregated onto and into the branching structures. Figure 19 (c) shows that over the course of the three days (day 6-8) of the FBR being on recycle, the small crystals stayed attached to the pellet and fixed themselves firmly to the branch structures. Figure 19 (a) and (d) are pellets from day 5 and 12, both covered in small crystals with very little difference from each other, except day 12 has fewer branches sticking out of the small crystals. During the entire experiment the small crystals did not build up on top of each other and completely cover the branching structure. This may be due to the crystals not bonding to the large flat surface of the top of the crystal branches or the crystals needing multiple surfaces, such as a corner that is ridged to bond to. It could also indicate the growth of the branches were faster than the agglomeration of the small crystals. Figure 19 (e) is a pellet after a 10-minute ultrasonic bath in methanol, where the majority of small crystals remained attached. This test confirmed physical bonding of the small crystals to the branching structures of the struvite pellet.  82  Figure 19. Outer surface of pellets from (a) [Sample57_bse_01] day 5 pH probe sample of a pellet after two hours of crystal addition 50x magnification; (b) [Sample69_bse_03] day 5 500x magnification; (c) [Sample60_bse_04] day 9 before crystal addition and after 2 days on recycle flow 50x magnification; (d) [Sample7_bse_01] 2.0mm pellet from day 12 25x magnification; (e) [Sample16_bse_01] day 12 harvested 2mm pellet from crystal addition ultrasonic in methanol 25x magnification; (f) [Sample7_bse_03] 1000x magnification of d. (a) (b) (c) (d) (e) (f)  83  The seed hopper agglomerate samples from days 5 and 9-12 show similar structures as days 2-4, with small crystals attached to the surfaces. Figure 20 (a) through (f) are samples with different magnifications from three different days of the experiment. Inside corners of the agglomerates tend to have more bonding occurring than on the flat surfaces, which is much more apparent in these smaller agglomerates than with the pellets. This was expected since there would be more attractive forces over the two surfaces in a corner, increasing the adhesion tendency (Pietsch, 1991). It could indicate that struvite requires multiple surface bonds to agglomerate, so roughness and crystal irregularities are important.  84  Figure 20. Seed hopper junction samples (a) [Sample58_bse_02] Day 5 100x magnification; (b) [Sample58_bse_03] inset in a 250x magnification; (c) [Sample64_bse_03] day 10 at 100x magnification; (d) [Sample64_bse_04] inset in c 250x magnification; (e) [Sample8_bse_02] day 12 150x magnification; (f) [Sample_8_bse03] inset in e 400x magnification.  (a) (b) (c) (d) (e) (f)  85 3.3.2.1 Crystal Retention To determine the amount of fine crystals that struvite pellets could agglomerate in an FBR, additional liquid samples were collected to measure if crystals remained in, or washed out of the FBR system. The samples were collected once an hour for the entire experimental period on days 2 to 5. Filtered and unfiltered liquid samples were collected from the seed hopper and the clarifier effluent. They were analyzed for phosphorus concentrations only, and the results were plotted. Sample numbering is in the order they were collected. Whole numbers represent the samples collected 2-4 minutes before the crystal injection pump turned on, and the samples half numbers (E.g. 1.5, 2.5 etc.) represent the samples collected 3-4 minutes after the pump turned on. Due to physical limitations, the four samples were collected over a 3-minute period. Figure 21 through Figure 24 shows the plots from the filtered effluent (EF), unfiltered effluent (EU), the filtered seed hopper (SHF), and the unfiltered seed hopper (SHU) samples from the four days.  The difference between the filtered and unfiltered samples indicates the extent of crystals exiting the FBR system. As the rate of crystal addition increased from days 2 to 4, for the continuous growth with limited crystal addition conditions, the graphs show an increase in the concentrations of crystals exiting the FBR. This is seen in the increase in the SHU concentrations and the difference between the EU and EF. Day 2 shows a peak in the SHU at 10 mg/L at sample 3, day 3 peaks at 21 mg/L at sample 2.5, and day 4 peaks at 25 mg/L at sample 3. The average difference between the EU and EF for day 2 is negligible, day 3 is 5 mg/L, and day 4 is 8 mg/L excluding the first several samples. The effluent samples exhibit less variation between samples than the seed hopper samples. This is a result of the clarifier HRT being eight minutes longer than one crystal pumping cycle. Therefore, the crystals did not have enough time to flush out of the clarifier or settle before the next cycle started.   86  Figure 21. Day 2 hourly liquid samples               Figure 22. Day 3 hourly liquid samples  0510152025300 1 2 3 4 5Concentration of P (mg/L)SampleEFEUSHFSHU0510152025300 2 4 6 8Concentration of P (mg/L)SampleEFEUSHFSHU 87              Figure 23. Day 4 hourly liquid samples              Figure 24. Day 5 hourly liquid samples  A dye test confirmed that the crystal pump took 40 seconds to pump into the FBR. Adding this to the HRT of the FBR, the time it would take for the crystals to reach the outlet of the FBR was 5.6-6.3 minutes after the crystal pump turned on. The samples collected after the crystal pump turned on would, therefore, be the lowest expected crystal 0510152025300 2 4 6 8Concentration of P (mg/L)SampleEFEUSHFSHU0510152025300 1 2 3 4 5 6 7 8Concentration of P (mg/L)SampleEFEUSHFSHU 88 concentration in the FBR, as the newly injected crystals would not be in the seed hopper for another 1-2 minutes. The samples collected before the pump turned on would then be higher crystal concentration values, because the crystals would not have completely washed out of the FBR. This high and low saw-tooth pattern can be seen in the SHU plots for days 4 and 5. Day 3 samples were not collected within the appropriate time intervals, so this pattern is not seen in the corresponding plot.  The SHU values do not decrease to the SHF values, indicating the cycle time was too short to allow for all the crystals to wash out of the FBR. The SHU plots for days 4 and 5 also show an increasing trend, indicating a potential build-up of crystals in the system. This increase is more noticeable at the beginning of the sampling, which could be a result of crystal retention, or agglomeration and then breakthrough, once crystal-pellet saturation is reached.  The plotted filtered samples from the clarifier (EF) and the seed hopper (SHF) for each day show they did not vary relative to each other. They also did not vary much from the control run. Figure 25 illustrates that the largest range from the control was +/- 3 mg/L of phosphorus, at the beginning of days 4 and 5; otherwise the values are negligible. This comparison confirms the injected crystals did not impact the reactor phosphorus content and operation by increasing the background concentrations.    89           Figure 25. Difference in effluent phosphorus concentration from the control   Dashed lines in Figure 22 and Figure 23, at the beginning of days 3 and 4, indicate a missed sample before crystals were injected. This missed sample most likely would have indicated a baseline at the same concentration as the filtered sample for each day. This baseline can be seen in day 5, which was sampled correctly, where all of the sample 1.0 values start at 7±1 mg/L. Assuming that baseline samples were taken for each day, then only day 2, at the 3.5 g/hour injection rate, did not show any breakthrough of crystals until 2 hours after starting the experiment. All of the higher injection rates indicate some crystals washing out of the FBR during the first cycle. Day 3, with the injection rate of 8.2 g/hour, exhibited the second lowest crystal concentration washing out during the initial cycle.  From these results, it is recommended that an intermittent crystal addition method with an injection rate between 3.5 and 8.2 g/hour would greatly enhance struvite pellet agglomeration processes in this FBR system. Further work would be required to determine the exact value and length of time between injections for optimal agglomeration. Suspension densities within the FBR would also be important to consider for crystal-pellet saturation values. -4-3-2-1012340 2 4 6 8Difference in Effluent P Concentration (mg/L)SampleDay 2Day 3Day 4Day 5 90 3.3.2.2 Pellet Agglomeration Potential An agglomeration potential could be calculated for struvite pellets in the form of a mass flux of what a pellet could agglomerate, as grams of crystal/m2 of pellet/time (g/m2*hour). This calculated value was used as evidence in the determination of the main mechanism of pellet formation, and was part of a total pellet mass growth rate determined in Ch.4. Section 4.3.1.4. The total struvite surface area contained within the control run FBR, before maximum FBR bed loading, was estimated from the masses of the last day’s harvest (seen in Appendix H). The masses were low on the 2.0 mm pellets, and would represent a growing bed after a harvest. Multiple pellets of 0.5, 1.0, and 2.0 mm from the control run were weighed. The fines were assumed to be an average size of approximately 200 μm, and their weight was estimated from 2/5th of the 0.5 mm pellet weight (see Table 7).  Table 7 Pellet Weights Pellet Size (mm) # of Pellets Mass (g) 2.0 100 0.4416 1.0 200 0.2046 0.5 400 0.1304  The total surface area in the control FBR was estimated by Equation 31.  𝐴𝐹𝐵𝑅 =  ∑ (𝑀𝑖𝐴𝑖𝑚𝑖)𝑖 (31) where Mi is the mass harvested from the FBR of size i; Ai is the surface area of the pellet of size i; and mi is the mass per pellet; where each size fraction i was 2.0, 1.0, 0.5, and 0.2 mm. The agglomeration flux was then calculated with Equation 32.  𝐽𝐴 =  𝑎𝐶𝑟𝐴𝐹𝐵𝑅 (32)  91 where a is the time in % that the crystal addition is being applied (intermittent injection of 5:10 on:off, a=0.33); Cr was the recommended crystal injection rate from Section 3.3.2.1. For the injection rates of 3.5 and 8.2 g/hour, the agglomeration flux was 0.74 and 1.73 g/m2hour, based on the total FBR pellet surface area of 1.57m2. 3.3.3 Comparison of Control and Crystal Addition This section will provide a brief comparison of the operation and performance of the two FBR SSRs, which will further validate the comparison of the pellet agglomeration results. A comparison of the harvested struvite masses and clarifier fines were expected to reveal the benefits or detriments of the crystal addition. 3.3.3.1 FBR Performance The in-reactor SSRs were calculated based on Equation 28, and were expected to be similar between FBRs. All operational data and sample concentrations for each FBR can be seen in Appendix F and Appendix G. The average difference between the unfiltered and filtered effluent samples for the control was ±1 mg/L of phosphorus, and for the crystal addition it varied between 0 to +8 mg/L with an average of +4 mg/L of phosphorus. The other ions varied in a similar manner.  Figure 26 is a comparison plot of the SSRs throughout the experiment for the crystal addition and the control with moving average trend-lines. There was a slight difference in the SSR between the control and the crystal addition for days 3 to 5, due to a slightly lower pH in the control FBR during those days. The crystal addition SSRs varied when calculated from the filtered and unfiltered effluent concentrations during days 9 to 12, due to the inability to sample without the added crystals being in solution. The true SSR for the crystal addition experiment was assumed to be the filtered SSR, based on the results of the control. There was good agreement between the control and the filtered  92 crystal addition SSRs during days 2 to 3, and 9 to 12. This ensured a valid comparison of the pellets from both reactors.            Figure 26. Comparison of the control and crystal addition FBR SSRs  Both FBRs performed very well and obtained an average phosphorus removal of 92%, based on filtered effluent samples. As shown in Figure 26, the in-reactor SSRs varied only slightly from each other, and on average the control SSR was 4.5 ± 1.5; whereas, the crystal addition was 4.9 ± 1.1. These SSRs were within the recommended values from Ch. 2 for limiting any repulsive forces, which would hinder agglomeration. The FBR nitrogen-to-phosphorus ratios were 90 ± 16 and 93 ± 10, and the magnesium-to-phosphorus ratios were 3.1 ± 0.4 and 2.8 ± 0.3, for the control and the crystal addition, respectively.  3.3.3.2 Struvite Production Struvite production between the control and the crystal addition FBR was also compared. It was hypothesized that pellet growth would be greater with the addition of 0.001.002.003.004.005.006.007.008.009.0010.000 2 4 6 8 10 12 14SSRDayControl FBRCrystal Addition FilteredCrystal Addition Unfiltered 93 crystals for agglomeration, but an excess of small crystals would be detrimental to overall pellet growth. The difference in the total harvested and sampled masses for different size fractions was analyzed. The small crystals added in the experiment (from Table 6) were subtracted from the total mass, for an adjusted total to compare growth separate from agglomeration. The initial seed added to the FBRs was neglected in this analysis because growth between specific times was being compared, not total production, and both FBRs require fully loaded beds for comparison. Harvests were carried out on the same days for each experiment. Table 8 outlines the differences in harvested masses between the control and the continuous growth with limited crystal addition on harvest days 3 and 4 for experiment days 2 and 3, and for the continuous growth during crystal addition for days 9-12. All struvite mass sample and harvest data can be found in Appendix H and Appendix I.    94 Table 8 Difference in harvested struvite masses between control and crystal addition FBRs (g) Struvite Description Experimental Day 3 4 9 to 12 Harvest 2.0 mm -20.3 -48.7 -0.8 Harvest 1.0 mm 19.9 39.4 148.6 Harvest 0.5 mm -10.9 41.6 -14.4 Harvest Fines -2.3 7.6 -144.7 pH Probe Sample -39.4 -5.8 5.1 Clarifier Fines - - -192.4 Total Mass Difference -61.6 31.7 -204.2 Crystals Added 13.0 41.9 309.9 Adjusted Total -48.6 73.6 105.7  The harvest from the control FBR on day 3 (for day 2 experiment) was approximately 150 g, and from the crystal addition FBR was approximately 200g. Adjusting for the amount of small crystals added to the experiment, the total percent difference in growth was 33% higher for the crystal addition FBR. This could be due to the slightly higher SSR than the control as mentioned in Section 3.3.3.1. The adjusted total assumes that all added small crystals were held within the FBR and were agglomerated into the pellets. This would be acceptable for day 2 experiments since the analysis of crystal retention showed very few crystals washing out of the FBR.  Day 4 harvest (for the day 3 experiment), with the higher crystal addition of 41.9 g, at 8.2 g/hour, showed a very different result. The control FBR harvest was approximately 200 g and the crystal addition FBR harvest was approximately 170 g. When the total was adjusted for the added crystals, the mass for the crystal addition FBR was 37% lower than the control. An assumption was made to demonstrate what effect the small crystals had  95 on the total growth in the FBR. It was assumed one-third of the injected crystals were retained in the agglomerates that were harvested, and the remaining injected crystals were retained in the smaller size-fractions within the upper sections of the FBR or they washed out. The one-third was derived from a combination of a comparison of the mass harvested on day 4 to the total FBR harvest on day 12 (neglecting clarifier fines), and an assumed higher agglomeration potential in the harvest zone. The harvested mass on day 4 was approximately 20% of the total mass in the FBR on day 12. A higher agglomeration potential in the harvest zone was then approximated to increase the retained crystal value to one-third; therefore, the  difference in mass for the crystal addition was 23% lower than the control for the harvest on day 4 (day 3 experiment). There were more 2.0 mm pellets harvested from the crystal addition FBR on day 4 than in the control, but there were less of the other size fractions than the control. This might just be the natural variation in FBRs, or due to the larger sample from the pH probe sampling port on day 3 in the crystal addition FBR affecting the size distribution. The control SSR for day 4 was also slightly lower than the crystal addition FBR, which should have resulted in lower growth for the control. This analysis demonstrates that the 42 g of added crystals, with continuous growth with limited crystal addition of 8.2 g/hour, had a slight, detrimental effect on the total growth within the FBR. For the continuous growth during crystal addition conditions on days 9 to 12, the total struvite removed from the FBR for the four days was compared to the control for the same period. The crystal addition for days 9 to 12 was on average 9.3 g/hour, for a total of 309.9 grams over 33.7 hours. The control FBR and clarifier were completely emptied at the end of day 12 and contained a total of 1177 grams of struvite. The crystal addition FBR and clarifier contained 1381 grams of struvite. The main differences in masses are seen in the individual size fractions. The control had larger amounts of the 1.0 mm pellets, whereas the crystal addition FBR had more fines. This size distribution difference seems  96 reasonable for the processes taking place in the two reactors. Lower pellet crystal growth would be expected for FBRs with an excess of fines. The extra surface area from the small crystals would grow and compete for ions in the FBR, limiting pellet growth. Adjusting the total struvite mass collected for the added crystals, the total difference in mass was 9% higher for the control. The fines accumulated in both the FBRs and in the clarifiers for the duration of the 12 days, not just for days 9-12. Therefore, there would be fines accumulated from the previous days of crystal injection. If the clarifier fines are removed from the analysis, the masses are 1106 g and 1118 g for the control and crystal addition FBRs, respectively. Adjusting these totals for the added crystals, the total difference in mass was then 27% higher for the control. The amount of time for just crystal growth was greatly reduced in the continuous growth during crystal addition conditions. The pellets were saturated with small crystals and not given time for growth. The smaller size agglomerates and the lower amount of growth, again confirms the negative influence of excess fines in the FBR for the formation of pellets.  3.3.4 Clarifier Particle Size Distribution The size distribution of the crystals that washed out of the FBRs into the clarifiers was measured to determine how they related to the crystal sizes retained in the FBR and crystals produced in the nucleation experiments in Ch. 2. It was hypothesized that primary nucleated crystals do not stay within the FBR, and are washed out into the clarifier if they do not agglomerate onto a pellet. Figure 27 shows the particle size distributions of the clarifier fines collected from the control and the crystal addition FBRs, compared to the small crystals added in the experiment. The d(0.5) of the small crystals and the crystal addition FBR were 29 and 28 μm respectively, and the control was 68 μm. The distributions show between 82-88% by  97 volume of the crystals in the clarifiers was less than 150 μm. It can be concluded that the fines in the clarifier, for the crystal addition FBR, were mostly made up of the small crystals that were added during the experiment, and did not appear to have grown to the size that the control clarifier crystals did.  Crystals from the nucleation study in Ch. 2 were < 20 μm for SSR 6 at the end of the 40-minute experiment. If primary nucleation occurred in the FBR, a crystal less than 20 μm could be produced since the retention time of the FBR is only six minutes. These crystals would be much too small to withstand the velocities in the FBR and would be flushed out and collected in the clarifier. The analysis of crystal sizes in the clarifier show that the birth of a pellet cannot occur from primary nucleation since they would be washed out of the FBR; the cores of pellets must originate from broken pellet branches, and will be discussed further in the following section.  Figure 27. Particle size distribution of clarifier fines compared to the injected small crystals   012345670.1 1 10 100 1000 10000Particle Size Distribution (Volume %)Particle Size (µm)Control FBRCrystal Addition FBRSmall Crystals 98 3.3.5 Pellet Development Struvite pellet formation in an FBR is a continuous, self-seeding process. It was hypothesized that a birth of a pellet was from larger attrition pieces of pellet branches that could withstand the velocities in the FBR, and not from primary nucleation. It is apparent from the seed hopper samples in this study that birth of a pellet occurs with a section of a branch breaking off a pellet, and then becoming the core of a new pellet. Figure 28 shows a pellet mounted in resin, cut approximately in half, and polished This slice of a pellet illustrates how the center appears to start from a compact core, with a different growth condition than the branches. This compact core is in contrast to that reported in the literature (Fattah et al., 2012; X. Ye et al., 2016; Z. Ye et al., 2018).   Figure 28.  [Sample13_bse_01] Resin mounted pellet at 24x magnification  If the newly born pellet is large and dense enough to have a settling velocity that is equivalent to the up-flow velocities in the top of the FBR, it will stay fluidized and will not be washed out. The samples from the junction of the fines zone and the seed hopper are approximately 200-300 μm and were fluidized at a velocity of 0.0067m/s (40 cm/min). Growth of the cores would be from small crystals agglomerating onto it, and/or it slowly growing branches by crystal growth.   99 It is hypothesized that the new pellet grows a loose branching structure close to the core in low SSR conditions at the top of the FBR. Once the pellet becomes heavier and moves lower in the FBR, it will experience higher SSRs and growth will accelerate. The branches then develop a tighter structure as they grow and experience higher SSRs and velocities in the lower sections of the FBR. Figure 29 illustrates this pellet growth concept in a 4-stage UBC struvite FBR.   Figure 29. Pellet development in FBR systems with primary nucleation occurring   Attrition Agglomeration Size Enlargement Migration High Growth Primary Nucleation      100 3.3.6 Hard Pellet Surfaces Pellet hardness was defined as one of the three most important qualities of the ready-to-use struvite fertilizer produced in the UBC FBR. The pellets need to be durable and able to withstand handling, shipping, and machine application. The commercially produced struvite pellets are much harder and denser than the ones produced in the pilot-scale FBRs used for research, so the physical structure was investigated in more detail to try to understand this difference.  A sample pellet from one of the commercial OstaraTM FBRs was imaged to compare it with pellets analyzed in this study. The Ostara pellet surface reveals similar characteristics to the pellets described in literature and to those grown in this study. Figure 30 (a) and (b) reveal a surface that appeared to be an agglomeration of very small crystals. There were crystals that resemble the tops of orthorhombic branch structures, but were at a much smaller scale than in the pellets previously discussed. Measurements of the smallest crystals on the surface were less than 10 μm, and the gaps and spaces between crystals, where agglomeration could take place were approximately the same size. The larger tops of crystals were 30-40 μm in length. When cut open and magnified 1000 times, a tight pattern, possibly resembling a branching structure was barely visible (Figure 30 c).    101  Figure 30. Hard Ostara pellet surface at (a) [Sample47_bse_01] 42x magnification; (b) [Sample47_bse_03] 600x magnification; (c) [Sample12_bse_02] cut and polished pellet at 1000x magnification.  For a hard coating to develop on struvite pellets, it has already been discussed that it is hypothesized that a higher velocity is required in the FBR. Figure 30 reveals that harder pellets contain very small crystals. The hard pellets have very small gaps and corners, so a very small crystal would be required for agglomeration. Attempting agglomeration with hard pellets would be futile, unless using primary nucleation to produce very small crystals. It is, therefore, speculated that harder pellets form by crystal growth and not agglomeration. 3.4 Chapter 3 Summary and Conclusions Struvite pellet formation has always been assumed to be an agglomeration of primary, nucleated crystals. This work aimed to de-couple struvite crystal growth from (a) (b) (c)  102 agglomeration processes within an FBR system. Low SSRs between 2-6 were required to promote agglomeration within the zeta potential range outlined in Ch.2 and restrict primary nucleation. The controlled addition of small crystals into an FBR proved that struvite pellet formation can be a combination of agglomeration and crystal growth, and the two processes can be separated. This means that the process of agglomeration can be controlled; therefore, pellet formation can also be controlled and fines losses from FBRs could be reduced.  Struvite pellets form a preferential morphology consisting of a branching structure at low SSRs where only growth processes are occurring. Agglomeration requires a source of smaller crystals and a larger crystal structure for them to adhere to and become fixed. Primary nucleated crystals are too small to withstand the up-flow velocities in the FBR, and are generally washed out of the FBR unless they become agglomerated onto a pellet. Therefore, struvite fines created by primary nucleation can only be used for agglomeration. Attrition creates larger fines than nucleation, which then turn into the cores of new pellets through agglomeration of primary nuclei and/or crystal growth. Crystal addition rates of between 3.5 and 8.2 g/hour, for an intermittent injection method, are recommended to promote agglomeration processes in FBRs operated in growth conditions. Excess fines were found to be detrimental to total struvite production. Further research is required to determine the optimal injection and growth cycle parameters for this method.  For researchers and industry utilizing FBRs for struvite pellet growth, the results of this study will allow them to: (1) analyze fines in the effluent to determine if primary or secondary nucleation is overwhelming the reactor; (2) analyze pellets to determine which formation processes are occurring; and (3) determine if operational conditions require adjusting.  This was the first study to separate struvite agglomeration and growth processes in an FBR. The separation of these processes showed that the underlining branching  103 structure found in this study was required for agglomeration of the small crystals. The low agglomeration potential, between 0.74 and 1.73 g/m2hour for 1.57 m2 of pellet surface area in the FBR, indicated that the main mechanism of pellet formation was growth, not agglomeration as previously thought. Individual crystal growth rates were investigated and reported on in Ch. 4 to further the knowledge of the pellet branching structures, and to provide additional evidence that growth is the main mechanism of struvite pellet formation in an FBR.      104 Chapter 4 Struvite Crystal Growth and Morphologies: The Influence of Concentrations, SSR and Velocity Phosphorus recovery in the form of magnesium-ammonium-phosphate-hexahydrate, more commonly known as struvite, is well on its way to becoming a mainstream process. To prevent eutrophication, countries are implementing tougher regulations to limit nutrient run-off and release into waterways. Technologies that recover struvite will become more efficient as demand increases and investment allows for more optimization. Currently, fine struvite crystals get washed out of fluidized bed reactors (FBR) instead of forming into pellets. The crystals are lost to recovery, and can account for 10-30% of struvite losses depending on FBR operational conditions (Mavinic et al., 2007; Shimamura et al., 2007; X. Ye et al., 2016). There is still a significant knowledge gap in understanding struvite pellet formation with respect to crystal growth. Struvite pellet formation has been accepted to be an agglomeration of crystals, but the physical growth and morphology have had little investigation. A pellet branching structure, identified in Ch.3, is the preferential crystal growth morphology for a struvite pellet at low SSRs. This newly identified branching structure was the motivation to investigate how growth rates and morphologies change with various FBR operating conditions. Physical understanding of how the crystals are growing will enable optimization of pellet formation, while reducing fines losses in FBRs.  In the worked described in this chapter, new methods were developed to obtain radial growth rates for individual struvite crystals and individual pellets, and morphologies were documented for varying conditions.  105 4.1 Background 4.1.1 Struvite Morphology There are many forms that struvite can take: pellets; agglomerates; single crystals; or multiple crystal formations. Crystal habit is defined as the morphologies or forms expressed by different crystal species. Struvite morphologies vary depending on the conditions for growth, but the crystal is classified under the orthorhombic crystal system with a hemimorphic habit along the c-axis. This difference on either end of the crystal establishes the typical coffin-shaped crystal habit seen in struvite studies. Figure 31 visually illustrates the different morphologies reported in the literature, and is followed by descriptions of each. There are two distinct sets of observations of struvite habit: the laboratory form; and the process, or in-situ habits. How the habits change is based on the system and solution conditions that produce the crystals. Most of the detailed laboratory morphology work has been undertaken in the field of urology, whereas the process work has been in the field of environmental engineering. Small single crystals are generally produced in a stirred vessel with twinning or dendrite crystals at high pH values. FBRs can produce single crystals, as well as agglomerates containing smaller crystals and branching structures. Figure 31 and the following definitions are current variations of struvite habits seen in the literature, and will be used to describe and identify crystals grown in this study.   106  Figure 31. Review of struvite morphologies from literature; laboratory observations, and process or in-situ observations1                                                1-16 see preface for copyright acknowledgments   107  Branching structure: Low SSR: compact, thin coffin-shaped or pyramidal crystal formations growing on top, and out from each other. This habit is found to radiate out from a central core of a pellet as seen in Ch. 3. An arborescent or “branching tree-like” crystal formation.  High SSR: combination of needle and dendrite crystals radiating outwards and growing from a larger, central needle, which acts as the base of the primary agglomerate (Z. Ye et al., 2014). Coffin shaped: A well-defined hemimorphic (non-symmetrical) morphology along the c-axis, where the (001) and (001̅) faces differ (see Figure 32). The (001) face is the smaller face of the two, which creates the coffin shape. The other symmetrical faces of the typical morphology (012) & (01̅2), (010) & (01̅0), (101̅) & (1̅01̅), (101) & (1̅01) can increase or decrease in growth depending on SSR, which influences the shape or existence of the (001) face (Prywer & Torzewska, 2009). This is a well-studied habit that is commonly produced in laboratory experiments at low SSRs.      108    Figure 32. Coffin shaped struvite crystal morphology along axis: a) b-a, b) b-c, c) a-c. Image from (Prywer & Torzewska, 2009) 2  Dendrites: Thin, needle-like crystals that can occur in patterns similar to feathers or snowflakes. They can be singular or in a branching structure, but less tree-like as the full branch description. Dendritic morphologies will occur in regions of high SSR (Sunagawa, 2005). Needles: Very thin, long individual needlelike structures; tips appear pointed. Higher SSR causes elongation into rods and needles. Little to no investigation of the physical structure has been carried out. Pellets: A round habit, greater than 0.5 mm, consisting of loose, or compact struvite crystals. Low SSR: branching structure as described above growing from a central core outwards. Medium SSR: a combination of the branching structure, and individual crystals bonded together into secondary agglomerates. (Reference Ch. 3) Plates: Flat plates, or block-like crystal structures have been found as part of larger dendritic structures as illustrated in Figure 33 (Prywer et al., 2012).                                                2 see preface for copyright acknowledgments  109   Figure 33. Plate structures identified by Prywer et al (2012) 3  Primary agglomerates: Growth of more than one crystal as a whole; not agglomerated by individual crystals. For example dendritic, star, or twin morphologies (Jones, 2002), where higher SSRs are required to form these examples of habits. Prismatic: A rare habit of struvite, which is not found in the engineering literature. Prism geometry, where two faces of the crystal are equal and parallel, and two sets of six angled sides join the parallel faces.  Pyramidal: A multi-sided base with a four-sided pyramid extending out from the base is a typical description of this type of habit in crystallography literature. Previously this crystallography class of struvite was not commonly seen in the engineering literature, but is described by crystallography experts (Hudson Institute of Mineralogy, 2018) and includes more faces than the coffin habit exhibits, as seen in Figure 34. Another depiction of the struvite pyramidal habit is modeled in Figure 35, which is similar to the coffin habit, but is shown with the (001) face non-existent as others have determined can occur depending on growth conditions (Prywer & Torzewska, 2009).                                                 3 see preface for copyright acknowledgments  110   Figure 34. Pyramidal 3D Model 36 of struvite by (Holtkamp, 2014) 4                                                                       Figure 35. Pyramidal 3D Model 47 of struvite by (Holtkamp, 2014) 5                                                 4 see preface for copyright acknowledgments 5 see preface for copyright acknowledgments  111 Stars: More than two hemimorphic crystals joined as penetration twins at 60 or 90 degrees from each other (Prywer & Torzewska, 2009). This habit is seen at very high SSRs. Rods: Not very well defined in literature. The example in Figure 31 appears to be 6-sided, and could be elongated coffins. Increasing the SSR will likely cause elongation into rods and needles. Twinned or X-Shape: Two hemimorphic crystals can join on the (001) face, called the twin axis, in a mirror type formation. They can also form penetration twins where one crystal is rotated 60 or 90 degrees and cuts through the other crystal (Prywer & Torzewska, 2009). This habit is seen at high SSRs. 4.1.2 Struvite Growth Struvite crystal formation is a precipitation and growth phenomenon from solution, and growth rates from solution can be determined for both single crystals or bulk crystals. A linear growth rate is generally used for growth of individual faces of single crystals, where the length (L) can be measured over time (t), as per Equation 33  𝐺 =  𝑑𝐿𝑑𝑡 (33) Either measuring the change in concentration, or change in mass over time, can determine bulk growth rates. Some determinations of bulk struvite growth rates reported in the literature have been carried out in stirred reactors (Ariyanto et al., 2014; Galbraith et al., 2014; Harrison et al., 2011; Mehta & Batstone, 2013), in FBRs (Bhuiyan et al., 2008), or in-situ with coupons in a wastewater treatment plant (Ohlinger et al., 1999). All of the bulk methods measure growth rates and growth coefficients indirectly, and require the addition of seed crystals to the reactor. The linear growth rate equation is converted into Equation 34, for use in bulk growth-rate determinations.  112  𝐺 =  𝐿𝑀𝑉3𝑊(−𝑑[𝐶𝑖]𝑑𝑡) (34) Where L is the average diameter of the crystals; M is the molecular weight of struvite; V is the solution volume; W is the total mass of the crystals; Ci is the limiting ion concentration; and t is time (Bhuiyan et al., 2008). Using a ‘birth and spread’ model of surface integration by nuclei formation and growth, the linear growth Equation 33 requires the addition of supersaturation, and is modified to Equation 35 and 36.   𝐺 =  𝑑𝐿𝑑𝑡=  𝑘(𝑆𝑆𝑅)𝑛 (35)   𝑛, 𝑘 =  𝑓(𝑇, 𝐿, 𝑁, … ) (36) Where k and n are growth kinetic constants which are a function of temperature (T), mixing intensity (N), and the mean crystal seed size (L) when industrial crystallizer conditions can be correlated (Ali & Schneider, 2008; Jones, 2002). Theoretical values used for n are dependent on the type of growth: diffusion controlled growth n=1; growth from screw dislocations n=1-2; and polynuclear growth n>2 (Jones, 2002). Relative supersaturation (σ = SSR-1) can be used instead of SSR, as this zeros out the equilibrium SSR value of 1 to equal 0, for ease of determining k. Struvite growth is assumed to occur by the two-stage, diffusion-reaction model. First, there is a diffusion step, and then a surface reaction or integration step, in series. The diffusion step encompasses molecules moving through the liquid and reaching the solid surface. This mass transport from the solution is controlled by fluid velocity, and can limit the integration step. The surface integration process requires the molecules to organize into the crystal lattice. The driving-force, SSR, controls the rate of the integration step. A supersaturated solution is required before either the diffusion or surface integration steps will proceed.   113 SSR has been studied in great detail for its impact on bulk growth rates, but there is a large knowledge gap for struvite growth relating to velocity. Experimental growth studies, for crystals other than struvite, utilizing a rotating single crystal through a solution, show low and high variations in the surface SSR (Derby, 2016). This demonstrates the importance of solution flows and how they can affect crystal growth. Liquid velocity affects crystal growth rates, which can be observed more easily at higher SSRs due to limited surface integration control. It has been found that different crystal faces grow at varying rates, contributing to different habits (Garside et al., 2002; Jones, 2002), which could also be due to velocity differences on those faces. A key piece of literature states that struvite crystal growth is highly influenced by mass transport, with the growth rate dependent on mixing intensity (Ohlinger et al., 1999). Another study similarly observed that with an increase in stirrer speed, struvite crystal growth increased (Ariyanto et al., 2014).  Since it is clear from the literature that the mass-transfer boundary layer and the mass-flux depend on fluid velocity (Jones, 2002), a brief discussion about diffusion flux will illustrate how the solution velocity can influence crystal growth rates. Consider a struvite pellet surface growing by the diffusion-reaction model of precipitation. The first step includes molecules diffusing through the diffusion-boundary layer to reach the pellet surface. The diffusion step is controlled by the thickness of the boundary layer and the magnitude of the driving force or gradient is applied. The rate of flow of ions per unit area through the boundary layer is the diffusive flux. The diffusive flux (J), from Fick’s first law, can be approximated by linearizing the concentration distribution as seen in Equation 37.  𝐽 =  −𝐷𝑑𝐶𝑑𝑥 = (𝐶𝐼 −  𝐶𝑆)𝐷𝐿 (37)  114 Where D is the diffusion coefficient; CI is the concentration at the interface between the surface and the solution; CS is the concentration in the bulk solution; and L is the boundary layer thickness (Jackson, 2010). This simplified example neglects geometrical spherical effects and assumes a steady state, so the flux is assumed constant at the interface of the crystal.  Figure 36 illustrates two concentration gradients for visual comparison: a) the solid line is a general depiction of a concentration gradient with the boundary thickness labeled Ls for slow velocity; and b) the dotted line is a concentration gradient with a faster solution velocity and boundary layer thickness labeled Lf. The faster velocity contributes to a steeper concentration gradient, which reduces the thickness of the diffusion-boundary layer, and increases the diffusion flux according to Equation 37. This increase in the diffusion flux for higher solution velocities could result in higher growth rates if the surface integration step is not the rate-limiting step.    Figure 36. Varying diffusion fields around a precipitating and growing crystal  115  It was discussed in Ch. 3, that higher velocities change the morphology of struvite pellets; therefore, it must also change the morphology of the struvite crystals making up the pellets, and possibly increase the crystal growth rates. No research to date has grown individual struvite crystals to determine how growth rates and morphologies change by varying typical parameters that are adjustable in FBRs (such as concentrations, SSR, and fluid velocity). Many laboratory urology studies, with bacteria-induced slow growth of single crystals, produce the well-known coffin-shaped struvite crystals, but the studies are limited in the information for industrial application. The objectives of this study were to determine growth rates and morphologies of stationary and rotating individual crystals under varying conditions, to see how they relate to the branching structures found in pellets, and thereby determine the best conditions for pellet growth while limiting fines losses. 4.2 Materials and Methods The work in this chapter was divided into three main sections: stationary crystal growth; rotational crystal growth; and bulk FBR pellet growth and their corresponding morphologies. 4.2.1 Stationary Crystal Growth  Stationary struvite crystal growth experiments were carried out to investigate morphology changes, and growth rates for varying SSR and velocity conditions for two chemical compositions. Crystals were attached to a rigid holder while sustained SSR and velocity conditions were systematically varied. The different experimental settings represented conditions pellets could experience within an FBR. Crystal holders were made out of a 12x10x20 mm stainless steel block, machined to allow flow through, and to support a wire. The wire was a tungsten needle cleaning wire  116 of diameter 0.089 mm, made by Hamilton Company, and was spot-welded onto the bottom of the crystal holder.  A flow cell was made from a 47 mm Fisherbrand petri dish. A hole was cut in the bottom of the petri dish and a cover slide was glued over it, allowing for the use of an inverted microscope to view the crystals growing. A cutout in the sidewall of the petri dish was fitted with a tube to allow liquid to flow out of the cell. Flow into the cell was from tubing attached to a Masterflex pump, and was supported externally. Figure 37 illustrates the setup.                                             Figure 37. Stationary crystal growth: crystal holder in flow cell   Struvite seed crystals used in the experiments were produced in the control FBR run described in Ch. 3. Crystals were harvested out of the seed hopper, and therefore, they represent the smallest crystals fluidized within the FBR. The crystals were formed by breakage of pellets, and they were identified as the beginning, or core, of a pellet within the FBR. These crystals were produced in growth conditions in which no agglomeration occurred. The crystals were not perfect single crystals, but were comprised of broken branching structures of various shapes, sizes, and defects (as seen in Figure 39 and Figure 40 in the Results Section 4.3.1). Growth of these crystals occurred on multiple faces; therefore, it represented the mean growth of a pellet surface, unlike typical single crystal growth experiments, whereby single faces are isolated. The crystals were glued  117 onto the wire of the holder with Advanced Krazy Glue(TM) on the side of the wire orientated into the flow. Crystal orientations were random; therefore, multiple faces were exposed to direct flow. The holder, with the crystals glued onto the wire, was placed on the cover slide in the flow cell. Two chemical compositions representing the extremes within an FBR, which a growing pellet could experience, were used in the experiments. A synthetic composition representing centrate from digested activated sludge was the basis for the two chemical compositions. The synthetic centrate with average concentrations of NH4-N of 800 mg/L and PO4-P of 100 mg/L was used directly, and was labeled the “no-mix” conditions. A “full-mix” condition was modeled after a fully-mixed FBR, with dilution from the recycle flow (recycle ratio of 7) to concentrations of NH4-N of 700 mg/L and PO4-P of 20 mg/L. The magnesium concentrations for the full-mix and no-mix conditions were set at 50 and 95 mg/L for a Mg:P ratio of 3.2 and 1.2, respectively. The full-mix composition incorporates the FBR recycle line diluting the centrate, and is the assumed normal FBR operation with complete mixing in the injector port; whereas, the no-mix at full strength could happen with improper mixing at a micro-scale, or if a single-pass FBR without recycle were used.  Two stock solutions for each concentration composition were prepared twice as strong in distilled water; one solution contained analytical grade of ammonium dihydrogen phosphate (NH4H2PO4) and ammonium chloride (NH4Cl), and the other solution magnesium chloride hexahydrate (MgCl26H20). The pH of each solution was raised to the required value for the specific SSR target with sodium hydroxide. The temperature was controlled between 24.5 ±1.5 °C. A Masterflex variable pump, with two #13 pump heads, provided flow from each chemical solution container into a joining T-fitting. The solutions were mixed in the T-fitting and pumped into the flow cell by tubing with an internal diameter of 0.8 mm. The tubing  118 outlet was positioned directly at the crystals on the wire, with approximately 12 mm distance away, with a length to width ratio of 15.  Two velocity regimes were selected to represent a range of what pellets could experience in the bottom of the harvest zone, close to the injector port of the UBC research FBRs. The velocities were 1.39 m/s and 0.70 m/s. The range was chosen to be approximately ±30% the average velocity exiting the injector port into the harvest zone. These values represent what a pellet could experience in the flow transition zone of the harvest zone with the lower velocity, and during an injector-plugging event, which would increase the velocity. Appendix J lists various models and dimensions of the UBC research FBRs, the back-calculated injector port velocities associated with the typical harvest zone velocity, and the percent reduction in injector port area for the 1.39 m/s velocity. Calculations for the injector port velocity (VI) based on the harvest zone velocity can be seen in Equation 38.  𝑉𝐼 =   𝑄𝐻𝐴𝐼=  𝑉𝐻 𝐴𝐻𝐴𝐼 (38) where QH is the flow through the harvest zone; VH and AH are the velocity and cross-sectional area of the harvest zone; and AI is the cross-sectional area of the injector port. Four sets of conditions, including full-mix and no-mix, each with high and low velocities, were examined at SSR ranges between 3-5, 6-9, and 13-15. Experimental duration varied due to the fragile nature of the crystals, especially at high SSRs where the crystals became too large and broke off. The appropriate duration was determined by visual inspection with a microscope during the experiments, and it varied from 7 to 90 minutes. Solution samples 50 mL volumes were collected from the outlet of the flow cell and acidified with one to two drops of a 1:1 HCl:H2O solution. Conductivity, pH, and temperature were measured in the effluent with a hand-held Horiba pH/conductivity D-54  119 meter. The holders with the crystals were rinsed immediately with methanol after they were removed from the solution. This was done to reduce any drying crystallization that could occur, and as a cleaning requirement for SEM imaging.  Two sets of images were taken before and after each experiment: one where the flow was from the side; and the other where the flow was into the page, or into the crystals. The images were captured with an Environmental Scanning Electron Microscope (SEM) FEI Quanta 650 VP using the secondary electron detector (indicated by LFD on the image), and some images were using the backscatter detector (indicated by CBS on the image) in low vacuum mode. Each experiment produced four sets of nine SEM images for the entire wire length.  4.2.2 Rotational Crystal Growth Pellets are not stationary in FBRs; they spin and move around in a random fashion. This spinning along with the fluid velocity was thought to be a contributing factor to the surface finish, either by producing small sizes of crystals, or filling in gaps between branches leading to a harder outer surface. The spinning of a pellet, and the relative velocity a pellet experiences, depends on the fluid velocity and the amount of turbulence in each FBR section, and these both decrease as the FBR height increases. It was thought that with more rotation or higher relative velocity, there could be a change in the crystal growth seen in the stationary crystal growth experiments; therefore, rotational pellet growth experiments were attempted.  Pellets, approximately 1 mm in size, from the control FBR run in Ch. 3 were glued, with Advanced Krazy Glue(TM), onto the end of a chromatographic syringe needle made by Hamilton Company. The needle, with the pellet, was inserted through a slit into the center of a vinyl tube with the internal diameter of ¼” (6.35 mm), and was held in place by pressure on the tube with a bumper glued onto the shaft of the needle. The tube was 170  120 mm long to where the needle was inserted, was mounted on a board to hold it in place, and had a T-fitting on the inlet to allow for the two solutions to mix, similar to the stationary growth experiments. A hose fitting was positioned over the end of the tubing to hold the needle in place and act as an outlet. The needle was clamped into a universal joint and splined to a Masterflex pump with an RPM digital readout. The needle was rotated at 400 RPM for all experiments. Flows in the tube ranged between 200 mL/min to 2200 mL/min, with SSRs between 5.5 and 8.7. Reported relative velocities, which the pellet experienced, were calculated by adding the rotational velocity to the average fluid velocity in the tubing. Solutions were mixed in the same manner as the stationary crystal growth experiments, but only the full-mix concentrations were used. Figure 38 illustrates the experimental setup.                                                Figure 38. Rotational pellet growth experimental setup  4.2.3 Bulk FBR Pellet Growth and Internal Structures Bulk growth rates were calculated from the difference between the added seed sizes, and the pellet sizes from the first FBR harvest of the control run from Ch. 3. The  121 internal structures of the pellets were further examined. Pellets were cut open and imaged with an Environmental Scanning Electron Microscope (SEM) FEI Quanta 650 VP. 4.2.4 Image Analysis  For the stationary crystal growth experiments, each set of nine images was stitched together into one integrated image with Adobe Photoshop CC. The “after image” was overlaid onto the “before image”, for each flow direction for each experiment. Matching crystals from before and after growth were masked for projected area and circularity measurements. Crystals that had fallen off the wire, moved, were obscured by the wire, or overlapped each other were not utilized. For the rotational crystal growth experiments, pellets from before and after growth were masked for projected area and circularity. Scale bars on all cropped images are 100 μm, unless otherwise specified.  4.2.5 Calculations  4.2.5.1 Supersaturation Ratio Struvite is a transparent to semi-transparent crystalline material made up of equal molar concentrations of magnesium, ammonium, and phosphate with six waters of hydration. Struvite precipitation is controlled by the saturation of solution with respect to ions that form the crystals lattice, and precipitates as per Equation 39.  𝑀𝑔2+ +  𝑁𝐻4+ +  𝑃𝑂43− + 6𝐻2𝑂 =  𝑀𝑔𝑁𝐻4𝑃𝑂4 ∙  6𝐻2𝑂 (39) The supersaturation ratio (SSR) represents the extent of crystallization that must occur in order for the system to reach equilibrium, and can be described as the driving force of the precipitation process. The SSR is the ratio between the ion activity product (IAP) and the solubility product (Ksp), as seen in Equation 40.  𝑆𝑆𝑅(𝑠𝑡𝑟𝑢𝑣𝑖𝑡𝑒) =  {𝑀𝑔2+}{𝑁𝐻4+}{𝑃𝑂43−}𝐾𝑠𝑝(𝑠𝑡𝑟𝑢𝑣𝑖𝑡𝑒)=𝐼𝐴𝑃𝐾𝑠𝑝 (40)  122 A pKsp value [𝑝𝐾𝑠𝑝 =  −𝑙𝑜𝑔10(𝐾𝑠𝑝)] of 13.47 was used in this project (Lobanov et al., 2013). All struvite SSR values were modeled using PHREEQC software. 4.2.5.2 Radial Crystal Growth The crystals in the experiments were assumed to be spherical, so a radial growth rate could be calculated from the projected area in the before and after crystal growth SEM images. The linear growth rate equation is generally used for growth of individual faces of single crystals, where the length (L) can be measured over time. In spherical coordinates the radius (r) is measured over time, as per Equation 41  𝐺 =  𝑑𝐿𝑑𝑡=  𝑑𝑟𝑑𝑡 (41) 4.2.5.3 Diffusion and Mass Flux The Law of Conservation of Mass states that mass cannot be created or destroyed, it can be accounted for in the products and reactants of chemical reactions within an isolated system. Struvite precipitation reactions convert mass of the individual ions in solution that make up the crystal lattice structure into solid crystal formations. When the diffusive flux, discussed in Section 4.1.2, limits crystal growth in the two-stage, diffusion-reaction model, the growth rate normalized by the surface area can be equated to the diffusive flux using conservation of mass according to Equation 42.    𝑑𝑀𝑑𝑡∙  1𝐴𝑠=  −𝐷𝑑𝐶𝑑𝑥  (42) The mass rate of growth of the pellet can be computed as an equivalent flux according to Equation 43 (Jackson, 2010).  𝐽𝑚 =  𝑑𝑀𝑑𝑡∙  1𝐴𝑠=  𝑑(𝑉𝜌𝑓𝑣)𝑑𝑡∙1𝐴𝑠=  𝑑𝑉𝑑𝑡∙𝜌𝑓𝑣𝐴𝑠 (43) Where Jm is in units of kg/(m2*s) or g/(m2*s) (mass of struvite/pellet surface area/time); dV/dt is the volume change of a spherical crystal; ρ is the density of struvite  123 (1.77 g/cm3 (Katz, 1972) ); 𝑓𝑣 is a void factor dependent on the crystal or pellet surface (determined from the SEM images); and As is the surface area of the sphere. Approximating the shape of the crystals as spherical, the volume change is found using Equation 44.  𝑑𝑉𝑑𝑡=  𝑑𝑉 𝑑𝑟∙𝑑𝑟𝑑𝑡=  𝑑𝑑𝑟(43𝜋𝑟3) ∙𝑑𝑟𝑑𝑡= 4𝜋𝑟2𝑑𝑟𝑑𝑡 (44) From the radial change in the SEM images, the radial growth rate was calculated, as per Equations 45 and 46  𝑑𝑟𝑑𝑡=(𝑟2 −  𝑟1)𝑡 (45)   𝑟𝑖 = √𝐴𝑖𝜋𝑛 (46) where i = 1 or 2; r2 is the radius after crystal growth; r1 is the radius before crystal growth; t is the time of the experiment; A is the projected area from the SEM images either before or after growth; and n is the number of crystals per experiment in the projected area measurement.  4.2.6 Analytical Methods Liquid samples were analyzed for orthophosphate and ammonia by using a LACHAT QuickChem FIA+ 8000 Series with a ASX 500 Series Autosampler, according to UBC method EE-SOP# A.04.01 ‘Ammonia, Nitrate and Phosphate Determination in Water by Continuous Flow Analysis’ with reference to Standard Method 4500-P G and 4500-NH3 H (Eaton et al., 1995). Magnesium was analyzed by atomic absorption spectroscopy with a 220 Fast Sequential Varian Spectra AA with a Varian Autosampler model SPS-5, in accordance with Standard Method 3111 B (Eaton et al., 1995).   124 4.3 Results and Discussion Struvite growth rates and morphologies vary depending on ion concentrations in solution, supersaturation levels, and velocity of the solution. The following three sections summarize the experimental findings from the stationary crystal growth experiments, rotational crystal growth experiments with individual pellets, and for bulk pellets grown in the FBR control run from Ch. 3, with the main focus on the stationary crystal growth experiments. A comparison of growth rates is then presented. 4.3.1 Stationary Crystal Growth Projected areas of the before and after crystal growth were measured from the two directions and averaged, for most of the experiments. Overall radial growth rates were then calculated for each experiment. A visual summary is compiled in Figure 39 and Figure 40 for the full-mix and no-mix conditions, respectively. Examples of individual crystal images from four experiments, for both the full-mix and no-mix conditions, are organized by low and high SSR and velocity. Each experiment is grouped into four images labeled (a) through (d). The (a) and (c) images are before growth, and the (b) and (d) images are after growth. The fluid flow is from the right for the (a) and (b) images, indicated with the arrow symbol in the direction of flow. The fluid flow is into the page for the (c) and (d) images, indicated with the cross symbol. The scale bar for all images is 100 μm, and the figure headings contain details about each experiment. It can be seen that all seed crystals vary in shape, and many have the distinct pellet branching structure observed in Ch. 3.  125  Figure 39. Full-mix concentration growth examples (exp#, SSR, mins, dr/dt):   2 – FMHV-7, 3.02, 55, 0.14;                              4 – FMHV-17, 12.88, 30, 0.55;                  1 – FMLV-15, 4.27, 60, 0.13;                             3 – FMLV-21, 12.88, 60, 0.40    126  Figure 40. No-mix concentration growth examples (exp#, SSR, mins, dr/dt): 2 – NMHV-27, 4.90, 25, 0.61;                               4 – NMHV-28, 13.49, 10, 4.40;              1 – NMLV-26, 4.90, 30, 0.48;                               3 – NMLV-11, 12.88, 10, 1.90  Many types of microscope imaging techniques were investigated in the attempt to collect real-time growth rates. No optical equipment was capable of this due to the struvite crystals being translucent in solution, and the surfaces reflecting and refracting light in all directions. The depth of field with the optical equipment was compromised for image quality, thereby limiting observation of the three-dimensional growth. Real-time growth  127 rate measurements could provide additional information in the initial stages of formation of the preferential habit of the crystals, and potentially enable real-time habit modification for optimal growth.  4.3.1.1 Growth rates Two chemical compositions were used to determine the influence of SSR and fluid velocity on crystal growth. Stationary struvite crystal growth rates were expected to increase with an increase in the chemical concentration, SSR, and fluid velocity.  All growth rates increased with an increase in SSR, and all but one condition showed a linear dependence on SSR. There was a slight exception to the linear dependence on SSR for the full-mix, high-velocity conditions due to inconsistent experimental durations. This anomaly will be discussed in detail in Section 4.3.1.6. Figure 41 and Figure 42 show the radial growth rates for the full-mix and no-mix conditions, respectively. All data and calculated values can be seen in Appendix K and Appendix L.  Figure 41. Full-mix radial growth rates   128  Figure 42. No-mix radial growth rates  The growth rates between the two solution concentrations are an order of magnitude different. This was expected since the phosphorus concentrations were 20 and 100 mg/L for the full-mix and no-mix conditions, and phosphorus is the limiting ion in the experiments. This confirms that struvite growth is not just a function of SSR, but also a function of solution concentrations or mass transport within the crystallization system. Compaction of the double-layer, with the higher concentrations of the no-mix solution, could also play a role in the increased growth rates. Increased ionic strength with the use of NaCl has been observed to decrease the diffusion boundary layer and increase the growth rate of struvite (Ariyanto et al., 2014). Figure 43 and Figure 44 are interpolated surface plots, utilizing the same growth rate data as above, with one exception; the very high SSR of 33.9 for the no-mix is removed to better show the trends, and to provide comparable SSR scales between both graph. The surface plots more clearly illustrate the SSR and velocity influences on growth for both concentrations. The no-mix is very linear as the SSR increases, whereas the full-mix is not for the higher velocity experiments, and will be discussed further in Section 4.3.1.6. Both sets of data show an increase in growth with an increase in velocity, especially at  129 higher SSRs. The lower SSRs do not show much difference in growth rates between velocity regimes for the same solution concentrations. This could be due to the short experimental timeframe and limited growth, where differences would not be noticeable.  The diffusion-integration growth model is validated by the experimental data due to the influence of velocity on the growth rates. The full-mix growth could possibly be controlled by the integration step at low SSRs, due to the low ion concentrations in solution and the resulting low driving force. Future work should look at higher velocities for at least the full-mix conditions to confirm the dominant growth mechanism at low SSRs. Testing growth at higher velocities than 1.39 m/s would define the boundary between diffusion and integration control for struvite crystal growth. For practical purposes, knowing the control mechanism of specific growth conditions would be very important for process design and calculations.  Figure 43. Full-mix interpolated surface plot of growth rates    130  Figure 44. No-mix interpolated surface plot of growth rates  4.3.1.2 Growth Coefficient Determination Growth data was used to determine the growth coefficient, which enables estimation of growth rates for varying SSR values. Linearizing the radial growth rate (dr/dt) versus the relative supersaturation ratio (σ = SSR-1), and setting the trendline intercept to zero, enabled the determination of the growth coefficients for each concentration and velocity combination, according to Equations 35 (see Appendix M for the graphs). As mentioned in Section 4.1.2, theoretical values used for n are dependent on the type of growth where for diffusion controlled growth n=1; if the growth is from screw dislocations n=1-2; and for polynuclear growth n>2 (Jones, 2002). The lower growth rates of the low velocity conditions compared to the high velocity conditions, for each chemical composition, show diffusion is the controlling growth step. It is assumed that the maximum diffusion flux was not reached for the high velocity conditions, so the reaction order of n=1 was used for the determination of the growth coefficients. Table 9 lists the growth coefficients, and the coefficient of determination (R2) for each concentration and velocity combination.   131 Table 9 Growth coefficients for varying conditions Concentration k (μm/min) k (m/s) R2 Full-Mix Low Velocity 0.0330 5.50x10-10 0.79 Full-Mix High Velocity 0.0678 1.13x10-9 0.66 No-Mix Low Velocity 0.1594 2.66x10-9 0.98 No-Mix High Velocity 0.3359 5.60x10-9 0.97  There are numerous studies for bulk growth of struvite crystals, but none on single crystals. Some examples of struvite growth rates reported in the literature involve in bulk growth in a stirred reactor (Ariyanto et al., 2014; Galbraith et al., 2014; Harrison et al., 2011; Mehta & Batstone, 2013), an FBR (Bhuiyan et al., 2008), or in-situ with coupons in a wastewater treatment plant (Ohlinger et al., 1999). These methods measure growth rates and growth coefficients indirectly. Reported values for growth range between 0.016 to 24 μm/min, and growth coefficients range from 10-13 to 10-5 m/s (see Appendix N for a tabulated review). This large variation in the literature could be explained by variations in ion concentrations, SSR, or velocity as seen from the results from the present study. Other reasons for this large variation in struvite growth rates and growth rate coefficients could include: the method used for precipitation of struvite, where either desupersaturation or constant pH could alter growth; the sensitivity of different methods for measuring and calculating bulk growth, which could include the change of mass, concentration, or crystal size; the assumptions used for the type of growth and value for n; the amount and size of seed crystals used could influence the total surface area for growth; the pKsp value used in calculating SSR; and what software used to calculate the SSR.  Typical bulk estimates in stirred reactors utilize SSR decay, or desupersaturation, at constant volume. These are inaccurate methods to determine growth for any continuous industrial application like an FBR, and they do not provide much detail into how the crystals  132 are growing. It has been reported that the amount of seed used in the batch reactors can change growth rate constants, as more or less surface area is available for growth, which is dependent on changes to population density (Ariyanto et al., 2014). The only growth rate constant similar in magnitude to those of the present study is from the study by Ariyanto et al. (2014), but it is unclear what solution concentrations were used and if they were comparable.  The present study directly measured crystal surfaces to determine growth rates, which provides a better understanding for actual physical, fundamental struvite crystal growth, based on constant concentrations, SSR, and velocities 4.3.1.3 Determining Void Factor A void factor was incorporated in Equation 43 to determine the actual solid crystal growth based on the different growth conditions. The void factor would adjust the growth rates, which otherwise assume solid surface growth. The void factor could also be used to determine the size of crystal required for agglomeration and injection into the FBR.  The stationary growth experiments were conducted at varying conditions and over different timeframes, which resulted in varying amounts of growth in the experiments. A more uniform amount of growth between experiments would be required to determine and develop a model for void factor application. These experiments were not suitable for determining void factors. However, a pellet from the FBR control run in Ch. 3, with an SSR of 4.5, low velocity, and concentration similar to the full-mix conditions in the stationary crystal growth experiments, was used instead to determine the void factor.  Figure 45 shows the initial SEM image and the modified image used in the image analysis. The image was first converted to a binary image with the threshold set arbitrarily to 128, to reduce the image to only black and white pixels. This binary image was then processed in Matlab to determine the quantity of white and black pixels. The ratio of the  133 black and white pixels is the void factor in percent. For the size of the voids, a developed Matlab code that draws 10,000 random lines was used to measure each different chord segment of binary data (either 1 or 0 representing black or white) for each line (MacIver & Pawlik, 2017).            Figure 45. a) Original SEM pellet image [Sample_1_bse_1] and b) converted to binary image  The resulting void factor for the pellet surface was 29% and was calculated from the total image pixel count (703,777) minus the white pixels (585,176), divided by the masked pellet area pixel count (411,361). The voids were of varying sizes with the mean of 23 μm, the minimum of 7 μm, the maximum of 311 μm, the 75th percentile at 30 μm, and the 95th percentile at 69 μm.  A cut pellet (Figure 15b from Ch. 3) was also analyzed for voids between the branches. Figure 46 shows the magnified section, converted to a binary image, of the sides of the branches being analyzed. The void fraction was 37%, and the size of voids ranged from a minimum of 7 μm to a maximum of 201 μm. The mean void size was 19 μm, the 75th percentile was 23 μm, and the 95th percentile was 50 μm.  a) b)  134                          Figure 46. Binary image of branch sides  The void sizes and void fractions estimated for the pellet surface and the branch structure are similar to each other, indicating surface measurements are representative of internal voids. Using pellets is a better representation of actual voids than the stationary crystals, but it is much harder to produce each exact condition in an FBR. Further, future work would require additional stationary crystal growth experiments, to obtain similar growth at varying conditions, for model development and void factor application. It is concluded for these specific FBR pellet growth conditions, crystal voids of approximately 30% would translate into a void factor, 𝑓𝑣 = (1 − 𝑣𝑜𝑖𝑑%), of 0.70. The mean void size of 20 um is suggested as the optimal crystal size for injection into the FBR to promote agglomeration. 4.3.1.4 Total Pellet Mass Growth Rate Combining the agglomeration potential (JA) developed in Section 3.3.2.2 with the crystal mass growth (Jm) from solution from Section 4.2.5.3 into a total pellet mass growth rate establishes an equation to simultaneously estimate both the agglomeration and growth processes occurring in an FBR. The objective was to compare the growth and agglomeration contributions to the overall pellet growth to further understand which was the dominant mechanism of pellet formation. Combining Equation 43 with Equation 32 from Ch. 3, estimates the total pellet mass growth rate, according to Equation 47   135  𝐽𝑇 =  𝐽𝑚 + 𝐽𝐴 =  𝑑𝑉𝑑𝑡∙𝜌 𝑓𝑣𝐴𝑠+  𝑎𝐶𝑟𝐴𝐹𝐵𝑅=  𝑑𝑟𝑑𝑡𝜌 𝑓𝑣 +𝑎𝐶𝑟𝐴𝐹𝐵𝑅  (47)  Estimating the total flux with an injection rate of 3.5 g/hr as per Section 3.3.2.2, an average growth rate (dr/dt) from stationary growth experiment 15 of 0.13 μm/min (similar conditions to the pellet growth in Ch. 3), and a void factor of (1 – 0.30) as calculated from the previous section, results in: JA = 0.74 g/m2hour; Jm = 9.66 g/m2hour; and a total pellet mass flux rate of 10.40 g/m2hr for an FBR with a low SSR and full-mixed conditions. The mass growth rate was one order magnitude higher than the agglomeration flux rate; therefore, this analysis provides additional evidence that growth is the main mechanism of struvite pellet formation in an FBR. 4.3.1.5 Stationary Crystal Growth Morphologies Morphology identification in this work is based on visual identification. No measurements of crystals were carried out, as this was beyond the scope of this work. The goal of identifying the changes in the morphologies, in relation to the varying growth conditions, was to determine which were more favorable conditions for optimal pellet formation. An optimal surface habit for a pellet would consist of no protruding crystals, but rather, compact, relatively flat, well-formed crystals that can withstand destructive forces in an FBR. The flow in FBRs is considered to be chaotic, due to turbulence and velocity fluctuations where the pellets are rotating, bouncing, and impacting other pellets and the walls of the FBR. A crystal formation that can withstand these physical interactions is required to optimize pellet formation. It is generally understood that rotating particles will form a spherical shape, and any protruding components tend to break off from physical forces. A spherical shape was seen in all of the pellet images from Ch. 3, with very few protruding crystals.   136 Figure 47 and Figure 48 visually summarize the morphologies and experimental timeframes of the stationary crystal growth experiments, with respect to conditions and growth rates. The tabulated data for each experiment are found in Appendix O, and are organized by chemical concentration, velocity, and SSR. There are many similarities in the morphologies between the high and low velocity regimes for the full-mix condition. At low SSRs the smooth crystal surfaces roughen, and only small crystals of boxy coffins or pyramidal habit form on the surface of the seed crystal. As the SSR increases, rough areas of the seed crystals fill in and smooth out into larger pyramidal habits, and protrusions grow out from the seed surface. The protrusions range from approximately 15 to 120 μm in length and develop into 4, 5, or 6-sided crystals.   Figure 47. Full-mix morphologies and experimental times in minutes    137  Figure 48. No-mix morphologies and experimental times in minutes  Crystals associated with the no-mix conditions vary more in morphology than with the full-mix conditions between the velocity regimes, but still have similar habits. At low SSRs there was some minor surface roughening and smoothing at low velocity, but also edge and surface growth. At the higher velocity and low SSR, there was bumpy roughening with pyramidal edge and surface growth. The typical 6-sided pyramidal habit occurred for both velocity regimes at low SSRs. The typical pyramidal crystal tops transformed into broader rectangular tops at higher SSRs, and became much larger for the lower velocity and extended out from the surface approximately 20 to 40μm. Protrusions of approximately 30-120 μm grew at the higher SSRs and higher velocities. These protrusions were 3 and 4-sided, and were partially or fully hollow.  Observations from the present study show similar trends to general crystal growth knowledge that states when the driving force is increased, a smooth surface will roughen and growth can take on different morphologies (Sunagawa, 2005). Some of the crystals  138 the present study developed rough surfaces, some rough surfaces smoothed out, while some conditions developed morphologies never seen before. Classifying the morphologies was difficult due to the fact that some experiments produced multiple crystal habits for the same condition. Figure 49 is an example of this, where rough surfaces smoothed out, smooth surfaces roughened, edge and dislocations grew, and flat surfaces did not grow. This multi-staged habit development could have been the seed crystal trying to equilibrate with the solution, and the full equilibrium habit might have required a longer timeframe to fully develop. The differing growth could also be due to the influence of the seed crystal surface, or orientation to the flow. The seed crystals were developed in the FBR control run at full-mix conditions, and an SSR of 4.5 ± 1.5. This was similar to conditions in experiment 15, in Figure 49 at SSR 4.27, although the prevailing velocities for the seeds produced in the FBR and the growth experiments were very different. The crystals grown in the FBR would have been fluidized at a velocity between 0.0019 to 0.0282 m/s, whereas the stationary growth was carried out at velocities approximately 100 times higher at 0.70 to 1.43 m/s. It is unknown if the velocity in which the seeds grew would influence the growth response in the experimental conditions.     139  Figure 49. Example of multiple morphologies at low SSR and low velocity: (a) before (b) after growth i) pyramidal crystals ii) edge growth iii) smoothening of rough surface and forming pyramidal habit [FMLF-15]  Overall, it was observed that the struvite crystal edges, macro-steps, and rough kink areas grew faster than large flat surfaces. Flat faces of crystals have little to no growth sites for unit cells to bond to, and they usually grow by a slow, layered growth (Markov, 2003). Figure 50 is an example of a large flat crystal where the edges show growth, but the flat surface from before (a) and after (b) growth look almost identical. Figure 39 (4c) and (4d), compared to Figure 51, illustrates how two crystals from the same experiment (FMHV-17) can exhibit different growth, due to the seed crystal surface. The rough seed surface in Figure 39-4 produced many individual pyramidal crystal protrusions, whereas with a flat surface as in Figure 51, the edge growth is more pronounced. This shows that growth rates are proportional to the density of growth sites, and morphologies differ depending on roughness and size of the underlying crystal surface.  a) b)  140   Figure 50. Example of no growth on large flat face a) before b) after growth [NMHF-27]   Figure 51. Growth influence from seed crystal experiment 17; a) before growth b) after growth [FMHF-17]  The preferential morphology of struvite growing at a low SSR, in a flow field, is pyramidal with the (001̅) face extending outwards. This habit can be seen in all the stationary growth experiments and cut pellet images, with variations depending on conditions. Seed surfaces that are already rough or have small protrusions on the surface a) b) a) b)  141 tend to develop this preferential habit more easily. The full-mix conditions generate slow growth rates, and develop a pyramidal, 6-sided top crystal from these protrusions, as illustrated in Figure 52. The pyramidal growth is also seen for higher SSRs in the full-mix conditions, as in Figure 39-3b. With the higher growth rates of the no-mix conditions, the pyramidal shape widens to form a 4-sided, rectangle top, like the model in Figure 35, but without the (103̅) face. Figure 40-3 illustrates this widened, rectangular pyramidal habit.   Figure 52. Development of pyramidal 6-sided habit with the (00?̅?) face extending outwards, Scale bar 100 μm across both images [FMLF-16]  Large protrusions are not a desired morphology in regards to pellet growth, as they are more than likely to break off and contribute to FBR fines losses. Protrusions occurred at SSR values over 8, under most of the experimental conditions, but they did not form in the no-mix, low-velocity, high SSR experiments. The only possible explanation for this exception could be slower growing faces of the rectangular pyramidal crystal top and sides that formed, or that the larger tops limited other crystals from growing outwards.  All the full-mix, high SSRs produced 4-6 sided rod/prismatic protrusions. Figure 53 (a) shows what appear to be 4-sided crystal protrusions with 6-faced, prismatic tops, but could be 6-sided, as seen in the morphologies from the rotational growth study in Section a) b)  142 4.3.2.2 .The SSRs between 11 and 13 for the full-mix, and at both velocities, generate fewer protrusions than the lower SSRs around 8 or 9 (determined on a qualitative visual basis only). This is opposite to what would be expected, but could be a result of different phosphate compounds forming in conjunction with struvite, with a different habit at the higher SSR conditions. However, experimental timeframes were not consistent, which could explain the fewer protrusions at the higher SSRs, and will be discussed further in Section 4.3.1.6. The no-mix conditions produced protrusions at the high-velocity, high SSR conditions, but not at the low-velocity, high SSR conditions. The habit in the low-velocity conditions changed to a wide rectangular pyramidal top as previously mentioned, so the only protrusions were the corners of the tops. For the high-velocity, high SSRs conditions, the entire seed surfaces were covered in protrusions growing into and perpendicular to the flow. Figure 53 (b) shows results for experiment 28, SSR 13.5, with 3-4 sided protrusions, which were partially hollow. Figure 53 (c) shows the protrusions produced from experiment 3, at high-velocity, SSR of 34, and the highest growth rate of 11μm/min. These crystals are large, 3-sided, hollow protrusions. If crystals are hollow, their density changes, leading to a lower settling velocity. This will cause the crystals to be washed out of the FBRs more easily, which will result in greater fines losses. Hollow crystals can be formed from the addition of impurities (Sunagawa, 2005). Since no impurities were introduced into the struvite experiments, these hollow crystals illustrate how individual crystal faces have different growth rates, and how these rates can completely change the crystal morphology. Identification of these faces was beyond the scope of this work.  143  Figure 53. Protrusion growth at high SSRs a) experiment 14 [image Protrusions FMLF-14] b) experiment 28 [image Protrusions NMHF-28] c) experiment 3 [image Protrusions NMHF-3]  Crystals from experiment 3 were the only crystals produced that were large enough to confirm they were struvite, by single crystal X-ray diffraction. The unit cell measurements were carried out with a Bruker X8 APEX diffractometer using Mo-K_alpha radiation. Table 10 compares the measurements from the sample against a reference for struvite, and confirms the sample is struvite.   a) b) c)  144 Table 10 Single crystal unit cell measurements/XRD results Parameter Struvite ref (Ferraris, 1986) Exp. 3 Single Crystal Sample  Space Group Pmn21 Pmn21 a 6.955 6.9365 b 6.142 6.1028 c 11.218 11.1774 a/b 1.13237 1.1366 c/b 1.82644 1.8315 Volume 479.21 473.17  It was intentional to use different timeframes for some of the same full-mix experimental conditions to see if morphologies changed. The longer experimental times produced more protrusions and more defined crystal structures, which was expected; however, it was not expected to see any effects on the growth rate, as discussed in the following section. Experiments 21 and 22 at the same SSR of 12.8, but with different growth times of 60 and 31 minutes, showed difference stages of surface growth. The shorter experiment’s surface had mostly roughening or small crystals forming on the surface, whereas, for the longer experiment, there was larger, more developed pyramidal structures. This progression of growth is also apparent between experiment 29 and 25 at similar SSRs of 10.2 and 11.0, and times of 60 and 30 minutes. The pyramidal crystals were more developed and larger for experiment 29 at the longer timeframe, and there were more protrusions. More growth at a longer timeframe was expected, but the increase in protrusions was not. This may not be of any consequence when growing in an FBR, as crystals are not stationary for these timeframes and directional growth would constantly change.  145 4.3.1.6 Errors in Deriving Growth Rates and Morphologies There were several difficulties with the newly developed method that influenced the exact value of the reported growth rates. Further, future refinement of the method will increase the accuracy by decreasing the influence from factors including: the crystal circularity; the attachment point of the crystals to the wire; the apparatus; the seed crystal surface influence on growth rates; and the duration of the experiments. Not all experiments performed were successful; therefore, the lists in the appendices are missing experiment numbers 1,2,4,5, and 6. The main cause for exclusion was obvious damage to the crystals, either from handling or too long a growth period. The circularity was calculated in Photoshop CC based on the masked area of each of the crystals in the experiments, and then all experiment circularities were averaged. It can be seen in Figure 39 and Figure 40 that the crystals were not all circular. The average circularity for all the crystals measured before growth was 0.55, with a minimum of 0.51 and maximum of 0.58. The average circularity after growth was 0.48, with a minimum of 0.39 and a maximum of 0.57. This indicates that the crystals were not spherical, as the calculations assumed. This causes some error in the estimated radial growth rate, but the order of magnitude should not be compromised. Manual measurement checks of the before and after SEM crystal images were performed to check the validity of the calculated linear growth rates, and it confirmed that the method was a good approximation; in fact, it probably underestimated the growth.  The images show that the crystals were often glued onto the wire on one side, which prevented crystal growth on that particular side. Approximately 25 to 30% of the crystal perimeters, accounted for in the projected area, were not growing. This applied to approximately 25 to 50% of the crystals per experiment, and contributed to a lower radial and mass flux growth rate than presented. Accounting for these inaccuracies would increase the crystal growth area by 6.25 to 15%, which in turn increases the radius and  146 the mass flux by 25 to 39%. If the mass flux was 0.24 g/m2*min, then it would increase to 0.30 to 0.33 g/m2*min. This is not a significant increase where orders of magnitudes are changing, but it would need to be taken into account if exact design calculations were required. This is likely the main cause of underestimating the growth rates noticed during the manual measurement checks. For some experiments, with high growth rates and where crystals were glued on the bottom side of the wire, the crystals grew to touch the glass cover slide in the flow cell. The crystals appeared to grow the same on either sides of the wire up until the glass interfered. This similar growth on either side of the wire indicates that the flow was evenly and uniformly distributed to the crystals, and there was no negative influence from being so close to the bottom of the flow cell, unless they touch it. The images for these occurrences were carefully examined for crystals that had not touched the glass slide for the growth rate determination and morphology characterization.  The crystal diameter compared to the wire diameter was a concern with respect to flow, especially if the crystal was smaller than the wire. If this were the case, the wire would influence the flow; however, Figure 39 and Figure 40 show that most of the crystals were larger than the wire. During image analysis this was taken into account and any crystals smaller than the wire were eliminated from the analysis. Dissolution occurred for three of the no-mix conditions at both low and high velocity, and SSR values between 2.57 and 3.02. This phenomenon was not expected since equilibrium conditions normally occurr around an SSR of one. No explanation can be confirmed, but an error in pH measurement is most likely the cause, or some interaction of the seed material that was grown in different conditions than those of the chemical solution. The three experiments were performed with the same chemical solutions, and the solution concentrations were double checked for measurement errors. The  147 experiments that showed dissolution were removed from the analysis and plots, but can be seen in the data in Appendix K and Appendix L.  It was found that protrusions skew the growth rates, and longer experiments have more protrusions. This can explain why the growth rate plot is not linear for the full-mix, high-velocity, higher SSR conditions. Figure 47 illustrates this with the experiments at SSR 8.71 and 10.23, which were carried out at 60 minutes; whereas, the other two, at SSR 10.96 and 13.18, were only 30 minutes long. The experiments at 30 minutes have a lower growth by approximately 0.1-0.2 μm/min. Low SSRs did not have protrusions, and there were no differences in growth rates with different experimental time frames. It was intentional to use different timeframes for some of the same experimental conditions to see if morphologies changed, but it was not expected to see any effects on the growth rate. It was also not possible to run all experiments at longer timeframes. This was especially true for very high growth, where the crystals were too delicate to handle if they were too large, or if they could break off during the experiment and lead to misinterpretation. Breakage could also be a reason for the lower growth at higher SSR, but this is not suspected based on careful examination of the images. One shortfall of this study is that experiments for the full-mix conditions should be extended for longer timeframes to see what morphology would develop and if they compare to higher SSRs, the no-mix conditions, or the pellet growth morphologies. Growth should also be carried out on crystals that have been acclimatized to the solution, to mimic FBR constant conditions and equilibrium morphologies.  4.3.2 Rotational Crystal Growth  In total, seven experiments were carried out to test the method presented. Challenges in sealing the tube became an issue when the fluid flow was increased from 200 mL/min to 850 mL/min and above. Instead of liquid leaking out, air was sucked in, due  148 to the low pressure in the tubing. The air came in along the backside of the needle, touching the pellet in varying amounts. This air leak was confirmed by small crystal formations seen in the images along the same path as the air. The small crystals were formed by an increase in the local SSR due to CO2 stripping, or from shear nucleation between the air and the solution. Due to time and financial constraints, re-designing the apparatus was not undertaken, so the data presented are only preliminary in nature. 4.3.2.1 Rotational Growth Rates Projected areas of the before and after pellet growth was measured and an overall radial growth rate was calculated for each experiment, similar to those for the stationary crystal growth experiments. Figure 54 and Figure 55 are plots of the data. Detailed data sheets are in Appendix P. The surface plot illustrates that growth rates increased with an increase in velocity, but it is not evident in the data set that SSR influenced growth rate, since the SSR was kept in a tight range and very few experiments were conducted.  Figure 54. Pellet radial growth rates   149  Figure 55. Interpolated surface plot of pellet radial growth rates  4.3.2.2 Rotational Crystal Growth Morphologies The morphology of the experimental pellets themselves was very similar to that of the pellets growing in the control FBR in Ch. 3. Since the rotational pellets were spinning, they continued to be spherical with the crystals growing out from the center. The SSRs were kept in the growth range of the full-mix concentrations, so the preferential pyramidal, 6-sided top habit of the struvite crystals continued to grow outward. A couple of pellets were used in consecutive experiments and as they grew larger the pyramidal structure continued to dominate. As the crystals developed in the consecutive experiments, the faces and tops did not exhibit much damage or dislocations compared to the pellets from the FBR, which could limit branch development. Many of the other rotational experiments showed crystals growing out as single crystals, and not forming many branching structures. This indicates that the physical impact damage occurring in the FBR is important for pellet branch development. The lack of damage is likely to change the habit to a protrusion or dendritic structure, as seen in aggregate studies (Z. Ye et al., 2014). Figure 56 a-d shows the progression of the pellet used for the three consecutive experiments. It can be seen in  150 (d), for experiment 5, when the flow was increased to 830 mL/min (velocity 0.46 m/s) small crystals formed on the needle and part of the pellet base from the air leak. The two other experiments, with flows of 200 mL/min (0.13 m/s), did not produce these small crystals, since no air was sucked into the tube.  Figure 56. Pellet glued to needle, used in consecutive experiments: a) seed pellet before growth [PelletGrowth_Needle1_003] b) after experiment 1 [PelletGrowth_Exp1_after_lsve_002] c) after experiment 2 [PelletGrowth_Exp2_after_lsve_001] d) after experiment 5 [Needle1_afterExp5_lsve_001]  The small crystals that have grown due to the air leak were not the typical pyramidal morphology. They grew outwards in the same habit as seen in the protrusions from the stationary crystal growth, with 4 or 6-sided crystal protrusions with 6-faced, prismatic tops similar to Figure 53 (a). This struvite morphology has not been seen in any prior literature until now. The 6-sided tops have 4-square sections where two have flat square faces, and a) b) c) d)  151 the other squares are split diagonally into triangles and cave inwards. Figure 57 shows the crystals from the top and from the side from different experiments. Since secondary nucleation is related to attrition and dependent on SSR (Jones, 2002), it is speculated that this might be secondary nucleation. It is likely caused by the higher SSR from CO2 stripping, leaving the crystals to then grow in the higher SSR conditions to produce this morphology.  Figure 58 shows clusters of these small crystals that appear to form in cracks or dislocations on the parent pellet crystals; again, illustrating the importance of damaged areas.  Figure 57. New morphology with 6-sided top a) side view [Needle2_afterExp3_lsve_007] b) top view [Needle3_afterExp6_lsve_013]  a) b)  152                    Figure 58. Clusters of small crystals [Needle1_afterExp5_lsve_005]  4.3.2.3 Relative Velocity Influence on Spinning Pellet Attempts were made to determine if the relative velocity between a pellet and the solution contributed to changes of the pellet surface. Manual crystal counts along six chord lengths per image were performed. A chord length of 450 μm was used, where three vertical and three horizontal lines were randomly placed on the images. Two experiments (2 and 7), which had little interference from air leaks and had similar SSRs, but different relative velocities, were compared. The velocities for experiments 2 and 7 were 0.13 m/s and 0.47 m/s, respectively.  Figure 59 visually shows the difference in growth between experiments 2 and 7. Experiment 7 appears to have more compact, smaller crystals from the higher velocity. The average of the three horizontal and vertical chord analysis are 12 and 14, and 17 and 20 for experiments 2 and 7, respectively. There are more crystals in the same amount of space for experiment 7, even though it has a slightly lower SSR than experiment 2. These initial experiments suggest that higher relative velocities reduce the size and increase the  153 compactness of the pellet crystals, and could lead to harder, smoother pellets. Further work is required to perfect the rotational growth method, and perform additional experiments to confirm these initial findings.  Figure 59. Comparison of pellet growth from a) experiment 2 [PelletGrowth_Exp2_after_lsve_002] b) experiment 7 [Needle4_afterExp7_lsve_002]  4.3.3 FBR Bulk Pellet Growth Rate and Internal Structures Growth rates in the FBR control run from Ch. 3, at full-mix concentrations, were calculated from the difference between the initial seed pellet size and the first harvest pellet sizes. The 86 grams of 0.5 mm seed pellets and 206 grams of seed crystals grew at an SSR of 4.9, and a much lower velocity than the stationary growth experiments (due to the small seed pellets being higher up in the FBR). The pellets grew for 20.5 hours until the first harvest was performed. Not all the pellets were removed from the FBR, but only the ones fluidized in the harvest zone. A total of 61.0 grams of 1.0 mm pellets, and 27.4 grams of 2.0 mm pellets, were removed and sieved from the FBR. Radial growth rates of 0.20 μm/min and 0.61 μm/min were calculated for the 1.0 mm and 2.0 mm harvested pellets, based off of a seed size of 0.5 mm. This could underestimate the 1.0 mm pellet growth because it is unknown what seed size they grew from. a) b)  154 Additional imaging of the pellet internal structures was carried out for pellets harvested from the FBR control run. Figure 60 to 62 show the branching structures of various formations and orientations, with the addition of arrows indicating the path of individual branches. Figure 62 illustrates a magnified, pyramidal crystal habit that the branches are comprised of. The crystals grow radially outward in the –c axial direction with the (001̅) face as the external surface of the crystal and pellet. The spontaneous polarization of the c-axis of struvite (-8.8 μC/cm2) may play a role in crystal growth and agglomeration processes. The faces of the c-axis have different growth rates due to this difference in electrical and physical properties (Romanowski et al., 2010), and is likely the reason for this preferred directional growth in pellets. This branching structure of pyramidal crystals growing out from each other is the preferential crystal morphology for a struvite pellet without agglomeration, at a low SSR (4.5 ± 1.5), and a velocity of 400 cm/min within the FBR harvest zone. Higher SSRs are thought to develop varying structures based on the stationary growth experiments, and should be investigated; however, it would be difficult to restrict nucleation and agglomeration that could cover up the underlying structure.   155                Figure 60. Cut pellet with branching structures [Sample 75-03]   Figure 61. a) Multi-branch formation [Sample 77-03]; b) Side of branch [Sample 77-05]  b) a)  156                         Figure 62. Typical crystals in branch formation [Sample 76-02], comparable to pyramidal model inset from (Holtkamp, 2014) 6  Branch development was seen forming on rough areas, kinks, or damaged areas identified in the stationary crystal growth experiments in Section 4.3.1. Figure 63 shows one of these experiments with two areas marked by arrows, from before (a) and after (b) growth, where new pyramidal crystals were formed. The upper arrow indicates growth of an already existing bumpy area, and the bottom one indicates growth from a kink, or damaged site. This shows the importance of steps, kinks, dislocations, and damaged areas for growth of these branches; therefore, it appears that a certain amount of damage is required in FBR operations to promote pellet branch growth.                                                6 see preface for copyright acknowledgments  157  Figure 63. Branch development from kink site or damage areas a) before growth b) after growth [NMHF-27]  If nucleation is occurring in the FBR during pellet growth, it is speculated that branch growth, or branch directional changes, could be initiated similar to twinning on the (001) face, or where the face disappears and forms an edge with either the (101) and (1̅01), or the (012) and (01̅2) crystal faces. The nucleating crystal could easily bond on an edge or step of the pellet surface and continue growing in a new direction. 4.3.4 Comparison of Growth Rates for Limited Conditions  A comparison between the stationary and rotational growth rates was attempted, even though there were no exact data points that can be compared between the two sets of data. Similar conditions for the full-mix, low and high velocity conditions used in the comparison are tabulated in Table 11. The percent difference between the stationary and rotational growth rates are approximately 50.9 and 77.2% for the low and high-velocity conditions, respectively. If the error of 25-39%, calculated in Section 4.3.1.6, in underestimating the stationary growth is taken into account, the percent difference between the stationary and rotational growth rates are approximately 31.6 to 71.9% and a) b)  158 34.1 to 40.5%, for the low and high-velocity conditions, respectively. The differences are significant, but statistically there is not enough data from the rotational growth experiments to make concrete conclusions. This comparison does, however, indicate that the rotation of a pellet has a significant impact on growth rates and warrants further investigation, or the stationary growth method is very inaccurate. Either way, both methods require refinement. Table 11 Comparable stationary and rotational growth rates Experiment # Velocity SSR Growth Rate (μm/min) Growth with 25% to 39% Error Stationary-15 0.73 4.27 0.13 0.16-0.18 Stationary-13 0.73 9.77 0.28 0.35-0.39 Rotational-4 0.73 5.62 0.57 - Stationary-30 1.43 8.71 0.88 1.1-1.22 Rotational-6 1.18 5.75 1.85 -  In addition, the stationary and rotational growth rates were also compared with the bulk growth rate calculated in Section 4.3.3, from the control FBR run in Ch. 3. All compared growth rates were for the full-mix concentrations at similar SSRs and low velocities. The FBR bulk growth rates were 0.20 μm/min and 0.61 μm/min for 1.0 mm and 2.0 mm pellets. The 1.0 mm pellet growth is similar to the stationary growth, and the 2.0 mm pellet growth is similar to the rotational growth. These values are on the low side of published growth rates seen in the review in Appendix N, but as mentioned previously, without knowing the exact conditions of growth in the published data it is hard to determine if they are comparable values. The maximum growth value of the 2.0 mm pellet from the FBR bulk growth is, more than likely, still overestimating the growth of a fully loaded FBR. Damage and attrition have not been studied to know the true growth of a pellet in a fully loaded FBR; however, now  159 there is a better understanding of the theoretical range based on physical measurements. The stationary crystal growth underestimates struvite growth rates, the rotational pellet growth overestimates pellet growth rates, whereas the bulk growth confirms they are in the right range, or order of magnitude, of growth. This is still a very limited comparison of only one set of operational conditions, but shows promise for further study. 4.4 Chapter 4 Summary and Conclusions Growth of struvite crystals was carried out to physically see how the crystals grow, to understand what influences growth rates and morphology changes, and to determine the best conditions for pellet growth in an FBR. It was found that single-crystal growth rates increased linearly with an increase in SSR. Stationary and rotational growth rates both showed an increase with an increase in fluid velocity. Growth rates also increase with an increase in ion concentration for comparable SSRs.  Struvite morphologies changed depending on conditions. At low SSRs and low growth, crystal surfaces initially roughen until they are stable, and form a pyramidal preferential habit similar to the habit identified in the pellet branches. Faces of the pyramidal habit changed and even disappeared with higher ion concentrations, forming rectangular tops. Protrusions form at SSRs over 8, and they were partially or fully hollow at higher ion concentrations. The documented morphologies that relate to specific growth rates can be utilized for growing specific pellet habits to limit protrusions that would otherwise break and add to fines losses. It is recommended to use SSRs lower than 8 to reduce the occurrence of protrusions, and higher velocities to maximize growth for all ion concentrations.     160 Chapter 5 Engineering Application and FBR Operation  To link the data and knowledge gained from this work to the operation of an FBR, and to highlight the importance of this new information to the design aspects, a comparison of two different sized FBRs is presented. An alternate FBR process-operation design is then suggested. The two FBRs utilized in the comparison are designated as a 2.54 and 7.60 cm reactor, which refers to the diameters of the harvest zones. Data from comparable experiments, with similar concentrations and SSRs, were used from the nucleation and ζ experiments to highlight where optimal agglomeration and nucleation would occur in the two different sizes of FBRs. The concentrations are similar to the no-mix concentrations used in Ch. 4 for the stationary growth experiments. Figure 64 illustrates a simplified schematic of the two FBRs, with the cumulative retention times at the end of each zone in each FBR, based on an up-flow velocity of 400 cm/min in the harvest zone.  The ζ experiments from the 1-B conditions have an SSR between 5 and 6. The ζ for these conditions start just below zero (see Figure 10a) and hold the same value for 1500 seconds. This time frame is greater than the 2.54 cm FBR retention time, so it has optimal agglomeration conditions throughout, whereas the 7.60 cm FBR has optimal agglomeration only up until the seed hopper. The induction time for primary nucleation, according to Ch. 2 for similar concentrations, at an SSR of 4.0-5.0 was 570 seconds, and at SSR 6.2-8.5 was 207 seconds (see Table 2). Figure 64 shows where the induction time of 207 seconds would be located in each FBR. A loaded bed of struvite pellets would reduce the solution SSR with an increase in height; therefore, at the SSR range of 6.2-8.5 it is doubtful that primary nucleation, at a 650 nm crystal size, is occurring in either FBR. Any fine crystal production would likely be from secondary nucleation and attrition, and could be identified by  161 examination of the crystals. If SSRs are increased, the induction time for primary nucleation decreases, and could result in primary nucleation occurring within the lower sections of the FBRs, as well as more secondary nucleation. Depending on what process is being designed, it may, or may not, require the production of nuclei within the FBR. Understanding the induction time for each specific wastewater being processed can control nuclei production.                          Figure 64. Dimensions and retention time comparison of a 2.54 and 7.60 cm diameter FBR  The 7.60 cm FBR would recovery more phosphorus than the 2.54 cm FBR because it has a longer retention time and bed length, which allows the solution to pass through more  162 pellets; therefore, reducing the solution concentrations to a greater extent. The recovery of phosphorus requires crystallization within the retention time of the FBR, so faster is better, and generally a higher SSR is required. Recovery of all the phosphorus is also wanted, but too high of an SSR creates problems for pellet growth with respect to repulsive forces and excess production of fine crystals. If a full reduction in concentrations is not achievable in the retention time of a particular FBR, an alternate method should be utilized. An additional step external to the pelletization in the FBR where the pH of the effluent is increased to precipitate the remaining phosphorus could be a solution. The struvite crystals could then be separated from colloidal material with a cyclone and re-injected into the FBR for agglomeration. Struvite has a higher density, between 1.77-1.82 g/cm3 for synthetic and composite struvite (Katz, 1972), than dead bacteria cells like Escherichia coli O157:H7 at 1.18 g/cm3 (Lewis et al., 2014), which is representative of a cell that could be found in wastewater. The density of other bacteria is not likely to be much greater than E. coli; therefore, cyclone separation could be a viable option. Separation was tested and is effective for Annacis Island clarifier sludge from a fourth-generation pilot FBR. The analysis of the fines that are washed out of FBRs could provide information for FBR operation and control. Image analysis could determine if primary or secondary nucleation is occurring, and if FBR operational parameters require adjusting. If single coffin, needle, twinned, or star-shaped crystals are being produced, primary nucleation is occurring. If there are broken or odd-shaped crystals, then secondary nucleation or attrition is more likely the main mechanism of fines production. Changing the SSR has been the typical way to control primary fines production. The hardness of pellets is related to up-flow velocity, but it is not yet completely understood how to control the process. If a means of increasing the relative velocities within the FBR is developed, without increasing the overall flow, it is thought that a hard  163 pellet surface could be produced. Growth rates can also be increased with higher relative velocities, so the recovery of phosphorus would occur at a faster rate and potentially decrease the size of the FBR. An increase in the relative velocity between pellets within the FBR could be accomplished with the addition of an internal recycle in the harvest zone from the active zone. This would impose a higher velocity in the active zone and push the pellets up. The pellets in the fines zone and seed hopper would likely push downwards, against the upward movement of pellets, theoretically resulting in pellets forced to experience a higher velocity.  Incorporating these alternate processing operations into an FBR system would enable a controllable, multi-operation FBR, instead of running at constant conditions and trying to integrate multiple operations at the same time. A schematic of this system can be seen in Figure 65, and would include: FBR operations using growth conditions with intermittent injection of crystals for agglomeration; raising the pH of the effluent external from the FBR and clarifier to extract all remaining phosphorus; separating fines from colloidal material from the clarifier and from the effluent using a cyclone, and using them for agglomeration injection; and lastly activating an internal recycle, once pellet sizes are appropriate, to add a hard coating before harvesting.   164                      Figure 65. Proposed FBR system  By optimizing FBR operations, struvite fines losses could be reduced and an additional 10-30% of struvite could be recovered. This would realize additional revenue for wastewater treatment facilities and would further reduce operational costs. Reducing the fines losses also benefits the environment by reducing any leaching of phosphorus from the disposal of the biosolids. It is optimistic, but 100% phosphorus recovery is achievable by improving the technologies and enhancing our understanding of struvite formation.   165 Chapter 6 Conclusions and Recommendations 6.1 Conclusions This compilation of work provides a greater understanding of fundamental knowledge of struvite formation, and the processes of pellet formation in an FBR. The goal of reducing struvite fines losses from FBRs, and achieving 100% phosphorus recovery is within reach with this additional knowledge. Each set of experiments in this work built on the knowledge generated in the previous ones, and resulted in similar conclusions and recommendations for optimal FBR operations for pellet formation. There is much work still to do, to fully understand all of the mechanisms of pellet growth; however, this dissertation brings together many aspects that have not been previously connected. The following are conclusions drawn from this research. 6.1.1 General An SSR recommendation between 2 and 6 is unanimous between chapter conclusions to prevent fines production, and maximize agglomeration potential and growth of pellets.  6.1.2 Nucleation A laser system that can sense 650 nm struvite crystals of approximately 1000 unit-cells in size enabled for the first time a means of relating a crystal size with struvite induction times. Knowing a crystal size was required for further experiments involving measuring ζ, and is important for agglomeration potential. The induction time relating to the crystal size can be used to identify where nucleation is occurring within an FBR.  6.1.3 Surface Charge The solution system has a profound effect on the surface charge of nucleating and growing struvite crystals. ζ values change over time as solution conditions change, and they become more negative as precipitation and growth progress. A lower starting SSR  166 can produce low magnitude ζ for an extended period. Magnesium can reduce the ζ magnitude and even cause a charge reversal if enough ions are present to overcome a negatively charged solution, or specifically the negative phosphate ions.  For agglomeration process-design purposes involving precipitation reactions, specific time frames relating to low magnitude ζ values need to be incorporated into FBR retention time calculations, or process logic and operations. If agglomeration is required right after mixing a lower initial SSR is required. High initial SSRs create higher ζ, and it takes longer for the solution system to reach a zero charge. If nuclei are required, but not maximum agglomeration, then a higher SSR should be used. SSR values between 2 and 6 are recommended to improve FBR agglomeration processes. 6.1.4 Crystal Growth This work was the first to grow individual struvite crystals and pellets, and measure the change in projected area to determine a radial growth rate and apply it to pellet growth. Growth rates increase with an increase in both SSR and relative velocities. Velocity has never been previously investigated to this degree and here it was shown to increase the potential to recover phosphorus at a faster rate. This means technologies could become smaller in size and yet still process the same amount of wastewater. New technology designs could also be realized that utilize fluid or a pellet relative velocity in completely different method than FBRs. A branching structure was identified as the main growth formation in struvite pellets grown in an FBR at low SSRs. This branching structure has been concealed until now by agglomerated fine crystals within pellets.  SSRs over 8 created protrusions in the stationary growth experiments. If protrusions grow on pellets in the same way, they would more than likely break off and produce the fines that are plaguing technology developers, by being lost to recovery.  167 Some of the protrusion habits presented have not been seen in detail before and are of crystallographic importance. At very high SSRs, the protrusions were hollow and only 3-sided. Air leaking into the rotational crystal growth experiments created protrusions with tops having 4 square sections, where two were flat, and the other squares were split diagonally into triangles caving inward. For resource recovery, it is important to understand the crystal forms that are produced under varying conditions, so as to limit unwanted habits. 6.1.5 Struvite Pellet Agglomeration and Growth By separating growth and agglomeration in the FBR, and growing individual crystals, a number of details could be pieced together to determine that agglomeration is not the main mechanism in struvite pellet formation; crystal growth has a more dominant role than agglomeration. This is apparent in the example calculation in Sections 3.3.2.2 and 4.3.1.4, where the agglomeration flux, with little breakthrough, was estimated to be 0.74 g/m2*hour, whereas, the mass growth rate was an order of magnitude higher at 9.66 g/m2*hour. The pellet voids of 30% compared to 70% for solid branches, seen in Section 4.3.1.3, also indicate growth is the dominant mechanism for pellet formation. 6.2 Recommendations for Future Work This work, as in all research, generated many questions and directions for future investigations that could not be answered in the timeframe or scope of this dissertation. The main objective of this research was to increase the fundamental knowledge of struvite pellet formation and mechanisms contributing to it within the UBC FBR system, with the goal to maximize growth rates while preventing fines losses. The following recommendations are intended to provide direction for further work on this research topic. 6.2.1 General All of the experiments in this dissertation were conducted with synthetic wastewater, which greatly reduces unknown variables and allows easier analysis of data;  168 however, it does not provide the complete picture. Future work should test real wastewaters to determine what variations occur in the growth rates and morphologies that have been presented here. Results would be waste specific, but would provide valuable knowledge. Developing a model to incorporate induction times, ζ influence on agglomeration, crystal injection rate or nucleation rate, and growth rates based on fluid or relative velocity would be an advantage in optimizing phosphorus recovery technologies. 6.2.2 Nucleation Struvite nucleation studies are limited to mainly using mixers at lower rpm ranges. The effects from the magnetic mixer used here were shown to be insignificant on the induction times in the present study, but due to the influence of velocity on the crystal growth in Ch.4, a more in-depth look at nucleation should be carried out. Small magnetic mixers do not simulate higher shear stresses that would occur in FBR injectors, in the fluid, or at the steps where mixing occurs. Induction time studies at higher mixing energies are needed to further the knowledge of how the fine crystals are being produced in FBRs, and if shear nucleation is occurring. Couette flow would induce simple fluid shear and could be a means of investigating this. Secondary nucleation is also not well understood for struvite in FBRs, and requires further study for the prevention of fines losses. Some of the protrusions seen in the growth experiments are thought to be secondary nuclei; thus, studies incorporating growth mechanisms and rates, forces required to break off the protrusions, and morphology changes in relation to varying conditions, are required.  From the agglomeration investigation in Ch. 3, a crystal injection rate was recommended for the FBR. Instead of using an injection approach, a well-defined and controlled nucleation rate could replace the external addition of small crystals for  169 maximum agglomeration with limited fines losses. This would have to take into account ζ as well as the agglomeration potential of the forming pellets. 6.2.3 Surface Charge Since this is the first study of surface charge for nucleating and growing struvite, additional work is recommended to determine stability ζ boundaries for the crystals based on ionic strength, or specific ions in solution. Research to see if ζ could be used as an online operating parameter is also recommended.  6.2.4 Crystal Growth Both the stationary and rotational growth experiment methods are in their infancy; but results show, with an increase in velocity, growth rates increase. Further research is required to expand on the struvite growth rates and determine limits with respect to velocity, especially for low SSRs where desirable morphologies are produced. It is also suggested that low SSR experiments be completed for longer time frames to confirm morphology development, and compare them to those at other conditions. Pellet branch growth is thought to require dislocations or damaged areas for continual growth. This is seen with the lack of branch formations in the rotational pellet growth images, where only clean straight crystal protrusions and surface growth occurred. This indicates the damage and breakage within the FBR is an important process for developing pellets. Studying how damage occurs, and the amount of damaged required for branch growth in relation to breakage and attrition, is recommended to further reduce fines losses. Attempts were made to determine how hard pellets form, but they were not conclusive. Relative velocities between pellets and the solution are thought to control the surface of a pellet by either reducing the size of the crystals, or filling in the gaps between the branches leading to a harder outer surface. Greater amounts of damage are also  170 thought to influence the size of crystals and could initiate harder pellet surfaces; therefore, it is suggested that the rotational single pellet growth method be improved to verify these theories.  6.2.5 FBR Operation Suspension densities, FBR bed loading, or bed length in relation to the amount of precipitation potential contained in the wastewater has never been analyzed. Suspension densities (% by mass) specific to individual reactor configurations, have a range where operation below, or above, it will increase the risk of primary nucleation or cause higher attrition, respectively (Beckmann, 2013). This would mean that below the range there wouldn’t be enough crystal growth in the FBR to decrease the SSR faster than the induction time. A loaded bed reduces the SSR with an increase in height; therefore, it reduces the potential for primary nucleation, but it is unknown how it affects secondary nucleation. The amount of growth in the FBR related to agglomerate surface area potentially limits the optimal SSR value. This information would be essential to know for designing any new processes or optimizing existing technology for complete removal of phosphorus from wastewaters, and could be a means of controlling unwanted struvite fines.     171 References Abbona, F., & Boistelle, R. (1985). Nucleation of struvite (MgNH4PO4.6H2O) single crystals and aggregates. Crystal Res. Technol., 20, 133–40. http://doi.org/10.1007/s00251-011-0563-3 Adnan, A., Koch, F. A., & Mavinic, D. S. (2003). Pilot-scale study of phosphorus recovery through struvite crystallization - II: Applying in-reactor supersaturation ratio as a process control parameter. Journal of Environmental Engineering and Science, 2(6), 473. http://doi.org/10.1139/S03-048 Agrawal, S., Guest, J. S., & Cusick, R. D. (2018). Elucidating the impacts of initial supersaturation and seed crystal loading on struvite precipitation kinetics, fines production, and crystal growth. Water Research, 132, 252–259. http://doi.org/10.1016/j.watres.2018.01.002 Ali, M. I., & Schneider, P. A. (2006). A fed-batch design approach of struvite system in controlled supersaturation. Chemical Engineering Science, 61(12), 3951–3961. http://doi.org/10.1016/j.ces.2006.01.028 Ali, M. I., & Schneider, P. A. (2008). An approach of estimating struvite growth kinetic incorporating thermodynamic and solution chemistry, kinetic and process description. Chemical Engineering Science, 63(13), 3514–3525. http://doi.org/10.1016/j.ces.2008.04.023 Ariyanto, E., Sen, T. K., & Ang, H. M. (2014). The influence of various physico-chemical process parameters on kinetics and growth mechanism of struvite crystallisation. Advanced Powder Technology, 25(2), 682–694. http://doi.org/10.1016/j.apt.2013.10.014 Ashley, K., Cordell, D., & Mavinic, D. (2011). A brief history of phosphorus: From the philosopher’s stone to nutrient recovery and reuse. Chemosphere, 84(6), 737–746.  172 http://doi.org/10.1016/j.chemosphere.2011.03.001 Beckmann, W. (Ed.). (2013). Crystallization: Basic Concepts and Industrial Applications. Weinheim, Germany: Wiley-VCH. Bhuiyan, M. I. H., Mavinic, D. S., & Beckie, R. D. (2008). Nucleation and growth kinetics of struvite in a fluidized bed reactor. Journal of Crystal Growth, 310(6), 1187–1194. http://doi.org/10.1016/j.jcrysgro.2007.12.054 Bhuiyan, M. I., Mavinic, D. S., & Koch, F. A. (2008). Phosphorus recovery from wastewater through struvite formation in fluidized bed reactors: a sustainable approach. Water Science Technology, 57(2), 175–181. Bhuiyan, Mavinic, & Beckie. (2007). A Solubility and Thermodynamic Study of Struvite. Environmental Technology, 289(28), 1015–1026. http://doi.org/10.1080/09593332808618857 Bouropoulos, N. C., & Koutsoukos, P. G. (2000). Spontaneous precipitation of struvite from aqueous solutions. Journal of Crystal Growth, 213(3–4), 381–388. http://doi.org/10.1016/S0022-0248(00)00351-1 Bowers, K. E. (2013). Technology Provider Presentations: Multiform Harvest Inc.,. Vancouver, BC, Canada: in: WEF/IWA Nutrient Removal and Recovery. Workshop A - Making Sence of Extractive Nutrient Recovery. Britton, A., Clark, D. R., & Sathyanarayana, R. P. M. (2015). Methods for fines control. USA: United States Patent and Trademark Office. Britton, A., Koch, F. A., Mavinic, D. S., Adnan, A., Oldham, W. K., & Udala, B. (2005). Pilot-scale struvite recovery from anaerobic digester supernatant at an enhance biological phosphorus removal wastewater treatment plant. Journal of Environmental Engineering and Science, 4(4), 265–277. Brunsteiner, M., Jones, A. G., Pratola, F., Price, S. L., & Simons, S. J. R. (2005). Toward a molecular understanding of crystal agglomeration. Crystal Growth and Design, 5(1),  173 3–16. http://doi.org/10.1021/cg049837m Chauhan, C. K., & Joshi, M. J. (2013). In vitro crystallization, characterization and growth-inhibition study of urinary type struvite crystals. Journal of Crystal Growth, 362(1), 330–337. http://doi.org/10.1016/j.jcrysgro.2011.11.008 Corbridge, D. E. . (2013). Phosphorus: Chemistry, Biochemistry and Technology (6th ed.). Boca Raton (FL): CRC/Taylor & Francis. Crutchik, D., Morales, N., Vázquez-Padín, J. R., & Garrido, J. M. (2017). Enhancement of struvite pellets crystallization in a full-scale plant using an industrial grade magnesium product. Water Science and Technology, 75(3), 609–618. http://doi.org/10.2166/wst.2016.527 Cusick, R. D., Ullery, M. L., Dempsey, B. A., & Logan, B. E. (2014). Electrochemical struvite precipitation from digestate with a fluidized bed cathode microbial electrolysis cell. Water Research, 54, 297–306. http://doi.org/10.1016/j.watres.2014.01.051 Derby, J. J. (2016). Fluid dynamics in crystal growth: The good, the bad, and the ugly. Progress in Crystal Growth and Characterization of Materials, 62(2), 286–301. http://doi.org/10.1016/j.pcrysgrow.2016.04.015 Derjaguin, B., & Landau, L. (1941). Theory of the stability of strongly charged lyophobic sols and of the adhesion of strongly charged particles in solutions of electrolytes. Acta Physico Chemica URSS, 14, 633. Eaton, A. D., Clesceri, L. S., & Greenberg, A. E. (Eds.). (1995). Standard Methods for the Examination of Water and Wastewater (19th ed.). Washington: American Public Health Association. Fattah, K. P., Mavinic, D. S., & Koch, F. a. (2012). Influence of Process Parameters on the Characteristics of Struvite Pellets. Journal of Environmental Engineering, 138(12), 1200–1209. http://doi.org/10.1061/(ASCE)EE.1943-7870.0000576 Fattah, K. P., Mavinic, D. S., Koch, F. A., & Jacob, C. (2008). Determining the feasibility  174 of phosphorus recovery as struvite from filter press centrate in a secondary wastewater treatment plant. Journal of Environmental Science and Health - Part A Toxic/Hazardous Substances and Environmental Engineering, 43(7), 756–764. http://doi.org/10.1080/10934520801960052 Ferraris, G. (1986). Sec B: Structural Science. Acta Crystallogr., 42, 253. Forrest, A. L. (2004). Process Optimization of a Technical Scale Phosphorus Recovery System Through Struvite Crystallization at the City of Penticton Advanced Wastewater Treatment Plant. University of British Columbia. Retrieved from http://hdl.handle.net/2429/15946 Forrest, A. L., Fattah, K. P., Mavinic, D. S., & Koch, F. A. (2008). Optimizing Struvite Production for Phosphate Recovery in WWTP. Journal of Environmental Engineering, 134(5), 395–402. http://doi.org/10.1061/(ASCE)0733-9372(2008)134:5(395) Galbraith, S. C., Schneider, P. A., & Flood, A. E. (2014). Model-driven experimental evaluation of struvite nucleation, growth and aggregation kinetics. Water Research, 56, 122–132. http://doi.org/10.1016/j.watres.2014.03.002 Garside, J., Mersmann, A., & Nyvlt, J. (Eds.). (2002). Measurement of Crystal Growth and Nucleation Rates (Second Edi). Rugby: IChemE. Ghosh, S., Lobanov, S., & Lo, V. K. (2018). Impact of supersaturation ratio on phosphorus recovery from synthetic anaerobic digester supernatant through a struvite crystallization fluidized bed reactor. Environmental Technology (United Kingdom), 0(0), 1–11. http://doi.org/10.1080/09593330.2018.1435734 Giavarina, D. (2015). Lessons in Biostatistics: Understanding Bland Altman Analysis. Biochemia Medica, 25(2), 141–151. http://doi.org/10.11613/BM.2013.003 Government of Canada, S. C. (2018). Canadian International Merchandise Trade Database. Retrieved October 24, 2018, from http://www5.statcan.gc.ca/cimt-cicm/commodities- 175 marchandises?lang=eng&chapterId=31&chapterName=Fertilizers&sectionId=6&sectionLabel=VI+-+Products+of+the+chemical+or+allied+industries.&refMonth=8&refYr=2018&freq=6&countryId=999&usaState=0&provId=1&dataTran Harrison, M. L., Johns, M. R., White, E. T., & Mehta, C. M. (2011). Growth Rate Kinetics for Struvite Crystallisation. Chemical Engineering Transactions, 25, 309–314. http://doi.org/10.3303/CET1125052 Healy, T. W., & Jellett, V. R. (1967). Adsorption-coagulation reactions of Zn (II) hydrolyzed species at the zinc oxide—water interface. Journal of Colloid and Interface Science, 24(1), 41–46. http://doi.org/10.1016/0021-9797(67)90274-3 Holtkamp, M. (2014). smorf crystal models. Retrieved December 11, 2018, from http://www.smorf.nl Huang, H., Mavinic, D. S., Lo, K. V, & Koch, F. A. (2006). Production and Basic Morphology of Struvite Crystals from a Pilot-Scale Crystallization Process. Environmental Technology, 27(3), 233–245. http://doi.org/10.1080/09593332708618637 Hudson Institute of Mineralogy. (2018). Struvite: Struvite mineral information and data. Retrieved October 18, 2018, from https://www.mindat.org/min-3811.html Hunter, R. J. (1981). Zeta Potential in Colloid Science: Principles and Applications. (R. . Ottewill & R. L. Rowell, Eds.). London: Academic Press. Jackson, K. A. (2010). Kinetic Processes: Crystal Growth, Diffusion, and Phase Transitions in Materials (Second Edi). Weinheim, Germany: Wiley-VCH. Jones, A. G. (2002). Crystallization Process Systems. Oxford: Butterworth-Heinemann. Jönsson, H., Stinzing, A. R., Vinneras, B., & Salomon, E. (2004). Guidelines on the Use of Urine and Faeces in Crop Production. EcoSanRes Publication Series, 2, 1–35. Katz, J. L. (1972). Elastic Properties of Urinary Stones: Some Experimental and  176 Theoretical Observations. In B. Finlayson, L. L. Hench, & L. H. Smith (Eds.), Urolithiasis: Physical Aspects (pp. 215–234). Washington, DC.: National Academy of Science. Retrieved from https://books.google.ca/books?id=gkArAAAAYAAJ&pg=PA215&lpg=PA215&dq=elastic+properties+of+urinary+stones:+some+experimental&source=bl&ots=MCVA6nEnAX&sig=ss9jN6R1MxycoraAwv7tSbLkpWI&hl=en&sa=X&ved=0ahUKEwj3qt3kltXOAhVW1GMKHQBuAQQQ6AEIHjAA#v=onepage&q=elast Kiriukhin, M. Y., & Collins, K. D. (2002). Dynamic hydration numbers for biologically important ions. Biophysical Chemistry, 99(2), 155–168. http://doi.org/10.1016/S0301-4622(02)00153-9 Koch, F. A., Mavinic, D. S., Yonemitsu, N., & Britton, A. T. (2008). Reactor Patent - US2008257826A1.pdf. Canada: United States. Kosmulski, M. (Ed.). (2009). Surface Charging and Points of Zero Charge. Boca Raton (FL): Taylor & Francis. Koutsoukos, P. G., & Kofina, A. N. (2007). Struvite Scale Formation and Inhibition in Wastewater. In NACE-07057 (pp. 1–15). Nashville, Tennessee: NACE International. Retrieved from https://www-onepetro-org.ezproxy.library.ubc.ca/conference-paper/NACE-07057 Lancia, A., & Musmarra, D. (1999). Measuring Induction Period for Calcium Sulfate Dihydrate Precipitation. AIChE Journal, 45(2), 390–397. http://doi.org/10.1002/aic.690450218 Latifian, M., Liu, J., & Mattiasson, B. (2012). Struvite-based fertilizer and its physical and chemical properties. Environmetnal Technology, 33(24), 2691–2697. http://doi.org/10.1080/09593330.2012.676073 Le Corre, K. S. (2006). Understanding Struvite Crystallisation and Recovery. Cranfield University.  177 Le Corre, K. S., Valsami-Jones, E., Hobbs, P., Jefferson, B., & Parsons, S. a. (2007). Agglomeration of struvite crystals. Water Research, 41(2), 419–25. http://doi.org/10.1016/j.watres.2006.10.025 Le Corre, K. S., Valsami-Jones, E., Hobbs, P., & Parsons, S. A. (2005). Impact of calcium on struvite crystal size, shape and purity. Journal of Crystal Growth, 283(3–4), 514–522. http://doi.org/10.1016/j.jcrysgro.2005.06.012 Le Corre, K. S., Valsami-Jones, E., Hobbs, P., & Parsons, S. a. (2007). Impact of reactor operation on success of struvite precipitation from synthetic liquors. Environmental Technology, 28(11), 1245–56. http://doi.org/10.1080/09593332808618885 Lewis, C. L., Craig, C. C., & Senecal, A. G. (2014). Mass and Density Measurements of Live and Dead Gram-Negative and Gram-Positive Bacterial Populations. Applied and Environmental Microbiology, 80(12), 3622–3631. http://doi.org/10.1128/AEM.00117-14 Li, H., Yao, Q. Z., Yu, S. H., Huang, Y. R., Chen, X. D., Fu, S. Q., & Zhou, G. T. (2017). Bacterially mediated morphogenesis of struvite and its implication for phosphorus recovery. American Mineralogist, 102(2), 381–390. http://doi.org/10.2138/am-2017-5687 Linnikov, O. D. (2008). Mechanism of aggregation and intergrowth of crystals during bulk crystallization from solutions. Crystal Research and Technology, 43(12), 1268–1277. http://doi.org/10.1002/crat.200800176 Linnikov, O. D., Grigorov, I. G., Rodina, I. V., & Polyakov, E. V. (2011). Mechanism of potassium nitrate crystal intergrowth during spontaneous crystallization from supersaturated aqueous solutions. Doklady Physical Chemistry, 439(1), 135–138. http://doi.org/10.1134/S0012501611070049 Liu, J. C. (2009). Recovery of phosphate and ammonium as struvite from semiconductor wastewater. Separation and Purification Technology, 64(3), 368–373.  178 http://doi.org/10.1016/j.seppur.2008.10.040 Lo, K. V., Ning, R., Kei, C., & Oliveira, Y. De. (2017). Application of Microwave Oxidation Process for Sewage Sludge Treatment in a Continuous-Flow System, 143(9), 1–9. http://doi.org/10.1061/(ASCE)EE.1943-7870.0001247. Lobanov, S., Koch, F. A., & Mavinic, D. S. (2013). The Implication of Aqueous Equilibrium Modelling to Evaluate the Potential for Nutrient Recovery from Wastewater Streams. In WEF/IWA Nutrient Removal and Recovery: Trends in Resource Recovery and Use (p. (USB proceedings)). Vancouver. MacIver, M. R., & Pawlik, M. (2017). Analysis of In Situ Microscopy Images of Flocculated Sediment Volumes. Chemical Engineering and Technology, 40(12), 2305–2313. http://doi.org/10.1002/ceat.201600523 Markov, I. V. (2003). Crystal Growth for Beginners: fundamentals of nucleation, crystal growth and epitaxy (2nd ed.). NJ: World Scientific. Mavinic, D. S., Koch, F. a, Huang, H., & Lo, K. V. (2007). Phosphorus recovery from anaerobic digester supernatants using a pilot-scale struvite crystallization process. Journal of Environmental Engineering and Science, 6(5), 561–571. http://doi.org/10.1139/S07-007 Mazziotti di Celso, G., Prisciandaro, M., Karatza, D., Lancia, A., & Musmarra, D. (2017). Salt scales on process equipment: Measurement of the induction time for calcium carbonate nucleation. Canadian Journal of Chemical Engineering, 95(9), 1671–1675. http://doi.org/10.1002/cjce.22878 Mehta, C. M., & Batstone, D. J. (2013). Nucleation and growth kinetics of struvite crystallization. Water Research, 47(8), 2890–900. http://doi.org/10.1016/j.watres.2013.03.007 Mihelcic, J. R., Fry, L. M., & Shaw, R. (2011). Global potential of phosphorus recovery from human urine and feces. Chemosphere, 84(6), 832–839.  179 http://doi.org/10.1016/j.chemosphere.2011.02.046 Ohlinger, K. N., Young, T. M., & Schroeder, E. D. (1999). KINETICS EFFECTS ON PREFERENTAIL STRUVITE ACCUMULATION IN WASTEWATER. Journal of Environmental Engineering, 125(8), 730–737. http://doi.org/10.1061/(ASCE)0733-9372(1999)125:8(730) Oleszkiewicz, J., Kruk, D., Devlin, T., Lashkarizadeh, M., & Yuan, Q. (2015). Options for Improved Nutrient Removal and Recovery from Municipal Wastewater in the Canadian Context. Waterloo ON. Retrieved from http://www.cwn-rce.ca/assets/resources/pdf/Oleszkiewicz-Nutrients/Oleszkiewicz-KI-Report-March-2015.pdf Ostara Nutrient Recovery Technologies Inc. (2017). Ostara Nutrient Recovery. Retrieved November 17, 2017, from http://ostara.com/about/ Pietsch, W. (1991). Size Enlargement by Agglomeration. Chichester: John Wiley & Sons. Ping, Q., Li, Y., Wu, X., Yang, L., & Wang, L. (2016). Characterization of morphology and component of struvite pellets crystallized from sludge dewatering liquor: Effects of total suspended solid and phosphate concentrations. Journal of Hazardous Materials, 310, 261–269. http://doi.org/10.1016/j.jhazmat.2016.02.047 Prywer, J., Sadowski, R. R., & Torzewska, A. (2015). Aggregation of struvite, carbonate apatite, and Proteus mirabilis as a key factor of infectious urinary stone formation. Crystal Growth and Design, 15(3), 1446–1451. http://doi.org/10.1021/cg5018032 Prywer, J., & Torzewska, A. (2009). Bacterially induced struvite growth from synthetic urine: Experimental and theoretical characterization of crystal morphology. Crystal Growth and Design, 9(8), 3538–3543. http://doi.org/10.1021/cg900281g Prywer, J., Torzewska, A., & Plociński, T. (2012). Unique surface and internal structure of struvite crystals formed by Proteus mirabilis. Urological Research, 40(6), 699–707. http://doi.org/10.1007/s00240-012-0501-3  180 Romanowski, Z., Kempisty, P., Prywer, J., Krukowski, S., & Torzewska, A. (2010). Density Functional Theory Determination of Structural and Electronic Properties of Struvite. Journal of Physical Chemistry, 114, 7800–7808. http://doi.org/10.1021/jp102887a Ronteltap, M., Maurer, M., Hausherr, R., & Gujer, W. (2010). Struvite precipitation from urine - Influencing factors on particle size. Water Research, 44(6), 2038–2046. http://doi.org/10.1016/j.watres.2009.12.015 Shimamura, K., Ishikawa, H., Tanaka, T., & Hirasawa, I. (2007). Use of a seeder reactor to manage crystal growth in the fluidized bed reactor for phosphorus recovery. Water Environment Research : A Research Publication of the Water Environment Federation, 79(4), 406–13. http://doi.org/10.2175/106143006X111899 Skuce, R. (2015). Bacteria – mineral – nanoparticle interactions in water and wastewater. University of Glasgow. Retrieved from http://theses.gla.ac.uk/6153 Snoeyink, V. L., & Jenkins, D. (1980). Water Chemistry. New York: John Wiley & Sons. Stumm, W., & Morgan, J. J. (1981). Aquatic Chemistry: An Introduction Emphasizing Chemical Equilibria in Natural Waters (2nd ed.). New York: John Wiley & Sons. Sun, W. D., Zhang, K. C., Wang, J. Y., & Wang, X. L. (2010). The chemical composition and ultrastructure of uroliths in Boer goats. Veterinary Journal, 186(1), 70–75. http://doi.org/10.1016/j.tvjl.2009.07.009 Sunagawa, I. (2005). Crystals: Growth, Morphology, and Perfection. Cambridge, England: Cambridge University Press. Tchobanoglous, George, Burton, Franklin L., Stensel, H. David, M. & E. (2003). Wastewater engineering: treatment and reuse / Metcalf & Eddy (4th ed.). New York: McGraw-Hill. Triger, A., Pic, J. S., & Cabassud, C. (2012). Determination of struvite crystallization mechanisms in urine using turbidity measurement. Water Research, 46(18), 6084–6094. http://doi.org/10.1016/j.watres.2012.08.030  181 Ueno, Y., & Fujii, M. (2001). Three years experience of operating and selling recovered struvite from full-scale plant. Environmental Technology, 22(11), 1373–1381. http://doi.org/10.1080/09593332208618196 United Nations. (2017). United Nations Department of Economic and Social Affairs. Retrieved December 13, 2017, from https://www.un.org/development/desa/publications/world-population-prospects-the-2017-revision.html US Geological Survey. (2018). Mineral Commodity Summaries 2018, Phosphate Rock. US. Retrieved from https://minerals.usgs.gov/minerals/pubs/commodity/phosphate_rock/mcs-2018-phosp.pdf Verwey, E., & Overbeek, J. (1948). Theory of the Stability of Lyophobic Colloids. Elsevier. Amsterdam: Elsevier. Vol’khin, V. V., Kazakov, D. A., Leont’eva, G. V., Andreeva, Y. V., Nosenko, E. A., & Siluyanova, M. Y. (2015). Synthesis of struvite (MgNH4PO4·6H2O) and its use for sorption of nickel ions. Russian Journal of Applied Chemistry, 88(12), 1986–1996. http://doi.org/10.1134/S10704272150120149 Ye, X., Chu, D., Lou, Y., Ye, Z. L., Wang, M. K., & Chen, S. (2017). Numerical simulation of flow hydrodynamics of struvite pellets in a liquid–solid fluidized bed. Journal of Environmental Sciences (China), 57, 391–401. http://doi.org/10.1016/j.jes.2016.11.019 Ye, X., Ye, Z. L., Lou, Y., Pan, S., Wang, X., Wang, M. K., & Chen, S. (2016). A comprehensive understanding of saturation index and upflow velocity in a pilot-scale fluidized bed reactor for struvite recovery from swine wastewater. Powder Technology, 295, 16–26. http://doi.org/10.1016/j.powtec.2016.03.022 Ye, Z.-L., Deng, Y., Ye, X., Lou, Y., & Chen, S. (2018). Application of image processing  182 on struvite recovery from swine wastewater by using the fluidized bed. Water Science and Technology, 77(1), 159–166. http://doi.org/10.2166/wst.2017.533 Ye, Z., Shen, Y., Ye, X., Zhang, Z., Chen, S., & Shi, J. (2014). Phosphorus recovery from wastewater by struvite crystallization: Property of aggregates. Journal of Environmental Sciences (China), 26(5), 991–1000. http://doi.org/10.1016/S1001-0742(13)60536-7        183 Appendices Appendix A Canadian Fertilizer Imports for 2017  HS Code Description Quantity (kg) Value (CAN$) 310311 Superphosphates fertilizer, cont by weight >= 35% diphosphorus pentaoxide (P2O5) 1,148,685 777,474 310319 Superphosphates fertilizer, nes 312,337 180,299 310390 Mineral or chemical fertilizers, phosphatic, nes, in packages weighing > 10 kg 6,123,226 6,819,029 310520 Fertilizers, cont nitrogen, phosphorus and potassium, in packages weighing > 10 kg 92,535,528 39,234,918 310530 Diammonium phosphate, in packages weighing > 10 kg 250,992,930 43,652,114 310540 Monoammonium phosphate and mx thereof, with diamonium phosphate, in pack > 10 kg 945,632,298 466,508,641 310551 Fertilizers, containing nitrates and phosphates, nes, in pack weighing > 10 kg 6,428,335 8,412,582 310559 Fertilizers, containing nitrogen and phosphorus, nes, in pack weighing > 10kg 555,425,156 278,123,172 310560 Fertilizers, containing phosphorus and potassium, in packages weighing > 10 kg 2,233,066 5,136,444 Total 1,860,831,561 848,844,673 Reference: (Government of Canada, 2018)     184 Appendix B XRD Results for Synthetic Struvite Crystals Used in Aged Zeta Potential Experiments                          00-036-1491 (*) - Dittmarite, syn - NH4MgPO4·H2O - WL: 1.5406 - Orthorhombic - a 5.61330 - b 8.76740 - c 4.79130 - alpha 90.000 - beta 90.000 - gamma 90.000 - Primitive - Pmnm (59) - 2 - 235.799 01-077-2303 (*) - Struvite, syn - MgNH4PO4(H2O)6 - WL: 1.5406 - Orthorhombic - a 6.95500 - b 6.14200 - c 11.21800 - alpha 90.000 - beta 90.000 - gamma 90.000 - Primitive - Pmn21 (31) - 2 - 479.20File: Marcia1.raw - Start: 5.022 ° - End: 90.015 ° - Step: 0.019 ° - Step time: 48. s - Anode: Cu - WL1: 1.5406 - WL2: 1.54439 - kA2 Ratio: 0.5 - Generator kV: 40 kV - Generator mA: 40 mA - Creation: 13Lin (Cps)01002003004005002-Theta - Scale5 10 20 30 40 50 60 70 80 90d=7.59697, 11.639 °d=3.79824, 23.402 ° 185 Appendix C Graph of Mixing Speed Influence on Induction Time   Forced slope to be 0.52       y = 0.5175x + 1.505R² = 0.7952y = 0.519x + 1.3759R² = 0.7991y = 0.5171x + 1.31R² = 0.787711.522.533.50 0.5 1 1.5 2 2.5 3 3.5 4Log(Induction Time)1/(Log SSR)^2RPM 200 RPM 400 RPM 600 186  Appendix D XRD Results from Various Nucleation Experiments        Run 400-015-0762 (*) - Struvite, syn - NH4MgPO4·6H2O - Y: 45.04 % - d x by: 1. - WL: 1.5406 - Orthorhombic - a 6.94500 - b 11.20800 - c 6.13550 - alpha 90.000 - beta 90.000 - gamma 90.000 - Primitive - Operations: ImportFile: Run4-M.raw - Start: 5.000 ° - End: 64.994 ° - Step: 0.019 ° - Step time: 36.4 s - Anode: Cu - WL1: 1.5406 - WL2: 1.54439 - kA2 Ratio: 0.5 - Generator kV: 40 kV - Generator mA: 40 mALin (Count s)010002000300040005000600070008000900010000110001200013000140001500016000170001800019000200002100022000230002400025000260002700028000290003000031000320002-Theta - Scale5 10 20 30 40 50 60Run 900-015-0762 (*) - Struvite, syn - NH4MgPO4·6H2O - Y: 50.74 % - d x by: 1. - WL: 1.5406 - Orthorhombic - a 6.94500 - b 11.20800 - c 6.13550 - alpha 90.000 - beta 90.000 - gamma 90.000 - Primitive - Operations: ImportFile: Run9-M.raw - Start: 5.000 ° - End: 64.994 ° - Step: 0.019 ° - Step time: 36.4 s - Anode: Cu - WL1: 1.5406 - WL2: 1.54439 - kA2 Ratio: 0.5 - Generator kV: 40 kV - Generator mA: 40 mALin (Counts)01000020000300002-Theta - Scale5 10 20 30 40 50 60Run 12 + 13 + 1500-015-0762 (*) - Struvite, syn - NH4MgPO4·6H2O - Y: 42.46 % - d x by: 1. - WL: 1.5406 - Orthorhombic - a 6.94500 - b 11.20800 - c 6.13550 - alpha 90.000 - beta 90.000 - gamma 90.000 - Primitive - Operations: ImportFile: Run12-13-15-M.raw - Start: 5.000 ° - End: 64.994 ° - Step: 0.019 ° - Step time: 36.4 s - Anode: Cu - WL1: 1.5406 - WL2: 1.54439 - kA2 Ratio: 0.5 - Generator kV: 40 kV - Generator mA: 40 mALin (Count s)0100020003000400050006000700080009000100001100012000130001400015000160001700018000190002000021000220002300024000250002600027000280002900030000310002-Theta - Scale5 10 20 30 40 50 60Run 1900-015-0762 (*) - Struvite, syn - NH4MgPO4·6H2O - Y: 47.61 % - d x by: 1. - WL: 1.5406 - Orthorhombic - a 6.94500 - b 11.20800 - c 6.13550 - alpha 90.000 - beta 90.000 - gamma 90.000 - Primitive - Operations: ImportFile: Run19-M.raw - Start: 5.000 ° - End: 64.994 ° - Step: 0.019 ° - Step time: 36.4 s - Anode: Cu - WL1: 1.5406 - WL2: 1.54439 - kA2 Ratio: 0.5 - Generator kV: 40 kV - Generator mA: 40 mALin (Counts)010002000300040005000600070008000900010000110001200013000140001500016000170001800019000200002100022000230002400025000260002700028000290003000031000320003300034000350002-Theta - Scale5 10 20 30 40 50 60Run 2200-015-0762 (*) - Struvite, syn - NH4MgPO4·6H2O - Y: 57.18 % - d x by: 1. - WL: 1.5406 - Orthorhombic - a 6.94500 - b 11.20800 - c 6.13550 - alpha 90.000 - beta 90.000 - gamma 90.000 - Primitive - Operations: ImportFile: Run22-M.raw - Start: 5.000 ° - End: 64.994 ° - Step: 0.019 ° - Step time: 36.4 s - Anode: Cu - WL1: 1.5406 - WL2: 1.54439 - kA2 Ratio: 0.5 - Generator kV: 40 kV - Generator mA: 40 mALin (Count s)010002000300040005000600070008000900010000110001200013000140001500016000170001800019000200002100022000230002400025000260002700028000290003000031000320002-Theta - Scale5 10 20 30 40 50 60Run 28 + 2900-015-0762 (*) - Struvite, syn - NH4MgPO4·6H2O - Y: 64.34 % - d x by: 1. - WL: 1.5406 - Orthorhombic - a 6.94500 - b 11.20800 - c 6.13550 - alpha 90.000 - beta 90.000 - gamma 90.000 - Primitive - Operations: ImportFile: Run28-29-M.raw - Start: 5.000 ° - End: 64.994 ° - Step: 0.019 ° - Step time: 36.4 s - Anode: Cu - WL1: 1.5406 - WL2: 1.54439 - kA2 Ratio: 0.5 - Generator kV: 40 kV - Generator mA: 40 mALin (Counts)01000200030004000500060007000800090001000011000120001300014000150001600017000180001900020000210002200023000240002500026000270002800029000300002-Theta - Scale5 10 20 30 40 50 60 187  Appendix E Zeta Potential Data Time (mins) pH Cond. (mS/cm) Temp (°C) ζ (mV) ζ SD (mV) Mg (mg/L) N (mg/L) P (mg/L) SSR Experiment 102 00:00:00 7.493 7.63 23.7     75 771 93 12.59 00:03:36 7.423 7.63 23.7     77 749 90 10.23 00:05:32       2 5.3         00:07:27       5.1 12.5         00:09:23 7.29     7.3 9         00:11:55 7.228 7.63 23.7     64 748 76 4.57 00:13:50       1.9 5.8         00:15:46       -6.5 5.6         00:17:41       -11.3 1.6         00:19:50 7.115 7.63 23.6     62 740 70 2.95 00:21:47       -10.5 1.6         00:23:43       -8.5 8         00:25:40       -10.5 1.6         00:28:05 7.07 7.63 23.6     61 740 72 2.63 00:30:02       -15 9         00:31:59       -13.8 1.6         00:33:56 7.053     -13.3 1.6         00:36:40 7.05 7.6 23.5     60 743 67 2.34 00:38:36       -19.1 1.6         00:40:33       -13.4 15         00:42:29       -20.9 18.8         Experiment 103 00:00:00 7.458 7.61 24     78 765 92 11.75 00:02:50 7.41 7.68 24     75 733 88 9.33 00:04:27       -2.4 5         00:06:05       6.4 11.5         00:07:42 7.278     4.6 9         00:09:17 7.222 7.64 24     68 753 80 4.90 00:10:54       1.2 2.6         00:12:31       -0.1 6.4         00:14:08 7.101     -9.7 2.6         00:16:22 7.072 7.61 23.9     62 746 73 2.75 00:17:59       -7.5 6.8         00:19:36       -14.6 1.6         00:21:13 7.035     -14.9 1.6         00:23:21 7.021 7.64 23.9     60   68    188 Time (mins) pH Cond. (mS/cm) Temp (°C) ζ (mV) ζ SD (mV) Mg (mg/L) N (mg/L) P (mg/L) SSR 00:24:57       -12.3 1.6         00:26:34       -15.5 1.6         00:28:10 7.006     -19 1.8         00:30:13 7.002 7.6 23.8     60 736 70 2.04 00:31:49       -20.1 1.6         00:33:25       -18.7 1.6         Experiment 104 00:00:00 7.47 7.66 24.2     76 770 98 12.59 00:02:07 7.39 7.66 24.2     77 763 91 9.55 00:09:28 7.233 7.67 24.2     70 753 87 5.50 00:11:03       3.9 1.6         00:12:38       6.1 5.9         00:14:13 7.148     -0.9 1.6         00:16:00 7.096 7.66 24.1     64 743 81 3.24 00:17:37       1.2 1.6         00:19:13       -4.4 1.6         00:20:50       -3.8 1.6         00:23:03 7.026 7.66 24.1     64 761 78 2.63 00:24:40       -7.1 1.6         00:26:16       -10.7 1.6         00:27:53 7.032     -14.9 1.6         00:29:49 7.024 7.62 24.1     63 754 70 2.34 00:31:27       -9.1 1.6         00:33:04       -9.1 5.8         Experiment 105 00:00:00 7.189 7.58 22.7     79 767 92 5.89 00:04:52 7.129 7.6 22.7     75 755 89 4.68 00:06:29       -9.9 21.2         00:08:05       -3.3 1.6         00:09:42 7.121     -2.4 8.3         00:12:06 7.118 7.62 22.8     75 758 89 4.57 00:13:42       3.3 9.8         00:15:18       -4 8.3         00:16:54 7.106     -3 1.6         00:19:00 7.099 7.61 22.8     74 756 88 4.17 00:20:37       -0.5 1.6         00:22:13       -0.7 1.7         00:23:50 7.077     -6.3 1.6         00:26:24 7.066 7.6 22.9     72 755 87 3.72  189 Time (mins) pH Cond. (mS/cm) Temp (°C) ζ (mV) ζ SD (mV) Mg (mg/L) N (mg/L) P (mg/L) SSR 00:28:00       2.4 1.6         00:29:37       -12.8 1.6         00:31:13 7.042     -7 1.6         00:34:17 7.028 7.62 22.9     70 749 83 3.09 00:35:53       -11 1.6         00:37:30       -8.6 1.6         00:39:06 7.01     -10.7 1.7         00:41:43 6.999 7.6 22.9     69 760 82 2.82 00:43:20       -6.2 1.6         00:44:58       -9 1.6         00:46:35 6.984     -14.7 1.6         Experiment 106 00:00:00 7.132 7.63 23.3     76 762 90 4.79 00:05:32 7.076 7.63 23.3     74 719 85 3.63 00:07:08       -0.1 9.3         00:08:44       -12 12.2         00:10:20 7.069     -6.7 10.8         00:12:13 7.063 7.64 23.3     73 747 85 3.55 00:13:49       0 1.6         00:15:26       -6.6 1.6         00:17:02 7.043     -10.2 1.6         00:19:04 7.031 7.62 23.3     71 755 87 3.24 00:20:40       -3.1 1.6         00:22:16       -7 1.6         00:23:52 7.003     -9.3 1.6         00:25:52 6.991 7.6 23.3     70 758 85 2.82 00:27:28       -8.3 1.6         00:29:05       -8.4 1.6         00:30:41 6.963     -6.5 1.6         00:33:10 6.952 7.61 23.3     68 755 83 2.40 00:34:46       -9.9 1.6         00:36:23       -8.6 1.6         00:37:59 6.93     -13.8 1.6         00:40:02 6.924 7.59 23.3     68 755 82 2.14 00:41:38       -11.9 1.6         00:43:13       -10.9 5.7         00:44:49 6.908     -9.5 1.6         Experiment 124 00:00:00 7.522 7.49 23.7     81 762 89 13.80  190 Time (mins) pH Cond. (mS/cm) Temp (°C) ζ (mV) ζ SD (mV) Mg (mg/L) N (mg/L) P (mg/L) SSR 00:02:52 7.471 7.64 23.7     81 754 86 11.75 00:04:28       1 14.4         00:06:04       11.9 1.6         00:07:40 7.375     6.3 1.6         00:09:13 7.328 7.61 23.7     72 743 74 6.46 00:10:49       13.2 6.2         00:12:25       11.4 1.6         00:14:01 7.21     7.8 1.6         00:15:44 7.189 7.57 23.6     69 747 75 4.27 00:17:20       3.2 1.6         00:18:55       -2.7 1.6         00:20:31 7.128     2.9 1.6         00:22:08 7.117 7.57 23.6     68 742 70 3.24 00:23:44       3.7 1.6         00:25:20       -4.7 1.6         00:26:56 7.094     -5.9 1.6         00:28:31 7.088 7.57 23.6     64 744 68 2.75 00:30:07       -1.3 10.1         00:31:43       -7.5 11.4         00:33:19 7.074     -7.9 1.6         00:35:01 7.071 7.55 23.6     63 739 67 2.57 00:36:37       -3.2 1.6         00:38:13       -10.1 1.6         00:39:49 7.062     0.1 12.1         Experiment 125 00:00:00 7.492 7.61 23.7     80 760 89 12.59 00:02:50 7.432 7.65 23.8     77 749 86 10.23 00:04:26       10.2 1.6         00:06:02       7.5 1.6         00:07:38 7.33     9.2 1.6         00:09:26 7.279 7.59 23.7     71 749 78 5.89 00:11:02       10.1 1.6         00:12:38       5.3 1.6         00:14:14 7.174     8.1 1.6         00:15:53 7.147 7.55 23.7     67 749 73 3.63 00:17:29       7.4 1.6         00:19:04       0.5 1.6         00:20:40 7.102     1.5 1.6         00:22:49 7.089 7.55 23.7     65 747 70 2.88  191 Time (mins) pH Cond. (mS/cm) Temp (°C) ζ (mV) ζ SD (mV) Mg (mg/L) N (mg/L) P (mg/L) SSR 00:24:25       4.4 1.6         00:26:00       -1.5 1.6         00:27:36 7.069     -2.6 1.6         00:29:31 7.062 7.54 23.6     63 744 69 2.57 00:31:07       0 1.6         00:32:42       -6 1.6         00:34:18 7.052     -5.6 6.4         00:36:05 7.048 7.55 23.6     64 747 69 2.51 00:37:41       -0.4 1.6         00:39:16       -11.6 16.8         00:40:52 7.038     1.8 5.8         Experiment 126 00:00:00 7.489 7.64 23.7     81 767 90 12.88 00:02:22 7.453 7.66 23.7     80 758 89 11.48 00:03:58       11.1 6.6         00:05:34       12.7 1.6         00:07:10 7.368     4.9 7.8         00:09:06 7.32 7.57 23.7     75 755 81 7.08 00:10:42       13 1.6         00:12:19       11.6 1.6         00:13:55 7.218     5.2 1.6         00:16:06 7.184 7.55 23.6     70 746 75 4.27 00:17:42       -0.8 6.5         00:19:17       3.5 5.1         00:20:53 7.134     6 7         00:22:40 7.119 7.57 23.6     68 747 72 3.39 00:24:16       1.7 1.6         00:25:52       4.2 1.6         00:27:28 7.097     -0.7 5.1         00:29:41 7.091 7.55 23.6     67 746 71 3.02 00:31:17       3.6 1.6         00:32:53       -4.8 1.6         00:34:29 7.078     -7.1 1.6         00:36:18 7.074 7.55 23.5     67 749 70 2.88 00:37:54       -1.6 1.6         00:39:31       -4.3 1.6         00:41:07 7.062     -12.1 9.7         Experiment 127 00:00:00 7.487 7.64 23.8     86 801 96 14.45  192 Time (mins) pH Cond. (mS/cm) Temp (°C) ζ (mV) ζ SD (mV) Mg (mg/L) N (mg/L) P (mg/L) SSR 00:02:30 7.442 7.63 23.8     82 759 90 11.22 00:04:06       3.8 1.6         00:05:42       2.1 8.7         00:07:18 7.379     -6.7 1.6         00:09:11 7.349 7.58 23.8     76 760 83 7.76 00:10:44       14.4 9.6         00:12:16       12.8 1.6         00:13:49 7.254     6.8 1.6         00:15:39 7.221 7.54 23.7     71 755 77 4.79 00:17:15       7 3.7         00:18:50       4.6 7         00:20:26 7.166     7.2 5.8         00:22:38 7.146 7.57 23.7     68 753 74 4.47 00:24:14       3.6 11.5         00:25:50       2.9 13         00:27:26 7.122     5.2 9.7         00:29:35 7.11 7.56 23.6     68 756 73 3.24 00:31:11       0.5 6.5         00:32:47       -6 1.6         00:34:23 7.098     -3.2 8.1         00:36:39 7.091 7.55 23.6     67 749 72 2.95 00:38:15       -0.3 6.2         00:39:51       -4.1 1.6         00:41:27 7.081     13.1 1.6         Experiment 209 00:00:00 6.32 10.83 24.2     363 983 523 6.03 00:02:17 6.3 10.96 24.1     357       00:03:53       12.2 1.9         00:05:30       2.5 1.9         00:07:06 6.289     63.8 1.9         00:08:52 6.276 10.89 24.1     363 971 508 4.79 00:10:28       4.6 4.2         00:12:05       10.9 4.2         00:13:41 6.233     13.4 4.2         00:15:45 6.219 10.84 24.1     349 971 498 3.72 00:17:21       -0.3 1.9         00:18:57       -0.4 1.9         00:20:33 6.194     -1.2 1.9         00:22:29 6.184 10.86 24     343 962 486 3.16  193 Time (mins) pH Cond. (mS/cm) Temp (°C) ζ (mV) ζ SD (mV) Mg (mg/L) N (mg/L) P (mg/L) SSR 00:24:05       0.9 4.2         00:25:40       3 4.1         00:27:16 6.171     -4.5 4.2         00:29:24 6.168 10.82 24     356 962 495 3.09 00:31:00       -4.8 1.9         00:32:36       -7.6 1.9         00:34:12 6.161     -6.7 1.9         00:36:15 6.156 10.84 24     341 966 486 2.82 00:37:50       -0.5 4.2         00:39:26       -9.1 4.2         00:41:01 6.15     2.2 4.2         Experiment 210 00:00:00 6.409 10.88 22     366 978 519 8.32 00:03:59 6.389 10.93 22.1     366 970 515 7.59 00:05:35       5.3 7.2         00:07:12       -1.3 1.9         00:08:48 6.368     1.6 1.9         00:10:26 6.352 10.84 22     360 952 496 6.31 00:12:02       4.7 1.9         00:13:39       -0.6 1.9         00:15:15 6.308     0.3 1.9         00:17:06 6.292 10.8 22     353 959 488 5.01 00:18:42       5.5 1.9         00:20:18       1.1 1.9         00:21:54 6.27     2.6 1.9         00:23:55 6.265 10.81 21.9     341 961 493   00:25:31       3.2 1.9         00:27:07       0.2 1.9         00:28:43 6.25     -2.4 1.9         00:30:41 6.245 10.78 21.8     344 954 476 4.07 00:32:17       2.2 5.5         00:33:54       -6.8 1.9         00:35:30 6.239     -0.8 1.9         00:37:27 6.237 10.76 21.7     350 959 475 4.07 00:39:03       -1.8 1.9         00:40:38       -3.7 1.9         00:42:14 6.231     -2.6 1.9         Experiment 312 00:00:00 7.769 4.7 21.4     37 441 43 6.03  194 Time (mins) pH Cond. (mS/cm) Temp (°C) ζ (mV) ζ SD (mV) Mg (mg/L) N (mg/L) P (mg/L) SSR 00:02:54 7.714 4.79 21.4     37 435 43 5.01 00:04:30       -0.5 9.6         00:06:06       -1.6 12         00:07:42 7.682     1.6 1.7         00:09:35 7.676 4.72 21.5     36 431 42 4.27 00:11:11       5.7 1.7         00:12:47       4.3 1.7         00:14:23 7.653     0.6 1.7         00:16:30 7.639 4.73 21.5     35 419 38 3.39 00:18:06       5 1.7         00:19:42       1.1 7.5         00:21:18 7.61     -0.4 1.7         00:23:23 7.597 4.72 21.5     32 434 37 2.88 00:24:59       1.8 1.7         00:26:35       -2 11.2         00:28:11 7.57     -5.3 1.7         00:29:56 7.56 4.71 21.5     32 430 35 2.45 00:31:32       -2.8 1.7         00:33:09       -4.9 1.7         00:34:45 7.537     -6.6 1.7         00:36:38 7.532 4.69 21.4     31 433 35 2.24 00:38:14       -4.6 1.7         00:39:50       -10.6 11.6         00:41:26 7.516     -10.2 1.7         Experiment 313 00:00:00 7.787 4.73 21.9     38 441 43 6.31 00:02:44 7.718 4.74 21.8     37 437 43 5.01 00:04:20       2.9 6.4         00:05:57       2.4 13.8         00:07:33 7.689     3.2 12.7         00:08:14 7.683 4.73 21.8     37 438 42 4.47 00:10:10       -10.8 1.7         00:12:06       3.6 7.2         00:14:02 7.67     3.4 8.4         00:16:19 7.658 4.72 21.8     35 434 43 4.07 00:17:55       5 1.7         00:19:31       4.3 1.7         00:21:07 7.633     1.8 1.7         00:23:06 7.62 4.71 21.8     34 430 40 3.39  195 Time (mins) pH Cond. (mS/cm) Temp (°C) ζ (mV) ζ SD (mV) Mg (mg/L) N (mg/L) P (mg/L) SSR 00:24:42       5.4 1.7         00:26:18       1.1 1.7         00:27:54 7.592     -1.4 7.8         00:30:08 7.579 4.71 21.8     32 442 37 2.75 00:31:44       0.3 1.7         00:33:20       3.3 1.7         00:34:56 7.561     -3 7.5         00:37:19 7.55 4.71 21.7     31 442 37 2.51 00:38:56       -4.1 1.7         00:40:32       -7.6 1.7         00:42:09 7.534     -9.7 1.7         Experiment 315 00:00:00 7.779 4.68 20.7     37 458 44 6.46 00:02:36 7.719 4.72 20.6     36 462 43 5.37 00:04:12       2.6 10.3         00:05:49       -0.6 14.7         00:07:25 7.693     0.5 16.8         00:09:46 7.688 4.73 20.6     36 460 43 4.90 00:11:22       6.7 1.7         00:12:57       4 11.1         00:14:33 7.672     5 7.1         00:16:19 7.665 4.71 20.5     34 450 41 4.17 00:17:55       7.1 1.7         00:19:30       6.6 1.7         00:21:06 7.637     5.3 1.7         00:23:19 7.623 4.71 20.5     33 451 39 3.47 00:24:55       6.3 1.7         00:26:31       5.2 1.7         00:28:07 7.596     4.2 1.7         00:30:25 7.583 4.71 20.4     31 458 37 2.88 00:32:01       6.3 1.7         00:33:38       0.6 1.7         00:35:14 7.562     1.5 1.7         00:37:09 7.554 4.71 20.4     30 454 36 2.51 00:38:45       5 1.7         00:40:21       2 1.7         00:41:57 7.536     0.1 1.7         Experiment 419 00:00:00 6.318 9.21 21     560 401 507 3.98  196 Time (mins) pH Cond. (mS/cm) Temp (°C) ζ (mV) ζ SD (mV) Mg (mg/L) N (mg/L) P (mg/L) SSR 00:03:29 6.313 9.32 21     558 396 498 3.47 00:05:05       5.3 1.8         00:06:41       3.6 1.8         00:08:17 6.315     1.6 1.8         00:09:50 6.301 9.3 20.9     548 394 475 3.09 00:11:26       2.9 1.8         00:13:03       6.7 1.8         00:14:39 6.254     5.4 1.8         00:16:25 6.237 9.23 20.9     543 385 470 2.40 00:18:01       3.2 1.7         00:19:36       0.2 1.8         00:21:12 6.213     11.1 1.8         00:22:51 6.208 9.22 20.9     541 384 460 2.09 00:24:27       4.5 1.7         00:26:02       4.6 6.9         00:27:38 6.196     3.8 1.8         00:29:46 6.19 9.19 20.8     536   470   00:31:22       0.2 1.7         00:32:57       3.2 1.8         00:34:33 6.184     0.7 3.9         00:36:26 6.182 9.23 20.8     539   479   00:38:02       1.7 1.7         00:39:38       0.6 1.8         00:41:14 6.175     1.3 1.8         Experiment 421 00:00:00 6.266 9.48 23.1     568 416 524 3.31 00:02:47 6.267 9.47 23     569 411 518 3.02 00:04:23       -2 7         00:05:59       -2.5 4.2         00:07:35 6.301     -3 1.8         00:09:27 6.31 9.48 22.9     556 409 515 3.39 00:11:03       2.6 9         00:12:39       -1.3 1.8         00:14:15 6.326     -3.5 1.8         00:16:16 6.329 9.46 22.9     578 410 517 3.72 00:17:52       5.3 1.7         00:19:28       0.4 1.8         00:21:04 6.337     -2.2 1.8         00:23:33 6.337 9.47 22.9     580 410 520 3.89  197 Time (mins) pH Cond. (mS/cm) Temp (°C) ζ (mV) ζ SD (mV) Mg (mg/L) N (mg/L) P (mg/L) SSR 00:25:09       1.7 1.7         00:26:44       -1.1 1.8         00:28:20 6.337     -2 1.8         00:30:15 6.337 9.47 22.9     603 423 534 4.17 00:31:51       2.4 1.7         00:33:28       -0.5 1.8         00:35:04 6.332     -1.6 1.8         00:37:03 6.329 9.45 22.9     591 414 515 3.89 00:38:39       1.4 1.7         00:40:15       -0.9 7.2         00:41:51 6.323     -2.9 1.8         Experiment 422 00:00:00 6.604 9.51 23.5     592 439 520 10.72 00:02:03 6.598 9.5 23.3     597   515   00:03:39       4.7 1.7         00:05:14       15.3 1.8         00:06:50 6.5     10 1.7         00:08:18 6.463 9.38 23.3     575   486   00:09:54       13.8 6.5         00:11:30       14.3 7.7         00:13:06 6.396     7.9 3.5         00:15:02 6.379 9.32 23.3     539 412 453 3.80 00:16:38       5.2 1.7         00:18:14       0.4 1.7         00:19:50 6.359     3.5 8.3         00:21:35 6.352 9.31 23.2     550   444   00:23:11       1.6 1.7         00:24:46       1.8 1.7         00:26:22 6.343     0.3 1.7         00:28:21 6.339 9.33 23.2     544 416 444 3.31 00:29:57       4.1 1.7         00:31:33       -1.5 5.7         00:33:09 6.33     1.7 1.7         00:35:17 6.328 9.3 23.2     543 415 442 3.16 00:36:53       2.8 1.7         00:38:29       -0.7 1.7         00:40:05 6.321     -1.3 1.8         Experiment 423 00:00:00 6.609 9.47 23.5     592 451 518 10.96  198 Time (mins) pH Cond. (mS/cm) Temp (°C) ζ (mV) ζ SD (mV) Mg (mg/L) N (mg/L) P (mg/L) SSR 00:01:51 6.602 9.48 23.4     586 459 503 10.00 00:03:27       6.3 7.8         00:05:04       13.1 1.8         00:06:40 6.491     5.1 1.8         00:08:43 6.434 9.33 23.4     562 431 465 5.01 00:10:19       6.1 4.4         00:11:54       6.3 1.8         00:13:30 6.375     10 9.2         00:15:33 6.364 9.32 23.3     542 422 455 3.72 00:17:09       4.4 7         00:18:45       5.1 1.7         00:20:21 6.346     0.4 7.5         00:22:27 6.342 9.34 23.3     533 420 444 3.31 00:24:03       2.9 1.7         00:25:38       2.7 1.7         00:27:14 6.332     3.9 1.7         00:29:16 6.33 9.3 23.3     544   451   00:30:52       3.1 1.7         00:32:28       0.1 1.7         00:34:04 6.324     -2.3 1.7         00:36:19 6.322 9.29 23.3     540 417 444 3.09 00:37:55       0.3 1.7         00:39:31       -1.8 1.7         00:41:07 6.317     -4 1.7         Experiment 528 00:00:00 8.005 3.93 23.4     40 340 43 9.33 00:02:18 7.95 3.93 23.3     40 342 43 8.13 00:03:54       -13.6 25.8         00:05:30       -30.9 24.7         00:07:06 7.908     -17.4 20.6         00:08:49 7.889 3.93 23.3     37 338 40 6.17 00:10:25       9.8 1.6         00:12:00       8.5 1.6         00:13:36 7.829     6.9 1.6         00:15:32 7.805 3.92 23.3     36 332 38 4.68 00:17:08       5.7 1.6         00:18:44       4.8 1.6         00:20:20 7.761     1.1 1.6         00:22:27 7.745 3.9 23.2     30 332 30 2.82  199 Time (mins) pH Cond. (mS/cm) Temp (°C) ζ (mV) ζ SD (mV) Mg (mg/L) N (mg/L) P (mg/L) SSR 00:24:03       3.5 1.6         00:25:39       -2 5.9         00:27:15 7.722     -1.5 1.6         00:29:09 7.712 3.9 23.2     29 334 29 2.45 00:30:45       0.9 1.6         00:32:21       -1.8 18.5         00:33:57 7.699     -4.2 1.6         00:36:52 7.69 3.9 23.1     28 333 28 2.19 00:38:28       -4.7 4.3         00:40:05       -8.9 8.2         00:41:41 7.682     -13.7 1.6         Experiment 529 00:00:00 8.009 3.96 23.7     40 345 44 9.33 00:02:00 7.959 3.96 23.5     38 329 38 6.92 00:03:36       -1.1 1.6         00:05:12       -2.9 1.6         00:06:48 7.913     1.4 1.6         00:08:34 7.908 3.94 23.6     39 339 42 6.92 00:10:10       8.3 5.2         00:11:46       3.3 5.7         00:13:22 7.882     4.8 14.4         00:15:28 7.869 3.89 23.6     37 342 40 5.89 00:17:04       9.8 7.7         00:18:39       6.7 1.6         00:20:15 7.837     5.7 1.6         00:22:17 7.821 3.94 23.6     34 337 36 4.37 00:23:53       6.8 1.6         00:25:29       3 1.6         00:27:05 7.793     -7.3 17.8         00:28:48 7.785 3.91 23.5     33 337 34 3.72 00:30:24       -1.4 9.6         00:32:00       -3.6 13.8         00:33:36 7.765     0 10.2         00:35:24 7.759 3.91 23.4     32 331 32 3.16 00:37:01       -8.4 16.7         00:38:37       -4.3 18.9         00:40:14 7.742     -12.6 9.6         Experiment 530 00:00:00 8.002 3.96 23.7     39 346 43 8.91  200 Time (mins) pH Cond. (mS/cm) Temp (°C) ζ (mV) ζ SD (mV) Mg (mg/L) N (mg/L) P (mg/L) SSR 00:01:44 7.957 3.95 23.5     39 341 43 7.94 00:03:20       0.8 1.6         00:04:57       3.5 1.6         00:06:33 7.915     3.1 1.6         00:08:24 7.913 3.93 23.5     38 340 42 6.92 00:10:00       0 1.6         00:11:36       0.8 1.6         00:13:12 7.91     2.4 11.4         00:15:30 7.91 3.95 23.4     38 342 42 6.92 00:17:06       3.1 1.6         00:18:42       3.2 1.6         00:20:18 7.909     1.4 1.6         00:22:22 7.909 3.94 23.4     38 342 42 6.92 00:23:58       -3.9 1.6         00:25:34       0.1 8.6         00:27:10 7.903     1.1 1.6         00:29:00 7.9 3.94 23.3     37 341 41 6.46 00:30:36       3.1 1.6         00:32:12       -2.5 1.6         00:33:48 7.896     -1 1.6             201 Induction Time - Zeta Potential Data  Experiment# Induction time (s) 102 98 103 82 104 112 105 323 106 327 124 38 125 94 126 90 127 140 209 110 210 132 312 191 313 275 315 192 419 110 421 358 422 33 423 35 528 123 529 273 530 198    Note: Laser method utilized     202 Appendix F FBR Control Data Sheet 1/4 Day Concentrated Mg (mg/L) Feed Lab Results (mg/L) Filtered Effluent Lab Results (mg/L) PO4-P NH4-N Mg PO4-P NH4-N Mg 1.0 1643 92 813 0 11 698 38 1.5 1643 92 813 0 9 672 34 2.0 1643 92 813 0 5 715 31 2.5 1643 92 813 0 7 668 32 3.0 1643 92 813 0 5 676 30 3.5 1643 92 813 0 9 704 33 4.0 1643 92 813 0 7 695 35 4.5 1643 92 813 0 10 664 36 5.0 1643 92 813 0 - - - 5.5 1643 92 813 0 8 702 33 9.0 1643 92 813 0 7 686 33 9.5 1643 92 813 0 7 721 34 10.0 1643 92 813 0 6 664 34 10.5 1643 92 813 0 5 682 33 11.0 1643 92 813 0 6 703 35 11.5 1643 92 813 0 5 672 34 12.0 1643 92 813 0 5 678 35 12.5 1643 92 813 0 5 696 34      203        Sheet 2/4  Day Unfiltered Effluent Lab Results (mg/L) pH Temp (°C)    Effluent Conductivity (mS) Mg Flow (mL/min) PO4-P NH4-N Mg 1.0 12 704 38 8.02 20.1 7.63 16 1.5 9 703 35 7.99 25.3 7.49 16 2.0 5 685 31 8.1 17.7 7.59 16 2.5 7 716 32 8.12 26.5 7.54 16 3.0 5 707 29 8.09 18.1 7.56 16 3.5 9     7.96 28.7 7.52 16 4.0 8 681 34 7.93 20.2 7.56 16 4.5 10 683 35 7.9 29.8 7.57 16.5 5.0 - - - 8.01 20.9 7.62 16 5.5 9 692 34 8.04 30 7.57 16 9.0 8 712 34 8.12 20.2 7.64 15.5 9.5 7 689 33 8.15 26.1 7.57 16 10.0 6 669 33 8.16 17.5 7.48 16 10.5 5 675 33 8.23 20.5 7.52 16 11.0 6 674 35 8.08 16.9 7.6 16 11.5 5 721 35 8.21 19.5 7.55 16 12.0 5 687 36 8.13 15.2 7.57 16 12.5 5 698 35 8.15 20.1 7.61 16      204   Sheet 3/4 Day Feed Flow (mL/min) Recycle Flow (mL/min) Total Flow (Mg  + Feed  + recycle) (mL/min) Filtered PO4-P In-Reactor (mg/L) From Feed From Recycle Total 1.0 246 1774 2036 11 10 21 1.5 246 1754 2016 11 8 19 2.0 256 1794 2066 11 4 16 2.5 247 1753 2016 11 6 18 3.0 252 1778 2046 11 4 16 3.5 252 1748 2016 12 7 19 4.0 260 1790 2066 12 6 17 4.5 248 1752 2016.5 11 8 20 5.0 252 1788 2056 - - - 5.5 253 1747 2016 12 7 18 9.0 251 1769 2035.5 11 6 18 9.5 250 1760 2026 11 6 18 10.0 250 1760 2026 11 5 17 10.5 250 1790 2056 11 5 16 11.0 250 1790 2056 11 5 16 11.5 253 1767 2036 11 4 16 12.0 253 1767 2036 11 4 16 12.5 247 1793 2056 11 4 15      205  Sheet 4/4 Day Filtered NH4-N In-Reactor (mg/L) Filtered Mg In-Reactor (mg/L) From Feed From Recycle Total Concentrated Mg From Recycle Total 1.0 98 608 706 13 33 46 1.5 99 584 684 13 30 43 2.0 101 621 722 13 27 40 2.5 100 580 680 13 28 41 3.0 100 587 687 13 26 39 3.5 102 610 712 13 29 42 4.0 102 602 704 13 30 43 4.5 100 577 677 13 31 44 5.0 - - - - - - 5.5 102 608 710 13 29 42 9.0 100 596 697 13 29 41 9.5 100 626 727 13 30 43 10.0 100 577 677 13 30 43 10.5 99 593 692 13 29 42 11.0 99 612 711 13 31 44 11.5 101 583 684 13 30 43 12.0 101 589 690 13 30 43 12.5 98 607 705 13 30 42      206 Appendix G FBR Crystal Addition Data  Sheet 1/4 Day Concentrated Mg (mg/L) Feed Lab Results (mg/L) Filtered Effluent Lab Results (mg/L) PO4-P NH4-N Mg PO4-P NH4-N Mg 1.0 1643 92 813 0 - - - 1.5 1643 92 813 0 17 698 12 2.0 1643 92 813 0 6 670 24 2.5 1643 92 813 0 - - - 3.0 1643 92 813 0 5 683 28 3.5 1643 92 813 0 8 692 30 4.0 1643 92 813 0 5 673 29 4.5 1643 92 813 0 - - - 5.0 1643 92 813 0 - - - 5.5 1643 92 813 0 - - - 9.0 1643 92 813 0 6 658 30 9.5 1643 92 813 0 6 678 28 10.0 1643 92 813 0 5 680 27 10.5 1643 92 813 0 5 680 28 11.0 1643 92 813 0 4 692 27 11.5 1643 92 813 0 5 665 27 12.0 1643 92 813 0 4 673 28 12.5 1643 92 813 0 - - -     207        Sheet 2/4 Day Unfiltered Effluent Lab Results (mg/L) pH Temp (°C) Effluent Conductivity (mS/cm) Mg Flow (mL/min) PO4-P NH4-N Mg 1.0 - - - - - - 15.5 1.5 20 690 14 8.26 25.7 7.4 15.5 2.0 6 691 24 8.13 18.7 7.53 15.5 2.5 - - - 8.27 27.4 7.43 15.5 3.0 5 681 29 8.16 18.5 7.56 15 3.5 11 669 33 8.14 28.9 7.48 15.5 4.0 6 694 32 8.11 19.2 7.52 15 4.5 - - - 8.06 20.9 7.52 16 5.0 - - - 8.14 21.6 7.55 15 5.5 - - - 8.13 29.6 7.48 15 9.0 10 666 34 8.16 21.6 7.54 14.5 9.5 14 680 36 8.29 29.2 7.53 15 10.0 9 673 31 8.23 17.4 7.53 15 10.5 11 683 32 8.3 20.6 7.51 15 11.0 8 674 31 8.21 17.1 7.53 15 11.5 13 675 33 8.28 19.6 7.5 15 12.0 11 688 33 8.2 15.4 7.56 15 12.5 - - - 8.15 20.7 7.61 15     208   Sheet 3/4 Day Feed Flow (mL/min) Recycle Flow (mL/min) Total Flow (Mg  + Feed  + recycle) (mL/min) Filtered PO4-P In-Reactor (mg/L) From Feed From Recycle Total 1.0 259 1711 1985.5 - - - 1.5 253 1747 2015.5 12 15 26 2.0 258 1742 2015.5 12 5 17 2.5 246 1704 1965.5 - - - 3.0 246 1954 2215 10 4 15 3.5 242 1758 2015.5 11 7 18 4.0 248 1832 2095 11 5 16 4.5 249 1701 1966 - - - 5.0 248 1712 1975 - - - 5.5 247 1703 1965 - - - 9.0 244 1716 1974.5 11 5 17 9.5 251 1649 1915 12 5 18 10.0 251 1649 1915 12 5 17 10.5 250 1720 1985 12 5 16 11.0 250 1720 1985 12 4 15 11.5 250 1700 1965 12 4 16 12.0 250 1700 1965 12 3 15 12.5 250 1700 1965 - - -       209        Sheet 4/4 Day Filtered NH4-N In-Reactor (mg/L) Filtered Mg In-Reactor (mg/L) From Feed From Recycle Total Concentrated Mg From Recycle Total 1.0 - - - - - - 1.5 102 605 707 13 10 23 2.0 104 579 683 13 21 33 2.5 - - - - - - 3.0 90 603 693 11 25 36 3.5 98 604 701 13 26 39 4.0 96 589 685 12 25 37 4.5 - - - - - - 5.0 - - - - - - 5.5 - - - - - - 9.0 100 572 673 12 26 38 9.5 107 583 690 13 24 37 10.0 107 586 692 13 23 36 10.5 102 589 692 12 24 37 11.0 102 600 702 12 23 36 11.5 103 575 679 13 23 36 12.0 103 583 686 13 24 37 12.5 - - - - - -       210 Appendix H Struvite Mass Removed from Control FBR  Day Harvest 2.0 mm Harvest 1.0 mm Harvest 0.5 mm Harvest Fines pH Probe Sample Seed Hopper Sample Reactor Cleaning Clarifier Fines 1 9.99 90.98 152.17 33.01 - - 97.07 - 2 29.42 65.04 12.03 77.50 - - - - 3 11.94 74.52 5.31 43.15 4.08 1.16 6.38 - 4 14.23 90.39 41.64 46.27 0.93 0.05 7.42 - 5 - - - - 1.07 0.08 - - 9 - 85.81 11.23 14.80 7.89 0.45 - - 10 - - - - 10.95 0.47 - - 11 - - - - 6.47 1.24 - - 12 2.76 107.34 291.97 546.35 4.36 1.39 12.93 70.59 Total 68.34 514.08 514.35 761.08 35.75 4.84 123.8 70.59    211 Appendix I Struvite Mass Removed from Crystal Addition FBR    Day Harvest 2.0 mm Harvest 1.0 mm Harvest 0.5 mm Harvest Fines pH Probe Sample Seed Hopper Sample Reactor Cleaning Clarifier Fines pH probe Harvest Harvest Valve Sample 1 - - - - - - - - - - 2 - - - - 0.77 1.26 - - - - 3 32.25 54.60 16.25 45.43 4.12 - 16.07 - 39.37 - 4 62.96 50.98 - 38.63 6.68 1.13 8.86 - - - 5 - - - - 28.5 0.14 - - - 30.18 9 - 17.23 27.83 12.06 6.98 0.13 7.20 - - - 10 - - - - 4.51 0.22 - - - - 11 - - - - 7.72 1.82 - - - - 12 3.51 27.31 289.78 693.75 5.32 1.39 12.64 263 - - Total 98.72 150.12 333.86 789.87 64.60 6.09 44.77 263.00 39.37 30.18        212 Appendix J FBR Velocities Based on Configuration  Reactor Description Harvest Zone Injector Port Between Zones Injector Area Reduction with 1.39 m/s (%) Dia (m) Velocity (m/s) Flow (m3/s) Dia (m) Velocity (m/s) Expansion Ratio Injector Port Reynolds Number  1.5” Bench Scale 0.0400 0.0667 8.378E-05 0.0095 1.1819 4.21 11,183 15 3” Pilot Scale 0.0760 0.0667 3.024E-04 0.0191 1.0555 3.98 20,080 24 6” Pilot Scale 0.1524 0.0667 1.216E-03 0.0381 1.0667 4.00 40,478 23        213 Appendix K Stationary Crystal Growth Experimental Data  Exp # pH Temp (°C) Conductivity (mS/cm) Flow (mL/min) NH4-N (mg/L) P04-P (mg/L) Mg (mg/L) SI SSR Stop Time (min) 3 7.87 25.8 7.63 42 736 88 93 1.53 33.88 7 7 7.89 23.6 6.24 43 778 14 47 0.48 3.02 55 8 7.9 23.7 6.49 43 773 15 46 0.51 3.24 60 9 7.87 23.4 6.46 43 755 15 48 0.49 3.09 50 10 7.6 23.6 7.56 22 862 88 86 1.14 13.80 10 11 7.57 23.3 7.64 22 841 88 87 1.11 12.88 10 12 7.57 23.6 7.53 22 832 86 87 1.09 12.30 10 13 8.3 23.2 7.02 22 720 18 55 0.99 9.77 58 14 8.24 23.9 7.1 22 698 18 55 0.91 8.13 90 15 7.95 23.2 7.12 22 705 18 55 0.63 4.27 60 16 7.97 23.3 7.1 22 672 18 55 0.63 4.27 80 17 8.47 24 6.97 42 676 18 55 1.11 13.18 30 18 7.08 23.6 7.78 22 693 88 94 0.48 3.02 15  214 Exp # pH Temp (°C) Conductivity (mS/cm) Flow (mL/min) NH4-N (mg/L) P04-P (mg/L) Mg (mg/L) SI SSR Stop Time (min) 19 7.04 24.1 7.68 22 695 89 94 0.43 2.69 20 20 7.02 24.6 7.62 42 692 90 95 0.41 2.57 15 21 8.54 24.3 7.04 22 675 16 54 1.11 12.88 60 22 8.53 23.7 6.93 22 675 16 53 1.08 12.88 31 23 7.36 24.5 9.07 43 870 79 98 0.76 6.61 10 24 7.34 23.9 9.13 22 736 79 97 0.73 5.37 10 25 8.37 25.2 7.6 42 746 18 55 0.96 10.96 30 26 7.28 25.1 9.18 22 735 88 98 0.69 4.90 30 27 7.29 25.5 9.13 42 873 88 97 0.69 5.75 25 28 7.69 25 9.04 42 721 86 98 1.13 13.49 10 29 8.26 23.9 6.53 43 630 21 63 1.01 10.23 60 30 8.21 24.3 6.49 43 630 21 60 0.94 8.71 60 31 7.38 24.4 7.71 43 728 107 115 0.99 9.77 7     215  Appendix L Stationary Growth Rate Data and Calculations      Growth of Crystals from Side View Exp # Velocity (m/s) Stop Time (min) SSR # of Crystals Measured Area Before Exp (μm2) Ave Circularity Before Area After Exp (μm2) Ave Circularity After Radius Growth (μm) Growth Rate (μm/min) 3 1.39 7 33.88 7 385964 0.41 994322 0.27 80.16 11.45 7 1.43 55 3.02 5 172924 0.59 207116 0.49 9.91 0.18 8 1.43 60 3.24 8 254689 0.53 280988 0.53 5.07 0.08 9 1.43 50 3.09 8 238593 0.55 269352 0.51 6.09 0.12 10 0.73 10 13.80 13 248247 0.57 388585 0.53 19.58 1.96 11 0.73 10 12.88 6 173251 0.56 239456 0.5 16.84 1.68 12 0.73 10 12.30 13 420806 0.58 580990 0.52 17.77 1.78 13 0.73 58 9.77 2 52020 0.59 68350 0.45 13.31 0.23 14 0.73 90 8.13 16 489048 0.56 799593 0.43 27.49 0.31 15 0.73 60 4.27 13 456823 0.5 517953 0.48 6.85 0.11 16 0.73 80 4.27 11 341580 0.52 417814 0.5 10.54 0.13 17 1.39 30 13.18 9 363684 0.59 493945 0.47 18.76 0.63  216     Growth of Crystals from Side View Exp # Velocity (m/s) Stop Time (min) SSR # of Crystals Measured Area Before Exp (μm2) Ave Circularity Before Area After Exp (μm2) Ave Circularity After Radius Growth (μm) Growth Rate (μm/min) 18 0.73 15 3.02 10 322553 0.56 323230 0.55 0.11 0.01 19 0.73 20 2.69 6 260228 0.54 247826 0.53 -2.83 -0.14 20 1.39 15 2.57 12 265444 0.58 259127 0.58 -1.00 -0.07 21 0.73 60 12.88 11 310636 0.53 480416 0.44 23.10 0.38 22 0.73 31 12.88 12 337409 0.55 413320 0.53 10.10 0.33 23 1.43 10 6.61 15 377852 0.6 451976 0.56 8.39 0.84 24 0.73 10 5.37 12 382419 0.56 427542 0.53 5.78 0.58 25 1.39 30 10.96 17 687190 0.53 902782 0.42 16.58 0.55 26 0.73 30 4.90 8 280900 0.53 365046 0.49 14.80 0.49 27 1.39 25 5.75 23 750098 0.56 975946 0.5 14.33 0.57 28 1.39 10 13.49 18 809428 0.54 1426282 0.41 39.17 3.92 29 1.43 60 10.23 6 330718 0.52 616568 0.38 48.40 0.81 30 1.43 60 8.71 6 158669 0.57 386951 0.4 51.53 0.86 31 1.39 7 9.77 9 355897 0.51 474313 0.48 17.33 2.48    217   Growth of Crystals from Flow Direction     Exp # # of Crystals Measured Area Before Exp (μm2) Ave Circularity Before Area After Exp (μm2) Ave Circularity After Radius Growth (μm) Growth Rate (μm/min) Average Growth Rate b/t Directions (μm/min) Average Circularity b/t Directions Before Growth Average Circularity b/t Directions After Growth Mass Flux  (g/m2 min) 3 - - - - - - - - - - 20.27 7 6 175347 0.55 195148 0.49 5.30 0.10 0.14 0.57 0.49 0.24 8 7 338592 0.48 346463 0.48 1.43 0.02 0.05 0.51 0.51 0.10 9 10 321963 0.56 367429 0.5 6.91 0.14 0.13 0.56 0.51 0.23 10 18 486475 0.56 745979 0.48 22.10 2.21 2.08 0.57 0.51 3.69 11 9 371338 0.55 521092 0.48 21.16 2.12 1.90 0.56 0.49 3.36 12 17 471196 0.54 686595 0.46 19.45 1.95 1.86 0.56 0.49 3.29 13 7 159821 0.57 242086 0.57 19.67 0.34 0.28 0.58 0.51 0.50 14 10 459882 0.56 719645 0.42 30.36 0.34 0.32 0.56 0.43 0.57 15 10 343546 0.56 400695 0.52 8.36 0.14 0.13 0.53 0.50 0.22 16 14 385164 0.54 480559 0.52 10.95 0.14 0.13 0.53 0.51 0.24 17 10 258888 0.55 346960 0.5 14.31 0.48 0.55 0.57 0.49 0.98 18 11 485967 0.54 481788 0.53 -0.51 -0.03 -0.01 0.55 0.54 -0.02  218  Growth of Crystals from Flow Direction     Exp # # of Crystals Measured Area Before Exp (μm2) Ave Circularity Before Area After Exp (μm2) Ave Circularity After Radius Growth (μm) Growth Rate (μm/min) Average Growth Rate b/t Directions (μm/min) Average Circularity b/t Directions Before Growth Average Circularity b/t Directions After Growth Mass Flux  (g/m2 min) 19 7 193628 0.56 194572 0.57 0.23 0.01 -0.07 0.55 0.55 -0.12 20 10 292331 0.56 290765 0.56 -0.26 -0.02 -0.04 0.57 0.57 -0.07 21 12 487097 0.5 719755 0.41 24.51 0.41 0.40 0.52 0.43 0.70 22 13 396062 0.51 482756 0.44 10.24 0.33 0.33 0.53 0.49 0.58 23 - - - - - - - - - - 1.49 24 8 401989 0.53 432705 0.49 4.74 0.47 0.53 0.55 0.51 0.93 25 19 678189 0.54 981349 0.43 21.63 0.72 0.64 0.54 0.43 1.13 26 17 1067359 0.53 1293165 0.47 14.24 0.47 0.48 0.53 0.48 0.86 27 23 732775 0.53 991180 0.49 16.42 0.66 0.61 0.55 0.50 1.09 28 14 497359 0.54 1058114 0.39 48.77 4.88 4.40 0.54 0.40 7.78 29 4 138784 0.59 273605 0.43 42.47 0.71 0.76 0.56 0.41 1.34 30 7 189745 0.55 473684 0.37 53.88 0.90 0.88 0.56 0.39 1.55 31 16 540008 0.56 713458 0.5 15.49 2.21 2.34 0.54 0.49 4.15  219 Appendix M Growth Coefficient Graphs   Graph 1:  Full-mix Conditions   Graph 2: No-mix Conditions    y = 0.0678xR² = 0.6608y = 0.033xR² = 0.793300.10.20.30.40.50.60.70.80.910 5 10 15Radial Growth Rate (um/min)SSR-1FMHVFMLVy = 0.3359xR² = 0.9671y = 0.1594xR² = 0.9789024681012140 10 20 30 40Radial Growth Rate (um/min)SSR-1NMHVNMLV 220 Appendix N Published Struvite Growth Rates and Coefficients   Reference SSR/pH Growth Rate kG/kd/kr Reactor Type Method Notes (M. I. H. Bhuiyan et al., 2008) uk/8.07 0.06-0.18 μm/min kd = 1.11x10-8 m/s; kr = 7.99x10-5 m/s FBR Deposition and mass calc, with drop in P conc., seeded (Mehta & Batstone, 2013) uk/8.0, 8.5, 9.0 0.06-0.30 μm/min/unit of σ kr= 0.09 ±0.04 mmol/m2*min = 0.022 g of struvite/m2*min 1L stirred reactor Deposition calc with σ, digester supernatant, seeded (Harrison et al., 2011) uk/8.5 10-24 μm/min kG 1.5x10-5 to 10-4 μm/min = 2.5x10-13 m/s Stirred reactor Volume equivalent size dist b/t time frames, and SSR, seeded (Ohlinger et al., 1999) uk 2-6 g/m2*d na WWTP Coupon – scale accumulation rate (Ariyanto et al., 2014) 2 to 10/ 8.0, 8.5, 9.0 various kG 1.1, 2.3, 3.3 x10-9 m/s 1L stirred reactor Change in size, conc.,seeded (Galbraith et al., 2014) pH 7.46 at SI . 0.37 and pH 7.62 at SI . 0.54 uk kG = 2.1x10-7 m/s 1L stirred reactor Size distr. and conc, seeded  σ = relative SSR kd = mass transfer coefficient (diffusion) kr = surface-reaction coefficient kG = overall growth coefficient SSR = 10SI   221 Appendix O Stationary Crystal Growth Morphology Data Exp # Code SSR Velocity (m/s) Stop Time (min) Crystal Growth Morphology/Surface Development Protrusion Habit Protrusion Frequency Protrusion Length (um) Roughing/ Smoothing 7 FMHV 3.02 1.43 55 boxy coffin/ short rod/pyramidal none - - R 9 FMHV 3.09 1.43 50 boxy coffin/ short rod/pyramidal none - - R 8 FMHV 3.24 1.43 60 bumpy/sm boxy coffin none - - R 30 FMHV 8.71 1.43 60 pyramidal/surface growth/some protrusions rod or elongated prismatic or pyramidal medium 40-140 S 29 FMHV 10.23 1.43 60 pyramidal/surface growth/some protrusions rod or elongated prismatic or pyramidal medium 40-120 S 25 FMHV 10.96 1.39 30 pyramidal/surface growth elongated coffins or rods minor 100 S 17 FMHV 13.18 1.39 30 pyramidal/elongated coffin surface growth elongated coffins or rods minor 15-40 S 15 FMLV 4.27 0.73 60 roughening on corners and edge growth/boxy coffin/ some pyramidal formations none - - R on edges 16 FMLV 4.27 0.73 80 boxy coffin/pyramidal/ surface growth none - - R  222 Exp # Code SSR Velocity (m/s) Stop Time (min) Crystal Growth Morphology/Surface Development Protrusion Habit Protrusion Frequency Protrusion Length (um) Roughing/ Smoothing 14 FMLV 8.13 0.73 90 surface and protrusion growth out from edges and surfaces/ protrusions also random/ pyramidal growth like pellets rod or elongated prismatic or pyramidal medium 60-100 S 13 FMLV 9.77 0.73 58 surface and protrusion growth out from edges and surfaces/ protrusions also random/ pyramidal growth like pellets rod or elongated prismatic or pyramidal medium 30-100 S 22 FMLV 12.88 0.73 31 surface roughening, small pyramidal, random protrusions rod or elongated prismatic or pyramidal minor 15-50 R 21 FMLV 12.88 0.73 60 surface and protrusion growth out from edges and surfaces/ not many protrusions also random/ pyramidal growth like pellets rod or elongated prismatic or pyramidal minor 30-100 S 20 NMHV 2.57 1.39 15 minor surface roughening and dissolution around edges and defects none - - R  223 Exp # Code SSR Velocity (m/s) Stop Time (min) Crystal Growth Morphology/Surface Development Protrusion Habit Protrusion Frequency Protrusion Length (um) Roughing/ Smoothing           27 NMHV 5.75 1.39 25 surface and growth of pyramidal out from edges and surfaces/ no  protrusions/ pyramidal growth like pellets none - - S mostly but some R on flat surfaces 23 NMHV 6.61 1.43 10 bumpy/sm boxy coffin none - - R 31 NMHV 9.77 1.39 7 some surface roughening/ small pyramidal with rectangular top none - - R,S 28 NMHV 13.49 1.39 10 surface roughening underneath, some pyramidal/ protrusions everywhere but mostly into and perpendicular to flow mostly 3 sided with occasional 4 sided-plate, partially hollow, much wider than FM high 30-100 R,S 3 NMHV 33.88 1.39 7 bumpy underneath where out of flow, large protrusions everywhere mostly 3 sided with occasional 4 sided-plate, most are hollow, much wider than FM extreme 70-120 R  224 Exp # Code SSR Velocity (m/s) Stop Time (min) Crystal Growth Morphology/Surface Development Protrusion Habit Protrusion Frequency Protrusion Length (um) Roughing/ Smoothing 19 NMLV 2.69 0.73 20 minor surface roughening and dissolution around edges and defects na - - R 18 NMLV 3.02 0.73 15 minor surface roughening and dissolution around edges and defects na - - R 26 NMLV 4.90 0.73 30 surface roughening and smoothing, bumpy, edge growth,  pyramidal none - - R,S 24 NMLV 5.37 0.73 10 roughening of surface/ bumpy/sm boxy coffin none - - R 12 NMLV 12.30 0.73 10 mostly surface growth, some roughening, smoothing, pyramidal with rectangular tops, plate, some smaller protrusions plate and triangle 3-4 sided, corner of tops minor 20-40 R,S 11 NMLV 12.88 0.73 10 mostly surface growth, pyramidal with rectangular tops corners of rectangle tops medium 20-40 S 10 NMLV 13.80 0.73 10 mostly surface growth, pyramidal with rectangular tops corners of rectangle tops medium 20-40 S    225 Appendix P Rotational Pellet Growth Experimental Data Exp # Needle Holder # pH Temp (°C) Conductivity  (mS/cm) Stop Time (min) Flow in Tubing (mL/min) Average Fluid Velocity (m/s) Rotational Velocity of Pellet (m/s) Total Velocity (m/s) NH4-N (mg/L) PO4-P (mg/L) Mg (mg/L) 1 1 8.04 22.0 8.02 60 200 0.11 0.02 0.13 761 18 66 2 1 from exp 1 8.20 22.2 7.42 60 200 0.11 0.02 0.13 694 18 65 3 2 8.07 23.0 7.73 60 850 0.45 0.02 0.47 747 21 47 4 3 8.06 23.0 7.81 60 1350 0.71 0.02 0.73 743 21 48 5 1 from exp 2 8.07 21.8 7.80 60 830 0.44 0.02 0.46 730 21 47 6 3 from exp 4 8.10 21.5 7.36 52 2200 1.16 0.02 1.18 632 19 53 7 4 8.22 21.8 7.31 60 860 0.45 0.02 0.47 650 18 53  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0378385/manifest

Comment

Related Items