Thermodynamics of Sodium Aluminosilicate Formation in Aqueous Alkaline Solutions relevant to Closed-cycle Kraft Pulp Mills HYEON PARK

Transcription

1 Thermodynamics of Sodium Aluminosilicate Formation in Aqueous Alkaline Solutions relevant to Closed-cycle Kraft Pulp Mills by HYEON PARK B.S., Hanyang University, Korea, 1989 M.S., Hanyang University, Korea, 1993 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUESTMENTS FOR THE DEGREE OF DOCTOR OF PHTLOSPHY in THE FACULTY OF GRADUATE STUDIES Department of Chemical and Bio-Resource Engineering We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA August, 1999 HyeonPark, 1999

2 in presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of 0)CVY\ Iced? W fifa-jxw-^u^ Krg<h ^jzr-l^j The University of British Columbia Vancouver, Canada Date /Uj.J^, DE-6 (2/88)

3 11 ABSTRACT Accumulation of Al and Si ions in the recovery cycle of a kraft pulp mill may cause sodium aluminosilicate scale formation. This glossy scale forms on process equipment and is very hard to remove. Thus, formation of the scale can create several operational problems in mills moving towards progressive system closure and should be prevented. The purpose of this study is to supply: (a) data on the precipitation conditions of sodium-aluminosilicates in green and white liquors of the recovery cycle; and (b) a model to predict such conditions. The data can either be used directly or to test process models for the design and optimization of progressive system closure strategies. The precipitation conditions of sodium aluminosilicates in synthetic green and white liquors at K (95 C) were determined. In the experiments, the effects of varying the Al/Si ratio and concentrations of OH", C0 3 2", SO4 2 ', and HS" were studied. The structure of the precipitates was identified by X-ray diffraction, thermogravimetry and chemical analysis. The precipitates were found to have the structure of hydroxysodalite dihydrate (Na8(AlSi04)6(OH) 2-2H20) except in a simulated green liquor system with low OH" and high Cf concentrations where sodalite dihydrate (Nag(AlSi04)6Cl2-2H20) was formed. The precipitation conditions in mill green and white liquors at K were also measured. The effects of varying the Al/Si ratio, NaOH, Na2C03, and Na2S concentrations were studied. The precipitates were found to have the structure of hydroxysodalite dihydrate. A thermodynamic model for sodium aluminosilicate formation in aqueous alkaline solutions was developed. Pitzer's method was adopted to calculate the activity of

4 Ill water and the activity coefficients of the other species in solution. The system under consideration contained the ions of Na + 2 2, Al(OH) 4 ', Si0 3 ", OH", C0 3 ", S0 4 2', Cl', HS" dissolved in water and in equilibrium with two possible solid phases (sodalite dihydrate : Na8(AlSi04)6Cl2-2H 2 0 and hydroxysodalite dihydrate : Na 8 (AlSi0 4 ) 6 (OH)2-2H 2 0) at K. The equilibrium constants of sodalite dihydrate and hydroxysodalite dihydrate formation reactions were determined using the thermodynamic properties of the species involved. Property values that were not available in the literature were estimated by group contribution methods. The model calculates the molality of all species at equilibrium including the amount of solid precipitates. The calculations were compared with published data and were found to be in good agreement. Meanwhile, since the system contains the Si03 2 " and Al(OH) 4 " ions, knowledge of the relevant Pitzer's model parameters is required. Osmotic coefficient and water activity data for Na2Si03 and mixed Na2Si03-NaOH aqueous solutions were obtained at K by employing an isopiestic method. The binary Pitzer's parameters, P (0), f3 (1), and C*, for Na2Si03 and the mixing parameters, Q0H-Si02- and N a + O H - s i o i -, were estimated using the osmotic coefficient data. In addition, osmotic coefficient data were obtained for the aqueous solutions of NaOH-NaCl-NaAl(OH) 4. The solutions were prepared by dissolving AICI36H2O in aqueous NaOH solutions. The osmotic coefficients of the solutions were measured by the isopiestic method at K. The osmotic coefficient data were used to evaluate the unknown binary and mixing parameters of Pitzer's model for the aqueous NaOH-NaCl-NaAl(OH) 4 system. The experimental osmotic coefficient data were correlated well with Pitzer's model using the parameters obtained.

5 iv TABLE OF CONTENTS ABSTRACT TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES ACKNOWLEDGEMENTS ii iv viii xi xiv CHAPTER 1. INTRODUCTION Environmental Impact of the Kraft Pulping Industry Closed-cycle Mill Operation Non-process Elements in the Kraft Pulping Process Sources and profile of non-process elements Adverse effects of non-process elements Recent Developments of Closed-cycle Technology Scale Formation in the Recovery Cycle Sodium Aluminosilicate Scale Formation Research Objectives 19 CHAPTER 2. THEORETICAL BACKGROUND Thermodynamic Equilibrium Constant Equilibrium Constant Calculation at Specific Temperature Activity of Water, Osmotic Coefficient, and Activity Coefficient Pitzer's Activity Coefficient Model Isopiestic Method 32 CHAPTER 3. MATERIALS AND METHODS Solubility Experiments using Synthetic Liquors of System A Experimental design Experimental procedure Identification of the precipitates 39

6 V 3.2. Solubility Experiments using Synthetic Liquors of System B Experimental design Experimental procedure and analysis Solubility Experiments using Mill Liquors Experimental design Experimental procedure and analysis Osmotic Coefficient Measurement for Na2Si03 and Na2Si03-NaOH 45 Systems Identification of silicic species by a titration method Apparatus and chemicals Experimental procedure Osmotic Coefficient Measurement for NaOH-NaCl-NaAl(OH) 4 System Apparatus and chemicals Experimental procedure 51 CHAPTER 4. THERMODYNAMIC MODELING OF SODIUM 53 ALUMINOSILICATE FORMATION 4.1. Model Equations Structure of the Thermodynamic Model 55 CHAPTER 5. OSMOTIC COEFFICIENT DATA FOR Na 2 Si0 3, Na 2 Si0 3 -NaOH, 57 AND NaOH-NaCl-NaAl(OH) 4 AQUEOUS SYSTEMS 5.1. Identification of Metasilicic Species by Titration Mole Fraction of Metasilicic Species Osmotic Coefficient Data for Na2Si0 3 Aqueous System Osmotic Coefficient Data for Na 2 Si0 3 -NaOH Aqueous System Osmotic Coefficient Data for NaOH-NaCl-NaAl(OH) 4 Aqueous System 70 CHAPTER 6. DETERMINATION OF PITZER' S PARAMETERS FOR Na 2 Si0 3, 73 Na 2 Si0 3 -NaOH, AND NaOH-NaCl-NaAl(OH) 4 AQUEOUS SYSTEMS 6.1. Pitzer's Parameters for Na2Si0 3 and Na2Si0 3 -NaOH Aqueous Systems 73

7 vi 6.2. Pitzer's Parameters for NaOH-NaCl-NaAl(OH) 4 Aqueous System Reliability of the Pitzer's Parameter Determination 81 CHAPTER 7. A PRIORI DETERMINATION OF MODEL PARAMETERS Pitzer's Parameters Equilibrium Constants Estimation of Thermodynamic Properties AHf and AGf of sodalite dihydrate and hydroxysodalite dihydrate C p of Sodalite dihydrate and hydroxysodalite dihydrate S of Si0 3 ' ( a q ) AH f of Si0 3 2"(aq) C p of Si0 3 - ( a q ) Calculation of K s o d at K Calculation of K h s o d at K Change of In Khsod with Temperature 99 CHAPTER 8. SOLUBILITY MAPS OF Al AND Si IN GREEN AND WHITE 101 LIQUORS 8.1. Synthetic Green and White Liquors of System A Synthetic Green and White Liquors of System B Mill Green and White Liquors Structure of Precipitates from synthetic liquors Structure of Precipitates from mill liquors Morphology of Precipitates 122 CHAPTER 9. PREDICTION OF THE PRECIPITATION CONDITIONS OF 125 SODIUM ALUMTNOSILICATES 9.1. Na Al(OH) 4 " - Si OFT - C0 3 " - S Cf - H 2 0 (System A) Na _ - Al(OH) 4 " - Si0 3 " - OH" - C0 3 - Cl' - HS" - H 2 0 (System B) 135 CHAPTER 10. CONCLUSIONS AND RECOMMENDATIONS 139

8 vii NOMENCLATURE 142 BIBLOGRAPHY 146 APPENDIX I. Chemical Analysis by Atomic Absorption Spectrophotometer 159 APPENDIX II. Standard Errors and Confidence Intervals for Experimental 161 Solubility Data APPENDIX III. Tables of Solubility Data 162 APPENDIX IV. Calculation of Titration Curve 165 APPENDIX V. Uncertainty of the Measured Osmotic Coefficient 167 APPENDIX VI. Uncertainty of the Estimated Thermodynamic Properties and 170 Equilibrium Constants APPENDIX VII. Comparison of Pitzer's Parameters with Published Values 174 APPENDIX VIII. Sensitivity Analysis of Pitzer's Parameters 176 APPENDIX IX. Computational Source Codes in FORTRAN Determination of Pitzer's binary parameters for single electrolyte system using 178 osmotic coefficient data Calculation of osmotic coefficient and activity coefficient using Pitzer's binary 183 parameters for single electrolyte system Determination of Pitzer's mixing parameters for multi-component electrolyte 186 system using osmotic coefficient data Calculation of osmotic coefficient and activity coefficient using Pitzer's binary 193 and mixing parameters for multi-component electrolyte system Computation of equilibrium state for NI^-NI^OH-Ff-HCl-NELiCl-Cr-Na*- 199 NaCl-K + -KCl system at K Computation of equilibrium state for sodium aluminosilicate formation 205

9 viii LIST OF TABLES Table 1.1 Amount of non-process elements from each source. 6 Table 1.2. Metal contents in the bleach effluent. 8 Table 1.3. Concentration range of process and non-process elements in green and 8 white liquors. Table 3.1. Solubility experiment design for synthetic liquors of system A. 36 Table 3.2. Solubility experiment design for synthetic liquors of system B. 41 Table 3.3. Analysis results of mill liquors. 43 Table 3.4. Solubility experiment design for mill liquors. 44 Table 5.1. Osmotic coefficients and water activities for the Na2Si03 aqueous 66 system at K. Table 5.2. Osmotic coefficients and water activities for the Na2Si0 3 -Na0H 68 aqueous system at K. Table 5.3. Osmotic coefficients of NaCl and KC1 as reference and standard 71 solutions at K. Table 5.4. Osmotic coefficients of NaOH-NaCl-NaAl(OH)4 aqueous solutions at K. Table 6.1. The Pitzer's parameters of Na2Si0 3 and Na 2 Si0 3 -NaOH systems at K. Table 6.2. The Pitzer's parameters at K available in the literature. 79 Table 6.3. The Pitzer's parameters of NaOH-NaCl-NaAl(OH) 4 aqueous system at K obtained in this study. Table 6.4. The Pitzer's parameters of NaTc0 4, NaTc0 4 -NaCl, and NaBr-NaC systems at K. Table 6.5. Comparison of measured osmotic coefficients with calculated those for 82 the NaTc0 4 system at K.

10 ix Table 7.1. Pitzer's binary parameters for the modeling of sodium aluminosilicate 85 formation. Table 7.2. Pitzer's mixing parameters for the modeling of sodium aluminosilicate 85 formation. Table 7.3. Thermodynamic data at K and 1 bar published in the literature. 87 Table 7.4. Estimations of AH f and AG f of anhydrous sodalite (Na 8 (AlSi0 4 ) 6 Cl 2 ). 89 Table 7.5. Estimated thermodynamic data at K and 1 bar. 90 Table 7.6. Estimation of C p of anhydrous sodalite (Na 8 (AlSi0 4 )6Cl 2 ). 92 Table 7.7. Contribution by 2H 2 0 for C p estimation. 92 Table 7.8. Contributions by Cl 2 and (OH) 2 for C p estimation. 93 Table 7.9. Prediction of the entropy of aqueous ions at K, cal/mol K. 94 Table Thermodynamic data of Na 2 SiC>3( S ), Na + (aq), and Si03 2 "(aq) for the 95 calculation of the AG 0, AH, and AS 0 of reaction (7.11). Table Heat capacities of aqueous oxy-anions at K, cal/mol K. 96 Table 8.1. Repeatability of the experiments in three runs. 102 Table 9.1. Comparison of the calculation results for the NH, - NHjOH - H* - HCl NH4CI - Cf - Na + - NaCl - K + - KC1 system at K. Table 9.2. Example of the modeling calculation results for the Na + - Al(OH) 4 " _ Si0 3 " - Off - C0 3 " - S0 4 - Cf - H 2 0 system at K. Table A Analytical data for the analysis of liquid phase alkaline samples. 159 Table A Analytical data for the analysis of samples from the solid precipitates. 160 Table A.3.1. Solubility data for synthetic liquors of system A. 162 Table A.3.2. Solubility data for synthetic liquors of system B. 163 Table A.3.3. Solubility data for mill liquors. 164 Table A.7.1. Comparison of Pitzer's parameters. 174 Table A.8.1. Sensitivity analysis of Pitzer's parameters for Na 2 Si0 3 aqueous 176 system.

11 X Table A.8.2. Sensitivity analysis of Pitzer's parameters for Na2SiC«3-NaOH aqueous 177 system. Table A.8.3. Sensitivity analysis of Pitzer's parameters for NaOH-NaCl- 177 NaAl(OH)4 aqueous system.

12 xi LIST OF FIGURES Figure 1.1. A diagram of the kraft pulping process. 2 Figure 1.2. Metal profile of pre-oxygen stage kraft pulp. 7 Figure 3.1. Equilibrium vessel for the solubility experiments. 38 Figure 3.2. The isopiestic apparatus. 47 Figure 4.1. A block diagram of the thermodynamic modeling. 56 Figure 5.1. Titration curve for the Na2SiC»3-NaOH solution with HC1 solution. 58 Figure 5.2. Calculated titration curve for the Na 2 Si0 3 -NaOH and Na 2 C0 3 -NaOH 62 solutions with HC1 solution. Figure 5.3. Mole fraction of metasilicic species with ph. 64 Figure 5.4. Osmotic coefficients of Na 2 Si0 3 and KCI aqueous solutions at K. Figure 5.5. Osmotic coefficients of mixed Na 2 Si0 3 -NaOH and KCI aqueous 69 solutions at K. Figure 6.1. Mean activity coefficients of Na 2 Si0 3 in Na 2 Si0 3 aqueous solution at K. Figure 6.2. Mean activity coefficients of Na 2 Si0 3 and NaOH in mixed Na 2 Si NaOH aqueous solution at K. Figure 7.1. Change of In Khsod with temperature. 99 Figure 8.1. Al and Si ions approaching equilibrium obtained from the base 102 experiment using synthetic green liquor of system A. Figure 8.2. Solubility map of Al and Si in synthetic green and white liquors of 104 system A. Figure 8.3. Effect of hydroxyl ions on the solubility limit in synthetic green and 105 white liquors of system A. Figure 8.4. Effect of carbonate ions on the solubility limit in synthetic green and 107

13 Xll white liquors of system A. Figure 8.5. Effect of sulfate ions on the solubility limit in synthetic green and white 108 liquors of system A. Figure 8.6. Solubility map of Al and Si in synthetic green and white liquors of 110 system B. Figure 8.7. Effect of Na 2 S on the solubility limit in synthetic green and white 111 liquors of system B. Figure 8.8. Solubility map of Al and Si in mill green and white liquors. 113 Figure 8.9. Effect of NaOH on the solubility limit of Al and Si in mill liquors. 114 Figure Effect of Na 2 C03 on the solubility limit of Al and Si in mill liquors. 115 Figure Effect of Na 2 S on the solubility limit of Al and Si in mill liquors. 116 Figure Solubility map comparing solubility limit of Al and Si in varying 117 liquors. Figure X-ray diffraction pattern of precipitates in synthetic green liquor of 119 system A. Figure Thermogravimetric analysis of hydroxysodalite dihydrate, 121 Na8(AlSi04)6(OH) 2-2H 2 0, precipitated in synthetic white liquor of system A. Figure Scanning electron micrographs of precipitates. 123 Figure 9.1. Equilibrium concentration of aluminum and silicon species. 127 Figure 9.2. Effect of hydroxyl ion concentration changes on the equilibrium 128 concentration of aluminum and silicon species. Figure 9.3. Effect of carbonate ion concentration changes on the equilibrium 130 concentration of aluminum and silicon species. Figure 9.4. Effect of sulfate ion concentration changes on the equilibrium 131 concentration of aluminum and silicon species. Figure 9.5. Effect of hydroxyl ion concentration changes on the equilibrium 132 concentration of aluminum and silicon species.

14 xiii Figure 9.6. Effect of carbonate ion concentration changes on the equilibrium 133 concentration of aluminum and silicon species. Figure 9.7. Effect of sulfate ion concentration changes on the equilibrium 134 concentration of aluminum and silicon species. Figure 9.8. Effect of hydrosulfide ion concentration changes on the equilibrium 135 concentration of aluminum and silicon species. Figure 9.9. Effect of hydrosulfide ion concentration changes on the equilibrium 136 concentration of aluminum and silicon species. Figure Comparison of model predictions with published correlations and 137 industrial data. Figure A.7.1. Osmotic coefficients for NaCl, Na 2 S0 4, and Na 2 S systems. 175

15 xiv ACKNOWLEDGEMENTS The author wishes to thank those who supplied the ideas, the time, and the friendship for this work. In particular, the author wishes to express his sincere appreciation to his supervisor, Dr. Peter Englezos, who guided him to the sea of electrolyte thermodynamics. His advice, patience, instruction and support are the principal contributions to the successful completion of this study. Sincere gratitude is extended to the thesis committee members, Dr. Charles Haynes for his fruitful discussions about the osmotic coefficient measurement as well as Dr. A. Paul Watkinson, and Dr. Chad P. J. Bennington for their valuable comments. The author is deeply indebted to Dr. Mati Raudsepp in the Department of Earth and Ocean Sciences for his kind help in performing the X-ray diffraction and SEM analysis. The author would like to acknowledge the help and support provided by Mike Towers, Vic Uloth, Brian Richardson, and Jim Wearing from PAPRICAN. Thanks are also expressed to Mr. Horace Lam, Ms. Rita Penco, Mr. Tim Paterson, Mr. Peter Taylor, Ms. Brenda Dutka, Mr. Ken Wong and all the staff in the Department of Chemical Engineering and Pulp and Paper Centre. John Bates Centennial fellowship as well as University Graduate Fellowship are gratefully acknowledged. On a more personal level, I wish to thank my best friends met in Canada: Tazim Rehmat for her kind friendship including a fortune chip from Las Vegas, Khizyr Khoultchaev for his unique instruction of titration methods, Peter Pang for his kind help in the lab, Geoff Bygrave for his useful discussion about thermodynamics, Isabelle

16 XV Pineault for sweet cookies and muffins, Ok Hyun Ahn for his great help with computers, Ah Hyung Park for her kindness to my family, Dal Hoon Lee for enjoying coffee breaks together, Jong Choon Lim and Wook Dong Kwon for sharing fresh atmosphere on the green at McCleery. I am also grateful to Prof. Kyong Ok Yoo, Prof. Doo Sub Kim, Prof. Hee Taik Kim, Prof. Sang June Choi, Prof. Dae Won Park, Prof. Tae Joo Choi, Dr. Dong Hyun Lee, and Dr. Byoung Moo Min for their grateful encouragement. I would pay special tribute to my mother, brothers, and parents in law for their unconditional love and support from my home, Korea. Twinkling eyes of my son, Si Young, have been a catalyst for this work. Most of all, I would like to express special thanks to my wife, Sun Mee Lee for her love, sacrifice and best dinners. She never doubted my decision to study abroad for many years. I would like to dedicate this work to my father in heaven. His confidence and encouragement in me have never weakened. Sincerely hope he is now proud of his son.

17 1 CHAPTER 1. INTRODUCTION Kraft pulp is made by reacting wood chips with a strongly alkaline solution. This process is called digestion in the pulp industry. The pulp fibres are then subjected to bleaching for brightness improvement. The kraft pulp process consists of three streams; brown fibre line, bleach plant fibre line and chemical recovery cycle (Smook, 1992). A simplified diagram of the kraft pulping process is shown in Figure 1.1. Wood chips are delignified in the digester by using a solution of NaOH (sodium hydroxide) and Na 2 S (sodium sulfide). After digestion, the delignified chips are discharged into a blow tank to be disintegrated into individual fibres. Separated weak black liquor from the brown stock washer is evaporated and burnt in the recovery boiler. Most of organic materials are incinerated. The remaining inorganic chemicals are entered into the smelt dissolving tank. Major portion of the smelt is a solid form of Na 2 C03 (sodium carbonate), Na 2 S and NaOH. The smelt is dissolved in water which forms green liquor. The dissolved Na 2 C03 in green liquor is converted to NaOH in white liquor by the reaction (1.1) in the causticizers. Na 2 C0 3 + Ca(OH) 2 -> 2NaOH + CaC0 3 (1.1) White liquor containing the cooking chemicals, NaOH and Na 2 S, is recycled to the digester and used for cooking chips (Smook, 1992; Grace etal, 1989).

18 2

19 3 The fibres separated at the brown stock washer do not have enough brightness. The lignin content of the fibres is reduced by 50 % in the oxygen delignification stage. Dissolved solids are recovered by washing the oxygen-delignified stock and they are recycled to the recovery boiler. The pulp is discharged to the bleach plant to increase the brightness (Grace et al, 1989). Several different sequences of bleach stages are used in modern kraft pulp mills. The conventional bleaching method using the chlorine gas is not used any more because of the dioxin formation problem. Most of kraft mills have ECF (Elementary Chlorine Free) or TCF (Totally Chlorine Free) bleach stages (Teder et al, 1990; Chandra, 1997). The kraft pulp is used to make high quality papers such as printing paper because of its better strength and brightness compared to mechanical pulp (Grace et al, 1989) Environmental Impact of the Kraft Pulping Industry Dioxin and furan detection in effluents from kraft mills has been reported (Servos et al, 1996; Norstrom et al, 1988; Bohn, 1998). In 1983, U.S. government reported that dioxin was found in the downstream of pulp mills in Maine, Wisconsin and Minnesota. Biologists of Canadian Wildlife Service found some of the highest dioxin levels in the egg shells near the pulp mill at Crofton on Vancouver Island in 1987 (Bohn, 1998). Norstrom et al. (1988) reported that ecologically significant amounts of chlorinated dioxins and furans were present in sediments and fish in the vicinity of B.C. mills. As a result of these findings, stringent pollution standards were set by the government and more efforts to minimize the environmental impact have been requested from the kraft pulp mill industry (Martin, 1998).

20 4 The large amount of water consumption is another environmental concern of the kraft pulp mills. A few decades ago, it was common for a bleached kraft mill to use up to 400 m 3 /ton pulp. In spite of the significant improvements in the water consumption, total water use still ranges from 30 to 150 m 3 /ton pulp (Bihani, 1996) Closed-cycle Mill Operation Closed-cycle mill operation has been considered as the ultimate method to achieve further effluent reduction and to minimize the environmental impact. Closedcycle in the context here means no liquid effluents from the mill. Of course, there might be some emissions as gaseous and solid wastes which can be handled in an environmentally safe manner. This is the concept of reducing pollution at the source by reusing more of the water, fibre, chemicals and energy contained in the waste stream. This may also reduce the load of the effluent treatment (Galloway et al, 1994). The first attempt at closed-cycle at the Great Lakes Forest Products Ltd. bleached kraft pulp mill at Thunder Bay, Ontario, Canada was unsuccessful due to corrosion in the white liquor evaporators, the recovery boiler superheater tubes, and the black liquor evaporator tubes, pitch deposition, scaling in the salt recovery process and other miscellaneous problems (Patrick et al, 1994). Several trials of closed-cycle kraft pulp production have been attempted but it has not been achieved yet (Galloway et al, 1994). The gradual move towards a closed mill is what the industry refers to a progressive system closure. Although closed-cycle operation of any kraft mill has not been achieved yet, a number of mills have stated this as their goal, and are working towards it. It should be noted that closed-cycle CTMP (chemi-thermomechanical pulp) mills are already a

21 5 commercial reality (Barnes et al, 1995). Such technological development of closed-cycle CTMP mills opened the possibility of virtually eliminating liquid discharges from bleached kraft pulp mills Non-process Elements in the Kraft Pulping Process One of the issues that prevent closed-cycle kraft mill operation is related to the problems caused by the build-up of non-process elements (NPEs). NPEs are ions such as aluminum (Al), silicon (Si), calcium (Ca), barium (Ba), potassium (K), chloride (Cl), magnesium (Mg), phosphorus (P), manganese (Mn), iron (Fe), and copper (Cu). They enter the system as part of the wood chips, water, and chemicals and are not required to manufacture pulp. In an open mill NPEs are withdrawn but they accumulate in a closed one (Galloway et al, 1994). The build-up of NPEs can cause severe process problems such as scaling, corrosion, and interference of bleaching (Galloway et al, 1994; Gleadow et al, 1997). The proper management of NPEs is prerequisite for the successful implementation of closed-cycle technology Sources and profile of non-process elements The major sources of the non-process elements are wood, make-up lime, and water. Table 1.1 shows the amount of non-process elements introduced to the process. Most of the non-process elements tend to concentrate in the bark, roots, and foliage. Therefore, unbarked wood or whole tree chips will contain higher concentrations of nonprocess elements. Sand, clay and contamination attached on the logs contain significant

22 6 amounts of silicon and aluminum. Logs transported by sea add extremely large amounts of chloride. Table 1.1. Amount of non-process elements from each source (Galloway et al, 1994). Element Wood chips Make-up lime Water (mg/kg a.d. pulp) (mg/kg a.d. pulp) (mg/kg a.d. pulp) Al <10 Si Ca K 1100 <10 22 Cl 629 <10 74 Mg Mn 159 <10 <10 Fe There are no major differences in NPEs between hardwoods and softwoods except for the calcium and potassium content. Calcium is much higher in hardwoods, especially in aspen. Potassium appears to be a little higher in hardwood. Calcium and magnesium are integral components of the wood (Magnusson et al, 1979). Make-up lime is one of the major sources for calcium. Silicon and magnesium contents in the make-up lime are not negligible. About 5 ~ 12.2 kg/ton a.d. pulp of the make-up lime is used for the kraft pulping. Dissolved organic and inorganic materials including non-process elements are washed in brown stock washer and enter into the recovery line. A portion of non-process elements are attached to the pulp, which is introduced to the bleaching line with pulp. The metal profile of the pre-oxygen stage kraft pulp of 15.3 % consistency and ph 10.3 from a mill (in Prince George, B.C.) was determined in our laboratory. The results are shown in Figure 1.2. The pulp sample of 20 g was dried in an oven at 105 C and ashed in a furnace at 575 C. The ash was digested in 6 N HC1 solution. The digested solution was

23 diluted with distilled and deionized water and filtered through 0.45 pm membrane filter. The filtrate was analyzed by the ICP (Inductively Coupled Plasma, ACME Analytical Lab. Vancouver, B.C.) to determine the metal contents in the pulp. Meanwhile, a portion of the pulp sample was squeezed to extract the liquor. The liquor was filtered through 0.45 pm membrane filter. The filtrate sample was analyzed by ICP to determine the metal content in the liquor (shaded bars in Figure 1.2). Metal content in the fibres (white bars in Figure 1.2) was calculated by subtracting the metal content in the liquor from that in the pulp. Figure 1.2. Metal profile of pre-oxygen stage kraft pulp Q. -ri o T Pulp 1 Liquor Fibre I? E, c o jjj A A Na K Ca Mg Si Al Fe Mn P Cu Ba In the pre-oxygen stage kraft pulp, large amounts of Na, K, Ca, and Mg elements were present. Amounts of Si, Al, Fe, Mn, P, Cu, and Ba were not negligible. Most of

24 8 elements were distributed in fibers and their content in liquor was very low except for sodium and potassium. Table 1.2 shows metal content in acid and alkaline effluents of an ECF bleaching plant. Metal ion level in the acid bleach effluent is higher than that in alkaline effluent. Table 1.2. Metal contents in the bleach effluent, mg/kg oven dried pulp (Towers, 1995). Na Al Si Ca K Mg Mn Fe Acid Alkaline ND ND : not detected Table 1.3. Concentration range of process and non-process elements in green and white liquors, mol/l. element Green liquor White liquor Na mol/l Al [ 3 ] m Si [ 1 ] Ca [ U Mg [ 1 ] [ 1 ] K [ 1 ] m 2 co l 4 ' 6 ] S0 4 2 " [ 4 5 ] [ 4 > 6 ] HS" [ 5 ] 0.7 t 5 > OH" [ 4 ' 5 ] , 4 1 cr m [ 1 ] [1] Towers 1995, [2] Ulmgren 1995, [3] Ulmgren 1987, [4] Magnusson etal. 1979, [5] Lindberg and Ulmgren 1982, [6] Grace etal Table 1.3 shows the concentration range of non-process elements as well as process elements in molarity. The Al, Si, and K are soluble elements in alkaline solutions and their concentration levels are higher than those of Ca and Mg.

25 Adverse effects of non-process elements Aluminum and silicon are found particularly in the bark of the pulpwood. These elements are a concern in the recovery cycle since no large purge points are available. Accumulation of aluminum and silicon in the sodium cycle can result in the formation of sodium aluminosilicate scales. Formation of these scales lowers heat transfer efficiency and reduces evaporation capacity of the black liquor evaporators (Ulmgren, 1982). An increase in the silicon content of lime mud lowers the reactivity of the burnt lime because alkali-silicon compounds could melt on the surface of the lime pellets thus reducing the porosity of lime (Magnusson et al, 1979). Calcium is considered a non-process element except in the recausticizing and lime kiln area. Calcium can lead to scaling problems in a number of mill areas (Gleadow et al, 1996). Recycling of acidic effluents to the alkaline brownstock increases calcium concentrations, which can cause equipment scales such as calcium carbonate, calcium oxalate and calcium silicate. When the calcium concentration in acid bleaching stages is high, calcium sulfate scales are formed because calcium sulfate is relatively insoluble at low ph (Bryant and Edwards, 1996). Accumulation of barium results in barium sulfate deposit problems in a ECF or TCF bleach plant (Ulmgren, 1996). Potassium enters the mill with the wood chips. It combines with chloride and leads to plugging related problems caused by sticky recovery boiler dust. Potassium has the greatest tendency to accumulate in the sodium cycle under closed-cycle operation. There is no effective purge for potassium since it is soluble in alkaline solutions. The major undesirable impact of potassium is in the recovery furnace where, in combination with chloride, it reduces the melting point of sodium salts that are entrained in the flue

26 10 gas and accumulate on the tubes of the boiler. Sodium and potassium produce slabbing and ring formation in the lime kiln at high concentrations because of the low melting points of sodium and potassium carbonate in the calcium cycle (Gilbert and Rapson, 1980). High chloride levels in the recovery cycle cause excessive corrosion of the superheater tubes in the recovery boiler, decreasing recovery boiler capacity, and allowing accumulation of dust on the boiler tubes (Blackwell and Hitzroth, 1992) Magnesium also enters as MgSCu used in the fiber line and with the make-up lime. Magnesium and phosphorus levels are important in the calcium cycle. Accumulation of these elements causes increased deadload, reduced settling rates, and poor filterability in the calcium cycle (Gleadow et al, 1996). Formation of Mg(OH)2, a gelatinuous material, causes plugging problems in white liquor and lime mud filters (Galloway et al, 1994). Transition metals such as Mn, Fe, and Cu interfere with bleaching by catalysing decomposition of hydrogen peroxide under alkaline conditions (Ulmgren, 1996) Recent Developments of Closed-cycle Technology A key element in progressing towards either a closed-cycle or effluent-free mill is the recycling of bleach plant effluents. Gleadow and his coworkers (1993) suggested a method to achieve complete recycling of the effluents. In this method, the alkaline bleach effluent is used in brownstock washing and the acidic effluent is concentrated and added to the black liquor. Evaporator condensates are treated for reuse as process water supply. If the acidic bleach effluent is recycled to the brown stock washers as shower water, some of the metals can be redeposited onto the pulp. This recycle of metals results

27 11 in an increase in transition metal content for the pulp that is sent to the bleach plant. In addition, the concentration of scale-inducing metal ions can be increased throughout the fiberline and in the weak black liquor (Bryant and Edwards, 1996). Richardson et al. (1995) suggested the recycling of bleach-plant effluents to the green liquor line. Although fresh water of 50 % was replaced with acidic bleaching effluents(dioo stage), causticizing efficiency was only decreased from 81.7 % to 77.4 %. The replacement of 50 % fresh water with alkaline bleaching effluents (E 0 stage) just decreased the causticizing efficiency from 81.7 % to 76.0 %. However, the recycling of bleaching effluents resulted in the build up of non-process elements such as silicon, aluminum, calcium, magnesium, and potassium. Maple et al. (1994) reported the development of the bleach filtrate recycle (BFR ) process for bleach plant closure. The important aspects of the process are chloride removal and metal removal. Alkaline bleach effluents are recycled for brown stock washing. Dissolved solids in the effluents are transported into the black liquor evaporators and fired in the recovery boiler. The chloride component in the electrostatic precipitator dust can be removed. During the chloride removal process, the dust is mixed with hot water. Because of greater solubility of sodium chloride than that of sodium sulfate, the sodium sulfate in the dust is precipitated out and recycled to the black liquor. Sodium chloride in aqueous form is disposed. Metal ions in acid bleach effluent are removed by chemical precipitation with sodium hydroxide and sodium carbonate or ion exchange. Treated effluent is recycled to the bleach plant. Ulmgren (1996) suggested a recycling strategy for a closed bleach plant. The alkaline bleach effluents are taken to the brown stock washer. The organic material in the

28 12 alkaline effluents can be removed by burning in the recovery boiler via the brown stock washer. The acidic effluents can be taken into the chemical recovery cycle for white liquor preparation. In the case of the acid effluent from an ECF bleach plant, the effluent contains a high level of chloride. These acidic effluents, rich in chloride, should be treated separately to prevent the intake of chloride ion. Gleadow et al. (1998) presented three closed-cycle case studies using process simulations followed by the analysis of elemental component behaviour in the kraft pulping process. Participating mills were a 1965 vintage B.C. coastal ECF kraft mill, a modern (1990s vintage) B.C. coastal ECF mill, and a 1965 vintage B.C. interior mill. The amount of bleach effluent, m 3 /a.d. metric ton, is a big portion of the total mill effluent, m 3 /a.d. metric ton. They suggested closed-cycle designs in which a part or all of the bleach plant effluent was concentrated in a bleach plant evaporator and the solids incinerated in the existing black liquor recovery boiler. Recycle of alkaline filtrate to brown stock was also considered. Accumulation of K and Cl can be controlled by purging precipitator catch in the recovery boiler. Simulation with these design options showed no increase in the corrosiveness of the recovery boiler deposits and the potential for plugging. Potassium and chloride levels were constant or decreased as a result of process modifications. Calcium and magnesium levels, however, in black liquor raised a concern. The concentration of these elements were increased up to 300 %. A new bleaching plant evaporator and a recovery boiler precipitator catch leaching system are required. Additional bleaching stages are be required to compensate for the decrease of bleaching efficiency in some cases. Capital cost requirements are similar to those of effluent treatment facilities.

29 Scale Formation in the Recovery Cycle Several types of scaling occur throughout the recovery cycle of kraft pulp mills. Materials insoluble in black liquor are fibres, sand, oil soap (at % solids), lignin (at ph below 11), calcium carbonate scale (at high temperature), sodium carbonate-sodium sulfate double salt (usually about 50% solids), and sodium aluminosilicate scale. Sand and fibres are usually found to be combined with deposits which plug tubes (Grace et al, 1989). Calcium scaling is very sensitive to temperature in black liquor evaporators. A complex is formed between calcium ion and lignin sub-groups in the liquor. At high temperature, the complex breaks down. The released calcium ions combine with carbonate ions on the hot surface, where they form calcium carbonate scale (Westervelt et al, 1982). Increased amount of calcium oxalate in black liquor significantly increases the tendency of the calcium carbonate scale formation (Grace et al, 1989). Sodium carbonate-sodium sulfate scales are also common in black liquor evaporators. The scales are soluble in black liquor. If their concentration exceeds the solubility limit, they form by crystallizing from a supersaturated solution. The scales are usually found to have a structure of burkeite, 2Na2S04-Na2CC>3. The solubility at high temperature is slightly lower than that at low temperature in the range 100 ~ 140 C Organic compounds in liquor do not greatly influence the solubility (Grace et al, 1989). The sodium carbonate-sulfate scaling can be aggravated by soap. Soap has a strong tendency to precipitate on fibre to form a sticky mass. Thus, soap and fibre should be kept out of the black liquor as much as possible (Uloth and Wong, 1985; Grace et al, 1989).

30 14 Sodium oxalate (Na2C204) scaling occurs sometimes in the kraft pulp mills using hardwood species, especially if weak black liquor oxidation is used to prevent odorous emissions of H2S gas. The sodium oxalate is moderately soluble in water (6.0 g/100 g at 100 C) and shows a normal solubility temperature relationship. Thus, solubility increases with increasing temperature. Solubility of the sodium oxalate varies with amounts of other sodium salts. It forms scales in black liquor evaporators by precipitation as the solid content in the liquor increases (Grace et al, 1989). Sodium aluminosilicate scales are hard and glossy. Since they have a very low thermal conductivity, a very thin layer is enough to cause a serious reduction of the evaporation capacity. Although these scales grow slowly, they are very tenacious and difficult to remove (Grace et al, 1989). Solubility increases slightly with increasing temperature in the range 95 ~ 150 C (Streisel, 1987). A higher solids content (decomposition products of lignin) in the black liquor results in a higher solubility. Thus, the liquor can tolerate more soluble Al and Si before sodium aluminosilicate precipitates. This is probably due to the increased concentration of decomposition products of lignin. These organic compounds form strong chelate complexes with Al. Such structures occur in black liquor as decomposition products of lignin (Ulmgren, 1985). Green liquor systems also have scaling problems in transfer lines, the clarifiers, and the dissolving tanks. These scales are usually found to have structures of pirsonnite (Na2C0 3Na2C0 3-2H 20) and gaylussite (Na 2C0 3 Na 2C0 3-5H 20). Scales seem to be favored by decreased temperatures (Frederick and Krishnan, 1990). Increased intake of aluminum and silicon can cause sodium aluminosilicate formation in green and white liquor lines (Wannenmacher et al, 1996).

31 Sodium aluminosilicate scale formation The sodium aluminosilicate scales may form in the recovery cycle of a kraft pulp mill and during the Bayer process for aluminum production (Gasteiger et al, 1992; Swaddle et al, 1994; Zheng et al, 1997; Gerson and Zheng, 1997). In the case of the kraft pulping process, aluminum and silicon ions enter the process with wood chips, water, and make-up lime as seen in Table 1.1. In particular, significant amounts of aluminum and silicon enter from unwashed chips since bark and soil on the chips are a major source of those ions (Galloway et al, 1994). If aluminum and silicon ions are introduced to the recovery cycle, those ions can not be purged out easily from the cycle since they are soluble in alkaline aqueous solutions. The build up of aluminum and silicon results in sodium aluminosilicate scale formation in the recovery cycle. The scale is glossy and hard to remove. It may lower the efficiency of heat exchangers and evaporation capacity of a kraft pulp mill (Ulmgren, 1982). The scale formation becomes more severe in closed-cycle plant operation because the effluent is recycled and the aluminum and silicon ion concentrations increase rapidly. Recycling of bleach plant effluents carries a lot of non-process elements to the recovery cycle as seen in Table 1.2. The Al and Si are not removed easily by precipitator dust removal or green and white liquor clarifications (Gleadow et al, 1998). Ulmgren (1982) studied the Na-Al-Silicate scale formation in black liquors of the recovery cycle. He measured the solubilities of Al and Si using the synthetic black liquors made of NaOH, Na 2 C0 3, Na 2 S0 4, and CH 3 COONa (sodium acetate) at C. The scale samples were found to have structures of NaAlSi04-l/3Na 2 C03 and/or

32 16 NaAlSi04-l/3Na 2 S04 which look like cancrinite. According to his results, one of the main factors governing formation of aluminosilicates was the Off concentration. More soluble species formed in solution as the Off concentration increased in the range of the Off concentration from 0.03 to 1.6 mol/l. In addition, high concentrations of Off caused scale formation on the surface of vessels and low concentrations resulted in precipitation as particles. Addition of HS" of 0.4 mol/l did not make any significant change in the solubility of Al and Si. The effect of temperature on the formation of sodium aluminosilicate was not significant at 120 C, 135 C, and 150 C. Solubility increased slightly with increasing temperature. Ulmgren also presented a model for the formation of Na-Al-Silicate scale in evaporators for black liquor in kraft pulp mills. For the chemical modeling, he assumed the Na-Al-Silicate in the evaporators were formed by precipitation from a solution saturated with ions involved such as Na +, Al(OH)4\ 2 2 Si0 2 (0H) 2 \ SiO(OH) 3 \ Off, C0 3 \ S0 4 2\ Cf and HS". Finally, he determined the equilibrium constant of the cancrinite formation using the experimental solubility data to correlate the precipitation conditions. Ulmgren (1987) suggested a method to reduce the concentration of aluminum in the recovery cycle by adding a magnesium salt such as magnesium sulfate and magnesium hydroxide carbonate to the smelt dissolving tank. The insoluble salt, hydrotalcite (Mgi. x Al x (0H) 2 (C0 3 )x/ 2 nh 2 0, 0.10 <x< 0.34), was formed which could be removed from the process together with the dregs by the green liquor clarification. In this way, sodium aluminosilicate scale formation by the build up of Al and Si could be avoided. He observed a decrease in the aluminum concentration from 1.5 to 0.5 mmol/l by the addition of magnesium salts with molar ratio of Mg/Al=6 into the mill green

33 17 liquor. He also observed that removal of aluminum by the addition of the magnesium salts could be expedited by reducing the OH" concentration and/or temperature of the liquor. Streisel (1987) studied the Na-Al-Silicate formation using synthetic liquors. He prepared the liquors using NaOH and NaCl. He measured the precipitation conditions of the sodium aluminosilicates by varying the following parameters: the Al/Si molar ratio from 0.08 to 275, the hydroxyl ion concentration from 0.1 to 4 N, and the ionic strength from 1 to 4 N. The temperature was set at 95 and 150 C. According to the results, as the hydroxide ion concentration increased, more soluble aluminum and silicon species formed in solution. The apparent solubility product of Al and Si decreased with increasing ionic strength. The apparent solubility product, [Al][Si] at 150 C was higher than the solubility product at 95 C. He found that the precipitates had a structure of sodalite (Na8(AlSi0 4 )6Cl 2 ) by the X-ray diffraction analysis. Streisel (1987) also studied the effects of the cationic ions (K, V, Fe, Mn, Ca, Mg) addition on the solubility of Al and Si. The potassium and vanadium are very soluble in alkaline solution and neither potassium or vanadium had any significant influence on the apparent solubility product of Al and Si. Iron and manganese are insoluble in alkaline solution. Iron and manganese precipitated as iron oxide (Fe 2 03) and manganese hydroxide (Mn(OH) 2 ). Neither iron or manganese had any significant influence on the apparent solubility product of Al and Si. Calcium is nearly insoluble in alkaline solution. The calcium has an effect on the rate of precipitation of Al and Si but not on the apparent solubility product. When the molar ratio of Al/Si is less than one, aluminum tobermorite (Ca5Si5Al(OH)n5H 2 0) was mainly formed. When the molar ratio was equal to one,

34 18 sodalite and calcium hydroxide were formed. By the addition of MgS0 4, hydrotalcite was formed. The higher Mg/Al ratios resulted in significantly more precipitation of aluminum. Streisel (1987) developed a model to predict the precipitation conditions in the concentration range of Off from 0.1 to 1.0 N for Na + - Cl" - Al(OH) 4 " - HSi0 4 3" - OH* - H2O system. He adopted Meissner's method (Zemaitis et al, 1986) to calculate the activity coefficients of species and extracted Meissner's model parameters from his experimental solubility data. His model was unable to calculate the concentration of aluminum at higher hydroxide concentrations because of a convergence problem. Gasteiger et al. (1992) upgraded Streisel's model. Pitzer's method (Pitzer, 1991) was adopted to calculate activity coefficients of the species and activity of water. The model contained five adjustable parameters, an equilibrium constant of sodalite/hydroxysodalite formation and four Pitzer's parameters of J3 (0) and for NaHSi0 4 and p (0) and p (1) for NaAl(OH) 4. They obtained the five adjustable parameters by correlating Streisel's experimental solubility data. They found that the precipitates had the structure of sodalite (Na8(AlSi0 4 )6Cl2) and/or hydroxysodalite (Nas(AlSi0 4 )6(OH)2) by X-ray diffraction analysis. Wannenmacher et al. (1996) reported the effect of CaO and MgS0 4 addition on the precipitation conditions of sodium aluminosilicate scale in an unclarified kraft mill green liquor at 95 C. For the 1:1 addition of CaO ([CaO]:[Si]), no change in the aluminum and silicon solubility was observed. The addition of MgS0 4 was found to lower the solubility of Al. They presented graphs showing the precipitation conditions of the Na-Al-Silicates for black, green and white liquors based on their experimental and

35 19 literature data. They also reported several equations to calculate apparent solubility product, [Al][Si] at equilibrium with sodalite (Na8(AlSiC»4)6Cl2) and/or hydroxysodalite (Na8(AlSi04)6(OH)2), cancrinite, natrolite, and hydrotalcite. The correlations were based on experimental solubility data. Activity coefficients were not considered for the calculations Research Objectives Knowledge of the sodium aluminosilicate scale precipitation conditions is required to design closed-cycle pulp mills. Although the formation of sodium aluminosilicate scales in alkaline solutions has been studied in the past (Ulmgren, 1982; Ulmgren, 1987; Streisel, 1987; Wannenmacher et al, 1996), the effects of Na 2 C0 3, Na2S04 and ISfoS concentration on the precipitation conditions in green and white liquors have not yet been investigated. Thus, one objective of this work was to measure the solubility of Al and Si in highly alkaline solutions at K (95 C) and determine the effects of Na2C0 3, Na2S04 and Na2S addition on the precipitation conditions. We used synthetic liquors of two different systems, A and B, that simulated mill green and white liquors. The synthetic green and white liquors of system A were prepared using NaOH, Na2C0 3, Na2S04, and NaCl. Those of system B were prepared using NaOH, Na2C0 3, and Na2S. The precipitation conditions of sodium aluminosilicate scale in mill green and white liquors were also measured. The effects of NaOH, Na2C0 3 and Na2S at different concentrations on the precipitation conditions in mill liquors were observed. Mill green and white liquors received from a mill in Prince George, B.C. were used for the study. In

36 20 addition, the structure of the precipitates were identified by X-ray, chemical and thermogravimetric analysis. The other objective of this work was to present a thermodynamics-based model for sodium aluminosilicate formation in aqueous alkaline solutions. Pitzer's activity coefficient method was used to calculate the activity of water and the activity coefficients of ions in the solution because it performs well even at high molalities (Pitzer, 1991; Zemaitis et al, 1986). All parameters needed by the model were obtained from independent experimental data or available property estimation methods. The effects of the anions of OH -, CO3 2, SO4 2 ", and HS - were taken into account. The model predicts the precipitation conditions of sodalite dihydrate and/or hydroxysodalite dihydrate. Since the system contains the Si03 2 ' and Al(OHV ions, knowledge of the relevant Pitzer's model parameters is required. Osmotic coefficient data for the Na2Si03 single electrolyte system and the Na2Si03-Na0H multi-component electrolyte system were obtained at K by an isopiestic method. The binary parameters, P (0), and C*, for Na 2 Si0 3 and the mixing parameters, 9 OH - siol - and ^Na+0H - Si0 2-, for Na2Si0 3 -Na0H system were determined using the osmotic coefficient data. In addition, the binary parameters, f3 (0), P (1), and C*, for NaAl(OH) 4 and the mixing parameters, OH-^^-, e crai (oh);. *N.-OH-AI(OH);» A N D ^N.-CTAKOHH a r e n e e d e d - 0 s m o t i c coefficient data for NaOH-NaCl-NaAl(OH)4 aqueous system were also obtained at K. The unknown Pitzer's parameters relevant for the description of the system were obtained using these osmotic coefficient data.

37 21 CHAPTER 2. THEORETICAL BACKGROUND This chapter presents the fundamental thermodynamic relations that govern the behaviour of electrolyte solutions as well as Pitzer's activity coefficient model. In addition, the isopiestic method for the measurement of osmotic coefficients is also presented Thermodynamic Equilibrium Constant Let us consider the following reaction in an aqueous electrolyte solution. aa + bb = cc + dd (2.1) The condition of chemical equilibrium can be denoted by au, A + bu, B = cue + duo (2.2) where p. is a chemical potential. It is convenient in aqueous solution thermodynamics to describe the chemical potential of a species i in terms of its activity ai. u i =H?+RTta(a i ) (2.3) where u, is a chemical potential at an arbitrarily chosen standard state. When the composition of the solution is described in terms of the molality scale, the standard state of ion species is that of a hypothetical one molar solution at the same temperature and pressure as the solution (Pitzer, 1991). For the solvent (water), the standard state refers to pure solvent (water) at the same temperature and pressure as the solution. The activity is

38 22 a measure of the difference between the component's chemical potential at the state of interest and its standard state. The deviation from ideality is defined by the activity coefficient, y = a/m. Substituting the activity of species i, ai to yjmj in equation (2.3) gives ^-ur+rtlncy.mj (2.4) Thus, the general expression of equation (2.2) for the equilibrium can be expanded as follows a(u +RTln( YA m A )) + b(k +RTln(y B m B )) = c(p. + RTln(y c m c )) + d(u D +RTln(y D m D )) (2.5) By combining terms au A +bu B -cn -du D =RTln (Yc m c) C (V D m D) D (Y A m A ) A (Y B m B ) B (2.6) Since the partial molar Gibbs free energy is also defined as the chemical potential, the u,, can be substituted by AG ; (standard Gibbs free energy of formation). aagf A + bag B -cag c -dag D = RTln (Ycm c) C (Y D m D ) D (Y A m A ) A (y B m B ) B (2.7) The solubility of A, B, C, and D species at equilibrium are expressed in molality, ITIA, me, mc, and mo respectively. Since the thermodynamic equilibrium constant for this reaction is defined as K = exp aag A +bag f B -cag? c -dag f t RT (2.8) the equation (2.8) can be written as follows

39 23 K = (Yc c) (YD D) - AG m C m D - exp.(yam A) (YBm B). RT A B (2.9) where AG 0 is the standard Gibbs free energy change of reaction (2.1) Equilibrium Constant Calculation at Specific Temperature The thermodynamic equilibrium constant K in equation (2.9) can also be written as follows. InK AG 0 RT (2.10) Differentiating the equation (2.10) with T at constant P and composition gives DdlnK d(ag /T) dt dt (2.11) The Gibbs-Helmholtz equation is given by d_ 5T f AG 0 > AH 0 (2.12) where AH 0 is the standard enthalpy change of reaction. Thus, the equation (2.11) can be written as follows. RdlnK = AH: dt T 2 The AH can be expressed as a function of temperature in terms of heat capacity, ACp. AH = AH + f T AC P dt (2.14) «r 0 Assuming constant ACp in the range of the To and T gives AH 0 = AH + AC (T-T 0 ) (2.15)

40 24 Combining (2.13) and (2.15) gives R d t a K = AH dt T 2 \T I +^ni] I 2 ) (2.16) Integrating the equation (2.16) between the limits of the reference temperature, To, and T gives AH 0 lnk-lnk n = - i n T X AC; f 1 1 N lnt-lnt 0 +T 0 (-- ) 1 A V o j (2.17) It can be written as follows since In K 0 = -AGQ /RT 0. lnk = V AGJ R T o J AH o c R I L AC o f l n ^ - 1 R V T 0 T (2.18) Equation (2.18) is used to calculate the equilibrium constant at any temperature, T (Anderson and Crerar, 1993) Activity of Water, Osmotic Coefficient, and Activity coefficient Activity and activity coefficients were introduced to describe the non-ideal behaviour of a component in a solution. Based on equation 2.3, the activity of solvent (water), a w, can be expressed by: exp RT (2.19) where p w is the chemical potential of the solvent, p^, is the standard state chemical potential of the solvent, R is the gas constant, and T is the absolute temperature. The standard state refers to pure solvent (water) at the same temperature and pressure as the solution (Pitzer 1991).

41 25 Activity, however, is not sensitive at low molalities and requires several significant digits to express the behaviour accurately. The practical osmotic coefficient was introduced to avoid this problem by exaggerating the deviation between real and ideal behaviour (Pitzer 1991, Zemaitis et al. 1986). The practical osmotic coefficient, <j>, is defined as follows na w (2.20a) M w 2 > i m i where, M w is the molecular weight of the solvent (water), Vi is number of ions produced by 1 mol of solute i, and m; is the molality of solute i. For a single aqueous electrolyte solution with a solute molality equal to m s and v ions produced when the electrolyte is dissolved the above equation is written as -looolna.,, M,vm. (2.20b) Let us now consider single electrolyte aqueous solution. The total Gibbs energy ng is a function of T, P, and the numbers of moles of the chemical species present (solute and water). Thus, the ng is given by ng = f(t,p,n s,n w ) (2.21) where n is number of moles, subscript w stands for solvent (water), and subscript s denotes the solute (electrolyte). The total differential of ng is d(ng) = d(ng) ap T,n dp + d(ng) err JP,n dt + d(ng) an. P.T,n dn. + a(ng) an... P.T.n, dn w (2.22) Since the partial molar Gibbs energy of species i is defined as follows, G: = a(ng) an ; (2.23) JT.P.tij

42 26 the equation (2.22) can be rewritten as follows. d(ng) = ~5(nG)~ dp + "d(ng)~ T,n 5T P,n dt + G dn +G. dn S 3 W W (2.24) Meanwhile, a general differential equation for (ng) at constant T and P is given by d(ng) = 2 n i d G, +2_G,dn l and d(ng) = n s dg s +n w dg w +G s dn s +G w dn w (2.25) Comparison of this equation (2.25) with (2.24) yields the Gibbs-Duhem equation. g(ng) cp -lt,n dp + d(ng) 5T P.n dt-n s dg s -n w dg w =0 (2.26) For the case of 1 kg of solvent (water) at constant T and P, the equation (2.26) becomes -m.dg -m,dg, =0 W W and m s dg s =-m w dg w (2.27) where m s and m w are the molalities of the electrolyte (solute) and water (solvent) respectively. Since the chemical potential is defined as the partial molar Gibbs energy, the equation (2.27) becomes m s dp s = - m w dp w (2.28) Taking the general formula of solute (electrolyte) to be C v A V j, the chemical potential of the solute is u.=v c u. e +v.u._ = v c p: + v.u. + v.rt ln(a c)+v a RT ln(a.) (2.29) = v c p: +v 8 p: +v c RTln(m c y c ) + v a RTln(m a Y a ) where subscript c stands for cation and subscript a denotes the anion.

43 27 Differentiating the equation (2.29) gives dn s =RTdln(m^m:-y:'y:-) (2.30) The mean molality and mean activity coefficient are defined as m ± = ( m > i - r = ( > : c > : - Y v = ^ > : - Y v (2.31) Y ± -(Y. V -Y. V, ),, V (2-32) where v = v c + v a. Thus, equation (2.30) becomes dn s =vrtdln(m ± Y ± ) (2.33) The chemical potential of solvent (water) is given by U w -U +RTlna w (2.34) Differentiating the equation (2.34) gives dp. w =RTdlna w (2.35) Substituting equations (2.33) and (2.35) to equation (2.28) gives vm s RTd ln(m ± Y ± ) = -m w RTd In a w and vm s dln(m ± Y ± ) = -m w dlna w (236) Another expression for equation (2.20b) is given as follows for single electrolyte system -M vm lna w = w s <b and lna w =^ m ^(j) (2.37) m w where m w = 1000/M W = moles/kg H 2 0 Thus, substituting the equation (2.37) to the equation (2.36) gives vm s dln(m ± Y ± ) = -m w d

44 28 m s dln(m ± Y ± ) = d(m 3 <t>) and m s d In (m±y±) = ( )dm s + m s d< ) (2.38) Equation (2.38) becomes dln(m ± y ± ) = dm s +d<t> m, dlnm ± +dlny ± =-^-dm s +d<j) Since dln(m ± )= dln(m s ) = dm L from equation (2.31), the above equation gives dlny ± = dm s +d< > (2.39) m s The above relation between mean activity coefficient of solute, y±, and osmotic coefficient, < >, can be obtained by integration of equation (2.39) from m s = 0 (where y± = 1 and < ) = 1) to m (Pitzer 1991). lny ± = r^dm s +(d)-l) (2.40) J o m 2.4. Pitzer's Activity Coefficient Model Pitzer developed a model to provide improved estimation of electrolyte solution properties by taking into account the effect of short-range forces. The Pitzer's model is based on the virial expansion of the excess Gibbs free energy. The first term in Pitzer's equation is a modification of the Debye-Huckel model for the electrostatic effect. The second term represents short range interaction in the presence of solvent between two solute species (ion-dipole interaction). The third term represents triple interaction

45 29 between solute species in solvent. Pitzer's model can predict the activity coefficients of ions accurately even at high molality of ions (Pitzer, 1991; Zemaitis et al, 1986). Pitzer expressed the excess Gibbs energy as a series of terms in increasing powers of molality to derive his equation (Pitzer, 1991). -f(i) + XSm 1 m J X ij (I) + XEZ m. m J W, K1 j j i j k m ^.jk+- (2.41) where, W w : number of Kg of water mi mj nik : molality of solute i j k I : ionic strength ( I = ~^m;zf ) f(i) : includes the Debye-Hiickel limiting law Xij(I) = X,ji(I) : short-range interaction between solute species i and j in solvent Uijic : triple interaction between solute species i, j and k in solvent The activity coefficient, Yi of solute species i and osmotic coefficient, < ) are obtained by differentiating equation (2.41) as follows. lny i = [3(GVW wrt)/an il (2.42) 4_l = _ ( a G e * / a w j n i / ( R T m i ) (2.43) Rewriting equation (2.41) in terms of experimentally determinable quantities B and C instead of the individual ion quantities X and p. gives the following equation (2.44). G e W RT f(i) + 2 m c m ( m r z r r C C ZZ mm. + ZZ mm, a a aa / J c caa (2.44) +2Y Y m n ml c + 2V Y m n ra,l + 2V V m m B,L n, +ymx m + / J / J n c nc / J / J n a na f J / tl f n n nn / * n nm

46 30 Pitzer derived semi-empirical equations for y and by taking the derivative of equation (2.44). For single electrolyte system, y± and are given by ln Y ± = z M z x f T +m(2v M v x /v)b^ +m 2 [2(v M v x ) 3/2 /v]c MX (2.45) 4.-lHz M z x f*+m(2v M v x /v)bt K +m 2 [2(v M v x ) 3/2 /v]c* lx (2.46) where, y± z M X v : mean activity coefficient : charge : cation : anion : number of ions produced by 1 mole of the solute f = -A4[I 1/2 /(l + bl 1/2 ) + (2/b)ln(l + bl 1/2 )] (2.47) f* = -A4 1/2 /(l + bl 1/2 ) (2.48) A4 : Debye-Huckel osmotic coefficient parameter b : universal parameter with the value of 1.2 (kg.mol) 1/2 B Y =B M X + B*MX (2.49) BMX =PSc+P (^g(a,i,/2 )+PScg(a 2 I,/2 ) (2.50) BL =P^x+3l!_ < exp(-a i r /2 ) + PS c exp(-a 2 I 1/2 ) (2.51) =3C* x /2 (2.52) The P (0) MX, 3 (1) MX, C'MX are tabulated binary parameters specific to the salt MX. The P (2) MX is a parameter to account for the ion pairing effect of 2-2 salts. When either

47 cation M or anion X is univalent, cti = 2.0. For 2-2, or higher valence pairs, cti = 1.4. The constant a 2 is equal to For multi-component electrolyte solutions, y and are given by 1 YM _ Z M a c 20 Mc +2>/F 1 Mca (2.54) a a' In y x = z x F + Z m c [2B cx + ZC cx ] + Z m 20 X a + m c ( cxa c a +ZSm c mj cc, x + Z x _T_>>c m a C (2.55) lnymx = z M z x F + Z m < 2B Ma +ZC Ma +^0 Xa V, +JJm e m,v '[2v M Z c a 2B cx +ZC cx +^0 Mc v x (2.56) d)-l = (2/Zm i )[f*i + ZZ m o m a(bl +ZCJ i c a + ZZm c m c,(ot +X m Xa) + Z Z m a ^ a «. + Z^caa)] c c' (2.57) The quantity F in thefirstterm of equations (2.54), (2.55), and (2.56) includes the Debye- Huckel term given by Also, F = f Y +ZZmcm a B; a +ZZm c m c,0: c,+zzniama.o; a. (2.58) c a c < c' a < a' z = Z m i l z i (2.59)

48 32 B c,=fc ) g'(a 1 I 1/2 )+3L 2) g'(a 2 I 1/2 )]/I (2.60) * ij =e ij + B e ij (o (2.61) (2.62) <D+=0> + K> IJ IJ IJ (2.63) Gij are tabulated mixing parameters specific to the cation-cation or anion-anion pairs. The is also tabulated mixing parameter specific to the cation-anion-anion or anion-cation-cation pairs. The parameters E 6ij(I) and E 0'ij(I) represent the effects of unsymmetrical mixing. These values are significant only for 3-1 or higher electrolytes (Pitzer, 1975). The g(x) and g'(x) are functions accounting for the ionic strength dependence of B M x and B'MX given by g(x) = (2.64) g'(x) = (2.65) The binary parameters P (0), and C* have larger effects than the mixing parameters, *F and 9 (Pitzer, 1991) Isopiestic Method There are several methods to determine the electrolyte thermodynamic properties such as activity coefficient, activity of solvent, and osmotic coefficient. The activity coefficient can be determined by the electromotive force (e.m.f.) method. The activity or osmotic coefficient of the solvent can be determined from the measurement of the vapor pressure of the solvent and from the isopiestic method. The e.m.f. method and vapor

49 33 pressure measurement are not suitable for this study. The e.m.f. method is one of the most precise measurements, but, proper reversible electrodes are not always available for many ions including silicic ions. In the case of vapor pressure measurements by the static method, very precise control of temperature is required because the vapor pressure of the solvent varies rapidly with even very small temperature change. The isopiestic method is simpler to perform and more accurate than direct vapor measurements. There are, however, some limitations for the isopiestic method. The solvent should be the only volatile component in the system. Furthermore, measurements below a molality of 0.1 do not show good reliability (Pitzer, 1991; Thiessen and Wilson, 1987). The osmotic coefficient is particularly useful for treating isopiestic data (Pitzer, 1991). Sample solutions and one or more reference solutions are prepared in separate open containers. For this study, two sample solutions, one reference solution (NaCl), and one standard solution (KC1) were prepared. The initial concentration and mass of each solution should be known. The reference and standard solutions are chosen as the electrolytes whose osmotic coefficient data are already known at that temperature. The data for the reference solution are used to determine the osmotic coefficients of the sample solutions. The standard solution is used to check the reliability of the experiments and calculate experimental errors. The containers are placed in a closed chamber where all solutions share the same vapor phase. The vapor space of the chamber is evacuated to contain only water vapor. Solvent is transferred through the vapor phase. The chamber containing the solutions is kept at isothermal conditions at specific temperature until no more change in the concentration of the solution is observed, thus, thermodynamic equilibrium is reached.

50 34 When the solutions are in thermodynamic equilibrium, then p*w M" w M*w M"w (2.66) where p. w is the chemical potential of water and superscripts, 1, 2, R, and S, refer to sample solution 1, sample solution 2, reference solution, and standard solution, respectively. Thus, the activity of the solvent is the same. In a' w = In a* = In a* = In a* (2.67) Substituting equation (2.67) to equation (2.20) gives the following relationship for the osmotic coefficients: (fr'^v.m, =f V j m. =^2v k m k = 4> 8 5> 1 m 1 (2.68) i j k 1 where, < > is the osmotic coefficient, v is the number of ions produced by 1 mole of the solute, and m is the molality of the solute. The subscripts, i, j, k, and 1, refer to each component in sample 1, sample 2, reference, and standard solutions. Following isopiestic equilibration, the solutions are weighed. The molality of each solution is calculated from the measured weight. Subsequently, the activity of solvent or osmotic coefficient of the standard solution at that molality is calculated by an activity coefficient model using known parameters from the literature. The activity of solvent and the osmotic coefficient of the other solutions at that molality can be determined using equations (2.67) and (2.68).

51 35 CHAPTER 3. MATERIALS AND METHODS Solubility experiments were performed using two types of synthetic liquors (system A and B) and mill liquors to measure the precipitation conditions of sodium aluminosilicates at K (95 C). The concentration levels of NaOH, Na2C03, Na2S04, NaCl, and Na 2 S in synthetic green and white liquors were designed based on those in various mill liquors in Table 1.3. The structure of the precipitates was identified by X-ray diffraction, chemical and thermogravimetry analysis. An isopiestic method was used to measure the osmotic coefficients for the Na 2 Si0 3, Na 2 Si0 3 -NaOH, and NaOH-NaCl-NaAl(OH) 4 aqueous systems at K. 3.1 Solubility Experiments Using Synthetic Liquors of System A The major difference in the chemical composition of green and white liquors is the concentration level of NaOH and Na 2 C0 3 (Smook, 1992). White liquor contains a higher concentration of OH - 2 and lower of C0 3 ' than those in green liquor as seen in Table 1.3. Two synthetic green and white liquor systems were prepared to study the precipitation conditions of sodium aluminosilicate complex in the systems Experimental design System A was prepared by dissolving the following six salts (reagent grade, Fisher Scientific, Vancouver, B.C.) in water: A1C1 3-6H 2 0, Na 2 Si0 3-9H 2 0, NaOH, NaCl, Na 2 C0 3, and Na 2 S0 4. Distilled water was used in all experiments after it was deionized

52 36 andfilteredthrough a 0.05 umfilterin an ELGASTAT UHP water purification apparatus (Fisher Scientific, Vancouver, B.C.). Based on typical mill concentration levels, a base concentration level was formulated and then perturbations around it gave the other compositions. The experimental design is shown in Table 3.1. The effects of the input molar ratio, Al/Si, and concentrations of NaOH, Na2C03, and Na2S04 on the solubility of aluminum and silicon were determined by using system A. Table 3.1. Solubility experiment design for synthetic liquors of system A. Synthetic green liquor Experiment Concentrations in initial solutions (mol/kg H2O) AICI36H2C» Na 2 SiO r 9H 2 0 NaOH Na 2 C0 3 Na 2 S0 4 NaCl Al* A2* A3* A A A A A A Synthetic white liquor A10* All* A12* A A A A A A * base concentration level experiments

53 37 The amounts of AICI36H2O and Na 2 Si03-9H20 that were added into white liquor were double those added in green liquor in order to obtain enough precipitate samples for X-ray diffraction, thermogravimetric, and chemical analysis Experimental procedure Two solutions were prepared by dissolving measured amounts of AlCl3-6H 2 0 and Na 2 Si03-9H_C) in 1 kg of deionized water respectively. Proper amounts of NaOH, Na2C03, Na2S04, and NaCl salts were added to the solutions to make the desired initial conditions for the experiments. The solutions were stirred until all salts were dissolved completely and kept at K (95 C). The solutions were then poured into a vessel made of stainless steel to create a supersaturated solution at the beginning of the experiments. Equilibrium vessels of capacity 4 L and 2.5 L were built for the experiments. A schematic of the vessel is shown in Figure 3.1. The inside of each vessel was coated with Teflon to prevent any chemical attack from the alkaline solutions. A variable speed stirrer was attached to the vessel. The stirrer shaft and blade were made of Teflon-coated steel. A thermocouple probe and a sampling port were built into the lid of the vessel. A Tefloncoated thermocouple probe was used. A sampling tube, which was also made of Teflon, was passed through the sampling port during sampling. An o-ring was placed between the lid and the vessel to prevent evaporation of the experimental solution. A Teflon packing gland was used around the stirrer shaft to prevent evaporation. The vessel containing the solution was covered with a lid. The lid of one vessel was made of stainless steel and coated with Teflon and that of the other vessel was made of plexiglass. The vessel was

54 38 then placed in a water bath at K. The air inside the vessel was replaced with nitrogen gas to prevent undesirable reactions such as those with the carbon dioxide in the air. The solution was stirred during the experiments. Figure 3.1. Equilibrium vessel for the solubility experiments. Thermometer Stirrer Packing Gland Sampling Tube O-Ring Shaft and Blade

55 39 Liquid phase samples were obtained from the vessel using a syringe. Each sample went through a Teflon pipe and a 0.5 um polyvinylidene fluoride membrane filter (Millipore Canada, Nepean, Ontario). It was found by dynamic light scattering (DLS) that all of the precipitate particles were larger than 3 um, meaning all portions of the precipitate were trapped by the filter. The DLS experiments were conducted by using a Malvern Zetasizer III apparatus (Malvern Instruments Inc., Malvern, UK). The filtrates were then analyzed by a GBC 904 Atomic Absorption Spectrophotometer (GBC Scientific Equipment Pty Ltd., Victoria, Australia) using the nitrous oxide-acetylene flame to measure the total soluble Al and Si. The procedure of Atomic Absorption analysis is explained in detail in APPENDIX I. The samples were taken over a period of time until no change in Al and Si concentration was observed. This was taken to be the equilibrium state of the system. It takes about seven days to reach equilibrium. The other approach of solubility experiment by adding sodium aluminosilicate precipitates into an unsaturated solution of Al and Si was not tried because dissolution of the solid precipitate was too slow as it was found by preliminary experiments. In particular, a small amount, g, of precipitates could not be dissolved significantly in 100 ml of alkaline solution containing NaOH of 1 mol/kg H 2 0 after four weeks Identification of the precipitates After equilibrium was reached, precipitates, which deposited on the wall, stirrer blade, and bottom of the vessel were collected, washed with deionized water several times and dried in an oven at 105 C. The structure of the dried precipitates was identified by X-ray diffraction (D5000 Diffractometer, Siemens Aktiengesellschaft, Germany) and

56 40 thermal analysis using a TGS-2 thermogravimetric analyzer (Perkin-Elmer, Norwalk, Connecticut, USA). The software, FIFFRAC / AT v.3.1 (Siemens Industrial Automation, Inc., Madison, Wisconsin, USA), was used to compare the X-ray patterns of samples with those of reference minerals. A portion of the solid materials was dissolved in nitric acid solution for analysis to obtain the molar ratios of Na, Al, and Si using the Atomic Absorption Spectrophotometer (see APPENDIX I). The presence of chlorine and sulfate in the solid were tested by mixing samples of dissolved solid with a AgNC>3 and BaCl 2 solution respectively (Greenberg et al, 1992; Masterton and Slowinski, 1972). The presence of carbonate in the structure of the solids was tested by dissolving the solid in an acidic solution to observe formation of bubbles of CO2 gas (Masterton and Slowinski, 1972). Finally, images of the precipitates were obtained on carbon-coated samples using a Philips XL30 scanning electron microscope (Philips Electronics, Eindhoven, Netherlands) Solubility Experiments using Synthetic Liquors of System B Experimental design The synthetic liquors of system B were prepared to study the effect of Na 2 S. The liquor concentrations were chosen to resemble closely mill liquors from a Kraft pulp mill in B.C., Canada. Two salts, Na 2 S04 and NaCl, were not added into the liquors of system B in order to make the system simpler with only major anions. It is also noted that the concentrations of SO4 2 - and Cl" were very small in the mill liquors compared with those of other anions such as OH", HS", and CO3 2 ' (Magnusson et al, 1979).

57 41 System B was prepared similarly with the following salts: AICI36H2O, Na2Si03-9H20, NaOH, Na2C03, and Na2S. The perturbations around the base concentration level gave the other compositions at which experiments were conducted as seen in Table 3.2. The liquors were prepared by controlling the added amount of Na2S for both green and white liquors. The amount of input Na2S was variedfrom0 to 1.0 mol/kg H2O to observe the effects of HS - and OH - concentrations in synthetic green liquor. It is known that Na 2 S dissociates to Na +, HS - and OH" in H 2 0 according to the following reaction (Smook, 1992): Na 2 S + H 2 0 -> 2Na + + HS" + OH". Table 3.2. Solubility experiment design for synthetic liquors of system B. Synthetic green liquor Experiment Concentrations in initial solutions (mol/kg H2O) AICI36H2O Na 2 Si0 3-9H 2 0 NaOH Na 2 C0 3 Na 2 S Bl* B2* B3* B B B B B B Synthetic white liquor BIO* Bll* B12* B B B B B B base concentration level experiments

58 Experimental procedure and analysis Two solutions were prepared by dissolving measured amounts of AICI36H2O and Na2Si03-9H20 in 1 kg of deionized water respectively. Proper amounts of NaOH, Na2C03, and Na 2 S salts were added to the solutions to make the desired initial conditions for the experiments. The subsequent experiments and analysis procedures are similar to those in sections and Solubility Experiments using Mill Liquors Prior to the solubility experiments, basic information on the mill liquors was obtained and is presented in Table 3.3. The liquor samples were passed through a 0.5 urn polyvinylidene fluoride membrane filter (Millipore Canada, Nepean, Ontario) to remove particles. The concentrations of Al, Si, and K were determined using a GBC 904 Atomic Absorption Spectrophotometer (GBC Scientific Equipment Pty Ltd., Victoria, Australia). The chlorine content was measured by titration with 0.1 M AgN0 3 solution using a Mettler Toledo DL25 titrator (Fisher Scientific, Vancouver, B.C.). A Mettler DM141 silver electrode (Fisher Scientific, Vancouver, B.C.) was used for the titration. The concentrations of NaOH, Na2C03, and Na2S were obtained from the ABC test (Grace et al, 1989) done by the Prince George Lab of the Pulp and Paper Research Institute of Canada. The ABC test is a titration method (TAPPI Standard Method T624 os-68) using HC1 solution to determine the amount of titratable alkali such as NaOH, Na2S, and Na2C03 in green and white liquors (Grace et al, 1989).

59 Experimental design Experiments were performed with samples prepared using the mill green and white liquors and adding appropriate chemicals to simulate: (a) progressive system closure: and (b) observed variations in the concentrations of carbonate and other ions in mill liquors. Table 3.3. Analysis results of mill liquors. C aemical concentrations (mol/kg H 2 0) Al Si K NaOH Na 2 C0 3 Na 2 S Cl Green liquor N.D (NaOH* 0.85) (NaHS* 0.56) White liquor N.D (NaOH* 2.63) (NaHS* 0.57) N.D. : not detected * : The concentrations of NaOH and Na 2 S can also be expressed as NaOH* and NaHS* since 1 mol of Na 2 S dissociates to 1 mol of NaHS and 1 mol of NaOH in water by the following reaction : Na 2 S + H 2 0 -> NaHS + NaOH (Smook, 1992). The detailed experimental design is shown in Table 3.4. First, experiments with base concentration levels were performed using the green liquors (experiments M1,M2, and M3) in Table 3.4. Perturbations around the base-case concentration level gave the composition for experiments M4 to M12 in order to observe the effects of NaOH, Na 2 C03, and Na 2 S on the precipitation conditions of sodium aluminosilicate. In the case of experiments with the mill white liquor, the base concentration level is the composition of liquors in experiments M13, M14, and M15. Again, perturbations around this base concentration level were done and gave the compositions for experiments Ml6 to M24.

60 Experimental procedure and analysis The subsequent experiments and analysis procedures are similar to those in sections and The sodium, aluminum, silicon, and potassium contents in the solid precipitates were analyzed using the Atomic Absorption Spectrophotometer after dissolving the solid samples in nitric acid solution. Table 3.4. Solubility experiment design for mill liquors. vlill green liquor Exp. Concentrations in initial solutions (mol/kg H 2 0) A1C1 3-6H 2 0 Na 2 Si0 3-9H 2 0 NaOH Na 2 C0 3 Na 2 S Ml* M2* M3* M M M M M M M M i l M Mill whil te liquor M13* M14* M15* M M M M M M M M M * base concentration level experiments

61 Osmotic Coefficient Measurement for Na2Si03 and Na2Si03-Na0H Systems Aqueous alkaline silicate solutions contain a variety of silicic ions, orthosilicic 4 3 _ 2 2 (Si0 4 \ HSi0 4, H 2 Si0 4 \ H 3 Si0 4 \ and FLjSiO^ or metasilicic species (Si0 3 \ HSi0 3 ', and H 2 Si03). Several reactions related to the silicic species take place in aqueous medium (Jendoubi et al, 1997). Although thermodynamic calculations have been done to check the stability between orthosilicic and metasilicic species (Babushkin, 1985), the nature of stable silicic species in alkaline solutions, when metasilicate salt is dissolved in water, has not yet been elucidated experimentally. Understanding the nature of silicic ions in alkaline solutions is a prerequisite to the study of electrolyte thermodynamic properties of those ions. A titration method was used to identify the species present when the sodium metasilicate is dissolved in aqueous alkaline solution Identification of silicic species by a titration method An aqueous solution of Na2Si0 3 -NaOH was prepared and titrated using HC1 solution. The concentration of Na2Si03 was 0.25 mol/l and that of NaOH was 1.00 mol/l. The solution was prepared by dissolving proper amounts of Na2Si03-9H20 (sodium metasilicate, Reagent grade, Fisher Scientific, Vancouver, B.C.) and NaOH (Reagent grade, Fisher Scientific, Vancouver, B.C.) in distilled and deionized water. The solution was then stirred using a magnetic bar coated with Teflon for a week at room temperature to allow for any possible reaction such as conversion of metasilicic species to orthosilicic species to occur (Gasteiger, 1988). The HC1 solution of 1 ± mol/l (Fisher Scientific, Vancouver, B.C.) was used as a titrant. The titration was conducted by

62 46 using a Mettler Toledo DL25 titrator (Fisher Scientific, Vancouver, B.C.) equipped with a Mettler DG111-SC ph electrode (Fisher Scientific, Vancouver, B.C.). The ph electrode was calibrated using two buffer solutions of ph 4 and ph 7 purchased from Fisher Scientific. The measured ph values were for NaOH solution of 0.1 mol/l and for that of 0.01 mol/l respectively. Calculated ph values using Pitzer's equation (2.46) were and respectively. This result shows that a reliable measurement was available in high ph range using the calibrated electrode. The titration proceeded until the ph reached a value of 2.5 with an increment of the titrant of 0.05 ml Apparatus and chemicals The isopiestic apparatus of four-neck type for osmotic coefficient measurement was designed by modifying a known three-neck type apparatus (Thiessen and Wilson, 1987). One set of the isopiestic apparatus is shown in Figure 3.2. The neck-type apparatus has been used by other researchers also (Lin et al, 1996; Ochs et al, 1990). This neck-type apparatus has several advantages compared to the conventional one. It has smaller capacity (140 ml). The solutions are in good thermal contact since the heat can be transferred through the glass from the water of the water-bath to the sample solutions. In the case of the conventional isopiestic apparatus, the heat is transferred through the vacuum, and it takes long time to reach the equilibrium. The time to reach the equilibrium can be substantially reduced by using a neck-type apparatus.

63 47 Figure 3.2. The isopiestic apparatus. sample bottle c a P

64 48 Four sets of the isopiestic apparatus were built by the Canadian Blowing Company (Richmond, B.C.) in order to speed up the work. One set of the isopiestic apparatus consists of a four-neck flask equipped with a vacuum stopcock, four sample bottles, and their caps. The capacity of the four-neck flask is 100 ml and each neck has a 14/20 joint of female connector. The four-necks are attached symmetrically to the flask. The high vacuum stopcock is attached to the top center of the flask. The capacity of the flat-bottom sample bottle is 10 ml and the neck has a 14/20 compatible joint of male connector. The water-bath (Model 1187, VWR Scientific, Richmond, B.C.) equipped with a digital temperature controller was used to keep the temperature constant. A circulation unit of the water-bath can minimize the temperature gradient inside the water-bath. The temperature stability of the water-bath was C. The readability of the temperature controller was C. Reagent grade salts (Fisher Scientific, Vancouver, B.C.) were used without further purification. The NaCl, KCI, and NaOH salts were dried in a oven at 105 C for 24 hours before use. Sodium metasilicate, Na 2 Si03-9H20, was used without further drying. Distilled water was used in all experiments after it was deionized and filtered through a 0.05 um filter in an ELGASTAT UHP water purification apparatus (Fisher Scientific, Vancouver, B.C.). The NaCl solution was used as a reference whereas the KCI solution was used as the standard to compare our results with data from the literature.

65 Experimental procedure The salt solutions were prepared to the desired concentrations by weighing and dissolving. Thefirstbottle of each apparatus contained the reference (NaCl) solution. The second bottle contained the standard (KG) solution. The remaining two bottles contained the sample solutions of same initial concentration. The sample solution of Na2Si03 was prepared by dissolving the sodium metasilicate, Na.Si03-9H20, salt in the water. The sample solution of Na2SiC>3-NaOH was prepared by dissolving proper amounts of Na 2 Si03-9H20 and NaOH salts. The molalities of Na2Si03 and NaOH were the same in the sample solution. The presence of Si03 2 ', metasilicic ion in alkaline aqueous solution, when the sodium metasilicate is dissolved, will be explained in section 5.1. The groundglass surfaces of each sample bottle were slightly coated by a silicon vacuum grease. The sample bottles were put into the four-neckflask.the apparatus was evacuated slowly to remove the air and the dissolved gas in the sample solutions using a vacuum pump. Four sets of the apparatus were available. In order to determine the equilibration time, the following procedure was used. The four sets were placed in the water-bath at K. Aiter two days, one of them was taken outfromthe water-bath. The solutions of each sample bottle were weighed. The remaining sets were taken out after two or three more days until no change in the solute concentration was observed. It was then assumed that the total elapsed time is the required equilibrium time. It was found that the equilibrium was reached in five days. It was then decided to set the equilibrium time to seven days. Subsequently, all four sets were used to obtain data. Following equilibration, the solutions were weighed. The molality of each solution was calculatedfromthe measured weight. The osmotic coefficient of the sample

66 50 solution was calculated from the molalities of the sample solution and of the reference (NaCl) solution and from the osmotic coefficient of the reference (NaCl) solution. The osmotic coefficient of the reference NaCl solution was calculated by Pitzer's model using equation (2.54) in section 2.5 with Pitzer's parameters, p (0) Naci = , P (1) Naci = , and c W i = at K (Pitzer, 1991). The measured osmotic coefficient of the KC1 solution was compared to calculated data using published Pitzer's parameters to ensure the accuracy of the experiments (Pitzer, 1991) Osmotic Coefficient Measurement for NaOH-NaCl-NaAl(OH)4 System Apparatus and chemicals The apparatus was the same one used in the previous osmotic coefficient study for the Na2Si03 and Na2Si03-NaOH systems. Reagent grade salts (Fisher Scientific, Vancouver, B.C.) were used without further purification. The NaCl, KC1, and NaOH salts were dried in an oven at 105 C for 24 hours before using. Aluminum chloride hydrate, AICI36H2O, was used without further drying. Distilled water was used in all experiments after it was deionized and filtered through a 0.05 pm filter in an ELGASTAT HHP water purification apparatus (Fisher Scientific, Vancouver, B.C.). The NaCl solution was used as reference whereas the KC1 solution was used as standard to compare our results with data from the literature. Aluminum ion speciation in aqueous solutions changes with both ph and concentration. In dilute solutions at ph above 9, it is commonly assumed that aluminate ion, Al(OHV, is the predominant one. The formation of other aluminum ions such as

67 51 Al20(OH)6 2 ' and Al(OH)6 3 " was not considered for this study since these ions can be present only at concentrations of aluminum above 1.5 mol/l and at extremely high ph (Swaddle etal, 1994; Pokrovskii andhelgeson, 1995). The sample NaOH-NaCl-NaAl(OH)4 aqueous solution was prepared by dissolving a proper amount of A1C13-6H20 in a NaOH aqueous solution. The concentrations of aluminum and hydroxyl ions were such that formation of other aluminum ions except for Al(OH)4' was prevented. In all solutions, the input molality ratio of AICI36H2O to NaOH was less than 1 : 5 so that all aluminum salts were completely dissolved and were present as Al(OH)4" ion. Thus, the species present in the sample solutions were assumed to be Na +, OH", Cl", and Al(OH)4" only Experimental procedure The experimental procedure for NaOH-NaCl-NaAl(OH)4 system was similar to that of the previous osmotic coefficient study for the Na2Si03 and Na2Si03-NaOH systems. Briefly, an experiment proceeded as follows. The solutions were prepared to the desired concentrations by weighing and dissolving. The first bottle of each apparatus contained the reference (NaCl) solution. The second bottle contained the standard (KCI) solution. The remaining two bottles contained the sample solutions of the same initial concentration. The ground-glass surfaces of each sample bottle were slightly coated by a silicon vacuum grease. The sample bottles were put into the four-neck flask. The apparatus was evacuated in order to remove the air and the dissolved gas in the sample solutions using a vacuum pump. The solutions in the apparatus were equilibrated in a well thermostated condition at K.

68 52 At equilibrium, the chemical potential of water is the same in the two sample solutions and in the reference and standard solution. This condition may be expressed as follows (Pitzer, 1991) T v m i i = * 2 I v i m i = * R I v A =4 S >. m. (2-54) i j k I where, <J> is the osmotic coefficient, v is the number of ions produced by 1 mole of the solute, and m is the molality of the solute. The superscripts, 1, 2, R, and S, refer to sample solution 1, sample solution 2, reference solution, and standard solution. The subscripts, i, j, k, and 1, refer to each component in sample 1, sample 2, reference, and standard solutions. Equilibrium in each experiment was reached in seven days. The measured osmotic coefficient of the KC1 solution was compared to calculated data using published Pitzer's parameters to ensure the accuracy of the experiments (Pitzer, 1991). Following equilibration, the solutions were weighed. The molality of the solutes in each solution was calculated from the measured weight. The osmotic coefficients of the sample solutions were then calculated from the measured molalities of the sample solutions and of the reference (NaCl) solution, and from the osmotic coefficient of the reference (NaCl) solution. The osmotic coefficient of the reference NaCl solution was calculated by Pitzer's model using equation (2.54) in section 2.5 with Pitzer's parameters, P (0) Nad= , P (1) Naci = , and C* Na ci = at K (Pitzer, 1991). The measured osmotic coefficient of the KC1 solution was compared to calculated data using published Pitzer's parameters (Pitzer, 1991) to ensure the accuracy of the experiments.

69 53 CHAPTER 4. THERMODYNAMIC MODELING OF SODIUM ALUMINOSILICATE FORMATION A thermodynamic equilibrium model has been developed to predict sodium aluminosilicate precipitation conditions in alkaline solutions at K. A method based on the equilibrium constants of the solid formation reactions was used for this modeling (Anderson and Crerar, 1993). Pitzer's method was adopted to calculate the activity of water and ion activity coefficients Model Equations The predominant aluminum and silicon species in solution were assumed to be 2 Al(OH)4' and Si0 3 ' respectively because the system is very alkaline (ph>13) (Babushkin et al, 1985; Pokrovskii and Helgeson, 1995; Swaddle et al, 1994). Thus, the model considers that eleven species are present at equilibrium at K and 1 atm: two solid species (Na 8 (AlSi0 4 ) 6 Cl 2-2H 2 0 and Na 8 (AlSi0 4 ) 6 (OH)2-2H 2 0), one liquid (H 2 0), and eight ions (Na + 2 2, Al(OH) 4 \ Si0 3 ', OH', C0 3 ', S0 4 2", HS", and Cl). The model equations consist of the reaction equilibrium equations for sodalite dihydrate and hydroxysodalite dihydrate, one charge balance equation, and the mass balance equations for the elements of Na, Al, Si, Cl, O, H, S, and C. Formulas for sodalite dihydrate and hydroxysodalite dihydrate formation in alkaline aqueous solutions can be written as following (Zheng et al, 1997).

70 54 8Na + (aq)+ 6Al(OH) 4(aq) + 6Si (aq) + 2Cr (aq) <-> Na_(AlSi0 4 ) 6 Cl 2-2H 2 0 (8) + 4H 2 0 (1) + 120H- (aq) (4.1) 8Na + (aq)+ 6Al(OH)4 _ (aq)+ 6Si (aq) + 20H" (aq) Na.(AlSi0 4 ) 6 (OH) r 2H 2 O w + 4H 2 0 (,)+ 120H- (aq) (4.2) Assuming the solids to be pure then their activity is unity and the thermodynamic equilibrium constant of sodalite dihydrate, K so _, can be written as follows K a Hp( m QH-'V'oH-) ^ 3) «a - ) «( O H ), YL (0 H), )«O 3 >- ylo? X m cr ylr ) where, a is the activity, m is the molality, and y is the activity coefficient of the species. Similarly, the equilibrium constant of hydroxysodalite dihydrate, Khsod, can be written as follows K A H 2 Q ("WVoH-) (A 4N ( m Na^Na+ )( M A 1 (OH), YAKOHK X m SiO>-Y S io?- ^ OW ) The mass balance equations for the Na, Al, Si, Cl, O, H, S, and C elements are given by ntoui = ( "^o ) ( ^ ^ 1000/ N a sod hsod = ^OOOnS.OlS^^^ + 6 m m «} ( 4 6 ) n ^ = ( id7^ ) ( m c m -) ^ " r = + ^OOo7 oi5 )(4m^ "><- + 3 "W + m o»- + 3 m coi- + 4m soj- (4.9) + 26m sod +28m hsod ) nr = 2n H20 +( 1 0 Q Q n / H )(4m A 1 ( O H ), + m oh _ + m Hs. +4m sod + 6 m h s o d ) (4.10)

71 55.OUl =, V w v ( 4 J jx s 1000/ soj HS v ; tow =, "H^O ) ( ^ ( ) c V 1000/ A c o ' ^ V ' The charge balance equation can be written as follows. m + = m,^tin. +2rn + ni _ + 2m,_ + 2m + m. _ + ni (4.13) Na + Al(OH) 4 SiOJ OH CO SOJ HS Cl where, n is the number of moles, superscript total stands for total number of moles of v ' element, subscript sod indicates sodalite dihydrate, subscript hsod indicates hydroxysodalite dihydrate, and m is the molality of the species Structure of the Thermodynamic Model A structure of the modeling is shown in Figure 4.1. The total amount of chemicals, Pitzer's parameters, and equilibrium constants of sodalite dihydrate and hydroxysodalite dihydrate formations were given for the modeling calculation. Model parameters such as Pitzer's parameters and equilibrium constants are discussed in more detail in Chapter 7. An initial guess for the molality of each species was also given as input. Water activity and activity coefficients of species were calculated using the Pitzer's equation in the subroutine ACTCOF. Pitzer's parameters used for this modeling will be discussed in section 7.1. The molalities, activity coefficients, and water activity were substituted into the model equations in the subroutine FUNCV. The model equations were a set of non-linear equations. A Newton-Raphson method (Numerical Recipies, 1995) was used to solve the non-linear equations. A constraint that the molality of any species can not be negative was included in the model. The molalities of

72 56 species at equilibrium were calculated by solving the model equations. The molalities of AJ(OHV and Si03 2 ' at equilibrium define the precipitation conditions. A computational source code (kshsod.for) of this modeling is given in APPENDIX IX together with examples of input and output. Figure 4.1. A block diagram of the thermodynamic modeling. INPUT Total amount of chemicals Pitzer's parameters Equilibrium constants (K^ and K^) Initial guess for molalities of species FUNCV Model equations, (4.3)~{4.13) ACTCOF Calculation of activity of water and activity coefficients using Pitzer's equations (2.40)~(2.43) OUTPUT Molalities of species (Na\ AI(OH)4", Si0 3 2-, OH", C03 2 -, S04 2 ", HS", CI-, H20, hsod, sod)

73 57 CHAPTER 5. OSMOTIC COEFFICIENT DATA FOR Na 2 Si0 3, Na 2 Si0 3 -NaOH, AND NaOH-NaCl-NaAI(OH) 4 AQUEOUS SYSTEMS Prior to the determination of Pitzer's parameters relevant to SiC»3 2 " and Al(OH)4', the osmotic coefficient data for the Na2SiC»3, Na 2 Si0 3 -NaOH, and NaOH-NaCl- NaAl(OH)4 aqueous systems were measured at K and the identification of silicic species in solution was done Identification of Metasilicic Species by Titration A titration curve of Na2Si0 3 -NaOH solution is shown in Figure 5.1. In the figure, 20 ml of the solution containing Na2SiC«3 of 0.25 mol/l and NaOH of 1.00 mol/l was titrated with a HC1 solution of mol/l. If we assume that metasilicic ion, Si03 2 ", and hydroxyl ion, OH', are the predominant ions in the solution and orthosilicic ion, HSi04 3 ', is not present, then the strong base, OH', should be titrated first. The Si03 2 " ion, then, should be converted to HSi0 3 " ion which in turn should be converted to H2Si03 by the reactions (5.1) and (5.2) with the addition of the titrant (Babushkin et al, 1985; Harris, 1991). Thus, three equivalence points should be displayed on the titration curve. Si0 3 2' + It -> HSi0 3 " (5.1) HSi0 3 ' + It -> H 2 Si0 3 (5.2) The first equivalence point, V e NaOH of 20 ml should correspond to the OFT since the OH" of 1.00 mol/l in the sample solution of 20 ml is titrated with 20 ml of HC1

74 58 solution of 1.00 mol/l. The second equivalence point, V e i should be observed at 25 ml since 5 ml of HC1 solution of 1.00 mol/l is needed to convert 20 ml of SiC^2' of 0.25 mol/l to the same amount of HSi0 3 '. The third equivalence point, V e 2 should be observed at 30 ml since 5 ml of HC1 solution of 1.00 mol/l is needed to convert 20 ml of HSiCV of 0.25 mol/l to the same amount of F^SiC^. Figure 5.1. Titration curve for the Na2Si03-NaOH solution with HC1 solution HCI solution (ml)

75 59 The ph values corresponding to the volumes of HC1 at which [Si03 2 "] equals [HSi0 3 '] and [HSi0 3 '] equals [H 2 Si0 3 ] are the pk a values of the reactions (5.1) and (5.2) according to the Henderson-Hasselbalch equation (Harris, 1991). The pk a values for the reactions (5.1) and (5.2) were calculated using the data of standard Gibbs free energy of formation, AGf (Babushkin, 1985). The standard Gibbs free energy change, AG 0 of the reaction (5.1) is equal to AG f (HSi0 3 ') - AG f (Si0 3 2') = kcal/mol - ( kcal/mol) = kcal/mol. The pk a2 of the reaction (5.1) is equal to log K(5.i) = - AG7(2.3RT) = -(-16.1 kcal/mol)/(2.3 x kcal/mol K x K) = Similarly, the standard Gibbs free energy change, AG 0 of reaction (5.2) is equal to AGf (H 2 Si0 3 ) - AG f (HSi0 3 ') = kcal/mol - ( kcal/mol) = kcal/mol. The pk a i of the reaction (5.2) is equal to log K<5.2) = -AG7(2.3RT) = -(-13.2 kcal/mol)/(2.3 x kcal/mol K x K) = Thus, the titration curve should pass from the following two points : (22.5 ml, ph 11.82), (27.5 ml, ph 9.69). If we assume that the metasilicic ion, Si0 3 2' is converted to orthosilicic ion, HSi0 4 ' by the reaction, Si0 3 " + OH' -> HSi0 4 " (Gasteiger, 1988), then the HSi0 4 3' of 0.25 mol/l and the OH* of 0.75 mol/l are present. In this case, the strong base, OH" 3 should be titrated first and the HSi0 4 ' ion should be converted to H 2 Si0 4 2" ion by the reaction (5.3) with the addition of the titrant. Subsequently, the H 2 Si0 4 2" ion is converted to H 3 Si0 4 ' ion which in turn is converted to H 4 Si0 4 according to the reactions, (5.4) and (5.5) with the addition of the titrant (Babushkin, 1985; Harris, 1991). Thus, four equivalence points should be present on the titration curve.

76 60 3 HSi0 4 " + rf -> H 2 Si0 2- ; 4 (5.3) 2 H 2 Si0 4 " + H+ -> H 3 Si0 4 ' (5.4) H 3 Si0 4 " + IT H4Si0 4 (5.5) The first equivalence point of 15 ml corresponds to the OFT since the OH" of 0.75 mol/l in the sample solution of 20 ml is titrated with 15 ml of HCl solution of 1.00 mol/l. The second equivalence point should correspond to 20 ml since 5 ml of HCl 3 solution of 1.00 mol/l is needed to convert 20 ml of HSi0 4 " of 0.25 mol/l to the same amount of H2Si0 4 2". The third equivalence point should be observed at 25 ml according to the conversion of H2Si0 4 2" to H 3 Si0 4 " and the fourth equivalence point should be shown at 30 ml according to the conversion of H 3 Si0 4 " to H4Si0 4. The pk a values of the reactions (5.3), (5.4), and (5.5) are 12.00, 11.70, and 9.77 respectively (Babushkin, 1985). Thus, the titration curve should pass from the following three points: (17.5 ml, ph 12.00), (22.5 ml, ph 11.70), and (27.5 ml, ph 9.77). As seen from Figure 5.1, the titration curve did not pass from the points: (17.5 ml, ph 12.00), (22.5 ml, ph 11.70) which correspond to the pk a values of orthosilicic 3 species (HSi0 4 ", H2Si0 4 2", H 3 Si0 4 ", and H»Si0 4 ). The curve, however, passed from (22.5 ml, ph 11.82) and (27.5 ml, ph 9.69) which correspond to the pk a values of metasilicic species (Si0 3 2", HSi0 3 ", and H2Si0 3 ). Only two equivalence points are shown that correspond to the volumes of 25 ml and 30 ml. The first equivalence point at 20 ml is 2 not significant. It can be explained by the reaction: Si0 3 ' + H2O <-» HSi0 3 " + OH". There 2 are two predominant anions, OH" dissociated from the NaOH, and Si0 3 ' from the Na2Si0 3, in the initial solution. The concentration of the strong base, OH" decreases with

77 61 the addition of the titrant, hydrochloric acid. As the concentration of OH" decreases, more Si03 2_ ions react with H_0 and produce HSiCV and OK to establish the equilibrium. The standard Gibbs free energy change, AG, of the reaction: SiC>3 2 ' + H_0 <-> HSiCV + OH" is equal to AG f (HSi0 3 ") + AG f (OH") - AG f (Si0 3 2") - AG f (H 2 0) = kcal/mol + ( kcal/mol) - ( kcal/mol) - ( kcal/mol) = +3.0 kcal/mol. The log K of the reaction is equal to -AG 0 /(2.3RT 0 ) = -3.0/(2.3 x kcal/mol K x K) = Thus the equilibrium constant, K is io' Since the equilibrium constant, 10" 2.2on Qf t n e r e a c tj o n i s high, the amount of OH" ions produced by the above reaction is enough to hide an inflection of the titration curve at the equivalence point. A calculated titration curve by a method available in the literature (Harris, 1991) is shown in Figure 5.2 with explanations in APPENDIX IV. As seen in the figure, the inflection at the equivalence point is not significant. Another calculated titration curve, when 20 ml of a solution containing Na 2 C03 of 0.25 mol/l and NaOH of 1.00 mol/l was titrated with a HCl solution of 1 mol/l, is shown in the same figure. This curve shows the effect of the equilibrium constant on the inflection point. The equilibrium constant, 10" 36769, of the reaction, C0 3 2" + H 2 0 <-> HC0 3 " + OH" gave a significant inflection at the equivalence point. Based on the above analysis, we conclude that when sodium metasilicate (Na 2 Si03-9H 2 0) is dissolved in water containing OH", the predominant species is metasilicic ion, Si03 2 ". Furthermore the metasilicic ion is not converted to orthosilicic ion (HSi04 3 ") under these conditions.

78 62 Figure 5.2. Calculated titration curve for the Na 2 Si0 3 -NaOH and Na 2 C0 3 -NaOH solutions with HC1 solution.

79 Mole Fraction of Metasilicic Species In addition to SiC>3 2 ', other anions such as OFT and HSiGV are also present, when the metasilicate salt (Na2Si03-9H20) is dissolved in water. The following equilibrium between each metasilicic species in the solution is established. Si0 3 2" + H 2 0 <-> HSiCV + OH* (5.6) Mole fractions of metasilicic species at various ph levels for the two solutions used in the osmotic coefficient experiments, Na2Si03 and mixed Na2Si03-NaOH, are shown in Figure 5.3. The ph values of the solutions were measured. The mole fractions of metasilicic species with varying ph were calculated by the method described in the literature (Lindsay, 1979) using the calculated equilibrium constants for the reactions (5.1) and (5.2). The mole fractions of H2Si03 and HSi03* are very low compared to that of Si03 2 ' in the concentration range of the osmotic coefficient experiments. Due to this reason, the existence of H2Si03 and HSKV can be ignored. Thus, Na2Si03 solution can be assumed to behave as the single electrolyte system of Na + -Si03 2 " and the mixed Na2Si03-NaOH solution can be assumed to behave as a ternary electrolyte system of Na + -Si0 3 -*-OH\ 5.3. Osmotic Coefficient Data for Na 2 Si0 3 Aqueous System Tables 5.1 contains the osmotic coefficient and water activity data and their uncertainties (standard deviations, a$) from eight experiments respectively for KCI (standard solution), Na2Si03 (sample solution 1), and Na2Si03 (sample solution 2).

80 Figure 5.3. Mole fraction of metasilicic species with ph. ph ph range of osmotic coefficient experiments llll Na 2 Si0 3 system Na 2 Si0 3 -NaOH system

81 65 The uncertainty of the measured osmotic coefficient was calculated in APPENDIX V from the uncertainty of the weight measurement, ± g, and uncertainty in osmotic coefficient of the reference NaCl solution, 0.01 (Pitzer, 1991), by a method available in the literature (Baird, 1995; Holman, 1994). The uncertainty was found to increase with decreasing molality of the standard KC1 and the sample solutions. The osmotic coefficient data and predictions by the Pitzer's model are shown in Figure 5.4. Comparison of the osmotic coefficient data of KC1 standard solutions with prediction by Pitzer's model using the published parameters shows the accuracy of the isopiestic method and our four-neck apparatus. The values of the published Pitzer's parameters are: P (0) KCI= , P (1) KCI = , and C* K ci = at K (Pitzer, 1991). The relative percent error, Rel Err %, was calculated by comparing the measured osmotic coefficients of the KC1 solutions with the calculated ones. For the entire set of experiments, the average relative percent error in the osmotic coefficients of KC1 (standard solution) was 0.26 % Osmotic Coefficient Data for Na2Si0 3 -NaOH Aqueous System Table 5.2 contains the osmotic coefficient and water activity data and their uncertainties from thirteen experiments respectively for KC1 (standard solution), Na 2 Si03 - NaOH (sample solution 1), and Na 2 Si03 - NaOH (sample solution 2). The uncertainties were also calculated and found to be increasing with decreasing molality. The osmotic coefficients are shown in Figure 5.5 together with values computed by using Pitzer's model. For the entire set of experiments, the average relative percent error in the osmotic coefficients of KC1 (standard solution) was 0.17 %.

82 66 Table 5.1. Osmotic coefficients and water activities for the Na2Si03 aqueous system at K. cole Salt Molality * Rel Err, % a w ** Rel Err, % a * w 1 NaCl KCI Na 2 Si Na 2 Si NaCl KCI Na 2 Si Na 2 Si NaCl KCI Na 2 Si Na 2 Si NaCl KCI Na 2 Si Na 2 Si NaCl KCI Na 2 Si Na 2 Si NaCl KCI Na 2 Si Na 2 Si NaCl KCI Na 2 Si Na 2 Si NaCl KCI Na 2 Si Na 2 Si * Relative Percent Error (Rel Err, %) = If"" - f H x 100 / ** Relative Percent Error (Rel Err, %) = aw"* - a w I x 100 / a w '

83 67 Figure 5.4. Osmotic coefficients ofna2si03 and KC1 aqueous solutions at K. Molality (mol/kg H 2 0)

84 68 Table 5.2. Osmotic coefficients and water activities for the Na 2 Si0 3 -NaOH aqueous system at K. Salt Molality ^calc * Rel Err, % calc o 6 aw aw * Rel Err,% 1 NaCl KC Na 2 SiO r NaOH Na 2 SiO r NaOH NaCl KC Na_SiO.-NaOH Na 2 SiO r NaOH NaCl KC Na 2 Si0 3 -NaOH Na 2 SiO r NaOH NaCl KC Na 2 SiOj-NaOH NajSiOj-NaOH NaCl KC Na 2 Si0 3 -NaOH Na 2 Si0 3 -NaOH NaCl KC Na 2 SiO r NaOH Na 2 SiO r NaOH NaCl KC Na 2 SiO r NaOH Na 2 Si0 3 -NaOH NaCl KC Na 2 Si0 3 -NaOH Na 2 SiO r NaOH NaCl KC Na 2 SiO r NaOH Na 2 Si03-NaOH NaCl KC Na 2 SiOj-NaOH Na 2 SiO r NaOH NaCl KC Na 2 Si0 3 -NaOH Na 2 Si0 3 -NaOH NaCl KC Na 2 Si0 3 -NaOH Na 2 SiO r NaOH NaCl KC Na 2 SiO r NaOH Na 2 Si0 3 -NaOH * Relative Percent Error, (Rel Err, %) = f ** - fh x 100 / f ** Relative Percent Error, (Rel Err, %) = a w exp - a w I x 100 / a w '

85 69 Figure 5.5 Osmotic coefficients of mixed Na2Si03-NaOH and KCI aqueous solutions at K. / Na 2 Si0 3 -NaOH o measured calculated by Pitzer's model with mixing parameters calculated by Pitzer's model (mixing parameters = 0) i i i i i i i i i i i I I I I I I I I I i i i i i i i i i i i r-p Molality (mol/kg H 2 0)

86 Osmotic Coefficient Data for NaOH-NaCl-NaAI(OH) 4 Aqueous System Nineteen experiments were performed. The osmotic coefficient data for the KCI standard solution and the two sample solutions were obtained from each set of experiments. Table 5.3 shows the osmotic coefficients for the reference (NaCl) and standard (KCI) solutions. Table 5.4 shows the results for two sample solutions of interest. The uncertainty (standard deviation, of the measured osmotic coefficient in Tables 5.3 and 5.4 was calculated from the uncertainty of the weight measurement, ± g, and uncertainty in osmotic coefficient of the reference NaCl solution, 0.01 (Pitzer, 1991), by a method available in the literature (Baird, 1995). In both tables, the calculated osmotic coefficients are also shown. In Table 5.3, comparison of the osmotic coefficient data of KCI standard solutions with calculated values obtained by Pitzer's model shows the accuracy of the isopiestic method of this study. For the entire set of experiments, the overall average relative percent error in the osmotic coefficients of KCI was about 0.20 %. The following values of published Pitzer's parameters, P (0) KCI = , p ( 1 ) Kci = , and C* K ci = at K were used (Pitzer, 1991) to calculate the osmotic coefficient for the KCI solution.

87 Table 5.3. Osmotic coefficients of NaCl and KCI as reference and standard solutions at K. Exp. No. Salt Molality * ^calc *Rel Err, % 1 NaCl KCI NaCl KCI NaCl KCI NaCl KCI NaCl KCI NaCl KCI NaCl KCI NaCl KCI NaCl KCI NaCl KCI NaCl KCI NaCl KCI NaCl KCI NaCl KCI NaCl KCI NaCl KCI NaCl KCI NaCl KCI NaCl KCI * Relative Percent Error (Rel Err, %) = If"" - f H x 100 / tf*

88 72 Table 5.4. Osmotic coefficients of NaOH-NaCl-NaAl(OH)4 aqueous solutions at K. Exp. No. **input ratio Molality *Rel Err, % NaOH NaCl NaAl(OH) * Relative Percent Error (Rel Err, %) = f " - f H x 100 / f " ** Input ratio : Input molality ratio of A1C1 3-6H 2 0 to NaOH

89 73 CHAPTER 6. DETERMINATION OF PITZER'S PARAMETERS FOR Na 2 Si0 3, Na 2 Si0 3 -NaOH, AND NaOH-NaCl-NaAl(OH) 4 AQUEOUS SYSTEMS Since the sodium aluminosilicate formation system contains the Si03 2 " and Al(OH)4* ions, knowledge of the relevant Pitzer's model parameters is required for the thermodynamic modeling. Unknown binary and mixing parameters of Pitzer's model were obtained using the osmotic coefficient data described in the previous chapter 5. Pitzer derived the semi-empirical equations of the osmotic coefficients and activity coefficients for the single and multi-component electrolyte systems (Pitzer, 1991). Pitzer's method is one of the most popular ones that can be used to calculate electrolyte thermodynamic properties for single and multi-component systems. This method works well even at high molalities (Pitzer, 1991; Zemaitis et al., 1986). For this reason, the Pitzer's method was adopted for the modeling Pitzer's Parameters for Na2SiOs and NaiSiOs-NaOH Aqueous Systems In Pitzer's model, there are four binary parameters, P (0) ca, P (1) ca, P (2) ca, and C* ca and eight mixing parameters, (W, B^-, E 6cc'(I), E 9aa'(I), E 6'cc'(I), ^'^'(I), ^cc* a, and ^caa-. The parameter, P (2) is important only for 2-2 or higher valence electrolyte (Pitzer, 1991). The effects of the parameters, E 9 CC <I), ^-(I), E 9'cc<I), and ^-(l), are significant only for 3-1 or more unsymmetrical types of mixing (Pitzer, 1975). These five

90 74 parameters, B^, E 0 CC <I), E Q^(l), E 9'cc(I), and E 0'aa-(I), were ignored for the modeling of the systems Na 2 Si0 3 and Na 2 Si0 3 -NaOH. The three Pitzer's parameters, p (0), P (I), and C*, were determined by least squares optimization using the experimental osmotic coefficient data of Table 5.1 for the Na 2 Si0 3 system. The Gauss-Newton method (Bard, 1974; Englezos, 1996) was used to minimize the following least squares objective function S(P (0),p (1),C+) = ( ( ) d 2 ( 6 1 > d=l where < ) exp is the measured and <J> calc is the calculated osmotic coefficient using equation (2.46). is the uncertainty. Sensitivity of Pitzer's parameters for Na 2 Si0 3 aqueous system is described in APPENDIX VIII. The computational code (siwpam.for) for the optimization is given in APPENDIX IV together with input and output files. Two sets of the binary parameters of NaOH and Na 2 Si0 3 systems and two mixing parameters, 6 - ov,i- and *F T +-. c..j. are required for the modeling of Na 2 Si0 3 -NaOH OH S1O3 Na OH S1O3 system. The binary parameters of Na2Si0 3 were already obtained as explained above and those of NaOH, p l NaOH = , P (1) NaOH , and C* N,OH = at K are available in the literature (Pitzer, 1991). The remaining two Pitzer's mixing parameters, 0 OH_ sjoi_ and ^ N l l + O H - s i o i - were determined by the least squares optimization method using the osmotic coefficient data of Table 5.2 for the Na2Si0 3 -NaOH system. The computational code (nasiwpam.for) for the optimization is given in APPENDIX IX together with input and output files. The binary and mixing parameters that were obtained are given in Table 6.1 with their standard deviations. The standard deviations were calculated according to methods described in the literature (Box et al, 1978). The

91 75 calculated osmotic coefficients using Pitzer's parameters have an average relative percent error of 0.33 % for Na2SiC<3 and 1.74 % for the Na2Si03-NaOH system respectively. The computational source codes (siwcal.for and nasiwcal.for) used for the calculations are available in APPENDIX IX. In Figures 5.4 and 5.5 found in the previous chapter, the solid lines are the calculated osmotic coefficients using the Pitzer's parameters obtained in this work. Minor inflections on the curves of calculated osmotic coefficients are observed at low molalities, 0 to 0.1 mol/kg H2O in the figures. The osmotic coefficient curve of K_Pt(CN)4 by Pitzer's model also shows a similar inflection at low molalities (Pitzer, 1991). Furthermore, it is known that the isopiestic method does not give reliable results below a concentration of 0.1 mol/kg FfeO (Pitzer, 1991). Table 6.1. The Pitzer's parameters of Na 2 Si0 3 and Na 2 Si0 3 -NaOH systems at K Binary parameters for the Na^SiOs system Parameter Standard Deviation p (0) = p (1) = C* = Mixing parameters for the Na2SiOs-NaOH system Parameter t_i n.- = OH SiOJ +. = Na OH Siof Standard Deviation

92 76 Pitzer introduced the mixing parameters, 0 and *F to account for the differences of ion interactions between multi-component and single salt electrolyte solutions. Thus, the mixing parameters have smaller effects than the binary parameters, p (0), P (1), and C* have (Pitzer and Kim, 1974). The contribution of the mixing parameters, however, was not negligible for the NaOH-Na2Si03 system. The dashed line in Figure 5.5 represents the calculated osmotic coefficients when the values of the mixing parameters, 0.-_. - and ^Na + oh-siofa r e s e t e c l u al t 0 z e r o- As shown in this Figure, the mixing parameters should be included for an improved calculation of the osmotic coefficient. Mean activity coefficients of Na2SiC»3 in water at K were calculated and are shown in Figure 6.1. Mean activity coefficients of Na2SiC>3 and NaOH in the Na 2 Si03- NaOH binary system were also calculated and are shown in Figure 6.2. Sensitivity of Pitzer's parameters for Na2Si03-Na0H aqueous system is described in Appendix VIII Pitzer's Parameters for NaOH-NaCI-NaAl(OH)4 Aqueous System Among the Pitzer's parameters, the parameter, P (2) ca, is important only for 2-2 or higher valence electrolytes. The parameters, E 0ee(I), E 0aa*(I), E 0'c C (I), and E 0' '(I), are significant only for 3-1 or for cases with more unsymmetrical mixing (Pitzer, 1975). Thus, these five parameters, P (2) c_, E 0ec'(I), E 0 '(I), E 0'cc'(I), and E 0' '(I), were ignored for the modeling of the system NaOH-NaCl-NaAl(OH)4 in water. The Gauss-Newton method was used to minimize the following least squares objective function N / i exp icak\2 S(pW > p(').c*) = _r^ d d=i 2 ) (6-1) cr.

93 77 Figure 6.1. Mean activity coefficients of Na2SiC«3 in Na2SiC«3 aqueous solution at K. 1.2 _ 10 C Q) O 0.8 O O 0.6 c CU H Na 2 Si "i i i i i i i i i i i i i i i i i i i i i i i i i i f Molality (mol/kg H 2 0)

94 Molality of Na 2 Si0 3 -NaOH (mol/kg H 2 0)

95 79 where <J> exp is the measured and (j) 03 ' 0 is the calculated osmotic coefficient using equation (2.46). c$ is the uncertainty. Three sets of the binary parameters, B ( 0 ), p (1), and C* of NaOH, NaCl, and NaAl(OH)4 systems and six mixing parameters, Q, 0 OH"Al(OH); ' Q Cl-Al(OH)7 ' ^oh-cr> VQH-AKOHK' A N D "WAKOH); are required for the modeling of the NaOH- NaCl-NaAl(OH)4 aqueous system. The binary parameters of the NaOH and NaCl systems and two mixing parameters and are available in the literature (Pitzer, 1991) as shown in Table 6.2. Table 6.2. The Pitzer's parameters at K available in the literature (Pitzer, 1991). Binary parameters System Parameter 3 (0) = NaOH p (1) = C* = p (0) = NaCl p (1) = C* Mixing parameters 0 = oircr Y + = First, the three Pitzer's parameters, P (0), p (1), and C* for the NaAl(OH) 4 system were determined by least squares optimization using the experimental data of Table 5.4

96 80 and taking into account the published Pitzer's parameters shown in Table 6.2. The estimated parameters are shown in Table 6.3 with their standard deviations. The mixing parameters, Q OH - M(OH) -, Q Ci - M(0H) -, ^ o i i - M m ) V and V C R A 1 ( 0 H ) - were set equal to zero during this optimization. Table 6.3. The Pitzer's parameters of NaOH-NaCl-NaAl(OH)4 aqueous system at K obtained in this study. Binary parameters of NaAl(OH)4 Parameter Standard deviation P (0) = p (1) = C* = Mixing parameters Parameter 0 = OH-Al(OH); Standard deviation = = Na + OH~AI(OH)i + = Na + CPAl(OH) After the determination of the binary parameters, the four Pitzer's mixing parameters, Q 0H - M(0H) -, 6 cra1(oh) -, V^-^-, and ^ N a + c r A 1 ( O H ) - were determined by

97 81 least squares optimization method using the osmotic coefficient data of Table 5.4 for the aqueous NaOH-NaCl-NaAl(OH)4 system. It was necessary to follow this approach because the simultaneous minimization search for the binary and ternary parameters did not converge. The mixing parameters that were obtained are also given in Table 6.3 with their standard deviations. The standard deviations were calculated according to methods described in the literature (Box et al, 1978). The calculated osmotic coefficients using Pitzer's parameters have an average relative percent deviation of 0.37 % from the experimental values. Sensitivity of Pitzer's parameters for NaOH-NaCl-NaAl(OFf)4 aqueous system is described in APPENDIX VIII Reliability of the Pitzer's Parameter Determination In order to check the reliability of the optimization procedure, Pitzer's parameters for the NaTcC«4, NaTcCVNaCl, and NaBr-NaC104 systems were obtained using the osmotic coefficient data from the literature (Hernandez-Luis, 1996; Konnecke, 1997). The parameters that we obtained were compared with the published ones. The published parameters for the systems are tabulated in Table 6.4 together with those obtained in this work. Except for the binary parameters for the NaTcC«4 system, all parameters obtained agree well with the published ones. Table 6.5 displays the published osmotic coefficient data and the calculated values for NaTcC>4 system. As seen from the table, the osmotic coefficients calculated by using the parameters obtained in this work match the experimental data slightly better than those obtained by using the published parameter values.

98 82 Table 6.4. The Pitzer's parameters of NaTcCu, NaTc0 4 -NaCl, and NaBr-NaClC>4 systems at K. Binary parameters for the NaTc04 system Parameter Konnecke, 1997 This work p(0) d) c* Mixing parameters for the NaTcO^NaCl system Parameter Konnecke, 1997 This work 6 TcO<Cl»P Na + TcOiCr Mixing parameters for the NaBr-NaCl04 system * * Parameter Hernandez-Luis, 1996 This work e Br'ClOi Na + Br"C10; * The parameters were obtained using the binary parameters, P ( ' = , p (1) = , and C* = determined by Konnecke et a/.(1997). Table 6.5. Comparison of measured osmotic coefficients with calculated those for the NaTc0 4 system at K. Konnecke, 1997 This work ( P (0) = , P (0) = , C*= ) (P (0) = , P (0) = , C*= ) Molality tf*f Rel Err, % ^calc Rel Err, % Ave Rel Err,% Ave Rel Err,% Relative Percent Error (Rel Err, %) = f * - fh x 100 /

99 83 Calculations were also performed for the NaCl, Na2SC>4, Na2S2C>3 and NaCl- Na2SC»4 (see APPENDIX VII). The calculated parameters were found to give exactly the same values for the osmotic coefficient of NaCl and slightly better values for the other systems.

100 84 CHAPTER 7. A PRIORI DETERMINATION OF MODEL PARAMETERS Model parameters such as Pitzer's parameters and equilibrium constants of sodalite dihydrate and hydroxysodalite dihydrate formation reactions are required for the thermodynamic modeling of the sodium aluminosilicate formation in alkaline solutions Pitzer's parameters Pitzer's binary and mixing relevant parameters for our system are shown in the Tables 7.1 and 7.2. A value of at K was used for the Debye-Huckel parameter (Pitzer, 1991). Pitzer introduced the mixing parameters, 9 and *F to account for the differences of ion interactions between in multi-component and single salt electrolyte solutions. Thus, the mixing parameters have smaller effects than the binary parameters, p (0), p (1), and C* have (Pitzer and Kim, 1974). The binary parameters for NaOH, NaCl, and Na2C03 for K were determined using the numerical expressions available in the literature (Pitzer, 1991; Silvester and Pitzer, 1977; Peiper and Pitzer, 1982). Although numerical expressions for the binary parameters for Na 2 S04 were available in the literature, they were not used because they were based on the value of 1.4 for the "a. parameter" instead of the usual value of 2.0 for a 1-2 electrolyte (Pitzer, 1991; Pabalan and Pitzer, 1988). Osmotic coefficient data for Na 2 S04 are available at , , , , and K (Pabalan and Pitzer, 1988).

101 85 Table 7.1. Pitzer's binary parameters for the modeling of sodium aluminosilicate formation. ion pair 3(0) 3d) c* Na + OH" m Na + Cl" P [ 2 ] Na + 2 C0 3 ' [ 3 ] [ 3 ] Na + 2 S0 4 ' [ 4 ] [41 Na + HS" [ 1 ] 0.0 w Na + 2 Si0 3 " [ 4 ] [ 4 ] [ 4 ] Na + A1(0H) 4 " [ 4 ] [ 4 ] [1] Pitzer (1991), [2] Silvester and Pitzer (1977), [3] Peiper and Pitzer (1982), [4] this study Table 7.2. Pitzer's mixing parameters for the modeling of sodium aluminosilicate formation. ion pair aa'na* OH"Cl" PI [ 1 ] 2 OH" C0 3 " 0.1 m OH" S0 4 " [ 1 ] OH" Si0 3 " [ 2 ] OH"Al(OH) 4 " [ 2 ] crco3 2 Cl" S0 " [ 1 ] m " 0.03 [ 1 ] C1" Al(OH) 4 " [ 2 ] [ 2 ] 2 2 C0 3 " S0 4 ' 0.02 m [ 1 ] [1] Pitzer (1991), [2] this study These data were interpolated at K and gave a set of values. The binary parameters of Na2S0 4 at K were then determined by using these interpolated osmotic coefficient data and minimizing the following least squares objective function

102 86 s(p (o),p (i),c*)= ;(c t -( )r 1 ) 2 d=l (7.i) where the (J)"* is the interpolated and the (J) 0 * 1 is the calculated osmotic coefficients using equation (2.46). Based on a study with NaCl, Pitzer noted that change in the parameter values from to K was very small. This justified the use of the values at K whenever values at K were not available. However, the Debye-Huckel parameter, A$, changes significantly and the value at K was used (Zemaitis et al, 1986). In the case that no Pitzer's parameters at K were available, those at K were used. Binary parameters for NaHS and mixing parameters for OH"-Cl', OH-CO3 2 ', OFT S0 4 ', C1'-C0 3 ", C1'-S0 4 ", and C0 3 "-S0 4 2" are available at K in the reference of Pitzer Finally, it is noted that binary parameters of Na 2 SiC>3 and NaAl(OH) 4 and mixing parameters of Na 2 Si0 3 -NaOH and NaAl(OH) 4 -NaOH-NaCl systems at K were determined from the osmotic coefficient measurements using isopiestic method as explained in chapters 5 and Equilibrium Constants Values of the equilibrium constants for sodalite dihydrate and hydroxysodalite dihydrate formation at K were not available in the literature. Assuming that heat capacity change of reaction, AC P, is not a function of temperature, the equilibrium constants at specific temperature can be calculated using by equation (2.18) (Anderson and Crerar, 1993) lnk = AG RT, 0 j AH" R 1 1 T T, AC" T T In +! R V T 0 T (2.18)

103 87 where K is equilibrium constant at temperature T, AGJJ is the Gibbs free energy change of reaction at reference state, AH is the enthalpy change of reaction at reference state, AC is the heat capacity change of reaction at reference state, R is the gas constant, and To is the reference temperature. Values for AG, AH^, and AC" can be obtained by the following equations AG = EvAGf products " EvAGf reactants (7.2) AH = EvAHf products - EvAHfVeactants (7.3) AC" = IVCP c products - c IvC P reactants (7.4) where the v is the number of moles of reactants or products. Table 7.3 shows property values for the species required to calculate the above equilibrium constants. Property values that were not available were estimated in this work as follows. Table 7.3. Thermodynamic data at K and 1 bar published in the literature. AG f (kj/mol) AH f (kj/mol) CP (J/mol-K ) Na + (aq) ±0.1 [ 1 ] ± 0.06 [1] [2] Cl'(aq) ± 0.1 [1] ±0.1 t l ] OH (aq) ±0.1 ( 1 ] ±0.04 m H2O (1) ±0.04 [ 1 ] ±0.04 m ±0.08 m Al(OH) 4 " (aq) ± [3] [4] [4] Si03 2 ( 8 q) [2] N/A N/A Na 8 (AlSi0 4 )6Cl 2-2H 2 0 w N/A [ 5 ] N/A N/A [5) NagtMSiOACOrD^rtO fl, N/A [ 5 ] N/A t 5 ] N/A 151 N/A : not available in the literature [1] Nordstrom and Munoz (1994), [2] Babushkin et al. (1985), [3] May et al. (1979), [4] Raizman (1985), [5] Hemingway (1997).

104 Estimation of Thermodynamic Properties AHf and AGr of sodalite dihydrate and hydroxysodalite dihydrate The AHf and AGf of sodalite dihydrate and hydroxysodalite dihydrate were estimated by the method of Mostafa et al. (1995). A mean error of their method was reported as 2.57 % for the estimation of standard enthalpy of formation and 2.60 % for that of standard Gibbs free energy of formation. They used the following functional forms for the standard enthalpy of formation and standard Gibbs free energy of formation of inorganic salts: AH = n j A H j (7.5) j AG^JXAo. (7.6) j where, nj is the number of groups of the jth type and AHJ and AQJ are contributions for the jth atomic or molecular group in the enthalpy and Gibbs free energy of formation respectively. The estimation procedure is quite simple. The molecular structural formula is broken into appropriate cationic, anionic, or ligand molecular structural groups. Then the numerical contribution of each group is obtained and multiplied by the number of occurences of the same group in the molecular structural formula. Sum of the numerical values of the various groups yields estimation for AHf and AGf. Table 7.4 shows the sequence of calculations for anhydrous sodalite. The published AHf of anhydrous sodalite (Nagl^AlSiO^Cb) is kj/mol (Komada et al, 1995) and predicted one is kj/mol. Thus, an error of the estimation is 2.57 %. The published AGf of

105 89 anhydrous sodalite (Na 8 (AlSi0 4 ) 6 Cl 2 ) is kj/mol (Komada et al, 1995) and predicted one is kj/mol. An error of the estimation of the Gibbs free energy is 2.63%. Table 7.4. Estimations of AHf and AGf of anhydrous sodalite (Na 8 (AlSi0 4 )6Cl 2 ). Na 8 (AlSi0 4 ) 6 Cl 2 group ni A H j n s Aci Na + 8 X X Al X X su + 6 X X o 2 24 X X cr 2 X X AHf = kj/mol AG f = kj/mol Since the AHf of anhydrous sodalite (Na 8 (AlSi0 4 )6Cl 2 ) is available and the only difference in the chemical formula of anhydrous sodalite (Na 8 (AlSi0 4 )6Cl 2 ) and that of sodalite dihydrate (Na 8 (AlSi0 4 )6Cl 2-2H 2 0) is two water molecules, the AHf of Na 8 (AlSi0 4 )6Cl 2-2H 2 0 can be calculated by adding the contribution term of 2H 2 0 to the published AH f of Na 8 (AlSi0 4 ) 6 Cl 2. The contribution by 2H 2 0 is equal to 2 x = kj/mol. The AH f of Na 8 (AlSi0 4 ) 6 Cl 2-2H 2 0 is equal to AH f (Na 8 (AlSi0 4 ) 6 Cl 2 ) + Contribution by 2H 2 0 = ( ) = kj/mol. The AG f of Na 8 (AlSi0 4 ) 6 Cl 2-2H 2 0 can be calculated by the same logic. The contribution by 2H 2 0 for the AG f is equal to 2 x = kj/mol. The AG f of Na 8 (AJSi0 4 ) 6 Cl 2-2H 2 0 is equal to AG f (Na 8 (AlSi0 4 ) 6 Cl 2 ) + Contribution by 2H 2 0 = ( ) = kj/mol. The uncertainties of the AH f and AG f of the Na 8 (AJSi0 4 ) 6 Cl 2-2H 2 0 were calculated from the uncertainties of the AHf and AGf of the Na 8 (AlSi0 4 )6Cl 2, ± kj/mol and kj/mol (Komada et al, 1995) and the uncertainties of the

106 90 contributions by 2H 2 0 by a method available in the literature (Baird, 1995; Holman, 1994). The uncertainties are shown in Table 7.5 together with the estimated values of the thermodynamic properties. Detailed calculations of the uncertainties are given in APPENDIX VI. Table 7.5. Estimated thermodynamic data at K and 1 bar. Si03 2 "( aq ) Na 8 (AlSi0 4 ) 6 Cl 2-2H 2 0 (s) Na 8 (AlSi04) 6 (OH) 2-2H 2 0 (. ) AGf ( kj/mol) AHf ( kj/mol) C P (J/molK ) ± ± ± ± ± ± ± ±5.37 The AHf and AG f of hydrated hydroxysodalite dihydrate (Na 8 (AlSi0 4 )6(OH) 2-2H 2 0) were calculated by the same method. The AHf and AGf of Na 8 (AlSi0 4 )6(OH) 2-2H 2 0 can be calculated by subtracting the contribution term of Cl 2 and adding the contribution term of (OH) 2 and 2H 2 0 to the published AHf and AGf of Na 8 (AlSi0 4 ) 6 Cl 2. The contribution of Cl 2 for the AH f is equal to 2 x = kj/mol and that of (OH) 2 is equal to 2 x = kj/mol. Thus, the AH/ of Na 8 (AlSi0 4 )6(OH) 2-2H 2 0 is equal to AH f (Na 8 (AlSi0 4 ) 6 Cl 2 ) - Contribution by Cl 2 + Contribution by (OH) 2 + Contribution by 2H 2 0 = ( ) + ( ) + ( ) = kj/mol. The contribution of Cl 2 for the AG f is equal to 2 x = kj/mol and that by (OH) 2 is equal to 2 x = kj/mol. Thus, the AG f of Na 8 (AlSi0 4 ) 6 (OH) 2-2H 2 0 is equal to AG f (Na 8 (AlSi0 4 ) 6 Cl 2 ) - Contribution by Cl 2 + Contribution by (OH) 2 + Contribution by 2H 2 0 = ( ) + ( ) + ( ) = kj/mol. The uncertainties of the AH f

107 91 and AGf of the Na8(AlSi04) 6 (OH)2-2H 2 0 were also calculated and are shown in Table 7.5 from the uncertainties of the AHf and AGf of the NagfAlSiO^eCb and the uncertainties of the contributions by 2H2O, CI2, and (OFfy groups C p of sodalite dihydrate and hydroxysodalite dihydrate Another group contribution technique was proposed by Mostafa et al. (1996) to predict the C p for solid inorganic salts. A mean error of their method was reported as 3.18 % when predicted values were compared with literature values for heat capacity at K. They used the following functional form for the heat capacities of solids: C P =2>Ai + r \ f \ l Z n i A b J xi 0 ' 3 M2>A* xl 6 IT^+I V J S " JADJ X L J V 1 6 J T (7.7) where j is atomic or molecular group, nj is the number of groups of the jth type, A is the contributions for the a, b, c, or d coefficient and T is in kelvin. A stepwise procedure for the estimation of C p is similar that of AGf and AHf estimations. The molecular structural formula for the solid inorganic salt is broken into appropriate cationic, anionic, or ligand molecular structural groups. Then the numerical contributions of each group are calculated and multiplied by the number of occurrences of the same group in the molecular structural formula. The sum of the numerical values of the various groups yields estimation for EpjAaj, SjnjAbj, EjnjAcj, and SjnjAdj. The C p can then be calculated by the equation (7.7) at the temperature of interest. Table 7.6 shows the sequence of calculations for the C p estimation of anhydrous sodalite (Na 8 (AlSi0 4 )6Cl 2 ). The published C p of anhydrous sodalite is J/mol K at

108 92 K (Komada et al, 1995) and predicted one is J/mol K. Thus, the estimation error is 0.48 %. Table 7.6. Estimation of C p of anhydrous sodalite (Na 8 (AJSi0 4 )6Ci2). step 1 group A* Acj Adj step 2 Na + 8 x x x x Al x x x x Si 4+ 6 x x x x O 2 24x x x x Cl 2 x x x x step 3 ZjnjAij step 4 C 0 = xl0-3 x ( )xl0 < ( )xl0' 6 x = J/mol K Since the chemical formula of Na 8 (AlSi04)6C-2 differs from that of Na8(AlSi04)6Cl2-2H_0 by two water molecules, the heat capacity of Nag(AlSi04)6Cl2-2H20 can be calculated by adding the contribution term of 2H 2 0 to the published heat capacity value of Na8(AlSi04)6C.2. The contribution by 2H2O is equal to as shown in Table 7.7. The C p of Na 8 (AlSi0 4 )6Cl 2 is J/mol K. Thus, the C p of Na 8 (AlSi04)6Cl2-2H 2 0 is equal to C p (Na 8 (AlSi0 4 )6Cl 2 ) + Contribution by 2H 2 0 = J/mol K. Table 7.7. Contribution by 2H 2 0 for C p estimation Group Aaj Abj Acj A_j H x x x x ZjnjAij Contribution of 2H 2 0 = xl0 3 x xl0 6 / ( )xl0^x = J/mol K

109 93 Similarly, the heat capacity of Na 8 (AlSi04)6(OH) 2-2H 2 0 can be calculated by subtracting the contribution term of Cl 2 and adding the contribution term of (OH) 2 and 2H 2 0 to the published heat capacity of Na 8 (AlSi04)6Cl 2. The contributions by Cl 2 and (OH) 2 are described in Table 7.8. The C p of Na 8 (AlSi0 4 )6(OH) 2-2H 2 0 is equal to C p (Na 8 (AlSi0 4 )6Cl 2 ) - Contribution by Cl 2 + Contribution by (OH) 2 + Contribution by 2H 2 0 = = J/mol K. The estimated C p values of Na 8 (AlSi0 4 )6Cl 2-2H 2 0 and Na 8 (AlSi0 4 )6(OH) 2-2H 2 0 are shown in Table 7.5 together with their uncertainties. The uncertainties were calculated from the uncertainty of C p (± 4.06 J/mol K) for Na 8 (AlSi04)eCl 2 and the uncertainty of the contributions by the 2H 2 0, Cl 2, and (OH) 2 groups following the method used before (Baird, 1995; Holman, 1994). Table 7.8. Contributions by Cl 2 and (OH) 2 for C p estimation. group Ajj Ay Acj Adj Cl 2 2 x x x x EjnjAij Contribution of Cl 2 = xl0 3 x (-0.502)xl0 6 / xl0^x = J/mol K (OH) 2 2 x x x x ZjnjAij Contribution of (OH) 2 = xl0" 3 x (-1.256)xl0 6 / xlQ- 6 x = J/mol K S of SiQ3%*) Prior to determining the AH/ and Cp of Si03 2 *, the S of SiC>3 2 " was estimated by the method of Couture and Laidler (1957). They found that the entropy of oxy-anions (XO n " m ) in aqueous solution at standard state is given by the following empirical relationship

110 94 S = 5.5z RlnM w?1^l_ (78) n r oxy where, z is the charge of ion, R is the gas constant (1.987cal/mol K), M w is the molecular weight of ion, noxy is the number of oxygen and r is the radius of a sphere that completely circumscribes the anion (r = rn ). The 1.40 A is the van der Waals radius of oxygen. The rn is the distance between the center of the central atom and the center of the surrounding oxygen atoms. The units of the calculated entropy by equation (7.8) are cal/mol K. The predicted entropies of several aqueous ions are compared with published data in the literature in Table 7.9. The published entropy data of aqueous ions were obtained from Babushkin et al. (1985). The average error of the estimation was 18.41%. The rn data except for that of SiC«3 2 ' are available in the literature (Couture and Laidler, 1957). The rn of Si03 2 ' is also available in the literature (Bailer et al, 1973). The entropy of the SiCb 2 ' aqueous ion was calculated as cal/mol K ( = J/mol K). The uncertainty of the entropy of the 2 Si0 3 ' was assumed to be, 18.41% (± 3.84 J/mol K), which is the average error of this estimation method from Table 7.9. Table 7.9. Prediction of the entropy of aqueous ions at K, cal/mol K. rn r M w fro xy z Calculated s Published S error CIO % N % C % so % Se % P % Si

111 AHf Of Si03 2 "(aq) The enthalpy of formation of SiC>3 2 ' aqueous ion was calculated by considering the following equation AH = AG 0 + TAS (7.9) where, AG is the standard Gibbs free energy change of reaction, AH 0 is the standard enthalpy change of formation, and AS 0 is the change of standard entropy. Let us consider the dissociation reaction of sodium metasilicate, Na2Si03(s) at K (Babushkin etal, 1985). Na 2 Si03 (s ) -> 2Na + (aq) + Si0 3 2'( aq) (7.10) Table 7.10 shows the thermodynamic data for Na2Si03( S ), Na + ( aq ), and Si03 2 "( aq ) needed in the calculation of the AG 0, AH, and AS of the reaction (7.10). Substituting the calculated AH, AG, and AS into equation (7.9) gives a value of kj/mol for AHf of S.O3 2 ". The uncertainty in the calculation of AHf for Si(_>3 2 " was found to be ± 1.14 kj/mol following a method available in the literature (Baird, 1995; Holman, 1994). Table Thermodynamic data of Na 2 Si03( S ), Na + (aq), and Si03 2 "( aq ) for the calculation of the AG 0, AH, and AS 0 of reaction (7.10). AG f (kj/mol) AH f (kj/mol) S (J/molK) Na 2 Si0 3(s ) m , Na (aq) ±0.1 [2] ± [2] Si03 "(aq) [1] [1] Babushkin etal. (1985), [2] Nordstrom and Munoz (1994), [3] calculated in section

112 C p Of Si03 2 "(aq) The estimation equation (7.11) for the heat capacities in cal/mol K for ionic solutes was suggested by Criss and Cobble (1964) C p = C p abs z = a + b(s - 5.0z) z (7.11) where, a is equal to -145 and b is equal to 2.20 for oxy-anions (XO n " m ) at K, and z is the charge of the ion. The Criss and Cobble's heat capacity estimation method was tested for several oxy-anions. The calculated values are compared with the published data in Table The published entropy and heat capacity data of aqueous ions were obtained from Babushkin et a/.(1985). The estimation error was not negligible for some ions such as Nb0 3 " and CO3 2 ". An average error of the estimations was %. The C p of SiC>3 2 " ion 2 is equal to cal/mol K (= J/mol K). The uncertainty of the C p of the Si0 3 " was assumed as the same, % (± J/mol K), as the average error of this estimation method from Table Table Heat capacities of aqueous oxy-anions at K, cal/mol K. Published Calculated Published z S r c 0 C 0 error 0 abs N % CKV % Nb % 2 C % 2 S % 2 Se % 2 S % 3 As % 2 Si

113 Calculation of at K Prior to calculating the equilibrium constant, Ksod, of sodalite dihydrate formation, the AGQ, AHQ, and AC of the reaction (4.1) were calculated with the thermodynamic data in Tables 7.3 and 7.5. The AG of reaction (4.1) is equal to AGf 0 (Na 8 (AlSi0 4 )6Cl 2-2H 2 0) + 4AG f (H 2 0) + 12AG f (OH) - SAG/fNa*) - 6AG f (Al(OH) 4 ") - 6AG f (Si0 3 2") - 2AGf (Cr) = x( ) + 12x(-157.2) - 8x(-262.0) - 6x( ) - 6x( ) - 2x(-131.2) = kj/mol from equation (7.2). The AH is equal to AHf (Na 8 (AlSi04)6Cl 2-2H 2 0) + 4AH f (H 2 0) +. 12AH f (OH-) - 8AH f (Na + ) - 6AHf 0 (Al(OH) 4 ") - 6AHf (Si0 3 2") - 2AH f (Cl') = x( ) + 12x( ) - 8x( ) - 6x( ) - 6x( ) - 2x(-167.1) = kj/mol from equation (7.3). The AC; is equal to C p (Na 8 (AlSi0 4 )6Cl 2-2H 2 0) + 4C P (H 2 0) + 12C p (OrT) - 8C p (Na + ) - 6C P (A1(0H) 4 -) - 6C p (Si0 3 2") - 2C p (Cf) = x(75.351) + 12x( ) - 8x(46.43) - 6x(241.44) - 6x( ) - 2x(-136.4) = J/mol K = kj/mol K from equation (7.4). Substituting AG, AH, and AC into equation (2.18) at T = K and T 0 = K gives In K s o c i = Thus, the equilibrium constant of sodalite dihydrate formation, K s o d, is equal to 1.82E+17. The uncertainty in the calculated values for In K s o _ was found to be ± 8.71 from the uncertainties of the estimated thermodynamic properties in Table 7.5 by the method of Baird (1995). The major contribution in the uncertainty of K S od was due to the uncertainties in the estimated thermodynamic data, especially that of the AGf (Na8(AlSi0 4 )6Cl 2-2H 2 0).

114 Calculation of K h s o d at K The AG of reaction (4.2) is equal to AG f 0(Na 8 (AlSi04)6(OH)2-2H 2 0) + 0 4AG f (H 2 0) + 12AG f (OFT) - 8AG f (Na + ) - 6AG f (AJ(OH) 4 ) - 6AG f (Si0 3 2") - 2AGf (OH") = = x( ) + 12x(-157.2) - 8x(-262.0) - 6x( ) - 6x( ) - 2x(-157.2) = kj/mol. The AH of reaction (4.2) is equal to AH f 0(Na 8 (AlSi0 4 )6(OH) 2-2H 2 0) + 4AH f (H 2 0) + 12AH f (OFT) - 8AH f (Na + ) - 6AH f (Al(OH) 4 ) - 6AH f (Si0 3 2') - 2AH f (OH-) = x( ) + 12x( ) - 8x( ) - 6x( ) - 6x( ) - 2x( ) = kj/mol. The AC of 0 reaction (4.2) is equal to C p (Na 8 (AlSi0 4 ) 6 (OH) 2-2H 2 0) + 4C P (H 2 0) + 12C p 0(OFT) - 8C p (Na + ) - 6C P (A1(0H) 4 -) - 6C p (Si0 3 2) - 2C p (OH") = x(75.351) + 12x( ) - 8x(46.43) - 6x(241.44) - 6x( ) - 2x(-148.5) = J/mol K = kj/mol K. Substituting AG 0, AH, and AC into equation (2.18) at T = K and T 0 = K gives In Khsod = Thus, the equilibrium constant of hydroxysodalite dihydrate formation, Khsod, is equal to 3.38E+38. The uncertainty in the calculated values for In Khsod was found to be ± Detailed calculation of the uncertainty is described in APPENDIX VI. Again major contribution in the uncertainty of K n s o d was due to the uncertainties of the estimated thermodynamic data, especially that of the AGf O (Na 8 (AlSi0 4 ) 6 (0H) 2-2H 2 0).

115 Change of In Khsod with Temperature 1 Values of In Khsod were calculated with varying temperature from K to K using equation (2.18). The results are shown in Figure 7.1. A solid line in the figure represents In Khsod at the temperature of interest. As seen in the figure, the In Khsod decreases with increasing temperature. Figure 7.1. Change of In Khsod with temperature. 130 gq I i i i i I i i i i I i i i i I i i I Temperature (K) The AC of the hydroxysodalite dihydrate formation reaction (4.2) is known at K as described in section but not at higher temperature. In this case, it is

116 100 generally better to assume that AC is constant as temperature increases from K to K (Anderson and Crerar, 1993). A dotted line in the Figure 7.1 represents In Khsod at the reference temperature, K. The area between the dotted line and the dashed line represents the contribution by the AC term in equation (2.18) on In Kh S O d. The area between the dashed line and the solid line represents the contribution by the AHQ term in the equation. As seen in the figure, the AH term has the largest contribution on the value of In Khsod- The contribution by the AC term is very small compared to that by AH. Thus, the assumption of constant AC in the range of temperature from To to T in the equation (2.18) is acceptable.

117 101 CHAPTER 8. SOLUBILITY MAPS OF Al AND Si IN GREEN AND WHITE LIQUORS Data of the solubility experiments using synthetic liquors (systems A and B) and mill liquors are shown in Tables A.3.1, A.3.2, and A.3.3. in APPENDIX Ul together with their standard error Synthetic Green and White Liquors of System A A solid phase started forming in a supersaturated solution of Al and Si within several hours from the beginning of the experiment. The solid precipitates were formed on the wall, stirrer blade, and bottom of the vessel. During the experiment, the concentrations of the soluble Al and Si decreased with time but after four of five days, they reached a constant value. After seven days, the samples were taken and considered to be in their equilibrium state (Figure 8.1). The base case experiment (Experiment Al) of synthetic green liquor of system A was carried out three times to check the repeatability of the experiments. The results of the experiment are tabulated on the Table 8.1. The standard deviations for Al and Si concentrations were found to be 6.37xl0' 5 and 6.52xl0" 5, respectively.

118 102 Figure 8.1. Al and Si ions approaching equilibrium obtained from the base experiment using synthetic green liquor of system A. (Experimental data were obtained from experiment Al using synthetic green liquors of system A at K and latm.) 800 CL 7 00# Q_ U tn 600 \- c o Total soluble Si O Total soluble Al (5 500 h o V) c 13 o t c 100 o O o Time, (days) Table 8.1. Repeatability of the experiments in three runs run Al Si [Al][Si] moles/kg H2O moles/kg H2O (moles/kg H 2 0) E E E E E E E E E-5

119 103 It is customary to present precipitation results in the form of a graph which is known as the solubility map (Wannenmacher et al, 1996). Figure 8.2 shows the solubility map for the synthetic green and white liquors of system A. In the solubility maps, the 95 % confidence interval for each data point has been plotted together with the experimental data points. The confidence interval calculations are explained in detail in APPENDIX II. This was calculated using the standard error of the five Atomic Absorption Spectrophotometer measurements for each sample. If the concentrations of Al and Si in the liquors correspond to a point on the right side of each line, precipitation will occur for this solution. The precipitation conditions in synthetic white liquor are higher than those in the synthetic green liquor. This result is attributed to the effect of [OFT]. White liquor contains more OFT. Higher [OFT] content makes Al and Si more soluble as we see in Figure 8.3 which shows the effect of changing the OFT concentration on the precipitation conditions. The change in the OH" concentration is accomplished by varying the input amount of NaOH during the preparation of the solution. The base concentration, given in Table 3.1, is 1.0 mol/kg H 2 0 in synthetic green liquor. The perturbations around the base case are also given in the figure. As seen from the plot, as the hydroxyl ion concentration increases, precipitation occurs at higher Al and Si concentrations. Zheng and coworkers also studied sodium aluminosilicate crystallisation in relation to the Bayer process at 100 C (Zheng et al, 1997). Their result is also plotted in Figure 8.3. The chemical compositions of the Bayer liquors are not exactly the same as in this work. The synthetic Bayer liquor contains 4.53 mol/l of NaOH.

120 104 Figure 8.2. Solubility map of Al and Si in synthetic green and white liquors of system A. (Experimental data were obtained from experiments Al, A2, and A3 using green liquors and from experiments A10, Al 1, A12 using white liquors at K and latm.)

121 105 Figure 8.3. Effect of hydroxyl ions on the solubility limit in synthetic green and white liquors of system A. (Experimental data were obtained at K and latm. Initial concentrations of NaOH in green liquors were 0.25 mol/kg H 2 0 for experiment A4, 1.0 mol/kg H 2 0 for experiments Al, A2, and A3, and 2.0 mol/kg H 2 0 for experiment A5. Those in white liquors were 2.0 mol/kg H 2 0 for experiment A13, 2.5 mol/kg H 2 0 for experiments A10, Al 1, and A12, and 3.0 mol/kg H 2 0 for experiment A14.) Si concentration (mol/kg H 2 0)

122 106 The effect of changing the concentration of the carbonate anion is shown in Figure 8.4. Increasing the initial carbonate ion concentration in the synthetic green liquor gives a lower solubility of Al and Si. In the case of synthetic white liquor, no significant change in the solubility was observed for the concentrations studied (0.1 to 0.5 mol/kg H_0). It is noteworthy that even though the synthetic green liquor contains more carbonate, it is sensitive to a perturbation in the carbonate concentrations whereas the white liquor is not, at least for a ± 0.2 mol/kg H2O change. The concentrations of the sulfate ions in real green and white liquors from kraft mills are in the range of 0.05 to 0.2 mol/l and 0.05 to 0.15 mol/l (Magnusson et al, 1979). The effect of the sulfate ions on the solubility of Al and Si was also measured at input sodium sulfate concentrations of 0.05, 0.1, and 0.2 mol/kg H2O for the synthetic green liquor and 0.05, 0.1, and 0.15 mol/kg H2O for the synthetic white liquor. As seen from Figure 8.5, no significant change in the solubility was observed Synthetic Green and White Liquors of System B The precipitation conditions in synthetic green and white liquors of system B are shown in Figure 8.6. The synthetic white liquor of higher [OFT] compared with the green liquor shows that precipitation occurs at higher Al and Si concentrations. The effect of input amount of Na2S on the precipitation condition is shown in Figure 8.7. As more Na2S is added into the system, precipitation occurs at higher Al and Si concentrations.

123 107 Figure 8.4. Effect of carbonate ions on the solubility limit in synthetic green and white liquors of system A. (Experimental data were obtained at K and latm. Initial concentrations of Na2CC>3 in green liquors were 0.5 mol/kg H2O for experiment A6, 1.0 mol/kg H2O for experiments Al, A2, and A3, and 1.5 mol/kg H 2 0 for experiment A7. Those in white liquors were 0.1 mol/kg H2O for experiment A15, 0.3 mol/kg H2O for experiments A10, Al 1, and A12, and 0.5 mol/kg H2O for experiment A16.) Si concentration (mol/kg H 2 0)

124 108 Figure 8.5. Effect of sulfate ions on the solubility limit in synthetic green and white liquors of system A. (Experimental data were obtained at K and latm. Initial concentrations of Na2S04 in green liquors were 0.05 mol/kg H2O for experiment A8, 0.1 mol/kg H2O for experiments Al, A2, and A3, and 0.2 mol/kg H2O for experiment A9. Those in white liquors were 0.05 mol/kg H2O for experiment A17, 0.1 mol/kg H2O for experiments A10, All, and A12, and 0.15 mol/kg H 2 0 for experiment A18.) 0.1 Si concentration (mol/kg H 2 0)

125 109 Hence, it becomes more difficult to precipitate solids. In other words, the window for normal operation expands. This may be attributed to the fact that Na 2 S dissociates when dissolved in water and releases OH" by the reaction: Na 2 S + H 2 0 -> 2Na + + HS' + OH" (Smook, 1992), that affects the precipitation conditions in a manner seen in Figure 8.3. The effect of [OFT] might be explained by considering the sodium aluminosilicate formation reaction as seen next. The sodium aluminosilicate precipitates were found to be hydroxysodalite dihydrate (Na 8 (AlSi0 4 ) 6 (OH) 2-2H 2 0) and/or sodalite dihydrate (Na 8 (AlSi0 4 ) 6 Cl 2-2H 2 0). The identification of the precipitates is explained in section 8.4. The chemical reaction for sodium aluminosilicate formation of this study can be written as 8Na + + 6A1(0H) 4 " + 6Si0 3 2" + 2X" <- Na 8 (AlSi0 4 )6X 2-2H 2 0(s) + 4H H" (8.1) where X" can be OH' or Cl" (Zheng et al, 1997). The solubility product of Al and Si can be written as (a AI(OH); K e<,(*w) 8 (a x -) 2 J (8.2) where a is the activity and K e q is the equilibrium constant". The solubility product increases with increasing [OH"] at constant concentrations of the other ions.

126 110 Figure 8.6. Solubility map of Al and Si in synthetic green and white liquors of system B. (Experimental data were obtained from experiments Bl, B2, and B3 using green liquors and from experiments BIO, Bl 1, B12 using white liquors at K and latm.) 0.1 \ \ \ o CM \ \ Precipitation D) O 0.01 E c o > \ \ \ < \ \ Q \ \ \ \ - < <D O C O Input ratio Al/Si green Iq. white Iq. 1/2 A 1/1 O 2/1 \ X \ \ \ \ \ \ N Si concentration (mol/kg H 2 0)

127 Ill Figure 8.7. Effect of NajS on the solubility limit in synthetic green and white liquors of system B. (Experimental data were obtained at K and latm. Initial concentrations of NazS in green liquors were 0.0 mol/kg H2O for experiments B4, B5, and B6, 0.5 mol/kg H2O for experiments Bl, B2, and B3, and 1.0 mol/kg H2O for experiments B7, B8, and B9. Those in white liquors were 0.0 mol/kg H2O for experiments B13, B14, and B15, 0.5 mol/kg H 2 0 for experiments B10, Bll, and B12, and 1.0 mol/kg H 2 0 for experiments B16, B17, and B18.) CM 0.1 \ - *\ \ \ \ green Iq. white Iq. Input Na 2 S 0.0 O 0.0 (moles/kg H-O) 0.5 (base} 0.5 (base) O 0.01 \ \ o \ q I c U c g \ m \ \ \ \ \ \ * \ \ \ _ l I ' I I I I I ' ' I I I I I I I I I I 'Ns. I I I I Si concentration (mol/kg H 2 0)

128 Mill Green and White Liquors Figure 8.8 shows the solubility map for the green and white liquors. In the solubility maps, the 95 % confidence interval for each data point has been plotted together with the experimental point. As seen from the graph, the data at three different A/Si ratios fall on the same line. If the concentrations of Al and Si in the liquors correspond to a point located on the right side of each line, precipitation will occur in this liquor. As seen, the precipitation conditions in white liquor are higher than that in the green liquor. This result is attributed to the effect of OFT. Higher OFT concentration makes Al and Si more soluble as seen in Figure 8.9. White liquor contains more OH" (2.63 mol/kg H 2 0) than the green liquor (0.85 mol/kg H 2 0). Figure 8.9 shows the effect of changing the OH" concentration on the precipitation conditions. As seen from the plot, higher concentrations of Al and Si are required to induce precipitation as the hydroxyl ion concentration increases. Thus, the liquor can tolerate increased levels of dissolved Al and Si. Increasing initial carbonate ion concentration in the mill liquors gave a lower solubility of Al and Si, which can be seen in Figure In other words, as the carbonate ion concentration increases, precipitation occurs at lower Al and Si concentrations. The effect of Na 2 S on the precipitation conditions is shown in Figure As more Na 2 S is added into the system^ precipitation occurs at higher Al and Si concentrations. This effect may be attributed to the fact that Na 2 S dissociates when dissolved in water and releases OH" (Ulmgren, 1982; Smook, 1992).

129 113 Figure 8.8. Solubility map of Al and Si in mill green and white liquors. (Experimental data were obtained from experiments Ml, M2, and M3 using green liquors and from experiments M13, M14, M15 using white liquors at K and latm.) 0.1 CM Precipitation o E c o TO 0.01 O c O o Green liquor (base) White liquor (base) Si concentration (mol/kg H 2 0)

130 114 Figure 8.9. Effect of NaOH on the solubility limit of Al and Si in mill liquors. (Experimental data were obtained at K and latm. Initial concentrations of NaOH in green liquors were 0.29 mol/kg H2O for experiments Ml, M2, and M3, and 1.29 mol/kg H2O for experiments M4, M5, and M6. Those in white liquors were 2.06 mol/kg H2O for experiments M13, M14, and M15 and 2.56 mol/kg H2O for experiments M16, M17, andm18.) Si concentration (mol/kg H 2 0)

131 Figure Effect of Na2C03 on the solubility limit of Al and Si in mill liquors. (Experimental data were obtained at K and latm. Initial concentrations of Na2CC«3 in green liquors were 1.05 mol/kg H 2 0 for experiments Ml, M2, and M3, and 1.55 mol/kg H_0 for experiments M7, M8, and M9. Those in white liquors were 0.23 mol/kg H 2 0 for experiments M13, M14, and M15 and 0.48 mol/kg H2O for experiments M19, M20, andm21.) 115

132 116 Figure Effect of Na 2 S on the solubility limit of Al and Si in mill liquors.(experimental data were obtained at K and latm. Initial concentrations of Na2S in green liquors were 0.56 mol/kg FfeO for experiments Ml, M2, and M3, and 1.06 mol/kg H2O for experiments M10, Mil, and M12. Those in white liquors were 0.57 mol/kg H 2 0 for experiments M13, M14, and Ml 5 and 1.07 mol/kg H 2 0 for experiments M22, M23, and M24.) Green liquor 0.1 Na 2 S [OH - ] (mol/kg H 2 0) , base C O CM White liauor Na 2 S [OH"] (mol/kg H 2 0) O , base o E, c o '~t > CO c o _1 1 1 I I I I _1 I I I I I I I J I I I 'Nl I I Si concentration (mol/kg H 2 0)

133 117 The results for the mill liquors with base concentrations were compared with data available in the literature and found in good agreement as seen in Figure The straight line on the graph represents a correlation of the precipitation conditions obtained from experiments with mill green liquors (Wannenmacher et al, 1996). The grey square in Figure 8.12 represents the concentrations of AJ and Si in unsaturated mill white liquors. These liquors contain Al and Si at a concentration level that is near the precipitation conditions. Figure Solubility map comparing solubility limit of Al and Si in varying liquors. (The data were obtained at K and 1 atm) 0.1 Green liquor : This work (base concentration) - Wannenmacher et. al.1996 D). O E c o I 8 c White liquor This work (base concentration) H Wannenmacher etal \ D \ I ' ' S i concentration (mol/kg H 2 0)

134 Structure of the Precipitates from Synthetic Liquors X-ray diffraction analysis of the precipitates from the solutions of system A and B showed that only sodalite (Na 8 (AlSi04)6Cl2nH 2 0) and/or hydroxysodalite (Nag(AlSi04)6(OH)2nH20) were present (Figure 8.13). There is no significant difference in the X-ray diffraction patterns for the sodalite and hydroxysodalite (Gasteiger et al, 1992). Ulmgren (1982) reported that samples of scales from Swedish pulp mills consist of NaAlSi04l/3Na 2 C0 3 and/or NaAlSi0 4 l/3na 2 S0 4 which look like cancrinite. Although the system of this study contained carbonate and sulfate ions, the X-ray diffraction patterns of the precipitates did not match to those of any minerals containing the carbonate and sulfate ions. A portion of the precipitates in the synthetic liquors of system A and B was dissolved in IN nitric acid and analyzed to determine the molar ratios of Na, Al, Si, Cl, and OH in the solid. If the precipitates contain carbonate in the structure, dissolving the solid in acid solution can cause release of carbonate ions to combine with hydrogen ions and be converted to C0 2 gas. However, no bubbles of CO2 gas were detected during the dissolution. Sulfate was not present in the precipitates obtained in the synthetic green and white liquors because the precipitation of BaS04 was not observed by the addition of BaCb solution to the dissolved precipitate solutions. Chloride was not present in the white liquor precipitates of system A and green and white liquor precipitates of system B because the precipitation of AgCl was not observed by the addition of AgN0 3 solution to the dissolved precipitate solutions. The green liquor precipitates of system A, however, had chloride since AgCl precipitated when AgN0 3 solution was added. The amount of

135 119 chloride ions in the dissolved precipitate solution was measured quantitatively using a titration method with a 0.1 N AgNC«3 solution. Figure X-ray diffraction pattern of precipitates in synthetic green liquor of system A. Hydroxysodalite (reference) 100 I i Sample i c 20 rptrrp' ''' r 'j' r r r ' l T* r r T! i'vt!"i i i 1rp'n'i pt'n p i rr^^trr^rrrt^'rrrp-tttttttinrtrrrp

136 120 Thus, X-ray diffraction analysis and chemical analysis revealed that the precipitates of the synthetic green liquor of system A contain the structure of sodalite and hydroxysodalite whereas those of the synthetic white liquor of system A and green and white liquors of system B contain the structure of hydroxysodalite only. In the case of the synthetic green liquor of system A, approximately 25 % of the precipitates on a molar basis are sodalite which has the Cl group and 75 % of the precipitates are hydroxysodalite which has the OH group. The molar ratio of Na : Al : Si is 8 : 6.06 ± 0.05 : 6.05 ± In the case of the synthetic white liquor of system A, the precipitate is hydroxysodalite. The molar ratio of Na : Al : Si is 8 : 5.96 ± 0.09 : 6.07 ± The precipitates of synthetic green and white liquors of system B are hydroxysodalite. The molar ratios of Na : Al : Si are 8 : 6.23 ± 0.20 : 6.15 ± 0.31 and 8 : 6.18 ± 0.13 : 5.89 ± 0.09, respectively. Thermogravimetry analyses were performed for four precipitates, each from the synthetic green and white liquor, of systems A and B. The analyses proceeded from 105 to 900 C at a heating rate of 10 C/min under a N 2 atmosphere. The data for synthetic white liquor precipitates from system A are shown in Figure Total weight loss was 5.73 % by heating up to 900 C and the loss is equivalent to 3.1 H 2 0 («3 H 2 0) molecules per unit cell of the precipitate. Engelhardt et al. (1992) studied the hydroxysodalite system with thermal analysis. According to their result, a total of n+1 water molecules per unit cell leave the structure of Na8(AlSi04)6(OH) 2 nh 2 0 by heating up to about 700 C. First of all, n water molecules leave to form Nag(AlSi04)6(OH) 2 and another water molecule from the remaining OH groups is released to change to a carnegieite-type phase

137 121 Na 2 0-6NaAlSi04. Therefore, the total water loss, three water molecules per unit cell in Figure 8.14, indicates that the structure of precipitate in synthetic white liquor is Na8(AlSi04)6(OF )2-2H 2 0, hydroxysodalite dihydrate. Other precipitates showed similar weight loss, 5.46 % of synthetic green liquor precipitate of system A, 5.22 % and 5.55 % of synthetic green and white liquor precipitates of system B. This fact means that all precipitates have the same number of hydrated water molecules, 2H2O, in their structure. Figure Thermogravimetric analysis of hydroxysodalite dihydrate, Na8(AlSi04)6(OH)2-2FJ.20, precipitated in synthetic white liquor of system A \ X 98 H ^ - > 97 H CO 'CD H 94 weight loss : 5.73 % loss of H 2 0/unit cell: 3.1 > 93 H T T T temperature ( C)

138 Structure of the Precipitates from Mill Liquors X-ray diffraction analysis revealed that the precipitates could be sodalite (Na 8 (AlSi0 4 )6Cl 2 nh 2 0) and/or hydroxysodalite (Na 8 (AlSiO 4 )6(OF0 2 nh 2 O). This analysis cannot distinguish between these two solids (Gasteiger et al, 1992). Therefore, it was complemented with chemical analysis and thermogravimetry. The chemical tests described earlier showed that there was not any chloride, sulfate or carbonate present in the precipitated solids. In addition, the atomic absorption spectroscopy tests showed that there was not any potassium present in the solid. The molar ratios of Na : Al : Si in the precipitates from the green and white liquors were found to be 8 : 6.17 ± 0.16 : 5.86 ± 0.08 and 8 : 5.97 ±0.11 : 5.99 ± 0.10 respectively. Thus, the chemical formula of the precipitate is that of Na 8 (AJSi04)6(OIT) 2 nh 2 0. Finally, the number of the hydrated water molecules was found to be equal to two by thermogravimetry. Hence the precipitate is hydroxysodalite dihydrate, Na8(AlSi04)6(OH) 2-2H 2 0. Structure of precipitates in mill liquors was found to be similar to that in synthetic liquors Morphology of precipitates The precipitates in green and white liquors consist of particles 5 to 30 pm in diameter. Aggregated particles, as shown in Figure 8.15, were found to be suspended in the solution or present as scale on the wall of the vessel and the stir blade. Most of green liquor precipitates were suspended in solution. Homogeneous nucleation might be the dominant nucleation mechanism in the green liquors (Mullin, 1972). The green liquor precipitates formed on the wall of vessel could be easily removed. The image of the green

139 123 l i q u o r precipitates i n F i g u r e 8.15 (a), (b), and (c) s h o w s a less p a c k e d structure o f the particles. F i g u r e S c a n n i n g electron m i c r o g r a p h s o f precipitates. (a) m Precipitates in (a) synthetic green liquor of system A (experiment A l ), (b) synthetic green liquor of system B (experiment Bl), (c) mill green liquor (experiment M l ), (d) synthetic white liquor of system A (experiment A10), (e) synthetic white liquor of system B (experiment BIO), and (f) mill white liquor (experiment M13). All scale bars in the images have the same length, 10 um.