Article pubs.acs.org/IECR
Solubility of Tricalcium Aluminate in Synthetic Spent Bayer Liquor Reza Salimi,* James Vaughan,* and Hong Peng The University of Queensland, School of Chemical Engineering Building (74), College Road, St. Lucia, Queensland 4072, Australia S Supporting Information *
ABSTRACT: Tricalcium aluminate hexahydrate (3CaO·Al2O3·6H2O, TCA) was synthesized via a reaction between slaked lime and sodium aluminate solution; the resulting solids were confirmed to be 3CaO·Al2O3·6H2O by X-ray diffraction. The TCA solubility was then determined over a range of industrially relevant conditions by measuring calcium concentration in solution when at equilibrium with excess solids. It was observed that the solubility increased with increasing sodium carbonate concentration. Slight shifts in the XRD pattern were observed when the solids were exposed to the carbonate-containing liquor. From the analysis of existing thermodynamic data, the dominant equilibrium reaction at the conditions tested was determined to be 3CaO·Al2O3·6H2O + 3CO32− + 4H+ = 3CaCO3(aq) + 2Al(OH)4− + 4H2O. From the experimental results and activity coefficient estimates, the relationship between the reaction equilibrium constant (K) and temperature (T) in the range of 308− 368 K is described by the relation log K = 80.56 − 0.105T.
1. INTRODUCTION Most of the alumina produced industrially is obtained via extraction from bauxite ore, using the Bayer process.1,2 Tricalcium aluminate hexahydrate (3CaO·Al2O3·6H2O, TCA) is used in the alumina industry as a filter aid when separating pregnant liquor from fine solids and colloidal material, prior to gibbsite crystallization. The main role of TCA is to reduce the tendency for the filtered solids to blind the cloth and shorten the cloth life by forming a porous solid coating on the cloth. Specific properties such as an appropriate particle size distribution are required from the TCA to ensure adequate filter throughput, lifetime, and effective solid separation from solution. In the Bayer process, TCA is typically produced on site through the addition of slaked lime to either spent liquor or pregnant liquor, depending on operation conditions.3 Variation in the TCA solubility may alter the crystal nucleation, growth, and agglomeration rates and the corresponding filter aid quality. The solubility as a function of key process variable must be established prior to developing an overall understanding of the crystallization process. No study of the solubility of TCA in spent Bayer liquor (Na2O−CaO−Al2O3−CO3−H2O) has been reported. Some relevant studies have included determining the solubilities of calcium hemicarboaluminate, calcium monocarboaluminate, and calcium tricarboaluminate.4−6 It was concluded that calcium hemicarboaluminate and calcium monocarboaluminate become stable phases at elevated carbonate concentration, and calcium tricarboaluminate was found to be unstable at 298 K. The solubility of calcium hydroxide in highly concentrated sodium hydroxide solution at 298 K also has been studied.7 The solubility of calcium hydroxide was found to steadily decrease with the increasing sodium hydroxide concentration. The solubility of TCA in the KOH−K2CO3−H2O system in the range of 30−80 °C has been studied. From this work, it was concluded that TCA converted to monocarbonate and calcium carbonate or calcium hydroxide, depending on the concentration of potassium hydroxide.8 © 2014 American Chemical Society
Solubility determination can be obtained by either the dissolution method or the precipitation method. 9 The experimental technique employed in this research is the most common method of determining solubility, which is by mixing an excess amount of solute into a solvent and gently mixing until equilibrium is reached then establishing the amount of solute that has dissolved.10 Calcium solution concentration was determined by assaying with inductively coupled plasma spectrometry−optical emission spectroscopy (ICP − OES), since it is sufficiently sensitive for the relatively low concentrations of aqueous calcium dissolved in these systems. For the thermodynamic predictions, relevant data was reviewed using critically assessed databases and specialist research papers where required. In order to make use of the thermodynamic predictions, the measured or known solution species concentrations are converted to aqueous species activities by way of an activity coefficient. The Debye−Hückel and Davies equations can be used to estimate the activity coefficients in dilute solution. For example, in 1996, Baron and Palmer employed the Davies equation to investigate the solubility products of jarosite (KFe3(SO4)2(OH)6) at 277− 308 K.11 However, the aqueous solution of interest in this research has high ionic strength, meaning the Debye−Hückel activity coefficient model is not applicable. Fortunately, in the last few decades, several advanced models have been developed for estimating the ionic activity coefficient in concentrated electrolyte solutions, such as the ENTRL equation, the Pitzer equation, the Bromley−Zemaitis equation, and the Meissner equation.12 In 2012, Gao and Li employed the Bromley− Zemaitis equation to investigate the solubility of Mg4Al2(OH)14·3H2O at various ionic strengths.13 In 1992, Gasteiger et al. employed the Pitzer model to account for activity coefficients to measure solubility of aluminosilicates in Received: Revised: Accepted: Published: 17499
July 6, 2014 October 9, 2014 October 13, 2014 October 13, 2014 dx.doi.org/10.1021/ie502693p | Ind. Eng. Chem. Res. 2014, 53, 17499−17505
Industrial & Engineering Chemistry Research
Article
Figure 1. Scanning electron microscopy (SEM) morphology of tricalcium aluminate (TCA).
alkaline solutions.14 The Pitzer model is an expansion of the Debye−Hückel method where “terms were added to account for the ionic strength dependence of the short-range forces effect in binary interactions”.15 In the present work, the Harveymodified Pitzer equation was used to estimate the activity coefficients. However, it does not seem that the temperature dependence of the activity coefficient for carbonate is considered in the model. The purpose of this research is to determine the solubility of well-characterized TCA in synthetic spent Bayer liquor over a range of temperatures (308−368 K) and solution compositions. Comparing the outcomes of the solubility experiments in these well-defined systems with thermodynamic predictions provides insight into the nature of the chemical equilibrium and also provides an indication of which published thermodynamic data more accurately describes the observed behavior for the situations where a range of data is available.
2.3. Characterization. X-ray diffraction (XRD) analysis of the synthesized solid TCA precipitate was performed with a Bruker Model D8 Avance XRD system with a LynxEye detector, and Cu radiation at 40 kV and 40 mA. The aluminum and calcium content of select solids was measured by fully digesting the solid sample in hydrochloric acid and analyzing the solution metal content using ICP-OES. The morphology and size of the synthesized crystallites were determined from images obtained from a Philips Model XL30 scanning electron microscopy (SEM) system. The particle size distribution of TCA was determined by an Accusizer, and the measurement is based on an optical pulse through a diluted slurry and analysis of the pulse height. In order to dilute the TCA slurry without dissolving the fine particles, the dilution solution used was a saturated solution of TCA, which was prepared by adding an excess amount of TCA to 100 mL of water and then filtering through a 0.22 μm membrane filter. For the particle size distribution analysis, a small amount of TCA slurry from the experiment (1 mL) was diluted to the appropriate particle concentration and fed into the Accusizer instrument. 2.4. Solution Characterization. Solution samples were isolated from solids via filtration, using 0.22-μm nylon membrane filters and separate aliquots allotted for metal concentration determination. The samples were diluted (10×) into a 1 m sodium hydroxide solution to stabilize the metals in solution and provide a liquid of sufficiently low viscosity to be fed directly to the ICP-OES equipment. ICP standards were matrix-matched to this solution. This method was employed as diluting the solution into the acid matrix would result in calcium concentrations too low to accurately measure and even with the current method the sensitivity limit of the instrument is being approached as the calcium concentration in the diluted solution is on the order of 1 mg/L. 2.5. Experiments. Solubility experiments were carried out over a temperature range of 308−368 K. Synthesized spent liquor (120 mL) was placed in a 250 mL Pyrex volumetric flask with gentle agitation provided by a magnetic stir bar. Once the temperature was equilibrated at the set-point by a feedback controlled hot plate, synthesized TCA (3 g) was added to the solution. The temperature fluctuated within ±2 °C of the setpoint. The thermocouple was calibrated against a glass thermometer and the performance of the thermocouple was also checked regularly. Nitrogen gas was purged through the solution for 20 min to expel carbon dioxide from the system before adding the TCA and prior to sealing the flasks for the duration of the experiment. During select experiments, samples were withdrawn periodically.
2. METHODS 2.1. Synthesis of Tricalcium Aluminate. Sodium hydroxide solution was first prepared by dissolving the compound in deionized and deaerated water in a volumetric flask, which was sealed and heated to 353 K to yield a final solution of 2.53 m (molal scale, defined as the number of moles of NaOH per kilogram of H2O (mol NaOH/kg H2O)). The molal concentration scale is convenient, because it does not vary with temperature due to changes in the density and it is also typically referred to in aqueous species activity coefficient models. The required mass of aluminum hydroxide (1.7 m) and calcium oxide slurry (as 100 g CaO/L H2O) were then added to the flask and the mixture was heated to 368 K. The reactor was completely sealed and constant and gentle stirring of solution was provided by a Teflon-coated magnet. After 2 h, the slurry was filtered and the TCA precipitate was quickly washed and dried in air at 373 K for 10 h. 2.2. Synthetic Bayer Liquor Preparation. Solutions were prepared from the following constituents: gibbsite (C-31 grade, 99.4% Al(OH)3 by weight), sodium hydroxide (2.2% Na2CO3 by weight), anhydrous sodium carbonate (purity of 99.9%), and deionized water. To prepare the sodium aluminate solution, the sodium hydroxide was dissolved into the water in a stainless steel beaker to prepare the required final considered concentration. The required mass of Al(OH)3 was then added to obtain NaAl(OH)4 (1.7 m). The mixture was heated to 368 K to yield an optically clear solution. In some experiments, the required mass of sodium carbonate was dissolved into the water and combined with the sodium aluminate solution. 17500
dx.doi.org/10.1021/ie502693p | Ind. Eng. Chem. Res. 2014, 53, 17499−17505
Industrial & Engineering Chemistry Research
Article
highly alkaline solutions.18 A series of experiments have been carried out to determine the effect of carbonate concentration on TCA solubilities and crystalline products at 368 K. The initial dissolved sodium hydroxide and aluminum hydroxide concentrations were kept constant and the effect of dissolved sodium carbonate within the Bayer range on the TCA equilibrium solubility was studied. Table 1 summarizes the experimental conditions and results. As shown in Figure 3, the solubility of TCA based on calcium concentration was observed to increase with carbonate concentration at 368 K. The presence of carbonate also caused a subtle structural shift in the TCA, as shown by the XRD patterns of the solids recovered from the solubility experiments in Figure 4. With no added sodium carbonate (Na2CO3 ≈ 0.03 m, from impurities in the source chemicals), the TCA peaks closely match those of the synthesized material. XRD analysis of the tricalcium aluminate indicated that the eight most intense peaks corresponded to the d-spacings of 5.1, 4.4, 3.3, 2.9, 2.8, 2.3, 2.0, and 1.7. In TCA, impurity reflections were also present in the diffraction pattern. They were detected at 2θ values of 34.5° and 36.0°, where the peak of 36.0° (3 Å) corresponds to the characteristic peak of Na2CO3·H2O. XRD analysis of the solids remaining after 24 h at 368 K (Figure 4) indicate the presence of some double diffraction peaks (denoted by inverted solid hexagonals, ⬟), which are caused by the presence of two unit cell sizes. This may be caused by the diffusion of carbonate into crystalline products. With increasing carbonate content, the shift of the peak to higher angles was observed (dotted line, ···). It was observed that the lattice parameters (a-axis, unit-cell volume) decreased from (12.57, 1986.12) to (12.56, 1983.75) when the carbonate was increased, which can be due to the substitution of CO32− for OH−. It is also obvious that the breadth of the peaks vary. From the XRD line broadening, we can observe that the presence of carbonate leads to a decrease in the peak intensities observed. Perhaps fine TCA particles are unstable and react to carbonate after 24 h. In the presence of carbonate, calcium monocarboaluminate is stable over a range of carbonate concentration. Thus, the TCA solubility and crystal structure are both sensitive to the concentration of solution carbonate. The XRD pattern from residual TCA after 2 weeks of equilibrium time (represented by the dotted line) showed that there was not any evidence of kinetic effects on the peak shifting, although the intensity of some of the peaks seemed to increase with time. Note that no evidence of monocarboaluminate was observed, since the main standard peaks for monocarboaluminate are observed at 2θ = 11.7°, 23.4°, 32.8°, and 40.7°, as shown in Figure 4. It was reported, upon the addition of calcium hydroxide, that there is an immediate reaction with the aluminate ion to produce the hemicarbonate.19 The formation of hemicarbonate was not observed in this study. We also measured the solids carbon content using LECO TruSpec CHN analyzer and no significant change was observed upon exposure to carbonate solution. The total carbon concentration of the TCA was ∼1.6 ± 0.2 wt %. Carbonate could enter the structure after drying and before LECO analysis.
3. EXPERIMENTAL RESULTS 3.1. Solid Characterization. XRD analysis of the synthesized TCA indicated that the nine most intense peaks corresponded to d-spacings of 5.1, 4.4, 3.3, 3.1, 2.8, 2.6, 2.5, 2.4, and 2. XRD patterns for the synthesized TCA is in close agreement with the reference XRD pattern.16 SEM analysis (Figure 1) showed that the majority of the synthesized TCA particles are very small (up to a few micrometers, with agglomerates up to 10 μm). The particles are round and agglomerated. These observations are similar to the morphology of TCA described by Whittington and Cardile.17 The ratio of calcium to aluminum was determined to be 1.5 by solids digestion/ICP solution assay, which confirm the TCA stoichiometry. The particle size distribution of the TCA, as determined by the Accusizer, is shown in Figure 2. It can be seen that the mean diameter is 9.0 μm. The distribution appears to be monomodal, with the exception of the fine particle fraction (
