Desilication Reactions at Digestion Conditions: An in Situ X-ray Diffraction Study D. Croker,* M. Loan, and B. K. Hodnett Materials & Surface Science Institute, UniVersity of Limerick, Ireland
CRYSTAL GROWTH & DESIGN 2008 VOL. 8, NO. 12 4499–4505
ReceiVed May 7, 2008; ReVised Manuscript ReceiVed June 25, 2008
ABSTRACT: The reactions of silica minerals at high temperature Bayer digestion conditions have been investigated using an in situ X-ray diffraction technique. Rapid precipitation of sodalite was observed following the addition of kaolin or quartz to Bayer liquor at 250 °C. The addition of calcium with kaolin/quartz resulted in the formation of calcium cancrinite. A reduction in the rate of both sodalite and calcium cancrinite formation was observed on decreasing the temperature in the range 180-250 °C. Seeding with sodalite/calcium cancrinite solids did not noticeably influence the reaction of quartz at 250 °C, but a slight increase in the formation of calcium cancrinite in the presence of calcium cancrinite seed was observed at 220 °C. The direct transformation of sodalite to calcium cancrinite was also investigated in the range 200-250 °C. The reaction proceeds quickly in the presence of calcium and hydrogarnet and is unaffected by the presence of titanium. Introduction Silica is an integral but undesirable component of Bayer process liquors. Bauxite typically contains 0.5-10 wt% silica, primarily in the form of kaolin (Al2O3 · 2SiO2 · 2H2O) and/or quartz (SiO2).1 Reactions involving kaolin/quartz are generally termed desilication reactions, and the products of these reactions identified as desilication reaction products (DSP). DSP is disposed from the Bayer circuit with the bauxite residue waste, constituting a caustic and aluminum loss from the process. Any silica not converted to DSP remains in the pregnant liquor and moves on to the precipitation stage where it can co-precipitate with gibbsite, causing product quality issues,2 and deposit as scale on the tube side of heat exchangers,3 reducing heat transfer efficiency. Kaolin dissolves quickly upon exposure to caustic and temperature ((1) > 80 °C) and begins to react at the predesilication stage of the Bayer process. The majority of kaolin will be transformed to DSP (2) within predesilication, but a certain amount of kaolin is held within the gibbsite matrix, and this will not react until the gibbsite itself dissolves at digestion conditions. The formation of DSP from kaolin consumes one tonne of caustic/ per tonne of SiO2 attacked. As there is alumina associated with kaolin, there is no net loss of alumina on the transformation of kaolin to DSP. DSP forming in predesilication will be in the form of sodalite (3(Na2O · Al2O3 · 2SiO2 · 2H2O) · Na2X). The sodalite structure is made up of cages of Al6Si6O24 which can accommodate other ions, X in eq 2 below.
Al2O3 · 2SiO2 · 2H2O + 6NaOH f 2Na2SiO3 + 2NaAlO2 + 5H2O (1) 6Na2SiO3 + 6NaAlO2 + 8Na+ + 2X- + (y - 6)H2O f 3(Na2O · Al2O3 · 2SiO2 · 2H2O) · Na2X + 2(2) 12NaOH, X ) 1 ⁄ 2CO23 , 1 ⁄ 2SO4 , OH , Cl
Roach and White (1980) investigated the dissolution of kaolin in synthetic sodium aluminate solution at 95 °C and found the reaction rate to be controlled primarily by the concentration of free caustic in solution.4 Increasing alumi* To whom correspondence should be addressed. E-mail: Denise.Croker@ ul.ie.
num concentrations (increasing A/C) were found to decrease the rate of kaolin dissolution slightly while the carbonate content of the liquor had a negligible effect on the rate of dissolution. An increase in reaction rate was reported on increasing the temperature in the range 60-150 °C and an activation energy of 93 kJ mol-1 reported for the dissolution of pure kaolin. The reaction was reported as being under chemical control within this temperature range as agitation had no effect on the rate of dissolution. Banvolgi et al. (1991) proposed a mechanism whereby kaolin transforms to sodium aluminum silicate without significant dissolution in the digestion of high silica bauxite (>5% SiO2) in caustic solution at 80-130 °C.5 Using SEM and IR techniques the author concluded that OH- and Na+ ions react with the outer layers of gibbsite in the kaolin lattice, but the Si-O bonds resist attack and transform gradually to sodalite. Sodium hydroxide concentration was again cited as the driving force for kaolin dissolution. The kaolin reaction has also been described as occurring in three stages at 100 °C.6 In the first 50 min kaolin dissolves gradually, increasing solution silica levels. At about 50 min, precipitation of a desilication product (no details given) begins. There is a rapid decrease in the silica concentration in solution and increase in the rate of kaolin dissolution. After 135 min, the rate of dissolution of kaolin slows as the concentration of DSP stabilized. In this way, the fastest rate of kaolin dissolution is observed once silica levels have exceeded supersaturation. Buhl et al. (1997) followed the hydrothermal transformation of kaolin to sodalite in pure sodium hydroxide using nuclear magnetic resonance (NMR) and report complete transformation of 2 g kaolin after 2 h at 200 °C.7 The rate of transformation was enhanced by increasing temperature (in the range 80-200 °C) and by the presence of carbonate in solution. The predesilicated bauxite slurry moves onto digestion where the remaining kaolin, the newly formed sodalite and the existing quartz are available to react. On dissolution of the gibbsite matrix, kaolin is free to react as before. Quartz is less susceptible to caustic attack than kaolin and only begins to react at 180 °C (3).2,8 Hence, quartz dissolution is restricted to high temperature digestion processes.
10.1021/cg8004739 CCC: $40.75 2008 American Chemical Society Published on Web 10/22/2008
4500 Crystal Growth & Design, Vol. 8, No. 12, 2008
Croker et al.
Figure 1. Sodalite to cancrinite transformation as described by AddaiMensah et al. (1997).12
Figure 2. Sodalite to Cancrinite transformation as described by Barnes et al. (1999).11
SiO2 + 2NaOH f Na2SiO3 + H2O
(3)
The rate of quartz dissolution is said to be much slower than that of kaolin and may be restricted by controlling the residence time in digestion. As there is no alumina associated with quartz to begin with, the conversion of quartz to DSP constitutes a loss of one ton alumina and one tonne caustic per tonne SiO2 processed and as such is extremely undesirable. Oku and Yamada (1971) studied the dissolution rate of quartz in the digestion of monohydrate bauxite at 180-240 °C.9 The dissolution of quartz was found to be first order with respect to quartz surface area, controlled by the chemical reaction of OHat the surface of the quartz and an activation energy of 19.7 kcal mol-1 (82 kJ mol-1) calculated for the reaction. The presence of dissolved aluminum ions has been reported to inhibit quartz dissolution.10 The high temperatures employed in digestion make possible formation of an additional form of DSP-cancrinite. Cancrinite is stoichiometrically identical to sodalite but it differs in structure, consisting of linear channels and aluminosilicate cages. Sodalite is less stable than cancrinite and has been reported to precipitate first and transform to cancrinite at a rate that increases with increasing temperature.3 The reaction will not proceed through solid state and requires a liquid medium.11 Two mechanisms have been described for the transformation. AddaiMensah et al. (1997) investigated desilication at 100/160 °C using sodium metasilicate in synthetic liquor of variable carbonate concentration and reported precipitation of two forms of sodalite prior to cancrinite formation (Figure 1).12 Sodalite1 has a higher carbonate content and larger unit cell (a ) 9.077 Å) than sodalite2 (a ) 8.988 Å) and is the first to precipitate. Sodalite2 subsequently transforms to cancrinite, this reaction being the slow step in the scheme. This is supported by Hermeler et al. (1991) who also report precipitation of two forms of sodalite differing in unit cell dimensions from 8.965 Å (Sodalite I) to 8.854 Å (Sodalite II).13 Again, the larger unit cell appears first and diminishes on size upon transformation to the second sodalite form. Barnes et al. (1999) report a different scheme whereby an amorphous phase is the first to precipitate (Figure 2).11 The authors proposed that sodalite dissolution is necessary for cancrinite precipitation but cancrinite may also coprecipitate with sodalite from the liquor. The addition of calcium to high temperature digestion can result in the formation of alternative DSP, namely, hydrogarnet and calcium cancrinite.14 Calcium cancrinite is formed by substitution of calcium for sodium in the (extra-framework) cancrinite structure. Its formation is beneficial in that is reduces the caustic loss associated with DSP formation by the calcium substitution. Hydrogarnet (Ca3Al2(SiO4)n(OH)12-4n) is a silicacontaining member of the tricalcium aluminate (TCA) group. In short, there is a wide range of reactions involving silica simultaneously occurring at high temperature Bayer digestion conditions. Although there is much literature available on the reactions of kaolin and quartz at Bayer conditions, little desilication
information pertaining specifically to high temperature digestion is in circulation. In situ X-ray diffraction offers the possibility to follow phase formation at digestion conditions.15,16 The application of this technique to Bayer digestion studies offers the opportunity to explore reactions as they occur in real time. This article aims to generate some understanding as to the behavior of kaolin and quartz at conditions specific to the high temperature digestion process practiced at Rusal Aughinish. Experimental Methodology All experiments were peformed at the 7T-MPW-EDDi beamline at the Berliner Elektronenspeicherung Gesellshaft for Synchrontronstrahlung (BESSY) in Berlin, Germany. This station provides a white beam (energy 10-150 keV, optimum flux between 20 and 80 keV) for sample irradiation. The diffracted beam is detected using one detector, the position of which (2θ) determines the range of d-spacing returned as per eq 4, where E ) energy of diffracted wavelength (keV), d ) d-spacing specific to sample (A°) and θ ) half the angle of detector (degrees).
E)
6.19926 d sin θ
(4)
For these experiments the detector position was maintained at 4 degrees 2θ, equating to d spacing 2.5-4.5 Å in the range 40-80 keV. The apparatus was previously described in Loan (2005).15 Slight adjustments have since improved the operation of the vessel. Addition of solids to liquid was achieved at temperature by application of ∼5 bar pressure to effectively “blow” solids into the solution. In this manner a blank “background” scan through the solution could be obtained prior to the addition of solids. Reagent grade kaolin (K7375, Sigma Aldrich) and quartz (342890, Sigma Aldrich) were used. The B.E.T surface areas were determined as 6 m2 g-1 (quartz) and 21 m2 g-1 (kaolin) using an ASAP 2010 instrument. Bayer industry nomenclature was maintained to represent the concentration of dissolved aluminum hydroxide (A, expressed as g L-1 Al2O3) and the concentration of dissolved sodium hydroxide (C, expressed as g L-1 Na2CO3)) of Bayer-type (sodium-aluminate) solutions. The A/C ratio represents the alumina/caustic ratio of a solution. Sodium-aluminate liquor was obtained from Rusal Aughinish Ltd. and adjusted to a working concentration of C ) 200 g L-1, A/C ) 0.7 using DI H2O and gibbsite (C31 Hydrate, Alcoa Australia Ltd.). This solution was used for all experiments unless otherwise stated. The starting silica concentration in solution was ∼0.5 g L-1. Sodalite was prepared from the reaction of 2 g of kaolin (K7375, Sigma-Aldrich) in 25 mL of Rusal Aughinish Ltd. process liquor (C ) 240 g L-1 as Na2CO3, A/C ) 0.4, 26 g L-1 Na2CO3) at 120 °C for 1 h. Calcium cancrinite was prepared from the stoichiometric reaction of this sodalite with calcium carbonate in sodium aluminate solution (C ) 240 g L-1 as Na2CO3, A/C ) 0.7, 20 g L-1 Na2CO3) at 250 °C for 1 h. Hydrogarnet was prepared by reacting lime and kaolin, (Ca/Si molar ratio 3:1) in sodium aluminate solution (C ) 240 g L-1 as Na2CO3, A/C ) 0.4, 5 g L-1 Na2CO3) at 200 °C for 1 h. The resulting solids were characterized using X-ray diffraction to ensure correct composition. The hydrogarnet solids showed a best match to PDF file 38-0368 (Ca3Al2(SiO4)(OH)8, “n” ) 1) and were found to contain slight impurities in the form of sodalite. All other reagents were laboratory grade (Sigma-Aldrich).
Results and Discussion The reaction of kaolin in the described liquor was observed by charging 0.6 g of kaolin to 15 mL of liquor at 250 °C (equivalent silica charge 19 g (SiO2) L-1). In-situ X-ray diffraction patterns were collected at intervals of 60s and demonstrate a rapid reaction to sodalite, as evidenced by emergence of the reflection at 3.6A° (Figure 3A). Kaolin dissolution was not observed as the primary kaolin reflection lies outside the range of detection at the selected 2θ value of 4. The position of the sodalite reflection remained relatively
Desilication Reactions at Digestion Conditions
Crystal Growth & Design, Vol. 8, No. 12, 2008 4501
Figure 3. In-situ X-ray diffraction patterns recorded following the addition of kaolin (0.6 g) to sodium aluminate solution (15 mL) at 250 °C (A). Also shown is the effect of temperature on the extent of reaction of the sodalite reflection and the calculated fits to the Avrami-Erofe’ev equation as per Table 1 (B).
Figure 4. In-situ X-ray diffraction patterns recorded following the addition of kaolin (0.6 g) and a calcium source (0.2 g) to sodium aluminate solution (15 mL) at 250 °C (A). Also shown is the effect of temperature on the extent of reaction of the sodalite reflection (B).
Table 1. Kinetic Parameters Calculated for the Formation of Sodalite Following the Addition of Kaolin (0.6 g) or Kaolin (0.6 g) and a Calcium Source (0.2 g) to Sodium Aluminate Solution, Using Least Square (LS) and Sharp-Hancock (SH) Analysis, as a Function of Temperature SODkaolin T ((2 °C) 180
analysis method
SH LS 200 SH LS 220 SH LS 250 SH LS EACT (kJ mol-1) a
k (s-1)
n 0.8 0.7 1.9 1.9 0.6 0.8 a a 56
( ( ( ( ( (
CANkaolin
0.3 0.2 0.7 0.4 0.5 0.3
3 3 6 6 1 1 a a
× × × × × ×
10-3 10-3 10-3 10-3 10-2 10-2
k (s-1)
n a a a a 1 ( 0.1 1.2 ( 0.1 1 ( 0.1 1.2 ( 0.1 a
a a a a 3 4 1 1
× × × ×
10-4 10-4 10-3 10-3
Insufficient data points for analysis.
constant for the duration of the experiment; two separate forms of sodalite could not be distinguished. Integration of the area under a reflection gives quantitative information on the amount of material present. This may be normalized to an extent of reaction function (R) by application of the formula R ) Ι/Ιmax, where Ι is the integrated peak intensity at any given time and Ιmax is the maximum integrated peak intensity achieved for that reflection. The extent of reaction of the sodalite reflection in the range 180-250 °C is displayed in Figure 3B. A reduction in the rate of sodalite formation was observed on decreasing the temperature, but a significant degree of sodalite formation was still observed at 180 °C. Information regarding reaction mechanism can be obtained through comparison of experimental data (R-time plots) with
Figure 5. In-situ X-ray diffraction patterns recorded following the addition of quartz (0.4 g) to sodium aluminate solution (15 mL) at 250 °C (A). Also shown is the extent of reaction of the sodalite and quartz reflections (B).
theoretical expressions relating R and time.17 The AvramiErofe’ev eq 5 is an expression for growth that has been found to hold over the greater part (0.05 < R < 0.9) of many reactions.
4502 Crystal Growth & Design, Vol. 8, No. 12, 2008
R ) 1 - e-(kt)
n
Croker et al.
(5)
Analysis of the growth of the sodalite reflection was attempted using a least-squares algorithm and the Avrami-Erofe’ev equation. The reaction was so rapid that analysis was restricted to less than 6 points in the region 180-220 °C, which reduces accuracy. The reaction at 250 °C was too fast to generate sufficient data points for analysis. Satisfactory fits to the data were achieved (Figure 3) in the range 0.15 < R < 0.7. These values were substantiated by Sharp-Hancock analysis of the data. The Sharp-Hancock eq 6 is a linearized version of the AvramiErofe’ev equation, the plotting of which (ln(-ln(1 - R)) against ln t) allows calculation of the kinetic parameters, n and k. The reaction exponent, n, is a function of nucleation and growth constants and k is the reaction rate.
ln(-ln(1 - R) ) n ln k + n ln t
(6)
The n and k values calculated at each temperature are presented in Table 1. The values of the exponent could be rounded to 1 or lower. A value of 1 would indicate a reaction under chemical control whereas a low value for the exponent is indicative of a diffusion-controlled reaction. The extremely fast rate of precipitation might suggest that the formation of sodalite is diffusion controlled. This was tested by plotting [1 - (1 - R)1/3]2 against time. This should result in a straight line for a 3-D diffusion controlled reaction. The resultant plots had a poor linear fit which does not support the hypothesis that the formation of sodalite is diffusion controlled. The Arrhenius equation was applied to the data by plotting ln k against 1/T and an activation energy of 56 kJ mol-1 calculated from the resulting straight line. Diffusion controlled reactions typically display activation energies of
