Znd. Eng. Chem. Res. 1994,33, 2107-2110
2107
Synthesis of Zeolite A from Calcined Diatomaceous Clay: Optimization Studies Biswajit Ghosh,t Dinesh C. Agrawa1,t and Subhash Bhatia'J Materials Science Programme and Department of Chemical Engineering, Indian Institute of Technology, P.O. Z.Z. T.,Kanpur 208 016 (UP),India
Optimization studies are carried out for the synthesis of zeolite A by a hydrothermal method using calcined diatomite. Fractional factorial design of experiments along with path of steepest ascent is used to optimize the composition and process parameters. A surface methodology technique is used to obtain the optimum value of the process parameters t o maximize the percentage crystallinity of zeolite. It is found that SiOdA1203 molar ratio of 1.31, NazO/SiOz molar ratio of 3.61, and temperature of hydrogel formation of 35.7 "C give maximum crystallinity of 92% ' at the crystallization temperature of 110 O C and reaction time of 51 h. 1. Introduction
2. Experimental Section
Zeolite A, a water-insoluble, finely dispersed ion exchanger, has been incorporated into detergents in many countries to replace phosphates in laundry detergents. All the ecological and toxicological criteria for detergent formation can be met by replacing sodium triphosphate by zeolite A. The rapid increase in the consumption of zeolite A calls for further work aiming at still cheaper raw materials for its synthesis. Clay minerals can constitute one such raw material. The synthesis of zeolites from kaolins is well-known and described by Barrer et al. (1974). Other raw materials such as highsilicabauxites, halloysites, and montmorillonites are also used for synthesis of zeolites (Aiello et al., 1984). Synthesis of zeolite A from other sources has also been reported (Breck, 1974;Fernandez et al., 1974; Drag et al., 1985; Van Erp et al., 1987; Szostak R., 1989; and Hu and Lee, 1990). Diatomite (also known as diatomaceous earth, kieselguhr, tripolite, etc.) is easily available in large quantities at an extremely low cost. Diatomite on complete calcination yields porous, cellular, and highly gray coloredmaterial which contains silica up to 89 w t % Being cellular, this silica is in a highly reactive state. As such, diatomaceous clay is an important source of silica, which on leaching with sodium hydroxide offers great potentialities for its use in the synthesis of zeolite catalysts. However, synthesis of zeolite from diatomite is a structurally and chemically complicated problem, and depends on a large number of factors. No efforts have been made in the past to use diatomite as a source of silica and alumina to synthesize zeolite A. In the present investigation the optimum conditions under which zeolite A can be synthesized from diatomite using the factorial design technique have been investigated. The important variables chosen in this work are the molar ratio of SiOz/A1203 of the starting mixture, NazO/A1203 ratio of the starting mixture, the temperature, and the time of formation of zeolite. With the use of design of experiments methodology an attempt has been made in the present work to find the optimum values of the different parameters needed for the synthesis of zeolite A. Crystallinity and yield of the zeolite sample have been maximized.
2.1. Materials. In the present study diatomite, sodium hydroxide, and aluminum hydroxide gel were used as the starting materials for the synthesis of zeolite A. Sodium hydroxide was in the form of pellets supplied by M/s Ranbaxy Laboratories, New Delhi. Aluminum hydroxide was supplied by M/s Robert Johnson, Bombay. The diatomaceous clay was obtained from M/s Clay & Refractories, Jodhpur. It was in the processed form and of 98% purity. Its chemical composition is given in Table 1. 2.2. Apparatus. The reaction and crystallization were conducted in a high-pressure Parr reactor. This unit consists of a stainless steel vessel of 1000-mL capacity, designed to withstand a pressure 100 atm and a temperature of 350 "C. A motor-driven stirrer is provided. A gas release valve is provided to release the pressure in the reactor at the end of the run. The autoclave has a temperature-indicator controller (Indotherm Model 401 supplied by Indotherm Instruments Co.,Bombay) together with an electrical heater assembly. This facility enables the control of the temperature of reaction mixture to an accuracy of f l "C. 2.3. Experimental Procedure of Synthesis. Diatomite was first calcined to 900 "C for 4 h, which is beyond the dehydroxylation range. During this process the diatomite loses the hydroxyl ions from its structure and an amorphous product is formed. Calcined diatomite is treated with 1:lHC1 to remove iron. The requirement of silica was met from diatomite itself, and no other source was used for the same. Hence silica/alumina ratio during the reaction was maintained by adding additional alumina in the form of aluminum hydroxide to the reaction mixture. An aluminosilicate gel was prepared by treating calcined diatomite and aluminum hydroxide gel with aqueous NaOH solution and with continuous stirring by a magnetic stirrer for 8-10 h a t temperatures ranging between 25 and 30 "C. This reaction mixture was charged into the autoclave for hydrothermal reaction after adding required amount of water and raising the temperature to the requisite extent. At the end of each run conducted for a specific period of reaction, the product was centrifuged and washed thoroughly with distilled water so that there is no sodium aluminate trapped in the pore of synthesized zeolite which is then oven dried at 120 "C for 7-8 h. In the synthesis procedure the temperature of reaction was varied from 90 to 100 "C, and the reaction time from
.
* To whom correspondence should be addressed. Materials Science Programme. t Department of Chemical Engineering.
t
0888-588519412633-2107$04.50/0 0 1994 American Chemical Society
2108 Ind. Eng. Chem. Res., Vol. 33, No. 9, 1994 Table 1. Chemical Composition of Supplied Diatomaceous Clay component silica alumina water iron (as Fe3+) calcium (as Ca2+) magnesium (as Mg2+)
wt%
82.5 5.9 10.4 1.2 negligible negligible
24 to 36 h. The molar ratio of SiOz/A1203was varied from 1to 1.5. The molar ratio of H20/A1203was kept constant as 210. The reaction products were characterized for the presence of zeolite A and percent crystallinity by X-ray diffraction. These and other characterization processes are reported separately. 3. Results and Discussion
In the present investigation, the optimum values of different variables were obtained in order to get maximum crystallinity of zeolite sample. Effects of different variables like SiOz/A1203 mole ratio, reaction temperature, and time of reaction on percent crystallinity and percent yield of zeolite A were studied. The percent crystallinity and percent yield are defined as area of XRD peak of product % crystallinity = area of XRD peak std sample % yield =
molar weight of zeolite A formed molar weight of aluminosilicate
Zeolite 4A was taken as the standard for percent crystallinity calculation. Response surface methodology (RSM) was used for this optimization study. This technique has been discussed elaborately by Cochran and Cox (19571, Hill and Hunter (1966), Kittrell and Erjavec (19661, Adler et al., (19751, Box et al. (1978), Davies (1978), and Khuri and Cornel1 (1987). The response surface methodology (RSM) cuts down the experimental effort by making use of experimental design which permits the experimenter to assess the strength of the interactions between the factors while varying them simultaneously. RSM consists of several steps: 1. The first step is to design a set of experiments and conduct them to get reliable estimates of the parameters. 2. A suitable mathematical model is proposed to fit experimental data, and then test for model adequacy is done through lack-of-fitF-tests (Draper and Smith, 1981). 3. Optimum values of the independent variables which will produce the maximum (or minimum) value of the response are estimated. At a point which is remote from the optimum, there is little curvature in the true response surface and first-order models will be satisfactory to describe the response surface. In the vicinity of the optimum, however, higher order models will be required due to the presence of curvature in the response surface. RSM is sequential in nature. Some experiments are carried out, valuable information is gathered, and the next stage is designed for getting better values of the response. Generally the method of steepest ascent (or descent) is applied for moving sequentially along the direction of maximum (or minimum) increase (or decrease) in response. 3.1. Variable Identification and Their Levels. Numerous parameters are involved in the synthesis of zeolites. Factors which influence the crystallization of zeolites are the source of starting materials, composition
Table 2. First Order Response Surface Strategy (First Move): Levels of Factors base lower higher code level level level 1.5 Si02/A1203(mole) ratio in starting mix x1 1.25 1 NazO/SiOz (mole) ratio in starting mix x2 3.45 2.5 4.4 ~3 13.5 13 14 pH of reaction mixture temp of hydrogel formation ("C) ~4 37.5 25 50 temp of Crystallization ("C) 315 95 90 100 reaction time (h) x6 30 24 36 factor
Table 3. The 2e2 First Order (First Move) Design Matrix and Values of the Responses factors in coded form xS
runno. 11 13 14 6 5 12 9 10 16 8 4 7 2 3 15 1 17 18 19
x1
x2
-
-
+ + +
+ + + + +
+ + -
+ + + + + +
x3
-
+ +
+ + + + + +
x4
(x&)
-
++-
-
+ + + + + + + +
-
++ + + -
-
values of responses x6
(xlx4)
+++-
+-
++++
+ + Repeat Trials + + + - + + + + - + + + + - +
% crystallinity 77 60 84 87 69 91 78 65 77 48 72 57 57 82 48 82
% yield 51 51 50 51 50 52 49 52 51 51 50 50 50 51 50 48
71 60 86
51 52 51
of the hydrogels (Si02/A1203,NazO/A1203, etc.), nature of the organic or inorganic template, nature of the mineralizing agent, gel concentration, gel alkalinity, sequence of addition of the reactants, rate and period of reactant mixing, temperature and duration of the aging phase, temperature and duration of the crystallization phase, stirring or not, seeding, etc. In a particular case, such as synthesis of zeolite A, only a few of these parameters are involved in the crystallization process; therefore the factorial design technique may be useful for reducing the time required for the optimization of the synthesis. The main parameters involved in the synthesis of zeolite A influencing the percent yield and percent crystallinity are the following: molar ratio of SiOz/A1203; molar ratio of NazO/SiOz;pH of the mixture; gel formation temperature; crystallization temperature; crystallization time. After knowledge of the important variable influencing the process, a base level had to be chosen within the experimental region. The base levels of the factors were chosen on the basis of a priori knowledge (Barrer, 1982). The variation interval of a factor when added or subtracted from the base level gives the upper or the lower level of the factor respectively. The levels of these variables in the first stage of design are given in Table 2. The responses considered were percent crystallinity and percent yield of zeolite A. 3.2. Selection of Experimental Design. Since there are six variables, 2'3-2 fractional factorial design around the base levels was employed. The design matrix is shown in Table 3. The basis for choosing x5 = xlx3 and x 6 = x1x4 is that these combinations are expected to have minimum interactions. Experiments were carried out in a randomized sequence to avoid bias, on the part of experimenter. The values of all the responses (percent crystallinity, percent yield) for the first move of experiments (runs
Ind. Eng. Chem. Res., Vol. 33, No. 9,1994 2109 Table 4. First Order Response Surface Strategy (First Move): Test for the Adequacy of the Model for Percent Crystallinity Model: 9, = 71.04 + 0.70021+ 0.5822 + 0.6423 - 5.4324 + 7.9626 + 9.36x6
ANOVA sum of squares
source residual pure error lack of fit
deg of freedom
4161.25 12 390.69 3 3770.55 9 F d = 418.951130.23 = 3.22 Fo.os(e.3)= 8.81
Table 6. First Order Response Surface Strategy (Second Move): Test for the Adequacy of the Model for Percent Yield Model: yy= 50.45 + 0.2521- 0.5422 - 0.1723 - 0.342, + 0 . 2 1 -~ ~ 0.3928 - 0.173Xz~- 0.17xg4 - 0.27X226 - 0.342226 0.482&
ANOVA
mean square 130.23 418.95
sum of sauares
source residual pure error lack of fit
Model:
yY= 50.45 + 0.2521-
0.53~2- 0.1628 - 0.3424
model is adequate
1.78 0.15 3 1.63 9 F d = 0.18/0.05 = 3.6 Fo.os(9.w= 8.81
sum of sauares
source residual pure error lack of fit
Fd model is not adequate
deg of freedom
+ 0.2125 -
10.00 0.15 3 9.85 9 1.09/0.05 21.77 Fo.os(s.3)= 8.81
mean sauare 0.05 1.09
Factor
60,
base level unit estd slope unit x ( b ) change in level per 5 "C change in x5
1.25 0.5 0.70 0.35
3.45 1.5 0.58 0.87
0.02 0.05
13.5 1 0.64 0.64 0.04
37.5 5 -5.43 -27.15 -1.70
95 10 7.96 79.63 5
30 12 9.36 112.33 7.05
Path of Steepest Ascent As Represented by a Series of Trial Pointa
numbered 1-16) as planned in the design matrix are given in Table 3. Experiment no. 10 was repeated four times so as to obtain an estimate of the error. 3.3. Model Fitting. The following first-order model has been fitted to the experimental data:
where
0.05 0.18
Table 7. First Order Response Surface Strategy (First Move): Calculation of the Path of Steepest Ascent
0.3926
ANOVA
mean sauare
~~
model is adequate
Table 5. First Order Response Surface Strategy (First Move) Test for the Adequacy of the Model for Percent Yield
deg of freedom
..., 6 6 are the best fitted values of coefficients,
Ri is the ith variable in its coded form, and Yis the predicted value of the response. The coefficients can be estimated by the least squares technique as
6 = (Y'm-lY'Y where X i s the design matrix. Xr is the transpose of the design matrix, Y is the vector responses, and 6 is the vector of coefficients. The details of the method are given by Draper and Smith (1981). This is then tested for model adequacy by a lack-of-fit F-test. With the experimental results given in Table 3, the fitted first-order models for percent crystallinity and percent yield are obtained as follows:
FC= 71.04 + 0.70~'+ 0 . 5 8 ~+ ~0.6x3- 5 . 4 3 +~ ~7 . 9 6 ~+~ 936x6 (3)
F,,= 50.45 + o.25X1 - o.537x2- 0.16~3- o.34x4 + o.21x50.39x6 (4) where Fcand p,,represent the predicted values of percent crystallinity and percent yield, respectively. The lackof-fit F-tests for the above models are shown in Tables 4 and 5, respectively. It is observed from Table 5 that a simple first-order model is not adequate to describe the percent yield data. Hence a higher order model incorporating cross-product terms has been proposed and tested for the lack-of-fit F-test. This is given in Table 6. 3.4. Calculation of the Path of Steepest Ascent and Conduct of Experiment along This Path. Information
trial point 1 2 3 4
x1 1.27 1.29 1.31 1.38
22
23
24
3.50 3.55 3.61 3.82
13.54 13.58 13.62 13.72
39.20 37.46 35.76 34.54
26
2-5
100 105 110 115
37.05 44.1 51.15 58.2
obtained from the models was used to locate the path of maximum increase in crystallinity. The method of steepest ascent is a procedure for moving sequentially along the direction of maximum increase in response. The direction of steepest ascent was determined using the relation
v4 = (64/6xl)nl + (64/6x2)2, +
+ (6$/6xk)Ck
(5)
where 4 is the function describing the response surface, V4 is the gradient of the response function, 6416x1 is the partial derivative of the function with respect to the ith factor, and Cl,62, ...,f i k are the unit vectors in the direction of the coordinate axes. It can be easily verified that the components of the gradient V4 Le., 6416x1 (i = 1,2, ...,k) are the same as the regression coefficient variables in eq 1. Thus by changing the independent variables x1-x6 in proportion to the values of their corresponding coefficients, the movement along the steepest path may be realized. From the coefficient of the fitted model for crystallinity (eq 31, it is easy to compute these factor-level combinations which predict an increase in the percent crystallinity of zeolite A. The calculation of this path is shown in Table 7. The details of these calculation are described below. The first two rows of the table indicate the values of the base level and variation intervals of different factors. In the third row the regression coefficients of theemode1 are given. These coefficients represents the gradient of the crystallinity in their respective directions. The fourth row was obtained by multiplying the unit of each factor with its coefficient so as to change the factor level in proportion to its slope (Le., its regression coefficient). In the fifth row step changes in the factorsxg and x1-xq were calculated corresponding to a change of 5 O C in the direction of factor
2110 Ind. Eng. Chem. Res., Vol. 33, No. 9, 1994
Table 8. First Order Response Surface Strategy (First Move): Results of Experiments along the Path of Steepest Ascent run no. 20 21 22 23
trial point 1 2 3 4
exptl (%) 89 91 92 90
predicted (%) 80f 11 82f 11 83f 11 84f 11
The choice of the size of the step change in factor xg is a matter of experimental convenience. The fifth row was then added (element by element) to the original base levels in the first row to get the elements of the first trial point in the path of the steepest ascent. The second trial point in the path is obtained by adding the first point to the elements of fifth row. This process is then iterated to get the other points of the path. Additional experiments were then performed (runs numbered 20, 21, 22, 23) according to trial points 1-4. The experimental results and values predicted by the first-order model (eq 3) are given in Table 8. From Table 8 it is observed that a maximum crystallinity of 92% can be achieved under the following condition: x5.
SiO,/Al,O, (mole) ratio ( x , ) = 1.31 Na20/Si02(mole) ratio (x,) = 3.61 pH (x,) = 13.6 temp of hydrogel formation (x,) = 35.7 OC temp of crystallization (x,) = 110 "C reaction time
(x6)
= 51 h
The optimum yield of zeolite A obtained was 52.0%.
Conclusions Synthesis of zeolite A has been carried out from calcined diatomite in the temperature range of 90-110 "C with reaction time varying from 8 to 52 h. Optimum values of different variables such as SiOdA1203molar ratio, Na2O/ Si02 molar ratio of reaction mixture, temperature, and time of reaction are obtained using fractional factorial design along with path of steepest calculation. The responses (percent crystallinity and percent yield) can be described in terms of fitted models.
Nomenclature 6 6. ..., 6 6 = parameters in eq 1 $'= :ector of coefficients in eq 2
xi
= ith factor in coded form
X = design matrix ? = predicted value of response Y = vector responses in eq 2 yc= predicted 5% crystallinity of zeolite A ?, = predicted ?4 yield of zeolite A 4 = function describing the response surface in eq 5
Literature Cited Adler, Yu. P.; Markova, E. V.; Granovsky, Yu. V. The Design of Experiments to Find Optimal Conditions; Mir Publishers: Moscow, 1975;pp 118,234. Aiello, R.; Colella, C.; Casey, D. G.; Sand, L. B. In Proceedings of the 5thZnternational Conference on Zeolites; Heyden: London, 1984; p 49. Barrer, R. M. Hydrothermal Chemistry of Zeolites; Academic Press: New York, 1982;Chapters 3-5. Barrer, R. M.; Beaumont, R.; Colella, C. Chemistry of Soil minerals Part XIV Action of some basic solutions of metakaolinite and kaolinite. J . Chem. SOC.,Dalton Trans. 1974, 934. Box, G. E. P.; Hunter, W. G.; Hunter, J. S. Statisticsfor Experiments; John Wiley & Sons: New York, 1978;Chapters 10 and 12. Breck, D. W.Zeolite Molecular Sieves; John Wiley & Sons: New York, 1974. Cochran, W. G.;Cox, G. M. Experimental Designs, 2nd ed.; John Wiley & Sons: New York, 1957. Davies, 0. L. The Design and Analysis of Industrial Experiments; Longman Group Limited: New York, 1978;Chapter 11. Drag, E. B.; Mieczikowski, A.; Abo-Lemon, A.; Rutkowski, M. Synthesis of A, X and Y Zeolites from Clay Minerals. In Zeolites Synthesis, Structure, Technology and Application; Drzay, B., Hocevar, S., Pejovnik, S., Eds.; Studies in Surface Science and Catalysis 24;Elsevier: Amsterdam, 1985;p 147. Draper, N. R.; Smith, H. Applied Regression Analysis, 2nd ed.; John Wiley & Sons: New York, 1981;pp 1-125. Fernandex, I.; Vivaldi, J. L. M.; Pozzuoli, A. Bol. Geol. Min. 1974, 8445,442. Hill, W. J.; Hunter, W. G. A Review of Response Surface Methodology: A. Literature Survey. Technometrics 1966,8 (4),571-590. Hsin, Hu C.; Tsug, Lee Y. Synthesis Kinetics of Zeolite A. Znd. Eng. Chem. Res. 1990, 29 (5),749. Kittrell, J. R.; Erjavec,J. Response Surface Methods in Heterogeneous Kinetics Modelling. Ind. Eng. Chem. Process Des. Dev. 1966, 7 (3),321-327. Khuri, A. I.; Cornell, J. A. Response Surfaces Designs and Analysis; Marcel Dekker, Inc.: New York, 1987;Chapters 1-3. Szostak, R. Molecular Sieves-Principles of Synthesis and Zdentification; Van Nostrand Reinhold New York, 1989. Van Erp, W. A.; Kouwenhoven,H. W.; Nanne, J. M. Zeolite Synthesis in Nonaqueous Solvents. Zeolite 1987, 7, 286. Received for review March 8, 1993 Accepted October 12, 1993' @
Abstract published in Advance ACSAbstracts, July 15,1994.
