Role of Binder and Preparation Pressure in Tubular Ceramic

Jul 10, 2014 - such as preparation pressure and binder contents on the porosity and ... composition due to the presence of binders in ceramic processi...
0 downloads 0 Views 2MB Size
Subscriber access provided by DALHOUSIE UNIV

Article

Role of binder and preparation pressure in tubular ceramic membrane processing: Design and optimization study using response surface methodology (RSM) Sujoy Bose, and Chandan Das Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/ie500792a • Publication Date (Web): 10 Jul 2014 Downloaded from http://pubs.acs.org on July 16, 2014

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Industrial & Engineering Chemistry Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Role of binder and preparation pressure in tubular ceramic membrane processing: Design and optimization study using response surface methodology (RSM)

Sujoy Bose and Chandan Das*

Department of Chemical Engineering, Indian Institute of Technology Guwahati Assam 781 039, India

*Corresponding author. Tel.: +91 361 2582268; Fax: +91 361 2582291 E-mail: [email protected] (C. Das)

1 ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 31

Abstract An empirical model using response surface methodology (RSM) is proposed to investigate the role of parameters, such as, preparation pressure and binder contents on the porosity and flexural strength of a tubular low-cost ceramic membrane based on central compact design (CCD). Preparation pressure and binder contents are selected as input parameters to obtain controlled porosity with considerable strength of membrane. The optimum preparation pressure is found to be at 9.81 MPa with the sodium metasilicate and boric acid contents of 7.50 % each providing a microfiltration range membrane. The optimization study reveals that the errors between the experimental and predicted values are below 2 %. FESEM images clearly unveil a consolidated microstructure of the ceramic membrane. Acid-alkali test conveys that there is no major change in elemental composition due to the presence of binders in ceramic processing. The cost of the optimized membrane is estimated at around $332/m2.

Keywords: Response surface methodology; Preparation pressure; Porosity; Flexural strength; Chemical stability

1. Introduction Tubular ceramic membranes are well-known for their advantages such as high flux obtainability, less probability of membrane fouling, less energy consumption, high durability. Preparation of tubular membranes is quite challenging in respective of materials and production processes that are complicated and expensive. A series of popular conventional methods extensively used for the production of tubular ceramic membranes are extrusion1, centrifugal casting2-8, sol-gel9-11 etc. But these processes face problems such as low surface smoothness, larger average pore size, highly expensive, high temperature requirement, possibility of nonhomogeneity,

contamination

from

containers

etc.

An

alternative

method

is

dry

pressing/compaction, is cheaper, reliable and economically feasible than all other conventional processes.12-14 Performance of membrane and its durability greatly depends on its morphology and mechanical strength. Therefore, membrane morphological properties such as porosity, pore size distribution and mechanical properties like flexural strength are important parameters in ceramic 2 ACS Paragon Plus Environment

Page 3 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

processing. These parameters mainly depend on binder content, preparation pressure, sintering temperature etc.15-18 Hence, it is essential to control these parameters for achieving better performance membrane as a function of morphology and mechanical strength. In our previous work19, we have fabricated and characterized a tubular ceramic membrane without binder. We have also presented the influence of different sintering temperatures on the morphology, mechanical and chemical stability of a membrane and optimized the sintering temperature experimentally on the basis of morphology, flexural strength and manufacturing cost of the membrane. The membrane fabricated at 850 °C shows 36 % porosity and 2 MPa strength. This observation is in close agreement with literatures in terms of morphology but not in terms of mechanical strength.20 It is clear from our study that mechanical strength is a key parameter in ceramic membrane fabrication and binder plays an important role to improve strength. Hence, it is necessary to add binders and optimize binder content that affects membrane morphology as well as membrane strength. Binder addition in an appropriate ratio and sintering temperature are vital parameters to maintain the uniformity into the membrane texture, thus, providing equal range of pore size and to enhance the mechanical strength. Effect of binder addition on microstructure, mechanical behaviour and electrical properties of Pb((Zn1/3Nb2/3)0.20(Zr0.50Ti0.50)0.80)O3 ceramics have been shown by Zheng et al.16 In our previous work19, we have presented the influence of addition of binder and sintering temperature on the morphology, mechanical and chemical stability of a membrane. This observation is in close agreement with Jana et al.20 who also prepared a ceramic membrane of 17.25 MPa strength with 37 % porosity by varying binder content and sintering temperature. It is clear from the above literatures that binders and sintering temperature play important roles in membrane fabrication. Hence, it is necessary to optimize these parameters that affect membrane morphology as well as membrane strength. Nowadays, several researchers have approached statistical and optimization study to improve membrane morphology21, strength and performances.22-24 Optimization and statistical tool such as response surface methodology (RSM) is found to be the most popular among the researchers and effective in evaluating the interaction between the independent process factors.21,25-26 For example, Khayet et al.27 proposed a design of direct contact membrane distillation (DCMD) using RSM for the prediction of permeate fluxes applied in the DCMD of salt (NaCl) aqueous solutions.

3 ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

In our present study, two binders (sodium metasilicate and boric acid) are added with kaolin, sawdust and feldspar to fabricate ceramic tubular membrane and two parameters i.e., preparation pressure and binder contents are optimized using response surface methodology (RSM). Our study also aims to understand the interaction between binder content and preparation pressure with membrane porosity and flexural strength in statistical analysis using central compact design (CCD) of RSM as no work on the effect of binder content and preparation pressure in ceramic membrane fabrication has been reported so far. A comparison between experimental and predicted results obtained from mathematical model are developed to validate the model and are able to predict the optimum independent parameters (binder content and preparation pressure) in fabricating microfiltration range membrane with high durability. In this study, we have conducted Field emission scanning electron microscope (FESEM), X-ray diffraction (XRD) and energy dispersive X-ray (EDX) analysis to perform a detailed surface morphology and chemical study of the fabricated membrane obtained from optimized composition. The gas permeation test has also been done to calculate the average pore diameter of the optimized membrane. The manufacturing cost of the optimized membrane has also been evaluated.

2. Materials and methods 2.1. Membrane fabrication The tubular shaped membranes (20 numbers) are fabricated by thorough mixing and grinding of sodium metasilicate (Na2SiO3) (Loba Chemie Pvt. Ltd., India) and boric acid (H3BO3) (Merck India Pvt. Ltd., India) along with other raw materials such as kaolin (Al2O3.2SiO2.2H2O) (Loba Chemie Pvt. Ltd, India), saw dust (mixture of Teak, Coco and Sal wood taken from local sawmill) and feldspar (KAlSiO3) (National Chemicals, India) of different weight percentages (with respect to binder content). The membranes are fabricated using mixer machine and ceramic mortar via dry compaction method. Sodium metasilicate (SM) and boric acid (BA) play the role of strength enhancer by forming silicate and metallic metaborates during sintering. According to the design expert, binders are varied from 4.21 % to 10.79 % of the total weight of the raw materials (total weight of sample mixture is 40g). Kaolin offered low plasticity and high refractory properties to the membrane support. In early stages of sintering process, 4 ACS Paragon Plus Environment

Page 4 of 31

Page 5 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

feldspar provided a glassy matrix to induce the bonding among the elements. Porous texture in the ceramics is realized by sawdust. Gyratory sieve shaker (I.S.S. 44) is used to separate the regular shaped particles of sawdust as the particle sizes vary in a wide range. The sample mixture is first placed into a cylindrical mold (50 mm height and 10 mm thickness) made by mild steel. Then the pressure is applied on the sample mixture using hydraulic press (Velan Engineering, Tamilnadu, India) and is varied from 7.23 MPa to 12.39 MPa. Then the mixture is kept under distributed forming pressure for 1 min to prevent the propagation deformation by keeping homogeneity in the raw material mixture. The fabricated membrane of 45 mm height and 10 mm thickness is removed cautiously from the mold and is subjected to heat treatment (sintered at 550, 700 and 850 °C) to harden and strengthen the microstructure. The reason behind the selection of three different sintering temperatures is to verify the change in morphology of the optimized ceramic membrane and phase transformation with sintering temperature. In this study, the optimized sintering temperature is considered as 850 °C on the basis of thermogravimetric analysis (TGA) of the raw materials along with pore former, mentioned in literature.19

2.2. Membrane characterization 2.2.1. Three-point bend test Three-point bend test (ASTM D790) is performed to measure the flexural strength of all prepared membrane using a Universal Tensile Test Machine (Deepak Polyplast Pvt. Ltd., India) applying 20 MPa load and maintaining 85 mm support span at a stroke rate of 0.5 mm/min. Rectangular bar-shaped specimens are prepared at thickness/length ratio of 1:16 for determining the flexural strength at room temperature.

2.2.2. Membrane porosity Meanwhile, the porosity of the membrane is calculated using volumetric porosity determination technique and the equation is: v 

void volume (W2  W1 ) /  H 2O   (W2  W1 ) /  H 2O .Vm bulk volume Vm

5 ACS Paragon Plus Environment

(1)

Industrial & Engineering Chemistry Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

where W2, W1,  H 2O and Vm are the weights of the wet and dry membrane samples, the density of the deionized water and the volume of the fabricated membrane respectively.

2.2.3. Surface morphology The surface morphology and the phase identification of the optimized membrane are analysed by field emission scanning electron microscope (FESEM) (∑igma, Zeiss, USA) and Xray diffraction (XRD) analysis, respectively. A portion of the specimen is cut into a small piece and mounted on an aluminium stab with carbon tape. Then the sample is coated with gold particles under vacuum and mounted on the FESEM plate. XRD study is performed using BrukerTM D8 Advanced series powder diffractometer with Cu Kα radiation of 40kV and 40mA in the 2θ range of 15-80° with 0.05° steps at a rate of 1s per step to verify the change in phase with change in temperature.

2.2.4. Average pore diameter and porosity An experimental set-up (Figure S1) is fabricated for gas permeation study, as shown in supporting information. The set-up used for this experiment consists of a tubular cell mounted on a flat rectangular base plate, made of perspex. The bottom end of the tubular cell is fixed with a flat rectangular plate and the top end is coupled with two rectangular flat plates by four flanges (black sticks). The fabricated membrane is fitted into the hollow sector inside the tubular cell by opening two top flat plates. Pressure gauges are attached with the membrane housing to maintain the gas flow rate at different transmembrane pressures. The gas is purged into the membrane unit through a pipe from the top plate. The gas is then permeated through the outer wall of the tubular cell unit, connected to the bubble flow meter for measuring the gas flow rate at different transmembrane pressures. A rubber balloon is attached at the one end of the bubble flow meter which contains soap solution. The gas permeation experiment is conducted at room temperature. The gas permeation data is utilized to estimate two vital membrane characteristics namely, average pore radius (rg) and effective porosity (ε/q2) according to the following expression 19

6 ACS Paragon Plus Environment

Page 6 of 31

Page 7 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

 rg k = 2.133  l

   v 2  1.6  q

 rg2  1       l    q 2  

  P 

(2)

where, P is the average pressure on the membrane in kPa, v is the molecular mean velocity of the operating gas (m/s), l is the pore length in m, q is the tortuosity factor of the membrane, η is the viscosity of the gas in Pa.s, k is the effective permeability factor in m/s. Eq. (2) simply characterised as a straight line in a graph drawn between k and P. In Eq. (2), the first term (intercept) and the second term (slope) resemble to Knudsen and viscous permeance, respectively. The values of the slope and intercept obtained from the graph are used to evaluate the pore diameter and porosity of the membrane. The effective permeability factor

k=

is calculated as:

QP2 SP

(3)

where, S is the permeable area of the membrane (m2), Q is the volumetric flow rate (m3/s), P2 is the membrane pressure at permeate side and ΔP is the transmembrane pressure drop. The average pore radius (rg) of the membrane is evaluated using the slope,  rg 2.133  l

   v 2  q

 rg2  1    1.6    2   l   q   

and intercept,

of the created graph expressed as:

B rg=1.333 v A

(4)

The gas permeance (p) is calculated as:

p=

Q SP

(5)

where, S is the permeable area of the membrane (m2), Q is the volumetric flow rate (m3/s).

7 ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

2.2.5. Chemical stability Chemical stability of the fabricated composite membrane is verified by immersing the composite into the acid (Conc. HCl, pH 2) and alkali (NaOH, pH 12) solution. Firstly, the membrane is weighed in dry condition and dipped into the solution of acid and alkali for one week at atmospheric conditions. Secondly, the membrane is weighed and estimated the porosity following the volumetric porosity determination technique. EDX (LEO 1430VP®, Oxford) analysis of the membranes before and after corrosion test is performed to verify the change in elemental composition.

2.3. Response surface methodology via design of expert Nowadays, mathematical modelling and optimization techniques are frequently used in the field of membrane technology, especially, statistical tools are well known in improving operating parameters and performances for membrane applications. In this study, design of expert software (Version 8.0.7.1, Stat-Ease, Inc. MN, USA) has been used to develop central composite design (CCD) with a full polynomial. CCD procedure is an effective design tool which works for experimental design analysis with lack of fit testing which compares the variation around the model with "pure" variation within simulated observations and measures the adequacy of the quadratic response surface model, process optimization, development of secondorder response models, estimating experimental error variance, etc. This tool follows three steps to conduct the experimental design and are (a) statistical design of experiments, (b) assessment of coefficients through a mathematical model with the prediction of response and (c) analysis of the models applicability.28 Design plan including three input factors i.e. preparation pressure, amount of SM and BA (in wt. %) and outputs (R1: flexural strength; R2: porosity) for each experiment is shown in Table 1. All the experimental response models are produced with respect to the input variables (experimental data). The experimental response models determine the flexural strength and porosity of membrane, thus, indicating role of binders in membrane fabrication. The three input factors have been chosen for designing purpose and their values are selected based on preliminary study and are varied over five levels: the high value (+1), the centre points (0), the low level (-1) and two outer points (-α and +α value), details are outlined in Table S1, in supporting information. 8 ACS Paragon Plus Environment

Page 8 of 31

Page 9 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

CCD contains design points and axial points that accompanied a total of 20 runs of experiment are used to study the data obtained from the experimental runs. These data are then used to optimize the amount of binders and preparation pressure. The response variables i.e. flexural strength and porosity of membrane are measured using a mathematical modelling selected from the CCD with significant terms and the model is not aliased.

2.4. Cost estimation Manufacturing cost is also estimated in terms of raw materials cost, electricity consumption, mold preparation cost and other miscellaneous expenses.

3. Results and discussion 3.1. Analysis of variance (ANOVA) and response surface 3.1.1. Membrane flexural strength The relationship between process factors (preparation pressures, SM and BA content) and desired responses (flexural strength and porosity) are plotted graphically based on mathematical analysis of the experimental data. Figure 1 presents three-dimensional response surface for membrane’s flexural strength. It is observed from the figure that all factors applied in the membrane fabrication showed non-linear effects on the flexural strength of the membrane. High flexural strength is obtained at high preparation pressure with higher amount of binders (Figures 1a, b and c). As SM and BA form silicate and metallic metaborates at higher temperature during sintering process, there causes an increases in particle-particle interconnectivity. On the other hand, among two binders, BA shows superior influence than SM to obtain high flexural strength (Figure 1c). Increase in particle-particle interconnection reduces the voids between membrane layers and thus, higher flexural strength is detected. Therefore, high pressure and high BA content are desirable condition for producing a membrane with high flexural strength. All three fabricating factors have significant model terms (A: preparation pressure, B: SM, C: BA) and show quadratic model on both membrane flexural strength and porosity. Table S2 in supporting information, presents the analysis of variance and regression model for all the 9 ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 31

three model terms where the Model F-value of 41.10 implies that the model is significant. It has been observed from the results that the preparation pressure and BA content are significant model terms where Prob>F values are less than 0.05. In Table S2, F-value for SM content (1.97) is far lower compared to the F-values of BA (119.07) and preparation pressure (159.61). It is confirmed that the effect of SM is less important in membrane flexural strength. The empirical model for membrane flexural strength in terms of actual factors obtained from ANOVA analysis is shown in Eq. (6). Flexural strength model: Sqrt ( R1)= -10.29102  1.87092  A  5.87759 E - 003  B   0.79157  C  - 1.30706 E - 003  A

 B

- 0.038743  A

C 

- 9.29759 E - 003  B 

 2   6.25176 E - 003( B2 ) - 0.013929(C 2 )

C 

(6)

- 0.070072 A

The dependability of regression models for flexural strength of membrane is explained 2 2 on the basis of high value of R2 (0.974), RAdj (0.950) and RPred (0.763) which approves that this

model is well-fitted to the experimental results, shown in Table S3 (supporting information).

3.1.2. Membrane porosity In Figure 2, the three-dimensional response surface plots demonstrate the influence of preparation pressure and binder content on membrane porosity. Maximum membrane porosity of 28 % is obtained at preparation pressure of 7.84 MPa and 5 % of binder content each whereas the minimum porosity of 10 % is attained at preparation pressure of 11.77 MPa at high binder content (10 % each), as shown in Figure 2 a and b. This result infers that both preparation pressure and the binder content contribute to membrane porosity. However, amount of BA seems to give a major effect on membrane porosity compared to SM content (refer Figure 2 c). At high binder contents, particles have a tendency to stay close to each other which cause a decrease in porosity. On the other hand, at maximum preparation pressure of 11.77 MPa (Run 3), it is observed that the porosity is high (26 %) compared to the membranes made by applying less pressure (9.81 MPa) (Run 2) which is listed in Table 1. This phenomenon is due to the presence

10 ACS Paragon Plus Environment

Page 11 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

of less binder content in the former though the forming pressure is high. Thus, binder looks to offer a major effect on membrane porosity than that of preparation pressure. Table S2 shows the effect of significant terms i.e. preparation pressure (A), amount of SM (B) and BA content (C), second-order effects of SM content (B2) and second-order effects of BA content for membrane porosity with the Prob>F less than 0.05. The F-value for BA content is quite higher (199.81) than that of preparation pressure (28.96) and SM content (38.68) which approves that BA content is more prominent in membrane porosity. A quadratic empirical model 2 for membrane porosity is expressed in Eq. (7) which provides high value of R2 (0.967), RAdj 2 (0.938) and RPred (0.750) as shown in Table S3 in supporting information.

Membrane porosity model: Sqrt ( R 2)  5.12793 - 0.069077  A  0.40462  B   0.069356  C  - 0.018886

 A  B 

 6.09732 E - 003  A

C 

 8.74352 E - 003  B 

 2  - 0.026644( B2 ) - 0.030356(C 2 )

C 

(7)

 1.96320 E - 003 A

3.2. Model verification on the basis of statistical analysis Interaction effect among all the input parameters is investigated using response surface methodology (RSM) software. This model has the advantage of identifying the effects of binary combination of combining two independent input factors. Figures 3 and 4 show the variable interaction for all responses. Figure 3 b shows non-parallel curvatures which suggest a strong interaction between the variables of BA content (C) and preparation pressure (A) for membrane flexural strength. As a result, AC seemed as one of the significant model terms and is mentioned in Table S3 of supporting information. In Figure 4, parallel curvatures refer that there is no such significant interaction between the independent parameters to each other but there is an obvious effect of individual parameters on the membrane porosity. The model predicts that the maximum flexural strength and maximum porosity are predicted around the region of middle preparation pressure, SM and BA content (Table S4 in supporting information). Residual analysis of response surface design is presented graphically in Figures 5 and 6 to ensure the fitting of analysis data with the statistical assumptions and to verify whether the 11 ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 31

standard deviations (SD) between the actual and predicted responses follow the distribution in normal way. Figures 5 (a and b) ensure that there is no indication of non-normality of the experimental results as the residual points fall near to a straight line showing normal distribution of points. Figures 6 a and b present the plots of the residuals versus predicted response for the membrane flexural strength and porosity, respectively. Figure 6 reveals a random scattering of all experimental data points across the horizontal line of residuals which suggests that the proposed models are adequate. The residuals between predicted and actual values are within the range of ±4 % and ±5 % for flexural strength and porosity, respectively. The predicted values obtained from regression model are compared with the actual values from experimental studies to confirm the consistency and acceptability of empirical models from individual responses (Figure 7). Figures 7 a and b show the comparison between actual values and predicted values for membrane flexural strength and porosity, respectively and indicate that all design points are distributed along the diagonal line above which are over- estimated and vice-versa. The over- and under- estimated data points for flexural strength are distributed along the diagonal line within the range of 5 %. But the over- and under- estimated data points for porosity lie along the diagonal line within the range of 7 %. This comparative study indicates a negligible experimental error as the responses from the empirical model are well-fitted in acceptable variance range. According to the data, the empirical model obtained from CCD can be used as an analyst for the optimization of binder content and preparation pressure in order to achieve porous ceramic membrane with high appreciable strength.

3.3. Optimization study An optimization study of all the independent input variables are performed through a desirability function (D) for two responses according to Eq. (8).29 The optimum preparation pressure and binder content for membrane fabrication predicted from all responses with high or low limit of inputs can be satisfied with desirability function (D). 1/

 N  D   diri   i 1 

 ri

(8) 12 ACS Paragon Plus Environment

Page 13 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

where, D is the desirability function, N is the number of responses, ri states to the significance of particular response and d i signifies the partial desirability function for specific responses. In Figure 8, the desirability plot ensures that the desirable preparation pressure is 9.81 MPa and the weight percentage of both SM and BA are 7.5 % of total weight of the sample mixture which gives optimized porosity of 21 % and flexural strength of 11.55 MPa, is in good agreement with literatures.30,20 This prediction of desirable input variables is also confirmed with the optimized input variables calculated from central composite design (listed in Table S4, supporting information). This optimization study shows that the desirability value of the model is 0.97. Table S5 (supporting information) explains the errors between the experimental and predicted values and the values are found to be under 2 % which is quite appreciable. This shows that using response surface methodology a formulation can be generated to obtain the role of binders and preparation pressures in ceramic membrane fabrication.

3.4. Study of phase transformation and microstructure of the optimized membrane The XRD patterns of the ceramic membrane using optimized binder contents obtained from RSM study, namely, C1 (SM: 7.50 % and BA: 7.50 %) along with C2 sintered at 550, 700 and 850 °C are shown in Figure S2 in supporting information. The additional composition i.e., C2 (SM: 5 % and BA: 5 %) is introduced to verify the phase transformation with the change in binder contents keeping the preparation pressure constant. The introduction of C2 is stated since it is difficult for the optimized membrane to identify any phase transformation with the variation in sintering temperature. A graphite free sample is formed as the temperature rises to 850 °C. The formation of crystalline inoyite and nephiline are detected at diffraction patterns at 2θ = 29.455° and 29.757°, respectively, due to the reaction between the binders and metakaolinite during heating. The XRD diffractrogram clearly reveals that inoyite and nephiline are the predominant crystalline component in the membrane matrix which signify membrane hardening with high crystallinity, is in good agreement with literatures.30 The microstructure of the ceramic matrix using FESEM images are shown in Figure S3, supporting information. Microstructure of the ceramic matrix mainly depends on two factors; one is sintering temperature and the other is binder content. At lower temperature (550 ºC), a 13 ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 31

muddy and patchy matrix (Figure S3 a) is observed which corresponds to an interfacial layer of partially decomposed raw materials (mainly hemicellulose, cellulose and lignin present in sawdust) during heating. A rough and pellet like surface texture (Figure S3 b) is detected when the membrane sintered at 700 °C owing to increase in the rate of decomposition of lignocellulosic material with increasing temperature. On the other hand, at 850 °C, the densification of ceramic matrix increases due to complete decomposition of raw materials and binder burn-out. Hence, the image shows a highly consolidated and well-bonded matrix grains (Figure S3 c). At 700 °C or higher temperature, the occurrence of consolidated surface texture evidently discloses the densification of ceramic body by forming silicate/metaborates or borosilicate bonds and hence increases the bonding strength, resulting in a higher flexural strength.

3.5. Gas permeation test of the optimized membrane The gas permeation experiment is implemented for the membrane prepared on the basis of optimized composition with three different sintering temperatures to determine the average pore diameter. Figure S4 (supporting information) illustrates the variation of effective permeability factor with average pressure and determines the value of slope and intercept for the membranes. The determined gas permeance and average pore diameters of the membranes sintered at 550, 700 and 850 °C are 3.78×10-3, 4.62×10-3, 1.96×10-3 m3 m-2 s-1 kPa-1 and 0.13, 0.12 and 0.21 µm, respectively. The porosity values (Table S6) obtained from volumetric porosity method (Eq. 1) are 0.34, 0.44 and 0.20 with increasing temperature, is in good agreement with literatures.30 Table S6 (supporting information) also defines the porosity values calculated from gas permeation study (Eq. 2) and the values are 0.33, 0.44 and 0.20 with increasing sintering temperature. Due to the uneven surface structure, the permeation rate of the membrane sintered at 700 °C is higher than that of sintered at 550 °C which is quite clear from FESEM images (Figure S3). This phenomena is due to partial construction of borosilicate and metaborates and partial decomposition of cellulose and lignin. The tortuosity factor is assumed as one (straight cylindrical pores) during calculation procedure.19,30 It is interesting to observe that the permeation rate of the membrane sintered at 850 °C is reduced compared to the membrane sintered at 700 °C. This behaviour is due to the complete decomposition of sawdust along with the formation of borosilicate and metaborates with 14 ACS Paragon Plus Environment

Page 15 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

sintering temperature. This condition clearly confirm that the formation of borosilicate and metasilicate bonds is completed after 700 °C which controls the morphology of the membrane superiorly than the decomposition of sawdust. Porosity values obtained from volumetric porosity method is slightly higher compared to the value calculated from gas permeation test as water remains intact into the interiors and the wall of the pores owing to swelling nature of kaolin based membrane.

3.6. Chemical stability test of the optimized membrane The porosity values of the ceramics during acid-alkali test are shown in Table S6, shown in supporting information. The weight loss of the membrane is under 6 % during acid-alkali test. However, increase in membrane porosity in the range of 9-10 % is observed in both acid and alkali media for the optimized membranes sintered at 850 °C. But 3 % decrease in porosity is detected for the membrane sintered at 550 and 700 °C during acid test due to the production of silicon dioxide while SM reacts with HCl whereas decrease in porosity in the range of 3-8 % is noted in alkali test due to the production of sodium metaborate while SM reacts with NaOH solution. EDX studies of the membranes obtained from optimization study before and after chemical stability test are shown in Table S7 (in supporting information). It is confirmed from the table that there is no significant change in elemental composition of the membrane in both acid-alkali media. Therefore, the observed trends in weight loss and EDX analysis during acidalkali tests convey that there is no major change in elemental composition due to the presence of binders in ceramic processing.

3.7. Cost analysis of the fabricated support membrane In our previous work 19, the economic feasibility assessment of the membrane is carried out on the basis of few parameters, such as, raw materials cost, energy consumption and membrane mold manufacturing cost. In this work, the cost analysis is performed including cost of two binders. The projected cost of the membrane is evaluated and presented in Table S8, 15 ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 31

supporting information. The estimated cost of the prepared C1 membrane is near about $302/m2. The total cost including membrane mold preparation and other miscellaneous expenditures (10 % of the estimated cost) is $332/m2. The prepared membrane is slightly expensive than the membrane prepared without binder

19

and cheaper than α-alumina based ceramic tube costing

$500-1000/m2 available in market. This study approves that the ceramic membrane prepared by kaolin, saw dust and feldspar along with binders would be attractive and useful than polymeric membranes as well as ceramic tubular membranes installed in industrial arrangements. The cost of the membranes may vary with their performance and stability in different process applications.

4. Conclusion Based on the response surface designs, it is confirmed that all the factors have an appreciable effect on the membrane porosity and flexural strength. Among the binders, BA is found to be the key factor in effecting flexural strength of membrane. However, the effects of both BA and SM are equally significant for the porosity of membrane but the influence of preparation pressure is less significant due to the introduction of binder. The optimum membrane fabricating condition is obtained at preparation pressure of 9.81 MPa, 7.50 % of SM and BA each with the desirability function value of 0.97 for microfiltration range membrane fabrication. The predicted values of flexural strength and porosity are 11.55 MPa and 20.59 %, respectively, which is in good agreement with the actual experimental data (flexural strength: 11.68 MPa, porosity: 21 %). Error analysis between actual and predicted values from CCD confirms that the deviation fell within 2 %. This result indicates that the proposed model optimized through response surface methodology on the basis of CCD is reliable and useful tool to understand the role of binder in the fabrication of membrane with desired morphological and mechanical behaviour. The morphological study of the optimized membrane shows appreciable range of average pore diameter along with good chemical stability. XRD patterns illustrates that the optimized membrane is free of graphite. These results suggest that the optimized membrane offer low manufacturing cost of the membrane as $332/m2 along with good mechanical strength, morphology and chemical stability. 16 ACS Paragon Plus Environment

Page 17 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Acknowledgements The authors are thankful to Central Instruments Facility (CIF), IIT Guwahati for their support. We are also thankful to DST-FIST for XRD (FIST, DST, SR/FST/ETII - 028/2010).

Associated contents Supporting information Gas permeation set up; XRD patterns of the ceramic membrane sintered at 550, 700 and 850 °C; FESEM images of the ceramic membrane surface sintered at (a) 550, (b) 700 and (c) 850 °C; Effective permeability factor vs. average pressure for the optimized membranes sintered at 550, 700 and 850 °C temperatures; Design summary; ANOVA for respective response surface quadratic models; ANOVA and regression analysis for flexural strength and porosity; Optimized input variables; Error analysis; Change in porosity of the optimized membrane in acid and alkali media as a function of temperature; EDX analysis; Cost assessment of the optimized membrane. This information is available free of charge via the Internet at http://pubs.acs.org/.

Nomenclature A

Model terms for preparation pressure

B

Model terms for SM content

C

Model terms for BA content

D

Desirability function

N

Number of responses

ri

Significance of particular response

di

Partial desirability function for specific responses

R1

Model response for membrane flexural strength

R2

Model response for membrane porosity 17 ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

R2

Coefficient of multiple regression

2 RAdj

Adjusted statistic coefficient

2 RPred

Predicted statistic coefficient

Vm

Volume of the fabricated membrane

W1

Weight of the dry membrane

W2

Weight of the wet membrane

Greek letters α

Axial points in CCD

H O 2

Density of the deionized water

Abbreviations ANOVA

Analysis of variances

CCD

Central composite design

RSM

Response surface methodology

Sqrt

Square root (a mathematical term)

Prob>F

Probability value greater than F-value

18 ACS Paragon Plus Environment

Page 18 of 31

Page 19 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

References (1) Nijmeijer, A.; Huiskes, C.; Sibelt, N.G.M.; Kruidhof, H.; Verweij, H. Tubular Membrane Supports by Centrifugal Casting. Am. Ceram. Soc. Bull. 1998, 77, 95. (2) Biesheuvel, P.M.; Breeveld, V.; Higler, A.; Verweij, H. Graded Membrane Supports Produced by Centrifugal Casting of a Slightly Polydisperse Suspension. Chem. Eng. Sci. 2001, 56, 3517. (3) Chen, C.H.; Takita, K.; Ishiguro, S.; Honda, S.; Awaji, H. Fabrication on Porous Alumina Tube by Centrifugal Molding. J. Eur. Ceram. Soc. 2005, 25, 3257. (4) Steenkamp, G.C.; Keizer, K.; Neomagus, H.W.J.P.; Krieg, H.M. Copper (II) Removal from Polluted Water with Alumina/Chitosan Composite Membranes. J. Membr. Sci. 2002, 197, 147. (5) Falamaki, C.; Veysizadeh, J. Taguchi Design of Experiments Approach to the Manufacture of One-Step Alumina Microfilter/Membrane Supports by the Centrifugal Casting Technique. Ceram. Int. 2008, 34, 1653. (6) Steenkamp, G.C.; Neomagus, H.W.J.P.; Krieg, H.M.; Keizer, K. Centrifugal Casting of Ceramic Membrane Tubes and the Coating with Chitosan. Sep. Purif. Technol. 2001, 25, 407. (7) Kim, K.H.; Choa, S.J.; Yoon, K.J.; Kim, J.J.; Ha, J.; Chun. D. Centrifugal Casting of Alumina Tube for Membrane Application. J. Membr. Sci., 2002, 199, 69. (8) Pinggen, R.; Mikio, I.; Takahiro, T.; Isao. K. Centrifugal Casting of Al2O3–15 wt.% ZrO2 Ceramic Composites. Ceram. Int. 2003, 29, 209. (9) Xomeritakis, G.; Tsai, C.Y.; Jiang, Y.B.; Brinker, C.J. Tubular ceramic-supported sol–gel silica-based membranes for flue gas carbon dioxide capture and sequestration. J. Membr. Sci. 2009, 341, 30. (10) Xomeritakis, G.; Tsai, C.Y.; Brinker, C.J. Microporous sol–gel derived amino-silicate membrane for enhanced carbon dioxide separation. Sep. Purif. Technol. 2005, 42, 249. (11) Brinker, C.J.; Sehgal, R.; Hietala, S.L.; Deshpande, R.; Smith, D.M.; Loy, D.; Ashley, C.S. Sol–gel strategies for controlled porosity inorganic materials. J. Membr. Sci. 1994, 94, 85. (12) Xia, C.; Liu, M. Low-temperature SOFCs based on Gd0.1Ce0.9O1.95 fabricated by dry pressing. Solid State Ionics 2001, 144, 249.

19 ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 31

(13) Xia, C.; Liu, M. A Simple and Cost-Effective Approach to Fabrication of Dense Ceramic Membranes on Porous Substrates. J. Am. Ceram. Soc. 2001, 84, 1903. (14) Zuo, C.; Zha, S.; Liu, M.; Hatano, M.; Uchiyama, M.

Ba(Zr0.1Ce0.7Y0.2)O3-δ as an

Electrolyte for Low-Temperature Solid-Oxide Fuel Cells. Adv. Mater. 2006, 18, 3318. (15) Das, N.; Maiti, H.S. Formatation of pore structure in tape-cast alumina membranes-effects of binder content and firing temperature. J. Membr. Sci. 1998, 140, 205. (16) Zheng, M.-P.; Hou, Y.-D.; Ge, H.-Y.; Zhu M.-K.; Yan H. Effect of NiO additive on microstructure, mechanical behavior and electrical properties of 0.2PZN–0.8PZT ceramics. J. Eur. Ceram. Soc. 2013, 33, 1447. (17) Sarbatly, R. Effect of Kaolin/pesf Ratio and Sintering Temperature on Pore Size and Porosity of the Kaolin Membrane Support. J. Appl. Sci. 2011, 11, 2306. (18) Yuan, Q.; Pu, Y. Effects of K0.5Bi0.5TiO3 addition on dielectric properties of BaTiO3 ceramics. Ceramics Int. 2013, 39, 3507. (19) Bose, S.; Das, C. Preparation and characterization of low cost tubular ceramic support membranes using sawdust as a pore-former. Mat. Lett. 2013, 110, 152. (20) Jana, S.; Purkait, M.K.; Mohanty, K. Preparation and Characterizations of Ceramic Microfiltration Membrane: Effect of Inorganic Precursors on Membrane Morphology. Sep. Sci. Technol. 2011, 46, 33. (21) Ahmad, A.L.; Low, S.C.; Abd Shukor, S.R.; Ismail, A. Optimization of membrane performance by thermal-mechanical stretching process using responses surface methodology (RSM). Sep. Purif. Technol. 2009, 66, 177. (22) Alventosa-deLara, E.; Barredo-Damas, S.; Alcaina-Miranda, M.I.; Iborra-Clar, M.I. Ultrafiltration technology with a ceramic membrane for reactive dye removal: Optimization of membrane performance. J. Hazard. Mater. 2012, 209-210, 492. (23) Xiangli, F.; Wei, W.; Chen, Y.; Jin, W.; Xu, N. Optimization of preparation conditions for polydimethylsiloxane (PDMS)/ceramic composite pervaporation membranes using response surface methodology. J. Membr. Sci. 2008, 311, 23. (24) Yeong, Y.F.; Abdullah, A. Z.; Ahmad, A. L.; Bhatia, S. Process optimization studies of pxylene separation from binary xylene mixture over silicalite-1 membrane using response surface methodology. J. Membr. Sci. 2009, 341, 96.

20 ACS Paragon Plus Environment

Page 21 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

(25) Vaez, M.; Moghaddam, A.Z.; Alijani, S. Optimization and Modeling of Photocatalytic Degradation of Azo Dye Using a Response Surface Methodology (RSM) Based on the Central Composite Design with Immobilized Titania Nanoparticles. Ind. Eng. Chem. Res. 2012, 51, 4199. (26) Gudena, K.; Rangaiah, G.P.; Lakshminarayanan, S. Modeling and Analysis of Hybrid Reactive Stripper-Membrane Process for Lactic Acid Recovery. Ind. Eng. Chem. Res., 2013, 52, 2907. (27) Khayet, M.; Cojocaru, C.; Garcıa-Payo, C. Application of Response Surface Methodology and Experimental Design in Direct Contact Membrane Distillation. Ind. Eng. Chem. Res. 2007, 46, 5673. (28) Guaracho, V.V.; Kaminari, N.M.S.; Ponte, M.J.J.S.; Ponte, H.A. Central Composite experimental design applied to removal of lead and nickel from sand. J. Hazard. Mater. 2009, 172, 1087. (29) Montgomery, D.C. Design and Analysis of Experiments; John Wiley & Sons: USA, 2001. (30) Nandi, B.K.; Uppaluri, R.; Purkait, M.K. Preparation and characterization of low cost ceramic membranes for micro-filtration applications. Appl. Clay Sci. 2008, 42, 102.

21 ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure Captions: Figure 1

Response surface plotted on (a) Sodium metasilicate (SM) content: preparation pressure; (b) Boric acid (BA) content: preparation pressure; (c) Sodium metasilicate (SM) content: Boric acid (BA) content for membrane flexural strength.

Figure 2

Response surface plotted on (a) Sodium metasilicate (SM) content: preparation pressure; (b) Boric acid (BA) content: preparation pressure; (c) Sodium metasilicate (SM) content: Boric acid (BA) content for membrane porosity.

Figure 3

Interaction via (a) Sodium metasilicate (SM) content: preparation pressure; (b) Boric acid (BA) content: preparation pressure; (c) Sodium metasilicate (SM) content: Boric acid (BA) content for membrane flexural strength (● design points).

Figure 4

Interaction via (a) Sodium metasilicate (SM) content: preparation pressure; (b) Boric acid (BA) content: preparation pressure; (c) Sodium metasilicate (SM) content: Boric acid (BA) content for membrane porosity (● design points).

Figure 5

Normal probability plot of residual for membrane (a) flexural strength; (b) porosity.

Figure 6

Plot between residual and predicted response for membrane (a) flexural strength; (b) porosity.

Figure 7

Plot of predicted versus actual values for membrane (a) flexural strength (R2=0.98); (b) porosity (R2=0.98).

Figure 8

Response surface plot of desirability operating region over SM:BA at preparation pressure set at 9.81 MPa.

Table captions: Table 1

Design arrangement and experimental responses for central composite design (CCD)

22 ACS Paragon Plus Environment

Page 22 of 31

Page 23 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

(a)

(b)

(c)

Figure 1. Response surface plotted on (a) Sodium metasilicate (SM) content: preparation pressure; (b) Boric acid (BA) content: preparation pressure; (c) Sodium metasilicate (SM) content: Boric acid (BA) content for membrane flexural strength.

23 ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(a)

(b)

(c)

Figure 2. Response surface plotted on (a) Sodium metasilicate (SM) content: preparation pressure; (b) Boric acid (BA) content: preparation pressure; (c) Sodium metasilicate (SM) content: Boric acid (BA) content for membrane porosity.

24 ACS Paragon Plus Environment

Page 24 of 31

Page 25 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Figure 3. Interaction via (a) Sodium metasilicate (SM) content: preparation pressure; (b) Boric acid (BA) content: preparation pressure; (c) Sodium metasilicate (SM) content: Boric acid (BA) content for membrane flexural strength (● design points).

25 ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4. Interaction via (a) Sodium metasilicate (SM) content: preparation pressure; (b) Boric acid (BA) content: preparation pressure; (c) Sodium metasilicate (SM) content: Boric acid (BA) content for membrane porosity (● design points).

26 ACS Paragon Plus Environment

Page 26 of 31

Page 27 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

(a)

(b)

Figure 5. Normal probability plot of residual for membrane (a) flexural strength; (b) porosity.

Figure 6. Plot between residual and predicted response for membrane (a) flexural strength; (b) porosity.

27 ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

4

6

8

10

12

10

14

(a)

Test Data of RSM 2 R =0.98

14

12

12

10

10

8

Predicted Porosity, %

14

15

20

25

30

(b)

Predicted Flexural Strength, MPa

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 31

30 30

Test data of RSM 2 R = 0.98

25

25

20

20

15

15

10

10

8

6

6 10

15

20

25

30

Actual Porosity, %

4

4 4

6

8

10

12

14

Actual Flexural Strength, MPa

Figure 7. Plot of predicted versus actual values for membrane (a) flexural strength (R2=0.98); (b) porosity (R2=0.98).

Figure 8. Response surface plot of desirability operating region over SM:BA at preparation pressure set at 9.81 MPa.

28 ACS Paragon Plus Environment

Page 29 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

29 ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 31

Table 1. Design arrangement and experimental responses for central composite design (CCD) Run

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Factor 1: A Pressure (MPa) 7.84 9.81 11.77 9.81 9.81 9.81 11.77 9.81 12.39 7.84 7.23 9.81 9.81 7.84 9.81 9.81 9.81 7.84 11.77 11.77

Factor 2: B Sodium metasilicate (SM) (%) 5.00 7.50 5.00 7.50 4.21 7.50 10.00 7.50 7.50 10.00 7.50 7.50 7.50 10.00 10.79 7.50 7.50 5.00 10.00 5.00

Factor 3: C Boric acid (BA) (%) 10.00 4.21 5.00 7.50 7.50 7.50 5.00 7.50 7.50 10.00 7.50 10.79 7.50 5.00 7.50 7.50 7.50 5.00 10.00 10.00

Response 1: R1 Flexural strength (MPa) 10.54 8.35 9.72 11.68 11.53 11.68 11.82 11.68 11.99 11.29 5.32 12.33 11.68 5.01 11.85 11.68 11.68 4.46 12.82 13.78

30 ACS Paragon Plus Environment

Response 1: R2 Porosity (%) 15 24 26 21 22 21 18 21 16 13 24 11 21 25 13 21 21 28 10 13

Page 31 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

Industrial & Engineering Chemistry Research

For Table of Contents Only

Sawdust

Sodium metasilicate

Feldspar

Boric acid

Tubular membrane Kaolin

ACS Paragon Plus Environment