Preparation of a Zirconia-Based Ceramic Membrane and Its Application for Drinking Water Treatment

This work concerns the preparation of a mineral membrane by the slip casting method based on zirconium oxide (ZrO2) and kaolin. The membrane support is produced from a mixture of clay (kaolin) and calcium carbonate (calcite) powders using heat treatment (sintering). Membrane and support characterization were performed by Scanning Electron Microscopy (SEM), X-ray Fluorescence (XRF), Fourier Transform Infrared Spectroscopy (FTIR), X-ray Diffraction (XRD), and Raman Spectroscopy. The prepared mineral membrane was tested to treat drinking water obtained from different zones of the El Athmania (Algeria) water station (raw, coagulated, decanted, and bio filtered water). Experimental parameters such as permeate flux, turbidity, and total coliforms were monitored. The results showed that the mineral membrane was mainly composed of SiO2 and Al2O3 and the outer surface, which represented the membrane support, was much more porous than the inner surface where the membrane was deposited. The permeate flux of the raw water decreased with filtration time, due to a rejection of the organic matters contained in the raw water. Moreover, the absence of total coliforms in the filtrate and the increase in concentration in the concentrate indicate that the prepared mineral membrane can be used for drinking water treatment.


Introduction
Membrane processes are increasingly used in industrial areas, covering a wide range of operating conditions and module designs. Using membrane technology, instead of a conventional separation or purification step that is part of an existing industrial process, can reduce overall energy consumption and produce acceptable results under softer conditions. In addition, membrane processes generally permit continuous operation and can be combined easily with other separation processes [1][2][3].
The significant development of membrane filtration processes has generated a great deal of interest in research on membrane materials. Although polymeric membranes have been used in industrial areas for many years, inorganic membranes, also known as ceramic membranes, are used for longer life, considering their better chemical and mechanical stabilities compared to polymeric membranes [4][5][6]. For this reason, many studies have been performed in recent years to develop new types of inorganic membranes [7,8].
The development of clay-based inorganic membranes can lead to an important new technological application that will add economic value to the enormous natural deposits of clay minerals, many of XRD, and Raman Spectroscopy. The membrane was used to remove organic matters from drinking water. The developed ceramic membrane was also tested for removal of Escherichia coli (E. coli). Variations in the permeate flux of distilled and raw drinking water versus filtration time were investigated. Moreover, the characterizations of raw drinking water, permeate and concentrate quality were also illustrated. Table 1 shows the chemical composition of the raw kaolin determined by X-ray fluorescence (XRF) analysis. A chemical composition of kaolin powders for the ceramic membrane support showed that SiO 2 , Al 2 O 3 , Fe 2 O 3 were major elements. However, K 2 O, Na 2 O, MgO, P 2 O 5 , TiO 2 , and MnO were detected as minor elements.

The Methods of the Membrane Support and Ceramic Membrane Preparation
In this part, we describe the methods of the membrane support from the clay powders (kaolin) and zirconia-based ceramic membrane preparations. The clay consisted mainly of aluminum silicate (Al 2 SiO 5 ), intended for the manufacture of the tubular microporous support. The method of tubular support preparation can be summarized in the following steps [28,29]: • Thermal treatment of the clay material at a temperature of 400-600 • C for 30 min for the removal of water contained in kaolin (dehydration) and the combustion of organic matters.

•
Grinding of the clay material to obtain small particles.

•
Sieving of the small particles to obtain particles smaller than 125 µm.

•
Addition of kaolin (75%) and calcium carbonate (CaCO 3 ) (22%) for the appearance of pores with an acceptable number and size in the final support.

•
Addition of an organic additive (Methocel) (3%) to improve the elastic properties of the dough and to facilitate the formation process.

•
Mixing of the above-mentioned materials with the presence of the solvent (distilled water), by using the mixer until a paste of good elastic properties was obtained. Then, the mixture was placed in a tightly closed plastic bag for 12 h to properly spread the water in the ceramic paste.

•
Extrusion of ceramic paste in tubular form.

•
Drying of the tubular support with ambient air by placing it on the machine containing the rotating cylinders to dry it uniformly and maintain its shape for 24 h.
• Sintering the components of the ceramic paste that forms the support at a temperature equal to 1100 • C, which will convert it to anorthite according to a series of reactions during a specific thermal program.
The sintering of the support was realized in the following steps. First, the temperature of the chamber was increased from the ambient temperature to the temperature of 250 • C, with a rise speed of 3 • C/min and a plateau of 15 min. During this step, water could be eliminated quickly. Second, the temperature was increased from 250 to 1100 • C with a rise speed of 3 • C/min and a plateau of 60 min. During this step, organic matters could be removed (Figure 1). The sintering of the support was realized in the following steps. First, the temperature of the chamber was increased from the ambient temperature to the temperature of 250 °C, with a rise speed of 3 °C/min and a plateau of 15 min. During this step, water could be eliminated quickly. Second, the temperature was increased from 250 to 1100 °C with a rise speed of 3 °C/min and a plateau of 60 min. During this step, organic matters could be removed (Figure 1).

The Method of the Slip Casting Membrane Preparation
The method consisted of suspending zirconia and polyvinyl alcohol in distilled water and pouring the produced slip inside the porous support ( Figure 2). The method results in the controlled diffusion process, which amounts to a simple loss of water from the suspension in the mass of the support; this causes the accumulation of zirconia particles on its surface. The support sintered with thermal sequences ensures that the material can withstand high pressures and chemical attacks. However, zircon oxide (ZrO2) accumulated in the inner part of the tubular support provides the formation of selective permeable membrane.
The method of zirconia-based ceramic membrane preparation can be summarized in the following steps: • Take 70% of the distilled water and add in 4% by weight of ZrO2 powder to mix the mixture until a good homogeneous mixture was obtained.

•
Place the mixture in an ultrasonic bath for 10 min to dispel the granules and dissolve the sediments.

•
Then, add 26% polyvinyl alcohol (PVA) and mix for 12 h to obtain the suspension solution.

•
The solution is poured into the support for 10 min and dried for 5 min.

The Method of the Slip Casting Membrane Preparation
The method consisted of suspending zirconia and polyvinyl alcohol in distilled water and pouring the produced slip inside the porous support ( Figure 2). The method results in the controlled diffusion process, which amounts to a simple loss of water from the suspension in the mass of the support; this causes the accumulation of zirconia particles on its surface. The support sintered with thermal sequences ensures that the material can withstand high pressures and chemical attacks. However, zircon oxide (ZrO 2 ) accumulated in the inner part of the tubular support provides the formation of selective permeable membrane.
The method of zirconia-based ceramic membrane preparation can be summarized in the following steps: • Take 70% of the distilled water and add in 4% by weight of ZrO 2 powder to mix the mixture until a good homogeneous mixture was obtained.

•
Place the mixture in an ultrasonic bath for 10 min to dispel the granules and dissolve the sediments.

•
Then, add 26% polyvinyl alcohol (PVA) and mix for 12 h to obtain the suspension solution.

•
The solution is poured into the support for 10 min and dried for 5 min.

Experimental Setup of Filtration
The prepared zirconia-based ceramic membrane was used for the treatment of drinking water obtained from Oued El Athmania water treatment plant (Mila, Algeria). The filtration experiments were carried out using a tangential filtration system ( Figure 3). The total volume of the reservoir was 5 L, and 3 L of drinking water was used for each experiment. The effective membrane area was 45 cm 2 and cross-flow velocity was 4.18 m/s. The concentrate was recycled back into the feed tank and the filtered water (permeate) was collected in an Erlenmeyer flask for analysis. The volume of filtrate was monitored as a function of time to determine the permeate flux (Jp).

Experimental Setup of Filtration
The prepared zirconia-based ceramic membrane was used for the treatment of drinking water obtained from Oued El Athmania water treatment plant (Mila, Algeria). The filtration experiments were carried out using a tangential filtration system ( Figure 3). The total volume of the reservoir was 5 L, and 3 L of drinking water was used for each experiment. The effective membrane area was 45 cm 2 and cross-flow velocity was 4.18 m/s. The concentrate was recycled back into the feed tank and the filtered water (permeate) was collected in an Erlenmeyer flask for analysis. The volume of filtrate was monitored as a function of time to determine the permeate flux (J p ).

Experimental Setup of Filtration
The prepared zirconia-based ceramic membrane was used for the treatment of drinking water obtained from Oued El Athmania water treatment plant (Mila, Algeria). The filtration experiments were carried out using a tangential filtration system ( Figure 3). The total volume of the reservoir was 5 L, and 3 L of drinking water was used for each experiment. The effective membrane area was 45 cm 2 and cross-flow velocity was 4.18 m/s. The concentrate was recycled back into the feed tank and the filtered water (permeate) was collected in an Erlenmeyer flask for analysis. The volume of filtrate was monitored as a function of time to determine the permeate flux (Jp).

Determination Method of Total Coliform Bacteria
Coliform bacteria filtration experiments were also carried out by prepared ceramic membrane. Coliform bacteria in water, thermo-tolerant coliforms, and E. coli were investigated and counted using the most probable number (MPN) technique in liquid medium. BCP (Bromocresol Purple) Lactose Broth Tubes were used for presumptive identification and enrichment of total coliforms, including thermotolerant coliforms. Water samples were collected in sterile glass bottles (1 L) in order to detect and count the final concentration of bacteria after filtration. The final reading was carried out according to the requirements of the MPN table, taking into account that E. Coli is a producer of gas and indole at 44 • C.

Characterization of the Support and Zirconia-Based Ceramic Membrane
The characterizations of the support and zirconia-based ceramic membrane are given in Tables 2  and 3.  Scanning Electron Microscopy allows for the observation of the morphology and cavities of the membrane-support (inner and outer surfaces). It should be remembered that the slip casting membrane was deposited inside the tubular support. Figure 4A,B show the outer surface of the support material. It could be seen from the figures that there were large and irregular pores on the support layer. Figure 4C,D show the inner surface of the ceramic membrane. The layered structure of zirconia and the asymmetric distribution of the pores can be seen from the figures. Moreover, cross-section images of the tubular membrane are seen in Figure 4E,F. The outer surface of the support had wider cavities compared to the inner surface. These cavities can be considered to be an advantage for deposition or adhesion of the solution inside the module.

X-Ray Fluorescence (XRF)
The X-Ray fluorescence analyses were carried out by a Panalytical Epsilon 3 spectrophotometer, for the evaluation of the most present elements in the membrane support. This energy dispersive X-Ray spectrophotometer was connected to a computer using the Omnian analysis software. The sample was placed under helium flow during the analysis. Fluorescence X-ray spectra were recorded under different excitation conditions. The use of a particular filter with a potential difference and a particular current allows for the better exploration of a particular region of the spectrum. For the support material, the first spectrum was realized with a potential difference (dp) of 5.00 keV and a current of 1000 μA ( Figure 5A). It allowed us to explore the region of energies up to about 4 keV. For the zirconia-based ceramic membrane, the second spectrum was realized with a silver filter with a thickness of 100 μ, a dp of 30.00 keV and a current of 300 μA ( Figure 5B). It allows the peaks of high energy to be observed. However, in this case, the peaks are strongly attenuated. From the spectrograms of Figure 5A,B, excitation energy spectra in the range of 1.486 to 1.597 keV and 1.739 to 1.836 keV corresponding to atoms (Al) and (Si), respectively, come from aluminum silicate (Al2SiO5) (the clay material). Moreover, in the spectrogram of Figure 5B, the excitation energy spectrum of 15.744 keV

X-ray Fluorescence (XRF)
The X-ray fluorescence analyses were carried out by a Panalytical Epsilon 3 spectrophotometer, for the evaluation of the most present elements in the membrane support. This energy dispersive X-ray spectrophotometer was connected to a computer using the Omnian analysis software. The sample was placed under helium flow during the analysis. Fluorescence X-ray spectra were recorded under different excitation conditions. The use of a particular filter with a potential difference and a particular current allows for the better exploration of a particular region of the spectrum. For the support material, the first spectrum was realized with a potential difference (d p ) of 5.00 keV and a current of 1000 µA ( Figure 5A). It allowed us to explore the region of energies up to about 4 keV. For the zirconia-based ceramic membrane, the second spectrum was realized with a silver filter with a thickness of 100 µ, a d p of 30.00 keV and a current of 300 µA ( Figure 5B). It allows the peaks of high energy to be observed. However, in this case, the peaks are strongly attenuated. From the spectrograms of Figure 5A,B, excitation energy spectra in the range of 1.486 to 1.597 keV and 1.739 to 1.836 keV corresponding to atoms (Al) and (Si), respectively, come from aluminum silicate (Al 2 SiO 5 ) (the clay material). Moreover, in the spectrogram of Figure 5B, the excitation energy spectrum of 15.744 keV corresponding to the zirconia atom (Zr) constituent of the zirconia oxide (ZrO 2 ) membrane was noted. Zirconia detected in the ceramic membrane showed that zirconia had entered the structure of the support material. atoms (Al) and (Si), respectively, come from aluminum silicate (Al2SiO5) (the clay material). Moreover, in the spectrogram of Figure 5B, the excitation energy spectrum of 15.744 keV corresponding to the zirconia atom (Zr) constituent of the zirconia oxide (ZrO2) membrane was noted. Zirconia detected in the ceramic membrane showed that zirconia had entered the structure of the support material.
It was also found from the spectrograms of Figure 5C,D, that the clay used in the preparation of the support contained other constituents such as Mn and Fe. We also noticed excitation energy spectra in the range of 3.5 to 4 keV, which confirmed the presence of the calcium (Ca). It composed the calcium oxide (CaO) by addition of calcite (CaCO3) transformed into (CaO) during the heat treatment of the support paste.

X-Ray Diffraction (XRD)
The support was characterized by a Panalytical Empyrian brand diffractometer operating under the following conditions: 40 mA, 45 kV with monochromatic radiation Kα = 1.54 A° of copper, equipped with a goniometer and an X-ray detector. Figure 6 represents XRD reflections of the support at a temperature of 1100 °C for 1 h. The main phase identified in the membrane support was the anorthite (CaO, Al2O3, 2SiO2) which was a predominant phase. This phase was very important because of its promising physical and mechanical properties [5,29]. It was also found from the spectrograms of Figure 5C,D, that the clay used in the preparation of the support contained other constituents such as Mn and Fe. We also noticed excitation energy spectra in the range of 3.5 to 4 keV, which confirmed the presence of the calcium (Ca). It composed the calcium oxide (CaO) by addition of calcite (CaCO 3 ) transformed into (CaO) during the heat treatment of the support paste.

X-ray Diffraction (XRD)
The support was characterized by a Panalytical Empyrian brand diffractometer operating under the following conditions: 40 mA, 45 kV with monochromatic radiation K α = 1.54 A • of copper, equipped with a goniometer and an X-ray detector. Figure 6 represents XRD reflections of the support at a temperature of 1100 • C for 1 h. The main phase identified in the membrane support was the anorthite (CaO, Al 2 O 3 , 2SiO 2 ) which was a predominant phase. This phase was very important because of its promising physical and mechanical properties [5,29].

Fourier-Transform Infrared Spectroscopy (FTIR)
Infrared spectra were recorded on a spectrophotometer equipped with an ATR accessory (Jasco FT/IR-4000). Infrared spectroscopy is a tool for mineralogists to characterize the crystallinity of materials, by observing the relative intensities of the hydroxyl (OH) vibration bands and that of the Si-O, Si-O-Si, Al-OH and Al-O in their structures [30,31].
The IR spectrogram of the clay support represented by Figure 7 was divided into two main zones. The first peaks corresponded to high frequency bands (wave numbers) between 3700 and 2800 cm −1 , and the second peaks corresponded to the lower frequencies located in the 1500-500 cm −1 area. The high frequency bands (zone II) concerned the vibration of OH hydroxyls, while the low frequency bands (zone I) related to the Si-O, Si-O-Si, Al-OH-Al, Al-OH, and Al-O bond networks [32].

Fourier-Transform Infrared Spectroscopy (FTIR)
Infrared spectra were recorded on a spectrophotometer equipped with an ATR accessory (Jasco FT/IR-4000). Infrared spectroscopy is a tool for mineralogists to characterize the crystallinity of materials, by observing the relative intensities of the hydroxyl (OH) vibration bands and that of the Si-O, Si-O-Si, Al-OH and Al-O in their structures [30,31].
The IR spectrogram of the clay support represented by Figure 7 was divided into two main zones. The first peaks corresponded to high frequency bands (wave numbers) between 3700 and 2800 cm −1 , and the second peaks corresponded to the lower frequencies located in the 1500-500 cm −1 area. The high frequency bands (zone II) concerned the vibration of OH hydroxyls, while the low frequency bands (zone I) related to the Si-O, Si-O-Si, Al-OH-Al, Al-OH, and Al-O bond networks [32]. The wave numbers (υ) of peaks and the functional groups corresponding to kaolin are summarized in Table 4. According to Table 2, in correlation with the literature, it was found that the clay used as support contained the different chemical elements with different percentages, such as Si, Al, Fe' and bonds with hydroxyls, which has been confirmed by several authors [33][34][35].  The wave numbers (υ) of peaks and the functional groups corresponding to kaolin are summarized in Table 4. According to Table 2, in correlation with the literature, it was found that the clay used as support contained the different chemical elements with different percentages, such as Si, Al, Fe' and bonds with hydroxyls, which has been confirmed by several authors [33][34][35].

Raman Spectroscopy
The Raman scattering spectra were collected by a Thermo Fisher DXR spectrometer equipped with an optical microscope, a three-grating monochromator (triple additive mode) and a CCD camera detector (Charge Coupled Device). The exciting radiation of a wavelength of 780 nm was delivered by the beam of a NIR diode laser. The beam was focused with a long frontal lens (×100 magnification) numerical aperture of 0.9 over 50 µm of the sample surface. The power irradiating the sample was about 10 mW. The scattered retro Raman spectrum was collected in confocal mode to avoid optical artifacts, particularly the signal from the glass slide above the sample cell. The spectral resolution was 1.9 cm −1 , with a precision on the measurement of the best wave number only 1 cm −1 . Figure 8 represents the Raman spectrum of the clay support, where several bands were observed. We noticed the appearance of new spectra at 1694, 2080, and 3372 cm −1 , which express the vibrations of ν (C=O), Si-H, and OH, respectively [36]. detector (Charge Coupled Device). The exciting radiation of a wavelength of 780 nm was delivered by the beam of a NIR diode laser. The beam was focused with a long frontal lens (×100 magnification) numerical aperture of 0.9 over 50 µm of the sample surface. The power irradiating the sample was about 10 mW. The scattered retro Raman spectrum was collected in confocal mode to avoid optical artifacts, particularly the signal from the glass slide above the sample cell. The spectral resolution was 1.9 cm −1 , with a precision on the measurement of the best wave number only 1 cm −1 . Figure 8 represents the Raman spectrum of the clay support, where several bands were observed. We noticed the appearance of new spectra at 1694, 2080, and 3372 cm −1 , which express the vibrations of ν (C = O), Si-H, and OH, respectively [36].    Figure 9A shows that the permeate flux (J p ) of distilled water increases with the increase in the transmembrane pressure (TMP) according to the Darcy law [37]. The variations of the permeate flux (J p ) versus time using distilled and raw drinking water are presented in Figure 9B. According to the obtained results, it can be seen that the permeate flux for the distilled and the raw water were not of the same order. This behavior can be explained by the retention of certain matters by the ceramic membrane causing a blockage of the pores, which leads to the reduction of the amount of water passing through the membrane (permeate). The straight line for distilled water explains that the flow of water was constant over the filtration time. On the other hand, the permeate flux of raw drinking water decreased during the first 20 min and then became constant. It could be explained by there being a certain amount of suspended matter which had deposited on the surface of the membrane.

Turbidity Variation Versus Time
Turbidity was determined using a 2100Q portable turbidimeter proposed by Hach, with a tungsten filament lamp equipped with a two-detector ratio optical system for accurate results during routine analyses. It brings greater measurement sensitivity over a wider range of 0 to 1000 NTU. The suspended solid in water causes turbidity. The membrane filtration experiments were carried out for raw drinking water obtained from Oued El Athmania water treatment plant. The turbidity of raw drinking water increased in the concentrate versus filtration time because the water was completely recycled into the feed tank ( Figure 10).

Turbidity Variation Versus Time
Turbidity was determined using a 2100Q portable turbidimeter proposed by Hach, with a tungsten filament lamp equipped with a two-detector ratio optical system for accurate results during routine analyses. It brings greater measurement sensitivity over a wider range of 0 to 1000 NTU. The suspended solid in water causes turbidity. The membrane filtration experiments were carried out for raw drinking water obtained from Oued El Athmania water treatment plant. The turbidity of raw drinking water increased in the concentrate versus filtration time because the water was completely recycled into the feed tank ( Figure 10).

Turbidity Variation Versus Time
Turbidity was determined using a 2100Q portable turbidimeter proposed by Hach, with a tungsten filament lamp equipped with a two-detector ratio optical system for accurate results during routine analyses. It brings greater measurement sensitivity over a wider range of 0 to 1000 NTU. The suspended solid in water causes turbidity. The membrane filtration experiments were carried out for raw drinking water obtained from Oued El Athmania water treatment plant. The turbidity of raw drinking water increased in the concentrate versus filtration time because the water was completely recycled into the feed tank ( Figure 10). The characterizations of raw drinking water, permeate and concentrate quality are presented in Table 5. The characterizations of raw drinking water, permeate and concentrate quality are presented in Table 5.

Total Coliform Bacteria Variation versus Time
Biological tests of total coliform bacteria were also performed in this work to test the retention capacity of the ceramic membrane. E. coli are considered as indicators of the microbial quality of drinking water [38]. In the filtered water (permeate), the number of coliforms was equal to zero during 60 min filtration, which depicted that all of the total coliforms were rejected by the zirconia-based ceramic membrane (Figure 11). On the other hand, in the concentrated water, an increase in the number of total coliforms from 15 to 35 in 100 mL was obtained after 60 min of filtration.

Total Coliform Bacteria Variation Versus Time
Biological tests of total coliform bacteria were also performed in this work to test the retention capacity of the ceramic membrane. E. coli are considered as indicators of the microbial quality of drinking water [38]. In the filtered water (permeate), the number of coliforms was equal to zero during 60 min filtration, which depicted that all of the total coliforms were rejected by the zirconiabased ceramic membrane ( Figure 11). On the other hand, in the concentrated water, an increase in the number of total coliforms from 15 to 35 in 100 mL was obtained after 60 min of filtration. Figure 11. Evolution of the number of total coliforms in the permeate and concentrate and versus time.

Conclusions
In this work, a zirconia-based ceramic membrane with a tubular configuration was prepared by the casting method. The anorthite support, which had favorable physical and mechanical properties, Figure 11. Evolution of the number of total coliforms in the permeate and concentrate and versus time.

Conclusions
In this work, a zirconia-based ceramic membrane with a tubular configuration was prepared by the casting method. The anorthite support, which had favorable physical and mechanical properties, was prepared by the extrusion method. The inner layer, containing smaller pores compared to the support, reduced the size of the pores and eliminated defects of the support. The membrane filtration results showed that there was an improvement in the physicochemical and bacteriological quality of raw drinking water. The prepared membrane retained all of the total coliforms. Using a ceramic membrane can help to obtain a good clarification, and can reduce the addition of chemical agents such as aluminum and chlorine used for coagulation and disinfection. These agents form an additional pollution, such as the presence of aluminum in the sludge from the settling basin, and an acceptable taste of water after the addition of a smaller amount of chlorine.