*Article* **Development of Red Clay Ultrafiltration Membranes for Oil-Water Separation**

**Saad A. Aljlil**

National Center for Water Treatment and Desalination Technology, King Abdulaziz City for Science and Technology, P.O. Box 6086, Riyadh 11442, Saudi Arabia; saljlil@kacst.edu.sa

**Abstract:** In this study, a red clay/nano-activated carbon membrane was investigated for the removal of oil from industrial wastewater. The sintering temperature was minimized using CaF<sup>2</sup> powder as a binder. The fabricated membrane was characterized by its mechanical properties, average pore size, and hydrophilicity. A contact angle of 67.3◦ and membrane spore size of 95.46 nm were obtained. The prepared membrane was tested by a cross-flow filtration process using an oil-water emulsion, and showed a promising permeate flux and oil rejection results. During the separation of oil from water, the flux increased from 191.38 to 284.99 L/m<sup>2</sup> on increasing the applied pressure from 3 to 6 bar. In addition, high water permeability was obtained for the fabricated membrane at low operating pressure. However, the membrane flux decreased from 490.28 to 367.32 L/m<sup>2</sup> ·h due to oil deposition on the membrane surface; regardless, the maximum oil rejection was 99.96% at an oil concentration of 80 NTU and a pressure of 5 bar. The fabricated membrane was negatively charged, as were the oil droplets, thereby facilitating membrane purification through backwashing. The obtained ceramic membrane functioned well as a hydrophilic membrane and showed potential for use in oil wastewater treatment.

**Keywords:** ultrafiltration; red clay; calcium fluoride powder; wastewater; oil separation

### **1. Introduction**

The daily generation of approximately 210 million barrels of water contaminated with oil can incur costs of \$45 B in water purification [1]. Recently the membrane technology has been applied to separate mixtures of oil and water, showing greater efficacy than traditional technologies [2,3]. It is well-known that membranes synthesized from polymers are unstable compared to ceramic membranes [4]. Ceramic membranes are often used in industrial applications, such as wastewater treatment, despite the high cost compared to those of polymeric membranes [5], owing to their extraordinary chemical, mechanical, and thermal properties [6]. They have the capability of backwashing, high flux, good toughness, resistance to bacterial growth, and thermal stability [4,7]. Materials like alumina, zirconia, titanium oxide and zeolite materials have been used in ceramic membranes that resist high pH and pressure to separate oil-water mixtures [4,8].

Additionally, microfiltration (MF) and ultrafiltration (UF) have been used as pressuredriven membrane methods [3]. UF is recognized as the most efficient in oil-water separation with significant advantages compared to conventional separation techniques: it requires no additional chemicals and its energy consumption is low [3]. High-flux MF membranes have also been used for oil-water separation; however, they carry the risk of oil penetration [3]. NaA zeolite was deposited on an α-Al2O<sup>3</sup> MF membrane by Cui et al. [9] and used in the separation of oil from water. The fabricated membrane had a pore diameter of 1.2 µm and showed 99% oil separation at a flow rate of 85 L/m<sup>2</sup> ·h and pressure of 50 kPa. A porous MF aluminum ceramic membrane was applied by Liu et al. for oil-water separation [10]; they reported 99.98% removal of emulsified oil.

A UF membrane was used to separate oil from water in an oilfield [3] with more than 96% oil rejection. Depositing TiO<sup>2</sup> on the surface of a UF ZrO<sup>2</sup> membrane to separate

**Citation:** Aljlil, S.A. Development of Red Clay Ultrafiltration Membranes for Oil-Water Separation. *Crystals* **2021**, *11*, 248. https://doi.org/ 10.3390/cryst11030248

Academic Editors: Abdolhamid Shafaroudi and Walid M. Daoushv

Received: 17 January 2021 Accepted: 25 February 2021 Published: 28 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

oil and water [11] revealed that the cohesion between the oil droplets and the membrane surface decreased. This was because less oil was adsorbed on the membrane surface, thus decreasing surface fouling. UF ceramic membranes were used [12] to separate an oil-water– anionic surfactant emulsion to investigate the effect of pH and flow rate. They reported an occurrence of concentration polarization phenomena at low flow rates. In addition, the membrane surface was positively charged at low pH and attracted the anionic surfactant in the oil-water emulsion, causing a reduction in the water flux rate. Luo et al. [13] used a UF hollow-fiber membrane for oil rejection from water, which yielded highly pure water with very low turbidity. Tubular UF membranes with a pore size of 10 nm were fabricated [14] and showed a flux rate of 200 L/m<sup>2</sup> ·h for wastewater purification. Issaoui and Lionel [15] have reported the fabrication of commercialized ceramic membranes based on expensive materials such as cordierite, titania, zirconia, and silicon carbide for various industrial applications.

Despite their several advantages and promising results, the high cost of ceramic components in manufacturing is a major disadvantage of ceramic membranes [4]. Therefore, experts have focused on developing innovative, low-cost, effective, and secure materials that can be used to prepare ceramic membranes, while still exhibiting the qualities required for wastewater purification, such as chemical stability, fouling resistance, and good mechanical attributes [2,3]. Efforts have been dedicated toward utilizing low-cost materials such as natural clay, apatite powder, dolomite, kaolin, bauxite, and mineral coal fly ash for the fabrication of ceramic membranes exhibiting good performance for wastewater and pollutant treatment. An MF tubular ceramic membrane was fabricated from low-cost clay and kaolin [16] to separate oil-water mixtures and showed 93% rejection of the oil emulsion. Zhu et al. [17] fabricated a ceramic membrane from fly ash and TiO<sup>2</sup> that showed 97% oil rejection. Liu et al [18] fabricated a bentonite clay membrane for use in the separation of oil from saltwater, but it was unsuitable for the high-salinity environment. Furthermore, kaolin, quartz, feldspar, sodium carbonate as low-cost MF membrane materials were tested by Nandi et al. [19]. They noted that the fabricated membranes showed good oil removal efficiency over a 60-min experiment, with 98.8% oil rejection at the flux of 5.36 <sup>×</sup> <sup>10</sup>−<sup>6</sup> <sup>m</sup>3/m<sup>2</sup> ·s and applied pressures in the range 68.95–275.8 kPa. Kakali and Pugazhenthi [20] focused on using lithium aluminosilicate for the fabrication of a ceramic membrane via the slip-casting method, using starch as the pore-forming material for removing bacteria and oil from wastewater; they achieved good performance in terms of removal of bacteria and oil from wastewater. Ben Amar and Oun [21] used Tunisian mud as a low-cost material for the fabrication of tubular ceramic ultrafiltration membranes for removing pollutants from wastewater. These membranes were obtained via the slip-casting technique followed by sintering at 650 ◦C. With a pore size of 11 nm, the membranes demonstrated a permeability of 90 L/h·m<sup>2</sup> ·bar and oil pollutant removal of 90%. Hubadillah et al. [22] used kaolin as a low-cost ceramic raw material for the fabrication of a ceramic hollow fiber membrane and achieved improved mechanical strength and excellent performance in terms of pollutant removal from wastewater.

Our goal is to fabricate a UF ceramic membrane for oil removal from water using red clay as a low-cost material, nano-activated carbon powder for pore-forming, and CaF<sup>2</sup> as a binder and nucleating agent to minimize the sintering temperature. The aforementioned low-cost materials were processed using an extrusion method, and the sintering was performed by carbon pyrolysis. To the best of our knowledge, all prior researchers have fabricated MF membranes as substrates and used interlayers and filtration layers to obtain UF membranes. Additionally, the use of nano-activated carbon as a pore-forming material and CaF<sup>2</sup> as a binder, rather than the conventional carboxymethyl cellulose is also novel. This study demonstrates the effective preparation of a novel tubular ceramic membrane for oil-water separation by using low-cost and locally sourced materials like red clay.

### **2. Methodology**

### *2.1. Raw Materials*

Table 1 shows that the red clay used in fabricating the membrane contains higher amounts of SiO<sup>2</sup> (47.63%), Al2O<sup>3</sup> (24.03%), and Fe2O<sup>3</sup> (9.57%) with low amounts of K2O (2.08%) and Na2O (1.29%). The red clay was collected from the Biadh plant in Riyadh, Saudi Arabia. The role of CaF<sup>2</sup> (Sigma Aldrich, St. Louis, MO, USA) is to minimize the sintering temperature of the red clay membrane and improve the mechanical strength. *Crystals* **2021**, *11*, x FOR PEER REVIEW 3 of 14 **2. Methodology** *2.1. Raw Materials*

**Table 1.** Chemical composition of Saudi red clay (wt%). Table 1 shows that the red clay used in fabricating the membrane contains higher amounts of SiO<sup>2</sup> (47.63%), Al2O<sup>3</sup> (24.03%), and Fe2O<sup>3</sup> (9.57%) with low amounts of K2O


Activated carbon (diameter = 4 mm; purity 98%) was obtained from Zhengzhou

Activated carbon (diameter = 4 mm; purity 98%) was obtained from Zhengzhou Company (Henan, China). Powdered activated carbon (65 µm) was obtained using a planetary ball mill at 300 rpm for 4 h, and was blended with water for 72 h and then treated ultrasonically (50 min, 540 W) using an SFX550 (Sonifier, Suwanee, GA, USA). Then, the collected suspension was centrifuged at 3500 rpm for 15 min to yield nano-activated carbon of size 91.6 nm (SEM, Figure 1). It is worth mentioning that activated carbon was converted to nanoactivated carbon for the sole purpose of a pore-forming material, i.e., to create nanopores in the body of the ceramic membrane. After sintering at temperatures above 500 ◦C, all of the activated carbon was burned from the ceramic membrane, which resulted in pore formation in the ceramic membrane (TGA, Figure 2). Company (Henan, China). Powdered activated carbon (65 μm) was obtained using a planetary ball mill at 300 rpm for 4 h, and was blended with water for 72 h and then treated ultrasonically (50 min, 540 W) using an SFX550 (Sonifier; Mexico). Then, the collected suspension was centrifuged at 3500 rpm for 15 min to yield nano-activated carbon of size 91.6 nm (SEM, Figure 1). It is worth mentioning that activated carbon was converted to nanoactivated carbon for the sole purpose of a pore-forming material, i.e., to create nanopores in the body of the ceramic membrane. After sintering at temperatures above 500 °C, all of the activated carbon was burned from the ceramic membrane, which resulted in pore formation in the ceramic membrane (TGA, Figure 2).

**Figure 1.** SEM of the nano-activated carbon prepared with 50000× magnification. **Figure 1.** SEM of the nano-activated carbon prepared with 50000× magnification.

The fabricated raw material for the membrane comprised of a powder blend of 90 wt.% red clay, 5 wt.% CaF<sup>2</sup> as a binder, and 5 wt.% nano-activated carbon powder as a pore former. The red clay was ground at 250 rpm for 3 h using a planetary ball mill to

water was slowly added. Through wet mixing, we obtained a paste with satisfactory plasticity, which was then fed into the extruder (Length=200 cm, Width= 50 cm, Die diameter=

**Loss on Ignition**

% 47.63 24.03 9.57 0.23 2.08 1.29 1.28 11.55

**Table 1.** Chemical composition of Saudi red clay (wt%).

**Oxides SiO<sup>2</sup> Al2O<sup>3</sup> Fe2O<sup>3</sup> Na2O K2O CaO TiO<sup>2</sup>**

*2.2. Membrane Fabrication* 

achieve the particle size of 100 μm. A

water, and the weight was determined using a balance to calculate the flux rate of the

where V is the water volume collected through the pores of the membrane, A is the

The turbidities of the water permeate and water in the feed tank were measured

*J* = V /A \* t (3)

*R*% = [1 − (*C*c/*C*i)] × 100, (4)

*J* = *K*c \* ΔP (5)

The water flux, *J* (L/m2·h), was determined from equation 2:

using a turbidimeter; the oil rejection (R%) was obtained from equation 3:

The membrane permeability (*P*) was obtained by equation 4:

where *C*<sup>i</sup> is the raw oil turbidity and *C*<sup>c</sup> is the turbidity after filtration.

where *K*c is the membrane permeability and ΔP is the applied pressure.

**Figure 2.** Sketch the experimental filtration process. **Figure 2.** Sketch the experimental filtration process.

membrane surface area, and t is the time.

#### **3. Results and Discussion** *2.2. Membrane Fabrication*

clean water.

*3.1. Characterizations* The zeta potential curves of the oil emulsion and membrane are presented in Figure 3. The isoelectric point of the oil droplets emulsion appears at pH 1, where the zeta potential was negative. The isoelectric point of the membrane was located at a slightly higher The fabricated raw material for the membrane comprised of a powder blend of 90 wt.% red clay, 5 wt.% CaF<sup>2</sup> as a binder, and 5 wt.% nano-activated carbon powder as a pore former. The red clay was ground at 250 rpm for 3 h using a planetary ball mill to achieve the particle size of 100 µm.

pH of 1.7, where the zeta potential was also negative. Since identical charges are known to experience electrostatic repulsion, it can be predicted that our fabricated membrane would prevent fouling during backwashing because of the electrostatic repulsion between the oil emulsion and the membrane. Mechanical testing of the ceramic membranes with and without CaF2 was performed using the three-point bending technique with a crosshead speed of 0.5 mm/min. The membrane without CaF<sup>2</sup> showed a lower bending strength of 49.53 MPa than the membrane with CaF<sup>2</sup> (54.13 MPa); the stress-strain relationship of both the membranes was linear. Further, a contact angle of 67.3° indicates that the membrane is hydrophilic. The membrane pore size distribution (Figure 4) indicated that a UF membrane was created without Az-mixer was used to combine the dry raw materials for 4 h. To this mixture, 400 mL water was slowly added. Through wet mixing, we obtained a paste with satisfactory plasticity, which was then fed into the extruder (Length = 200 cm, Width = 50 cm, Die diameter = 20 cm). Two sintering stages were applied: first, sintering was performed from 30 to 500 ◦C at an average heating rate of 1.5 ◦C/min to burn the organic material, thereby creating pores in the prepared membrane. Figure 2 presents the TGA data in an air of the nano-activated carbon powder, clarifying the sintering operation. It is clear from Figure 2 that complete burning of the nano-activated carbon powder was obtained at 450 ◦C. Subsequently, the produced membrane was densified by sintering in a furnace at temperatures from 400 ◦C to 1000 ◦C at a rate of 2 ◦C/min for 4 h.

#### using a coating layer. The pore size of this membrane ranged from 40 nm to 110 nm with *2.3. Characterization of Ceramic Membrane*

#### an average pore size of 96 nm, and 88% of the total pores were smaller than 96 nm. In 2.3.1. Scanning Electron Microscopy (SEM)

addition, the measured porosity of the membrane was 32.56%. SEM was used to determine the morphology of the fabricated membrane (Figure 5). It was noted that the absence Scanning electron microscopy (SEM) (model NNL-200, Philips, 1-nm resolution) was used for morphological characterization and microstructural analysis of the fabricated sintered membrane. Samples smaller than 10 mm were obtained by cutting the sintered membrane samples by a cutting machine with a diamond cutting disc. The membrane sample was dried for 24 h in a vacuum oven, and then etched by 1% HF + 1% HNO<sup>3</sup> solution for 30 s. Finally, the membrane sample was coated by a thin gold layer using a sputter coater (SPI Inc., Lakewood, WA, USA) to increase the conductivity of the membrane and to obtain a clear image. SEM images of the membrane sample were obtained by scanning it with a focused beam of electrons, which interact with the electrons in the membrane sample, producing various detectable signals containing information about the sample's surface topography.

### 2.3.2. Apparent Porosity

The apparent porosity of the ceramic structures was determined by the standard test method (ISO EN 993-1) for ceramic structures using the Archimedes buoyancy technique with dry weights, soaked weights, and immersed weights in water. The membrane sample was dried in an oven at 105 ◦C for 24 h to eliminate the absorbed water. The dried membrane sample was weighed by the balance and the weight is recorded as Md. Then the membrane sample was placed in a water-filled container for 24 h at room temperature. After that, the membrane sample has weighed and the weight recorded as Mw. In addition, the

membrane sample was weighed inside water and the weight recorded as M<sup>a</sup> (suspension weight). Finally, the apparent porosity can be calculated from equation 1:

$$Apparent\ porosity = 100 \times \left(\frac{M\_{\overline{w}} - M\_d}{M\_{\overline{w}} - M\_d}\right) \tag{1}$$

### 2.3.3. Contact Angle Measurements

The capillary force liquid weight gain, which occurs when wetting the membrane sample, was precisely measured using a K100 force tensiometer (Kruss, Wissenschaftliche Laborgeräte, Borsteler Chaussee 85, Germany) with a particularly high resolution to obtain reliable and accurate contact angle (θ) data of the membrane sample. Before measuring, the membrane sample was dried in a vacuum oven for 24 h at 100 ◦C. The main procedure to measure the membrane contact angle using Kruss K100 involves sealing off the open ends of the tubular membrane sample with epoxy resins, hanging the sample on the microbalance in the K100 force tensiometer (Kruss, Wissenschaftliche Laborgeräte, Borsteler Chaussee 85, Germany)), and immersing the sample gradually into deionized water. The rate of immersion has to be adjusted to ~6 mm/min. Finally, the contact angle will be calculated from the forces acting on the membrane surface.

### 2.3.4. Mechanical Test

The mechanical properties of the ceramic membrane were determined by the threepoint bending strength test using a Shimadzu-Universal testing machine (AGS-X, Riverwood Drive Columbia, MD 21046, USA) with a capacity of 5 kN, a total grip distance of 690 mm, a crosshead speed of 0.5 mm/min, a potential of 200 V, and power of 60 Hz. The stress-strain relationship of the membrane was found to be linear. Membrane samples with a length of at least 4 cm were obtained by cutting the samples by a cutting machine with an artificial diamond disc, the result is obtained by using the equation

$$\delta\_f = \frac{\text{FL}}{\pi \text{( $d\_2 - d\_1$ )}^3} \tag{2}$$

where "*δf*"is the bending strength "F" is the flexural load in newton and "L", "*d*2" and "*d*1" is the span length, outer diameter, and inner diameter respectively.

### 2.3.5. X-Ray Diffraction

The X-ray diffraction technique (diffractometer used: model D8AD VANCE, BRUKER, Billerica, MA, USA) was used to identify the crystalline phases of the membrane sample. In this technique, the scattered intensity of an X-ray beam, generated upon hitting the membrane sample, is measured as a function of incident angle. The membrane sample for XRD analysis was dried for 24 h in a vacuum oven at 105 ◦C to eliminate any moisture present in the material. Then, the membrane sample was powdered and spread on the glass holder with a gap of 0.5 mm. The holder with the sample was then placed in the X-ray chamber and scanned at a constant temperature and a speed of 2◦/min using CuKα radiation, over a diffraction angle (2θ) range from 10◦ to 80◦ , with a step size of 10◦ . The Joint Committee on Powder Diffraction Standards (JCPDS) diffraction file cards (2001) are used as reference for interpretation of the X-ray patterns obtained in the experiment.

### 2.3.6. Pore size Distribution Measurements

The pore size distribution of the membrane was determined using a constant-pressure fluid-fluid porometer (IFTS advanced fluid-fluid porometer, Institut de la Filtration et des Techniques Séparatives, Rue Marcel Pagnol, Foulayronnes, France).

### *2.4. Oil Emulsion Characterization*

The performance of the prepared ceramic membrane was characterized by implementing it in the separation of oil from an oil-water mixture, where ultra-pure paraffin oil

was used as a synthetic oil in the absence of a surfactant. The emulsion was vigorously and continuously mixed by using an agitator for 50 min at 1350 rpm and remained stable for several days unaffected by gravitational forces. The initial concentration of the oil emulsion was measured in terms of turbidity at 64 NTU. In addition, an oil-water mixture was obtained from an Aramco oilfield. Furthermore, a Zetasizer was used to obtain the zeta potential curve and to determine the oil droplet charge. To define the efficiency of the prepared membrane in oil-water separation, a turbidity meter was used to determine the oil turbidity.

### *2.5. Ultrafiltration Testing*

Figure 2 shows the cross-flow filtration set-up, which consists of a feed tank, pump, pressure gauges, agitator, and a tubular-type ceramic membrane. In the flow circuit, feeds with a volumetric flow rate of 55 L/h at 25 ◦C and a specific pressure of a known and constant composition (water and oil emulsion) was pumped continuously through the cross-flow ultra-filtration membrane (diameter = 0.5 cm, length = 20 cm) at a specified cross flow pressure of 5 bar. In addition, a secondary agitator was used to provide mixing effects and to ensure emulsion stability. A water tank was used to collect the permeated water, and the weight was determined using a balance to calculate the flux rate of the clean water.

The water flux, *J* (L/m<sup>2</sup> ·h), was determined from Equation (2):

$$J = \mathbf{V} / \mathbf{A} \, ^\ast \mathbf{t} \tag{3}$$

where V is the water volume collected through the pores of the membrane, A is the membrane surface area, and t is the time.

The turbidities of the water permeate and water in the feed tank were measured using a turbidimeter; the oil rejection (R%) was obtained from Equation (3):

$$R\text{\textquotedblleft} = \left[1 - \left(\mathbb{C}\_{\text{c}}/\mathbb{C}\_{\text{i}}\right)\right] \times 100,\tag{4}$$

where *C*<sup>i</sup> is the raw oil turbidity and *C*<sup>c</sup> is the turbidity after filtration. The membrane permeability (*P*) was obtained by Equation (4):

$$J = K\_{\mathbb{C}} \, \text{\* } \Delta \text{P} \tag{5}$$

where *K*<sup>c</sup> is the membrane permeability and ∆P is the applied pressure.

### **3. Results and Discussion**

### *3.1. Characterizations*

The zeta potential curves of the oil emulsion and membrane are presented in Figure 3. The isoelectric point of the oil droplets emulsion appears at pH 1, where the zeta potential was negative. The isoelectric point of the membrane was located at a slightly higher pH of 1.7, where the zeta potential was also negative. Since identical charges are known to experience electrostatic repulsion, it can be predicted that our fabricated membrane would prevent fouling during backwashing because of the electrostatic repulsion between the oil emulsion and the membrane.

Mechanical testing of the ceramic membranes with and without CaF<sup>2</sup> was performed using the three-point bending technique with a crosshead speed of 0.5 mm/min. The membrane without CaF<sup>2</sup> showed a lower bending strength of 49.53 MPa than the membrane with CaF<sup>2</sup> (54.13 MPa); the stress-strain relationship of both the membranes was linear. Further, a contact angle of 67.3◦ indicates that the membrane is hydrophilic. The membrane pore size distribution (Figure 4) indicated that a UF membrane was created without using a coating layer. The pore size of this membrane ranged from 40 nm to 110 nm with an average pore size of 96 nm, and 88% of the total pores were smaller than 96 nm. In addition, the measured porosity of the membrane was 32.56%. SEM was used to determine the morphology of the fabricated membrane (Figure 5). It was noted that the absence of cracks

of cracks in the fabricated membrane was indicative of its high quality and good material

the residual weight was ~94 wt.%. Crystal water decomposition was found to occur at 500–600 °C, and the standard decomposition or recrystallization possible during heat

Figure 6 also presents the TGA data for the red clay paste, with a residual weight of 85 wt.% at 1000 °C. The holding time of the sintering procedure after treatment at 1000 °C was 2 h. CaF<sup>2</sup> as a binder and nucleating agent to minimize the sintering temperature was used. The sintering temperature of the ceramic membranes with and without CaF2 was

of cracks in the fabricated membrane was indicative of its high quality and good material

Figure 6 also presents the TGA data for the red clay paste, with a residual weight of 85 wt.% at 1000 °C. The holding time of the sintering procedure after treatment at 1000 °C was 2 h. CaF<sup>2</sup> as a binder and nucleating agent to minimize the sintering temperature was used. The sintering temperature of the ceramic membranes with and without CaF2 was performed. The membrane with CaF<sup>2</sup> showed a lower sintering temperature of 1000 °C than the membrane without CaF<sup>2</sup> (1150 °C). The TGA data for CaF<sup>2</sup> (Figure 2) showed that the residual weight was ~94 wt.%. Crystal water decomposition was found to occur at

properties in agreement with the results of the three-point bending strength test.

properties in agreement with the results of the three-point bending strength test.

*Crystals* **2021**, *11*, x FOR PEER REVIEW 7 of 14

treatment occurs at 600–1100 °C.

*Crystals* **2021**, *11*, x FOR PEER REVIEW 7 of 14

in the fabricated membrane was indicative of its high quality and good material properties in agreement with the results of the three-point bending strength test. treatment occurs at 600–1100 °C.

**Figure 3.** Zeta potential of oil droplets and the red clay membrane as a function of pH. **Figure 3.** Zeta potential of oil droplets and the red clay membrane as a function of pH. **Figure 3.** Zeta potential of oil droplets and the red clay membrane as a function of pH.

**Figure 4.** The pore size distribution of the fabricated membrane. **Figure 4.** The pore size distribution of the fabricated membrane.

**Figure 4.** The pore size distribution of the fabricated membrane.

**Figure 5.** SEM of the tubular ceramic membrane: (**a**) cross-section of the membrane and (**b**) the top **Figure 5.** SEM of the tubular ceramic membrane: (**a**) cross-section of the membrane and (**b**) the top view of the membrane.

view of the membrane. Figure 6 also presents the TGA data for the red clay paste, with a residual weight of 85 wt.% at 1000 ◦C. The holding time of the sintering procedure after treatment at 1000 ◦C was 2 h. CaF<sup>2</sup> as a binder and nucleating agent to minimize the sintering temperature was used. The sintering temperature of the ceramic membranes with and without CaF<sup>2</sup> was performed. The membrane with CaF<sup>2</sup> showed a lower sintering temperature of 1000 ◦C than the membrane without CaF<sup>2</sup> (1150 ◦C). The TGA data for CaF<sup>2</sup> (Figure 2) showed that the residual weight was ~94 wt.%. Crystal water decomposition was found to occur at 500–600 ◦C, and the standard decomposition or recrystallization possible during heat treatment occurs at 600–1100 ◦C.

> The XRD pattern of the fabricated membrane (Figure 7), which contains red clay and CaF2, shows that red clay is a highly illitic kaolinite-type clay. It includes illite, kaolinite, and hematite, which gives it a red color. Additionally, the clay contains some amount of free quartz. CaF<sup>2</sup> could be seen in the XRD.

> Sintering at 1000 ◦C led to the decomposition of kaolinite and illite (clay minerals). Some amount of CaF<sup>2</sup> was also decomposed, while the remaining was observed in the membrane structure by XRD analysis. The calcium released from the CaF<sup>2</sup> decomposition and the aluminum silicate from the clay minerals were reacted to create an anorthite phase. During sintering, the free quartz and hematite remained stable, as evidenced by the XRD pattern of the sintered membrane. Most of the decomposed kaolinite was transformed to mullite. In addition, large amounts of amorphous phase were present in the sintered membrane, which was determined by the broad peak between 2θ = 20◦ and 2θ = 40◦ .

free quartz. CaF<sup>2</sup> could be seen in the XRD.

The XRD pattern of the fabricated membrane (Figure 7), which contains red clay and CaF2, shows that red clay is a highly illitic kaolinite-type clay. It includes illite, kaolinite, and hematite, which gives it a red color. Additionally, the clay contains some amount of

view of the membrane.

40°.

(b)

(a)

**Figure 5.** SEM of the tubular ceramic membrane: (**a**) cross-section of the membrane and (**b**) the top

**Figure 6.** Thermogravimetric analysis of red clay membrane, activated carbon, and CaF<sup>2</sup> under air. **Figure 6.** Thermogravimetric analysis of red clay membrane, activated carbon, and CaF<sup>2</sup> under air. sintered membrane, which was determined by the broad peak between 2θ = 20° and 2θ =

**Figure 7.** XRD for fabricated membrane: (**a**) without sintering operation and (**b**) with sintering **Figure 7.** XRD for fabricated membrane: (**a**) without sintering operation and (**b**) with sintering operation at 1000 ◦C.

operation at 1000 °C.

tion 3 (Figure 8).

The water flux was obtained and the water permeate was collected for 4 h at different operating pressures (3, 4, 5, and 6 bar, corresponding to 300, 400, 500, and 600 kPa) and using ultra-pure paraffin oil as a feed at 64 NTU concentration. The water flux rate was calculated from equation 1 and the membrane permeability was determined from equa-

3.2.1. Evaluation with Synthetic Oil Emulsion

### *3.2. Evaluation of Fabricated Membrane*

#### 3.2.1. Evaluation with Synthetic Oil Emulsion *Crystals* **2021**, *11*, x FOR PEER REVIEW 10 of 14

The water flux was obtained and the water permeate was collected for 4 h at different operating pressures (3, 4, 5, and 6 bar, corresponding to 300, 400, 500, and 600 kPa) and using ultra-pure paraffin oil as a feed at 64 NTU concentration. The water flux rate was calculated from Equation (1) and the membrane permeability was determined from Equation (3) (Figure 8). *Crystals* **2021**, *11*, x FOR PEER REVIEW 10 of 14

**Figure 8.** Water flux rate and permeability of fabricated membrane at different pressures. **Figure 8.** Water flux rate and permeability of fabricated membrane at different pressures. for the fabricated membrane at low operating pressures. The obtained results were com-

As shown in Figure 8, the flux changed from 191.38 to 284.99 L/m2·h on increasing the applied pressure from 3 to 6 bar. The data in Figure 9 were fitted with Darcy's law [23] and the membrane permeabilities were obtained. High water permeability was observed for the fabricated membrane at low operating pressures. The obtained results were comparable to those from the RO process [24]. Based on Figure 8, the standard deviation (5.36) is lower than the mean (179.84), indicating that the data is reliable. In addition, a 90% confidence level is 185.10–174.58, i.e., we are 90% certain that the mean lies between 185.10 As shown in Figure 8, the flux changed from 191.38 to 284.99 L/m<sup>2</sup> ·h on increasing the applied pressure from 3 to 6 bar. The data in Figure 9 were fitted with Darcy's law [23] and the membrane permeabilities were obtained. High water permeability was observed for the fabricated membrane at low operating pressures. The obtained results were comparable to those from the RO process [24]. Based on Figure 8, the standard deviation (5.36) is lower than the mean (179.84), indicating that the data is reliable. In addition, a 90% confidence level is 185.10–174.58, i.e., we are 90% certain that the mean lies between 185.10 and 174.58 with a small margin of error. parable to those from the RO process [24]. Based on Figure 8, the standard deviation (5.36) is lower than the mean (179.84), indicating that the data is reliable. In addition, a 90% confidence level is 185.10–174.58, i.e., we are 90% certain that the mean lies between 185.10 and 174.58 with a small margin of error. The oil emulsion was tested at 5 bar (500 kPa) to measure the flux over time using a feed concentration of 64 NTU. Based on the standards of the industry, UF can be conducted in the range 4–7 bar (400–700 kPa) [25]. Hence, to determine a perfect water flux, the process was run at 5 bar (500 kPa). The water flux rate was calculated from equation 1 and the rejection was determined from equation 2 (Figure 9).

**Figure 9.** Water/synthetic oil flux rate through membrane and percentage rejection by a fabricated membrane at different times. brane over the course of testing, resulting in a decreased flux through membrane fouling. **Figure 9.** Water/synthetic oil flux rate through membrane and percentage rejection by a fabricated membrane at different times.

**Figure 9.** Water/synthetic oil flux rate through membrane and percentage rejection by a fabricated

Figure 9 shows that the water flux rate decreased over 4 h from 490.28 to 367.32 L/m2·h. This behavior could be due to a considerable quantity of oil deposited on the membrane over the course of testing, resulting in a decreased flux through membrane fouling.

membrane at different times.

The oil emulsion was tested at 5 bar (500 kPa) to measure the flux over time using a feed concentration of 64 NTU. Based on the standards of the industry, UF can be conducted in the range 4–7 bar (400–700 kPa) [25]. Hence, to determine a perfect water flux, the process was run at 5 bar (500 kPa). The water flux rate was calculated from equation 1 and the rejection was determined from equation 2 (Figure 9). *Crystals* **2021**, *11*, x FOR PEER REVIEW 11 of 14

Figure 9 shows that the water flux rate decreased over 4 h from 490.28 to 367.32 L/m<sup>2</sup> ·h. This behavior could be due to a considerable quantity of oil deposited on the membrane over the course of testing, resulting in a decreased flux through membrane fouling. Nandi et al. [26] observed similar results warranting an efficient cleaning procedure to remove foulants from the membrane. Nandi et al. [26] observed similar results warranting an efficient cleaning procedure to remove foulants from the membrane.

The standard deviation (3.65) is lower than the mean (404.25), indicating that the data is reliable. In addition, the 90% confidence level is 404.26–404.24, i.e., we are 90% certain that the mean is between 404.26 and 404.24 with a small margin of error. The standard deviation (3.65) is lower than the mean (404.25), indicating that the data is reliable. In addition, the 90% confidence level is 404.26–404.24, i.e., we are 90% certain that the mean is between 404.26 and 404.24 with a small margin of error.

### 3.2.2. Evaluation with Aramco Oil-Contaminated Water 3.2.2. Evaluation with Aramco Oil-Contaminated Water

Contaminated water obtained from Aramco was tested at 5 bar (500 kPa) to measure the flux over time using a feed concentration of 80 NTU. The water flux rate was calculated from Equation (1) and the rejection was determined from Equation (2) (Figure 10). Contaminated water obtained from Aramco was tested at 5 bar (500 kPa) to measure the flux over time using a feed concentration of 80 NTU. The water flux rate was calculated from equation 1 and the rejection was determined from equation 2 (Figure 10).

It was observed that the water flux decreased because of oil deposition on the membrane surface causing membrane fouling, while the oil rejection changed from 99.23 to 99.96%. The decrease in the water flux was related to oil precipitation on the membrane. Based on Figure 10, the standard deviation (3.64) is lower than the mean (318.19) indicat-It was observed that the water flux decreased because of oil deposition on the membrane surface causing membrane fouling, while the oil rejection changed from 99.23 to 99.96%. The decrease in the water flux was related to oil precipitation on the membrane. Based on Figure 10, the standard deviation (3.64) is lower than the mean (318.19) indicating that the data is reliable; the 90% confidence level is 318.19–318.15.

### ing that the data is reliable; the 90% confidence level is 318.19–318.15. *3.3. Cleaning Mechanism of a Fouled Membrane and Cyclic Filtering Test*

*3.3. Cleaning Mechanism of a Fouled Membrane and Cyclic Filtering Test* Membrane fouling has detrimental effects on membrane performance. During oilwater separation, fouling occurs due to the interaction between the membrane and oil droplets in the wastewater; the cohesion between the foulant and membrane surface depends on membrane surface properties, such as its zeta potential and hydrophilicity [23]. Membrane fouling has detrimental effects on membrane performance. During oilwater separation, fouling occurs due to the interaction between the membrane and oil droplets in the wastewater; the cohesion between the foulant and membrane surface depends on membrane surface properties, such as its zeta potential and hydrophilicity [23]. Here, the fouling problem on the membrane was investigated via a cyclic filtrating test performed for 1 h, and the flux rate was determined. Next, backwashing was performed

Here, the fouling problem on the membrane was investigated via a cyclic filtrating test

membrane. Then, the water flux was recalculated, and a total of seven experimental cycles

Performing the filtration periodically can prevent the tendency of fouling and maintain a nearly constant flux rate value owing to the repeated backwash purification of the membrane. The plot of the water flux rate as a function of the number of cycles (Figure 11) revealed that the membrane retained a significant negative surface charge, as shown in the zeta potential plot of the fabricated membrane (Figure 3), resulting in an almost constant water flux. This also allowed for the membrane to be efficiently cleaned. It is

were performed (Figure 11).

for 10 min to clean foulants from the membrane by pushing water mixed with air into the membrane. Then, the water flux was recalculated, and a total of seven experimental cycles were performed (Figure 11). brane suitable for use in oil-water separation. Figure 11 shows that the standard deviation (6.24) is lower than the mean (287.89); therefore, the data is reliable. In addition, the 90% confidence level is 296.32–279.45.

possible to purify and reform the negatively charged membrane during backwashing because the oil droplets are also negatively charged (Figure 3), yielding suitable water flux with sufficient cycling. The backwashing potential arising from the repulsion between the negatively charged oil droplets and the membrane surface renders the developed mem-

*Crystals* **2021**, *11*, x FOR PEER REVIEW 12 of 14

**Figure 11.** The water flux of seven experimental cycles filtration for oil separation by fabricated **Figure 11.** The water flux of seven experimental cycles filtration for oil separation by fabricated membrane.

membrane.

[30]

The effectiveness of the fabricated membrane was similar to those of membranes described in the literature [9,27–30], and some comparisons are presented in Table 2. It is evident that the fabricated membrane presented favorable performance with red clay, which is locally available, and an inexpensive material. Table 2 also presents data for some expensive membrane materials, such as NaA zeolite deposited on α-Al2O3. Additionally, our tubular ceramic membrane has a higher water flux (367.32 L/m<sup>2</sup> h) than the tubular PVDF-UF at the same operating pressure (5–6 bar, 309 L/m<sup>2</sup> h). **Table 2.** Performance comparison of the fabricated membrane in this study with membranes in-Performing the filtration periodically can prevent the tendency of fouling and maintain a nearly constant flux rate value owing to the repeated backwash purification of the membrane. The plot of the water flux rate as a function of the number of cycles (Figure 11) revealed that the membrane retained a significant negative surface charge, as shown in the zeta potential plot of the fabricated membrane (Figure 3), resulting in an almost constant water flux. This also allowed for the membrane to be efficiently cleaned. It is possible to purify and reform the negatively charged membrane during backwashing because the oil droplets are also negatively charged (Figure 3), yielding suitable water flux with sufficient cycling. The backwashing potential arising from the repulsion between the negatively charged oil droplets and the membrane surface renders the developed membrane suitable for use in oil-water separation.

vestigated in the literature. Figure 11 shows that the standard deviation (6.24) is lower than the mean (287.89); therefore, the data is reliable. In addition, the 90% confidence level is 296.32–279.45.

> **Permeation Flux (L/m<sup>2</sup> h)**

0.1 810

1.37 66

1 224

**Ref Membrane Type Solution Pressure (bar)** [27] cellulose microfiltration membranes Synthetic produced water 3 200 The effectiveness of the fabricated membrane was similar to those of membranes described in the literature [9,27–30], and some comparisons are presented in Table 2. It is evident that the fabricated membrane presented favorable performance with red clay, which is locally available, and an inexpensive material. Table 2 also presents data for some expensive membrane materials, such as NaA zeolite deposited on α-Al2O3. Additionally, our tubular ceramic membrane has a higher water flux (367.32 L/m<sup>2</sup> h) than the tubular PVDF-UF at the same operating pressure (5–6 bar, 309 L/m<sup>2</sup> h).

[28] PAN nanofiber membrane Synthetic pro-

[29] a-Al2O<sup>3</sup> ceramic membrane Synthetic pro-

Magnesium bentonite hollow fiber ceramic membrane

duced water

duced water

Synthetic produced water


**Table 2.** Performance comparison of the fabricated membrane in this study with membranes investigated in the literature.

The practical implication of the membrane fabricated in this study is the clean, oil-free water, which can be used in agriculture and cooling systems. The scientific contribution of this study is the use of new materials, such as CaF<sup>2</sup> and nano-activated carbon, to fabricate a ceramic membrane from a low-cost material like red clay. Our working hypotheses of using low-cost materials for ceramic membrane fabrication were confirmed by the obtained results. However, the membrane fabricated in this study is only limited to separating oil and water. To expand its for use in industry, its performance must be investigated regarding different types of industrial pollutants. In addition, irreversible fouling also requires further research.

### **4. Conclusions**

An efficient new membrane was fabricated from red clay combined with CaF<sup>2</sup> as a binder and nano-activated carbon as a pore former. These materials were used to fabricate a porous membrane by the extrusion technique, and the membrane was applied for the purification of oil-contaminated water. CaF<sup>2</sup> was used to minimize the sintering temperature of the red clay membrane and increase the mechanical strength of the membrane as a nucleation promoter.

The fabricated membrane was tested using both, a synthetic oil-water emulsion and water produced from an oilfield from Aramco. The fabricated membrane had an average pore size of 95.46 nm; thus, it qualified as a UF membrane. It showed a good bending strength of 54.13 MPa and a contact angle of 67.3◦ , indicating hydrophilicity. The performance of the fabricated membrane complied with the standards of the national wastewater. The clean water also met the desired standards.

In the separation of oil from water, the flux was increased upon increasing the applied pressure. High water permeability was obtained for the fabricated membrane under low operating pressure, and this result was fitted with Darcy's law. The membrane flux decreased by oil deposition on the membrane surface; regardless, the maximum oil rejection was 99.96% at the oil concentration of 80 NTU and pressure of 5 bar (500 kPa). The prepared membrane showed high efficiency in removing foulants by the backwash technique because of the charge repulsion forces between the oil molecules and the negatively charged membrane. The fabricated membrane showed good potential applicability in oil-water separation treatments.

**Funding:** This research received no external funding.

**Acknowledgments:** The author appreciates KACST for their encouragement during this study.

**Conflicts of Interest:** The author declare no conflict of interest.

### **References**

