Continuous and efficient oil/water separation by special wettability granular filter media

Oily wastewater is a kind of industrial wastewater with a large volume and wide range. It mainly comes from the food, textile, petroleum, natural gas, petrochemical, and metal processing industries (Shannon et al. 2008). Without effective treatment, oily wastewater will cause serious harm to the environment and human health (Kintisch 2010). Traditional oil/water separation technology includes gravity separation (Hafiz et al. 2005), coagulation (Zaroual et al. 2006; Calvo et al. 2010), skimming (Song et al. 2015), flotation (Moosai & Dawe 2003), membrane separation (Tanudjaja et al. 2019), adsorption method (Pintor et al. 2016), and so on. However, these separation technologies have problems, such as low separation efficiencies, high energy consumption, complicated operational processes, and easy secondary pollution (Zhang et al. 2013). Therefore, more efficient and low-energy consumption oil/water separation technology has become an issue of extensive concern (Li et al. 2017; Lian et al. 2018; Du et al. 2019).

In recent years, special wettability materials have been widely researched (Wang et al. 2015). Because the surface tensions of oil and water are different, the wettability of oil and water on the same solid surface may also be different. Based on this feature, some researchers have developed oil/water separation materials with special wettability, such as selective filtration membranes and three-dimensional adsorption materials (Wang et al. 2016). The substrates of the selective filtration membranes include iron mesh (Cheng et al. 2017), stainless mesh (Jiang et al. 2019; Shang et al. 2019; Tang et al. 2019; Tian et al. 2019), copper mesh (Tudu & Kumar 2019; Wang & Liu 2020; Yin et al. 2020a), nickel mesh (Yin et al. 2020b), aluminum mesh (Zhang et al. 2019), commercial fabrics (Yao et al. 2019; Meng et al. 2020), ethylene terephthalate nonwoven fabric (Yagoub et al. 2019), polyvinylidene difluoride (Obaid et al. 2017), and so on. The selective filtration membrane allows lyophilic liquid to pass through the membrane hole and prevents the lyophobic liquid from passing through to achieve oil/water separation. Selective filtration has the advantage of high separation efficiency, but the membrane hole is small, usually at the nanometer/micron level; as a result, the membrane flux is small, and the separation process even requires additional pressure. In addition, once the membrane hole is contaminated, it is difficult to clean. The substrates of the three-dimensional adsorption materials include fibers (Li et al. 2014; Zang et al. 2015), sponges (Ge et al. 2014; Ge et al. 2015), cotton (Cheng et al. 2019), filter paper (Du et al. 2014), and so on. The separation method is as follows: first, the adsorption material is immersed in the oil/water mixture, and then, the adsorption material is removed after the lyophilic liquid is adsorbed. This technology also has a great separation efficiency. To reuse the adsorption material, the adsorbed lyophilic liquid needs to be extruded by external forces, such as extrusion, so it has the disadvantage of a complicated operational process. These shortcomings hinder the practical application of selective filtration membranes and three-dimensional adsorption materials.

To overcome these shortcomings, we chose a hard granular filter medium as the substrate, modified it into a special wettability filter medium, and packed it into the filter column for oil/water separation. The filter layer allows lyophilic liquid to flow through the porous channel and intercepts lyophobic liquid. When the filter layer is polluted by the lyophobic liquid, it can be backwashed. Therefore, the operation of the oil/water separation process is very simple. The oil/water mixture involves two liquid phases, namely the water phase and the oil phase. If only one special wettability filter media is used for oil/water separation, the lyophobic liquid will accumulate in the separator, thus blocking the lyophilic liquid from the filter media; as a result, the separation process cannot be continued. For those reasons, we developed a separator, which has two filtration columns, one column packed with superhydrophilic and underwater superoleophobic filter media that allows water to pass through but not oil, and another column packed with superhydrophobic and superoleophilic filter media that allows oil to pass through but not water. Consequently, this separator can separate oil/water mixtures continuously.

Here, the two special wettability granular filter media mentioned above were fabricated using quartz sand as the substrate. Quartz sand is widely used in water treatment because of its high hardness, good stability, and low price (Liu et al. 2018a). We have studied the superhydrophobic and superoleophilic quartz sand filter media in an earlier stage (Liu et al. 2018b) and directly adopted the same fabrication method in this paper. For superhydrophilic and underwater superoleophobic media, the modifiers mainly include hydrogel (Teng et al. 2014), polymer reagents (Yang et al. 2015), ZnO nanoparticles (Budnyak et al. 2015), graphene oxide (Huang et al. 2015), and agro-waste. Because of its advantages, such as its ease of acquisition, large quantity, and low cost, agro-waste has attracted extensive attention from researchers in the field of environmental engineering and science in recent years. For example, Jing et al. (2019) using corn bracts, prepared biomass carbon (BC) aerogels with large specific surface areas and superhydrophobicity. The BC aerogels exhibited an extremely high absorption capacity and excellent reusability for oil and could be used to separate oil from water. Jamalludin et al. (2019) prepared a hybrid green ceramic membrane using agricultural-sugarcane bagasse waste for the adsorbent/separation of real oily wastewater. Potato residue is a kind of waste produced in the food processing industry. Potato residue is an ideal choice for designing new underwater superoleophobic materials because it contains a large amount of hydrophilic hydrogen bonding. Therefore, we fabricated superhydrophilic and underwater superoleophobic quartz sand filter media by coating potato residue on the quartz sand surface.

This paper studies the separation performance of the separator, including the separation efficiency, intrusion pressure, hydraulic conductivity, backwash performance, and so on. The results show that the separator can separate oil/water mixtures continuously and efficiently and that its separation efficiency is more than 99.4%.

Quartz sand was purchased from Henan Hongda Filter Media Co., Ltd, Henan Province. Potato residue was provided by Yinguang Starch Factory, Lintao County, Gansu Province. Waterborne polyurethane was produced by Shenzhen Jitian Chemical Co., Ltd. Petroleum ether, cyclohexane, sodium hydroxide, ethanol, methylene chloride, and trichloroethane were all produced by Rionlon Bohua (Tianjin) Pharmaceutical & Chemical Co., Ltd. Engine oil (10 W-40) was produced by the Lubricating Oil Branch of China Petroleum Corporation. Diesel was purchased from local petrol stations. Methylene blue was produced by Tianjin Damao Chemical Reagent Factory. Oil red O was purchased from Shanghai Zhongqin Chemical Reagent Co., Ltd. Deionized water was made in the laboratory. The relevant parameters of the several oils and water used for separation are shown in Table S1 (Supporting Information).

Quartz sand was sieved with stainless steel, washed with distilled water, dried in an oven at 110 °C, washed twice with ethanol, and dried to obtain pretreated quartz sand. Twenty grams of potato residue was sieved with a 400-mesh stainless steel sieve, immersed in 500 mL of a 9% NaOH aqueous solution at room temperature, ultrasonically shaken with an ultrasonic oscillator (KQ5200DB, Kunshan Ultrasonic Instrument Co., Ltd, China) for 2 h, sealed and placed in a 0 °C environment for 12 h. After drying at 100 °C for 1 h, activated potato residue was obtained. Activated potato residue was added to 500 mL of ethanol and magnetically stirred at room temperature for 4 h. Then, 4 g of waterborne polyurethane was added and stirred for 15 min. Finally, the pretreated quartz sand was immersed in the prepared mixed solution, sealed, and immersed at room temperature for 48 h; dried at 80 °C; repeatedly washed with deionized water, and dried again at 80 °C. Superhydrophilic and underwater superoleophobic quartz sand filter media (hereafter referred to as PR@QS) were obtained.

Superhydrophobic and superoleophilic quartz sand filter media were prepared based on our previous research (Liu et al. 2018b). Briefly, 25 g tris(hydroxymethyl)aminomethane solid particles were dissolved in 500 mL of an ethanol solution, and then 1.7 g ZnO nanoparticles and 1.0 mg dopamine hydrochloride were added and stirred. Next, 40 g pretreatment quartz sand was added and ultrasonically dispersed at 40 kHz for 5 min using an ultrasonic oscillator and then stirred for 16 h at 650 rpm using a JB-2 thermostatic stirrer. The quartz sand filter was finally leached and dried in an oven at 80 °C to obtain ZnO nanoparticle-roughened quartz sand. ZnO nanoparticle-roughened quartz sand was added to 10 mL octadecyltrichlorosilane (OTS) dissolved in 100 mL of a toluene solution, stirred at 650 rpm for 2 h, leached, and dried again in an oven at 80 °C. Superhydrophobic and superoleophilic quartz sand filter media (hereafter referred to as OTS-ZnO@QS) were obtained.

A contact angle meter (KRÜSS DSA100, Germany) was used to measure the static contact angle of the PR@QS surface at room temperature. PR@QS was placed into water, 5 μL oil was dropped on the surface of PR@QS, photos were taken after 15 s, and the contact angle was measured. All contact angles were tested at different positions at least three times. X-ray energy spectroscopy (SEM, Kevex, USA) was used to study the surface morphology of the filter media. A PHI-5702 X-ray photoelectron spectrometer (XPS) was used to analyze the surface elements of the filter media. An infrared spectrum analyzer (FTIR) was used to analyze the chemical structure of the filter media surface. Because the characterization of OTS-ZnO@QS has been studied in previous papers (Liu et al. 2018b), we do not repeat it here. Therefore, this paper only characterizes PR@QS.

PR@QS was immersed in an ethanol solution and placed in an ultrasonic oscillator. After ultrasonic oscillation for 12, 24, and 48 h at 40 kHz, PR@QS was removed and washed with deionized water five times. Then, PR@QS was dried, and the contact angle was measured to investigate the mechanical wear resistance of the filter media. Similarly, to study the acid and alkali resistance, PR@QS was immersed in aqueous solutions with pH values of 1, 5, 10, and 14 for 48 h. To study the surfactant resistance, PR@QS was immersed in a sodium dodecylbenzene sulfonate (SDBS) aqueous solution for 48 h. PR@QS was exposed to air at room temperature, and the contact angles were measured after 30, 60, and 90 d to study the air exposure stability of the filter media.

Four different particle sizes of filter media (PR@QS and OTS-ZnO@QS) were packed into the filter column. The range of the particle diameters, equivalent particle diameters ⁠, and the capillary radius of packed filter layer are shown in Table S2 (Supporting Information), where the equivalent particle sizes are represented by the geometric mean particle sizes.

During the separation process of the oil/water mixture, if the oil concentration in the water outlet sharply increased, it indicated that the oil phase had intruded the superhydrophilic PR@QS column. Similarly, if the water concentration in the oil outlet sharply increased, it indicated that the water phase had intruded the superoleophilic OTS-ZnO@QS column. When the filter column was intruded, it needed to be backwashed. For the OTS-ZnO@QS column, the operational process of backwashing is shown in Figure 1(c) as follows: oil is pumped from the oil outlet into the OTS-ZnO@QS column and flows into the intake chamber, and is then discharged from the intake chamber. For the PR@QS column, water is pumped. The amount of lyophilic liquid used in each backwashing was approximately five times the volume of the filter layer, that is, approximately 500 mL. The backwashing liquid could be re-separated.

The potato residue is a by-product of cell fragments and residual starch particles produced in the production of potato starch. The main component is cellulose. The intramolecular hydrogen bonds and intermolecular hydrogen bonds of cellulose aggregate into different fibril structures with high crystallinity, which makes its highly reactive hydrogen bonds locked in the crystallization zone and difficult to contact with water. Therefore, intramolecular hydrogen bonds should be released by activation. The activated potato residue has more hydroxyl groups on its surface, which is easy to blend with waterborne polyurethane to form potato residue/waterborne polyurethane composite emulsion. When the quartz sand is immersed in the emulsion, the waterborne polyurethane adheres to the surface of the quartz sand, thus introducing potato residue into the quartz sand filter media surface.

As seen from the scanning electron microscope (SEM) images of PR@QS shown in Figure 2, the surface of pristine potato residue is smooth and complete without defects. After combined activation by alkali and ultrasonication, the surface of the potato residue becomes uneven, with the formation of more pores and cracks, indicating that the activation process is successful. The surface of pristine quartz sand has more edges and corners, and its surface porosity is not developed. While the surface of PR@QS is covered with a thin film and there are micro/nano scale pits, it should be a potato residue coating layer, which indicates that there is a rough structure when potato residue is introduced.

Potato residue is mainly composed of cellulose, hemicellulose, and lignin. The –CH₂OH group of activated cellulose will form a condensation reaction with the –OH of quartz sand to graft the hydrophilic group onto the surface of quartz sand. It can be seen from the FTIR spectra (Figure 3(a)) that, for PR@QS, there is a strong stretching vibration peak of C–O belonging to alcohol, phenol, ether, or ester at 1,079 cm⁻¹. At 2,921 cm⁻¹, there is an aliphatic group –CH stretching vibration peak (Deng et al. 2011). These groups belong to the characteristic peak of potato residue (Zhang et al. 2015) and do not exist in the spectrogram of pristine quartz sand, indicating that potato residue has been coated to the surface of quartz sand. At 3,427 cm⁻¹, the stretching vibration peak of O–H is wider and stronger (Fang et al. 2008), indicating that there are more hydroxyl groups on the surface of PR@QS than on pristine quartz sand.

The same results can be obtained from XPS analysis (as shown in Figure 3(b)). Before activation, potato residue was mainly composed of C (288.0 eV, 77.98%), O (535.6 eV, 19.70%), and N (411.6 eV, 1.32%), with a small amount of Si (103.5 eV, 0.99%). After activation, the Na (501.2 eV) element content reached 14.97%, which was induced by NaOH activation. The concentration of O (31.30%) was higher than that before activation, which was generated by the introduction of –OH and cellulose molecules. Pristine quartz sand is mainly composed of O (532.8 eV) and Si (103.5 eV), with atomic concentrations of 69.55 and 30.45%, respectively. For PR@QS, Na (501.2 eV, 1.04%) and C (288.0 eV, 38.78%), the characteristic elements of activated potato were introduced, and the content of Si (103.5 eV, 15.23%) was reduced, indicating that the surface of PR@QS was coated with potato residue, thus reducing Si exposure.

The wettability of a solid surface can be represented by the contact angle. According to our previous study, the water contact angle of OTS-ZnO@QS is 154.1°, while the oil contact angle is approximately 0°. OTS-ZnO@QS has superhydrophobic and superoleophilic characteristics (Liu et al. 2018b).

For PR@QS, both water and oil (dichloromethane) can rapidly spread on the PR@QS surface in air, and the contact angle is approximately 0°, as shown in Figure 4(a), indicating that PR@QS is superhydrophilic and superoleophilic in air. Placing PR@QS in water and then dropping oil onto its surface leads to the formation of spherical shaped oil, as shown in Figure 4(b), indicating that PR@QS has underwater superoleophobic characteristics. The underwater oil contact angles of all four light oils and two heavy oils are shown in Figure 4(c). Figure 4(c) shows that the underwater engine oil contact angle is the smallest at 151.6° ± 1.7°, while that of trichloromethane is the highest at 160.9° ± 1.7°.

For example, the underwater contact angle of cyclohexane is 155.4°. Substituting it into Equation (4), it can be calculated that ⁠, that is, 95.5% of the filter surface is covered by water film.

The wetting stability of PR@QS was also studied, as shown in Figure 4(d). Ultrasound oscillation is often used to clean impurities on the surfaces of substances because the ultrasonic process can physically damage these impurities. Therefore, oscillation is often used to study the mechanical stability of superwettable surfaces (Milionis et al. 2016). Figure 4(d) shows that the underwater oil contact angle remains at 150.2° ± 0.5° after 48 h of oscillation, indicating that PR@QS has strong mechanical stability. Under acidic and alkaline conditions, the underwater superoleophobicity of PR@QS slightly decreases. The underwater oil contact angles were 141.6° ± 1.3° and 144.9° ± 1.7° at pH 1 and pH 5, respectively, and 149.3° ± 0.7° and 142.9° ± 1.3° at pH 10 and pH 14, respectively. These findings indicate that PR@QS does not have strong acid resistance but has good alkali resistance. After washing with SDBS, the underwater oil contact angle of PR@QS decreased to 150.8° ± 0.9°, still maintaining a good underwater superoleophobic property. This finding indicates that PR@QS has strong resistance to surfactants. After a long exposure to air, the underwater oil contact angle of PR@QS decreased to some extent, but the decrease was not significant. After 90 d, the underwater oil contact angle was 144.5° ± 1.11°, which still had good underwater oleophobicity. Therefore, PR@QS could be stored for a long time.

When the particle sizes of the two-filter media are 0.355–0.577 mm and the length of the filter columns are 10 cm, the separation efficiencies of the five oil/water mixtures are shown in Figure 5. As seen from Figure 5(a) and (b), water is invisible to the naked eye in the oil outlet beaker and the oil cannot be observed with the naked eye at the water outlet beaker. After separating the 1 L mixture, the oil content at the water outlet and the water content at the oil outlet were determined. The oil and water separation efficiencies of all five kinds of oils were between 99.40 and 99.99%. These high separation efficiencies can be explained by the capillary pressure of the liquid in the filter layer, as shown in Figure 6 (Bigui et al. 2019). As seen from Figure 6, for superhydrophobic and superoleophilic OTS-ZnO@QS, the capillary pressure of the oil points to the outlet side of the filter layer, driving the oil phase flow through the filter layer, while the capillary pressure of the water points to the inlet side, preventing the water phase from passing through the filter layer (Liu et al. 2018b). Therefore, the outlet of the OTS-ZnO@QS filter column is oil. Similarly, for superhydrophilic and underwater superoleophobic PR@QS, the capillary pressure drives the water phase flow through the filter layer and prevents the oil phase from passing through (Yong et al. 2016), so the outlet is water.

During the separation process, due to the height of the liquid level in the intake chamber, hydrostatic pressure will be generated on the filter layer. Theoretically, when the hydrostatic pressure is less than the theoretical capillary pressure, the lyophobic liquid will not intrude on the filter layer; when the hydrostatic pressure exceeds the theoretical capillary pressure, the lyophobic liquid will intrude the filter layer (Li et al. 2018a). When the filter layer is just intruded, the hydrostatic pressure is called the theoretical intrusion pressure, which is equal to the capillary force (Li et al. 2018a).

The theoretical and experimental values of the intrusion pressure of several lyophobic liquids on the two filter layers are slightly different, as shown in Figure 7(b). As seen from Equation (5) and Figure 7(b), the theoretical intrusion pressure is related to the surface tension and contact angle of the lyophobic liquid. Since the surface tension of water is the maximum, it has the maximum intrusion pressure. The experimental and theoretical values are up to 0.29 ± 0.01 and 0.31 kPa, respectively. The theoretical intrusion pressure of engine oil is the lowest, 0.14 kPa, but the experimental intrusion pressure is as high as 0.27 ± 0.02 kPa. The reason for this difference is that the viscosity of engine oil is too high and the flow speed is too slow, so the experimental error is relatively large.

Figure 7(c) shows that the experimental hydraulic conductivities of the six liquids in the filter layer are slightly different from those of the theoretical data, but the trends are consistent. As seen from Equation (8), the properties of the liquid (such as density and viscosity) have effects on the hydraulic conductivity. For the 0.355–0.577 mm particle size filter media layer, the experimental hydraulic conductivity of water in the PR@QS layer was up to 7.2 ± 1.7 m/h, slightly higher than the calculated value of 5.9 m/h. For OTS-ZnO@QS, the hydraulic conductivity of petroleum ether is the largest, 14.4 ± 0.3 m/h, due to the minimum viscosity. The value of methylene chloride is 13.13 m/h, followed by cyclohexane at 6.18 m/h and diesel at 3.56 m/h. Because the engine oil viscosity is too large, approximately 350 times that of petroleum ether, the hydraulic conductivity is very small, only 0.034 ± 0.001 m/h.

The theoretical and experimental intrusion pressures of filter media with different particle sizes are shown in Figure 8(b). For the OTS-ZnO@QS layer, the experimental intrusion pressure of water is slightly less than the calculated value. For the PR@QS layer, the oil intrusion pressure of the experimental value is consistent with the theoretical value when the equivalent diameter is greater than 0.4 mm, while the experimental value is far greater than the theoretical value when the equivalent diameter is less than 0.4 mm, but the trends are identical. From Equation (8), the hydraulic conductivities of the filter layer are proportional to the square of the particle size. The experimental and theoretical hydraulic conductivities of the filter layer with different particle sizes are shown in Figure 8(c). The experimental values are consistent with the theoretical values.

Through comprehensive analysis of the influences of particle size on separation efficiency, intrusion pressure, and hydraulic conductivity, it can be seen that if the filter media particle size is too large, it is not conducive to practical application for oil/water separation. Therefore, it is recommended that the equivalent particle size should not be greater than 0.6 mm.

Equation (10) is the calculation formula for the water head loss of the clean filter bed. Equation (10) shows that the head loss is inversely proportional to the length of the filter column L. With the same flow rate, the head loss increases proportionally as the filter length increases. With the same head loss, the flow rate decreases with increasing filter length. Therefore, the length of the filter column should not be too long. In the actual separation, the filter layer length is recommended to be 10 cm.

To research the continuous separation performance of the light oil/water mixture, 0.355–0.577 mm filter media and 10 cm-long filtration columns were used to conduct continuous separation of the diesel and water mixture. The experimental results are shown in Figure 9 and Figure 10. Figure 9 shows that the oil and water separation efficiencies are greater than 99%. Once the filter layer is intruded by a lyophobic liquid, the oil concentration in the water effluent (or the water concentration in the oil effluent) sharply increases, and the oil (or water) droplets can be seen by the naked eye (as shown in the inset figures in Figure 9). At this time, the filter column should be backwashed. The PR@QS column was intruded by diesel when separating 20 and 40 L mixtures. After backwashing, the separation efficiency is more than 99.5%. This is because when the lyophilic liquid reverse flows through the filter column, it should extrude the lyophobic liquid that has intruded into the filter column. The OTS-ZnO@QS column was intruded by water at the 12th, 24th, and 42nd L of separation, and the oil separation efficiency remained above 99.4%. It can also be seen from Figure 9 that after multiple intrusion/backwashing cycles, the separation efficiency remained unchanged, indicating that the filter column has good recovery performance and can continuously separate light oil/water in large quantities.

As the separation continues, the hydraulic conductivity of the filter column gradually decreases. The hydraulic conductivity of water gradually decreases from 10.1 m/h at the beginning to 4.0 m/h at the 5th L mixture separated and then fluctuates between 2.1 and 6.1 m/h. For diesel, the initial hydraulic conductivity is up to 29.6 m/h, gradually falls to 6.2 m/h at the 8th L mixture separated, and then fluctuates between 4.8 and 7.1 m/h. The initial stage in which the hydraulic conductivity decreases can be called the ‘ripening stage’. When the lyophilic fluid flows through the filter layer, it produces a shear force on the filter layer, thus making the metastable filter media move, making the filter layer denser and the pores smaller. When the filter layer is ripening, the head loss tends to be stable.

In the same way, 0.355–0.577 mm filter media and 10 cm-long filtration columns are used to conduct continuous separation of the dichloromethane and water mixture. The experimental results are shown in Figure 11. From Figure 11, it can be seen that the separation efficiencies of oil and water are all greater than 99.6%. The PR@QS column is intruded by dichloromethane when separating the 5 and 23 L mixtures, and the OTS-ZnO@QS column is intruded by water only once, at the 7th L of separation. Compared with diesel, dichloromethane and water are less intruded when separated, which may be caused by two reasons. The first reason is that the theoretical intrusion pressure of dichloromethane is 1.3 times that of diesel, so the PR@QS column is more likely to be intruded by diesel than by dichloromethane. Second, diesel oil is a mixture separated light oil, it is in the upper layer in the intake chamber, and water is in the lower layer. As a result, the hydrostatic pressure of water is higher than that of the dichloromethane/water mixture, so water more easily intrudes on the OTS-ZnO@QS column.

Similar to the process of separating the diesel/water mixture, the hydraulic conductivity of water gradually decreases from 10.3 m/h at the beginning to 5.5 m/h at the 13th L of the mixture separated and then fluctuates between 2.0 and 6.2 m/h. For dichloromethane, the hydraulic conductivity decreases from 13.5 to 3.0 m/h at 13th L of the mixture separated and then fluctuates between 2.4 and 3.9 m/h. The ‘ripening stage’ is before filtering the 8th L of the mixture is separated. The experimental results show that the separator can separate heavy oil and water mixtures continuously and efficiently in large quantities.

Some researchers have also studied the repeated separation of oil/water mixtures with special wettability materials, and the relevant research results are listed in Table 1. Compared with previous research results, this paper still has greater separation efficiency under the condition of a larger separation volume.

In summary, we report a method for continuous oil/water separation using special wettability materials. Two separation columns were filled with two kinds of superwettable granular filters with opposite wettabilities, and the two separation columns were connected with the same separation chamber. The superhydrophobic and superoleophilic filter column only allows the oil phase to pass through, not the water phase. The superhydrophilic and underwater superoleophobic filter column only allows the water phase to pass through, not the oil phase. Consequently, both the oil phase and water phase of the oil/water mixture in the separation chamber can flow out from its own filter column, which realizes the separation of oil and water continuously. Superhydrophobic and superoleophilic filter media were prepared according to previous research results by our team, and superhydrophilic and underwater superoleophobic filter media were prepared by activated potato residue coated with quartz sand. The experimental results show that all the separation efficiencies of the five oil/water mixtures (petroleum ether, engine oil, diesel oil, cyclohexane, methylene chloride, and trichloromethane) were greater than 99.4%. It is recommended that the equivalent diameter of the granular filter should not be greater than 0.6 mm and the length of the filter layer should be 10 cm. In the continuous separation of oil/water mixtures, the backwashing method can be used to extrude the lyophobic liquid that intrudes the filter layer and to recover the separation efficiency of the filter layer. Therefore, the oil/water separator can separate oil/water mixtures continuously and efficiently and can be applied to practical industrial production.

This work was supported by the National Natural Science Foundation of China (Nos 51668032 and 52060014) and the Foundation of A Hundred Youth Talents Training Program of Lanzhou Jiaotong University (No. 152022).

All relevant data are included in the paper or its Supplementary Information.