*2.4. The Performance of the Membranes with Synthetic Feed Waters*

After ensuring the good performance of the module by conducting the hot/cold water test, the main test was conducted at the atmospheric pressure of one for 6 h in two phases with direct contact membrane distillation setup (Figure 4 and Figure S1). The first phase included operations using an unmodified membrane and the second phase involved operations with a modified membrane. The temperature difference of the feed and the permeate was kept constant during the process (about 50 degrees on average). First, water containing nitrate was synthesized by adding 5 wt % potassium nitrate.

**Figure 4.** Lab-scale schematic diagram of direct contact membrane distillation setup.

Parallel counter-currents (the hot-water feed containing nitrate ions and the cold water containing pure permeate) were formed using two peristaltic pumps with a constant speed of 37 rpm. The feed container was placed in a hot water tank whose temperature was controlled. The permeate container was placed in a tank containing a mixture of water and ice and its temperature was controlled. The temperatures of the feed and permeate were controlled by isolating the system. The water vapor passed through the membrane from the feed side to the permeate side. In the cold side, vapor lost its energy and became water. During the test, the weight and volume of the feed gradually decreased, while the volume and weight of the permeate increased. The amount of ion in the feed and permeate containers was measured by an electrical conductivity meter every hour. The weights of the feed and permeate were also measured by a scale every hour.

#### *2.5. Pilot Process Simulation*

Dong et al. [33] created a mathematical approach by pairing tanks-in-series and a black box to investigate all the main factors of the DCMD process versus the length of the membrane. The designed simulator was used to predict and evaluate the performance of the flat-sheet DCMD at an industrial scale, as well as the impact of the physical characteristics of the membrane, module dimensions, and the implementation of conditions on the performance of the large-scale DCMD module. This research was conducted to obtain the design considerations for the production of pure water. In their study, the software required operational and laboratory-scale data to produce the desired outputs in large-scale countercurrents (Tables S1 and S2 and Figure S2). In the current research, the experimental conditions of Dong et al.'s study were reconstructed. However, nitrate was used instead of NaCl. Due to the similarities between the physical properties of nitrate and NaCl, Dong's MATLAB codes were used to convert experimental results to large-scale results.

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

#### *3.1. Surface Morphology*

The morphology of membranes was characterized by FE-SEM (QUANTA FEG 450, Hillsboro, OR, USA) operating at 15 kV. The comparison of several scanning electron microscopy (SEM) images of the FTCS–TiO2–PVDF membrane and the PVDF membrane (Figure 5a,b) shows that a multi-level hierarchical roughness was created on the surface of the membrane at the end of the process which prevented water droplets from penetrating the surface. Furthermore, while the contact angle increased substantially, the average pore size did not change significantly (Figure 5c–f). It was expected that the performance of the modified membrane was better than that of the unmodified membrane. Furthermore, the microscopic changes on the surface of the membrane were investigated in different stages.

As can be seen in (Figure 5g,h), coating TiO<sup>2</sup> nanoparticles did not significantly change the porosity, pore size, and surface structure of the membrane. These nanoparticles are important in the creation of superhydrophobic properties in two regards: first, they increase the surface roughness; second, they are an anchor for linking the FTCS molecules on the surface.

The microscopic images of coating FTCS on the membrane (Figure 5i,j) show that the self-assembly method was able to create a more uniform coating on the surface than previously reported methods. As can be seen in this figure, despite the penetration of FTCS into the pores, no significant surface structure changes occurred, and the size of the pores did not change. It should also be noted that preserving the size and structure of the pores is crucial in the performance of DCMD. FTCS was used to reduce surface energy chemically. However, as can be observed in (Figure 5i,j) FTCS also increased hydrophobicity by creating roughness. FTCS crystals increased the surface roughness (Figure 5k,i). Furthermore, the images show that the sizes of the pores did not change substantially (Figure 5i–q) SEM images of different samples of the FTCS–PVDF membrane in different magnifications (Figure 5m–q).

**Figure 5.** The FESEM images of (**a**) the virgin PVDF membrane; (**b**–**f**) the FTCS–TiO2–PVDF membrane at different magnifications; (**g**,**h**) the TiO2–PVDF membrane in different magnifications; (**i**–**p**) the FTCS–PVDF membrane in different magnifications.

In a study carried out by Razmjou et al. [29], the increase of the contact angle in the stage of coating FTCS on PVDF was reported as 146◦ ± 5 ◦ (the contact angle for the virgin membrane was 125◦ ± 1 ◦ ) while in the current work which was done by the self-assembly method, the contact angle was 142◦ ± 20◦ (the contact angle for the virgin membrane was 89◦ ± 8 ◦ ). The contact angles formed by water droplet on the membranes surface were measured by the sessile drop technique (CA-500A instrument, Yasin Pajooh Co. Ltd., Isfahan, Iran).

## *3.2. Investigating the Durability and Photocatalytic Activity of TiO<sup>2</sup> Nanoparticles Coated on the Surface*

The KI test was employed to ascertain the durability of TiO<sup>2</sup> nanoparticles on the surface of the PVDF membrane (Figure 6). If the coated membrane has a photocatalytic activity, it can oxidize ion (I−) according to Equation (1) and convert it to (I2) and change the color of the colorless solution KI to yellow color.

$$\begin{aligned} \mathrm{I}^- + \mathrm{h}^+ &\to \mathbb{V} \mathrm{I}\_2 \\ \mathrm{I}\_2 + \mathrm{I}^- &\to \mathrm{I}\_3^- \end{aligned} \tag{1}$$

**Figure 6.** Investigating the durability of TiO<sup>2</sup> nanoparticles on the surface of the PVDF membrane using the KI test: (**a**) preparing the KI solution; (**b**) the change in the color of the colorless solution by entering the TiO2–PVDF membrane; (**c**) the UV-Vis spectrophotometer to show the solution wavelengths.

The color change qualitatively indicates the photocatalytic activity of the sample coated by TiO<sup>2</sup> nanoparticles. The UV-Vis absorption spectroscopy device was used for the quantitative evaluation of the photocatalytic properties of the sample. The UV-Vis spectrophotometer showed the highest amount of absorption at the wavelengths of 288 and 351 nm which represent the presence of I<sup>2</sup> and I − 3 atoms, respectively. To check the durability of TiO<sup>2</sup> nanoparticles, the KI test may be done under UV radiation in different periods. In the solution containing the unmodified membrane, no change of color and no absorption occurred after 6 h of UV radiation, while in the solution containing the membrane coated with TiO<sup>2</sup> nanoparticles both color change to yellow and absorption peak were observed. As is shown in Figure 7 absorption in each period for both samples remained relatively unchanged. An unchanged absorption spectrum during the period not only reflects its excellent photocatalytic activity, but is also proof of the suitable durability of TiO<sup>2</sup> coatings.

**Figure 7.** The absorption diagram of the KI solution containing the TiO2–PVDF membrane at the wavelengths of 351 and 288 nm after it was exposed to UV for 6 h and for 6 periods.

#### *3.3. Membrane Hydrophobicity*

As was mentioned above, in this research, the hydrophobic properties of the membrane were optimized in two stages: first, changing the topography of the surface to enhance roughness by coating TiO<sup>2</sup> nanoparticles on the surface of the membrane; second, changing the chemical properties of the surface to repel water more using the functional agent FTCS. Then the modified membrane was used in a DCMD membrane module at the experimental scale.

Although the coating of TiO<sup>2</sup> nanoparticles on the surface of the PVDF membrane can reduce its hydrophobic properties due to the hydrophilic properties of TiO2, in this study, the purpose of using titanium oxide nanoparticles was to create roughness and the hydrophilic modification of the membrane was not the aim. A multi-layer roughness can change the wettability properties of a membrane [34,35]. An increase in surface roughness (after the surface is coated with TiO<sup>2</sup> nanoparticles and a multi-layer roughness is formed on it) can lead to a strong capillary water suction on the membrane surface which expands the droplet on the surface until saturation [36]. Therefore, the hydrophilic properties of the surface increase.

Then the PVDF membrane was coated by FTCS and a substantial reduction was observed in the hysteresis angle from 18◦ for the unmodified membrane to about 6◦ for the FTCS–PVDF membrane. The increase in the water contact angle for the FTCS–PVDF membrane surface may be because of the notable reduction of the free energy of the membrane surface and its increased roughness.

Hydrophobic surfaces repel water and an air gap is created between the membrane and water droplets. This gap decreases the wettability of the membrane surface. In this case, water droplets can only have contact with the tip of the roughness; therefore, the physical contact between water droplets and the surface is considerably reduced. This is due to the fact that an air gap is created among the roughness bumps and surface tensile forces. Water droplets on superhydrophobic surfaces have small hysteresis contact angles and lie on the membrane surface spherically with a good approximation. Thus, they can roll even on surfaces with low slopes and remove pollutants from the surface (called the 'self-cleaning' property).

To increase the hydrophobic properties of the membrane, the TiO2–PVDF membrane was coated by FTCS. The unmodified PVDF membrane had the contact angle of 89◦ ± 8 ◦ degrees with water. It is worth noting that the surface modification (FTCS–TiO2–PVDF) created a contact angle of about 174◦ ± 10◦ . The unmodified and TiO2–PVDF membranes lacked self-cleaning properties on steep slopes, while the FTCS–TiO2–PVDF membrane showed self-cleaning properties even at low slopes.

#### *3.4. Membrane Distillation at the Experimental Scale*

Tests with a 5 wt % potassium nitrate solution were conducted for the unmodified and modified membranes. When the unmodified membrane was used, the cross-membrane water flux was 2 × 10−<sup>3</sup> Kg/m<sup>2</sup> s and when the FTCS–TiO2–PVDF membrane was employed, the cross-membrane water flux was 6.4 × 10−<sup>4</sup> Kg/m<sup>2</sup> s. When each of the membranes was used, the electrical conductivity of the permeate was zero. In general, flux reduction was observed in both cases over time which could have two reasons. The first reason may be the blockage or fouling of the membrane pores and the reduced surface area available for the passage of water vapor; the second reason could be the decrease of the driving force (i.e., the reduction of temperature difference between the feed and permeate) [37]. To remove the fouling of the pores, each membrane was placed for 15 min in a NaOH 0.2 wt % solution with the pH of 12 at room temperature (25 ◦C). After being washed and reused, the modified membrane had the recovered flux of up to 95% and the unmodified membrane had the recovered flux of about 60%. After the membranes were washed, colors and sediments were removed from the modified membrane, while in the unmodified membrane, the color change was not resolved requiring more time for washing and sometimes backwashing. This phenomenon can show the anti-sedimentation property of the modified membrane. In the current research, the coat is easy to produce due to the low production temperature and the easy methods selected for coating. Hence, in case of damage to the coat of the membrane during washing, the coat can be easily restored.

#### *3.5. Physical and Chemical Characteristics of the Permeate*

As can be observed in Figure 8a, the electrical conductivity of the feed had an ascending trend and increased from 200 to 244 µs during a 6 h continuous test. Nevertheless, the electrical conductivity of the permeate remained constant throughout the test and was equal to that of distilled water. This result indicates that no ion was able to penetrate to the other side of the membrane and the permeate was pure.

**Figure 8.** (**a**) The electrical conductivity-time diagram of the membrane distillation process using the PVDF membrane; (**b**) Mass-time diagram of the membrane distillation process using the PVDF membrane; (**c**) Electrical conductivity-time diagram of the membrane distillation process using the FTCS–TiO2–PVDF membrane; (**d**) Mass-time diagram of the membrane distillation process using the FTCS–TiO2–PVDF membrane.

Moreover, the mass of the feed had a descending trend and the mass of the permeate had an ascending trend indicating the transfer of mass through the membrane from the feed side to the permeate side Figure 8b.

In what follows, the above experiment was conducted using the FTCS–TiO2–PVDF membrane. As can be seen in Figure 8c, during 6 h of continuous testing, the electrical conductivity of the feed had an ascending trend and increased from 200 to 208.6 µs, while the electrical conductivity of the permeate remained constant during the experiment and was equal to that of distilled water.

Moreover, the mass of the feed had a descending trend and the mass of the permeate had an ascending trend indicating the transfer of mass through the membrane from the feed side to the permeate side Figure 8d.

Comparing Figure 8a,c shows that the slope of the feed of the PVDF membrane is higher than that of the FTCS–TiO2–PVDF membrane which may be due to the higher flux in the PVDF membrane. Interestingly, the slope of the permeate is almost constant in Figure 8b,d, indicating the purity of the permeate.

The interesting point about the performance of the unmodified and modified membranes in the two tests mentioned above is the difference between the increased electrical conductivity of the feed in them over 6 h. According to the mass-time diagrams, this difference could be due to the lower mass reduction of the modified membrane Figure 8b than that of the unmodified membrane Figure 8d. There are detailed explanations regarding mass transfer reduction. As can be observed in Figure 8d, the flux is not as expected which is a disadvantage of this system and needs to be improved. Future studies may focus on solving this problem.
