The effect of process temperature on permeate flux is best described by Antoine equation, where α, β, and γ are constants relating to the specific substance and P_i is the vapour pressure (Pa) and T is the temperature (K). P i T = e α − β γ + T

According to Antoine equation, vapour pressure exponentially increases with temperature (Choudhury et al., 2019). Furthermore, water flux is directly proportional to feed temperature (Banat and Simandl, 1998). However, flux increments are limited by the process temperature and declines once an optimum has been attained (Banat and Simandl, 1998). Moreover, Attia et al. (2017) evaluated the effect of temperature using synthetic electrospun PVDF, superhydrophobic alumina, and commercial PVDF membranes in a comparative AGMD process. A direct relationship between permeate flux and feed temperature was established (Attia et al., 2017). Liu et al. (2022) assessed the effect of temperature and flow velocity to obtain lithium chloride from air-conditioning systems via DCMD. Reportedly, an increase in feed temperature improved solute generation although the membrane’s hydrophobicity was altered. However, the increase in solute concentration reduced the water flux due to a decreased partial vapour pressure (Liu et al., 2022). Although high water fluxes are obtained at higher temperatures, the water recovery factor is reduced due to salt precipitation (Zhu et al., 2021). The effect of feed temperature on process operation is summarized in Table 2.

The capability of the MD technique to progressively concentrate a feed solution to supersaturation gave rise to MDC (Yadav et al., 2022). The gradual passage of water vapor from the feed stream to the distillate results in the eventual concentration of the feed solution to its critical saturation. Further increase in feed supersaturation enables the recovery of crystal salts from the crystallization reactor (Das et al., 2021). Importantly, this process facilitates the recovery of higher quality mineral crystals in terms of size and purity. Other benefits include controlled rate of supersaturation and nucleation (Yadav et al., 2022). However, the increase in feed concentration towards solution supersaturation induces temperature and concentration polarization, thus reducing the permeate flux (Martínez, 2004). Moreover, pore blockage occurs due to the formation of crystals on the surface of membrane (Yadav et al., 2022). Martínez (2004) investigated the effect of feed concentration on the permeate flux using a flat sheet PTFE membrane and feed solutions of pure water, sodium chloride, and sucrose. Notably, the pure water flux remained stable towards supersaturation. However, the deposition of sodium chloride and sucrose crystal on the membrane surface resulted in a decrease in the permeate flux (Martínez, 2004). When supersaturation is attained in the bulk feed solution, nucleation is then induced which is succeeded by crystallization (Yadav et al., 2022). Moreover, a higher feed temperature increases rate of solvent evaporation, thus facilitating an increased rate of supersaturation compared to that experienced at low feed temperatures (Edwie and Chung, 2013).

The formation and crystal growth are influenced by the solubility of the salt, rate of water recovery and process temperature. For instance, feed solutions with low concentration containing extremely soluble solutes requires a lengthy period to form crystals (Rudolph, 2010; Liu et al., 2021). Additionally, slow crystal growth rate facilitates formation of large crystals. Therefore, longer crystallization periods give rise to larger crystals (Alvarez et al., 2020). In their study, Wagstaff et al. (1964) evaluated the impact of crystallization duration to the size of cristobalite. Based on their findings, the size of the crystals increased quadratically upon increase in duration of process crystallization (Wagstaff et al., 1964). Essentially, the rate of crystal growth is governed by the various factors including flow of latent heat from the growing crystal, diffusion and reactions occurring at the crystal interface (Rudolph, 2010). In MDC processes, the inclusion of the membrane provides a site for heterogeneous nucleation. The Gibbs free energy is lower at the membrane-solution interface, thus favoring heterogeneous nucleation rather than homogenous nucleation (Ruiz Salmón and Luis, 2018). According to Edwie and Chung (2013), a high feed temperature encourages a higher rate of evaporation resulting in a lower average crystal size. Once nucleation has been established, the nuclei begin to grow until the critical cluster size has been achieved. Thereafter, crystals form and grow in saturation zones (i.e., metastable and unstable growth zones) (Yadav et al., 2022). Technically, rate of supersaturation and nucleation affect crystal network growth, consequently the duration of crystallization (Das et al., 2021).

High recovery rates of MDC processes are realized at higher recirculation rates (Swaminathan and Lienhard, 2018). For an efficient and high performing MDC process, the overall recovery factor should be greater than that of a single pass process (Lokare et al., 2018). To achieve high recovery factors, the retentate is mixed with the new feed solution prior to crystallization (Lokare et al., 2018). In addition to high recoveries, an increase in the recirculation rate enhances the heat transfer coefficient. Consequently, this minimizes the boundary layer thus improving the permeate flux (Srisurichan et al., 2006). Due to the improvement of water turbulence, a high recirculation rate reduces temperature polarization and membrane fouling, thus ensuring the stable water flux (Lokare et al., 2018).

Flux decline caused by membrane fouling requires frequent membrane cleaning and possibly replacement, consequently increasing operating and maintenance costs (OPEX). Therefore, wastewater pre-treatment integrated to MDC improves process performance. Primarily, pre-treatment strategies limit fouling by reducing foulants concentration in the feed water. The choice of pre-treatment depends on the feed water. Typically, a combination of pre-treatment strategies is required to improve efficiency of foulant removal from the feed solution. These combinations involve physical and chemical processes such as low-pressure membrane filtration, coagulation and flocculation, adsorption, pH adjustments and the addition of anti-scalants.

Mechanical pre-treatments consist of membrane processes such as microfiltration (MF), ultrafiltration (UF), and nanofiltration (NF). Particularly, NF is used for water softening and reduction of natural organic matter (NOM). The UF and MF reduces colloidal, suspended and biological matter (Alkhatib et al., 2021). These pre-treatment methods have been evaluated in water processing of various complexities (Nthunya et al., 2021). El-Abbassi et al. (2013) studied coagulation-flocculation and MF pre-treatment of olive mill wastewater in DCMD. Coagulation-flocculation pre-treatment reduced the concentration of TDS and phenolic compounds by 23% and 18%, respectively. The TDS removal was improved to 30% while that of phenolic compounds was reduced to 4.8% upon the MF treatment (El-Abbassi et al., 2013). In another study, Karakulski and Gryta (2005) investigated NF pre-treatment of tap water for use in MD. Reportedly, untreated feed water caused membrane scaling leading to rapid flux decay. However, NF pre-treatment removed scalants thus ensuring high process performance (Karakulski and Gryta, 2005). Additionally, adsorption has been proven to effectively remove organic matter prior to MD water purification. Nthunya et al. (2019c) reported removal of phenolic compounds from feed wastewater using a candle filter (pore size ∼100 µm) equipped with polyethyleneimine-functionalized polyacrylonitrile nanofibre membranes. The membranes presented 39.9 mg g⁻¹ adsorption capacity (Nthunya et al., 2019b). Notably, MD process performance remained relatively stable upon feeding with pre-treated wastewater. Coagulation-flocculation is another process proven to effectively remove foulants prior to MD water processing. In this process, foulant particles are converted into larger flocs, thus reducing their adhesive interaction with the membranes. Moreover, coagulation-flocculation coupled with conventional treatment or membrane filtration processes remove the flocs from the feed water (Alkhatib et al., 2021). Li et al. (2016) investigated the purification of biologically treated coking wastewater using MD coupled with coagulation pre-treatment. A poly-aluminium chloride (PACl) flocculant reduced the foulants thus promoting the stable performance in MD (Li et al., 2016). Lastly, pH-adjustments have been extensively used to treat feed solutions in membrane processes. The increase in feed pH promotes formation of metal precipitates which are removed as insoluble metal hydroxides prior to MDC. Similarly, the feed solution is acidified to dissolve the foulant, thus impeding their interaction with the membranes (Karakulski and Gryta, 2005). A summary of MDC pre-treatment processes is presented in Figure 3.

Anti-scalants are precipitation-inhibiting chemicals impeding nucleation or crystal growth of scalants on membrane surfaces. Anti-scalants adsorb on the nuclei surface to obstruct the rate of crystal growth and agglomeration (Lin and Singer, 2005; Gloede and Melin, 2008; Abdel-Karim et al., 2021). The anti-scaling mechanism of action takes place through ligand exchange or electrostatic interactions (Horseman et al., 2021). Commonly used anti-scalants include organophosphates, polyelectrolytes and polyphosphates (Ketrane et al., 2009). Yin et al. (2021) evaluated gypsum anti-scaling in reverse osmosis (RO) coupled with MD using Poly (acrylic) acid (PAA). A 1,300 min test recorded 95% water flux decay in the absence of an antiscalant. However, the decay was reduced by 30% upon addition of anti-scalant, thus corresponding to 40% water recovery (Yin et al., 2021). Lin and Singer (2005) utilized polyphosphates to minimize calcite crystal growth in MD. The process performance remained stable with minimal flux decay recorded. Though anti-scalants improve MDC processes, their addition beyond maximum threshold promote membrane biofouling (Tijing et al., 2015). Therefore, the anti-scalant dosage should be optimized to meet the process requirement upon treatment of a specific feed solution.

Membrane flushing and gas bubbling are classified as physical fouling mitigation strategies. Flushing is often carried out to remove adsorbed solutes from the membrane surface using deionized water. Nonetheless, flushing fails to remove solutes within the membrane pores (Alkhatib et al., 2021). Flushing is often operated in two modes namely, forward and backwashing. Technically, deionized water is pumped in a forward direction during forward flushing while the reverse is true for backflushing (Alkhatib et al., 2021). Gas bubbling enhances shear rate and fluid dynamics thus reducing temperature and concentration polarization (Alkhatib et al., 2021). Reportedly, finely dispersed bubbles are more efficient compared to course bubbles (Lu et al., 2008). Choi et al. (2020) assessed the recovery of sodium sulfate from seawater brine using a hollow fiber PVDF membrane in fractionally submerged MD crystallization. Two cleaning procedures were used, namely air backwashing and deionized water flushing in the presence of ammonium sulfate. Air backwashing enabled 90% flux recovery. Similarly, flushing recovered 82% water flux from the original level. However, multiple air backwashing caused progressive permeate flux decline (Choi et al., 2020). To reduce scaling of a commercial PTFE membrane supported on polypropylene (PP), Nghiem and Cath (2011) used MilliQ water. Five cycles of membrane flushing recovered 30% of the original flux (Nghiem and Cath, 2011). Though flushing is more efficient for removal of inorganic foulants, it can also be used for removal of organic foulants upon treatment of an oil-contaminated feed (Gryta, 2020).

Temperature and flow reversal (backflow) techniques are novel methods used to mitigate fouling in MD/MDC. This experimental procedure was evaluated by Hickenbottom and Cath (2014) to minimize scaling while ensuring stable process performance. The temperature swap between the feed and distillate effectively reversed the driving force across the membrane, thus reducing the surface interactions between the membrane and scalants. Water flux and rejection efficiencies were recovered to 95%. Remarkably, both methods minimized scaling, thus ensuring stable fluxes and maintaining high salt rejection (Hickenbottom and Cath, 2014). Notably, these mitigation strategies avoid the use of expensive and toxic chemicals. Therefore, temperature and flow reversal are attractive alternative measures to control membrane fouling. However, extensive research is required to ascertain their sustainability at an industrial scale.

Chemical cleaning is the most evaluated reactive measure used to control membrane fouling. The mechanism of action involves breaking foulant-membrane interactions (Alkhatib et al., 2021). Chemical reagents include acids, bases, surfactants, chelating agents, enzymes, and oxidants (Al-Amoudi and Lovitt, 2007; Porcelli and Judd, 2010). Typically, bases and surfactants are used to address organic and biofouling (Alkhatib et al., 2021; Char et al., 2021) while acids and chelating agents are true for inorganic fouling (Alkhatib et al., 2021; Gryta, 2021). To ensure a synergistic cleaning process, a combination of chemicals is generally used during the treatment of complex feed solutions characterized by various foulants (Alkhatib et al., 2021). Charfi et al. (2021) optimized cleaning procedures of the MD process during treatment of anaerobic digestate. Reportedly, deionized water flushing was followed by 0.2% NaOCl and 3% citric acid for 60 min. NaOCl and citric acid were effective for organic and inorganic foulant removal, respectively thus ensuring 75.5% flux recovery. Furthermore, the cleaning process recovered 87% of membrane hydrophobicity with minimal membrane wetting (Char et al., 2021). In another study Guillen-Burrieza et al. (2014) evaluated a variety of cleaning agents in long-term scaling control in MD processes. As per reported findings, a combination of 0.1 wt% oxalic acid and 0.8 wt% citric acid recovered 97% of the membrane WCA. Furthermore, formic, and sulfuric acid recovered 96.7% and 94.6% of the membrane WCA respectively. Although these processes restored WCA, the integrity and mechanical strength of the membranes were affected (Guillen-Burrieza et al., 2014). The destruction of membrane integrity depends on cleaning conditions including the concentration of reagents, duration, and cleaning frequency. To understand the impact of chemical cleaning pertaining to physicochemical properties, various characterization techniques should be employed. These include chemical, morphological, topological, hydrophobic/hydrophilic, and mechanical analysis of the membrane. Puspitasari et al. (2010) investigated the cleaning and ageing of PVDF membranes using oxidative sodium hypochlorite (NaOCl). The effect of chemical concentration on the cleaning and ageing of the membrane is presented in Figure 4A. The cleaning efficiency improved with an increase in NaOCl. The same trend was observed for cyclical membrane cleaning. However, following the cleaning protocol, SEM micrographs showed presence of foulants on the membrane surface. Moreover, FTIR results presented changes in the chemical functional groups of membrane, thus alluding to an ageing effect. Furthermore, higher concentrations NaOCl damaged the integrity of the membrane (Figure 4B) (Puspitasari et al., 2010). To minimize the damage, a combination of cleaning reagents and anti-scalants is commonly used. This was evaluated by Peng et al. (2015) during the MD treatment of RO concentrated brine. A series of chemicals namely, NaCl, NaOH, KCOOH, citric acid, and EDTA-4Na were used. While operating at elevated temperatures, EDTA-4Na enabled highest flux recovery. Improved recovery was associated to chelation of calcium ions, thus reducing their interactions with the membranes (Peng et al., 2015). In another study, Zhang J et al. (2021) used a combination of organic phosphoric acid and hexamethylene diamine tetra (methylene phosphonic acid) (HDTMPA) during treatment of landfill leachate in FO/MD system. A combination of these chemicals reduced the foulant-membrane interactions, thus improving the process performance. Although 90% of flux was recovered in the first cycle, continuous cleaning did not show significant increase in performance recovery (Zhang P et al., 2021).

Some foulants bind strongly on membrane surfaces, thus causing irreversible fouling. This phenomenon was reported by Naidu et al. (2015) upon NaOH cleaning MD membranes fouled by humic substances. Partial regeneration of the membrane with 19% hydrophobicity recovery was reported (Naidu et al., 2015). Further improvements in chemical cleaning involves the use of 3D spacers. Spacers amplify flow turbulence, thus reducing foulant-membrane interaction. In their study, Castillo et al. (2019) investigated a step-wise cleaning of MD membrane using citric acid and water in the presence of spacers. Upon cleaning, 87% of membrane WCA was recovered (Castillo et al., 2019). The effect of various cleaning strategies is presented in Table 4.

Membrane modification improves resistance to fouling and wetting. Typically, modification is achieved through systematic manipulation. Currently, superhydrophobic membranes characterized by self-cleaning properties are explored with low success rate. To improve membrane resistance to fouling while retaining high salt rejection, omniphobic and Janus membranes are also reported (Wang and Lin, 2017; Yao et al., 2020; Tjale et al., 2022). These membranes are characterized by asymmetric wettability to minimize fouling while retaining process stability (Afsari et al., 2021). Xiao et al. (2020) prepared omniphobic membranes through incorporation of silica nanoparticles (SiNPs)-coated micropillars (MP) to PVDF. Reportedly, SiNPs-MP-PVDF membrane reduced scaling and fouling, thus maintaining the process performance over a longer period. Figure 5 1) and 2) present the role of membrane modification towards preventing flux decay.

In another study, Toh et al. (2019) modified PVDF-co-hexafluropropylene membranes using silica nanoparticles to improve their resistance to wetting and fouling. The modified membranes were characterized by high WCA and low surface energy (Toh et al., 2019). Zhang J et al. (2021) reported hydrophilic surface modification of PVDF hollow fibre membrane through co-deposition of polydopamine (PDA) and poly (MPC-co-2-aminoethyl methacrylate hydrochloride) (MPC-co-AEMA). The smooth hydrophilic thin layer reduced the foulant-membrane interaction (Zhang P et al., 2021). In addition to hydrophilic coating, antimicrobial nanoparticles are embedded on hydrophobic membranes to combat organic, inorganic and biofouling. These additives include silver nanoparticles, cellulose nanocrystals and carbon nanotubes (Nthunya et al., 2019a; Nthunya et al., 2020). Membrane modifications processes addressing fouling are presented in Table 5.