Abstract

Polyvinylidene fluoride (PVDF) membrane was improved using TiO2 nanoparticles and nanocellulose for membrane distillation crystallization in this work. Besides the addition of TiO2 nanoparticles and nanocellulose, PVDF membrane was post-modified with octadecyltrichlorosilane after phase inversion using a dual coagulation bath. The addition of hydrophilic TiO2 nanoparticles and nanocellulose reduced membrane hydrophobicity, but the dispersed TiO2 nanoparticles assisted silane modification to improve surface hydrophobicity. Besides reducing the agglomeration of TiO2 nanoparticles, nanocellulose induced the formation of larger pore size and higher porosity as proven in SEM images and gravimetric measurement, respectively. The abundant moieties of nanocellulose accelerated the exchange between solvent and non-solvent during phase inversion for the formation of large pore size and porosity, but membrane thickness increased due to the thickening effects. The modified membrane showed higher water permeate flux in membrane distillation with salt rejection greater than 97%. Severe fouling in membrane distillation crystallization was not observed.

HIGHLIGHTS

  • TiO2 nanoparticles and nanocellulose were incorporated into PVDF membrane.

  • Nanocellulose induced the formation of larger pore size and higher porosity.

  • Silane modification enhanced membrane hydrophobicity.

  • Permeate flux increased in membrane distillation.

  • Insignificant fouling in membrane distillation crystallization was observed.

Graphical Abstract

Graphical Abstract
Graphical Abstract

INTRODUCTION

Brine is the major waste generated from desalination processes. The salinity of brine depends on the recovery in the desalination process. At 40% recovery of water, the salinity of brine is about 1.66 times higher than sea water. A great amount of brine is also produced at the same time, approximately 1.5 L of brine per litre of potable water (Lee et al. 2019). Brine management is important because brine is not only corrosive, it also contains toxic chemicals such as anti-foulants, anti-scalants and surfactants added during desalination. Surface discharge, deep well injection and solar evaporation are the common practices in brine management. However, the discharged brine can disrupt the sea ecosystem due to a drastic change of salinity that affects aquatic organisms. Deep well injection preserves surface water, but usually increases the contamination risk for underground water sources. Without proper lining, the brine in evaporation ponds may penetrate the water aquifer beneath the pond and affect the water aquifer as well. Instead of brine discharge, simultaneous recovery of water and salt from brine could be implemented using membrane distillation crystallization (MDC) for improving the sustainability of water supply.

In recent years, MDC has been extensively studied to recover water and salts from brine. Hydrophobic and macroporous membrane works as the barrier of two phases in MDC, allowing only the water vapor to be removed from the hot brine and transported through the membrane pores into the other phase at lower temperature or pressure. The partial pressure difference between the hot feed and permeate serves as the driving force for this separation process. Besides water recovery, the brine can be further concentrated in MDC until reaching crystallization. MDC involves more than water vapor removal; further increment of concentration beyond the salt solubility can initiate nucleation, crystal growth and precipitation. MDC requires less energy than the conventional evaporative crystallizers since MDC can be conducted at lower temperature (Ruiz Salmón & Luis 2018). Solar heat or waste heat can be used to increase the temperature of brine before it enters an MDC system (Koschikowski et al. 2003; Shahu & Thombre 2020). In addition, the highly porous structure of the membrane provides a high interfacial surface for optimum water vaporization, increasing the overall water recovery and collection of valuable salts.

Researchers have proposed different types of MDC systems which can be mainly categorized into ex situ crystallization systems or in situ crystallization systems. Zou et al. (2019) reported on vacuum MDC with submerged polytetrafluoroethylene (PTFE) hollow fibers. Under stirring or aeration, the crystallization could be changed from heterogeneous crystallization on the membrane surface into bulk crystallization in the feed solution with higher critical volume concentration factor and salt recovery. Ko et al. (2018) modified a tubular alumina support using hydrophobic polymethylsilsesquioxane aerogel, while the alumina hollow fibers were modified using fluoroalkylsilanes. Both hydrophobic membranes with water contact angle (WCA) of about 138° achieved stable permeate flux without wetting during crystallization, but the modified alumina hollow fibers exhibited higher water permeate flux. Comparing Hyflo/polyvinylidene fluoride (PVDF) membranes with varied pore size (0.47–0.80 μm) but similar WCA (132°–137°), Cui et al. (2018) found that the membrane with larger pore size showed a reduction of nucleation time and increment of crystal growth rate. A high permeate flux is preferable to promote supersaturation degree, supersaturation rate, nucleation rate and secondary nucleation. MDC was successfully applied to recover fertilizer (Quist-Jensen et al. 2018). Struvite precipitate and NH4+-rich solution was recovered from sewage using a membrane distillation (MD) system consisting of a polypropylene hollow fiber membrane. Besides water recovery, calcite and halite could be also recovered from shale-gas-produced water by MDC. However, Kim et al. (2018) commented that inorganic scaling on PP membrane should be controlled by reducing the crystallization temperature.

To elevate the efficiency of MDC, several new strategies of membrane improvement, namely pore enlargement and hydrophobicity enhancement using sustainable and low-cost filler, should be proposed. Recently, nanocellulose has been extensively used to improve membrane properties for oil and water separation (Kollarigowda et al. 2017), gas separation (Dai et al. 2019) and water purification (Karim et al. 2016). Leitch et al. (2016) developed a superhydrophobic aerogel membrane from bacterial nanocellulose through supercritical drying and chemical vapor deposition. The superhydrophobic aerogel membrane with high porosity and low bulk thermal conductivity achieved higher water permeability and a lower temperature polarization coefficient than PVDF membrane in MD. Kollarigowda et al. (2017) functionalized hydrophilic cellulose with block-co-polymers of silane and myrcene monomers via reversible-fragmentation chain transfer (RAFT). A water contact angle greater than 160° was achieved and water was effectively removed from the oil phase in filtration. Cheng et al. (2018) coated cellulose nanocrystal onto cotton fabric to enhance surface roughness before silane modification for hydrophobicity improvement. The superhydrophobic cotton fabric could separate oil from water with an efficiency higher than 98%. A self-cleaning membrane with underwater superhydrophobicity could be easily made by filtering nanocellulose decorated with TiO2 for oil–water separation (Zhan et al. 2018). The hydrolysis of titanium oxysulfate induced the in situ formation of TiO2 nanoparticles on the surface nanocellulose. Without chemical modification to reduce surface energy, the resultant membrane was endowed with underwater superoleophobocity for achieving high water flux.

In this work, near-superhydrophobic membranes were developed for the effective recovery of salt and water via MDC. Microfibrillated cellulose was used to improve the morphology of PVDF membrane incorporated with TiO2 nanoparticles. The membrane was further post-modified with non-fluorinated silane after phase inversion in a dual coagulation bath. In our previous work (Hamzah & Leo 2016), the silanated PVDF/TiO2 membrane exhibited non-wetting, anti-fouling and self-cleaning properties in MD, but the water recovery was limited by the small pores and particle agglomeration. As reported by others (Zhang et al. 2018b), microfibrillated cellulose with abundant hydroxyl groups could improve the demixing process during phase inversion, provide more active sites for silanation and assist the dispersion of TiO2 nanoparticles. It is important to further understand the effects of nanocellulose on membrane wetting. The proper adjustment of membrane propeties such as pore size and wetting is important to comprehend MDC.

MATERIALS AND METHODS

Materials

PVDF (Solef® 6010 PVDF) from Solvay Solexis (France) was dried at 100 °C before the preparation of the membrane dope solution. The polymer solvent, N-methyl-2-pyrrolidone (NMP) (>99.5%), was supplied by Merck (Darmstadt, Germany) while the inorganic filler, TiO2 nanoparticles (21 nm primary particle size, >99.5% trace metal basis), were supplied by Sigma-Aldrich (St Louis, MO, USA). The other additives of the dope solution included ortho-phosphoric acid (H3PO4) (>85%), lithium chloride (LiCl) and acetone acquired from Merck (Darmstadt, Germany), but the nanocellulose, microfibrillated cellulose (Exilva F01-V, 10%w/w) with a length up to 100 μm and diameter at 10–100 nm, was sponsored by Borregaard Cellulose Fibrils. The coagulation bath and silanation used ethanol acquired from Merck (>99.9%, Darmstadt, Germany). Octadecyltrichlorosilane (Gelest Inc., Morrisville, PA, USA) was utilized as the hydrophobic agent in the post-modification of membranes. For MD and MDC, sodium chloride (NaCl, Sigma-Aldrich, St Louis, MO, USA) was used to prepare the saline feed.

Synthesis and modification of membrane

Membranes were synthesized according to our previous work (Hamzah & Leo 2016). The dried PVDF (15 wt%) was dissolved into NMP solvent (74 wt%) containing non-solvent additives such as acetone (5 wt%) and H3PO4 (3 wt%). For the preparation of PVDF-T membrane, TiO2 nanoparticles (3 wt%) were dispersed in the solvent before blending with polymer and non-solvent additives. On the other hand, nanocellulose was dispersed in the solvent containing LiCl at 165 °C before being blended into the dope solution of PVDF-TN membranes. To achieve homogeneity, the dope solution was stirred for 24 h at 50 °C and degassed 24 h before casting. The solution was cast on a glass plate covered with woven support at a casting gap of 400 μm (Elcometer 4340 automatic). The wet film was immersed into a coagulation bath containing pure ethanol for 20 min. The supported membrane was removed and immersed into a second coagulation bath containing distilled water for 24 h. The membrane was dried at 40 °C in the oven for 72 h before silane modification.

The membranes were modified using silane in ethanol at a volumetric ratio of 1 mL silane to 50 mL ethanol. The mixture was stirred for 30 min and dried membrane was immersed in the silane solution for 5 min. The modified membranes were rinsed using ethanol and they were dried in the oven before further characterization and separation testing using the MD system. Table 1 summarizes the dope composition of all the membranes prepared in this work. Additional samples of the membrane outperformed in MD were modified using silane solution with higher concentration at 3 mL silane in 50 mL of ethanol. The membrane was characterized and tested in MDC.

Table 1

The composition of membrane dope solution

MembranePVDF (wt%)NMP (wt%)H3PO4 (wt%)LiCl (wt%)Acetone (wt%)TiO2 (wt%)Nanocellulose (wt%)
PVDF 13 77 – – 
PVDF1a 13 77 – – 
PVDF-T 13 74 – 
PVDF-T1a 13 74 – 
PVDF-TN 13 74 2.8 0.2 
PVDF-TN1a 13 74 2.8 0.2 
PVDF-TN3b 13 74 2.8 0.2 
MembranePVDF (wt%)NMP (wt%)H3PO4 (wt%)LiCl (wt%)Acetone (wt%)TiO2 (wt%)Nanocellulose (wt%)
PVDF 13 77 – – 
PVDF1a 13 77 – – 
PVDF-T 13 74 – 
PVDF-T1a 13 74 – 
PVDF-TN 13 74 2.8 0.2 
PVDF-TN1a 13 74 2.8 0.2 
PVDF-TN3b 13 74 2.8 0.2 

aThe membrane was modified using silane in ethanol at a volumetric ratio of 1 mL: 50 mL.

bThe membrane was modified using silane in ethanol at a volumetric ratio of 3 mL: 50 mL.

Membrane characterization

The water contact angle on the membrane surface was measured using a goniometer (Ramé-Hart Instruments Co., USA) in order to study the changes of membrane hydrophobicity. Deionized (DI) water was dropped through a syringe onto the membrane surface at room temperature. Replicate measurement was conducted at five different positions on the membrane surface. The surface morphology and cross-section of membranes where studied using scanning electron microscopy (SEM) (HITACHI S-3000N, Hitachi Ltd, Japan). The membrane thickness was measured from SEM images. The mean pore size of the membranes was measured using a porometer (Porolux 1000, IB-FT GmbH, Germany) while the membrane porosity was determined using a gravimetric method. The dried membrane sample was immersed into Porefil solution for one hour. The membrane was weighed before and after wetting to determine the volume of entrapped Porefil solution for the calculation of membrane porosity: 
formula
(1)
where wwet is the weight of the wetted sample (g), wdry is the weight of the dry (g), ρp is the density of Porefil (1.8695 g·cm−3), As is the surface area of the membrane sample in cm2 and l is the thickness of the membrane sample in cm. The chemical properties of the unmodified and modified membranes were studied using Fourier transform infrared spectroscopy (FTIR) (Nicolet iS10, Thermo Scientific, USA), recording the spectra from 600 cm−1 to 3,800 cm−1.

Membrane distillation (MD) and membrane distillation crystallization (MDC)

The performance of the membranes synthesized in this work was evaluated using a direct contact membrane distillation (DCMD) system as described in our previous work (Hamzah & Leo 2016). The membrane sample was placed in a membrane module to isolate the hot feed from cold permeate. The hot feed stream was NaCl aqueous solution with a concentration of 35 g/L at 60 °C and the cold permeate stream was distilled water at 20 °C. Both streams were counter-currently circulated into the separation system at 300 mL/min using two peristaltic pumps. The salt concentrations of the feed and permeate solutions were measured with a portable conductivity meter (TDS-EC-Tester) for the determination of salt rejection. The permeate flux, J (kg/m2·h) through the membrane during MD was calculated by the following equation: 
formula
(2)
where ΔW is the difference of distillation water mass (kg), A is the effective area of flat-sheet membrane (m2) and Δt is the sampling time (h). The rejection coefficient, R (%) was calculated as follows: 
formula
(3)
where Cf is the concentration of the feed (g/L) and Cp is the concentration of permeate (g/L).

In MDC, all the operating conditions were unchanged except a feed concentration of 100 g/L NaCl was used.

RESULTS AND DISCUSSION

Membrane characteristics

Water contact angle indicates the surface hydrophobicity, which is important in the study of membrane wetting during MDC. As shown in Figure 1, all the PVDF membranes without silane modification showed satisfactory hydrophobicity with a WCA more than 90°. The hydrophobicity was created with different roughness creation strategies. The surface roughness of neat PVDF membrane was enhanced using ethanol as the soft-coagulation bath, which promotes the formation of spongy structure (Hamzah & Leo 2017). The surface roughness of PVDF-T and PVDF-TN membranes were further enhanced by adding TiO2 nanoparticles and nanocellulose respectively. As reported in our previous work (Hamzah et al. 2019), the surface roughness of PVDF/TiO2 membrane after silane modification is 0.62 ± 0.02 μm, which is higher than the silane-modified neat PVDF membrane at 0.41 ± 0.01 μm. The addition of nanocellulose is expected to further increase the surface roughness due to its accelerated solvent–non-solvent exchange rate (Bai et al. 2019). Besides that, the nanoparticles tend to create macrostructures on the membrane surface which in turn increase the surface roughness. However, surface roughness was not affected much by silane modification (Hamzah & Leo 2017). The enhancement of membrane surface roughness is important to achieve the Cassie–Baxter state. In the Cassie–Baxter state, the air pockets existing between the liquid and membrane surface will prevent the penetration of liquid, which could thus minimize the wetting of the membrane. However, the hydrophilic nature of TiO2 nanoparticles and NC reduced WCAs on both unmodified PVDF-T and PVDF-TN membrane significantly before silane modification.

Figure 1

Water contact angle measurements of membranes before and after silane modification.

Figure 1

Water contact angle measurements of membranes before and after silane modification.

Octadecyltrichlorosilane with very low critical surface tension of 20–24 dynes/cm was expected to enhance the water contact angle of the modified surface up to 102°–109° (Arkles 2011). After post-modification, these membranes exhibited WCAs greater than 120° except the PVDF-T1 membrane. TiO2 nanoparticles and nanocellulose provided abundant hydroxyl groups to bond with hydrolyzed silane covalently, but the modification effects could be limited by the particle agglomeration resulting from the TiO2 nanoparticles. Nanocellulose could improve the dispersion of TiO2 nanoparticles in fiber entanglement as reported by others (Ng & Leo 2019), thus enhancing silane modification to improve surface hydrophobicity. Superhydrophobic membranes could be easily created using fluorinated (Hamzah & Leo 2017) and non-fluorinated silanes (Sun et al. 2016) as long as sufficient low-energy groups could be grafted. PVDF-TN membrane was further modified with a higher concentration of silane (3 mL silane : 50 mL ethanol) to increase membrane hydrophobicity. Near-superhydrophobicity with an average WCA of 144.2° was achieved.

SEM images in Figure 2 were used to study the surface morphology of membranes. PVDF membrane showed spongy and interconnected structure with irregular pore size before and after silane modification, which was also displayed in another study that used ethanol as the first coagulation bath. Asymmetric membrane structure with a porous support layer and dense skin layer would form when water was used as the coagulation bath in phase inversion (Zhang et al. 2018a). Ethanol reduced the demixing rate, resulting in a spongy symmetric structure instead of asymmetric structure (Hai et al. 2019). Compared with PVDF membrane, smaller voids and cavities were observed in the SEM image of PVDF-T membrane. TiO2 nanoparticle agglomerates also appeared in PVDF-T membrane. Méricq et al. (2015) reported on TiO2 agglomerates in PVDF membrane in their studies. They concluded that TiO2 nanoparticles could lead to thermodynamic instability and rheological hindrance in phase inversion. Consequently, TiO2 were easily agglomerated and entrapped in small pores. When nanocellulose was added, particle agglomeration was greatly minimized in PVDF-TN membrane, as demonstrated in the SEM images. The interaction and coherence between TiO2 nanoparticles and cellulose fibers promoted the dispersion of nanoparticles and then reduced the undesirable agglomeration (Ng & Leo 2019).

Figure 2

SEM images of membrane surface before and after silane modification.

Figure 2

SEM images of membrane surface before and after silane modification.

PVDF-TN membrane showed distinct porous surface structure with obvious interconnected spherical crystallites. The microfibrils observed in SEM images of PVDF-TN membrane indicate the presence of cellulose nanofibers on the membrane surface. It can be clearly seen that PVDF-TN membrane exhibited larger pores as compared with PVDF and PVDF-T membranes. Bai et al. (2012) and (Zhang et al. 2018b) commented that the presence of nanocellulose could influence the precipitation kinetics in phase inversion and change the membrane morphology subsequently. The strong hydrophilicity of nanocellulose due to its abundant moieties accelerated the exchange between solvent and non-solvent during phase inversion, promoting pore growth in the polymer-poor phase. The addition of nanocellulose greatly increased the viscosity of the solution even at a small amount, causing high resistance in mass transfer and limiting the demixing rate for pore growth in the PVDF-TN membrane (Ismail et al. 2017).

Comparing all the membranes to the silane-modified membranes, the pores of both PVDF1 and PVDF-T1 membranes were slightly hindered after silane modification. The hindrance could be due to the aggregation of silane arising from self-crosslinking that would eventually block the pores, resulting in reduction of permeation flux (Xu et al. 2015). The silanol groups could undergo self-crosslinking via siloxane bonds (Si-O-Si), which were detected in FTIR spectra. For this reason, the porosities of silane PVDF1 and PVDF-T1 membranes were also found to be reduced marginally (Table 2). However, it is worth noticing that the pores of the PVDF-TN membrane were not severely affected by silane modification. PVDF-TN3 membrane modified using highly concentrated silane showed pore blockage which required further measurement of porosity for confirmation.

The membrane cross-section was further studied using SEM images (Figure 2). All membranes showed symmetrical spongy structure before and after silane modification. Since the rate of ethanol-induced phase inversion is slower, it allows ample time for polymer precipitation and pore growth symmetrically as discussed previously. The membrane thicknesses without fabric support were measured using SEM and are tabulated in Table 3. PVDF membrane is the thinnest, followed by PVDF-T and PVDF-TN membranes. The increment of membrane thickness could be attributed to the viscosity increment of the dope solution (Elizalde et al. 2018) after adding TiO2 nanoparticles and nanocellulose. Cellulose has been extensively used because of its thickening effect which depends on the degree of the hydrogen bond (Islam et al. 2018). Li et al. (2004) observed the growth of macro-void size in thick membranes, which was caused by critical structure-transition thickness. Table 2 shows that PVDF-TN, PVDF-TN1 and PVDF-TN3 membranes incorporated with TiO2 and nanocellulose exhibited larger mean pore size and higher porosity than the other fabricated membranes in this work. The changes in pore size and porosity of membrane could be related to the alteration of demixing rate during phase inversion. This work confirmed that the membrane pore size and porosity could be improved by adding nanocellulose into PVDF/TiO2 membrane. Nanocellulose enhanced the demixing process and silanation process, owing to its rich hydroxyl groups (Zhang et al. 2018b). Inorganic particles were widely used to create a rough surface for membrane hydrophobicity improvement (Xu & Wang 2018; Gu et al. 2020), and nanocellulose could be the green substituent to promote pore formation during membrane fabrication. The membrane with large pores is favored for achieving a high water flux, but still the thin membrane is also desired to reduce the mass transfer resistance in MD and MDC. In this work, silane modification did not affect the membrane thickness significantly or the membrane mean pore size and porosity. PVDF-TN3 membrane modified using highly concentrated silane attained a mean pore size of 0.39 μm and porosity of 41.79 ± 6.71%, demonstrating a small reduction of pore size and porosity after silane modification as reported in our work (Hamzah & Leo 2017). The selection of silane modification is important to reduce the negative impact of silane on membrane pore size and porosity.

Table 2

The porosity of unmodified and modified membranes

MembraneMean pore size (μm)Porosity (%)
PVDF 0.18 31.44 ± 3.87 
PVDF1 0.16 28.63 ± 3.81 
PVDF-T 0.22 34.43 ± 3.00 
PVDF-T1 0.21 31.01 ± 2.43 
PVDF-TN 0.36 36.79 ± 8.21 
PVDF-TN1 0.41 44.43 ± 0.97 
PVDF-TN3 0.39 41.79 ± 6.71 
MembraneMean pore size (μm)Porosity (%)
PVDF 0.18 31.44 ± 3.87 
PVDF1 0.16 28.63 ± 3.81 
PVDF-T 0.22 34.43 ± 3.00 
PVDF-T1 0.21 31.01 ± 2.43 
PVDF-TN 0.36 36.79 ± 8.21 
PVDF-TN1 0.41 44.43 ± 0.97 
PVDF-TN3 0.39 41.79 ± 6.71 
Table 3

The thickness of unmodified and modified membranes

MembraneMembrane thickness (μm)
PVDF 23.5 ± 0.3 
PVDF1 25.1 ± 0.0 
PVDF-T 38.6 ± 4.5 
PVDF-T1 38.9 ± 3.0 
PVDF-TN 82.6 ± 2.6 
PVDF-TN1 84.5 ± 1.4 
MembraneMembrane thickness (μm)
PVDF 23.5 ± 0.3 
PVDF1 25.1 ± 0.0 
PVDF-T 38.6 ± 4.5 
PVDF-T1 38.9 ± 3.0 
PVDF-TN 82.6 ± 2.6 
PVDF-TN1 84.5 ± 1.4 

The FTIR spectra in Figure 3 show the changes in chemical composition of the unmodified and modified membranes. The bands related to the CH2 wagging vibration at 1,401 cm−1 and C-C bonding at 1,168 cm−1 indicate the chemical property of PVDF. The two peaks observed at 839 cm−1 and 878 cm−1 correspond to the C-F stretching vibration and C-C-C asymmetrical stretching vibration of PVDF, respectively (Bai et al. 2012). After modification, the FTIR spectrum of the PVDF1 membrane showed the existence of extra peaks at 2,849 cm−1 and 2,917 cm−1 which resulted from the CH2 symmetric and asymmetric stretching of the PVDF and silane coupling agent. The silane coupling agent, rich in methylene groups, caused the peaks to appear to be more intense. Changes were also observed in the range of 1,150–1,000 cm−1 due to the existence of Si-O-Si groups of silane in the range of 1,055–1,020 cm−1 and O-Si-O groups at the peak of 1,071 cm−1 (Ahmad et al. 2018). This proved that silanol groups of octadecyltrichlorosilane were successfully grafted on the membrane surface. However, the Si-Cl band at 625–425 cm−1 (Launer & Arkles 2013) was not observed in the FTIR spectra. This indicated that silane had been completely reacted during hydrolysis to form silanol. The PVDF-T membrane showed a similar FTIR pattern to the PVDF membrane. Besides the PVDF characteristic peaks, the typical vibrations for Ti-O bonds for TiOn compounds with n < 6 ranging from 770 to 800 cm−1 (Collazzo et al. 2011) appeared in the FTIR spectra of the PVDF-T membrane before and after silane modification due to the presence of TiO2 nanoparticles. After silane modification, the appearance of Si-O-Si and O-Si-O groups were confirmed by the peak located between 1,150 cm−1 and 1,000 cm−1 in the FTIR spectrum of the modified PVDF-T1 membrane. The FTIR spectrum of the PVDF-TN membrane showed a similar pattern to the PVDF-T membrane even with nanocellulose incorporated into the membrane. The characteristic peaks of PVDF were displayed in the FTIR spectra of the PVDF-TN and PVDF-TN1 membranes. The presence of TiO2 nanoparticles was also proven by the peaks in the range of 650 cm−1 to 800 cm−1. Changes of bands (1,150 cm−1 to 1,000 cm−1) due to silanation were still observed in the PVDF-TN1 membrane. Different from the PVDF and PVDF-T membranes, the PVDF-TN membrane showed the extra peak of the O-H stretching vibration of cellulose at 3,341 cm−1 (Bai et al. 2012).

Figure 3

FTIR spectra of unmodified (dotted line) and modified (solid line) membranes.

Figure 3

FTIR spectra of unmodified (dotted line) and modified (solid line) membranes.

Membrane distillation (MD) and membrane distillation crystallization (MDC)

The unmodified and modified PVDF, PVDF-T and PVDF-TN membranes were first tested in MD before membrane selection for MDC. Figure 4 summarizes the permeate flux of water vapor from the hot feed solution at 60 °C through these membranes into the cold permeate water at 20 °C. These membranes were tested in MD for 6 h using the feed containing 35 g/L NaCl and distilled water which were counter-currently circulated at 300 mL/min. Based on Figure 4(a)(i), all the unmodified membranes show almost the same performance with the initial permeate flux as high as 5.0–6.0 kg·m−2·h−1. After 6 h of operation, the permeate flux falls to a range of 0.5–1.5 kg·m−2·h−1. The reduction of flux could be related to membrane wetting as reported by others (Munirasu et al. 2017). When wetting occurred in a hydrophobic membrane, the liquid-filled pores introduced extra resistance to the transfer of water vapor.

Figure 4

Permeate flux of (a) unmodified and (b) modified membranes in (i) MD and (ii) MDC.

Figure 4

Permeate flux of (a) unmodified and (b) modified membranes in (i) MD and (ii) MDC.

However, the membranes modified with silane showed variation of permeate flux in MD as shown in Figure 4(b)(i). The modified PVDF1 and PVDF-T1 membranes demonstrated a significant decline in permeate flux to 1.0 kg·m−2·h−1 after 6 h. The pore-size reduction in modified PVDF1 and PVDF-T1 could be the reason for permeation reduction in comparison with PVDF and PVDF-T, respectively. However, the permeate flux around 2.5 kg·m−2·h−1 was achieved after 6 h when the modified PVDF-TN1 was applied in MD. The improvement of permeate flux was mainly contributed by its large pore size and improved hydrophobicity among all the fabricated membranes. In terms of salt rejection (Figure 5), more than 99.0% of NaCl salt was being rejected when using PVDF, PVDF1, PVDF-T and PVDF-T1 membranes. The percentage of salt rejection was slightly reduced to around 97.7% when using PVDF-TN and PVDF-TN1 membranes with larger pore size and higher porosity.

Figure 5

Salt rejection of unmodified and modified membranes.

Figure 5

Salt rejection of unmodified and modified membranes.

Highly concentrated NaCl solution (100 g/L) was used as the feed in the MDC system. To minimize membrane fouling, the PVDF-TN membrane was further modified into PVDF-TN3 membrane using concentrated silane solution (3 mL silane to 50 mL ethanol). The permeate flux of PVDF-TN3 membrane in MDC is presented in Figure 4(b)(ii). With all the same operating conditions as in MD, MDC was conducted at a feed concentration increased to 100 g/L NaCl for crystallization to occur. The membrane achieved salt rejection of 99.4% and an initial permeate flux of 4.5 kg·m−2·h−1, but the permeate flux reduced gradually near to 2.0 kg·m−2·h−1 after 6 h of operation. The PVDF-TN3 membrane achieved lower initial permeate flux than the PVDF-TN1 membrane due to the high feed concentration in MDC. The gradual flux decline was principally related to the reduction of vapor pressure that resulted from the increasing feed concentration, along with the effect of concentration and temperature polarization as reported by Zou et al. (2019). The water flux of membranes developed in this work is comparable to our previous research (Hamzah & Leo 2016) and other studies (Leitch et al. 2016; Zou et al. 2019; Gu et al. 2020) (Table 4). Severe membrane fouling was not observed in this work. Figure 6 shows that the salt crystals were hardly found on the surface of the PVDF-TN3 membrane.

Table 4

Comparison between studies of membrane properties and performances in different applications

MembraneaMembrane characteristicsbOperating conditionscFlux (kg·m−2·h−1)Rejection (%)Reference
PTFE hollow fiber WCA: – F: 100 g/L NaCl 5.0 – Zou et al. (2019)  
Φ: 41.0% P: Vacuum 
dp: 0.186 μm TF: 75 °C 
 TP: – 
PVDF/TiO2 WCA: 140.0° F: 100 g/L gallic acid 2.5 99.9 Hamzah & Leo (2016)  
Φ: 49.9% P: Distilled water 
dp: 0.400 μm TF: 40 °C 
 TP: 20 °C 
PU/SiO2 WCA: 152.1° F: Oil – > 98.5 Gu et al. (2020)  
Φ: 62.5% PF: 0.9 bar 
dp: 1.796 μm  
Bacterial NC Aerogel WCA: 156.0° F : – 8.4 99.9 Leitch et al. (2016)  
Φ: 98.0% P : – 
dp: 0.115 μm TF: 40 °C 
 TP: 20 °C 
PVDF/TiO2/NC WCA: 124.1° F: 35 g/L NaCl 2.5 97.7 This work 
Φ: 44.4% P: Distilled water 
dp: 0.41 μm TF: 60 °C 
 TP: 20 °C 
PVDF/TiO2/NC WCA: 144.2° F: 100 g/L NaCl 2.0 99.4 This work 
Φ: 41.79% P: Distilled water 
dp: 0.39 μm TF: 60 °C 
 TP: 20 °C 
MembraneaMembrane characteristicsbOperating conditionscFlux (kg·m−2·h−1)Rejection (%)Reference
PTFE hollow fiber WCA: – F: 100 g/L NaCl 5.0 – Zou et al. (2019)  
Φ: 41.0% P: Vacuum 
dp: 0.186 μm TF: 75 °C 
 TP: – 
PVDF/TiO2 WCA: 140.0° F: 100 g/L gallic acid 2.5 99.9 Hamzah & Leo (2016)  
Φ: 49.9% P: Distilled water 
dp: 0.400 μm TF: 40 °C 
 TP: 20 °C 
PU/SiO2 WCA: 152.1° F: Oil – > 98.5 Gu et al. (2020)  
Φ: 62.5% PF: 0.9 bar 
dp: 1.796 μm  
Bacterial NC Aerogel WCA: 156.0° F : – 8.4 99.9 Leitch et al. (2016)  
Φ: 98.0% P : – 
dp: 0.115 μm TF: 40 °C 
 TP: 20 °C 
PVDF/TiO2/NC WCA: 124.1° F: 35 g/L NaCl 2.5 97.7 This work 
Φ: 44.4% P: Distilled water 
dp: 0.41 μm TF: 60 °C 
 TP: 20 °C 
PVDF/TiO2/NC WCA: 144.2° F: 100 g/L NaCl 2.0 99.4 This work 
Φ: 41.79% P: Distilled water 
dp: 0.39 μm TF: 60 °C 
 TP: 20 °C 

aNC is nanocellulose.

bWCA is water contact angle (°); Φ is porosity (%); dp is mean pore size (μm).

cF is feed; P is permeate; TF is feed temperature; TP is permeate temperature: PF is feed pressure.

Figure 6

Salt crystals formed during MDC using PVDF-TN3 membrane.

Figure 6

Salt crystals formed during MDC using PVDF-TN3 membrane.

CONCLUSIONS

TiO2 nanoparticles and nanocellulose were successfully blended into PVDF membrane to improve membrane characteristics for water and salt recovery via MD and MDC. The post-modified membranes with non-fluorinated silane showed great improvement of hydrophobicity up to a water contact angle of 144.2°. However, silane tended to cause pore blockage in the membranes, which in turn created more resistance in mass transfer and reduced water permeate flux in MD and MDC. The incorporation of nanocellulose in the membrane induced the formation of larger pore size and greater porosity, leading to a high water permeate flux of 2.5 kg·m−2·h−1 being achieved by the modified PVDF-TN membrane with salt rejection as high as 97% after 6 h of operation. In MDC, the high feed concentration reduced the permeability of water vapor. Hence, more future studies are required to improve the permeability, supersaturation degree, and nucleation rate for crystal formation in MDC.

ACKNOWLEDGEMENTS

The authors would like to acknowledge the financial support, LRGS (203/PJKIMIA/67215002) from the Ministry of Education Malaysia. The authors would like to thank Borregaard Cellulose Fibrils for providing microfibrillated cellulose.

REFERENCES

REFERENCES
Ahmad
N. A.
,
Leo
C. P.
&
Ahmad
A. L.
2018
Amine wetting evaluation on hydrophobic silane modified polyvinylidene fluoride/silicoaluminophosphate zeolite membrane for membrane gas absorption
.
Journal of Natural Gas Science and Engineering
58
,
115
125
.
https://doi.org/10.1016/j.jngse.2018.08.003
.
Arkles
B.
2011
Hydrophobicity, Hydrophilicity and Silane Surface Modification
, 2nd edn.
Gelest Inc
,
Morrisville, PA, USA
.
Bai
H.
,
Wang
X.
,
Zhou
Y.
&
Zhang
L.
2012
Preparation and characterization of poly(vinylidene fluoride) composite membranes blended with nano-crystalline cellulose
.
Progress in Natural Science: Materials International
22
(
3
),
250
257
.
https://doi.org/10.1016/j.pnsc.2012.04.011
.
Bai
L.
,
Liu
Y.
,
Ding
A.
,
Ren
N.
,
Li
G.
&
Liang
H.
2019
Surface coating of UF membranes to improve antifouling properties: a comparison study between cellulose nanocrystals (CNCs) and cellulose nanofibrils (CNFs)
.
Chemosphere
217
,
76
84
.
https://doi.org/10.1016/j.chemosphere.2018.10.219
.
Cheng
Q.-Y.
,
Guan
C.-S.
,
Wang
M.
,
Li
Y.-D.
&
Zeng
J.-B.
2018
Cellulose nanocrystal coated cotton fabric with superhydrophobicity for efficient oil/water separation
.
Carbohydrate Polymers
199
,
390
396
.
https://doi.org/10.1016/j.carbpol.2018.07.046
.
Collazzo
G. C.
,
Jahn
S. L.
,
Carreño
N. L. V.
&
Foletto
E. L.
2011
Temperature and reaction time effects on the structural properties of titanium dioxide nanopowders obtained via the hydrothermal method
.
Brazilian Journal of Chemical Engineering
28
,
265
272
.
http://dx.doi.org/10.1590/S0104-66322011000200011
.
Cui
Z.
,
Li
X.
,
Zhang
Y.
,
Wang
Z.
,
Gugliuzza
A.
,
Militano
F.
,
Drioli
E.
&
Macedonio
F.
2018
Testing of three different PVDF membranes in membrane assisted-crystallization process: influence of membrane structural-properties on process performance
.
Desalination
440
,
68
77
.
https://doi.org/10.1016/j.desal.2017.12.038
.
Dai
Z.
,
Deng
J.
,
Yu
Q.
,
Helberg
R. M. L.
,
Janakiram
S.
,
Ansaloni
L.
&
Deng
L.
2019
Fabrication and evaluation of bio-based nanocomposite TFC hollow fiber membranes for enhanced CO2 capture
.
ACS Applied Materials & Interfaces
11
(
11
),
10874
10882
.
https://doi.org/10.1021/acsami.8b19651
.
Elizalde
C. N. B.
,
Al-Gharabli
S.
,
Kujawa
J.
,
Mavukkandy
M.
,
Hasan
S. W.
&
Arafat
H. A.
2018
Fabrication of blend polyvinylidene fluoride/chitosan membranes for enhanced flux and fouling resistance
.
Separation and Purification Technology
190
,
68
76
.
https://doi.org/10.1016/j.seppur.2017.08.053
.
Gu
H.
,
Li
G.
,
Li
P.
,
Liu
H.
,
Chadyagondo
T. T.
,
Li
N.
&
Xiong
J.
2020
Superhydrophobic and breathable SiO2/polyurethane porous membrane for durable water repellent application and oil–water separation
.
Applied Surface Science
512
,
144837
.
https://doi.org/10.1016/j.apsusc.2019.144837
.
Hai
A.
,
Durrani
A. A.
,
Selvaraj
M.
,
Banat
F.
&
Haija
M. A.
2019
Oil–water emulsion separation using intrinsically superoleophilic and superhydrophobic PVDF membrane
.
Separation and Purification Technology
212
,
388
395
.
https://doi.org/10.1016/j.seppur.2018.10.001
.
Hamzah
N.
&
Leo
C. P.
2016
Fouling prevention in the membrane distillation of phenolic-rich solution using superhydrophobic PVDF membrane incorporated with TiO2 nanoparticles
.
Separation and Purification Technology
167
,
79
87
.
https://doi.org/10.1016/j.seppur.2016.05.005
.
Hamzah
N.
,
Leo
C. P.
&
Ooi
B. S.
2019
Superhydrophobic PVDF/TiO2–SiO2 membrane with hierarchical roughness in membrane distillation for water recovery from phenolic rich solution containing surfactant
.
Chinese Journal of Polymer Science
37
(
6
),
609
616
.
https://doi.org/10.1007/s10118-019-2235-y
.
Islam
E.
,
Kaur
G.
,
Bhattacharjee
D.
,
Singh
S.
,
Biswas
I.
&
Verma
S. K.
2018
Effect of cellulose beads on shear-thickening behavior in concentrated polymer dispersions
.
Colloid and Polymer Science
296
(
5
),
883
893
.
https://doi.org/10.1007/s00396-018-4299-6
.
Ismail
N. M.
,
Jakariah
N. R.
,
Bolong
N.
,
Anissuzaman
S. M.
,
Nordin
N. A. H. M.
&
Razali
A. R.
2017
Effect of polymer concentration on the morphology and mechanical properties of asymmetric polysulfone (PSf) membrane
.
Journal of Applied Membrane Science & Technology
21
(
1
),
33
41
.
https://doi.org/10.11113/amst.v21i1.107
.
Karim
Z.
,
Claudpierre
S.
,
Grahn
M.
,
Oksman
K.
&
Mathew
A. P.
2016
Nanocellulose based functional membranes for water cleaning: tailoring of mechanical properties, porosity and metal ion capture
.
Journal of Membrane Science
514
,
418
428
.
https://doi.org/10.1016/j.memsci.2016.05.018
.
Kim
J.
,
Kim
J.
&
Hong
S.
2018
Recovery of water and minerals from shale gas produced water by membrane distillation crystallization
.
Water Research
129
,
447
459
.
https://doi.org/10.1016/j.watres.2017.11.017
.
Ko
C.-C.
,
Ali
A.
,
Drioli
E.
,
Tung
K.-L.
,
Chen
C.-H.
,
Chen
Y.-R.
&
Macedonio
F.
2018
Performance of ceramic membrane in vacuum membrane distillation and in vacuum membrane crystallization
.
Desalination
440
,
48
58
.
https://doi.org/10.1016/j.desal.2018.03.011
.
Kollarigowda
R. H.
,
Abraham
S.
&
Montemagno
C. D.
2017
Antifouling cellulose hybrid biomembrane for effective oil/water separation
.
ACS Applied Materials & Interfaces
9
(
35
),
29812
29819
.
https://doi.org/10.1021/acsami.7b09087
.
Koschikowski
J.
,
Wieghaus
M.
&
Rommel
M.
2003
Solar thermal driven desalination plants based on membrane distillation
.
Water Supply
3
(
5–6
),
49
55
.
https://doi.org/10.2166/ws.2003.0149
.
Launer
P. J.
&
Arkles
B.
2013
Infrared analysis of organosilicon compounds
. In:
Silicon compounds: Silanes and Silicones
.
(B. Arkles & G. L. Larson, eds)
,
Gelest Inc
,
Morrisville, PA, USA
, pp.
177
180
.
Lee
S.
,
Choi
J.
,
Park
Y.-G.
,
Shon
H.
,
Ahn
C. H.
&
Kim
S.-H.
2019
Hybrid desalination processes for beneficial use of reverse osmosis brine: current status and future prospects
.
Desalination
454
,
104
111
.
https://doi.org/10.1016/j.desal.2018.02.002
.
Leitch
M. E.
,
Li
C.
,
Ikkala
O.
,
Mauter
M. S.
&
Lowry
G. V.
2016
Bacterial nanocellulose aerogel membranes: novel high-porosity materials for membrane distillation
.
Environmental Science & Technology Letters
3
(
3
),
85
91
.
https://doi.org/10.1021/acs.estlett.6b00030
.
Li
D.
,
Chung
T.-S.
,
Ren
J.
&
Wang
R.
2004
Thickness dependence of macrovoid evolution in wet phase-inversion asymmetric membranes
.
Industrial & Engineering Chemistry Research
43
(
6
),
1553
1556
.
https://doi.org/10.1021/ie034264 g
.
Méricq
J. P.
,
Mendret
J.
,
Brosillon
S.
&
Faur
C.
2015
High performance PVDF–TiO2 membranes for water treatment
.
Chemical Engineering Science
123
,
283
291
.
https://doi.org/10.1016/j.ces.2014.10.047
.
Munirasu
S.
,
Banat
F.
,
Durrani
A. A.
&
Haija
M. A.
2017
Intrinsically superhydrophobic PVDF membrane by phase inversion for membrane distillation
.
Desalination
417
,
77
86
.
https://doi.org/10.1016/j.desal.2017.05.019
.
Ng
H. K. M.
&
Leo
C. P.
2019
The coherence between TiO2 nanoparticles and microfibrillated cellulose in thin film for enhanced dispersal and photodegradation of dye
.
Progress in Organic Coatings
132
,
70
75
.
https://doi.org/10.1016/j.porgcoat.2019.02.017
.
Quist-Jensen
C. A.
,
Sørensen
J. M.
,
Svenstrup
A.
,
Scarpa
L.
,
Carlsen
T. S.
,
Jensen
H. C.
,
Wybrandt
L.
&
Christensen
M. L.
2018
Membrane crystallization for phosphorus recovery and ammonia stripping from reject water from sludge dewatering process
.
Desalination
440
,
156
160
.
https://doi.org/10.1016/j.desal.2017.11.034
.
Ruiz Salmón
I.
&
Luis
P.
2018
Membrane crystallization via membrane distillation
.
Chemical Engineering and Processing – Process Intensification
123
,
258
271
.
https://doi.org/10.1016/j.cep.2017.11.017
.
Shahu
V. T.
&
Thombre
S. B.
2020
Analysis and optimization of a new cylindrical air gap membrane distillation system
.
Water Supply
20
(
1
),
361
371
.
https://doi.org/10.2166/ws.2019.164
.
Sun
H.
,
Xu
Y.
,
Zhou
Y.
,
Gao
W.
,
Zhao
H.
&
Wang
W.
2016
Preparation of superhydrophobic nanocomposite fiber membranes by electrospinning poly(vinylidene fluoride)/silane coupling agent modified SiO2 nanoparticles
.
Journal of Applied Polymer Science
134
,
44501
.
https://doi.org/10.1002/app.44501
.
Xu
C.-L.
&
Wang
Y.-Z.
2018
Self-assembly of stearic acid into nano flowers induces the tunable surface wettability of polyimide film
.
Materials & Design
138
,
30
38
.
https://doi.org/10.1016/j.matdes.2017.10.057
.
Xu
W.-T.
,
Zhao
Z.-P.
,
Liu
M.
&
Chen
K.-C.
2015
Morphological and hydrophobic modifications of PVDF flat membrane with silane coupling agent grafting via plasma flow for VMD of ethanol–water mixture
.
Journal of Membrane Science
491
,
110
120
.
https://doi.org/10.1016/j.memsci.2015.05.024
.
Zhan
H.
,
Peng
N.
,
Lei
X.
,
Huang
Y.
,
Li
D.
,
Tao
R.
&
Chang
C.
2018
UV-induced self-cleanable TiO2/nanocellulose membrane for selective separation of oil/water emulsion
.
Carbohydrate Polymers
201
,
464
470
.
https://doi.org/10.1016/j.carbpol.2018.08.093
.
Zhang
W.
,
Zhong
L.
,
Wang
T.
,
Jiang
Z.
,
Gao
X.
&
Zhang
L.
2018
Surface modification of cellulose nanofibers and their effects on the morphology and properties of polysulfone membranes
.
IOP Conference Series: Materials Science and Engineering
397
,
012016
.
https://doi.org/10.1088/1757-899X/397/1/012016
.
Zou
T.
,
Kang
G.
,
Zhou
M.
,
Li
M.
&
Cao
Y.
2019
Submerged vacuum membrane distillation crystallization (S-VMDC) with turbulent intensification for the concentration of NaCl solution
.
Separation and Purification Technology
211
,
151
161
.
https://doi.org/10.1016/j.seppur.2018.09.072
.