inorganic nanoparticles on performance of polyurethane nanocomposites for potential wound dressing applications

Author’s Accepted Manuscript Effect of organic/inorganic nanoparticles on performance of polyurethane nanocomposites for potential wound dressing applications Arman Jafari, Shadi Hassanajili, Mohammad Bagher Karimi, Amir Emami, Farnaz Ghaffari, Negar Azarpira www.elsevier.com/locate/jmbbm

PII: DOI: Reference:

S1751-6161(18)30908-1 https://doi.org/10.1016/j.jmbbm.2018.09.001 JMBBM2966

To appear in: Journal of the Mechanical Behavior of Biomedical Materials Received date: 17 June 2018 Revised date: 31 August 2018 Accepted date: 2 September 2018 Cite this article as: Arman Jafari, Shadi Hassanajili, Mohammad Bagher Karimi, Amir Emami, Farnaz Ghaffari and Negar Azarpira, Effect of organic/inorganic nanoparticles on performance of polyurethane nanocomposites for potential wound dressing applications, Journal of the Mechanical Behavior of Biomedical Materials, https://doi.org/10.1016/j.jmbbm.2018.09.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Effect of organic/inorganic nanoparticles on performance of polyurethane nanocomposites for potential wound dressing applications Arman Jafaria, Shadi Hassanajilia,, Mohammad Bagher Karimib, Amir Emamic, Farnaz Ghaffaria, Negar azarpirad a

Department of Chemical Engineering, School of Chemical and Petroleum Engineering, Shiraz University, Shiraz 71348-51154, Iran b

c

Burn & Wound Healing Research Center, Microbiology Department, Shiraz University of Medical Science, Shiraz, 71345-1978, Iran d



Iran Polymer and Petrochemical Institute, 14965-115, Tehran, Iran

Transplant Research Center, Shiraz University of Medical Science, Shiraz, 71345-1978, Iran

Corresponding author: Shadi Hassanajili, e-mail: [email protected] Tel: +98 713 6133779

Abstract This study focuses on the evaluation and modification of polyurethane (PU) membranes containing organic and inorganic nanoparticles for potential use as a wound dressing. For the purpose of PU nanocomposite preparation, chitosan (CS) was converted into nanoparticles by the ionic-gelation method to improve its blending capability with the PU matrix. These CS nanoparticles (nano-CS) were obtained as a hyrdophilic organic filler with different contents and were utilized along with inorganic titanium dioxide (TiO2) nanoparticles in the nanocomposite membrane preparation. The membranes were prepared using phase inversion technique and their microstructure was controlled by manipulating the solvent non-solvent exchange rate. Obtained results demonstrate that addition of polymer solvent to nonsolvent induced a microstructure alteration from finger-like to sponge-like, which is more suitable for fluid uptake and

1

consequently more useful for wound dressing applications. Similar results were obtained by introduction of nanoparticles to membranes. Due to the polar nature of nanoparticles and their effects on PU structure, prepared membranes showed 71.5% improve in swelling when compared to neat PU. Moreover, the reinforcement effect of nanoparticles caused an 18.94% increase in ultimate tensile strength in comparison with bare PU film, while elongation at break was not affected considerably. In addition, prepared PU nanocomposite films showed suitable antibacterial activity of 69% against Staphylococcus aureus and did not show any toxicity to human fibroblast cells. Based on these results, simultaneous use of TiO2 and chitosan nanoparticles can improve both physical and antibacterial properties of PU as an ideal wound dressing.

Keywords: polyurethane; wound dressing; nano-chitosan; phase inversion 1. Introduction The first line of the human body against invasion of harmful external factors is the skin (Hess, 2012). The skin has barrier properties against microbial agents to prevent microbial infections. When the skin is injured or destroyed, it loses its efficacy and becomes vulnerable to microbial attack. After skin destruction, the wound needs protection from environmental stimuli until skin can regenerate itself. This protection can be achieved by wound dressing (Kamoun et al., 2017). Based on medical criteria, an ideal wound dressing needs a suitable water vapor transmission rate (WVTR), high oxygen permeation, low adhesion to wound site, suitable exudate uptake, should appropriately hinder the perforation of infections, and should possess good antibacterial

2

properties to diminish the bacteria growth around the wound (Chen et al., 2011; Di et al., 2017; Ovington, 2007). Polymers, both synthetic and natural, are used as a new generation of wound dressings due to their availability, manufacturability, and broad range of physical, mechanical and biological properties. Among polymers, PU is abundantly used in painting, coatings, adhesives, and film preparation due to its exceptional mechanical properties (Aung et al., 2014; Karimi and Hassanajili, 2017). PU is a block copolymer composed of hard and soft segments (Karimi et al., 2018). Hard segments, which are the crystalline part of PU, are responsible for mechanical strength, while soft segments are rubbery and give the polymer elastic behavior and flexibility (Hassanajili et al., 2014). The mechanical strength of PU, as well as its biocompatibility, make PU a good candidate for bioprinting and fabrication of wound dressing (Aljohani et al., 2017). For instance, Khil et al. produced nanofibers of PU with ultrafine pores and reported that their membrane has good oxygen permeability without any toxic effect as a wound dressing (Khil et al., 2003). In another work, Unnithan and colleagues prepared a wound dressing using PU and dextran with enhanced cell attachment and proliferation (Unnithan et al., 2012). Along with laboratory investigations on PU based bandages, there are also commercially available PU dressings including TegadermTM. Nano-scaled materials have generated great interest in recent years (Pitkethly, 2004). In biomedical applications, they are used for cancer therapy, imaging, drug delivery, and more (Biju, 2014; Lin et al., 2016; Peng et al., 2014). Antibacterial activity of nano ZnO, Ag, TiO2, CuO have already been shown (Ren et al., 2009; Sirelkhatim et al., 2015; You et al., 2017). TiO2 is conventionally used as a catalyst due to its photocatalytic activity (Fotiou et al., 2015; Yu et

3

al., 2003). Previously, different research groups have demonstrated the antibacterial activity of TiO2 nanoparticles against both gram-positive and gram-negative bacteria (Fu et al., 2005; Joost et al., 2015; Podporska-Carroll et al., 2015; Woo et al., 2015). Micro or nano-scale chitosan particles have generated interest for medical applications (Ghadi et al., 2014; Jayakumar et al., 2010). Chitosan, which is derived from chitin, the second-most abundant natural biopolymer, is distinguished by properties like biocompatibility, hydrophilicity, swelling ability, antibacterial susceptibility, and hemostatic properties (Chung et al., 2004; Dash et al., 2011; Rázga et al., 2016; Sudheesh Kumar et al., 2012). The reaction of chitosan with sodium tripolyphosphate (TPP) is a simple way to produce nano-chitosan particles. Several types of research showed that these nanoparticles have improved antibacterial activity with reduced cytotoxicity (Malathi et al., 2015). In addition, nanoscale chitosan can be blended with hydrophobic polymers to improve their hydrophilicity and biocompatibility. Jung et al. (Jung et al., 2015) prepared polycaprolactone-based nanofibers and incorporated nano-CS particles as drug carriers. They showed that this nanocomposite is nontoxic and that nanoparticles are suitable carriers for drug delivery. In this work, we tried to improve the antibacterial performance of polyurethane using organic and inorganic nanoparticles (nano-CS and TiO2, respectively). As polymer microstructure has a great impact on wound dressing performance, we tried to adjust the polymer microstructure by manipulating phase inversion speed. Obtained results revealed that PU membrane with a spongelike morphology is more suitable for wound dressing applications and that the presence of nanoparticles can effectively improve their properties. 2. Materials and methods

4

2.1. Materials Polyether based thermoplastic PU was supplied by Coim Group, Italy (trade name: LPR9060EF, density: 1.20 g/cm3, Shore hardness A: 90). Its soft segment is comprised of poly (oxytetramethylene) (PTMG). Chitosan (CS, high molecular weight (to be approximately 310000), >75% deacetylation), Thiazolyl Blue Tetrazolium Bromide (MTT), and nano-TiO2 (<25 nm particle size) were bought from Sigma-Aldrich Co. Human fibroblast cells were supplied by Pasteur Institute of Iran. Sodium tripolyphosphate (TPP), N, N-Dimethylformamide (DMF) and acetic acid (AA) anhydrous were purchased from Merck Co. Deionized water was used for all of the experiments. 2.2. Synthesis of nano-chitosan particles Nano-CS particles were synthesized by ionic-gelation method with a modified procedure described before (Anitha et al., 2009). Briefly, CS was dissolved in acetic acid to obtain a 1 v/v% solution. Crosslinking agent (TPP) was also dissolved in acetic acid to obtain 1 w/v% and added dropwise to CS solution under magnetic stirring of 1000 rpm, at room temperature and with a volume ratio of 1/10 TPP/CS. Nanoparticles were formed spontaneously by this addition. When all of the TPP solutions were added, the mixture was stirred for another 30min. After this time, the solution was centrifuged to recover nanoparticles and then lyophilized at -47C and 0.055 mbar to dry synthesized nano-CS completely. 2.3. Preparation of membranes The dry/wet phase inversion method was used for membrane preparation. Different membranes were prepared based on Box-Behnken design technique for rational procedure for preparation of

5

different membranes. The result of this design is given in Table 1. The percentage of TiO2 was fixed at 5% based on previous investigations (Chen et al., 2011; He et al., 2016; Yan et al., 2011). All of the PU solutions were prepared at 20% (w/v) (Based on pretests taken, higher concentrations (e.g. 30%) make the solution too viscous and lower concentrations (e.g. 10%) do not result in good membranes). PU was dissolved in DMF and acetic acid solution under magnetic stirring at 70°C to obtain a 20 wt% solution. Acetic acid was added to DMF to improve dispersion of chitosan nanoparticles. After PU completely dissolved in the solvent, it was cooled to room temperature. Subsequently, the desired amount of both chitosan and TiO2 nanopowders were added to the solution and were stirred with the PU solution for a while at room temperature. After nanopowders were dispersed moderately, the solution was sonicated to disperse the nanopowders completely. The sonicated mixture was cast on a glass plate using a film casting blade. Table 1. Box-Behnken design for three specified variables. Experiment No.*

%Nano-CS (wt %)

%water (v %)

%AA (v %)

1

7.5

85

0

2

7.5

85

5

3

10

92.5

5

4

7.5

100

5

5

5

92.5

5

6

5

92.5

0

7

5

85

2.5

8

10

85

2.5

9

5

100

2.5

10

10

100

2.5

11

10

92.5

0

12

7.5

100

0

13

7.5

92.5

2.5

14

7.5

92.5

2.5

6

15 *

7.5

92.5

2.5

Constant conditions for all experiments: 20% PU solution and 5% TiO2 nanoparticle

The casted mixture was then immersed in the coagulation bath containing a mixture of deionized water and DMF. After 24 h the membranes were removed from the coagulation bath, washed with deionized water to remove any remained DMF, then dried for 24 h at 50 °C and kept for further analysis. This process is shown in Scheme 1. The thickness of fabricated nanocomposites was 70 ± 5 µm.

Scheme 1. Schematic representation of the preparation of the wound dressing material.

2.4. Characterization of chitosan nanoparticles and wound dressings

7

FT-IR spectra of CS, nano-CS particles, PU, and nanocomposites were taken with Perkin Elmer (model RX1) Fourier transform infrared spectrophotometer in range of 400-4000 cm-1 with resolution of 2 cm-1. Particle size distribution of nanoparticles were measured using dynamic light scattering (SZ-1, Horiba, Japan) measurement at scattering angle of 173 and temperature of 25 °C. 2.5. Scanning electron microscopy Scanning electron microscopy (SEM, TESCAN-Vega3, Czech Republic) was used to investigate the top surface and cross-section morphology of the membranes. The cross-section of the samples was prepared by breaking the membranes while they were immersed in liquid nitrogen and then were coated by gold with sputtering coater (Q150R-ES, Quorum Technologies, England) before placing in the SEM equipment. To determine the mean particle size for nano-CS particles, a specified amount of nano-CS was weighed and then dispersed in alcohol using probe sonicator, and finally, SEM images were taken for further investigations. 2.6. Swelling measurement To determine the extent of fabricated membranes’ swelling, they were immersed in a solution of phosphate buffer saline (PBS) with pH of 7.4 at a temperature of 37°C. The measurements were taken two different times of 1 h and 24 h. For tests, a specific amount of the membranes was cut and then weighed, subsequently, they were submerged in PBS solution and kept at 37°C. After an hour, membranes were removed from the solution, and their surfaces were dried using a filter paper to remove excess PBS on the surface and then weighed again to calculate the swelling percentage using the following formula:

8

(

(1)

)

Where SW is a percentage of swelling, WW is the weight of membrane soaked in PBS solution and Wd is the weight of dry membrane weighed before starting the test. The same procedure was conducted for 24 h measurements. All of the tests were repeated three times, and an average of these test is reported. 2.7. Water vapor transmission rate (WVTR) measurement To evaluate the permeability of the membranes to moisture, the WVTR test was taken by modified ASTM E96. In this method, a cup was filled with some amount of water, and the membrane was put on the opening of the cup and sealed so that no vapor could leave the cup except via the membrane. After that, the cup was weighed and placed in a desiccator, which is filled with silica gel to maintain the humidity of the desiccator at zero. The temperature was fixed at 37C. The cups were removed from desiccator after 24 h and were weighed again to calculate the WVTR. 2.8. Antibacterial studies Antibacterial studies of prepared membranes were done based on the modified ASTM 2149-01 for a gram-negative bacteria Pseudomonas aeruginosa (ATCC 27853) and a gram-positive bacteria Staphylococcus aureus (ATCC 25923). Briefly, 0.1 g of each membrane was cut and weighed. Bacteria were cultured overnight, and then a suspension of approximately 106 CFU/mL

9

was prepared. Each weighed sample was then added to 50 mL of suspension and was put in a shaking incubator at a temperature of 37°C for 2 h with a shaking rate of 170 rpm. A time point of 2 h was chosen to bypass any possible lag time for bacteria growth. At time 0 and 2 h, a certain amount of each suspension was collected, serial dilutions were performed, and the standard plate count was used to investigate the antibacterial activity in triplicate for each sample and bacteria. PU-only dressing was used as a control specimen. 2.9. In vitro biocompatibility Biocompatibility of wound dressings was evaluated using MTT assay. In brief, dressings were punched to fit in 24-well cell culture plates, sterilized, seeded with 3×104 of human fibroblast cells (passage 4) per well and were incubated at 37 ± 0.1 °C with 5% CO2. Media for cell culture was Dulbecco’s modified Eagle medium/nutrient mixture F-12 containing 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. After a specified time (24h, 48h, and 72h), plates were taken out from the incubator, the medium was removed, and 500 µL of MTT solution (2.5 mg/mL) was added to each well and again incubated for three hours. After this time interval, plates were centrifuged and solutions were discarded and 500 µL of dimethyl sulfoxide (DMSO) was added to each well and shaken for 15min to dissolve all of the synthesized formazans. Optical density (O.D.) of each sample was read at a wavelength of 570 and 630 nm by a microplate reader. The difference between these two data points was used for cytotoxicity analysis. Cell morphology, attachment, and proliferation of wound dressings after one day of incubation were assessed using SEM imaging. For this purpose, cells were seeded on samples as was done for MTT assay, and after 24 h, cells were fixed using 2.5% glutaraldehyde followed by 1%

10

osmium tetroxide for 2 h. After that, samples were washed with water and dehydrated in a series of ethanol (25%, 50%, 75%, and 95%) for 30 min. After complete dehydration, samples were coated with gold and examined with a TESCAN-Vega3 scanning electron microscopy (SEM). 2.10. Mechanical Properties Mechanical properties of dressings were tested using the universal testing machine (Santam STM-20, Iran). Films were cut into a rectangle (5.5 cm × 1 cm) and stretched at a rate of 20 mm/min. Ultimate tensile strength (UTS), as well as elongation at break (EB) of samples, were measured in both dry and wet states. For the wet state, samples were swollen in PBS for 24 h to reach maximum swelling before starting the test. 2.11. Statistical Analysis All tests were conducted in triplicate (n=3), and results are given in mean ± standard deviation. Statistical analysis was done using one-way ANOVA by GraphPad Prism 7.03. 3. Results and discussion 3.1. Characterization of chitosan nanoparticles Synthesized particles formed by cross-linking appeared between CS and TPP. Dissolution of chitosan in acidic solvents occurs due to protonation of amine groups in the chitosan chain. Formation of these positive charges makes a repulsive force between chitosan chains, which helps their dissolution. Besides, when sodium tripolyphosphate dissolves in water, it forms an anion with five negative charges. By addition of TPP to chitosan, positive charges of chitosan can make ionic attraction with negatively charged TPP, and consequently ionic-gelation happens

11

between these two materials as nano-CS forms and repulsion force between CS reduces (Jing et al., 2016). A graphical sketch of this reaction between CS and TPP is given in Scheme 2.

Scheme 2. Ionic interaction between chitosan and TPP. The FT-IR spectra of both CS and CS-TPP particles, shown in Fig. 1a, were recorded to investigate the chemical structure of synthesized nanoparticles. Three main functional groups can be used for characterization of pure CS. These functional groups are hydroxyl (-OH), amine (NH2) and ether which appear at 3424, 1650 and 1092 cm-1, respectively (Nawrotek et al., 2016). Although nanoparticles are formed from CS, there is some difference between their characteristic peaks in FT-IR analysis, which can be used to identify the creation of CS-TPP nanoparticles. The main difference is shifting of CS characterization peak to another wavenumber and appearance of new peaks related to the presence of the cross-linker agent (TPP) in chitosan structure. Presence of TPP in chitosan structure can be identified by its P=O groups (Wu et al., 2017). By looking at Fig.1a, it can be seen that a new peak appears at 1229 cm-1 in FT-IR spectra of CSTPP which confirms the presence of P = O groups (Wu et al., 2017). Moreover, a peak observed 12

at 1650 cm-1 related to amine groups of CS is shifted to 1542 cm-1, and a new peak appears at 1645 cm-1 related to C-O stretching bond. In CS spectra there is a strong width peak at the range of 3200 to 3600 cm-1 that depicts both N-H and O-H groups stretching vibration (Wu et al., 2005). This peak is shifted to higher wavenumbers and gets wider for nano-CS, which shows improvement in hydrogen bonding. These changes demonstrate the CS and TPP interaction to form nanoparticles and are in agreement with previous reports (Anitha et al., 2009; Luo et al., 2010).

13

Fig. 1. FT-IR spectra of (a) CS and nano-CS and (b) PU, PU/TiO2 (5% TiO2) , and sample 3 (5% TiO2, 10% nano-CS). In order to investigate the shape and size distribution of synthesized chitosan nanoparticles, SEM and DLS (Fig. 2) were used. SEM image (Fig. 2a) shows that nanoparticles have almost oval shape. Based on DLS data (Fig. 2b), an average size of 141.2 nm is obtained for nanoparticles.

Fig. 2. (a) SEM images of chitosan nanoparticles and (b) Nano-CS particle size distribution.

3.3. Characterization of wound dressings 3.3.1. FT-IR analysis Fig. 1b shows FT-IR spectra of PU, TiO2 loaded PU, and TiO2/nano-CS loaded PU films. PU film shows a peak at 3328 cm-1 relating to N-H stretching vibration (Tetteh et al., 2014). Other characterizing peaks were observed at 1704 and 1602 cm-1, which are assigned to carbonyl and

14

amide groups (Asefnejad et al., 2011; Barrioni et al., 2015). Incorporation of TiO2 into PU dressing shifted the amide peak to 1598 cm-1 due to interaction of TiO2 with amide groups of PU (Siripatrawan and Kaewklin, 2018). It also created new peaks at 614 cm-1, confirming the presence of TiO2 nanoparticles (Ahmed et al., 2018). In FT-IR spectra of sample 3, the recorded peak of N-H is shifted to higher wavenumber of 3350 cm-1 and broadened due to the addition of 10% chitosan nanoparticles and presence of O-H groups. 3.3.2. SEM images To investigate the membranes’ morphology, SEM analysis was used. Fig. 3a-e show crosssection views of PU membranes prepared by immersing the PU solution into a water coagulation bath. For wound dressing applications it is preferred that membranes have a porous structure. This porosity helps the dressing to absorb higher amount of exudates and make the wound site cleaner. In case of fabricated dressings, the porous structure is obvious from these images. Addition of nanoparticles as well as changes in coagulation bath concentration altered crosssectional morphology of dressings from a finger-like structure with perforated macrovoids to a sponge-like structure, which is even more suitable for wound dressing applications. One of the most important factors which affects membrane morphology in dry/wet technique is solvent non solvent exchange rate, as membranes with a lower solvent non solvent exchange rate show more sponge like structures. In a model represented by Young and Chen (Young and Chen, 1995), they defined diffusion ratio (k) as the flux of solvent into each layer (insolvent) to the flux of nonsolvent out of it (nnonsolvent) or: (2)

15

They used this parameter qualitatively to discuss morphology variation within membranes. It is noteworthy to mention that when a nonsolvent penetrates into the membrane, it divides the solution to a polymer poor phase covered with a polymer rich phase. This polymer poor phase is called the nucleus. This nucleus grows by absorption of nonsolvent until its border becomes rigid. Formation of macrovoids, finger-like structures, and sponge-like structures depends on them. The high affinity between bath and solvent causes the polymer poor phase to absorb a higher amount of nonsolvent until its border solidifies. Therefore, final morphology includes macrovoids and finger-like structures. On the other hand, lower affinity or driving force reduces nonsolvent penetration speed and give more time for the creation of new nuclei. Increase in nuclei number limits their growth and therefore structure shifts to a sponge shape. Moreover, it has been reported that addition of solvent to the coagulation bath can reduce the affinity of solvent to nonsolvent and therefore reduce k value. Lower k value gives time for films to create more polymer poor nuclei and consequently for macrovoids change to sponge-like structure (Kim et al., 1999). Scheme 3 graphically shows this phenomenon. Nanoparticles can also change morphology in two ways. First, nanoparticles increase the viscosity of solutions. Increasing viscosity can restrict solvent/nonsolvent replacement and reduce phase inversion speed to give more time for the nucleation process to occur. Second, nanoparticles can change hydrophilicity of composite that results in alteration of solvent/nonsolvent diffusion ratio (Razmjou et al., 2011; Yuliwati and Ismail, 2011).

16

Scheme 3. Effect of coagulation bath composition on polymer morphology.

In the PU-only sample, the coagulation bath is pure water (without solvent). Therefore, the driving force for demixing is at its highest point. Addition of DMF to the coagulation bath manipulates driving force between solvent and nonsolvent and alters k value. Comparing PUonly film with a sample containing nanoparticles but with the same coagulation bath composition (sample No. 9), shows some sponge structure at the bottom of the sample which is the result of increased viscosity due to incorporation of nanoparticles. Considering the effect of nonsolvent composition, we have observed that addition of 7.5% DMF to the coagulation bath could change the morphology of films slightly to form finger-like structures along with macrovoids and some small voids (Fig. 3b). Increasing DMF content to 15% completely changed the morphology of the films (Fig. 3c). This amount of DMF reduced the driving force of phase separation dramatically and thus diffusion of nonsolvent in and solvent out of films are within one order of magnitude. Therefore, there is enough time to create and grow new nuclei. The combination of these two phenomena results in a dominant sponge-like structure inside wound dressings. Kim et al. (Kim et al., 1999) prepared PU membranes and investigated the effect of inclusion of solvent

17

to coagulation bath on final morphology. They have observed the same structural change from macrovoids to a sponge-like structure for their membranes. However, they have used higher amounts of solvent in their coagulation bath (50%) to reach the completely sponge-like morphology. Having a film with sponge-like structure can improve its mechanical strength along with its capability for fluid uptake. As a wound dressing material, higher fluid uptake translates to a cleaner wound cite which can improve wound healing efficiency. SEM images with higher magnifications in Fig. 3e-f show dispersion of nanoparticles in the skin and cross section area of the prepared nanocomposites, respectively. It can be seen that the nanoparticles were appropriately dispersed within the film and also over the surface of the membrane. Even though some agglomeration occurred, particles were still nanoparticle-sized.

18

Fig. 3. SEM images of (a) PU-only sample, (b) sample 9 (5% TiO2, 5% nano-CS, 2.5% AA, and 100% water) , (c) sample 13 (5% TiO2, 7.5% nano-CS, 2.5% AA, and 92.5% water), (d) sample 2 (5% TiO2, 7.5% nano-CS, 5% AA, and 85% water), (e) dispersion of nanoparticles within the membrane, and (f) top surface of sample 13.

3.3.3. Swelling measurements Improvement in exudate uptake ability of modified membranes was assessed by swelling measurement in two different time intervals. The 1 h swelling shows the capability of dressings for fast absorption and addresses whether it can be used for short time usage. 24 h measurement shows its equilibrium swelling and represents its ultimate absorption capability. Test results are given in Fig. 4. High swelling capability, as stated above, is an essential character of a good

19

wound dressing. As one can notice, in 1 h measurements an immense improvement in swelling percentage is observed ranging from 124.46 to 551.66% with an average of 264.89% as compared to PU-only dressing. Based on statistical analysis, swelling percentages for high concentrations of nano-CS (>7.5%) have significant differences compared to the control sample (p < 0.05). This can be attributed to the hydrophilic nature of nano-CS, which tends to absorb water faster than hydrophobic PU and therefore reach equilibrium swelling in much shorter time. A similar behavior was observed at the 24 h measurement. At that time interval, growth in swelling was in the range of 5.28 to 245.32% in comparison with neat PU while the average increase was 71.15%. This improvement can be related to the improved hydrophilicity of PU in addition to its capability for swelling which resulted in the higher swelling ratio for prepared dressings. In addition, high amounts of DMF in the coagulation bath caused delayed demixing and formation of a sponge-like structure which causes the formation of membranes with higher porosity. A significant difference (p < 0.05) with the PU-only sample is observed for samples with the highest concentration of nano-CS and solvent in the coagulation bath. Main effect plots are represented in Fig. 5 for a detailed discussion about the impact of each parameter. Fig. 5a shows the effect of acetic acid on swelling behavior of prepared membranes. Obtained results showed that higher amounts of acetic acid resulted in higher swelling percentages. This may be due to better dispersion of nano-CS particles. Acetic acid can protonate amine groups of chitosan and create a repulsion force preventing agglomeration of nanoparticles. Effect of coagulation bath concentration on swelling can also be assessed using Fig. 5. Here, a sophisticated trend is observed. Reduction in swelling by the introduction of DMF in the coagulation bath may be caused by some dense parts which appeared within membranes while the morphology started to change. Although DMF can promote nuclei formation and hence produce a sponge-like structure,

20

it causes the polymer poor phase to absorb more solvent from its neighbor and therefore creates some dense parts. However, larger amounts of DMF (e.g. 15 %), as discussed in the previous section, could create a sponge-like structure and improve swelling capability. The water vapor transmission rate is another critical aspect of bandages. It demonstrates dressing’s ability to exchange gases. In the next section, we have studied this property for prepared wound dressings.

Fig. 4. (a) Swelling 1 h, (b) Swelling 24 h, and (c) WVTR test results for prepared samples and PU-only. Statistical analysis were done using one-way ANOVA by comparing results with control (PU-only), (*p <0.05, **p <0.01, ***p <0.001, ****p < 0.0001).

21

Fig. 5. Main effects plot of (a) acetic acid, (b) nano-CS, and (c) coagulation bath for 24 h swelling.

3.3.4. Water vapor transmission rate Fig. 4c gives the results of WVTR test. Here, WVTR of all samples reduced in comparison with PU-only dressing. Statistical analysis shows that all of the samples have a significant difference with PU sample (p < 0.0001). In addition, WVTR is in range of 891 to 1635 g/m2/day which shows that the selected parameters can regulate WVTR for specified conditions. Reduction in WVTR is caused because macrovoids and finger-like structure of neat PU sample shifted towards a sponge-like structure which can increase dead-end pores and consequently hinder water permeation. This phenomenon has also been observed previously by Deshmukh and Li (Deshmukh and Li, 1998). Fig. 6 shows the main effects plot of different parameters for WVTR. Fig. 6a shows that variation of acetic acid concentration does not have a significant impact on the water vapor transmission rate. Plots for nano-CS and DMF concentration show the same trend (reduction at first followed by an increase). This can be related to the effect of these parameters on morphology of dressings. As described before, small amounts of DMF in coagulation bath creates some dense parts, which blocks transmission of vapor while sponge structure in the

22

presence of enough amount of DMF is more desirable for transmission. Nano-CS concentration shows the same effect. The hydrophilic nature of these nanoparticles, as well as an increase in viscosity of the solution, can change the driving force for water penetration and change morphology the same way coagulation bath does.

Fig. 6. Main effect plots of (a) actic acid, (b) nano-CS, and (c) coagulation bath for water vapor transmission rate. 3.3.5. Biocompatibility In vitro cytotoxicity of nanocomposites was assessed using an MTT test. Samples 1 to 5 were taken for this test based on swelling measurements as well as to cover all the nano-CS contents in order to assess the effect of nano-CS concentration on cell viability. Results show that all samples except sample number 4 have higher O.D. value than control for 72 h. As shown by SEM images, nanocomposites have a porous structure with a rough surface which can improve cell attachment. Comparison of O.D. values for the samples depicts an increasing trend with time which indicates that cells can proliferate on the surface of dressings easily. Statistical analysis of comparing nanocomposites with PU-only sample showed that they were not significantly

23

different (P<0.05). Hence, nanoparticles did not have an adverse effect on the biocompatibility of wound dressings. Based on MTT assay results, we expect that all nanocomposites will have higher cell adhesion and proliferation properties than PU-only film. Fig. 7 shows cell adhesion and proliferation on PU-only and sample No. 2, as the best sample of MTT assay, after 24 h of cell cultivation. It is obvious that fibroblast cells maintain their normal shape with organized microfilament bundles. SEM images show that the number of adhered and proliferated cells on sample 2 is higher than that of PU-only sample. This higher cell attachment and proliferation can be attributed to higher hydrophilicity and biocompatibility caused by the addition of nano-CS particles and morphology change toward sponge structure. These images along with MTT assay show that fabricated wound dressings are completely compatible with fibroblast cells.

Fig. 7. (a) Cytotoxicity test results of PU-only and nanocomposites for 1, 2, and 3 days. SEM images of human fibroblast cells attached to (b, c) sample 2 (5% TiO2, 7.5% nano-CS, 5% AA,

24

and 85% water), (d, e) PU-only sample. Data is shown in mean ± SD (*, +, # p < 0.05, **, ++, ## p < 0.01, ***, +++, ### p < 0.001).

3.3.6. Antibacterial activity One of the main characteristics of an ideal wound dressing is antibacterial nature. This property helps prevention of infections in the wound area. The antibacterial activity of prepared dressings (samples 1 to 5) was assessed using the shake flask method (Fu et al., 2016) against a grampositive and gram-negative bacteria. Results for P. aeruginosa indicate that all of the tested samples have a good antibacterial activity with the reduction rate in the range of 63 to 69%. This close range of antibacterial effects reveals that concentrations of chitosan nanoparticles higher than 5 percent show little effect on this gram-negative bacteria. On the other hand, the antibacterial activity of tested wound dressing films on Staphylococcus aureus is dependent on the amount of nano-CS. Our results indicate that increase in nano-CS concentration or increase in positively charged amine groups improves wound dressing performance against S. aureus. Chitosan and TiO2 nanoparticles have different approaches for their antibacterial activity. While chitosan shows antibacterial activity mainly due to its positively charged amine groups, it is believed that TiO2 nanoparticles show antibacterial susceptibility because of production of reactive oxygen species or production of ions (Kong et al., 2010; Ullah et al., 2016). It is noteworthy that the dense layer at top of the dressings along with morphology change from big macrovoids to smaller sponge-like structure which may be small enough to block bacteria penetration can enhance membranes’ ability to provide a clean environment for wound healing. For example, P. aeruginosa is rod shaped with a size of almost one µm wide and 1-5 µm long;

25

therefore pores with lower dimensions can block bacteria penetration to the wound site. This change, on the other hand, may also cause entrapment of antibacterial agents within nondegradable PU film. Some of the nanoparticles are trapped in small voids and as a result, bacteria are not able to have any contact with them. This may be main reason for the moderate antibacterial activity of dressings. Results of the antibacterial tests are given in Fig. 8.

Fig. 8. Antibacterial activity data of selected samples on tested bacteria.

3.3.7. Mechanical Properties

Results of tensile test for PU-only and sample No. 2 in both dry and wet phase are depicted in Fig. 9. UTS for samples in dry condition were 4.54 MPa and 5.40 MPa for PU-only and sample 2, respectively. The nanoparticle-loaded dressing has an 18.94% higher UTS. Considering elongation at break, samples have EB in range of 300 to 400%. Higher UTS of nanocomposite comes from the interaction between nanoparticles and polymer chains. Polymer chains can twist around nanoparticles that reduce their mobility (Hong et al., 2007; Luo et al., 2012). For the PU-

26

only sample, comparing UTS and EB of wet sample with the dry one show a slight reduction of UTS and little increase in elongation at break. Based on swelling study, neat PU has low water uptake and shows a low tendency to absorb water among its chains. Therefore, water could slightly alter physical properties of films to act as a plasticizer. Addition of nanoparticles with a hydrophilic nature caused higher water uptake at the interface of nanoparticles. This “excess” water resulted in sites with higher water content in comparison with the bulk (matrix) and therefore formed a heterogeneous structure and as a result reduced mechanical stability and properties when compared with the dry state. These results also show that mechanical properties of the PU-only sample were not significantly affected by the selected parameters (nanoparticle addition and change in coagulation bath compostion). Thus, we can suppose that other samples show mechanical properties almost equal to the tested nanocomposite, which means higher UTS and EB than PU-only in dry state while these parameters may be slightly lower than those of in the wet state. Moreover, Lee et al. (Lee et al., 2016) examined the mechanical properties of commercially available PU dressings. Based on their results, UTS was in the range of 0.011 to 0.248 kgf/mm2 (0.108 to 2.43 MPa), and elongation at break was in the range of 180 to 1101%. Our results indicate that the prepared wound dressing has significantly higher ultimate tensile strength while elongation at break is within mid-range of the examined PU wound dressings.

27

Fig. 9. (a) Ultimate tensile strength and (b) elongation at break of prepared wound dressings. Data were obtained for the dry and wet condition of samples.

4. Conclusions PU nanocomposite membranes with asymmetric microstructure were fabricated using a dry/wet phase inversion technique for use in wound dressing applications. Membrane morphology was controlled by solvent/non-solvent exchange rate. Antibacterial as well as swelling properties of membranes were tailored by incorporating organic and inorganic nanoparticles to achieve an ideal wound dressing. Chitosan was converted to nanoparticles by the ionic-gelation method to improve its blending capability with the hydrophobic PU. Formation of nano-CS was confirmed by SEM, DLS, and FT-IR analyses. Based on the results, by reducing the solvent/non-solvent exchange rate, membrane microstructure shifts toward a sponge-like structure, which is more appropriate for wound dressing applications. Swelling measurements demonstrated a significant increase in swelling ratio up to 5 times in 1 h, which was mainly due to addition of hydrophilic nano-CS to the hydrophobic PU matrix. An antibacterial test also confirmed the antibacterial

28

characteristics of prepared wound dressings against both gram-positive and gram-negative bacteria. Cytotoxicity test using human fibroblast cells showed the biocompatibility of wound dressings so that prepared wound dressings had higher cell proliferation in comparison with neat PU. Among all prepared samples, sample 2 showed the greatest improvement towards an ideal wound dressing and was designated as the optimum based on all investigated parameters. Its 24 h swelling increased about 245%, showed higher VWTR compared to other samples, and demonstrated good antibacterial activity. In the mechanical test, this sample also showed proper tensile strength and elasticity in comparison with other commercial PU-based wound coverings. These improvements are related to creation of a sponge-like microstructure which has a higher exudate absorbance capability, creating small voids that block bacteria penetration as well as the presence of nanoparticles in the membranes. References: Ahmed, M.A., Abdelbar, N.M., Mohamed, A.A., 2018. Molecular imprinted chitosan-TiO2 nanocomposite for the selective removal of Rose Bengal from wastewater. International journal of biological macromolecules 107, 1046-1053. Aljohani, W., Ullah, M.W., Zhang, X., Yang, G., 2017. Bioprinting and its applications in tissue engineering and regenerative medicine. International journal of biological macromolecules, 261-275. Anitha, A., Divya Rani, V.V., Krishna, R., Sreeja, V., Selvamurugan, N., Nair, S.V., Tamura, H., Jayakumar, R., 2009. Synthesis, characterization, cytotoxicity and antibacterial studies of chitosan, O-carboxymethyl and N,O-carboxymethyl chitosan nanoparticles. Carbohydrate Polymers 78, 672-677. Asefnejad, A., Khorasani, M.T., Behnamghader, A., Farsadzadeh, B., Bonakdar, S., 2011. Manufacturing of biodegradable polyurethane scaffolds based on polycaprolactone using a phase separation method: physical properties and in vitro assay. International journal of nanomedicine 6, 2375-2384. Aung, M.M., Yaakob, Z., Kamarudin, S., Abdullah, L.C., 2014. Synthesis and characterization of Jatropha (Jatropha curcas L.) oil-based polyurethane wood adhesive. Industrial Crops and Products 60, 177-185. Barrioni, B.R., de Carvalho, S.M., Oréfice, R.L., de Oliveira, A.A.R., de Magalhães Pereira, M., 2015. Synthesis and characterization of biodegradable polyurethane films based on HDI with hydrolyzable crosslinked bonds and a homogeneous structure for biomedical applications. Materials Science and Engineering: C 52, 22-30. Biju, V., 2014. Chemical modifications and bioconjugate reactions of nanomaterials for sensing, imaging, drug delivery and therapy. Chemical Society Reviews 43, 744-764. Chen, Y., Yan, L., Yuan, T., Zhang, Q., Fan, H., 2011. Asymmetric polyurethane membrane with in situgenerated nano-TiO2 as wound dressing. Journal of Applied Polymer Science 119, 1532-1541.

29

Chung, Y.-C., Su, Y.P., Chen, C.-C., Jia, G., Wang, H.L., Wu, J.G., Lin, J.G., 2004. Relationship between antibacterial activity of chitosan and surface characteristics of cell wall. Acta pharmacologica sinica 25, 932-936. Dash, M., Chiellini, F., Ottenbrite, R.M., Chiellini, E., 2011. Chitosan—A versatile semi-synthetic polymer in biomedical applications. Progress in Polymer Science 36, 981-1014. Deshmukh, S., Li, K., 1998. Effect of ethanol composition in water coagulation bath on morphology of PVDF hollow fibre membranes. Journal of Membrane Science 150, 75-85. Di, Z., Shi, Z., Ullah, M.W., Li, S., Yang, G., 2017. A transparent wound dressing based on bacterial cellulose whisker and poly (2-hydroxyethyl methacrylate). International journal of biological macromolecules 105, 638-644. Fotiou, T., Triantis, T., Kaloudis, T., Hiskia, A., 2015. Evaluation of the photocatalytic activity of TiO2 based catalysts for the degradation and mineralization of cyanobacterial toxins and water off-odor compounds under UV-A, solar and visible light. Chemical Engineering Journal 261, 17-26. Fu, G., Vary, P.S., Lin, C.-T., 2005. Anatase TiO2 nanocomposites for antimicrobial coatings. The Journal of Physical Chemistry B 109, 8889-8898. Fu, H., Wang, Y., Li, X., Chen, W., 2016. Synthesis of vegetable oil-based waterborne polyurethane/silverhalloysite antibacterial nanocomposites. Composites Science and Technology 126, 86-93. Ghadi, A., Mahjoub, S., Tabandeh, F., Talebnia, F., 2014. Synthesis and optimization of chitosan nanoparticles: Potential applications in nanomedicine and biomedical engineering. Caspian journal of internal medicine 5, 156-161. Hassanajili, S., Khademi, M., Keshavarz, P., 2014. Influence of various types of silica nanoparticles on permeation properties of polyurethane/silica mixed matrix membranes. Journal of Membrane Science 453, 369-383. He, Q., Zhang, Y., Cai, X., Wang, S., 2016. Fabrication of gelatin–TiO2 nanocomposite film and its structural, antibacterial and physical properties. International journal of biological macromolecules 84, 153-160. Hess, C.T., 2012. Clinical guide to skin and wound care. Lippincott Williams & Wilkins, USA. Hong, R., Fu, H., Zhang, Y., Liu, L., Wang, J., Li, H., Zheng, Y., 2007. Surface‐modified silica nanoparticles for reinforcement of PMMA. Journal of Applied Polymer Science 105, 2176-2184. Jayakumar, R., Menon, D., Manzoor, K., Nair, S., Tamura, H., 2010. Biomedical applications of chitin and chitosan based nanomaterials—A short review. Carbohydrate Polymers 82, 227-232. Jing, Z.-W., Jia, Y.-Y., Wan, N., Luo, M., Huan, M.-L., Kang, T.-B., Zhou, S.-Y., Zhang, B.-L., 2016. Design and evaluation of novel pH-sensitive ureido-conjugated chitosan/TPP nanoparticles targeted to Helicobacter pylori. Biomaterials 84, 276-285. Joost, U., Juganson, K., Visnapuu, M., Mortimer, M., Kahru, A., Nõmmiste, E., Joost, U., Kisand, V., Ivask, A., 2015. Photocatalytic antibacterial activity of nano-TiO2 (anatase)-based thin films: effects on Escherichia coli cells and fatty acids. Journal of Photochemistry and Photobiology B: Biology 142, 178185. Jung, S.-M., Yoon, G.H., Lee, H.C., Shin, H.S., 2015. Chitosan nanoparticle/PCL nanofiber composite for wound dressing and drug delivery. Journal of Biomaterials Science, Polymer Edition 26, 252-263. Kamoun, E.A., Kenawy, E.-R.S., Chen, X., 2017. A review on polymeric hydrogel membranes for wound dressing applications: PVA-based hydrogel dressings. Journal of advanced research 8, 217-233. Karimi, M.B., Hassanajili, S., 2017. Short fiber/polyurethane composite membrane for gas separation. Journal of Membrane Science 543, 40-48. Karimi, M.B., Khanbabaei, G., Mir Mohamad Sadeghi, G., Jafari, A., 2018. Effect of nano‐silica on gas permeation properties of polyether‐based polyurethane membrane in the presence of esterified canola oil diol as a nucleation agent for hard segments. Journal of Applied Polymer Science 135, 45979.

30

Khil, M.S., Cha, D.I., Kim, H.Y., Kim, I.S., Bhattarai, N., 2003. Electrospun nanofibrous polyurethane membrane as wound dressing. Journal of Biomedical Materials Research Part B: Applied Biomaterials 67, 675-679. Kim, Y.D., Kim, J.Y., Lee, H.K., Kim, S.C., 1999. Formation of polyurethane membranes by immersion precipitation. II. Morphology formation. Journal of applied polymer science 74, 2124-2132. Kong, M., Chen, X.G., Xing, K., Park, H.J., 2010. Antimicrobial properties of chitosan and mode of action: a state of the art review. International journal of food microbiology 144, 51-63. Lee, S.M., Park, I.K., Kim, Y.S., Kim, H.J., Moon, H., Mueller, S., Jeong, Y.-I., 2016. Physical, morphological, and wound healing properties of a polyurethane foam-film dressing. Biomaterials research 20, 15. Lin, J., Chen, X., Huang, P., 2016. Graphene-based nanomaterials for bioimaging. Advanced drug delivery reviews 105, 242-254. Luo, Y., Zhang, B., Cheng, W.-H., Wang, Q., 2010. Preparation, characterization and evaluation of selenite-loaded chitosan/TPP nanoparticles with or without zein coating. Carbohydrate Polymers 82, 942-951. Luo, Z., Cai, X., Hong, R., Wang, L., Feng, W., 2012. Preparation of silica nanoparticles using silicon tetrachloride for reinforcement of PU. Chemical engineering journal 187, 357-366. Malathi, B., Mona, S., Thiyagarajan, D., Kaliraj, P., 2015. Immunopotentiating nano-chitosan as potent vaccine carter for efficacious prophylaxis of filarial antigens. International journal of biological macromolecules 73, 131-137. Nawrotek, K., Tylman, M., Rudnicka, K., Gatkowska, J., Balcerzak, J., 2016. Tubular electrodeposition of chitosan–carbon nanotube implants enriched with calcium ions. Journal of the mechanical behavior of biomedical materials 60, 256-266. Ovington, L.G., 2007. Advances in wound dressings. Clinics in dermatology 25, 33-38. Peng, F., Su, Y., Zhong, Y., Fan, C., Lee, S.-T., He, Y., 2014. Silicon nanomaterials platform for bioimaging, biosensing, and cancer therapy. Accounts of chemical research 47, 612-623. Pitkethly, M.J., 2004. Nanomaterials–the driving force. Materials today 7, 20-29. Podporska-Carroll, J., Panaitescu, E., Quilty, B., Wang, L., Menon, L., Pillai, S.C., 2015. Antimicrobial properties of highly efficient photocatalytic TiO2 nanotubes. Applied Catalysis B: Environmental 176, 7075. Rázga, F., Vnuková, D., Némethová, V., Mazancová, P., Lacík, I., 2016. Preparation of chitosan-TPP submicron particles: Critical evaluation and derived recommendations. Carbohydrate polymers 151, 488499. Razmjou, A., Mansouri, J., Chen, V., 2011. The effects of mechanical and chemical modification of TiO2 nanoparticles on the surface chemistry, structure and fouling performance of PES ultrafiltration membranes. Journal of membrane science 378, 73-84. Ren, G., Hu, D., Cheng, E.W., Vargas-Reus, M.A., Reip, P., Allaker, R.P., 2009. Characterisation of copper oxide nanoparticles for antimicrobial applications. International journal of antimicrobial agents 33, 587590. Sirelkhatim, A., Mahmud, S., Seeni, A., Kaus, N.H.M., Ann, L.C., Bakhori, S.K.M., Hasan, H., Mohamad, D., 2015. Review on zinc oxide nanoparticles: antibacterial activity and toxicity mechanism. Nano-Micro Letters 7, 219-242. Siripatrawan, U., Kaewklin, P., 2018. Fabrication and characterization of multifunctional active food packaging from chitosan-titanium dioxide nanocomposite as ethylene scavenging and antimicrobial film. Food Hydrocolloids, 125-134. Sudheesh Kumar, P., Lakshmanan, V.-K., Anilkumar, T., Ramya, C., Reshmi, P., Unnikrishnan, A., Nair, S.V., Jayakumar, R., 2012. Flexible and microporous chitosan hydrogel/nano ZnO composite bandages for wound dressing: in vitro and in vivo evaluation. ACS applied materials & interfaces 4, 2618-2629.

31

Tetteh, G., Khan, A., Delaine-Smith, R., Reilly, G., Rehman, I., 2014. Electrospun polyurethane/hydroxyapatite bioactive Scaffolds for bone tissue engineering: The role of solvent and hydroxyapatite particles. journal of the mechanical behavior of biomedical materials 39, 95-110. Ullah, M.W., Ul-Islam, M., Khan, S., Kim, Y., Jang, J.H., Park, J.K., 2016. In situ synthesis of a biocellulose/titanium dioxide nanocomposite by using a cell-free system. RSC Advances 6, 22424-22435. Unnithan, A.R., Barakat, N.A., Pichiah, P.B., Gnanasekaran, G., Nirmala, R., Cha, Y.S., Jung, C.H., ElNewehy, M., Kim, H.Y., 2012. Wound-dressing materials with antibacterial activity from electrospun polyurethane-dextran nanofiber mats containing ciprofloxacin HCl. Carbohydr Polym 90, 1786-1793. Woo, C.H., Choi, Y.C., Choi, J.S., Lee, H.Y., Cho, Y.W., 2015. A bilayer composite composed of TiO2incorporated electrospun chitosan membrane and human extracellular matrix sheet as a wound dressing. Journal of Biomaterials Science, Polymer Edition 26, 841-854. Wu, J., Wang, Y., Yang, H., Liu, X., Lu, Z., 2017. Preparation and biological activity studies of resveratrol loaded ionically cross-linked chitosan-TPP nanoparticles. Carbohydrate polymers 175, 170-177. Wu, Y., Yang, W., Wang, C., Hu, J., Fu, S., 2005. Chitosan nanoparticles as a novel delivery system for ammonium glycyrrhizinate. International journal of pharmaceutics 295, 235-245. Yan, L., Si, S., Chen, Y., Yuan, T., Fan, H., Yao, Y., Zhang, Q., 2011. Electrospun in-situ hybrid polyurethane/nano-TiO 2 as wound dressings. Fibers and Polymers 12, 207-213. You, C., Li, Q., Wang, X., Wu, P., Ho, J.K., Jin, R., Zhang, L., Shao, H., Han, C., 2017. Silver nanoparticle loaded collagen/chitosan scaffolds promote wound healing via regulating fibroblast migration and macrophage activation. Sci Rep 7, 10489. Young, T.-H., Chen, L.-W., 1995. Pore formation mechanism of membranes from phase inversion process. Desalination 103, 233-247. Yu, J.-G., Yu, H.-G., Cheng, B., Zhao, X.-J., Yu, J.C., Ho, W.-K., 2003. The effect of calcination temperature on the surface microstructure and photocatalytic activity of TiO2 thin films prepared by liquid phase deposition. The Journal of Physical Chemistry B 107, 13871-13879. Yuliwati, E., Ismail, A., 2011. Effect of additives concentration on the surface properties and performance of PVDF ultrafiltration membranes for refinery produced wastewater treatment. Desalination 273, 226-234.

32

33