The importance of nanofiber hydrophobicity for effective fog water collection

To increase fog collection efficiency in a fiber system, controlled wetting properties are desirable. In this work, hydrophobic (PA11) and hydrophilic (PA6) polyamides were tested to verify the surface wetting effect on fog water collection rate. Highly porous fiber meshes were obtained from both polymer solutions. Randomly oriented fibers with average diameter of approximately 150 nm were observed with a scanning electron microscope (SEM). Despite the similar geometry and zeta potential of PA6 and PA11 meshes, it was shown that the hydrophobic PA11 nanofibers are more effective at water collection than hydrophilic PA6. These results indicate that wetting properties of electrospun nanofiber mesh have a significant effect on the process of draining from the mesh, as discussed in this paper. The results obtained are crucial for designing more efficient fog water collectors that include nanofibers in their construction.


Introduction
Water collection from fog is an effective and low-cost method of water harvesting in places where access to traditional water sources is limited. [1][2][3] Commercial Fog Water Collectors (FWCs) are usually placed in foggy and windy regions, such as Chile, Morocco, Nepal and Eritrea. 4 FWCs are specially designed meshes from polyethylene (PE) or polypropylene (PP) mounted on steel stands. They consist of a double layer of mesh with total opening range from 35% to 70%. 3,[5][6][7] The mechanism of water collection is relatively simple. Fog consists of small water droplets, which are moved by the wind that randomly passes through the collectors and become trapped in the mesh. 8,9 Average diameter of water droplets can range from 17 mm at the formation stage up to 47 mm in dense fog. 10 Such small droplets need a dedicated mesh design for satisfactory water harvesting. Aer impact, droplets remain on the meshes and concentrate to larger agglomerates, 11,12 which then run down to containers under the inuence of gravity and wind. 4 Notably, for brous meshes, the remaining challenge is the frequent blockage of pores by growing droplets, which reduces their efficiency for collecting water. 12,13 Oen, lower water contact angle hysteresis results in a more efficient drainage system allowing faster removal of droplets, thus reducing the time of pore blockage. 14 Water collection solutions have been also inspired by nature e.g. Namib Desert beetles (Stenocara gracilipes) where the combination of hydrophobic-hydrophilic properties is important. [15][16][17] This combination of wetting properties in the form of Janus system is in the focus of research which has shown signicant increase in the water collection efficiency. [18][19][20] Many studies have shown the possibility and suitability of nano-bers with hydrophobic and hydrophilic properties to catch fog and water droplets from humid air. [21][22][23][24] Nanobers are commonly produced via electrospinning, where a voltage is applied between the nozzle and grounded collector. [25][26][27][28] The electrostatic eld causes elongation of the polymer solution into a jet, which moves in a spiral motion and becomes unstable due to solvent evaporation. 29,30 Instability caused by solvent evaporation usually results in a random distribution of bers that form the mesh. Micro and nanobers can also be integrated into commercially available Raschel meshes, resulting in enhanced water collection. 31,32 The mechanism of water collection in FWCs, including the draining process of deposited droplets, depends on the wetting properties of the ber surface. For ber meshes with randomly oriented bers, spreading of droplets is dictated by the surface wettability, texture of the mesh and individual bers inuencing the water contact angle. Wetting properties considering roughness are described by Wenzel regime, or its theoretical modication, for hydrophilic surfaces and in a Cassie-Baxter state, or its theoretical modication, for hydrophobic surfaces. [33][34][35] Therefore, the surface properties, size of bers and ber fraction, together with their shape and roughness, affect the wettability of a mesh, which translates to fog collection efficiency. 23 Interestingly, a combination of hydrophobic and hydrophilic properties can be used to enhance the water collection rate 36,37 of FWCs by changing the wetting and drainage system from the vertically placed meshes where the gravity has its effect. 38 Another aspect in fog collection is the surface charges commonly used in electrostatic collectors. 39 The inuence of an electric eld causes polarization of water molecules which, in consequence, accelerates their coalescence and growth. 40 This approach is based on a high electric eld, which imparts a net charge to the incoming droplets, and they are directed to the collector by electrostatic force. 41 The optimization of the brous meshes is crucial for efficient water drainage and high-water collection rate in FWCs. Water behavior in contact with material at macro, micro and nanoscale may be drastically different. 42 In this work we want to compare two materials with a signicant difference in wetting properties, but with similar average ber diameters and chemical structure (Fig. 1). Therefore, we investigated polyamides (PA): hydrophobic PA11 and hydrophilic PA6 electrospun nanobers with similar diameters. The PA meshes have similar pore size and porosity, however, the difference in the number of carbon atoms in the polymer chain structure changes the wettability of PA meshes. By investigating the mechanisms of water collection for both systems and their effects on water collection efficiency, we are able to determine, which wetting properties of materials are more favorable for efficient fog water collection for nano-scale materials at very low wind conditions.

Materials and sample preparation
Materials and electrospinning. Polyamide 6 (PA6, (C 6 H 11 NO) n , M w ¼ 24 000 g mol À1 ; BASF, Germany) and polyamide 11 (PA11, (C 11 H 21 NO) n , M w ¼ 204.31 g mol À1 ; Sigma Aldrich, USA) (see Fig. 1e) were used to prepare electrospinning solutions with concentrations of 12% and 6%, respectively. To evaporate water from the hygroscopic PA6 granules, they were dried at 40 C until constant weight was obtained. The polymer was then dissolved in a mixture of formic and acetic acids in 1 : 1 volume ratio (CH 2 O 2 > 98%; C 2 H 4 O 2 > 99.5%, Avantor Performance Materials Poland S.A., Poland) at T ¼ 25 C, according to previous protocols. 23,31,43 The PA11 was dissolved in formic acid at T ¼ 60 C. The solutions were stirred for 4 h at a constant speed of 500 rpm for PA6 and 700 rpm for PA11 (RCT basic, IKA, Germany).
PA nanobers were produced via electrospinning in a chamber with environmental control (IME Technologies, The Netherlands). 44 The electrospinning parameters are listed in Table 1. A hypodermic injection needle was used as a nozzle (hypodermic injection needle KD Fine 0.8 Â 40 mm 21 G Â 1 1/ 2 00 green, KD Medical GmbH Hospital Products, Germany). The meshes were electrospun on a rotating drum (diameter ¼ 9 cm, length ¼ 18 cm) spinning at 10 rpm. Fibers were electrospun onto baking paper for easier removal of the mesh from the collector. For static contact angle (q S ) measurement, the samples were deposited on glass slides. Additionally, thin lms for the q S measurements were prepared by placing 0.1 ml of solution on a glass slide (16 Â 16 mm) and spin-coating (L2001A v.3, Ossila, Sheffield, UK) at a rotation speed of 1000 rpm for 30 s. Then, lms were dried in a fume cupboard for 24 h at 25 C.

Mesh characterization
SEM and image analysis. A scanning electron microscope (SEM, Merlin Gemini II, ZEISS, Germany) was used to analyze nanober and lm morphology. Prior to imaging, samples were coated with a 10 nm gold layer using a rotary pump sputter coater (Q150RS, Quorum Technologies, UK). The SEM imaging was performed with an accelerating voltage of 2.5 kV and current of 110 pA at a working distance of 7 mm. ImageJ soware (version 1.50i, National Institutes of Health, USA) was utilized to determine nanober diameters and calculate ber (F f ) and pore (P f ) fractions from SEM macrographs. The sum of F f and pore P f should reach about 100%. The F f supports the similar information about the meshes as the commonly used shade coefficient in FWC. 5,13 The average nanober diameter was calculated from 100 measurements presented in histograms prepared using OriginPro (2018b, OriginLab, USA). The fraction of ber and pore analysis were performed using the particle function in ImageJ based on the images showed in Fig. 1c and d. The pore size was calculated from SEM binary images prepared using the Li thresholding method in ImageJ. 45,46 The threshold was set to 0-90 for P f and pore size calculation. The F f was calculated from inverted binary 2D images.

Wetting properties
Static contact angle. The static contact angle (q S ) was measured in the horizontal position using the sessile drop method at 25 C and 40% RH. The deionized water (DI, Spring 5UV purication system -Hydrolab, Poland) was applied as 3 ml volume droplets. The images were taken by a camera with macro lens (EOS 700D, EF-S 60 mm f/2.8 Macro USM, Canon, Japan) 3 s aer the droplets were placed. The q S was determined from the images using ImageJ.
Streaming zeta potential. A high-end electrokinetic analyzer was used for zeta potential analysis (SurPASS 3, Anton Paar, Austria). The streaming potential was measured between two meshes with dimensions 20 Â 10 mm placed in the cell with the adjustable gap set to 110 mm. The pH value was controlled in acidic (5.6-3.0) and basic (5.6-8.0) ranges with pH steps of 0.3. The titrations were done by the progressive addition of 0.05 M HCl or 0.05 M NaOH to a 0.01 M KCl solution. The zeta potential measurement was repeated 4 times at every pH value with two rinse cycles performed before each test.
Fog collection experiment. The meshes used for fog collection (10 Â 10 cm) were placed in a specially designed setup. 31,32,43 Fog was produced by a humidier (Beurer GmbH, Germany) at a ow rate set to 400 ml h À1 . The humidier output was set at 90 angle to, and 6 cm distance from, the vertically placed mesh, which led to a humidity of 95% during tests. 43 This is a typical setup used in the laboratory conditions to simulate fog, 41,47 which is characterized with the water droplet size of 250 nm and above. 48 The experiments are performed in very low wind conditions, with the fog ow velocity of 0.19 m s À1 . The water recovered from fog was collected in a beaker placed underneath the mesh and weighed every 30 min over 3 h of fog collection. The amount of water collected was obtained by normalizing the mass of water to the area, and the water collection rate was calculated by dividing the water collected by time. 14,43 The pH of collected water was measured using the pH-metric electrode from the electrokinetic analyzer (SurPASS 3, Anton Paar, Austria).

Material characterization
The randomly oriented electrospun polyamide meshes observed with an SEM showed similar average ber diameters of 150 AE 20 nm and 151 AE 30 nm for PA6 and PA11, respectively ( Fig. 1a and b). The distributions of ber diameter are presented as histograms (Fig. 1f). The analysis of ber and pore fraction based on the 2D SEM images conrmed the geometrical similarity of PA6 and PA11 meshes ( Fig. 1c and d), as the F f and P f for the two meshes were almost identical (Fig. 1g). Due to the small ber diameter, the PA meshes presented a closely packed structure when observed by the SEM, Fig. 1a and b, which extended to nearly 70% of the membrane area. The analysis of pore size resulted in an average value of 9800 AE 2135 nm 2 for PA6 and 10 800 AE 2135 nm 2 for PA11. Importantly, this 2D analysis can be extended to the 3D investigation using the dual beam microscopy based on focused ion beam (FIB) and SEM to obtain the 3D reconstructions of electrospun bers network. 49 The previous study using the 3D analysis based on FIB-SEM microscopy found that the porosity reached 96% for similar PA6 membranes. 50 Importantly, the different wetting of PA6 and PA11 was conrmed by the static contact angle measurements 23,51-53 (Fig. 1h). Contact angles measured on the lms were 27 AE 2 for PA6 and 102 AE 4 for PA11. Measurements on electrospun bers showed an increase in contact angle for both polymers to 42 AE 3 and 123 AE 3 for PA6 and PA11, respectively. A similar value for electrospun PA6 has been reported previously. 43 Increase of contact angle was caused by geometry and roughness effects of the wetted bers, 22,23 which falls in line with Wenzel and Cassie Baxter models. The chemical analyses of both polyamides have previously been investigated for electrospun PA6 (ref. 54) indicating the characteristic peaks at the wavenumber range 1000-3500 cm À1 . The amide I and II peaks found between 1500 and 1700 cm À1 , and at 3300 cm À1 and 2850-2950 cm À1 represent hydrogenbonded N-H stretching and CH 2 asymmetric and symmetric stretching. For PA11 (ref. 55) the peaks are in the wavenumber range 2600-3600 cm À1 , where N-H stretching amide I is observed at 3300 cm À1 . PA6 nanobers were also characterized in term of their surface free energy showing enhanced wetting behavior in comparison to PA6 lms. 53

Zeta potential measurement
To verify the surface charge effect of electrospun meshes on the water collection efficiency, streaming zeta potential measurements were performed. The zeta potential analysis showed a decrease in the streaming potential with an increase of pH for both types of polyamide samples (Fig. 1i). The isoelectric points for hydrophilic and hydrophobic meshes were similar and occurred at pH ¼ 4.94 for PA6 and pH ¼ 4.89 for PA11. Only at these pH values, does the zeta potential of the meshes equal 0 mV. 56 Such behavior is important as the fog contains air pollution, which inuences pH. A typical range of pH for fog water is 3.5-6. 3,57,58 The pH range chosen in this study allowed prediction of the electrical potential of meshes in contact with natural fog water with known pH. The electrostatic interactions with water were the same for both PA meshes. The zeta potential measurements were performed to conrm the electrical neutrality of PA meshes and eliminate the possibility that the surface charges have an impact on the water collection efficiency for each case.

Fog water collection
Water collection curves ( Fig. 2a and b) for PA11 increased faster than for PA6, as reected in water collection rate. The pH of collected water was 5.9, because the fog can absorb the CO 2 from the air. Water collection rate including water collected from mesh and retained between bers was 41 AE 2 mg cm À2 h À1 for PA6 and 58 AE 2 mg cm À2 h À1 for PA11. Hydrophobic PA11 mesh was more efficient at water collection than hydrophilic PA6. These results show that higher values are due to the hydrophobic material as the average ber diameters and pore sizes were similar. However, blocking of pores by water droplets was observed for hydrophilic PA6. The observation of water droplets remaining on the mesh during fog collection (Fig. 2c  and d) shows different wetting properties of the two materials investigated. Hydrophilic PA6 nanobers allow permeation of water into empty spaces between them. Due to the hydrophobic nature of PA11, the droplets can drain more easily to the beaker due to reduction of blocking in the pores during fog droplet collection. The analysis of the SEM micrographs, ber fractions and zeta potential results indicated similar ber diameter and surface potential for the PA meshes. This suggests that the water collection of the two PA meshes is mainly dependent on the wetting properties of the material, which is inuenced by their chemical structure. The number of carbon atoms in a molecule is greater for hydrophobic PA11 (11 at. C) than for hydrophilic PA6 (6 at. C). This is consistent with the expected increase in a given q s for a given polyamide with the number of carbon atoms in a single molecule of the polymer chain. 51 The water collection rate obtained for PA6 is similar to results from a previous study for hydrophilic cellulose acetate brous meshes with average ber diameter of 0.54 AE 0.16 mm, which reached 45 mg cm À2 h À1 . 59 Also, the water collection rate for hydrophobic polystyrene (PS) microbers with average diameter of 4.80 AE 0.22 mm achieved 59 mg cm À2 h À1 , 59 which is comparable to PA11. However, the water transport from PS bers was problematic due to high P f (60 AE 3%). Additionally, the PS mesh had much larger pore size of 67.59 AE 8.30 mm 2 , 59 which is three orders of magnitude higher than for PA11 mesh. The large spaces between the microbers trap the water droplets for longer, slowing down the drainage, which decreases the water collection efficiency. 4,13 The clogging of mesh was reduced with a thinner ber layer, similarly to the system inspired by plants, 60 where water collection depends on the thickness of a thin brous layer. The homogeneous PA11 nanober mesh investigated in this work allows faster water drainage from the mesh while maintaining high collection efficiency.
The shape of the water droplets shown in Fig. 2c-h clearly indicates the different wetting and drainage mechanism of PA meshes. The side view images of water droplets on the vertical meshes during the fog collection (Fig. 2e for PA6 and Fig. 2h for PA11) indicate clear variations in the contact angle hysteresis, which is approximately 3 for hydrophilic PA6 and 21 for hydrophobic PA11 on representative images. This drastic difference in wetting behavior resulted in different water removal processes for the samples measured. It is also observed in the retained water aer water collection experiment, where for hydrophobic PA11 is higher than for hydrophilic PA6, see Fig. 2b. Importantly, the number of droplets collected on PA11 bers is higher than on PA6 bers, as indicated on the images in Fig. 2c and f. The water has not managed to ow down to the beaker as more water droplets are collected on the mesh, see Fig. 2f. Fig. 3 The polyamide meshes (a-d) PA6 and (e-h) PA11 before water collection, 0 h, and after 1, 2 and 3 h of fog collection.
Using polyethylene terephthalate (PET) bers, Azad et al. 61 showed a directional water droplet transport on the surface of mesh, proving the effect of surface structures on the efficiency of fog collection. The hierarchical structure of hydrophobic PVDF bers improved the water collection efficiency of the membrane. PVDF bers with average diameter of 650 AE 25 nm were modied to obtain nanopillars on their surface. The nanopillars were 140 AE 20 nm high and had a diameter around 150 nm, which is comparable with the PA6 and PA11 nanober diameters in this work. The water collection rate for just neat PVDF nanobers and modied PVDF nanobers with nanopillars reached 27 mg cm À2 h À1 and 81 mg cm À2 h À1 , respectively. The investigated here hydrophobic PA11 mesh had a lower water collection rate than the PVDF with nanopillars, but higher than the just neat PVDF bers with a 4 times larger diameter presented in the work of Ganesh et al. 62 This suggests that solutions based on materials in the nanometric scale allow an increased fog collection rate. The nano-roughness of electrospun bers is known to increase their hydrophobicity, 63,64 which is crucial to obtain a faster water drainage mechanism, thus increasing water collection efficiency. The fog collection mechanism based on the hydrophobic PA11 allows a higher water collection rate to be obtained due to the faster removal of water droplets. These ndings are opposite to what was observed in a previous study on microber meshes. 43,59 The water capture and blockage effect can also be observed on meshes containing, not only nanobers, but also microbers. The PS-PA6 composites accumulated a similar amount of water as PS. The addition of PA6 nanobers ameliorated poor drainage and the water was running off the mesh instead of clogging the pores. 43 In Fig. 3 the photographs of water collected on PA6 meshes show a poor drainage system opposite to PA11 meshes, where the water droplets form dripping channels. However, the hydrophilicity of PA6 takes the advantage in attracting water molecules as it has been discussed in other systems. 31 Here, we show that in terms of attracting water molecules in the electrospun mashes the main effect increasing their water collection efficiency is related to the sufficient drainage system, leaving the space for the next water droplet to be captured.

Conclusions
The water collection properties of two meshes were measured. The hydrophobic PA11 mesh had a 40% higher fog water collection rate than the hydrophilic PA6 mesh, despite the fact that both materials had a similar value of the streaming zeta potential in the given pH range. The wetting properties of polyamides are the main factor inuencing the water collected as both meshes have similar ber diameters. The surface chemistry driven by the polymer structure and the amount of carbon in the polymer chains determines the wetting properties of polyamides. The results obtained indicate the importance of wetting properties of the material selected for mesh production and the effect of hydrophobicity at the nanoscale. The fog collection mechanism based on the hydrophobic PA11 allows a higher water collection rate to be obtained due to the faster removal of water droplets. These ndings are opposite to what was observed in a previous study on microber meshes. 43 Fundamental knowledge for the further development of fog collectors, including nanotechnology, is provided, as there is a growing need to seek novel approaches to the global water crisis.

Data availability
Any additional data from this work are available from the corresponding author upon reasonable request.

Conflicts of interest
The authors declare that they have no known competing nancial interests or personal relationships that could have appeared to inuence the work reported in this paper.