Dual-scale TiO2 and SiO2 particles in combination with a fluoroalkylsilane and polydimethylsiloxane superhydrophobic/superoleophilic coating for efficient solvent–water separation

Surfaces that have unique wettabilities and are simultaneously superhydrophobic with water contact angles > 150°, and superoleophilic with oil contact angles < 5°, are of critical importance in the oil/solvent–water separation field. This work details the facile preparation of highly efficient oil–water separation devices that successfully combine hierarchical surface roughening particles and low surface energy components with porous substrates. Coatings were generated using TiO2 and hydrophobic-SiO2 micro/nanoparticle loadings which were then embedded within polydimethylsiloxane, commercially known as Sylgard® 184, and 1H,1H,2H,2H-perfluorooctyltriethoxysilane (FAS) polymer mixtures. The resulting slurries were dip coated onto copper meshes with varying pore diameters (30, 60 and 100 meshes had 595, 250 and 149 μm pore dimensions respectively). Functional testing proved that mesh substrates coated in the lowest Sylgard® 184 : FAS polymer ratio formulations displayed heightened water repellency and retained their superoleophilic properties upon repeat testing. The largest average water contact angle of 145 ± 1°, was recorded on a copper 30 mesh substrate with a coating comprising H-SiO2 microparticles and TiO2 nanoparticles in a 1 : 9 polymer mixture of Sylgard® and FAS. The coating's extreme oil affinity was supported by high solvent–water separation efficiencies (≥99%) which withstood numerous testing/washing cycles.


Introduction
The increasing volume of oil and/or solvent contamination in industrial wastewater is a lasting contribution to the deterioration of our marine and terrestrial ecosystems. The food, leather, textile and petrochemical industries all generate signicant amounts of unwanted oily water during daily production. 1 When these continuous waste outputs couple with freak accidents, such as oil spills, the natural environment suffers severe implications for up to one decade aer each isolated event. 2 During the period between 2010 and 2017 there was more than 47 000 tonnes of petrochemical oil spilt in aquatic environments, 80% of this amount was lost in just ten incidents. 3 It is therefore of great importance to engineer efficient oil-water separation techniques capable of rapidly cleaning the environment in the event of future spills and to decontaminate oily industrial waste.
Interactions between liquids and surfaces are central to this eld of research. Superhydrophobic materials oen make use of nano and/or microscale protrusions combined with very low surface energy modications to achieve static water contact angles $ 150 . 4,5 Water droplets upon these materials can exist in a Wenzel state, whereby the droplets 'stick' to all parts of the surface, or Cassie-Baxter regime where droplets 'slip' on a lubricating air layer trapped within surface asperities. 6,7 The literature details various ways of fabricating novel water repellent surfaces inspired by nature's waterproof materials such as lotus leaves buttery wings and geckos' feet. 5 Plasma enhanced chemical vapor deposition (PECVD) was carried out by Bico et al. to generate carbon ake-like multiscale surface structures. 8 Aerosol-assisted chemical vapour deposition (AACVD) of Sylgard® 184 or polydimethylsiloxane (PDMS) and tetraethyl orthosilicate (TEOS) afforded appropriately rough and transparent coatings on glass substrates. [9][10][11] Alternative superhydrophobic generating methods such as chemical etching, 12 electrospinning 13,14 and lithographic imprinting/templatation 15 have enhanced the functionality of oil-water separation, selfcleaning, anti-icing and anti-corrosion materials. 12,[16][17][18] Whilst water repellency is crucial to certain oil-water separation devices, materials with two wettability regimes have been known to improve separation efficiency. A surface displaying both superhydrophobic 'water-hating' and superoleophilic 'oilloving' properties can be tuned for selective separation providing its surface energy resides between that of the solvent/ oil and water it aims to separate (approximately between 30 mN m À1 and 70 mN m À1 ). 19 Filtration and/or absorption are successful types of oil-removing techniques that encompass both size exclusion and wettability selectivity mechanisms. Pore sizes are commonly designed to either let oil run straight through the mesh/membrane or for the material to absorb the hydrocarbon whilst water is repelled. 20 Copper meshes have been etched with nitric acid and alkali (NaOH/K 2 S 2 O 8 ) solutions before modication with selfassembled monolayers of hexadecanethiol and dodecanethiol respectively. High water contact angles, $150 , and low oil contact angles, <5 , produced materials that separated at least 97% of water from an oily solution. 21,22 Vertically-aligned multiwalled carbon nanotubes were synthesised on a stainless steel mesh via CVD. Needle-likes tubes created a high surface area and aided capillary action; the oil contact angle was recorded at 0 and oil penetrated the mesh in 0.4 s. 23 Li et al. fabricated candle soot and silica coated meshes that repelled hot water and corrosive liquids. These coated meshes worked as effective gravity separation devices with $99% of organic solvents permeating through the device. 24 Nano protrusions have also been introduced by electrochemical etching, a copper mesh was anodized in a NaOH solution (1 M) to create a needle-like Cu(OH) 2 lm. The roughened substrate was then coated in a self-assembled monolayer of 1H,1H,2H,2H-per-uorooctyltriethoxysilane (FAS) to form a durable solvent/water separator. 25 Spin coating metal mesh substrates with ZnO nanorods has been yet another effective fabrication method. 26 Swapping a porous substrate for porous structures (on special materials) enhances robustness by removing the possibility for coating stripping. Tu et al. designed a micro-bead and nanober lm created in one step by spraying onto any given substrate. The water contact angle on these surfaces was not as high as other but the cost and fabrication time were drastically reduced. 27 A thermoplastic polyurethane mat, created via electrospinning, immersed in a hexadecyltrimethoxysilane (HDTMS) modied nanosilica solution also generated an efficient structured separation material. 28 Several other surface modication methods such as electrostatic deposition, graing polymerisation, spray drying and photo-initiated polymerisation have gained traction by utilising various polymers, gels and biomaterials. 29,30 Herein we present an extremely facile one-pot method to fabricate highly efficient oil-water separation devices, >99% efficiency. Combinations of hierarchical surface roughening particles and low surface energy components upon porous substrates lead to enhanced robustness and functionality when tested with a variety of common industrial solvents; this elevated device effectiveness across a substrate and solvent range has substantially advanced work in this area. Here, TiO 2 and hydrophobic-SiO 2 micro/nanoparticles were embedded within polydimethylsiloxane (Sylgard® 184) and 1H,1H,2H,2H-peruorooctyltriethoxysilane (FAS) polymer mixtures. Resulting slurries were dip coated onto copper meshes with varying pore diameters (30, 60 and 100 meshes had 595, 250 and 149 mm pore dimensions respectively). Water contact angles, solvent separation efficiencies and mechanical stability were found to outperform many of the existing optimum superhydrophobic/ superoleophilic ltration surface previously documented.

Preparation of polymer stock solutions
Sylgard® 184 curing agent and elastomer base in a 1 : 10 mass ratio (30.00 g combined mass) were dissolved in ethanol (150.00 g), Fig. 1a. The polydimethylsiloxane polymer was magnetically stirred for 2 hours. A colourless, viscous mixture was achieved.

Pre-functionalisation of SiO 2 particles (H-SiO 2 )
Due to the innate hydrophilicity of unrened silica, 0.5-1.0 mm diameter SiO 2 particles (5.00 g) and 5-15 mm diameter SiO 2 particles were sonicated (40 kHz) for 30 min in the uoroalkylsilane (FAS) stock solution (25 mL). The cloudy mixture was oven dried at 60 C for 2 hours to allow for complete solvent evaporation, a ne white hydrophobic-SiO 2 (H-SiO 2 ) powder remained. The TiO 2 particles did not require this prefunctionalisation step.
For ease of reference the following particle combinations and loadings (embedded in a polymer mixture) have been abbreviated, Table 1. Optimised functional coatings have also been labelled, Table 2.

Characterisation
X-ray photoelectron spectroscopy (XPS) was performed on a Thermo Scientic XPS K-Alpha X-ray Photoelectron Spectrometer with a monochromated AL Ka X-ray source at 1486.6 eV. Fourier transform infrared (FT-IR) spectroscopy was carried out using Bruker Alpha Platinum-ATR equipment (650 to 4000 cm À1 ). Surface morphologies were investigated using a JEOL JSM-6301F scanning electron microscope (SEM) with an acceleration voltage of 5 or 10 kV.

Functional testing
Three water contact angles were measured per coating at ambient temperature via the sessile-drop method using a FTA 1000 optical contact angle meter (5 mL water droplet). An average value and associated error were calculated for each sample. The tilting angle, dened as the angle at which a water droplet readily slides off a slanted surface (xed droplet volume of 0.5 mL), was recorded using a digital angle nder. Averages and standard deviations were calculated.
According to the experimental setup in Fig. 2, solvent-water separation efficiencies were carried out by rapidly pouring oily mixtures over the selected separation device and recording the mass of solvent collected aer 30 s (later converted to a percentage of the total solvent content to be separated). Three repeat readings were recorded per separation solution and the initial solvent mass present in the 10 mL mixture was known. One part solvent three parts water (25%), solvent and water in equal parts (50%) and three parts solvent one part water (75%) mixtures of toluene-methylene blue water, hexane-methylene blue water and dichloromethane-methylene blue water (total volume, 10 mL) were the chosen testing solutions.  Meshes were washed with ethanol (5 mL) before and aer same solvent separations and new functional meshes were used when solvents were interchanged. Methylene blue water was utilised to prove no water had permeated through the separation mesh as the blue coloured water was easily distinguished from the colourless solvent. Sample durability was further assessed using the Scotch tape test and acid and base baths: Scotch tape was rmly applied to cover the whole sample surface and then ripped/removed at speed 3 times.

Results and discussion
A one-pot synthesis was used to generate a range of highly functional 'water hating' and 'oil loving' coatings for application in oil-water separation. Various iterations of TiO 2 and hydrophobic-SiO 2 (H-SiO 2 ) micro/nanoparticles were dispersed in Sylgard® 184 and 1H,1H,2H,2H-peruorooctyltriethoxysilane (FAS) mixtures, as detailed in Tables S1, † 1 and 2. The resulting coatings were dip coated onto 3 types of copper meshes (30, 60 and 100 meshes had 595, 250 and 149 mm pore dimensions respectively).

Characterisation
X-ray photoelectron spectroscopy (XPS) was carried out on the functional coatings to identify the oxidation state of Ti and chemical environments surrounding Si atoms, Fig. 3. Elemental scans of the dried optimised TiO 2 particle containing coatings (D, E, F and G (formulations listed in Table 2)) conrmed that Ti was constantly in the +4 oxidation state; a 2p doublet was characteristic of the 459.00 eV Ti 2p 3/2 and the 464.70 eV Ti 2p 1/2 peaks. Additionally, optimised coating formulations D, E and G also contained SiO 2 particles. For sample D, consisting of SiO 2 5-15 mm particles (0.6 g) and TiO 2 21 nm particles (0.6 g) in a 1 : 9 polymer mixture of Sylgard® 184 and FAS, the Si 2p peak was deconvoluted into a 104.00 eV Si-O 2 and a 101.82 eV Si-OR environment. Values agreed with the literature and were consistent with other TiO 2 and SiO 2 particle containing samples. [31][32][33][34] The X-ray diffraction pattern of coating D displays a high intensity TiO 2 anatase peak at 25.4 and subsequently less intense peaks at 36.7 , 37.8 , 38.8 , 47.9 , 54.6 , 55.1 and 62.5 . A medium intensity rutile peak was identied at 27.4 and SiO 2 particles were attributed to the quartz peak at 20.8 . As coating F was absent of SiO 2 particles the only detectable peaks were characteristic of the anatase and rutile phases of TiO 2 . All coating patterns presented in Fig. S1 † agreed with particle standards and the Inorganic Crystal Structure Database.
Fourier transform infrared (FT-IR) analysis, consistent with the NIST Standard Reference Database, resulted in the identi-cation of several bending and stretching modes characteristic of the Sylgard® 184 and 1H,1H,2H,2H-per-uorooctyltriethoxysilane (FAS) polymer mixtures, Fig. 4. Samples containing hydrophobic-SiO 2 (H-SiO 2 ) and/or TiO 2 particles embedded in 1 : 0, 0 : 1, 1 : 1, 1 : 4 and 1 : 9 Sylgard® 184 : FAS ratios all had similar polymer transmittance bands. A sharp peak at 3120 cm À1 was seen in the spectra of coatings D, E, F and G which depicted the C-H alkane stretch present in both polymers. Varying intensities of the following peaks were observed: 1257 cm À1 (sh, w) intense out of phase vinyl ether stretch in Sylgard® 184, 35 1010 cm À1 (s) Si-OR Sylgard® 184 and FAS stretch, 790 cm À1 (m) alkene out of plane bending in Syl-gard® 184, 622 cm À1 (w) C-F FAS stretch and 457 cm À1 (w) antisymmetric Si-(CH 3 ) Sylgard® 184 stretch. 35 The presence of these C-F FAS stretches were synonymous with functional effectiveness; uorine's extreme electronegativity made it only weakly susceptible to eeting dipoles that form the basis of van der Waals forces. Consequently, the samples' uorinated carbon chains had small intermolecular forces and therefore the low surface energy hydrophobic requirement. Scanning electron microscopy (SEM) images provided visual information about the topography of separation materials, particle size/distribution and the extent of copper mesh substrate pore blockage. This qualitative evidence was attributed to water contact angle measurements as well as separation efficiency results. Initial trial coatings containing TiO 2 particles in high ratios of Sylgard® 184 to FAS (1 : 1 or 1 : 0) resulted in almost complete pore blockage, Fig. S2. † As expected, pore obstruction physically impeded the passage of any solvent through the membrane and resulted in poor separation efficiencies < 25%. Coatings with higher proportions of the more viscous Sylgard® 184 polymer were also seen to reduce the prominence of particle roughening structures thus had reduced average water contact angle values, <110 . Fig. 5a presents SEM images of the unobstructed pores of a 30 mesh (595 mm pore dimension) copper substrate functionalised with coating D (SiO 2 5-15 mm particles (0.6 g) with TiO 2 21 nm particles (0.6 g) in a 1 : 9 polymer mixture of Syl-gard® 184 and FAS). Dual scale surface structures were accentuated in this coating due to the reduced quantity of Sylgard® 184 and were reected in the high average water contact angle, 145 AE 1 . However, solvent-water separation testing indicated that the pores were too large to selectively lter solvents from water and so the use of copper 30 mesh substrates were discontinued. Fig. 5b displays a section of coating D adhered to a copper 60 mesh substrate (250 mm pore dimension). Once again, imaging proved that pore openings were clear. The average water contact angle remained relatively high, $135 , and the toluene-water separation efficiency in equal parts was recorded as 100%. Fig. 5c shows images of coating E (SiO 2 5-15 mm particles (0.6 g) with TiO 2 21 nm particles (0.6 g) in a 0 : 1 polymer mixture of Sylgard® 184 and FAS) on copper 100 mesh substrates (149 mm pore dimensions). Open pore structures were conserved, due to the absence of Sylgard® 184, and there was a clear presence of protruding micro and nanoscale particles. The associated average water contact angle and toluene-water separation efficiency in equal parts were 147 AE 1 and 99 AE 1% respectively. Lastly, Fig. 5d displays coating F on a copper 100 mesh substrate (TiO 2 60-200 nm particles (1.5 g) with TiO 2 21 nm particles (1.5 g) in a 1 : 4 polymer mixture of Sylgard® 184 and FAS). The small pores were $50% blocked by this polymer mixture making this an unsuitable solvent-water separation device despite preserving dual scale surface contours, average water contact angle of 146 AE 1 . SEM analysis lead to the assumption that coatings with high proportions of the non-viscous uorinated FAS polymer mixture created highly structured surface topographies and maximised the appropriate pore area of copper 60 mesh (and in some cases copper 100 mesh) substrates. This in turn elevated the average contact angle values and enhanced efficiency of solvent separation; low Sylgard® 184 levels were tolerated to preserve device coating robustness.

Functional testing
Average water contact angle results are plotted in Fig. 6 to demonstrate the impact of changing particle loading/ combination and polymer mixture on hydrophobicity. The highly durable, viscous and adhesive Sylgard® 184 polymer had to be delicately balanced with the easily abraded, non-viscous and highly water repellent uoroalkylsilane (FAS) polymer to achieve a formulation that wouldn't compromise membrane pore size (for effective solvent-water separation) but maintain functionality and durability. Coatings were prepared with 1 : 0, 0 : 1, 1 : 1, 1 : 4 and 1 : 9 Sylgard® 184 : FAS polymer ratios with embedded hydrophobic-SiO 2 (H-SiO 2 ) and/or TiO 2 loading combinations and functionally contrasted. With the majority of polymer combinations dual scale H-SiO 2 particles produced the lowest average water contact angle values, for example 125 AE 10 was obtained with the 0 : 1 Sylgard® 184 : FAS polymer ratio mixture. Dual scale TiO 2 particles consistently generated the highest average water contact angle results for each coating composition (with the exception of the Sylgard® 184 : FAS ratio of 1 : 0), a maximum of 151 AE 7 was achieved with the 1 : 9 Sylgard® 184 : FAS polymer mixture. The coatings that contained the highest concentration of Sylgard® 184 oen performed comparatively worse than mesh coatings with a larger FAS content. These ndings were expected, the abundance of -C-F polymer bonds in FAS is known to effectively reduce a material's surface energy and consequently increase the average water contact angle value. All samples were in the Cassie-Baxter regime, assumed from low tilting angle results, and improved upon the hydrophobicity of separation meshes detailed in the literature. Fore example. Xue's group generated a superhydrophilic mesh whilst Zhang's coated mesh afforded average water contact angles < 130 . 36,37 Average percentage of toluene separated from 25%, 50% and 75% toluene solutions (75%, 50% and 25% water respectively) were contrasted on coated 60 mesh copper substrates, via the setup in Fig. S3. † The H-SiO 2 and TiO 2 particle combination, B, embedded in a 1 to 9 Sylgard® 184 : FAS mixture was the most effective at separating 25% toluene-50% water solutions on copper 60 mesh substrates (95 AE 6% separation efficiency). Fig. 7 illustrates that all H-SiO 2 /TiO 2 particle combinations, A, B and C, reached 100% separation efficiencies for 50% toluene-50% water solutions when coupled with their optimised Syl-gard® 184 : FAS polymer mixture. Subsequently, high separation efficiencies of 99 AE 1% were achieved on dual scale H-SiO 2 as well as H-SiO 2 /TiO 2 containing particle systems embedded in a 1 to 9 Sylgard® 184 : FAS mixture when tested with 75% toluene-25% water solutions. All particle and polymer mixture combinations had slightly reduced separation efficiencies for 25% toluene-75% water solutions but remained >90%.
Final separation efficiencies on optimised coatings on copper 60 and 100 separation meshes have been recorded in Table S2 Table S2 † documents solvent separation efficiencies for toluene-, hexane-and dichloromethanewater solutions along with associated errors.
The results indicated that coated copper 60 mesh substrates again were most likely to generate a favourable separation efficiency when compared to coated copper 100 mesh substrates. The larger pore diameter remained unblocked aer coating with even the most viscous of polymer mixtures and therefore allowed toluene, hexane and dichloromethane to lter through the mesh; small pore 100 mesh substrate blockages most strongly inhibited the densest solvent, dichloromethane, from ltering through the device and consequently has not been featured in Table S2. † Despite the exceptional performance of all optimised coatings D-G, formulation F on 60 mesh substrates was an extremely effective separation coating with 100 AE 0%, 85 AE 0% and 97 AE 0% efficiencies for toluene-, hexane-and dichloromethane-water solutions respectively. The innately hydrophobic dual scale TiO 2 particles embedded coating F's 1 : 4 Sylgard® 184 : FAS mixture helped elevate the average water contact angle, 146 AE 1 . This favourable particle combination coupled with the relatively high proportion of FAS in the coating elevated average contact angles and improved separation potential. Various other separation materials that used longer more elaborate fabrication procedures have been well documented over recent years. Separation efficiencies $ 95% were achieved but the vast number of readily available oils/ solvents and the absence of a standardized separation method makes it difficult to directly contrast like results from across the eld. [36][37][38][39][40][41] Due to the adhesive properties of Sylgard® 184, optimised coatings D, F and G remained comparably functional, within the original error limits represented in Fig. 6, aer 3 washseparation-wash cycles and the Scotch tape test. Fig. 8 displays the water contact angle image post separation testing of coating D; unaltered water repellency further supports coating robustness. The 1 : 9 and 1 : 4 Sylgard® 184 : FAS ratios provided the Fig. 6 Average water contact angle data on coatings containing various particle combinations and loadings embedded in a Sylgard® 184 and FAS polymer mixture. Abbreviated particle combination and loadings are as follows; A -TiO 2 60-200 nm particles (0.6 g) with TiO 2 21 nm particles (0.6 g), B -SiO 2 5-15 mm particles (0.6 g) with TiO 2 21 nm particles (0.6 g) and C -TiO 2 60-200 nm particles (0.6 g) with TiO 2 21 nm particles (0.6 g). Each particle combination was embedded in 0 : 1, 1 : 0, 1 : 1, 1 : 4 and 1 : 9 Sylgard 184 : FAS ratios. Coatings were applied to copper 60 mesh substrates prior to oven drying. Error bars show the maximum and minimum values obtained after three repeat readings. Fig. 7 Average percentage of toluene separated from a 50% toluene-water solution on coatings containing various particle combinations and loadings embedded in a Sylgard® 184 and FAS polymer mixture. Abbreviated particle combination and loadings are as follows; A -TiO 2 60-200 nm particles (0.6 g) with TiO 2 21 nm particles (0.6 g), B -SiO 2 5-15 mm particles (0.6 g) with TiO 2 21 nm particles (0.6 g) and C -TiO 2 60-200 nm particles (0.6 g) with TiO 2 21 nm particles (0.6 g). Each particle combination was embedded in 0 : 1, 1 : 0, 1 : 1, 1 : 4 and 1 : 9 Sylgard 184 : FAS ratios. Coatings were applied to copper 60 mesh substrates prior to oven drying. Error bars show the maximum and minimum values obtained after three repeat readings.
ideal particle embedded polymer systems to preserve surface morphology and pore structures (copper 60 mesh substrates), establish preferred wettabilities and achieve surface durability.
The most favorable separation meshes were identied as a result of ensuring surface morphologies and open pore structures were preserved whilst tailoring the surface energy to reside between that of oil and water (approximately between 30 mN m À1 and 70 mN m À1 ). 42 Trends were identied that indicated solvent density had a signicant part to play. For example, copper 100 mesh separation rates were substantially higher using toluene-water solutions as opposed to hexane for identical surface coatings. This was attributed to toluene's increased density of 865 kg m À3 , 43 as opposed to 672 kg m À3 for hexane. 44 A denser solvent was able to more quickly permeate a mesh substrate when originally mixed with water.

Conclusion
A facile route to fabricate a superior (super)hydrophobic and oleophilic separation surface was devised throughout this research. Hierarchical surface roughening in combination with silicon-based polymers generated water repellent coatings with oil-loving properties. Numerous different combinations and concentrations of dual scale TiO 2 and/or hydrophobic-SiO 2 (H-SiO 2 ) particles in varying Sylgard®184 : FAS polymer ratio mixtures were explored in order to enhance surface wettabilities. Average water contact angles were linked to improved solvent separation efficiencies, which, in turn, were most favourable on samples with higher quantities of the uorinerich uoroalkylsilane, FAS, polymer. For example, coating F's dual scale TiO 2 particle system embedded in a 1 : 4 Sylgard® 184 : FAS polymer mixture was applied to copper 60 mesh substrates (pore diameter of 250 mm). The resulting device was highly functional with 100 AE 0%, 85 AE 0% and 97 AE 0% separation efficiencies for toluene-, hexane-and dichloromethanewater solutions respectively. A trade-off between the highly adhesive and durable Sylgard® 184 polymer and the low surface energy FAS, in a 1 : 4 ratio, afforded robust separation coatings that showed no deviation in functionality aer 3 wash-separation-wash cycles nor aer the Scotch tape test.

Conflicts of interest
There are no conicts to declare. Fig. 8 Water droplet on a copper 60 mesh substrate with a coating D surface (SiO 2 5-15 mm particles (0.6 g) with TiO 2 21 nm particles (0.6 g) in a 1 : 9 polymer mixture of Sylgard® 184 and FAS). Image was recorded post separation functional testing.