Surface Chemistry Enhancements for the Tunable Super-Liquid Repellency of Low-Surface-Tension Liquids

Super-hydrophobic, super-oleo(amphi)phobic, and super-omniphobic materials are universally important in the fields of science and engineering. Despite rapid advancements, gaps of understanding still exist between each distinctive wetting state. The transition of super-hydrophobicity to super-(oleo-, amphi-, and omni-)phobicity typically requires the use of re-entrant features. Today, re-entrant geometry induced super-(amphi- and omni-)phobicity is well-supported by both experiments and theory. However, owing to geometrical complexities, the concept of re-entrant geometry forms a dogma that limits the industrial progress of these unique states of wettability. Moreover, a key fundamental question remains unanswered: are extreme surface chemistry enhancements able to influence super-liquid repellency? Here, this was rigorously tested via an alternative pathway that does not require explicit designer re-entrant features. Highly controllable and tunable vertical network polymerization and functionalization were used to achieve fluoroalkyl densification on nanoparticles. For the first time, relative fluoro-functionalization densities are quantitatively tuned and correlated to super-liquid repellency performance. Step-wise tunable super-amphiphobic nanoparticle films with a Cassie–Baxter state (contact angle of >150° and sliding angle of <10°) against various liquids is demonstrated. This was tested down to very low surface tension liquids to a minimum of ca. 23.8 mN/m. Such findings could eventually lead to the future development of super-(amphi)omniphobic materials that transcend the sole use of re-entrant geometry.


Supersaturation Functionalization of SiO 2
The silanol density on silica surfaces is quite variable, occurring within a range of 2-12 silanol groups per nm 2 , depending of the method of synthesis. [1][2][3] Regardless of such variations, grafting densities have rarely exceeded 1-3 fluoroalkyl chains per nm 2 . [4][5][6][7][8][9] In this work, we used a wellinvestigated source of fumed silica, with a particle diameter of 7 nm and a Brunauer-Emmett-Teller (BET) surface area of ca. 395 m 2 .g -1 . This source of silica is typically assumed to have a surface silanol density of 4 µmol.m -2 or 2.4 silanol groups per nm 2 . 4 In theory, the maximum possible silanol surface coverage of a periodic silica surface is 3.5 groups per nm 2 , modelled based on the perfect silica configuration highlighted in the polyhedral oligomeric silsesquioxane (POSS) molecule. 10 However, owing to steric hindrance, monolayer surface functionalization rarely achieves complete coverage. This study re-examines the process of creating heavily grafted nanoparticles though alternative multilayer mechanisms, aiming to close the gap of fluorination density against the fluorinated POSS limit. Fumed silica was chosen because it possesses extremely high specific surface area (SSA), thus holding the most potential for enabling densified grafting conformations. 4 Use of nanoparticles with different roughness could be interesting, but the investigation can be very extensive and currently remains outside the scope of this manuscript. More importantly, it is difficult and complex to directly compare effects of surface geometry vs. chemistry on liquid repellency. The influence of geometry, such as roughness, pitch distances, re-entrant geometries cannot be easily compared to chemistry, which includes factors such as chain length, chain chemistry, network functionalization mechanism etc. Changes made to one variable, such as roughness or pitch distances can influence geometry or chemistry -linked parameters during synthesis and vice versa. 4  The following procedure was performed in an argon-filled glovebox. A purged reactor was first charged with 30 mL of dry chloroform (Sigma Aldrich, ≥ 99%) and purged with argon for 30 min. 1 g of fumed silica nanoparticles (Sigma Aldrich, 7 nm) with an effective surface area of 395 m 2 g -1 was then added into the flask under gentle stirring. The fumed silica was not previously dried but was stored at an equilibrium of 40% relative humidity at 25 °C. Variable amounts of heptadecafluoro-1,1,2,2-tetrahydrodecyl trichlorosilane were then added. Reaction was then allowed to proceed at 25 °C at a stirring rate of 500 RPM for 96 h, under argon purge, in an oil bath to commence and complete the process. Grafting density was calculated based on μmol of silane per m 2 silica. The theoretical limit is known to be at 4 μmol monochlorosilane per m 2 . 3,5 However, as a trichlorosilane was utilized in this work, the theoretical limit should be closer to 1.33 μmol trichlorosilane per m 2 . Based on these assumptions, the volume of silane added ranged from a graft density of 1 (G1), to 20 (G20). Fluoro-functionalized silica (F-SiO 2 ) were then washed and dried at 50 °C for 24 h. F-SiO 2 was then re-suspended and re-dispersed in acetone (Sigma Aldrich, ≥ 99.5%), at a concentration of 10 mg mL -1 . The re-dispersion process requires sonication in acetone for at least 1 hour, 80 °C (~ 10 mL of solution). Grafting densities were correspondingly analyzed via the use of thermogravimetric analysis (see below), based on the work by Campos et al. 4 In contrast to Campos' work, we have chosen to investigate the use of trichlorosilanes due to their potential for network formation. Campos et. al. chose not to investigate the use of trichlorosilanes due to poor grafting performance (ca. 16.1 % w/w) during their preliminary studies. 4 The analysis of intrinsic wettability (on a flat surface) was not possible owing to the following limitations in synthesis: For 1.88 x 10 -3 m 2 (surface area of a glass slide), the amount of silane used would be 0.000282 mL, or 0.282 µL at a reaction graft ratio of 1:1, and 1.7 µL at a reaction graft ratio of 6:1. Scaling the volume of reaction medium, chloroform would give only just 76 µL of reaction volume. These values are currently too small to replicate the reaction at scale, especially since 76 µL does not fully wet a 2.5 cm by 7.5 cm glass slide (notwithstanding further errors due to evaporative losses). Human errors caused by pipetting sub 1 microliter volumes will further exacerbate the problem. Moreover, functionalization of nanoparticles/bulk surfaces (even between fumed vs. precipitated nano-silica) 4 differs greatly and drastic differences in the functionalization of plain glass slides may occur. Results obtained from these experiments is likely unreliable owing to the above listed issues.

Spectroscopic Analysis
Fourier Transform Infrared-Attenuated Total Reflectance (FTIR-ATR, Bruker-Alpha, U.S.A) was performed (32 scans from 400 to 4000cm -1 ) on all grafted samples to qualitatively verify extents of functionalization. Spectroscopic analysis confirmed the successful functionalization through the formation of peaks from 500 cm -1 to 800 cm -1 and 1200 cm -1 , which are indicative of -CF 2 and -CF 3 groups. 11 Graft Density Analysis Thermogravimetric (TGA) and differential thermogravimetry (DTG) analysis were conducted using the STA 8000 (Perkin Elmer, U.S.A) from 50 to 900 °C at 10 °C min -1 ramp under an air atmosphere. A holding step at 120 °C was performed before the actual thermal decomposition procedure that ranged from 120 to 900 °C. Results were presented as the final weight percentage of the calcined ashes, over the initial value at 120 °C. As fumed silica does not decompose during thermal treatment, loss in weight during thermogravimetric tests enables direct computation of grafting density. Samples ranging from a grafting ratio of 1 to 20 μmol silane per m 2 were used to map the effective grafting density (µmol.m -2 ). The molecular weight (M r ) of the organic graft (heptadecafluoro-1,1,2,2-tetrahydrodecyl-) is 347.1. Considering a SSA of 395 m 2 g -1 , the surface coverage of silanes can be represented in μmol silane per m 2 , or silane chains per nm 2 . 4 The latter indirectly represents the functionalized silane groups per nm 2 of the raw material.
Our reference material, the fluorinated POSS molecule, is typically synthesized with 8 distinct fluorinated chains. In this study, we compare our findings and performance with this limit, exemplified by the (1H,1H,2H,2H-heptadecafluorodecyl) 8  Supporting these results, BET-SSA also showcased gradual decline in specific surface area of the grafted material, decreasing from ca. 400 m 2 /g to just ca. 100 m 2 /g at saturation (G6 variant).
Increase in grafting density and coating thicknesses lead to decrease in specific surface areas.
These are methods that are conventionally used in the field for analyzing nanosilica material. 4 Unfortunately, owing to the inherent nature of fractal fumed silica, hydrodynamic radii cannot be reasonably abstracted from the data.

Spray coating of F-SiO 2
F-SiO 2 in acetone suspensions (10 mg mL -1 ) were sprayed onto soda lime glass substrates at 3 bars with a flow rate of 0.2 mL s -1 from a 10 cm working distance using an artist's air brush (nozzle diameter, 0.2 mm). 10 mL of the suspension was sprayed a dimensional area of 2.5 cm by 10 cm. A traverse rate of ca. 10 cm s -1 was maintained using guide rails on a custom-built spray rig. All coatings were stored for at least 24 h in darkness prior to the commencement of testing.

Wetting Analysis
Superhydro(oleo)phobicity was assessed through the measurement of static contact angles contact angle goniometer (Finland) with a heliopan ES43 camera (Japan). The CA, SA and CAH were computed by a commercially available (CAM2008) program. Data was presented as mean ± standard deviations.

Droplet Impact Dynamics
The droplet impact dynamics was evaluated in an ambient environment, at 25 °C with 60% relative humidity. Liquid drops of ∼ 5 µL (corresponding to a radius (R o ) of ∼ 1.0 ± 0.05 mm) were released from a pre-determined height of 5. of 3% to cover the maximum peak-to-trough heights of hierarchical coatings averaging 3 crossbatch repeats respectively. Data was presented as mean ± standard errors.

Evaluation of Optical Data
The presence of a networked fluoroalkyl layer results in contrasting refractive indices at the silica-fluoroalkyl interfaces, aggravated by the high surface areas inherent to nanomaterials.
This, in turn, results in increased scattering that is observed in both interferometric and transmittance studies ( Figure S4d-e). However, quantitative optical analysis of these surfaces suggests distinctive optical variations. Firstly, white light interferometric analysis of the surfaces revealed increasing root-mean-square roughness at higher reaction grafting ratios, from 231 ± 18 nm at 1:1 to an equilibrium maxima of ca. 650 ± 100 nm at 6:1 ( Figure S4d).
Secondly, UV-vis analysis revealed a gradual drop in transmittance properties, despite the same material loading (thickness, Figure S4a-c), down to ca. 50-60 % at and above the reaction grafting ratio of 6:1 ( Figure S4e). Even though surface roughness and transmittance appear to have an influence with various grafting ratios, these are optical effects related to increasing functionalization. The increasingly dense fluoroalkyl layer around the nanoparticles ( Figure   S4d-e) leads to refractive index mismatching. For reference, refractive indices of pure fumed silica (n = 1.46) and heavily fluorine-doped silica (1.38 < n << 1.46) can be quite different. 18 This is a well-known effect in coated or shell-like nanoparticles. 19,20 The    1. c) Peak-to-trough distances for surfaces appear to have a gradual rise with increasing grafting ratios, but error bars (standard deviation) overlap significantly, and does not support the trend of an increasingly thickened coating. Moreover, the "gradient" is not consistent. Nanoparticle coatings analyzed across reaction grafting ratios (1:1 to 20:1), with d) root-meansquare roughness and e) UV-vis transmittance indicative of maxima at 6:1 with a tapering effect from 6:1 to 12:1. A sharp step-change between 4:1 and 6:1 is evident.