Photo Initiated Chemical Vapour Deposition To Increase Polymer Hydrophobicity

Apple growers face new challenges to produce organic apples and now many cover orchards with high-density polyethylene (HDPE) nets to exclude insects, rather than spraying insecticides. However, rainwater- associated wetness favours the development of apple scabs, Venturia inaequalis, whose lesions accumulate on the leaves and fruit causing unsightly spots. Treating the nets with a superhydrophobic coating should reduce the amount of water that passes through the net. Here we treat HDPE and polyethylene terephthalate using photo-initiated chemical vapour deposition (PICVD). We placed polymer samples in a quartz tube and passed a mixture of H2 and CO through it while a UVC lamp (254 nm) illuminated the surface. After the treatment, the contact angle between water droplets and the surface increased by an average of 20°. The contact angle of samples placed 70 cm from the entrance of the tube was higher than those at 45 cm and 20 cm. The PICVD-treated HDPE achieved a contact angle of 124°. Nets spray coated with a solvent-based commercial product achieved 180° but water ingress was, surprisingly, higher than that for nets with a lower contact angle.

Chemical modification is an alternative to this problem since the surface is functionalized. News groups are chemically bonded to the surface, which should prevent leaching. Available technologies included sol-gel 15,23 and chemical vapour deposition (CVD) 24 . By opting for a gas-phase approach like CVD, it is possible to further prevent residual solvent leaching. Further, CVD coatings typically exhibit high film durability 25 and are readily scaled up 26,27 . The deposition can be initiated by heat (thermally activated/TACVD) 28,29 , by plasma (plasma enhanced / PECVD) 30,31 or by light (photo-initiated/PICVD) 32,33 . Heat initiation is problematic with temperature-sensitive substrates such as polymers 25,34 , whereas plasma is plagued with scale-up issues due to specific operating requirements 12,25,[35][36][37] . PICVD is an alternative with its low energy treatment and wide spectrum of possible variations 24,35 . Indeed, depending on the light source used, it can be operated at ambient temperature and pressure conditions, without the use of highly specialized equipment. This article focuses on reducing the permeability of high-density polyethylene (HDPE) and polyethylene terephthalate (PET) nets via PICVD and evaluating their behaviour under simulated field conditions.

Methods
Experiments. Rain simulator. To demonstrate that a superhydrophobic net reduces water ingress during rainy days, we designed and built a rain simulator 38 (see Supplementary Fig. S1). A peristaltic pump feeds water to a nozzle at a rate of 250 ml/h. Droplets with a diameter of 0.6 mm form at a needle placed at the end of the nozzle 39 . The droplets fall on a 1 mm by 0.6 mm exclusion net set at tilt angles ranging from 0° to 80°. The superhydrophobic coating (Rustoleum NeverWet) was applied following 2 steps. First, a base coat was sprayed approximately 15 cm to 30 cm from the net. After 30 min, a topcoat was sprayed at the same distance. After an additional 30 min the test was initiated.
Photo initiated chemical vapour deposition (PICVD). Polymer samples are placed at various lengths along a 25 mm ID quartz tube (see Supplementary Fig. S2). Argon (400 ml min −1 ) purged the tube for 2 min to scavenge residual oxygen. Brooks mass flow controllers metered the gases to the reactor (400 ml min −1 ). Two 30 W UVC lamps (253.7 nm, irradiance of 5.5 × 10 −4 μ W cm −2 , at 4.5 cm) illuminated the surface of the polymer samples. A precision valve controlled the pressure in the tube. The line was connected to a three-way valve leading to vacuum to test sub atmospheric pressures. For further details, see Dion et al. 40 . The factors tested to functionalize the polymer surfaces included time, pressure, H 2 /CO (syngas) ratio, photo initiator feed rate (H 2 O 2 ) as well as sample position in the reactor (3 to 87 cm from the inlet). Ranges for each factor can be found in Supplementary Table S1. We tested both HDPE, which apple growers use to cover their trees, as well as PET. HDPE exclusion nets typically contain a UV protector (used to extend the life-span of the net itself), but that protector limits the reaction rate initiated by the UVC lamp. Therefore, PICVD assays were conducted with PET, which has superior UV resistance, and HDPE sheets without the UV protector package.

Materials
Samples included: Protek Net 80 g/m 2 HDPE exclusion net (Dubois Agrinovation) with a 1 mm by 0.6 mm mesh size, which is optimal for apple tree applications; sheet extrusion HDPE (McMaster) for PICVD experiments cut, to 15 mm by 38 mm by 1.3 mm; and film PET (McMaster) for PICVD tests, cut to 15 mm by 38 mm by 0.3 mm. CO, H 2 and argon were purchased from Air Liquide and hydrogen peroxide (50%) from Sigma-Aldrich.
All polymer samples were washed in a Fisher Scientific ultrasonic bath (model FS110H) in two steps. First, samples were submerged in water for 15 min, followed by an acetone bath. Samples were then dried in a vacuum desiccator for at least 2 h.
Characterization. Contact angle measurement. After each experiment, 5 contact angle measurements were taken per sample and the average value is reported. 2 μ L of distilled water is deposited on the untreated or treated surface to measure the sessile drop contact angle. FDS tensiometer OCA DataPhysics TBU 90E recorded all contact angles after one minute from when the drop reaches the surface.

Fourier Transform Infrared Spectroscopy (FTIR).
A Thermo Scientific Nicolet iS5 Spectrometer equipped with an ATR module measured the infrared absorbance as a function of wave number. The analysis range varied from 400 cm −1 to 4000 cm −1 and the number of scans was set to 16 with a resolution of 4 cm −1 .
Atomic Force Microscopy (AFM). Atomic force microscopy images were collected with a multiMode8 AFM, Bruker/Santa Barbara NanoScope V, using standard tapping mode with a Tespa-type needle, in air at room temperature. Intermittent contact imaging was performed at a scan rate of 1 Hz using etched silicon cantilevers (ACTA from AppNano) with a resonance frequency around 300 kHz (spring constant of ≈ 42 N/m and tip radius of < 10 nm). Images were taken with a 512 × 512 pixel resolution over a 5 μ m × 5 μ m square with medium tip oscillation damping (20-30%).

Results and Discussion
Hypothesis validation. Despite the Lee et al. work, it is not intuitive that a superhydrophobic surface can prevent water ingress through a net. To prove this concept, a rain simulator was built to simulate rain fall on real HDPE exclusion nets. Figures 1 and 2 show how a superhydrophobic net treated with a silicon-based coating can limit water ingress. The commercial treatment reduces the amount of water passing through the net by 70%, even in a horizontal position. The major problem with this solution is treatment longevity. After about 2 h the commercial coating disappears. Moreover, the treated HDPE net loses its mechanical properties such as elasticity-the net Scientific RepoRts | 6:31574 | DOI: 10.1038/srep31574 becomes visibly stiffer (Supplementary Video S1). This would be problematic when it becomes windy. Given that PICVD coatings are thinner than silicon-based treatments, the degradation rates should be reduced considerably.

Experimental
Design. An experimental design was conducted and related to statistical models to gain a better understanding of the effect of operating conditions as a function of the substrate selected. These models were employed to target the optimal conditions for the highest contact angle. The contact angle results following an experimental design using multiple approaches are presented in Supplementary Table S2. Initially, the contact angle on HDPE without a treatment is 96° ± 3° and for PET it is 85° ± 3°. A combination of 3 experimental plans was developed to account for quadratic effects. We designed a fractional factorial design (FFD) of resolution V to account for linear affects. Resolution V allows having fewer experiments with greater precision 41 . The central composite design (CCD) is complementary to the FFD, as it takes into account quadratic effects 42 . The Box-Benhken plan added experiments to identify additional quadratic effects 43 . Finally, midpoint experiments fixed the model in space 44 . Contact angle results presented are the average from 5 measurements per sample. When a value is aberrant according to Chauvenet's criterion, it is deleted 45 .

Model Equation.
Dion et al. 40 presented a statistical model for the PICVD treatment of copper with a similar experimental design. The major contributing factors to the model were position and syngas ratio. We tested five factors-pressure, time, position, H 2 /CO ratio and H 2 O 2 and developed a regression model based on the Marquardt-Levenberg algorithm to identify a model that maximizes the multiple R 2 (SigmaPlot ® ). Interactions between each variable and the possibility of second-order effects were verified. Only the significant variables and interactions were kept in the model (equation (1) for HDPE and equation (2) for PET). Parameter values can be found as Supplementary Tables S3 and S4. In each model, an aberrant point was deleted according to a large deviation from the quartile 46 and we normalized the factors (supplementary information). The fact that a value has been removed from the model leads to the conclusion that these models are an approximation of the behaviour after the surface modification. Despite the high value of the regression coefficients (R 2 ), the models are accurate only inside the range studied (Fig. 3).    Comparison of the two models. The two models seem very complex, evidenced by a large number of different factors that influence the reaction. Moreover, the parameters are different, indicating that the reaction depends on the substrate. The majority of experiments for HDPE led to hydrophobic behaviour, explaining the mass of data points between 100 and 120. For PET, the variance is slightly higher and the results are distributed over all contact angles values. Position (Po) appears to be a key factor in the surface modification process, having an individual effect on the reaction, as well as many interactions with others factors. Most of these interactions are linear, except for one quadratic effect in the PET model. Pressure (Pr), H 2 /CO ratio (r) and hydrogen peroxide (H 2 O 2 ) must be combined with other factors to influence the reaction. In both models, the interaction between the five factors cannot be omitted.
First approach to kinetic modelling. Position is the most significant factor in the PICVD reaction and must be analyzed in depth. To more clearly illustrate its interaction with treatment time (and infer reaction kinetics), samples were placed in the reactor for 3 min, the contact angle measurement was recorded and the samples were returned to the reactor for treatment for an additional time increment and so on. Figure 4 present the contact angle as a function of time for HDPE. Supplementary Fig. S3 shows similar figure for PET. For each experiment, the ratio H 2 /CO was 0.12 with 1 ml/h peroxide injected at 10 kPa relative pressure.
We can express the evolution in contact angle as a function of time as a sum of two reactions, the first one decreasing the contact angle (k 1 ) and the second one increasing it (k 2 ). Assuming first order reactions, we obtain: where θ 0 is the initial contact angle and Δ θ is the coefficient of variation of the contact angle. The fit parameters are tabulated in Table 1. Despite somewhat low R 2 values (attributed to data variation), this preliminary kinetic modelling highlights the two different reactions that occur in PICVD. First, k 2 is always higher than k 1 which indicates that the reaction leading to a contact angle increase is the fastest. On the other hand, the decreasing reaction's pre-exponential factor, θ 0 is higher, implying that it will become dominant over time. As position in the reactor increases, the coefficients k 1 and k 2 stabilize -in other words, the difference in contact angle will be minimal if the samples are further inserted from the gas inlet. For the two polymers, the k 1 and k 2 coefficients are similar, hence the same measured increase in contact angle (approximately 20°). The coefficient of variation of the contact angle (Δ θ 0 ) is identical for all substrates and positions, and represents the maximum variation attainable. Moreover, there are diminishing returns on treatment: the contact angle will not change significantly after a certain treatment time. A significant decrease in contact angle is observable at position Po = 20 cm for each polymer after 1 hour. The high variability between the model presented and these data points can be explained by hydrophobic recovery. Indeed, Extreme values. The experimental plan allowed for a mapping of the various surface properties that can be obtained on HDPE and PET following PICVD treatment. Table 2 and Fig. 5 highlight the extreme values (i.e. most hydrophobic and most hydrophilic behaviour) that could be obtained by PICVD.
Many experimental conditions can yield hydrophilic surfaces on both polymers, but this effect is unstable because of hydrophobic recovery 47 . Moreover, while the typical target for superhydrophobicity is a contact angle higher than 150°, the more moderate increase in hydrophobicity achievable by PICVD does affect water ingress (see section "exclusion net experiments").

Fourier Transform Infrared Spectroscopy (FTIR). HDPE is a linear molecule with only C-H groups.
After hydrophobic PICVD treatment, new functional groups appear (see Fig. 6). Indeed, C-O links are present after treatment, evidenced by new peaks around 1100 cm −1 , which can be attributed to alcohol, or ether groups. Moreover, C= O bonds are identified around 1600 cm −1 , in the form of ketone or aldehydes. The combination of peaks ca. 1100 cm −1 and 1600 cm −1 are a strong indication of carboxylic functionality 48,49 .
Despite the measured contact angle difference, no visible differences can be measured by FTIR for PET, likely due to the presence of oxygen atoms in its molecular structure before any treatment. A more surface-sensitive analysis technique such as X-ray Photoelectron Spectroscopy (XPS) would be needed to probe the functional groups imparted by PICVD.   (equation (3)).

Contact angle [θ HDPE ] (°) Standard deviation (°) Contact angle [θ PET ] (°) Standard deviation (°)
Hydrophobic treated 124 ± 3 112 ± 3 Hydrophilic treated 14 ± 2 13 ± 1 Atomic Force Microscope (AFM). It is somewhat counter-intuitive that contact angles increase with the addition of oxygen-containing functional groups. However, roughness plays a significant role in hydrophobicity. This can be measured through AFM (see Fig. 7). While both polymers are initially quite flat, PICVD treatment leads to the appearance of several "islands". This radical change in roughness increases the formation of air pockets below the water droplets, thus leading to the observed hydrophobic behaviour 50,51 . For PET, the AFM images provide conclusive evidence of surface modification, despite the absence of change in the FTIR spectra. PET roughness increased after PICVD treatment, while this change was more moderate for HDPE. Quantitative characterisation for both polymers is presented in supplementary Table S5.
Exclusion net experiments. The contact angle reached a maximum of 124° by PICVD on HDPE (without any UV protection package). Because the HDPE exclusion nets contain a UV protector, they could not be treated in the PICVD reactor. However, applying the commercial single step "Rustoleum NeverWet for Outdoor Fabric Treatment" formulation could simulate this hydrophobic effect. We used this product instead of the original NeverWet product because the contact angle on the exclusion nets reached 120°. After applying the coating in a single step and drying for 24 h, rain tests were conducted (see Supplementary Fig. S1). The results for this "less" hydrophobic treatment are nearly identical to that of the superhydrophobic treatment (see Fig. 8). Although the HDPE nets could not be directly treated by PICVD, this simulated behaviour demonstrates that, if the UV  protector issue can be curtailed, a 120° contact angle should be sufficient to prevent water ingress on exclusion nets, at least under conditions similar to those of our rain simulator. The results for superhydrophobic and hydrophobic treatment are statistically indistinguishable above 20°: the error bars overlap and the average water fraction that penetrates is within 5%. At 10°, the superphydrophobic surface allows more water to pass through the net (p < 0.05). On the superhydrophobic treated nets, the droplets bounce many times on the surface. This may increase the chance that smaller droplets form during rebound and these pass through the holes. So, a hydrophobic surface with a contact angle of 120° is good enough to minimize water passing through the studied exclusion nets.

Conclusions
High contact angle exclusion nets for organic apple application are an alternative to minimize fungicides that are otherwise needed to control apple scabs. The PICVD reaction renders polymer surfaces hydrophobic, which is desirable to minimize rainwater ingress to the trees, but also hydrophilic depending on the treatment conditions.  To optimize the hydrophobicity, we derived a mathematical relationship for each polymer based on an experimental design. We measured the contact angle as a function of exposure time and fit the data to a differential equation. FTIR and AFM confirmed that the surface was chemically modified. Rain test results showed that a superhydrophobic or a hydrophobic coating minimized water ingress through a ProtekNet 80g (mesh size 1.0 × 0.60 mm). A more precise study on surface degradation with time, sun, humidity and wind is required to establish the suitability of this technology for commercialization.