Nanofabrication of silicon surfaces for reduced virus adhesion

Enteric viruses are highly prevalent, accounting for more than 58% of all foodborne illnesses and 95% of nonbacterial acute gastroenteritis in the United States each year.^1,2 The majority of foodborne viral outbreaks were caused by noroviruses, whose attachment and detachment are critical steps leading to contamination and spreading prior to infection of the host.^1,3–6 Noroviruses are negatively charged biological nanoparticles at neutral pH.^7,8 It has been reported that electrostatic interaction and hydrophobic interaction contribute significantly to virus adhesion on food processing surfaces.^9–14 While the chemical property of a material plays essential roles in virus adhesion, the contribution of surface topography is also nontrivial. Murine norovirus and male-specific coliphage (MS2) have been commonly used as the surrogates of human norovirus in laboratories due to human norovirus’ highly infectious nature, the infeasibility of in vitro culture, and similarities of the surrogates in geometry and capsid proteins.^2,15–17 In a recent study, we evaluated MS2 adhesion to various food contact surfaces.¹⁴ While MS2 demonstrated stronger adhesion to positively charged, hydrophobic surfaces, high-resolution imaging of MS2-inoculated polyvinyl chloride (PVC) surfaces revealed that the randomly distributed pores, 60–80 nm wide and 20–30 nm deep, acted as nano-containers to retain and trap the virions, resulting in eight times faster adsorption and longer term retention of MS2 in the porous areas than in the smooth areas of the same substrate.¹⁴ The strength of MS2 adhesion was twice higher on porous than nonporous surfaces, most likely due to the increased contact area of a rougher surface.

It is known that for a surface with a roughness on the nanoscopic or microscopic scale, a liquid can either wet the surface cavities (Wenzel state) or suspend atop the cavities without penetration (Cassie–Baxter state).^18–20 The wetting behavior is pertinent to the free energy of a droplet on a solid substrate associated with the interface energies and interface areas. Apparently, the MS2 suspension on the porous PVC surface was in the Wenzel state, allowing the liquid suspension to wet the interior wall of the pores, leading to increased MS2 adhesion. In general, a droplet is in the Wenzel state if the cavities are large and shallow, and the Cassie–Baxter (CB) state is preferred if the cavities are small and deep. However, there is no definitive dimension of the cavities that warrants the CB state over the Wenzel state. A droplet may adopt a CB state on surfaces with either microscopic cavities or nanoscopic cavities. The chemical composition and other morphological parameters, such as cavity density and relative roughness of the flat surface, also contribute to the liquid-vapor, liquid-solid, and solid-vapor interface energies and areas, thus playing important roles in determining the surface wettability.^18–20 We hypothesize that a proper control of the surface morphology of a material’s surface can promote the Cassie–Baxter state of a liquid’s wettability, hence rejecting the entry of the virus suspension to the cavities. The pores on the PVC surface had irregular shapes and sizes and were randomly distributed at a low density. Here, highly ordered arrays of nanoholes were generated on the silicon wafer to explore conditions that could minimize virus adhesion.

Nanofabrication is one of the most widely used techniques for surface modification to create desirable patterns in nanoscale. It is useful for a wide range of applications in electronics, photonics, material engineering, and biomedical applications.²¹ Electron-beam lithography (EBL), ion beam lithography, and colloid monolayer lithography have been widely used in nanofabrication.^21–24 EBL is the prominent technique of creating nanoscale patterns on silicon (Si).²¹ Nanofabrication typically involves the processes of resist coating, lithography, development, etching, and resist removal. Each step needs to be optimized to achieve the desirable structures.^25–28 It was reported that nanoarrays at the sub-10 nm scale were successfully generated by EBL on Si substrates.^21,29–32 In this exploratory work, we applied EBL to generate arrays of nanoholes with a diameter of 50 nm, a depth of 22 nm, and a pitch distance of 100 nm on Si substrates. Nanostructured and smooth Si surfaces were compared for the strength of MS2 adhesion as well as for prevention of MS2 entrapment.

A 2 × 2 cm² precleaned Si chip was spin coated with e-beam resist GL2000-anisole (1:5) at 2000 rpm for 35 s and then baked on a hot plate at 150 °C for 3 min. The coating thickness of 33.6 ± 2.6 nm was measured using a Filmetrics® F-20 (San Diego, CA). Computer assisted design of 25 patterned areas, 1 × 1 mm² of each containing an array of nanoholes with a diameter of 50 nm and a pitch distance of 100 nm, was achieved using the CAD (L-Edit™).³³ The EBL was carried out using a JEOL® JBX-8100FS (Peabody, MA) at a dose of 225 μC/cm² and a current of 1 nA. Following the e-beam exposure, the e-beam resist was developed in n-amyl acetate at 20 °C for 45 s and rinsed with isopropyl alcohol for 30 s. The resist nanoscale patterns were transferred to the Si substrate by dry etching in an Oxford® PlasmaLab System 100 (Bristol, UK) at 250 W, 20 °C for 30 s with O₂ flow of 1.0 standard cubic centimeters per minute (sccm) and an HBr flow of 5.0 sccm. The etched Si chip was placed in a Remover 1165 (MICROPOSIT™, Dow Inc., Midland, MI) at 70 °C overnight in order to remove residual e-beam resist [Fig. 1(a)].

The nanofabricated Si wafer was examined by a Raith® 150 SEM (Raith America Inc., Troy, NY) operated under 10 kV, 20 μm aperture and 7 mm working distance. The images were captured in the scan size of 500 nm. A Bruker® AFM (ScanAsyst™, Bruker Corporation, Billerica, MA) was also used to examine the surface in tapping mode using an Si probe (TESPA-V2, Bruker Corporation, Billerica, MA) with the 125 μm cantilever, whose spring constant and resonance frequency were 42 N/m and 320 kHz, respectively. The AFM was operated at a scan rate of 0.8 μm/s. The root-mean-square (RMS) surface roughness was derived from AFM measurements using the AFM software.

The interaction force between MS2 and the nanostructured or smooth Si surfaces was measured using an Agilent 5500 AFM (Agilent Technologies. Santa Clara, CA). MS2 was prepared and purified following the previous protocol,^13,14 and the concentration was approximately 10⁹ PFU/ml as determined by plaque assay. MS2 was conjugated to a colloidal gold probe (1.6 ± 0.3 μm in diameter) via a cross linker, succinimidyl-6-[3-(2-pyridyldithio) propionamido]hexanoate, as routinely performed in our lab.^14,34–37 The force-distance curves were collected with the MS2-conjugated probe in fluid contact mode in phosphate buffered saline at pH 7.4 at a z-scan rate of 0.5 μm/s. The spring constant of 0.17 ± 0.02 N/m was evaluated by using reference cantilevers with known spring constants.^34–37 Typically, force-distance curves were obtained in three different areas for each surface. 320 data points were collected and analyzed statistically by one-way analysis of variance (ANOVA) test.

Images of MS2-inoculated Si surfaces in the absence and presence of nanoholes were captured to examine the population of adhered MS2 on the distinctive surfaces. The substrates were inoculated with 100 μl MS2 suspension of 10⁹ PFU/ml and stored at 4 °C for 2 h followed by thoroughly rinsing (thrice) with de-ionized water and dried by N₂ gas. The images were collected in air tapping mode at a scan rate of 0.4 μm/s and a driving frequency of 275–280 kHz at room temperature.^14,36 MS2 particle density was evaluated using ImageJ® software with a threshold particle diameter of 24 nm. ANOVA was conducted to evaluate the statistical significance of differences.

The wettability of the nanostructured and smooth Si surfaces was examined by water contact angles measured by the sessile drop technique^38,39 using a Model 210 Ramé-Hart goniometer (Succasunna, NJ). The measurements were performed at 22 ± 2 °C with 0.2 μl droplets of de-ionized water or MS2 aqueous suspension. The droplet was small enough to land on the 1 × 1 mm² patterned area, and the diameter of the projected area of the droplet varied narrowly from 0.8 to 1.0 mm [see inset of Fig. 2(b)]. Images of the air, water, and solid interfaces were captured within 20 s after a droplet was discharged to a substrate to avoid the distortion of the drop shape and size due to water vaporization. The DROPimage Pro software was applied to measure the contact angles. For comparison, contact angles were also measured at the smooth regions of the same substrate. The reported contact angle of each substrate was the average of the left and right angles from 48 different measurements.

Figure 1(a) illustrates the flow chart of nanoholes fabrication. The surface was characterized following each step. Both AFM and SEM images were collected after resist development and RIE. Nanoholes of 48 ± 1 nm in diameter and 100 ± 2 nm in pitch distance were measured by SEM after resist development. In good agreement, nanoholes of 51 ± 5 nm in diameter and 100 ± 4 nm in pitch distance were measured in AFM images. No significant difference was found between the two sets of data (p > 0.05). From the AFM images, the depth of the holes was measured to be 28.1 ± 3.6 nm. The thickness of coated e-beam resist was 33.6 ± 2.6 nm. While the complete removal by the developer of the e-beam resist in the exposed structures was border-line, it was within the errors of measurements. If any resist was left, it was removed during the RIE process due to the presence of oxygen in the gas mixture.

Figures 1(b) and 1(c) illustrate the nanofabricated Si surfaces after dry etching and removal of the e-beam resist. Nanoholes of 48 ± 2 nm in diameter and 99 ± 2 nm in pitch distance were determined by SEM. Consistently, nanoholes of 49 ± 2 nm in diameter and 100 ± 2 nm in pitch were measured by AFM. No significant difference was found (p > 0.05). From the high-resolution AFM images, the mean surface roughness of a nanostructured surface and a smooth surface was 1.7 nm and 0.46 nm, respectively. The cross-sectional analysis of high-resolution AFM images indicated the depth of the holes to be 22.1 ± 0.4 nm [Fig. 1(d)]. Thus, we achieved the goal of generating an array of nanoholes with a minimum depth of 20 nm. The clean Si surface was confirmed by the signature 520 cm⁻¹. Raman shift and the lack of spectral feature in the range of 1200–3000 cm⁻¹ in the Raman spectra.

A fascinating and powerful effect of nanofabrication is the tunability of a material’s property. As shown in Fig. 2, the water contact angle on a nanostructured surface was higher than on a smooth surface, and the difference is statistically significant (Table I). In the Cassie–Baxter model, the contact angle θ is expressed by cos θ = f(cos θ₀+ 1) − 1, where θ₀ is Young’s contact angle (for an ideally smooth surface) and f is the ratio of solid-liquid interfacial area to the projected surface area. It predicts the increase of water contact angle on a nanostructured surface, because f < 1 as water does not wet the cavities in the Cassie–Baxter model due to the presence of air pockets in the cavities.^18–20 The nanoholes were 50 nm in diameter and 100 nm in pitch distance. Theoretically, f is about 80% if the cavities are completely unwetted. Taking θ₀= 62.3°, the mean value of the contact angles measured on smooth Si surfaces, and the calculated water contact angle θ at a CB state is 80.1° for the nanostructured surface. However, the measured contact angle was 68.1 ± 3.7°. We surmise that the surface under the current nanofabrication condition was likely in an intermediate wetting state,^18,19 in which the pores were partially wetted due to the relatively rough boundary of the pores [as seen in Fig. 1(d)] and the relatively high surface energy of Si against water.

When water was replaced by the MS2 aqueous suspension, the contact angle measured on a smooth Si surface increased from 62.3° to 66.4° (mean value). It implies that the surface energy of Si dropped in the presence of MS2. The contact angle of MS2 suspension increased to 71.1° on the nanostructured surface, demonstrating a similar surface wetting behavior of the MS2 suspension to that of pure water.

To compare MS2 adsorption on the smooth and nanostructured surfaces, AFM imaging was carried out after 2 h of MS2 incubation on the surfaces, followed by three rinses and drying. MS2 particles were identified and located in height images and confirmed in the simultaneously captured phase or amplitude images¹⁴ (see supplementary material⁴⁷ for identifying MS2 virions on a nanostructured Si substrate with combined height and amplitude images). As shown in Fig. 3(b), randomly distributed particles were observed on a smooth Si substrate. Particles were also observed on the nanostructured Si surface; however, they were mostly present in regions between adjacent nanoholes [Fig. 3(d)]. These particles are MS2 virions, evidenced by their absence on surfaces without MS2 inoculation [Figs. 3(a) and 3(c)]. It is further supported by the observation of bright green spots in fluorescence images of the MS2-inoculated surfaces upon the staining of SYBR gold, which emits a strong green fluorescence after binds to the RNA of MS2.^14,40,41 Noticeably, MS2 particles were found in less than 0.3% nanoholes after 2 h of MS2 incubation. Even after 24 h inoculation, MS2 particles were present in less than 2% nanoholes. We ascribe it at least partially to the pseudo-Cassie–Baxter state of the surface wetting behavior, in which the nanoholes were partially occupied by air pockets, preventing the MS2 entry to the holes and invoking their residence in the flat region between the holes. This is in striking contrast to the observation of virions in every pore of a PVC surface in our previous study.¹⁴ It implies that virions in the aqueous suspension had full access to the pores on PVC, but were denied by pores on the nanostructured Si surface.

In the comparison of the two porous surfaces, despite the difference in chemical composition, the surfaces demonstrated similar water contact angles (68.1° on Si and 69.3° on PVC) indicating a similar surface energy against water. The cavities on the PVC surface were 60–80 nm wide, 20–30 nm deep, and irregularly shaped. It was frequently observed that some pores were inter-connected and clustered, forming larger cavities but distributed randomly at a low density (8.8 pores per μm²).¹⁴ The nanoholes on the fabricated Si surface were 49 ± 2 nm wide, 22.1 ± 0.4 nm deep, and periodically spaced with a pitch distance of 100 ± 2 nm (equivalent to 100 pores per μm²). Even though the pore density was much higher on the nanostructured Si substrate, the overall surface roughness was 1.7 nm in contrast to 10.8 nm for the porous PVC surface. Therefore, in addition to the larger cavities, morphological heterogeneity was present in the background of the PVC surface. Both the large pore size and the extra surface roughness had the effect of increasing the surface contact area and favored MS2 adsorption. The irregularities in cavity shape, size, and distribution brought complications to the system, making it difficult to evaluate the effects of distinctive factors. Nanofabrication, however, offers precise control of a large set of surface parameters for systematic studies to reveal the mechanism and optimize the surface properties for minimizing virus adsorption.

ImageJ analysis of the AFM images was carried out, and the particle densities were 2.2 and 0.87 particles per μm² on the smooth and nanostructured Si surfaces, respectively (Table I). Due to MS2’s icosahedral symmetry, an MS2 can be treated as a sphere of 26 nm in diameter^42–45 with a projected area of 531 nm². Based on the projected areas of the particles in AFM images, the particle size distribution was summarized in the histogram in Fig. 4. Evidently, most MS2 virions were present as individual virions on a nanostructured surface; however, they were largely present as aggregates (the projected areas were multiples of 531 nm²) on a smooth surface. With 50 nm holes and a pitch distance of 100 nm, the nanostructured surface offered a limited space between the adjacent holes for the virions to land on, hence, prevented the formation of aggregates. Individually adhered MS2 are expected to be less stable than the MS2 aggregates due to the spatial isolation. Additionally, MS2 capsid proteins are relatively hydrophobic.¹⁴ In an aqueous medium, the aggregates can recruit additional virions more efficiently to induce MS2 accumulation. Taken together, the nanostructured surface effectively reduced the number of adhered virions.

A virion can adsorb on a surface via weak or strong interactions. To quantitatively differentiate the strength of MS2 adhesion on the smooth and nanostructured surfaces, the strength of MS2 adhesion was determined by AFM force measurement using MS2-conjugated colloidal probes [Fig. 5(a)]. 320 force-distance curves were collected on either substrate, and the adhesion force was measured from the retraction curve [Fig. 5(b)]. As a control, the adhesion forces were also measured using a bare colloidal probe. On both surfaces, the presence of MS2 on an AFM probe induced a significantly higher adhesion force [Figs. 5(c) and 5(d)]. This is consistent with the results from our previous work,^13,14 and verified that the strong adhesion force originated from MS2. The mean value of the strength of MS2 adhesion to a smooth Si surface was 1.75 times stronger than that to a nanostructured surface, and the difference is statistically significant (Table I). Therefore, the nanostructured surface effectively reduced the strength of MS2 adhesion. It was also found that an increase of water rinses from 3 times to 8 times reduced the number of adsorbed MS2 particles by 60% and 25%, respectively, from the nanostructured and smooth Si surfaces. It implies a weaker interaction between the attached MS2 and the nanostructured surface.

On the other hand, we noticed that the strength of MS2 adhesion to a smooth PVC surface (5.1 nN) was at the same level as that of a smooth Si surface (5.6 nN), but was 2.4 times weaker than that to the porous PVC substrate.¹⁴ The distinctive surface topography likely played a key role. MS2 could access the pores on the PVC surface and interacted with the substrate at an area greater than the apparent surface area, prompting a stronger interaction. On the contrary, the cavities on the nanostructured Si surface were inaccessible, leading to the reduced contact area between MS2 and the substrate, and consequently, a weaker interaction. This assumption was further verified by the observation that the interaction force between a control bare probe (in the absence of MS2 conjugation) and a nanostructured Si surface (1.04 ± 0.06 nN) was lower than that on a smooth Si surface (2.1 ± 0.2 nN), whereas the opposite was observed for PVC substrates.¹⁴ 

The effect of Si surface nanofabrication on MS2 adhesion is threefold: (1) the presence of the 50 nm nanoholes reduced the solid contact area by 20% to reduce the overall interaction between MS2 and the substrate; (2) the diameter, depth, and density of the nanoholes granted the system to adopt a pseudo-Cassie–Baxter state of surface wettability, preventing the entry and entrapment of MS2 in the pores to reduce MS2 adsorption; (3) the 100 nm pitch distance equally spaced the nanoholes across the surface and limited the unit landing area for MS2 particles, reducing the number of adsorbed virions. As a result, the nanostructured Si substrate was much less adhesive than the smooth surface.

Nevertheless, the capsid of the MS2 surrogate used in this work is different from that of human norovirus (the major foodborne illness etiology). The quantitative results derived in this study may not necessarily correlate directly to the strength of human norovirus adhesion to the same surfaces. On the other hand, similar to MS2, human norovirus particles are also 27–40 nm in diameter with a negative surface charge at neutral pH, and the percentage of hydrophobic amino acids is high in the protruding domains of the capsid protein.⁴⁶ Thus, the chemical and physical features of a material’s surface are expected to play comparable yet distinctive roles in human norovirus adhesion. Quantitative evaluation can be carried out using a similar method.

Given the dimension of the virions, we anticipate that nanofabrication, with precise control of surface topography, could be a versatile tool to abate virus adhesion. Further study is ongoing to optimize the pitch distance, hole size, and hole density of the nanostructured surface to minimize virus adhesion. Surface silanization can effectively alter the chemical property of a nanostructured surface and is expected to modulate the effect of nanofabrication on virus adhesion. A systematic study with smart design of both chemical composition and nanofabrication will provide insightful information for understanding the surface design principle that will serve to diminish virus adhesion, hence facilitating the development of improved mitigations for controlling viral contamination and transmission.

It should be noted that EBL is not a practical technique to process food contact materials. It is expensive and less accessible. However, it produces precise and reliable nanostructures, whose surface chemistry can be conveniently varied via surface functionalization, for testing hypotheses, exploring ideas, and establishing fundamental understanding on small scales.²² Alternative lithographic techniques, such as nanoimprinting and soft-lithography, are more accessible and cost-effective for the creation of nanostructures on a large scale. It is a realistic option, applicable to various materials when combined with advanced pattern transfer techniques. Chemical and biological bottom-up approaches are also favorable as nanostructures can be developed rapidly with less efforts through self-assembly and self-organization of individual molecular building blocks into desired nanostructures. Defined materials and a controlled reaction environment must be chosen to warrant food safety.

Use of the Center for Nanoscale Materials, an Office of Science user facility, was supported by the U.S. Department of Energy (DOE), Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-06CH11357. It was also supported by U.S. Food and Drug Administration (FDA-CFSAN-IF01673). The authors would like to acknowledge Yue Li for her assistance in AFM operation at ANL.

The data that support the findings of this study are available within the article and its supplementary material.