Always on the bright side of life: anti-adhesive properties of insect ommatidia grating

The insect compound eye consists of 50 (Xenos peckii), 800 (Drosophila melanogaster) or even 10,000 (Aeshna mixta) individual eye facets (ommatidia), each with its own lens and a set of photoreceptor cells, packed in a hexagonal array. Compared with vertebrate camera-type eyes, insect compound eyes feature a large variation in the facet size and number and in corneal curvature. All these features contribute to a large variation in spatial resolution of the eye (Duparre and Wippermann, 2006). Insect eyes can detect fast movement (Olberg et al., 2007) and, in some cases, the polarisation of light (Labhart and Meyer, 1999). The biconvex, transparent and colourless corneal lens, partly consisting of resilin (Viswanathan and Varadaraj, 1985), forms the main part of the ommatidium cuticle and contributes to absorbing and filtering the suitable spectral range of light.

Surfaces of insect ommatidia contain nanoscale protuberances termed ‘corneal nipples’ or ‘ommatidia gratings’ (Bernhard and Miller, 1962; Miskimen and Rodriguez, 1981; Stalleicken et al., 2006). They are generally accepted to play a fundamental role in the anti-reflective properties of the optical system of the eye (Bernhard et al., 1965; Parker et al., 1998; Stavenga et al., 2006).

Eyes of many groups of pollinating insects, such as bees and flies, as well as eyes of subterranean beetles, may be negatively affected by pollen and dust contamination (Singer and Cocucci, 1997). However, a clean eye surface is crucial to the visually orientated insects. Our preliminary observations show that, in contrast to the rest of the body, ommatidia of various insects remain clean even in a heavy contaminated environment. In numerous insect species, the eye contamination problem is solved due to active cleaning systems (Hlavac, 1975; Jander, 1976). For example, highly elaborate grooming systems have been repeatedly reported in the literature for representatives of Hymenoptera (Schonitzer and Renner, 1984), Diptera (Szebenyi, 1969) and Coleoptera (Valentine, 1973).

By contrast, in addition to previously reported anti-reflection properties of ommatidial gratings, a self-cleaning mechanism based on cuticle geometry and surface energy is present on insect ommatidia. The principal aims of this study were to investigate (a) whether there is a significant variance in the adhesion properties between ommatidia of different insect species; (b) whether the variance in adhesion is due to a difference in surface topography; and (c) whether these structures could be biomimetically reproduced.

To investigate these aims, we measured the adhesive properties of insect ommatidia and of their polymer moulds, and compared these data with control surfaces that had the same curvature radii as ommatidia but lacked the nanostructure of corneal nipples. For this purpose, atomic force microscopy (AFM) was applied. It has been previously demonstrated that AFM is an excellent method to study the structure of insect surfaces at the highest resolution under near-living conditions and to evaluate surface adhesion properties (Voigt et al., 2009). The adhesive properties of ommatidia moulds of representatives from three different insect groups (Odonata, Lepidoptera and Diptera) were compared. While diurnal Aeshna mixta (Odonata) lacks protuberances, nocturnal Laothoe populi (Lepidoptera) and diurnal Volucella pellucens (Diptera) bear corneal nipples of different shapes.

The corneae of dead, dried dragonflies Aeshna mixta described by Pierre André Latreille 1805 (Odonata, Aeshnidae), moths Laothoe populi L.1758 (Lepidoptera, Sphingidae) and flies Volucella pellucens L.1758 (Diptera, Syrphidae) were taken from different collections and prepared using a scalpel and tweezers. The samples were rehydrated in aqua Millipore (Millipore GmbH, Schwalbach/Ts., Germany) for 10 min. For imaging, 3 mm×3 mm pieces of corneal cuticle were cut out and fixed on a glass slide using double-sided adhesive tape. Prior to and between measurements, all samples were kept in closed Petri dishes. Transparent corneae allowed their observation in an inverted light microscope (Zeiss Axiovert 135, Carl Zeiss MicroImaging GmbH, Göttingen, Germany) equipped with a AFM scanner head (JPK Instruments, Berlin, Germany). Due to the necessity of the sputter-coating procedure for scanning electron microscopy (SEM) and therewith an influence on the results of adhesion measurements, one (un-sputtered) cornea hemisphere of each insect was studied with AFM whereas the second hemisphere (sputtered) was used in SEM studies.

To gain additional information about the dimensions of the corneal nipples and to verify AFM images, a SEM Hitachi S-4800 (Hitachi High-Technologies Corp., Tokyo, Japan), operating at an accelerating voltage of 2 kV, was used. Samples were mounted on aluminium stubs using double-sided, carbon-containing, adhesive tape and were coated with gold–palladium (thickness 6 nm) in a Bal-tec SCD 500 Sputter Coater (Bal-tec AG, Balzers, Liechtenstein).

It is well known that surface chemistry influences the adhesive properties of the surface (Kendall, 2001). As different insect species and control surfaces have different surface chemistry, we were challenged to exclude the chemistry effect, in order to explore solely geometry effects, caused by ommatidia gratings, on adhesion. This is why we compared the adhesive properties not only of original insect ommatidia but also of resin moulds of compound eyes and those of control surfaces.

Negative moulds of the cornea samples, obtained using Affinis® dental wax (Coltène/Whaledent GmbH&Co. KG, Langenau, Germany), were subsequently moulded using epoxy resin, which was polymerised for 8 h at 70°C. We applied a low viscosity epoxy resin (Spurr, 1969), which is able to replicate even the smallest and rather complex biological surface features (Gorb, 2007).

To find smooth control surfaces with curvatures corresponding to those of the ommatidia, surface profiles (virtual cross-sections) of 10 randomly selected ommatidia of each insect species were obtained from AFM scans and averaged to obtain a mean curvature. The curvatures obtained for each insect species were then compared with the diameters of commercially available microspheres (Duke Scientific Corporation, Palo Alto, CA, USA), resulting in dimensions from 70μm for V. pellucens, 140μm for L. populi, up to 400μm for A. mixta. These microspheres were used as templates to prepare smooth control surfaces with curvatures similar to that of ommatidia of the selected insect species. Spurr moulds of the control surfaces were fabricated as mentioned above.

For imaging and adhesion measurements, we used a NanoWizard® atomic force microscope (JPK Instruments), mounted on an inverted light microscope (Zeiss Axiovert 135, Carl Zeiss MicroImaging GmbH). The AFM intermittent contact mode was applied to visualise the corneal nipples of the insects. The error channel (also known as the amplitude channel) visualises the change in damping of the cantilever amplitude while scanning the surface. Only images obtained with the error channel are shown, because this visualisation method permits a better comparison with SEM micrographs and enables a more vivid imaging of the surface topography. The force–distance curves were obtained on the resin moulds of the biological surfaces and control surface moulds. Scans were carried out in air (room temperature 24°C, relative humidity 41%) at a 1 Hz scan rate and a resolution of 1024×1024 pixels with an intermittent contact mode cantilever (c=50 N m^–1, NST-NCHF, Nascatec GmbH, Stuttgart, Germany). NanoWizard® SPM software 3.3.23 (JPK Instruments) was used to obtain AFM images. NanoWizard® image processing software 3.3.25 was applied to extract 3-D surface profiles and adhesive forces measured on each surface.

Adhesion measurements were carried out in air using a low frequency, non-contact mode cantilever (NST-NCLF) with a 30μm certified traceable size standard sphere (Duke Scientific Corporation) attached to the cantilever tip. The spring constant of the cantilever (30 N m^–1) was ascertained using the thermal noise method (Hutter and Bechhoefer, 1993), before the sphere was attached to it.

On eyes of each of the three insect species studied, 10 ommatidia were chosen for adhesion measurements. To prevent a double contact of the sphere with two adjacent ommatidia, a 10μm×10μm scan pattern (force map) on top of each ommatidium was defined. Adhesion force measurements were carried out on these 100 selected sites within the 100μm² area on each ommatidium. Adhesion force values of each of the 10 ommatidia were then averaged and statistically compared with the values obtained for the smooth control surface moulds. In all, 1000 individual force–distance curves for eyes of each insect species were collected and analysed.

The comparison of SEM and AFM images of insect corneal nipples (Fig. 1) revealed that the heights of the gratings ranged from 25±6 nm (L. populi) to 70±10 nm (V. pellucens), packed in a hexagonal manner, as are the ommatidia themselves. However, packing of the respective protuberances is rather dissimilar in L. populi and V. pellucens. Whereas L. populi perpetuates the hexagonally dense packing at the nanoscale, V. pellucens, on the contrary, features laterally distinct nipples with valley distances of about 40 nm.

While in A. mixta, a regular grating was absent, a surface roughness (r.m.s.=1.3 nm) was observed. The aspect ratio of single nipples ranged from 0.1 for L. populi to 0.32 in V. pellucens with a peak-to-peak distance of 260±10 nm (L. populi) and 200±10 nm (V. pellucens).

Resin moulds of original surfaces demonstrated their excellent quality (Fig. 2). Differences in protuberance shape were negligible, permitting usage of moulds for adhesion measurements. The sphere moulds, used as control samples, featured similar roughness as the ommatidium surface of A. mixta.

Pull-off forces measured on the moulds of ommatidia of L. populi and V. pellucens were significantly lower than those obtained on the smooth control surface moulds (Fig. 3). A statistically significant difference in pull-off force of ommatidia moulds (Mann–Whitney rank sum test, P≥0.001) for L. populi (77±13 nN) and V. pellucens (19±1 nN) was found when compared with their smooth control surface mould values (L. populi control: 0.9μN and V. pellucens control: 1.9μN). Values obtained for A. mixta ommatidia moulds (178±40 nN) and their control samples (1.2±0.5μN) suggest a difference in adhesion, which could not be statistically verified. A comparison between obtained adhesion forces for each surface is given in Fig. 4.

To calculate the real contact area between the ommatidium surface and the sphere attached to the AFM cantilever, the JKR model (Johnson et al., 1971) was used. With R₁ (radius of the sphere attached to the cantilever) kept constant at 30μm and changing R₂, depending on the differing curvature values of the ommatidia known from cross-section, Eqn 2 gives the reduced radius R needed to determine contact radius a₀ (Eqn 1).

Reduced E-modulus K=11.3 GPa was derived for the equilibrium state from Eqn 3 by setting E_a=7 GPa (E-modulus of polymerised Spurr resin), E_b=70 GPa (E-modulus of glass) and Poisson ratios of both materials ν_a=ν_b=0.5. Substituting R, K and γ=0.05 N m (Spolenak et al., 2005) in Eqn 1, a₀ is obtained. The A_C can be calculated according to Eqn 4. Table 1 depicts the R, a₀ and A_C values obtained for each type of surface–probe contact.

The relevant contact area for adhesion is further reduced, due to corneal grating within the calculated area of contact. Fig. 5 shows a visualisation of the calculated contact area (white circle) within an AFM error channel image of L. populi. It is clearly visible that only a few corneal nipples are able to make contact with the sphere whereas the rest of the calculated area remains untouched. For this reason, we recalculated the contact area and pull-off force, taking the number of possible contacts into account (Table 2).

The data on topography and adhesion properties of the eyes of three different insect species, obtained in the present study, clearly demonstrate an anti-adhesive function of insect corneal protuberances. Our SEM and AFM images revealed that our measurements of L. populi and V. pellucens protuberance height and width and distribution of corneal nipples are in good agreement with previously published values (Bernhard and Miller, 1962; Bernhard et al., 1970). It is noteworthy that protuberance shape and packing density on the ommatidium surface are rather different in these two species, while A. mixta shows no grating at all. Whereas L. populi features a hexagonal close packing (h.c.p.) with a round and almost sphere-like shape of nipples, V. pellucens shows a non-close packing (n.c.p.) and a more spike-like shape of single nipples. These two packing patterns were previously described for the mosquito Culex pipiens (Gao, X. F. et al., 2007). However, in C. pipiens, both patterns appear in the same individual and at different dimensions.

We found distribution and shape varieties in corneal grating correlated with measured adhesion. As aspect ratio increases, adhesion is reduced. The rather smooth, non-nippled ommatidia of A. mixta (r.m.s.=1.3 nm) demonstrated no statistical significant difference in adhesion when compared with their smooth control samples. Results obtained from the insect with smooth ommatidia emphasises the role of corneal grating in adhesion reduction.

It has previously been assumed that certain micro- and nanostructured surfaces, similar to the eye corneal grating, may be responsible for adhesion reduction between contaminating particles and biological surfaces, resulting in a self-cleaning property (Wagner et al., 1996). Literature data on other biological surfaces show that both surface structure and chemistry are involved in such a mechanism of anti-contamination (Barthlott and Neinhuis, 1997; Bhushan and Jung, 2008; Watson et al., 2008).

Numerous studies in the field of contact mechanics show a strong correlation between surface roughness and adhesion reduction due to a minimised area of contact between surfaces. It was found that a small surface roughness can completely remove adhesion (Kendall, 1971; Fuller and Tabor, 1975; Persson and Tosatti, 2001). Despite these findings, corneal grating has so far only been discussed with respect to super hydrophobicity (Gao, X. F. et al., 2007) or anti-reflective properties (Bernhard et al., 1965; Parker et al., 1998). In the present study, the remarkably low adhesion values were revealed on the ommatidia possessing corneal nipples, and the results of the calculations according to the JKR contact model lead to the conclusion that the real contact area in nippled ommatidia is reduced, when compared with the smooth ones. This means that the van der Waals forces, responsible for adhesion, may only occur on top of a few protuberances within the imaginary area of contact. The measured adhesion force is much lower compared with the mass of an average, for example, Salvia glutinosa pollen grain (0.041μg), resulting in a non-sticky state. This phenomenon can also be applied to dust particles, etc. That is why the nippled ommatidia surfaces have self-cleaning properties mainly due to this particular hierarchical surface geometry. Additionally, during flight, wind drag and cuticle vibrations may contribute to the self-cleaning properties of the ommatidia.

One can assume that the presence of self-cleaning corneal surfaces may negatively correlate with the presence of structures actively used for grooming. As most insects strongly depend on their visual perception, which can be affected by contamination (Singer and Cocucci, 1997), some of them, like honeybees, developed a highly elaborate cleaning device combined with grooming behaviour (Thorp, 1979), while in others, such as representatives of Lepidoptera, grooming is infrequently observed (Jander, 1976). This is why the lack of corneal grating may have pushed the development of more complex cleaning devices and corresponding grooming behaviour in some insect groups.

In conclusion, our observations of reduced adhesion due to ommatidia gratings and the results of our JKR model calculations of real contact area for different eye geometries indicate that the self-cleaning property of some insect eyes might be another biological function of ommatidia gratings in addition to their widely accepted anti-reflective properties and ability to reduce glare to predators. As our measurements on the polymer moulds of insect eyes show, such a strong anti-adhesive effect might be interesting for industrial applications. Corresponding technology for the development of surfaces having two hierarchical radii of curvature and simultaneously bearing self-cleaning and anti-reflective properties was recently described in the literature (Gao, H. J. et al., 2007).

Sincere thanks to Christian Löbbe (JPK Instruments, Berlin, Germany) for his permanent support and help in the experimental setup. Valuable discussions with Lars Heepe on contact mechanics are acknowledged.