The impact behaviour of silk cocoons

Fundamental research on silkworms tends to focus on increasing our understanding of the material and biochemical properties of their silk fibres (Porter and Vollrath, 2009). More applied silkworm research tends to focus on methods to best extract the silk from the cocoons of the domesticated silkworm Bombyx mori or from wild cocoons (Gheysens et al., 2011; Kundu et al., 2012), or indeed on turning those silks into useful textiles or other products. In contrast, the structure of the cocoon itself has largely been ignored, although recent research shows that cocoons can be considered as hierarchical composites composed of silk fibres and sericin bonds, which confers a number of unique properties to the cocoon (Chen et al., 2012a; Chen et al., 2012b; Chen et al., 2012c).

Cocoon properties may be of great interest in the biomimetic development of flexible, damage-resistant, lightweight composites. For example, many silkworms have silk cocoons with a laminated structure (see examples of B. mori, Antheraea pernyi and Opodiphthera eucalypti cocoons in Fig. 1). Although the cocoons of each species have their own individual features, they all have a multi-layer structure with fewer fibres connecting layers than aligned in the individual layers. The interlayer bonding is much weaker than the intralayer bonding. The porosity of the cocoon layers decreases from inner to outer layers (Chen et al., 2012b). This structure shares a morphological similarity with conventional laminated composites, in which the fibre prepregs are stacked and bonded by resin matrix between the layers. Furthermore, cocoon structures (and therefore cocoon properties) are optimised for combinations of different biological functions across the species range against a range of different threats and environmental challenges. Importantly, the animal uses a very limited number of material components, usually only silk fibre and sericin. Some cocoons also have additional features, such as extra calcium oxalate crystals on the cocoon surface in A. pernyi cocoons or the incorporation of hairs and spines (Chen et al., 2012a). In addition, some cocoons might display rather large holes, e.g. the cocoon walls of O. eucalypti (Fig. 1, insets), which will be relevant for discussions later in this paper. As the predominant nonwoven structures in a number of silk cocoons have a similar microstructure to other stochastic fibrous materials (such as paper, nonwoven textiles and electrospun polymer mats), the bioengineering design principles evolved in silkworm cocoons make them excellent natural prototypes and models for the study of structural composites (Chen et al., 2010b).

However, in order to fully utilise the biomimetic potential of the cocoons, we need to better understand the natural (native) relationships between their structure and function, which have been ‘optimised’ over millions of years of evolution. All cocoons appear to be multifunctional and possible functions include camouflage (Danks, 2004), thermoregulation (Lyon and Cartar, 1996) and humidity control (Blossman-Myer and Burggren, 2010). However, despite surprisingly little experimental data, the main function of all cocoons is likely to be the mechanical protection of the enclosed pupae from predators (Scarbrough et al., 1972; Waldbauer, 1967), similar to the protection against both general and specialised predators provided by the silken cocoons enclosing spider egg sacs (Hieber, 1992). Clearly, the exact way in which the cocoon offers protection depends on the nature of its main predators.

There seem to be three main strategies that predators employ to penetrate a cocoon. (1) Penetration of the cocoon wall by drilling. This strategy is employed by ichneumonid wasps, which lay eggs in or on the pupae that develop into larvae that parasitise the pupae (Marsh, 1937). (2) Creating a hole in the cocoon by removing silk by tearing and shearing movements. This strategy is employed by mice and other small mammals (Scarbrough et al., 1972). (3) Creating a hole in the cocoon by breaking individual fibres and the cocoon wall by making an indentation in the cocoon. This strategy is employed by birds, which either use their beaks to compress the cocoon walls or, like woodpeckers, attempt to directly hack holes in the cocoons (Waldbauer, 1967).

All three strategies involve impact strikes on the outer cocoon wall. Defences against these strategies would include the ability to either resist the impact with little damage, or to tolerate the damage by maintaining the structure. As a cocoon acts as a protective structure for the pupa, it should evolve to have engineering benefits such as impact resistance and damage tolerance by combining the structure and material properties. In the present study, we focus on how cocoons might have evolved counter-measures against the third predator strategy (creating a hole) using an engineering approach. We describe how cocoons deform during quasi-static localised indentation, and we analyse these observations with the help of a simple finite element analysis (FEA) model designed to simulate this behaviour and highlight key features of the cocoon deformation and failure mechanisms. This combination of experiment and simulation provides important insights into the defensive strategy of silkworms, whose cocoons have structural and materials properties that appear optimised for achieving a high impact resistance and damage tolerance.

Importantly, FEA has previously been extremely useful to better understand the form and function relationships of spider webs (Hesselberg and Vollrath, 2012; Ko and Jovicic, 2004; Lin et al., 1995). Indeed, one might go as far as to say that without this engineering approach such insights would have been impossible (Vollrath, 2000). After all, FEA requires detailed statements of the mechanical properties of all component parts, as well as (for truthing) solid empirical data on the behaviour of the composite itself in addition to a good understanding of the biological and abiotic (environmental) frame parameter under which the structure operates and for which it has evolved. While spider's webs are lightweight, airborne net structures that fish for aerial prey, silkworm cocoons are complex, rather solid composite constructions that house a delicate inhabitant. Yet in both cases FEA provides key insights into structure–property–function relationships.

The cocoons used in this study were obtained through commercial trade. Bombyx mori (Linnaeus 1758) (family Bombycidae) cocoons were acquired from ICIPE (Nairobi, Kenya). Antheraea pernyi (Guérin-Méneville 1855) and Opodiphthera eucalypti (Scott 1864) (both family Saturniidae) cocoons from Korea and Australia were purchased through Worldwide Butterflies (Sherbourne, UK). All the cocoons were spun in free space without attaching cocoon walls to the substrates. In order to keep the intact structure of the cocoons, samples were frozen at −5°C for 24 h to kill the pupae inside before the emerging adults tunnelled through the cocoon and broke the structure.

The cocoons were then cut into two halves to release the pupae without contaminating the cocoon materials. This was done to isolate the effects of the material properties of the cocoon from any potentially confounding effects of the pupae. The geometry of the cocoons replicated in the models was measured using a Mitutoyo IP65 micrometer (Andover, UK).

The tensile strength and modulus of the cocoon materials were measured with a crosshead speed of 2 mm min⁻¹ in an Instron 5542 (Instron, High Wycombe, UK). Test specimens were individually cut from the cocoon wall with a sharp blade into dogbone-shaped samples having a width of 5 mm and a length of 15 mm. Opodiphthera eucalypti cocoon samples were cut to avoid the fabricated holes on the cocoon wall. The gauge length of samples in the tensile test was 5 mm. The samples were aligned using a rectangular guiding tool from the Instron instrument. Three samples of each cocoon species were tested.

Each half of the cocoon samples was affixed (with superglue) at its cut edge onto a glass slide. The central points of the samples were measured and marked to be the compressive points. Opodiphthera eucalypti cocoon samples were cut to avoid the fabricated holes on the cocoon wall. The tests were carried out in a Zwick/Roell Z0.5 test machine (Ulm, Germany) under displacement control at a loading rate of 5 mm min⁻¹. The indentation load was applied through a steel rod with a round flat end (1.56 mm diameter). The unloading was performed at a speed of 15 mm min⁻¹. The samples flexed back but did not recover completely their un-deformed shape. The cocoon samples respond to form a concave area under the indentor. The shapes of cocoons crushed during the indentation until the indentor reached the slide underneath the sample were recorded.

For the finite element modelling, we performed simulations of the quasi-static response of cocoons in order to understand the failure modes caused by the local indentation described above. In previous work we have shown that models based upon open-cell foam structures describe the mechanical properties of silk cocoons (Chen et al., 2010b). Here we therefore use the finite element software ABAQUS/CAE 6.8-4 (Dassault Systèmes, Vélizy-Villacoublay, France) to simulate the indentation behaviour of the cocoons in this study, as previous work on simulating deformation in foam materials was carried out using ABAQUS (Gilchrist and Mills, 2001; Li et al., 2000; Rizov et al., 2005).

An axisymmetrical idealisation was used for the geometry of the three cocoon types. The mesh consisted of 94,400 finite elements with two elements through the wall thickness (Fig. 2). All degrees of freedom at the boundary of the specimen bottom were constrained, simulating a rigid fixed support. The indenter was modelled as a force applied to a circular area on the upper surface of the cocoon in the program. All degrees of freedom of the indenter were constrained, except in the vertical direction.

Four types of silk cocoon were categorised according to their different mechanical behaviours in a previous paper (Chen et al., 2012c): (1) ‘lattice’ cocoons, with a loose scaffold structure, (2) ‘weak’ cocoons, with high porosity and weak interlayer bonding properties, (3) ‘brittle’ cocoons, with low porosity, strong interlayer bonding and brittle mechanical properties, and (4) ‘tough’ cocoons, with medium porosity, interlayer bonding and tough mechanical properties. Here a typical cocoon was selected from each of the latter (2–4) three nonwoven cocoon categories in order to investigate their failure mechanisms. Bombyx mori cocoons have high porosity: the inter-fibre bonding breaks gradually with the increasing strain, leading to a reducing modulus; the stress drops when the breakage of the structure reaches a percolation point, then the unbonded fibres unravel from the structure. Opodiphthera eucalypti cocoons, in comparison, have low porosity and strong bonding between the fibres; a crack starts from the sericin binding matrix and propagates to the fibres; the structure fails when the fibres break perpendicular to the tensile load, which is seen in the stress–strain curves as a dramatic drop of the stress (Fig. 3). The A. pernyi cocoon has a medium porosity compared with the other two cocoons: it exhibits yielding below 20% strain, which absorbs the energy to produce a tough behaviour; the structure fails when more than half of the sericin binder matrix fragments; the fibres then pull out from the structure, which is seen as a stress tail in stress–strain curves.

We have previously shown that B. mori cocoons have a graded layer structure, where the innermost layer exhibits a strong but brittle behaviour, while the outermost layer is usually the weakest layer (Chen et al., 2012b). This is due to the gradual increase of porosity and the decrease of the number of sericin binding points from the innermost to the outermost layer. In this paper, a similar behaviour of A. pernyi cocoons is shown. The fibres break at the innermost layer and disentangle in the outer layers. The graded morphology observed from the scanning electron micrographs in Fig. 1 confirms the same mechanism as in the B. mori cocoons. Opodiphthera eucalypti cocoons show a clean failure in the structure because of their low porosity, even in the outermost layer. However, the sericin bonds still start at the outermost layer and then propagate into the inner layers. We conclude that all three nonwoven cocoons tested have a graded layer structures in which the mechanical strength increases from the outer to the inner layers.

Typical load–indentation behaviour and load–deflection curves for the three cocoon types are shown in Figs 4 and 5, respectively, with deflection given as a fraction of the maximum strain until the indenter reached the glass plate and with load shown on the same scale for all the cocoons for comparison. The B. mori cocoon load–deflection curve shows a gradual decrease in stiffness before the stress reaches a plateau level with the emission of a noise (cracking sounds) (Fig. 5), which are attributed to the extensive sericin breakage and delamination in the cocoon sample (Fig. 6). The damage process is also reflected in the fluctuations in the load response. Bombyx mori cocoons have the largest delamination areas among the three cocoons and nearly all the interconnectivity between layers in the concave area is broken (Fig. 6). The B. mori response is relatively consistent, and the five curves shown in Fig. 5 reflect the full range of responses, which are attributed here to relatively consistent size and shape of the cultivated cocoons, compared with the other wild cocoons.

The indentation behaviour of A. pernyi and O. eucalypti cocoons has much more variation that that of B. mori cocoons (Fig. 5). This is attributed to their inconsistent geometrical shapes and sizes arising from them being spun in the wild, in contrast to the cultivated B. mori cocoons. Generally, both cocoons also have nonlinear behaviour (Fig. 5). The onset of inter-laminar delamination of the cocoons is again associated with the gradual decrease of the stiffness in the load–deflection curves. After that, the cocoons are extensively damaged, which can be seen from the pronounced fluctuations in the load–deflection curves. Antheraea pernyi cocoons have extensive sericin breakage and some delamination after indentation tests. In contrast, general fibre and layer breakages are found in O. eucalypti cocoons with less damage propagation and large-scale failure found close to the indentation area. The indentation terminates with a clear penetration of fibre and layer breakage in O. eucalypti cocoons (Fig. 6).

One of each of the three experimental tests from both A. pernyi and O. eucalypti cocoons were selected for finite element modelling. Their specific geometrical data were collected for FEA calculations, and their experimental curves were used for comparison with the simulation results.

An open-cell foam model was used to describe the tensile properties of cocoon walls based upon the similarity of cocoon morphology with that of cellular materials (Chen et al., 2010b). The compression behaviour of cocoon shells can be described by a comparable open-cell foam model using the appropriate deformation geometry of the cell walls under compression (Chen et al., 2010b). To complicate things further, the inter-laminar debonding contributes to the mechanical properties of cocoon walls and thereby must be included in the deformation and failure mechanisms, particularly where shear (hence bending) deformation is involved through the wall thickness (Chen et al., 2010b).

The plastic compressibility of foams results from the collapse of the cell walls due to buckling, breaking or yielding (Rizov et al., 2005). A localised progressive crushing mechanism for foams is described by Li et al. (Li et al., 2000), where the plastic deformation is usually localised into the weakest cell layer and the strain increases rapidly in this layer and then triggers the next layer to crush. This process is very similar to the bond fission and delamination processes of cocoon damage. Because such foam compression simulations have previously been calculated and modelled using ABAQUS (Li et al., 2000), we expected to find an appropriate constitutive model in the software that would implicitly include damage occurring in the plasticity of the irreversible process of cocoon indentation

All these parameters come from fitting experimental tensile stress–strain data, as shown in Fig. 7 and Table 1. As this is a deformation plasticity model, which is actually an elastic model modified to include yielding behaviour, it does not include the rapid drop in stress at the percolation damage point in each experimental curve. This model is chosen to simply suggest the mechanism for initiation of the cocoon collapse in areas where the modulus reduces significantly under load without including the damage propagation.

Fig. 8 shows the evolution of the stress on the outer surface of the cocoons at three increasing indentation loads and also the stress through the thickness of the cocoon wall in a cross-section through the long axis, with red being the highest stress and blue the lowest. Three different sets of indentation loads were selected to show the cocoon deformations evolved from elastic, initial yield and plastic stages. The cocoon sample has a highly deformed zone, which is localised around the indenter, while other parts of the cocoon undergo far less deformation. The highest tensile stress is concentrated on the outer surface of the minor axis of the cocoon.

Most of our experiments and simulations of cocoon indentation were carried out on B. mori cocoons and hence we concentrate on these results in order to deduce general trends. The measured and simulated static indentation responses of B. mori cocoon samples are shown in Fig. 9. Simulation can generally reproduce the concave shapes of the cocoons seen in experiments at small deflections. However, with high deflection, the shape deformation in the simulations tends to concentrate too much around the indenter. In the experiments, the indenter gives a broader crater area of deformation of the cocoon, thereby giving a lower crater height.

Different constitutive models for the cocoon wall material were tried, including hyper-elastic models and even simple rubber models, on the B. mori cocoon FEA (data not shown). The general concave shape of the cocoon indentation crater can be reproduced almost universally, but clearly with poor predictions of the load, which suggests that the general cocoon response behaviour is generic and independent of the material properties. The geometric shape of the cocoon appears to have more effect on its indentation behaviour.

The measured and the simulated static indentation load responses with the same cocoon geometries of B. mori are given in Fig. 10, but using different sets of material parameters in the constitutive model. For the B. mori material data shown in Fig. 10, the initial elastic slope of the simulated load–indentation curve is in good agreement with the experimental results. However, at large deflections the numerical simulations do not reproduce the overall ‘yielding’ profile; the finite element modelling and the experimental study diverge at approximately 20% deflection. Using the constitutive model parameters for A. pernyi simply shifts the simulation to higher loads due to the higher modulus. The simulations of A. pernyi and O. eucalypti cocoon also reproduce the experimental results before the materials yield (Fig. 11).

Fig. 12 shows the distribution of the stress and strain that corresponds to the divergent point between the experimental results and the numerical simulation at approximately 20% B. mori cocoon deflection. The inner surface of the cocoon centre, which is localized under the indenter, has a highly deformed zone. The outer surface of the edge of the bending concave crater (circled) also develops the highest general strain, which is locally above 4% strain due to the tensile force distribution on the outer surface. This happens to be the breaking strain of sericin (Fig. 13) (Chen et al., 2010a).

The experimental observations suggest that this divergence between simulation and experiments indicates that the model cannot predict the initiation of damage. In the cocoon, the initiation of the damage is the sericin breakage between the layers at the surface of the cocoon (Chen et al., 2010b). To explore this hypothesis, a B. mori sericin model was used in the indentation simulations. This model uses the sericin properties (Fig. 13) and has the same stress–strain behaviour as B. mori cocoon material at low strain, but is constrained to yield at 4% strain as for pure sericin (Fig. 14). The intention was to examine whether the local stress–strain response in the cocoon walls is dominated by sericin failure.

The sericin model can reproduce the experimental results at low strain shown in Fig. 10, but the simulations would not converge above a deflection of approximately 30%, which corresponds to a local strain at the outer surface of 4.5% and suggests failure of the sericin binder material. This suggests that the point at which the simulations diverge from the experiments is the initiation of the sericin breakage, which is an irreversible damage in the cocoon structure that is not included in the constitutive model. The point at which the sericin damage threshold is reached should correspond to delamination of the cocoon and loss of connectivity in the planar interfibre bonding. At this point the composite material cannot sustain load, and the stress is transferred in a cascade to ever-deeper layers of the cocoon structure to propagate the damage process through the full cocoon thickness, resulting in no more increase in the indenter load.

The present study demonstrates that the silk moth cocoons show adaptations to resist standardised impact forces in the laboratory, which suggests that cocoons could similarly be adapted to resist impact forces generated by predators in the wild. The moth larvae appear to employ a number of different strategies while building the cocoons to increase their impact resistance and damage tolerance by combining their macro-structural shape, micro-layer structure and properties, and component materials. These hierarchical organisation and complicated engineering properties are inherent to their design.

The B. mori cocoon is the most comprehensively studied type of silkworm cocoon in terms of structure and properties (Chen et al., 2012a). So it was used as the main example for analysing impact resistance and damage tolerance in cocoons. Our experimental and modelling results both suggest that the silk cocoon may indeed be mechanically suited to withstand impact force. This is because of: (1) the ellipsoid shape of the cocoon, (2) the component material properties, (3) the laminated layer structure and (4) the graded layer properties. Both the experiments and models show that loading the cocoon centre leads to a concave shape of deformation (Fig. 9). The ellipsoid shape of the cocoon allows a certain level of elastic deformation of the cocoon wall, and then restricts the deformation to avoid propagation of damage. In contrast, FEA simulations suggest the deformation process and the deformed shape of the cocoon is relatively independent of its material properties (Fig. 10). The damage is localized at the edge of the bending concave crater without excessive propagation of breakage. The material properties affect only the elastic deformation rather than the damage area. This feature helps to keep an intact cocoon structure, and provides a good support to avoid massive deformation onto the pupa inside the cocoon by using limited materials.

In biological structures, the component material properties usually closely link with the structure, as they evolve together to fit the biological functions (Vincent, 2012; Meyers et al., 2008; Vogel, 2003). This is also found in the B. mori cocoon in its relationship between component properties and laminated structure (Chen et al., 2012b). The damage of the B. mori cocoon can be described by two coupled failure modes: sericin bonds cracking and delamination. FEA shows that sericin cracking is the first fracture event around the edge of the cocoon concave during an impact process due to its relatively low breaking strain compared with the fibres (Figs 12, 13). After extensive breakage of sericin bonds, delamination (the crack in the sericin bonds between layers) develops, inducing separation of the individual layers, which has been observed from our experimental results (Fig. 6). This sacrificial breakage allows each fibre layer to deform semi-independently rather than transferring the stress through the layers. In contrast, silk fibres with a high strain to failure adapt within this individual layer structure (Fig. 13). The fibre layer can have an elastic deformation without damaging the fibre to maintain its own mechanical properties in the fibre directions. This gives maximum energy absorption and increases the toughness of the structure.

The graded layer structure is another feature that may have evolved to improve the impact resistance of the cocoon. This is based on the FEA simulation results that the largest deformation of the cocoon happens at the innermost layer of the area under the impacter (Fig. 12). After delamination damage, the individual layer would be able to deform independently, and the properties of each layer would determine its own damage mechanism. We have learnt from our previous publication that the mechanical properties of cocoon layers decrease from the inner to the outer direction (Chen et al., 2012b). This is also confirmed by tensile tests of the cocoons in this project (Fig. 3). FEA simulation, in contrast, shows that the deformation of layers decreases from inner to outer layers (Fig. 12). It has been pointed out by Zhao et al. (Zhao et al., 2005) that this material–structure relationship may have evolved to adapt to the cocoon bending behaviour, which is part of the impact deformation, as the inner layer would sustain more tensile force than the outer layers.

The A. pernyi cocoon has a breakage mechanism similar to that of the B. mori cocoon, where the damage initiates at the brittle sericin and leads to an extensive delamination between the layers (Fig. 6). The separated layers deform to absorb the impact energy and the relatively strong inner layers provide stiffness and stress to the structure. Given the species-specific geometry and identical loading regimes, peak stress was lower in the A. pernyi cocoon because of its bigger size, leading to less damage and higher impact resistance (Fig. 5). Further, the calcium oxalates crystals found on the cocoon surface (Chen et al., 2012a) may also help to improve the hardness of the materials and therefore the impact resistance.

In comparison, the O. eucalypti cocoon seems to employ another strategy to protect the pupa. It has a high stiffness material property from its fibre connectivity structure because of its low porosity (Figs 1, 3). This reduces the ability to deform the structure and damage of the materials, resulting in a high impact resistance. Apart from sericin cracking, fibre breakage is its significant energy-absorbing mechanism. It leads to a catastrophic penetration when the sericin crack propagates to the fibres and the failure reaches a critical extent (Fig. 6). This mechanism localises the damage under the impacter, avoids excessive damage and maintains the structure integrity in other undamaged areas. In nature, O. eucalypti seems to have tolerance of holes in the cocoon, as it fabricates woven holes around its attachment in the cocoon construction (Fig. 1). This may be a compensating strategy for ventilation as it has very low porosity in the cocoon walls compared with other species (Chen et al., 2012c). These fabricated holes do not appear to interfere with the structure and mechanical properties of the rest of the cocoon wall, unlike other nonwoven cocoons. In this case, it is not surprising that the O. eucalypti cocoon would employ a penetration-breaking mode in its cocoon impacting behaviour.

The impact behaviour of cocoon composites is very similar to that of conventional composite laminates, where layers of aligned unidirectional carbon fibres stack with epoxy matrix bonding between the fibres (Chen and Hodgkinson, 2011). Materials and structural design are the key tools for controlling energy absorption and toughness of composites. These design principles in composites engineering can be used to understand the design criteria of cocoon structure by connecting the materials and structure with their engineering functionality. On the one hand, comparative analyses of cocoons from a broader sample of wild species are essential to test whether the observed cocoon properties and structures are a result of adaptations to predators employing different strategies to penetrate the cocoon walls. On the other hand, as a complicated engineering product arisen from millions of years of evolution to fit their individual functions, silk cocoons may be able to provide a wealth of examples of how to apply biomechanical engineering principles in the design of synthetic structural composites.