The housing tube material of the marine worm Chaetopterus sp. exhibits thermal stability up to 250°C, similar to other biological materials such as mulberry silkworm cocoons. Interestingly, however, dynamic mechanical thermal analysis conducted in both air and water elucidated the lack of a glass transition in the organic tube wall material. In fact, the viscoelastic properties of the anhydrous and undried tube were remarkably stable (i.e. constant and reversible) between –75°C and 200°C in air, and 5°C and 75°C in water, respectively. Moreover, it was found that hydration and associated-water plasticization were key to the rubber-like flexible properties of the tube; dehydration transformed the material behaviour to glass-like. The tube is made of bionanocomposite fibrils in highly oriented arrangement, which we argue favours the biomaterial to be highly crystalline or cross-linked, with extensive hydrogen and/or covalent bonds. Mechanical property characterization in the longitudinal and transverse directions ascertained that the tubes were not quasi-isotropic structures. In general, the higher stiffness and strength in the transverse direction implied that there were more nanofibrils orientated at ±45° and ±65° than at 0° to the tube axis. The order of the mechanical properties of the soft–tough tubes was similar to synthetic rubber-like elastomers and even some viscid silks. The complex structure–property relations observed indicated that the worm has evolved to produce a tubular housing structure which can (i) function stably over a broad range of temperatures, (ii) endure mechanical stresses from specific planes/axes, and (iii) facilitate rapid growth or repair.
The marine worm Chaetopterus variopedatus (Polychaete, Chaetopteridae) lives in self-constructed cylindrical U-shaped parchment-like tubes typically buried in soft seafloor sediment with both the tapered extremities exposed. Regarded as one of the most structurally specialized of all annelids , Chaetopterus animals have a cosmopolitan distribution, typically occurring in mud- or sand-flats in the intertidal zone or in shallow coastal waters (less than 50 m deep) [2,3]. Notably, dense aggregations of similar chaetopterids and their characteristic tubes have been found at even deep-sea sites (greater than 1000 m deep), near hydrothermal vents, around mud volcanoes, at cold seep sites and around whale-falls . This suggests that not only the chaetopterid worm, but also their self-secreted tubes have a broad range of environmental tolerance, including high subsea pressure, temperature, salinity and pH levels . Our own anecdotal observations of tubes kept in aquaria of solutions of seawater, 70% ethanol or various pH, for several years also draw attention to the sturdy nature of the tubes as these show no apparent change in shape, size, texture, among other properties. Given that the polychaete expends considerable energy building the tube, in fact up to double the energy spent on somatic growth and gamete production combined [3,5], the tubes represent both an energy and organic material investment.
Other than the broad environmental occurrence of Chaetopterus tubes, their perceived mechanical function has never been characterized. In Nature, these tubes experience a combination of mechanical stresses. These include compressive stress or external pressure from sediment weight, and cylinder stresses (i.e. hoop, radial and axial stresses) from (i) hydrostatic loads from water pressure, particularly at deep-sea sites and (ii) the axial flow of water through the tube. The flow is actively produced by the worm via a pressure differential created by vigorous rhythmic peristaltic contractions and expansions (in radial and axial directions) of the piston-like pumping body segments [1,6], enabling feeding, respiration and tube-cleaning [1,6–8].
Previously, studies have examined the highly oriented lamellofibrillar ultrastructure of Chaetopterus tubes [4,9–11]. The composite tubes were found to consist of up to 20 multi-axially oriented non-woven plies of aligned nanoscale (diameter: 50–1000 nm) fibrils bound together with a diffuse organic matrix (figure 1). While the orientation angle of the nanofibrils in the plies varied between tubes, and even further along the same tube (figure 1), possibly owing to a new layer or growth being added constantly or periodically, quantitative orientation distribution analyses revealed that nanofibrils were principally oriented at 0°, ±45° and ±65° to the tube axis .
As in man-made fibre-reinforced composite materials, in the tubes, the orientation of the high aspect ratio and (therefore) anisotropic reinforcing fibres has a significant effect on material stiffness and strength . Importantly, the determined dominant orientation angles (namely 0°, ±45° and ±65°) are known to be optimal winding angles for fibre-reinforced composite pipes subjected to bending and axial loads, torsion/shear loads and combined external pressure and buckling loads, respectively . It is therefore justified to hypothesize that the biocomposite tube structures have evolved to confer a high degree of mechanical stability and resistance against deformation and tearing forces. Here, we examine this hypothesis.
Owing to the fineness of the nanofibril layers, we were not able to measure the fibre volume fraction in the tube structure, let alone the actual fraction of filaments or layers in each orientation . Therefore, while we know that the filaments are preferentially aligned, we cannot determine whether the structures are indeed quasi-isotropic. Tube anisotropy, however, can be evaluated by testing tensile properties in orthogonal directions. If properties in the orthogonal directions were nominally equivalent, then it could be claimed that the structure would be quasi-isotropic and that the worm is not reaping benefits of an oriented structure by ‘deliberately’ aligning anisotropy along a principally loaded axis/plane. In that case, the worm would be producing oriented layers for some other, if not another, incentive.
Indeed, lamellar tubes orientated in the manner observed could be ‘grown’ quickly [11,13] and at lower energy consumption in comparison with unoriented or semioriented tubes, by rapid lay-up of a ‘self-assembled’ mat over previously deposited mats. However, this would imply the presence of a mechanism by which the worm is capable of detecting or using as a nucleating (or initiation) template the orientation of previously deposited layers . Unless, of course, all fibrillar meshes of a regular set of orientations (e.g. [0°, ±45°, ±65°]) were sequentially added in one interlinked set of action patterns, which would ensure that a preferred orientation (or lack of) is maintained, by using the first or most recent deposited layer as a template that coordinates the next or all subsequent layers .
To investigate some of the issues raised above, we characterized the thermal mechanical properties of Chaetopterus tubes. Thermogravimetric analysis (TGA) and dynamic mechanical thermal analysis (DMTA) were employed to assess the thermal degradation profile and the thermomechanical stability of the tubes. The tensile mechanical properties of the tubes in both longitudinal and transverse directions were determined and compared. Tensile tests were also conducted on wet and dry samples, and on carefully separated layers of the tube structure. In addition, we studied the variation in tensile properties of ring samples obtained from the top (new growth), middle and bottom (older growth) sections of the tubes. A discussion of the measured properties in relation to the tube material, morphology and function is also presented.
2. Material and methods
2.1. Tube collection
All the Chaetopterus tubes used in this study originated from the La Jolla submarine canyon in San Diego, CA, USA as described elsewhere [11,14]. In spring 2013, bundles of tubes were hand collected by scuba at 20–30 m depth and taken to the Marine Biology Experimental Aquarium Facility at Scripps Institution of Oceanography. The worms were left to grow in their tubes, kept in circulating seawater at ambient temperature, with no additional food. Most tubes rapidly showed new growth. Sections of tubes (from the freshly grown tip but also from older sections) were then carefully cut off using a scalpel blade and shipped to the Department of Zoology at the University of Oxford, in Falcon tubes with 70% ethanol. Once received, the tubes were kept in cold (7°C) fresh marine water (34.5 PSU).
2.2. Thermogravimetric analysis
To ascertain the thermal and thermo-oxidative stability of the tubes, TGA was performed on ambient-dried and equilibrated samples using a TGA Q500 (TA Instruments) from ambient to 800°C with a heating rate of 10°C min−1 under both a 100 ml min−1 nitrogen (i.e. inert) and air (i.e. oxidative) gas flow, respectively. The weight loss and the derivative weight loss during temperature scanning were recorded. Using the multiple curves averaging function in OriginPro 9 (linear interpolation over the common temperature range), an average curve of three replicate samples was obtained.
2.3. Dynamic mechanical thermal analysis
DMTA was conducted using a DMTA Q800 (TA Instruments) in film-tension multi-frequency strain mode, to investigate the evolution of the tensile storage modulus (E′), the loss modulus (E′′) and the damping factor tan δ (=E′′/E′) of the tubes with temperature. We prepared specimens for analysis by cutting rectangular strips (20 mm long, 6 mm wide, 10 mm gauge length) from various positions along the tube length. Specimens were loaded in the wet state at initiation of the analysis.
Static-dynamic tests were conducted in air, applying single and double thermal cycles. For the former, specimens were subjected to a thermal scan from –25°C to 300°C, whereas for the latter, specimens were subjected to double thermal cycles between –75°C and 200°C to specifically study hysteresis effects resulting from moisture loss. Specimens were first equilibrated for 5 min at the low temperature. In both test regimes, the following parameters were used: preload of 0.01 N, heating/cooling rate of 3°C min−1, frequency of 1 Hz and applied strain of 0.1%.
In addition to the tests conducted on wet samples in air, the DMTA instrument was employed to run double thermal cycles between 5°C and 75°C on samples submerged in water throughout the test duration. This enabled the study of the thermomechanical properties of the ‘undried’ tube wall material in its typical environment. The following parameters were used: preload of 0.01 N, heating/cooling rate of 0.5°C min−1, frequency of 1 Hz and applied strain of 0.1%.
2.4. Mechanical tensile testing
The longitudinal (i.e. along the tube axis) tensile properties of the parchment tubes were measured using three different methodologies, listed in table 1, for comparative purposes. At least five samples were tested for each methodology.
Tensile tests on various other specimen types were carried out in film-tension mode using the DMTA instrument (samples not submerged), with the same methodology as in table 1. After carefully peeling longitudinal tube samples into two layers, the tensile properties of the different layers were evaluated. Furthermore, ring samples from the top section (new growth), middle section and bottom section (older growth) of parchment tubes were also subjected to tensile tests to determine properties in the transverse direction. Care was taken to ensure that the samples were wet throughout the duration of the test. For comparison, longitudinal samples dried under ambient conditions over 48 h were tensile tested.
Stress was calculated assuming a uniform rectangular cross section. For all the tensile tests, the stress–strain curve of each sample was used to determine the tensile modulus (in the strain range of 3–4%), tensile strength and elongation at break. For any statistical comparison, difference of means one-tailed t-tests (at α = 0.05) were conducted.
2.5. Scanning electron microscopy
Tensile fracture surfaces were examined using a JCM-5000 NeoScope (JEOL) scanning electron microscope (SEM), at an acceleration voltage of 10 kV under high vacuum. Sample preparation included drying and equilibrating of the tube in ambient conditions for 48 h, followed by sputter coating with gold/palladium (Au/Pd) alloy.
3. Results and discussion
3.1. Thermal and thermo-oxidative stability
The thermal degradation profile of the ambient-dried tubes is presented in figure 2. The initial weight loss of 7–9% up to 200°C was principally related to moisture evaporation. A comparable water-related mass loss is observed in various biological materials, including the cocoons of Bombyx mori silkworms (see electronic supplementary material, figure S1). While weight loss stabilized after 100°C in an inert atmosphere, it increased fairly linearly in an oxidative atmosphere (figure 2). The onset of both thermal and thermo-oxidative degradation was at approximately 250°C, with the maximum rate of degradation at approximately 270°C. The significant weight loss between 200°C and 450°C was attributed to decomposition of organic material (i.e. nanofibrils and matrix), possibly protein or polysaccharide end chains. In comparison, the cocoons of mulberry silkworms, chitin and collagen are all thermally stable up to 250–300°C [15,16] (see electronic supplementary material, figure S1).
Further weight losses between (i) 450 and 650°C and (ii) above 650°C were attributed to (i) charring or further denaturation of organic compounds or breakdown/volatilization of inorganic compounds and (ii) decomposition of only inorganic compounds, respectively (figure 2). It is interesting to note that mulberry silkworm cocoons exhibit three thermal decomposition peaks in an oxidative atmosphere: a distinct peak at about 300°C, a shallow peak at about 440°C and a broad dual peak between 550°C and 600°C (electronic supplementary material, figure S1) . All these peaks and transitions match those of the tubes at approximately 270°C, approximately 400°C and 566°C (electronic supplementary material figure S1), which suggests that the thermal degradation of the organic material in mulberry silks and Chaetopterus tubes follows comparable patterns, which in turn suggests that the structure or composition of the two organic materials may be analogous.
Cocoons of mulberry silkworms, chitin and collagen have a residual weight of less than 5% (see electronic supplementary material, figure S1) [15,16]. The high residual weight of 50–70% of the marine worm tubes and the presence of weight loss peaks much above 600°C (figure 2) represent the high content of inorganic contaminants such as sand particles, attached to the surface of the tubes .
3.2. Thermomechanical properties
TGA revealed the mass loss profile of the tubes when subjected to high temperatures. The underlying design of a structure can be further examined by measuring also the mechanical property profile when subjected to a range of temperatures: from subzero to above thermal decomposition temperatures.
3.2.1. Thermal scanning in air
The viscoelastic properties of the tubes were scanned in air from –25°C to 300°C (figure 3), after a 5 min isothermal hold at –25°C. The transitions in properties were well marked by the (rate of) changes in the specimen length (figure 3). The storage and loss moduli gradually reduced with increasing temperature to a minimum value at about 20°C, owing to the reduction of the stiffness of ice with increasing temperature  and the subsequent melting of ice above 0°C.
At ambient temperature (approx. 20°C), the storage modulus, loss modulus and loss factor measured using DMTA were 10.3, 1.5 and 0.14 MPa, respectively. The loss of moisture, indicated by contraction in specimen length (figure 3), dictated much of the material's above-ambient temperature properties. In particular, moisture-loss-related specimen contraction was rapid between ambient and 65°C, but gradual between 65°C and 200°C. The changes in viscoelastic properties followed a similar trend. This indicates that at approximately 65°C perhaps a critical threshold was reached, above which further loss in moisture and thus reduction in ‘free volume’ occupied by bound water in the porous tube structure  had a relatively insignificant effect on the mechanical performance of the tube material.
Specifically, an increase in temperature above ambient led to rapid increases in both the storage and loss moduli; the loss modulus peaked to 9.1 MPa at 40°C, whereas the storage modulus was 50.1 MPa at the same temperature (figure 3). Notably, the loss modulus peak coincided with the loss tangent peak (tan δ ≈ 0.19). Above 40°C, the loss tangent collapsed and stabilized to about 25% (tan δ ∼ 0.05) of the peak value, and the loss (dissipation) modulus collapsed and stabilized to about 67% (E′′ ∼ 6 MPa) of the peak value. This was indicative of the significantly reduced damping performance of the samples above 40°C. Moreover, the storage (dynamic) modulus continued to increase rapidly until about 65°C, reaching a value of 125 MPa, thereafter increasing at a gradual rate. In essence, below 40°C, the tube worm wall material was rubber-like (i.e. soft, flexible, damping), whereas above 40°C, the material behaviour was glass-like (i.e. rigid, brittle, non-damping).
Between 65°C and 200°C, the loss modulus and the loss factor were fairly constant, whereas the storage modulus increased at a gradual rate by approximately 20% owing to moisture-loss-related stiffening (figure 3). Importantly, this feature demonstrates the high thermomechanical stability of the tube structure. Above 200°C, (i) a slightly increased rate of specimen length contraction, (ii) a reduction in the storage modulus, (iii) an increase in the loss modulus, and (iv) an increase in the loss factor were observed (figure 3), which may be the onset of mechanical thermal degradation. This result matched well with the previous finding of a significant thermal degradation peak of the organic tube material at an onset temperature of approximately 250°C (figure 2).
Importantly, the lack of a glass transition in the DMTA curve (figure 3) implies that the tube material would have some form of highly crystalline or highly cross-linked, hydrogen- or even covalent-bonded structure. Indeed, only slow (rather than rapid) changes in viscoelastic and physical (i.e. specimen length) properties above thermal decomposition temperatures (figure 3) reinforce this assertion.
In comparison with the cocoons of mulberry silkworms, the thermomechanical performance of Chaetopterus sp. tubes in air is generically different in the temperature range studied (electronic supplementary material, figure S2). Unlike the tubes, silkworm cocoons exhibit a glass transition temperature of approximately 200°C , above which the material becomes softer indicated by a significant increase in the loss factor and a significant reduction in storage modulus (electronic supplementary material, figure S2). Notably, while the loss factor increases from 0.03 to 0.17 between 40°C and 200°C for silkworm cocoons, the loss factor decreases from 0.19 to 0.05 between the same temperature range for the tubes. In addition, the loss modulus peaks at about 200°C for silkworm cocoons , compared with 40°C for the tubes. In essence, at least in the temperature range studied, the tubes became more glass-like (less damping) with increasing temperature, whereas silkworm cocoons become more rubber-like (more damping).
In general, the lack of a distinct glass transition peak indicates that the Chaetopterus tube material is unlikely to have amorphous phases, which is characteristic to silks for example. The property differences of the anti-predatory tubes and cocoons built by the annelid and insect, respectively, are likely to be related to the difference in materials (their chemical nature and their physical arrangement). Most apparent is the fact that the Chaetopterus tube material is fluorescent (typically in the green range (525 nm), when excited at 390 or 470 nm) , whereas the cocoons of mulberry silkworms are not. Of course, the materials have evolved to function in very different environments: while the tubes must function in marine environments, the cocoons are exclusively terrestrial. Indeed, the role of water in the properties of the marine worm tubes was made clear from the significant contraction of the specimens tested in air; the terrestrial cocoons do not show such contractions in length (electronic supplementary material, figure S2). In addition, the post-larval Chaetopterus animals construct the tube to actively generate a microenvironment for feeding, gas exchange and protection . The Bombyx silkworm larvae not only construct their cocoons for protection during the vulnerable phase of ecdysis, but also to regulate a gas exchange and control their microenvironment .
3.2.2. Thermal cycling in air
Thermal cycling tests were carried out in air to further study the thermomechanical stability of the tubes and to specifically observe hysteresis effects in the viscoelastic properties of the ‘dehydrated’ tube material (figure 4). The high peak temperature of 200°C was selected noting that at this temperature moisture was completely removed (figure 2), but thermal mechanical degradation had not yet commenced (figures 2 and 3).
Tests revealed that in the first phase, subzero temperature properties of the wet tube material were dictated by the properties of ice (figure 4). It is known that the elastic (dynamic) modulus of ice is a function of temperature and decreases as the temperature increases: at –40°C, E ≈ 9.5 GPa and at –2°C, E ≈ 8.6 GPa . It was also confirmed that a subzero glass transition did not exist. After a peak in the loss modulus and loss factor around the melting point of ice (at –5°C and 5°C, respectively), the loss modulus fell by almost three orders of magnitude to approximately 1.4 MPa, whereas the loss factor dropped by a factor of five to approximately 0.11 at ambient temperature (20°C). The storage modulus was about 13 MPa at 20°C. All viscoelastic properties remained fairly constant between 20°C and 60°C. Specimen length contraction and therefore moisture loss commenced at 60°C (figure 4), much later than observed in the single thermal scan tests (figure 3). However, differences between the property profiles in the thermal scan and the first phase of the cyclic test were attributed to the stress history of the sample; specifically, whereas the tubes subjected to a single scan were first equilibrated at –25°C, tubes subjected to the double cycle were first equilibrated at a much lower temperature of –75°C. Nonetheless, once the specimen started to dry with increasing temperature, the same evolution in viscoelastic properties was observed as described previously for the thermal scanning tests where the properties remained fairly constant between 100°C and 200°C.
At the end of the first phase, the sample would be completely dried. This dehydrated sample was cooled down to –75°C in the second phase and then re-heated to 200°C in the third phase (figure 4). Notably, the viscoelastic properties and the specimen length of the dehydrated material were highly reversible over the entire temperature range, as results from the second and third phase coincided (figure 4). In addition, while the storage modulus decreased linearly with increasing temperature (rate of approx. 0.6 MPa °C−1), the loss modulus and loss factor remained fairly constant between –75°C and 200°C. The storage modulus ranged between 300 and 475 MPa, whereas the loss modulus and loss factor were measured to be 10–14 MPa and 0.025–0.055, respectively. The observed thermomechanical stability and reversibility are indicative of the tolerance of the chemical and physical structure of the anhydrous tube wall material over a wide temperature range of –75 to 200°C. Such a tolerance in an organic material is possible provided it were highly cross-linked and strongly bonded at an intramolecular level.
However, if the tube were to consist of highly networked components, then the elastic modulus of the dry tube structure would be higher than what is observed: of the order of 2–3 GPa rather than the observed 0.3–0.5 GPa. The low elastic modulus of the dehydrated tube may be a result of (i) a porous ultrastructure resulting in a lower fibre volume fraction than envisaged, (ii) some intricate fibre substructure such as hollow nanofibrils, and/or (iii) a larger fraction of nanofibrillar mats oriented in the ±45° and ±65° orientation rather than in the 0° orientation. Previously conducted ultrastructure studies  have revealed that the porosity volume fraction in dried Chaetopterus tubes ranges between 3% and 7%. In fibre-reinforced composites, porosity has a substantial detrimental effect on mechanical properties, including elastic modulus. In fact, the effect of porosity on composite stiffness may be approximated by a factor of (1 − vp)2, where vp is the porosity volume fraction [21,22]. That is, 3–7% porosity content would lead to a 6–14% reduction in elastic modulus, which was much less than observed. Hence, the porous structure on its own does not seem to account for the short-fall in elastic stiffness. While an investigation on the substructure of the nanofibrils may provide some explanation, the effect of misorientation of nanofibrillar mats would have a much larger contribution. In fibre-reinforced composites, Krenchel's orientation efficiency factor η  is often used to model the effect of misorientation of fibres to composite mechanical properties. η is proportional to cos4(θ), where θ is the angle between the fibre axis and the applied load. While a fibre aligned along the loading direction (i.e. at 0°) would contribute 100% of its reinforcing potential, fibres aligned at ±45° and ±65° to the loading direction would contribute only 25% and 3% of their reinforcing potential, respectively. Therefore, equal proportions of nanofibrillar mats in the three orientation angles (0°, ±45°, ±65°) would result in a 58% drop in elastic modulus, in comparison with a tube constructed entirely of mats at 0°. More fibrils orientated at ±45° and ±65° would reduce the elastic modulus further. Tensile tests results on longitudinal and transverse samples (presented later in §3.3) conclusively show that indeed there were more fibrils in the tube wall that were oriented at ±45° and ±65° than at 0° to the tube axis.
3.2.3. Thermal cycling in submersion mode
Noting that the natural environment of the Chaetopterus tubes collected for this study is (marine) water (typically 6–15°C) and that moisture loss has a significant effect on the mechanical properties of the tube (figures 3 and 4), thermal cycling tests between 5°C and 75°C were conducted in submersion mode (figure 5).
As expected, no (significant) length contraction was observed (figure 5) implying that the tube sample remained ‘undried’ throughout the test duration. We were, however, surprised to find that the viscoelastic properties of the undried tube wall material remained effectively constant across the entire temperature range: E′ = 13.3 ± 0.6, E′′ = 1.50 ± 0.06 and tan δ = 0.113 ± 0.001. In the second and third phases, the loss tangent was slightly lower but constant, whereas the storage and loss moduli were higher at lower temperatures, possibly owing to gelation. Notably, the viscoelastic properties of the undried tube were reversible across the entire temperature range. The results clearly demonstrate the exceptional mechanical stability of the tubes over a wide temperature range in their natural environment (i.e. submerged in water). While the Chaetopterus tubes typically experience temperatures of 6–15°C at depths of less than 50 m, the results show that, at least for short exposure times, the tubes may be able to survive temperatures at even deep-sea sites (more than 1000 m deep, 2–3°C) and near hydrothermal vents (more than 60°C). It is, therefore, not surprising that similar chaetopterids and their tubes have been found at such locations .
Comparing the results of thermal cycling tests conducted in air (figure 4) and water (figure 5), we can compare the difference in properties of the anhydrous and undried tube. While the dehydrated tube material was glassy (i.e. rigid, brittle and non-damping), the undried tube was rubbery (i.e. soft, flexible and damping). This exemplifies the critical role of water and (therefore) hydrogen bonding and water plasticization in maintaining the rubber-like properties of the organic tube material. Importantly, the tube in both its dried and undried state (in air and water, respectively) was stable over a large temperature range with viscoelastic properties being constant and reversible. This is a great incentive for our future work to focus on the chemical identity and structure of the building-blocks (the nanofibrils) making the tube.
Moreover, of interest is the finding that chitin-based polysaccharide materials produced by other marine animals tend to show a similar DMTA profile and property trends [24–26] to the Chaetopterus tubes. The chitin-based films also exhibit thermomechanical reversibility in both the dry and wet (i.e. submerged) state, as well as a remarkable difference in magnitude in the modulus in the wet and dry states [24–26]. The latter highlights the critical role of bound water and its mobility (as a function of temperature) in the properties of such biomaterials.
3.3. Mechanical properties
The tensile stress–strain curves of the various tube samples tested are given in figure 6, with the box-and-whiskers chart in figure 7 summarizing the extracted tensile properties, i.e. stiffness, strength and failure strain.
3.3.1. Longitudinal tensile properties
Comparison between the different methodologies. The longitudinal properties of the tubes measured using three different test modes are presented in figures 6a and 7. In general, the stress–strain curves (figure 6a) overlay each other indicating the similarity in properties measured from the different techniques. However, figure 7 reveals considerable variability in stiffness and failure strain, which might account for the absence of statistical support (p > 0.2) for differences in thicknesses and the tensile strengths actually measured in our samples. Measurements on the Zwick tensile tester and on the DMTA instrument in submerged mode were more comparable to each other. However, the results obtained on the DMTA instrument in standard film-tension mode showed much more consistency and repeatability with coefficient of variations of up to 30%, compared with coefficient of variations of up to 65% and 50% for tests conducted on the DMTA instrument in submerged mode and on the Zwick, respectively. In addition, specimen loading and humidity control were easier on the DMTA instrument in standard film-tension mode. As various other tests were (therefore) conducted in this mode, to set a benchmark, we discuss further the longitudinal tensile properties obtained from this test regime.
The longitudinal tensile stiffness, strength and failure strain of the tubes were measured to be 6.8 ± 2.2 Mpa, 1.1 ± 0.2 Mpa and 14.0 ± 1.2%, respectively. The order of these mechanical properties is similar to rubber-like elastomers. In fact, even spiders produce specific elastomeric (viscid) silks with a modulus of the order 1 MPa, which also have the mechanical properties of a rubber-like polymer . The high failure strain or toughness of the tubes is a result of distributed deformation which is feasible from (i) the complex orientation of the nanofibrils, (ii) stress-transfer and strain-accumulation at the large surface area fibre–matrix interface, and (iii) non-woven laminate structure.
Tensile failure surfaces were observed under an SEM to identify fracture mechanisms. Failure mechanisms varied from (i) ductile fracture with delamination and extensive fibrillation resulting from pull-out or fracture of fibres principally oriented in the loading direction (figure 8a) to (ii) brittle fracture resulting from rapid crack propagation transverse to the loading direction, i.e. transverse tearing (figure 8b) to (iii) ductile–brittle fracture resulting from uniquely oriented layers failing in distinctly different manners (figure 8c). The orientation of the nanofibrils in the different layers of the tubes would have had a considerable effect on the tensile properties and fracture mechanism . Typically, fibre-dominated ductile failure through fibre fracture or fibre pull-out is characteristic of fibres oriented in the loading direction (i.e. 0–5°), whereas laminates with fibres oriented at 5–45° to the loading direction fail owing to interlaminar shear stress and transverse tensile stresses, leading to delamination and transverse tensile fracture of fibres . Laminates with fibres oriented at between 45° and 90° to the loading direction exhibit brittle failure principally owing to transverse tensile fracture of fibres . The presence of all three (i.e. mixed-mode) fracture mechanisms (figure 8) is expected as the non-woven layers of the Chaetopterus tubes employ nanofibrils principally oriented at 0°, ±45° and ±65° to the tube axis .
Effect of drying. The TGA and DMTA (§§3.1 and 3.2) clearly demonstrate the occurrence and effect of moisture loss on tube thermomechanical properties. Hydration was key to the rubber-like flexible properties of the tube; for samples tested in air, increase in temperature and subsequent moisture loss transformed the material behaviour to glass-like: stiff and brittle. This was also evident from the tensile test results shown in figures 6b and 7. Drying significantly (p < 0.001) increased the stiffness (by 125% to 15.2 MPa), but reduced the failure strain (by 75% to 3.4%). The 50% statistically significant (p < 0.001) reduction in strength from drying was due to brittle fracture resulting from rapid transverse crack growth.
It is notable that while warpage and folds formed quickly, drying did not significantly change the thickness of the tube. Not only was testing of such dried samples difficult, the folds were likely sites for stress concentration, leading to premature brittle failure. Surprisingly, we observed that rehydrating the tubes reversed the warps and folds that had formed during drying. It would be of interest therefore to test in the future the effect of rehydrating on mechanical properties. Indeed, as Chaetopterus tubes are also found in intertidal zones, resistance to drying–rehydrating cycles would be of import.
Properties of layers. The tensile properties of separated layers are presented in figures 6b and 7. It was expectedly self-evident that the layers had a lower thickness than the whole tube sections. The stress–strain curves (figure 6b) depict the differing tensile behaviour of the separated layers and the tube, where the layers showed higher stress than the tube at the same strain. In fact, the layers had significantly (p < 0.01) higher stiffness (29.3 ± 15.8 MPa) and strength (2.6 ± 0.8 MPa) than the whole tube that had stiffness and strength of 6.8 ± 2.2 and 1.1 ± 0.2 MPa, respectively (figure 7). In addition, the failure strain was lower for the layers (p < 0.02).
We conclude that the observed result may be attributed primarily to size (i.e. thickness) effects. Polymer composites comprising multiple plies of multi-axially oriented fabrics consistently show that thicker laminates have lower stiffness and strength, with higher strain to failure [12,29]. Thicker laminates delaminate more readily, particularly owing to reduced interactions between adjacent cracked and uncracked plies , larger shear forces, and more flaws.
3.3.2. Transverse tensile properties
Comparison between top, middle and bottom ring samples. Ring samples from three different regions of the tubes were tested to study transverse tensile properties. The obtained stress–strain curves and extracted properties are presented in figures 6c and 7, respectively.
The thickness of the samples from different sections of the tube increased in the following order: top < middle < bottom, probably because more layers are deposited in the bottom section in comparison with the middle and top sections. The transparency (or rather opaqueness) of samples from different sections in the tubes supports this interpretation .
It is evident from the stress–strain curves in figure 6c that the different sections in the tube exhibited distinctly different mechanical properties. In general, the tensile stress–strain curve became steeper in the following order: top < middle < bottom. In particular, the bottom ring sections had higher stiffness and strength, but lower failure strain in comparison with ring sections from the top (figure 7). It is notable that such a transition in stress–strain profiles and properties is observed in polymer composites when fibre orientation is changed [12,28]. Steeper slopes indicate fibres aligned closer to the loading direction, whereas gentle slopes indicate fibres aligned at an angle to the loading direction [12,28]. It is already known that the fibrils are principally oriented at 0°, ±45° and ±65° to the tube axis . Hence, the nature of the stress–strain curves in figure 6c may be interpreted as follows: the top section (with fewer layers) had fibrils that were primarily inclined away from the loading direction (i.e. at 0° and ±45° to the tube axis), and the middle and bottom ring sections were deposited with additional layers employing fibrils which were increasingly more inclined closer to the loading direction (i.e. at ±65° to the tube axis). That is, in terms of volume fraction, the top section of the tubes had more plies with fibrils oriented at 0° and ±45° than at ±65° to the tube axis, and the middle and bottom sections had an increasing fraction of fibrils that are oriented at ±65° to the tube axis.
Comparison with longitudinal properties. The longitudinal tensile stiffness, strength and failure strain of the tubes were measured to be 6.8 ± 2.2 Mpa, 1.1 ± 0.2 Mpa and 14.0 ± 1.2%, respectively. The transverse tensile stiffness, strength and failure strain of the tubes were in the ranges 8–20 Mpa, 1.4–2.6 Mpa and 10–20%, respectively (figure 7). While the longitudinal stiffness of the tubes was comparable to the transverse stiffness of tubes from the top section, it was significantly lower than the transverse stiffness of samples from the middle and bottom sections (figure 7). In addition, the transverse strength of the tubes from all sections was found to be higher than the longitudinal strength (figure 7). The longitudinal failure strain was (i) comparable to, (ii) higher than or (iii) lower than the transverse failure strains of the (i) bottom, (ii) middle and (iii) top sections, respectively (figure 7).
The dissimilarities between the longitudinal and transverse properties suggest that the tube lacks a quasi-isotropic structure. In general, the higher stiffness and strength in the transverse direction implies that there were more fibrils orientated at ±45° and ±65° than at 0° to the tube axis. In this case, the tubes would be more resistant to torsion/shear loads (from ±45° oriented fibrils) and combined external pressure and buckling loads (from ±65° oriented fibrils), than from bending and axial tension/compression loads (from 0° oriented fibrils) . Importantly, this result is in agreement with the previous assertion (in §3.2.2) explaining the low elastic modulus of the highly cross-linked tube material.
The thermomechanical properties of the self-constructed housing tube of the marine worm Chaetopterus sp. provide important insights. Both thermal and thermo-oxidative degradation began around approximately 250°C, with the maximum rate of degradation at approximately 270°C. All three thermal decomposition peaks and transitions of the tubes matched those of mulberry silkworm cocoons, indicating potential similarity/analogy between the organic materials. However, unlike in silkworm cocoons, dynamic mechanical thermal analysis conducted in both air and water elucidated the lack of a glass transition in the tube material, indicating that the organic material may not have amorphous phases, but rather consist of highly crystalline or highly cross-linked, hydrogen- or covalent-bonded molecules. This assertion was supported by the impressively stable (i.e. constant and reversible) viscoelastic properties of the anhydrous and undried tube between –75 and 200°C in air and 5 and 75°C in water, respectively. Moreover, it was found that hydration (and associated-water plasticization) was key to the rubber-like flexible properties of the tube; dehydration transformed the material behaviour to glass-like (i.e. rigid, brittle and undamping). Structural features including porosity, complex fibre substructure (e.g. hollow nanofibrils) and the multi-axial orientation of nanofibrils are the suggested likely sources of the low elastic modulus of the tube material. Nonetheless, the stability of the organic tube wall material over the broad temperature range studied makes future studies on its physico-chemical structure and production of great interest.
Tensile mechanical property characterization of the tubes revealed intriguing similarities to synthetic rubber-like elastomers and even viscid spider silks. The high failure strain or toughness of the tubes was a result of distributed deformation, feasible from (i) the complex orientation of the fibrils, (ii) stress-transfer and strain-accumulation at the large surface area fibre–matrix interface, and (iii) non-woven laminate structure. Notably, the complex orientation of the filaments had a governing effect on the mixed-mode fracture mechanism. Dissimilarities between the longitudinal and transverse properties established a lack of quasi-isotropic structure in the tube walls. In general, the higher stiffness and strength in the transverse direction implied that there were more fibrils oriented at ±45° and ±65° than at 0° to the tube axis. The complex structure–property relations observed indicate that the worm has evolved to produce such a structure to endure mechanical stresses from specific planes/axes, in addition to enable rapid tube growth and repair.
We thank the US Air Force Office for Scientific Research (AFOSR grant number F49620-03-1-0111 to F.V. and FA9550-10-1-0112 to D.D.D.) and the European Research Council Advanced Grant (SP2-GA-2008–233409) for generous funding.
We are grateful to Phil Zerofski, marine collector at SIO, who kept a constant supply of the tube worms.
- Received May 19, 2014.
- Accepted June 19, 2014.
- © 2014 The Author(s) Published by the Royal Society. All rights reserved.