Page 1 of 27

Cite as: Shah DU, Porter D, Vollrath F. Composites Science and Technology (2014). 101: p.

173-183. http://dx.doi.org/10.1016/j.compscitech.2014.07.015

Can silk become an effective reinforcing fibre? A property

comparison with flax and glass reinforced composites

Darshil U. Shah*, David Porter, Fritz Vollrath

Oxford Silk Group, Dept. of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS, UK

* Corresponding author; E-mail: darshil.shah@zoo.ox.ac.uk, Tel: +44 (0)1865271216

Abstract

With the growing interest in bio-based composites as alternatives to traditional glass

fibre reinforced composites (GFRPs), there has been a persistent rise in the commercial use

of plant fibre composites (PFRPs). In contrast, nature’s ‘wonder-fibre’ silk has had no

commercial applications, and only limited scientific investigations, as a composite

reinforcement. To produce silk fibre composites (SFRPs) with useful properties, three key

recommendations from our critical literature review were followed: i) a high-failure strain,

low-processing temperature thermoset matrix was used to a) maximise the reinforcing effect

of low-stiffness, ductile silk, and b) facilitate impregnation and avoid fibre degradation, ii)

high fibre volume fractions were employed to ensure that fibres carried a larger fraction of

the load, and iii) given the lack of studies investigating fracture energy dissipation

mechanisms in SFRPs, interface modification was avoided due to its complex, sometimes

detrimental, effects on toughness. In directly addressing the question, ‘is there a case for silks

as polymer reinforcements?’, we evaluated various mechanical properties of nonwoven and

plain woven SFRPs against similar flax and glass composites. In all cases, woven composites

performed better than nonwoven composites. While SFRPs were weak in terms of stiffness,

their flexural and tensile strength was comparable to PFRPs, but much below that of GFRPs.

Notably, the low density of SFRPs, like PFRPs, made them comparable to GFRPs in terms of

specific flexural properties. Woven SFRPs exhibited much higher fracture strain capacities

than both flax and glass composites, making SFRPs suitable for applications where high

compliance is required. The Achilles’ heels of PFRPs have been their reportedly i) inadequate

interfacial properties, ii) inferior impact properties, iii) poor strength performance, and iv)

high moisture sensitivity. We found that SFRPs outperformed their flax counterparts in areas

i)-iii), and were more comparable to, but not better than, GFRPs. While concerns such as cost

and ‘sustainability’ of silk are acknowledged, potential applications for SFRPs are discussed.

Page 2 of 27

Keywords: A. Glass fibres; A. Polymer-matrix composites (PMCs); B. Mechanical properties; Silk and other natural fibres

1 Introduction

Biocomposites reinforced with plant fibres such as flax, jute and hemp have been

widely investigated in literature as potential eco-friendly alternatives to synthetic fibre

reinforced composites [1-4]. In general, the low cost, low density and sustainable nature of

plant fibres make them attractive in comparison to the commonly used reinforcing fibre, E-

glass (Table 1). While the mechanical strength (absolute and specific) of plant fibre

reinforced composites (PFRPs) is generally lower than that of glass fibre reinforced

composites (GFRPs), PFRPs may be suitable replacements to GFRPs in stiffness-critical

applications (Fig. 1) [5, 6]. Consequently, a persistent rise in the commercial use of PFRPs,

primarily in the automotive industry, has been observed over the past several years [7].

Fig. 1. Ashby plot comparing the absolute and specific tensile properties of various plant fibre reinforced composites (PFRPs) and glass fibre reinforced composites (GFRPs). Our results on silk fibre reinforced composites (SFRPs) are also included for comparison. Adapted from [5, 8].

Page 3 of 27

In contrast, silk, the only natural fibre to exist as a continuous filament, has had no

commercial applications, and only limited scientific investigations, as a reinforcement for

non-biomedical composites. The question arises: is there a case for silks as suitable polymer

reinforcements? More specifically, what advantages do silks and their composites offer in

comparison to plant fibres and their composites, and glass fibres and their composites?

1.1 The case for silk fibres as reinforcing agents

Many arthropod animals, including silkworms, spiders, scorpions, mites and fleas,

have evolved to spin task-specific fibrous protein polymers into fibres for a variety of

functional uses: from protection (through structural cocoons or sacs) to prey capture (using

webs) [9-12]. It is this large group of fibres that we call silks. Silk from the cocoons of the

domesticated mulberry silkworm, Bombyx mori is of particular economic importance and is

generally used in luxurious textiles.

Importantly, the biocompatibility and bioresorbable properties of silks, their

amenability to aqueous or organic solvent processing into various ‘regenerated’ forms

(including aqueous solutions, films, hydrogels, porous sponges, regenerated fibres and cords,

and nonwoven mats), alongside their unique combination of high strength and toughness,

make them ideal for a wide range of clinical applications: from braided suture threads for

surgical options, to porous, reinforced-composite scaffolds for cartilage and bone repair [9-

12]. Naturally, considerable research has focussed on biocomposites based on regenerated

silks for such biomedical applications [9-12].

Nevertheless, many of the properties of native silks (as opposed to regenerated silks)

also make them potential sustainable alternative reinforcement materials, alongside plant

fibres, for engineering (i.e. non-biomedical) composites. This forms the focus of our research.

Table 1 compares the economic, technical and ecological properties of silks with plant and

glass fibres. In general, the primary disadvantages of silks in comparison to plant and glass

fibres are: i) higher cost, ii) lower annual production, iii) higher moisture absorption, iv)

lower softening (and therefore processing) temperatures, v) poor stiffness, and vi) high

embodied energy for processed materials (e.g. fabrics). However, they possess i) lower

density (than even plant fibres), ii) natural flame resistance iii) moderate strength, iv)

unparalleled toughness (higher than even Kevlar), and v) a generally favourable

environmental profile of the raw material. Other technical advantages of silks specific to

composites applications include i) their naturally continuous length, and ii) the high

Page 4 of 27

compactibility of silk preforms [13]. While the former would translate to a high fibre length

distribution factor ηl and therefore reinforcing effect in composites, the latter provides an

opportunity to produce high fibre volume fraction natural fibre composites [13].

Table 1. Comparison of the economic, technical and ecological properties of silk, plant and glass fibres.

Properties Silk fibres a Plant fibres b Glass fibres c

Eco

nom

y Annual global production of fibres [tonnes] 150,000 31,000,000 4,000,000

Distribution of fibres for FRPs in EU [tonnes] 0 60,000 600,000

Cost of commercial raw fibre [£/kg] 2.0-30.0 0.5-1.5 1.3-20.0

Tec

hn

ical

Chemical nature proteinaceous lignocellulosic silica-based

Fibre length continuous discrete continuous

Fibre diameter (apparent) [μm] 1-15 (8-15) 15-600 (15-30) 5-25

Density [gcm-3] 1.25-1.35 1.35-1.55 2.40-2.70

Moisture absorption [%] 5-35 (20-35) 7-25 (7-10) 0-1

Softening temperature [°C] 170-220 190-230 700-1,100

Tensile stiffness [GPa] 5-25 (5-15) 30-80 (50-80) 70-85

Tensile strength [GPa] 0.2-1.8 (0.3-0.6) 0.4-1.5 (0.5-0.9) 2.0-3.7

Specific tensile stiffness [GPa/gcm-3] 4-20 (4-12) 20-60 (30-60) 27-34

Specific tensile strength [GPa/gcm-3] 0.1-1.5 (0.3-0.7) 0.3-1.1 (0.3-0.7) 0.7-1.5

Tensile failure strain [%] 15-60 (15-25) 2-30 (2-4) 2.5-5.3

Toughness [MJm-3] 25-250 (70) 5-35 (7-14) 40-50

Specific toughness [MJm-3/gcm-3] 20-185 (50-55) 3-26 (4-10) 16-19

Abrasive to machines No No Yes

Eco

logi

cal

Embodied energy of commercial raw fibre [MJ/kg] d 50-100 4-15 30-50

Renewable source Yes Yes No

Recyclable Yes Yes Partly

Biodegradable Yes Yes No

Hazardous/toxic (upon inhalation) No No Yes a Includes silks from various spiders and silkworms. As most of the commercial silk is cultivated from the Bombyx mori silkmoth, figures in brackets present the typical properties of this variety of silk. The composites manufactured in this study also employ B. mori silk. Data from [14-18] (and references therein). b Includes bast, leaf and seed fibres, but does not include wood and grass/reed fibres. Figures in brackets present the typical properties of flax fibre. Data from [1, 19] (and references therein). c Includes E- and S-glass fibres. Properties for E-glass are in the lower range, in comparison to S-glass. Data from [1, 19] (and references therein). d The conversion of silk fibres in cocoons into reeled slivers and later aligned textile products can further increase the cumulative energy demand, for instance, to up to 1850 MJ/kg for raw silk slivers [15]. Similarly, while the energy required in the cultivation of plant fibres is low (4-15 MJ/kg), further

Page 5 of 27

processing steps (e.g. retting and spinning) can significantly increase the cumulative energy demand, for instance, to up to 146 MJ/kg for flax yarn [3, 20, 21]. Glass fibres, on the other hand, are produced through an extrusion process and can be converted into reinforcements for composites (in the form of chopped strand mats or aligned fabrics, for instance) without significant energy input.

1.2 A critical literature review on silk fibre composites

From a general perspective, the limited literature available on mulberry silk fibre

reinforced polymers (SFRPs) principally attempts the investigation of two types of

composites: i) biodegradable or bio-based composites for non-structural applications, and ii)

tough composites for energy-absorbing and crashworthy structures. Most studies have

employed low fibre weight fractions, ranging between 1 and 30%.

In the first case, short silk fibres (0.5-10 mm in length) have been incorporated as

reinforcements for i) thermoplastic polymers (such as biodegradable polylactic acid and

polybutylene succinate, and non-biodegradable polypropylene) [22-27], or ii) elastomeric

rubbers (both natural and synthetic) [28, 29], via extrusion/injection moulding processes.

Notably, the use of screws and mixers in such manufacturing processes leads to i) the 3D

dispersion and spatial ‘random’ orientation of the anisotropic fibres, and ii) the breakage of

chopped short silk fibres (< 10 mm in length) into even shorter fibres (lf = 0.3-2.0 mm in

length) [1, 28-30]. The former leads to fibre orientation distribution factors ηo in the range of

0.20 (nominally-3D fibre dispersion) to 0.37 (some preferred orientation of fibres) [1].

Reinforcing fibre length lf, on the other hand, affects the fibre length distribution factor of

discontinuous fibre composites [1, 31]. For the purposes of this discussion, we refer to the

length distribution factor ηl for predicting composite strength (Eq. (1)). Sub-critical length

fibres (lf < lc) will not carry the maximum load, and it can be shown, through Eq. (1), that

provided the fibre length is more than ten times the critical fibre length (i.e. lf > 10lc), ηl >

0.95 (i.e. ηl ~ 1) can be achieved [1, 31]. For reference, PFRPs have critical fibre lengths lc in

the range of 0.2-3 mm and corresponding length distribution factors ηl for extrusion/injection

moulded short fibre composites of <0.3 [1]. Critical lengths lc for silk composites have been

measured (through microbond tests) by only one study in literature [32], to be in the range of

0.1-0.4 mm. This is comparable to the estimated critical length lc of 0.1-0.3 mm, calculated

using Eq. (1), and assuming i) an interfacial shear strength τ of 15-30 MPa for the silk

composite [32], and ii) uniform silk fibres with tensile strength σf and apparent diameter df of

300-600 MPa and 8-15 μm (Table 1), respectively. For silk fibre lengths lf of 0.3-2.0 mm, the

aforementioned critical lengths lc yield length distribution factors ηl of 0.03-0.6. The

combined length and orientation distribution factors ηl·ηo of short fibre silk composites lead

Page 6 of 27

to composite mechanical properties that are only marginally improved in comparison to the

neat matrix.

τσ

ηη2

],1,0[,2/

2/1 ffcl

cfcf

cffcl

dlwhere

llforllllforll

=∈

≤

≥−= Eq. (1)

In the second case, to produce tough composites, woven textile fabrics have been

incorporated as reinforcements for typically brittle thermosetting resins (such as epoxy) by

employing compression moulding (i.e. hot press moulding) manufacturing techniques (in

some cases, for further compaction/consolidation after an initial hand-layup or filament

winding process) [33-41]. Here, the use of continuous silk filaments leads to length

distribution factors ηl close to unity, as lf >> lc. The multiaxial fabrics have a range of fibre

orientation distribution factors ηo, depending on the ply orientation. For composites with

balanced biaxial reinforcements in a (0,90)n and (±45)n stacking sequence, it can be shown

that ηo = 0.5 and 0.25, respectively [1, 31].

1.2.1 Designing silk fibre reinforced polymer composites

In our opinion, there are two key issues that require specific attention in the

development of SFRPs with useful properties: i) selection of a complementary matrix, and ii)

understanding the role of the interface.

The fundamental philosophy in fibre reinforced polymers is the development of a

brittle fibre–ductile matrix system [31], where the fibres, compared to the matrix, have a

much lower failure strain coupled with a much higher stiffness. Commonly used polymer

matrices (including thermosets like epoxy, and thermoplastics like polypropylene) have a

relatively low stiffness (<4 GPa) and high failure strain (>5%, even up to 1000%). The strong

and stiff fibres are required to carry most of the load (Fig. 2), while the ductile matrix

provides crack blunting and bridging mechanisms [31]. Silk fibres are different from

traditional reinforcing fibres (including E-glass and flax) in that mulberry silk fibres have a

relatively low stiffness (5-15 GPa) and high failure strain (15-25%), while traditional

reinforcing fibres have a high stiffness (>50 GPa) and low failure strain (1-5%).

Referring to the schematic in Fig. 2, the comparable stiffness of silk fibres to

commonly used matrices (Ef/Em = 1-5 for silk-epoxy) implies that fibre volume fractions vf in,

for instance unidirectional SFRPs need to be i) >20-40% to ensure that fibres carry at least

equal load as the matrix (i.e. Pf/Pm ≥ 1), and ii) >60% to ensure that fibres carry at least 10

Page 7 of 27

times more load than the matrix (i.e. Pf/Pm ≥ 10). The required fibre volume fractions are

much lower for traditional reinforcing fibres due to their higher stiffness (Fig. 2). It is telling

that most studies in literature on SFRPs have targeted fibre volume fractions of <30%. To our

knowledge, only two studies [39, 41] have examined SFRPs with fibre volume fractions of

~40%. Notably, in both studies SFRPs were manufactured via compression moulding at high

compaction pressures of ~15 bar. Of interest is our recent assessment [13] on the compaction

behaviour of silk textiles where we have shown that even at low compaction pressures (as in

vacuum infusion processes) silk reinforcements are significantly more compressible than

plant and glass fibre textiles. For instance, at 2.0 bar compaction, fibre volume fractions of

50-60% could be achieved with silk textiles [13].

Fig. 2. The ratio of the load carried by unidirectional fibres to the load carried by the diffuse matrix, Pf/Pm is proportional to the ratio of the elastic moduli, Ef/Em and the volumetric ratio, vf/vm. For maximum reinforcement (i.e. fibre taking most of the load), Ef/Em and vf/vm should be maximised. Due to the low stiffness of silk, its combination with traditional matrices (such as epoxy) results in low Ef/Em ratios (1-5 for silk-epoxy). Stiff fibres such as flax, E-glass and carbon, impart higher Ef/Em ratios (12-25, 17-25 and 40-75 for flax-epoxy, E-glass-epoxy and carbon-epoxy, respectively). Consequently, while unidirectional flax and E-glass fibres would carry at least 10 times the load carried by an epoxy matrix at fibre volume fractions above 30% (i.e. Pf/Pm ≥ 10 for vf > 0.3), unidirectional silk fibres would carry at least 10 times the load carried by an epoxy matrix at fibre volume fractions only above 60% (i.e. Pf/Pm ≥ 10 for vf > 0.6).

Page 8 of 27

Parallel to the discussion on comparable stiffness ratios is the discussion of failure-

strain mismatch. As silk fibres are ductile in comparison to many traditional polymer

matrices (particularly thermosets like epoxy and polyester), the resulting composite would be

a ductile fibre–brittle matrix system, where the fracture mechanisms may be unfavourable to

produce a tough composite as the relatively brittle matrix relies on the fibres to act as crack

stoppers [31]. Having said that, concrete and ceramics, which are quasi-brittle materials with

a low strain capacity, can be toughened with ductile reinforcing fibres, like steel, by

exploiting the pseudo strain-hardening and matrix multiple cracking phenomenon [42, 43],

where the fibres i) inhibit the initiation and growth of micro-cracks, and more importantly ii)

effectively bridge multiple, parallel cracks in the matrix. While most fibre reinforced

polymers (FRPs) do not exhibit such work-hardening phenomena, and the load bearing

ability of the FRPs typically falls quite rapidly after peak load, a high work of fracture may

still be needed to complete fracture due to a gradual degradation process (post-peak load)

[31]. Most studies looking to produce tough SFRPs [33-40], with the exception of van Vuure

et al. [41], have employed a brittle epoxy as the matrix with a failure strain lower than that of

the reinforcing silk fibres. More appropriately, van Vuure et al. [41] investigated woven silk

fabrics as reinforcements for various high failure strain thermoplastic resins, predictably

finding that the absorbed impact energy of the SFRPs increased almost proportionally to the

failure strain of the unreinforced polymer matrix.

It is clear, therefore, that to produce tough SFRPs high fibre volume fractions and a

complementary matrix (that would produce a brittle fibre–ductile matrix system) are critical.

While most thermoplastic matrices have high failure strains, they may prove to be difficult to

process with silk fibres due to the latter’s low thermal degradation temperature (of 170 °C).

Matrices with high failure strains and low processing temperatures would be ideal for SFRPs.

As with PFRPs, the fibre/matrix interface in SFRPs has been a common discussion

point in several studies, including [25, 32, 38, 41, 44, 45]. The widespread effort has been to

improve the fibre/matrix interface in SFRPs by considering fibre surface modification

through chemical and enzymatic ‘degumming’ of silkworm silk fibres (e.g. [44-46]).

Degumming is a common process in the silk textile industry and involves removal of the

gum-like sericin protein coating to reveal the silk fibroin protein filament (Fig. 3). So far,

researchers have almost exclusively investigated the effect of degumming processes on silk

fibres [11, 44-48], rather than silk-reinforced composites [45]. It is well-known that

degumming processes have significant detrimental effects on the mechanical properties of the

Page 9 of 27

silk fibres [11, 44-48]. The reduction in fibre properties (viz. tensile stiffness and strength)

post-treatment are attributed to i) dimensional changes, ii) degradation of protein chains, and

iii) weakening of inter- and intra-molecular hydrogen bonds. A limited number of researchers

have investigated the use of physical surface treatments (e.g. ionized radiation [25]) and

compatibalisers [38] to promote fibre/matrix adhesion in SFRPs, reporting some

improvement in composite mechanical properties.

There is a noticeable lack of comprehensive studies in literature on fracture and

energy dissipation mechanisms in SFRPs, and the effects of interface modification. In

general, reinforcing fibre length and fibre/matrix adhesion are intimately related, and have a

combined, complex effect on composite mechanical properties [31]. A simplistic analysis

based on Eq. (1) (for predicting tensile strength of discontinuous unidirectional fibre

composites) suggests that longer fibres are more capable of transferring load from the matrix,

and that provided the fibre length is more than ten times the critical length (i.e. lf > 10lc), fibre

length distribution factors of ηl > 0.95 (i.e. ηl ~ 1) can be achieved [1, 31]. Backed by

experimental findings of Craven et al. [32], we previously estimated critical lengths lc for silk

composites to be in the range of 0.1-0.4 mm. Consequently, employing unidirectional silk

fibre reinforcements longer than 4 mm would ensure that the maximum fibre tensile strength

is exploited.

The toughness of a composite depends on the various energy dissipation mechanisms

during failure; fibre pull-out and debonding dissipate more energy than fibre fracture, for

instance [31]. Notably, fibre pull-out and debonding are indicative of poor fibre/matrix

adhesion [31]. It is well-known that as fibre/matrix adhesion improves, the work of fracture

first increases with increasing interfacial shear strength τ (as more energy is required for

frictional work during and post- fibre debonding), and then decreases as the critical fibre

length lc, which is related to interfacial shear strength τ through Eq. (1), becomes very small,

leading to fibre fracture being more prevalent than fibre pull-out [31]. Similar observations

have been made for SFRPs and PFRPs [41, 49, 50]. Given the lack of studies investigating

fracture energy dissipation mechanisms in SFRPs, interface modification was avoided in our

study, due to its complex, sometimes detrimental, effects on toughness.

Following the recommendations from our critical literature review, we examined the

mechanical properties of high fibre volume fraction (up to 45%) SFRPs. A high failure strain

epoxy was used as a compatible matrix. The low viscosity of the resin enabled composite

manufacture via liquid composite moulding, ensuring good reinforcement impregnation

Page 10 of 27

without any external compaction pressure. Silk fibre reinforcements in the form of nonwoven

(i.e. nominally randomly-oriented fibre) mats and plain woven textiles were considered. The

silk composites were then evaluated against similar flax and glass fibre reinforced composites

(using literature data) to determine appropriate applications. Such a comprehensive cross-

comparative study on sustainable composites is not found in literature so far.

While we characterise the mechanical properties of silk composites, and evaluate

them against similar flax and glass composites, it is to be noted that the manuscript concerns

a subset of composite types, namely nonwoven mat and plain woven fabric reinforced

thermoset-matrix composites. We acknowledge that composites, including PFRPs and

GFRPs, have a wide range of properties [1, 5, 8], as illustrated in Fig. 1, which can be

tailored by changing various parameters, including fibre type and form (twisted yarn, sliver,

tow), textile weave, matrix type, and manufacturing technique.

2 Experiments and methodology

2.1 Materials

EPIKOTE™ Resin RIMR135 with EPIKURE™ Curing Agent RIMH137, supplied

by Momentive Specialty Chemicals GmbH (Stuttgart, Germany), was used as the epoxy

matrix system. The low viscosity of the mixed resin (200-300 mPas) enabled composite

manufacture via vacuum-driven resin transfer moulding. According to the manufacturer, the

epoxy resin had a cured density ρm of 1.15 gcm-3, tensile modulus Em of 2.7-3.2 GPa, tensile

strength σm of 60-75 MPa, tensile failure strain εm of 12-16%, and impact strength of 70-80

kJm-2.

As silk fibre reinforcements, i) nonwoven mats (304 ± 32 gm-2; produced from B.

mori cocoons that were sourced from the Sericulture and Agriculture Experiment Station

(Vratsa, Bulgaria)) and ii) a balanced plain weave fabric (88 ± 8 gm-2; supplied by Stephen

Walters & Sons Ltd. (Suffolk, UK)), were used. The density of silk fibre ρf was assumed to

be 1.3 gcm-3 (Table 1).

2.1.1 Fabrication of nonwoven silk mat

Nonwoven silk mats (Fig. 3) were produced by hot-pressing (at 100-130 °C) wet

cocoon shells. The process, described in [51], flattens individual cocoon shells, and binds

adjacent shells due to the melt flow of a natural binder (namely, sericin [12, 14]) in the

Page 11 of 27

cocoons. Given the in-plane nominally random orientation of silk fibres in a cocoon [52], the

produced mat has a true nonwoven structure (Fig. 3).

The conventional process in the production of nonwoven fibre mats involves

chopping fibres to discrete lengths, and then chemically (using powders or emulsions) or

mechanically (by needle-punching, for instance) binding the fibres. The methodology we

employ maintains the continuous length of silk fibres ensuring high length distribution

factors, and minimises processing steps in mat production thereby reducing cumulative

energy demand. However, we do appreciate that the resulting nonwoven silk mats had lower

permeability than the unprocessed cocoon shells as the melt flow of sericin binder fused

together adjacent cocoons as well as adjacent fibres (Fig. 3).

Fig. 3. The Bombyx mori cocoon seen in (a) is a hierarchical 3D natural polymer composite structure with multiple nonwoven porous laminas of 2D randomly-oriented silk fibroin fibres that are bonded by sericin protein (b), where the fibres exist as dual-strands (c). Our nonwoven mat reinforcements (d) were produced by hot-pressing wet cocoon shells. Hot-pressing causes the melt flow of the natural sericin binder, which consequentially fuses adjacent fibres (e, f) and cocoon shells (d). Consequently, the nonwoven mats had lower permeability than the natural cocoon shell.

2.2 Composite manufacture

Composite laminates (150 mm square, 2.6-2.9 mm thick) were fabricated using the

vacuum-driven resin transfer moulding technique in a rigid aluminium mould tool with a

fixed-depth cavity. The reinforcement nonwoven mats and woven textile fabrics were used

Page 12 of 27

as-produced and as-received, respectively (i.e. without any preconditioning). Resin infusion

was carried out under vacuum at 50-100 mbar absolute pressure. Laminates were made with

the epoxy system. Post cure was carried out at 70 °C for 15 h after ambient cure for 24 h.

Assuming no porosity, the estimated fibre volume fraction vf and density ρc of the nonwoven

and biaxial SFRPs was 36.2% and 1.20 gcm-3, and 45.2% and 1.22 gcm-3, respectively.

2.3 Testing of mechanical properties

The laminates were machined under dry conditions (i.e. without liquid coolant) using

a fine tooth band saw to prepare test specimens for mechanical property characterisation. The

plain woven silk composites were tested along the warp (0°) direction. For each test, at least

five specimens of each composite type were tested.

Tensile tests were conducted according to ISO 527-4:1997 using an Instron 5582

testing machine equipped with a 100 kN calibrated load cell and a 25 mm clip-on

extensometer. A cross-head speed of 2 mm/min was used. The tensile modulus (in the strain

range of 0.025–0.100% [53]) Ec, ultimate tensile strength σc, and ultimate tensile failure strain

(at maximum stress) εc were measured from the stress-strain curves.

Three-point bending flexural tests were performed according to ISO 178:1997 using a

Hounsfield testing machine equipped with a three-point bending fixture and 10 kN calibrated

load cell. A cross-head speed of 2 mm/min was used. The flexural modulus, flexural strength,

and ultimate flexural strain at maximum stress were measured from the stress-strain curves,

where stress and strain were calculated at the outer surface (i.e. convex or tension side) of the

test specimen at mid-span.

Short-beam shear tests were carried out according to ASTM D2344, where un-

notched specimens were loaded in a three-point bending configuration at a cross-head speed

of 2 mm/min. A Zwick Z150 testing machine equipped with a 150 kN calibrated load cell

was used for these tests. A span-to-thickness ratio of 4:1 was used to encourage failure of

specimen through interlaminar shear along the neutral axis. Five specimens were tested for

each type of composite and the mean value for the ‘apparent’ interlaminar shear strength

(ILSS) was calculated.

The impact strength (or work of fracture) of the composites was determined according

to ISO 179:1997, using a Zwick-Roell Hit 50P Charpy impact test machine. The un-notched

specimens were loaded flat-wise with a 50 J hammer at a striking velocity of 3.8 ms-1.

Page 13 of 27

2.4 Reference data on PFRPs and GFRPs

The measured mechanical properties of nonwoven and plain woven silk-epoxy

composites were compared with the typical properties of nonwoven and plain woven flax-

and glass-epoxy composites manufactured using liquid moulding process found in literature

(Table 2). Note that to enable a true comparison to the silk composites, which employ a high

failure strain epoxy, our referenced data on flax and glass composites include results for plain

woven reinforcements that also employ the same resin system. In general, while the fibre

volume fractions of the PFRPs and GFRPs were in a similar range to our SFRPs, the densities

of the composites were generically in the following order: SFRP ≤ PFRP < GFRP (Table 2).

3 Results and discussion

3.1 Tensile and flexural properties

The stress-strain curves in Fig. 4 qualitatively describe the behaviour of nonwoven

and plain woven silk composites subjected to tensile and flexural loads. Two particular trends

were noticed. Firstly, the flexural stress-strain curves i) initially (up to ~1% strain) lay on, but

then ii) extended above and beyond (in terms of stress and strain, respectively), the tensile

stress-strain curves. While the former indicated that the flexural and tensile moduli were

likely to be quantitatively similar, the latter advised that the fracture strength and strain were

much higher in flexural mode. Flexural and tensile properties would be identical for a

homogenous isotropic material, however for heterogeneous, anisotropic fibre reinforced

composites, such as our silk composites, such a trend between flexural and tensile properties

is expected [54, 55] (as also seen in data for flax and glass composites in Fig. 5). This is

mainly because under flexural loads specimens are subjected to mixed-mode conditions with

tensile and compressive loads on opposite faces and shear loads close to the neutral axis, and

the post-cracking (i.e. post-elastic region) stress-strain curve in tension is different from that

in compression [54, 55]. Nonetheless, as flexural tests typically produce a tension-dominated

failure, as was the case for our silk composites, qualitative strength and damage trends concur

between flexural and tensile data.

A second trend observed from Fig. 4 was that plain woven silk composites endured

larger stresses and strains than nonwoven silk composites. The initial slope of the stress-strain

curves was also higher for woven composites. Such a trend is commonly observed in other

fibre reinforced composites [1, 31] (as also seen in data for flax and glass composites in Fig.

5). It is principally a result of the higher orientation distribution factor ηo of a balanced plain

Page 14 of 27

weave fabric (neglecting crimp) with ηo = 1/2 (= 0.50) in comparison to a nominally random

nonwoven mat with ηo = 3/8 (≈ 0.38). The greater reinforcing effect in woven composites due

to a typically higher fibre volume fraction than nonwoven composites also plays a role (Table

2). In addition, nonwoven composites typically experience matrix- and interface-dominated

brittle failure due to limited crack-stopping capacities, while woven composites exhibit better

post-cracking strain capacities [31, 56].

Fig. 4. Typical stress-strain curves of the nonwoven (solid line) and plain woven (dotted line) silk composites subjected to tensile (red line) and flexural (blue line) loads.

The measured tensile and flexural properties of the silk composites are presented in

Fig. 5 and Table 2, alongside typical properties of flax and glass composites reported in

literature.

Page 15 of 27

Table 2. Comparison of the physical and mechanical properties of the manufactured silk-epoxy composites with typical properties of flax-epoxy and glass-epoxy composites found in literature.

Physical properties Tensile mechanical properties Flexural mechanical properties

Composite Fibre volume fraction

Density

Stiffness Ultimate strength

Ultimate strain

Stiffness Ultimate strength

Ultimate strain

ILSS Impact strength

Source

[%] [gcm-3] [GPa] [MPa] [%] [GPa] [MPa] [%] [MPa] [kJm-2]

Nonwoven silk-epoxy 36.2 1.20 5.4 ± 0.2 60 ± 5 1.3 ± 0.1 5.2 ± 0.2 143 ± 10 3.4 ± 0.4 31.0 ± 3.7 16 ± 1 This study

Plain woven silk-epoxy 45.2 1.22 6.5 ± 0.1 111 ± 2 5.2 ± 0.2 6.4 ± 0.4 250 ± 4 6.9 ± 0.2 42.6 ± 5.9 115 ± 7 This study

Nonwoven flax-epoxy 15-35 1.20-1.26 5.8–9.8 37–75 0.8–1.6 4.8–6.7 55–91 2.1–3.2 13.6–26.7 8–15 [1, 5, 57-60]

Plain woven flax-epoxy 30-55 1.24-1.32 7.3–11.2 63–89 1.5–2.9 2.1–10.1 57–195 3.3–4.9 9.7–23.3 23–36 [1, 5, 59, 61-65]

Nonwoven glass-epoxy 15-45 1.36-1.80 10.2–16.7 123–241 1.0–2.1 9.0–11.4 192–325 3.0–4.0 25.0–35.0 73–107 [1, 5, 8, 56, 60, 66]

Plain woven glass-epoxy 30-65 1.58-2.09 17.0–24.0 350–500 2.1–2.5 13.2–22.0 370–560 3.5–4.0 38.0–52.0 165–280 [1, 5, 8, 62, 65-67]

Page 16 of 27

3.1.1 Tensile and flexural moduli

The tensile and flexural moduli of the i) nonwoven and ii) plain woven silk

composites were in the range of i) 5.2-5.4 GPa, and ii) 6.4-6.5 GPa. The silk fibre tensile

stiffness Ef was back-calculated using the rule-of-mixtures (Eq. (2); refer to [1]), assuming a)

length distribution factors of unity, and b) orientation distribution factors ηo of 0.38 for the

nonwoven composite and 0.50 for the plain woven composite. The estimated silk fibre

stiffness Ef was in the range of 24.7-27.1 GPa from nonwoven silk composites, and 21.0-21.8

GPa from woven silk composites. Notably, the back-calculated stiffness of B. mori silk was

found to be much higher than values typically quoted in literature of 5-15 GPa (Table 1).

Several articles on plant fibres and their composites have raised a similar observation [68-71]

and suggested that the discrepancy is likely due to the use of an incorrect (over-estimated)

fibre cross-section area of non-circular natural fibres when measuring the fibre properties.

σηη

strengthorEstiffnessisMwherev

MvMM

fol

mfcf ,,

)1( −−= Eq. (2)

The higher nominal back-calculated fibre stiffness of silk fibres in the nonwoven

composites compared to woven composites may be due to a number of possible reasons.

Firstly, fibres in the nonwoven mat were un-degummed (i.e. natural sericin coating was

present) while fibres in the commercial woven fabric were degummed (for improved

dyeability), and degumming processes have notable detrimental effects on the tensile

properties of the silk fibres [11, 44-48]. Secondly, perhaps silk fibres in the woven fabric

were damaged during mechanical processing (e.g. reeling, spinning or weaving). This has

been shown to be the case for plant fibre reinforcements, where every additional processing

step increases the number of defects (in the form of kink bands) and subsequently reduces the

fibre mechanical properties [1, 72]. Thirdly, perhaps the presence of hydrophilic sericin on

the surface of silk fibres in the nonwoven mat may induce better bonding with the polar

epoxy matrix than the degummed silk fibres in the woven fabric.

Fig. 5 also compares the stiffness of SFRPs with similar PFRPs and GFRPs. The

tensile modulus of nonwoven silk composites (5.4 ± 0.2 GPa) was just about comparable to

nonwoven flax composites (5.8-9.8 GPa), but much lower than nonwoven glass composites

(10.2-16.7 GPa). Similarly, the tensile modulus of woven silk composites (6.5 ± 0.1GPa) was

almost comparable to woven flax composites (7.3-11.2 GPa), but much lower than woven

glass composites (17.0-24.0 GPa). The flexural modulus followed a similar trend. In all cases,

Page 17 of 27

the stiffness was enhanced by fibre reinforcement in comparison to the neat matrix. The

results demonstrated that multiaxial silk composites could not replace flax composites, and

certainly not glass composites, in absolute stiffness-critical applications, like automotive

interior components.

Fig. 5. The tensile (left) and flexural (right) properties of nonwoven and plain woven silk composites (measured; error bars indicate ±1 std. dev.) in comparison to nonwoven and plain woven flax and glass composites (from literature; upper error bar indicates typical range).

Page 18 of 27

3.1.2 Tensile and flexural strengths

The tensile strength of nonwoven and woven silk composites was 60 ± 5 MPa and

111 ± 2 MPa, respectively. The flexural strengths were 2.3-2.4 times higher than the tensile

strengths. It was noted that the tensile strength of the nonwoven silk composites was lower

than that of the matrix (60-75 MPa); that is, the silk fibres in the nonwoven mat were not

reinforcing the matrix as such. This, however, is common for traditional nonwoven

composites [1, 31].

Similarly as before, the silk fibre tensile strength σf could be back-calculated using the

rule-of-mixtures (Eq. (2)). The estimated silk fibre strength σf was in the range of 90-160

MPa from nonwoven silk composites, and 309-346 MPa from woven silk composites. While

the back-calculated strength of B. mori silk from woven composites did lie in the typical

range quoted in literature (0.3-0.7 GPa; Table 1), the value obtained from nonwoven

composites was much lower. However, it was noted that as silk fibres in the nonwoven mat

were fused together by the sericin matrix (Fig. 3), the reinforcement may behave more like a

bulk material rather than as individual fibres. It is well known that large scale polymer

components have lower failure initiation stresses than the polymer fibres [18]. In addition, we

believe that perhaps the low permeability of the nonwoven silk mat may have resulted in a

heterogeneous distribution of fibres, like in wood-epoxy laminates. This would result in the

development of considerable shear stresses at the interface and assist in the premature brittle

fracture observed.

Fig. 5 illustrates that the nonwoven and woven silk composites had tensile and

flexural strengths comparable to flax composites but lower than glass composites. Thus our

measurements demonstrated that multiaxial silk composites could replace similar flax

composites, but not glass composites, in terms of absolute strength.

3.1.3 Tensile and flexural strains at maximum stress

The high toughness of silk fibre results from its high cohesive energy density

(resulting from its high hydrogen bond energy density) [18] and is visible as a large area

under the stress-strain curve. To transfer the high ductility of silk fibres into the composites,

we had employed a high failure strain epoxy matrix. As shown in Fig. 5, we found that

woven silk composites exhibited high strains at maximum stress of 5.2% in tension and 6.9%

in flexure. In fact, to determine the flexural properties of the woven silk composites

corrections for large deflections had to be made, as per ISO 178:1997. In comparison, woven

Page 19 of 27

flax and glass composites had much lower strains at maximum stress of 1.5-2.9% in tension

and 3.3-4.9% in flexure (Fig. 5). Flax and glass fibres have similar failure strains (Table 1);

consequently their composites also showed comparable strains at maximum stress, as

composite failure was dominated by fibre fracture [53, 73]. Our results, therefore, indicate

that plain woven SFRPs may be useful in applications where high compliance is required.

Notably, nonwoven composites had consistently lower strain capacities than woven

composites. In addition, nonwoven composites of silk, flax and glass fibres had comparable

failure strains.

3.1.4 Specific tensile and flexural properties

Generally, minimising material weight (density ρ), cost and/or eco-impact are key

objectives for industrial products. The key mechanical parameters, defined by the component

function and constraint, are typically stiffness E and strength σ. Using Ashby’s approach [74],

product engineers would consequently use materials performance indices, defined in [74], for

the comparison and selection of various materials. For reference, Shah [1, 5] has previously

produced comprehensive Ashby-type materials selection charts for evaluating PFRPs against

other engineering materials.

Here, we consider reducing component weight as a primary objective. The critical

material performance indices that need to be maximised for a light-weight beam/plate loaded

in pure tension are specific tensile stiffness E/ρ and specific tensile strength σ/ρ. For a

beam/plate loaded in bending mode, specific flexural stiffness E1/3/ρ and strength σ1/2/ρ need

to be maximised.

The lower density of silk composites in comparison to flax and glass composites

(Table 2) is advantageous for the former. Fig. 6 compares the specific tensile and flexural

properties of silk, flax and glass composites. It was observed that silk composites (both

nonwoven and woven) had specific tensile properties (both stiffness and strength)

comparable to flax composites, but were still inferior to glass composites. However, density

has an appreciably larger effect on specific flexural properties than specific tensile properties

[74]. Notably, silk composites (both nonwoven and woven) had comparable, if not higher,

flexural properties (both stiffness and strength) than both flax and glass composites.

Therefore, multiaxial silk composites may be investigated as alternative materials to similar

flax and even glass composites for flexural stiffness- or strength-limited design at minimum

mass. Example applications would include i) structural load floors in automotives, ii)

Page 20 of 27

composite construction beams and roof panels, iii) sports and leisure products like surf

boards, and iv) small rotor blades.

Certainly, depending on the application, other factors such as materials and

manufacturing cost (both, economic and environmental), and component operating conditions

and design life (i.e. materials environmental aging properties) may require scrutiny. This

needs to be evaluated for natural fibre composites and for SFRPs, in particular.

Fig. 6. The specific tensile (left) and flexural (right) properties of nonwoven and plain woven silk composites (measured) in comparison to nonwoven and plain woven flax and glass composites (from literature; upper error bar indicates typical range).

3.2 Interlaminar shear strength

Fig. 7 and Table 2 present the ‘apparent’ interlaminar shear strength (ILSS) of the

composites measured through short beam shear tests. While the ILSS is a measure of the

strength of the matrix plus the interface, it is sensitive to several parameters (e.g. specimen

dimensions) and cannot be directly compared with interfacial strengths measured through

other techniques [31]. ILSS values presented for flax and glass composites were from studies

that have used similar test parameters as in our study.

It was observed that nonwoven composites had lower ILSS than woven composites.

Nonwoven silk composites had an ILSS of 31.0 ± 3.7 MPa, while woven silk composites had

an ILSS of 42.6 ± 5.9 MPa. Notably, the ILSS of SFRPs was much higher than that of PFRPs

Page 21 of 27

and comparable to that of GFRPs (Fig. 7). The high interfacial bonding between the silk

fibres and the polar epoxy matrix achieved without any active surface modification was

revealing as several articles in literature refer to poor fibre/matrix adhesion as a typical

feature of natural fibre composites. Given that the tensile and flexural stress-strain curves of

woven silk composites extended to large strains with only small increments in stress (Fig. 4),

it was likely that the high interfacial strength of SFRPs provided strength-maximising tensile

and flexural strain capacity, post-cracking (i.e. post-yield).

Fig. 7. The apparent interlaminar shear strength (ILSS), and absolute and specific impact strengths of nonwoven and plain woven silk composites (measured; error bars indicate ±1 std. dev.) in comparison to nonwoven and plain woven flax and glass composites (from literature; upper error bar indicates typical range).

Page 22 of 27

The critical fibre lengths lc for the various composites can be estimated using Eq. (1),

where the ILSS as taken as an indicative value for interfacial shear strength τ, and respective

fibre properties are taken from (Table 1). We find that our silk composites have critical fibre

lengths lc ~ 0.1 mm, which is in the range of lc ≈ 0.1-0.3 mm estimated from literature in

Section 1.2. The flax and glass composites are both found to have critical fibre lengths in the

range of lc ≈ 0.2-0.6 mm; this is also in the range typically quoted in literature [1, 31].

Referring to Eq. (1), it is noted that the lower tensile strength σf and diameter df of silk fibres

(in comparison to flax and E-glass) play a factor in yielding the lower critical length lc for

SFRPs.

3.3 Impact strength

The toughness of a material can be defined as its ability to absorb energy (per unit

volume) without fracturing. The Charpy impact strength is a measure of material toughness.

The impact strength of the nonwoven and woven silk composites is compared to that of flax

and glass composites in Fig. 7 and Table 2. As a general trend, woven composites performed

much better than nonwoven composites. For instance, the impact strength of nonwoven and

woven silk composites was 17 kJm-2 and 115 kJm-2, respectively. The difference in

properties, however, was much lower in flax and glass composites, where nonwoven and

woven flax composites, for instance, had impact strength of 8-15 kJm-2 and 23-36 kJm-2,

respectively. The much lower impact strength of nonwoven silk composites was likely due to

premature brittle fracture resulting from catastrophic crack-propagation in a heterogeneous

microstructure as the epoxy matrix may not be well-distributed in the nonwoven mat.

The woven silk composite had a 44-64% higher impact strength than that of the neat

matrix (70-80 kJm-2). Typically, impact strength increases, alongside tensile and flexural

properties, with increases in fibre content [31]. Preparing woven silk composites with higher

fibre volume fraction (up to 60%), possibly using low (< 2 bar) external compaction pressure

[13] through a vacuum-assisted resin transfer moulding process or autoclave curing, would be

a next suitable step.

We also found that the silk composites had better (absolute and specific) impact

strength than flax composites (Fig. 7). Plant fibre composites are often quoted to have poor

impact performance [49, 50, 75]. Amongst plant fibres, owing to their high cellulose

microfibril angle, sisal and coir have high failure strains and toughness, but their low

Page 23 of 27

strengths results in composites with still inferior toughness [1]. Silk fibres, on the other hand,

offer a good balance of strength and ductility. In fact, while the impact strength of silk

composites was lower than that of glass composites, woven silk composites had almost

comparable specific impact strength to woven glass composites (Fig. 7). While woven silk

composites will certainly not replace similar glass composites, the former do offer a unique

opportunity to produce light-weight, tough components from natural precursor materials.

Potential applications may include light-weight, crashworthy and impact-critical components

such as in automotives (e.g. structural parts of agricultural vehicles), defence (e.g. light-

weight drones), and safety (e.g. high performance helmets).

3.4 Comments on sustainability

Considering the increasing renewed interest in engineering materials of natural origin,

silks seem to be a strong natural fibre candidate for reinforcements in polymer composites.

From an environmental perspective, however, as raw silk cocoons are produced in a two-step

process [15, 76], namely mulberry plant production and silkworm farming, the cumulative

energy demand of silk cocoon production is i) much higher than that of plant fibre

production, which relies on a single-step agricultural process, and ii) comparable to that of

glass fibre productions. Processed silk textiles, like processed plant fibre textiles, have much

higher production energy demands than glass fibre textile productions [1, 3, 20]. Therefore, a

detailed life cycle assessment should be carried out to examine the sustainability of such

natural fibre composites in comparison to glass composites. However, it is important to

mention that socio-economic aspects, such as job creation, also need to be considered as

natural fibres are agricultural fibres with notable benefit for the fibre crop growers and their

communities [15, 76].

4 Conclusions

We have highlighted three key recommendations in the development of SFRPs with

useful properties: i) for compatibility (viz. property and processing) with the fibre, a matrix

with high-failure strain and low-processing temperature needs to be selected, ii) high fibre

volume fractions (ideally above 40%, if not 60%) need to be employed to ensure the fibres

carry a larger fraction of the load, and iii) long (>> 4mm) silk fibres that are not actively

surface treated may provide adequate fibre/matrix adhesion.

Through a cross-comparative study, we found that the specific flexural properties

(strength and stiffness), interlaminar shear strength and specific impact strength of nonwoven

Page 24 of 27

and plain woven SFRPs was higher than that of flax composites, and more comparable to, but

not better than that of GFRPs. The absolute stiffness (tensile and flexural) of SFRPs was

lower than that of PFRPs and GFRPs. However, plain woven SFRPs demonstrated high

fracture strain capacities, which may be particularly attractive in applications where

progressive failure or high compliance is required.

In general, it is evident that SFRPs offer some unique property advantages over

PFRPs, and while they do not suffice to replace GFRPs, SFRPs are an interesting sustainable

materials option. Certainly, depending on the application, factors such as materials and

manufacturing cost, and component operating conditions and design life (i.e. materials

environmental aging properties) may require scrutiny.

Acknowledgements

We thank the US Air Force Office for Scientific Research (AFOSR Grant Number

F49620-03-1-0111) and the European Research Council Advanced Grant (SP2-GA-2008-

233409) for generous funding. We thank Momentive Specialty Chemicals GmbH (Stuttgart,

Germany) for in-kind supply of the matrix resin. We thank Dr Clive Siviour and Dr Igor

Dyson (University of Oxford) for providing access to tensile and flexural testing machines.

We also thank Dr Fujia Chen (Williams F1, UK) and Dr Chris Holland (University of

Sheffield) for kindly running Charpy impact tests and short beam shear tests on our numerous

specimens. We also thank the anonymous referees for suggesting valuable improvements to

the manuscript.

References

1. Shah D. Developing plant fibre composites for structural applications by optimising composite parameters: a critical review. Journal of Materials Science, 2013, 48(18): p. 6083-6107.

2. Misnon M, Islam MM, Epaarachchi JA, Lau KT. Potentiality of utilising natural textile materials for engineering composites applications. Materials & Design, 2014, 59: p. 359-368.

3. Summerscales J, Dissanayake N, Virk AS, Hall W. A review of bast fibres and their composites. Part 2 – Composites. Composites Part A: Applied Science and Manufacturing, 2010, 41(10): p. 1336-1344.

4. Faruk O, Bledzki AK, Fink HP, Sain M. Biocomposites reinforced with natural fibres: 2000-2010. Progress in Polymer Science, 2012, 37(11): p. 1552-1596.

5. Shah D. Natural fibre composites: Comprehensive Ashby-type materials selection charts. Materials & Design, 2014, 62: p. 21-31.

6. Dicker M, Duckworth PF, Baker AB, Francois G, Hazzard MK, Weaver PM. Green composites: A review of material attributes and complementary applications. Composites Part A: Applied Science and Manufacturing, 2014, 56: p. 280-289.

7. Carus M, Eder A, Dammer L, Korte H, Scholz L, Essel R, Breitmayer E. Wood-plastic composites (WPC) and natural fibre composites (NFC): European and global Markets 2012 and future trends. WPC/NFC Market Study 2014-03, 2014. nova-Institut GmbH: Hürth, Germany.

8. Sims G. 2.05 - Glass fiber reinforced plastics - Properties, in Comprehensive composite materials, Kelly A, Zweben C, 2000. Elsevier. p. 151–197.

9. Vepari C, Kaplan DL. Silk as a biomaterial. Progress in Polymer Science, 2007, 32: p. 991-1007.

Page 25 of 27

10. Hardy J, Scheibel TR. Composite materials based on silk proteins. Progress in Polymer Science, 2010, 35: p. 1093-1115.

11. Altman G, Diaz F, Jakuba C, Calabro T, Horan RL, Chen J, Lu H, Richmond J, Kaplan DL. Silk-based biomaterials. Biomaterials, 2003, 24: p. 401-416.

12. Hakimi O, Knight DP, Vollrath F, Vadgama P. Spider and mulberry silkworm silks as compatible biomaterials. Composites: Part B, 2007, 38: p. 324-337.

13. Shah D, Porter D, Vollrath F. Opportunities for silk textiles in reinforced biocomposites: Studying through-thickness compaction behaviour. Composites Part A: Applied Science and Manufacturing, 2014, 62: p. 1-10.

14. Lewin M. Handbook of fiber chemistry. Third ed, 2007. Boca Raton: CRC Press LLC. 15. Vollrath F, Carter R, Rajesh GK, Thalwitz G, Astudillo MF. Life cycle analysis of cumulative energy

demand on sericulture in Karnataka, India, in 6th BACSA International Conference: Building Value Chains in Sericulture (BISERICA 2013). 2013. Padua, Italy.

16. Agnarsson I, Kuntner M, Blackledge TA. Bioprospecting finds the toughest biological material: extraordinary silk from a giant riverine orb spider. PLOS ONE, 2010, 5(9): p. e11234.

17. Swanson B, Blackledge TA, Hayashi CY. Spider capture silk: performance implications of variation in an exceptional biomaterial. Journal of Experimental Zoology Part A: Ecological Genetics and Physiology, 2007, 307A: p. 654-666.

18. Porter D, Guan J, Vollrath F. Spider silk: Super material or thin fibre? Advanced Materials, 2013, 25(9): p. 1275-1279.

19. Shah D, Schubel PJ, Licence P, Clifford MJ Hydroxyethylcellulose surface treatment of natural fibres: the new ‘twist’ in yarn preparation and optimization for composites applicability. Journal of Materials Science, 2012, 47: p. 2700-2711.

20. Dissanayake N, Summerscales J, Grove SM, Singh MM. Energy use in the production of flax fiber for the reinforcement of composites. Journal of Natural Fibers, 2009, 6(4): p. 331-346.

21. Dissanayake N, Summerscales J, Grove SM, Singh MM. Life cycle impact assessment of flax fibre for the reinforcement of composites. Journal of Biobased Materials and Bioenergy, 2009, 3(3): p. 1-4.

22. Manjula K, Kumar MNS, Soare BG, Picciani P, Siddaramaiah. Biobased chain extended polyurethane and its composites with silk fiber. Polymer Engineering and Science, 2010: p. 851-856.

23. Cheung H, Lau KT, Pow YF, Zhao YQ, Hui D. Biodegradation of a silkworm silk/PLA composite. Composites Part B: Engineering, 2010, 41: p. 223-228.

24. Ho M, Lau KT, Wang H, Bhattacharyya D. Characteristics of a silk fibre reinforced biodegradable plastic. Composites Part B: Engineering, 2011, 42: p. 117-122.

25. Shubhra Q, Alam AKMM. Effect of gamma radiation on the mechanical properties of natural silk fiber and synthetic E-glass fiber reinforced polypropylene composites: A comparative study. Radiation Physics and Chemistry, 2011, 80: p. 1228-1232.

26. Han S, Ahn HJ, Cho D. Hygrothermal effect on henequen or silk fiber reinforced poly(butylene succinate) biocomposites. Composites Part B: Engineering, 2010, 41: p. 491-497.

27. Zhao Y, Cheung HY, Lau KT, Xu CL, Zhao DD, Li HL. Silkworm silk/poly(lactic acid) biocomposites: Dynamic mechanical, thermal and biodegradable properties. Polymer Degradation and Stability, 2010, 95: p. 1978-1987.

28. Setua D, Dutta B. Short silk fiber-reinforced polychloroprene rubber composites. Journal of Applied Polymer Science, 1984, 29: p. 3097-3114.

29. Setua D, De SK. Short silk fibre reinforced nitrile rubber composites. Journal of Materials Science, 1984, 19: p. 983-999.

30. Setua D, De SK. Effect of short fibres on critical cut length in tensile failure of rubber vulcanizates. Journal of Materials Science, 1985, 20: p. 2653-2660.

31. Harris B. Engineering composite materials, 1999. London: The Institute of Materials. 32. Craven J, Cripps R, Viney C. Evaluating the silk/epoxy interface by means of the Microbond Test.

Composites Part A: Applied Science and Manufacturing, 2000, 31: p. 653-660. 33. Eshkoor R, Oshkovr SA, Sulong AB, Zulkifli R, Ariffin AK, Azhari CH. Comparative research on the

crashworthiness characteristics of woven natural silk/epoxy composite tubes. Materials and Design, 2013, 47: p. 248-257.

34. Ude A, Eshkoor RA, Zulkifili R, Ariffin AK, Dzuraidah AW, Azhari CH. Bombyx mori silk fibre and its composite: A review of contemporary developments. Materials & Design, 2014, 57: p. 298-305.

35. Ataollahi S, Taher ST, Eshkoor RA, Ariffin AK, Azhari CH. Energy absorption and failure response of silk/epoxy composite square tubes: Experimental. Composites Part B: Engineering, 2012, 43: p. 542-548.

Page 26 of 27

36. Eshkoor R, Ude AU, Oshkovr SA, Sulong AB, Zulkifli R, Ariffin AK, Azhari CH. Failure mechanism of woven natural silk/epoxy rectangular composite tubes under axial quasi-static crushing test using trigger mechanism. International Journal of Impact Engineering, 2014, 64: p. 53-61.

37. Ude A, Ariffin AK, Azhari CH. Impact damage characteristics in reinforced woven natural silk/epoxy composite face-sheet and sandwich foam, coremat and honeycomb materials. International Journal of Impact Engineering, 2013, 58(31-38).

38. Zulkifli R, Azhari CH, Ghazali MJ, Ismail AR, Sulong AB. Interlaminar fracture toughness of multi-layer woven silk/epoxy composites treated with coupling agent. European Journal of Scientific Research, 2009, 27(3): p. 454-462.

39. Loh K, Tan CKW. Natural silkworm silk-epoxy resin composite for high performance application, in Metal, ceramic and polymeric composites for various uses, Cuppoletti J, 2011. InTech.

40. Priya S, Ramakrishna HV, Rai SK, Rajulu AV. Tensile, flexural, and chemical resistance properties of waste silk fabric-reinforced epoxy laminates. Journal of Reinforced Plastics and Composites, 2005, 24: p. 643-648.

41. van Vuure A, Vanderbeke J, Osorio L, Trujillo R, Fuentes C, Verpoest I. Natural fibre composites: Tough (silk) and strong (bamboo), in 17th International Conference on Composite Materials (ICCM17). 2009. Edinburgh, UK.

42. Cuypers H, Wastiels J. Stochastic matrix-cracking model for textile reinforced cementitious composites under tensile loading. Materials and Structures, 2006, 39: p. 777-786.

43. Wu H, Li VC. Stochastic process of multiple cracking in discontinuous random fiber reinforced brittle matrix composites. International Journal of Damage Mechanics, 1995, 4: p. 83-102.

44. Ho M, Wang H, Lau KT. Effect of degumming time on silkworm silk fibre for biodegradable polymer composites. Applied Surface Science, 2012, 258: p. 3948-3955.

45. Ho M, Wang H, Lau KT, Lee JH, Hui D. Interfacial bonding and degumming effects on silk fibre/polymer biocomposites. Composites Part B: Engineering, 2012, 43: p. 2801-2812.

46. Mahmoodi N, Arami M, Mazaheri F, Rahimi S. Degradation of sericin (degumming) of Persian silk by ultrasound and enzymes as a cleaner and environmentally friendly process. Journal of Cleaner Production, 2010, 18: p. 146-151.

47. Khan M, Tsukada M, Gotoh Y, Morikawa H, Freddi G, Shiozaki H. Physical properties and dyeability of silk fibers degummed with citric acid. Bioresource Technology, 2010, 101: p. 8439-8445.

48. Jiang P, Liu H, Wang C, Wu L, Huang J, Guo C. Tensile behavior and morphology of differently degummed silkworm (Bombyx mori) cocoon silk fibres. Materials Letters, 2006, 60: p. 919-925.

49. Wambua P, Ivens J, Verpoest I. Natural fibres: can they replace glass in fibre reinforced plastics? Composites Science and Technology, 2003, 63: p. 1259-1264.

50. Pavithran C, Mukherjee PS, Brahmakumar M, Damodaran AD. Impact properties of natural fibre composites. Journal of Materials Science Letters, 1987, 6: p. 882-884.

51. Carter R, Scott LA, Holland CA, Vollrath F. WO 2012172085 A2: Non-woven silk materials and methods for manufacture thereof, 2012.

52. Chen F, Porter D, Vollrath F. Silk cocoon (Bombyx mori): Multi-layer structure and mechanical properties. Acta Biomaterialia, 2012, 8: p. 2620-2627.

53. Shah D, Schubel PJ, Clifford MJ, Licence P. The tensile behavior of off-axis loaded plant fiber composites: an insight on the non-linear stress-strain response. Polymer Composites, 2012, 33(9): p. 1494-1504.

54. Chamis C, ed. Test methods and design allowables for fibrous composites. 1989. American Society for Testing and Materials (ASTM): Baltimore, USA.

55. Kelly A, ed. Concise encyclopedia of composite materials. 1994. Elsevier Science Ltd.: Oxford, UK. 56. Sanders B, ed. Short fiber reinforced composite materials. 1982. American Society for Testing and

Materials (ASTM): Baltimore, USA. 57. George J, Ivens J, Verpoest I. Mechanical properties of flax fibre reinforced epoxy composites. Die

Angewandte Makromolekulare Chemie, 1999, 272(1): p. 41-45. 58. Bos H. The potential of flax fibres as reinforcement for composite materials. PhD, 2004. Technische

Universiteit Eindhoven: Eindhoven, Netherlands. 59. Goutianos S, Peijs T, Nystrom B, Skrifvars M. Development of flax fibre based textile reinforcements

for composite applications. Applied Composite Materials, 2006, 13(4): p. 199-215. 60. Rodríguez E, Petrucci R, Puglia D, Kenny JM, Vazquez A. Characterization of composites based on

natural and glass fibers obtained by vacuum infusion. Journal of Composite Materials, 2005, 39(5): p. 265-282.

61. Meredith J, Coles SR, Powe R, Collings E, Cozien-Cazuc S, Weager B, Müssig J, Kirwan K. On the static and dynamic properties of flax and Cordenka epoxy composites. Composites Science and Technology, 2013, 80: p. 31-38.

Page 27 of 27

62. Liang S, Gning, PB, Guillaumat, L. A comparative study of fatigue behaviour of flax/epoxy and glass/epoxy composites. Composites Science and Technology, 2012, 72(5): p. 535-543.

63. Adekunle K, Cho SW, Patzelt C, Blomfeldt T, Skrifvars M. Impact and flexural properties of flax fabrics and Lyocell fiber-reinforced bio-based thermoset. Journal of Reinforced Plastics and Composites, 2011, 30: p. 685-697.

64. Muralidhar B, Giridev VR, Raghunathan K. Flexural and impact properties of flax woven, knitted and sequentially stacked knitted/woven preform reinforced epoxy composites. Journal of Reinforced Plastics and Composites, 2012, 31(6): p. 379-388.

65. Dissanayake N, Summerscales J, Grove SM, Singh MM. Infusion of natural vs. Synthetic fibre composites with similar reinforcement architecture in the context of a LCA, in The 9th International Conference on Flow Processes in Composite Materials (FPCM-9). 2008. Montreal, Canada.

66. CES EduPack, 2012. Granta Design Limited: Cambridge, UK. 67. , Technical Data Sheet - EPIKOTE Resin MGS RIMR 135 and EPIKURE Curing Agent MGS RIMH

134–RIMH 137, 2006. Momentive Specialty Chemicals GmbH: Stuttgart, Germany. 68. Charlet K, Baley C, Morvan C, Jernot JP, Gomina M, Bréard J. Characteristics of Herme`s flax fibres

as a function of their location in the stem and properties of the derived unidirectional composites. Composites Part A: Applied Science and Manufacturing, 2007, 28: p. 1912-1921.

69. Charlet K, Jernot JP, Gomina M, Bizet L, Bréard J. Mechanical properties of flax fibers and of the derived unidirectional composites. Journal of Composite Materials, 2010, 44(24): p. 2887-2896.

70. Virk A, Hall W, Summerscales J. Modulus and strength prediction for natural fibre composites. Materials Science and Technology, 2012, 28(7): p. 864-871.

71. Oksman K, Wallstrom L, Berglund LA, Filho RDT. Morphology and mechanical properties of unidirectional sisal-epoxy composites. Journal of Applied Polymer Science, 2002, 84: p. 2358-2365.

72. Hanninen T, Thygesen A, Mehmood S, Madsen B, Hughes M. Mechanical processing of bast fibres: The occurrence of damage and its effect on fibre structure. Industrial Crops and Products, 2012, 39: p. 7-11.

73. Shah D, Schubel PJ, Clifford MJ, Licence P. Mechanical property characterization of aligned plant yarn reinforced thermoset matrix composites manufactured via vacuum infusion. Polymer-Plastics Technology and Engineering, 2014, 53: p. 239-253.

74. Ashby M. Materials selection in mechanical design, 1992. Oxford, UK: Pergamon Press. 75. Santulli C. Impact properties of glass/plant fibre hybrid laminates. Journal of Materials Science, 2007,

42: p. 3699-3707. 76. Akovali G. Handbook of composite fabrication, 2001: Rapra Technology Ltd.