You are on page 1of 8

Eur. Phys. J.

E 16, 199206 (2005)


DOI: 10.1140/epje/e2005-00021-2

THE EUROPEAN
PHYSICAL JOURNAL E

Predicting the mechanical properties of spider silk as a model


nanostructured polymer
D. Porter1,a , F. Vollrath1 , and Z. Shao2
1
2

Department of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS, UK
Department of Macromolecular Science and Key Laboratory of Molecular Engineering of Polymers of Ministry of Education,
Fudan University, 200433 Shanghai, P.R. China
Received 4 August 2004 / Received in nal form 6 October 2004
c EDP Sciences, Societ`
Published online 22 February 2005 
a Italiana di Fisica, Springer-Verlag 2005
Abstract. Spider silk is attractive because it is strong and tough. Moreover, an enormous range of mechanical properties can be achieved with only small changes in chemical structure. Our research shows that
the full range of thermo-mechanical properties of silk bres can be predicted from mean eld theory for
polymers in terms of chemical composition and the degree of order in the polymer structure. Thus, we can
demonstrate an inherent simplicity at a macromolecular level in the design principles of natural materials.
This surprising observation allows in depth comparison of natural with man-made materials.
PACS. 87.15.La Mechanical properties 81.05.Lg Polymers and plastics; rubber; synthetic and natural
bers; organometallic and organic materials

1 Introduction
Spider silk is a strong and tough polypeptide that is optimised by nature to full a wide range of functions using
subtle changes in chemical composition and, importantly,
morphological structure at a nanometer scale [1,2]. Attempts to emulate the attractive properties of silk and
other natural polymers are frustrated by a lack of quantitative structure-property relations, which are the subject
of this work.
In order to develop a quantitative model for silk, let us
take the radical step of looking at silk from the perspective
of a user. Silk bres are produced by a spider to manage
mechanical energy for dierent tasks without breaking: to
store elastic energy in supporting its own weight or for the
structural framework of a web, and to absorb kinetic energy to capture ying insects. Here we identify the mechanisms at a molecular level that dictate energy storage
and dissipation in a polymer and derive straightforward
analytical relations for the full range of mechanical properties that are possible in silk. These relations are expressed
in terms of a small number of energy-based parameters
with a direct fundamental link to chemical composition
and morphological order. In this way, we hope to elucidate some of the key design principles in natural polymers.
Moreover, this approach can be applied also to much simpler man-made polymers, thereby oering the potential
for the design of improved synthetic polymers.


Contribution presented at the World Polymer Congress


MACRO 2004, Paris, France, July 2004
a
e-mail: DPORTER@qinetiq.com

Why look specically at spider silk? Although silkworm silk is strong, the oscillatory motion of the worm
during spinning leads to great uctuations in mechanical properties [3,4]. Spider silks, on the other hand, are
evenly spun, which provides a model natural bre with
superb combinations of strength and toughness. Such silks
range from drag-line with a modulus and strength of about
10 GPa and 1 GPa respectively to capture thread with a
more modest strength, but because of its high extensibility
a comparably high toughness.
Figure 1 shows a schematic diagram of the stressstrain prole of a silk bre through a general loading cycle and also the deformation through to break. The initial modulus, Ei , takes an upper value for dragline silk of
about 15 GPa, down to about 4 GPa for capture thread.
The yield strain, y , under ambient conditions is usually
about 2%. Other parameters of yield stress, y , post-yield
modulus, Ey , and stress and strain to break, b and b respectively, are dependent upon the detailed composition
and morphology of the silk polymer. Generally, a high initial modulus and failure stress has a lower strain to failure,
and vice versa.
Recent experimental measurements of dynamic mechanical and stress-strain properties of single bres of
Nephila Major Ampullate drag-line silk (over a wide range
of temperatures from 100 to +350 C) suggest that the
mechanisms in silk deformation are no more complex than
for any other semi-crystalline polymer [5]. The small insert
in Figure 1 is a typical stress-strain plot at 15 C while
Figure 2 shows the temperature dependence of dynamic
tensile modulus and loss tangent for the same silk, with

200

The European Physical Journal E

Fig. 1. Schematic diagram of the stress-strain prole of a spider silk in a deformation cycle and during strain to break.
The insert shows experimental curves for a dragline silk discussed in the text. The numbered regions relate to the following
deformation processes: 1. Initial elastic zone through to yield: amorphous domains change from glass to rubberlike states at
yield. 2. Post-yield elastic zone of a crystal-reinforced rubber. 3. Plastic zone as mechanical energy transforms rubber states
back to glass. 4. Recoverable elastic energy: approximately the sum of zones 1 and 2.

Fig. 2. Experimental DMTA data for tensile modulus and loss tangent and initial modulus from tensile tests for the mechanical
properties of a dragline silk discussed in the text.

points also showing the initial modulus from stress-strain


plots at dierent temperatures; loss tangent here is the ratio of energy dissipated to energy stored in a deformation
cycle, and plays a key role in the toughness of a polymer.
The semicrystalline morphology of spider silk has been
discussed extensively in literature, but the nanometer
scale of the dierent domains are dicult to dene in
terms of specic crystallographic forms with the same
clarity as synthetic polymers. Grubb proposes only 12%
crystallinity in the dragline silk of Nephilia clavipes using
wide angle X-ray diraction, which clearly represents the
-sheet domains of poly(alanine) segments [6]. Van Beek is
more circumspect, again identifying the -sheet domains
using NMR in silk from Nephilia edulis, but also a large
fraction of helical structures in the glycine residues with
a 3-fold symmetry alongside the truly disordered rigid
amorphous phase [7]. The term non-periodic lattice is
a useful way to describe order in silk, since neighbouring

polymer chain backbones can be aligned very regularly


with hydrogen bonding between the chains, but the irregular distribution of side-chains along the backbone chain
length leads to a lack of precise order perpendicular to the
chains [8].
Previous models have provided limited insights into
the mechanical properties of silk bres [9,10], but are
generally weak in their link to chemical composition and
semicrystalline morphology. For example, the network
model of Termonia simulates properties in remarkable
agreement with observation. This model is based upon
a hypothesis of a hydrogen-bonded rubber matrix reinforced by a fraction of rigid crystal domains, acting as
physical crosslink sites [11]; the sti hydrogen bonds are
rst broken to allow the dynamic rubber phase to redistribute the deformation eld to predict the nonlinear
large strain deformation. The NMR studies of van Beek
show that the non-crystalline phase in dry silk is not

D. Porter et al.: Mechanical properties of spider silk

rubber, but glassy, which means that care must be exercised in the use of the term rubber in any model [7]. The
conventional alternative to network models for polymers
is detailed molecular dynamics simulations [12]. However,
structures such as silk require enormous model calculations to simulate dimensions and time scales far shorter
than those of experiment. Consequently understanding
the mechanisms that quantitatively control complex mechanical properties are at present virtually impossible
with such a black-box approach. The same comments on
structure-property relations can be applied equally well to
semicrystalline synthetic polymers and bres, which are
dominated by phenomenological models, based upon continuum theory [13,14].

2 Model
Our work is based on two premises of energy storage and
energy dissipation. The rst implies that the high modulus
and great strength of silk are due to the high cohesive
energy density of hydrogen bonding between the amide
segments of the backbone in the primary silk protein
(Spidroin) molecules. The second premise implies that silk
toughness is due to the high energy density absorbed in
the post-yield work-hardening phase through to break.
The premises are derived from the dierent molecular
level mechanisms characterised in the stress-strain prole of Figure 1 i.e. 1) the change at yield of the amorphous domains from glass to rubber 2), the post-yield
complex modulus 3), the plastic behaviour as the rubbery
state converts back to glass and 4) the recoverable elastic
energy.
The analysis uses a mean-eld modelling method
called Group Interaction Modelling (GIM) that has been
used before to predict the engineering properties of polymers like poly(styrene) and poly(carbonate) [15]. GIM applies the ensemble average values of the thermodynamic
energy terms of a characteristic mer unit (at a molecular
level) to a potential function that quanties the relationship between intermolecular energy and separation distance between polymer chains. Thus, GIM allows analytical structure-property relations for mechanical properties
to be derived by a combination of mathematical processing
of the potential function and application of continuumlevel physics of a materials energy-storage and energyloss. Hence, the quantitative physical links between structure and properties are transparent at all stages of the
modelling process.
To avoid unnecessary controversy over crystalline fractions (see introduction), let us take the straightforward
position that silk consists of ordered and disordered
phases, and that the more rigid ordered domains are dispersed at a nanometer scale within the less rigid amorphous phase. The nanoscale morphology allows the two
phases to share energy at the molecular level and behave
in bulk as a homogeneous material (e.g. [16]). Energy sharing at a nanoscale is a key aspect in natural materials, and
has been discussed quantitatively for bone, which consists
of organic and inorganic layers to form a rigid and tough
natural hybrid nanocomposite [17].

201

A detailed analysis of the component groups in the


primary structure of both Spidroin I and II suggests that
poly(alanine) can be taken as a hypothetical model polymer with ensemble average parameter values that represent the chemical composition of Spidroin [18]; the calculation of model parameters for Spidroin I is given in
the appendix to illustrate the use of ensemble averaged
parameter values.

The alanine segment consists of a highly polar amide


group and a non-polar hydrocarbon group, such that the
interactions at a molecular level are most likely to be between like groups of atoms.
The model poly(alanine) exists in ordered or disordered states with fractions ford and fdis respectively. The
main dierence here between the ordered and disordered
structures is the number of hydrogen bonds per segment.
The ordered segments have two hydrogen bonds per segment due to favourable alignment of adjacent chain backbones. Disorder reduces the inter-chain bonding to one
hydrogen bond per segment due to unfavourable alignment of backbone chains. However, both ordered and disordered phases can be oriented in the bre axis, with the
disordered phase simply having more non-minimum energy conformers.
The polymer group structure is quantied by a limited
number of model parameters that allow the main energy
terms at a molecular level to be calculated for use in the
potential function. A set of parameters for poly(alanine)
have been calculated using a combination of group additivity tables [15,19], connectivity indices [20], and molecular modelling [21,22]. The dierence between ordered and
disordered phases is only in the cohesive energy, using
10 kJ/mol per hydrogen bond, and the overall value of
cohesive energy for any silk is simply the sum of the ordered and disordered fractions. The parameter N is the
number of degrees of freedom per group and plays a key
role in the energy dissipation model. The value of N is
given for the glassy or crystal state of matter as skeletal
vibrations perpendicular to the chain axis, but N increases
by 50% above the glass or crystal melt transition as the
polymer becomes mobile in the chain axis, which is reected in the change in heat capacity through the glass
transition [19,23]
Cohesive energy
Van der Waals volume
Molecular weight
Degrees of freedom
Debye temperature

Co = 52 kJ/mol ordered or
42 kJ/mol disordered
Vw = 40 cc/mol
M = 72
N =8
= 400 K

The Lennard-Jones potential function for the binding energy, C, per group for a strongly bonded onedimensional polymer chain in a weak three-dimensionally

202

The European Physical Journal E

bonded lattice can be written in terms of the thermal energy of skeletal mode vibrations, HT , and the congurational energy, Hc , by
C = Co + Hc + HT = Co + Hc +
 
 3 
6
Vo
Vo
= Co
2
V
V

N R T D1 (T )
4
(1)

where Vo = 1.26 Vw is the volume at the minimum of the


global potential well, R is the gas constant, and D1 (T )
is the one-dimensional Debye function consistent with the
structural model for the potential function [24,25]. The
congurational energy dierentiates between ordered and
disordered states of matter and takes fractional values of
the cohesive energy of 0.04 Co and 0.106 Co respectively;
greater order has fewer excited conformer states, and allows the dierence between crystal and amorphous states
of matter to be quantied.
The most direct predictions from the potential function are volume, V , and the bulk elastic modulus, B, as a
function of temperature. Volume as a function of temperature can be obtained by solving the potential function,
and gives density values in the range 1.2 to 1.3 g/cc. Bulk
modulus is particularly important here as a reference parameter, and is given by
B = V

d2 C
C
= 18
2
dV
V

(2)

which shows that elastic modulus is determined directly


by the cohesive energy density and supports the rst
premise of this model.

3 Elastic modulus and loss


The results of our Dynamic Mechanical Thermal Analysis of a single bre of dragline silk are shown in Figure 2.
They are consistent with the observation for polymers in
general that the steepest negative gradient in elastic modulus is associated with peaks of loss tangent due to relaxation processes at a molecular level [26]. Qualitatively,
elastic modulus is a measure of the elastic energy stored
in a material, so greater energy dissipation or loss must
be reected in a lower value of elastic modulus. Bondi
pointed out that the empirical proportionality between
loss tangent and the temperature gradient of elastic modulus could be used to predict elastic modulus if the mechanism for mechanical energy transformation to heat could
be identied [27]. Porter derived a general expression for
this proportionality in terms of the structural model parameters and the chain length per group, L 0.3 nm,
with A as the proportionality constant [15]. The derivation relates the change in thermal energy in the polymer to
the mechanical energy input during mechanical deformation via its eect on the theta temperature in the Tarasov
form of the Debye theory for heat capacity; loss tangent

is then the fraction of mechanical energy converted irreversibly to heat. The most appropriate form of the model
expression is


1.5 105 L B dE
B dE
=
.
(3)
tan A
E dT
M
E dT
The parameter A takes a general value of about unity in
units of K/GPa, and specically for poly(alanine) A
1.5 K/GPa. This expression can be inverted to allow elastic modulus to be calculated from the loss prole of a polymer, which is easier to predict from the chemical structure
using models for relaxation mechanisms. Note that bulk
modulus is the reference value of elastic modulus before
any relaxation mechanisms allow transient spatial redistribution of groups of atoms under strain to reduce the
value of modulus
 

tan dT
E = B exp
.
(4)
AB
The loss tangent peaks of the silk in Figure 2 are due to relaxation events in the dierent atomic groups of the polymer chain. In particular, the peaks below about 300 C
are local phase transition events in the disordered fraction.
The peak at about 70 C is conventionally attributed to
hydrocarbon interactions [26], and the higher temperature
peak at about 200 C is shown below to be attributed to
amide segment interactions. The temperature of the peaks
can be predicted as local glass transition temperature, Tg ,
events in terms of model parameters for the specic group
interactions by the relation [15]
Tg = 0.224 + 0.0513

Co
.
N

(5)

The model expression for the glass transition is derived


using the Born criterion for a transition, from the energy
at which the force between chains is a maximum using the
Lennard-Jones potential function, and has been validated
for a dataset of about 250 polymers. The expression shows
that stier chains with a higher skeletal theta temperature
and higher cohesive binding energy give a higher glass
transition temperature, but that greater thermal energy
through more degrees of freedom reduces Tg .
The predicted transition temperatures are given in Table 1 and are in good general agreement with the peak positions in Figure 2. Note that the extra degrees of freedom
equal to fam , in the amide group are due to the extra degrees of freedom developed in the alanine segment above
the hydrocarbon transition temperature and the value of
Tg = 479 K is for the silk used in this section as an example for predicting mechanical properties with ford = 0.66.
Each loss tangent peak is modelled as a Gaussian distribution of transition events over a distribution of temperatures with a standard deviation, s, that is broader for
a greater range of dierent structural group interactions;
Figure 2 suggests a value of s 30 degrees. The total area
under each Gaussian peak n is called the cumulative loss
tangent, tann , such that the area up to a temperature T
is expressed in terms of the activated fraction of groups up

D. Porter et al.: Mechanical properties of spider silk

203

Table 1. Segment interaction parameters and properties.


Segment

Co (J/mol)

Tg (K/ C)

tan

Fraction

Hydrocarbon

9000

205/68

19

0.5

Amide: 1 H-bond

33000

4 + fdis

479/206

68

0.5

Fig. 3. Predicted dynamic mechanical properties for a model silk with ford = 0.66, to be compared with the experimental plots
in Figure 2.

to T , fn , in the disordered fraction, fdis , and the model


parameters in the GIM expression for tan

Con
tan n dT = tan n fn fam = 0.0085
fn fam . (6)
Nn
The values of tann are given in Table 1 for the hydrocarbon and amide interactions. The relation for tan =
0.0085 Co /N is derived again from the irreversible change
in thermal energy due to new degrees of freedom, N/2,
that are invoked by forcing the structural groups through
a transient glass to rubber phase transition by mechanical
energy of deformation. The fractions of the hydrocarbon
and amide interactions are their volume fractions, which
gives the correct value of tan = 45 for the model alanine
segment overall.
Figure 3 uses equations (2), (4), and (6) to reproduce
the loss tangent and the bulk and tensile elastic moduli
of a model silk with an ordered fraction ford = 0.66. Two
curves are shown for tensile modulus, which are the limiting upper and lower predicted values. The upper value
assumes that the activated hydrocarbon groups are pinned
by the amide groups and does not allow energy dissipation until the model alanine segment as a whole becomes
mobile. The lower value assumes spatial redistribution to
occur; the following discussion uses the upper curve. Note
that the dynamic mechanical property calculations stop
at the point where observed modulus starts to rise at high
temperatures, since this marks the onset of thermally induced crystallisation in the sample, with a loss peak developing with the increasing modulus in Figure 2.
The value of ordered fraction used in the predictions
for Figure 3 was chosen ad-hoc to make the initially best
match with the measurements shown in Figure 2 for a
dragline silk. By changing the single variable of the frac-

tion of the ordered phase, we propose that the model can


reproduce the full range of mechanical properties of spider silks. Qualitatively, a greater degree of order has fewer
groups that undergo phase changes to dissipate energy,
and is thereby stier and stronger. For example, a high
proline content in a silk such as agelliform is less sti
and more extensible due to the disorder induced by the
proline. The next step in the model is to translate the dynamic mechanical property predictions into engineering
stress-strain proles over the range of possible values for
ordered fraction.

4 Stress-strain profile
Energy in the potential function used to predict thermomechanical properties in silk is dependent upon the dimensions occupied by the interacting groups of atoms.
This can be used to make a transformation of the temperature dependent dynamic mechanical properties to predictions of stress-strain curves to break. To do this, temperature, T , is used as a dummy variable to calculate
consistent pairs of values of strain, , and tensile stress, ,
using the linear coecient of thermal expansion, , that
is predicted in GIM by the relation derived by dierentiating the potential function of equation (1) with respect
to temperature
0.45

N
0.0001 K1
Co

T
=

T
dT

(7)

E dT .

(8)

Five key parameters are needed to map the overall stressstrain proles that are possible with spider silk: initial

204

The European Physical Journal E

(a)

(b)

Fig. 4. (a) Predicted stress-strain proles of silks with dierent degrees of order, ford . (b) Experimental stress-strain curves
for silks within the predicted envelope of properties shown as a dashed line: silks curves identied by markers at break point:
 various dragline silks (4),  Bombyx mori drawn evenly at dierent rates [3, 4], Argiope trifasciata dragline silk [31],
 Laboratory spun recombinant silk based upon Spidroin II [32].

modulus, Ei , yield strain, y , yield stress, y , post-yield


modulus, Ey , and strain to break, b . These parameters
are shown in Figure 1.
Taking the ambient temperature to be 20 C, the yield
strain can be estimated by associating the yield point
with maximum rate of reduction in modulus through the
peak in loss tangent of the amide groups in the disordered
phase, at a temperature about 200 degrees above ambient; yield is thereby considered to be a mechanical form
of the glass transition condition, where a transient glass
to rubber phase change occurs in the disordered fraction
to dissipate elastic energy. This suggests a yield strain of
about 2%. The yield stress is taken as a rst approximation to be the product i Ei .
The initial and post yield moduli are the values of E at
ambient temperature and at the upper end of the amide
loss tangent peak at about 300 C, and are calculated
using appropriate values of B and cumulative loss values
in equation (4) after each of the two relaxation peaks;
namely, tan values of 19 and 44 respectively. The upper
and lower limiting values of Ei for totally ordered and
disordered silks are 14 and 2.6 GPa respectively, which
is in good general agreement with observation. The postyield modulus has upper and lower values of 12 GPa and
the rubberlike plateau modulus of about 10 MPa. Note
that work hardening of the polymer above yield is not
included in this rst presentation of the model, but Ey
increases with strain as rubberlike states are transformed
back to glassy states.
Finally, the strain and stress to break is calculated
from the mechanical energy absorbed in the post-yield deformation, shown as regions 2 and 3 in Figure 1. In this
region, mechanical energy transforms rubberlike states of
matter back into glassy states as the polymer chains are
forced closer together by the tensile strain, thus removing the extra skeletal degrees of freedom above the yield
or glass transition condition, Ng = 0.5 N. The strain
to break is calculated from the equivalence of mechanical

and thermal energy density using as a rst approximation


y (b y ) +

Ey (b y ) Ng R T D1 (T )
=
2
V
1.2 108 J/m3 . (9)

The left hand side of equation (9) is the mechanical energy density as the sum of plastic and elastic terms, which
are regions 2 and 3 in Figure 1. The right hand side of
equation (9) is the change in energy due to the change in
degrees of freedom through the phase change, and its numerical value is a direct quantitative measure of the toughness of a polymer as the area under a stress-strain curve
to break. For example, synthetic engineering polymers
poly(carbonate) and poly(ether ether ketone) (PEEK)
have values 0.5 108 and 0.4 108 J/m3 respectively, and
are thus far less tough than spider silk by a factor of more
than two. Polymers such as poly(ethylene) and Nylon may
have comparable model measures of toughness, but they
require a high degree of molecular orientation to achieve
the stiness of dragline silk and do not have the higher
temperature stability imparted by the hydrogen bonding
in the high amide group concentration in dry silk.
For ease of calculation, the Debye function takes a
value of 0.7 for the model = 400 K and the volume
is taken to be V 1.5Vw = 60 cc/mol. Figure 4a shows
the full range of stress strain proles predicted for the
model silk at dierent values of the single model variable ford , and all other parameters are calculated from
ford using the structure-property relations derived above.
Experimental plots of stress-strain for some typical silk
types are plotted in Figure 4b for comparison with predictions, with each curve labelled or explained in the gure
caption. The overall limiting envelope of predicted stressstrain proles has the correct general form, and future
work will explore the detailed bounds of the failure envelope due to energy absorption mechanisms post-yield,
such as the formation of new hydrogen bonds in the more
oriented strained structure.

D. Porter et al.: Mechanical properties of spider silk

205

Table A.1. Calculation of the average group parameters for Spidroin I.


Vw
(cc/mol)
29

Co
(kJ/mol)
24.3

56

16
10
3
1
2
1
1

0
4.5
28.8
35.8
18
45
10.8

0
10.2
36.1
58.5
40.8
52.8
15.2

0
15
72
107
57
101
31

0
2
5
4
4
7
3

7.7
32

12
41

12
68

2
8

Group

Structure

Number

Peptide base
-R groups
Glycine (G)
Alanine (A)
Glutamine (Q)
Tyrosine (Y)
Leucine (L)
Arginine (R)
Serine (S)

- CCONH -H
-CH3
-CH2 CH2 CONH2
-CH2 PhOH
-CH2 C(CH3 )2
-CH2 CH2 CH2 NHC(NH2 )2
-CH2 OH

Average -R
Total average

5 Conclusions and discussion


Mean eld theory for polymers has been used to predict structure-property relations for the mechanical properties of spider silk in terms of chemical composition
and morphological structure. The chemical composition
of Spidroin proteins is embodied in the model parameters
of poly(alanine), whose morphology is quantied by the
single variable parameter of the fraction of ordered phase.
The dynamic mechanical properties of silk are derived in
terms of the ordered fraction, and are then transformed
into predictions for the whole range of stress-strain proles that are possible in silk. The predicted envelope of
silk properties is in good quantitative agreement with experimental observation. Hence the surprising outcome of
our approach was the demonstration that spider silk, apparently so complex in morphology, can be modelled accurately with relatively few and rather robust assumptions.
Also, a key to the silks strength clearly lies in the peculiar
molecular and nano-scale structure of its morphology.
The initial premise of this work assumed that stiness
and strength, on the one hand, are due to the high cohesive
energy density of hydrogen bonding. Toughness, on the
other hand, is due to the high energy absorption during
post-yield deformation. The model shows both premises to
form a useful basis for understanding the structural origins
of the properties of spider silk: but not just spider silk. Our
model is fully portable not only to other bio-polymers but
also to the much less complex non-biological, man-made
engineering polymers such as poly(carbonate) and PEEK.
Thus, with a curious twist, insights into silk, which is the
oldest commercial high-strength bre, will in the future
enable us to better understand its modern (although still
much less sophisticated) man-made derivatives.
Although the main variable parameter of degree of
order may appear vague and no assumptions have been
made as to the morphology associated with order, it does
force us to reconsider the rather narrow description of
crystal and amorphous states in a protein. Ongoing work
is looking at how processing the silk (rate, temperature
and chemistry [28]) and cyclical loading of silk bres [29]
changes the observed degree of order. Group Interaction

Modelling was originally formulated to develop structureproperty relations for polymer processing rheology and
to include such non-Newtonian eects as shear-thinning
and strain-induced crystallisation due to macromolecular
strain [30]. Thus, we expect that the model presented here
for silk properties will provide a valuable direct link to help
understand the production of silk bres with controlled
mechanical properties.
For funding we thank the British EPSRC (grant
GR/NO1538/01) and BBSRC (S12778) as well as the
European Commission (grant G5RD-CT-2002-00738), the
National Natural Science Foundation of China (NSFC,
No. 20244005) and the AFSOR of the United States of
America (grant F49620-03-1-0111).

Appendix A: Model parameters for spidroin


Spidroin I has the repeat sequence:

The structure of each peptide segment is dierent in the


R group, such that highly irregular sequences of side
chains are attached to a very regular chain structure.

The detailed GIM model parameters for Spidroin I are


tabulated above (Tab. A.1) and then averaged to show
that the ensemble average parameter set can be approximated very well by those of poly(alanine), except the cohesive energy, which is take to be the average of all peptide

206

The European Physical Journal E

segments. In addition, the cohesive energy must also be


supplemented by hydrogen bonding energy, at 10 kJ/mol
per hydrogen bond. Thus, the disordered and ordered fractions have +10 and +20 kJ/mol extra due to hydrogen
bonds respectively. Hydrogen bonding between side chain
groups has been ignored at this stage, but would tend to
increase the average cohesive energy value by a maximum
of 3 kJ/mol.
The Debye theta temperature is calculated from the
average molecular weight per segment to be 1 = 400 K,
and the average length per segment in the chain is taken
to be L 0.3 nm.

11.
12.
13.
14.
15.
16.
17.
18.
19.
20.

References
1. D. Fox, New Scientist, 24 April 1999, pp. 3841
2. F. Vollrath, Reviews in Molecular Biotechnology 74, 67
(2000)
3. Z. Shao, F. Vollrath, Nature 418, 741 (2002)
4. F. Vollrath, D. Knight, in Handbook of Biopolymers, edited
by A. Steinb
uchel, S. Fahnestock (Wiley-VCH, Heidelberg
and New York, 2003), Chap. 2, p. 25
5. Z. Shao, Y. Yang, X. Chen, P. Zhou, D. Knight, D. Porter,
F. Vollrath, Spider silk performs as much tougher material
at low temperatures, Adv. Mater. (in press)
6. D.T. Grubb, L.W. Jelinski, Macromolecules 30, 2860
(1997)
7. J.D. van Beek et al., PNAS 99, 10266 (2002)
8. B.L. Thiel et al., Biopolymers 34, 1089 (1994); B.L. Thiel,
Biopolymers 41, 703 (1997)
9. J.M. Gosline et al., in Silk Polymers: Materials Science and
Biotechnology, edited by Kaplan et al., ACS Symposium
Series 544 (ACS Press, New York, 1994), Chap. 27, p. 328
10. M.A. Becker et al., in Silk Polymers: Materials Science and
Biotechnology, edited by Kaplan et al., ACS Symposium
Series 544 (ACS Press, New York, 1994), Chap. 17, p. 185

21.
22.

23.

24.
25.
26.

27.

28.
29.
30.
31.
32.

Y. Termonia, Macromolecules 27, 7378 (1994)


M.J. Forster, Micron 33, 365 (2002)
J.J.M. Baltussen, M.G. Northolt, Polymer 40, 6113 (1999)
A. Galeski, Prog. Polym. Sci. 28, 1643 (2003)
D. Porter, Group Interaction Modelling of Polymer
Properties (Marcel Dekker, New York, 1995)
M.J. Buehler, F.F. Abraham, H. Gao, Nature 426, 141
(2003)
D. Porter, Mat. Sci. Eng. A 365, 38 (2004)
M. Xu, R.V. Lewis, Proc. Natl. Acad. Sci. USA 87, 7120
(1990)
D.W. van Krevelen, Properties of Polymers (Elsevier,
Amsterdam, 1993)
J. Bicerano, Prediction of Polymer Properties (Marcel
Dekker, New York, 1993)
J. Rossmeisl et al., J. Chem. Phys. 118, 9783 (2003)
Molecular mechanics and dynamics simulations performed on the Cerius2 system of Accelrys Inc.: see
http://www.Accelrys.com
B.
Wunderlich,
S.Z.D.
Cheng,
K.
Loufakis,
Thermodynamic Properties in Encyclopedia of Polymer
Science and Engineering, Vol. 16 (Wiley-Interscience,
New York, 1989)
H.S. Bu, S.Z.D. Cheng, B. Wunderlich, J. Phys. Chem. 91,
4179 (1987)
V.V. Tarasov, G.A. Yunitskill, Russian J. Phys. Chem. 39,
1109 (1965)
N.G. McCrum, B.E. Read, G. Williams, Anelastic and
Dielectric Eects in Polymeric Solids (John Wiley and
Sons, London, 1967)
A. Bondi, Physical Properties of Molecular Crystals,
Liquids, and Glasses (John Wiley and Sons Inc., New
York, 1969), p. 401
F. Vollrath, Proc. R. Soc. Lond. B 268, 2339 (2001)
Z. Shao, F. Vollrath, Polymer 40, 1799 (1999)
D. Porter, J. Non-Newtonian Fluid Mech. 68, 141 (1997)
J. Perez-Rigueiro et al., J. Appl. Polym. Sci. 82, 2245
(2001)
A. Lazaris et al., Science 295, 472 (2002)

You might also like