Professional Documents
Culture Documents
solidi
physica
Phys. Status Solidi B, 1–20 (2013) / DOI 10.1002/pssb.201248550
status
www.pss-b.com
basic solid state physics
National Centre of Advanced Tribology at Southampton (nCATS), University of Southampton, University Road, Southampton,
SO17 1BJ, UK
Auxetics are a modern class of material fabricated by altering particular emphasis to the auxetic foams, due to their low price,
the material microstructure. Unlike conventional materials, easy availability and desirable mechanical properties. Key
auxetics exhibit a negative Poisson’s ratio when subjected to areas discussed include the fabrication method, the effects
tensile loading. These materials have gained popularity within played by different parameters (temperature, heating time, cell
the research community because of their enhanced properties, shape and size and volumetric compression ratio), micro-
such as density, stiffness, fracture toughness and dampening. structural models, mechanical properties and potential appli-
This paper provides a critical oversight of the auxetic field with cations.
1 Introduction Poisson’s ratio (n) is defined as the [11, 12]. These materials demonstrate unique and enhanced
ratio between the longitudinal expansion and the lateral mechanical properties and for this reason numerous
contraction of a material during loading [1]. Conventional researches have been carried out to understand the
materials present a positive Poisson’s ratio and their cross- mechanisms that render a material auxetic and to reproduce
section becomes larger in compression and smaller in tension these mechanisms and properties in man-made materials.
(Fig. 1a) [2, 3]. Thermodynamic considerations of strain To date a wide range of auxetic materials, such as polymers,
energy in the theory of elasticity demonstrate that the metals, ceramics, composites, laminates and fibres have
Poisson’s ratio for a homogeneous solid isotropic material been manufactured (see Fig. 2a) using a particular fabrica-
could be between 1 and 0.5, thus allowing the existence of tion process that results in a change of the material structure
materials with a negative Poisson’s ratio [1, 4–6]. This class [11–14]. A timeline for the discovery and the areas in which
of materials is identified with the term ‘auxetics’, that derives man-made and natural auxetic materials are available (with
from the Greek word ‘auxetikos’ which means ‘that tends to their length-scale) are reported in Fig. 2a and b, respectively.
increase’ and Poisson’s ratios are as low as 0.7 for The increased interest in the research and applications of
polymers and 0.8 for metals [7, 8]. Auxetic materials are auxetic materials is demonstrated by the increased number of
characterised by a counterintuitive behaviour, which is patent filed (Fig. 3a) and research paper published (Fig. 3b)
evident by applying a tensile load in one direction they since the late 1980s. Auxetic materials can be currently
expand in all directions (Fig. 1b) [9, 10], or more simply they found in commercially available products such as polytetra-
become fatter, laterally, when stretched lengthwise. fluorethylene (PTFE) and GoreTex [15]; since their initial
Auxetic materials constitute a new class of materials proposal by Love, numerous organisations including
that can be found in nature, i.e. cubic elemental metals, Toyota, Yamaha, Mitsubishi, AlliedSignal Inc., BNFL and
a-cristobalite (high temperature polymorphic mineral), the US Office of Naval Research have filed numerous patents
and biological tissues, i.e. cat skin and cow teat skin [15, 16].
solidi
physica
status
2 R. Critchley et al.: Manufacture, mechanical properties and applications of auxetic foams
Figure 1 Differences between (a) a positive and (b) a negative Poisson’s ratio material (adapted from Ref. [11]).
Several reviews on auxetic materials have been pub- ratio (the change in volume from the sample original
lished dedicating particular attention to known auxetic dimensions to the new dimensions obtained after the
materials, different types of auxetic microstructures and their conversion expressed as a ratio) could differ up to 52%
applications [11, 13, 17–21]. The aim of this paper is to from the originally applied value while the sample density
provide critical overview of auxetic foams, their processing was susceptible to a time-dependant change of up to 30%
methods, deformation mechanisms and applications. Auxe- after a week. This behaviour is most likely the result of creep,
tic foams have been chosen from all the various auxetic which occurs as a consequence of the long-term exposure to
materials since the auxetic effect and the mechanisms stresses and not sufficient heating times employed during
involved have been widely studied using open cell the fabrication process. The surface creasing and wrinkles
polyurethane foam; their low price, easy availability and observed in the more deformable regions of the foam
fabrication method further increase their attractiveness as specimen are due to the volumetric compression ratio
materials. The mechanical properties and the possible applied during fabrication [24, 25]. In order to resolve this
applications of auxetic foams have also been reviewed, problem, two solutions have been suggested: lubrication of
while new areas of study that have yet to be explored are the mould (the lubricant should not be an oil-derivative or a
proposed. distilled oil due to their instability at high temperatures and
the production of unpleasant smells) [6, 24, 26] and the use of
2 Methods to convert conventional foams into wires or tweezers inside the moulds to pull the foam instead
auxetic The manufacturing of auxetic foams was first of pushing it [10]. A further solution would be to redesign
reported in 1987 by Lakes [3] (although the concept of the moulds so that the tri-axial compression is applied at a
auxetic foams were proposed in 1985 by Kolpakov) [23], uniform rate and no creasing areas form. Another solution
who suggested two different manufacturing procedures: one has been reported by Chan and Evans in 1997 [24], they
for polymeric and one for metallic foams. attempted to resolve the formation of wrinkles on the sample
surface by applying the volumetric compression in several
2.1 Methods for the conversion of polymeric stages in order to produce a more homogeneous auxetic
foams For polymeric materials, Lakes started with a material.
conventional open-cell foam and made it auxetic by making Since 1987, this production process has been applied by
the ribs of each cell protrude inward producing a re-entrant numerous investigations and a number of modifications have
structure [3]. This change is achieved in a three-step process been reported [2–5, 10, 24–32]. While each has been altered
in which the open cell foam is tri-axially compressed, heated either to produce differences within the polymeric materials
to a temperature slightly higher than the softening tempera- or to optimise the process the overall principle has remained
ture of the material and cooled at room temperature to store the same, i.e. volumetric compression followed by heating
potential energy in the compressed ribs [3, 13, 24]. The and cooling [4, 24, 31]. These modifications are described
compression stage of the process changes the shape of below.
the cellular ribs and the heating and cooling stages soften
and fix the ribs in a new position. 2.2 Multi-phase auxetic fabrication Bianchi et al.
The main problems that have been identified as [5] further modified the fabrication process by incorporating
dependent on the auxetic conversion process are: long-term a re-conversion back to conventional foam via a shape
instability with the samples reverting back to their original memory polymer (SMP) process followed by a second
shape and structure, severe surface creasing and inability to auxetic conversion. Auxetic materials can be considered as
produce large samples. For example Bianchi et al. [5] noticed an SMP even if the fabrication process is usually intended as
that following the auxetic conversion, within hours, the a permanent structural change. Indeed it has been observed
samples would naturally attempt to return to their original that the application of an external trigger, such as exposure
dimensions. This behaviour was also reported by Scarpa to a solvent or heat, reverts the auxetic foam back to its
et al. [10], who noticed that the final volumetric compression conventional state and dimensions. The re-conversion
Figure 2 (a) Timeline for the discovery of natural and man-made auxetic materials, (b) areas where auxetic materials have been discovered
and their size (adapted from Ref. [11]).
solidi
physica
status
4 R. Critchley et al.: Manufacture, mechanical properties and applications of auxetic foams
suggested to be directly caused by the non-homogenous followed by numerous studies attempting to determine the
temperature distribution in the foam body, where only methodology and the justification behind this choice of
the higher temperatures penetrate the external regions of the temperature. In 1997, Chan and Evans employed a
sample, and not the interior. Densities were also found to thermocouple placed in the centre of a foam specimen to
be affected based upon the time a sample spent waiting to determine the softening temperature of the material when the
be removed, with those waiting longer exhibiting higher cell ribs collapsed [24]. This experiment suggested that the
densities. This feature is likely to be the result of the samples temperature employed for the auxetic conversion should
being tri-axially constrained whilst cooling, thus stopping be between 5 and 20 8C lower than the softening temperature
the ribs from protruding towards their original structure. of the material. Conversely Evans and Alderson [2, 11],
Samples fabricated through this novel route were also found supported by Friis et al. [2] employed a heating temperature
to only exhibit a negative Poisson’s ratio under compression above the softening point and Wang et al. [32] suggested that
due to the non-homogenous microstructure caused in the conversion temperature should be equal or greater than
manufacture. The greatest negative Poisson’s ratio achieved the softening temperature of the material. From the results
in this instance was 0.34. reported, it is reasonable to assume that Chan and Evans’
suggestion of keeping the conversion temperature below the
2.3 Methods for conversion of metallic materials softening temperature (e.g. 180 8C) was incorrect
foams For metallic materials, Lakes stated that reticulated and instead the temperature employed should in fact be equal
metal foam specimens were transformable into auxetics by or above the softening temperature of the conventional
compressing the material successively in each of the three polymeric foam, as this allows for the softening of the
perpendicular directions at room temperature (18–23 8C). polymeric ribs to occur [24].
The foams subjected to this procedure exhibited a re-entrant Whilst attempting to determine the reasons behind the
cell structure [3]. This methodology has also been applied by temperature values published by Lakes [3], Chan and Evans
other authors, with little to no known changes being reported several interesting observations [24]. They
introduced [2, 36]. However, it has been suggested by Friis observed that if the heating time was too long the foam
et al. [2] that, in principle, the thermal transformation would either decompose or melt with the cell ribs sticking
technique used for thermoplastic foams could be applied also together to form a dense block of material, and if the heating
to metallic foams, although this may be difficult due to the time was too short the foam could not be ‘set’ into its new re-
higher and sharper melting points of metals. entrant structure and would soon begin to expand back to its
original size [24]. Choi and Lakes and Bianchi et al. [4, 29]
3 Parameters affecting manufacture of auxetic further confirmed this behaviour with the rate at which the
foams From the literature [4, 5, 28, 29, 32], it is clear that negative Poisson’s ratio was lost fluctuating from a few days
there are many variables influencing the manufacture of to a few months. Furthermore, Chan and Evans noticed that
auxetic foams, such as the composition of the material, its the heating time needed to turn a conventional porous
relative density and cell size, the processing temperature and material into an auxetic one depended on the exact type of
heating-time, the humidity and the volumetric compression foam and hence initial base material and porosity [24]. Wang
ratio applied [29]. There is still much debate regarding which et al. confirmed these observations and further suggested that
are the key factors influencing the process. For example the cell size was also an important factor influencing the
Wang et al. [32] stated that the main physical parameters auxetic fabrication. It was found that polyurethane foams
influencing the auxetic transformation process are the having a smaller cell size required higher heating tempera-
volumetric compression ratio, the processing temperature tures and heating times compared to foams with larger pore
and the heating time. However, other authors [4, 5, 28] size. It was speculated that this behaviour could be due to
consider only the volumetric compression ratio as the main either the surface tension effects within smaller cells or to a
parameter responsible for a successful conversion, with the possible difference in the chemistry of the material [32].
other parameters acting as secondary influence. Although Other notable effects of the temperature on the auxetic
an exact relationship between the effects that each of the conversion process have been reported when studying the
parameters play in the transformation is still unknown, the early stages of heating and cooling. During the early stages of
influences that the different process parameters have on heating at temperatures of 100–135 8C, Bianchi et al. [28]
the final results are reviewed below. There have been also noticed using thermogravimetric analysis that conventional
attempts to identify the relationships between the manufac- foams experience an approximate 2% loss in weight. It is
turing parameters and the final material properties (density, suggested that the loss in weight could potentially explain the
Poisson’s ratio, stiffness and energy dissipation) [4]. difference in mechanical properties between conventional
and auxetic materials during the multi-phase conversion
3.1 Influence of temperature and heating (phase 1: conventional, phase 2: first auxetic, phase 3:
times Lakes reported that for a polyester open cell foam returned conventional and phase 4: second auxetic). During
the ideal temperature to be applied in the conversion process cooling, numerous studies have shown that the specimens
was between 163 and 171 8C, just above the softening should be cooled at room temperature, as set out by Lakes [3,
temperature of the material [3]. This statement was then 24, 29]. In 2008, Bianchi et al. [4] introduced an alternative
solidi
physica
status
6 R. Critchley et al.: Manufacture, mechanical properties and applications of auxetic foams
Chan and Evans 1997 closed-cell polyester urethane foam 60 39.9 [24]
Chan and Evans 1997 reticulated polyester urethane foam 60 33.7 [24]
Chan and Evans 1997 open-cell polyether urethane foam 10 24.1 [24]
Chan and Evans 1997 open-cell polyether urethane foam 30 24.5 [24]
Chan and Evans 1997 open-cell polyether urethane foam 60 21.7 [24]
Bianchi et al. 2008 conventional grey open-cell polyurethane 30–35 27.2 [4]
Scarpa et al. 2002 open-cell polyurethane foam not given 32 [26]
Wang et al. 2001 open-cell polyurethane foam 20 30 [32]
Wang et al. 2001 open-cell polyurethane foam 65 30 [32]
Wang et al. 2001 open-cell polyurethane foam 100 33 [32]
Scarpa et al. 2005 conventional grey open-cells polyurethane 30–35 27 [10]
Bianchi et al. 2010 grey-coloured open-cell polyurethane-based foam 30–35 27.2 [5]
Bianchi et al. 2010 conventional light-blue-coloured open-cell polyurethane-based foams 52–57 27 [5]
Bianchi et al. 2010 grey-coloured open-cell polyurethane-based foams 30–35 27.2 [28]
Bianchi et al. 2010 conventional light-blue-coloured open-cell polyurethane-based foams 52–57 27 [28]
Scarpa et al. 2004 open cell polyurethane grey foam 30–35 32 [27]
One of the more comprehensive studies on the effect that bow-tie re-entrant microstructure (Fig. 6). Both models for
cell shape and size have on the properties of the foam was the auxetic and conventional pores are described in more
reported by Chan and Evans [37]. They specifically studied detail below.
the microstructure and the deformation mechanisms of
conventional and auxetic open-cell foams and observed that 3.3 Conventional model For conventional foams
the cell size was not the dominating factor in determining the reported in Fig. 5, the cell length can be calculated using the
mechanical properties, while the cell shape and geometry following equations [37]:
were found to be greater contributor, as also reported in Ref.
[38]. The cell geometry is influenced by the manufacture Maximum: AD ¼ h þ 4L sinu; (1)
process of the conventional foam that is formed by injecting Minimum: BC ¼ h þ 2L sinu; (2)
gas in the semi-liquid material, the gas dissipates throughout
the material then the liquid is cooled to the solid state Average: y ¼ h þ 3L sinu; (3)
retaining the air pockets formed by the gas. The air pockets
are somewhat elongated and therefore the foam could be and the cell width at any point can be taken as:
anisotropic [37, 39]. By assuming that the mechanical Normal: EF ¼ 2L cosu; (4)
properties of the foams are similar to those of the solid
material employed for their production, the main variable Average: x ¼ 2L cosu: (5)
affecting the foam mechanical properties is its cellular
geometry [37]. It should be noted that although the foams can
be treated similarly to their solid counterparts, they are 3.4 Auxetic model For auxetic foams the cell length
considerably less stiff [1, 7, 40]. can be given by the following equations (all the parameters
Chan and Evans studied and modelled the cell are specifically related to Fig. 6):
geometries and volumes in two and three dimensions for Maximum: GE ¼ h; (6)
conventional (see Fig. 5) and auxetic foams (see Fig. 6),
where nearly 100 specimens were measured to achieve a Minimum: BC ¼ h 2L sinu; (7)
sufficient statistical analysis of the data set [37]. They
observed that once the conventional foam has been Average: H ¼ h L sinu; (8)
converted to an auxetic structure the cell geometry no longer
the auxetic cell width can be obtained by:
resembles that of an ellipsoid with the smaller side within the
direction of rise (Fig. 5), but becomes more reminiscent of Normal: EF ¼ 2L cosu; (9)
Figure 5 Schematic representation of (a) a three-dimensional (3-D) geometry for a conventional foam and (b) a two-dimensional (2-D)
geometry of the same pore type (adapted from Ref. [37]).
solidi
physica
status
8 R. Critchley et al.: Manufacture, mechanical properties and applications of auxetic foams
Figure 6 Diagrammatic representation of (a) 3-D geometry of an auxetic pore and (b) 2-D geometry of the same auxetic pore (adapted from
Ref. [37]).
and the auxetic mid-plane waist height is calculated using: subjected to a screening process where samples were
manufactured at temperatures between 120 and 280 8C and
Maximum: JK ¼ h 2L sinu; (10) processing time between 5 and 21 min at a volumetric
compression ratio of approximately 2.5, yielding a total of
Minimum: AD ¼ h 4L sinu; (11) 48 samples, the data is shown in Fig. 7.
It was found that for the 100 PPI foam, the optimal
Average: W ¼ h 3L sinu: (12) processing parameters were a heating time between 8 and
12 min, a temperature between 210 and 230 8C. Whereas the
It is important to underline that although these models best conditions for the 20 and 60 PPI foam specimens were a
allow estimating the cell geometry, real samples contain heating times of 17 and 13 min and a temperature of 170 and
numerous cellular and structural anomalies. A foam speci- 190 8C, respectively. These results show that smaller cell
men contains a variety of cell shapes with four to six sides size foams prefer higher temperatures and shorter heating
and a wide range of cell sizes, some of which are so small that time.
deformation is not possible and therefore act as junction
points within the foam [37]. They also observed that the
foams tested were slightly elongated due to the foam
manufacturing process. While each of these anomalies
discredits somewhat the proposed mathematical model,
it should be underlined that the model should not be
disregarded, since it can still provide insightful information
for the cell geometry.
One area that Chan and Evans failed to address in this
study was to understand how the cellular size affected the
overall negative Poisson’s ratio of the foam [37]. Wang et al.
[32] found that for a number of polymeric foam specimens,
the lower the number of pores per inch (PPI), i.e. the larger
the cell size, then the greater the negative Poisson’s ratio.
3.6 Effect of applied volumetric compression observed by Bianchi et al. [28], where it was further noticed
ratio The volumetric compression ratio (Rc) is defined as that lower volumetric compression ratios produced greater
the ratio between the original volume (Vo) and the final tangent modulus values.
volume (Vf) after the auxetic conversion of the specimen Metallic foams are also affected by the volumetric
[24]. The volumetric compression ratio plays a critical role in compression ratio applied during the conversion process.
the auxetic conversion process and it has been observed that During a study of the non-linear properties of a copper-based
failure to apply a value within the optimum range can result foam, Choi and Lakes reported an optimal volumetric
in an unsuccessful conversion process. It has been reported compression ratio of 3.6 [36]. This value was very similar to
in the literature that different types of materials exhibit those reported for polymeric foams with comparable density.
different maximum volumetric compression ratio values. Interestingly, Choi and Lakes found that the metallic foams
For example, Lakes stated that for the open-cell polymeric were subjected to the same problems reported for polymeric
foam he employed the optimal volumetric compression ratio foams, e.g. when a volumetric compression ratio above the
was between 1.4 and 4 [3]. More recently, Choi and Lakes optimum range is applied the auxetic behaviour became
reported that in order to have a successful auxetic conversion obstruct because the ribs come into contact hindering
the volumetric compression ratio had to be between 2 and 5 the unfolding mechanism [36]. It should be noted that to
[32, 36]. Although optimum compression ratio ranges have the author’s knowledge little to no work has been published
been suggested, it has to be remembered that the optimum on how the porosity of metallic foams affects the overall
permanent compression to achieve the best negative volumetric compression ratio.
Poisson’s ratio depends on the initial foam density and
appears to have a purely geometrical origin, as the same 4 Foam structure and deformation mechanism
negative Poisson’s ratio value was observed for metallic and The cellular structure of conventional polymeric foams is
polymeric foams with the same relative density [36]. The well-known; during the auxetic conversion process the
majority of Rc values reported in the literature fell within cellular structure completely changes and therefore it is
this range [2, 7, 10, 27, 28, 41]; however, experiments with fundamental to understand the deformation mechanisms
greater volumetric compression ratios have also been involved. In order to understand the counterintuitive
reported but only with respect to mechanical testing and behaviour of auxetic foams, numerous studies have analysed
not volumetric effects [31]. how the cell ribs and the foam structure changes under
The compression ratio has also been described in terms compression and tensile loading and proposed numerous
of volumetric percentage change [10, 42], which differs from theories that will be summarised below.
the volumetric compression ratio, as no optimum percentage
range has been outlined. However, from estimates given 4.1 Bow-tie or re-entrant cell model In the ear-
in the literature, the optimal percentage range may be liest studies reported by Lakes the permanent inward
approximated as being between 30 and 94% of the original protrusion of the cell ribs during the conversion process
volume; converting these values into volumetric com- forming a ‘bow-tie’ structure was highlighted; this structure
pression ratios, it can be found that the optimal range is is also known as re-entrant cellular structure [3]. Further
between 1.43 and 16.1. While this range seems too wide to be work by Lakes confirmed that the performance of the auxetic
considered as a useful optimal range, it is in agreement with foams is a result of the rib behaviour [43]. Following these
the commonly accepted 2–5 range previously defined by observations, the re-entrant cell model shown in Fig. 8 was
Lakes [32, 36], which as a percentage equates to 50–80% of established; according to this model, the application of a
the original volume. tensile load results in deformation of the cell ribs by bending,
The volumetric compression ratio is the most important which in turn leads to unfolding of the cells to their original
and influential parameter in the auxetic manufacture [4, 5, structure. This is further demonstrated by following the rib
28, 32], in fact it has been observed to influence both the movement in the scheme reported in Fig. 8. Applying a
auxetic microstructure and the mechanical properties tensile load, points A and B expand further apart attempting
of the foam. However, the volumetric compression ratio
is also susceptible to the effects of other variables, such
as temperature and pore size. For example, Choi and Lakes
reported that when a large volumetric compression ratio is
applied together with time and temperature values outside
the optimum conditions, the ribs in the foam become stuck
together [29]. It has been observed that applying smaller
volumetric compression ratios produces samples with a very
low negative or almost positive Poisson’s ratio values [27].
However, Wang et al. [32] noted that, within the optimum
volumetric compression range of 2–5, specimens with lower
volumetric compression ratios were found to exhibit greater Figure 8 Model describing the rib behaviour under tensile loading
negative Poisson’s ratios. This behaviour was once again adapted from Lakes et al. [43].
solidi
physica
status
10 R. Critchley et al.: Manufacture, mechanical properties and applications of auxetic foams
to achieve the original hexagonal geometry of the cell, while the auxetic foam with a repeatable bow-tie structure (as
points C and D maintain a fixed position; the expansion of shown in Fig. 9b), as also proposed by Lakes [43]. In order
A-B produces an increase in volume. Lakes also observed that for this model to be valid it was assumed that the foam
the bending was not crucial, with the same effect generated by presented a state of cellular symmetry, since it is known that
employing stretchable spring elements, with no rotational experimental foams are asymmetric and still convertible to
constrains, that are allowed to pivot freely. Rothenburg et al. auxetic it could be concluded that the cellular symmetry is
[44] further proposed a similar approach in which each not required for auxetic conversion to occur [5, 45]. When
elastic element within a randomly oriented network can be tensile loading is applied in the y-direction, the hexagonal
represented as shock absorbing unit. By assuming that all the cells in the conventional foam become elongated along the
orientations of the elastic elements have equal probability stretch path and contract in the x-axis producing a positive
of occurring, deformation can be said to be similar to that of Poisson’s ratio (see Fig. 9a). By applying the same tensile
isotropic elastic materials; therefore the Poisson’s ratio is loading along the y-axis, the bow-tie cell in the auxetic foam
shown to be determined by the ratio between normal and elongate in both the direction of stretch and the perpendicular
tangential stiffness (l), where a negative Poisson’s ratio is direction. However, this behaviour is anisotropic, as loading
achieved when l > 1. in the y-axis will not yield the same results as loading in the
x-axis [37]. On the other hand, Lakes stated that although
1l
v¼ : (13) some materials experienced anisotropic behaviour not all
3þl auxetic materials are anisotropic [3, 37]. Moreover, it has
Within this arrangement, should three elastic elements been suggested that auxetic foams produced via an isotropic
form a triangular structure, a compressive or tensile load in compression ratio have a tendency to exhibit superior
one direction will cause the structure to either compress or properties compared to their anisotropic counterparts [5].
expand. Interestingly, by introducing a structural kink into The 2-D honeycomb model explained above has been
an elastic element, the axial stiffness of the collapsed bar is widely accepted by the research community for under-
greatly reduced, where the kink angle determines the stiffness standing the cellular structure and deformation mechanism
reduction. It is on this principle that the re-entrant structure of 2-D foams [18, 25, 27, 46]. However, alternative models
proposed by Lakes achieves a negative Poisson’s ratio. have been proposed to explain the auxetic behaviour, most of
which work in conjunction with the bow-tie model.
4.2 Honeycomb model In 2000, Evans and Alder-
son presented a two-dimensional (2-D) model that described 4.3 Missing rib model Smith et al. [25] proposed the
the different structures of conventional and auxetic foams missing rib foam model because, at the time of publication,
under tensile loading [11]. This model represented the they believed that the models proposed by other groups were
conventional foams with a repeatable honeycomb structure satisfactory in describing the stress–strain behaviour of
made of hexagonal geometries (as shown in Fig. 9a) and conventional foams but were limited in describing the
auxetic behaviour. In this model the cellular internal angles
are not subjected to angular changes, but a fraction of the cell
ribs are removed, as shown in Fig. 10. This model is best
employed in the calculation of strain-dependent Poisson’s
ratio function, opposed to small-strain Poisson’s ratios, as
defined by Berthelot and Reinf [47]. The advantage of this
model is that both auxetic and conventional materials can
be described using realistic geometries and stress–strain
behaviours could be predicted slightly more accurately. The
disadvantages of this model result from both the assumption
that needs to be made (real world foams contain already
broken ribs which are randomly distributed) and its inability
to accurately describe the compression/heating effects on
fabrication. From this model Smith et al. [25] suggested that
during several scenarios of sample fabrication, it is plausible
to presume that there are multiple mechanisms causing
the auxetic conversion, i.e. concaved re-entrant cells and
missing ribs.
solidi
physica
status
12 R. Critchley et al.: Manufacture, mechanical properties and applications of auxetic foams
Figure 12 A finite element sample (2.2 mm3) of an auxetic foam of 45 PPI, demonstrating the response to tensile loading conditions,
undertaken at the University of Southampton.
imaging techniques such as computed tomography (CT), a some authors have also studied or speculated the possible
more accurate model could be produced. Unlike optical deformation mechanisms in metallic foams. Friis et al. [2]
microscopy and SEM, CT is capable of producing true 3-D suggested, based on the ideal foam theory, that the main
data, not only of the surface but throughout the volume of a deformation mechanism in copper foams would be plastic
body. For auxetic foams this is extremely important, as the hinging. Microscopy studies of re-entrant foam structures
auxetic ribs deform in three-dimensions. CT is capable of demonstrated that plastic rib buckling was also present.
analysing large samples, but the achievable resolution is Successively, Choi and Lakes suggested that the behaviour
affected by both the area to be scanned and the type of of polymeric and metallic foams is attributed to the ribs of the
equipment employed. CT is a non-destructive technique and material, with both materials experiencing re-entrant cellular
therefore allows studying the changes in microstructure structures as a result of the auxetic conversion, resulting in
under various experimental conditions. CT can also be a 3-D auxetic behaviour [1, 3, 7]. It was further found for
utilised in the application of finite element analysis, an area copper foams that the ribs could form plastic hinges by
rarely studied in current literature. An example of this work yielding at a relatively small strain, unlike polymeric foam
can be seen in Fig. 12, where a real world auxetic foam which are elastomer [36].
sample of 45 PPI at a volumetric compression ratio of
4.88 was scanned, meshed and analysed in an attempt to 4.7 Granular mechanics Rothenburg et al. [44]
understand localised rib movement under tensile loading studied the behaviour of negative Poisson’s ratio by
conditions, while investigating how a simulated response assuming materials to be granular, for instance, consisting
corresponds to the model demonstrated in Fig. 13. Some of randomly packed smooth stiff spheres connected by
prelininary work using the CT technique has been reported imaginary normal and tangential springs. Within this system,
by Elliott et al. [53] and Gaspar et al. [54]. the interactions occur through frictional interfaces, resist-
ance to compression (stiffness) and force transmission along
4.6 Models for metallic foams While the majority the axis of the connecting particles centres and tangential to
of the research has been carried out on polymeric foams, the plane of contact, as per real regular granular materials. By
Figure 13 Diagram showing the key deformation mechanisms in conventional foam under compression and tension, (A) conventional
foam vertex and ribs, (B) forces acting on the ribs under compression, (C) deformation mechanism for each of the ribs during compression,
(D) forces acting on the ribs under tension, (E) deformation mechanism for each of the ribs during tension, where blue arrows represent
forces upon the ribs and dotted red lines represent previous positions of the ribs before force.
considering the material as a 2-D isotropic plane, the able to be greater than the normal stiffness resulting in
Poisson’s ratio [Eq. (14)] is once more shown to be negative Poisson’s ratio.
controlled by the ratio between normal and tangential Pasternak and Dyskin expanded the granular mechanics
stiffness, approach to produce a 3-D granular system with the inclusion
of imaginary rotational elastic springs, where the ratio between
1l normal stiffness (kn) [N m1] to shear stiffness (ks) [N m1]
v¼ : (14) was again shown to be controlling parameter in determining
4þl
Poisson’s ratio [Eq. (15)] [56]. The inclusion of rotational
freedom failed to add any additional controlling parameters.
Shufrin et al. [55] later explored the effects of negative
Poisson’s ratios in planar isotropic structures through kn ks
v¼ : (15)
reference units (Fig. 14), each comprising of an infinitely 4kn þ ks
rigid hexagonal core surrounded by six linear elastic circular
frame arches of conventional Poisson’s ratio. By tailoring Following this relationship, Pasternak and Dysken
the units through various frame and core shapes, internal proceeded to present two isotropic structures (ball link
structures and bonding types, shear stiffness is once more model and thin shell model) capable of achieving a
solidi
physica
status
14 R. Critchley et al.: Manufacture, mechanical properties and applications of auxetic foams
Figure 15 Ball link unit model where (a) unloaded and (b) tensile
loaded, adapted from Pasternak and Dysken [56].
Figure 16 Thin shell model consisting of seven hollow balls glued together: (a) before loading and (b) bi-axial compression, adapted
from Pasternak and Dysken [56].
resistance [1, 2, 7, 11, 24, 26, 58–60], the fracture toughness modulus is greater than the shear modulus. In addition, Lakes
[1, 7, 9, 11, 24, 43, 61], the compression [2, 24], the shear studied the effects of indention on a wrestling mat made
modulus [24, 43, 44, 55, 56], the stiffness [9, 10, 44, 55, 56], from elastomeric foams (n 1/3) [8]. He assumed that the
acoustic dampening [1, 7, 43, 62, 63], the dynamic indentation could be described as indentation rigidity:
performance [24], the optical passive turning of structural
P E
vibration [64] and the viscoelastic loss [27, 65–67]. ¼ 1 v2 ; (21)
The improvements in mechanical properties observed w 2a
after the auxetic conversion have led numerous researchers to where a is the radius of a circular localised pressure [mm], P
consider many possible applications for auxetic foams the localised pressure [Pa] and w is the indentation depth
ranging from smart filtration systems to protective sports [mm], as described by Timoshenko and Goodier [69]. The
equipment [1, 11]. The mechanical enhancements observed indentations experienced by the mat were further analysed
are due to the changes to the four elastic constraints of the for small and large impacts [8]. For small impacts (narrow)
material: Young’s modulus (E) [Pa], shear modulus (G) [Pa], it was assumed that the circular pressure distribution could
bulk modulus (K) [Pa] and Poisson’s ratio, which respectively be taken as an elastic half space yielding:
measure stiffness, rigidity, compressibility and volumetric
change under strain [11]. For these constraints there is not a F Gan
¼ ; (22)
discernible characteristic length scale, which implies that a u narrow ð1 vÞ
microstructural dimensions smaller than 1 mm could exhibit
where F is the indentation force [N], u and an are the
a negative Poisson’s ratio [3]. The following equations show
maximum displacement [mm] and the radius for narrow
how the elastic constraints are related to each other and why
indentation [mm], respectively. For impacts greater than the
altering a constraint will affect the others [68]:
mat thickness (wide), it is assumed that both compression
9KG and force are uniformly distributed, with the Poisson’s ratio
E¼ ; (16) effect controlled giving:
3K þ G
E Ga2w
G¼ ; (17) F 2hð1þvÞ
2ð 1 þ v Þ ¼ ; (23)
u wide ð1 2vÞ
E where h is the mat thickness measured in mm and aw is the
K¼ ; (18)
3ð1 2vÞ radius for wide indentation [mm].
From this study, Lakes suggested that the continuum
1 3K 2G theory of elasticity was probably not the best approach to
v¼ : (19)
2 3K þ G describe the auxetic behaviour and proposed using the
Cosserat theory of elasticity that takes into account both
These equations also provide understanding of the the maximum stresses and a natural length scale to allow
property enhancements exhibited by auxetic materials. The microstructural size to be incorporated into the prediction of
two main mechanical properties of auxetic foams studied in failure [8, 70, 71].
the literature have been hardness (indentation) and toughness Subsequently, Evans and Alderson confirmed the theory
(energy absorption), both properties are well-established as presented by Lakes treating the indentation as an effect of
the primary applications of the conventional foams are uniform pressure distribution that for an isotropic material
packaging and cushioning [29]. can be described as proportional to [(1 – n2)/E] 1 (Fig. 17)
[11]. By applying the classic theory of elasticity, which states
5.1 Indentation of auxetic foams Lakes first that a material can have a Poisson’s ratio varying from 1 to
studied the effects of indentation on an auxetic foam by þ0.5, it can be seen that the indentation resistance increases
assuming a localised pressure distribution, where the pressure towards infinity for a negative Poisson’s ratio.
is proportional to (1 n2)/E, and from this expression it is Furthermore, Evans and Alderson presented a relation-
evident that for a material with a Poisson’s ratio close to 1 ship to describe how the elastic constraints affect one another
any indentation is very difficult [3, 69]. At the same time this with respect to the shear modulus (G) [11].
material becomes extremely compressible because the shear
modulus exceeds the bulk modulus. This behaviour is due to 3K ð1 2vÞ
G¼ : (24)
the connection between the bulk modulus, the shear modulus 2ð1þvÞ
and the Poisson’s ratio through the equation:
In order to keep the same representation employed by
2Gð1þvÞ Lakes [Eq. (20)], this equation can be re-written in terms of
K¼ : (20)
ð1 2vÞ the bulk modulus (K) to give [3]:
solidi
physica
status
16 R. Critchley et al.: Manufacture, mechanical properties and applications of auxetic foams
Figure 17 Reaction of conventional and auxetic materials to loading under Hertzian indentation, adapted from Ref. [11].
In direct response to Lakes’ original paper [3], Burns where e and ef are the strain and the strain upon failure and s
presented a relationship for the elastic constrains in term of is the stress [Pa].
the bulk modulus [72], similar to that already given by Evans In 1987 Lakes observed that the Poisson’s ratio affects
and Alderson [Eq. (25)] [11]; however, the discrepancies the toughness of the auxetic materials and that as the
between the original work of Lakes [Eq. (20)] and the later Poisson’s ratio approaches 1, the material is expected to
works published by other authors [Eq. (25)] should be noted become extremely tough [3]. Using the critical tensile stress:
[11, 72]. Even if all the three sources employed the same
theory and treat the indentation as a uniform pressure pET
s¼ ; (27)
distribution, the equations obtained are different. When 2r ð1 v2 Þ
the original literature is compared to the work of Evans,
Alderson and Burns, it can be seen that the uniform pressure where T is the surface tension [N m1] and r the circular
distribution only accounts for a negative Poisson’s ratio, crack radius [mm], this behaviour can be explained [3, 8,
unlike in the later works, Poisson’s ratio could be either 73]. Further to elastic constraints, the material toughness
positive or negative [3, 11, 72]. Furthermore, the equation and the Young’s modulus are also affected by both its non-
presented in the later literature is set to the power of 1, a linear properties and structural aspects [43]. This initial
value unaccounted for in the original work [3, 11]. The work was continued by Lakes and Choi who outlined the
discrepancies continue when looking at the relationship differences between conventional and auxetic foams in
between the elastic constraints; with the work presented by terms of material toughness; they observed that re-entrant
Lakes lacking of a 3 in the denominator, compared to the foams have an initial lower stiffness (Young’s modulus)
equations reported by Burns and Evans and Alderson [3, 11, compared to conventional ones, but as a result of non-linear
72]. It is unknown why these discrepancies occurred, it is behaviour during a large deformation, the energy density is
possible that they could be the result of either an error within higher [29]. Furthermore, the energy absorption can be
the text or a failure to apply the correct concepts. Because of increased by using conventional foams with higher
these discrepancies found within the original literature the densities; this however is not always desirable as it results
work by Burns, Evans and Alderson and the equations that in a higher stiffness [26, 29]. In order to understand the full
they present should be considered to be correct. effect of energy absorption for auxetic and conventional
foams, Lakes and Choi studied the behaviour of foams
5.2 Toughness Toughness is an important mechan- subjected to a range of volumetric compression ratios under
ical property for a polymeric porous material since it compressive and tensile stresses [29]. They found that the
determines the maximum energy absorption of the foam material behaviour was substantially different for each of
per unit, and can be determined by taking the integral of a the re-entrant specimens with toughness factors (1.7, 2.1,
stress–strain curve, and applying the following equation: 2.3, 2.6 and 3.2) increasing with increasing volumetric
Z ef compression ratio (2.0, 2.6, 3.2, 3.7 and 4.2) [29].
Energy Likewise, Choi and Lakes analysed the material
¼ sde; (26)
Volume 0 toughness by applying a methodology similar to that
where emin and emax are the minimum and maximum strain.
The graphical data reported in Fig. 18, where loading
levels (r), are defined as [74]:
Umax Figure 18 Load vs. displacement for different levels of loading for
r¼ ; (29) (a) auxetic and (b) conventional foams, adapted from Ref. [9].
Ur
where Umax and Ur are the maximum displacement at a
particular level and at failure (measured in mm), show foams exhibit energy dissipation greater than that of auxetic
greater energy absorption for the hysteresis loops of the foams and deduced that the overall compression is the
auxetic foams, which with support of the stress–strain controlling factor in the final Poisson’s ratio.
curves of static testing demonstrates the increase in To further understand the mechanical properties of
resistance to failure and mechanical resilience. The auxetic foams, Bianchi et al. [5] studied how the energy
hysteresis cycle areas for both auxetic and conventional dissipation during both tension and compression is affected
foam increased with the load applied. On average auxetic when the foam is subjected to a four phase fabrication
foams dissipate 2.4 times the amount of energy dissipated by process (as described in Section 2.1.1), as shown in Fig. 19.
conventional foams for the first cycle under different An interesting observation was that the energy dissipation is
loadings. Furthermore, Bezazi and Scarpa stated that there on average greater for second phase auxetics than for first
was different stiffness degradation behaviour for tension– phase auxetics, although first phase auxetic experienced far
tension and compression–compression loading, although it less scattering of the results. It was further observed, for both
was found that the difference between each was not directly tensional and compressive loading, that the conventional
linked to the energy absorption [9]. phase demonstrate on average smaller energy dissipation
Bianchi et al. [4] continued to study the energy compared to the auxetic counterparts. However, the
dissipation during quasi-static and cyclic loading under conventional phases exhibited mechanical stiffness of up
tension. They observed that when the negative Poisson’s to an order of magnitude higher when compared to the
ratio approached zero the energy dissipation decreased. auxetic phases [25, 27, 37, 60, 75].
Furthermore, this study demonstrated that there was no In addition to quasi-static cyclic loading, Scarpa et al.
significant statistical correlation between the various auxetic [10] have studied the energy absorption of conventional,
sample batches analysed. Interestingly, Bianchi et al. auxetic and iso-volumetric materials at different strain rates
reported that during quasi-static cyclic tests, conventional under tensile loading. Strain rates of 8, 10 and 12 s1 and
solidi
physica
status
18 R. Critchley et al.: Manufacture, mechanical properties and applications of auxetic foams
Other possible applications include structural integrity [4] M. Bianchi, F. L. Scarpa, and C. W. Smith, J. Mater. Sci. 43,
structures [6, 10], sandwich components [1, 7, 10, 13, 64], 5851 (2008).
smart structures [11, 27], biomedical components [18, 77], [5] M. Bianchi, F. Scarpa, and C. W. Smith, Acta Mater. 58, 858
dynamic and multi-physics applications [26, 27, 78–81], (2010).
knee pads [1], tear resistant sponges [1] and sound absorbing [6] K. E. Evans, Endeavour 15, 170 (1991).
[7] J. B. Choi and R. Lakes, Cell Polym. 10, 205 (1991).
materials [1].
[8] R. S. Lakes, J. Mech. Des. 115, 696 (1993).
[9] A. Bezazi and F. Scarpa, Int. J. Fatigue 31, 488 (2009).
7 Summary [10] F. Scarpa, P. Pastorino, A. Farelli, S. Patsias, and A. Ruzzene,
Phys. Status Solidi B 242, 681 (2005).
(i) Conventional polyurethane foam cells can be repre- [11] K. E. Evans and A. Alderson, Adv. Mater. 12, 617 (2000).
sented as pentagonal and hexagonal structures. [12] N. Raviral, A. Alderson, and K. L. Alderson, J. Mater. Sci. 42,
However, in reality the pores are slightly elongated 7433, (2007).
and not perfectly symmetrical. [13] W. Yang, Z. M. Li, W. Shi, H. Xie, and M. B. Yang, J. Mater.
(ii) Auxetic polyurethane foam cells can be represented as Sci. 39, 3269 (2004).
either re-entrant cells or random rib structures. [14] A. Andersson, S. Lundmark, A. Magnusson, and F. H. J.
(iii) Both 2-D and 3-D models have been proposed to Maurer, J. Appl. Polym. Sci. 111, 2290 (2009).
[15] P. J. Stott, R. Mitchell, K. L. Alderson, and A. Alderson,
explain the auxetic behaviour. These models include:
Materials World 8, 12 (2000).
the rigid triangle, honeycomb, missing rib and bow-tie [16] A. E. H. Love, A Treatise on Mathematical Theory of
models. Although each of the models account for some Elasticity (Dover Publisher, New York, 1927), p. 104.
of the behaviours observed, no one of these models [17] K. E. Evans and K. L. Alderson, Eng. Sci. Educ. J. 9, 148
alone can account for all behaviours observed and (2000).
therefore they should be considered together. [18] G. E. Stavroulakis, Phys. Status Solidi B 242, 710 (2005).
(iv) The auxetic fabrication requires a temperature equal or [19] Q. Liu, Literature review: materials with negative Poisson’s
greater than the softening temperature of the polymeric ratio and potential applications to aerospace and defence.
material. Defence Science and Technology Organisation, Victoria,
(v) A literature review indicates that an optimum heating Australia (2006), p. 44.
[20] A. Alderson and K. L. Alderson, Proc. Inst. Mech. Eng.,
time for the process has not been identified yet and that
Part G – J. Aerosp. Eng. 221, 565 (2007).
heating time ranging from 6 to 50 min has been [21] Y. P. Liu and H. Hu, Sci. Res. Essays 5, 1052 (2010).
reported. This is likely due to the authors utilising [22] A. Alderson, Soc. Chem. Ind. 24, 18 (2011).
different equipment, materials and different sizes. [23] A. G. Kolpakov, J. Appl. Math. Mech. 49, 739 (1985).
(vi) Failure to apply a correct volumetric compression ratio [24] N. Chan and K. E. Evans, J. Mater. Sci. 32, 5945 (1997).
will result in transitional samples, where both conven- [25] C. W. Smith, J. N. Grima, and K. E. Evans, Acta Mater. 48,
tional and auxetic cell units are present but without 4349 (2000).
presenting auxetic behaviour. [26] F. Scarpa, J. R. Yates, L. G. Ciffo, and S. Patsias, Proc. Inst.
(vii) The optimum volumetric compression ratio is between Mech. Eng. C 216, 1153 (2002).
2 and 5 or between 50 and 80% of the original size. [27] F. Scarpa, L. G. Ciffo, and J. R. Yates, Smart Mater. Struct.
(viii) Pore size has been found to be a secondary factor 13, 49 (2004).
[28] M. Bianchi, F. Scarpa, C. W. Smith, and G. R. Whittell,
influencing the auxetic behaviour whilst the cell
J. Mater. Sci. 45, 341 (2010).
geometry is believed to be of primary influence. [29] J. B. Choi, and R. S. Lakes, J. Mater. Sci. 27, 4678 (1992).
(ix) Auxetic materials have been found to demonstrate [30] S. A. McDonald, N. Ravirala, P. J. Withers, and A. Alderson,
enhanced properties including mechanical hardness, Scr. Mater. 60, 232 (2009).
toughness, stiffness and dampening. [31] F. Cadamagnani, S. Frontoni, M. Bianchi, and F. Scarpa,
(x) There are a numerous potential applications for auxetic Phys. Status Solidi B 246, 2118 (2009).
samples including smart filters, biomedical apparatus, [32] Y. C. Wang, R. Lakes, and A. Butenhoff, Cell Polym. 20, 373
protective sports clothing, packaging and numerous (2001).
other smart systems. [33] J. N. Grima, D. Attard, R. Gatt, and R. N. Cassar, Adv. Eng.
Mater. 11, 533 (2009).
[34] M. Bianchi, F. Scarpa, M. Banse, and C. W. Smith, Acta
Mater. 59, 686 (2011).
Acknowledgements The authors gratefully acknowledge [35] M. Bianchi, S. Frontoni, F. Scarpa, and C. W. Smith, Phys.
the financial support of EPSRC and Dstl (EP/G042195/1). Status Solidi B 248, 30 (2011).
[36] J. B. Choi and R. Lakes, J. Mater. Sci. 27, 5375 (1992).
References [37] N. Chan and K. Evans, J. Mater. Sci. 32, 5725 (1997).
[38] L. J. Gibson and M. F. Ashby, Cellular Solids: Structure
[1] R. Lakes, Nature 358, 713 (1992). and Properties (Cambridge University Press, London,
[2] E. A. Friis, R. Lakes, and J. B. Park, J. Mater. 23, 4406 1988).
(1988). [39] N. Mills, Polymer Foams Handbook (Butterworth Heinemann,
[3] R. Lakes, Science 235, 1038 (1986). 2007), p. 535
solidi
physica
status
20 R. Critchley et al.: Manufacture, mechanical properties and applications of auxetic foams
[40] K. E. Evans, M. A. Nkansah, I. J. Hutchinson, and S. C. [61] A. Bezazi and F. Scarpa, Int. J. Fatigue 29, 922 (2007).
Rogers, Nature 353, 124 (1991). [62] F. Scarpa, W. A. Bullough, and P. Lumley, J. Mech. Eng. Sci.
[41] A. Alderson, J. Rasburn, and K. E. Evans, Phys. Status Solidi 218, 241 (2004).
B 244, 817 (2007). [63] F. Scarpa and F. C. Smith, J. Intell. Mater. Syst. Struct. 15,
[42] J. N. Grima, R. Gatt, N. Ravirala, A. Alderson, and K. E. 973 (2004).
Evans, Mater. Sci. Eng. 423, 214 (2006). [64] F. Scarpa and G. Tomlinson, J. Sound Vib. 230, 45 (2000).
[43] R. Lakes, J. Mater. Sci. 26, 2287 (1991). [65] C. P. Chen and R. Lakes, Cell Polym. 8, 343 (1989).
[44] L. Rothenburg, A. I. Berlin, and R. J. Bathurst, Nature 354, [66] R. Lakes, Cell Polym. 10, 466 (1991).
470 (1991). [67] C. P. Chen and R. Lakes, J. Eng. Mater. Technol. 118, 285
[45] J. B. Choi and R. Lakes, J. Compos. Mater. 29, 113 (1995). (1996).
[46] I. G. Masters and K. E. Evans, Compos. Struct. 35, 403 (1996). [68] F. A. McClintock and A. S. Argon, Mechanical Behaviour
[47] J. M. Berthelot, J. Reinf. Plast. Compos. 7, 284 (1988). of Materials (Addison-Wesley, Reading, Massachusetts,
[48] J. B. Choi and R. Lakes, Int. J. Mech. Sci. 37, 51 (1995). 1966).
[49] K. E. Evans, M. A. Nkansah, and I. J. Hutchinson, Acta [69] S. P. Timoshenko and J. N. Goodier, Theory of Elasticity
Metall. Mater. 42, 1289 (1994). (McGraw-Hill Book Company, New York, 1970).
[50] F. R. Attenborough, The modelling of network polymers. [70] R. D. Mindlin, Int. J. Solids Struct. 1, 265 (1965).
PhD thesis, University of Liverpool, Liverpool (1997). [71] R. Lakes, Int. J. Solids Struct. 22, 55 (1986).
[51] C. P. Chen and R. S. Lakes, Cell Polym. 14, 186 (1995). [72] S. Burns, Science 238, 551 (1987).
[52] H. X. Zhu, J. F. Knott, and N. J. Mills, J. Mech. Phys. Solids [73] I. N. Sneddon, Fourier Transforms (McGraw-Hill, New York,
45, 319 (1997). 1951).
[53] J. A. Elliott, A. H. Windle, J. R. Hobdell, G. Eeckhaut, R. J. [74] A. R. Bezazi, A. El Mahi, J. M. Berthelot, and B. Bezzazi,
Oldman, W. Ludwig, E. Boller, P. Cloetens, and J. Baruchel, Strength Mater. 35, 149 (2003).
J. Mater. Sci. 37, 1547 (2002). [75] N. Chan and K. E. Evans, J. Cell. Plast. 35, 130 (1999).
[54] N. Gaspar, C. W. Smith, E. A. Miller, G. T. Seidler, and K. E. [76] H. Askes, D. Nguyen, and A. Tyas, Comput. Mech. 47, 657–
Evans, Phys. Status Solidi B 242, 550 (2005). 667 (2011).
[55] I. Shufrin, E. Pasternak, and A. V. Dyskin, Int. J. Solids [77] B. D. Caddock and K. E. Evans, Biomaterials 35, 130 (1995).
Struct. 49, 2239 (2012). [78] M. Ruzzene and F. Scarpa, J. Intell. Mater. Syst. Struct. 14,
[56] E. Pasternak and A. V. Dyskin, Int. J. Eng. Sci. 52, 103 (2012). 443 (2003).
[57] F. Scarpa and P. J. Tomlin, Fatigue Fract. Eng. Mater. Struct. [79] F. Scarpa, F. C. Smith, B. Chambers, and G. Burriesci,
23, 717 (2000). Aeronaut. J. 107, 175 (2003).
[58] A. Alderson, Chem. Ind. 384, (1999). [80] F. C. Smith and F. Scarpa, IEE Proc. Sci. Meas. Technol. 151,
[59] R. Lakes and K. Elms, J. Compos. Mater. 27, 1193 (1993). 9 (2004).
[60] C. W. Smith, F. Lehman, R. J. Wootton, and K. E. Evans, Cell [81] J. P. M. Whitty, A. Alderson, P. Myler, and B. Kandola,
Polym. 18, 79 (1999). Composites A 34, 525 (2003).