Composites Part B 250 (2023) 110433

Contents lists available at ScienceDirect

Composites Part B
journal homepage: www.elsevier.com/locate/compositesb

Fracture performance of fibre-reinforced epoxy foam

George Irven a, b, Declan Carolan a, b, *, Alexander Fergusson a, b, John P. Dear b
a
FAC Technology, 53 Lydden Grove, London, SW18 4LW, UK
b
Department of Mechanical Engineering, Imperial College London, London, SW7 2AZ, UK

A R T I C L E I N F O A B S T R A C T

Keywords: Low density aramid and carbon fibre-reinforced epoxy foam has been synthesised with the aim of improving
Epoxy foam mechanical properties, principally fracture performance. The foam properties measured were fracture energy,
Fracture toughness compressive strength, and density. The influence of fibre type, loading, and length was investigated. In addition,
Fibre-reinforced
composite face-sheet bond tests were performed to ascertain how effective toughness transferred from individual
Fractography
component to composite structure. In general, the addition of fibres improved the mechanical performance of
reinforced samples compared to the control foam. Increases in compressive strength were moderate whilst
fracture energy was increased by up to 107% from 124 J/m2 to 256 J/m2 by the addition of 0.75 mm aramid
fibres. Increased fracture energy of the foam and the presence of fibres on the foam surface, caused an increase in
face-sheet bond propagation fracture toughness of 50% from 277 J/m2 to 416 J/m2.

1. Introduction variety of researchers in recent years, Song et al. [6] produced short
carbon fibre-reinforced epoxy foams via mechanical frothing and ach­
The combination of high specific strength, corrosion resistance and ieved an increase in toughness of up to 113% at densities ranging from
low radar signature makes composite sandwich structures an attractive 362 to 475 kg/m3. Alonso et al. [7] synthesised short fibre-reinforced
structural choice for many disciplines [1]. However, the brittle nature of epoxy foam with both glass and aramid fibres at a density of 300
composite materials can lead to substantial overdesign of sandwich kg/m3. Improvements in shear and compressive strength were sub­
structure components, counteracting their weight and cost savings stantial, especially in the foam rise direction. Alonso et al. highlight the
benefits. The toughness of a composite sandwich structure, which is a importance of using a suitable coupling agent between fibre and matrix
critical parameter for it to withstand damage, can be improved by to ensure effective strengthening. They also worked on modelling of the
improving the toughness of the individual components. Previous compressive properties of glass fibre-reinforced epoxy foam using a
research by this group has demonstrated that by altering the layup of the statistical approach [8]. The foams ranged from 250 to 550 kg/m3 and
individual plies in the composite face-sheets, a significant increase in the variables that were studied were density, fibre weight fraction, and
impact resistance of the structure can be obtained [2]. More recently, fibre length. They note that more variables such as fibre aspect ratio and
impact performance of sandwich structures has been improved through stiffness could be included for increased accuracy.
toughening the matrix of the sandwich face-sheet [3]. During the impact Fibre modification of polymer foams has been the subject of research
testing of the epoxy-foam-core sandwich structures in the previous for some time. Cotgreave and Shortall [9–11] investigated polyurethane
research, extensive cracking within the core was found. As such, this foams reinforced with chopped glass fibres. Increases in tensile strength
project builds on this research further and details a systematic study of were moderate, up to 22%, while increases in toughness were more
the effect of reinforcing epoxy foam with short-cut carbon and aramid pronounced, up to 45%. Fibres were found to lie within struts and shown
fibres with the aim of increasing strength and toughness. Toughness is a to arrest or deflect cracks propagating through the foam matrix causing
crucial property for foams as in tension they fail by the propagation of a fibre pull-out to occur. They found that individual filaments well
single crack [4]. Marsavina and Linul [5] recently conducted a distributed within the foam were more effective than fibre bundles.
comprehensive review of the fracture toughness of polymer foams Carling [12] tested polyurethane foams reinforced with 7 mm glass fi­
including reinforcement methods. bres at a density of 80 kg/m3 Limited gains in fracture energy were
Reinforcement of epoxy foam has experienced attention from a realised, however, modulus and critical stress intensity factor were

* Corresponding author. FAC Technology, 53 Lydden Grove, London, SW18 4LW, UK.
E-mail address: d.carolan@imperial.ac.uk (D. Carolan).

https://doi.org/10.1016/j.compositesb.2022.110433
Received 28 March 2022; Received in revised form 9 October 2022; Accepted 17 November 2022
Available online 18 November 2022
1359-8368/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
G. Irven et al. Composites Part B 250 (2023) 110433

Fig. 1. SCB schematic.

improved. It was noted that improvements in mechanical performance temperature is required when manufacturing foams as the viscosity,
were limited due to the fibres being bundled together as opposed to particularly the thixotropic response, of the foam is critical to devel­
being individually distributed throughout the foam. Significant work oping and maintaining an optimal microstructure at a target density.
has been carried out on fibre-reinforced phenolic foams [13–16], im­ The fibre loadings used in parts per hundred resin (PHR) were 0.5, 1,
provements in compressive, tensile, shear, and friability properties were 2.5, and 5. For 3 mm and 6 mm fibres, it was only possible to load the
substantial. resin with 0.5 and 1 PHR, higher loading caused a drastic increase in
The failure mode of face-core debonding can cause significant de­ viscosity and resulted in a poor-quality foam.
creases in the structural integrity of sandwich panels as it prevents shear Carbon fibre-reinforced polymer (CFRP) composite sandwich panels
transfer between face sheets [17]. Recent single cantilever beam (SCB) were manufactured for single cantilever beam (SCB) testing. The carbon
tests by Irven et al. [3] found that systematically modifying the matrix fibre used was 385 gsm of H2550 fibres in 12 k tows in a 0/90 non-crimp
resin of the sandwich face-sheet caused a change in failure mode at the fabric, the SCB sandwich panels used [[0/90]4]s for 8 plies in each face-
face-core interface. When the matrix resin has high strength, the failure sheet. An amine-cured epoxy system formed the matrix of the SCB
mode involves significant foam fracture whereby portions of foam panels. The resin was a standard diglycidyl ether of bisphenol-A
remain on the face after testing. As such, it was expected that an (DGEBA) with an epoxide equivalent weight of 185 g/eq. This was
improvement in foam fracture performance caused by the addition of cured with a stoichiometric amount of a difunctional primary amine
short fibres would translate to an improvement in SCB interface (JEFFAMINE D-230) from Huntsman, UK. The composite sandwich
toughness. Furthermore, Shen et al. [16] reported a seven-fold increase panels were laid up on a flat release-coated aluminium plate and sealed
in peel resistance with the addition of 3 wt percentage (wt%) 6.4 mm using a vacuum bag. The infusion then took place over a period of 5–10
aramid fibres in phenolic foam. min and was then cured under vacuum for 10 h at 40 ◦ C and 10 h at
The current research investigates reinforced epoxy foams at a nom­ 55 ◦ C.
inal density of 170 kg/m3 with the aim of improving the mechanical
properties of the epoxy foam. A key aim of this work was to maintain a 3. Methods
low density for all reinforced foams to maximise utility in real-world
applications. Previous research in the literature has mainly focussed Single-edge notched bending (SENB) tests in three-point bend
on high density epoxy foams [6,7]. configuration were conducted to determine the fracture energy, Gc, in
accordance with ASTM D-5045 [18]. Multiple studies have confirmed
2. Materials the SENB specimen to be suitable for fracture toughness testing in foam
[19,20]. In order to satisfy the plane strain condition, test specimens
A commercially sensitive epoxy based foaming resin formed the basis were machined with dimensions 140 mm × 28 mm × 14 mm. These
of the materials investigated in the current work. The nominal density of specimens were notched to a depth of 14 mm with a razor blade held in a
the cured foam in the current work is 170 kg/m3. Short-cut para-aramid custom device that was fixed to a drill press, allowing a consistent notch
and carbon fibres were sourced from Barnet Europe. Aramid fibres had depth to be achieved. Razor blades are frequently used in the literature
cut lengths of 0.75 mm, 3 mm, and 6 mm and an average diameter of 12 to create cracks in foam fracture toughness specimens [11,12,17,21,22].
μm. Carbon fibres had cut lengths of 1.5 mm, 3 mm, and 6 mm and an All mechanical testing was conducted using an ‘Instron 4466’
average diameter of 8 μm. Manufacturing limitations prevented 0.75 screw-driven universal testing machine fitted with a 10 kN load cell. The
mm carbon fibres being cut. Both short-cut fibre types were appropri­ SENB specimens were tested at a constant crosshead displacement rate
ately sized for use with epoxy. Milled carbon fibres were sourced from of 1 mm/min. The fracture energy, GIc was calculated using the energy
ELG Carbon Fibre with average length 0.6 mm and an average diameter method via:
of 7 μm. The milled fibres used were unsized prior to milling and were
U
used as received. GIc (bulk) = (1)
Reinforced epoxy foams were synthesised via a dispersing homoge­ bwφ
nising blade attached to a mounted mixer with a maximum rotational
Where U is the energy under the corrected load-displacement curve and
speed of 2800 rpm to disperse fibres in the foaming resin. High-speed
φ is an energy calibration factor as defined in the ASTM standard (b and
mixing took place with a resin temperature of 70 ◦ C to minimise vis­
w are the breadth and width of the sample respectively) [18]. At least
cosity and ensure the good dispersion of the various fibres. A stoichio­
five replicate specimens were tested for each formulation.
metric amount of a commercially sensitive amine-based hardener was
The compressive properties of the epoxy foam were tested according
then added, these constituents were then mixed thoroughly again. The
to ASTM D1621 [23] using a screw-driven universal testing machine.
mixture was noted to begin foaming shortly after mixing and was poured
Samples with dimensions of 30 × 30 × 30 mm3 cubes were cut from
into a rectangular mould and cured at 21 ◦ C for 24 h, followed by a 24 h
foam panels with a diamond saw. The samples were placed between
post-cure at 40 ◦ C. The long cure cycle at a precisely controlled
stainless steel testing platens, and a load was then applied with a

2
G. Irven et al. Composites Part B 250 (2023) 110433

Fig. 2. SEM images of polished aramid foam samples. (a): Control (b): 0.75 mm 2.5 PHR (c): 3 mm 0.5 PHR. (d): 6 mm 1 PHR.

crosshead rate of 2 mm/min. Compressive strength was calculated from mode-I dominant but not pure mode-I. It should be noted that it is ex­
the maximum stress within a strain of 10%. The compression samples pected that the mode-mixity for the SCB test used in this research is
were also used to calculate density. The sampls.es were measured and minimal. Adams et al. [27,28] conducted finite element analyses to
weighed with an electronic balance. Measuring multiple samples evaluate the mode-mixity of various face-sheet debonding test methods.
allowed for variations in density across a foam panel to be monitored. They found that the SCB test method was the most appropriate for
Imaging of both polished and fractured foam samples was conducted minimising bending stresses in the core, eliminating crack kinking, and
using a Tescan Mira scanning electron microscope (SEM). Polished minimising any mode-II component at the crack tip to below 5%. In fact,
samples were prepared using a standard wet grinding technique up to they found that the mode-I component is over 98% for the sandwich
2000 grit sandpaper. Imaging of the polished samples revealed valuable configurations evaluated.
information regarding the morphology and microstructure of the foams. SCB tests were conducted with both the control foam and the 2.5
A tight distribution of void size is typically a good indicator of subse­ PHR 0.75 mm Aramid fibre foam. Test specimens of 25 mm × 185 mm
quent good mechanical performance. Additional images of the fractured were cut from a panel with a core thickness of 10 mm and a face-sheet
samples were also taken with a high-resolution digital camera. The thickness of 3.2 mm. A 12 μm thick PTFE crack starter film of length
camera used was a Canon EOS 5DS R with a 100 mm macro lens. 55 mm was used to ensure an appropriately sharp starter crack. The
Many test methods have been developed to test mode I debonding of corrected beam theory (CBT) method was employed to calculate both
a sandwich skin, they are summarised by Ratcliffe [24,25] who also the initiation fracture energy, Gc,init, and the steady-state propagation
developed a test using a lengthened loading arm in an attempt to stan­ fracture energy, Gc,prop, of the composites. Each specimen is clamped to
dardise the test. However, the focus here will be on the SCB test as used a roller that is free to move on a track perpendicular to the crosshead
by Glaessgen et al. [26] illustrated in Fig. 1. This method allows an direction but is otherwise built-in. The tests were conducted at a con­
apparent mode I critical strain energy release rate to be measured most stant crosshead displacement rate of 2 mm/min using a screw-driven
accurately as discussed by Ratcliffe [24] and Adams [27,28]. Ratcliffe tensile testing machine. The loads and displacements were recorded,
also carried out a sizing study which proposes an algorithm, based on and the crack lengths monitored using a high-resolution digital camera
limitations of the material used, to determine appropriate dimensions setup for magnification and periodic imaging. At least five replicate
for SCB specimens [24,25]. Dimension limitations such as sizing the specimens were tested for each foam used.
initial debond length to ensure bending is the dominant deformation
mode of the loaded face sheet are outlined in a step-by-step manner. This 4. Results
sizing system has been followed for the material properties of the
sandwich structures in the current research. There are difficulties to 4.1. Microscopy: polished aramid-reinforced foam
measuring pure mode-I fracture properties of an interface between dis­
similar materials; indeed, the measured mode is often mixed [29]. The Each foam sample was cut and polished to give a smooth surface to
difference of elastic properties between, in this instance, the skin and image using an SEM. An SEM image of one foam containing each aramid
core, will disrupt the symmetry even if the geometry and loading are fibre length is shown in Fig. 2. The images show cells surrounded by
symmetric. The mismatch in modulus will couple the normal and shear nodes connected by cell walls. The images show the foams have bi-
deformations ahead of the debond front [24]. Consequently, the modal cell size distributions with many small cells surrounding fewer
measured critical strain energy release rate will not be referred to as larger cells. The cell sizes do not appear to change significantly across
mode-I, GIc , but Gc and should be considered as a fracture energy that is the aramid fibre-reinforced formulations. However, the small cells in the

3
G. Irven et al. Composites Part B 250 (2023) 110433

Fig. 3. SEM images of polished aramid foam samples. (a): Large void in 6 mm 1 PHR. (b): Epoxy covered fibres in 0.75 mm 2.5 PHR.

Fig. 4. SEM images of polished carbon foam samples. (a): Milled fibre 2.5 PHR. (b): 1.5 mm 1 PHR. (c): 3 mm 1 PHR. (d): 6 mm 1 PHR.

aramid fibre-reinforced foams appear somewhat smaller than in the microscope. Large voids in fibre-reinforced foam with fibres of similar
control foam. The images also show that there is a slight loss of spher­ length have been previously reported in the literature [6,12].
oidicity in the cells between Fig. 2 (a) and Fig. 2 (b), (c), and (d). In the 3 During the manufacturing process of the foam, all fibres are
mm and 6 mm aramid foams, very large voids could be found. This is completely wetted with liquid epoxy. Furthermore, the aramid fibres are
shown clearly in Fig. 3 (a). These voids could measure up to 6 mm across sized appropriately for use in epoxy. As a result, the fibres in the finished
and were difficult to get an entire void within the field of view of the foam are covered in epoxy. This is the case even if the fibres do not

Table 1
Measured properties of aramid reinforced foam.
Fibre Length Fibre Loading Fracture Energy GIc % Compressive Yield Strength % Density [kg/ % Compressive Modulus
[mm] [PHR] [J/m2] change [MPa] change m3] change [MPa]

– – 124 ± 13 00 2.33 ± 0.03 00 175.1 ± 0.8 00 86.5 ± 4.9

0.75 0.5 113 ± 14 − 08 2.46 ± 0.01 +06 187.8 ± 0.8 +07 81.3 ± 2.5
1 137 ± 25 +11 2.52 ± 0.05 +08 191.7 ± 1.4 +09 98.3 ± 2.3
2.5 208 ± 45 +68 2.56 ± 0.06 +10 199.6 ± 5.7 +14 99.0 ± 4.8
5 256 ± 30 +107 2.28 ± 0.04 - 02 200.1 ± 0.6 +14 89.9 ± 1.7
3 0.5 133 ± 23 +07 2.46 ± 0.02 +06 189.8 ± 1.5 +08 94.3 ± 2.8
1 165 ± 28 +33 2.44 ± 0.07 +05 194.0 ± 1.1 +11 95.5 ± 0.8
6 0.5 143 ± 31 +16 2.33 ± 0.09 00 190.6 ± 5.1 +09 88.4 ± 6.3
1 204 ± 44 +65 2.43 ± 0.05 +04 203.3 ± 3.5 +16 94.2 ± 4.3

4
G. Irven et al. Composites Part B 250 (2023) 110433

Table 2
Measured properties of carbon reinforced foam.
Fibre Length Fibre Loading Fracture Energy GIc % Compressive Yield Strength % Density [kg/ % Compressive Modulus
[mm] [PHR] [J/m2] change [MPa] change m3] change [MPa]

– – 124 ± 13 00 2.33 ± 0.03 00 175.1 ± 0.8 00 86.5 ± 4.9

0.6 0.5 161 ± 37 +30 2.48 ± 0.02 +07 175.4 ± 0.4 00 84.9 ± 2.9
1 168 ± 40 +36 2.55 ± 0.03 +10 176.2 ± 1.0 +01 92.6 ± 1.6
2.5 180 ± 33 +45 2.58 ± 0.04 +11 178.4 ± 0.8 +02 91.9 ± 3.2
5 245 ± 59 +98 2.38 ± 0.09 +02 184.5 ± 0.9 +05 93.3 ± 5.3
1.5 0.5 140 ± 26 +13 2.41 ± 0.03 +04 176.5 ± 0.2 +01 89.7 ± 2.3
1 157 ± 33 +27 2.48 ± 0.05 +06 177.7 ± 0.5 +01 95.8 ± 2.6
3 0.5 137 ± 16 +10 2.15 ± 0.05 - 08 172.5 ± 0.6 - 01 69.3 ± 1.2
1 201 ± 13 +62 1.87 ± 0.05 - 20 173.6 ± 1.2 - 01 65.2 ± 0.9
6 0.5 159 ± 57 +29 1.93 ± 0.04 - 17 172.3 ± 1.4 - 02 69.0 ± 4.1
1 187 ± 43 +51 1.83 ± 0.01 - 21 176.7 ± 0.9 +01 74.3 ± 3.8

completely lie within a natural cell wall, epoxy will encapsulate the fibre
and bridge to the cell wall, an example of this is shown in Fig. 3 (b).
Exposed fibres, with no epoxy sheath, are very rare within the polished
samples.

4.2. Microscopy: polished carbon-reinforced foam

Fig. 4 shows polished samples of foams containing each carbon fibre

length. Again, there is a slight loss of spheroidicity in the cells between
the control foam in Fig. 2 (a) and the four foams in Fig. 4. However,
there appears to be less of a bi-modal distribution of cell sizes within the
carbon-reinforced foams. As with the aramid foams, the 3 and 6 mm
foams have some large voids which were avoided for these SEM images
as the voids could measure up to 6 mm across. Similar to the chopped
aramid fibre foams, the chopped carbon foams do not show exposed
fibres when polished. However, some exposed fibres can be seen in the
milled carbon fibre foam in Fig. 4 (a), this is related to the lack of sizing
on the milled carbon fibres.

4.3. Compression

The compressive yield strengths of the aramid and carbon fibre-

reinforced foams are presented in Table 1 and Table 2, and are plotted
Fig. 5. Compressive yield versus fibre loading for aramid fibre-
in Figs. 5 and 6 respectively. The compressive yield strength of the foam
reinforced foams.
is increased by up to 10% by adding fibres. All aramid fibre lengths give
rise to an increase in compressive strength, however, adding 5 PHR 0.75
mm aramid fibre causes the compressive yield to drop below that of the
control. A similar trend is observed when adding 0.6 and 1.5 mm carbon
fibres, 5 PHR 0.6 mm causes strength to drop back to that of the control.
It is clear that, to maximise compressive yield, approximately 2.5 PHR of
the shortest fibre length is optimal for both fibre types. While short
carbon fibres improve compressive yield, the 3 and 6 mm carbon fibres
reduce the yield strength of the foam. A visual inspection reveals the 3
and 6 mm carbon fibre-reinforced foams have some large voids such as
the one in Fig. 3. These large voids cause a drop in yield strength by
reducing the total load bearing area of a foam sample. In contrast, short
fibres do not reduce the quality of the foam in the same way. While the
long aramid fibre-reinforced foams also display large voids, the increase
in density of these foams prevents a potential drop in yield strength. The
compressive modulus values of the aramid and carbon fibre-reinforced
foams are also presented in Tables 1 and 2. These values follow the
same trends as the compressive strength values. However, the standard
deviations of the modulus values are larger than the standard deviations
of the strength values. There are significant difficulties with accurately
measuring strain in a compressive test of structural foam as discussed in
detail by Rajput et al. [30]. Primarily, parts of the sample can begin to
fail and form a crush band, as a result, the strain throughout the sample
can vary massively. Furthermore, when a foam sample is machined, the
Fig. 6. Compressive yield versus fibre loading for carbon fibre- top and bottom layer of cells are broken and are therefore very
reinforced foams. compliant. Consequently, the measured modulus will both be lower than
the true modulus and it will be sample size dependent.

5
G. Irven et al. Composites Part B 250 (2023) 110433

4.4. Density

It is interesting to note the effect on density associated with each

fibre type, length, and loading. The density of aramid and carbon fibre-
reinforced foam can be found in Tables 1 and 2 respectively. The addi­
tion of carbon fibres does little to increase the density of the foam,
clearly a benefit in any potential application. However, the addition of
aramid fibres moderately increases the density of the foam. The addition
of 5 PHR 0.75 mm aramid fibre produces a foam of 200 kg/m3, 14%
denser than the control. However, this increase in density is far out­
weighed by the 107% increase in fracture energy. Furthermore, longer
aramid fibres have an increased effect on the density with 1 PHR 6 mm
producing a foam with a density of 203 kg/m3. The increase in density is
partially caused by the addition of the denser fibres. Both aramid and
carbon fibres are denser than the epoxy polymer. The density is pri­
marily affected by the inhibiting effect the fibres have on the foaming
process due to the increased viscosity of the fibre resin mixture. It can be
inferred that carbon fibres impede the foaming process significantly less
than aramid fibres. In fact, the large voids in the 3 and 6 mm carbon
fibre-reinforced foams have caused the overall density to be lower than
the control. In these foams, the longer fibres have a tendency to mat
together forming a barrier in the resin and effectively trap more gas
Fig. 7. Fracture energy versus fibre loading for Aramid fibres. during the foaming process than the resin would otherwise, forming
large voids in the process.

4.5. Fracture properties

4.5.1. Fracture energy: Aramid-Reinforced Foam

Fracture energies for all the manufactured aramid fibre-reinforced
foams are given in Table 1, all data shows the calculated average,
standard deviation, and percentage difference to the control. The frac­
ture energy, GIc, versus fibre loading of each of the aramid reinforced
foams is shown in Fig. 7. The addition of aramid fibres was found to
greatly increase the toughness of the foam from 124 J/m2 to a maximum
of 256 J/m2. It is clear that, for a similar fibre loading, longer fibres
induce a greater increase in toughness in the foam. For example, a
loading of 1 PHR 6 mm aramid fibres produces a similar toughness to a
loading of 2.5 PHR 0.75 mm aramid fibres. It was observed from SENB
raw data curves that foams with long fibres failed in a progressive
manner with fibres arresting the crack progression. In contrast, the
control foam failed in a catastrophic manner, as displayed in Fig. 8. The
progressive failure of the long fibre specimens does not affect the
measured fracture energy; however, it is clearly a benefit for a material
in service to fail in such a manner as the total energy absorbed during a
complete failure of the foam is increased. Shen and Nutt [15] have also
reported fibre bridging in aramid-reinforced sandwich core flexure tests.
Fig. 8. Load-displacement curves for the SENB testing of aramid 6 mm
They noted a significant increase in load after initial failure, again this
and control.
did not affect the measured property. However, they note the potential
for fibre reinforcement to be used to avoid catastrophic failure of

Fig. 9. SEM images of SENB fracture surfaces. (a): Control sample. (b): 6 mm 1 PHR aramid sample showing exposed aramid fibres up to 1 mm in length.

6
G. Irven et al. Composites Part B 250 (2023) 110433

Fig. 10. SENB fracture surfaces 0.75 mm Aramid. (a): Fibres exposed after pull-out and fracture. (b): Fibres exposed after surrounding matrix failure.

Fig. 11. SENB fracture surfaces 6 mm Aramid. (a): Fibre pulled out with epoxy remaining on the end. (b): Image showing how long aramid fibres fray when broken.

sandwich structures. This is otherwise achieved through increased safety plane around the base of a pulled-out fibre. Fig. 10 (a) exhibits a case
factors leading to an increase in structure weight. where a microcrack is formed as a fibre debonds from the matrix. In
Fig. 10 (b) fibres are seen bridging across where a secondary crack has
4.5.2. Fracture morphology: Aramid-Reinforced Foam caused epoxy to fracture off around them. These images clearly
It can be seen in Fig. 9 that the toughening mechanisms of aramid demonstrate that the presence of the fibres cause the crack path to
foams include fibre pull-out and fracture as well as crack deflection. The deflect and lengthen which require additional energy to propagate [31]
crack direction in all fracture micrographs is from bottom to top. Fig. 9 and for additional lesser cracks to form leading to an increase in fracture
(a) shows the fracture surface of a control sample. The foam structure is energy.
clearly visible and not dissimilar to the polished samples examined in Fig. 11 (a) shows an image of a pulled-out fibre while (b) shows a
Fig. 2. Moreover, the fracture surface was observed to have no signifi­ series of fractured fibres, the bases of which are situated in nodes and
cant deviations from the initial crack plane over the entire crack prop­ cell walls. It is clear from Fig. 11 (a) that, while the fibre is not
agation length. Fig. 9 (b) shows the fracture surface of a foam modified completely encased in polymer, there is a significant amount of epoxy in
with 1 PHR 6 mm aramid fibres. Significant deviations from the initial small pieces remaining adhered to the end of the fibre. This contrasts
crack plane were noted, indicating that the presence of the fibres in the with the clean exposed fibres shown in Fig. 10. The cutting process to
foam caused the crack to deflect. In addition, many exposed fibres can be prepare the fibres is an aggressive process. It is probable that it affects
observed, up to 1 mm long. On the contrary, in the polished sample the sizing of the virgin fibre, especially close to the cutting plane. The
images, Fig. 2, no exposed fibres were observed. Thus, the visible fibres shorter 0.75 mm fibres will therefore be more affected by the cutting
in Fig. 9 (b) are as a consequence of the fracture process, i.e., they were process than the longer fibres. The broken fibres in Fig. 11 (b) have split
pulled out of the opposite fracture surface. Fibre pull-out appears to be a and frayed upon fracture. Furthermore, some amount of pull-out has
principal toughening mechanism within all the aramid fibre-reinforced occurred before failure as the fibres are exposed. As the fibres are pulled
foams studied here. The fibres shown in Fig. 9 (b) are solitary fibres out, they bridge the crack before failure. Sequential failure of groups of
and appear well dispersed within the foam rather than manifesting as these fibres are responsible for the progressive load-displacement traces,
bundles or agglomerates. This represents a key development on fibre- such as the one shown in Fig. 8.
reinforced foams previously reported in the literature where bundles
of fibres are mentioned as a manufacturing issue and being less effective 4.5.3. Fracture energy: carbon-Reinforced Foam
at improving mechanical and fracture properties [10,12]. The experimentally determined fracture energies for all of the
The 0.75 mm aramid fibres caused large increases in the fracture manufactured carbon fibre-reinforced foams are given in Table 2. The
energy of the foam and allowed for a higher fibre loading than longer addition of carbon fibres was found to greatly increase the toughness of
fibres. Fig. 10 presents examples of fibres on SENB fracture surfaces, the foam from 124 J/m2 to a maximum of 245 J/m2. All the carbon fibre-
Fig. 10 (a) demonstrates that fibres are pulled out but does not show reinforced foams showed an increase in fracture energy over the un­
significant amounts of epoxy left on the surfaces of the fibres. It can be modified foams. However, the link between fibre length and increase in
seen that the crack is deflected by a fibre as it deviates from its main fracture energy is not as clearly defined as with the foams modified with

7
G. Irven et al. Composites Part B 250 (2023) 110433

Fig. 12. Photos of 6 mm carbon foam SENB samples. (a): Zoomed in image of the front of a sample during a test. The crack can be seen to initially cross the path of
fibres, fracturing them, then deflect to follow a path between aligned fibres within the foam. (b): Fracture surface of a sample showing a crack that has been deflected
multiple times and fractured fibres protruding from the fracture surface.

pouring the mixture into a separate foaming and casting mould, there is
no prevailing global direction for these aligned fibres within the foaming
and casting mould. The alignment of fibres is only local. However, since
fracture initiation is a local process, this local alignment of fibres is
critically important. In Fig. 12 (a), the fracture initially progresses up­
wards from the initial crack tip from point A to point B, fracturing the
fibres in its path. After approximately 2 mm of crack propagation, it
straightens to progress mainly upwards to point C before deflecting
again to progress upwards to point D, taking a less obstructive path
between the aligned fibres within the foam. Fig. 13 plots the raw load
versus displacement data for this sample, as the load climbs from 20 to
35 N, small sharp dips in load are clearly visible. During this period of
the test, cracking noises were heard, however, periodic imaging
revealed no advancement of the main crack during this period. The
control foam did not produce these cracking noises; therefore, the noises
pre-failure in the fibre reinforced foams are attributable to the failure of
the fibres. Based on the analysis of Fig. 12 (a), Fig. 13, the periodic
imaging of the test, and the noises heard, the fibres fail before the crack
advances. The failure strain of the carbon fibres is lower than that of the
epoxy. As a result, when the sample is loaded and the area ahead of the
crack is strained past the fibre failure strain, the fibres fail sequentially.
When the epoxy failure strain is reached the broken fibres provide a path
for the crack to travel through until intact fibres prevent further prop­
Fig. 13. Load-displacement curves for the SENB testing of carbon 6 mm
and control. agation in that direction. The crack then deflects along the aligned fibres
and the sample fails suddenly. This deviation from the main plane of the
crack will give rise to mode II contributions and a resultant increase in
aramid fibres.
fracture energy. The crack path in this sample is also longer than an
unreinforced sample, again causing an increase in fracture energy.
4.5.4. Fracture morphology: carbon-Reinforced Foam
The toughening mechanisms taking place within this sample
From fractographic analyses conducted on SENB samples, it can be
depended greatly upon the state of fibre-reinforcement ahead of the
seen that the toughening mechanisms for carbon fibre-reinforced foams
crack tip, in this instance the arrangement of the fibres was conducive to
also include fibre pull-out and fracture as well as crack deflection.
a high fracture energy. Some samples did not have fibres positioned in
Fig. 12 shows high-fidelity photographs of two 6 mm carbon 1 PHR
such a way as to give as high a fracture energy. This fact explains the
samples. Fig. 12 (a) shows fibres aligned beneath the surface of the foam
larger standard deviations in Tables 1 and 2 for the fracture data of long
of one of the samples. These fibres are 6 mm long and span the whole
fibre foams. Fig. 12 (b) shows the fracture surface of a different sample,
width of the image. The process of manufacturing the foam involves
it is clear from the upper image that the crack path has been deflected.
mixing the resin and fibres together with a rotating blade. During this
The lower image shows broken fibres protruding from the surface, the
process, the fibres tend to align tangential to the tip of the mixing blade
fibres appear to be individual and well dispersed. From these images it is
within the foam mixture. This is especially relevant for longer fibres.
clear that fibre breakage is a key toughening mechanism. Similar pho­
As a result of the circular mixing motion and the subsequent step of
tographs are not shown for aramid foams as the yellow fibres are

8
G. Irven et al. Composites Part B 250 (2023) 110433

Fig. 14. SENB fracture surfaces of carbon foams. (a): Empty fibre track from a 3 mm fibre. (b) Example of both a pulled out and fractured 3 mm fibre. (c): Example of
two 6 mm fibres pulled out.

Fig. 15. SENB fracture surfaces of carbon foam. (a) 3 mm fibres bridging after epoxy has fractured off. (b): 1.5 mm fibre pulled out with epoxy remaining. (c): Pair of
6 mm fibres pulled out coated with epoxy.

wher7e a fibre has been pulled during the fracture event. The fibre was
encased in epoxy, this is evident as the track is 2 μm wider than the 8-μm
diameter fibres. The rough surface of the fibre track reinforces this, a
smooth track would indicate that the fibre was not well adhered. Fig. 14
(b) demonstrates an example of how fibres can lie within a cell wall. One
fibre is fractured at the crack plane while the second fibre has been
pulled-out and is mostly encased in epoxy. Fig. 14 (c) exhibits a pair of
fibres lying within a node that have been pulled out and are also both
encased in epoxy. Fig. 15 (a) and (b) show long exposed fibres with
significant parts of cell walls remaining after fracture. The extra fracture
surfaces created by these fibres, and the associated energy absorption,
represents a significant toughening mechanism. The main toughening
from fibre pull-out is from the shear at the interface due to interfacial
fictional sliding [32]. Fig. 15 (c) shows a pair of fibres pulled out encased
in epoxy showing clear signs of epoxy shear failure on the surface. While
the aramid fibres were found by themselves, it was not uncommon to
find longer carbon fibres in pairs such as in Fig. 14 (c) and Fig. 15 (a) and
(c). Furthermore, while aramid fibres frayed and split when broken,
carbon fibres exhibited clean fractures.
Fig. 16. SEM image of a SENB fracture surface showing milled carbon fibres Fig. 16 shows a series of milled fibres protruding from the fracture
pulled out. Note the relatively clean surfaces of the milled fibres. surface. A key difference between these fibres and the longer, chopped
carbon fibres discussed earlier is the lack of epoxy polymer adhering to
difficult to distinguish. the fibre. The chopped fibres were originally sized for use with epoxy as
The toughening mechanisms in carbon fibre-reinforced foams are the matrix and so are encased in epoxy when pulled out, whereas the
highlighted in Fig. 14. Fig. 14 (a) shows an empty fibre track, from milled fibres did not have any appropriate sizing and so pull out with a
very clean surface, almost devoid of polymer. Despite this, the milled

9
G. Irven et al. Composites Part B 250 (2023) 110433

Fig. 17. SEM image of a 6 mm fibre below the surface of a fractured node
having been loaded.

fibres provide benefit as the pull-out of clean fibres still absorbs energy
through interfacial frictional sliding [32].
Fig. 17 shows a carbon fibre below the surface of a fractured node
from an SENB sample. From the striations leading to the cell wall and Fig. 18. Load-displacement curves for the SCB testing of a control and a 2.5
node, it is clear that the fibre and the surrounding polymer material was PHR 0.75 mm aramid sample.
loaded as the crack passed along the main fracture plane, providing a
stiffening effect. It is interesting to note that the fibres within the foam
causing this stiffening effect and increasing constraint may well be
reducing the fracture energy if they are not within the crack path trig­ Fig. 21. Exposed fibres on the surface can be clearly observed. This is the
gering energy absorption mechanisms. surface that the face-sheet will ultimately be infused on. The aramid
fibres are much tougher and difficult to machine than the relatively
friable polymer foam structure. As a result, during the machining pro­
4.6. Effect on face-core interface cess, they tend to deflect under the cutting blade and bounce back once
the blade has passed. Fibres exposed during foam machining leave an
Once the improvements in fracture performance of fibre-reinforced ideal surface for the face-sheet matrix resin to bond to. Control foam
foams had been confirmed, a set of SCB face-core debond tests were samples showed failure partially within the foam. It was hypothesised
carried out on both a control foam and a 2.5 PHR 0.75 mm aramid foam. that were the foam made tougher, the failure of the SCB specimens
The results of this SCB testing can be found in Table 3. An example of a would be mostly interfacial and the foam would be left intact. However,
load versus displacement for each type of sample is shown in Fig. 18. it is clear from the results presented here, that in adding fibres to the
Adding 0.75 mm 2.5 PHR aramid fibres to the foam core of a sandwich foam the interfacial bond itself has also been improved due to the
structure increases face-core propagation fracture toughness by 50% exposed fibres on the surface.
and initiation fracture toughness by 30%. Fig. 19 shows images taken of
the side of both types of SCB samples during testing, the images show 5. Conclusion
that foam is fractured off the core and remains on the face-sheet surface.
As a result, similar toughening mechanisms observed in SENB foam Low density short carbon and aramid fibre-reinforced epoxy foams
fracture have caused the increase in toughness of this interface. These have been successfully synthesised for the first time. Improvements in
mechanisms are clear in face-sheet -side fracture surface images in the fracture performance of epoxy foams have been achieved through
Fig. 20 (a) and (b), exposed fibres that have been pulled out are clear in the addition of these fibres. The effect of fibre type, length and loading
both photographs and SEM images. There is an increase in the difference have been investigated. The incorporation of 5 PHR 0.75 mm aramid led
between the propagation and initiation fracture toughness with the to a maximum increase in fracture energy of 107% from 124 J/m2 to
addition of aramid fibres. This increase suggests an increase in the R- 256 J/m2. The main toughening mechanisms observed were fibre pull-
curve behaviour. This is due to material, mainly the fibres, bridging the out, fibre fracture and crack deflection and arrest. The addition of 2.5
gap behind the crack tip. As the aramid foam fractures, the fibres are PHR 0.75 mm aramid fibre and 0.6 mm carbon fibre both cause an in­
often pulled out as seen in the side-on image in Fig. 19, and both the SEM crease in compressive yield strength of ~10%, further additions cause a
micrograph and microscope image of the fracture surfaces in Fig. 20. drop in strength. The addition of aramid fibre causes an increase in foam
Therefore, the fibres are being pulled out of the foam behind the crack density of up to 16%, short carbon fibres only increased density up to 5%
tip, causing an increase in R-curve behaviour and a large increase in the while 3 and 6 mm carbon fibres reduced density by up to 2%. When used
difference between propagation and initiation fracture toughness. in a composite sandwich structure, 0.75 mm 2.5 PHR aramid fibre-
To further demonstrate the role of fibre bridging on improving the reinforced foam improves face-core initiation and propagation fracture
toughness of the face-core interface, a side-on micrograph was taken toughness by 30 and 50% respectively. This investigation has shown
using SEM of the surface of an ‘as-machined’ foam. This is shown in that fibre reinforced epoxy foams can outperform their unreinforced

Table 3
SCB results for aramid 0.75 mm 2.5 PHR foam and control foam.
Fibre Length [mm] Fibre Loading [PHR] Initiation Gc [J/m2] % change Propagation Gc [J/m2] % change Prop – Init Gc [J/m2] % change

– – 231 ± 43 – 277 ± 38 – 46 ± 7 –
0.75 2.5 301 ± 52 +30 416 ± 24 +50 115 ± 54 +150

10
G. Irven et al. Composites Part B 250 (2023) 110433

Fig. 19. Side view of SCB tests showing exposed fibres in the aramid foam sample.

Fig. 20. SCB face-sheet side fracture surface. (a) Digital photograph. (b) SEM image.

Funding Acquisition, John P. Dear: Writing – Review and Editing, Su­

pervision, Funding Acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial

interests or personal relationships that could have appeared to influence
the work reported in this paper.

Data availability

Data will be made available on request.

Acknowledgements
Fig. 21. SEM image showing the edge of a machined sample of aramid foam
showing exposed fibres on the surface. George Irven would like to acknowledge an EPSRC faculty CASE PhD
studentship with FAC Technology. Grant number [EP/R513052/1].
counterparts with few drawbacks other than a slight increase in density Declan Carolan acknowledges the support of UKRI, Future Leaders
and increased manufacturing complexity. There are more gains to be Fellowship. Grant number [MR/T023406/1].
realised if longer fibres can be used to manufacture low density foams
without large voids which reduce compressive strength. Large voids will References
also cause increased resin uptake during a resin infusion or require more
[1] Gibson LJ, Ashby MF. Cellular solids. Cambridge University Press; 1997.
adhesive during bonded sandwich structure manufacture. For this
[2] Rolfe E, Kaboglu C, Quinn R, Hooper PA, Arora H, Dear JP. High velocity impact
reason, the most attractive option is to make use of short carbon and and blast loading of composite sandwich panels with novel carbon and glass
aramid fibres in epoxy foams. construction. J Dyn Behav Mater 2018;4:359–72. https://doi.org/10.1007/
s40870-018-0163-5.
[3] Irven G, Duncan A, Whitehouse A, Carolan D, Fergusson AD, Dear JP. Impact
Author statement response of composite sandwich structures with toughened matrices. Mater Des
2021;203:109629. https://doi.org/10.1016/j.matdes.2021.109629.
George Irven: Conceptualization, Methodology, Investigation, [4] Marșavina L. Fracture mechanics of cellular solids. In: CISM int Cent Mech Sci
Courses lect, 521. Vienna: Springer; 2010. p. 1–46. https://doi.org/10.1007/978-
Formal analysis, Writing – Original Draft, Writing – Review and Editing, 3-7091-0297-8_1.
Declan Carolan: Conceptualization, Conceptualization, Methodology, [5] Marșavina L, Linul E. Fracture toughness of rigid polymeric foams: a review. Fatig
Formal Analysis, Writing – Review and Editing, Supervision, Alexander Fract Eng Mater Struct 2020;43:2483–514. https://doi.org/10.1111/ffe.13327.
[6] Song W, Konstantellos G, Li D, Lee KY. Short carbon fibre-reinforced epoxy foams
Fergusson: Writing – Review and Editing, Supervision, Resources, with isotropic cellular structure and anisotropic mechanical response produced

11
G. Irven et al. Composites Part B 250 (2023) 110433

from liquid foam templates. Compos Sci Technol 2019;184. https://doi.org/ [20] Pugna A, Negrea R, Linul E, Marșavina L. Is fracture toughness of pur foams a
10.1016/j.compscitech.2019.107871. material property? Stat . Materials Mater (Basel) 2020;13:1–13. https://doi.org/
[7] Alonso MV, Auad ML, Nutt SR. Short-fiber-reinforced epoxy foams. Compos Part A 10.3390/ma13214868.
Appl Sci Manuf 2006;37:1952–60. https://doi.org/10.1016/j. [21] Zenkert D, Bäcklund J. PVC sandwich core materials: mode I fracture toughness.
compositesa.2005.12.011. Compos Sci Technol 1989;34:225–42. https://doi.org/10.1016/0266-3538(89)
[8] Alonso MV, Auad ML, Nutt SR. Modeling the compressive properties of glass fiber 90030-4.
reinforced epoxy foam using the analysis of variance approach. Compos Sci [22] Saenz EE, Carlsson LA, Karlsson AM. In situ analysis of fatigue crack propagation in
Technol 2006;66:2126–34. https://doi.org/10.1016/j.compscitech.2005.12.016. polymer foams. Eng Fract Mech 2013;101:23–32. https://doi.org/10.1016/j.
[9] Cotgreave TC, Shortall JB. Failure mechanisms in fibre reinforced rigid engfracmech.2012.10.009.
polyurethane foam. J Cell Plast 1977;13:240–4. https://doi.org/10.1177/ [23] ASTM D1621-16. Standard test method for compressive properties of rigid cellular
0021955X7701300401. plastics. West Conshohocken, USA: American Society for Testing and Materials;
[10] Cotgreave TC, Shortall JB. The mechanism of reinforcement of polyurethane foam 2016. https://doi.org/10.1520/D1621-10.2. n.d.
by high-modulus chopped fibres. J Mater Sci 1977;12:708–17. https://doi.org/ [24] Ratcliffe JG. Sizing a single cantilever beam specimen for characterizing facesheet-
10.1007/BF00548161. core debonding in sandwich structure. 2010. https://doi.org/10.1177/
[11] Cotgreave TC, Shortall JB. The fracture toughness of reinforced polyurethane foam. 0021998311401116.
J Mater Sci 1978;13:722–30. https://doi.org/10.1007/BF00570506. [25] Ratcliffe JG, Reeder JR. Sizing a single cantilever beam specimen for characterizing
[12] Carling MJ. Fracture mechanics of short fibre composites. Imperial College London; facesheet-core debonding in sandwich structure. J Compos Mater 2011;45:
1988. 2669–84. https://doi.org/10.1177/0021998311401116.
[13] Desai AA, Nutt SR, Alonso MV. Modeling of hybrid composite foams. J Cell Plast [26] Glaessgen EH, Reeder JR, Sleight DW, Wang JT, Raju IS, Harris CE. Debonding
2010;46:113–28. https://doi.org/10.1177/0021955X09351195. failure of sandwich-composite cryogenic fuel tank with internal core pressure.
[14] Desai AA, Auad ML, Hongbin Shen H, Nutt SR. Mechanical behavior of hybrid J Spacecraft Rockets 2005;42:613–27. https://doi.org/10.2514/1.5567.
composite phenolic foam. J Cell Plast 2008;44:15–36. https://doi.org/10.1177/ [27] Adams D, Nelson J, Bluth Z, Hansen C. Development and evaluation of fracture
0021955X07078021. mechanics test methods for sandwich composites. Baltimore: FAA JAMS Tech. Rev.
[15] Shen H, Nutt SR. Mechanical characterization of short fiber reinforced phenolic Meet.; 2012.
foam. Compos Part A Appl Sci Manuf 2003;34:899–906. https://doi.org/10.1016/ [28] Adams D, Nelson J, Bluth Z. Development and evaluation of fracture mechanics
S1359-835X(03)00136-2. test methods for sandwich composites. FAA Technical Review: https://pdfs.semant
[16] Shen H, Lavoie AJ, Nutt SR. Enhanced peel resistance of fiber reinforced phenolic icscholar.org/bd21/65b6c116ba8c3c818907d34c0f57351da09a.pdf; 2011.
foams. Compos Part A Appl Sci Manuf 2003;34:941–8. https://doi.org/10.1016/ [29] Suo Z, Hutchinson JW. Sandwich test specimens for measuring interface crack
S1359-835X(03)00210-0. toughness. Mater Sci Eng, A 1989;107:135–43. https://doi.org/10.1016/0921-
[17] Viana GM, Carlsson LA. Influences of foam density and core thickness on debond 5093(89)90382-1.
toughness of sandwich specimens with PVC foam core. J Sandw Struct Mater 2016; [30] Rajput MS, Burman M. K\"{o}ll J, Hallstr\"{o}m S. Compression of structural foam
5:103–18. https://doi.org/10.1177/1099636203005002020. materials – experimental and numerical assessment of test procedure and specimen
[18] ASTM D5045-14. Standard test methods for plane-strain fracture toughness and size effects. J Sandw Struct Mater 2019;21:260–88. https://doi.org/10.1177/
strain energy release rate of plastic materials. West Conshohocken, USA: American 1099636217690500.
Society for Testing and Materials; 2014. https://doi.org/10.1520/D5045-14. n.d. [31] Faber KT, Evans AG. Crack deflection processes-II. Experiment. Acta Metall 1983;
[19] Viana GM, Carlsson LA. Mechanical properties and fracture characterization of 31:577–84. https://doi.org/10.1016/0001-6160(83)90047-0.
cross-linked PVC foams. J Sandw Struct Mater 2002;4:99–113. https://doi.org/ [32] Hull D, Clyne TW. An introduction to composite materials. Cambridge University
10.1177/1099636202004002227. Press; 1996. https://doi.org/10.1017/cbo9781139170130.

12