High Performance Carbon Fibre Reinforced Epoxy Composites with

Controllable StiffnessHenry A Maples a,b, Steven Wakefield a, Paul Robinson c,*, Alexander Bismarck a,b,*

a Polymer and Composite Engineering (PaCE) Group, Department of Chemical Engineering, Imperial College London, South Kensington Campus, London SW7 2AZ, UK

b Polymer and Composite Engineering (PaCE) Group, Institute of Materials Chemistry and Research, Faculty of Chemistry, University of Vienna, Währinger Straße 42, A-1090 Vienna, Austria

c The Composites Centre, Department of Aeronautics, Imperial College London, South Kensington Campus, London SW7 2AZ, UK

*Corresponding Authors, +43(1)427771301, a.bismarck@imperial.ac.uk; alexander.bismarck@univie.ac.at (A. Bismarck), +44 (0)20 7594 5073, p.robinson@imperial.ac.uk (P. Robinson)Keywords: Carbon fibres (A); polymer-matrix composites (A); structural composites (A); interleaved composites; high-temperature properties (B).

Abstract

The mechanical properties of polystyrene-interleaved carbon fibre reinforced epoxy

composites, which exhibit controllable stiffness, have been investigated. DMTA and flexural

tests showed that the storage modulus and flexural stiffness of these composites could be

reduced by up to 98% when heated from 20°C to 120°C and the stiffness was fully

recoverable on cooling. The flexural stiffness of the interleaved composites at room and

elevated temperatures were predicted using simple beam theory and were found to be in good

agreement with the measured values. Compressive and tensile performances were

significantly reduced at 120°C due to the presence of the softened polystyrene interleaves.

Flexural strength tests at 20°C indicate that there is a need for improvement of the adhesion

between polystyrene and carbon fibre reinforced epoxy plies.

1. Introduction

Controllable stiffness materials possess the characteristic that the stiffness can be changed on

demand. Such materials have potential applications in deployable structures, e.g.

instrumentation booms on satellites [1, 2], and in shape adaptive structures such as morphing

wings [3-6]. In both cases the ability to reduce the stiffness prior to deployment, or prior to

the required shape change, can significantly reduce the requirements of the actuation system.

Various forms of controllable stiffness materials have been proposed such as a controllable

stiffness composite consisting of an elastomer matrix containing braided composite tubes

which could be pressurised to change the stiffness and shape of the composite [7-9]. However

this paper will focus on materials in which the stiffness can be varied by controlling the

1

resistance to shear displacement between constant stiffness elements. One such material,

proposed by McKnight and Henry [10], consists of a laminate formed by layers of

discontinuous steel plates (the constant stiffness material) separated by, and bonded to, layers

of a variable stiffness, polyurethane-based shape memory polymer (SMP). The gaps between

the steel plates in a layer of this composite were positioned to be ‘out of phase’ with those in

adjacent layers so that the membrane and flexural stiffnesses of the composite are a function

of the shear distortion of the SMP in the overlap regions between the steel plates in adjacent

plies. On heating to temperatures above the transformation temperature of the SMP, the shear

modulus of the SMP reduced significantly and so the stiffness of the composite was also

reduced. High losses (up to 99%) in storage modulus were measured.

McKnight and Barvosa-Carter subsequently patented concepts for variable stiffness

structures [11]. The patent includes various configurations using combinations of constant

and variable stiffness materials; among these is an interleaved laminate form consisting of

continuous constant stiffness layers separated by, and bonded to, variable stiffness interleaf

layers. Fig. 1 illustrates, for an interleaved laminate consisting of carbon fibre reinforced

polymer (CFRP) layers and thermoplastic interleaves, how the flexural stiffness of the

laminate depends on the shear stiffness of the interleaf material. The thermoplastic interleaf

material is chosen so that its glass transition temperature (Tg-t) is less than that of the fibre

reinforced composite plies (Tg-c). At temperatures less than Tg-t the laminate is in a high

flexural stiffness state but when the temperature is increased to above Tg-t (but less than Tg-c)

the loss of shear stiffness of the interleaf layers results in a reduced flexural stiffness [12].

Maples et al. [12, 13] have conducted preliminary experimental investigations of a

polystyrene (PS)-interleaved carbon fibre reinforced epoxy composite, which indicated that

large reductions of over 90% in flexural stiffness are possible. In addition to simply

controlling the flexural stiffness, Raither et al. [14] were able to demonstrate that the bend-

twist coupling could be reduced by a factor 10 in a CFRP multidirectional laminate

containing elastomer interleaf layers when heated above the Tg of the elastomer.

Controllable stiffness interleaved configurations have also been examined by Ghandi et al.

[15, 16] who investigated a laminated beam consisting of aluminium plates separated by

layers of cast acrylic or polyvinyl chloride (PVC). Ultra-thin electric heating blankets were

embedded into the polymer layers to heat them through Tg and the resulting flexural stiffness

were reported to reduce by a factor of between two and four in initial experiments, depending

on the geometry. Subsequent finite element modelling has shown that much greater

reductions (over 95%) can be achieved in other interleaved beam configurations [17].

2

The interleaved laminate strategy uses a variable stiffness interleaf layer to control the

relative shear displacement between adjacent constant stiffness layers. Bergamini et al. [18]

took an alternative approach to control the shear displacement by using electrostatic coupling.

A laminated beam consisting of CFRP and glass fibre reinforced polymer layers was

manufactured in which the layers were not bonded together but an electric field was applied

through the thickness to prevent or allow relative shear displacement between adjacent layers.

Experiments were performed to demonstrate that this approach could be used to adjust the

flexural stiffness of such a beam to suppress vibration due to resonance.

Another strategy to control the stiffness of a material by exploiting the temperature dependent

shear stiffness of a polymer layer has been investigated by us [19, 20]. A controllable

stiffness composite was manufactured, which consisted of polyacrylamide (PAAm) coated

carbon fibres in an epoxy matrix. A current was passed through the carbon fibres to heat the

PAAm interphase through its Tg and so permit relative shear displacement between fibres and

matrix. This resulted in an 88% reduction in flexural stiffness of the composite. The flexural

stiffness was fully recovered when the composite was cooled to room temperature.

This paper presents a detailed investigation of carbon fibre reinforced epoxy laminates

containing PS interleaves. Simple beam theory is used to predict the flexural behaviour of the

materials and this is compared to the performance observed in the tests.

2. Predictions of flexural stiffness and strength

Simple beam theory can be used to approximate the bending behaviour of the interleaved

composites [12, 18]. At temperatures T, where T < Tg-t, the elastic modulus of the

thermoplastic interleaf (PS in this case) will be very small when compared to the fibre

direction stiffness of the unidirectional (UD) CFRP plies but is assumed to be sufficiently

large to ensure that the composite plies act as an integral structural element and that sections

initially plane and normal to the axis of the beam remain so when the beam is flexed. At a

temperature T, where Tg-t < T < Tg-c, the stiffness of the PS is so low that the CFRP plies act

as independent structural elements able to effectively slide freely relative to each other so that

initially plane sections no longer remain plane.

2.1 Analysis of room temperature bending behaviour of interleaved composite containing 0°

plies

For a symmetric layup, the apparent flexural modulus, EfRT, of the beam material at room

temperature assuming the beam is homogenous, can be calculated using Eq. 1, where Ec is the

elastic modulus of the composite ply in the 0° direction. The definitions of the terms are

shown in Fig. 2. Note that a composite layer can consist of more than one composite ply.

3

E fRT=

12E c

h3 ∑i=1

N ( ti3

12+ ti zi

2) (1)

An expression for the apparent flexural strength, σ*, can also be derived where, again,

‘apparent’ indicates that this is the strength of the beam if it is treated as homogenous.

σ ¿=

12 σu ∑i=1

N ( ti3

12+ ti . Z i

2)Composite layers only

h3

(2)

Using σu as the longitudinal flexural strength measured in a pure CFRP specimen, Eq. 2 gives

the predicted apparent flexural strength, σ*, for an interleaved composite.

2.2 Analysis of high Temp. bending behaviour of interleaved composites containing 0° plies

Assuming the layers are free to slide, the apparent flexural modulus at a T above Tg of PS but

less than that of the epoxy of the interleaved composite is given by Eq. 3.

E fHT=

Ec∑i

N

t i3

Composite layers onlyh3

(3)

Failure will occur when the maximum stress in any of the CFRP layers exceeds the strength.

Since all layers can be assumed to have the same curvature about their own centroidal axis

(i.e. no extension or compression at the mid-plane of each CFRP layer) then the highest stress

occurs in the thickest CFRP layer of thickness ti|max. Eq. 4 can therefore be derived for the

apparent flexural strength, σ*, at high temperature.

σ ¿=

σ u∑i=1

N

t i3

Composite layers onlyh2 ti∨max

(4)

in which σu is again taken as the measured longitudinal flexural strength of the pure CFRP

specimen.

3. Experimental details

3.1 Materials

Carbon fibre-epoxy UD prepreg (HexPly, 914C-TS-5-34%) was supplied by Hexcel (UK).

PS pellets (ST316310) were purchased from Goodfellow (UK). Tg of the epoxy was 196°C

and that of the PS was much lower at around 100°C.

3.2 Fabrication of polystyrene films

Composite layers only

4

5 g of PS pellets were distributed evenly in a circular pattern (75 mm diameter) in the centre

of a layer of polyimide release film (250 mm × 250 mm × 0.025 mm, Upilex-25S, UBE,

Osaka, Japan). Two more release film layers of the same size but containing a central 190

mm diameter circular cut-out were aligned on top of the first film. Finally a complete layer of

release film was placed on top to enclose the PS pellets within the release film layers. The

release film assembly was then placed between two stainless steel plates (250 mm × 240 mm

× 3 mm) and put into a hot press (Model 4126 Manual, Hydraulic Press, Carver, Indiana,

USA) for 10 mins at 250 °C without pressure then at 3 ton for 40 minutes. The resulting PS

film was 126 ± 9 μm thick. A 240 ± 20 μm thick PS film was produced using the same

method described above except that a pressure of 2 ton was applied for 60 minutes. A 73 ± 10

μm thick PS film was manufactured by pressing 2 g PS pellets at 3 ton for 40 minutes. Lastly

a 113 ± 25 μm thick PS film was fabricated by pressing 10 g of PS pellets at 3 ton for 40

minutes, except that a 220 mm diameter circular cut-out had been removed from the centre of

the two inner release film layers.

3.3 Fabrication of composite laminates

A pure CFRP laminate and four interleaved composite laminates (A, A*, B and C) were

manufactured. Composite A consisted of 8 PS layers (126 ± 9 μm thick) and 9 UD carbon

fibre reinforced epoxy plies (nominal cured ply thickness = 126 μm) arranged in an

alternating sequence (Fig. 2). Composite A* was manufactured for tensile testing and had the

same alternating sequence as Composite A but with fewer layers. It consisted of 3 layers of

PS (113 ± 25 μm thick) and 4 UD carbon fibre reinforced epoxy plies ordered in an

alternating sequence. Composite B contained three layers of PS (240 ± 20 μm thick) and four

blocks of carbon fibre reinforced epoxy (each block consisting of 3 prepreg plies) arranged in

an alternating sequence as shown in Fig. 2. Composite C consisted of three layers of PS (73 ±

10 μm thick) and four blocks of carbon fibre reinforced epoxy (each block consisting of 4

prepreg plies) also arranged in an alternating sequence (Fig. 2). Two pure CFRP 0° control

laminates were manufactured. One consisted of 17 plies of CFRP and was used for

comparison with the mechanical property testing of laminates A, B and C, and also to

provide measurements of Young’s modulus and flexural strength needed for the beam theory

predictions. The other control laminate consisted of 8 CFRP 0° plies and was to be used for

comparison with the tensile tests performed on laminate A*. All the composite plates were

100 mm long in the 0° fibre direction and 150 mm wide, except the tensile test specimen

plates that were 210 mm long in the 0° fibre direction and 120 mm wide. The laminates were

cured in a hot press (G969, George E Moore & Sons, UK) at 175°C and 100 psi for 1 h. The

5

thickness of the thick and thin 0° control laminates were 2.14 ± 0.06 mm and 0.99 ± 0.02

mm, respectively. The cured thickness of composite panels A, A*, B and C were 2 ± 0.04

mm, 0.81 ± 0.02 mm, 2.12 ± 0.05 mm and 2.22 ± 0.07 mm, respectively. The thickness of the

CFRP plies within the interleaved laminates remained 126 μm, but the PS layer thickness had

decreased. Using the total laminate thickness and the CFRP ply thickness the average

thickness of the PS layers for laminates A, A*, B and C was 109 μm, 102 μm, 196 μm and 68

μm, respectively. The change in thickness of the PS layers during curing is due to flow of the

material at high temperature and pressure causing flushing from the laminate edges. To determine whether our interleaved composite concept can be integrated into a structural

solution interleaved composites with ‘constrained’ interleaf layers were manufactured. A

special interleaved laminate in which the interleaf layers were only present in the central 28

mm of the flexure specimen (Fig. 3) was prepared. This length was equal to the beam span in

order to maintain a span-to-thickness ratio of approximately 32:1. The interleaved zone of the

specimen consisted of 3 PS layers (126 ± 9 μm thick) and 4 CFRP plies arranged in an

alternating sequence (i.e. same as laminate A*). Beyond the interleaved zone the PS layers

were replaced with CFRP plies. The thickness of the PS layers in the cured laminate was 109

μm. This laminate was termed ‘constrained’ as the CFRP plies were restricted from relative

sliding at the ends of the flexure specimen. This laminate was developed as there may be

applications in which the interleaved composites need to be clamped at elevated temperature

(e.g. in a bolted joint) without the risk of compressing the PS layers or loading only the outer

CFRP plies.

3.4 Preparation of DMTA, flexural and short beam shear tests specimens

Interleaved composites A, B and C and the 17-ply 0° control laminate were cut into 40 mm ×

5 mm coupons for DMTA, 80 mm × 10 mm coupons for flexural testing and 20 mm × 10 mm

coupons for interlaminar shear testing using a diamond bladed cutter (Diadisc 4200,

Mutronic, Germany). The constrained laminate was cut into 40 mm × 10 mm coupons. In all

cases the 0° fibre direction was parallel to the longer specimen dimension.

3.5 Preparation of tensile test specimens

The composite A* laminate (3 layers of PS and 4 UD carbon fibre reinforced epoxy plies)

and the 0° control laminate were trimmed to 200 mm × 100 mm using a diamond blade cutter

with the longer direction parallel to the 0° direction. Four end tab plates (100 mm × 50 mm ×

1.6 mm) of glass fibre reinforced epoxy with 45° chamfers were bonded using epoxy

adhesive (2014-1 Araldite, Huntsman, UK) onto the end regions the composite surface,

6

which had been previously grit blasted. The epoxy was cured under vacuum for 24 h at room

temperature. The resulting gauge length was 100 mm. The end-tabbed plate was then cut into

200 mm × 12 mm specimens using a diamond blade cutter and strain gauges (QFLA-2-11,

Strain Gauges, Tokyo Sokki Kenkyujo Co., Japan) glued onto one side of the composite at

the centre of the gauge length using adhesive (NP-50, TML Strain Gauge Adhesive, Tokyo

Sokki Kenkyujo Co., Japan).

3.6 Preparation of compression test specimens

Composite A was cut to the required dimensions (140 mm × 91 mm) using a diamond blade

cutter. In this case the 0° direction is parallel to the shorter direction of the plate. Four glass

epoxy laminate end tab plates (140 mm × 40 mm x 1.6 mm) with inverse chamfers were

glued onto the grit blasted composite surface using epoxy glue. To ensure the gauge length

remained at approximately 10 mm during the epoxy curing, a silicone rubber shim was

placed along the length of the gauge area to control the shape of the adhesive fillet. After

curing under vacuum for 24 h at room temperature the shim was removed and the composite

ground to a width of 90 mm using a grinder. The composite plate was then cut into 90 mm ×

10 mm specimens using a diamond blade cutter and strain gauges (QFLA-2-11) glued to both

sides of the composite at the centre of the gauge length using adhesive.

3.7 Testing procedures for mechanical properties

3.7.1 Evaluation of viscoelastic properties of composites

The viscoelastic behaviour of the composites was investigated using DMTA on a Tritec 2000

(Triton Technology Ltd., UK). Three-point bending mode testing was performed at 1 Hz and

with a gauge length of 15 mm. For each laminate type, three specimens were tested. A

heating rate of 5°C/min from room temperature up to 230°C was used to measure the storage

modulus (E'), loss modulus (E'') and tan δ of the composite samples.

3.7.2 Flexural testing

The flexural properties of the composites at both ambient and elevated temperatures were

determined using three-point bending tests according to ASTM D7264-07. The tests were

carried out on an Instron 4505 (Bucks, UK) using a 1 kN load cell. A span-to-thickness ratio

of 32:1 was used. The specimen dimensions and test parameters for each of the mechanical

tests are summerised in Table 1. The diameter of the loading nose and supports was 6 mm

and a crosshead speed of 1 mm/min was used. Measurements were taken at room temperature

and then at 120 °C using an environmental chamber (SFL, Eurotherm, UK). In the first test of

a specimen at room temperature (RT1), the specimen was loaded to a maximum deflection of

1 mm and then unloaded. The test was then repeated at 120°C in the environmental chamber

7

and then again at room temperature (RT2). The flexural strength was determined at room and

elevated temperature by loading samples to failure that had not previously been tested. For

some specimens of Composite A and the pure CFRP control sample a direct current was

passed through the carbon fibres as an alternative method of heating. To achieve this, the

upper and lower surfaces at each end of the samples were abraded with sand paper to expose

the carbon fibres and were then painted with silver conductive paint (186-3600, RS

Components, UK). Copper tape was then wrapped around the silver paint and the loose tape

ends connected to a power supply (HY3003-3, Digimess, UK). To heat a specimen to 120°C

ca. 1.8A and 3V were needed for the interleaved composite and 1.3A and 4V for the pure

CFRP composite.

To obtain the average flexural modulus and flexural strength of the composites, five

specimens of each layup were tested for each of the test temperatures.

The apparent flexural modulus, E, of the specimens was calculated using Eq. 5.

E= L3 m4 bh3 (5)

where L is the support span, b the beam width, h the beam thickness and m the gradient of the

load - displacement curve. The apparent flexural strength was calculated using Eq. 6.

σ= 3 PL2bh2 (6)

where P is the load at failure. The maximum shear stress τmax in the flexural test specimens

was calculated using Eq. 7.

τ max=P

2 (bh ). 3

2 (7)

3.7.3 Short beam shear testing

The interlaminar shear strengths (ILSS) of the composites were determined using short beam

shear tests according to ASTM D2344-00 at both room temperature and 120°C. An

environmental chamber was used to heat the composite to 120°C. To determine the average

ILSS of the composites, 5 specimens of each sample were tested. The tests were carried out

on an Instron 4505 using a 10 kN load cell. A span-to-thickness ratio of 4:1 was used. The

diameters of the loading nose and supports were 6 mm and 3 mm, respectively. A crosshead

speed of 1 mm/ min was used. The short beam shear strength was calculated using Eq. 7.

3.7.4 Tensile testing

Tensile tests were performed at room temperature and at 120°C (using a Eurotherm chamber)

on an Instron 4505 with a 100 kN load cell. The tests were carried out in accordance with

8

ASTM D3039-07. The testing speed was 2 mm/min. The tensile strength and modulus were

determined by loading 5 specimens to failure.

3.7.5 Compression testing

Compression tests were performed at room temperature and at 120°C according to the

Imperial College Compression Test method [21]. The tests were performed on an Instron

4505 with a 100 kN load cell. The testing speed was 1 mm/min. To calculate the average

compressive strength and modulus 5 specimens were tested.

3.8 Preparation of microsection specimens

Microsection specimens were prepared in order to analyse the cross-sections of the

composites. Composite samples were embedded into an epoxy resin (Epoxicure, Buehler,

UK) and cured for 8 h at room temperature. The samples were polished using a grinder-

polisher (MetaServ/MetaPol, Buehler, UK). Four sand papers with varying grit sizes were

used initially (Si P120, P320, P800 and P2500) followed by 3 diamond based dispersions (6

μm, 3μm and 1μm). The samples were then examined using an optical microscope (BX51M/

DP70, Olympus, Japan).

4. Results and Discussion

4.1 Viscoelastic properties of interleaved composites

The storage modulus E′ of the pure 17-ply carbon fibre reinforced epoxy composite remained

at approx. 123 GPa when heated from 30°C to 130°C (Fig. 3). Above this temperature E′

decreased because of epoxy resin Tg at 196°C. Composite A was heated from 30°C to 120°C

(Fig. 3). E′ at 30°C was 58 GPa. Tg of the PS, determined from the peak in the loss modulus,

was 103°C. Above the Tg, at 120°C, E′ dropped by 98% to 1.2 GPa. The test was repeated

three times on the same sample, from room temperature to 120°C, to determine whether the

process was reversible. The same stiffness loss and E′ at room temperature was observed in

each case, showing that under these test conditions the process was completely reversible. At

room temperature the storage modulus of Composite B was 96 GPa. Compared to Composite

A this modulus is considerably higher and can be explained by the greater content of CFRP

plies within Composite B. At 120°C the E′ dropped by 90% to 10 GPa. Similarly, Composite

C underwent an 87% loss in storage modulus from 102 GPa at 30°C to 13 GPa at 120°C.

4.2 Flexural properties of interleaved composites

As stated earlier, the flexural modulus of each sample was determined using three-point

bending tests at room temperature (RT1), at 120°C (using an environmental chamber) and

again at room temperature (RT2). Selected samples of the control composite and Composite

9

A were heated to 120°C by applying a direct current to the carbon fibres. Representative

load-displacement curves for the 0° control sample are shown in Figs. 4a, b. The gradients of

the lines are approx. the same despite the changes in temperature and heating technique.

Load-displacement curves for Composite A tested in three-point bending can be seen in Figs.

4c,d. The gradient is significantly lower at 120°C than at room temperature and fully returns

to its original value when cooled to room temperature. Again, the method of heating had an

insignificant effect on the behaviour at high temperature. Although a displacement of only 1

mm is shown, a maximum deflection of 17.8 mm was applied to one specimen at 120°C

without initiation of failure. When this deformed sample was cooled to room temperature

while under load and then unloaded the specimen retained nearly all the applied deformation

(Fig. 5). The composite then returned to its original shape (<0.5 mm residual deflection)

when heated back up to 120°C without load. This shape recovery occurred as the internal

stresses in the deformed reinforced layers were released as a result of the PS softening.

Composites B and C were heated to 120°C using the environmental chamber and not through

an applied current. Both of these composites exhibited a similar change due to temperature to

that observed for Composite A.

The flexural Young’s moduli (apparent moduli for interleaved composites) of the composites

are summarised in Table 2. The values are in good agreement with storage modulus values

obtained from DMTA. The flexural modulus of the pure CFRP was approx. 116 GPa at both

room temperature and 120°C. The flexural modulus of Composite A at room temperature was

65 GPa. At 120°C, above the Tg of PS, the modulus fell by 98% to 1 GPa. When re-tested at

room temperature no loss in flexural modulus could be seen. The values of flexural modulus

obtained from the environmental chamber and resistance heating are very similar suggesting

that an applied current does not adversely affect the mechanical properties of the composite.

However, PS could be seen flowing out at the ends of Composite A where the electrical

contacts were connected, as the temperature was significantly higher at these positions. This

problem did not occur when using the environmental chamber.

The flexural Young’s modulus of Composite B at room temperature was 90 GPa. At 120°C

this fell by 96% to 4 GPa and when re-tested at room temperature returned to 90 GPa. The

flexural modulus of Composite C at room temperature was 107 GPa. When heated to 120°C

the modulus dropped by 93% to 7 GPa. The modulus returned to its original value prior to

heating once cooled to room temperature.

10

Micrographs are shown in Fig. 5 of Composite B before testing and after cooling in a

deflected state achieved in a flexural test at 120°C. The PS interleaf layers can clearly be seen

between the reinforced layers. The deflected and cooled specimen showed that the reinforced

epoxy plies had slid relative to each other as a result of PS softening.

Load-displacement curves of the control sample and interleaved composites, tested to failure,

can be seen in Fig. 6. For Composite B the kink in the load displacement curve at approx. 400

N was taken as the point of failure. At room temperature the flexural strength of the 0°

control sample was 1550 ± 11 MPa and failed in compression at the mid-span. The apparent

flexural strengths of composites A, B and C were 966 ± 18 MPa, 894 ± 12 MPa and 987 ± 20

MPa, respectively (Table 3) and all exhibited interlaminar shear failure. Table 3 includes the

maximum interlaminar shear stress τmax in each of the composites at failure. The consistently

low τmax in the interleaved specimens indicated that the PS/epoxy bond was poor.

At 120°C the flexural strength of the 0° control sample dropped by 59 % to 636 ± 25 MPa

and the failure mode was compressive (as at room temperature). The flexural strength of the

interleaved composites could not be measured at 120°C since at large deflections the sides of

the loading nose fixture came in contact with the composite before failure occurred.

4.3 Flexural properties of constrained interleaved composites

At room temperature the flexural modulus and strength of the constrained laminate was 69 ±

6 GPa and 942 ± 31 MPa, respectively (Table 2 and 3). The maximum shear stress in

constrained laminate (13 MPa) was also similar to that measured in the standard interleaved

laminates A, B and C, all of which also failed in shear at room temperature. When the

constrained laminate was heated to 120 °C, the stiffness of the composite was reduced by 90

% to 7 ± 0.4 GPa. The percentage loss in stiffness when heated to 120 °C was smaller than

observed in the standard interleaved laminates because relative sliding of the CFRP plies is

prevented at the ends of the interleaved zone. Upon cooling, the stiffness of the constrained

laminate returned to its original value prior to heating. However, the constrained composite

shows that free edges allowing for the sliding of the composite plies (Fig. 5 bottom) are not

required to achieve a significant loss in stiffness. This also indicates that our composite with

controllable stiffness could be integrated into structural components. Unlike the standard

interleaved specimens, it was possible to measure the apparent flexural strength of the

constrained specimens at elevated temperature and this was found to be 122 ± 23 MPa. The

failure mode was delamination growth (at a small applied displacement of approximately

3mm) into the solid CFRP zone from one end of a PS interleaf. This failure mode would need

careful consideration in a real application.

11

4.3 Predicted flexural stiffness and strength of standard and constrained interleaved

composite specimens

The predicted bending behaviour (Table 2 and 3) of the UD interleaved composites at room

temperature and 120°C was calculated using beam theory (Eq. 1, 2, 3 and 4). The predicted

% loss in apparent flexural Young’s modulus for each composite was very similar to the

measured values. However, the predicted values of flexural modulus at room temperature are

higher than the measured values from flexural tests. For example, the measured flexural

modulus of Composite A at room temperature was 65 ± 3 GPa but the predicted value is 71.3

GPa. A possible explanation for this difference is that the simple beam theory used for the

predictions does not include shear distortion of PS layers. There are also differences between

the actual and predicted stiffness values at high temperature. The predicted values are lower

than the measured values at 120°C. For example, for Composite A, the measured flexural

modulus at 120°C is 1 ± 0.2 GPa and the predicted value is 0.3 GPa. This is because the

theory disregards the residual shear stiffness of the softened PS layers.

The apparent flexural moduli of the constrained composites at room and elevated temperature

were also predicted (Table 2) using simple beam theory [12]. Again the beam theory

overpredicts the stiffness at room temperature, presumably because the theory does not

include shear distortion of the PS layers, and underpredicts the modulus at high temperature

values because theory completely disregards the residual shear stiffness of the softened

thermoplastic layers.

Table 3 shows there are significant differences between the measured and predicted apparent

flexural strength of the interleaved composite specimens A, B and C. (As measured earlier,

apparent flexural strength refers to the strength that is calculated from the failure load

assuming the beam is homogeneous.) The measured flexural strength of the 0° control sample

was used to predict the apparent flexural strength of the interleaved composites. However, as

noted earlier, in all three cases the specimen failure mode was not due to the compressive

flexural stresses (as observed in the control specimen) but was caused by interlaminar shear

stresses. τmax (Table 3) show that the interlaminar shear stress in each of the interleaved

specimens at failure was around 15 MPa and indicate that the PS/epoxy interface was weak.

To demonstrate that higher flexural stresses could be achieved in the interleaved composites

if the interlaminar shear stress was reduced, a modified flexural specimen was manufactured

from Composite C and tested in 4-pt bending. The composite was bonded to aluminium

plates at either end of the specimen to increase its length. The loading span was larger than

the composite length to eliminate shear force in the composite part of the extended beam

12

specimen. The apparent bending strength of this extended specimen calculated using the

measured failure load was 1107 MPa but the failure mode was still due to interlaminar shears

induced in the composite at the junction with aluminium plates. However this apparent

bending strength is closer to the predicted value of 1474 MPa confirming that shear failure

was responsible for the low apparent flexural strength values in the standard flexural

specimens.

The apparent flexural strength of the interleaved laminates is predicted to dramatically reduce

at 120°C. For example, the apparent flexural strength of Composite A is predicted to reduce

by 97.6% from 936.4 MPa at room temperature to 22.6 MPa at 120°C. However, to reach

these stresses would require very large deflections which, as noted earlier, were not possible

with the three-point loading fixture used in these tests.

4.4 Interlaminar shear properties of interleaved composites

The ILSS at room temperature of composites A, B and C were 71 ± 6 MPa, 79 ± 4 MPa and

52 ± 2 MPa, respectively (Table 3). The ILSS of the 0° control sample was 111 ± 2 MPa. The

lower ILSS of the interleaved composites is associated with poor PS/epoxy adhesion at the

interface, where delamination took place. From flexural tests τmax in the 0° control sample

was calculated as 24 ± 0.2 MPa using Eq. 7 and no shear failure was observed, as expected.

For composites A, B and C τmax at failure of the flexural specimens was 15 ± 0.3 MPa, 14 ±

0.2 MPa and 15 ± 0.3 MPa, respectively (Table 3). These values are much smaller than the

measured ILSS values but for these interleaved laminates the failure mode, as noted in the

previous section, in all cases was interlaminar shear. It is known that the short beam shear test

can give artificially high values of ILSS due to the compressive forces that exist in this test

[22, 23]. ILSS testing was also performed on the interleaved composites at 120°C. However,

the composites did not fail because, due to the large deflections, these specimens became

locally compressed between the loading nose fixture and the support pins rather than being

loaded in the desired 3-point bending mode.

4.5 Tensile properties of interleaved composites

Both the interleaved composite and 0° control sample failed within the gauge length at room

temperature and exhibited brooming as result of fibre fracture. At 120°C, only the outer

CFRP plies of the interleaved composite failed as they carried the majority of the load.

Failure of these plies occurred at the border between the end tab and the gauge length. The

outer plies then slid with the attached end tabs. The large clamping force of the grips caused

PS and CFRP layers to extrude from around the sides of the end tab in the interleaved

specimens. The control sample failed at 120°C the same as it had done at room temperature.

13

The tensile properties of the interleaved composite (Composite A*, consisting of 4 0° CFRP

plies and 3 PS interleaves) and the 4-ply control sample at room temperature and 120°C are

shown in Table 4. At room temperature the Young’s modulus of the interleaved composite is

61% of that of the control specimen which is very close to the theoretically expected value

based on the 62% volume fraction of CFRP plies in the interleaved composite. However the

average tensile strength of the interleaved composite, at 52% of the average strength of the

control sample, is lower than expected from simple theory. This may be due to variations in

the quality of the interleaved composite as indicated by the significantly higher scatter in the

strength data of this material. The tensile properties of the control composite remained

unchanged at elevated temperature as previously observed [24]. The strength of the

interleaved composite at 120°C dropped by 55% of the room temperature value and this is

because the inner CFRP plies were effectively disconnected when the PS layers had softened.

However, at 120°C there was only a 38% drop in Young’s modulus of the interleaved

composite compared to the room temperature case. As noted earlier, the modulus was

determined using data from a central strain gauge, and the fact that the stiffness drop is less

than 50% indicates that at this distance from the specimen grips some load had transferred

across the softened PS layers to the inner CFRP plies of the interleaved composite. This also

suggests that if longer flexural specimens were tested then the stiffness loss at elevated

temperature may be less than predicted by the simple beam theory used here.

4.6 Compression properties of interleaved composites

At room temperature the both the control sample and interleaved composite A failed

catastrophically in the gauge length. At 120°C the control specimens failed catastrophically

in the gauge length but the interleaved specimens did not show any material failure and

instead buckled at a very low load.

The measured compression properties of the control sample and the interleaved composite A

at room temperature and 120°C are given in Table 4. The compressive Young’s modulus and

strength at room temperature for the control sample were 127 ± 3 GPa and 1264 ± 132 MPa,

respectively and for the interleaved composite were 66 ± 1 GPa and 786 ± 64 MPa,

respectively. The interleaved values of modulus and strength are 52% and 62% of the control

sample values, which, in view of the variability of the data, are reasonably close to the

expected value of 57% i.e. the volume fraction of CFRP plies in the interleaved specimen.

At 120°C the modulus of the control sample changed very little but the strength decreased by

48% to 657 ± 60 MPa. This loss of strength is expected due to the temperature dependence of

the matrix shear properties [25]. This reduction in compressive strength at high temperature

14

was studied previously [24]. A similar, but larger, strength reduction was observed in the 120

°C flexural tests of the control specimens, which also exhibited a compressive failure mode.

As noted earlier the interleaved specimens did not exhibit any failure but instead buckled at a

low load. As consequence no values for Young’s modulus and strength were determined.

5. Conclusions

We investigated the mechanical properties of PS-interleaved carbon fibre reinforced epoxy

composites with controllable stiffness. Such composites have potential for use in applications

such as morphing wings, where a stiff material is required that can be deformed on demand

with acceptable actuation forces.

Up to 98% reduction in storage modulus and flexural Young’s modulus could be achieved

when the interleaved composites were heated from room temperature to 120°C. When the

temperature was maintained at 120°C without an applied load, the composites returned to

their original shape prior to bending. When re-tested at room temperature, no loss in flexural

stiffness was observed. The flexural stiffness of the interleaved composites at room and elevated

temperatures were predicted using simple beam theory and were found to be in good agreement

with the measured values from three-point bending tests.

The room temperature compressive and tensile properties were largely as expected from the

volume fraction of the carbon reinforced plies in the interleaved material. At 120°C there was

a significant drop, as expected, in mechanical performance in compression and tension due to

the presence of the softened PS interleaves. The room temperature interlaminar shear strength

of the interleaved composites was observed to be reasonably good but the maximum bending

stress that could be achieved in three-point flexural tests of the interleaved material at room

temperature was found to be limited by premature interlaminar failure. This is most likely

due to poor adhesion between PS and epoxy composite plies and could be remedied either by

suitable surface treatment of PS layers or by use of another interleaving material with better

adhesion to the epoxy composite.

Acknowledgements

This research was supported financially through a CASE award by MBDA and the UK

Engineering and Physical Science Research Council (EPSRC). We thank Stephen Smith

(Imperial College, 2012) for preparing and analysing the constrained interleaved laminates.

References

15

1. Thill C, Etches J, Bond I, Potter K, Weaver P. Morphing skins. The Aeronautics Journal.

2008;112:117-39.

2. Henry C, McKnight G. Deformable variable-stiffness cellular structures, Patent US 7678440

B1; 2010.

3. Kornbluh RD, Prahlad H, Pelrine R, Stanford S, Rosenthal MA, von Guggenberg PA. Rubber

to rigid, clamped to undamped: Toward composite materials with wide-range controllable

stiffness and damping. In: Smart structures and materials 2004: Industrial and commercial

applications of smart structures technologies, San Diego, USA; March 2004.

4. Sofla AYN, Meguid SA, Tan KT, Yeo WK. Shape morphing of aircraft wing: Status and

challenges. Materials and Design. 2010;31:1284-92.

5. James T, Menner A, Bismarck A, Iannucci L. Morphing skins: Development of new hybrid

materials. In: Fourth SEAS DTC technical conference, Edinburgh, UK; July 2009.

6. Kuder IK, Arrieta AF, Raither WE, Ermanni P. Variable stiffness materials and structural

concepts for morphing applications. Progress in Aerospace Sciences. 2013;63:33-55.

7. Chen Y, Yin W, Liu Y, Leng J. Structural design and analysis of morphing skin embedded

with pneumatic muscle fibers. Smart Materials and Structures. 2011;20:085033.

8. Hinshaw TL. Analysis and design of a morphing wing tip using multicellular flexible matrix

composite adaptive skins. Masters Thesis. Virginia Polytechnic Institute; 2009.

9. Shan Y, Philen M, Lotfi A, Li S, Bakis CE, Rahn CD, Wang KW. Variable stiffness

structures utilizing fluidic flexible matrix composites. Journal of Intelligent Material Systems

and Structures. 2009;20:443-56.

10. McKnight G, Henry C. Variable stiffness materials for reconfigurable surface applications.

In: Smart structures and materials 2005: Active materials: Behavior and mechanics, San

Diego, USA; March 2005.

11. McKnight H, Barvosa-Carter W. Variable stiffness structure, Patent US 7892630 B1; 2011.

12. Robinson P, Maples HA, Gaite O, Smith S, Bismarck A. Carbon fibre reinforced epoxy

composites with variable stiffness for use in morphing aerostructures. ICCM19, Montreal,

Canada; July 2013.

13. Maples HA, Robinson P, Bismarck A. High performance carbon fibre reinforced polymer

composites with controllable stiffness: Can it be done? ECCM15, Venice, Italy; June 2012.

14. Raither W, Bergamini A, Gandhi F, Ermanni, P. Adaptive bending-twist coupling in

laminated composite plates by controllable shear stress transfer. Composites Part A: Applied

Science and Manufacturing. 2012;43:1709-16.

16

15. Gandhi F, Kang SG. Beams with controllable flexural stiffness. Smart Materials and

Structures. 2007;16:1179-84.

16. Murray G, Gandhi F. Multi-layered controllable stiffness beams for morphing: energy,

actuation force, and material strain considerations. Smart Materials and Structures.

2010;19:045002.

17. Gandhi F, Murray G, Kang SG. Flexural stiffness control of multilayered beams. AIAA

Journal. 2009;47:757-66.

18. Bergamini A, Christen R, Motavalli M. Electrostatically tunable bending stiffness in a GFRP-

CFRP composite beam. Smart Materials and Structures. 2007;16:575-82.

19. Tridech C. Smart fibre coatings: stiffness control in composite structure. Ph.D. Thesis.

Imperial College London; 2011.

20. Tridech C, Maples HA, Robinson P, Bismarck A. High performance composites with active

stiffness control. ACS Applied Materials and Interfaces. 2013;5:9111-19.

21. Häberle JG, Godwin EW. The Imperial College method for testing composite materials in

compression, Imperial College London, London; 1999.

22. Turner S. General principles and perspectives. In: Mechanical testing of advanced fibre

composites; Hodgkinson JM., Ed. Cambridge: Woodhouse Publishing; 2000. p. 27.

23. Feraboli PJ, Kedward KT. In search of the true interlaminar shear strength. In: Nineteenth

technical conference of the joint American society for composites/American society for

testing and materials committee D30, Atlanta, USA; October 2004.

24. Ankara A, Weisgerber D, Vilsmeir J. Influence of moisture on the mechanical properties of

graphite epoxy system. In: International Conference on Advanced Composite Materials and

Structures, Taipei, Taiwan; May 1986.

25. Soutis C, Turkmen D. Moisture and temperature effects of the compressive failure of CFRP

unidirectional laminates. Journal of Composite Materials. 1997;31:832-49.

17

Figure 1: Flexural stiffness control in an interleaved composite laminate

0�

0�

0�

0�

0�

0�

0�0�

0�

0�

0�0�

0�

0�0�

0�

0�0�

0�0�0�

0�

0�0�

0�

0�

0�0�

0�

0�

0�0�

0�

0�

0�0�

0�

Mid-plane of laminate

h/2

h/2

123

N

iMid-plane of layer i

zi

Composite ply

Polystyrene ply

b

ti

0 �

0 �

0 �

0 �

0 �

0 �

0 �

0 �

0 �

0�

0�

Figure 2: Cross-section parameters of interleaved composite and layups

Figure 3: Geometry and layup of the constrained laminate specimen

Composite A Composite B Composite C

18

50 100 150 2000

20406080

100120140

E/ G

Pa

Temperature/ °C40 60 80 100 120

0

20

40

60

E/ G

Pa

Temperature/ °C

E' tan

E''0.00

0.02

0.04

0.06

tan

E' tan

E''0.0

0.2

0.4

0.6

0.8

1.0

tan

(a) (b)

Figure 3: DMTA curves showing the storage modulus (E′), loss modulus (E″) and tan δ as a

function of temperature for (a) 17-ply 0° control sample (b) Composite A

0.0 0.2 0.4 0.6 0.8 1.0

0

20

40

60

80

100

120

140

160

0.0 0.2 0.4 0.6 0.8 1.0

0

20

40

60

80

100

120

140

160

Load

/ N

Displacement/ mm

Room temperature 1 120 °C Room temperature 2

Displacement/ mm

Room temperature 1 120 °C Room temperature 2

(a) 0° control sample (environmental chamber) (b) 0° control sample (resistance heating)

0.0 0.2 0.4 0.6 0.8 1.0

0102030405060708090

0.0 0.2 0.4 0.6 0.8 1.0

0102030405060708090

Load

/ N

Displacement/ mm

Room temperature 1 120 °C Room temperature 2

Displacement/ mm

Room temperature 1 120 °C Room temperature 2

(c) Composite A (environmental chamber) (d) Composite A (resistance heating)

Figure 4: Typical load-displacement curves from flexure tests on 17-ply control sample and

Composite A

19

Figure 5: Flexure specimen of Composite A before testing and after cooling in deflected state

(top), cross section of Composite B before flexure testing (bottom left) and in its deformed

state after cooling under load (bottom right)

0 1 2 3 4 5 60

100

200

300

400

500

600

700

800

Load

/ N

Displacement/ mm

Control sample Composite A Composite B Composite C

Figure 6: Load-displacement curves from flexure tests to failure of 0° control sample and

interleaved specimens at room temperature

CFRP ply

Polystyrene ply

20

Table 1: Specimen dimensions and geometrical parameters for mechanical tests

Composite h / mm†Geometrical parameters for mechanical tests / mm

Flexural SBS† Tensile Compressionl† b†† L** l b L†† l b lo

† l b lo

17-ply 0° control 2.14 ± 0.06 8

0 10 68.5 20 10 8.6 - - - 90 10 10

A 2.00 ± 0.04 80 10 64 20 10 8 - - - 90 1

0 10

B 2.12 ± 0.05 80 10 67.8 20 10 8.5 - - - - - -

C 2.22 ± 0.07 80 10 71.0 20 10 8.9 - - - - - -

Constrained 0.83 ± 0.04 40 10 26.6 - - - - - - - - -

8-ply 0° control 0.99 ± 0.02 - - - - - - 200 12 100 - - -

A* 0.81 ± 0.02 - - - - - - 200 12 100 - - -† h: sample thickness, l: sample length, b: sample width, SBS: Short Beam Shear, lo: gauge length

†† L: support span length. In the table the value for L is an average value. Please note that the actual support span length was always adjusted with respect to the actual sample thickness.

Table 2: Flexural moduli of the pure and interleaved CFRP composites.

CompositeApparent Flexural Modulus/ GPaEnvironmental Chamber Applied CurrentRT 1 120 °C % loss RT 2 RT 1 120 °C % loss RT 2

17-ply 0° control Measured 118 ± 3 112 ± 2 5.1 116 ± 3 113 ± 11 110 ± 8 2.7 114 ± 7

A Measured 65 ± 371.3

1 ± 0.20.3

98.5 65 ± 4-

68 ± 3-

1 ± 0.2-

98.5 66 ± 3-Predicted 99.6 -

B Measured 90 ± 197.4

4 ± 12.6

95.6 90 ± 2-

--

--

- --Predicted 97.3 -

C Measured 107 ± 7112.2

7 ± 0.025.2

93.5 106 ± 3-

--

--

- --Predicted 95.4 -

Con-strained

Measured 69 ± 6 7 ± 0.4 89.9 65 ± 5 - - - -Predicted 87.5 6.2 92.9 - - - - -

† RT1 & 2: first and second room temperature tests

21

Table 3: Flexural and interlaminar shear strengths of pure and interleaved CFRP composites.

Composite

Flexural tests SBS testsApparent Flexural Strength/ MPa

τmax at Failure/ MPa

ILSS/ MPa

RT 120 °C % loss RT RT17-ply 0° control Measured 1550 ± 11 636 ± 25 59.0 24 ± 0.2 111 ± 2

A Measured 966 ± 18 936.4

-22.6

- 15 ± 0.3- 71 ± 6Predicted 97.6

B Measured 894 ± 12 1279.9

-81.0

- 14 ± 0.2- 79 ± 4Predicted 93.7

C Measured 987 ± 20 1474.3

-131.5

- 15 ± 0.3- 52 ± 2Predicted 91.1

Con-strained

Measured 942 ± 31 122 ± 23 87 13 ± 0.3 -Predicted - - - -† RT: Room temperature test, SBS: Short beam shear

Table 4: Tensile and compressive properties of the pure and interleaved CFRP composites.

CompositeTensile CompressiveModulus/ GPa Strength/ MPa Modulus/ GPa Strength/ MPaRT 120 °C RT 120 °C RT 120 °C RT 120 °C

8-ply 0° control 145 ± 6 146 ± 13 1682 ± 67 1629 ± 94 - - - -

A* 88 ± 12 55 ± 6 867 ± 114 395 ± 62 - - - -17-ply 0° control - - - - 127 ± 3 136 ± 5 1264 ± 132 657 ± 60

A - - - - 66 ± 1 - 786 ± 64 -†RT: Room temperature test

22