lable at ScienceDirect

Composites Science and Technology 119 (2015) 26e33

Contents lists avai

Composites Science and Technology

journal homepage: http: / /www.elsevier .com/locate /compscitech

Mechanical properties of carbon fiber composite octet-truss latticestructures

Liang Dong*, Haydn WadleyDepartment of Materials Science and Engineering, University of Virginia, Charlottesville, VA 22903, United States

a r t i c l e i n f o

Article history:Received 16 June 2015Received in revised form1 September 2015Accepted 22 September 2015Available online 28 September 2015

Keywords:Octet-truss latticeElastic stiffnessStrengthCarbon fiber composite

* Corresponding author.E-mail address: ld5fy@virginia.edu (L. Dong).

http://dx.doi.org/10.1016/j.compscitech.2015.09.0220266-3538/© 2015 Elsevier Ltd. All rights reserved.

a b s t r a c t

Octet-truss lattice structures have been made from balanced [0/90] carbon fiber reinforced polymer(CFRP) laminates using a simple snap-fit method. The lattice structures moduli and strengths have beenmeasured during [001] and [100] directions free compressions as a function of the lattice relative density.Core failure occurred by either (i) Euler buckling ðr<5%Þ or (ii) delamination dominated failure ðr>5%Þof the struts. The measurements are shown to be well predicted by micromechanics models of thesecomposite strut failure modes. Snap-fit CFRP octet-truss lattice structures are found to exhibit me-chanical properties competitive with other cellular materials and topologies. Their isotropic responsemay provide new opportunities for ultra-lightweight multiaxial loaded structures.

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

The octet-truss lattice structure first proposed by Fuller [1]provides a method for filling 3-D space with a structurally effi-cient truss structure of an arbitrary cell size. This stretch-dominated structure has a high nodal connectivity of 12 [2], andan almost isotropic yield surface [3]. Recent alloy casting ap-proaches have demonstrated the possibility of making octet-trusslattices with strut lengths in the 5e10 mm range [3]. Wrought ti-tanium alloy octet-truss lattices have also been recently fabricated[4] while self-propagating waveguide or laser based stereo-lithographic methods, when combined with electroless nickelplating or vapor deposition, have enabled fabrication of micrometerscale structures [5e7].

Material property charts provide a useful way to compare themechanical properties of these low density materials. Fig. 1 com-pares the density dependent moduli and strengths of compres-sively loaded polymer and metal foams and lattice structures madeby investment cast of aluminum [3] and titanium alloys [8,9],electrodeposition of Nie7P [5,6], carbon fiber composites via areversible assembly technique [10], photosensitive HDDA polymers[6], and by the vapor deposition of alumina [6,7]. Recently reportedTie6Ale4V octet-truss lattice structures [4], balsa wood [11],

polymer [12] and metallic [13] syntactic foams are also included inthese Ashby maps. Foams have a low nodal connectivity, and arebending governed structures, and therefore compliant and weakerthan lattice topology counterparts. When made from high specificstrength materials, the octet-truss lattice is a highly weight effi-cient, multiaxial stress supporting structure, with a stiffness andstrength that scale linearly with the lattice relative density, r (thedensity of the lattice structure divided by that of the material fromwhich it was made) [14] if the struts failed by plastic deformation.

Carbon fiber composites (CFRP) have a high specific strengthand stiffness and is therefore a promising material for making stiffand potentially strong cellular structures. Here, we explore theapplication of a simple “snap-fit” assembly method [15] for fabri-cating octet-truss cellular materials from CFRP laminate materials.The [001] and [100] directions compressive properties of the octet-truss lattices have been characterized as a function of the latticerelative density and compared to micromechanical predictions.

2. Lattice fabrication and laminate material properties

The as-received [0/90] CFRP laminate sheets (McMaster-Carr)have a thickness t ¼ 1.59 mm, and a density 1.44 Mg/m3. Theycontained 55 vol% 228 GPa carbon fiber in a vinyl ester matrix. Thelaminate was comprised of 7 plies. The 2 surface plies were madefrom plain weave fabrics while the 5 interior plies were unidirec-tional and laid up in a [0/90]5 arrangement (Supplementaryinformation S.1). The laminate sheets were fabricated in such a

Fig. 1. Material property charts showing (a) stiffness and (b) strength versus densityfor octet-truss lattices and several other topology cellular structures.

L. Dong, H. Wadley / Composites Science and Technology 119 (2015) 26e33 27

way that the volume fraction of fibers along the laminate longitu-dinal direction (parallel to 0� unidirectional plies) is equivalent tothat of fibers along the laminate transverse direction (parallel to90� unidirectional plies). However, the 0� unidirectional plies had avolume approximately 1.5 times that of the 90� unidirectional plies.This volume difference was compensated by different volumes ofwarp and weft tows in the surface plies. Laminate sheets withwoven plies on the outer surfaces were selected for the presentstudy since they have better toughness and impact strength, andare less susceptible to delamination during machining operationssuch as drilling, side milling and slotting [16]. Thus, the use oflaminate sheets with woven surface plies was driven by amanufacturing constraint: the need to avoid delamination failuresduring fabrication and assembly of the lattice structures.

The CFRP octet-truss lattice structures were fabricated from theas-received [0/90] CFRP laminate sheets in a three step process,Fig. 2. The pyramidal truss row patterns, Fig. 2(a), and intermediatefaces, Fig. 2(c), were water jet and CNC mill cut separately from thelaminate sheets so that half the fibers were parallel to the strutsaxes. Note that the strut axis was chosen to be parallel either to thelaminates longitudinal or transverse directions. We subsequently

use “longitudinal strut” and “transverse strut” to denote octet strutswith axes parallel to the longitudinal and transverse directions ofthe laminate sheet from which they were cut. Rows of pyramidaltrusses were collinearly aligned and a second collinear arrayattached to their top, forming a [0/90] arrangement; Fig. 2(b). Thepyramidal truss layers were then snap-fitted into the crosses of theintermediate faces to form the octet-truss lattice; Fig. 2(d). AHYSOL®E-120HP™ (Loctite®Brand, Westlakes, OH) high strengthepoxy was finally applied to the nodal regions of the assembledstructure. Octet-truss lattice structures were fabricated with arelative density (octahedral cell relative density of the snap-fitlattice) ranging from 1.7 to ~16% by allowing the strut length l,defined in Fig. 2(a), to vary between 8 and 33 mm. All the latticestructures had a strut thickness t ¼ 1.59 mm (t ¼ w) and node di-mensions b ¼ 4.76 mm, c ¼ 2.24 mm, h ¼ 0.95 mm, htab ¼ 1.59 mm,t0 ¼ 1.27 mm, m ¼ 2.77 mm, R ¼ 5.08 mm, and u ¼ 45� (the geo-metric design variables shown in Fig. 2(a) and (c)). The expressionused to calculate the relative density, r of the snap-fit octahedralcell (Fig. 2(e)) is given in the Supplementary information S.2.Photographs taken at several orientations of the assembled sam-ples are shown in Fig. 3.

In order to predict lattice mechanical properties, the longitudi-nal and transverse compressive and tensile moduli and strengths ofthe laminate material were first determined (Supplementaryinformation S.3), and the laminate material was experimentallyconfirmed to be orthotropic material. Results are shown in Fig. 4(a)and summarized in Table 1. The laminate compressive strengthswere different in different loading conditions due to different fail-ure mechanisms: in CLC compression, failure was controlled byplastic fiber micro-buckling, whereas the failure was dominated bydelamination in unclamped compression; as observed optically(Fig. 4(c) and (d)) and confirmed by meXCT analysis (Fig. 4 (e) and(f)). In both cases, kink bands formed in the plies whose fibers wereparallel to the loading direction. The damage modes (kink bandsand delamination) were both initiated within the plain weavesurface plies where fiber misalignment was the greatest. This initialdamage can disturb the subsequent loading condition in unclam-ped compression by introducing bendingmoments at the specimenfree ends, and stress concentrations can also trigger delaminationor matrix cracking near the damage zones prior to plastic fibermicro-buckling of the interior unidirectional plies. In contrast, theCLC test fixture eliminated the end effects, allowing the interiorunidirectional plies along the loading direction to fully contributethe plastic fiber micro-buckling strength, and this resulted in ahigher compressive strength.

3. Compression responses

The compressive stressestrain responses of the CFRP octet-trusssamples are measured (Supplementary information S.4) and shownin Fig. 5(a) and (b). In all cases an initial linear response is observedfollowed by a regime of nonlinear responses. Typically, the peakstress was attained as strut failure was first observed. The stressthen decreased rapidlywith increasing strainwith serrations on thestressestrain curve associated with a series of strut failures. Pho-tographs of some of the lattice structures after testing are shown inFig. 5(c)e(g).

The lowest relative density sample ðr ¼ 1:7%Þ failed by strutsEuler buckling. Samples of higher densities failed by delaminationof the struts supporting compressive loads. Delamination typicallyinitiated near the ends of the struts and then propagated along thestrut axis. During [001] direction free compression, the interme-diate faces supported tensile stresses and remained intact after thetests (Fig. 5(c)e(e)) since the tensile strength of the laminate ma-terial is much higher than that in compression. During [100]

Fig. 2. Schematic illustration of the truss fabrication and “snap-fit” assembly method for making an octet-truss lattice from the [0/90] CFRP laminate sheets. The geometries of (a)pyramidal truss and (b) intermediate face with relevant core design variables are identified. (e) An octahedral cell with a Cartesian co-ordinate system and miller index loadingdirections.

Fig. 3. Photographs of snap-fit CFRP octet-truss specimens. (a) View in the [001] compression loading direction, (b) a side view showing the pyramidal trusses and (c) an isometricview of a sample ðr ¼ 5:4%Þ. (d) An isometric view of a sample ðr ¼ 1:7%Þ with a higher strut slenderness ratio (l/t). (e) Shows a close-up of an assembled node at the side of ther ¼ 5:4% sample. Sample size is defined by side length L(s), and height H(s).

L. Dong, H. Wadley / Composites Science and Technology 119 (2015) 26e3328

direction free compression, intermediate faces struts supportedcompressive loads, and eventually failed by delamination (Fig. 5(f)and (g)). Nodal rotations were observed after the peak stress wasachieved and the loading symmetry at the nodes was broken.Table 2 summarizes the compressive moduli and strengths of theCFRP octet-truss specimens as a function of their relative densitiesunder both [001] and [100] directions free compressions.

4. Analytical models for compression responses

Deshpande et al. [3]. have analyzed an ideal octahedral cell(cubic symmetrywith vanishing node volume)made from isotropicmaterials, and the results apply to the octet-truss lattice con-structed by 3-D stacking of such an octahedral cell. For small t/l, thecontribution to overall lattice stiffness from struts bending is

Fig. 4. (a) Uniaxial stressestrain responses of the [0/90] CFRP laminate material along one of the fiber directions (“L” ¼ longitudinal; “T” ¼ transverse). Results from both theCelanese compression (CLC) and unclamped compression are included. (b) The geometry of the laminate specimen used for unclamped compression testing. (c) Delaminationfailure after unclamped compression. (d) Plastic fiber micro-buckling failure after CLC compression. m-XCT images of the laminate showing the failure modes after (e) unclampedand (f) CLC compressions.

L. Dong, H. Wadley / Composites Science and Technology 119 (2015) 26e33 29

negligible compared to that from struts stretching [3], and thus theDFA model assumed pin-joined struts to simplify the analysis. Weemployed the same assumption for the analysis of the snap-fit CFRPoctet-truss lattice manufactured here.

The octahedral cell designed in the present study contains extranodal masses, as shown in Fig. 2(e), and has a transverse isotropicsymmetry due to its nodal geometry. The linear elastic stress, s andstrain, ε tensor relationship takes the form (with Cartesian indices);

26666664

sxxsyyszzsyzsxzsxy

37777775¼

266666666666664

Cxxxx Cxxyy Cxxzz 0 0 0

Cxxyy Cxxxx Cxxzz 0 0 0

Cxxzz Cxxzz Czzzz 0 0 0

0 0 0 Cyzyz 0 0

0 0 0 0 Cyzyz 0

0 0 0 0 0Cxxxx � Cxxyy

2

377777777777775

�

26666664

εxxεyyεzzεyzεxzεxy

37777775

(1)

with the principal material axes (x,y,z) are defined in Fig. 2(e).There are five independent elastic stiffness constants, Cij (the

contracted indices i and j are ordered pairs of Cartesian indices),fC11; C12; C13; C33; C44g, equivalent to fCxxxx; Cxxyy; Cxxzz; Czzzz;Cyzyzgwith contracted indices. In [001] direction free (or confined)compression, 1=Szzzz (or Czzzz) is determined, while in [100] di-rection free (or confined) compression, 1=Sxxxx (or Cxxxx) isdetermined.

The octahedral cell shown in Fig. 2(e) contains 4 longitudinaland 4 transverse struts that are loaded in compression. The unit cellcompressive stiffness therefore depends on both the longitudinaland transverse struts compressive moduli, whereas the unit cellcompressive strength will be governed by the compressive strengthof the (weakest) transverse struts. Since the longitudinal strutshave higher moduli than the transverse struts, these octet strutswill suffer different displacements when the octahedral cell is un-der free compression, and static equilibrium is achieved by(compensating) nonuniform nodal displacements. However, therequirement for static equilibrium dictates that all the octet strutshave the same internal stress when the octahedral cell is under freecompression. To simplify the analysis, we assumed that all thestruts have the same compressive, EaveC and tensile, EaveT moduli,where EaveC ð¼ 1=2ðELC þ ETCÞ � 26GPaÞ and EaveT ð¼ 1=2ðELT þ ETT Þ �60GPaÞ are the averages of longitudinal, and transverse compres-sive and tensile moduli of the laminate material, respectively.

4.1. Compressive modulus

As shown in Supplementary information S5, the octahedral cellcompressive modulus Ezz under [001] direction free compression is

Table 1The measured properties of the as-received [0/90] CFRP laminate along both the longitudinal and transverse directions. Standard deviations are given based on multiplemeasurements for each mechanical property.

Compressive modulus(Unclamped)

Compressive modulus(CLC)

Micro-bucklingstrength (CLC)

Delamination strength(Unclamped)

Tensile strength Tensile modulus

ECðUCÞ (GPa) ECðCLCÞ (GPa) smb (MPa) sdl (MPa) sT (MPa) ET (GPa)

Longitudinal direction (L) 32.5 ± 1.8 (15 times) 33.6 ± 2.1 (7 times) 640 ± 36 (7 times) 457 ± 76 (21 times) 949 ± 34 (6 times) 76 ± 3.5 (7 times)Transverse direction (T) 19.8 ± 2.3 (15 times) 20.5 ± 1.4 (7 times) 428 ± 42 (6 times) 305 ± 63 (28 times) 497 ± 10 (5 times) 45 ± 1.9 (5 times)

L. Dong, H. Wadley / Composites Science and Technology 119 (2015) 26e3330

Ezz ¼2EaveT EaveC�

EaveC þ 2EaveT

� t2H½001�lA½001�

(2)

where H½001� ¼ 2ðl sin uþ hþ 2htabÞ is the [001] direction octahe-dral cell height, A 001½ � ¼ 2l cos uþ bþ cð Þ2=2 is the cross-sectionalarea perpendicular to the [001] direction.

If we define Es ¼ EaveC , Equation (2) can be expressed in the formof relative compressive modulus as

Fig. 5. Compressive stressestrain responses of CFRP octet-truss samples of various relativcompression: (c) Pyramidal strut delamination in the r ¼ 5:4% specimen, and (d) a higher mlowest density ðr ¼ 1:7%Þ specimen. (f) Failure mechanism under [100] direction free compface strut.

EzzEs

¼ 2EaveT�EaveC þ 2EaveT

� t2H½001�lA½001�

¼ KEt2H½001�lA½001�

(3)

with KE ¼ 2EaveT =ðEaveC þ 2EaveT Þ. It is noted here that Equations (2)and (3) were derived assuming free compression; in lateralconfined compression KE ¼ 1.

Similarly, it can be shown that the unit cell compressivemodulus Exx under [100] direction free compression is

e densities. (c), (d) and (e) show the failure mechanisms under [001] direction freeagnification view of the delamination. (e) Euler buckling initiated strut fracture in theression ðr ¼ 5:4%Þ, and (g) a higher magnification view of a delaminated intermediate

Table 2Relative densities, experimental and unit cell compressive moduli and compressive strengths under [001] and [100] directions compression of the manufactured snap-fit CFRPoctet-truss specimens.

Length(l, mm)

Relativedensity ðrÞ

Sample compressivestiffness (GPa)

Unit cell compressivestiffness (GPa)

Sample compressivestrength (MPa)

Unit cell compressivestrength (MPa)

[001] [100] [001] [100] [001] [100] [001] [100]

33.020 1.7% 0.075 0.107 0.060 0.079 0.73 0.75 0.60 0.5716.891 5.4% 0.295 0.301 0.237 0.230 4.39 4.01 3.69 3.1412.014 9.4% 0.556 0.546 0.447 0.422 7.98 7.01 6.81 5.609.728 13.0% 0.753 0.773 0.606 0.602 9.90 9.53 8.54 7.718.433 15.9% 0.983 1.033 0.792 0.811 11.39 10.95 9.89 8.95

L. Dong, H. Wadley / Composites Science and Technology 119 (2015) 26e33 31

Exx ¼2EaveT EaveC�

EaveC þ 2EaveT

� t2H½100�lA½100�

(4)

and the corresponding relative compressive modulus

ExxEs

¼ 2EaveT�EaveC þ 2EaveT

� t2H½100�lA½100�

¼ KEt2H½100�lA½100�

(5)

where, H½100� ¼ffiffiffi2

plþ bþ c, is the [100] direction octahedral cell

height, A½100� ¼ ðffiffiffi2

plþ bþ cÞðl=

ffiffiffi2

pþ hþ 2htabÞ is the cross-

sectional area perpendicular to the [100] direction,KE ¼ 2EaveT =ðEaveC þ 2EaveT Þ. Again, KE ¼ 1, in the case of [100] direc-tion lateral confined compression.

4.2. Compressive strength

The octahedral cell compressive strength under [001] directionfree compression is given by (Supplementary information S.5)

spkzz ¼ 2

ffiffiffi2

pt2

A½001�sc (6)

where, sc is the collapse strength of a single strut. If we define ss ¼savemb , where savemb ¼ 1=2ðsLmb þ sTmbÞ � 535MPa, the average oflaminate longitudinal and transverse compressive strengths,Equation (6) can be expressed in the form of a relative compressivestrength

spkzz

ss¼ 2

ffiffiffi2

pt2sc

savembA½001�¼ Ks

t2scA½001�

(7)

Where, Ks is a constant ¼ 2ffiffiffi2

p=savemb .

Similarly, it can be shown that the [100] direction unit cellcompressive strength is

spkxx ¼ 2

ffiffiffi2

pt2

A½100�sc (8)

the relative compressive strength in this direction is

spkxx

ss¼ 2

ffiffiffi2

pt2sc

savembA½100�¼ Ks

t2scA½100�

(9)

where, Ks is a constant ¼ 2ffiffiffi2

p=savemb .

During either [001] or [100] direction compression, a CFRPoctet-truss lattice can collapse by either (i) Euler buckling (ii)delamination or (iii) plastic fiber micro-buckling of the struts. Thecollapse strength of a single strut due to different collapse mech-anisms can be found in Supplementary information S.6.

5. Comparison of measurements and predictions

Themanufactured samples have extra edge struts that belong tothe unit cells of a larger area sample. These edge struts of partialunit cells contribute both stiffness and strength to the samplesmechanical response. The measured compressive properties werethus adjusted to account for the edge effect in order to makecomparison with modeled properties. The properties of a latticewithout extra edge struts are subsequently defined as the unit cellstiffness or strength (Supplementary information S.7), and aresummarized in Table 2.

The measured unit cell compressive elastic moduli in both [001]and [100] directions are plotted against the relative density, r, inFig. 6(a) and (b), and the compressive strength data in Fig. 6(c) and(d). The model predictions are also plotted on all the figures. Therelative moduli and strengths were normalized by the averagecompressive modulus Es ¼ EaveC ð26GPaÞ and strength governed byplastic fiber micro-buckling failure, ss ¼ savemb ð535MPaÞ of thelaminate material. The delamination and plastic fiber micro-buckling models used conservative transverse compressivestrengths, sTdlðminÞ ¼ 240MPa and sTmbðminÞ ¼ 386MPa, of thelaminatematerial for predictions. An average compressivemodulusEaveC ð26GPaÞ of compression struts was used for the Euler bucklingfailure model predictions.

In Fig. 6(a) and (b), results are shown for the relativecompressive moduli in both laterally confined compression (Czzzz/Es, equivalent to C33/Es for [001] compression and Cxxxx/Es,equivalent to, C11/Es for [100] compression) and free compression((1/Szzzz)/Es, that is (1/S33)/Es for [001] compression and (1/Sxxxx)/Es, or (1/S11)/Es for [100] compression) using Es ¼ EaveC ð26GPaÞ. Theunit cell data is found to be well predicted by the free compressionmodels for [001] direction compression, but is slight under-estimated by the free compression model in the case of [100] di-rection compression. In [100] direction free compression, theunmodelled flexural stiffness of the intermediate faces providedlateral constraint in such a loading configuration. Also, the nodaldesign of the intermediate faces improved the nodal rotationalstiffness, and k was therefore likely to lie between 1 and 2 for theactual struts end conditions.

The experimental data of relative compressive strengths(Fig. 6(c) and (d)) follow the Euler buckling model at low relativedensity (r<0:054), and are then consistent with the delaminationmodel when r>0:054. However, the model predicted a failuremode transition from Euler buckling to delamination near rz0:07rather than the measured transition near rz0:054. This is attrib-uted to k ¼ 1 pin-joined struts model assumption; whereas, theprecise rotational stiffness of the designed nodes lies between 1and 2. Due to insufficient node constraint, the plastic fiber micro-buckling failure mode of the struts was not activated, and thelattice properties are inferior to the corresponding modelpredictions.

L. Dong, H. Wadley / Composites Science and Technology 119 (2015) 26e3332

6. Comparisons with other materials

Themeasured [001] direction unit cell compressive responses ofthe snap-fit CFRP octet-truss lattices were included in the materialproperty charts shown in Fig. 1. The properties of the snap-fit CFRPoctet-truss lattices are clearly superior to the counterpartsmade viathe reversible assembly technique [10], and lattices made by in-vestment casting using aluminum [3] and titanium alloys [8,9].They also out-perform electrodeposition deposited Nie7P [5],photosensitive HDDA polymers and vapor deposited alumina (withsolid trusses) [6], and are quite competitive with the Tie6Ale4V [4]and Al2O3-polymer hybrid octet-truss lattices [7].

For densities lower than 0.04 Mg/m3 (where Euler bucklingdominates), the compressive strengths of the snap-fit CFRP octet-truss lattices are inferior to electrodeposited alumina and Nie7Poctet-truss lattices with hollow trusses [6] due to a lower secondarea moment of the solid struts. Even so, it is evident that thepredicted compressive moduli of the snap-fit CFRP octet-truss lat-tices are superior to the hollow truss Nie7P octet-truss lattices [6]and very competitive with the hollow truss alumina octet-trusslattices [6] in the lower density range (<0.04 Mg/m3).

The model predictions clearly reveal the potential of CFRP octet-truss lattices for filling gaps in material property space. The me-chanical properties of a cellular material are determined by itsrelative density, by its parent material mechanical properties andby the cell topology. The strength of a stretch-dominated cellularlattice, spk, can be expressed in the general form [17];

Fig. 6. Comparisons between measured (symbols) and predicted relative compressive moddensity. Error bars represent the maximum and minimum values obtained from 3 experim

spk ¼ Sssr (10)

where, ss is the strength of the solid material fromwhich the latticeis made, r is the relative density of the lattice and S is a latticetopology dependent “strength coefficient”. The gap between thelinear extrapolation of lattice property data points and solid ma-terial property is attributed to this topology dependent strengthcoefficient S (since when, r ¼ 1, spk ¼ Sss). For an octet-truss lat-tice, if the node volume vanishes, S ¼ 1=3 [3]; but in the case of thesnap-fitted octet-truss lattice designed in the present study, extranode mass was present, and so the topology dependent strengthcoefficient becomes a function of the node geometry, and for thecases investigated here was less than 1/3.

The minimum cell size that can be produced by the snap-fitroute was dictated by the precision of the machining operation,by the variability in laminate thickness and by the tools used forthe assembly of the lattice. If one seeks to make samples ofsimilar relative density to those reported here, but with a smallercell size, it would be necessary to keep the width to length ratioof the struts fixed. This in turn would require a reduction in thethickness of the starting laminate (in proportion to the reductionin cell length). If the thickness of the plies used to fabricate thelaminate remained fixed, this would require a reduction in thenumber of plies. However, recently developed very thin plycomposite laminates could enable significantly thinner laminatesto be used [18].

uli and strengths of the snap-fit CFRP octet-truss lattices as a function of the relativeents.

L. Dong, H. Wadley / Composites Science and Technology 119 (2015) 26e33 33

It is possible that the delamination failure mode could be sup-pressed with the use of a higher shear strength matrix [19], or by astitching [20] or 3D weaving/braiding approach [21,22] to improvethe delamination resistance and increase the fraction of fibers inthe load carrying directions. Finally, if these strategies delay theonset of buckling and delamination failure modes, the use ofnanoscopic reinforcements of the polymer matrix [23,24], mightenable the stress at which the plastic fiber micro-buckling failuremechanism is activated to be raised.

7. Conclusion remarks

CFRP octet-truss lattice structures made from [0/90] CFRPlaminate sheets with relative densities ðrÞ in the range 1.7e16%have been successfully manufactured by employing a simple snap-fitting method. The lattice structures moduli and strengths havebeen characterized under both [001] and [100] directions freecompressions as a function of the relative density ðrÞ. The failuremechanism changed from Euler buckling to delamination ofcompression struts at a relative density of about 0.054. Thecompressive moduli and strengths of the lattice structures are wellpredicted by a micromechanics model adapted to account for extranode volume.

The CFRP octet-truss lattices exhibit mechanical propertiescompetitive with other isotropic materialecell topology combina-tions, and appear to be a promising candidate for ultra-lightweightload bearing applications. However, the current designs undergodelamination failures of the struts and do not achieve the full po-tential of composite cores as predicted by the plastic fiber micro-buckling failure models.

Acknowledgments

We are grateful to the DARPA MCMA program (Grant NumberW91CRB-10-1-005) for the financial support of this research (Pro-gram manager, Dr. Judah Goldwasser).

Appendix A. Supplementary information

Supplementary information related to this article can be foundat http://dx.doi.org/10.1016/j.compscitech.2015.09.022.

References

[1] R.B. Fuller, Octet truss. 1961, U.S. Patent No. 2, 986, 241.

[2] V.S. Deshpande, M.F. Ashby, N.A. Fleck, Foam topology: bending versusstretching dominated architectures, Acta. Mater. 49 (2001) 1035e1040.

[3] V.S. Deshpande, N.A. Fleck, M.F. Ashby, Effective properties of the octet-trusslattice material, J. Mech. Phys. Solids. 49 (2001) 1747e1769.

[4] L. Dong, V.S. Deshpande, H.N.G. Wadley, Mechanical response of Tie6Ale4Voctet-truss lattice structures, Int. J. Solids. Struct. 60 (2015) 107e124.

[5] A. Torrents, T.A. Schaedler, A.J. Jacobsen, W.B. Carter, L. Valdevit, Character-ization of nickel-based microlattice materials with structural hierarchy fromthe nanometer to the millimeter scale, Acta Mater. 60 (2012) 3511e3523.

[6] X. Zheng, H. Lee, T.H. Weisgraber, M. Shusteff, J. DeOtte, E.B. Duoss, et al.,Ultralight, ultra-stiff mechanical metamaterials, Science 344 (2014)1373e1377.

[7] J. Bauer, S. Hengsbach, I. Tesari, R. Schwaiger, O. Kraft, High-strength cellularceramic composites with 3D microarchitecture, Proc. Natl. Acad. Sci. 111(2014) 2453e2458.

[8] Q. Li, E.Y. Chen, D.R. Bice, D.C. Dunand, Mechanical properties of cast Ti-6Al-4Vlattice block structures, Metall. Mater. Trans. A 39 (2008) 441e449.

[9] Q. Li, E.Y. Chen, D.R. Bice, D.C. Dunand, Mechanical properties of cast Ti-6Al-2Sn-4Zr-2Mo lattice block structures, Adv. Eng. Mater. 10 (2008) 939e942.

[10] K.C. Cheung, N. Gershenfeld, Reversibly assembled cellular composite mate-rials, Science 341 (2013) 1219e1221.

[11] A. Da Silva, S. Kyriakides, Compressive response and failure of balsa wood, Int.J. Solids Struct. 44 (2007) 8685e8717.

[12] D. Klempner, K.C. Frisch, in: Handbook of Polymeric Foams and Foam Tech-nology, vol. 404, Munich etc.: Hanser, 1991.

[13] P.K. Rohatgi, N. Gupta, B.F. Schultz, D.D. Luong, The synthesis, compressiveproperties, and applications of metal matrix syntactic foams, JOM 63 (2011)36e42.

[14] H.N.G. Wadley, N.A. Fleck, A.G. Evans, Fabrication and structural performanceof periodic cellular metal sandwich structures, Compos. Sci. Technol. 63(2003) 2331e2343.

[15] K. Finnegan, G. Kooistra, H.N.G. Wadley, V.S. Deshpande, The compressiveresponse of carbon fiber composite pyramidal truss sandwich cores, Int. J.Mater. Res. 98 (2007) 1264e1272.

[16] J.K. Kim, M.L. Sham, Impact and delamination failure of woven-fabric com-posites, Compos. Sci. Technol. 60 (2000) 745e761.

[17] H.N.G. Wadley, Multifunctional periodic cellular metals, Philos. Trans. A Math.Phys. Eng. Sci. 364 (2006) 31e68.

[18] J.D. Fuller, M.R. Wisnom, Pseudo-ductility and damage suppression in thin plyCFRP angle-ply laminates, Compos. Part A. Appl. Sci. Manuf. 69 (2015) 64e71.

[19] B.D. Garland, I.J. Beyerlein, L.S. Schadler, The development of compressiondamage zones in fibrous composites, Compos. Sci. Technol. 61 (2001)2461e2480.

[20] K. Dransfield, C. Baillie, Y.W. Mai, Improving the delamination resistance ofCFRP by stitchingda review, Compos. Sci. Technol. 50 (1994) 305e317.

[21] T. George, V.S. Deshpande, K. Sharp, H.N.G. Wadley, Hybrid core carbon fibercomposite sandwich panels: fabrication and mechanical response, Compos.Struct. 108 (2014) 696e710.

[22] A.J. Malcom, M.T. Aronson, H.N.G. Wadley, Three-dimensionally woven glassfiber composite struts: characterization and mechanical response in tensionand compression, J. Compos. Mater. (2015), 0021998315569751.

[23] V.P. Veedu, A. Cao, X. Li, K. Ma, C. Soldano, S. Kar, P.M. Ajayan, M.N.G. Nejhad,Multifunctional composites using reinforced laminae with carbon-nanotubeforests, Nat. Mater. 5 (2006) 457e462.

[24] S.S. Wicks, R.G. de Villoria, B.L. Wardle, Interlaminar and intralaminar rein-forcement of composite laminates with aligned carbon nanotubes”, Compos.Sci. Technol. 70 (2010) 20e28.