You are on page 1of 6

Materials and Design 67 (2015) 159–164

Contents lists available at ScienceDirect

Materials and Design


journal homepage: www.elsevier.com/locate/matdes

Short Communication

Designable dual-material auxetic metamaterials using


three-dimensional printing
Kan Wang a,⇑, Yung-Hang Chang a, YiWen Chen b, Chuck Zhang a, Ben Wang a
a
H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA
b
3D Printing Medical Research Center, China Medical University Hospital, Taichung City 40447, Taiwan

a r t i c l e i n f o a b s t r a c t

Article history: Metamaterials are inherently advantageous in achieving designable auxeticity since their Poisson’ ratios
Received 19 September 2014 are determined by the geometry of their unit cells. A family of auxetic metamaterials was created by
Accepted 22 November 2014 computer-aided design (CAD) and dual-material three-dimensional (3D) printing. The effects of material
Available online 28 November 2014
selections and stiff material fraction on the Poisson’s ratio, equivalent Young’s Modulus, and maximum
volume reduction were investigated. The results from finite element analysis (FEA) and mechanical test-
ing indicated that the auxeticity and mechanical properties of this dual-material auxetic metamaterial
(DMAM) are distinctly different from those of traditional single-material auxetic metamaterials
(SMAMs). The interesting properties of DMAMs could be valuable to various engineering applications
such as smart materials, biomedical components, and shock-resistant components.
Published by Elsevier Ltd.

1. Introduction Elipe and Lantada have developed a computer-aided design


(CAD) library of auxetic geometries and compared Poisson’s ratios,
Most conventional materials have positive Poisson’s ratios, maximum area/volume reductions, and equivalent Young’s moduli
meaning when an extension is applied to the axial direction of of 25 two-dimensional (2D) patterns and 7 three-dimensional (3D)
the material, a contraction will be observed along the transverse patterns [19]. An interesting finding in this comparative study is
direction, and vice versa. This is rational since it reflects volume that there are trade-offs among those three characteristics. For
conservation. Auxetic materials are those that unexpectedly dis- example, the ‘Re-entrant cuboid’ has the largest negative Poisson’s
play a lateral expansion/contraction with applied longitudinal ratio at 1.7 while most typical values for other patterns range
stretch/compression [1–5] and [6]. Despite being counterintuitive, from 0.1 to 0.9. However, it also has the lowest maximum vol-
auxetic materials are not so uncommon and auxeticity has been ume reduction of 16.57% while others are around 30%. Similarly,
reported for a wide range of materials including metals [6], ceram- ‘Re-entrant tetracaidecahedron’ has the highest equivalent Young’s
ics [7], polymers [8], composites [9], fibers [10] and biological tis- modulus of 1.46 MPa comparing to values of other patterns from
sues [11]. While the theoretical lower limit of Poisson’s ratio for 0.37 to 0.74 MPa. At the same time its negative Poisson’s ratio is
isotropic solid material is 1, metamaterials can exceed the limit only 0.053. In other studies, the effects of geometrical changes
by manipulating the geometry of the unit cell. on basic auxetic patterns have been systematically investigated,
Metamaterials were first introduced as novel electromagnetic including cell shape [20], beam alignment [21], and angles at
(EM) materials and their characteristic structural length is one or hinges [22]. Nevertheless, the attempt to increase one specific
more orders smaller than the EM wavelengths [12,13]. The concept property means there are compromises in other properties. This
of metamaterials has recently been extended to include any mate- is inherently true for single-material auxetic metamaterials
rial whose effective properties are delivered by its structure rather (SMAMs) since there are underlying contradictions in structural
than the bulk behavior of the materials that composed it [14]. design. Higher equivalent Young’s modulus implies stiffer beams/
Much research has been conducted to build auxetic metamaterials walls that essentially mean thicker beams/walls if it is built from
from conventional materials, such as SU-8 epoxy [14], liquid crys- a single material. On the other hand, the auxeticity of metamaterial
talline polymers [15], metal [16], composite [17] and foam [18]. results from the flexibility of the joints, i.e., the capability of the
angle change between two connected beams/walls (Fig. 1a). As
⇑ Corresponding author at: 813 Ferst Drive, Room 375, Atlanta, GA 30332, USA. the thickness of beams/walls increases, one can either simulta-
Tel.: +1 8505246420. neously increase the thickness at the joints, which lowers the flex-
E-mail address: kan.wang@gatech.edu (K. Wang). ibility (Fig. 1b), or keep the joints at thinner geometries, which

http://dx.doi.org/10.1016/j.matdes.2014.11.033
0261-3069/Published by Elsevier Ltd.
160 K. Wang et al. / Materials and Design 67 (2015) 159–164

leaves them as structural weak points when the entire structure is


deformed (Fig. 1c). Both scenarios result in a decrease in negative
Poisson ratio or maximum area or volume reduction.
The introduction of dual-material 3D printing has offered an
opportunity to solve this dilemma. Connex™ series 3D printers,
which are using PolyJetÒ technology developed by Objet, manufac-
tured by Stratasys Ltd., are capable of printing two types of mate-
rials simultaneously in the same part. The mechanical properties of
their printable materials vary from rigid (VeroX family) to rubber-
like (TangoX family). It also allows mix-printing of any two mate-
rials to create so called ‘digital materials’ to deliver any mechanical
properties in between. By assigning a flexible material at the joints
and a rigid material at the beams/walls, the abovementioned
design dilemma can be overcome and a new dimension in the
design space for auxetic metamaterials can be revealed. In this
study, a collection of dual-material auxetic metamaterials
(DMAMs) were designed with CAD, and influences of design
parameters on auxeticity and equivalent Young’s modulus were Fig. 2. A typical re-entrant structural cell: (a) isometric view; and (b) cross-section
investigated with finite element analysis (FEA). Validation samples view.
were fabricated and experimental testing results were reported.

2. Materials and methods metrical parameters such as d,r  h/l, and h (Fig. 2b), while maxi-
mum volume reduction is determined by material properties
2.1. Design and FEA simulation such as strength and maximum strain at break. Equivalent Young’s
modulus is determined by both cell geometry and material proper-
Although the method introduced can be applied to any auxetic ties. In practice, SMAMs cannot achieve theoretic properties due to
patterns, the re-entrant cuboid pattern was chosen as an example beam/wall buckling and other structural failures.
in this study because of its geometrical simplicity and relatively For DMAMs (Fig. 3a), more parameters are introduced, such as
high negative Poisson’s ratio. All models used in this study were the length fraction of the stiff section, or fs (as defined in Fig. 3b),
developed in SolidWorksÒ. A typical cell structure is shown in and the material choices for the elastic part and stiff part. Under
Fig. 2. the definition of stiff section fraction, SMAMs can be seen as
Theoretical relationships between geometrical parameters and fs = 1. When fs < 1, the beams/walls always buckle at the elastic
mechanic properties of this typical re-entrant cell have been thor- regions because of their lower stiffness. By changing the fraction
oughly studied. Those parameters include angles at hinges [22] and of stiff regions, the buckling position and its impact on geometry
[23], diameters of beams [24] and length ratio of beams [25]. Those change can be controlled. It should be noted that the fraction of
theoretic models usually assume stiff cell walls and elastic joints, stiff regions on inclined beams/walls can be controlled indepen-
0
which is the concept that the dual-material design aims to deliver. dently, and a parameter fs = ls/l can be defined in the similar way
0
Generally speaking, Poisson’s ratio is largely determined by geo- as fs. For simplicity, the relationship fs  fs was kept in this study.

Fig. 1. (a) Angle changes at joints when auxetic metamaterial deforms; (b) thickness at the joints increases with beams/walls; and (c) thickness at the joints keeps the same
when thickness of the beams/walls increases.
K. Wang et al. / Materials and Design 67 (2015) 159–164 161

Fig. 3. An example of DMAM. Darker regions represent elastic material. Lighter regions represent stiff material. (a) 5  5  5 array of repetitive cells; and (b) cross-section of a
dual material cell with fs = 0.5.

Table 1
joints are expected to be the stress concentration points. The
Mechanical properties of materials for dual-material 3D printing. observation that structures always fail at joints in experiments
confirmed this assumption. The results of FEA simulations are dis-
Material Material trade Young’s modulus Tensile strength
code name (MPa) (MPa)
cussed in Section 3.

S1 VeroWhitePlus 2500 57.5


S2 RGD8520-DM 2000 50.0 2.2. Materials and equipment for printing validation samples
S3 RGD8530-DM 1400 33.5
E1 TangoBlackPlus 0.5 1.2
E2 FLX9840-DM 0.8 1.6 The DMAM samples were fabricated on a Connex350Ò 3D prin-
E3 FLX9860-DM 1.4 3.3 ter. The layer thickness of printed part is 16 lm, and the in-plane
accuracy is 0.1 mm. The beam/wall thickness of DMAM samples
was designed as the thinnest printable to achieve the smallest pos-
sible features.
The default materials used for stiff regions and elastic regions
Numerical simulations were conducted in COMSOL Multiphys- are VeroWhitePlusÒ and TangoBlackPlusÒ, respectively. These
icsÒ 4.4 with finite element method (FEM). CAD models were two materials represent the two extremes of printable materials
imported from SolidWorks to COMSOL via LiveLink™. Periodic with VeroWhitePlus being the stiffest and TangoBlackPlus the
boundary conditions were assigned to the cell walls to simulate most elastic. The Connex350 can also mix those two base materials
the 3D array in the metamaterial. The materials properties at certain ratio and print them simultaneously to form digital
assigned to elastic and stiff regions are based on printable materi- materials which have mechanical properties between the base
als listed in Section 2.2. Poisson’s ratio and equivalent Young’s materials. In this study three materials each were selected for
modulus can be directly deduced from simulation results. The the stiff region and the elastic region, which are coded as S1–S3
maximum volume reduction was calculated by comparing the ori- and E1–E3, respectively. Table 1 summarizes the Young’s moduli
ginal volume of cell and the volume when the global maximal and tensile strengths of two base materials and selected digital
stress reaches the strength of the elastic material. The reason of materials. The values are median numbers from the material
using the strength of elastic material as failure criteria is that the datasheet provided by Stratasys. For Young’s moduli of E1, E2,

Fig. 4. An example of 2D cellular sample. (a) CAD model; (b and c) microscopic images of the 3D printed sample.
162 K. Wang et al. / Materials and Design 67 (2015) 159–164

Table 2 experimentally. Those parameters were expected to have similar


Parameters of validation samples. effects on the material properties of DMAMs. Therefore, the scope
Group Factora Levels of this study was focused to the two design parameters introduced
C1/T1 C2/T2 C3/T3 C4/T4 C5/T5 by the dual-material design. Experiments were designed to investi-
gate the single-factor effects. Two groups of samples were printed,
G1 Materials S1–E1 S1–E2 S1–E3 S2–E1 S3–E1
G2 fs 0.6 0.7 0.8 0.9 1b
one for each factor. The samples used in mechanical testing shared
a default setting of r = 2, h = p/4, fs = 0.5 with the material combina-
a
This column indicates the changing factor within each group. Other three fac- tion of S1–E1. For each sample group, five different levels were
tors were kept at their default levels for all five samples in that group.
b
fs = 1 means this sample is single-material.
selected for one factor and the other three factors were kept at
the default setting. Table 2 summarizes the 20 settings used for val-
and E3, the values are determined by dynamic mechanical analysis idation. Note that the first samples in both groups are identical as
(DMA). they are all under the default setting. The default water jet washing
Although the FEA simulation is based on 3D auxetic cells, both procedure for removing support materials was not used because it
2D cellular structure and 3D truss structure were used in sample might have destroyed the delicate structure of the DMAMs. The
design. The 2D cellular samples were designed as 5  16 2D re- printed samples were immersed in 5% sodium hydroxide (NaOH)
entrant cells with wall height of 10 mm. The 3D truss samples solution and stirred at 120 rpm on a CimarecÒ SP131325 stirrer
were designed as 5  5  10 re-entrant cells. The dimensions of for 24 h. After all support materials were agitated, the samples were
samples are different for different geometric parameters of cells. rinsed by water and dried in air at room temperature.
The average wall thickness of the 2D samples are 202.4 lm and For each setup, one 2D cellular sample (C1–5) and one 3D truss
the average beam diameter of 3D samples are 454.5 lm based on model (T1–5) were printed for comparison, which yielded 20 sam-
20 random measurements each. Figs. 3 and 4 demonstrate exam- ples in total. In this paper the samples are noted as G[group num-
ples of a 2D cellular sample (r = 2, h = p/6, fs = 0.9) and a 3D truss ber][sample type][sample number]. For example, G1C5 refers the
sample (r = 2, h = p/4, fs = 0.8). cellular sample 5 of group 1 and G2T2 refers the truss sample 2
of group 2.
2.3. 3D printed samples
3. Results and discussion
The goal of the mechanical testing is to validate the simulation
results on the effects of material choices and fs on the Poisson’s 3.1. Effects of material choices
ratio, maximum volume reduction, and equivalent Young’s Modu-
lus of DMAMs. The effects of other design factors that exist in tradi- The simulation and experimental results of sample group 1 are
tional SMAMs, such as r and h, have been studied extensively in the summarized in Fig. 5. In experiments, the Poisson’s ratio was mea-
literature [7,15,22,25–27] and [28], both computationally and sured every 5% strain along z-direction up to 20% (most samples

Fig. 5. Computational and experimental results for sample group 1. (a) Simulation results indicated that stress was concentrated in elastic regions; (b) Poisson’s ratio vs. axial
strain curves; (c) equivalent Young’s modulus; and (d) maximum volume reduction.
K. Wang et al. / Materials and Design 67 (2015) 159–164 163

Fig. 6. Computational and experimental results for sample group 1. (a) Poisson’s ratio vs. axial strain curves; (b) equivalent Young’s modulus; and (c) maximum volume
reduction.

failed between 20% and 25%). The Poisson’s ratio is largely deter- Young’s modulus for the DMAM structure. Since the material
mined by the geometry of the auxetic cell, therefore no significant choice does not affect the Poisson’s ratio after the ‘‘run-in’’ phase,
difference showed in the simulation results for all 10 samples in this implies the equivalent Young’s modulus can be tuned inde-
group 1. Comparing with previous studies on SMAMs in the litera- pendently by varying the material stiffness of the elastic region.
ture [19,29] and [30], the DMAMs demonstrate a unique feature in The maximum volume change is not significantly affected by
simulation which is the constant Poisson’s ratio for different strain material choices. In experiments, DMAMs tend to fail before reach-
level from 5% to 20%. It is due to the controlled deformation at ing the theoretical strain limits because of the defects in the 3D
elastic regions that prevent stiff regions from buckling, which printed part (mostly in the inclined beams/walls). Therefore, the
keeps the auxeticity of DMAMs at the same level as the strain values for maximum volume reduction are much lower than the
increases. However, during strain levels 0–5%, the large deforma- simulation results.
tion of joint regions leads to a rapid change from positive Poisson’s
ratio at the beginning to the same level as the constant negative 3.2. Effects of fs
Poisson’s ratio around 0.5. This ‘‘run-in’’ phase is longer in exper-
imental testing. It can be observed that the durations of the ‘‘run- The simulation and experimental results of sample group 1 are
in’’ phase are shorter for G1T2/G1C2 and G1T3/G1C3, which summarized in Fig. 6. The SMAM samples, G2T5 and G2C5, demon-
implies that the auxeticity at the small strain level can be strated very different properties comparing to DMAM samples.
improved by using relatively stiffer material at elastic regions. In Without the flexible joints, beams and walls tend to buckle around
the simulation, the lowest Poisson’s ratio achieved was around their middle sections, which causes the rapidly decreasing auxetic-
0.5, which is much higher than the value of similar structure in ity (Fig. 6a). On the other hand, the controlled deformation at the
Ref. [19] (around 1.7), but comparable to the simulation results joints in all DMAMs samples allows a stable negative Poisson’s
in Ref. [29] (around 0.5) and the prediction of the model in Ref. ratio when the strain is large. In experiment, G2T5 failed at around
[25] (around 0.55). A possible explanation is that the value in 13% strain and G2C5 failed at around 16% strain. The much stiffer
Ref. [19] was calculated at a different strain level. For the results joints of SMAM samples resulted in much higher equivalent
of physical tests, the Poisson’s ratios of all samples after the Young’s Moduli (Fig. 6b). However, the lack of elasticity limited
‘‘run-in’’ phase stayed around 0.4, whereas those values in Ref. the maximum volume reduction (Fig. 6c). For DMAM samples,
[30] changed from 1.18 to 0.3. This has shown that the DMAMs higher fs yields higher negative Poisson’s ratio at the stable stage
are distinctively different to SMAMs in the stability of auxeticity. and higher equivalent Young’s modulus. The effect of fs on maxi-
The material selection for the elastic region also affects the mum volume reduction is not significant.
equivalent Young’s modulus. Stiffer material for elastic region
yields higher equivalent Young’s modulus. The ratio of equivalent 4. Conclusions
Young’s modulus to Young’s modulus of the elastic material is
about the same (0.078) for all three elastic materials, meaning Auxetic metamaterials with elastic joints and stiff beams or
relatively stiffer material in elastic region yields higher equivalent walls were designed, built, tested, and analyzed. This unique
164 K. Wang et al. / Materials and Design 67 (2015) 159–164

design allows metamaterials to deform without the beam or wall [11] Bhullar S, Ko J, Ahmed F, Jun M. Design and fabrication of stent with negative
Poisson’s ratio. Int J Mech Ind Sci Eng ISSN 2014:2231–6477.
buckling issue that is inevitable in single material designs. Those
[12] Pendry JB, Holden AJ, Stewart WJ, Youngs I. Extremely low frequency plasmons
DMAMs can be fabricated by dual-material 3D printing, which in metallic mesostructures. Phys Rev Lett 1996;76(25):4773–6.
has been demonstrated in this report. Some unique properties, [13] Pendry JB, Holden AJ, Robbins DJ, Stewart WJ. Magnetism from conductors and
such as stable negative Poisson’s ratio at large strain levels, distin- enhanced nonlinear phenomena. IEEE Trans Microw Theory Tech
1999;47(11):2075–84.
guish DMAMs from SMAMs. The effects of two new design param- [14] Lee JH, Singer JP, Thomas EL. Micro-/nanostructured mechanical
eters introduced by the dual-material nature, namely material metamaterials. Adv Mater 2012;24(36):4782–810.
selections and fraction of stiff regions, were investigated both com- [15] Grima JN, Caruana-Gauci R, Dudek MR, Wojciechowski KW, Gatt R. Smart
metamaterials with tunable auxetic and other properties. Smart Mater Struct
putationally and experimentally. Those new parameters added 2013;22(8).
more dimensions to tune the auxeticity and mechanical properties [16] Grujicic M, Galgalikar R, Snipes JS, Yavari R, Ramaswami S. Multi-physics
of DMAMs. More importantly, those parameters affect the proper- modeling of the fabrication and dynamic performance of all-metal auxetic-
hexagonal sandwich-structures. Mater Des 2013;51:113–30.
ties without changing the overall geometry of an auxetic cell. With [17] Subramani P, Rana S, Oliveira DV, Fangueiro R, Xavier J. Development of novel
the flexibility of dual-material 3D printing technology, innovative auxetic structures based on braided composites. Mater Des 2014;61:286–95.
applications, such as metamaterials with uniform geometry but [18] Smardzewski J, Klos R, Fabisiak B. Design of small auxetic springs for furniture.
Mater Des 2013;51:723–8.
varying auxeticity, are enabled by DMAMs. [19] Elipe JCA, Lantada AD. Comparative study of auxetic geometries by means of
computer-aided design and engineering. Smart Mater Struct 2012;21(10).
[20] Xu B, Arias F, Brittain ST, Zhao XM, Grzybowski B, Torquato S, et al. Making
References negative Poisson’s ratio microstructures by soft lithography. Adv Mater
1999;11(14):1186–9.
[21] Smith CW, Grima JN, Evans KE. A novel mechanism for generating auxetic
[1] Lakes R. Foam structures with a negative poissons ratio. Science
behaviour in reticulated foams: missing rib foam model. Acta Mater
1987;235(4792):1038–40.
2000;48(17):4349–56.
[2] Evans KE. Auxetic polymers – a new range of materials. Endeavour
[22] Masters IG, Evans KE. Models for the elastic deformation of honeycombs.
1991;15(4):170–4.
Compos Struct 1996;35(4):403–22.
[3] Liu YP, Hu H. A review on auxetic structures and polymeric materials. Sci Res
[23] Warren TL. Negative Poisson’s ratio in a transversely isotropic foam structure. J
Essays 2010;5(10):1052–63.
Appl Phys 1990;67(12):7591–4.
[4] Bezazi A, Scarpa F. Tensile fatigue of conventional and negative Poisson’s ratio
[24] Gibson LJ, Ashby MF. Cellular solids: structure and properties. Cambridge
open cell PU foams. Int J Fatigue 2009;31(3):488–94.
University Press; 1999.
[5] He CB, Liu PW, McMullan PJ, Griffin AC. Toward molecular auxetics: main
[25] Wan H, Ohtaki H, Kotosaka S, Hu GM. A study of negative Poisson’s ratios in
chain liquid crystalline polymers consisting of laterally attached para-
auxetic honeycombs based on a large deflection model. Eur J Mech – Solids
quaterphenyls. Phys Stat Solidi B – Basic Solid State Phys 2005;242(3):576–84.
2004;23(1):95–106.
[6] Critchley R, Corni I, Wharton JA, Walsh FC, Wood RJK, Stokes KR. A review of
[26] Lakes R. Deformation mechanisms in negative poisson ratio materials –
the manufacture, mechanical properties and potential applications of auxetic
structural aspects. J Mater Sci 1991;26(9):2287–92.
foams. Phys Stat Solidi B – Basic Solid State Phys 2013;250(10):1963–82.
[27] Gaspar N, Smith CW, Alderson A, Grima JN, Evans KE. A generalised three-
[7] Evans KE, Alderson A. Auxetic materials: functional materials and structures
dimensional tethered-nodule model for auxetic materials. J Mater Sci
from lateral thinking! Adv Mater 2000;12(9):617–28.
2011;46(2):372–84.
[8] Andersson A, Lundmark S, Magnusson A, Maurer FHJ. Shear behavior of flexible
[28] Evans KE. Tensile network microstructures exhibiting negative poisson ratios. J
polyurethane foams under uniaxial compression. J Appl Polym Sci
Phys D – Appl Phys 1989;22(12):1870–6.
2009;111(5):2290–8.
[29] Babaee S, Shim J, Weaver JC, Chen ER, Patel N, Bertoldi K. 3D soft
[9] Smith DR, Padilla WJ, Vier DC, Nemat-Nasser SC, Schultz S. Composite medium
metamaterials with negative Poisson’s ratio. Adv Mater 2013;25(36):5044–9.
with simultaneously negative permeability and permittivity. Phys Rev Lett
[30] Critchley R, Corni I, Wharton JA, Walsh FC, Wood RJK, Stokes KR. The
2000;84(18):4184–7.
preparation of auxetic foams by three-dimensional printing and their
[10] Ravirala N, Alderson A, Alderson KL. Interlocking hexagons model for auxetic
characteristics. Adv Eng Mater 2013;15(10):980–5.
behaviour. J Mater Sci 2007;42(17):7433–45.

You might also like