You are on page 1of 4

pss

solidi
physica
Phys. Status Solidi B 247, No. 2, 269–272 (2010) / DOI 10.1002/pssb.200945513

status
b
www.pss-b.com
basic solid state physics
Auxetic cellular structures through
selective electron-beam melting
,1 2 2 2
J. Schwerdtfeger* , P. Heinl , R. F. Singer , and C. Körner
1
Institute of Advanced Materials and Processes (ZMP), University of Erlangen-Nürnberg, Dr.-Mack-Str. 81, 90762 Fürth, Germany
2
Institute of Materials Science and Technology (WTM), University of Erlangen-Nürnberg, Martensstr. 5, 91058 Erlangen, Germany

Received 15 July 2009, revised 4 November 2009, accepted 20 November 2009


Published online 22 December 2009

PACS 62.20.de, 62.20.dj, 81.05.Bx, 81.05.Rm

* Corresponding author: e-mail jan.schwerdtfeger@zmp.uni-erlangen.de

This paper is concerned with the build up and characterization objects. Using SEBM we are able to produce structures of
of well-defined auxetic structures (negative Poisson ratio) from arbitrary geometry in a well-defined manner. Here, we
Ti-6Al-4V through selective electron-beam melting (SEBM). introduce a self-designed 3D-auxetic structure and determine
SEBM is a rapid prototyping/manufacturing technique allow- its mechanical properties. We also address the dependence of
ing for the direct translation of CAD models to real world Young’s modulus on relative density.

ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction Auxetic cellular structures have in a highly controlled manner. This not only opens up a whole
attracted considerable interest in recent years due to their new field of possible structures, but allows for a high degree
unique mechanical properties resulting from a negative of control over the resulting mechanical properties. In this
Poisson’s ratio n [1–4]. The value of the Poisson’s ratio paper we present a self-designed three-dimensional auxetic
controls the elastic behavior of a material to the same extent structure and its build up in the SEBM process using
as, for example, the Young’s modulus E. A negative value of Ti-6Al-4V. Results for the mechanical characterization of
n leads to higher indentation resistance [5], shear resistance, the structure by compression testing are shown.
[6] and fracture toughness [7] to name only a few affected
properties. Potential applications for metal auxetics might be 2 The structure This is built up from basic elements
found in a wide area of applications as these materials allow that can be envisaged as an inverted tetrapod. These elements
for new combinations of material properties and present a are arranged in building blocks that themselves are used
new class of materials [6]. To give an example, in high- to fill a plane. The planes are stacked on top of each other
temperature systems the combination of low forces due to by turning the stacking layer by 608 (see also Fig. 1a).
thermal expansion (low E) and high shear resistance are of The resulting structure has an abab stacking order and is
great interest. Previously, auxetic cellulars have been mainly anisotropic, which makes it necessary to determine the
produced through volumetric compression of conventional mechanical properties in different directions. The coordinate
open porous polymer or metal foams (e.g. [8, 9]). The system used in the following to describe the orientation of
volumetric compression results in a complex microstructure the structure is shown in Fig. 1b.
that leads to negative Poisson ratios. The exact mechanisms
are still a matter of debate, but involve the unhinging of 3 Experimental The samples for mechanical testing
crumbled cells and rotation of rigid units (e.g. [10, 11]) were built in a SEBM system (ARCAM A2) using Ti-6Al-
within the microstructure as well as possibly more 4V powder with a maximum particle size of 105 mm and a
complicated modes of deformation [12]. Unfortunately, this mean particle diameter of 70 mm. In the process, parts are
compression process allows for very limited control over the built layerwise by depositing 100-mm layers of metal powder
resulting mechanical properties including n. However, by on a steel start plate using a rake system and melting the
using selective electron-beam melting (SEBM) we are able powder in the desired places with an electron beam gun
to build well-defined auxetic cellulars of arbitrary geometry (acceleration voltage 60 kV). After each layer the start plate

ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim


pss b

solidi
physica

status
270 J. Schwerdtfeger et al.: Auxetic cellular structures through selective electron beam melting

Figure 1 (online color at: www.pss-


b.com) (a) Building plan of the auxetic
structure with dimensions A and B indi-
cated. (b) CAD model of a small cube
made of the auxetic structure with pro-
jections of three faces and the coordi-
nate system used to describe the
orientations are shown.

is lowered and a new powder layer is deposited. The whole samples were build up in 3 directions to avoid differences
process produces low impurity concentrations and good in their mechanical behavior caused by the formation process
mechanical properties comparable to cast material or better. itself. For compression, an updated Instron testing machine
A more detailed description of the process can, for example, (Inspekt Retrofit 100) and a crosshead speed of 1 mm/min
be found in Ref. [13]. was used. Strain was measured by an extensometer setup to
The geometric data of the part to be built are given as a measure the distance between the compression stamps.
CAD model. In the present study the relative density of the The Poisson’s ratio n was determined on larger samples
structures was varied by changing the energy deposited by as this allowed for better accuracy (for exact dimensions see
the electron beam per unit length (constant scan speed of Table 2). To measure the transverse strain an extensometer
160 mm/s; variable beam current of 1.9–6 mA). More energy was attached to the sample using two sharp blades. Then they
translates to a larger amount of metal to be molten. This were strained in compression with a stress up to 1.7 MPa,
results in thicker struts and higher relative densities. Relative which was situated within their elastic range and the
densities were determined geometrically, by determining resulting transverse strains measured. On both types of
sample dimensions using callipers and weighing. samples, care was taken to ensure contact planes with the
For the determination of the Young’s modulus E compression stamps as flat as possible by polishing their
rectangular samples of slightly varying dimensions were faces before testing.
used. For different orientations of the structure different
Table 1 Test specimen dimensions (compression).
dimensions were necessary in order to not cut through cells.
The exact dimensions of the compression samples are listed orientation height [mm] width [mm] thickness [mm]
in Table 1. The dimensions of a single inverted tetrapod
given by the CAD-file were A ¼ 3.75 mm and B ¼ 3 mm (see Orientation 1 39.5 17.4 17.5
Orientation 2 39.3 17.5 17.5
Fig. 1a). Four different line energies were used to build the
Orientation 3 39.6 17.4 17.5
structures resulting in four different relative densities. All

ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-b.com


Original
Paper

Phys. Status Solidi B 247, No. 2 (2010) 271

Table 2 Test specimen dimensions (Poisson’s ratio).


orientation height [mm] width [mm] thickness [mm]

Orientation 1 49.5 34.7 34


Orientation 2 50.5 34.5 35
Orientation 3 49.5 34.5 35

Additionally, a CT-scan (resolution 20 mm) was taken


of a sample built with parameters in the same range as the
compression samples to assess the build quality.

4 Results and discussion Figure 2a shows three


large samples used for the determination of the Poisson’s
ratio. Figure 2b is a comparison between the original CAD
file (left) and a CT-scan of the resulting structure (right).
Figure 2b also illustrates the approach used for Figure 3 (online color at: www.pss-b.com) Young’s modulus for
controlling the relative density of the structure. The struts the three different sample orientations determined by compression
in the CAD file are thinner than what can be realized by the testing plotted against relative density. The lines shown correspond
SEBM system and their thickness is then only controlled by to an exponent of m ¼ 2.5.
the energy deposited by the electron beam. The layered
nature of the forming process is clearly visible in the
CT-scan, but the geometry of the CAD file has been with an exponent m around 2.5 for all three sample
translated well to the real structure. orientations. For the orientations 1 and 2 the values for
The results for the Young’s modulus are compiled in the Young’s modulus are very similar while Orientation 3
log–log diagram shown in Fig. 3. An increase in relative shows lower stiffness throughout the relative density range
density rrel leads to an increase in E. The dependence of E on covered here. Keeping the structure in mind this is not too
rrel can be described by a power law of the form surprising, as in projection Orientations 1 and 2 correspond
to honeycombs, while the third orientation has a fundamen-
E / rm
rel ; (1) tally different geometry.
The exponent found here is slightly higher than the
exponent predicted by the model of Gibson and Ashby [11,
12] formulated for open-cell foams. Their model, which is
based on a periodic cellular structure deforming by beam
bending only, postulates an exponent of m ¼ 2. There are
several possible reasons for this deviation. One thing to be
considered is the relatively large surface roughness of the
struts caused by the SEBM process visible in the CT-scan in
Fig. 2b. This roughness adds material to the struts without
contributing to the actual stiffness of the structure. With
increasing strut thickness the effect of the surface roughness
becomes smaller as it does not increase to the same extent as
the load-bearing part of the structure. This might lead to a
tilting of the power law to higher exponents.
It is also possible that apart from the beam bending
considered by Gibson and Ashby other deformation modes
play a role for the auxetic structure. Apart from beam-
bending contributions, stretching of the struts in the force
direction and the formation of elastic/plastic hinges in the
nodes where struts meet have been considered in the
literature [14–16]. It is generally agreed that stretching only
gives a significant contribution for relatively dense cellular
structures [15], unless the axial stiffness of the cell walls is
Figure 2 (online color at: www.pss-b.com) (a) The actual samples decoupled from their bending stiffness [16] as in the case of
used for determining Poisson’s ratio (three different orientations). wavy cell walls or molecular networks. The formation of
(b) Comparison between original CAD-file (left) and CT-scan (right) hinges usually does not play a significant role in elastic
of resulting structure. foam deformation either. However, for cellulars where the

www.pss-b.com ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim


pss b

solidi
physica

status
272 J. Schwerdtfeger et al.: Auxetic cellular structures through selective electron beam melting

the faces resembling a wavy honeycomb structure (n12 and


n21) we find positive values of n in the range of 0.45–0.5 and
no auxetic behavior.
For all other faces we find negative Poisson ratios in the
range of –0.2 to –0.4 depending on the orientation and,
hence, auxetic behavior. These values show that when the
structure is compressed along the 3 directions the whole
sample contracts. In all other orientations tested here two
faces show a contraction under compressive stress, while one
face (the wavy honeycomb face) expands.

5 Conclusions We have designed and built an auxetic


cellular structure using the SEBM process and were able to
demonstrate the auxetic nature of the structure by com-
Figure 4 (online color at: www.pss-b.com) Poisson’s ratio deter-
pression testing. For the dependence of Young’s modulus on
mined on the three samples shown in Figure 3 (three different relative density of the structure a power-law exponent of
orientations) plotted against strain. about 2.5 was found. A more thorough analysis of the
deformation mechanisms involved is needed to interpret this
exponent, which will be the focus of future work.
material in the nodes has been weakened it can contribute to
the overall axial strain [16]. For foams this can usually be
Acknowledgements We would like to thank Frank
ruled out as their nodes tend to be thicker and hence stiffer
Nachtrab from the ‘‘Fraunhofer Entwicklungszentrum
than their struts. In comparison the cellular structure Röntgentechnik’’ who recorded the CT-scan. The authors also
presented here has nodes that are not reinforced, but rather gratefully acknowledge the funding of the German Research
slightly weakened. This is a result of the production process; Council (DFG), which, within the framework of its ‘‘Excellence
every strut is formed by a series of melt points similar to Initiative’’ supports the Cluster of Excellence ‘‘Engineering of
pearls on a string. A certain overlap ensures the mechanical Advanced Materials’’ at the University of Erlangen-Nuremberg.
integrity of the struts, but between two melt points an area of
smaller diameter exists. As a result, nodes are connected to
References
their surrounding struts through slightly thinner areas where
hinges might form. In order to assess whether one of these [1] R. Lakes, Science 235, 1038 (1987).
mechanisms gives a significant contribution to the axial [2] Y. Wang and R. Lakes, Solids Struct. 39, 4825 (2002).
strain a more thorough investigation of the deformation [3] P. Pastorino, F. Scarpa, S. Patsias, J. Yates, S. Haake, and
behavior of the cells is required, which will be the focus of M. Ruzzene, Phys. Status Solidi B 244, 955 (2007).
future investigations. [4] V. Salit and T. Weller, Acta Mater. 57, 125 (2009).
[5] R. Lakes and K. Elms, J. Compos. Mater. 27, 1193 (1993).
Figure 4 shows the results for the Poisson ratio measured
[6] A. Alderson and K. Alderson, Proc. ImechE Part G, J.
in different orientations of the structure. The Poisson’s ratio Aerospace Eng. 221, 565 (2007).
is given by [7] J. Choi and R. Lakes, Int. J. Fract. 80, 73 (1996).
[8] E. Friis, R. Lakes, and J. Park, J. Mater. Sci. 23, 4406 (1988).
ei [9] R. Lakes, Science 235, 1038 (1987).
nij ¼  ; (2)
ej [10] N. Chan and K. Evans, J. Mater. Sci. 32, 5945 (1997).
[11] J. Grima, A. Alderson, and K. Evans, Phys. Status Solidi B
with ei being the resulting transverse strain caused by the 242(3), 561 (2005).
imposed strain ej and i, j the direction of the measured strain. [12] S. McDonald, N. Ravirala, P. Withers, and A. Alderson, Scr.
All samples used for determining n had a relative density of Mater. 60, 232 (2009).
[13] P. Heinl, A. Rottmair, C. Körner, and R. Singer, Adv. Eng.
about 0.1. The Poisson ratio does not depend appreciably
Mater. 9, 361 (2007).
on strain for the narrow strain window covered in our [14] L. Gibson and M. Ashby, Cellular Solids, second ed.
experiments. For higher strains the dependence would be (Cambridge University Press, Cambridge, UK, 1997).
expected as n is only the result of the internal structure that is [15] L. Gibson and M. Ashby, Proc. R. Soc. Lond. A 382, 43
changing during deformation. The absolute values are (1982).
different for the different orientations of the structure. On [16] I. Masters and K. Evans, Compos. Struct. 35, 403 (1996).

ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-b.com

You might also like