Vous êtes sur la page 1sur 6

Materials Science and Engineering A283 (2000) 105 – 110

www.elsevier.com/locate/msea

Effect of cell morphology on the compressive properties of


open-cell aluminum foams
T.G. Nieh a,*, K. Higashi b, J. Wadsworth a
a
Lawrence Li6ermore National Laboratory, L-350, PO Box 808, Li6ermore, CA 94551, USA
b
Department of Metallurgy and Materials Science, College of Engineering, Osaka Prefecture Uni6ersity, 1 – 1 Gakuen-cho, Sakai,
Osaka 599 -8531, Japan
Received 12 October 1999; received in revised form 10 December 1999

Abstract

The mechanical properties of open-cell 6101 aluminum foams with different densities (  5 – 10%) and morphologies (4–16 cells
cm − 1) were characterized in compression. It was found that density is the primary variable controlling the modulus and yield
strength of foams. The effects of other variables such as cell size and shape were also studied. Whereas the cell size appears to
have a negligible effect on the strength of foams, at a fixed density, the cell shape was shown to effect the strength of foams. In
the present paper, theoretical models are offered to explain the differences in modulus and strength caused by the differences in
cell shape and size. © 2000 Elsevier Science S.A. All rights reserved.

Keywords: Metallic foams; Compression; Tomography; Cellular structure

1. Introduction core structures for high strength aircraft wing panels,


energy absorbers for shaped charges, blast shock waves,
Cellular solids (also known as porous solids) are a and auto bumpers (mechanical) [9]. Their widespread
special class of material that exists commonly in nature: uses exploit the special combination of properties which
wood, cork, sponges, coral and human bones are exam- can be derived from the cellular structure. Currently,
ples. Man has also made his own cellular solids. For there is a high interest in using metallic foams (e.g. Al
example, polymeric foams have been used in everything and Mg) for automotive, railway, aerospace, and chem-
from disposable coffee cups, and packaging materials, ical applications where weight reduction, chemical pol-
to the crash padding of an aircraft cockpit. Advanced lutant minimization, and improvement in comfort and
techniques now exist for foaming not only polymers, safety are needed. Metallic foams have been produced
but also metals [1 – 3] and ceramics [4,5]. In fact, many mainly by casting [10–14], although some have also
honeycomb-like metals, made up of parallel, prismatic been produced by powder metallurgy [15,16].
cells, are used for lightweight aerospace structural com- Many studies have recently been carried out on
ponents. Also, catalytic converters for automobiles are, metallic foams. For example, McCullough et al. [17,18]
sometimes, made from zirconia honeycombs. New characterized the uniaxial deformation behavior and
foams are increasingly used — for porous electrodes, toughness properties of two closed cell aluminum al-
catalytic surfaces (chemical), preforms for metal–ma- loys, Al–lMgl–0.6Si and Al–1Mg1–10 Si (wt.%).
trix composites, heat sinks for electronic components, Harte et al. [19] investigated the fatigue behavior of an
heat exchangers, thermal insulators for fire retardation open-cell A1 6101-T6 foam, and a closed cell ‘Alporas’
[6,7] and thermal shock resistant materials, sound foam of composition Al–5Ca–3Ti (wt.%). Even the
dampers (acoustic) [8], cushions, vibration reducers, high-temperature properties, e.g. creep, of some alu-
minum foams have been investigated [20,21]. There is
* Corresponding author. Tel.: +1-925-4239802; fax: + 1-925-
also a growing interest in studying the energy absorp-
4238034. tion of metallic foams. For example, Mukai et al.
E-mail address: nieh1@llnl.gov (T.G. Nieh) [14,22] studied the deformation behavior of a closed cell

0921-5093/00/$ - see front matter © 2000 Elsevier Science S.A. All rights reserved.
PII: S 0 9 2 1 - 5 0 9 3 ( 0 0 ) 0 0 6 2 3 - 7
106 T.G. Nieh et al. / Materials Science and Engineering A283 (2000) 105–110

Al–5 Ca–3Ti alloy foam and a open-cell AZ91 (Mg–9 from A1 6101. The use of end plates is to assure
wt.%Al–l wt.% Zn – 0.2 wt.%Mn) alloy foam under a loading alignment between a test sample and the upper/
dynamic strain rate (\103 s − 1). In both cases, it was lower platforms of testing frame, and the ease of han-
found that the plateau stress of the foam materials dling low-density foams. Since the modulus and yield
exhibited a high strain rate sensitivity. Also, the absorp- strength increase rapidly with relative density (Eqs. (1)
tion energy at the dynamic strain rate of 103 s − 1 is and (2)), the facing plates can be approximately treated
significantly higher (by about 150 – 200%) than that as rigid bodies. The constraint to lateral expansion of
obtained at the quasi-static strain rate of 10 − 3 s − 1. the foam during testing may be appreciable at a large
In attempts to understand the mechanical response of plastic strain, but is expected to be small in the initial
foam materials under load, a number of micromechani- loading where data of elastic modulus and yield
cal models have been developed [23 – 25]. For example, strength were taken. For discussion purposes, these
considering bending, Young’s modulus, E, for an open- foams will be designated by: relative density/cpi/orien-
cell foam is related to its relative density through the tation, herein. For example, a foam with a 9% relative


equation: density, 40 cell per inch, and that is tested in a direction
perpendicular to the solidification direction is desig-
E r 2
=A (1) nated by 9/40/T.
E0 r0
For synchrotron X-ray tomography, foam samples
where A 1 [26], E0 is the modulus of the fully dense with dimensions of 10× 10× 10 mm were sliced. X-ray
solid, r is the density of the foam, and r0 is the density measurements were carried out using a beam source at
of the solid from which the foam is made. The plastic SSRL (Stanford Synchrotron Radiation Laboratory).
yield stress, s*pl, is also a function of the relative den- The morphology of the foams was also characterized


sity, specifically, using a conventional scanning electron microscope
(SEM). The pore size is defined as the average value of
spl r 3/2
:0.3 (2) the linear dimension of cell across cell faces. The aver-
sys r0 age value was determined by measuring the statistical
where sys stands for the yield stress of the fully dense values from X-ray tomographs.
material [26]. It is obvious from the above two equa- Compression tests were performed directly from the
tions that the single most important structural charac- as-received specimens. The direct measurement of the
teristic of a cellular solid is its relative density, r/r0. elastic modulus from a foam material has subtle aspects
Recently, Nieh et al. [2] studied the elastic properties because of the difficulties associated with sample align-
of an open-cell aluminum alloy (A1 6101-T6). The ment at low loads and the attachment of a strain gage
synchrotron tomographic images indicated that the 3D to the sample. In the present investigation, three exten-
cell structure was slightly anisotropic (8%), and spe- someters were attached, each 120° apart, to measure the
cifically cells were found to be slightly elongated in one displacement of the two faceplates. The crosshead
orientation. Interestingly, both the calculated and mea- movement was monitored by a computer equipped with
sured Young’s moduli along the ‘hard’ axis of the foam a data acquisition system that controlled tests under
was also found to be 5 – 10% larger than that along the either constant crosshead speed or constant true strain
other two ‘soft’ axes. Apparently, cellular morphology rate conditions. Average strain was obtained from the
affects the properties of a foam. In the present paper, readings of the three extensometers. Several loading-un-
we present the effects of cell size and shape, in addition loading steps were taken to obtain the Young’s mod-
to density, on the compressive properties of open-cell ulus and the variation of data is about 9 25 MPa. It
AA6101 aluminum foams. will be shown in the Results section, that this variation
may be acceptable for the ‘high-density’ foams but is
quite significant for the ‘low-density’ foams. Compres-
2. Experimental procedures sion tests were terminated typically at a compression
strain of about 0.8 at which point extensive densifica-
In this study, 6101 aluminum foams (composition tion of the test sample had occurred.
(wt.%) A1–0.6 Mg – 0.5 Si) were fabricated by ERG,
Inc. using a directional solidification technique. These
foams have a relative density ranging from 5 to 10%, 3. Results
cell size from 10 to 40 cell inch − 1 (cpi), and an orienta-
tion that is either longitudinal (L) or transverse (T) to 3.1. Structure and morphology
the solidification direction. The foams were provided in
a cubic shape (dimension: 75×75 ×50 mm). Both the An SEM micrograph of a foam with a relative den-
top and bottom of these compression samples are sity of 6% is shown in Fig. 1. The foam has an open-cell
bonded with 12.5 mm-thick facing plates, also made structure and the cell geometry is remarkably uniform.
T.G. Nieh et al. / Materials Science and Engineering A283 (2000) 105–110 107

The pore size and ligament diameter are about 2.5 and
0.4 mm, respectively. The microstructure of the foam is
shown in Fig. 2, which is a typical cast microstructure
containing large grains and coarse inclusions. Grains
are so large that they often extend across the entire
ligament between cells. The inclusions, analyzed by
EDX, are Fe–Si – A1 particles commonly observed in
6000-series aluminum alloys. No internal voids were
observed within these ligaments.
To examine the 3D connectivity of the foam, syn-
chrotron X-ray tomography was conducted. X-ray to-
mographs of a porous aluminum foam (sample
dimension: 10× 10 × 10 mm) were readily thresholded
into binary images of metal and pore space (see Fig. 3).
Because of the exceptionally high contrast between
metal and pore space there is a broad range of possible
threshold values which lead to nearly identical struc-

Fig. 3. Synchrotron tomographic image of AA6101 Al foam at a


resolution of 23 mm. The structure is predominately open-celled. The
rough surface of cell walls (edges) is a result of the casting process.

tures. The foam shown in Fig. 3 is predominantly


open-cell, but there are a number of partially closed
cells, solidified in place before the cell walls could drain
away, indicating the non-equilibrium nature of the cast-
ing process. Although it is not obvious in Fig. 3, cells
were found to be slightly elongated along the solidifica-
tion direction, resulting from the directional solidifica-
tion process.

Fig. 1. SEM micrograph of AA6101 aluminum foam with a 6%


relative density.
3.2. Mechanical properties

Compressive tests were performed on aluminum


foams with different orientations (morphologies) and
densities. The compressive stress-strain curve obtained
from a 20 cpi foam sample with a density of 9.2%
(9.2/20/T) is shown in Fig. 4. The foam exhibits a small
elastic region, followed by a stress plateau, caused by
the plastic yielding and bending of cell ligaments. Fur-
ther straining causes the collapse of the ligaments and
densification (i.e. a sharp rise in stress) of the overall
structure. The densification strain, od, defined by the
intersection of the two asymptotic curves shown in Fig.
4, occurs when the compressive strain is about 0.72.
The plastic yield stress is about 1.9 MPa.
Experimental data obtained from foams with differ-
ent densities and morphologies tested in different orien-
tations are summarized in Table 1. Also included in the
Fig. 2. Open-celled AA6101 Al foam shows a typical cast microstruc-
table are the analytical predictions for elastic modulus
ture containing large grains and coarse inclusions (indicated by and yield strength, according to Eqs. (1) and (2). The
arrows). modulus and yield strength of A1 6101-T6 are 70.3
108 T.G. Nieh et al. / Materials Science and Engineering A283 (2000) 105–110

Fig. 5. Compressive stress-strain curve obtained from Al 6101-T6


Fig. 4. Compressive stress-strain curve obtained from a 9.2%-density foam with different orientations.
Al 6101-T6 foam.
In addition, there is a slight yield drop occurring in a
GPa and 193 MPa, respectively [27]. As shown in the gradual manner in the longitudinal samples. Also, the
table, the predicted moduli are in reasonable agreement strength difference between the longitudinal and trans-
with experimental values, but the predicted yield stresses verse directions is greater for the 10 cpi sample than that
are noted to be much higher than experimental values. for the 20 cpi samples. The cellular morphology, and
In fact, the yield stress predictions are over one order of particularly the size, appears to have little effect on
magnitude higher than the experimental values. The strength, provided foams have a similar density. For
relative low experimental values must be, in part, associ- example, the strength difference between the 10 and 20
ated with the high plasticity of metals as compared to cpi samples is negligible (B 0.5 MPa), within the exper-
polymeric or ceramic foams, which are practically elas- imental variations.
tic. It is also evident in the table that the densification
strain increases with decreasing density. For example, 3.4. Density
the densification strains are 0.68 and 0.81 for the 10.8%
and 4.8% dense aluminum foams, respectively. Compressive stress-strain curves obtained from
AA6101 A1 foams with different densities are included
3.3. Morphology (different cell size) and orientation in Fig. 6. For clarity, only the portions up to strains of
(longitudinal 6ersus trans6erse) 0.1 are displayed. It is evident in Fig. 6 that a denser
foam exhibits a higher strength. This general trend is
The early stages of the stress – strain curves of two consistent with the statement of Gibson and Ashby [26]
AA6101 A1 foams, 20 and 10 cpi, loaded in two that density is by far the most important variable
perpendicular orientations are shown in Fig. 5. Some influencing the properties of a foam. Although orienta-
observations are readily made. Both the Young’s mod- tion (longitudinal versus transverse) and morphology
ulus and yield stress are noted to be higher for the (cpi) also play certain roles, their effects are considered
longitudinal direction than for the transverse direction. secondary.

Table 1
Properties of AA6101 aluminum foams with different morphologiesa

Sample ID Modulus Predicted modulus/ Stress at 10% Densification strain


(MPa) yield stress [23] (MPa) (%)

10.8/40/L 752 820/2.1 3.0 0.68


9.2/20/L 572 595/1.6 2.2 0.73
9.2/10/L 670 595/1.6 2.4 0.72
4.8/40/L 110 162/0.8 0.9 0.81
8.4/40/T 452 496/1.4 1.3 0.74
9.2/20/T 554 595/1.6 1.9 0.72
9.1/10/T 576 582/1.6 1.7 0.72
5.8/40/T 183 236/0.6 0.5 0.79

a
Modulus and yield strength of Al 6101-T6 are 70.3 GPa and 193 MPa, respectively.
T.G. Nieh et al. / Materials Science and Engineering A283 (2000) 105–110 109

Fig. 6. Compressive stress–strain curves obtained from Al 6101-T6


foam with different orientations.

Fig. 8. A rectangular unit cell loaded in the transverse direction.


4. Discussion
to t 4, and Cl is a geometrical constant. It is pointed out
In the present study Young’s modulus and yield that Eq. (3) readily reduces to Eq. (1) for equiaxed cell.
stress of a foam were observed to be different in two When the cell is loaded in the transverse direction
orientations. Specifically, the longitudinal direction ex- (Fig. 8), Young’s modulus can be calculated as follows.
hibited a higher Young’s modulus and yield stress than Standard beam theory gives the deflection caused by F
the transverse direction. The effect can be explained as to be proportional to (Fh%3)/(3E0I). The force is related
follows. to the remote compressive stress, s, specifically F8
Let us consider an elongated cubic unit cell with shl%. The strain is related to the displacement deflection
square ligament (cell edge), as shown in Fig. 7. The d by o: d/h. It follows immediately that Young’s mod-
sides of the ligament are t and t%, and dimensions of cell ulus, Et, for the foam in the transverse direction is given
are h×h and h× h%, in the two perpendicular projec- by
tions, respectively. When the cell is loaded at the mid-
point of the cell in the longitudinal direction (Fig. 8), s EI
Et = =Ct 04 (4)
one can readily obtain the Young’s modulus, El, by o h%
adopting the approach used by Gibson and Ashby [26], Since I for the cell edge with length h% is proportional to

  
as: t%4, Eq. (4) is reduced to
4
El h% t Et t%4
= Cl (3) = Ct 4 (5)
E0 h h E0 h%
where I is the moment of inertia which is proportional Combine Eqs. (3) and (4), and assume Ct =Cl, it


follows that
Et h (t%/h%)4
= · (6)
El h% (t/h)4
Further assume a constant local density, i.e. t%2h%=t 2h,

 
Eq. (6) can be readily reduced to
5 4
Et h t%
= · B1 (7)
El h% t
since h/h% B1 (and t/t% B1). That is, the longitudinal
modulus is higher than the transverse modulus. In a
similar manner, it can be readily demonstrated that the
elastic collapse stress (yield stress) in the longitudinal
direction is higher than that in the transverse direction,
i.e.
s*ys(t) B s*ys(l) (8)
This explains why the longitudinal orientation has a
Fig. 7. A rectangular unit cell loaded in the longitudinal direction. higher strength than the transverse orientation. In the
110 T.G. Nieh et al. / Materials Science and Engineering A283 (2000) 105–110

present study, it was also found that the Young’s Acknowledgements


modulus and yield stress for foams with different mor-
phologies (e.g. 10 versus 20 cpi) are similar, provided This work was performed under the auspices of the
their densities are the same. This can be explained as US Department of Energy by LLNL (contract no.
follows. W-7405-Eng-48). KH is supported by the Ministry of
The density of the foam, as shown in Fig. 8, can be Education, Science, Culture and Sports, Japan as a
expressed by Grant-in-Aid for Scientific Research on Priority Area
‘Platform Science and Technology for Advanced Mag-
1 8t 2h+ 4t%2h% 3t 2 nesium Alloys’ (11225101 and 11225209) and by the US
r= = (9)
4 h 2h% hh% Army Research Office (grant no. N62649-99-1-0006).
where 6 in the equation is the coordination number. In
the case of 10 versus 20 cpi, 2h %2 =h %1 and 2h2 = h1,
References
where the subscripts 1 and 2 stand for the 10 and 20 cpi
foams, respectively. Since the two foams have the same [1] V. Shapovalov, MRS Bull. 19 (4) (1994) 24.
density, i.e. r1 = r2, it is straightforward to show from [2] T.G. Nieh, J. Kinney, A.J.C. Ladd, J. Wadsworth, Scr. Mater. 38
Eq. (9) that (10) (1998) 1487.
[3] Y. Yamada, K. Shimojima, Y. Sakaguchi, M. Mabuchi, M.
t1 =2t2. (10) Nakamura, T. Asahina, T. Mukai, H. Kanahashi, K. Higashi,
Mater. Sci. Eng. Lett. A272 (1999) 455.
By substituting Eq. (10) into Eqs. (3) and (5), one [4] P. Sepulveda, Ceram. Bull. 76 (10) (1997) 61.
[5] Y. Yamada, K. Shimojima, M. Mabuchi, M. Nakamura, T.
obtains
Asahina, T. Mukai, H. Kanahashi, K. Higashi, J. Mater. Sci. Lett.
18 (18) (1999) 1477 – 1480.
E 1l =E 2l and E 1t =E 2t (11)
[6] T.J. Lu, H.A. Stone, M.F. Ashby, Acta Mater. 46 (10) (1998) 3619.
[7] T.J. Lu, C. Chen, Acta Mater. 47 (5) (1999) 1469.
where superscripts 1 and 2 stand for the 10 and 20 cpi
[8] T.J. Lu, A. Hess, M.F. Ashby, J. Appl. Phys. 85 (11) (1999) 7528.
foams, respectively. Likewise, one can readily show that [9] A.G. Evans, J.W. Hutchinson, M.F. Ashby, Prog. Mater. Sci. 43
(3) (1998) 171.
s*ys(l)
1 2
=s*ys(l) and s*ys(t)
1 2
=s*ys(t) (12) [10] S. Akiyama, H. Ueno, K. Imagawa, A. Kitahara, S. Nagata, T.
Morimoto, T. Nishikawa, M. Itoh, Foamed metal and method of
It is pointed out that the usual assumptions for a producing same, in US Patent no. 4713277, 1987, Shinko Kosen
low-density foam t h and t% h% are used in the above Kogyo Kabushiki Kaisha, Amagasaki, Japan.
derivations. This offers an explanation as to why two [11] V. Shapovalov, Method for manufacturing porous articles, in US
foams with different morphologies but the same density Patent no. 5181549, 1993, Dmk Tek, Inc.
[12] A.E. Simone, L.J. Gibson, Acta Mater. 44 (1996) 1437.
have similar Young’s modulus and yield strength. [13] A.E. Simone, L.J. Gibson, Mater. Sci. Eng. A229 (1997) 55.
[14] T. Mukai, H. Kanahashi, Y. Yamada, K. Shimojima, M.
Mabuchi, T.G. Nieh, K. Higashi, Scr. Mater. 41 (4) (1999) 365.
5. Summary [15] J. Baumeister, H. Schrader, Methods for manufacturing formable
metal bodies, in US Patent no. 5151246, 1992, Fraunhofer-
Gesellschaft zur Forderung der angewandten Forschung e.V.,
AA6101 aluminum foams (composition (wt.%): Al– Munich, Germany.
0.6 Mg–0.5Si) with different densities (5 – 10%) and [16] Y. Sugimura, J. Meyer, M.Y. He, H. Bart-Smith, J. Grenstadt,
morphologies (10 – 40 cells inch − 1) were fabricated by A.G. Evans, Acta Mater. 45 (12) (1997) 5245.
using a directional solidification technique. The 3D [17] K.Y.G. McCullough, N.A. Fleck, M.A. Ashby, Acta Mater. 47
(8) (1999) 2323.
morphology of the foams was examined using synchro-
[18] K.Y.G. McCullough, N.A. Fleck, M.A. Ashby, Acta Mater. 47
tron X-ray tomography and the results revealed rela- (8) (1999) 2331.
tively uniform, open-cell structures. The compression [19] A.-M. Harte, N.A. Fleck, M.A. Ashby, Acta Mater. 47 (8) (1999)
stress–strain curves exhibited a typical three-stage be- 2511.
havior: elastic, nearly-perfect plastic (i.e. with a stress [20] E.W. Andrews, L.J. Gibson, M.A. Ashby, Acta Mater. 47 (10)
(1999) 2853.
plateau), and densification. It was shown that density is [21] E.W. Andrews, J.S. Huang, L.J. Gibson, Acta Mater. 47 (10)
by far the most important variable; a denser foam has (1999) 2927.
a higher modulus and yield strength. In terms of mor- [22] T. Mukai, H. Kanahashi, T. Miyoshi, M. Mabuchi, T.G. Nich,
phology, whereas the cell size appears to have a negligi- K. Higashi, Scr. Mater. 40 (8) (1999) 921.
[23] W.E. Warren, A.M. Kraynik, J. Appl. Mech. 64 (1997) 787.
ble effect, provided the density is the same, the cell [24] H.X. Zhu, N.J. Mills, J.F. Knott, J. Mech. Phys. Solids 45 (11–12)
shape was demonstrated to produce certain influences. (1997) 1875.
In this paper, theoretical models were developed to [25] R.M. Christensen, J. Solids Structures 37 (1 – 2) (2000) 93.
describe the differences in modulus and strength of [26] L.J. Gibson, M.F. Ashby, Cellular Solids: Structure and Proper-
ties, 2nd edn, Cambridge University Press, UK, 1997.
foams caused by the differences in cell shape and size.
[27] ASM, Metals Handbook, 9th edn., vol. 2, Properties and Selec-
Also, as expected, the strain at which densification tion: Nonferrous Alloys and Pure Metals, American Society for
occurs decreases with the density of a foam. Metals, 1980.

Vous aimerez peut-être aussi