Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Hans-Peter Degischer,
Brigitte Kriszt (Editors)
Handbook of Cellular Metals
Production, Processing,
Applications
Handbook of Cellular Metals: Production, Processing, Applications.
Edited by H.-P. Degischer and B. Kriszt
Copyright c 2002 Wiley-VCH Verlag GmbH & Co. KGaA
ISBNs: 3-527-30339-1 (Hardback); 3-527-60055-8 (Electronic)
Continuous Casting
ISBN 3-527-30283-2
G. Santhoff
Intermetallics
ISBN 3-527-29320-5
Handbook of Cellular Metals: Production, Processing, Applications.
Edited by H.-P. Degischer and B. Kriszt
Copyright c 2002 Wiley-VCH Verlag GmbH & Co. KGaA
ISBNs: 3-527-30339-1 (Hardback); 3-527-60055-8 (Electronic)
Edited by
Hans-Peter Degischer, Brigitte Kriszt
Handbook of Cellular Metals: Production, Processing, Applications.
Edited by H.-P. Degischer and B. Kriszt
Copyright c 2002 Wiley-VCH Verlag GmbH & Co. KGaA
ISBNs: 3-527-30339-1 (Hardback); 3-527-60055-8 (Electronic)
ISBN 3-527-30339-1
Handbook of Cellular Metals: Production, Processing, Applications.
Edited by H.-P. Degischer and B. Kriszt
Copyright c 2002 Wiley-VCH Verlag GmbH & Co. KGaA
ISBNs: 3-527-30339-1 (Hardback); 3-527-60055-8 (Electronic) V
Preface
There have been a few international and national research and development pro-
grams on cellular metals in recent years: the US Multidisciplinary Research Initia-
tive (MURI) on ultralight metal structures since 1996; a few European research
projects funded within the 4th and 5th EU Framework Program; and a focussed
research program funded by the German Research Council, which began in
1998. These cooperative activities add to the original research activities at the
Fraunhofergesellschaft IFAM in Bremen, the University of Cambridge, Ranshofen
and Vienna in Austria, the Slovac Academy of Science in Bratislava, and several
other places in Europe. A considerable quantity of data has been produced and
presented at international conferences. One of these meetings was organized by
the German Society of Materials (DGM) on 28/29 February 2000 at the Vienna
University of Technology. The proceedings are mainly in German [1] and the
authors wanted another opportunity to publish their results in context with present
knowledge. The reader of proceedings might be overwhelmed by very specific re-
search results on selected topics, obscuring a general understanding of this ex-
panding field of research. Besides the classical books dealing with cellular metals
[2,3], an overview of the state of the art in the year 2000 was needed, covering pri-
mary and secondary processing, characterization of cellular metals, properties,
modeling, and exploitation.
As a result, some of the contributors to the Vienna symposium ªMetallschaÈume
2000º have been asked to extend their papers, by referring to related results of
other researchers and giving a review of their particular topic, whilst maintaining
the detailed specialist knowledge of the author. These contributions are introduced
by coordinators who describe the state of the art in that field. Foamed metals are
described more extensively than other cellular metals because of the actual re-
search activities prevalent in Europe. The European development goal is the appli-
cation of cellular metals in components for motor vehicles, a field in which price
limits and consistent high quality are essential. This handbook aims to give a more
detailed overview of the present state of the art in research and development on
cellular metals with specific emphasis on processing and characterization of
foamed metals than the recent survey [4]. The handbook gives a starting point
VI Preface
for researchers new to the field, and references to topics adjacent to their own spe-
cialty for experts already engaged. Engineers and potential users are encouraged to
consider the application of cellular metals, taking into account the specific peculia-
rities of each material to avoid failures due to miscalculation of processing require-
ments and performance. A guide to the multitude of different types of cellular me-
tals is provided with an indication of particular differences in properties. Research
is ongoing and it is to be hoped that experience with applications, that the book
intends to promote and stimulate, is expanding.
The editors thank all contributors and acknowledge their assistance in adjusting
the content of their contributions in cooperation with each other to produce a con-
cise overview of the state of the art in cellular metals from a European viewpoint.
References
Foreword
T. W. Clyne
Interest in metallic foams dates from the 1940s, when Sosnick filed a patent on a
production method involving vaporization of low melting point constituents of me-
tallic alloys [1]. Other publications and patents followed in the subsequent two de-
cades, covering concepts such as that of injecting metal into the interstices around
ªspace holderº particles, which are subsequently removed, and the dispersion of
particles, which will release gas via chemical reaction or thermal decomposition.
However, in general, research on the production and performance of cellular me-
tals remained at a low level until the 1990s, when a substantial acceleration took
place. It is now a research topic receiving a high level of attention and various in-
dustrial applications are currently being explored. Updated information about
these activities is available from certain websites [2] and a comprehensive review
has recently been published [3].
Further research is certainly necessary into the development of improved pro-
cessing methods, since much of the material produced hitherto has been of rela-
tively poor quality and/or inherently rather expensive. However, in this context it
is very important to understand the processing-microstructure-property inter-rela-
tionships for cellular metals and the relevance of these to the property combina-
tions required for various applications. Making a foam from a metal, as opposed
to a polymer, boosts the stiffness, range of operating temperature and resistance
to many (organic) solvents, while, in comparison with a ceramic foam, important
advantages are expected in terms of toughness, (thermal and electrical) conductiv-
ity and formability. However, when considering the detailed characteristics, a clear
distinction should be drawn between open-cellular and closed-cellular metal, since,
not only are these two materials made by different processing routes, but in gen-
eral rather separate types of application can be identified for them. Of course, all
types of cellular materials tend to be relatively light, and to have a high specific
stiffness, but these features usually depend primarily on the pore content, whereas
many other properties are much more sensitive to microstructure (cell structure
and nature of cell walls).
VIII Foreword
Closed-Cell Foams
Since the most promising methods for cheap production of bulk material tend to
generate closed-cell foams (commonly of aluminum), attention has been concen-
trated on these for many structural applications. These include various compo-
nents designed to absorb energy progressively under relatively low applied (com-
pressive) loads. In principle, the reduced constraint on cell walls, compared with
solid metal, should mean that large plastic strains can arise throughout via concer-
tina-like deformation, with substantial absorption of energy. However, the perfor-
mance of such components has often been a little disappointing hitherto, with pre-
mature failure commonly occurring within shear bands and such poor tensile duc-
tility exhibited that components tend to fracture readily if any tensile stresses arise,
for example under bending moments. It is now becoming clear that these prob-
lems are substantially reduced if the cell size can be kept fine and uniform pre-
ferably to sub-mm levels. There is thus a strong incentive to develop processing
techniques capable of producing such material in bulk. Furthermore, it is com-
monly the case that the cell walls in these foams contain severely embrittling con-
stituents, such as large ceramic particles and thick oxide films. Some such consti-
tuents are often deliberately introduced during processing in order to raise the
melt viscosity and thus inhibit cell coarsening and drainage [4]. Recent work has
confirmed that such constituents can have highly deleterious effects on the me-
chanical characteristics of the foam and has outlined the mechanisms responsible
for this [5]. Further work is required both on understanding these effects in more
detail and on developing processing routes in which these constituents are suitably
modified or eliminated. Closed-cell foams are also of interest for other types of ap-
plication, such as thermal barriers, although ceramic materials would often be pre-
ferred for these.
Open-Cellular Metals
Suggested and actual applications include filters, catalyst supports, heat exchang-
ers, fluid flow damping conduits (including various types of shock wave dissipation
devices), biomedical prostheses, internally-cooled shape memory actuators, air bat-
teries, and protective permeable membranes and sheathes. Such functional compo-
nents tend to incorporate higher added value than those in purely structural appli-
cations, which is appropriate in view of the generally higher costs of producing
open-cellular metal. The scale of the cell structure is often important for the func-
tional characteristics. This will clearly be a basic specification for filters and fluid
flow limiting devices, but a fine cell size would often be preferred for heat exchang-
ers etc. (subject to limitations imposed by any danger of pore clogging), while bone
in-growth into prosthetic implants might require relatively coarse pores. However,
these properties would often be needed in combination with a minimum strength
and ductility requirement, so a relatively fine, uniform cell structure and a defect-
free cell struts' microstructure might be beneficial from that point of view. A wide
range of metals is being investigated for applications of open-cellular metals, bring-
ing a requirement for improved understanding of process optimization issues in
various alloy systems.
Foreword IX
References
Contents
Index 365
Handbook of Cellular Metals: Production, Processing, Applications.
Edited by H.-P. Degischer and B. Kriszt
Copyright c 2002 Wiley-VCH Verlag GmbH & Co. KGaA
ISBNs: 3-527-30339-1 (Hardback); 3-527-60055-8 (Electronic) XIX
List of Contributors
O. Andersen F. BaumgaÈrtner
Frauenhofer-Institut fuÈr Schunk Sinter Metalltechnik GmbH
Fertigungstechnik und Postfach 10 09 51
Materialforschung (IFAM) 35339 Gieûen
Auûenstelle fuÈr Pulvermetallurgie Germany
und Verbundstoffe
Winterberstr. 28 C. Beichelt
01277 Dresden Wilhelm KARMANN GmbH
Germany Karmannstr. 1
Postfach 26 09
M. Arnold 49084 OsnabruÈck
Lehrstuhl fuÈr Werkstoffwissenschaften Germany
UniversitaÈt Erlangen-NuÈrnberg
Martensstr. 5 T. Bernard
91058 Erlangen Neue Materialien Bayreuth GmbH
Germany UniversitaÈtsstr. 30
94447 Bayreuth
M. F. Ashby Germany
Engineering Design Centre
University of Cambridge H. W. Bergmann²
Engineering Department, UniversitaÈt Bayreuth
Trumpington Street Lehrstuhl Metallische Werkstoffe
Cambridge CB2 1PZ Ludwig-Thoma-Str. 36b
UK 94440 Bayreuth
Germany
J. Banhart
Hahn-Meitner-Institute H. J. BoÈhm
Dept. of Materials ± SF3 Institute of Lightweight Structures and
Glienicker Str. 100 Aerospace Engineering
14109 Berlin Vienna University of Technology
Germany Guûhausstr. 27±29
1040 Vienna
Austria
XX List of Contributors
R. Braune C. Haberling
Lehrstuhl fuÈr Fertigungstechnologie Firma Audi AG
UniversitaÈt Erlangen-NuÈrnberg Abteilung
Egerlandstr. 11 Werkstoffe/Verfahren/Recycling I/EG-34
91058 Erlangen 85045 Ingolstadt
Germany Germany
T. W. Clyne F. Heinrich
Department of Materials Science Lehrstuhl fuÈr Werkstoffwissenschaften
University of Cambridge UniversitaÈt Erlangen-NuÈrnberg
Pembroke Street Martensstr. 5
Cambridge CB2 3QZ 91058 Erlangen
UK Germany
M. C. Hahn A. Kottar
Lehrstuhl fuÈr Fertigungstechnologie Institute of Materials Science
UniversitaÈt Erlangen-NuÈrnberg and Testing
Egerlandstr. 11 Vienna University of Technology
91058 Erlangen Karlsplatz 13
Germany 1040 Vienna
Austria
List of Contributors XXI
B. Kriszt A. Mortensen
Institute of Materials Science Laboratoire de MeÂtallurgie MeÂcanique
and Testing Ecole Polytechnique FeÂdeÂrale
Vienna University of Technology de Lausanne
Karlsplatz 13 1015 Lausanne
1040 Vienna Switzerland
Austria
U. Mosler
R. Kretz Institut fuÈr Metallkunde
ARC Leichtmetallkompetenzzentrum TU Bergakademie Freiberg
Ranshofen GmbH Gustuv-Zeuner-Str. 5
Postfach 26 09596 Freiberg
5282 Ranshofen Germany
Austria
C. Motz
E. M. A. Maine Erich-Schmid-Institute of Material
Centre for Technology Management Science
Engineering Design Centre Austrian Academy of Science
Engineering Department Jahnstr. 12
University of Cambridge 8700 Leoben
Trumpington Street Austria
Cambridge CB2 1PZ
UK R. Neugebauer
Frauenhofer Institut fuÈr Werkzeug-
E. Maire maschinen und Umformtechnik (IWU)
CR1 CNRS Reichenhainerstr. 88
GEMPPM 09126 Chemnitz
Batiment Saint ExupeÂry Germany
23 Avenue Capelle
69621 Villeurbanne cedex A. Otto
France Lehrstuhl fuÈr Fertigungstechnologie
UniversitaÈt Erlangen-NuÈrnberg
U. Martin Egerlandstr. 11
Institute fuÈr Metallkunde 91058 Erlangen
TU Bergakademie Freiberg Germany
Gustuv-Zeuner-Str. 5
09596 Freiberg R. Pippan
Germany Erich-Schmid-Institute of Material
Science
H. Mayer Austrian Academy of Science
Institute of Meteorology and Physics Jahnstr. 12
University of Agricultural Sciences 8700 Leoben
TuÈrkenschanzstr. 18 Austria
1180 Vienna
Austria
XXII List of Contributors
G. Rausch R. F. Singer
Frauenhofer Institut fuÈr angewandte Institut fuÈr Werkstoffwissenschaften
Materialforschung (IFAM) UnivertitaÈt Erlangen-NuÈrnberg
Lesumer Heerstr. 36 Martensstr. 5
28717 Bremen 91058 Erlangen
Germany Germany
F. G. Rammerstorfer G. Stephani
Institute of Lightweight Structures Frauenhofer-Institut fuÈr
and Aerospace Engineering Fertigungstechnik und
Vienna University of Technology Materialforschung (IFAM)
Guûhausstr. 27±29 Auûenstelle fuÈr Pulvermetallurgie
1040 Vienna und Verbundstoffe
Austria Winterberstr. 28
01277 Dresden
W. Seeliger Germany
Wilhelm KARMANN GmbH
Karmannstr. 1 M. Thies
Postfach 26 09 Lehrstuhl fuÈr Werkstoffwissenschaften
49084 OsnabruÈck UnivertitaÈt Erlangen-NuÈrnberg
Germany Martensstr. 5
91058 Erlangen
M. Seitzberger Germany
Institute of Lightweight Structures
and Aerospace Engineering B. Zettl
Vienna University of Technology Institute of Meteorology and Physics
Guûhausstr. 27±29 University of Agricultural Sciences
1040 Vienna TuÈrkenschanzstr. 18
Austria 1180 Vienna
Austria
F. Simancik
Institute of Materials and Machine
Mechanics
Slovak Academy of Science
Racianska 75
P.O. Box 95
83008 Bratislava
Slovakia
Handbook of Cellular Metals: Production, Processing, Applications.
Edited by H.-P. Degischer and B. Kriszt
Copyright c 2002 Wiley-VCH Verlag GmbH & Co. KGaA
ISBNs: 3-527-30339-1 (Hardback); 3-527-60055-8 (Electronic) XXIII
List of Abbreviations
1
Introduction: The Strange World of Cellular Metals
F. Simancik
When Nature builds large load-bearing structures,
She generally uses cellular materials: wood, bone, coral.
There must be good reasons for it.
M. F. Ashby
It is well known that porous structures are good for insulation, packaging, or filter-
ing, but few people believe that they can also be very effective in structural applica-
tions. Thousands of scientific publications deal with the minimization of porosity
in load-bearing parts. Engineers work hard to eliminate pores from castings, pow-
der metallurgy parts, weld joints, or coatings, thinking that a defect-free part is a
pore-free one. With this attitude it is difficult for someone to accept that a load-
bearing material can include pores, even quite large ones. However, large natural
structures of porous materials have existed for thousands of years, demonstrating
how evolution has generated cellular structures that optimize mechanical proper-
ties and structural function for minimum weight.
Mankind tries to learn from nature. Understanding the benefits of natural struc-
tures gives us information to help us produce man-made cellular solids. The cell-
wall material has to be chosen very carefully if the structure is expected to carry
loads. Polymers appear to be insufficiently rigid and ceramics are too brittle. Per-
haps metals could be the right choice. Several of the engineering properties of me-
tallic foams are superior to those of polymeric ones: they are stiffer by an order of
magnitude, they are stable at elevated temperatures, they possess superior fire re-
sistance, and they do not produce toxic fumes in a fire. Moreover, these materials
are fully recyclable without any pollution or waste problems. The latter fact can no
longer be ignored, because the production, disposal, and use of stronger and stiffer
materials in new products often have negative environmental impacts over the
product life cycle.
Owing to their pores, cellular metals possess a set of unusual properties
compared with bulk structural materials: they are crushable, they exhibit a plateau
stress if compressed, and they exhibit a change in Poisson ratio on deformation.
The excellent combination of good mechanical properties (mostly strength and
stiffness) and low weight is the prime advantage. In addition, cellular metals
2 1 Introduction: The Strange World of Cellular Metals
absorb high impact energies regardless of the impact direction, and are very
efficient in sound absorption, electromagnetic shielding, and vibration damping.
Most of the mechanical properties of foam materials can be achieved with other
materials, sometimes more effectively, but foams can offer a unique combination
of several (apparently contradictory) properties that cannot be obtained in one con-
ventional material at the same time (e. g., ultra-low density, high stiffness, the capa-
bility to absorb crash energy, low thermal conductivity, low magnetic permeability,
and good vibration damping). Cellular metals are thus promising in applications
where several of these functions can be combined.
These properties depend significantly on the porosity, so that a desired portfolio
of properties can be tailored by changing the foam density. This is one of the most
attractive features of these remarkable materials.
Cellular material properties also depend on the pore structure. This influence,
imperfectly understood at present, is a topic of intense study. Various constitutive
laws have been suggested for the characterization and modeling of this relation-
ship. These laws, originally developed for polymeric foams, are usually based
on the relative density of the foam, and therefore suppose uniform cellular struc-
ture, at least at a macroscopic level. However, metallic foams are dramatically dif-
ferent from polymeric foams: polymeric foams generally have a regular micro-
structure, whereas metallic foams may be highly disordered with a wide dis-
persion of cell size and shape. Moreover, many imperfections exist in a cell
structure, such as cracks or holes in the cell walls, corrugated cells etc. These
effects are inevitable due to manufacturing at significantly higher temperatures
than in the case of polymers. If these features are not taken into account and
the properties of the foam are characterized only in relation to apparent density,
a higher scatter of properties is to be expected. This is why it is still widely
believed that acceptable reproducibility of the properties of metallic foams is
questionable.
The structure of metallic foams is often non-uniform, especially in the case of
complex 3D parts. It should be noted that a uniform structure is not necessary
for obtaining acceptable and reproducible properties. Anisotropic or gradient
pore structures allow the distribution of load bearing material according to load
conditions (simulating the optimum bone-like structure), without a need to in-
crease the overall weight or volume of the component. Therefore, the challenge
for manufacturing is not to produce a uniform structure, but to achieve reproduc-
ible properties with a controlled non-uniform structure.
If the non-uniform structure is optimal, the crucial question ªWhat is the mate-
rial?º should be answered. It is really difficult to distinguish between material and
structure. If a cellular metal is a material, it is very problematic to define geometry-
independent material characteristics (the strength or elasticity modulus); if it is a
structure made of a certain metal it is almost impossible to define its random
geometry.
Cellular metals can be prepared by various processing methods. They may all be
called ªmetallic foamsº, but they are very different materials, depending on the
manufacturing technique. The production method affects the distribution of the
1 Introduction: The Strange World of Cellular Metals 3
cell-wall material in such a way that the properties of differently manufactured ma-
terials are not comparable.
Metallic foams result from the nucleation and subsequent growth of gas bubbles
in a liquid or semi-liquid metal. They usually have a non-uniform pore structure
(variable pore size and sometimes preferred orientation of pores). The pores are
initially closed, but some defects always appear on cooling, owing to shrinkage of
solidifying metal and gas-pressure reduction in pores. Many solidified cell faces
have non-uniform curvature or are corrugated and have occasional broken walls
that still hang in place. These are the main features of this kind of cellular solid.
Other cellular metals may be manufactured by casting or deposition of the metal
onto templates or place holders, which have to be removed from the final product,
thus creating a porous structure. These cellular structures usually have adjustable
distribution of pore size (according to the template) and their cells are always open.
The manufacturing process dictates not only the properties but also the potential
applications of the foam. Thus foams prepared by the powder compact foaming
(PCF) technique (usually with a dense skin) can be effectively used as net-shape
components, stiffening cores in castings, or in complicated hollow profiles,
whereas the foams prepared by the ªmolten metal routeº (typically large blocks
or panels) can be effectively used as voluminous energy absorbers, cores for sand-
wiches, or for blast protection. The open cellular structures made by investment
casting are good for heat exchangers, sound absorbers, or for electrodes in bat-
teries. The properties arising from the cellular structure produced by a certain
manufacturing technique cannot be effectively achieved using another method.
This also means that cellular metals manufactured differently are not necessarily
competitive materials.
The first attempt to foam a metal was performed by B. Sosnik in 1943 [1]. In
order to create pores, he added mercury to molten aluminum. In 1956 J. C. Elliot
replaced mercury by foaming agents generating gas by thermal decomposition [2],
so now modern scientists and engineers can develop metallic foams without
having to deal with the toxicity of mercury.
In 1959 B. C. Allen [3] invented the PCF route for manufacturing metallic foams
and the basic processing techniques were thus completed.
The success in the preparation of the first metallic foams and discovery of their
remarkable properties started a euphoric enthusiasm for these materials. In 1957
J. Bjorksten stated [4]: ªFoamed metals offer great market potential and might con-
ceivably account for 10 % of all metals produced within 20 yearsº. However he also
said: ªA lot of work still remains to bring about production on a large scale, such as
closer control of density and dimensions.º Unfortunately, only the second of his
statements turned out to be realistic.
Although it is many years since the first patents concerning the manufacture of
metallic foams appeared, the material has not been put into large-scale commercial
production yet. This discouraging fact can be attributed to inadequate design of
components, low reproducibility of properties, a lack of testing procedures and cal-
culation approaches, absence of concepts for secondary treatment, as well as the
production technologies being too complicated and relatively expensive.
4 1 Introduction: The Strange World of Cellular Metals
References
1. B. Sosnik, US Patent 2 434 775, 1948. 6. J. Iljoon et al., US Patent 5 115 697, 1992.
2. J. C. Elliot, US Patent 2 751 289, 1956. 7. Ultralight Metal Structure Project, MURI
3. B. C. Allen, US Patent 3 087 807, 1963. Grant No. N00014-1-96-1O28, Washington
4. Modern Metals, 1957, October. 1996.
5. S. Akiyama et al., European Patent 0 210 803, 8. Zellulare metallische Werkstoffe, DFG-Priority
1986. Program 1075, Bonn 1999.
Handbook of Cellular Metals: Production, Processing, Applications.
Edited by H.-P. Degischer and B. Kriszt
Copyright c 2002 Wiley-VCH Verlag GmbH & Co. KGaA
ISBNs: 3-527-30339-1 (Hardback); 3-527-60055-8 (Electronic) 5
2
Material Definitions, Processing, and Recycling
H. P. Degischer
Figure 2-1. Processing techniques for cellular metals classified according to the state of the
metal, the formation of the cellular architecture, and the pore-forming ingredients.
6 2 Material Definitions, Processing, and Recycling
density of less than 0.3 [1]. Materials with higher relative densities are called po-
rous materials (for instance, powder compact greens), most of which also can be
produced by the processing techniques listed in Fig. 2-1. The most widely devel-
oped and investigated cellular metals are based on aluminum and its alloys.
The general aspects of processing are described elsewhere [1 3] and specific pre-
sentations can be found in the proceedings of the conferences dealing with cellular
metals [4 10]. The most important processing techniques are described in detail in
subsequent chapters.
Cellular metals are heterogeneous materials formed by a 3D metallic matrix with
gas-containing pores occupying more than 70 vol.-%. Cellular metals are classified
according to the following criteria (Fig. 2-1).
x The metal condition during production of porosity: liquid, solution or emulsion,
solid.
x The forming process involved: casting, foaming, deposition, sintering (including
precursor slurry).
x The method of pore formation: incorporating hollow substrates, removable sub-
strates, or gas (either directly, dissolved, or by means of a dissociating agent).
Open porosity structures can be formed by replication (Duocel [6]), deposition (In-
cofoam [6]) or by construction of solid ingredients with space between (for in-
stance, the 3D networks of lattice block materials (LBM) [5,6] or those prepared
by rapid prototyping techniques [11]). Open-cell solid structures may be called
sponges. Closed cells are produced by embedding or cementation of hollow ingre-
dients (ªsyntactic foamsº), or by foaming in the liquid state.
The expression ªmetal foamsº, strictly valid only for the liquid phase, is often
used to describe the solid product. Mixtures of metal powders and blowing agents
are compacted by extrusion or hot pressing providing a precursor material foamed
above the solidus temperature, a method called ªpowder compact foamingº (Alu-
light [5 10], IFAM-Foaminal [3,5 10], Alufoam [5,7]). The Formgrip material
[3,7] is made by remelting a stir-cast foamable precursor metal matrix composite.
Low-density-core material (LDC [3,6]) is produced by pore formation in the solid
state by the high gas pressure of entrapped dissociating blowing agents. A blowing
agent powder is mixed into the melt either in a crucible (Alporas [1±7]) or in the
gate in pressure die casting (Buehler [10]). Foaming of particle-reinforced metals
takes advantage of the stabilization of gas bubbles by the ceramic ingredients prod-
ucing a cellular metal matrix composite from the melt (Cymat [1 7], COMBAL
[10]). The formation of a gas metal eutectic is the principle for the production
of Gasar foams [1,2,6]. A slip reaction foam technique based on foaming of precur-
sor slurry by chemical reaction and a reaction sintering process for aluminides has
been described [10]. The originally closed cells of foamed metals may not be gas
tight after solidification owing to cracks in their walls.
The resulting cellular metal products can be differentiated by their structural fea-
tures. The term ªstructureº is used for the description of cellular materials at dif-
ferent levels of observation (structology): the geometric architecture of the solid
(skeleton) in the individual cells and their 3D arrangement, the variations of that
2 Material Definitions, Processing, and Recycling 7
References
2.1
Foaming Processes for Al
The melt route for processing closed-cell metal foams is very attractive since this
approach allows economic handling of large quantities of material. Melt-route pro-
cesses are also well suited to the use of scrap as feedstock. In order to foam the
melt properly, gas must be introduced. This can be either done by gas injection
or by in-situ gas generation by a chemical decomposition of a foaming agent.
For the production of homogenous foams some prerequisites have to be fulfilled.
If foaming is by in-situ gas generation uniform dispersion of the foaming agent in
the melt within a time that is short compared to the decomposition or reaction
time of the additive is required. In addition, the escape of gas during the foaming
process has to be prevented. One important step to meet these requirements is to
increase the viscosity of the melt. There are several approaches to do this: foaming
in the semi-liquid state, incorporation of ceramic particles [1] or oxidation [2]. The
effect of the particles added is always twofold: beside increasing the melt viscosity
they stabilize the cell walls.
Independent of the specific foaming process it is found that porosity, quantified
by density, and mean cell size are intimately related. Generally, for a particular alloy
or composite the mean cell size can not be chosen independently of the porosity.
2.1.1
Gas Injection: the Cymat/Alcan and Norsk Hydro Process
Figure 2.1-1. Principle of the melt-foaming route employed by Cymat. The foam casting process
for producing flat panels consists of melting and holding furnaces, the foaming box and foaming
equipment, and a twin-belt caster [1].
2 Material Definitions, Processing, and Recycling 9
particles) is used as a starting material. The starting material is molten with con-
ventional foundry equipment and transferred to a tundish where gas, typically air,
is injected via small nozzles incorporated into a rotating impeller, thus forming a
dispersion of small gas bubbles. The bubble size can be controlled by adjusting the
gas flow rate, the impeller design (number of nozzles and their size), and the speed
of rotation of the impeller. The gas bubbles rise to the surface where they accumu-
late. The ceramic particles are trapping gas bubbles owing to the favorable interface
energy and serve as stabilizer of the cell walls and delay their coalescence. They
also reduce the velocity of the rising bubbles by increasing the viscosity of the
melt. That is, the particles reduce the kinetic energy of the rising bubbles and
hence the danger of mechanical rupture when they arrive at the surface. The re-
sulting metal foam, which is still liquid, is carried away by means of a conveyor
belt where it solidifies and cools. The relative density is predominantly controlled
by the process parameters, such as rotor speed, gas flow, and the amount of parti-
cles in the melt, and finally the solidification condition.
This process of casting aluminum foam is capable of producing slabs with a re-
lative density in the range 2 20 % (0.05 0.55 g/cm3). The average cell size is in-
versely related to the density (Fig. 2.1-2a) and is in the range 2.5 30 mm [1].
a)
Figure 2.1-3. Optical micrographs of Cymat foam produced by the gas-injection method: left)
cell structure (density about 0.3 g/cm3), inhomogeneous and anisotropic; right) foam surface
(density about 0.05 g/cm3).
The production facility set up by Cymat is capable of casting foam panels in con-
tinuous length at an average rate of 900 kg/h up to 1.5 m wide with a thickness
range of 25 150 mm. This shows that the process is relatively straightforward
and economical. Challenges that may require more work in the future include
the variation in cell size, the density gradient, and the anisotropy of the cell struc-
ture, which results from mechanical forces from the conveyor belt (Fig. 2.1-3).
Especially earlier prototype have been compressed by the conveyor belt producing
flattened pores causing poor stiffness and strength values along the thickness of
the slabs. It should also be noted that the preparation of the composite feedstock
requires relatively tedious long time stirring processes to achieve the proper homo-
genous distribution of the particles in the melt.
In principle, this foam generation technology also allows the casting of nonrec-
tangular, 2D profiles as well as 3D shapes.
2.1.2
In-situ Gas Generation: the Shinko Wire Process and the FORMGRIP process
The gas-injection method suffers from the fact that a relatively small number of
large bubbles is generated, which leads to rather coarse and irregularly shaped
pore distribution. Two techniques are described below, in which the foaming gas
results from a thermal decomposition of solid ingredient. In this way a huge num-
ber of bubble nuclei is created throughout the melt.
the viscosity by the formation of oxides: CaO, Al2O3, CaAl2O4. There is an appro-
priate stirring resistance for optimizing the foaming ratio [4]. The thickened alumi-
num is poured into a casting mold and stirred with an admixture of 1.6 % TiH2 as a
foaming agent. While vigorously stirring, the TiH2 dissociates and H2 -bubbles are
formed causing molten material to expand and to fill the mold. Then, the foamed
material is cooled by fans to solidify in the casting mold. After removal from the
casting mold. An Alporas block 450 mm wide, 2050 mm long, and 650 mm
high is sliced into plates.
Alporas is an ultra-light material with a closed-cell architecture (Fig. 2.1-5). The
density of the product is 0.18 0.24 g/cm3, the mean cell size is about 4.5 mm. Al-
poras is regarded as the best commercially available aluminum foam in terms of
regular cell micro structure. This is to a certain extent because transfer and defor-
mation of the still liquid foam, as in the Cymat/Norsk Hydro process, is avoided.
The cell architecture is the outcome of a growing process where bubble expand and
coalesce due to cell wall rupture.
Figure 2.1-5. Typical cell structure of an Alporas foam. As a result of the growth process most
cells are far from equiaxed. The global homogeneity is superior to other commercially available
aluminum foams.
12 2.1 Foaming Processes for Al
Figure 2.1-6. Diagram of the melt-based FORMGRIP process for production of near net-shape
metal foam parts [7].
2 Material Definitions, Processing, and Recycling 13
Figure 2.1-7. Optical micrographs illustrating the distribution of SiC particles in Al-9Si alloy
based FORMGRIP foams, showing sections through: a) cell wall, b) a node. A significant fraction
of particles is located at the gas/melt interface [5].
Figure 2.1-8. Examples of cross sections of aluminum alloy FORMGRIP foams baked under
different conditions. Porosity, P, levels and the mean cell sizes, d, are: a) P 69 %, d 1.1 mm;
b) P 79 %, d 1.9 mm; c) P 88 %, d 3.1 mm [7].
The relation between relative foam density and average cell diameter fits that de-
rived for Cymat foams. Fig. 2.1-2b shows the mean cell diameter as a function of
the reciprocal density (see Section 4.1). The mean cell diameter is inversely propor-
tional to the density indicating that foam expansion is governed by cell coalescence
and the mean cell wall thickness is constant. The dependence of the cell diameter
on the density is the same for both contents of SiC particles. The influence of par-
ticle size has not yet been investigated. Theoretical work of Kaptay [9] and experi-
mental work of Weigand [10] indicate that a reduction of particle size will not lead
to a higher stabilization and therefore to a smaller critical cell wall thickness.
In terms of geometrical complexity and simultaneous microstructural control,
the FORMGRIP process surpasses the Shinko Wire one. There is no need to trans-
fer material from a mold into a die while the foaming process is going on. In ad-
dition, the precursor material can be shaped before the final baking step. The eco-
nomics of the FORMGRIP process, however, are clearly inferior to the other pro-
cesses discussed in this chapter. This is due to the discontinuous nature of the pro-
cess as well as its various number of processing steps.
14 2.2 Industrialization of Powder-Compact Foaming Technique
References
2.2
Industrialization of Powder-Compact Foaming Technique
There are many different ways to manufacture cellular materials [1]. One of the
available processes has become increasingly popular in the past few years and is
at the stage of industrial implementation now. The method is sometimes loosely
called the ªpowder-metallurgical routeº, but the term ªpowder-compact foaming
techniqueº seems more appropriate.
2.2.1
Principles of Foam Production
The technique consists of mixing aluminum or aluminum alloy powders with ap-
propriate foaming agents, which get entrapped by compacting this mix to a dense
product called ªfoamable precursor materialª. The powder mix can be compacted di-
rectly by hot pressing, conform extrusion, or powder rolling. Alternatively the pow-
2 Material Definitions, Processing, and Recycling 15
der may be cold compacted for better handing in conventional extrusion or rolling
(Fig. 2.2-1). Heating the precursor above its solidus temperature releases the pres-
sure on the foaming agent allowing decomposition and formation of bubbles. After
cooling a low-density foam structure of originally closed cells is obtained [2,3].
The method is not restricted to aluminum and its alloys: tin, zinc, lead, gold, and
some other metals and alloys can also be foamed by choosing appropriate foaming
agents and process parameters (see Section 2.1.2). The most common alloys for
foaming, however, are pure aluminum or wrought alloys such as aluminum
2xxx, 6xxx, or 7xxx alloys, e. g. AA 2014, 6060, 6061, 6082, or 7075. Casting
alloys such as AlSi7Mg (A356) and AlSi12 are also frequently used because of
their low melting point and good foaming properties, although in principle vir-
tually any aluminum alloy can be foamed by carefully adjusting the process para-
meters.
Quite complex-shaped metal foam parts can be manufactured by expanding the
foam inside a mold, thus confining spatial expansion. An example for one such
part is shown in Fig. 2.2-2. The part, developed in the framework of a feasibility
study, is a novel pantograph horn for an electrical locomotive. This light-weight so-
lution based on aluminum foam replaces traditional cast aluminum parts saving
30 % weight.
A nice feature of the technique is that composite structures consisting of an alu-
minum foam and bulk metal parts can be made without using adhesives. Exam-
ples are foam-filled aluminum sections and sandwich panels with an aluminum-
foam core and metallically bonded steel, aluminum, or even titanium face sheets.
For making such composites the foamable precursor material is first bonded to the
solid section or sheet by co-extrusion or roll-cladding, after which the foamable
core layer is expanded by heat treatment [4,5] (see also Section 3.3).
16 2.2 Industrialization of Powder-Compact Foaming Technique
The advantages of the powder-compact route are obvious and are listed in
Table 2.2-1. Beside the first two features already mentioned, the flexibility arising
from the preparation of the precursor from powders is important. Alloys can be
made simply by mixing inexpensive elementary powders. No ceramic additives
are needed to stabilize the foam, in contrast to some of the melt-route foaming pro-
cesses in which up to 15 % silicon carbide has to be added [1,6]. However, if re-
quired, ceramic powders, metal fibers, or ceramic fibers can be added to the pow-
der blend for special applications, such as for reinforcement or to increase wear
resistance. Naturally, there are also some disadvantages that are inherent to the
process. Metal powder is more expensive than bulk metal and requires effort for
compaction. This rules out applications that require very cheap materials. More-
over, the size of aluminum foam parts that can be manufactured is limited by
the size of the baking furnace, and is therefore smaller than for some of the com-
peting melt-foaming processes. The largest sandwich components that have been
manufactured using the powder-compact foaming technique are about 2 m q 1 m
q 1 cm in size (possibly larger in future). At LKR in Ranshofen a part of similar
size was produced without face sheets (Fig. 2.2-3). True 3D-volume parts are
usually not thicker than 30 cm, a limit which is difficult to shift to higher values.
A large aluminum foam column produced at Fraunhofer IFAM was 1 m high and
18 cm in diameter, weighing 13 kg. In contrast, the liquid-metal route allows for
making panels 15 m in length [7] and 100 cm thickness [8]. However, as these pro-
cesses cannot be used for near-net-shape production and only permit very simple
geometries, they are appropriate for different fields of application. Continuous
foaming of long products is under investigation [9].
The middle column of Table 2.2-1 lists some of the problems that are still en-
countered when foaming aluminum with the powder-compact melting method
but which can, in principle, be solved with further research.
2 Material Definitions, Processing, and Recycling 17
2.2.2
Practical Aspects of Foam Production
2.2.2.2 Mixing
The mixing procedure should yield a homogeneous distribution of alloying ele-
ments and the foaming agent to ensure that high-quality foams with uniform
pore-size distributions are obtained. Powders are mixed in batches of 500 kg at
Schunk-Honsel in commercial large-scale tumbling mixers with parameters deter-
mined in technological tests. Alternatively, powder mixes can be obtained by aero-
dynamic mixing. For example, Alulight International GmbH Austria mixes alumi-
num and titanium hydride in large containers with 50 80 short pulses of pressur-
ized nitrogen gas.
2.2.2.3 Densification
Powder consolidation can be carried out by various techniques. It has to be ensured
that the foaming agent is completely embedded in the metal matrix and no resi-
dual open porosity remains. One way to obtain foamable precursor material with
nearly 100 % theoretical density is the combined use of cold isostatic pressing
(CIP) and ram extrusion. CIP is first applied to consolidate the powder mix to bil-
lets with a relative density of 70 80 % and a mass of typically 50 kg. These billets
are used in the subsequent extrusion step. Although CIPping is not absolutely ne-
cessary (powders have been filled into thin-walled aluminum cartouches and in-
serted into the extrusion machine without prior consolidation) it has additional ad-
vantages such as the prevention of powder contamination and powder de-mixing.
The CIP billets themselves are not foamable because of their large content of resi-
dual porosity, which causes massive hydrogen losses when the material is heated.
To obtain foamable material, the billets are preheated to 350 hC and extruded as
rods or any other profile. For this a horizontal direct extrusion machine is used
(25MN Schunk-Honsel). The extrusion machine is operated in cycles with a new
billet inserted after each extrusion step. This way rather high outputs can be
achieved.
Foamable material has also been manufactured by rotary continuous extrusion
in the so-called CONFORM process by Mepura (Ranshofen) [10]. Here a rotating
wheel is used to drag the powder into the consolidation chamber from which it is
pulled off in radial direction as a compacted wire. Foamable wires of about 8 mm
diameter were manufactured from wrought alloys containing titanium hydride.
2 Material Definitions, Processing, and Recycling 19
2.2.2.5 Foaming
Heat treatment at temperatures above the solidus temperature of the foamable ma-
trix is necessary to produce the foam structure. The gas released by the decompos-
ing foaming agent may form pores in the solid state but only above the solidus are
bubbles formed and the matrix expands up to a maximum volume, that is to a
minimum density. The density and density distribution of the growing foam can
be controlled by several parameters. The foaming agent content in the precursor
material is obviously important, but furnace temperatures and heating rates also
have an influence [12]. The mold material, the mold shape, and the type of furnace
naturally influence the heating rate and have therefore also to be considered. A
careful control of the heating conditions during foaming is essential for obtaining
high-quality foams. The difficulty is that the liquid foam is thermodynamically un-
stable and conditions change constantly during foaming. There are various inter-
mediate stages: at first only the mold is heated directly, whereas the foamable ma-
terial receives heat only indirectly via heat conduction through the mold. Initially
there are merely some point contacts between the piece of foamable material in the
mold and the mold walls. However, as the temperature increases, the precursor
softens and assumes the contour of the mold thus increasing the transfer of
heat. Moreover, heat transfer via radiation gains importance with rising tempera-
tures. The reflectivity of the mold and precursor surfaces may change during the
process and add a further variable. Finally, after foaming has started, the thermal
conductivity of the precursor rapidly decreases thus reducing heat flow. As soon as
the mold has been filled with foam it has to be cooled down below its solidus tem-
perature to stabilize the foam structure. The phenomena during cooling are also
quite complex and difficult to describe for reasons similar to those mentioned
for the heating phase.
Typical densities of aluminum foams are in the range 0.4 0.8 g/cm3 including
the closed skin around the foam body. The final density of a foamed part can be
simply predicted if the volume of the hollow mold and the mass of the inserted
precursor material are known. The foaming mold may be loaded with several
small pieces or one single piece of precursor. Choosing the latter method (which
is preferred by LKR and SAS [13]) one has to take into account that each piece
of the expanding precursor material has a dense aluminum oxide layer on its sur-
20 2.2 Industrialization of Powder-Compact Foaming Technique
Figure 2.2-4. Left) top of aluminum foam part made by inserting various pieces of foamable
material into the mold (dark figures indicate original size of the precursor pieces). Right) foam
part made of two pieces of precursor without achieving bonding between the two pieces.
face, which has to be broken up by expansion of the individual foam pieces. Incom-
plete foaming may cause the foamed pieces to remain separated even after the
foaming process (Fig. 2.2-4b). A relative movement of the foam pieces to each
other helps to break up the oxide films. Fig. 2.2-4 shows an example of a successful
formation of a foamed body from various pieces of the precursor and an example
of failure. In the former case the location of the original individual foam pieces can
still be identified from the contrast in gray scales between the various regions:
darker gray identifies oxide layers of extruded surfaces, brighter gray is the new
(expanded) surface. This effect is currently exploited to create foam panels and
other foamed parts for making designer objects.
2.2.3
State of Commercialization
Currently the foaming technique described is still in the stage of industrial imple-
mentation. Nevertheless, a number of companies have already made commitments
for a future production and are building up facilities [14]. The joint effort of Schunk
Sintermetalltechnik (Gieûen) and Honsel GmbH&Co KG (Meschede) is one exam-
ple. Owing to their collaboration with Karmann the activities are preferentially di-
rected towards foam and foam sandwich parts with a complex 3D geometry (see
Section 3.3). Alulight International GmbH is another example. It is a joint venture
of SHW (Germany) and Eckart Austria. The company offers aluminum foam
panels in sizes up to 625 mm q 625 mm, with thickness of 8 25 mm. Neuman
Alufoam, another Austrian company, also offers foamable precursor material
(extrusions) and foamed parts.
2 Material Definitions, Processing, and Recycling 21
References
2.3
Making Cellular Metals from Metals other than Aluminum
The previous section was dedicated exclusively to aluminum foams. For many ap-
plications one would like to use cellular materials made from metals or alloys other
than aluminum. There have been some attempts to manufacture metal foams by
simply adapting the powder-compact process originally developed for aluminum to
other metals by adjusting the properties of the foaming agent and the process para-
meters. This procedure was successful in some cases. However, for high-melting
alloys the powder-compact foaming technique is difficult to implement and espe-
cially for titanium no promising results could be obtained. Here alternative routes
based on advanced powder metallurgy yielded better results. Therefore, in the cur-
rent section the topic will be slightly extended from ªfoamedº to ªcellular or porous
metalsº in a more general sense.
22 2.3 Making Cellular Metals from Metals other than Aluminum
2.3.1
Zinc
2.3.2
Lead
Lead and lead alloys such as Pb Sn and Pb Sb can be foamed by another mod-
ification of the process. TiH2 and ZrH2 cannot be used as foaming agents because
of the low melting temperatures of pure lead (327 hC) and even lower solidus tem-
perature of the alloys. Quite good foams have been obtained by using lead(II) car-
bonate as a foaming agent: it decomposes above about 275 hC and releases CO2 and
water, which act as foaming gas. Fig. 2.3-1b shows an example of a lead foam
2.3.3
Titanium
Owing to its high melting temperature (1670 hC) and relatively low density (4.51 g/
cm3), titanium and its alloys are excellent materials for lightweight applications at
elevated temperatures and are widely used in aeronautical applications. Porous ti-
a) b)
Figure 2.3-1. Zn and Pb foams (width of sample is about 5 cm).
2 Material Definitions, Processing, and Recycling 23
tanium structures have an additional potential for weight reduction and could even
be suitable for functional applications if the pore structure were open.
In principle, there are many possible production methods for cellular materials
based on titanium (see Section 2.4), most of them starting from metal powders.
1. Consolidation of slurry-saturated plastic foam.
2. Foaming and sintering of powder slurries.
3. Reaction sintering of elemental powder mixtures.
4. Foaming of powder compacts containing foaming agents (powder-compact
melting process).
5. Hot isostatic pressing and creep expansion of titanium compacts with en-
trapped inert gas.
6. Sintering of hollow spheres.
7. Sintering of compacted or loose powder filler mixtures.
While some of these methods (1 3) have not yet been investigated very intensively,
the feasibility of the foaming agent process (4) for titanium has been demonstrated
[1]. However, owing to the high temperatures during foaming titanium, the reac-
tivity of this metal with practically any non-inert gas and the lack of appropriate
foaming molds, this method is not suitable for producing shaped titanium foam
components. Hot isostatic pressing of titanium powder with gas entrapment (5)
has been successfully developed for some aircraft applications [2]. Metal hollow
spheres (6) can be produced using wet chemical methods for coating Styrofoam
spheres [3]. Shaping and sintering of these hollow structures typically result in ma-
terials with very low porosity.
One of the most promising methods for manufacturing open porous titanium
materials is the sintering of compacted or extruded mixtures of powders and fillers
that contain removable space-holder materials. The materials are mixed and
shaped by conventional PM techniques. After removal of the space holder the
green samples are sintered at temperatures of 1100 1400 hC. Bram and coworkers
use urea and ammonium hydrogen carbonate as space holders [4], which can be
removed by thermal treatment below 200 hC. Depending on the size and shape
Figure 2.3-2. Open porous titanium made by space-holder technique: left) pore size 1 4 mm;
right) pore size about 500 mm, porosity 55 80 %.
24 2.3 Making Cellular Metals from Metals other than Aluminum
Figure 2.3-4. Strength and Young's modulus as a function of density obtained from bending and
tension tests [5].
2 Material Definitions, Processing, and Recycling 25
of the space-holder powder, spherical and angular pores in the range 0.1 2.5 mm
can be obtained, resulting in overall porosities of 70 80 %. It was found that the
sintering activity can be increased by partially substituting titanium by titanium hy-
dride, thus yielding an increased compression strength.
At Fraunhofer IFAM, polymer granules were used as the space holder. They
were removed by a chemical process at temperatures around 130 hC, after pressing.
After space-holder removal, samples are sintered in vacuum at temperatures of
1100 1250 hC. Depending on the particle size of the granules, average pore dia-
meters in the range 200 3000 mm can be obtained. Fig. 2.3-2 shows some typical
samples. Fig. 2.3-3 shows the typical pore structure of samples based on spherical
space-holder granules. Beyond the primary pore structure, some microporosity
(secondary pores) inside the sintered network is visible. It was shown that the sec-
ondary porosity has a strong influence on the overall strength of the samples and
can be reduced by either changing the sintering parameters and/or partially replac-
ing titanium powder by titanium hydride [4]. As for all porous materials, the me-
chanical properties of cellular titanium are a function of density. Fig. 2.3-4 shows
the strength and Young's modulus obtained from bending and tension tests as a
function of density.
2.3.4
Steel
cursor material, extrusion has been successfully used for compacting iron powder
mixtures. The resulting samples are shown in Fig. 2.3-5.
Experiments with powder mixtures of iron and carbon have shown that free car-
bon without any additional foaming agent already leads to a certain degree of po-
rosity. Carbon is oxidized during the foaming process and the resulting gaseous
CO and CO2 creates pores. However, pore size distributions are not uniform and
pore shape is usually rather irregular. The porosity mainly results from large, iso-
lated pores (Fig. 2.3-6). Adding 0.25 % SrCO3 leads to an increase of porosity to
55.5 % (Fig. 2.3-7). The pore structure at this composition appears to be more
homogeneous and the average pore size is obviously lower. Increasing the amount
of SrCO3 results in a further increased porosity (64.3 %). From that it can be con-
cluded that SrCO3 has a significant influence on the achievable porosity and the
maximum expansion. The foaming agent technique has therefore been shown to
be feasible for steel. However, foaming of stainless steel or even superalloys has
not yet been successful and the general state-of-the-art of foaming steel with the
foaming agent method is still far behind the aluminum foaming technology.
Figure 2.3-8.SEM images of sintered specimen: left) stainless steel 316L, 1100 hC, 1 h, particle
size I16 mm; right) Inconel 600, 1250 hC, 1 h, particle size 100 200 mm [4].
28 2.4 Recycling of Cellular Metals
References
2.4
Recycling of Cellular Metals
H. P. Degischer
Cellular metals compete with polymers for some applications. The recyclability of
metals is one of their benefits, enabling ecologically sustainable product life cycles
[1]. Compared to bulk metal products, there are two complications to be tackled in
remelting cellular metals:
x The high surface-to-volume ratio of the order of 100/length unit increases the ex-
tent of surface adsorptions and reactions.
x The low average density, due to the high porosity filled with gas, makes the cel-
lular material float on its melt.
2.4.1
The Remelting of Cellular Metals
Heating cellular aluminum in air enhances the growth of the oxide film not
only in open cellular structures, but also in those foamed by formation of
closed pores in the liquid, or semi-solid state, which usually become permeable
to gas after solidification and cooling. When heating the cellular aluminum up
to the melting point in a short time, about 1 h, the thickness of the oxide film
may reach about 10 mm and any cracks formed will be covered quickly by
2 Material Definitions, Processing, and Recycling 29
ongoing oxidation. The growth will slow down after prolonged heating, when
the thickness reaches about 100 mm [2], which is similar to the cell-wall thickness.
The dehydrated oxide is stable up to about 2000 hC, so the cellular aluminum part is
slowly converted to a cellular alumina structure, which may maintain its macro-
scopic shape if it does not break up under its weight, or an external force.
It is difficult to submerge a cellular part in a melt because it tends to float
and contains a lot of air, which would have to be replaced by melt. Therefore
the cellular structures have to be compressed as much as possible, as is done
with packaging foils, or shredder scrap before melting [1,3]. Cellular metals
can be shredded too and treated as normal shredder scrap. Nevertheless, the
high content of surface oxides reduces the efficiency of metal reclamation.
The only experience is from small batch laboratory trials on scrap from pow-
der-compact foamed aluminum [4], in which efficiencies up to 80 % have
been reached. Oxides and other impurities have to be removed by the usual
cleaning process for melts to the extent necessary for the later production pro-
cess [3]. The purity requirements may be very low for the production of cellular
metals. Impurities in precursor material prepared for powder-compact foaming,
as well as for foaming in the melt, might be advantageous, as long as they act
as nuclei for the formation of pores. There might be even an up-grading of low-
quality scrap when used for the production of foamed metals [5]. The Ti remain-
ing from the blowing agent (usually I0.5 wt.-%) is soluble below 0.12 wt.-% at
the peritectic temperature of the binary aluminum melt [1] and does not de-
grade the quality of the alloy, but provides a grain-refining effect during solidi-
fication. ALPORAS foam contains about 2 wt.-% Ca, which is an element to be
restricted below 0.1 wt.-% in all cast alloys. Ca will be partly oxidized and thus
transferred into the dross, but some care has to be taken in the mix of ingre-
dient scrap for not exceeding the specified impurity levels for the secondary
alloy. The impurity levels of secondary aluminum cast alloys are not as stringent
[1,3], so that scrap of the common types of cellular aluminum can be added to
the remelting furnace.
2.4.2
Recycling of Cellular Metal Matrix Composites
for the production of metallic foams, may not be as stringent as placed on MMC
for bulk components, thus higher quantities of oxide skins and other impurities
may be tolerated, or are even advantageous in foamable aluminum. Figure 2.4-1
shows the material flow cycle for discontinuously reinforced metals and cellular
metals made from MMC indicating the possible interactions between these two
cycles:
x Remelted MMC, with, or without cleaning treatments, can be used as a base ma-
terial for the melt foaming processes and for FORMGRIP.
x Any secondary alloy, including that reclaimed from MMC melts, can be used for
production of cellular structures by any of the known production methods.
x Scrap from MMC foams can be introduced into the MMC cycle, if quality
requirements are met.
The same rules have to be obeyed for the recycling of MMC foams as for the MMC
matrix [11] in addition to the pretreatment (compaction and drying) to reduce gas
evolution. Two main problems have to be tackled when remelting MMC.
x The reactivity of the reinforcement with the melt increases with temperature.
x The dewetting tendency of the reinforcement limits conventional melt cleaning
methods.
Figure 2.4 -1. Recycling of discontinuously reinforced metals and of metal foams based on MMC,
indicating the reuse of secondary MMC for the processing of foamed metals either by the powder-
compact method, or by the melt-foaming technique.
2 Material Definitions, Processing, and Recycling 31
The reaction between 6xxx type wrought alloys and alumina reinforcements is
driven by the Mg content forming MgAl2O4, spinel. An original level of approxi-
mately 2 vol.-% spinel seems to stabilize at just above 3 vol.-% after several remelt-
ing cycles; this does not influence the mechanical properties significantly [12], but
may increase wettability and, consequently, reduce foamability. SiC-reinforced
wrought alloys cannot be remelted without severe aluminum carbide formation
at the interfaces. The formation of Al4C3 would be detrimental as it significantly
reduces the corrosion resistance of the component. Furthermore, it affects the
foamability by increasing the viscosity of the melt and above all by providing wet-
tability. Nonwettability of the additives is necessary for the trapping of gas bubbles.
In the case of Al casting alloys containing 7 12 wt.-% Si reinforced by SiC, there is
the chance to conserve the integrity of SiC by keeping the melt temperature below
750 hC as recommended for primary foundry technology [13].
A detailed study of the recycling of SiC-particle reinforced Al Si casting alloys is
given elsewhere [14], where the quality criteria of the melt are given and remelting,
recycling, and holding practices are described.
x Dry, pre-heated scrap can be added to the melt between 700 and 750 hC.
x During a rest period, usually more dross is formed than on primary material. It
contains mainly oxides and SiC-particles without reducing significantly their con-
tent in the melt.
x The dross has to be skimmed.
x An impeller is introduced to produce strong movement under the surface skin.
x Fluxing and degassing with argon (SF6 may also be used) to remove oxides and
reduce the hydrogen content.
x The melt is allowed to sit and then skimmed.
x The melt is mechanically agitated, without forming a vortex, to distribute the SiC
particles homogeneously.
No loss in particle content was reported and the removal of porosity, oxide films,
and hydrogen were efficient. The amount of dross generated is relatively high
and may amount to more than 10 % of the total weight.
If the ceramic particles should be removed from the aluminum melt, conven-
tional salt addition, or fluxing techniques (as are executed to remove oxide films
[14]) can be applied. Gravity settling allows the fluxed ceramic ingredients to
float to the dross at the top of the aluminum melt. Rotary salt furnace technology
is an established reclamation process to recover aluminum from various mixtures,
including particle reinforced metals; however, this requires 20 50 wt.-% salt [13].
Both wrought and foundry alloys and even machining chips can be recovered
using this technique. Recovery of about 80 % of the available aluminum can be ex-
pected [15].
The efficiency of particle removal by fluxing is related to the probability of con-
tact between the ceramic constituent and the flux. Duralcan, a supplier of particu-
late-reinforced aluminum, proposes to incorporate the salt into a melt agitated by
gas injection [16]. Thus dewetting is achieved with much smaller salt additions
(I1 wt.-% for alumina and about 1.5 wt.-% for SiC) and in combination with
32 2.4 Recycling of Cellular Metals
2.4.3
Conclusions
There are the following possibilities for recycling, or reuse of cellular metals (espe-
cially cellular aluminum).
x Recycling of unreinforced cellular metals can be carried out as for packaging ma-
terial (foils), or shredder scrap, by remelting, with an efficiency probably reduced
to 80 % to produce bulk cast products from the secondary metal. In the case of
recycling ALPORAS, the impurity level of Ca has to be controlled.
x Recycling of MMC foams can be done by extracting the matrix metal for conven-
tional secondary products. The reinforcement may be segregated to the dross and
deposited.
x Recycling of MMC foams can alternatively be achieved by remelting according to
the precautions for MMC recycling, but at a reduced efficiency due to the in-
creased oxide content to produce MMC foams again.
x The foam production itself may be based on an up-grading of secondary material
with higher contents of impurities that increase foamability. In particular, the
cycles for MMC and MMC foams can be closely related.
References
12. D. M. Schuster, M. D. Skibo, R. S. Bruski, and Fabrication of Light Metals and Metal
R. Provencher, G. Riverin, ªThe recycling and Matrix Composites, Montreal, TMS, 1992,
reclamation of metal-matrix compositesº p. 598 604.
J. Metals 1993, 45(May), 26 30. 15. T. F. Klimowicz, ªThe large scale commer-
13. Duralcan, Duralcan Composite Casting cialization of aluminum natrix compositesº
Guidelines, Duralcan USA, San Diego, CA J. Metals 1994, Nov, 49 53.
1990. 16. Duralcan, Aluminium Recovery from Metal
14. R. Provencher, G. Riverin, C. Celik, Matrix Composite Scrap, Duralcan USA, Novi,
ªRecycling of Duralcan Aluminium Metal Michigan 1996.
Matrix Compositesº in Proc. Adv. Production
2.5
The Physics of Foaming: Structure Formation and Stability
A foam is a dispersion of gas bubbles in a liquid in which the bubbles are de-
formed due to their mutual interaction. Cellular metals produced via gas bubbles
in liquid metal are commonly named metal foams although they do not strictly
meet the definition above after solidification. Owing to the surface energy neces-
sary to form the metal±gas interface a foam is never in equilibrium and hence per-
manently trying to lower the internal energy by reducing the internal surface. That
is, the cellular structural state of a foam evolves with time and its actual structure is
a function of its history including all thermal and mechanical influences. The foam
structure is normally strongly disordered and evolves by some combination of
three basic mechanisms: bubble coalescence via film rupture; bubble coarsening
via diffusion of gas from smaller to larger bubbles; and drainage downwards
and out from the foam in response to gravity [1]. It is only in recent years that
some progress has been made towards an understanding of the basic mechanisms
governing the temporal evolution of foams [1±5].
Owing to their opacity and the high temperatures, in-situ observation of the
structure evolution of metal foams is difficult [6]. Fortunately, the solidification pro-
cess is in general much faster than the evolution processes at least as far as the
evolution is not caused by expansion. That is, nearly full information of the
foam formation process can be extracted from ex-situ investigation of foamed sam-
ples in different stages of expansion.
34 2.5 The Physics of Foaming: Structure Formation and Stability
2.5.1
Isolated Gas Bubble in a Melt
S 3 _2 R_ 1 2s
RR R S 4v (Pbubble s s PT ) (1)
2 R r R
That is, inertia effects delaying bubble expansion can be neglected if foam forma-
tion is on a time scale of seconds. The contribution of viscous forces for liquid me-
tals is normally very small, for example about 4nr(R/RÇ) z 10 6 bar with RÇ 1 mm/
s; R 100 mm; n 1 mm2/s. On the other hand, if foaming takes place in the
semi-liquid state, where the viscosity of the metal is several orders of magnitude
higher than in the liquid state, viscous forces might delay bubble expansion.
Bubble expansion takes generally more than one second for commercially known
foaming methods for metals [9]. In this case, the viscous and inertia forces can be
neglected and the Rayleigh equation reduces to pressure equilibrium at the gas±li-
quid interface
s s
Pbubble PT S 2 P0 S rgh S 2 (3)
R R
For a pure aluminum melt and bubble radii R of 100 mm and 10 mm the rising
velocity is about 1 cm/s and 100 mm/s, respectively. Additives in the fluid like SiC
or Al2O3 particles influence the movement of the bubbles and are able to stabilize
them [10±12]. They have an effect on both the viscosity of the melt and the surface
tension. How these particles actually operate and how their action can be opti-
mized is not yet understood and still a matter of research.
A gas bubble grows or shrinks due to gas exchange with the surrounding melt
[13]. There is a gas flow from the liquid into the bubble if the concentration of dis-
solved gas in the liquid, as a result of the decomposition of a foaming agent for
example, is higher than the equilibrium concentration in the liquid given by
Henry's law for a given gas pressure in the bubble. Since hydrogen, which is pre-
ferentially used as foaming gas, dissociates when dissolved in aluminum the equi-
librium hydrogen concentration, ceq, at the gas±liquid interface is governed by
Sievert's law, a special form of Henry's law [14]
2760
p mol
ceq 1:4 q 10s3 q 10s T Pbubble p (5)
cm3 bar
2.5.2
Agglomeration of Bubbles: Foam
Figure 2.5-3. 2D-lattice Boltzmann simulation critical lower value. The melt boils, the number
of the growth of bubble nuclei by in-situ gas of bubbles decreases, and gas is lost to the
generation. The cell walls are not stabilized, so environment. The structure collapses once a
two bubbles coalesce if the cell wall reaches a critical expansion factor is reached [31].
Figure 2.5-4. Expansion stages of an AlMg1 completely liquid. Already at the very beginning
foam with 0.4 % TiH2 produced by powder of foam expansion coalescence occurs fre-
compaction. By applying a high ambient pres- quently indicating that the initial number of
sure of 100 bar during heating bubble nuclea- nuclei will not have much influence on the
tion was suppressed until the sample was residual foam structure.
cell walls are curved due to the pressure difference of neighboring cells. One effec-
tive way to characterize the cell structure is by the shape factor, F, calculated from a
2D cut through the foam [23] (see also Section 4.1)
4p X
n
ai
F (8)
n i li2
last expression is deduced from a simple cubic plate model for the cells with the
assumption that the volume of the cell-forming material per cell is constant. Actu-
ally, the experimental data (see Fig. 2.5-7) is much better fitted by
d 3d
D p for rrel II 1 (9)
1s 3
1 s rrel rrel
which follows from a cubic plate model assuming a constant mean cell-wall thick-
ness, d. That is, the mean cell-wall thickness is constant during expansion. It de-
pends on the alloy and is found between 50 mm and 300 mm for aluminum
foams. Consequently, given a fixed alloy cell size and relative density are intimately
related and can hardly be influenced by processing parameters [23]. In order to
generate aluminum foams with a lower density and smaller cells the mean cell-
wall thickness has to be reduced. How the cell walls of metal foams are stabilized
by the particles and which properties of them (quantity, size, form) determine the
mean cell-wall thickness is not yet understood and a matter of flow research [18].
Coalescence, rupture of a cell wall, is a mechanical process that depends on the
mechanical stability of the membrane. It has statistical character and occurs if a
local fluctuation of the film thickness leads to faster local thinning because it is
not compensated by restoring forces. The different physical mechanisms that are
responsible for the mechanical stability of a membrane have been extensively in-
vestigated for aqueous films [5,15]. Experimental studies of the stability of metal
membranes are not known in literature. Physical forces between the two surfaces
of the film like the van der Waals or electromagnetic interaction are only relevant
for very thin films (I1 mm). For thicker films, as it is the case for metal foams,
there are two effects that might lead to a kind of elasticity of the membrane.
The Gibbs' effect [15] takes into account that a local thinning of the membrane
leads to a local thinning of the surfactants. As a result, the surface energy increases
at this point leading to a restoring force. The Marangoni effect [15] describes that a
gradient of the surface energy leads to a flow of surfactants taking with it a thin
film of fluid in the direction of lower concentration. In order to maximize both ef-
fects the surfactant should minimize the surface energy. At the moment, it is not
clear whether the stability of metal membranes is based on one of these two me-
chanisms. In metals, melt surfaces are covered with oxides and other particles that
might have the required effects on the surface energy.
The origin for cell-wall instability is cell-wall thinning, which can be traced back
to different mechanisms. The first one is simply related to foam expansion itself. If
a cell expands the cell-wall thickness decreases due to mass conservation. During
foam expansion the principal mechanism for cell-wall thinning is by growth. The
second mechanism of cell-wall thinning is drainage due to gravity and capillary
forces. Drainage determines the foam stability in the absence of further expansion
due to gas release, that is, the time that a given foam structure is maintained with-
out destruction. Gravity induces a melt flow from the top to the bottom of the foam
and generates a density gradient [4,5]. Neglecting capillary forces, the draining ve-
locity, vD, can be estimated in a first approximation by the balance of gravity and
viscous stress from shear flow within the Plateau borders [1]
grd2edge
vD Z (10)
h
That is, the draining velocity increases with the cell-edge thickness, dedge.
Capillary forces arise due to the different curvature of the interfaces. They force
the melt from the cell membranes to the cell edges (Fig. 2.5-8). The rate of cell-wall
thinning due to this pressure gradient can be modeled using a Reynolds-type
equation for the flow between two circular parallel discs with immobile wall sur-
faces [5]
2sx 3 (11)
x_ s
3hR2 RPB
meter. As a result, pores are deformed and also destroyed during mold filling by
mechanical forces or a non-uniform foaming velocity (Fig. 2.5-10). Hence, the
internal foam structure is strongly influenced by the particular production pro-
cedure.
References
2.6
Infiltration and the Replication Process for Producing Metal Sponges
The process of replicating a structure has been known to the metallurgist for cen-
turies. For example, casting can be thought of as a replication procedure in which a
metal is used to reproduce the negative form of the mold. In the study of porous
materials, however, replication refers to a process used to replicate the open-pore
architecture of a porous material.
44 2.6 Infiltration and the Replication Process for Producing Metal Sponges
2.6.1
Replication
More specifically, this process can be defined as a general three-step procedure for
the production of highly porous materials (Fig. 2.6-1). The three steps are:
1. preparation of a removable pattern;
2. infiltration of the pattern followed by solidification (or an alternative process,
such as curing or cross-linking, which rigidifies the infiltrate); and
3. removal of the pattern.
In some cases, a fourth step, such as pyrolysis, may be required to transform the
porous network into the desired phase. In this text, we focus on the use of the re-
plication process for the production of cellular metals with open porosity. Several
authors have noted that metal (or metallic) sponge is perhaps the most appropriate
terminology for these materials as both the term ªfoamº and ªcellularº imply a
closed-cell architecture [1,2], while porous metals produced by replication must
have an open-pore network. We will thus refer to metals with open porosity of
any density I0.8, produced by replication, as metal sponge.
This process is distinct from deposition techniques, such as those used by Inco
Limited (Swansea SA6 5QR, Wales, UK) to produce porous nickel substrates, called
Incofoam [3]. In that process, a temporary pattern is used, but the pore space is not
replicated: instead, the pattern is coated and the pore space only partially filled.
Sintering methods that utilize a space holder are also not considered replication
for two reasons: the micro-architectural features of the pores are determined by
the metal powder size in addition to the space-holding material and infiltration
is not one of the processing steps. The so-called lattice block materials (LBM)
are prepared using polymer patterns of truss structures and investment tech-
niques; thus the lattice block is rather an engineered structure and cannot really
be considered a metal sponge, which can be ªshapedº and ªformedº while main-
taining the statistics of its properties, as long as the cell sizes are of millimeter
or more dimensions. For more details on lattice block materials contact JAMCORP
(17 Jonspin Road, Wilmington, MA 01887-1020 USA). Although we limit this dis-
cussion to metals, we note in passing that the polymeric sponge technique used to
produce porous ceramics also does not replicate the pattern and should not be con-
sidered a replication process.
In a metallurgical context, the replication technique can be used to produce
metal sponge that is difficult or impossible to produce by other processing meth-
odologies. Replicated metal sponge, for example, by necessity has an open-pore
structure, which is a fundamental difference compared to most commercially avail-
able foamed metals. The structure or architecture of metal sponge produced by re-
plication is flexible and determined by the pattern: porosity as high as 98 % [4] and
as low as 55 % [2] for different pattern materials have been reported and pore sizes
as small as 10 mm have been achieved [5]. In addition, metal sponge can be
produced from virtually any alloy that can be cast (provided a suitable pattern ma-
terial exists). The replication process has been used extensively for the production
of various porous materials including carbon [6], silicon carbide [7], aluminum al-
loys [2,4,5,8±13], magnesium alloys [14,15] and alloys based on iron and nickel
[4,16].
In the following sections we provide details about the production of metal
sponges by replication techniques. We first describe the important physical phe-
nomena that govern each of the three basic steps of the replication process (with
some emphasis on sponge produced from salt patterns for which the most infor-
mation exists in the literature). We then provide some physical and mechanical
properties of metal sponge produced in this manner.
46 2.6 Infiltration and the Replication Process for Producing Metal Sponges
2.6.2
The Replication Process: General Principles
is determined by the size of the initial spheres (or particles) and the ratio of poly-
styrene to resin. This type of pattern has the advantage that it is not necessary
to physically remove solid grains from the structure after infiltration, as the pattern
is essentially burned away at the end of the process. On the other hand, this pat-
tern must be infiltrated relatively cold to avoid burning of the pattern during infil-
tration (i. e., while the metal is still liquid) and is thus limited to low-melting alloys
such as aluminum, magnesium, and zinc.
The fourth class of pattern material is leachable: these patterns are removed by
dissolution in an appropriate solvent. The process of dissolution (or leaching) is
less restrictive on size and density than insoluble refractory materials. In practice
however, patterns of leachable material are created from granules or powder, thus
the density is related to the packing efficiency of the granules, which itself depends
on the size (and size distribution) of the granules.
Sodium chloride (NaCl) has been employed most extensively as a leachable pat-
tern, because it is inexpensive and easy to handle. Sodium chloride patterns used
for the production of porous aluminum were first reported [8,9] and patented [10]
in the 1960s, and have advanced more recently [5] (there are also reports of their
use for carbon sponge [6] and for SiC sponge [7], both with a pyrolysis step).
Salt patterns are limited by their melting point; NaCl, for example, is limited to
aluminum and lower-melting alloys, while NaF could potentially be used at tem-
peratures greater than 900 hC. Additionally, highly concentrated saline solutions
are generated during dissolution and can cause significant corrosion in some al-
loys.
Salt grains also have the advantage that they can be sintered to enhance the con-
nectivity of the salt, change the structure of the pattern, and create a free standing
preform that can be handled and subsequently infiltrated [5±7]. The sintering step
is not necessary, however, for two reasons: transient bonding of the grains (perhaps
due to humidity in the salt [8]) may occur during preheating the salt prior to and
during infiltration, and incomplete penetration of the melt locally where individual
grains meet, provides the necessary open-network for salt removal.
Sintering processes depend on a number of parameters, including particle size,
temperature, atmosphere, residual stresses, and time. Such numerous process
parameters might explain some of the varying observations in the literature with
regard to the sintering of sodium chloride. Incidentally, these varying observations
point to the fact that sodium chloride may be a good choice for a pattern material
because of flexibility in controlling the sintering mechanisms and thus the micro-
architectural features of the pattern.
There is a consensus in the literature that large sodium chloride particles sinter
predominantly by an evaporation±condensation mechanism, in which the necks
between the particles grow without densification of the salt compact. Small particle
compacts, on the other hand, can densify significantly during sintering, due to the
predominance of other mechanisms where the center-to-center distances between
particles can decrease, such as bulk diffusion [18±21]. The minimum particle size
that can sinter without densification has been estimated to be about 150 mm
[19,20]. Sintering in vacuum may promote the evaporation±condensation mechan-
48 2.6 Infiltration and the Replication Process for Producing Metal Sponges
ism for smaller particles [22] as supported by the pressure dependence that was
noted for the rate of neck growth when non-densifying sintering mechanisms
are dominant [23].
There are advantages to both classes of sintering mechanisms with respect to
producing metal sponge. Non-densifying mechanisms, such as evaporation±con-
densation should develop a structure dictated by capillary equilibrium: these are
well known [24]. The pore architecture in such a case is expected to be similar
to the regular structures of polymeric sponge. Densifying mechanisms, on the
other hand, provide a means to achieve relative densities significantly greater
(and thus significantly lower for the sponge) than those attainable by packing
and pressing of particles. Sintered salt densities as high as 90 % have been reported
with nearly all the pores as open porosity, although for a particle size of 1 mm,
which would pose challenges in the subsequent processing steps [25].
2.6.2.2 Infiltration
Ideally, infiltration under the influence of gravity is preferred as it does not require
special equipment or procedures. In most cases, however, the metal does not wet
the pattern and some additional force must be applied to the melt to promote (or
assist) infiltration and effect a uniform distribution of metal within the pattern.
Vacuum and low-pressure assisted casting procedures are standard foundry tech-
nologies and, in many cases, these are sufficient for infiltration. Indeed, it has
been reported that spontaneous infiltration is possible when infiltrating salt grains
greater than 4 mm in size. Vibration slightly reduces the size of salt grains that can
be infiltrated under gravity, while vacuum applied to the pattern with a slight over-
pressure on the melt has also been used to infiltrate grains as small as a few hun-
dred micrometers in diameter [8].
The specific pressure required to initiate infiltration (the threshold pressure, P0)
depends to a large extent on the size of the pores to be infiltrated and the volume
fraction of these pores. A detailed physical description of infiltration is beyond the
scope of this work; more detailed information is available elsewhere [26±28]. To the
authors' knowledge, the pressure required to initiate infiltration has not been ex-
plored with specific emphasis on patterns used for replication processes; however,
expressions derived for the threshold pressure in the infiltration of packed ceramic
beds with pure aluminum should be directly relevant. Garcia-Cordovilla et al. offer
a simple semi-empirical relationship [27]
f
P0 MPa z 16 s 0:09
(1 s f )D
where f is the volume fraction solid (or in this context the fractional density of the
pattern) and D is a characteristic diameter of the solid in micrometers. This rela-
tionship is based on data for particles in the range 10±100 mm packed to a density
of 50±60 % and infiltrated with pure aluminum, but it should provide reasonable
magnitude estimates of the pressures required for infiltration over a broader range
2 Material Definitions, Processing, and Recycling 49
Polymer pattern materials can be removed by burning in air as is the case for
polystyrene-resin patterns [11]. The cleanliness of this burning is an important
issue in an industrial setting. These patterns, however, have the advantage that
complex parts can be machined before removal of the pattern and without signifi-
cant damage to the structure of the sponge.
In some cases, leachable patterns can also be machined to complex geometries
prior to leaching and without damaging the architecture of the sponge. In the case
of salt, removal of the pattern is by immersion in water, causing dissolution of the
salt. Dissolution is primarily a diffusive process, as the dissolved salt ions diffuse
from their place-holding position into the water bath through the narrow channels
in the sponge where the water is essentially stagnant. Thus, the rate and total time
of dissolution depend strongly on the size of the specimen. Salt is more difficult to
remove from small pores, particularly if gas is entrapped in the pores; this is the
so-called ªgas lockº phenomenon, which can block the penetration of fluid into the
porous network [8].
Corrosion of metal sponge during leaching of the pattern can be a problem due
to the large surface area of the porous network, thus the salt concentration and the
immersion time must be minimized. This is accomplished by simply changing the
water periodically, however, it is important that the sponge is not drained at this
point, as this will cause gas to be entrapped in pores and promote ªgas lock.º Ad-
ditionally, impurities in the water should be controlled to avoid precipitation of
other salts in these concentrated solutions. Aluminum alloys (the metals for
which salt patterns are most applicable) tend to be relatively corrosion-resistant
in salt solutions. The aluminum±copper alloys are among the least corrosion-resis-
tant in salt solutions, but sponge can still be prepared from these alloys without
degradation by corrosion. Electrochemical protection could also offer additional
protection from corrosion.
A large volume of water is not necessary as salt-saturated water is denser than
distilled water, and hence falls to the bottom of the container where it remains
in the absence of convection. As a result, during the dissolution process a large
concentration gradient is maintained between the outer surface of the salt±metal
composite suspended near the surface of the dissolution bath (but completely im-
mersed) and the bottom of the container. For large pores and low-density sponge,
mixing may modestly enhance dissolution if the velocity of the fluid in the porous
network is affected. In most cases, however, this condition is not met and mixing
the water primarily distributes the salt concentration and reduces the diffusion gra-
dient in the water bath, in turn slowing the leaching process.
The final stage of salt removal from the sponge is rinsing. This can be accom-
plished by flowing water through the structure, or simply by changing the water
several times to incrementally remove the saturated fluid; this process is greatly
facilitated by the higher density of salt solutions. Pressurized gas is also effective
for removing the residual fluid in the sponge such that dissolved salts are not
left in the structure when this water evaporates.
Other methods may contribute to more efficient (faster) removal of salt, but
these must enhance the rate of diffusion of the salt in the fluid. For example, ul-
2 Material Definitions, Processing, and Recycling 51
trasonic vibration has been proposed as a useful method as it should increase the
rate of diffusion with the added benefit of freeing gas bubbles entrapped in the po-
rous network [8].
2.6.3
Physical and Mechanical Properties of Metal Sponge
Table 2.6-1. Summary of replication techniques used for the production of metal sponge.
Pattern material Metal Infiltration technique Pattern Metal sponge characteristics* Reference
removal
Pore size (mm) Porosity (%)
Investment of poly- Al vacuum assisted water spray 4.5 93±96 [12, 13]
urethane sponge (SG91A)
Investment of poly- Al (Cu, Zn, Fe, vacuum or pressure (not reported) 0.85±2.5 92±98 [4]
urethane sponge Ni, Co) assisted
Pelleted casting sand pig Fe vacuum assisted pressurized water 1±5 60±63 [15]
Polystyrene and resin Al modified die casting burning 1.2±3.7 73±86 [11]
NaCl Al gravity, vibration assisted, leaching 0.15±5 (60±63) full range [8]
pressure assisted by not reported
vacuum
2.6 Infiltration and the Replication Process for Producing Metal Sponges
Figure 2.6-2. Scanning electron micrograph of aluminum sponge produced by replication from
sintered NaCl patterns.
2.6.4
Conclusions
Replication processing provides a versatile and attractive method for the produc-
tion of metal sponge (often called open-celled metal foam). Advantages of this pro-
cess comprise considerable control of the pore size, density, and internal architec-
ture of the sponge, the capacity for net or near-net shape component production,
and the potential for production of components that are part dense, part porous.
Also, the open-porosity that is intrinsic to metal sponge can be put to advantage
in several applications that require fluid flow through the porous metal, such as
filters and heat-exchangers.
The replication process exists in a few variants, which differ mostly according to
the nature of the pattern: continuous refractory, discontinuous refractory, burnable,
and leachable patterns are four classes of patterns that have been employed to date.
The range of density and pore size that can be achieved in replicated metal sponges
depends largely on the pattern material and also on the infiltration process. With
leachable patterns, in particular, these ranges have been shown to be comparatively
wide; at present these also seem to be complementary to what can be achieved with
the methods utilizing insoluble patterns. There is, though, significant room for in-
novation in replication technology as related to the production of metal sponge and
there exists perhaps even opportunities for cross-pollination between the methods
described here.
References
1. D. Weaire, M. A. Fortes, Adv. Phys. 1994, 43, 8. L. Polonsky, S. Lipson, H. Markus, Modern
685. Castings 1961, 39, 57.
2. J. Banhart, Adv. Eng. Mater. 2000, 2, 188. 9. H. Seliger, U. Deuther, Freiburger For-
3. J. Babjak, V. A. Ettel, V. Paserin, US Patent schungshefte 1965, 103, 129.
4 957 543, 1990. 10. H. A. Kuchek, US Patent 3 236 706, 1966.
4. I. Wagner, C. Hintz, P. R. Sahm, in Metal 11. L. Ma, Z. Song, D. He, Scripta Mater. 1999,
Matrix Composites and Metallic Foams, Proc. 41, 785.
Euromat 99, Munich, Germany, T. W. Clyne, F. 12. Y. Yamada, K. Shimojima, Y. Sakaguchi, M.
Simancik (eds), DGM/Wiley-VCH, Weinheim Mabuchi, M. Nakamura, T. Asahina, T.
2000, p. 40. Mukai, H. Kanahashi, K. Higashi, Mater. Sci.
5. C. San Marchi, J.-F. Despois, A. Mortensen, Eng. 1999, A272, 455.
in Metal Matrix Composites and Metallic 13. Y. Yamada, K. Shimojima, Y. Sakaguchi, M.
Foams, Proc. Euromat 99, Munich, Germany, Mabuchi, M. Nakamura, T. Asahina, T.
T. W. Clyne, F. Simancik (eds), DGM/Wiley- Mukai, H. Kanahashi, K. Higashi, Mater. Sci.
VCH, Weinheim 2000, p. 34. Eng. 2000, A280, 225.
6. R. W. Pekala, R. W. Hopper, J. Mater. Sci. 14. Y. Yamada, K. Shimojima, Y. Sakaguchi, M.
1987, 22, 1840. Mabuchi, M. Nakamura, T. Asahina, T.
7. T. J. Fitzgerald, V. J. Michaud, A. Mortensen, Mukai, H. Kanahashi, K. Higashi, J. Mater.
J. Mater. Sci. 1995, 30, 1037. Sci. Lett. 1999, 18, 1477.
56 2.7 Solid-State and Deposition Methods
15. Y. Yamada, K. Shimojima, Y. Sakaguchi, M. 24. P. J. Wray, Acta Metall. 1976, 24, 125.
Mabuchi, M. Nakamura, T. Asahina, T. 25. R. W. Davidge, Proc. Br. Ceram. Soc. 1969,
Mukai, H. Kanahashi, K. Higashi, Adv. Eng. 12, 245.
Mater. 2000, 2, 184. 26. A. Mortensen, I. Jin, Int. Mater. Rev. 1992,
16. U. Mohr, W. Bleck, in Metal Matrix Com- 37, 101.
posites and Metallic Foams, Proc. Euromat 99, 27. C. Garcia-Cordovilla, E. Louis, J. Narciso,
Munich, Germany, T. W. Clyne, F. Simancik Acta Mater. 1999, 47, 4461.
(eds), Wiley-VCH, Weinheim 2000, p. 28. 28. A. Mortensen, in Metal Matrix Composites,
17. R. M. German, Particle Packing Character- A. Kelly, C. Zweben (eds), Vol. 3 of Compre-
istics, Metal Powder Industries Federation, hensive Composite Materials, T. W. Clyne (ed.),
Princeton, NJ 1989. Pergamon, Oxford, UK 2000, Ch. 21.
18. C. S. Morgan, L. L. Hall, C. S. Yust, J. Am. 29. A. Mortensen, J. A. Cornie, Metall. Trans.
Ceram. Soc. 1963, 46, 559. 1987, 18A, 1160.
19. A. A. Ammar, D. W. Budworth, Proc. Br. 30. A. Mortensen, T. Wong, Metall. Trans. 1990,
Ceram. Soc. 1965, 3, 185. 21A, 2257.
20. W. J. Tomlinson, G. Astle, J. Mater. Sci. 31. T. Mukai, H. Kanahashi, Y. Yamada, K.
1976, 11, 2162. Shimojima, M. Mabuchi, T. G. Nieh, K.
21. R. J. Thompson, Z. A. Munir, J. Am. Ceram. Higashi, Scripta Mater. 1999, 41, 365.
Soc. 1982, 65, 312. 32. H. Kanahashi, T. Mukai, Y. Yamada, K.
22. W. D. Kingery, M. Berg, J. Appl. Phys. 1955, Shimojima, M. Mabuchi, T. G. Nieh, K.
26, 1205. Higashi, Mater. Sci. Eng. 2000, A280, 349.
23. J. B. Moser, D. H. Whitmore, J. Appl. Phys. 33. W. Thiele, Metal Mater. 1972, 6, 349.
1960, 31, 488.
2.7
Solid-State and Deposition Methods
In this chapter, solid-state and deposition methods for the manufacturing of cel-
lular metals are reviewed with regard to their most important processing features.
Discussion will be limited to methods producing predominantly 3D structures.
Typical values of the cell size, cell wall thickness, and obtainable relative densities
are cited wherever available. Earlier overviews including both methods of manufac-
turing are available [1,2].
The term ªsolid-state methodº means that directly before the formation of the
cellular structure, the metal is in the solid state. However, in some cases limited
amounts of liquid phase appear in the process: during liquid-phase sintering or
brazing of single cells. Except for foaming of solids, all solid-state methods require
a sintering step before the final cellular metallic structure is obtained. Naturally,
powder metallurgical methods play a dominant role in this area.
2 Material Definitions, Processing, and Recycling 57
Porosity in cellular structures made from powders often exhibits two different
size scales: a macroporosity of the order of the targeted cell size, and a micropo-
rosity in the order of the primary particle or, in case of reactive sintering, the
grain size.
For all powder metallurgical processes, the cost of the powders is a major con-
cern. Some of the methods, especially those using slurries of metal powders dis-
persed in a dissolved binder, are capable of using cheap oxides and thus have
the potential for a significant cost reduction.
Deposition methods need to mobilize the metal in order to transport it onto a
porous substrate. This is normally done by transforming the metal into the ionic
or vapor state. From the method of manufacture, it can be deduced that the deposi-
tion methods are only capable of generating an open-cell morphology. Usually, they
also imply a heat treatment for the removal of the precursor on which the metal is
deposited.
A practical differentiation between the methods in question can be made with the
help of two manufacturing parameters, as is shown in Fig. 2.7-1. The chosen criteria,
ªformation of cellular structureº and ªformation of cellsº have direct implications for
the processing itself, as well as for the resulting properties of the structure.
According to this classification, the formation of the cellular structure can be ac-
complished by building it up from single cells, which do not necessarily have to be
in the final metallic state. Alternatively, the structure can be built up from bulk.
The formation of the cells themselves can either be carried out with the help of
a lost core, or in a coreless manner. Within this framework the deposition meth-
ods, for example, constitute a small subset of the lost core bulk processes.
Solid state
and deposition methods
coreless methods
Generally, formation from single cells has the potential of a very close control of
the cellular properties. However, in most cases the single cell processes require
more processing steps than bulk formation processes.
Processes using lost cores usually are prone to introducing unwanted interstitials
like carbon and oxygen into the base material, since the cores are predominantly hy-
drocarbons, which have to be removed thermally. Another disadvantage compared to
coreless processes is the additional time required for the removal of the core.
Generally speaking, there is a remarkable variety of solid-state and deposition
methods for making cellular metals, which all have their specific strengths and
weaknesses. It therefore depends very much on the final application, which
method may be best suited in a given situation.
2.7.1
Formation from Single Cells: Coreless Methods
a) Surface 1.0 mm
achieved with stainless steel structures made from reduced iron and chromium
oxide powders [9], Fig. 2.7-3.
2.7.2
Formation from Single Cells: Lost Core Methods
Figure 2.7-4. Cross section of a bronze hollow-sphere structure made by cementation and sin-
tering (courtesy Institute for Chemical Technology of Inorganic Materials, Technical University of
Vienna).
2 Material Definitions, Processing, and Recycling 61
Figure 2.7-5. Brazed nickel hollow-sphere structures with sphere diameter of 4.5 mm made from
galvanically manufactured hollow spheres (courtesy ATECA, Montauban, France).
are heat treated in the same way as the single spheres. In principle, this allows for
the production of hollow-sphere structures from arbitrary metals and alloys.
In order to enhance the mechanical properties of the resulting structures, large
contact areas between the spheres are achieved by applying a slight pressure to the
green spheres during the forming step. In this stage, the elastic Styrofoam core
helps to prevent buckling and leads to a more uniform distribution of the cell
pre-deformation. A typical example of the resulting structure is shown in
Fig. 2.7-7.
The attainable range of the sphere diameters is 0.5±10 mm. Wall thicknesses
have been made from 10 to 1000 mm. Relative densities can be as low as 0.03. Re-
cent overviews of the processing and properties of such structures are given else-
where [17,18].
Materials currently under investigation are numerous, with a focus on stainless
steel 316L [19] and low-cost hollow-sphere structures made from iron oxides [20].
2.7.3
Bulk Formation: Coreless Methods
Figure 2.7-8. Cross-sectional cut of sintered fiber structure made from melt-extracted short
fibers from FeCrAl 23.15. From left to right: relative density 0.3, 0.2, 0.1 (courtesy IFAM,
Department of Powder Metallurgy and Composite Materials, Dresden).
64 2.7 Solid-State and Deposition Methods
Using the melt extraction process for making short fibers directly from the melt,
the materials limitations of the mechanical fiber manufacturing routes can be over-
come [24,25]. Highly porous sintered structures from intermetallics such as Ni3Al
[26] and FeAl, as well as from many other ferrous and nonferrous alloys have been
made. Equivalent fiber diameters are 50±150 mm, fiber length may vary between
3 and 25 mm. This allows for relative densities between 0.5 and 0.03 in the final
structure, Fig. 2.7-8.
Figure 2.7-9.Porous TiAl3 structure prepared from elemental powder mixture. Large pores
correspond to primary Al particles (courtesy IFAM, Department of Powder Metallurgy and
Composites).
b) HIP Consolidation
(1040oC, 100 - 200MPa, 2h)
Isolated
pressurized
voids
c) Hot Rolling
(approx. 850oC, 6-20 passes)
Thinning of
facesheet
Internal gas
Changes in
pressure
pore shape,
matrix
microstructure
Vacuum
furnace
Isolated
porosity
Sandwich
(<40%)
panel
Facesheet
Figure 2.7-10. LDC process scheme (courtesy IPML, University of Virginia, Charlottesville).
2 Material Definitions, Processing, and Recycling 67
2.7.4
Bulk Formation: Lost Core Methods
Figure 2.7-11. Titanium sample with relative density of 0.3 made by space holder method
(courtesy Research Centre Ju
È lich, Institute for Materials and Processes in Energy Systems).
68 2.7 Solid-State and Deposition Methods
References
3
Secondary Treatment of Cellular Metals
F. Simancik
The introduction of new materials into practice requires suitable technologies for
their machining, forming, surface coating, or joining. Many papers in the literature
deal with the preparation of cellular metals and with the evaluation of their me-
chanical and physical properties, but only limited attention is paid to their second-
ary treatment. There is no doubt that the following unusual features of cellular me-
tals require specific approaches for shaping and fabrication.
x Crushability of the structure (difficult forming and machining).
x High sensitivity to tensile loading (difficult forming).
x Defects in cell walls and in surface skin (difficult coating).
x Outer surface that is usually covered with oxides (difficult coating, brazing, weld-
ing).
x Presence of the melt stabilizing ceramic particles (difficult machining).
Components made of metallic foam are usually prepared in near-net-shapes by
foaming in the mold. These porous components are always covered with a compact
surface skin. However, the skin usually contains various defects like small holes or
microcracks, so it cannot be considered fully dense. The skin significantly im-
proves the properties and the appearance of the foamed part, so it is undesirable
to remove it. In addition, removing the skin would be expensive.
There are other cellular metals made using various templates or placeholders.
They have open-cell structures and no outer skin (the template must be removed
from the structure after solidification). It is obvious that the secondary treatment of
these two kinds of structures will be very different.
Machining
Machining of net-shape foams breaks the dense surface skin and reveals the inner
structure of the pores. Machining is best avoided at the design stage. Nevertheless,
even well-designed components will need to be cut and drilled. Cellular metals can,
in principle, be cut by all conventional methods, but achieving a high-quality sur-
face is substantially more costly than for traditional materials. If ceramic particles
are used for stabilizing the liquid foam, the excessive wear on the machine tools
72 3 Secondary Treatment of Cellular Metals
has to be taken into account, especially if SiC particles are present. Conventional
machining causes bending of thin pore walls in ductile alloys and fracture in brittle
alloys. Pore walls yield or fracture at their thinnest point, which is often out of the
cutting plane. This leads to poor quality of the cutting surface and low precision.
The accuracy and quality of the cutting plane are significantly affected by partial
melting of the thin pore walls and their subsequent diffusion bonding to the cut-
ting tool. High porosity and the resultant low thermal conductivity do not allow for
efficient cooling by transport of heat into the surroundings. On the other hand,
melting and bending of pore walls can be utilized for the strengthening of the cut-
ting surface (bent pore walls are pressed into the surface pores, which improves
their apparent density).
If high precision and a good quality machined surface is required, traditional
methods of machining cannot be used. Electric discharge machining was demon-
strated to be most suitable, as it avoids the mechanical effect on the pore walls and
thus their bending and failure. Material in the cutting plane simply burns out. The
result is a clean cutting plane (without material pressed into the surface pores) and
good dimensional accuracy. The speed of electric discharge machining is substan-
tially greater in porous materials than with dense ones. The main disadvantage of
the process is the need to use dielectric fluid, which penetrates into the porous
structure and needs to be removed after machining by burning: this is environ-
mentally harmful. If a simple cut is required, wire electric discharge machining
in water can be used to overcome this disadvantage. A water beam with abrasive
can also be used for the precise cutting of cellular metals. The quality of the cutting
plane is substantially better than in the case of classical band saws (the bending of
pore walls does not take place). As the cutting speed and the thickness of the plate
increases, the quality of the bottom edge of the cutting plane deteriorates owing to
scatter of the beam. The same effect takes place in plasma or laser cutting.
If the component is foamed, its final shape derives from the mold and forming
is usually not necessary. From the practical point of view, forming can be consid-
ered with respect to flat plates. Sometimes bending or roll-bending of these panels
is required. During bending, cracks initiate in the tensile surface and then they ra-
pidly propagate across the whole section. The bending that can be achieved without
damage is limited by the plasticity of the thin surface skin and the thickness of the
plate (as the thickness of the plate grows, the deformation of the tensile loaded part
of the sample for a given bending radius increases). Panel bending is therefore pos-
sible only for foams made of ductile alloys and/or at higher temperatures (see
Section 3.1).
Surface coating
For surface coating the same basic principles can be applied as for the bulk cell-
wall material, but ªsoakingº of the porous structure should be taken into account
as it significantly restricts those surface treatment procedures in which compo-
nents are immersed (acid pickling, anodizing, etc.). Aluminum foam can be ano-
dized like aluminum, but removal of the electrolyte from the structure is required
to avoid corrosion. Owing to the microporosity of the surface skin, painting with
3 Secondary Treatment of Cellular Metals 73
viscous paints is not problematic. Baking of the paint can result in the formation of
bubbles (blisters in the surface paint) caused by the expansion of gases in pores
and their attempts to escape through surface microcracks. For this reason it is re-
commended to leave some bare areas to enable the foam to ªbreatheº on tempera-
ture changes. If the foam has to be sealed completely, the coating process should
be performed in vacuum until the coating solidifies.
Joining
For practical applications of cellular metals, joining techniques are required. As the
structure of cellular metals is similar to that of wood, methods used for joining
wooden parts, especially gluing and bolting, can also be used with cellular metals.
Moreover, methods of joining metal parts, such as welding and soldering, are also
possible.
Joining cellular metal parts with bolts is simple, quick, flexible, and cost effec-
tive: the joint is also detachable. The strength of the joint depends primarily on
the density and pore size of the foam and increases if foam with a surface skin
is used. Higher strength is achieved with foams having finer porosity (finer poros-
ity provides a larger contact surface for the thread of the bolt). The strength of the
joint can be significantly improved if the drilled hole is filled with a suitable adhe-
sive.
Glued Joints
Metallic foams can be glued with all types of adhesive recommended for gluing the
cell-wall metal. These joints can be quickly and easily made and their strength
often exceeds the strength of the basic foam structure. Because of the simplicity
and flexibility, glued joints can be recommended for preparation of permanent
joints.
74 3 Secondary Treatment of Cellular Metals
However, glued joints also have disadvantages: low thermal stability, different
coefficients of thermal expansion for the adhesive and the bonded material, de-
creasing strength of the glued joint due to aging of the adhesive, and formation
of an insulating layer (thermal and electrical) are some of them. Moreover, if
foams without a surface skin are glued together, a greater consumption of adhesive
(weight increase, cost) has to be expected because of the significantly enlarged sur-
face area.
Welding
Metallic foams can be also welded, if the weldability of the cell-wall material is
good. Welded joints are very promising, but the techniques need further develop-
ment and the costs are still relatively high for practical application. Welding prob-
lems arise mainly from the very low and variable thickness of the surface layer.
This obstacle can be partially overcome using controlled heat input into the
joint, by laser-welding for example. Good joints can be obtained if the foamable
material is used as filler. Welding in combination with such a filler can provide
a method for repair of foamed parts.
Another possibility for joining cellular metals is the use of integral foams con-
taining the parts made from bulk material, which are built into the foamed struc-
ture and if possible diffusion bonded to the foam during foaming. These parts pro-
vide the contact counterparts of the foam for a joint. The bulk material can be
either locally placed inside the foam if the position of joint is known before foam-
ing (Fig. 3-1a), or it can create a large portion of the foam's surface (Fig. 3-1b). The
material of the built-in element should be compatible with the cell-wall alloy with
respect to corrosion resistance and thermal expansion. Locally distributed elements
need precise positioning and suitable fastening to avoid displacement due to foam
motion during foaming, which makes the foaming procedure more expensive. The
use of various coversheets that are diffusion bonded to the foam during foaming
can eliminate the higher costs and lower flexibility of the built-in elements, on
the other hand the higher overall weight must be taken into account.
a) b)
Figure 3-1. Joints made using bulk aluminum parts diffusion bonded to the AlSi12 foam as:
a) locally placed element, b) 3 mm thick Al coversheet.
3 Secondary Treatment of Cellular Metals 75
Further Reading
3.1
Forming, Machining, and Coating
Cellular metals are a new class of materials with low densities and novel physical,
mechanical, thermal, electrical, and acoustic properties, which offer potential for
lightweight structures, for energy absorption, and for thermal management [1].
In order for their outstanding material properties to be used, they have to be ma-
chined and processed to build structural components. The relevant techniques are
described in this section.
3.1.1
High-Temperature Forming
At present cellular metals produced from the melt, by deposition, and by solid-state
processes are available. They have very valuable properties in terms of density, stiff-
ness, and energy absorption, but their potential usefulness is limited by the diffi-
culty of forming or fabricating finished parts. A forming process is necessary for
the manufacture of 3D structural parts and a joining technique is required to fab-
ricate such parts into composite components.
Foams produced from melts, in contrast, are much less expensive but there are
problems with the forming properties of foamed cellular metals. At ambient tem-
perature, forming the foams gives rise to fractures or to the collapse of the pore
structure in the deformed zone. Local stress and strain concentrations can occur
causing failure at the weakest parts of the cellular structure [2].
Forming at elevated temperatures facilitates the shaping of metallic foams with-
out failure and with the pore structure being retained. At high temperatures the
yield stress and the deformation limits can be significantly reduced so that yielding
can occur. Local stress concentrations can be reduced by creep relaxation. The de-
tails of the process are described in the following section.
Figure 3.1-2. Left) working area in Al Si phase diagram [3]. Right) workpiece formed in a steel
mold [3].
3 Secondary Treatment of Cellular Metals 77
The workpiece in Fig. 3.1-2 was formed using a steel mold in a chamber furnace
in the temperature range of the solidus. The forming force in this case was gravity.
The mold was reproduced well, including the corrugations, but a disadvantageous
temperature management and locally higher stresses at the beginning caused a col-
lapse in contact areas at the edges.
Figure 3.1-3. TiH2 (left) and gas injection foamed (right) foams: influence of temperature and
foaming agent on cell structure.
78 3.1 Forming, Machining, and Coating
weight of the bending wing. As can be seen the material can be formed without
collapse of the cell structure, and relatively large bending angles are possible.
Owing to the properties mentioned above, the temperature range and with it the
working area and the process window can be expanded. Thus, this material is very
well suited to high-temperature forming process in contrast to the TiH2 foamed
materials, which only show a small process window.
3.1.2
Machining
Conventional cutting and machining techniques like sawing, milling, or drilling all
are suitable for processing metal foams. But using conventional machine tools
causes surface distortion or damage to low-density metal foams [5]. Nevertheless,
the conventional methods are satisfactory for most applications in which metal
foams have to be machined [6, 7].
For accurate cutting electrodischarge machining (EDM), chemical milling, water-
jet cutting, high speed cutting, or diamond sawing are recommended [5]. This is
especially necessary for machining metallic foam specimens for testing, to avoid
cell damage that would influence the properties of the samples.
Owing to the cellular structure of metal foams there are different methods of
machining the material. The appropriate technique should be selected according
to the required quality of the cutting surface. For cutting of metallic foams rein-
forced with hard particles like SiC, a diamond cutting tool might be required.
3.1.3
Coating
Owing to their properties, metal foams have great potential in lightweight con-
struction. At present there are few applications for metallic foams because
foams with open surfaces show relatively low mechanical strength and are difficult
to join to other parts. So metallic foams are not yet well suited to structural appli-
cations [8]. To further enhance the stiffness and strength of the foams they are
often used as composites with roll-plated or adhesive bonded cover sheets. The
cut surface of metal foam displays open cells and a rough texture, and is vulnerable
to local damage [5]. It can be filled with an epoxy or other resin, or, in the case of
syntactic structures that have a natural skin, the surface can be coated by conven-
tional methods. These manufacturing methods are mostly limited to plane sur-
faces and do not allow complex shaped parts. By using thermal spraying tech-
niques almost any desired form of foam-spray-deposit composites can be produced.
Furthermore, there are considerable advantages concerning recycling and usage at
elevated temperatures compared with adhesive-bonded composites [8].
80 3.1 Forming, Machining, and Coating
treated to roughen and to activate it. This is generally done by corundum blasting
[8].
Currently available foamed semi-finished products made via the powder-metal-
lurgical route have a thin smooth surface skin. This surface is often damaged dur-
ing the blasting process, which opens the underlying pores. Thus, a surface treat-
ment that is more suitable for foams is required because the spraying process is
not able to fill the open cells and form a new plane surface. A suitable approach
may be microblasting, in which the semi-finished foam products are blasted
with low pressure and with particles less than 100 mm in diameter. Although the
process parameters have not been fully investigated yet, the first results of the mi-
croblasting technique have been very satisfying.
Figure 3.1-6. High (left) and low (right) oxide NiAl5 layers [8].
82 3.1 Forming, Machining, and Coating
ses and leads to a relatively small reduction of the hardness of the spray deposit.
One disadvantage is the high density of NiAl5 compared with aluminum alloys.
The use of highly tensile solidifiable aluminum alloys as spray materials is the sub-
ject of investigations at the present time [8].
Another technique more complicated than wire arc spraying but producing
much better spray deposits is plasma spraying (Fig. 3.1-7). The spraying material
is injected into an inert gas plasma as a powder and then heated and accelerated in
the direction of the substrate. The material is an AlSi30 powder, which in conjunc-
tion with suitable process parameters leads to relatively hard deposits when
sprayed. A practical thermal treatment of these spray deposits and their further me-
chanical properties are currently being investigated.
Thermal spraying is a suitable and flexible method of producing composites with
metallic foam cores. But investigations are still necessary with regard to the en-
hancement of the mechanical properties of the deposits as well as the pretreatment
of the metallic foams.
Figure 3.1-7. Left) spray deposit on aluminum foam [8]. Right) plasma sprayed deposit [8].
References
3.2
Joining Technologies for Structures Including Cellular Aluminum
3.2.1
Introduction
3.2.2
Feasible Joining Technologies
This section summarizes conventional joining techniques that are feasible for com-
pound structures utilizing aluminum foams.
which have the advantage that no additional hole has to be drilled into the sheet
material.
The performance of the mechanical joints can be improved if the fastening ele-
ments are additionally smeared with epoxy adhesives before driving them into the
foam [4,5].
A further possibility for joining foams with mechanical fasteners is the applica-
tion of integral foams [3], where the foam contains parts made from bulk material.
These parts, screw sockets for example, are integrated into the foamed structure
and bonded to it during the foaming process. However, the locally distributed ele-
ments need precise positioning and suitable fastening in order to avoid a displace-
ment caused by the foam motion during foaming. Consequently, not only is the
weight of the cellular structure increased by integrating these parts, but the cost
of production as well.
3.2.2.2 Gluing
Gluing has been described as a feasible joining technology for aluminum foams
[6]. Different gluing systems have to be distinguished. One-component systems
are, for instance, epoxy-based and cure for approximately 30 min at 180 hC. The
process conditions of these adhesives make them very interesting for car manufac-
turers, as the glue can cure during cathodic electrocoating and no further working
steps are necessary. Epoxy-based systems are available as conventional glues or as
expanding glues. The latter ones are very convenient for bridging high tolerances
and for integrating foam bodies into hollow extruded profiles. In case a thermal
curing process is undesired, two-component systems are available that cure at
room temperature. By combining riveting and gluing it is easily possible to fix
the foam body in the structure while the curing of the glue takes place. Addition-
ally, the properties of the composite are improved in certain loading cases.
3.2.2.3 Welding
Welding processes that have been shown suitable for joining aluminum foams are
laser-beam welding [6] and ultrasonic (US) welding [7].
With the latter process, various geometries of the joining area are possible. They
reach from quadratic joints with a maximum area of 100 mm2 to roller seam welds
with a width of 1 5 mm and a maximum length of 500 mm. Finally circular bond-
ing areas with typical diameters of 18 mm are producible. A characteristic of the
US welding process is that it is a joining process in the solid state based on friction
and consequently not a real welding process, where both joining partners are trans-
ferred into the liquid state for joining [7].
Laser-beam welding has been shown as a viable joining technology for alumi-
num foams [6]. Owing to its characteristics such as the keyhole effect and the
minimal heat input, deeper parts of the structure can be integrated into the
joint as well. Fig. 3.2-1 shows the principle mechanism of surface welding and
the key-hole welding effect for bulk and foam material.
3 Secondary Treatment of Cellular Metals 85
Figure 3.2-1. Surface welding and keyhole effect during laser-beam welding.
Soldering
Various Sn- and Zn-based solders can be used for fabricating structures using Al
foams [3]. Reliable joints are said to be possible without the use of fluxes, just
by scratching the oxide layer.
Another suitable technology is the S-Bond process, developed by Euromat [8].
This solder has a chemical composition of SnAg4Ti4 (wt.-%) and a melting
range of 220 229 hC. The S-Bond process is a flux-free soldering process that is
characterized by the mechanical ripping of the oxide layer that has formed around
the molten soldering material. The oxides of the substrate layer are partially ripped
during the soldering process, so that a metallurgical interaction of the soldering
material and the substrate can take place. Additionally, the soldering material
creeps below the oxide layer of the substrate and wets its surface totally. S-Bond
is available as wire, rod or foil, which makes this process an interesting alternative
for extensive joints.
Though flux-free soldering technology can be transferred to aluminum foams,
corrosion preventing measures have to be taken as the Sn- and Zn-based alloys
can cause corrosion problems in combination with aluminum.
86 3.2 Joining Technologies for Structures Including Cellular Aluminum
Brazing
Brazing (Ti450 hC) is a feasible joining process for aluminum foams [9]. Though
noncorrosive fluxes are available for brazing, it is advisable to coat the structure to
protect it from humidity. The brazes are preferably Al-based and the furnace atmo-
sphere should be inert.
An alternative is the use of a porous filler material [9], which expands during the
brazing process (Fig. 3.2-2). Thus a sudden change in porosity from foam to bulk
material is prevented. The working temperature of the filler material has to be low
enough that the porous base material does not deform during the expansion of the
filler material.
As outlined, there is a variety of suitable joining technologies available for fabri-
cating structures utilizing aluminum foams. However, for a practical investigation
the number of technologies has to be limited to one or two processes per category.
Table 3.2-1 gives an overview and some characteristic properties of relevant joining
technologies.
Figure 3.2-2. Joining of Al foam and solid Al by brazing with a filler that expands to a foam [9].
3.2.3
Foam Foam Joints
Table 3.2-2. Feasibility of the joining technologies for various foam-foam joining geometries
Blind riveting ± ± I
Screwing ± I I
Gluing I I I
Riveting/gluing ± ± I
Soldering I I I
3.2.4
Foam Sheet Joints
The following text concentrates on processes for producing foam sheet compo-
sites [2,11]. Though there are other processes, such as the roll-bonding process
[1], further technologies have to be investigated, for integrating foam bodies into
hollow extruded profiles or for the partial stiffening of structural parts. In most
88 3.2 Joining Technologies for Structures Including Cellular Aluminum
cases, foam sheet composites offer only one-sided accessibility. This has to be
taken into consideration when selecting a feasible joining technology. Under
these circumstances, all joining technologies listed in Table 3.2-1 are feasible.
b) Riveting
c) Gluing
d) Soldering
e) Laser-beam welding
Figure 3.2-3. Micrographs of the investigated foam sheet joints.
90 3.2 Joining Technologies for Structures Including Cellular Aluminum
or the substrate. The bonding of the glue and the surface might either be caused by
ionic bonding or polar interactions [12]. Gluing is a feasible joining technology for
foamed metals, independent of the presence of a casting skin. Though the area in
direct contact with the sheet material is drastically reduced in the absence of a cast-
ing skin, the viscosity of the glue is low enough to fill the pores. The loss of surface
can be compensated for by an additional form-fitting joining mechanism.
Figure 3.2-4. Geometry of the samples for tensile-shear and pull-out testing.
Pull-Out Tests
Pull-out tests were carried out with structures utilizing mechanical fastening ele-
ments. After preliminary investigations with a selection of conventional fastening
elements, all fasteners were excluded that form the joint according to a frictional-
fitting joining mechanism. Though good results with frictional-fitting fastening
elements could be measured with large volume samples [5], the thickness of
foams for technical applications such as the partial stiffening of structures should
be limited to less than 10 mm. In this case however, there is not enough material
left for the frictional interaction between cell edges and the fastening element, so
that it will be torn out at low loads. Figure 3.2-5 shows some characteristic
load deflection curves of joints utilizing mechanical fasteners with an additional
form-fitting joining mechanism.
Riveted samples show a larger range of deformation as the bulge has to be pulled
through a hole of smaller diameter. This means that a great amount of deforma-
tion work has to be done. In contrast to this, the diameter of the investigated
screws reduces to the tip of the screw. As a consequence, not much deformation
work can be done when pulling out the screw, after the maximum load has
been reached.
Obviously, the pull-out load is dependent on the density of the foam material.
The effect of relative density on the maximum pull-out load, Fr, is nonlinear
with a power of about 1.5 [5]
Fr Ar3=2 (1)
Figure 3.2-5. Qualitative load deflection curves for screwed and riveted samples in tensile-shear
loading.
Tensile-Shear Tests
Tensile-shear tests were carried out for the joining technologies listed in Table 3.2-1.
The maximum shear loads for the joining technologies are given in Table 3.2-4.
There are two different failure mechanisms that apply to point-contact joining
technologies or area/line-contact joining technologies. Laser-beam welded and
3 Secondary Treatment of Cellular Metals 93
Table 3.2-4. Influence of the joining technology on the maximum load in tensile-shear loading
glued samples tend to fail in the foam close to the joining zone. Consequently, the
mechanical properties of the cellular structure influence the behavior under tensile
shear loading. This is observable at a maximum load that represents the mechan-
ical properties of the foam. After the maximum load is passed, the foam tears al-
most brittle. Laser-beam welded samples show a lower maximum load than glued
samples as their joining area is smaller. Consequently the failure load is highly
dependent on the number of welding seams.
In contrast, riveted and screwed samples show ductile failure behavior under
tensile-shear loading. The fastening element compresses parts of the cellular struc-
ture during the deformation. When the remaining cross section of the foam body
is too small, the foam breaks, which is shown by an abrupt decrease of the load.
Influence of the Foam Density on the Joint Strength for Line-Contact Joints
Figure 3.2-6 shows that the strength of welded samples is highly dependent on the
relative density of the foam [2], i. e. the number of the cell walls in the area of the
joint. In these experiments the casting skin was removed before joining so that the
resulting joining mechanism is based on a fusion of the sheet with the cell struc-
tures as well as by form-fitting due to a flow of the melt into the porous structure.
The failure of the joint is observed to start close to the welding seam.
The failure behavior can be described by a correlation between the foam density
and the strength of the joint.
s Metalfoam
max rmin S ar (3)
There are two ways of assessing the dependence of the strength of laser-beam
welded composites on the foam density. For example, the strength of massive
sheet sheet joints of about 210 N/mm2 (Fig. 3.2-6) would result in a relationship
according to a power law (Eq. 2). In this study however, the joining partner is not a
massive material but a powder-metallurgic produced aluminum. As a conse-
quence, the values of the massive material cannot be reached with a density of
94 3.2 Joining Technologies for Structures Including Cellular Aluminum
a)
b)
Figure 3.2-6. Influence of the density of the Al foam on the strength of laser-beam welded joints.
2.7 g/cm3. The values of composites utilizing a metal foam with a casting skin in-
dicate a linear relationship between the foam density and the strength of the joint.
Here, the casting skin must be assessed as a massive joining partner and the
resulting density of the casting skin is 1.8 g/cm3 (linear regression of Eq. 3).
The solution of the power law equation gives a lower value of 1.1 g/cm3.
Bend Tests
Figure 3.2-7. Influence of the joining technology on the energy absorbing behavior of a sandwich
structure in four-point bending after a deflection of 10 mm (see Table 3.2-3).
for integrating the rivets. Welded and riveted/glued samples show an average stiff-
ness, like soldered specimens.
As the integral of the load deflection curve (equal to the energy absorbed dur-
ing the deformation) increases with rising stiffness, the stiffness of the composi-
tes is mirrored in the values of energy absorption up to a deflection of 10 mm
(Fig. 3.2-7).
Related to their mass, structures utilizing a steel sheet (applied joining technol-
ogy: gluing) show the worst properties owing to the high weight of the sheet
material. Composites utilizing a tissue on the tensile side of the structure are
the lightest, but they have a low stiffness owing to the tissue. In consequence,
the amount of energy that can be absorbed is comparatively low.
Four-point bend tests with a constant bending moment along the sample serve
for evaluating the stiffness of a sample. In the case of a three-point bend test
the maximum bending moment is in the area of the bending stamp. Thus,
three-point bend tests serve for evaluating the strength of the structure.
Figure 3.2-8. Influence of the joining technology on the energy absorbing behavior of a sandwich
structure in three-point bending after a deflection of 10 mm (see Table 3.2-3).
meet this requirement. Though the strength of the tissues is high enough, it fails
after a comparatively low deformation owing to catastrophic fracture.
Displaying the amount of absorbed energy in Fig. 3.2-7 to 3.2-10 shows that all key
figures located on the same straight line through the origin of the diagram have
the same mass specific value. With an increasing slope of the line, the mass-related
properties increase as well. The goal of optimization is therefore to increase the
slope of the line on which the values are located.
This cannot only be achieved by a variation of the joining technology, but also by
the density of the foam core.
Influence of the Density of the Foam Core on the Properties in Three-Point Bending
The results of three-point bend tests with structures utilizing a foam body without
casting skin and an average density from 0.3 to 1.1 g/cm3 show that there is an almost
linear dependence of the maximum load on the density of the foam core in the inves-
tigated density range.
Figure 3.2-9. Influence of the weight of the sandwich structure on the value of the absorbed
energy up to a deflection of 10 mm (see Table 3.2-3)
3 Secondary Treatment of Cellular Metals 97
Comparing the performance of sandwich panels where the casting skin was still
on the core material to those without casting skin reveals that the sandwich panels
with core material without casting skin have a higher specific energy absorption
(Fig. 3.2-9).
As the foam is compressed in the area of the bending stamps during three-point
bending, the loading mode in this area is consequently similar to a compression
test. However, the casting skin is orientated perpendicular to the load and cannot
enhance the energy absorbing properties. Thus the casting skin just decreases the
mass-related energy absorbing properties in bending.
Figure 3.2-10. Influence of the joining technology on the energy absorbing behavior of a sand-
wich structure in dynamic three-point bending (see Table 3.2-3).
98 3.2 Joining Technologies for Structures Including Cellular Aluminum
3.2.5
Transferability to Structural Parts
The basic fields of application of aluminum foams are energy absorbing structures
and stiffening of structural elements. The following sections show the transferabil-
ity of the described joining technologies to structural elements.
3.2.5.1 Tubes
The integration of foam bodies in tubes for energy absorbing structures is well de-
scribed elsewhere [13,14]. The results show that gluing as well as laser-beam weld-
ing are feasible joining technologies for this application. Figure 3.2-11 shows an
X-ray of a crash-absorbing element partially filled with Al foam and a detail of
the joining area.
Regarding the triggering of the sequential buckling process, laser-beam welding
shows some synergetic effects in axial loading, as shown previously [13,15]. The
initial peak load of the characteristic load deflection curve in axial loading can
be reduced, which leads to an increase in the effectiveness of the composite struc-
ture in axial loading.
Figure 3.2-11. Foam body integrated in an Al tube by laser-beam welding for an application as a
crash element in the front side member of a vehicle [13].
3 Secondary Treatment of Cellular Metals 99
3.2.5.2 Hat-Profiles
Figure 3.2-12. Integration of a foam body into a hat profile for improving the torsion, bending,
and energy absorbing behavior of the structure.
Figure 3.2-13. Influence of the joining strategy (one- or three-sided) and the joining technology
on the deformation behavior of a hat-profile in dynamic three-point bending.
100 3.2 Joining Technologies for Structures Including Cellular Aluminum
Owing to the optimum contact area of glued samples, the sheet material stays
in contact with the foam during the whole deformation of the composite struc-
ture. In the welded sample, however, the sheet can deform independently of the
foam in large areas, which reduces the energy-absorbing potential of the struc-
ture in a defined interval of deformation. Independently of the joining technol-
ogy, three-sided joining improves the energy-absorbing properties of the struc-
ture.
Torsion Tests
The torsion tests shown in Fig. 3.2-14 show that glued samples have better proper-
ties than welded ones. Fewer folds form during the deformation process, which
means that more deformation energy can be absorbed by the porous structure.
In laser-beam welded samples, the welding seams that fix the foam core are the
weakest part of the structure and tear during the loading process.
Corresponding to the bend tests, the experiments with foam-filled hat profiles
show that with increasing area of the joint, the properties of the composites in-
crease.
Figure 3.2-14. Influence of the joining strategy (one- or three-sided) and the joining technology
on the deformation behavior of a hat-profile in torsion.
3.2.6
Summary
In a final overview, Table 3.2-5 and Table 3.2-6 show the properties of the investi-
gated joining technologies for different cases of loading, including a technological
assessment of the joining technologies.
3 Secondary Treatment of Cellular Metals 101
Table 3.2-5. Summary of the properties of the investigated joining technologies in the different
loading cases (++ very good; + good; o overage; ± bad;/ not carried out)
Riveting + ++ ± ± ±
Screwing + + ± ± ±
Gluing + / ++ + +
Riveting/ ++ / o ++ ++
Gluing
Laser beam o / + + o
Welding
Table 3.2-6. Final evaluation of the investigated joining technologies (++ very good; + good;
o overage; ± bad)
Riveting ++ ++ ++ + o +
Screwing ++ ++ ++ + + +
Gluing o o o o ++ ++
Riveting/ ± o o ++ o +
gluing
Soldering o ± + + ++ +
Laser-beam + ± ++ o o ++
welding
102 3.2 Joining Technologies for Structures Including Cellular Aluminum
References
3.3
Encasing by Casting
Foam parts prepared by expansion in a mold show a closed surface skin with a
thickness comparable to the cell-wall thickness of about 200 mm. Even in cases
where the surface near cells are compressed yielding a few layers of cell walls
along the outer skin, it is not suitable for carrying high loads, as required for struc-
tural applications. Casting aluminum around a foam can create components where
a low density foam core is completely surrounded by a massive outer shell. The
shell can be designed in such a way that additional functions besides its load carry-
ing ability can be fulfilled. As only one processing step is required to produce such
a foam core component, production is expected to be very economical.
Potential applications for this kind of foam core castings are space frame nodes,
knuckles, control arms, cross-members, and stiffness providing structural compo-
nents. The described method also offers the potential to produce quasi-hollow cast-
ings where conventional salt or sand cores are replaced by permanent foam cores
in complex geometries.
The various casting methods for the massive outer shell differ by the acting pres-
sures, the die-filling time, and the solidification time, which is mainly determined
by the temperature of the die. Generally, casting of the shell is not completely
straightforward due to the limited compressive strength of the foam cores and
the very thin surface skin, which is not without flaws [1]. A further difficulty results
from the fact that methods have to be developed for fixing the cores.
The following text is based on three publications: squeeze-casting, Kretz and
Kaufmann [2]: high-pressure die-casting, Heinrich, KoÈrner, and Singer [3,4]; and
gravity die-casting, Simancik and Schoerghuber [5].
3.3.1
Foam Cores for Encasing by Casting
tion-molding technique is thicker than that of the furnace baking method. This is
an advantage for encasing but it also increases the weight of the core. Figure 3.3-1
shows several core geometries used for different encasing techniques.
Figure 3.3-3. Global compressive strength s CG versus density of several Al foam samples from
plates produced via the furnace-baking method [4].
106 3.3 Encasing by Casting
Figure 3.3-4. Average values of the local compressive strength versus density of several foam
samples produced via the furnace-baking method [4].
There is a strong scatter of strength values due to the inhomogeneous cell struc-
ture of the foams.
The results in Fig. 3.3-4 demonstrate that the strength of the foam surface skin
has to be improved for squeeze and die casting, which are characterized by a very
high pressure. For gravity die casting the global strength as well as the local
strength are not so important.
a) b) c)
Figure 3.3-5. Optical micrographs showing breakage of the surface at high pressure levels;
foam inserts after the local compressive test: c) Al plasma sprayed coating, d 200 mm, local
a) no coating, surface damage occurs at low damage and overall collapse by compression
pressure level preferentially at locations with deformation, d 400 mm, compression [4].
large cells; b) ceramic coating with d 300 mm,
3.3.2
Shell Casting Processes
cores are not infiltrated and not deformed. However, the processing window for
achieving good samples without core infiltration or core deformation is a very nar-
row one.
Figure 3.3-10. Two different demonstration parts with foam core inserts encased by die casting:
a) trapezoid, b) plate [4].
As core materials the casting alloy AlSi10 and the wrought alloy AlMg1 were
used. The upper bound of the pressure inside the die is limited by the global com-
pression strength of the foam core. It is about 60 100 bar and can only be raised
by increasing the foam density.
a)
Casting experiments with coated and uncoated foam cores were performed for a
plate-like geometry with different specific casting pressures pNS. The specific cast-
ing pressure is the operating dwell pressure in the casting chamber. It has to be
taken into account that the actual acting pressure inside the die is much lower
because of friction. Figure 3.3-11 shows the infiltrated pore volume VI and the
porosity in the skin versus pNS.
As expected infiltration increases with increasing pNS whereas porosity in the
skin decreases. By using coated foam cores a significant reduction of core infiltra-
tion is possible. In accordance with the tests to determine the local compression
strength, the plasma-sprayed plates show the best results. More recent results
achieved with optimized die design and process conditions (evacuation of the
die) demonstrate that the porosity of the skin can be reduced even to less than
5 % without infiltration of the core. The minimum skin thickness is about 2 mm
for die geometries evaluated so far.
Figure 3.3-12. Pressure-less casting: Foam core, encased core, and cross section with foam core
(height of the foamed part 25 cm, maximum width of the cross section 12 cm) [5].
112 3.3 Encasing by Casting
Figure 3.3-13. Aluminum casting with different foam cores: left, AlSi12 foam; right, AlMg1Si0.6
foam (diameter of the foam filled tubes 6 cm) [5].
Another advantage of preheating the cores is that cold shuts, which appear in
regions of the castings with the lowest thickness of the shell, can be avoided.
The preheating of the cores leads to a lower heat flow from the melt into the
core, thus enabling reduction of the wall thickness in comparison to usual sand
cores, which can not be preheated. With preheating temperatures of 400 hC the
minimum shell thickness is found to be about 1.5 mm.
References
3.4
Sandwich Panels
For structural applications metal foams are often used in combination with con-
ventional dense metal structures such as sheets, columns, or more complex-shaped
hollow metal structures. This allows for optimized mechanical properties in a
given loading situation [1]. It also facilitates ªhidingº the metal foam inside a closed
and dense structure, which again is advantageous for corrosion protection. Such
composites containing aluminum foam may be manufactured in various ways.
The most obvious and straightforward one is achieved by adhesive bonding of
pre-fabricated aluminum foams and flat face sheets. However, this approach has
certain disadvantages and is not feasible in all cases. An alternative and preferable
method consists in establishing composites during the foaming process. Foam-
filled columns can be produced by inserting foamable precursor material into a col-
umn and then heating. The foam will eventually rise and fill the column. Another
possibility is to start with a foamed part and to coat it with aluminum by thermal
spraying [2] thus establishing a dense outer skin. Yet another way is to use the
foam as a core for die-casting [3,4]. Aluminum foam sandwich (AFS) panels
may be manufactured very elegantly by roll-cladding face sheets to a sheet of foam-
able precursor material, then creating the desired shape in an optional working
step, and by finally foaming the entire composite [5,6] (Fig. 3.4-1). Foaming will
create a highly porous core structure without melting the face sheets if the melting
points of the foam and the face sheets are different and process parameters are
chosen appropriately.
114 3.4 Sandwich Panels
Figure 3.4 -1. Process steps for making sandwich panels with aluminum foam cores.
3.4.1
Sandwich Foaming Process
shows no structure in the low magnification chosen. The interface of foam and
face sheet lies on a straight line and is well defined.
Finally, the foamed sample on the right hand side of Fig. 3.4-3 represents an
even later stage of expansion. It exhibits a notably coarser grain size distribution
in the foam and a slightly diffuse boundary between foam and face sheet. The
eutectic phase has grown into the former face sheet material by diffusion pro-
cesses and has locally amalgamated with the face sheet alloy: one reason for
the excellent bonding between foam cores and face sheets in properly manu-
factured aluminum foam sandwich parts and the explanation of the absence
Figure 3.4 -3. Metallographic images of sand- shortly before maximum expansion; right)
wich structures with an AlSi7 core. Three dif- sandwich at the onset of face sheet melting.
ferent expansion stages are shown: left) un- The width of each image is 0.75 mm.
foamed precursor material; middle) sandwich
116 3.4 Sandwich Panels
3.4.2
Industrial Application
Figure 3.4 -4. ªDetroit Show Car 1998º developed by Karmann (Osnabru Èck) for demonstrating
potential uses of aluminum foam sandwich panels. The rear bulkhead and the front firewall (not
directly visible) are made of AFS.
3 Secondary Treatment of Cellular Metals 117
Figure 3.4-5. ªEUROC 99º concept racing car: left) CAD model of entire car; right) space-frame
structure with AFS parts.
ing parts. The application of AFS parts may also lead to an elimination of noise
attenuation materials because of the low structure-borne sound characteristics of
AFS.
Another area of interest to be investigated in the future is that of exterior panel
closures, e. g., doors, hoods, and decklids. This application depends on the
achievement of a Class-A surface with stamped AFS panels. Taking a hood
stamped out of AFS as an example, there is no longer any need for an inner
panel due to the inherent stiffness of the AFS outer layers, thus reducing ex-
penses for material, tools, and the complete assembly. In spite of higher material
expenses, an AFS bonnet may (depending on the shape) be more cost-effective
than a steel hood up to a production volume of 100 000 units. This is attributed
to the reduction of manufacturing and tooling expenses. This means that light
weight AFS constructions may be applied economically in low and middle pro-
duction volumes due to the reduced investment costs compared to conventional
steel components.
In general applications, possible additional costs of AFS panels should be com-
pared to the advantages that can be expected. Even if only one of the improved
characteristics of AFS is required for a particular application, the associated in-
crease of expenditure may be of secondary interest. In general, the application of
AFS components will increase overall material costs, so that an enhanced perfor-
mance must be achieved as justification. In the future, this situation may change
as vehicle operating costs gain more significance due to the higher energy costs. As
a consequence, this implies that the amount of money available for the use of light
materials, which is right now about 2.5 5 Euro per kg weight reduction, will
increase.
Acoustic Properties
Considering the customers comfort requirements, a new lightweight body mate-
rial is not allowed to show inferior acoustic properties than steel. Therefore, the
good acoustic properties of AFS should be emphasized. Especially cars with alu-
minum bodies show poor damping properties. To improve this situation, a large
amount of damping material must be added, thus sacrificing the mass saving po-
tential.
A joint research by the TU Dresden and Karmann has revealed that AFS offers a
significantly better acoustic behavior especially in the range 50 400 Hz. Additional
insulation measures may be reduced: weight is saved. The acoustic performance of
special insulation materials will of course not be attained, so that AFS cannot ex-
clusively serve as a sound damping material.
Thermal Properties
Thermal conductivity is another important aspect when selecting body materials.
Due to the entrained air bubbles, the heat transport capabilities of foams are
low. Depending on the density, the thermal conductivity of AFS is reduced to
1/12 1/20 of the conductivity of bulk aluminum. Furthermore, AFS satisfies
most of the fire protection regulations. No adhesives are contained and the AFS
components maintain their shape up to the melting point of 600 hC and in
some cases even above. The exceptional welding characteristics together with mini-
mal distortion justify to characterize AFS as ªthermally very stableº.
Robustness
Sandwich components, as known from aerospace technology, are relatively vulner-
able in impact situations. Even small damages of honeycomb panels may lead to a
complete breakdown of the core panel structure. As a result of the metal link be-
tween core and panels, this does not occur in an AFS panel. Cracks may only occur
in the core and their expansion is limited. A delamination of core and panels has
not been detected with parts manufactured and tested to date. This is very impor-
tant since structural body components are not subject to special checks during the
product life cycle
Other Properties
Additional properties of AFS that enhance product performance are: good energy
absorption, recyclability, and low manufacturing time periods for the sandwich compo-
nents. The foaming process, e. g., takes only 30 45 s even for large parts. There-
fore, mass production with a comparably low number of parallel production lines
is possible.
3.4.3
Joining Technology of AFS
Process Details
Riveting nuts/screws M4 M8
Flow drilling M4 M8
a)
b)
c)
has no impact on the quality of the bonding as the joining zone is obviously lim-
ited to the face sheet.
a)
b)
c)
Figure 3.4-8. Different joining techniques for AFS: a) punch riveting, b) riveting nuts and screws,
c) flow drilling.
3 Secondary Treatment of Cellular Metals 123
pressed, its tensile strength drops to a value of 50 % plastic deformation, but re-
gains the original mechanical values of the nondistorted sandwich at maximum
compression. This property is a result of the increasing mechanical clutching of
the collapsed cell structures and the likewise increasing friction. The mechanical
values of compression strength and shear stress do decrease in this range but
they stabilize again at a low level under static as well as under dynamic load.
3.4.3.7 Riveting
AFS parts may also be riveted. Especially in mixed constructions, riveting is very
well suitable. It is important to choose a relatively large diameter of the rivet
head since too small a diameter leads to a fastening pressure of the rivet that com-
presses the core. Due to the high surface pressure one will find plastic deforma-
tions on the face sheets.
3.4.3.8 Bonding
AFS with face sheets of aluminum may be bonded with the same technology as
conventional aluminum sheets (Fig. 3.4-9). The same parameters have to be re-
spected as there are the creation of a defined surface, a construction adapt for
bonding and especially the choice of an adequate bonding system. The strength
of today's bonding systems partly exceeds the physical values of the aluminum
foam cores. One of the main advantages of bonding lies in the optimum transmis-
sion of the applied forces. This technology allows the AFS parts to be excellently
integrated into the surrounding structure.
124 3.4 Sandwich Panels
3.4.4
Cutting of AFS
Cutting of AFS structures, the unfoamed precursor material as well as the foamed
sandwich panels, can be quite a challenge, especially in the latter case. Here, con-
ventional mechanical cutting techniques cannot be applied in a straight-forward
way because of the danger of uncontrollable deformation of the material. There-
fore, two alternative methods have been evaluated.
References
4
Characterization of Cellular Metals
B. Kriszt
We are familiar with natural cellular structures such as wood and bone: these nat-
ural materials are structured hierarchically.
In engineering, man-made cellular materials (porous construction materials and
cellular polymers) have been used for a long time because of their remarkable ther-
mal insulation. A new group of cellular materials has been invented more recently:
the cellular metals. When we study different man-made cellular materials we soon
recognize that they also show a hierarchical architecture. As in nature, every pro-
duction process leads to a different cellular structure, so clearly cellular structures
are determined by the process by which they are made. However, when we want to
describe a cellular metal we say Alporas, Alulight, or Inco foam, and do not de-
scribe exact structural quantities.
So before we can characterize cellular metals, we need to understand the signif-
icant parameters necessary to describe the architecture and the hierarchy.
First, we can think of a cellular metal as a heterogeneous composite material of a
metal and a gas. If only the metal matrix is observed, the microstructure consisting
of grains, precipitates, dendrites, or different phases can be seen, but there is no
information on the architecture of the cellular metal. This hierarchical level is
often called the microstructure of the metal.
The density of the cellular metal depends on the volume fraction of metal and
gas. Consequently, cellular metals are characterized by their density compared to
the parent metal: the relative density. Typical metal foams have relative densities
of less than 0.3. Typical porous metals have relative densities greater than 0.3
but less than 1.The density of cellular metals, in absolute units or as relative den-
sity, is one of their main characteristics. From the density of a cellular metal part,
we have an instrument for describing the material at the ªmacroº level.
The second main characteristic is the cellular architecture. Often fluctuations in
cellular architecture lead to inhomogeneity in density, so the local density distribu-
tion can describe a cellular metal more precisely. This way of describing cellular
structures belongs to the ªmesoº level.
128 4 Characterization of Cellular Metals
However, the definition of macro, meso, and micro is not yet standardized: in
Section 4.1 a scheme is discussed that helps to clarify the classification of the hier-
archical structure.
Real cellular metals do not have perfect ideal cells, but the analysis of ideal cells
is nevertheless helpful for describing the architecture of cellular metals. The ideal
cells model describes the whole structure in terms of at least one space-filling
three-dimensional unit cell; in most cases we can think of polygonal cells. For ex-
ample, honeycombs are networks of channels having hexagonal cross section; me-
tallic foams, sponges, or lattice block material are representatives of real 3D poly-
gonal cellular metals.
Cellular structures in which adjacent cells are separated by cell walls or faces are
called closed-cell foams or metals: the other group, open-cell foams or metals, are
permeable. Sometimes this classification is not clear, because in some cellular me-
tals both closed cells and open cells can be observed.
To get a feeling for a 3-D structure, we conceptualize a space-filling unit cell of
an ideal structure (Fig. 4-1). From a geometric point of view it is helpful to think of
a cellular structure formed by a skeleton of cell edges (struts), which merge at
nodes (vertices).
One space-filling unit cell is the Kelvin tetradaidecahedron cell. This cell has a
low surface area per unit volume. Most foams made by expanding a gas in a liquid
tend to form such a cell shape. For this reason, this cell type is often chosen for
modeling cellular metals [1 6]. Knowing the geometry of the unit cell allows us
to calculate the relative density of the cellular metal. From the example of the Kel-
vin cell, we can elaborate the main structural features used for characterization.
Certain parameters are needed to describe the cell shape, the cell size, the topol-
ogy, the geometry of cell struts, nodes, and walls, and their relation to neighboring
cells. Often it is helpful to define also parameters that describe the imperfections
or the defects of a cellular material compared with the ideal unit cell (see
Section 4.1).
A long list of parameter can be defined, but we have to ask critically if these serve
a useful purpose. If we want to understand relations between processing para-
meters and cell structure (Chapter 2, and Section 4.1), as well as between cell struc-
ture and the property profile (Section 4.1 and Chapter 5), we need to investigate
and quantify the cells and their imperfections as precisely as possible.
Another task that requires structural parameters is modeling the mechanical be-
havior of cellular metals (Section 4.1 and Chapter 6). Input data derived from char-
acteristic cell features of real foams are needed for generating virtual cellular net-
works, such as the Voronoi network, for examples.
References
4.1
Characterization of Cellular and Foamed Metals
4.1.1
Definition of Structural Features of a Cellular Metal and Influence on Property Profile
As can be seen in Fig. 4.1-1, we can define a list of more than 30 structural para-
meters for describing the geometrical structure (architecture) and the microstruc-
ture of cellular materials.
As already discussed in the introduction of Chapter 4, cellular metals are hetero-
geneous materials, like composites, because the pores are embedded in the metal
cell skeleton. The shape and arrangement of the cells is called the geometrical
structure or architecture. In some cellular metals the geometrical architecture
Figure 4.1-1.List of parameters for describing structure of metallic foam; depending on the re-
solution, geometrical structure and microstructure can be defined.
4 Characterization of Cellular Metals 131
can be observed visually, so the length scale of cell size can range from some tenths
of millimeters to some centimeters. For acquiring images of the outside appear-
ance of the geometrical structure either digital video cameras or ordinary scanners
can be used. Typical magnifications that can be reached with these systems are up
to tenfold.
If we zoom in on the cell material, we will see another hierarchical level or sub-
structure, the microstructure of the cellular skeleton. We can observe the structure
of metallic cell material, micropores, and cracks at a resolution of microns by using
light microscopy or scanning electron microscopy. If transmission electron micro-
scopy is used, the lateral resolution is increased to nanometers and we may observe
dislocation structure or small precipitates.
From this, it becomes clear that cellular materials are more than heterogeneous
materials: they are structured hierarchically. When describing cellular materials,
one has to be aware that different structural features that can influence the proper-
ties of the cellular material can be defined for each hierarchical level.
Understanding such a complex system is rather difficult. As a consequence, no
general material rules based on real structures can be developed to explain the in-
fluences of structure and architecture on properties. Furthermore, no guidelines
for the characterization of cellular metals have been established. In recent years
structural features that influence the properties have been identified but their in-
teractions and effects on real metal foams are not known quantitatively.
Figure 4.1-2. Dependence of density on structural features of pores and cell skeleton.
the same density, but one samples can have one or two big pores that represent
most of the volume fraction of the gaseous phase, and the other sample (for in-
stance made by a replication technique or investment casting) can consist of a net-
work of nearly uniform small pores. Figure 4.1.-3 shows differences in structures
of cellular material.
b)
Cell Skeleton
After characterizing the pore structure, it still remains to define quantities for de-
scribing the cell skeleton. As explained in the introduction of this chapter, the ske-
leton consists of cell nodes interconnected by cell edges. In a closed-cell structure,
the cell edges are connected by cell walls that separate the cells.
The volume fraction of metal in nodes, edges, and walls defines the density of
the cellular material. The features that can be described are number of nodes,
number of cell edges and walls per cell, length, thickness, and curvature of cell
edges, angle between nodes and cell edges, and shape of cell edge, number of
cell walls connected of edges and of edges connected at nodes (connectivily). A
drawback of quantitative image analysis of nontransparent samples is that it is
difficult to define the exact position and extent of the cell walls, edges, and
nodes, as well as the range of these features.
All these parameters have an influence on mechanical properties. As long as the
cellular material is nearly uniform and perfect, these are the most important para-
meters, but in real foams parameters are necessary to characterize agglomeration
of solid material. Klocker proposed the analysis of covariance function for detection
of clustering [10]. In a comparison of undeformed and crushed Alporas samples,
clustering of cells could be shown by a drop in covariance. Another way of charac-
terizing clustering is the pair correlation function [16], but this method has not yet
been applied for describing cell clustering in metallic foams.
cellular material [11 14]. Unfortunately, most metallic foams tend to show many
of these defects.
Defects in cell edges and walls have been identified as reducing the elastic mod-
ulus of metallic foams significantly. Two different typical defects are known: curved
cell edges and corrugated cell walls and edges. Corrugated cell walls are typically
found in metal foams such as Cymat (Alcan) or Hydro foams, which have rather
thin cell walls compared to the cell size (Fig. 4.1-5). Curved or corrugated cell
walls occur during forming or on solidification.
In modeling, cell edges or walls are often described by beam or shell elements
having an ideal shape and constant thickness. In real foams, the metal is thicker
at cell nodes and edges because of the dynamics of metal flow during foaming.
Flow of the melt also causes the rupture of cell walls. During solidification, dendri-
tic growth at a free surface and shrinkage of the melt can induce roughening of cell
walls, edges, or nodes. Figure 4.1-6 shows the surface of a solidified cell wall. The
shrinkage of the melt during solidification and cooling leads to cell-wall wriggling
and also cracks in cell walls. The influence of initial cracks on toughness or fatigue
behavior is still under discussion.
The formation of micropores is also connected to sucking of liquid into the
nodes (Plateau effect). Owing to the increased amount of liquid in nodes, the
solidification starts at the surface of the nodes caused by the heat flow. While the
nodes solidify the melt is consumed and, as know from casting, small voids occur
in the middle of the node. This can be observed as micropores in cell nodes or
thicker cell edges.
4.1.2
Characterization Methods and Quantities of Geometric Architecture of Real Metallic
Foams
Although XCT is now established for the investigation of metallic foams (see
Section 4.2), the characterization of metallic foams is usually performed using
quantitative 2D image analysis methods. Most papers concentrate on the character-
ization of cell structure derived from analysis of cross sections. For further reading
on the principles of stereology the reader is referred to the basic literature of Exner
[15] and Ohser [16].
In the following section the structure of cellular metals and metallic foams are
described and typical structural features are discussed. Before going into detail on
structural analysis, some information on the preparation techniques for metallic
foams is given.
a)
a)
b)
as Alulight is the non-uniform distribution of solid metal in the foamed part (see
Fig. 4.1-3a and 4.1-4). Zones having an enrichment of solid material are referred as
hard spots in the literature [18].
Measurements of the pores size for both foam types have shown that they have
the same average equivalent diameter of about 3 mm. The distribution curves of
equivalent diameter versus frequency of pores of both foam types show a bimodal
shape (Fig. 4.1-8a and Fig. 4.1-9a). The large fraction of small pores is caused by
solidification shrinkage. Interestingly, it seems that the Alulight material always
shows a broader distribution curve. Because of the high number of small pores,
the cell-size distribution is dominated by the small pores. The presentation of
area fraction of a pore class over equivalent diameter helps to emphasise the con-
tribution of big pores to geometric structure (Fig. 4.1-8b and Fig. 4.1-9b). The
presence of outliers can be made more visible if this method of presenting the re-
sults is used. Outliers having a pores size bigger than five times the average pore
size can be detected in nearly all samples produced by PCF techniques, such as
Alulight. This is not the case in metallic foams produced by the melt route.
4 Characterization of Cellular Metals 139
Figure 4.1-10. Density as a function of pore size for various foam types (data for FORMGRIP
process from [20]).
From quantitative structology different methods for defining shape factors are
known. Klocker defined the ratio between the largest and the smallest principle
cell dimension as the shape anisotropy ratio [10]. A similar definition was used
by Simone (ellipse shape ratio) [22]. Another shape factor is circularity, which is
the ratio of pore area and the square of pore perimeter [23]. For all mentioned
shape factors a value of 1 corresponds to a circular shape. The circularity allows
one to interpret the shape of the cell; the circularity of a rectangle is 1.4 and
that of a regular pentagon is 1.1.
Klocker has shown that the shape anisotropy ratio of Alporas and Alulight has an
average ratio of 1.5 [10]. The ellipse shape ratio of Alulight foam is Gaussian dis-
tributed, and has an average value of 0.6 [17]. In Cymat foam an ellipsoidal shape
ratio of a factor of 1.8 was found [22]. Studies of evolution of pore shape during
foaming revealed that foams having a low expansion rate (high density) tend to
have pores with a circular shape. As the foaming process proceeds the shape of
the pores becomes rectangular or ellipsoidal [23].
4.1.3
Characterization of Microstructure of Massive Cell Material
Another interesting result was found by MuÈller et al. when studying the micro-
structure of Fraunhofer foams made of the casting alloy AlSi7 [30]. They found
some Ti-rich particles after foaming (Fig. 4.1-13). It can be assumed that these
particles are residuals of the TiH2 blowing agent, which are stabilized by the
formation of a diffusion barrier. Braune et al. (Section 3.1) reported that
they observed refoaming in some foamed samples when the samples were
heated before hot forming. Both observations indicate that due to the forma-
tion of intermetallic layers surrounding the TiH2 particle, it is likely that the
dissociation decelerated.
4.1.4
Conclusions
But with the ongoing research on cellular metals the investigation of microstruc-
ture is important. Often there are typical defects, such as irregularity in cell shape,
cracks in cellular structure, or micropores caused by the intrinsic properties of the
massive metal. So in future, the emphasis in foam research will be put on the
microstructure, because not all material properties can be explained by the
architecture of the cellular metal and knowledge of microstructure is essential
for understanding processing.
References
23. F. Berger, ªEinfluss der Prozessparameter Materialforschung und -PruÈfung, Bad Nau-
auf die Schaumtruktur von Al-SchaÈumen auf heim, Germany 1998, p. 167.
pulvermetallurgischer Baisisº, Thesis, Lehr- 27. D. Lehmhus, C. Marschner, J. Banhart, in
stuhl fuÈr Werkstoffkunde und Technologie MetallschaÈume, H. P. Degischer (ed), Wiley
der Metalle, UniversitaÈt Erlangen, Germany VCH, Weinheim 2000, p. 474 477.
2000. 28. B. Kriszt, O. Kraft, H. Clemens, Materwiss.
24. U. Mosler, G. Heinzel, U. Martin, H. Oet- Werkstofftechn. 2000, 31, 432.
tel, Materwiss. Werkstofftechn. 2000, 31, 29. Y. Sigimura, J. Meyer, M. Y. He, H. Barth-
519 522. Smith, J. Grenstedt, A. E. Evans, Acta Mater.
25. B. Kriszt, B. Foroughi, K. Faure, H. P. De- 1999, 45, 5245 5259.
gischer, Mater. Sci. Technol. 2000, 16, 792. 30. A. MuÈller, U. Mosler, Fortschritte in der
26. B. Kriszt, R. Gradinger, B. Zettl, Metallographie, G. Petzow (ed), Prakt. Met.
ªUntersuchung von Aluminiumschaum, im Sonderband, Vol. 32, DGM-Informations-
speziellen der mechanischen Eigenschaftenº gesellschaft, Frankfurt 2001, p. 279 284.
in WerkstoffpruÈ fung, Deutscher Verband fuÈr
4.2
Computed X-ray Tomography
E. Maire
4.2.1
Principle of the Technique
a)
b)
Figure 4.2-1. a) Diagram of the principle of the X-ray radiography technique used in the present
chapter with a diverging beam for a medium resolution. b) Diagram of the X-ray tomography
technique.
where K is a constant, r and Z are the density and the atomic number of the in-
vestigated material, and E is the energy of the incident photons. Above 60 keV
the Compton effect dominates, where the attenuation becomes nearly proportional
to the density.
This attenuation law explains the contrast observed in the X-ray radiograph of a
bulky material (see Fig. 4.2-1a) because each point of a detector placed behind the
sample is situated in front of a different path: if the material is heterogeneous, the
integral value of m(x,y,z) varies also with x and y.
4.2.2
Set-ups
Different set-ups can be used. They all include a source, a rotation stage on which
the object is fixed, and an X-ray detector. The easiest way of getting digitized
images is to use directly a two-dimensional radioscopic detector composed of a
screen transforming the X-rays into visible light, which is then transferred by ap-
propriate optic lenses to a cooled CCD camera. The common geometric constraint
of the different set-ups is that the axis of rotation of the sample must be parallel to
the plane of the detector and its image must be aligned with one of the columns of
the CCD (preferably the central column). The crucial point in applying tomography
to materials science is the achievable spatial resolution r defined as the size of the
pixel of the detector in reference of the sample. Its limit value is mainly governed
by the available photon flux at the level of the sample and by the set-up, as de-
scribed in the next two sections. For medical tomography (XCT scanner) this
limit of resolution is of the order of 300 mm. Materials scientists wishing to see
and distinguish features with a size of the order of 1 10 mm had to develop appro-
priate tools.
4.2.3
Experimental Results
a) b)
Figure 4.2-2. a) Slice extracted from the tomographic reconstructed volume of a sample of 6 %
relative density Norsk Hydro foam. b) Slice extracted from the tomographic reconstructed volume
of a sample of 11 % relative density Norsk Hydro foam.
of the sample). The internal structure and cell-wall arrangement is clearly imaged
and the difference between the two samples is easy to visualize. The relative den-
sities of the two materials shown in these figures are 6 and 11 %. A 3D rendering
of the 6 % foam is shown in Fig. 4.2-3. This rendering is obtained by removing the
pixels located in the pores, showing then only the material located in the walls of
the foam. This virtual image really looks like an optical picture of foam, as can be
seen in different sections of this book. From such images, 3D morphological char-
acteristics can be measured such as the wall thickness, the wall dimensions, the
cell size and so forth. These characteristics have a strong influence on the mechan-
ical properties.
Note also that these images can be used with a lower resolution to assess the
density fluctuations inside the material (m is directly related to the density). It is
perfectly well established that the value of the global relative density of a sample
of metallic foam has a direct effect on its mechanical properties but the density
fluctuations have also a strong influence and especially on the scatter of these prop-
erties. This aspect is developed in Chapter 6 of this book.
a)
b)
c)
The compression set-up used for this study was an Adamel Lomargy standard
mechanical frame. The sample was placed between two plates and the force/dis-
placement curve was recorded. The resulting s(e) conventional curve was found
to be classical for this material with a yield stress of 0.7 MPa and a slight increase
in the stress during the plastic collapse. We made three steps during the compres-
sion loading to allow the characterization of the internal structure by tomography.
The three steps were made at increasing values of the remanant plastic true strain:
0.065, 0.2, and 0.6, measured directly from the change in dimensions of the paral-
lelepiped after unloading.
Figure 4.2-4 shows a set of 2D reconstructed slices of the same zone of the sam-
ple numerically and nondestructively extracted from the volume at the four com-
pression stages from 0 to 0.6 of remanant true strain. These slices are parallel to
the compression axis, which is vertical on the figure. One can clearly see the defor-
mation mechanism of the studied foam in compression by comparing figures (a)
(initial state) and (b) (remnant strain 0.065). The large value of the plastic strain
at this stage is clearly not due to a homogeneous plastic straining of the whole
sample, but to the local buckling of several walls. Between the second and the
third step, we observe that the number of new buckled walls is small. The defor-
mation process leads first to the complete closing of the collapsed cells instead
of the appearance of new ones. From the observation of the entire population of
buckled walls, we have observed that they are situated in a band perpendicular
to the loading axis (see Fig. 4.2-4c). These 3D nondestructive observations confirm
those made using other techniques [9] or with the same technique by other authors
[3,5,6].
4.2.4
Micromodeling of a Foam by Finite Elements
The internal structure of a metal foam is extremely complex: the cell size and the
length of the walls are parameters that are widely spread. The nonperiodic charac-
ter of the structure leads also to a very difficult calculation of the actual force ap-
plied to each wall compared to the case of honeycombs. The complexity of this en-
tanglement suggests analyzing the distribution of the stress and deformation in
each wall using an appropriate tool: finite element analysis. In the present study
we have a 3D picture of the actual microstructure of the foam. In what follows,
we present an easy way to use this picture to generate meshed models of the actual
microstructure. These models are readable by the Abaqus commercial code, which
we have used to perform calculations, firstly in 2D and then in 3D.
a cubic element in a finite element mesh. The cubic elements located in the voids
are then removed from the model. Given the experimental conditions we used, we
considered that the boundary between the steel plates of the compression rig and
the aluminum foam was perfect. The displacement of the lower nodes were fixed
in the three directions. We generated a steel plate perfectly bonded to the upper
nodes and applied a displacement along the compression direction to the upper
nodes of this steel plate.
4.2.4.2 Results
All the calculations were performed assuming elastic behavior for the walls.
Two-dimensional (2D)
The procedure described above can be very simply applied to any slice by generat-
ing only one layer of four-node 2D solid elements. We applied this technique for
several slices. The elements used were plane-strain four-nodes elements (PE4 Aba-
qus elements).
Figure 4.2-5 shows a contour plot of the s22 (i. e., the normal component in the
direction of loading) component of the stress tensor in the walls calculated accord-
ing to our procedure for a slice with PE elements. The higher compressive stresses
are shown in black on the figure. What we want to reach with this calculation is not
the absolute value of the stress but the determination of the sign of this stress so
the color bar has been intentionally simplified to distinguish tension and compres-
Figure 4.2-5. Stress map (arbitrary units) calculated using the Abaqus FE code by converting the
actual microstructure of a 2D slice into a mesh using the local procedure described in this
chapter. The stress component shown here is the normal component.
4 Characterization of Cellular Metals 153
Figure 4.2-6.Enlargement of a region of the stress map shown in Fig. 4.2-5, which shows the
bending character of the stress field when calculated in 2D.
sion. Note that the values of the stresses in the walls seem to form paths of high
compression loading between the top and the bottom of the section. These paths
define ªpillarsº where the compression load is transferred while some other re-
gions are much less loaded. This permits us to locate highly loaded regions in
the material.
If we zoom in a single wall situated in one of these pillars and oriented along the
compression axis, we observe a very special stress state composed of zones in com-
pression (black) and in tension (light gray), which indicate a high bending charac-
ter of the stress (see Fig. 4.2-6) with 2D assumptions. Plane-strain elements imply
that the cells that are defined in Fig. 4.2-6 are section of tubes having an infinite
dimension along the third axis perpendicular to the figure. This tube shape pro-
motes a high bending stress state within the plane of the slice only.
Three-dimensional (3D)
A complete 3D calculation of the entire foam with our technique would require a
1 000 000 element mesh. This amount of data is not tractable with our computer
today. A region of 1003 voxels was selected from the initial raw data. In order to
reduce the number of elements in the final mesh, the voxels of the region were
grouped eight by eight (two by two in the three directions), which led to an increase
in the actual resolution in the final mesh leading to a size of 300 mm q 300 mm q
300 mm for each element. This 503 voxels 3D block was transformed into a mesh
according to the previously described procedure and the mesh was slightly com-
pressed (0.02 % of elastic strain). The calculation took around 140 h on a standard
workstation. This allowed us to analyze the stresses and strains in the walls as illus-
154 4.2 Computed X-ray Tomography
trated in Fig. 4.2-7 and Fig. 4.2-8, which show a contour plot of the normal stress
and of the Tresca shear stress in a layer. As for 2D simulation, the color map has
been simplified to emphasize the sign of the stress rather than its absolute value.
Note that Fig. 4.2-7 shows much less region loaded in tension (black) than the
equivalent 2D calculation, shown for instance in Fig. 4.2-6. This indicates that
when calculated in 3D, the stress state in the walls is completely different than
Figure 4.2-7. Stress map (arbitrary units) cal- applied strain was 0.02 %. The amount of ten-
culated using 3D elements on a representative sile stress is much lower than for the 2D cal-
region of the actual compressed block. The culation (compare with Fig. 4.2-6).
normal stress component is shown here. The
Figure 4.2-8. A map of the value of the Tresca's shear stress calculated in the same 3D condi-
tions as for Fig. 4.2-7.
4 Characterization of Cellular Metals 155
4.2.5
Conclusions
References
4.3
Considerations on Quality Features
The quality of a material is defined by its properties with respect to the expected
performance. A general demand for engineering materials (not necessarily for ob-
jects of arts) is reproducibility in the sense that the product exhibits the same prop-
erties within specified limits delivered in different samples or batches in order to
be predictable in its performance and an engineer can design a reliable component
suitable for series production. Material science relates properties to microstructural
features of bulk materials, for which quality criteria can be deduced defining limits
for structural variations (for instance grain size) and for inhomogeneities (for in-
stance frequency and size of inclusions). Usually the processing steps producing
the microstructure are specified to guarantee the correlated properties.
4.3.1
Introduction
Cellular metals are a new class of materials on the brink to industrial application
and therefore the producers have to develop specifications for their product with
respect to the application aimed for. In Chapter 7 various applications are pro-
posed, each requiring different property profiles. Applications in architecture ob-
viously will require different quality criteria than heat exchangers or crash absor-
bers. A design guide [1] based on the available properties of different cellular me-
tals and material laws has been offered to the engineers for product development.
The great variety of processes and products presents a considerable challenge for
the selection of the appropriate cellular material for a potential application.
4.3.2
Non-Uniformity of Cellular Metals
Regular cellular structures like those considered for modeling [2] can hardly be pro-
duced. The most regular 3D structures are achieved in lattice block material (LBM)
[3], which can be simulated by a 3D periodic array of volume elements that repre-
sent the main structural features and exhibit the corresponding properties. The be-
havior of any macroscopic part made of LBM can be calculated (see Chapter 6) by
means of known properties of the representative volume (RV), which contains just
as many cells that the volume related properties do not change by adding more
cells, but would change if one cell were taken off. RV is the smallest volume exhi-
biting the properties of the corresponding cellular structure of a given material
with a certain cell architecture yielding an uniform average density. Consequently,
no dimension of a component should be smaller than the RV, if the performance of
the specified cellular structure were required. Any disturbance in the regularity of
4 Characterization of Cellular Metals 157
their specific loading conditions related to the geometry and to the surface. In a
first approach, the cellular metal or part is defined by its processing route, although
the interaction of the different processing parameters cannot be fully correlated to
the produced cellular structure yet. Each type of product may be characterized by
investigating the statistics of achieved properties and the correlated structural fea-
tures accessible to quality control [7]. The range of scatter of properties acceptable
for the envisaged service performance has to be defined, which can be related to
the scatter of structural features. Beyond specification limits, defects can be de-
fined, which have to be weighted in their influence on the service properties. Ac-
cording to the actual development stage, it can be expected that some defects could
be eliminated or considerably reduced by improved processing techniques.
A hierarchical approach is proposed for defining quality criteria of cellular me-
tals by addressing different description levels according to the characterization le-
vels described in Section 4.1: macro-, meso-, and microscopic. Corresponding fea-
tures are to be defined that can be quantified and validated with respect to material
properties.
The average density of a part or component is an example of a macroscopic level.
The question is: How reproducible is the density of a given part in series produc-
tion? How much variation is tolerable with regard to the scatter of related proper-
ties? Even for a given average density of a part the properties might scatter owing
to the local scatter of mass densities within the part, which may vary on a meso-
scopic scale. Going into further structural details on the microscopic level, the var-
iations in shape factors, in dimensional and microstructural features of the cell
walls and nodes have to be quantified statistically and assessed with respect to tol-
erable variations or irregularities beyond.
4 Characterization of Cellular Metals 159
4.3.3
Macroscopic Parameters
All those parameters are referred to as macroscopic, which can be assigned to the
whole cellular part without looking into the structure in detail: the general appear-
ance of a cellular architecture, the composition of the ingredients, and the process-
ing method.
In Section 7.3 the problem of deforming the cell walls during mechanical cutting
operations is addressed. The degree of such deformations could be of importance
for following joining operations. The enlargement of outer surface in comparison
with a carefully prepared cross section could be employed to quantify the smearing
of cross sections of elements of the cellular structure by machining.
Foamed and cast cellular metals are usually surrounded by a metal skin. The
oxide content will be much higher than on interior cell surfaces, because that
skin was exposed directly to the atmosphere in the liquid state and during solidi-
fication and cooling. The roughness depends greatly on the used mold and its coat-
ing and may be qualified similar to sand casting quality. Attempts to measure the
thickness of the skin by eddy current tests did not succeed so far. A skin region of
about 1.5 mm thickness can be deduced from Fig. 4.3-2 of an Alulight sample,
where the local density is almost twice of that in the center of the plate. Still the
skin cannot be regarded as solid metal, but consists of porous material.
There are as well pores open to the environment in such skins, which can be
detected by dye penetrant tests [6]. Although these pores may be in the range of
tenths of a millimeter in diameter, they form a passage to the cells below, which
might be of a few millimeters in diameter. Such open pores might adsorb humidity
and thus enhance corrosion attack. These pores prohibit chemical treatments like
anodizing, because the electrolyte penetrates and can hardly be removed. On the
other hand, if those pores are sealed by a coating, which has to be baked, it
might form blisters when the temperature rises and the internal gas pressure in-
creases. Ambient temperature coating processes with a strong sealing effect can be
applied successfully [9]. The number of surface pores per surface area and their
distribution of diameters are quantifiable surface quality features. A pore density
of 1-3 pores/cm2 has been measured on Alulight samples [6].
The dimensional accuracy of net shape metal foam parts depends on the thermal
stability of the mold and the uniform shrinkage of the foam during solidification
and cooling. The cell-size distribution, especially along the surface of the part, af-
fects the surface roughness. Big pores (in the range of one centimeter) may cause
dents due to the drastic reduction of pressure of the gas in the closed pores during
cooling. The achievable tolerances for net shape metal foams may be in the range
of percent, but no detailed study has been published so far.
4.3.3.4 Properties
Macroscopic properties of test samples are described in Chapter 5, where they are
usually related to the apparent density of the specimens. The observed scatter in
mechanical properties, like the plateau stress for instance, depends very much
on the uniformity of the cellular structure. In general, the properties are closely
related to the quality of the cellular architecture. The scatter in the test results in-
dicate the variations in structure of the individual test specimens. Most of the tests
are executed without knowing the microgeometry of the test sample exactly, there-
fore there is little evidence on the correlation with structural features, especially
when the tests are destructive. A systematic correlation of properties to the
three-dimensional cellular architecture is highly desirable in order to give rise to
process developments aiming for improvements in the reliability of the service
related performance of the foams.
4.3.4
Microscopic Features
In terms of microscopic features the individual cells are to be considered: the mi-
crostructure of the metal forming the cell walls and their topological parameters
described in Section 4.1.
162 4.3 Considerations on Quality Features
Figure 4.3-3. Cross sections of mass distribu- Alulight AlMgSiCu of r 0.39 g/cm3, c) Alu-
tions computed from X-ray microtomograms light AlSi10Mg r 0.37 g/cm3, d) same mate-
(voxel size 40 mm q 40 mm q 40 mm) of foamed rial as (c), but specially foamed small cell ar-
aluminum: a) Alporas of r 0.45 g/cm3, b) chitecture r 0.56 g/cm3.
164 4.3 Considerations on Quality Features
Figure 4.3-4. Mass distributions computed polygonal cell walls (same magnification as
from micro X-ray tomograms of foamed Al- Fig. 4.3-3; note that not all of them are end-on),
Si9Mg/SiC/20p melt of r 0.11 g/cm3 (voxel b) 3D cell reconstruction of a cuboid volume
size 0.1 mm q 0.1 mm q 0.1 mm): a) cross element (22 mm q 18 mm q 16 mm) indi-
section showing different plateau borders and cating curved shapes of the cell walls.
4.3.4.3 Microdefects
Of course, pores created in cellular metals due to processing imperfections like
shrinkage pores, imperfect sintering, spraying pores etc., may not to be considered
as essential defects. But in general the consistence of the percolating solid should
be of similar metallic properties throughout the cellular structure. Cellular struc-
tures processed in the melt or at elevated temperatures suffer thermal stresses dur-
ing solidification and cooling. Hot tearing and cracking is very probable because of
the multi-axial stress situations created in walls of little strength (see Section 4.1).
Almost 100 % connectivity of the pores have been determined within powder com-
pact foamed aluminum [5], although almost 40 % of the solid is located within the
cell walls. Cell-wall cracks have been observed frequently in powder compact foams
[19].
The material properties of the metal in the cellular structure have to be consid-
ered to be of poorer quality than those of the same metal in bulk. For those pro-
duced from the melt it is expected that the alloy exhibits the properties of a low
quality casting. If the frequency of microdefects depends on the position within
a sample, their effect should be considered as deviation from uniformity.
4 Characterization of Cellular Metals 165
4.3.5
Mesoscopic Features
The mesoscopic description of real cellular metals assumes that a sample or part
can be described as an arrangement of volume elements, each representing a char-
acteristic microgeometry. The local inhomogeneities in architecture and in mass
distributions can be classified by different mean cellular structures approximating
the properties of each sub-domain of the material. A mesoscopic model for uniaxial
compression has been proposed [20,21], which divides the samples into layers of
cellular materials with different apparent densities. Generalizing this concept,
any sample or part can be divided into a 3D arrangement of sub-domains of differ-
ent cellular phases of known features and properties. Thus the cellular sample is
considered as a multiphase material (see Fig. 4.3-1), where a cellular phase is
defined by its composition, microgeometry, and mass distribution occupying the
volume of a sub-domain.
The mesoscopic behavior of each sub-domain corresponds to that of the asso-
ciated RV, so that each sub-domain behaves like a cellular material to which macro-
scopic material laws of cellular metals can be assigned. In the extreme case, where
the sub-domains were smaller than RVs (i. e., most cells were essentially different
from their neighbors), such a mesoscopic approximation would not make sense
and complex finite element modeling of the real microstructure would be neces-
sary as shown in Section 4.2. In the best case of samples of uniform microgeo-
metry like Duocel, Incofoam, or those cast by replication, the mesoscopic consid-
eration is superfluous and the macroscopic treatment would result satisfactorily.
In general, the range of variations in size, volume fractions and properties of
the mesoscopic sub-domains can be quantified providing a quality measure on uni-
formity of the considered part.
4.3.6
Systematics of Quality Features
The features characterizing parts or samples made of cellular metals are correlated
to the structural hierarchy described above in a matrix presented in Table 4.3-1,
from which quantifiable magnitudes for quality considerations are deduced.
Most of the microscopical parameters are described in Section 4.1 and require con-
siderable effort to be determined statistically, especially in 3D. The macroscopic
magnitudes can be obtained from the manufacturer and can be complemented
by nondestructive tests except the mechanical properties. The mechanical proper-
ties of uniform cellular metal parts can be calculated by simulating the geometrical
condition with input data from test samples of equal materials. The properties of
non-uniform cellular parts (either intentionally graded or inhomogeneous due to
the processing conditions) can be estimated from their mesoscopic structure
when the scaling laws with respect to the relevant microstructural features are
known.
Mesoscopic parameters can be defined on the condition that ensembles of cells
exhibit similar microstructural features, which form more or less uniform sub-do-
mains of a sample of cellular metal. An attempt is made to characterize a cellular
material without determining microscopic features quantitatively. Irregularities in
a certain depth from the surface can be detected by ultrasonic spectroscopy [26] as
Table 4.3-1. Matrix of features of considered quality aspects and correlated quantities within the structural hierarchy of cellular metals.
Cell Cell size, shape Cell volume histo- Definition of vol- 3D data set of arrange- Processing method, Processing para-
architecture and orientation gram shape, shape umes of sub-do- ment of sub-domains cell type, permeabil- meters, category of
of pores and factors and orienta- mains of similar in size and shape com- ity architecture, perme-
cell walls, tion distributions, microgeometry posing the cellular ability coefficient,
connectivity connectivity para- component surface/volume ratio
meter
Dimensional Dimensions of Statistics of dimen- Sensitivity of sub- Dependence of sub-do- Achievable toler- Lateral dimensions,
accuracy constituents of sions of cell edges/ domain volumes main arrangement on ances in dimensions radii at corners
cells, irregulari- faces; buckling, on averaging para- the averaging limits for
ties wiggles, circularity, meters microstructural features
sphericity, cracks
Surfaces Composition, Definition of surface Interface between Boundary conditions Surface formation Fabrication method,
structure, thick- region/skin, form of sub-domains or for interfaces of sub- and topology properties, rough-
ness of outer and constituents, chemical sections of outer domains; specific phase ness measure, sur-
inner surface composition, topology surface for outer surface region face porosity index
Metal matrix Grain structure, Grain size and inter- Similar condition Correspondence of sub- Alloy composition, Chemical analysis of
phase distribu- dendritic spacing, in- (microstructure) of domain volumes with additives, metal con- ingredients, thermo-
tions clusions, precipitates the solid uniform relevant mi- dition mechanical manu-
in function of position crostructure facturing para-
in cell/sample meters
Mass Fraction of mass Mass distribution Similar solid dis- Correspondence of sub- Average, apparent Absolute or relative
distribution in cell constitu- coefficient between tribution, uniform domain volumes with mass density; center mass density, differ-
ents faces, Plateau borders apparent density continuous uniform of gravity ence of centers of
and nodes mass distributions gravity of geometry
and of mass
4 Characterization of Cellular Metals
Properties Metal properties Physical and mechan- Properties of an Criteria for definition of Service properties of Experimental data
167
on microscopic ical properties of solid ensemble of simi- properties of sub- component and simulation re-
scale different in function of position lar cells domains or RV, scaling sults with scatter
from bulk in cell and sample laws and their accuracy and validity limits
168 4.3 Considerations on Quality Features
long as the cell architecture is essentially uniform. In the case of regular Duocel
samples relative density variations could be observed by eddy current measure-
ments [27]. The following paragraphs describe the possibilities to determine meso-
scopic parameters of non-uniform cellular metals and discuss their applicability for
quality criteria that can predict, if a sample's behavior complies with the specifica-
tions of service requirements.
4.3.7
Approximation of a Cellular Structure by a Continuum
For simplicity it is assumed that the microstructure of the metal and the type of
cell architecture are essentially defined by the processing method. The remaining
variables are the local density distribution and the properties of the solid material,
which have to be calibrated. A non-uniform object can be divided into sub-domains
to which the appropriate material laws can be applied to simulate the whole
part's performance. For this purpose, it is convenient to transform the cellular
structure into a continuum, which behaves like a cellular architecture (crushable
material). The experimental basis is provided by computed X-ray tomography
[13] of the 3D metal distribution of a sample, which will be locally averaged to
yield density values in each point of the sample smoothing away the originally
cellular structure.
where
R g is a weight function that has to fulfill the normalization condition
g(u)d3u = 1. Equation (1) is also written in the form ra g*r, which denotes
the convolution of the function r with g. Cellular structures can be investigated
by X-ray radiography or by X-ray computed tomography as described in
Section 4.2. The density distribution is represented by a voxelized data set mXCT(rijk)
that can be used to calculate a density mapping as defined in Eq. (1). Due to the
limited resolution of the XCT equipment mXCT(rijk) is related to the linear mass
attenuation m(r) of the object by
Z
mXCT (rijk ) PSF(rijk s r 0 )m(r 0 )d3 r 0 PSF m (2)
4 Characterization of Cellular Metals 169
where the point spread function PSF describes the ªblurringº of the object function
m(r) [28].
For example, the micro XCT slices in Fig. 4.3-5a and Fig. 4.3-5d are a very good
representation of the real foam (i. e., the PSF is a sharp peak), whereas the corre-
sponding slices obtained by medical XCT (Fig. 4.3-5b and Fig. 4.3-5e) are of signif-
icantly lower resolution (the PSF is relatively broad). Since the metallic foams in-
vestigated herein consist of a single alloy, m(r) is proportional to r(r) and analo-
gously to Eq.(2), rXCT PSF*r can be defined. Averaging these rXCT values with
the weight function w yields raaccording Eq.(1)
X X Z
w(rijk s rlmn )rXCT (rlmn w(rijk s rlmn ) PSF(rlmn s r 0 )r(r 0 )d3 r 0
l;m;n l;m;n
Z X Z
w(rijk s rlmn )PSF(rlmn s r 0 )r(r 0 )d3 r 0 g(r 0 srijk )r(r 0 )d3 r 0 ra (rijk ) (3)
l;m;n
It should be noted that rXCT is already a function of the form of Eq. (1), i. e., rXCT
can be understood as a density mapping performed by the tomograph itself and a
Figure 4.3-5. Comparison of a micro XCT with c) calculated density map using averaging
pictures obtained from a human tomograph volumes of 6.8 mm q 6.8 mm q 6.8 mm.
and calculated density distributions for Alulight Alporas of r 0.45 g/cm3 d) micro XCT slice of
Si12Mg of r 0.5 g/cm3: a) high-resolution 40 mm, e) human XCT of a 3 mm slice (0.9 mm
micro XCT for a slice thickness of 40 mm (40 mm q 0.9 mm q 3 mm voxel), f) density map
q 40 mm q 40 mm voxel), b) human XCT of a calculated using the same averaging volume
1 mm slice (0.6 mm q 0.6 mm q 1 mm voxel), as in (c).
170 4.3 Considerations on Quality Features
second mathematical averaging process reveals the final mapping ra (see Fig. 4.3-
5c and Fig. 4.3-5f). The density mappings in this section are calculated by arith-
metic mean values of the voxels rXCT(rijk) over a cuboid (see Fig. 4.3-6)
1X
ra (rijk ) rXCT (rlmn ) (4)
N l;m;n
where l, m, n run over the voxel indices of the selected cuboidal surrounding of the
location rijk and N is the number of voxels, i. e., the weight function in Eq. (3) is w
1/N within the cuboid and w 0 otherwise.
Figure 4.3-7 shows a one-dimensional model r(x) of a cellular metal with walls of
the solid density rS and embedded empty pores. Convolution of r(x) with the PSF
results in the depicted rXCT(xi) values and further averaging over intervals [xi 34,
xi 34] with N 69 yields the mean local density function ra(xi).
As long as the averaging volume of the density mapping is bigger than the reso-
lution of the tomography, the resulting density map is almost independent of the
XCT resolution [29]. Using averaging cuboids approximately as big as the cells
suppresses microstructural details, but allows to determine local differences in
Figure 4.3-7. One-dimensional model of the Eq.(4) for each voxel with an averaging interval
density mapping procedure: the real mass of xi e 34 voxels yields the smoothed density
distribution r(x) is imaged by the XCT data values ra(xi). rm is the average density; the
rXCT(xi) of the voxels, which are resulting from variations e Dr correspond to e rm/3.
the convolution of r(x) with the PSF; applying
4 Characterization of Cellular Metals 171
Figure 4.3-8. Example of the effect of the size 10 mm apart; 2nd to 5th column, calculated
of the averaging volume on the presentation of density distributions of the same planes for
density variations in a AlMgSiCu/Al2O3/20p averaging volumes of 4 mm q 4 mm q 4 mm,
melt foamed sample of r 0.25 g/cm3 [15]: 1st 6 mm q 6 mm q 6 mm, 9.5 mm q 9.5 mm q
column, microscopic XCT images (0.1 mm q 9.5 mm, and 15 mm q 15 mm q 15 mm.
0.1 mm q 0.1 mm voxel) of parallel planes
a)
b)
Figure 4.3-9. Illustration of the 3D local den- r 0.45 g/cm3; b) Alulight AlSi12Mg of
sity distribution obtained from 4 mm q 4 mm r 0.5 g/cm3; the gray value represents the
q 4 mm averaging volumes for a series indicated deviation from the apparent density
of six layers at distances of 3.7 mm of the same rm (white high density).
samples as in Fig. 4.3-5: a) Alporas of
mass densities for the considered averaging volume size vary in these examples be-
tween 40 % and 160 % of the apparent density of the Alporas sample, but between
0 and more than twice of rm of the Alulight sample. The gray scale presentation in
Fig. 4.3-9 divides the samples into density classes of 0.1 rm. It is clear from the
density distribution of the Alulight sample, that no uniform material behavior
can be expected.
Figure 4.3-11. Presentation of hard and soft sity maps with averaging volumes of 6 mm q
regions within an Alulight AlSi12Mg sample (a 6 mm q 6 mm: a) iso-surface of hard sub-do-
b 22 mm, c 30 mm) of apparent rm mains with r i 1.33 rm; b) iso-surface of soft
0.5 g/cm3 by iso-surfaces computed from den- sub-domains with r I 0.67 rm.
part [30]. If there were cells bigger than the chosen averaging volume, so that
points result in the sample where no contrast is found within the surrounding cu-
boid, the density mapping would contain co-ordinates of zero density. The approx-
imate size, volume fraction, and position of pores bigger than the averaging vo-
lume can be determined by means of the calculated 3D density map. An example
for an Alulight specimen with over-sized pores is shown elsewhere [6].
X-ray radiography can be applied to flat samples. The intensity of the transmitted
columns contains the density integrated over the thickness of the sample yielding
the contrast of pixels in the projected plane. This can be transformed into 2D den-
sity maps by applying the analogous procedure of averaging over m q n pixels in
the image plane. Examples given elsewhere [6] illustrate an improved detectability
of inhomogeneities (like ensembles of big pores, drainage effects, or dense skins)
174 4.3 Considerations on Quality Features
by planar density mapping. Of course, radiography does not provide any informa-
tion on density variations in the direction of the thickness. A compromise between
3D tomography and radiography is provided by X-ray laminography [31], where the
density distribution of parallel layers of a sample can be determined with moderate
resolution.
4.3.8
Proposal of Quality Criteria
Cell architecture Reproducible and similar Reproducible processing Unsatisfactorily controlled Process records, assessment of
throughout the sample parameters parameters relevant parameter variations
Dimensional accuracy Tolerances of outer Specified variations in Dimensions beyond Contour measurements
dimensions achieved lengths and radii tolerances
Surfaces Reproducible surface Properties, roughness, Dents and blisters Roughness measurement,
features open porosity Surface porosity index Dye penetrant tests,
Chemical analysis, hardness
Metal matrix Defined chemical Uniform solid properties Scatter in chemical analysis Chemical analysis of ingredients
composition of different positions and product
Mass distribution Reproducible density Apparent mass density Scatter in average mass Density measurements,
within specifications density; asymmetry of mass Determination of centers of
distribution gravity of shape and mass
Properties Properties unambiguously Property values within Scatter beyond specifications Representative tests,
related to structural quality specifications Correlation to structure,
Service performance
4 Characterization of Cellular Metals
175
Table 4.3-3. Matrix of features of cellular metals and correlated mesoscopic quality criteria proposing their verification.
Cell architecture Reproducible space filling Master curves of local Histogram shape, Volume Histogram of mass density dis-
sub-domain arrangement density distributions fractions below and above tribution from XCT; Difference
certain limits of rm and histogram's maxi-
mum; Volume fraction of
r I rm Dr and of r i rm Dr
Dimensional accuracy Sub-domain volumes contain Convergence criterion for Averaging volume size too Averaging volume bigger than
statistically representative size of averaging volume big in relation to sample dominant cell sizes, smaller
number of cells deduced from width of size, large sub-domains of than 1/3 of sample's minimum
density histogram only few not representative thickness; Sensibility tests for
over-sized cells variations of averaging volume
size
Surfaces Outer surface boundary con- Free surface except area of Load transfer area smaller Comparison of modeling with
4.3 Considerations on Quality Features
ditions according to loading, load actuation than interacting sub- deformation experiments re-
independent of interface Spatial displacement of domain corded by XCT or digital video
boundaries sub-domain interfaces Limited correspondence images
of loading, sub-domains,
FE-mesh
Metal matrix Properties insensitive to Experience from microstruc- Definition of critical property Estimated differences of solid
microstructural variations tural examinations of same relevant irregularities properties compared with bulk
within averaging limits type of cellular metal (calibration of input properties)
Mass distribution Properties of sub-domains Determination of location Definition of critical dimen- Definition of critical Dr
represented by the local of soft and hard regions sions, connectivity and posi- Shape, connectivity and size of
average mass density of the tions extreme densities (iso-surfaces)
corresponding volume Check of gradients in main
directions
Properties Appropriate scaling laws and Calibration by experimental Establishing confidentiality Testing stiffness in different di-
constitutive equations tests limits for different types of rections, comparison with com-
known materials puter simulations (e DE(x,y,z)
for small strains)
4 Characterization of Cellular Metals 177
togram, the volume fractions of the sample with densities outside a certain speci-
fication and the local position of those extremely soft or hard regions. Even the
existence of over-sized pores can be indicated.
The following uniformity criteria are proposed for the example of Alulight
foams:
x the local density variations should be within e30 % around the mean apparent
density;
x regions of smaller mass densities than the specified range (including zero for
over-sized pores) should in any direction not be bigger than 1/3 of the thickness
of the sample;
x the sample's regions with mass densities beyond the specified limits should not
be connected in any preferential orientation causing anisotropy, and should oc-
cupy less than 5 vol.-%.
A stiffness test in different directions is proposed as an experimental check of the
mechanical behavior of a part in order to reveal any significant irregularity and
macroscopic anisotropy. The absolute stiffness values can be compared with simu-
lations of the elastic performance of that part (see Section 6.2). Up to now there is
little experience on the applicability of any quality criterion for cellular metals and
specifications will have to be developed on the basis of the experience that will be
gained from the first series products.
Acknowledgements
References
5
Material Properties
R. Pippan
Material Al SiC Al Al Al Ni
Relative density, r/r0 0.02 0.2 0.1 0.35 0.08 0.1 0.05 0.1 0.03 0.04
Structure Closed cell Closed cell Closed cell Open cell Open cell
Young's modulus, 0.02 2.0 1.7 12 0.4 1.0 0.06 0.3 0.4 1.0
E [GPa]
Poisson's ratio, n 0.31 0.34 0.31 .34 0.31 0.34 0.31 0.34 0.31 0.34
Compressive strength, 0.04 7.0 1.9 14.0 1.3 1.7 0.9 3.0 0.6 1.1
s pl [MPa]
Tensile elastic limit, 0.04 7.0 2.0 20 1.6 1.8 0.9 2.7 0.6 1.1
s y [MPa]
Tensile strength, 0.05 8.5 2.2 30 1.6 1.9 1.9 3.5 1.0 2.4
s UTS [MPa]
Endurance limit, 0.02 3.6 0.95 13 0.9 1.0 0.45 1.5 0.3 0.6
s ce [MPa]
Densification strain, eD 0.6 0.9 0.4 0.8 0.7 0.82 0.8 0.9 0.9 0.94
Tensile ductility, eUTS 0.01 0.02 0.002 0.04 0.01 0.06 0.1 0.2 0.03 0.1
Fracture toughness, 0.03 0.5 0.3 1.6 0.1 0.9 0.1 0.2 0.6 1.0
KcIC [MPa.m1/2]
Resistivity, R [10 8Vpm] 90 3000 20 200 210 250 180 450 300 500
Some scaling relations have been developed for idealized structures [2], between
the properties of the solid, the density, and the type of cellular structure: a few
of them are listed here. These relations clearly point out that the most important
structural characteristic of a cellular material is the relative density r/rs.
The importance of other structural parameters can be easily seen in the density
dependence of the thermal and electrical conductivity. The thermal conductivity of
metal foam is mainly dominated by the conductivity through the solid. The contri-
bution of the gas, radiation across the pores, and convection within the cell play a
minor role. Therefore, the thermal and the electrical conductivity should behave in
the same manner. The conductivity of the cellular metal should be equal to the
conductivity of the dense solid times its volume fraction (r/rs) multiplied by an ef-
ficiency factor, which is given by the geometry and takes into account the tortuous
path of the heat flow. If we assume that a variation of the relative density is caused
by a proportional change of the thickness of the cell wall and struts, or an increase
of the size of the cells maintaining the cell wall thickness the efficiency factor
should be constant. For Al foams it was shown that a variation of the efficiency fac-
tor can be expressed also in a power law of the relative density, it is about propor-
tional (r/rs)0.5, hence the conductivity is about proportional (r/rs)1.5.
In contrast to thermal and electrical properties, the influence of the density and
the architecture of the cellular metal on the mechanical properties is much stron-
ger and more complex. For load bearing structures, packaging, and crash elements
their importance is evident, however, for functional applications, for example, heat-
exchange systems, damping, and filtering the structural stability is also essential.
In general, one can distinguish between elastomeric, elastoplastic, and brittle
foams (this is not a classification of the solid material, it describes the behavior
of the cellular structure).
In compression, these three types of foam show a similar behavior [2] but for
different reasons. Metallic foams, having a relative densities of r/rsI0.2, exhibit
after an elastic loading a more or less clear plateau regime, leading into a final re-
gime of steeply rising stress. The differences are visible in the mechanism of de-
formation or by unloading of the compressed cell structure: in elastomeric
foams the collapse is by elastic buckling of the cell walls and is more or less reco-
verable; in plastic foams it is the formation of plastic hinges; and in brittle foams it
is the brittle fracture of the cell wall (crush) that causes the plateau region: the last
two are not recoverable.
Cellular metals are usually ductile. Figure 5-1 compares schematically the
stress strain curve of a cellular and a solid metal in tension and compression.
The foam shows a linear elastic regime, which is not so clearly established as in
a solid; followed by a ªhardeningº, which is not only hardening as in a solid, it
is induced by a redistribution of deformed regions; in compression one reaches
then a plateau regime and a final densification; in tension the maximum stress
is reached at relative small strains, typically 1 4 %, which is much smaller than
in a solid metal. The characteristic values are the Young's modulus E (strictly
the stiffness of the cellular structure), the elastic limit s y (which may be slightly
different in tension and compression), the tensile strength, s UTS, the plateau stress
182 5 Material Properties
Figure 5-1. Schematic comparison of the stress strain behavior of a metal and a metal foam.
5 Material Properties 183
References
1. L. J. Gibson, M. F. Ashby, Cellular Solids, 14. J. Banhart (ed.), MetallschaÈume, MIT Verlag,
Cambridge University Press, UK 1997. Bremen, Germany: The proceedings of a
2. W. E. Warren, A. M. Kraynik, J. Appl. Mech. conference held in Bremen in March 1997 (in
1988, 55, 341 346. German).
3. T. G. Nieh, K. Higashi, J. Wadsworth, Mater. 15. J. Banhart, M. F. Ashby, N. A. Fleck (eds.),
Sci. Eng. A 1999, 283, 105 110. Metal Foams and Foam Metal Structures, Proc.
4. J. Grenestedt, K. Tanaka, Scripta Mater. 1999, Int. Conf. Metfoam `99, 14 16 June 1999,
40, 71 77. MIT Verlag, Bremen 1999.
5. J. Grenestedt, F. Bassinet, Int. J. Mech. Sci. 16. A. G. Evans (ed.), Ultralight Metal Structure,
2000, 42, 1327 1338. Division of Applied Sciences, Harvard Uni-
6. A.-F. Bastawros, H. Bart-Smith, A. G. Evans, versity, Cambridge, MA, USA: the annual
J. Mech. Phys. Solids 2000, 48, 301 322. report on the MURI programme sponsored
7. H. Bart-Smith, A.-F. Bastawros, D. R. by the Defense Advanced Research Projects
Mumm, A. G. Evans, D. J. Sypeck, H. N. G. Agency and Office of Naval Research, 1998.
Wadley, Acta Mater. 1998, 46, 3583 3592. 17. D. S. Schwartz, D. S. Shih, A. G. Evans,
8. J. Banhart, J. Baumeister, J. Mater. Sci. 1998, H. N. G. Wadley (eds.), Porous and Cellular
33, 1431 1442. Materials for Structural Application, MRS
9. Y. Yamada et al., Mater. Sci. Eng. A 2000, 277, Proceedings Vol. 521, MRS, Warrendale,
213 217. PA, USA, 1998.
10. K. A. Dannemann, J. Lankford Jr., Mater. 18. H. P. Degischer (ed.), MetallschaÈume,
Sci. Eng. A 2000, 293, 157 164. Proceedings of the DGM Symposium
11. A. Paul, U. Ramamurty, Mater. Sci. Eng. A ªMetallschaÈumeº, 28 29 February 2000,
2000, 281, 1 7. Vienna, Austria, MATWER 31, 6, Wiley-VCH,
12. M. MuÈnch, M. Schlimmer, Mater. Sci. Eng. Weinheim 2000 (in German).
Tech. 2000, 6, 544 546. 19. B. Zettl, H. Mayer, S. E. Stanzl-Tschegg,
13. O. B. Olurin, N. A. Fleck, M. F. Ashby, H. P. Degischer, Mater. Sci. Eng. A 2000, 292,
Mater. Sci. Eng. A 2000, 291, 136 146. 1 7.
5.1
Mechanical Properties and Determination
5.1.1
Young's Modulus
One of the most important properties of structural materials is the linear elastic
behavior, which is described by a set of moduli. For isotropic materials two moduli
are needed: the Young's modulus, E, and the shear modulus, G, to characterize the
linear elastic response. Some foams are not isotropic owing to the manufacturing
processes and therefore more than two moduli are necessary to describe the linear
elastic behavior.
184 5.1 Mechanical Properties and Determination
In the isotropic case, which will be accounted for in this section owing to sim-
plification and because many closed-cell metal foams are more, or less isotropic,
the following well known relations between Young's modulus E, shear modulus
G, bulk modulus K, and elastic Poisson's ratio n exist
E E
G , K (1)
2(1 S v) 3(1 s 2v)
In the situation of cellular material, or foams it is in general not appropriate to use
the term ªmodulusº, because a foam is a structure (like a framework) and one
should use ªstiffnessº instead. However, it is now usual to consider that new
class of materials as a continuum, hence common material properties are used
to characterize them. But it is important to note that the Young's modulus of a
cellular structure is not a material constant and depends therefore mainly on the
architecture of the cellular metal.
The subscript ªsº indicates the properties of the solid from which the cell walls
and cell edges are made. F is the fraction of solid that is contained in the cell edges
for closed-cell foams. Although these equations are based on a simple model of a
cubic foam structure, which considerably deviates from a real foam, there is a fair
agreement with experimental results (Fig. 5.1-1). The equations are valid as long as
the main deformation mechanism is bending of cell edges. For higher-density
foams extension and compression of cell edges becomes more important and
therefore a deviation from these relations should be observed.
Figure 5.1-1. Dependence of the Young's modulus on the relative foam density for different types
of closed-cell (closed symbols) and open-cell (open symbols) aluminum foams [6].
186 5.1 Mechanical Properties and Determination
Figure 5.1-2. Stress strain response with measured unloading moduli indicated for a closed-cell
aluminum foam (Alporas, rr 0.1).
5 Material Properties 187
5.1.2
Compression Behavior
In compression, cellular metals show a unique stress strain response with a pla-
teau region in which the stress is nearly constant over a wide range of strain
(Fig. 5.1-4). This behavior makes cellular metals interesting for energy absorbing
applications where at a relatively low constant stress a large amount of deformation
can be absorbed. More about energy absorbing and crush behavior can be found in
Section 5.1.3. Depending on the material from which the foam is made, different
deformation mechanisms (elastomeric, brittle, and ductile behavior) can be ob-
served. Metallic foams usually show a considerable ductile behavior.
Many investigations have been performed on compression behavior and energy
absorbing performance of metal foams in recent years. The behavior is similar to
many polymer foams. Thus the basic deformation mechanisms and the stress
strain response are well known. Compression loading at small strains of ductile
foams yields to bending and extension/compression of cell edges and cell walls.
If the stresses in the edges and walls exceed the yield stress s y,s of the solid, the
onset of plastification is reached and the deformation is no longer reversible.
Caused by the inhomogeneous structure of real cellular metals the stress concen-
trations exceed the yield stress in some elements at relatively low strains, which
leads to an early ªplastificationº of the foam. Therefore the linear elastic part of
the stress strain curve of a ductile foam is in general hard developed. Increasing
the load on the foam causes to buckling of cell edges and walls in weaker regions
of the foam. A deformation band perpendicular to the loading direction develops,
in which plastic collapse of the cells takes place. This is accompanied by the begin-
ning of the plateau region in the stress strain curve. With increasing strain, addi-
tional deformation bands are formed until most of the cells have collapsed and the
188 5.1 Mechanical Properties and Determination
Stress strain response in compression for a ductile aluminum foam and a brittle
Figure 5.1-4.
alumina foam [11].
densification region is reached [8,9]. Depending on the cellular structure and the
properties of the solid, the plateau region of the stress strain curve may not be
flat, but a slight slope, or waviness of the curve can occur. Two extreme cases
are depicted in Fig. 5.1-4, the behavior of a ductile aluminum foam and a brittle
alumina foam, showing a smooth and a jagged plateau, respectively. For the esti-
mation of the plateau stress s pl depending on the foam density r and the yield
stress s y,s of the solid, Gibson and Ashby used a simple cubic-cell model for the
foam. The plateau stress is reached, when the moment exerted to the cell edges
exceeds their fully plastic moment
!3=2
s pl r
z 0:3 (open-cell)
s y; s rs
(3)
!3=2
s pl r r
z 0:3 F S0:4(1 s F) (closed-cell)
s y; s rs rs
Figure 5.1-5 shows the stress strain curves in compression of Alulight alumi-
num foams with different densities.
The increase of the plateau stress with increasing density is clearly obvious. Dif-
ferent methods exist for the measurement of the plateau stress depending on the
course of the stress strain curve. If there is a stress peak at the beginning of the
plateau, this peak or a clear flat section stress will be taken as plateau stress [10]. In
5 Material Properties 189
Figure 5.1-5. Stress strain curves of aluminum foams (Alulight) with different densities in
compression.
the absence of a stress peak the plateau region will be extrapolated to lower strains
and the intersection with the initial elastic line, or the ordinate (Rpe) delivers the
plateau stress [10,27]. The plateau stress plays an important role to characterize
the energy absorbing behavior and is a good material property for the compression
performance of a foam. However, for structural design with foams the plateau
stress may not be applicable, because of the relatively high deformation and
damage of the foam that is accompanied with this stress level. The determination
of a yield stress s y is problematic owing to the usually poorly developed linear
elastic regime of the stress strain curve. To compromise, an offset yield stress,
like s 0.2, or Rp02 should be used, depending on the mechanical response of the
cellular metal.
For the experimental determination of a stress strain curve, standard equip-
ment for compression tests can be used. If the measurement of the Young's
modulus E and the 0.2 % offset yield stress s 0.2 is needed, the displacement should
be measured directly at the specimen with a clip gauge, or a videoextensiometer. To
avoid errors arising from inhomogeneous deformation of the foam the displace-
ment should be measured at more than one position on the specimen. The
specimen size must exceed the average cell size by 7 to 10 times in each principal
direction to reduce the influence of the inhomogeneous cell structure.
190 5.1 Mechanical Properties and Determination
5.1.3
Energy-Absorbing and Impact Behavior
Good energy absorbers should support large strains at a relatively low constant
stress level. Owing to their compression behavior, cellular metals (especially
foams) and also some polymer foams, are especially suited for energy absorbing
applications. The advantages of metal foams for these applications are their rela-
tively low density, adjustable plateau stress level, good temperature resistance
and non-flammability. Thus their application fields are the transport and packaging
industries.
Figure 5.1-6. Comparison of stress strain re- the same for both. The peak stress of the ab-
sponse of two different energy absorbers. The sorber that has no plateau region is higher and
amount of absorbed energy per volume in a this will enhance the impact forces.
certain strain interval is indicated as EV and is
5 Material Properties 191
Figure 5.1-7. The energy absorbing efficiency, h, is the ratio between absorbed energy of the
material EV and the absorbed energy of an ideal plastic material EV,ideal in a certain strain range.
The absorbed energy per unit volume EV in a certain strain interval [e1, e2] is
equal to the area below the stress strain curve and can be expressed as
Z e2
EV s(e)de (4)
e1
The efficiency of energy absorption, h, is the ratio between the absorbed energy of
the real material and the absorbed energy of an ideal absorber that is an ideal plas-
tic material
R e2
s(e)de
h e1 (5)
s 0 (e2 s e1 )
Most cellular metals have a nearly flat plateau region and therefore a high energy-
absorbing efficiency depending on the used strain interval, and in general, a good
absorbing capacity (Fig. 5.1-7).
sure and gas flow inside the cells of a closed-cell foam. Both effects can increase
the flow stress and therefore also the plateau stress of the foam with increasing
strain rate [12]. In open-cell structures no significant increase of flow stress is ob-
served at higher strain rates, which suggests that the inertial masses of the struc-
tural elements have no contribution to the flow stress [12]. In closed-cell foams the
situation is different, because some of the metal foams show a strain-rate depen-
dence. Real closed-cell metal foams are, in general, not completely closed-cell struc-
tures because they frequently contain some initial defects in the cell walls arising
from the manufacturing process. Additional defects are generated during the defor-
mation process when cell walls rupture. The gas flow between the cells through
these defects is strain rate dependent. Thus the gas pressure inside the cells in-
creases during high-speed deformation and contributes to the flow stress [13].
Owing to the strain localization in deformation bands, the local strain rate is
much higher than the measured global one, which will increase the strain rate
effect. As a result, the plateau stress and the energy absorbing capacity of closed
cell metals foams ise [12,13]. Investigations on aluminum foams show an increase
in the plateau stress of 20 90 % when the strain rate increases from 10 3 to 103 s 1
(Fig. 5.1-8). The strain rate effect is mainly controlled by the cell morphology and
the alloy from which the foam is made.
Figure 5.1-8. Strain-rate dependence of the plastic strength of two types of aluminum foams [12].
5 Material Properties 193
5.1.4
Tension Behavior
where C is a constant, l is the cell size, a is the critical initial defect size, and s fs,s is
the fracture strength of the solid. For closed-cell metals the situation is more com-
plicated.
In ductile cellular metals (such as aluminum foams) cracking of cell walls and
cell edges can also be observed before a main crack starts to propagate. With in-
creasing deformation, a fracture-process zone starts to develop within which the
whole deformation is concentrated. In closed-cell metals, several cell walls rupture
inside this fracture-process zone. Further deformation leads to development and
propagation of a main crack along the weakest path in the cellular structure.
Only the strongest cell edges remain intact and hold the two foam parts together.
This process is accompanied by a decrease in flow stress, which is evident in the
stress strain curve depicted in Fig. 5.1-10. Figure 5.1-9 shows the corresponding
194 5.1 Mechanical Properties and Determination
Figure 5.1-9.Failure mechanism in aluminum foam in tension. Crack initiation in some weak cell
walls and subsequent development of a fracture-process zone (stress strain curve in Fig. 5.1-10).
Figure 5.1-10. Stress strain response in tension of aluminum foam with some unloading moduli
indicated (Alporas, 3r 0.18).
5 Material Properties 195
Figure 5.1-11. Influence of the notch depth ratio, a/W, on the stress strain response of a ductile
aluminum foam in tension.
196 5.1 Mechanical Properties and Determination
5.1.5
Torsion Behavior
In torsion the cell edges are subjected to bending and the cell walls to a combi-
nation of bending and shearing. The shear behavior is important for the develop-
ment of material laws and also for the application in sandwich structures, where
core shear may occur in particular load situations. Alternative methods for testing
the shear strength are, for example, the double-lap shear test or the ASTM C-273
Figure 5.1-12. Stress strain response of aluminum foam in torsion with no axial stress (black
line) and with tensile axial stress (gray line) [17].
5 Material Properties 197
5.1.6
Fracture Behavior
Fracture behavior and fracture toughness values may play a major role for the ap-
plication of cellular metals foams in load-bearing structural components. The frac-
ture of brittle foams, where linear elastic fracture mechanic (LEFM) can be applied,
were discussed briefly in the section on tensile behavior.
The situation in closed-cell ductile metal foams is more complicated, where
LEFM usually can not be applied, which is considered in this section.
Figure 5.1-13. Diagram of the load displacement curve and the corresponding SEM images to
demonstrate crack initiation and crack propagation in a ductile aluminum foam acquired from in-
situ fracture tests.
to obtain valid fracture toughness values should be thought about because, for
example, the required specimen size also depends on the cell size, l, of the foam.
For the determination of J Da curves according to ASTM E813 or ASTM E1152,
the following parameters must be measured: the force F, the load-line displace-
ment vLLD, and the crack extension Da. Figure 5.1-14 shows a load versus load-
line displacement curve obtained in a fracture mechanics test and the crack exten-
sion as a function of the load-line displacement of a ductile aluminum foam. The
calculated J Da curve is plotted in Fig. 5.1-15. The ASTM standard provides that
the initial J-value, J0.2, is obtained by the intersection of the blunting line at
0.2 mm crack extension with the straight line achieved from a linear regression
of the J Da curve in the Da interval 0.15 1.5 mm.
A crack extension of 0.2 mm is rather low for metal foams, because it is one
magnitude smaller than the usual cell size (the characteristic microstructural
length) and can be caused by other effects (microcracking, localized yielding)
rather than crack extension, depending on the measurement method used. Fleck
and Ashby [19] introduced a steady-state J-value, JSS, which is measured as the pla-
teau J in the J Da curve. They assumed that this JSS is a measure of the fracture
toughness for cracks, where the bridging zone is fully developed. However, not all
cellular metals show a typical plateau region in the J Da curve and the plateau is
reached at rather high crack extensions. The standardization of the evaluation of
initial, or steady state fracture toughness values from J Da curves is a problem
that has not yet been solved satisfactorily.
Gibson and Ashby [6] established the following relation between the fracture
toughness and the foam density for ductile metal foams
Figure 5.1-14. Load and crack extension versus load-line displacement curve for a ductile metal
foam (Alporas, rr 0.1).
200 5.1 Mechanical Properties and Determination
Figure 5.1-15. J-integral versus crack extension plot with fracture toughness values JSS and J0.2
indicated (J-values obtained from Fig. 5.1-14).
!3=2
r
JIC z Cs y; s l (7)
rs
This equation correlates the fracture toughness with the cell size (compare with
Eq. 8, it is assumed that COD l), which is demonstrated only for a limited num-
ber of foams and densities. Other fracture toughness values, like KIC, or CODC, can
be derived from this equation using the well-known fracture mechanics relations.
An alternative method for the determination of the fracture toughness of a duc-
tile material is the measurement of the critical crack opening displacement (COD).
Although this is a direct method, it is rarely used owing to the difficulties in
experimental determination. Schwalbe [20] suggested to measure the COD 5 mm
away from the crack tip, COD5, which can be more easily performed. We have
adapted this technique to foams. Figure 5.1-16 shows COD5 Da and CTOD Da
curves of a ductile aluminum foam (CTOD crack tip opening displacement).
The advantage of this method is the similarity of the COD Da curves for different
foams and the easier evaluation of critical fracture toughness values. The relation
between COD and the J-integral is given by the following equation
J k s COD (8)
where s * is the mean value between yield stress s y and ultimate tensile stress s UTS
and k is 1 for plain stress state and 1,3...2,6 for plain strain state for common
metals. In the case of metal foams the value of k for the (macro) plain strain
state depends on the Poisson's ratio of the foam in the plastic regime. It is im-
5 Material Properties 201
Figure 5.1-16. Crack opening displacement measured behind the crack tip and reaches a
(COD) in terms of COD5 and crack tip opening constant value after a certain crack extension,
displacement CTOD versus crack extension whereas COD5 is measured at a fixed position
curves for the same foam as shown in 5 mm behind the initial crack tip.
Fig. 5.1-14 and Fig. 5.1-15. The CTOD value is
portant to note that we have to distinguish the macro- and micro- stress states (the
stress state in the cell wall). Since npl is much smaller than in solid metals, k should
be only somewhat larger than 1.
References
5.2
Fatigue Properties and Endurance Limit of Aluminum Foams
5.2.1
Literature Survey of Endurance Data
quently, the load ratio R is 1 for fully reversed loading, 0 J R I 1 for cyclic ten-
sion loading and 1 J R J T for cyclic compression loading. The amplitude of a
stress cycle Ds/2 (s max s min)/2 is used to describe the cyclic stress range.
A summary of S N curves described in the literature is shown in Fig. 5.2-1.
The S N data are approximated by power-law curve-fits with a 50 % survival
probability.
Ds (1)
Nn C
2
The values of the exponent, n, and the coefficient, C, (in MPa) have been evaluated
and are shown in Table 5.2-1. Different symbols and lines refer to different mate-
rials and loading conditions (cyclic compression, cyclic tension, or fully reversed
loading). Fatigue properties of the different aluminum foams show significant var-
iations. The cyclic stress amplitudes that lead to mean lifetimes of 105 cycles, for
example, may vary by a factor of 200, depending on the investigated foam and the
load ratio.
The main reasons for the large variations of the fatigue properties of aluminum
foams are their different mass densities and structures. Literature data show a sig-
nificant trend, that fatigue properties of foams increase with increasing plateau
stress. As an example, the stress amplitude for 50 % survival probability of 105 cy-
cles and of 107 cycles, respectively, are shown in Fig. 5.2-2. Irrespective of foam
type and loading condition (cyclic tension, cyclic compression, or fully reversed),
the fatigue strength increases with the plateau stress. The linear relation shown
in Fig. 5.2-2 indicates, that the stress amplitude for mean lifetimes of 105 cycles
is approximately 0.35 of the plateau stress, and Ds/2s pl 0.27 leads to mean life-
times of approximately 107 cycles. This means that the fatigue lifetime of an alu-
minum foam is mainly determined by the cyclic stress amplitude, Ds/2 with re-
spect to the plateau stress, s pl, and Ds/2s pl is an appropriate measure of the mag-
nitude of cyclic loading.
To compare different aluminum foams it is reasonable to normalize the cyclic
properties with respect to the plateau stress s pl. Eq. (1) may be rewritten as follows
Ds C
Nn (2)
2s pl s pl
In Fig. 5.2-3, literature data obtained under cyclic compression, under cyclic ten-
sion, and under fully reversed loading conditions are summarized. Additionally,
mean S N curves for these three loading conditions are shown. The algebraic ex-
pression of the mean curves of the literature data is
Ds C
N nmean C0 (3)
2s pl s pl
The values of the exponent, nmean, and the coefficient, C0, of the mean curves are
shown in Table 5.2-2. The largest absolute value of the exponent nmean is found for
fully reversed loading and the smallest for cyclic compression loading.
Using the relative stress amplitude, Ds/2s pl to describe the magnitude of load-
ing, Duocel foam shows good fatigue properties under cyclic compression
Table 5.2-2. Coefficients for mean curves (Eq. 3) under different loading conditions.
a) b)
(Fig. 5.2-3a: #2) and cyclic tension (Fig. 5.2-3b: #12). Duocel foam was the only
open-cell foam where fatigue data were found in the literature. Similar relative fa-
tigue strength is visible for Alporas foam (Fig. 5.2-3a: #3 #7, Fig. 5.2-3b:
#13 #16), Alulight foam (Fig. 5.2-3a: #8,#10, Fig. 5.2-3b: #17, Fig. 5.2-3c: #20)
and IFAM foam (Fig. 5.2-3a: #9, Fig. 5.2-3c: #19). Hydro aluminum foam has
been tested under fully reversed loading conditions only (Fig. 5.2-3c: #18) and
shows the highest relative strength at R 1. Alcan foam, which is the material
with the lowest density, shows the lowest relative fatigue strength under cyclic
compression and cyclic tension (Fig. 5.2-3a: #1, Fig. 5.2-3b: #11).
5.2.2
High Cycle Fatigue Properties and Endurance Limit
To evaluate the high cycle fatigue properties and to determine an eventual fatigue
limit for aluminum foams, specimens have to be tested up to very high numbers of
cycles. These investigations are time consuming using conventional fatigue testing
equipment working at low frequencies, since several specimens have to be tested to
obtain statistically relevant data of this rather inhomogeneous material. Therefore,
the ultrasonic fatigue testing method has been used in this study to investigate the
high cycle regime in an efficient and time saving manner.
5.2.2.2 Results
Results of endurance tests with AlMg1Si0.6 foam are shown in Fig. 5.2-4a. Num-
bers of cycles to failure are related to the experimentally determined cyclic strain
amplitude, De/2. Assuming a mean Young's modulus of 3.9 GPa, the stress am-
plitude, Ds/2 can be calculated from the measured cyclic strain as shown in
Fig. 5.2-4a (right ordinate). Specimens that did not fail within 109 cycles (runouts)
are marked with an arrow. Fatigue data are subjected to pronounced scatter, which
is typical for inhomogeneous materials. The solid line indicates a mean fracture
probability of 50 % assuming power-law dependence between stress amplitudes
and cycles to failure.
The S N measurements of AlMg1Si0.6 foam indicate a well-defined fatigue
limit: Specimens either fail before about 107 cycles are reached or they survive
109 cycles or more. The median endurance limit (failure probability of 50 %) is in-
a) b)
c) d)
Figure 5.2-4. Results of endurance fatigue tests under fully reversed tension compression cy-
cling: a) AlMg1Si0.6 foam, b) AlMg1Si0.6/T5 foam, c) AlMg0.6Si0.3 foam, d) AlSi12 foam.
210 5.2 Fatigue Properties and Endurance Limit of Aluminum Foams
dicated with a line parallel to the abscissa in Fig. 5.2-4a at a stress amplitude of
1.3 MPa, which is approximately 23 % of the plateau stress. However, specimens
may fail at significantly lower cyclic stresses due to the statistical distribution of
defects and areas of stress concentrations. The highest stress amplitude, where
no specimen failed within 109 cycles, is 1 MPa.
Endurance tests on AlMg1Si0.6 T5 foam are summarized in Fig. 5.2-4b. The
heat treatment leads to longer mean lifetimes at all stress levels in comparison
with the not heat treated material. Additionally, the endurance limit increased.
The maximum stress amplitude, where no specimen failed, is 1.2 MPa, and the
median fatigue strength at 109 cycles is 1.4 MPa (approximately 20 % of the plateau
stress).
The results of endurance tests with AlMg0.6Si0.3 and AlSi12 foams are shown in
Fig. 5.2-4c and d, respectively. At a stress amplitude of 1 MPa, one (of five tested)
AlMg0.6Si0.3 foam specimens failed. The fatigue limit for 50 % survival is 1.1 MPa
(approximately 22 % of the plateau stress). The median fatigue strength at 109 cy-
cles of AlSi12 foam is 1.3 MPa (approximately 19 % of the plateau stress). Fatigue
data of AlSi12 foam are subjected to the most pronounced scatter among the inves-
tigated Alulight foams. At stress amplitudes of 1.2 MPa three specimens did not
fail within 109 cycles, whereas one specimen failed at 1 MPa.
5.2.3
Mechanism of Crack Initiation
Figure 5.2-5. Cell-wall defects in AlMg1Si0.6 foam introduced: a) during foaming, b) during
cooling.
SEM studies at higher magnification show, that the fatigue fracture surface is re-
latively smooth and transcrystalline. The general fracture appearance is ductile,
and striations are visible in some areas. Figure 5.2-7 shows both, a fatigue fracture
surface in a cell wall (Fig. 5.2-7a and b) and a fracture surface that has been formed
by static rupture of the specimen after fatigue testing (Fig. 5.2-7c). Fatigue cracks
show clear differences compared with cracks formed by static loading or during
cooling of the melt due to the far lower plastic deformation of the material around
the crack path. Fatigue crack paths preferentially follow areas where the cell wall is
thin. In Fig. 5.2-7, the thicknesses of cracked cell walls are approximately
30 60 mm, whereas the most frequent wall thickness obtained by image-analysis
is 250 500 mm [10]. SEM studies of the fracture surface do not show deformation
bands or plastic buckling as it could be found for uniaxial compression or for cyclic
compression fatigue loading.
The role of cell nodes, where fatigue cracks may be trapped, helps to understand
the influence of different foam structures on the fatigue properties. Duocel is an
212 5.2 Fatigue Properties and Endurance Limit of Aluminum Foams
Figure 5.2-6. Typical fatigue cracks: a) fatigue crack inside a cell wall, length about 300 mm,
b) crack starting from a cell-wall defect, length about 150 mm, c) fatigue crack causing rupture
of a specimen.
5 Material Properties 213
Figure 5.2-7. Fracture surfaces of AlMg1Si0.6 T5 (Ds/2 1.96 MPa, number of cycles to failure
8.3 q 104): a) fatigue fracture surface, b) striations indicating crack growth direction, c) fracture
surface formed by static rupture.
214 5.2 Fatigue Properties and Endurance Limit of Aluminum Foams
open-cell foam with r 0.15 g/cm3 and consists of cell nodes only and approxi-
mately the same amount of material is distributed to nodes and cell walls inside
Alcan foam (r 0.19 g/cm3). Fatigue cracks initiated in Alcan foam may find
lower growth resistance since cell nodes contain less material compared with Duo-
cel foam. This may be one reason, why Duocel foam (Fig. 5.2-3b, #11) shows far
better cyclic tension properties than Alcan foam (Fig. 5.2-3b, #12).
5.2.4
Summary
A survey of literature data (20 S N curves determined with six different aluminum
foams under cyclic compression, cyclic tension, and fully reversed loading) shows
that the fatigue strength is related to the plateau stress, s pl.
x The stress amplitude, Ds/2 for mean lifetimes of 105 cycles (107 cycles) is ap-
proximately 35 % (27 %) of the plateau stress.
x Numbers of cycles to failure and relative stress amplitudes, Ds/2s pl may be cor-
related by a power law function. The largest gradient is found for fully reversed
loading and the smallest for cyclic compression loading.
To investigate the high cycle fatigue properties of aluminum foams, three Alulight
foams have been investigated at numbers of cycles between 104 and 109. Cylindrical
rods with a continuous surface layer and a closed cell structure were tested. The
foam density is 0.56 g/cm3.
x The investigated Alulight foams show a fatigue limit, and cycles to failure above
107 are rare. The median endurance limits, Ds/2 are 1.1 1.4 MPa, which is
19 23 % of the plateau stress.
x Fatigue damage under fully reversed tension-compression loading is governed by
the formation of cracks. Preferential areas for crack initiation are initial defects
like precracks or holes in the interior sections of cell walls. No strain localization
or formation of deformation bands (as reported for cyclic compression tests) is
found.
x Fatigue cracks preferentially grow in the inner sections of cell walls where wall
thickness is small. Below the endurance limit eventual cracks are trapped at
the nodes of cells.
Acknowledgements
The authors acknowledge financial support by the ªFonds zur FoÈrderung der wis-
senschaftlichen Forschung (FWF)º, Vienna, (project no. P13230PHY) and the LKR-
Center of Competence on Light Metals, Ranshofen, Austria, for the grant through
the K-plus project and for supplying the Alulight foams.
5 Material Properties 215
References
5.3
Electrical, Thermal, and Acoustic Properties of Cellular Metals
It is evident that the presence of gas bubbles will significantly affect also electrical,
thermal, and acoustic properties of cellular solids. In the case of polymer foams
these properties play an important role: thermal insulation and noise attenuation
are beside packaging the main fields of application for this kind of materials. How-
ever, only a little attention is paid to the mentioned properties when cellular metals
are considered, although the multifunctionality is often the prerequisite for their
successful application. The aim of this chapter is therefore to reveal and critically
discuss the potential of cellular metals in this field with respect to published mod-
els and rarely reported experimental data. For better understanding of considered
properties, the basic physical background will be also briefly introduced.
5.3.1
Electrical Properties
Cellular metals are in distinction to polymer and most ceramic matrices electrically
conductive. This simple fact changes the typical applications drastically: ceramic
and polymeric foams can be used for insulation or for structural enclosures that
216 5.3 Electrical, Thermal, and Acoustic Properties of Cellular Metals
must transmit microwaves, while the metallic counterparts are good for the oppo-
site purpose. The electrical conductivity of foamed metals, although considerably
reduced by the porosity, is still sufficient enough to provide good electrical ground-
ing, low voltage contacts, or capability to absorb electromagnetic waves. In the case
of metallic sponges with open cell structure, the conductivity allows to distribute
the electric potential on large accessible area, which makes them attractive for bat-
tery electrodes. Cellular nickel is extensively used in this application [1].
It is generally accepted, that cellular material inherits electrical characteristics of
the solid of which it is made. The electrical conductivity s of material is defined
according to Ohm's law
l I
s (1)
S U
where U is a potential drop along the distance l when current I is applied. As can
be seen conductivity depends strongly on the cross-sectional area S of the sample.
In a case of cellular metals only a small portion of cross section is electrically con-
ductive. These are the cell walls with roughly the conductivity of the base alloy. The
main part of the cross section is created by pores, which are usually filled with di-
electric air. The cell walls form a complex 3D network of electric conductors of
varying cross section, whereas cell walls oriented normally to the potential gradient
do not contribute to conduction because of no potential difference. Moreover, the
cells are mostly continuously covered with electrically nonconductive metal oxides.
The cellular structure provides therefore much lower electrical conductivity than
bulk material.
To calculate the electrical conductivity according to Eq. (1) the effective length l of
the cell walls conducting along the potential difference and in the cross-sectional
area S effectively contributing to conduction has to be known. Unfortunately, it
is extremely difficult to determine it in real samples, although for example com-
puted tomography offers this possibility. More reasonable and cheaper is to con-
sider the sample as a black box of given outer geometry with macroscopically
homogeneous and continuos distribution of a material within the box and to inves-
tigate the so called effective electrical conductivity.
It is evident that the dependence of effective cross-sectional area on relative
density of the structure is non linear. One of the first models of this dependence
was introduced by Ashby et al. for the case of open-cell sponge [2]. The structure
was idealized with narrow struts of given length and thickness. The struts were
arranged in a perfect cubic lattice, with nodes linked by struts. Only struts oriented
parallel to potential gradient contributed to conduction. For higher relative den-
sity they added also the contribution of the nodes and obtained the following
relation
! !3=2
s r r
f S(1 s f) (2)
s0 r0 r0
5 Material Properties 217
where s and r are the effective property and density of the foam respectively, while
s 0 and r0 are the corresponding properties of the cell-wall material. The first term
of this equation describes the contribution of struts and the second one that of
nodes. Therefore, a structural parameter f was introduced, which characterizes
how the material is distributed between cell nodes and struts. However, in real
sponges the ratio between volume of cell nodes and cell struts is not constant
and varies also with the density. Thus f in Eq. (2) must be obtained experimentally
for each material and is valid only for certain density interval.
Another approach for modeling of the relationship between relative density and
selected physical property uses the principles of percolation theory [3 6] and
covers the whole density range. In this case the cellular metal is considered as
an ªinfiniteº cluster of randomly distributed gas pores in the metallic matrix.
According to percolation theory the effective property near the connectivity
threshold (in this case set to zero) behaves like a power-law function of a relative
density
!t
s r
z (3)
s0 r0
where s and r are the effective property and density of the foam respectively, while
s 0 and r0 are the corresponding properties of the cell-wall material. Constant t
(often called as a critical exponent) can be theoretically predicted for the specific
property. The theoretically estimated value for electrical conductivity in 3D is t
2.00 [7]. The constant z ought to be 1, because for bulk material r r0 and the
effective property s s 0.
Also this approach has some difficulties: It does not take into account defects
such as cracks in pore walls, which influence the conductivity but not the relative
density. Moreover, the theoretical predictions of critical exponents in 3D are still
under development and optimization, especially when finite size clusters or aniso-
tropic orientations are considered.
Only little data for electrical conductivity of cellular metals have been reported
[8 12] up to now (see also Table 5.3-1). Therefore, a serious verification of recent
theoretical models is very difficult, if possible at all.
f 0.33 and f 0.05 were obtained for ERG open-cell sponge and for Alulight
closed-cell foam, respectively, applying Eq. (2) to fit the reported data [2]. The con-
siderable difference in a, confirms an assumption that a is strongly material and
structure dependent and for each case has to be obtained experimentally.
The power law seems to be a good characterization of the relationship between
experimental data of electrical conductivity and density measured on Alulight sam-
ples [3 5] (see Fig. 5.3-1 and Table 5.3-2). The characteristic exponents obtained
experimentally have been found slightly lower than the critical exponent value the-
oretically predicted by the percolation theory. It can be explained by the fact that in
distinction to the theory, where an infinite cluster is taken into account, the real
experimental samples have always a finite size and in the referred case possess
a surface skin. The volume fraction of the surface skin increases with decreasing
218 5.3 Electrical, Thermal, and Acoustic Properties of Cellular Metals
Table 5.3-1.Overview of the electrical and thermal conductivity of cellular aluminum with various
density from suppliers data sheets [8 12] measured at room temperature.
400 ± 5.0
300 ± 3.8
200 ± 2.6
94 ± 1.0
400 10 2.6
410 0.7 ±
270 0.4 ±
160 0.36 ±
68 0.28 ±
250 ± 1.6
200 ± 1.2
140 ± 0.9
size of the sample. The high portion of surface skin can diminish the effect of the
porosity on the reduction of electrical conductivity of the foam. If the surface skin
is removed from the sample the reduction of the conductivity due to increasing
porosity is enhanced and characteristic exponent becomes higher. The similar ef-
fect can be observed when the sample size increases. It implies that the conductiv-
ity of the foam is not only a function of the porosity, but it is strongly affected also
by the shape and size of the foamed part.
5 Material Properties 219
Table 5.3-2. The obtained fitting parameters for the dependence of normalized electrical
conductivity on the relative density for Alulight Al 99.7. x 2 is a minimization function [7].
z t x2
5
s/0 (mix.) [4] 0.999 e 0.009 1.55 e 0.02 7.31 qx 10
4
s/s0 (Al 99.7) [4] 0.979 e 0.025 1.75 e 0.06 6.54 qx 10
Although the behavior can be qualitatively understood by noting that with de-
creasing density the cross section available for conduction decreases and the tortu-
osity of the current path increases, a quantitative theory is still lacking [13].
Good electrical conductivity of cellular metals minimizes the penetration of elec-
tromagnetic waves into the structure, especially in the case of closed-cell foam
parts covered by a surface skin. This fact can be successfully utilized for the pro-
tection of electronic devices or humans from the effect of electromagnetic noise.
The ability for electromagnetic wave shielding can be defined by shield effective-
ness depending on the wave frequency and the material thickness. The main
advantage of metallic foam is a possibility to achieve a thickness required for
low magnetic permeability at lower weight than in the case of bulk metals. The
experimental measurements on closed cell Alulight and Si-steel sheet of the
same weight are illustrated in Fig. 5.3-2 and Fig. 5.3-3 [9]. The shielding effectivity
of the foam is even slightly better than that of a bulk aluminum plate of the same
thickness, additionally the foam weight was approximately five times lower in this
case (see Fig. 5.3-2). Cellular aluminum foam exhibits much better magnetic field
Figure 5.3-1. The power-law dependence of normalized electrical conductivity on the relative
density for Alulight Al 99.7 foam (cylindrical samples with surface skin, s 0 37.6 q 106 S m 1,
r0 2700 kg m 3) [4], for the fit parameter see Table 5.3-2.
220 5.3 Electrical, Thermal, and Acoustic Properties of Cellular Metals
Figure 5.3-2. Magnetic field shielding effectiveness as a function of frequency for cellular
aluminum (Alulight) and Si-steel samples of the same weight and a massive Al sheet of the same
thickness (t 8.5 mm) as the aluminum foam. Sample size 140 q 140 q t, density of foam
500 kg m 3 (KEC-method) [9].
Figure 5.3-3. Electric field shielding effectiveness as a function of frequency for cellular alumi-
num (Alulight) and Si-steel samples of the same weight. Sample size 140 q 140 q t, the density
of foam 500 kg m 3 with t 8.5 mm (KEC-method) [9].
shielding properties than Si-steel sheet of the same weight. The electrical shielding
effectivity of Alulight is comparable with the properties of Si steel (see Fig. 5.3-3)
up to 10 MHz. For higher frequencies the aluminum foam is superior. Similar
results have been reported for Alporas foam [10]
5 Material Properties 221
5.3.2
Thermal Properties
The main thermal properties of practical importance with respect to cellular metals
are the melting point, specific heat, coefficient of thermal expansion, thermal con-
ductivity, thermal diffusivity, emissivity of the surface, fire and thermal shock resis-
tance. Some of these properties are almost the same as for bulk materials, other
ones depend strongly on the porous structure.
Melting point of cellular metals is practically the same as for an alloy from which
it is made. However, the surface is often covered by a continuous oxide layer, the
melting point of which is usually considerably higher: for example in the case of
cellular aluminum. The surface area covered with oxide layer increases with in-
creasing porosity and decreasing pore size. When the oxide layer is sufficiently
thick it can support the porous structure even above the melting point of the
alloy, provided that no significant external force is applied on the structure. If
the structure is heated in air or oxidizing atmosphere at the temperature close
to melting point for sufficiently long time, the thickness of oxide layer increases
and material becomes more stable. After a threshold thickness depending on the
mechanical stability of the sample, the metallic cell walls oxidize completely by
subsequent heating converting the cellular metal into a ceramic foam. This behav-
ior is utilized for instance for the preparation of ceramic sponges from open-cell
metallic precursors like Duocell sponges [14].
The energy needed to increase the temperature of cellular structure by a unit
temperature is almost the same as required for cell-wall materials when the unit
mass is considered. The slight difference can be accounted to the presence of
thin surface oxides and the contained air. However, the specific heat per unit volume
Cv of a cellular structures is of cause significantly lower. It makes them attractive
for applications where low thermal capacity is required, for example for rapid heat-
ing and cooling systems.
The coefficient of thermal expansion a of cellular metals is also almost the same as
for the cell-wall material, while the thermal conductivity l is much lower. Higher
thermally induced distortion has to be expected that increases with raising a/l ratio.
To avoid it, the temperature differences in the cellular structure on heating or cool-
ing should be minimized.
On the another hand if the temperature shock resistance is considered, the situa-
tion is not so unambiguous. When a sudden temperature difference occurs be-
tween outer surface and inner material regions the thermal expansion causes ther-
mally induced strains, the related elastic stress depend on the modulus of elasticity
of the material. The stresses depend on the product a E and as the modulus of elas-
ticity of cellular metal decreases with decreasing relative density the same tempera-
ture difference induces smaller thermal stresses the higher the porosity of the cel-
lular structure, assuming the coefficients of thermal expansion is constant. How-
ever, with decreasing density also the collapse or tensile stress of the material de-
creases. If this stress scales with relative density at smaller exponent than the mod-
ulus of elasticity does, the temperature shock resistance of the foam will improve.
222 5.3 Electrical, Thermal, and Acoustic Properties of Cellular Metals
If not, the foam may yield or even break earlier than bulk material. In distinction to
bulk materials, cellular metals are able to withstand much higher compressive
strains without fracture because of the existence of a plateau stress. Therefore
their temperature shock resistance is usually considered to be higher than that
of bulk metals. It should be noted, that the same sudden change of ambient tem-
perature causes higher temperature difference within a foam than within solid
metal because of significantly smaller thermal conductivity of the cellular metal.
It is evident that the thermal conductivity, l, of cellular metal will be significantly
lower than the corresponding conductivity of bulk material. In comparison with
electrical conductivity this property is more complex. Apparent thermal conductiv-
ity of porous structure has four contributions [13]: besides the thermal conductivity
of the solid cell walls also the conductivity of enclosed gas, convection and radiation
effects should be taken into account
l lS S lG S lc S lr (4)
where indexes S, G, c, and r denote the contribution to thermal conductivity of the
solid cell wall material, the gas within pores, convection within the cells, and
radiation through the cell walls. The thermal conduction is additionally affected
by the presence of low conductive metal oxides on the surface of pore walls and
cracks.
However, in a case of cellular metals only the first term in Eq. (4) is important,
the last three are practically negligible of moderate temperature levels. The contri-
bution of conduction through gas lG is very low in comparison with the conduction
through metal, because of low thermal conductivity of gas (thermal conductivity of
air is 0.025 W m 1 K 1). Convection within the cells is important only when the
Grashof number (the ratio between force driving convection and opposing viscous
force) is greater than about 1000 [13]. This value is achieved with air filled pores at
room temperature, when the pore diameter exceeds 10 mm. However, the pore size
in real metallic foams usually is smaller, so the convection should be negligible in
this case.
Radiation through the cell walls is not possible in a case of optically nontranspar-
ent metals. Radiation within the cells can be ignored as well, when the conduction
of the cell-wall material is greater than 20 W m 1 K 1 [15].
The contribution of the metal to the thermal conductivity of cellular structure (ls)
depends on relative density r/r0 and similarly as for electrical conductivity it can be
expressed by power law [3 6]
!t
r
lS l0 (5)
r0
where l0 denotes the thermal conductivity of solid metal. Constant t is the critical
exponent for thermal conductivity and according to percolation theory its theoreti-
cally predicted value in 3D is t 2.0 [7]. The characteristic exponents obtained
experimentally are listed in Table 5.3-3. The power-law dependence is illustrated
in Fig. 5.3-4.
5 Material Properties 223
Table 5.3-3. The obtained fitting parameters for the dependence of normalized thermal conduc-
tivity on the relative density for Alulight foam at various temperature. x 2 is a minimization function.
z t x2
5
l/l0 (20 hC) [5] 1.000 e 0.006 1.60 e 0.03 2.97 qx 10
5
l/l0 (100 hC) [5] 1.000 e 0.008 1.55 e 0.03 5.66 qx 10
5
l/l0 (300 hC) [5] 1.000 e 0.010 1.48 e 0.04 9.10 qx 10
5
l/l0 (400 hC) [5] 1.000 e 0.005 1.51 e 0.03 4.00 qx 10
Figure 5.3-4. The power-law dependence of normalized thermal conductivity on the relative
density for Alulight foam at 20 hC. The corresponding fit parameters are given in Table 5.3-3
(cylindrical samples with surface skin, l0 232 W m 1 K 1, r0 2700 kg m 3) [3].
224 5.3 Electrical, Thermal, and Acoustic Properties of Cellular Metals
Figure 5.3-5. Thermal conductivity of Alulight (ls 200 W m 1 K 1); v, AlMgSi0.6 base alloy,
foams made of various alloys as a function of r 500 kg m 3, V, AlSi12 base alloy,
testing temperature and density r: y, Al 99.6 r 440 kg m 3 (ls 120 W m 1 K 1), x, AlSi12
base alloy, r 320 kg m 3 (ls 230 W m 1 K 1); base alloy, r 330 kg m 3
Y, AlMgSi0.6 base alloy, r 420 kg m 3
Thermal diffusivity a governs heat flow through a material before a steady state
flow by thermal conduction is achieved. It is defined as a ratio l/Cv between ther-
mal conductivity l and volumetric specific heat Cv. Foams have usually higher ther-
mal diffusivity than bulk cell-wall materials and thus they achieve a steady state heat
flow earlier. This property makes the cellular metals attractive for use as heat ex-
changers.
The emissivity of the foam surface is almost always higher than that one of the
metal from which it is made of. As the actual surface area of the foam is also
much larger at equal outer dimensions the foam will emit or absorb more thermal
radiation than bulk material. This property in combination with small specific heat
per unit volume can be utilized in applications where rapid heating or cooling by
radiation is required.
Although polymer and ceramic foams are used more for thermal insulation than
for any other purpose, application of cellular metals in this field is unlike because
their thermal conductivity l is about two orders of magnitude higher than poly-
mers (see Table 5.3-1). However, higher thermal conductivity and diffusivity of cel-
lular metals can be successfully utilized in heat exchangers or heat sinks for power
electronics, air-cooled condenser towers and regenerators. This application is re-
stricted only to open cell structures enabling flow of cooling gas or fluid without
significant pressure drop. The main design principles for heat exchangers made
from cellular metals are [2]: high conductivity of metal (preferably Cu, Al), turbu-
lent flow of fluid to enhance local heat transfer; and minimized low-pressure drop
between fluid inlet and outlet. For any system there is a trade-off between heat flux
and pressure drop. Optimum can be achieved with cell sizes in the mesorange and
with relative densities of order r 0.2. [16]
5 Material Properties 225
Bi hL=l (6)
where h is heat transfer coefficient, L is the thickness, and l is the thermal conduc-
tivity of the panel. For given Biot number the fire resistance is infinite if the im-
pose fire temperature is smaller than a threshold value TF
TF TC S Bi(TC s T0 ) (7)
where TC and T0 are critical and initial temperatures of the panel, respectively. The
Biot number for a panel made of metallic foam is always higher than that one for a
corresponding metal sheet of the same weight. For example, it takes 8 min to heat
up the opposite side of a Cymat sheet (density 160 kg m 3, geometry 24 cm q
30 cm q 5.5 cm) from room temperature up to 500 hC using oxy/acetylene torch
(theoretical temperature 3100 hC) rotating at 15 cm distance from the foam front
side. The burn-through time for corresponding aluminium sheet of the same
weight (density 2700 kg m 3, geometry 24 cm q 30 cm q 0.34 cm) is only 23 s
[11]. This illustrates the opportunities of metallic foams for this application.
5.3.3
Acoustic Properties
For this reason the sound travels at velocity of about 5000 m/s in solid steel or alu-
minum, and only 343 m/s in air at the sea level. The wave velocity of sound is
related to the wavelength l and frequency f of sound wave by c lf. The normal
human ear responds to the frequencies from about 20 Hz up to 20 kHz, which cor-
responds to wavelengths of 17 m to 17 mm. For the noise control and reduction it
is often enough to reduce sound pressure mainly in the range of 500 4000 Hz.
The amplitude of sound wave determines the loudness of sound. The amplitude
of sound pressure is measured in Pascals (Pa), but it is more convenient to use
the logarithmic scale in decibels (dB) [2].
226 5.3 Electrical, Thermal, and Acoustic Properties of Cellular Metals
In general, the three basic cases can be distinguished with respect to noise at-
tenuation materials:
x materials for sound insulation: they keep airborne sound out of the protected
space,
x materials for sound absorption: they reduce or fully absorb the sound intensity
within the room where the sound is generated,
x materials for damping: they reduce the amplitude of the structure acoustic re-
sponse and hence the sound intensity born by the structure.
In the following the physical principles that determine the potential use of cellular
metals for sound management purposes will be briefly reviewed considering all
three mentioned cases. Guidelines for material selection and procedures ensuring
successful practical applications will be suggested.
aw
sl
Resonance controlled
s
Mass controlled
ma
Stiffness controlled
f
iono
te ns
Sound reduction index [dB]
Ex
Coincidence dip
Effect of damping
Critical Frequency Hz
frequency
Figure 5.3-6. Principal dependence of sound reduction index of a panel as a function of
frequency for given material and structure geometry [17].
5 Material Properties 227
At very low frequencies the sound reduction is controlled by the stiffness of the
panel. In this case the application of cellular metals can be very attractive because
of their excellent stiffness to weight ratio (see parameter E/r3 in Table 5.3-4).
At somewhat higher frequencies the sound transmission is determined by the
natural resonance of the panel depending on the used material, panel size, and
mounting conditions. Also in this frequency range the cellular metals can be use-
ful, because of their better damping properties in comparison with bulk metals
(see also later in this Section). Moreover, they can shift the first resonant frequency
towards higher values and thus extend the stiffness controlled frequency range (see
Table 5.3-4). This can be important when the panels of smaller size are used.
However, for typical sound insulation, the most important is the frequency
range, where the sound reduction is controlled by mass of the panel according to
where TL0 in dB is the sound reduction index for normal incidence, m is the super-
ficial density in kg m 2 of the soft impervious panel and f is the frequency in Hz.
This is known as Mass Law [17]: the sound transmission loss increases with in-
creasing mass of the panel. Therefore, the application of cellular metals for
sound insulation in this frequency range is highly disadvantageous.
Table 5.3-4. Design parameters for the materials suitable for sound insulation. Thickness h, 1st
resonant frequency f1 of fully clamped plate and critical frequency fc are calculated for the square
plate of given weight (10 kg) and area of 1 m q 1 m. Parameter E/33 demonstrates the bending
stiffness at minimum weight of the panel. AlSi12 aluminum foam of density 500 kg m 3 and
modulus of elasticity 5 GPa was introduced for comparison.
* 1st resonant frequency calculated for the panel size of 0.5 m q 0.5 m q h m in order to demon-
strate the size effect
228 5.3 Electrical, Thermal, and Acoustic Properties of Cellular Metals
Beside bulk absorbers two types of resonant absorbers are in general use, espe-
cially for sound absorption at lower frequencies. The panel absorber is a resonant
system formed by the mass spring combination of the facing panel and the stiff-
ness of enclosed air. The resonant frequency of panel absorber at which the max-
imum absorption is reached is given by
600
f p (11)
md
where m is the superficial density of the panel in kg m 2 and d is the depth of air
gap in cm.
The Helmholtz resonator or cavity resonator is essentially a vessel in which the
mass of air in the neck is driven in and out in resonance upon the stiffness of en-
closed air volume. The resonant frequency is given by
r
S
f 55 (12)
LV
where S is the cross-sectional area of the neck, L is the length of the neck, and V is
the enclosed volume of air.
The perforated panel often used for protection of porous absorbers can provide
an additional resonance effect, similarly, as it is in a case of Helmholtz resonator.
Such semiresonant performance is found when the percentage of open area of
front panel is below 20 30 %. Above this value, the system behaves primarily as
a simple porous absorber [17]. The faced material itself can be constructed more
sophisticated containing a plenty of small Helmholtz resonators (see Fig. 5.3-7).
The frequency at which the maximum sound absorption is reached can be
tuned by cavity geometry. Such combination usually shifts the maximum absorp-
tion performance towards low frequencies, preferably in the range 100 1000 Hz.
Figure 5.3-7. Typical construction of sound absorber: thick glass fiber absorber (1), thin pro-
tective and damping foil (2), facing wood panel with resonator cavity (3), and outer design (4)
[18].
230 5.3 Electrical, Thermal, and Acoustic Properties of Cellular Metals
Accordingly, the main parameter for good sound absorption is the permeability
of the absorber. Therefore, cellular metals can be used for this purpose only if they
meet this fundamental requirement. The closed-cell metallic foams are too stiff to
convert sound energy into heat by vibration of their cell walls. According to Eq. (11),
they cannot be effectively applied for panel resonators because of their very low
weight.
Unfortunately, the experimental measurements performed in impedance tube
on various types of cellular metals [19,20] are in a good coincidence with these dis-
couraging expectations (see Fig. 5.3-8). The best sound absorbing performance was
obtained using Access-sponge metals having open-cell structure and thus very low
flow resistance [21]. Such structure absorbs the sound energy mainly by viscous
losses as the sound pressure wave pumps air in and out of the interconnected
pores. Alporas closed cell foam (made by melt-route process) [22] is manufactured
in large blocks from which the absorbers are machined. Machining results in
rough open-pore surface that slightly improves the sound absorption. Alulight
closed-cell aluminum foam [9] is prepared in net shape and therefore is covered
with a surface skin. This relatively flat and pore free skin reduces the sound ab-
sorption performance almost to zero: the sound is entirely reflected.
Figure 5.3-8. Sound absorption coefficient of various types of cellular aluminum foams in
comparison with fiberglass sound absorbing material (directly on rigid background).
Sample Access sponge Alporas Alulight Fiberglass
Thickness [mm] 20 10 9 48
Access material has a good potential to be applied for sound absorption purposes
after optimization of the structure. In a case demonstrated in Fig. 5.3-8 the flow
resistance was too low to provide sufficient viscous losses. However, the manufac-
turing process allows to adjust the optimum pore size or even to obtain gradient
pore structure in relatively easy and inexpensive way [21]. Figure 5.3-9 demon-
strates the effect of the reduction of open area at the surface on the sound absorp-
tion performance of this material.
Although the sound absorption performance of cellular metals is not very im-
pressive, it can be significantly improved by optimal opening of close cell structure.
In this case the foams can serve as multiple cavity resonator (see Eq. 12) with
much longer ªneckº than in the case of bulk materials of the same weight. The
variability of the pore size will provide a wider frequency range of good sound ab-
sorption than typical resonators. Various methods, such as cutting of surface skin,
sand blasting, compressing, rolling, and drilling of holes were examined to open
the foam structure [19,20]. It was shown that sound absorption of as received Al-
poras aluminum foam can be significantly improved by partial compression, roll-
ing, or hole drilling. However, the combination of these method offers only a little
additional benefits [19].
Similar results were obtained for Alulight foams covered with surface skin (see
Fig. 5.3-10). The significant improvement of the sound absorption was reached al-
ready after opening of only 1.4 % of the surface area by drilling of holes. The open-
ing of the surface area can be more economically realized by simple punching of
the foam surface without the need to make the throughgoing holes (Fig. 5.3-11).
Figure 5.3-9. Sound absorption coefficient of Access aluminum sponge without and with
perforated facing Al sheet (air gap of 48 mm behind the absorber, Access sponge: density
1180 kg m 3, thickness 20 mm; facing Al sheet: thickness 1 mm, 61 holes diameter 2 mm).
232 5.3 Electrical, Thermal, and Acoustic Properties of Cellular Metals
Figure 5.3-10. Sound absorption coefficient as a function of opened surface area for Alulight
foam (8.9 mm thick AlSi12 foam, density 580 kg m 3, holes drilled through the sample, 20 mm air
gap between sample and rigid wall).
Surface opening 0.07 % 0.19 % 0.61 % 1.40 % 2.49 %
Number of holes 7 19 61 61 61
As expected [23], the increasing thickness of absorber shifts the maximum of the
sound absorption coefficient towards lower frequencies (see Fig. 5.3-12), while the
shape of the curves remains almost unchanged provided, that the density and pore
size do not depend on thickness. At constant sample thickness, the foams with low
density were found to be better sound absorbers than the foams with high densi-
ties in the frequency range of 200 1700 Hz [19], however the obtained results are
affected by an increasing pore size with decreasing density. Another experimental
measurement [24] indicates that there exists an optimal pore size for a sound ab-
sorption with respect to sample thickness at constant density. This experimental
finding agrees with theoretical calculations made by Wang and Lu [25] for 2D cel-
lular solids of Voronoi structure: the sound absorption performance will be im-
proved with increasing pore size for sample, which is infinitely thick. It implies
that the sound absorption coefficient is a complex function of foam thickness, den-
sity, and pore size.
It is not always feasible to increase the thickness of absorber, because of the
weight limitations. The use of a thinner section of absorber spaced by an air gap
gives very similar results to a thick section (see Fig. 5.3-13), however at signifi-
cantly lower weight. Therefore, the frequency range of maximum sound absorption
5 Material Properties 233
Figure 5.3-11. Sound absorption coefficient as a function of opened surface area for Alulight
foam (AlSi12 foam 15 mm thick, density 380 kg m 3, diameter of holes 1 mm, holes were punched
into the depth of 8 mm, directly on rigid background).
Figure 5.3-12. Sound absorption coefficient as a function of absorber thickness (Access AlSi
sponge, average pore size 4 5 mm, density 1000 kg m 3, directly on rigid background).
234 5.3 Electrical, Thermal, and Acoustic Properties of Cellular Metals
Figure 5.3-13. Comparison of sound absorp- is thicker foam at zero distance from rigid wall
tion for two designs of sound absorbing panels (Access AlSi sponge, density 1000 kg m 3,
at two constant thicknesses: 1st type is the average pore size 4 5 mm).
foam at given distance from rigid wall, 2nd type
can be tailored by creating an optimum air gap between the absorber and rigid wall
[19,20,24] without need to increase weight. The sound absorption can also be en-
hanced for wider frequency range by the combination of several foam plates
with an air gap between them [20].
Although the cellular metals cannot provide a high level of sound absorption
itself, they can be attractively used as a facing materials, especially in combination
with fibrous absorbers, provided, that the structure is sufficiently opened. The fre-
quency corresponding to the maximal sound absorption can be tuned by the thick-
ness of foam panel or by changing of pore size. The sound absorbing performance
is then comparable with typical solutions using facing wood panels (see Fig. 5.3-14).
The cellular metals offer additionally very high stiffness at low weight, are self-
supporting and stable at elevated temperature, does not release toxic gases when
subjected to excessive heat, possess high durability under the effects of air flow
or vibration, exhibit sufficient fire resistance and can be recycled. Moreover, as a
relatively new material with impressive surface design, it can be very attractive
for architects and designers of building interiors (see Table 5.3-5).
The potential applications are expected in sterile (antiseptic, dust free) environ-
ment, in the structures where nonflammability is important (airplanes, hotels,
commercial, and industrial buildings) or in interiors under difficult conditions (ele-
vated temperature, moisture, dust, flowing gas, and vibrations).
5 Material Properties 235
Figure 5.3-14. Sound absorption coefficient of fiber glass itself and faced with AlSi12 Alulight
(thickness 13.2 mm, density 330 kg m 3, diameter of holes 2 mm) with Helmholtz resonators
inside (cavity diameter 6.5 mm, length 8 mm). For comparison the sound absorption charac-
teristic of facing sheet of Topperfo acoustic panels TP 16/16/12-2 [18] is presented.
Structure
236 5.3 Electrical, Thermal, and Acoustic Properties of Cellular Metals
h Df =fn (13)
The second ªdecay-rateº method is used for materials with lower damping. The
loss factor is calculated from the vibration amplitude decay D in dB s 1 at resonant
frequency after instant stopping of exciting force, according to
The typical structural materials with high strength (for example steel, cast iron,
aluminum alloys) have unfortunately very little damping. On the another hand,
the high damping materials, such as lead, rubber, or soft plastic, have little
strength and cannot be regarded as structural materials [17].
The cellular metals exhibit at least one order of magnitude higher values of the
loss factor h [20]. The dissipation of the vibrations mainly results from the friction
between the attaching surfaces of the cracks appearing in the structure and
partially due to the vibration of the thin pore walls. Thus, the damping can be
enhanced by the reduction of the cell-wall thickness or/and by introducing of
imperfections into the structure. Higher loss factor values are therefore obtained
with for example foams made of casting aluminum alloys, which can be prepared
with very thin cell walls and contain a lot of cracks. If the nonsoluble ceramic
particles, for example SiC, Al2O3, or graphite are additionally introduced into the
structure of metallic cell walls, the damping will be further improved. New sliding
surfaces born in the cell walls in this way are responsible for this behavior (see
5 Material Properties 237
Figure 5.3-15. Effect of addition of nonsoluble particles on the damping properties of aluminum
foams and steel tubes filled with them (30 mm diameter q 300 mm, foam density 600 kg m 3).
Fig. 5.3-15). Nevertheless, the loss factor of typical cellular metals is still too low
compared with standard damping materials having h in the range 0.01 0.1.
Recently, there is some effort to apply the cellular metals as stiffeners or cores in
metallic hollow parts, castings, or sandwiches. Beside the stiffening and the in-
creased capability to absorb impact energy, also the reduction of noise and vibra-
tions is expected (see Fig. 5.3-16) [27]. In this case the structural damping is sig-
nificantly affected by the quality of contact between cellular metal and bulk metallic
shell. If the attached surfaces can slide on each other under pressure an additional
interface damping will be obtained. The damping effect is produced by dry friction,
which can provide very effective attenuation of excessive vibrations.
Figure 5.3-16. Frequency spectra of reciprocal mass for a hollow profile and a profile foamed
with various types of Alulight foams (steel tube 22 mm diameter q 245 mm filled with 18 mm
diameter q 245 mm foams of density 700 kg m 3).
238 5.3 Electrical, Thermal, and Acoustic Properties of Cellular Metals
Figure 5.3-17. Vibrational damping in complex epoxy (foam density approximately 700 kg m 3).
aluminum casting and its variations due to the For comparison the damping solution with
presence of Alulight foam insert injected or bitumen layer (3 mm thick) is also plotted.
fastened with either elastic adhesive or cured
Figure 5.3-18. Effect of the foaming on the values of resonant frequency and damping charac-
teristics of hollow steel profile.
5 Material Properties 239
Figure 5.3-19. Dependence of the first resonant frequency on the apparent density of Alulight
foams: longitudinal vibrations of AlSi12 foam 25 mm diameter q 300 mm (left axis) transverse
vibrations of Al 99.7 foam 140 mm q 140 mm q 8.5 mm (right axis).
Figure 5.3-20. Property gain obtained by use of material was used (weight 390 kg, self deflec-
aluminium foam for stiffening of hollow ma- tion 4 mm, first resonant frequency 240 Hz,
chine table (800 mm q 720 mm q 250 mm) damping at first resonant frequency 0.15 %)
[28]. As a reference the table made of bulk
240 5.3 Electrical, Thermal, and Acoustic Properties of Cellular Metals
Beside mentioned approaches there is also a possibility to avoid the effect of re-
sonant vibrations by stiffening of the structure. The stiffening arrangements do not
damp the resonances, they merely shift them towards higher frequencies. If the
resonances can be shifted to frequencies that will not be excited during normal op-
eration of the equipment this solution to the problem of reducing vibrations may
be very efficient.
The resonant frequencies generally depend on the modulus of elasticity E and
the density r of the material of which the component is made and, of course, on
its geometry. For example, the first resonant frequency f1 of longitudinal vibration
of cantilever is
s
3:52 EI
f1 (15)
2p rSL4
where I is area moment of inertia of beam cross section S and L is the length of the
beam.
To shift a frequency either the geometry or material ought to be altered. The
changes of the part's geometry are often not possible. The altering of the mate-
rial may not be always successful at least in the case of metallic parts, because
E/r ratio of many metals is almost the same. Here the cellular structure can be
very attractive, while both the modulus of elasticity as well as the density can be
varied without changing the outer dimensions of the component (see Fig. 5.3-19).
This feature can be very effectively utilized for shifting the resonant frequencies in
hollow machine components by filling them with cellular metals. Figure 5.3-20
shows the effect of the utilization of aluminum foam as a filler in hollow machine
table. Beside stiffening and improving of the damping a significant shift of the first
resonant frequency towards higher values has been achieved [28].
References
12. Norsk Hydro, Data Sheets, Hydro Alumi- 21. F. Grote, P. Busse, Giesserai 1999, 10,
nium Ltd., Sunndalsora, Norway, 1995. 75 78.
13. L. J. Gibson, M. F. Ashby, Cellular Solids, 22. T. Miyoshi, M. Itoh, S. Akiyama, A. Kita-
Pergamon Press, Oxford, 1988. hara, in Metal Foams and Porous Metal Struc-
14. E. S. Park, S. D. Poste, US Patent 4 808 558, tures, J. Banhart, M. F. Ashby, N. A. Fleck
1989. (eds), MIT Verlag, Bremen 1999, p. 125.
15. T. J. Lu, C. Chen, in Metal Foams and Porous 23. H. Utsumo, T. Tanaka, T. Fujikawa,
Metal Structures, J. Banhart, M. F. Ashby, J. Acoust. Soc. Am. 1989, 86, 637.
N. A. Fleck (eds), MIT Verlag, Bremen 1999, 24. F. Chen, D. P. He, in Metal Foams and
p. 391. Porous Metal Structures, J. Banhart, M. F.
16. A. G. Evans, J. W. Hutchinson, in Metal Ashby, N. A. Fleck (eds), MIT Verlag, Bremen
Foams and Porous Metal Structures, J. Banhart, 1999, p. 163.
M. F. Ashby, N. A. Fleck (eds), MIT Verlag, 25. X. Wang, T. J. Lu, J. Acoust. Soc. Am. 1999,
Bremen 1999, p. 45. 106, 1.
17. P. D. Wheeler, in Industrial and Machinery 26. Bruel & Kjaer, Mechanical Vibration and
Noise Control Practice, Institute of Sound & Shock Measurements, K. Larsen & Son, So-
Vibration Research, University of South- borg, Denmark 1980.
ampton, UK 1978, p. 11.1. 27. F. Simancik, F. Schoerghuber, in Porous
18. n'H Akustik Design AG, Data Sheet, and Cellular Materials for Structural Applica-
Lungen, Switzerland, 1999. tion, D. S. Schwartz, D. S. Smith, A. G.
19. T. J. Lu, A. Hess, M. F. Ashby, J. Appl. Phys. Evans, H. N. G. Wadley (eds), MRS Symp.
1999, 85, 7528. Proc. Vol. 521, MRS, Warrendale, PA 1998,
20. J. Kovacik, P. Tobolka, F. Simancik, in p. 151.
Metal Foams and Porous Metal Structures, 28. R. Neugebauer, Th. Hipke, in Me-
J. Banhart, M. F. Ashby, N. A. Fleck (eds), tallschaÈume, H. P. Degischer (ed), Wiley
MIT Verlag, Bremen 1999, p. 405. VCH, Weinheim 2000, p. 515.
Handbook of Cellular Metals: Production, Processing, Applications.
Edited by H.-P. Degischer and B. Kriszt
Copyright c 2002 Wiley-VCH Verlag GmbH & Co. KGaA
ISBNs: 3-527-30339-1 (Hardback); 3-527-60055-8 (Electronic) 243
6
Modeling and Simulation
The modeling and simulation of cellular metals is a field of research that has at-
tracted considerable interest from researchers, research groups, foam manufac-
turers, and industrial end-users. Two major directions of research can be distin-
guished: the design of metallic foams on the one hand and the design of compo-
nents made of cellular metals on the other.
6.1
Modeling of Cellular Metals
In the course of the last decade the modeling and simulation of cellular materials
have gained in importance. Well-established representatives of this group, such as
honeycombs or polymer foams, as well as recently developed materials, such as
metallic foams, have been the subjects of various studies. Cellular materials,
which consist of a solid skeleton in the form of struts and/or cell walls and a
high volume fraction of voids, are highly inhomogeneous. This heterogeneity
leads to thermomechanical responses that are markedly different from those of
bulk solids and it gives rise to material properties that have made cellular materials
attractive for many engineering applications. It also presents an obvious target for
modeling studies aimed at gaining an improved understanding of the mechanical
behavior of cellular materials and structures made of them.
246 6.1 Modeling of Cellular Metals
6.1.1
Motivation
Metallic foams, which are the subjects of the present discussion, have entered the
stage of practical application, their most important mechanical properties being ex-
cellent energy absorption capacity as well as high specific strength and stiffness.
Studying their mechanical behavior has attracted considerable research interest,
both from the theoretical and the applications points of view.
Simulations of the mechanical responses of cellular metals may be carried out at
different length scales. In micromechanical approaches the inhomogeneous struc-
ture of these materials is accounted for at the level of individual cells, cell walls,
struts, and vertices either via statistical arguments or by discrete geometrical mod-
els. The corresponding length scale, which is of the order of the size of individual
cells, is called the microscale in the following. Studies at this level provide informa-
tion on the local deformation and load-transfer behavior that can then be correlated
to the mechanical behavior of the material at the structural level. A good under-
standing of the mechanics of cellular metals on the microscale is especially impor-
tant for identifying advantageous microgeometrical parameters for materials de-
sign and development. Microscopic models for metallic foams are the focus of
Section 6.1.2.
For studying samples and components that are, say, two or three orders of mag-
nitude larger than the size of an individual cell, only the overall thermomechanical
behavior is of interest and local details of the metallic foam do not have to be ac-
counted for explicitly. In other words, at the macroscale the foam is treated as a
homogeneous continuum rather than an inhomogeneous medium. Modeling ap-
proaches appropriate for describing metallic foams at the macroscopic level,
which are often rather application specific, will be discussed in Section 6.1.5. In
addition, questions involving the spatial variations or gradients of cell sizes and
shapes within a given sample or structure may be studied at length scales that
are intermediate between microscale and macroscale. Some simple models in-
volving such a mesoscale will be presented in Section 6.1.4, while Section 6.2 is
largely devoted to mesoscopic approaches.
In the following some important aspects of the present state of the modeling and
simulation of metallic cellular materials are discussed. It should be noted, however,
that no attempt is made to give a full overview of this research topic. For compre-
hensive information on the behavior of cellular materials and of its modeling the
reader is referred to the books by Gibson and Ashby [1,2] and to an overview by
Weaire and Fortes [3]. A recent review of the mechanical behavior of metallic
foams was given by Gibson [4].
6 Modeling and Simulation 247
6.1.2
Micromechanical Modeling of Cellular Materials: Basics
From the point of view of micromechanical modeling, cellular metals fall into three
groups: honeycombs, which can be studied by 2D models, open-cell foams, the
solid scaffold of which is dominated by beam-like members, and closed-cell
foams, in which membrane- or shell-like cell walls are present. All of these materi-
als typically show a limited elastic range and their mechanical behavior tends to be
dominated by local deformation mechanisms such as bending, buckling, plastic
yielding, and fracture of cell walls and struts. These deformation mechanisms,
in turn, are highly sensitive to details of the microgeometry, which, in practice,
may be quite regular for some honeycombs, but tends to be complex and highly
non-uniform in the case of metallic foams.
The mean field and Hashin±Shtrikman-type methods that play an important role
in continuum micromechanics of composites and materials with small volume
fractions of pores [5], have seen only limited use for cellular materials. Hashin±
Shtrikman upper bounds for the elastic moduli can be evaluated for macroscopi-
cally isotropic cellular materials by prescribing vanishing stiffness for the void
phase (the lower bounds, however, vanish trivially). Typically, the elastic moduli
of metallic foams lie considerably below these upper bounds, but for honeycombs
it was possible to identify microgeometries that realize the upper bounds [6].
As a consequence, most micromechanical studies of cellular materials in general
and of metallic foams in particular have been based on discrete microgeometrical
models. Owing to the high geometrical complexity and irregularity of actual metal-
lic foams such model microgeometries tend to be highly idealized. One common
modeling strategy is based on studying periodic ªmodel foamsº, the thermomechan-
ical behavior of which is fully described by appropriate unit cells. Alternatively, cells
or geometrical units may be studied in isolation without requiring them to be
space filling. A third type of approach employs a geometrically fully resolved micro-
region (ªcoreº) that is embedded in a much larger region in which the microgeo-
metry is not resolved and smeared-out material behavior is used (ªembedded cell
modelsº). Discrete microgeometrical modeling approaches facilitate the controlled
variation of selected geometrical parameters in order to assess their effect on the
mechanical response. These modeling strategies, which have well-established
equivalents in continuum micromechanics of composite materials, allow studying
both the local deformation mechanisms and the corresponding overall behavior,
which can be obtained by homogenization.
The above concept can also be used to follow the evolution of the yield surface as
a given macroscopic loading path is traced. To do this, the loads are applied incre-
mentally and complete unloading is carried out after each step followed by the eva-
luation of a new yield surface. This may again be done by superimposing elastic
solutions. It should be noted, however, that the application of linear superposition
analyses for such purposes is limited to the small deformation regime. In cellular
metals geometrical nonlinearities (caused, for instance, by marked bending of cell
walls) may give rise to noticeably nonlinear overall responses even before yielding
sets in.
As a foam is subjected to increasing compressive stresses a load will be reached
at which the first collapse of a cell occurs. In metallic foams the initial collapse of a
cell under these conditions tends to be followed by the growth of the collapsed re-
gion, which typically takes place at stress levels that show only limited variation,
giving rise to a so-called plateau region in the overall stress versus strain behavior.
When a considerable percentage of cells have been ªconsumedº the foam densifies,
the plateau region ends, and a much stiffer response sets in. The accumulation of
considerable compressive strains at nearly constant stresses is of particular interest
for cellular metals that are to be employed for impact energy absorption. Collapse
stresses under uniaxial loading have been studied for many cellular morphologies
and for various imperfections. For general load cases the initial collapse stress
states form an envelope surrounding the initial yield surface, which may be called
a collapse surface. For the point-wise evaluation of collapse surfaces fully nonlinear
analyses along individual load paths are required.
Relatively little work has been reported on the modeling of fracture of metallic
foams, which is the dominant failure mechanism under macroscopic tensile load-
ing. Embedded cell models provide a flexible approach for studying cracks in ide-
alized cellular metals, see Section 6.1.3.8.
6.1.3
Selected Results of Micromechanical Simulations
Simone and Gibson [24] developed unit cells for hexagonal honeycombs and tet-
rakaidecahedral foams in which Plateau borders were modeled by appropriate
curved and parallel regions. Their results indicate that the distribution of material
in the cell walls has little influence on the Young's modulus and only a moderate
effect on the uniaxial yield strength of closed-cell metallic foams. A different ap-
proach was followed by Chen et al. [17], who studied honeycombs with wall thick-
nesses that increase linearly from the middle of the walls to the vertices. This sim-
plifies the mathematics of the problem sufficiently for analytical solutions for the
overall yield surfaces of the models to be obtained.
Investigations of the deformation patterns of honeycombs in the large strain re-
gime, which determine the materials' energy absorption capacity, can also be car-
ried out with unit-cell models. In such simulations the initiation of densification
can be identified from the occurrence of self-contact of the surfaces of the collapsed
voids, the modeling of which requires special provisions in terms of the unit-cell
geometries. Unit cells with all boundaries running within the cell walls have
been found to answer well, see Fig. 6.1-5, which shows symmetric and periodic de-
formation modes before the onset of densification. The main advantage of unit
cells of the above type is that all free surfaces of the voids face inwards, which al-
lows a straightforward use of contact algorithms. Obviously, such models can also
be adapted for studying the dependence of the deformation patterns on prescribed
distributions of mass between cell walls and vertices.
Drainage of the material into Plateau borders reduces the cell wall thickness,
which is the governing parameter for the strength and the stiffness of the arrange-
Figure 6.1-5. Predicted deformation modes of hexagonal honeycombs with Plateau borders: left)
symmetric, right) periodic.
254 6.1 Modeling of Cellular Metals
ment. This is especially important in the case of foams of very low density, in
which the cell wall thicknesses approach zero. At higher densities, however, the re-
distribution of material towards the cell vertices and, therefore, into regions sub-
jected to higher bending moments can be beneficial for the overall properties of
foams [17,24].
As more material is accumulated in the vertices, the plastic hinges under bend-
ing tend to form closer to the mid-region of the cell edges, so that the size of the
nearly rigid regions around the vertex is increased. Although such a zone can ro-
tate during compression it still takes up more space if the distance between the
hinges and the vertex centers is increased. Together with changes in the folding
kinematics this causes the cell walls to contact each other at lower strains than
is the case for microgeometries with less pronounced plateau borders. This effect
dominates the energy absorption potential as shown in Fig. 6.1-6, Fig. 6.1-7, and
Fig. 6.1-8. Even in cases where the collapse and pre-contact loads are increased
by the material redistribution, the decrease of the usable deformation length re-
duces the amount of energy that can be dissipated by plastic deformation before
the densification regime is reached.
Microscale fluctuations of the material distribution in cellular materials may, on
the one hand, be due to thickness variations within individual cell walls as dis-
cussed above in connection with Plateau borders or, on the other hand, be caused
by the individual cell walls in a microgeometry having different thicknesses. The
latter problem was addressed by Grenestedt and Bassinet [36], who developed a
3D tetrakaidecahedral model of a closed-cell foam that contains a total of 112
Figure 6.1-6. Predicted nominal strain at first cell wall contact as a function of the material
distribution between walls and cell vertices (Plateau borders) for three different apparent
densities.
6 Modeling and Simulation 255
Figure 6.1-7. Predicted normalized collapse stress as a function of the material distribution
between cell walls and cell vertices (Plateau borders) for three different apparent densities.
Figure 6.1-8. Predicted normalized absorbed energy (up to first cell wall contact) as a function of
the material distribution between cell walls and cell vertices (Plateau borders) for three different
apparent densities.
cell walls, the thicknesses of which could be assigned individually. The results
show that the stiffness of regular closed-cell arrangements is rather insensitive
to the presence of cell walls of different thickness. This was explained by noting
that such microgeometries deform primarily by cell wall stretching, which is
less sensitive to wall thickness effects than bending modes induced, for example,
by corrugated cell walls.
foams, however, curved and corrugated (ªwavyº) cell walls typically are present, see
Fig. 6.1-4.
Grenestedt [16] studied the influence of wavy imperfections of the walls on the
elastic stiffness of cellular solids and found that the bulk modulus decreases by
some 50 % when corrugations with an amplitude of twice the thickness of the
cell walls are present. Simone and Gibson [22] documented the detrimental effect
of curved and corrugated cell walls on the stiffness and strength of regular honey-
combs and perfectly tetrakaidecahedral unit-cell models. The Young's modulus was
shown to be more adversely affected by these imperfections than the collapse
stress. Chen et al. [17] gave analytical expressions for the yield surface of regular
honeycombs with wavy cell walls. They found that corrugations significantly re-
duce the hydrostatic yield strength of honeycombs, whereas the deviatoric yield
stress is hardly affected. Evidently, cell-wall bending is activated in addition to
membrane deformations under overall hydrostatic loading.
a) b)
c) d)
Figure 6.1-9. Predicted yield surfaces for regular and irregular honeycombs with different types and
degrees of imperfections [29]. Dashed and dash-dotted lines denote elastic buckling stress states.
6 Modeling and Simulation 257
The influence of corrugated and curved cell walls on the onset of nonlinear be-
havior was studied by Daxner et al. [29] by evaluating initial yield surfaces.
Figure 6.1-9 shows such curves that represent overall stress states that give rise
to the local onset of yielding or to local elastic buckling in regular and perturbed
honeycombs that have curved and wavy cell walls of varying curvature. In perfect
regular honeycombs in-plane hydrostatic compression leads to pure membrane
stresses in the cell walls, which give rise to an elongated overall initial yield surface
in the s x s y stress plane. For low-density, perfect honeycombs, this yield surface is
truncated by the failure surface for elastic cell wall buckling, see Fig. 6.1-9a, which
shows a kink owing to a change of the buckling mode. This type of behavior is
maintained in the presence of small corrugations, but when the amplitudes of wig-
gles exceed about 5 % of the wall thickness yielding alone determines the onset of
nonlinearity.
The influence of wiggles and curved cell walls on the yield surfaces of honey-
comb microgeometries that are irregular from the outset is much less pronounced,
see Fig. 6-1.9b±d. Perturbing the geometry by randomly displacing cell vertices
causes the yield surface to shrink even more than the presence of severe wiggles,
see Section 6.1.3.3. Corrugated or curved cell walls generally have little effect on
the uniaxial yield strength, but severe wiggles can reduce the in-plane hydrostatic
yield stress by some 35 %. Curved cell walls typically give rise to similar effects as
wiggles, see Fig. 6.1-9c, but cell walls incorporating only slight curvature can lead
to minor increases of the in-plane hydrostatic yield stress.
Figure 6.1-10. Honeycomb geometries for studying the interaction of cells of different sizes [29]:
Regular hexagonal arrangement (HC, left), a cluster of small cells surrounded by big cells (SBS,
center), and big cells surrounded by small cells (BSB, right).
curves. If, in contrast, cell wall buckling and extensive layer-wise collapse perpen-
dicular to the loading direction are the dominant mechanisms, the stress strain
relationship shows marked oscillations on account of the sequential collapse of
ªcell rowsº that are incapable of accommodating subsequent deformation by slid-
ing. Figure 6.1-11 shows deformation modes of the above types for the case of an
imperfect hexagonal honeycomb, with the corresponding stress versus strain
curves being given in Fig. 6.1-12.
In addition to these general observations on collapse modes some specific rules
for the interaction of small and big cells can be identified. Comparing the deforma-
tion patterns developed by arrangements SBS and BSB, see Fig. 6.1-13, one can
observe that in the latter the clusters of small cells do not contribute very much
to the overall deformation and may become obstacles in the densification regime.
The framework of small cells surrounding the large ones in model SBS, however,
provides more uniform stiffness and deforms more evenly. In general, clusters of
small cells tend to be detrimental to the deformation and energy absorption poten-
tial of cellular metals as they typically deform less than their bigger neighbors, the
longer struts of which confer lower bending stiffness and strength.
Figure 6.1-11. Undeformed and deformed loads applied in the horizontal and vertical
configurations (as predicted by unit cells with directions. Note the occurrence of shear loca-
periodic boundary conditions) of an imperfect lization (left) and layer-wise collapse (right),
hexagonal honeycomb subjected to uniaxial respectively.
260 6.1 Modeling of Cellular Metals
Figure 6.1-13. Deformed configurations predicted for imperfect models describing small cells
surrounding big cells (BSB, left) and big cells surrounding small cells (SBS, right) based on the
configurations presented in Fig. 6.1-10.
These observations lead to the conclusion that whenever smooth stress versus
strain relationships are desired, the cell size distribution should be as uniform
as possible and the cells should be small compared to the sample or structure to
either prevent localization or to limit its detrimental effects.
Experiments by Prakash et al. [28] have shown that the filling of some cells leads
to local strengthening of honeycombs, increases the elastic modulus and the de-
gree of strain hardening, but reduces the densification strain. The opposite effect
is caused by removing individual cells or whole cell clusters as demonstrated by
Guo and Gibson [45] in a FE study on intact and damaged honeycombs. They re-
ported on correlations between the undamaged cross-sectional area perpendicular
to the loading direction and the elastic buckling load as well as the plastic collapse
strength. The interaction between separate defects of this type was found to have a
range of about ten cell diameters.
Chen et al. [26] investigated the influence of solid inclusions as well as holes on
the stiffness and on the uniaxial and in-plane hydrostatic yield strengths of perfect
honeycombs and of 2D arrangements in which some 5 % of the cell walls were ran-
domly fractured. Solid inclusions were shown to lead to minor increases of the
elastic stiffness and to have a negligible effect on the uniaxial and in-plane hydro-
static yield strength of both the otherwise perfect and the fractured honeycombs.
Because the solid inclusions introduce additional mass, however, the specific prop-
erties were negatively affected. Large holes were found to induce cell wall bending
in otherwise perfect honeycombs, leading to significant reductions in the bulk
modulus and the hydrostatic yield strength. For prefractured honeycombs the de-
crease in overall stiffness due to holes could be estimated from the reduced overall
relative density of the honeycomb.
Figure 6.1-14.Unit cell for a periodic ªreal structureº honeycomb model [32] adapted from a
micrograph [16]. Note the wide range of cell sizes, cell shapes, and cell wall imperfections.
6 Modeling and Simulation 263
Figure 6.1-15. Overall yield and collapse surfaces [32] predicted for the real structure honeycomb
model presented in Fig. 6.1-14. The collapse surface is not evaluated for load cases for which a
distinct maximum could not be found in the overall stress versus strain relationship.
264 6.1 Modeling of Cellular Metals
isotropic hardening [47]. It assumes that the origin and the shape of the yield sur-
face stay the same, but its size increases driven by the accumulated volumetric
plastic strains. In the Crushable Foam model [48] mixed hardening is assumed,
where the yield surface keeps the same shape with increasing size (isotropic hard-
ening), but shifts its origin (kinematic hardening) so that the hydrostatic tensile
yield stress remains constant.
Micromechanical analyses based on the real structure geometry shown in
Fig. 6.1-14 paint a somewhat more complex picture. Figure 6.1-16 displays predic-
tions for the evolution of the overall yield surface corresponding to progressive uni-
axial compressive loading in the x-direction. At several points during the loading
sequence the analysis was interrupted, the model was unloaded and then subjected
to linear superposition analyses to estimate the onset of nonlinear behavior. This
way the yield surface corresponding to the actual state of hardening can be gener-
ated. The initial hardening behavior was predicted to be essentially kinematic, the
yield surface being shifted in the direction of the applied stress without major
changes in shape or size. At elevated compressive stresses, however, the evolving
yield surface changes its shape and contracts as well. Because Fig. 6.1-16 is
based on the superposition of solutions from linear analyses that cannot account
for the geometrical nonlinearities, which may be expected to play an increasing
role as compressive collapse is approached, its reliability in the latter regime is
not fully clear. Accordingly, the results can be interpreted as clearly favoring kine-
Figure 6.1-16. Evolution of the overall yield surface under uniaxial compressive loading in
x-direction predicted for the real structure honeycomb model presented in Fig. 6.1-14.
6 Modeling and Simulation 265
matic hardening models for low inelastic strains and hinting at possible changes in
the hardening behavior in the high strain regime.
As mentioned before, planar models are directly applicable to honeycombs, but
are not necessarily directly transferable to real foams, which require 3D simula-
tions for quantitative predictions. Denzer developed 3D closed-cell microgeome-
tries [32] based on a body-centered cubic arrangement of tetrakaidecahedral cells,
using symmetry boundary conditions to limit the computational costs. Initial over-
all yield surfaces under compressive loading were evaluated for configurations with
perfect cell geometry as well as for models with statistically perturbed vertex posi-
tions, see Fig. 6.1-3 and Fig. 6.1-17. The general shapes of these yield surfaces in
principal stress space are ellipsoid-like and their major axes are aligned with the
hydrostatic axis, see Fig. 6.1-18. This can be explained by the stress states induced
in the cell walls, which are mainly membrane-like for hydrostatic loads (even
though regular tetrakaidecahedra are not elastically isotropic) whereas uniaxial
loads tend to give rise to bending close to the vertices. These effects are less pro-
nounced for the imperfect models, where the ªaspect ratioº of the yield surface
tends to be lower. This behavior is qualitatively similar to predictions obtained
from honeycomb models, but the yield surfaces for the perfect tetrakaidecahedra
are less elongated along the hydrostatic axis, see Fig. 6.1-9.
By projecting selected points of the yield surface shown in Fig. 6.1-18 (left)
onto the equivalent stress versus mean stress plane a standard representation is
obtained, see Fig. 6.1-19, that can be readily compared with other results. The pro-
jected points can be seen to be scattered around an ellipse the major axis of which
coincides with the mean stress axis, the hydrostatic yield stress being about 2.5
Figure 6.1-17. Irregular tetrakaidecahedral model for a closed-cell foam [32]. As can be seen from
the planar section, perturbations were introduced only in the core of the model to allow for the
use of symmetry boundary conditions.
266 6.1 Modeling of Cellular Metals
Figure 6.1-18. Yield surfaces predicted for regular (Fig. 6.1-3) and irregular (Fig. 6.1-17)
tetrakaidecahedral closed-cell foams [32].
Figure 6.1-19. Projection of the yield surface shown in Fig. 6.1-18 (left) onto the von Mises
equivalent stress versus mean stress surface.
times larger than the uniaxial yield stress. It may be noted that many material laws
for foams use yield surfaces that give rise to ellipses when projected to the equiva-
lent stress versus mean stress plane, among them the model of Deshpande and
Fleck [47].
Figure 6.1-21. Macroscopic force versus displacement curves of CT specimens made of open-cell
aluminum foams predicted by embedded cell models, see Fig. 6.1-20, for different effective
densities and different aluminum alloys [35].
Figure 6.1-22. Macroscopic crack resistance curves of CT specimens made of open-cell alumi-
num foams predicted by embedded cell models, see Fig. 6.1-20, for different effective densities
and different aluminum alloys [35].
were obtained. For this purpose, the macroscopic stress intensity factor KI was eval-
uated at each maximum of the force versus displacement curves according to stan-
dard expressions for CT specimens, the crack length increments Da being dictated
by the selected microgeometry. Both Fig. 6.1-21 and Fig. 6.1-22 show results for
three values of the apparent density of the foam and for two aluminum alloys dif-
fering in ductility.
6 Modeling and Simulation 269
Within such a modeling approach parameters such as the mass density of the
foam (which determines the void volume fraction), the size of the cells (which in-
fluences the fracture behavior in terms of an internal length scale), the yield
strength, and the strain to fracture of the bulk material (which govern the ductile
damage and failure of the struts) can be varied easily in order to study their respec-
tive influences on the fracture behavior of open-cell metallic foams.
6.1.4
Modeling of Mesoscopic Density Inhomogeneities
Foams are not only heterogeneous in terms of consisting of solid regions and
voids, but in many cases also show an inhomogeneous distribution of their appar-
ent density, mean cell size and other geometrical parameters. These variations cor-
respond to length scales larger than that of the voids (the microscale) and smaller
than that of the sample or component (the macroscale) and are, accordingly,
termed mesoscopic inhomogeneities in the following. Fig. 6.1-23 (left) shows
such density variations in a cross section of a foam sample.
Shim et al. [51] introduced uniaxial lumped mass±spring mesoscopic models for
the simulation of the dynamic uniaxial crushing of foam samples with uniform ap-
parent density. Systems of elastic-plastic springs were employed by Gradinger and
Rammerstorfer [52] for studying the compressive force displacement behavior of
crush specimens with deterministically distributed meso-inhomogeneities. These
two approaches were combined by Daxner et al. [53] to assess the influence of in-
homogeneous density distributions on the impact response of foam plates with ge-
neric variations in their apparent density, the interaction of density inhomogene-
Figure 6.1-23. Micrographical cross-section of an aluminum foam with three regions of differing
apparent density (left) and sketch of the corresponding mesoscopic simulation model (right) [53].
270 6.1 Modeling of Cellular Metals
ities in directions normal to the orientation of the load being also accounted for,
see Fig. 6.1-23 (right).
The influence of mesoscale inhomogeneities in the apparent density is of partic-
ular interest when the metal foam is applied as protective padding. Strong mesoin-
homogeneities reduce the energy absorption efficiency, because they lead to strain
localization resulting in a decrease of the initial plateau stress and a less pro-
nounced plateau region [52,53], as can be seen in Fig. 6.1-24. This holds true for
density variations occurring in the loading direction and perpendicularly to it
[53]. Accordingly, for such applications it is desirable to tune foam production tech-
nologies for achieving uniform density distributions.
When a metallic foam is treated as a homogenized material, inertia effects [54]
lead to wave propagation phenomena that are controlled by the interaction of foam
density, layer thickness, impact energy and impact velocity [53]. At moderate im-
pact speeds those inertia effects give rise to an impact resistance that only slightly
exceeds the static yield strength, see Fig. 6.1-25. Nevertheless, the dynamic effects
of wave reflection and superposition are noticeable under these conditions, as is
evident in Fig. 6.1-26, where the stacking order of foam layers with different den-
sity is shown to significantly influence the time history of impact stresses. The re-
sults obtained with simulations that treat cellular metals as homogenized materi-
als, however, will have to be compared to dynamic tests to assure that no artificial
inertia effects are introduced by the homogenization approach. Possibly, the mass
density used for the homogenized material may have to be adjusted to closely fit
the experimental results for dynamic events.
Figure 6.1-24. Stress strain diagrams pre- a meso-homogeneous plate, the open circles
dicted for foam plates that consist of layers of denote a two-layer structure and all other cases
different density while maintaining the same are three-layered arrangements [53].
overall density. The solid line corresponds to
6 Modeling and Simulation 271
Figure 6.1-25. Predicted influence of impact velocity and foam density on the initial stress
response s dyn normalized by the static collapse stress s 0,stat [53].
Figure 6.1-26. Predicted contact stress between an impacting mass and a foam cube composed
of two equally thick layers of different densities [53]. The two possible stacking orders of the layers
were simulated.
6.1.5
Macroscopic Modeling and Simulation
From the perspective of modeling samples and structural components made of cel-
lular metals, discretization down to the level of individual cells is not only infeasi-
ble with current computer and software technology, but also appears to be imprac-
ticable with regard to the required modeling effort. As a consequence, macroscopic
constitutive descriptions of metallic foams have been developed. The main condi-
tion for their use is that the overall dimensions of the sample or structure must be
significantly larger than the dimensions of single foam cells.
Structural analysis of cellular metals in the elastic range is relatively straightfor-
ward in that only the homogenized elasticity tensor must be known. For isotropic
metallic foams, accordingly, two elastic moduli have to be calibrated; more material
parameters are required for foams of lower symmetry. Experiments, however, have
shown that metallic foams may not display a distinct elastic regime under com-
pressive loading, but rather begin to deform irreversibly at stresses that are low
compared to their collapse or plateau stress. Thus, it is not fully clear to what extent
cellular metals can be treated as linear elastic materials in structural analyses.
Practically relevant constitutive models for metallic foams, accordingly, must be
capable of accounting for nonlinear macroscopic material responses. If, for exam-
ple, the FE method is to be used in the engineering analysis of components made
of cellular metals, constitutive descriptions are required at the integration point
level, which may have to be coded specifically [55].
Macroscopic yield surfaces appropriate for metallic foams typically can be repre-
sented by ellipses in the mean-effective stress plane. The yield surface proposed by
Deshpande and Fleck [47] is symmetrical with respect to compressive and tensile
stress states. Furthermore, it evolves by uniform scaling in all directions (isotropic
hardening) and uses an associated flow rule, which implies that the plastic Poisson
ratio is generally not equal to zero. In contrast, the Crushable Foam material model
[48] is characterized by a constant tensile hydrostatic yield stress, which causes the
origin of the yield surface to move along the hydrostatic axis as the yield surface
expands during hardening. The non-associated flow rule governing this material
model is formulated to always prevent a plastic Poisson effect. The accumulated
plastic volumetric strain is the internal variable driving the hardening process.
For additional macroscopic constitutive models proposed for metallic foams see
[56 58].
For the above material models additional parameters determining the shape of
the yield surface and the hardening behavior must be provided by the user, requir-
ing information on the mechanical behavior of the material under multiaxial load-
ing conditions. Since multiaxial experimental data [47,59] is scarce, the user very
often must rely on appropriate assumptions, which can be derived either from
microstructural FE simulations or via parameter identification techniques (that
is, by minimizing the discrepancy between simulations and experimental results).
Fortunately, applications that are dominated by hydrostatic loading conditions
are rare. To illustrate this point, in the next section a case study will be pre-
6 Modeling and Simulation 273
sented in which the uniaxial compressive stress strain relationship is the most
important material characteristic: the impact of a large sphere on an aluminum
foam pad.
Figure 6.1-27. Predicted acceleration versus time history of a typical impact event.
274 6.1 Modeling of Cellular Metals
Figure 6.1-28. Schematic view of an impact event (left) and predicted deformed configuration
after the impact (right).
ing due to fracture of cell walls may also occur when stress states enter tensile re-
gimes.
The evolution of local stress states can be visualized in the mean versus equiva-
lent stress diagram used commonly for describing pressure dependent material
laws. Figure 6.1-29 shows such stress paths for three different material points.
These stress paths lie mostly between the axis of uniaxial compression (inclination
3:1) and the vertical axis that represents pure shear, indicating that these two types
of loading will dominate the local deformation of the foam. The `Axis, top' line in
Fig. 6.1-29 is associated with a point under the center of the impact mass. It is sur-
rounded by an ellipse, which is the projection of the Crushable Foam yield surface
corresponding to the highest value of the hardening parameter, namely the accu-
mulated volumetric plastic strain. Uniaxial stress states can be seen to prevail in
the impact event considered here, indicating that the simulation should not be
very sensitive to changes in the hydrostatic parameters of the material. This is com-
forting, since only limited information is available on the behavior of foams under
hydrostatic pressure. Furthermore, the principal stress axes maintain roughly the
same orientation throughout the impact event, so that the selected hardening
model does not play a major role. These findings narrow down the requirements
on improved material laws for crash simulations involving a comparatively thin
padding and a large impactor.
6 Modeling and Simulation 275
2.5 111111111111
000000000000
000000000000
111111111111
Initial yield surface 000000000000
111111111111
000000000000
111111111111
Axis, bottom
000000000000
111111111111
2.0
Axis, top
Top, Border
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
Effective Stress [MPa]
Uniaxial Compression
Hardened yield surface 000000000000
111111111111
000000000000
111111111111
1.5 000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
1.0 000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
00
11
000000000000
111111111111
00
11
000000000000
111111111111
0.5
00
11
0.0
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
100
FE analysis / filled
FE analysis / empty
3
No. 1396 / 0.52 g/cm
80 No. 1404 / empty
60
Load [kN]
40
20
0
0 20 40 60 80 100
Displacement [mm]
Figure 6.1-30. FE models of the crushing of filled and empty square tubes: Comparison of
computed and measured load±compression curves [61] (left) and stages of compressive axial
deformation (right).
Figure 6.1-31. Folding mechanisms for empty prismatic profile (left) and simplified deformation
kinematics of foam core (right).
6.1.6
Design Optimization for Cellular Metals
Metallic foams show some potential for being produced with controlled spatial
variations of their density, introducing inhomogeneities on a mesoscale as defined
in Section 6.1.4. In the case of foams produced via powder metallurgical routes the
foaming process, and, as a consequence, the final density distribution may be con-
trolled by suitable choices of the distribution of the blowing agent and of process
parameters such as temperature and pressure. This suggests employing such
foams as graded materials in space filling lightweight structures designed in anal-
ogy to cancellous bone, a natural functionally graded cellular material that displays
increased density in regions of high loading, see Fig. 6.1-32. Reiter et al. [72,73]
presented algorithms for the optimization of composite materials that are based
on procedures for simulating the natural adaptation of bone to applied loads.
Daxner et al. [74] implemented the algorithm of Reiter [73] for a self-adapting
material, providing options for converging towards a uniform distribution of the
local strain energy density or towards the satisfaction of a local yield criterion by
iteratively adapting the local density. The former choice gives rise to configurations
of increased stiffness and the latter to density distributions in which the material
shows uniform safety against plastic failure.
It was shown that strength optimization with material parameters that are typi-
cal for metallic foams leads to continuous solutions showing density gradients, see
Fig. 6.1-33, which may indeed be realizable by suitable processing technologies. At-
tempts to optimize the specific stiffness of foam structures resulted in the forma-
tion of discrete structures such as frameworks of struts. This indicates that cellular
metals may not be ideal for stiffness-optimized structures, at least in the presence
of concentrated loads. For sandwich beams with foam cores as studied by Vonach
et al. [75] smooth mesoscopic density distributions were also obtained for stiffness
optimized configurations under distributed loads.
6.1.7
Outlook
Figure 6.1-33. Strength optimized symmetric beam with fixed supports under a central load.
Highly stressed regions such as the supports and the lower and upper regions in the middle of
the beam accumulate the highest local foam density [74]
In view of the engineering potential of cellular metals and of the research chal-
lenges they pose, it appears safe to say that studies of these materials on all length
scales discussed above will remain a fertile field of study for many years to come.
Acknowledgements
References
6.2
Mesomodel of Real Cellular Structures
6.2.1
Introduction
Cellular metals are a class of material that offers special properties as described in
Chapter 5. Their successful introduction into industrial application requires the
development of design methods that predict the responses of these materials to
external loads.
The discrete mass distribution of cellular metals causes discontinuities in their
local properties. To determine the response of a highly porous solid to external
loading, such a cellular structure can be modeled as a network of curved beams
(open-cell foam) or shells (closed-cell foam). This model can be analyzed using
available numerical solutions such as finite elements or boundary elements meth-
ods. The 3D cellular structure of a real, foamed metal (see Chapter 4) is too com-
plex to be easily handled with such a model due to excessive calculation time.
Another method is to approximate the cellular structure by a continuum model.
Such a homogenization method was used by some of the theoretical studies [1]. In
this approach the cellular structure is homogenized over the whole sample or com-
ponent, a scale much larger than the typical microstructure.
Theoretical studies [1] suggest that the mechanical properties of an ideally homo-
geneous foam are scaled by rn. In these relations that are also called ªscaling lawsº
the r is the relative density of the foam and n is a constant that depends on the
mechanism governing the deformation of the cell walls. For example, the elastic
modulus of an open-cell foam with low relative density ( r J 0.1) is governed by
bending of struts and scales with r2. At higher relative densities (0.1 J r J 0.3)
the effect of shear and axial deformation of struts shall be considered [2]. Correc-
tion of the scaling law shows that the influence of these effects on the final
response of an open-cell foam in the elastic regime is small. Prediction of such
scaling laws and their correction at medium densities are compared in Fig. 6.2-1.
In closed-cell foams, bending of struts is accompanied by stretching of the
cell faces, which is described by an additional linear term. The equation for the
theoretical elastic modulus of a closed-cell foam is expressed as [1]
r)2 S C2 (1 s f)
E=Es C1 (f r (1)
where Es is the elastic modulus of the solid material of struts and cell faces. f is the
volume fraction of solid contained in the cell edges, the remaining fraction (1 f) is
in the cell faces. The uniaxial compression (plateau) strength, s, of low-density open-
cell foams is scaled with r3/2. At higher relative densities the influence of the thick-
ness of the struts and the size of the corner becomes considerable [2].
282 6.2 Mesomodel of Real Cellular Structures
As shown in Fig. 6.2-1b the correction of the scaling law at medium densities
has a significant effect on the prediction of the value of plateau strength. The
theoretical value of plateau strength is modified for a closed-cell foam with the
stretching effect of cell faces as
r)3=2 S C4 (1 s f)
s=s ys C3 (f r (2)
where s ys is the yield stress of solid material and C1, ... C4 are constants depending
on the cell geometry [1]. These equations describe the behavior of a material that is
assumed to be continuous and homogeneous. In this study, our interest lies in
local variations of density and other dependent fields over scales not very much lar-
ger than the typical microscale. This may be accomplished with the help of a me-
soscale window (averaging volume), which becomes the classical representative vol-
ume element (RVE) in the infinite medium. Mechanical properties at such a me-
soscale can not be uniquely approximated. But they depend on the essential or nat-
ural boundary conditions of the averaging volume, which bounds the material re-
sponse from above and below [3]. This bound will be tighter with increasing size of
mesoscale or, in other words, with decreasing effect of boundary conditions on the
response of the averaging volume. In the mesoscale, the relative density can be de-
fined as a function of a local point X in an undeformed stress-free reference con-
figuration, r r(X). This allows us to define an inhomogeneous mass distribution
in the foamed samples. Studies on the effects of inhomogeneous or gradient
distributions of mass on the behavior of metallic foams are limited. Beals and
Thomson reported that density gradients have a significant effect on compression
properties of Alcan aluminum foams [4]. Huschka et al. modeled foam materials
with a stacking of layers of different densities [5]. Gradinger and Rammerstorfer
developed a 1D model to analyze the effect of density gradient in loading direction
[6]. The effect of 1D inhomogeneities on the stress strain response and on the
crush energy absorption of aluminum foams has been studied. Daxner et al.
improved this method to a 2D model [7]. In this model, the material is re-
Figure 6.2-1. Correction of scaling laws at medium densities: a) elastic modulus; b) plateau
strength.
6 Modeling and Simulation 283
0.05
Alulight cast
Alulight wrought alloy Scaling law (Correc.)
0.04 Alporas φ = 0.96
σ /σ ys
0.03
Scaling law (Correc.)
φ = 0.9
0.02
0
0.08 0.1 0.12 0.14 0.16 0.18 0.2
Figure 6.2-2. Comparison of measured and predicted plateau strength of Alporas and Alulight
samples. The plateau strengths have been normalized by the yield stress of the cell wall.
6.2.2
3D Mesomodel
The density mapping method was used to approximate the cellular structure by a
continuous, 3D density distribution at the mesolevel. The density recorded by
XCT is homogenized over an averaging volume. The dimension of the averaging
volume specifies the size of mesoscale where the cellular structure is assumed
to be homogeneous. Overlapping of averaging volumes allows to increase the reso-
lution of continuous density distribution up to the resolution of used XCT data.
Details of the density mapping method have been explained in Chapter 4.3 and
previously [8,10]. The influence of an averaging volume size (AVS) on the result
of the developed mesomodel will be discussed in Section 6.2.4.
The obtained inhomogeneous continuum body is then implemented by FE anal-
ysis. The mean density of each element is obtained by averaging the values of den-
sity mapping for all XCT voxels within the FE. Figure 6.2.-3. shows 2D XCT pixels
overlapping with a triangular FE. The next step is to describe the inhomogeneous
body as a multiple sub-domain problem. The material properties (elastic modulus
and Poisson's ratio in the elastic regime and plateau strength in the plastic regime)
depend on the local relative density of each element. In a general case, each ele-
ment will have a mechanical behavior that is different from the behavior of neigh-
boring elements. This high variation in the material properties can complicate the
analysis of the model. To overcome this difficulty the relative density variation,
rmin, rmax], is divided into N intervals. All elements, in which the relative densities
[
lie in the interval i (rFE [ri, ri 1]) form a sub-domain Vi (i 1,2,....,N). Therefore
the material consists of N sub-domains each of which is assumed to be homoge-
neous and isotropic. Their mechanical properties depend on the average value of
the relative densities of the elements lying within the sub-domain.
Although the mechanical properties in each sub-domain are assumed to be con-
stant, they show discontinuities across any interface between sub-domains. The in-
terfacial conditions can lead to deviation from uniaxiality of the loading condition
in the sub-domains even in the case of uniaxial compression of samples. Because
of this, the general form of constitutive laws is used for the description of elastic
and plastic behavior of the sub-domains.
where eie, s j are the elastic strain and stress tensor, respectively, and Sij is the com-
pliance matrix. This matrix is characterized by a set of moduli. Two independent
moduli describe an isotropic and homogeneous material. The Young's modulus
and the Poisson's ratio are used for this purpose. The Young's modulus of each
sub-domain is derived from the scaling law (Eq. 1) with the assumption that the
material behaves like a closed-cell foam. The volume fraction, f, should be ob-
tained by fitting the experimental results.
The Poisson's ratio is assumed to be 0.35 in the elastic regime. There is no ex-
perimental evidence for this parameter but the computational modeling of a reg-
ular cellular structure (Kelvin cells) suggests 0.35 as a reliable value [15].
Classical J2 Plasticity
The classical J2 plasticity is the simplest and most common model to predict the
plastic deformation in dense metals. The yield surface in this model is expressed as
f se s Y 0 (4)
where Y is the yield strength of the material under uniaxial load and s e is the von
Mises effective stress defined by
r
3
se sij sij (5)
2
sij is the deviatoric stress tensor. This yield surface is independent of the mean
stress. Such a material can not fail due to hydrostatic load and is plastically incom-
pressible (epvol) when an associate flow rule is assumed [19]. Because of plastic com-
pressibility of cellular structure, the J2 plasticity is used to estimate the mechanical
behavior of foams at small global strain values and to determine the position where
local deformation begins to form a band. Strictly, J2 plasticity is not valid for
286 6.2 Mesomodel of Real Cellular Structures
where
" ! !3 #
1 1 1 r
t se 1 S s 1s (6b)
2 K K se
where pc and pt are the strength of material in hydrostatic compression and ten-
sion, respectively. M is the slope of critical state line that depends on pt and on the
initial yield strength under uniaxial load, Y, as well as under initial hydrostatic com-
pression strength, pc0. K is a material parameter that defines the shape of yield surface
in the deviatoric plane (Fig. 6.2-4a), and r is the third deviatoric invariant of sij.
The yield surface is depicted in Fig. 6.2-4b. Non-associated flow is assumed for
this constitutive model. The flow potential, g, is defined in term of hydrostatic and
deviatoric effective stress, and chosen in this model as
r
9 2
g p S s 2e (7)
2
This potential gives the zero plastic Poisson's ratio. This means that loading in
any direction causes insignificant deformation in the other directions perpendicu-
lar to the loading. Experiments don't always confirm this property for foamed me-
tals, specially those that have high relative densities [16]. This is a limit of the
crushable foam model to describe the plastic deformation of foamed metals. An-
other restriction is the calibration of the model with the experimental results.
6 Modeling and Simulation 287
S1 t hardened surface
K=1.0
K=0.8
pt pc|0 pc p
S3 softened surface
S2
original surface
(a) (b)
Figure 6.2-4. Crushable foam model: a) effect of material parameter K; b) yield surface in t p
plane.
b) densification regime
pl pl
pc k2 eaevol eD J evol (8b)
k1, k2, and a are material parameters that depend only on relative density. They
must be evaluated by fitting the p e curves obtained by hydrostatic compression
tests. Parameter eD (densification strain) is adjusted to satisfy the continuity of
the function pc(epl
vol) in the passage from zone (a) to zone (b).
6.2.3
Modeling of Uniaxial Compression
Uniaxial compression tests are simulated with respect to the presented model. The
compression samples are rectangular prisms with dimensions of 20 q 20 q
40 mm3. ABAQUS/Standard is employed for computation of the field variable (dis-
placement, strain, and stress). 3D solid (continuum) elements with eight nodes of
type C3D8 are used to model the sub-domains. The nodes of this element have
three translational DOFs (degrees of freedom) but no rotational DOFs. The bound-
ary conditions are chosen so that they represent the conditions of uniaxial com-
pression tests. For this purpose, one clamping surface (bottom face) is assumed
to be fully fixed within the clamping plane, so all nodes are fixed in this plane
(x-, y-, z- translational displacements are assumed to be zero). Note that because
of the friction effect, in practice, the lateral displacement of a sample on the clamp-
ing surface is also negligible. Nodes in the other clamping surface (top face) are
fixed in the lateral directions (x and y) whereas the displacement in the axial direc-
tion (z) is given as loading.
6 Modeling and Simulation 289
Figure 6.2-5. Surface images of an Alulight cells are marked with white arrows; c) calcu-
cast alloy sample of r 0.13: a) cellular struc- lated local plastic strains in loading direction
ture before deformation; b) cellular structure using presented model at 4 % nominal strain,
after deformation, 4 % nominal stain, deformed yielding the first deformation band.
290 6.2 Mesomodel of Real Cellular Structures
Figure 6.2-6. Results of deformation modeling using the crushable foam model for densification
of first deformation band and forming the second band.
6 Modeling and Simulation 291
Figure 6.2-7. Simulation of large deformation of an Alulight sample using the simplified self-
similar model.
292 6.2 Mesomodel of Real Cellular Structures
To compute the effect of inhomogeneity the investigated sample was divided into
nine sub-domains. The densities of the sub-domains were 220, 240, 260, 280, 330,
400, 500, 620, and 720 kg/m3. The calculated plastic strains in the loading direction
happening at a nominal strain of 4 % are depicted in the Fig. 6.2-5c.
Very good agreement is seen between the position of the observed and the com-
puted deformation band. Densification of this band and forming the second defor-
mation band was modeled using the crushable foam model as shown in Fig. 6.2-6.
Large deformation of another Alulight specimen was simulated using the simpli-
fied self-similar model as shown in Fig. 6.2-7. It should be noted that field para-
meters (such as stress, strain, and displacement) are computed for a continuum.
They are average values and must not be considered as microscopical field para-
meters related to deformation of the cell walls.
Figure 6.2-8. Comparison of the measured stress strain curve of a typical Alporas sample with
the prediction of the mesomodel.
6 Modeling and Simulation 293
s 0:507r1:56, d 2:0, eD
true 1:0 s 2:13 r (11)
The effect of inhomogeneous mass distribution on the mechanical properties can
also be calculated using the developed mesomodel. For example, the Young's mod-
uli and compressive strengths of various Alulight samples have been predicted. As
described in Section 6.2.2, the mechanical properties of each sub-domain were ob-
tained from Eq. 1 and Eq. 2 assuming f 0.94 [22]. Computed and measured val-
ues as well as the prediction of scaling laws are depicted in Fig. 6.2-9. Comparison
of computed values with measured ones suggests the effect of inhomogeneity. It
0.08
0.07
0.06
0.05
E/Es
0.04
0.03
Exp.
0.02 mesomodel
Phi=1
Phi=0.9
0.01
Phi=0.94
(a) 0
0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26
0.06
0.05
0.04
σp /σ ys
0.03
0.02
Exp.
mesomodel
Phi=1
0.01
Phi=0.94
Phi=0.94 (correc.)
(b)
0
0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26
Figure 6.2-9. Comparison of measured values with predication of the mesomodel based on the
simplified self-similar model: a) elastic modulus; b) plateau strength.
294 6.2 Mesomodel of Real Cellular Structures
can be seen that this effect on elastic modulus is small, while the plateau strength
can be significantly affected from inhomogeneous mass distribution. However, the
overall response of foamed samples depends on the size and shape of soft (sub-do-
mains with low densities) and hard (sub-domains with high densities) regions. The
plateau strength will be decreased specially when a soft region is oriented perpen-
dicular to the external compressive load [10,12].
6.2.4
Discussion
3.5
2.5
Stress [MPa]
1.5
Exp.
0.5 AVS=11 pixel
AVS=23 pixel
ADS=33 pixel
0
(b) 0 1 2 3 4 5 6
Figure 6.2-11. Influence of averaging volume size (AVS) on the density distribution derived from
XCT data (0.588 mm/pixel).
6 Modeling and Simulation 297
6.2.5
Conclusions
A mesomodel has been developed to explain the localized deformation behavior and
to calculate the mechanical properties on the basis of the mass distribution of the
cellular metals measured by CT data. Density mapping method is employed to ap-
proximate the mesoscopical continuous distribution of density in investigated sam-
ples. The model confirms the correlation between the inhomogeneous density dis-
tribution and the localization of plastic strains yielding deformation bands. The po-
sition of forming and propagation of the deformation band is predicted accurately.
The influence of inhomogeneity on the elastic modulus (the stiffness of the cel-
lular structure) and yield strength of metallic foams can be demonstrated. The in-
fluence on the elastic modulus is small while the plateau strength can be affected
significantly. The discrepancy between the calculated and the measured plateau
strength is caused by the uncertainties for the parameters used for the scaling
laws for Alulight and effects due to the assumption of isotropy.
References
1. L. J. Gibson, M. F. Ashby, Cellular Solids: 12. B. Kriszt, B. Foroughi, K. Faure, H. P.
Structure and Properties, 2nd edn, Cambridge Degischer, Mater. Sci. Technol. 2000, 16,
University Press, UK 1997, Chapter 5. 792 796.
2. L. J. Gibson, M. F. Ashby, Proc. R. Soc. Lond. 13. T. Miyoshi, M. Itoh, S. Akiyama, A. Kita-
1982, A382, 43 59 hara, Adv. Eng. Mater. 2000, 2, 179 183.
3. M. Ostoja-starzewski, Int. J. Sol. Struct. 1998, 14. J. Banhart, J. Baumeister, J. Mater. Sci.
35, 2429 2455. 1998, 33, 1431 1440.
4. J. T. Beals, M. S. Thomson, J. Mater. Sci. 15. A. Ableidinger, Thesis, University of
1997, 32, 3595 3600. Vienna, 2000.
5. S. Huschka, S. Hicken, F. J. Arendts, in Proc. 16. V. S. Deshpande, N. A. Fleck, J. Mech. Phys.
MetallschaÈume, J. Banhart (ed.), MIT, Bremen Solids 2000, 48, 1253 1283.
1997, p. 189±197. 17. R. E. Miller, Int. J. Mech. Sci. 2000, 42,
6. R. Gradinger, F. G. Rammerstorfer, Acta 729 754.
Mater. 1999, 47, 143 148. 18. W. Ehlers, H. MuÈllerschoÈn, O. Klar, in
7. T. Daxner, H. J. BoÈhm, F. G. Rammerstorfer, Metal Foams and Porous Metal Structures,
Comput. Mater. Sci. 1999, 16, 61 91. J. Banhart, M. F. Ashby, N. A. Fleck
8. H. P. Degischer, A. Kottar, in Metal Foams (eds), MIT Verlag, Bremen 1999,
and Porous Metal Structures, J. Banhart, M. F. p. 255 262.
Ashby, N. A. Fleck (eds), MIT Verlag, Bremen 19. J. Lubliner, Plasticity Theory, Macmillan,
1999, p. 213 220. New York 1990.
9. A. Kottar, H. P. Degischer, B. Kriszt, Mater- 20. HKS ABAQUS/Standard user manual,
wiss. Werkstofftechn. 2000, 6, 465 469. Version 5.8, Hibbit Karlsson & Sorensen Inc.,
10. B. Kriszt, B. Foroughi, K. Faure, H. P. Nantucket, RI 1998.
Degischer, in Metal Foams and Porous Metal 21. B. Foroughi, Thesis Vienna, University of
Structures, J. Banhart, M. F. Ashby, N. A. Fleck Technology, Austria 2001.
(eds), MIT Verlag, Bremen 1999, p. 241 246. 22. K. Y. G. McCullough, N. A. Fleck, M. F.
11. B. Kriszt, B. Foroughi, A. Kottar, H. P. Ashby, Acta Mater. 1999, 47, 2323 2330.
Degischer, Proc. Euromat 99, Vol. 5, Wiley- 23. E. W. Andrews, P. R. Oneck, L. J. Gibson,
VCH, Weinheim, 2000, p. 4 82. Int. J. Mech. Sci. 2001, 43, 701 713.
Handbook of Cellular Metals: Production, Processing, Applications.
Edited by H.-P. Degischer and B. Kriszt
Copyright c 2002 Wiley-VCH Verlag GmbH & Co. KGaA
ISBNs: 3-527-30339-1 (Hardback); 3-527-60055-8 (Electronic) 299
7
Service Properties and Exploitability
7.1
The Range of Applications of Structural Foams Based on Cellular Metals
and Alternative Polymer Solutions
C. Haberling
7.1.1
Introduction
Plate structures:
p
3
E Z flexual stiffness
r
(plate)
Rp0,2
Z tensile strength
r
p
E
Z bending stiffness
r (bar)
Hollow structures:
q
3
Rp0,22
Z bending strength
r
(bar)
Rp0,2
Z tensile strength
r
Figure 7.1-1. Typical loading of car body structure (Audi A8): strategy for structural foam rein-
forcements.
7.1.2
Potential Areas of Use
The areas where these kinds of material are of potential use can be derived at an
early stage in design using a finite element (FE) analysis of typical stresses in a car
body. The result makes it possible to evaluate which structural components in the
body structure can play a multifunctional role and what measures can be adopted
to achieve the desired result cost-effectively, taking into account space availability
and the factors inherent in lightweight construction. As illustrated in the car
body of the Audi A8 (Fig. 7.1-1), there are many areas in the bodywork where struc-
tural foams can be used. They are particularly suitable for hollow structures where
space is too restricted for other more cost-effective measures. Particular areas
where the multifunctional applications profile can be exploited to beneficial effect
are the side-members, the boundaries of the floor of the passenger compartment,
and in the region of the sills and pillars.
7.1.3
Material Properties
The application profile of a structural foam can be derived from the density of the
material, its compressive strength, and its elastic modulus. These are illustrated
(Fig. 7.1-2) for a single construction method. It is quite possible to adapt the behav-
ior of the material specifically to the local stress relationships within the structure
of the vehicle owing to the wide range of material densities, types of material, and
7 Service Properties and Exploitability 301
heat treatment methods that are available. Typically, metallic foams have a higher
elastic modulus, whereas epoxy foams exhibit a greater compressive strength at
similar densities for temperatures up to 80 hC.
7.1.4
Main Component Configurations
Using lightweight construction criteria [3], it is possible to use these material char-
acteristics to assess which type of material is most suited for what kind of stress
conditions (Fig. 7.1-1). Compared with compact materials, however, the efficiency
of foam materials in single construction methods where the available building
space is not an issue is somewhat limited. It is only where a differential method
of construction is used with compact materials and in limited spaces, which is
the predominant case encountered in automobile construction, where the effective-
ness of these types of foam materials comes into its own. Even greater advantages
are gained when they are used in conjunction with compact materials. In the first
place, all foam materials exhibit an extremely variable behavior under tensile and
compressive loads. While the compressive range is characterized by a broad, uni-
form, and even deformation behavior (plateau) under constant load, under tension,
structural foams inherently become brittle and fail at very low loads (Fig. 7.1-3). In
addition, inhomogeneous regions are formed in the material that do not particu-
larly affect the overall behavior under compressive stress (reaching the ultimate
strength either later or at a lower level) under purely axial loads, but may cause
buckling or failure of the foam structure if there is an uneven distribution across
302 7.1 The Range of Applications of Structural Foams Based on Cellular Metals
Figure 7.1-3. Deformation behavior of cellular metals under compressive and tensile stresses.
Figure 7.1-4. Buckling due to non-uniform loads on cellular metals: testing and FE simulation.
the stressed cross section, and thus lead to a premature total failure (Fig. 7.1-4).
This would represent a risk particularly in the case of the complex stress patterns
in a car body that can certainly be avoided by using a carrier structure.
The foam materials on the market today that are used for structural stiffening
are ideally used inside hollow components or inside a sandwich composite struc-
ture (Fig. 7.1-5). The function of these carrier materials is to avoid high local ten-
sile stresses in the foam and to apply the forces acting externally into the core of
the foam, directing them during deformation. It is of primary importance, particu-
larly in circumstances where energy absorption is required, to obtain the correct
geometric balance in the carrier/foam system. A completely foamed tubular struc-
ture that is subject to bending can serve as an example. While the empty tube de-
7 Service Properties and Exploitability 303
forms considerably under a low load, when it is filled completely with foam, its
flexural resistance is increased considerably, but, at the same time, it will fail earlier
with smaller deformations. This can be attributed to excessive local stresses in the
outer tube structure that is overloaded owing to the restricted plastic deformation
locally, causing brittle fracture. It is more advisable to provide a partially foam-filled
structure where material is saved in the region of the neutral axis or inside the less
active tensile zones in the case of structural foams. In this way, weight can be
Figure 7.1-6. Foam-filled tubes: variation of foam geometries in impact bending tests.
304 7.1 The Range of Applications of Structural Foams Based on Cellular Metals
7.1.5
Application and Attachment Techniques
In general, there are many ways of applying and attaching structural foams. The
effectiveness depends essentially on the attachment of the foam materials to the
surrounding or adjacent components. This is particularly important in structures
that derive their properties from a composite sandwich structure. Frequently, how-
ever, these requirements conflict with production-related conditions such as man-
ufacturing tolerances. Several additional factors affect the stiffening of hollow
bodywork structures in car construction. In one such instance, it must be ensured
that the process liquids in the paint shop flow through the hollow cross sections of
the bodywork, so that, in general, this objective is defeated if larger hollow sections
are completely filled with foam. To prevent corrosion zones, narrow gaps must be
avoided between foam elements and the basic structure, particularly where mixed
construction methods are used, such as the employment of aluminum foams in a
steel body. Also, incomplete contact must be avoided for acoustic reasons. The con-
tact must be such that either it is so intimate that contact is permanently assured
over the entire surface, or both parts must be protected separately against corrosion
before assembly, for example by painting. However, this solution generally incurs
considerable extra costs. In the case of cellular metals, both cementing and casting
are feasible means of bonding surfaces. Local joining can be done using thermal
and mechanical bonding processes.
7.1.5.1 Casting
The casting of aluminum foams using sand and permanent molds can be regarded
as state of the art. Cast-in molding in pressure die-casting is possible using die-
casting machines controlled in real-time to avoid pressures in the molten metal be-
coming too high. Most die-cast parts used in car construction have a typical wall
thickness of less than 5 mm. In order to deal with all aspects of lightweight con-
struction in this range of applications, it is necessary to explore at what maximum
pressures and casting speeds aluminum foam inserts of minimum density can be
cast-in with the thinnest possible casting skin while avoiding both foam infiltration
and incomplete filling of the mold (Fig. 7.1-7). This is particularly important for
components with large hollow sections (such as swivel bearings) that are normally
produced by gravity molding with a lost-sand core (see Section 3.3).
Apart from casting, gluing is a process that can be used to join together the sur-
faces of foam parts. In car production it is especially important to achieve a suffi-
cient adherence and durability against corrosion and aging. Expanding adhesives
can be used to compensate for even greater manufacturing tolerances, so that, for
instance, it is possible to insert a foam part in an extruded section (Fig. 7.1-7)
(see Section 3.2).
7 Service Properties and Exploitability 305
ances of the parts being joined. Other welding methods require appropriate pre-
paration of the carrier system, such as localized perforations, since the depth of
the weld is shallow. Welding with and without filler materials has been checked
in numerous ways. When employed in fabricating car bodies, there still remain
questions regarding the corrosion behavior in narrow gaps along the weld seam.
Furthermore, the positions of the seams must be adjusted to match the stress pro-
file of the component. This is necessary to ensure the maximum effectiveness of
the foam inlay, and, at the same time, to minimize as much as possible the weak-
ness of the surrounding sheet metal along the seam, since typically in aluminum
welds the zone around the weld is subject to softening.
7.1.6
Effectiveness
Physical characteristics can be checked using simple test pieces for bending, im-
pact, torsional, and axial stresses that mirror the geometry of how hollow struc-
tures are used in a vehicle. In doing so, the assumption is made that the outer
dimensions of the construction space that is being optimized are not allowed to
change since an increase in the cross section of the carrier profile generally repre-
sents the most effective lightweight construction solution. This practical necessity
occurs frequently in vehicle production owing to space restrictions. The test piece
selected was a closed spot-welded hat section made from DC-04 sheet steel, filled
with U- and O-shaped foam pads with a density r 0.55 g/cm3 made from alu-
minum or epoxy foam. The properties evaluated were stiffness under bending,
or flexural strength, impact strength behavior, torsional stiffness, and torsional
strength. This was compared with a section of the same weight with thickened
walls (Fig. 7.1-9).
creased by using an O-shaped structural foam insert [5]. Nevertheless, the effective-
ness is considerably less than when a bending stress is applied. Removing the ef-
fect of weight (t max,elast/m), reinforcement using 4 mm thick epoxy foam produces a
20 % improvement compared with an initial section without the insert. Using an
8 mm thick aluminum foam pad (this reflects the minimum wall thickness cur-
rently possible under production conditions when producing more complex
parts) results in a torsional stiffness that is scarcely any different than if one
used a 4 mm thick epoxy foam of the same density, with adhesion across the
complete surface. This is despite the higher elastic modulus and the greater
wall thickness, and it occurs because the surfaces for gluing the upper and
lower sides are reduced for technical reasons. In this case, therefore, owing to
the total additional weight, the aluminum foam does not offer an advantage
for lightweight construction. This applies also to unreinforced sections with
increased wall thicknesses used for this application, where the increased weight
make them unsuitable for lightweight construction since, again, they offer no
advantages.
Figure 7.1-11. Loading of empty and foam-reinforced test pieces in an impact bending test.
the external plate, the failure load and the energy absorption are more than
doubled compared with an unreinforced hollow section of the same weight and
with thickened walls. This applies also to epoxy foam variants and for aluminum
foam elements glued on all sides. If the same aluminum foam element is glued
only locally, however, such as along the cover plate, the stiffening effect is reduced
considerably (Fig. 7.1-11). If the energy absorption is spread through the body of
the composite component, the same pattern emerges. However, it is also evident
that the effectiveness in relation to the mass used falls off again when the foam
thicknesses are too high.
7.1.6.4 Acoustics
Acoustic effectiveness has been tested on different models (see also Section 3.4.3.1).
Large AFS panels display good absorption qualities below 500 Hz [4]. Other com-
pact foam parts also provide good absorption characteristics but only within a lim-
ited frequency range. Tests performed on foam-filled engine supports (Fig. 7.1-13)
A: diecast B: aluminium
engine support foam-insert
20 mm
m n
C: combination D: alternative
of A and B solution
Figure 7.1-13. Aluminum-foam filled engine support for acoustic means and alternative solution.
7 Service Properties and Exploitability 311
revealed high acoustic effectiveness, but were considerably heavier than alternative
solutions using organic materials since the minimum thickness of the foam was
8 mm. At present, therefore, the acoustic efficiency provided by closed pore cellular
metals when used in car construction can only be regarded as an added benefit
when taken in the context of a multifunctional role.
7.1.7
Outlook
,
ls
Costs for lightweigt construction $/kg
e ria s al
at e ti
m ss en on Cellular
e w roce iffer ucti metals and
n p d str
w d n structural
ne - an f co foams
o o
on d
M t ho Mixed
e
m materials
lighter
materials
Structural
Starting modification
point I=bh /12
W=bh /6
On the other hand, structural foams based on plastics are now found in numer-
ous applications in car manufacture and are used in the structural regions illustrat-
ed in (Fig. 7.1-1). Large-scale use of these materials is, however, prevented in the
current cost situation, although many of the additional costs associated with metal-
lic foams, such as extra joining operations and corrosion protection do not apply in
the case of these materials. It is possible to use them when the space available ex-
cludes any other solutions, or where immediate additional stiffening measures
need to be taken, or when, as in the new Audi A4, additional costs or variants
can be avoided in another place.
AFS sandwich panels in prototypes have shown that there is a potential for using
them also [7]. Even for this material, however, series production has raised new
questions about joining in mixed construction methods. Furthermore, vehicle de-
sign must be adapted to the AFS sandwich in order to extract the maximum effi-
ciency from this material.
First and foremost, the quality of the material has to be improved before cellular
metals can be used more in vehicle technology. It is already possible, however, to
use them inside sandwich structures or in conjunction with carrier materials. De-
velopment effort is important to produce an efficient cost-effective joining method,
as well as investigations on corrosion protection for using them inside steel struc-
tures that are subject to corrosive conditions. Furthermore it is essential to reduce
further the present minimum wall thicknesses and the manufacturing costs in
order to be able to compete with comparable polymer-based materials and other
approaches.
References
7.2
Functional Applications
J. Banhart
7.2.1
General Considerations
The first point is, in particular, crucial for any evaluation of applications for cellular
metallic materials. Many applications require that a medium, either liquid or gas-
eous, be able to pass through the cellular material. There may be a need for various
degrees of ªopennessº, ranging from ªvery openº for high rate fluid flow to
ªcompletely closedº for load-bearing structural applications, and appropriate mate-
rials satisfying these conditions have to be found. Figure 7.2-1 shows which types
of porosity the various application fields require. Normally, a difference is made be-
tween whether an application is ªfunctionalº or ªstructuralº, but there is consider-
able overlap between these two notions.
The question of from which metals or alloys a given type of cellular structure can
be manufactured is also important. Structural load-bearing parts have to be light
because otherwise they would be made from conventional massive metals or alloys.
Therefore, aluminum, magnesium, or titanium cellular or porous metals are pre-
ferred for such applications. For medical applications, titanium may be preferred
because of its compatibility with tissue. Stainless steel or titanium is required
for applications in which aggressive media are involved or high temperatures
occur.
Finally, processing and cost issues have to be considered. The technology must
be available to bring the selected cellular metal into the required shape and to in-
corporate it into the machine or vehicle where it has its function. A technology for
making cellular metal will be futile if the required component cannot be manufac-
tured at a reasonable price.
In the following sections, applications for cellular metals are discussed that are
predominantly ªfunctionalº in the sense discussed. Traditional powder metallurgy
314 7.2 Functional Applications
structural load-bearing
components
bio-medical energy
implants absorbers
catalyst
kind of supports
silencers
application
sound
absorbers
filters bearings
heat
exchangers
functional
open partially open closed
type of porosity
(PM) has created porous sintered metals for a wide range of applications [1 3]. It
is therefore not surprising to find very similar applications for the cellular metals
described in the present text, provided they have a certain degree of open porosity.
7.2.2
Biomedical Implants
Biomedical applications often have both a structural and a functional aspect, and
are therefore very challenging. Titanium or cobalt chromium alloys are used for
prostheses or dental implants because of their biocompatibility. To ensure ingrowth
of tissue, one usually produces a porous layer of the same or another biocompatible
material on the prosthesis by thermal spraying or other methods [4]. Alternatively,
one could use porous titanium or titanium foam for such applications and tailor
the density distribution to meet the required strength and moduli of such compo-
nents. There is no unanimity about how implants should be designed to ensure
maximum durability and functionality. According to one opinion, the modulus
of, say, dental implants should match the modulus of the jaw bone. Knowing
the relationship between modulus and density of metallic foams, one could easily
7 Service Properties and Exploitability 315
7.2.3
Filtration and Separation
There are two types of filters: filters holding back and separating solid particles or
fibers dispersed in a liquid (suspensions) or filters holding back solid or liquid par-
ticles dispersed in a gas (smoke or fog). Examples of the first type are filters for
cleaning recycled polymer melts, for removing yeast from beer, or for contami-
nated oil. The second type includes filtration of diesel fumes or water removal
in air lines. Important filter properties are fine filtration capacity, good particle re-
tention, cleanability, mechanical properties, corrosion resistance, and cost. Some of
the cellular metals described in this text possess a combination of properties not
covered by the traditional PM materials and might therefore be considered as com-
plementary to these, such as the materials described in Sections 2.1.2, 2.3 and 2.4.
7.2.4
Heat Exchangers and Cooling Machines
Highly conductive foams based on copper or aluminum can be used as heat ex-
changers [6]. In this case open-cell structures are needed such as those described
in Sections 2.3 and 2.4. Heat can be removed from or added to gases or liquids
by letting them flow through the foam and cooling or heating the foam at the
same time. Owing to the open porosity, pressure drops can be minimized (see
Fig. 7.2-2). An example of such an application is a compact heat sink for cooling,
for example, in microelectronic devices with a high power dissipation density such
as computer chips or power electronic components. Nowadays, fin-pin arrays are
the standard solution in such cases. Metal foams or ordered cellular metals of
the type shown Fig. 7.2-3 can perform better if they are selected so that thermal
conductivity is kept as high as possible with their flow resistance maintained
low. These two requirements are contradictory. Therefore, an optimization problem
has to be solved [7,9,10]. The variables are the dimensions of the individual struts
and their spacing in the example given. A denser arrangement of metal naturally
gives rise to a better heat conduction but also increases pressure drop. The opti-
mum determined is represented by a certain area in the plane of the diagram
spanned by the heat dissipation and the pressure drop, both expressed as dimen-
sionless quantities, in Fig. 7.2-3.
Another application field for open-cellular materials is transpiration cooling. The
high surface area, low flow resistivity, and good thermal conductivity of some of the
materials used make them promising candidates for such purposes.
316 7.2 Functional Applications
Figure 7.2-2. Pressure drop in partially open-porous cellular solids of the type shown in the
middle of Fig. 7.2-1 [8].
7.2.5
Supports for Catalysts
7.2.6
Storage and Transfer of Liquids
7.2.7
Fluid Flow Control
Porous materials can be used for controlling the flow of liquids and gases [1]. It is
known that PM flow restrictors are more reliable and accurate than conventional
micrometering valves. Because so many degrees of ªopennessº of cellular metals
are available, one could find tailored solutions for even more applications by appro-
priately selecting a cellular metal. Metal foams have already been used as flow
straighteners in wind tunnels [12] or flow distributors in valves [13].
318 7.2 Functional Applications
7.2.8
Silencers
7.2.9
Spargers
7.2.10
Battery Electrodes
Lead foams could serve as supports for the active material in lead acid batteries in
replacement of conventional lead gratings, thus allowing for constructing very
light electrodes [14]. The electrochemically active mass, a paste containing very
fine lead oxide powders, could be filled into the open voids of a lead foam
where it would contact the electrolyte (sulfuric acid). The lead foam acts as a highly
conductive lattice leading to a low internal resistance of the battery. The nickel
7 Service Properties and Exploitability 319
foams described in Section 2.4 are already used as electrodes in pasted rechargeable
NiCd or NiMeH batteries where weight savings and a higher energy density can be
achieved [15 17]. Porous PM materials with their extremely high surface area are
being used in fuel cells [1,15].
7.2.11
Electrochemical Applications
7.2.12
Flame Arresters
Cellular metals with high thermal conductivities of the cell-wall material can be
used to stop flame propagation in combustible gases. Open-cell foams of the
type described in Section 2.4 have been shown to be capable of arresting flames
even when they were travelling at velocities up to 550 m/s. In practice, long
runs of pipes transporting combustible gases are protected close to possible
sources of ignition so that, if ignition does occur, the flame cannot accelerate to
high velocities [13].
7.2.13
Water Purification
7.2.14
Acoustic Control
foams have been studied for their suitability as impedance adapters for ultrasound
sources. Noise reduction functions are described in Section 5.3.
References
7.3
Machinery Applications
Cellular metals have existed as particularly innovative new materials for several
years [1 3]. Since the development of foaming technology, a number of prototypes
for use in the car and machine tool industries have been demonstrated [4 6].
This Section describes possible applications of foamed metals in mechanical en-
gineering.
7 Service Properties and Exploitability 321
7.3.1
Parameters
0,030
Coefficient of thermal expansion
0,025
x-direction
0,020 z-direction
[µm/(K*mm)]
0,015
0,010
0,005
0,000
Alu Alu Steel Steel
1 mm 2 mm 1 mm 2 mm
Figure 7.3-1. Expansion coefficients as a function of cover plate material and plate thickness for
a two-face sandwich (density 0.65 g/cm3; sandwich 25 mm).
322 7.3 Machinery Applications
Recoil, screwed
Figure 7.3-2. Maximal forces for different connecting elements (diameter M6, Al foam density
0.7 g/cm3).
7 Service Properties and Exploitability 323
shape. The joint length with the screws was always 20 mm, but the nuts and in-
serts had a joint length that ranges between 6 and 20 mm.
7.3.2
Examples of Application
tion sensor. From this curve, we obtain the logarithmic damping decrement and
consequently damping, D
!
X^1 1
D ln (1)
X^z 2p(z s 1)
Table 7.3-1 Natural frequencies of hollow profiles and foam-filled profiles. Vibration type: W
local oscillation of pipe walls; X oscillation in x direction (direction of minimum area moment
of inertia).
Table 7.3-2. Frequencies and characteristic types in comparison: pipe 160 q 80 q 3, length
1000, foam-filled/hollow.
damping values are to be integrated into the simulation runs, it is of particular im-
portance that the calculated and the measured values coincide. This is the only way
making sure that the measured damping values can be used for the calculation of
component dimensions
Damping of Structure
Damping was determined at the same time as the frequency was measured. Con-
trarily to expectations, the location of the supports did not influence the frequency
and the damping parameters. This is because, in a dynamic understanding, the
ropes used are soft. So the ropes would only influence a much lower frequency
range than that investigated here.
Having kept all the other conditions constant, even the scanning rate was with-
out any obvious influence: when comparing scanning procedures at 2.5 and
1.25 kHz, the difference between both damping values was merely 0.46 %.
The evaluation showed that the damping factor was slightly higher for the first
vibrations than for more attenuated vibrations. In one experiment, damping be-
tween peak 200 and 700 was D 0.66 q 10 5, and between 1000 and 1500, it
was D 0.64 q 10 5. This effect is caused by the higher damping influence of
the surrounding air for high amplitudes. To minimize this influence of about
3 %, the evaluation range was selected as widely as possible.
In the follow-up investigation of pipes, the damping values shown in Fig. 7.3-4
were determined for the corresponding frequencies. In the investigation, the de-
pendency of damping on the pipe cross sections or the wall thickness could not
be recognized. Instead the damping values are related to the natural frequency.
It makes sense to compare the damping of equal pipe frequencies before and
after foaming.
326 7.3 Machinery Applications
800
The increase in damping by foaming varied: the mean change ranged from 133
to 645 %, or 799 % for actual, rather than mean, values. Figure 7.3-4 shows that the
majority of measured values are located below 300 % (see line); few values lie
above, but these are notable. We suppose that the damping mechanisms that
acted in both types of pipes (above and below the 300 % line) are different. For
the types below the line, material damping is the active mechanism, assuming
that the foamed cores and the steel pipes do not move relative to each other,
thereby excluding any frictional or joint damping. These conditions are obtained
for an optimal foaming procedure with clean steel pipes, so a direct metal connec-
tion between the steel pipe and the foam core is brought about, without any gap
that would permit a relative motion.
If the steel pipes are slightly rusty, then metal-to-metal contact between the pipe
and the foam core may be incomplete, and some frictional damping may then
occur. Frictional damping depends on many factors such as pressure, surface
roughness, the medium in the gap, and others. In pipes lying above the 300 %
limit, a combination of material and frictional vibration can be assumed.
We cut sample sections from the foamed pipes, which confirmed that in some
there was good metal-to-metal contact and in others there was a visible gap
between the foam and the pipe wall. It is too simple to classify the foamed
pipes into ªgoodº and ªpoorº categories, but the following statements can be pos-
tulated.
x For foam-filled pipes in which the foam adheres to the inner steel wall in an op-
timum manner, damping is increased by up to 300 % in comparison with hollow
pipes.
x Damping beyond this limit suggests that the foam has not been completely
bonded. This phenomenon could be used as the basis of a nondestructive quality
inspection of the foaming process.
7 Service Properties and Exploitability 327
Comparing the measured damping values with those taken from literature, we
note that a damping for steel of D 3.7 q 10 4 [7] is in the range of our steel
pipes. The values for metal foam range from D 9 q 10 4 to 2.1 q 10 3 and
are clearly higher than those of steel. Taking both materials together, a
higher total damping should be provided. However, a direct comparison is not
possible, since damping does not only depend on the material and load, but also
the shape of the specimen [8]. For that reason, the damping measured on a work-
piece is sometimes defined as the workpiece vibration instead of the material
vibration.
Mass [kg]
Solid table
300
Stell frame
200
Foamed table
100
25
20
Deflection [µm]
Solid table
15
Steel frame
10 Foamed table
5
Figure 7.3-8. Comparison of maximum
0 deflection.
600
Natural frequency [Hz]
500
400 Solid table
300 Steel frame
200 Foamed table
100
Figure 7.3-9. Frequencies of the first charac-
0 teristic type.
Very good results were obtained for dynamic parameters: the first natural fre-
quency was increased by 80 % (Fig. 7.3-9), which is mostly due to wall vibrations
being almost eliminated by the foam. Such vibrations can only be detected at very
high frequencies.
The damping was multiply increased for the foamed version: for the first natural
bending frequency, a 10 times improvement was achieved (Fig. 7.3-10). Increased
damping arises from the material damping of the metal foam and from the proper-
ties of the adhesive that bonds the steel and the foamed parts.
This result is of special interest for applying the technique in mechanical engi-
neering. In present machine tools, assemblies move with enormous accelerations.
To meet the increased demands for manufacturing accuracy, surface quality, and
tool life, the vibrations resulting from deceleration, acceleration, and the manufac-
turing process, have to be damped in an efficient manner.
2
Damping [%]
1,5
Solid table
1 Steel frame
Foamed table
0,5
Figure 7.3-10. Damping of the first natural
0 frequency.
7 Service Properties and Exploitability 329
7.3.2.3 Cross-Slide
As a second example, the cross-slide of a high-speed milling machine constructed
from steel and foamed aluminum is shown after fabrication (Fig. 7.3-11). The
cross-slide represents the connecting element between the z slide spindle and
the machine bed.
Obvious improvements compared with conventional castings can be anticipated
from simulations (Fig. 7.3-12). Thus, weight is reduced by 46 %. The deformation
deteriorated by 12 %. This worsening could be fully compensated by increasing the
wall thickness. Even then, an evident mass reduction would remain.
Nevertheless, the first natural frequency is increasing by 35 % in any case, which
means a clear enhancement in the field of mechanical engineering. Apart from
this, the position of the natural frequencies can be selected by varying the foam
density.
140
120
100
Change [%]
80
135
60 112
40
54
20
0
Mass Total deformation 1. natural frequency
Figure 7.3-12. Comparison of variants: original design (light) and foamed construction (dark).
330 7.4 Prototypes by Powder Compact Foaming
7.3.3
Conclusions
Metal foams are characterized by very useful physical and mechanical characteris-
tics. Examples from mechanical engineering show some beneficial properties of
hollow components and sandwiches when filled the metal foam material and
how to make use of them in innovative products.
References
7.4
Prototypes by Powder Compact Foaming
R. Kretz
When metallic foams were revived by new manufacturing technologies such as pow-
der metallurgy, investment casting, and other processes in the early nineties, the
new material was introduced to the industry and to potential customers in a very en-
thusiastic manner and with great advertising expense. Information was widely
spread by companies and institutes who worked on the development of metallic
foams. Metallic foams became highlights at industrial fairs and at many conferences
researchers, designers, and materials engineers discussed the structures and the
properties of cellular metals that are relevant to industrial application. The foam de-
velopers missed no chance to present their work and to introduce their new product,
7 Service Properties and Exploitability 331
although the first material samples were only demonstrators: they were small, had
very irregular pore sizes and pore distributions, and a high degree of scattering in
their properties. They were all ªhand madeº and their producers were very proud
of their creations from porous metals. With the growing knowledge and experience,
manufacturing methods were further improved and very soon the presented foams
were of a respectable quality with a certain reproducibility. Whenever people held the
new material in their hands, they were impressed and almost everybody had a new
idea concerning an application of metal foams.
7.4.1
Introduction
But very soon the discussion came round to real parts or components and then the
foam manufacturers had to think about manufacturing net-shape parts, consider-
ing particular problems of geometry or function. Since real net-shape parts are not
flat plates, cubes, or cylinders, new manufacturing methods using mold and fur-
nace technologies had to be introduced. The manufacturing of the powder metal-
lurgical (PM) precursor material as well as the foaming technique for the com-
pacted powder are called the powder compact foaming technique (PCF), which
is practiced by IFAM, Bremen, Germany, Leichtmetallkompetenzzentrum, Rans-
hofen (LKR), Austria, and the Slovak Academy of Science (SAS), Bratislava, Slova-
7.4.2
Methods, Machines, and Molds
From the very beginning the application of the PCF route for making net-shape
parts was the goal. This method, using compacted aluminum powder and titanium
hydride as a foaming agent, is described by several authors [1,2].
7 Service Properties and Exploitability 333
very high and a uniform temperature distribution could be achieved. This method
worked quite well and the aluminum-foam samples yielded good properties, but
chemical attack of the molds by the Al melt caused short lifetimes of the molds
and the mold attachments. Furthermore, the handling of the molds in liquid alu-
minum is dangerous and problematic.
High-convection furnaces (air-circulating furnaces) are used in industry for rapid
heating processes. A ventilator in the furnace effects very high heat transfer from
the heat source to the mold. The rapid heat transfer, caused by high flow velocity,
causes high heating rates, but the gas flow is influenced by the geometry of the
mold and that leads to uneven temperature distribution within the mold. This ef-
fect can be avoided by the installation of baffle plates, which must be adapted for
every new mold that is used in the furnace. Therefore air-circulating furnaces can
be employed in series production, where high numbers of aluminum-foam parts
have to be made with the same mold geometry, but they are not suitable for pro-
totype manufacturing, where flexible and universal furnace systems are required.
For prototype manufacturing it is essential to cover a wide range of different
components. For this purpose a special chamber furnace with adequate heating ca-
pacity had to be developed. One example of a foaming furnace especially built for
foaming of big components, is the 100 kW Nabertherm Chamber furnace at LKR,
This method was used for foaming car structural parts, the largest prototypes so far
manufactured in net shape.
Besides structural steel and cast iron, high-temperature tool steel, copper, and
ceramics were also used as mold materials in a large number of prototype devel-
opment projects. Table 7.4-1 shows an overview of different mold materials and
their properties.
To avoid sticking of the aluminum-foam product in the molds, the cavity of the
die was coated with common facing materials.
7.4.3
Prototypes and their Applications
It is a long way from the first idea to the first prototype, and it is even further from
the prototype to a series product. Some of the product ideas had to be rejected be-
cause the aluminum-foam properties were over-estimated. The outstanding me-
chanical properties of closed-cell metal foams are their energy absorbing potential
and specific stiffness. Figure 7.4-4 shows a couple of aluminum-foam samples
made for testing mechanical and technological properties of the material.
The feasible applications are spread in a wide range of various industrial fields.
The automotive and transportation industries are interested in metallic foams as
well as mechanical engineering, architecture, and building construction companies.
High specific structural parts such as intermediate panels or front panels are fea-
sible applications. Figure 7.4-7 shows a couple of structural part prototypes includ-
ing an intermediate panel [9] that was made for DaimlerChrysler in 1995.
The largest aluminum-foam prototype part made by LKR was a front panel that
was foamed by order of Opel in 1999. The dimensions of this panel, shown in
Fig. 7.4-8, are about 140 cm q 50 cm; it is a net-shape foamed in a cast-iron
mold [10].
Up until now sandwich sheets have had to be made in two process steps. A foam
plate is made and the face sheets are applied by adhesives. Manufacturing of sand-
wich parts and foam filled aluminum profiles by forming the foam between two
aluminum sheets is reported in Section 3.3.
Aluminum-foam filled steel profiles can also be used as energy absorbing
components in cars, trucks, trains, and other vehicles. Manufacturing of these
profiles is relatively easy because the steel profile can be used as a foaming mold.
In Fig. 7.4-10 a couple of foam-filled profiles and sandwich specimens are shown.
Aluminum or magnesium castings with permanent aluminum-foam cores
should meet the requirements of improved energy absorbing potential, higher
stiffness than hollow castings and good vibration damping behavior. The use of
foam cores can also improve the manufacturing process in a number of ways: no
sand cores have to be removed from the casting, complicated quasi-hollow shapes
can be cast, and improved solidification conditions can be achieved. Target applica-
tions for castings with foam cores, as shown in Fig. 7.4-11, are suspension and struc-
342 7.4 Prototypes by Powder Compact Foaming
Foam cores, cubes, ring foundries, cores in castings casting conditions, damping
behavior, weight,
energy absorption,
7 Service Properties and Exploitability 343
tural parts with improved energy absorbing properties, and wheels. Also mechanical
engineering components with good damping behavior and good castability [11].
In addition to the applications already described a large variety of other potential
technical applications of aluminum foam have been discussed. Table 7.4-2 con-
tains an overview of feasible aluminum-foam applications.
A very important task of foam developers and foam manufacturers must there-
fore be to give their customers and potential material users up-to-date and accurate
information about the material properties and convenient applications, but also its
limits. Otherwise they run the risk of losing interest in this material altogether.
References
7.5
Applying the Investment Methodology for Materials (IMM) to Aluminum Foams
7.5.1
Introduction: The Investment Methodology for Materials (IMM)
IMM consists of three linked segments: viability analysis, market assessment, and
value capture (Fig. 7.5-1). A material is viable in an application if the balance be-
tween its technical and economic attributes is favorable. Assessing viability in-
volves technical modeling of the application, cost modeling of manufacturing,
input from the market assessment, and value analysis. Market assessment deploys
techniques for identifying promising market applications and for forecasting fu-
ture production volumes. Likelihood of value capture is assessed through an
7.5.2
Initial Market Scan for Potential Applications for Al Foams
Aluminum foams have several unusual features, among them, exceptional en-
ergy absorbed per unit volume and, particularly, energy absorbed per unit weight [5].
Potential applications are those in which these functions are important; those
that also exploit secondary attributes of metal foams (such as flame resistance,
heat dissipation, noise reduction, water resistance, blast amelioration) are parti-
cularly attractive. The strongest opportunities for short-term take up of alu-
minum foam lie with substitution into existing markets, since for these the
design requirements are well known. Radically new designs taking best advan-
tage of their properties are slowed by lack of designer awareness and confi-
dence, and by current manufacturing difficulties with achieving high toler-
ances.
There are existing aerospace and automotive applications that require high en-
ergy absorption per unit volume. The large-scale production of aluminum foams
cannot yet meet the tolerances required for aerospace applications (other than
the expensive Duocel, which finds a market in defense and space vehicles); thus,
automotive applications are prioritized, specifically, the ªA-pillarº designed for oc-
cupant safety, and the redesigned hood and front bumper regions to meet new pe-
destrian protection legislation.
7.5.3
Material Assessment
Figure 7.5-2. Stress strain curves of closed-cell Alporas and open-cell ERG metal foams
(Duocell).
Figure 7.5-3. Production costs for processing aluminum foam by three methods.
7 Service Properties and Exploitability 349
and capital write-off. The output of the models (Fig. 7.5-3) shows the way in which
the cost of the manufactured component depends on production volume. Cost dri-
vers can also be determined. In this instance the batch powder metallurgical pro-
cess is the most economical option for annual production volumes of up to 20 000
parts [2,9].
By comparing the cheapest aluminum foam component (at the required produc-
tion volume) with an incumbent material for any given application, the cost differ-
ential between a new and an incumbent material solution can be established. This
provides input into the value assessment.
Figure 7.5-4. Selection of metal foams and polymer foams by volume and cost with s pl 3.5 MPa.
350 7.5 Applying the Investment Methodology for Materials (IMM) to Aluminum Foams
7.5.4
Market Forecast
A technical assessment indicates that metal foams are technically viable as energy
absorption elements within the A-pillar [8]. The next step is to examine the market
size and rate of take-up.
Figure 7.5-5. Scenarios for aluminum foam substitution into A-pillar market.
7.5.5
Value Capture
rights to the process, which is not difficult to reproduce. Their position is made
less attractive because they must fight both against competitors who are commer-
cializing alternative processes for making metal foams (the lower end liquid alumi-
num foam process, for instance) and against substitutes for energy absorption in
automotive applications, such as polymeric foams, shaped aluminum sections, and
fiber-reinforced polymer composite sections. Additionally, the automotive compa-
nies exert very strong buyer power that will be difficult for the small producer to
counter (Fig. 7.5-6).
7.5.6
Conclusions: Applying IMM to Aluminum Foams
As the chances of value capture are relatively low and the payback period is rela-
tively long, a small firm would be disadvantaged in commercializing this innova-
tion unless as a joint venture with a metal supplier or automotive producer. A lar-
ger company might be interested in pursuing this opportunity if it was in a good
position to capture the value created. Alternatively, a government-sponsored initia-
tive for pedestrian or occupant safety might subsidize the commercialization of
such an innovation.
Acknowledgements
We wish to acknowledge the support of the KoÈrber Foundation, the Cambridge Ca-
nadian Trust, the UK EPSRC through the support of the Engineering Design Cen-
tre at Cambridge, and Granta Design, Cambridge, who developed the software
illustrated in this article.
References
1. D. Wield, R. Roy, ªR&D and corporate stra- 5. M. F. Ashby, A. Evans, N. Fleck, L. J. Gibson,
tegies in UK materials-innovating J. W. Hutchinson, H. N. G. Wadley, Metal
companiesº Technovation 1995, 15(4). Foams: A Design Guide, Butterworth-Heine-
2. E. M. A. Maine, ªInnovation and Adoption of mann, Oxford 2000.
New Materialsº PhD Thesis, Cambridge 6. CES, The Cambridge Engineering Selector,
University 2000. Granta Design, Cambridge 1999, http://
3. C. Freeman, L. Soete, The Economics of In- www.granta.co.uk
dustrial Innovation, 3rd ed, Pinter, London 7. M. F. Ashby, Materials Selection in Mechanical
1997, p. 237. Design, Pergamon Press, Oxford 1992.
4. E. M. A. Maine, M. F. Ashby, ªAn invest- 8. E. M. A. Maine, M. F. Ashby, ªAn invest-
ment methodology for materials. Part 1: The ment methodology for materials. Part 2: Ap-
methodologyº Mater. Design, Elsevier 2002 plications of the methodologyº Mater. Design,
in press. Elsevier 2002 in press.
354 7.5 Applying the Investment Methodology for Materials (IMM) to Aluminum Foams
8
Strengths, Weaknesses, and Opportunities
H. P. Degischer
The research interest in cellular metals and the activities of manufacturers in var-
ious market sectors illustrate the appeal of these materials: they are attractive be-
cause of their unusual specific properties and appealing property profiles. Expecta-
tions for so-far unachievable metallic light-weight structures have created ideas for
the substitution of existing materials in weight-sensitive components such as
floating metals, light stiff panels for automotive and aerospace applications, crash-
absorbing safety components, electromagnetic shielding, structural components
with low absorption for nuclear particle radiation, housings with high acoustic
damping, and decorative design objects. Many different markets have been
addressed, but how long will it take for a market breakthrough? Tables 8-1 8-3
resulted from a discussion at the end of the Symposium ªMetallschaÈumeº in
Vienna, 29 February 2000. The tables show the advantages of cellular metals and
the achievements of research as well as their disadvantages, and the remaining un-
answered questions: i. e. cellular metals' strengths, weaknesses, and opportunities.
8.1
Processing
Table 8-1. Achievements in the processing of cellular metals and remaining questions.
x Mass densities about 10 1 of the bulk x Other metals only in laboratory scale;
metal (mainly aluminum alloys); x relation of processing parameters with
x powder compact foaming and melt cell architecture not fully understood,
foaming processes; therefore unsatisfactory control of
x various brands of foamed aluminum uniformity of cell architecture;
Foamed metals
very low densities with the highest degree of regularity and reproducibility of cell
architecture. They seem to be only appropriate for small volume components for
high-tech and high-price applications.
Foaming of metals in the liquid state promises the most economic processing
route with the drawback of difficult control of cell architecture. Products of remark-
able size and complexity have been made for structural applications. Although
foaming produces closed bubbles in the liquid, during solidification the cell
walls are usually damaged giving the components a certain permeability and neg-
ligible mechanical stiffening. Even the characteristic surface skin of foamed metal
parts can not be considered as defect-free and sufficiently protective. Processing of
closed-cell structures with enclosed hollow substrates or sintered hollow spheres is
still in the prototyping stage.
There are no defined specifications for cellular metals. They are essentially char-
acterized by their processing method. The dependence of material properties on
the different techniques of production remain the subject of research and develop-
ment. There is great potential in improving reproducibility by scaling-up produc-
tion, allowing better control of processing parameters. Some successful examples
of industrial production are quoted in Table 8-1.
The chances of any of the cellular metals being used in components depends
mainly on the availability of appropriate bonding and joining techniques. Second-
ary processing like forming, machining, bonding, and surface treatment require
further technological developments, which have to be governed by the service re-
quirements of each case. A significant advantage of cellular metals compared
with polymers is their recyclability and the potential of up-grading scrap metal,
which may gain increasing importance in the future.
358 8.2 Properties
8.2
Properties
The processing method for a cellular metal component defines the achievable
property profile in the first instance. The density achieved by a given production
route determines the magnitude of the properties. Physical principals govern
those property relations, which reveal the limits as well. The introduction of
pores into a metal cannot improve its mechanical properties even when related
to the specific mass. Tensile strength is limited by fracture mechanics due to inho-
mogeneous stress distributions within the cellular architecture containing many
crack-initiation sites. Properties depending nonlinearly on the dimension of the
part offer benefits for design application, for example specific stiffness for bending
of beams or plates.
The close relation between processing and properties is enhanced by the impor-
tance of the mesoscopic 3D-structure formed by the metal matrix with its specific
microstructure. The scatter of material properties is assumed to be caused by irre-
gularities in the cellular architecture of real products. Some of these can be toler-
ated, but there is still the ambiguity of extremely poor property values not yet cor-
related with structural parameters. Sensitivity studies are required to relate the in-
fluence of meso- and microscopic defects to the properties. Consequently the reli-
ability of the properties needs verification by testing series of products.
The properties depend on the variations in cellular architecture and their unifor-
mity. The actual achievements are contrasted in Table 8-2 with the deficiencies still
existing for most of the available cellular metals. The average density does not
seem sufficient for the characterization of a cellular component. The properties
of cellular materials are not only determined by the matrix metal, but essentially
Table 8-2. Achieved properties and characterization methods compared with demands.
x Average mass densities in the range x Uniformity only for cellular structures
10 1 10 2 of the bulk determined by with porous substrates, and solid state
classic densitometry; processed;
x medium and high resolution 2D- x architectural defects' assessment re-
microscopy; quired; oversized pores as well as
x thermo-acoustic nondestructive test mass concentrations cause degrada-
Cellular structure
x Thermal stability usually better than x High thermal gradients due to low
Environmental properties
by the local material distribution: the architecture of the cell structure. X-ray com-
puted tomography offers the possibility of 3D characterization, but requires signif-
icant effort. Nondestructive testing is one of the main activities of the research ne-
cessary to correlate properties with cellular architecture. Furthermore, the proper-
ties of the matrix metal itself cannot be simply transferred from bulk material, but
need to be investigated for each processing technique. In the case of age-hardening
metals, the influence of thermal treatments during production has been indicated,
but may offer chances for further material development.
Table 8.2 lists the most appealing properties, especially attractive in relation with
the low specific mass. The potential of cellular metals depends on combinations of
these specific properties according to particular service requirements. The mechan-
ical properties for structural applications have to be defined specifically for these
cellular structures. Elastic properties, for example, have to be considered as the
stiffness properties of the architecture of the metal; similarly plastic deformation
and damage mechanisms. Complications arise from the internally multi-axial
load transfer, which require special attention regarding the test conditions provid-
ing material data. Both test conditions and the significance of the results have to be
defined properly for cellular metals.
Life-time predictions require further research to complete the achieved results
for exposure conditions simulating long-term service. Damage mechanisms are re-
vealed qualitatively and first attempts of quantification need verification in use. The
physical properties deserve considerable attention to underline the perspectives of
multi-functional application. They have to be determined quantitatively individu-
ally for each type of cellular metal, based on the principally existing potential.
There are essential advantages of cellular metals in comparison with polymers re-
garding environmental properties. Of course, the environmental service conditions
have to be considered for specific cases. The high fire resistance and the space
compatibility have not yet been exploited for aerospace applications.
8.3
Design and Application
The technical advantages of cellular metals cannot be fully exploited just by substi-
tution of materials, but require material-appropriate engineering solutions and de-
sign concepts. The simulation of the performance of components relies on the ma-
terial laws and constitutive equations appropriate to cellular metals and their load-
ing conditions. The actual situation in modeling of cellular metals and in design-
ing components thereof is summarized in Table 8-3. Material models for cellular
metals are required, which have to reflect the 3D behavior of the highly porous,
compressible material. Theoretical studies help to identify the important para-
meters. The transfer of the theoretical results to real architectures requires experi-
mental calibration of the material laws. As stated before, even micromechanical
computations require the appropriate metal matrix properties, which are very dif-
ficult to determine in the processing condition of cells. Reasonable approximations
8 Strengths, Weaknesses, and Opportunities 361
Table 8-3. State of applicability of modeling, simulation, and design methods for different
products.
x Applications where the value of weight x Price estimations actually more than
saving is high (means of transport, 15 U/kg or 5 U/liter (which is more
highly accelerated machine parts and than the automotive market is pre-
sporting goods); pared to pay);
x aerospace applications; x advanced material image to be im-
x small series rail, public transport and proved;
Market
Index
d F-type ATPases 23 ff
des(diserylglycyl)ferrirhodin 470 families of transporters 1 ff, 5 ff
dipeptide permease (Dpp) 144 Fe3 -citrate 306
diphtheria toxin 304 Fe3 -dicitrate 300
DMT1 329 Fe3 regulatory mechanism 306
Dpp-type 159 Fe3 -siderophore transport 291 ff
DppA 89 Fe-enterobactin transporter 473
DTAP domain, in ArsA 386 FecI and FecR proteins 306
DtxR, crystal structures 305 feo genes 304
DtxR/IdeR family 338 FepA 264 ff
b-barrel 265 ff
e conserved residues 284
EI 118 extracellular loops 270 ff
enantio-ferrichrome 470 N-terminal domain 267 ff
enantio-rhizoferrin 468 ribbon diagram 264
ENB1 transporter 472 structure determination of 275
energy transduction, ATP synthase 31 ff FepA mutants 285
enteric bacteria, peptide transport in 143 ff Ferric Uptake Repressor (Fur) 337
Enterococcus hirae cop operon 368 ferric citrate transport system 306
Enterococcus hirae CopB copper ATPase ferrichrome 470
363 ferrichrome A 470
Escherichia coli HlyA 221 ferrichrome transporter 471
aquaporin Z 250, 255 ferrichrysin 470
ars operon 378, 391 ferricrocin 470
ABC transporters 80 ff ferrirhodin 470
ArsA 388 ferrirubin 470
ATP synthase 24 ff, 36 ff ferrous iron transport system 304
ATP transporter binding proteins 90 FET3 448
copper transport in 371 FET4 450
FhuABCD activities 292 FET5 450
FhuD protein 297 FhuA 264 ff
LamB 240 barrel structure 266
MalK 105 conserved residues 284
melibiose transporter 54 extracellular loops 270 ff
Na/melibiose transporter 58 N-terminal domain 267 ff
Na/proline transporter 55, 59 ff, 65 ff structures with ligand 272 ff
Na/substrate transport in 52 FhuA protein 265, 292
Nik system 407 FhuABCD activities, of E. coli 292
OmpF 234 FhuB transport protein 298
peptide transport 143 ff FhuC, ATP hydrolysis by 298
Tat pathway in 174 FhuD 89, 295 ff
TolC trimer 242 crystal structure 297
extracellular loops Fibrobacter succinogenes 50
in FepA 270 ff FRE genes, of Saccharomyces cerevisiae
in FhuA 270 ff 448
FRE reductases, in siderophore
f transport 474 ff
F1 subunits 25 ff FRE1 447
F1/FO interface 31 ff FRE2 447
F1FO ATP synthase 23 ff FRE3 448
model of energy transduction 39 ff FRE4 448
FO subunits 33 ff FTR1 448
proton translocation pathway 42 fungal ornithine-N5 -oxygenase 466
F-ATPases 23 ff fungi, siderophore transport in 463 ff
484 Index
i m
IIA 118 magnesium transporters 347 ff
IIAGlc subunit 120 CorA 351 ff
regulatory role 131 MgtE 350
IIB 118 Major Intrinsic Proteins (MIPs)
IIC 118 superfamily 247
IICBGlc mutants 124 ff malK dimer, asymmetry within 105 ff
listing of 125 ff MalK from Thermococcus litoralis 101 ff
IICBGlc subunit 121 ff maltoporin 237
regulatory role 132 maltose binding protein from E. coli 90
topology of 122 maltose binding protein MBP or MalE 239
insecticidal crystal (Cry) proteins 223 manganese, in bacteria 326
integral transmembrane domains (TMDs) manganese accumulation 325 ff
ABC transporters 91 ff manganese homeostasis 452
sequence 96 manganese homeostasis protein,
ion selectivity of porins 233 Ccc1p 456
Index 485
manganese transport n
Saccharomyces cerevisiae genes involved N-terminal domain
in 458 of FepA 267 ff
in Saccharomyces cerevisiae 452 ff of FhuA 267 ff
in bacteria 330 ff Na cycle 52
manganese transport ATPase, Pmr1p 455 Na/citrate transporter (CitS) of
manganese transporters, in yeast 447 ff K. pneumoniae 56, 59, 61, 69
manganese uptake, in bacteria 325 Na/H antiporter 52
MBP 89 Na/K ATPase 23
MBP (MalE) 89 Na/melibiose transporter of E. coli 58
melibiose transporter of E. coli (EcMelB) Na/proline transporter 53
54, 62 ff Na/proline transporter (PutP) of E. coli
melibiose transporter of K. pneumoniae 59 ff, 55, 65 ff
(KnMelB) 54, 62 Na/substrate transport in Escherichia coli
membrane topology of CPx-type ATPases 363 52
metal ABC transporters 316 Na/substrate transport systems 48 ff
metalloenzymes, nickel as cofactor 398 ff Natural Resistance-Associated Macrophage
Methanobacterium thermoautotrophicum 49 Protein (Nramp1) 329
Methanococcus jannaschii 49, 353 Ncc (nickel cobalt cadmium resis-
MFS-type nickel exporter (NrsD) 406 tance) 406
Mg2 transporter families: MgtE, CorA, NhlF 410
and MgtA/B 347 Nic1p, of Schizosaccharomyces pombe 412
MgtA/MgtB Mg2 transporters 355 ff nickel, as a cofactor of metalloenzymes
MgtC protein 357 398 ff
MgtE magnesium transporters 350 nickel/cobalt transporter family 408 ff
microbial nickel transport 397 ff hydropathy profile alignment 409
microbial siderophores 465 in Bradyrhizobium japonicum 412
MIP-like channel genes 252 nickel homeostasis 403 ff
MIP-related aquaporins 248 nickel resistance 401
MIP-related sequences, phylogenetic nickel toxicity 401
tree 252 nickel transport 397 ff
mitochondrial copper ion transport nickel transporter (UreH, UreI)
419 ff of thermophilic Bacillus sp. 412
mitochondrial cytochrome c oxidase nickel transporters, ABC-type- 407
422 ff nickel uptake, high-affinity systems
MMT1 451 406
MMT2 451 nickel/cobalt transporters, metabolic
Mn(II) ABC transporters 328 functions 410
Mn(II) as a Fur co-repressor 338 NiCoT in Bradyrhizobium japonicum,
Mn(II) primary transporter 328 nickel/cobalt transporter in 412
Mn(II) permeases 328 Nik system, Escherichia coli 407
Mn(II) transport in S. aureus 330 Nik-related transporters 408
MntABC uptake system 333 nikABCDE operon 407
MntH 329 nikR gene 407
MntH proteins 336 nitrous oxide reductase (N2OR),
ModA 89 in P. stutzeri 439
molecular recognition templates NixA 410
(MRT) 156 ff (Nramp)/Divalent Metal Transporter
molecular recognition templates (MRTs), (DMT) 329
optimal features of 159 Nramp family 452
MRP1 (multidrug resistance-associated Nramp1 protein 339
protein) 393 Nramp2 329
multi-copper oxidase 448 nucleotide binding sites, in ArsA
Mycobacterium tuberculosis (MtNicT) 413 381 ff
486 Index
u y
ubiquinol oxidase 436 yeast
uniporters 8 ff ACR genes 393
Urbs1 in Ustilago 466 AFT1 in 466
ureases 399 iron transporters 447 ff
Ustilago, Urbs1 466 manganese transporters 447 ff
yeast oxidase complex 422
v yeast vacuole and manganese 457
Vibrio alginolyticus 50 Yersinia pestis YfeABCD system 334
Vibrio parahaemolyticus 50 YfeABCD system, of Yersinia pestis 334
virulence factors 304
vitamin B12, binding of 294 z
Zinc transport 313 ff
w Zinc uptake, low-affinity systems 321
Walker A and B sequences 98 ZIP transporters 405
Walker A sequence 386 Zn2 export 314
water channels 247 Zn2 transport systems, regulators of 323
Wilson copper ATPase (ATP7B) 367 Zn2 uptake, binding protein-dependent-
316 ff
znuA (zinc uptake) gene 321