Materials Science & Engineering A 558 (2012) 343–348

Contents lists available at SciVerse ScienceDirect

Materials Science & Engineering A

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

Production of high porosity metal foams using EPS beads as space holders
T. Shimizu a,n, K. Matsuzaki a, H. Nagai a, N. Kanetake b
a
AIST (National Institute of Advanced Industrial Science and Technology), Namiki 1-2-1, Tsukuba, Ibaraki 305-8564, Japan
b
Department of Material Science and Engineering, Graduate School of Engineering, Nagoya University, Furou-cyo, Nagoya, Aichi 464-8601, Japan

a r t i c l e i n f o abstract

Article history: High porosity metal foams were produced using a SWS (Space holders & Whipped Slurry) process that
Received 19 February 2012 uses EPS (Expanded polystyrene) beads as space holders and a whipped slurry of a metal powder and a
Received in revised form water solution of PVA (Polyvinyl alcohol). First, the slurry is whipped and mixed with EPS beads, then
25 July 2012
molded, and finally frozen and left for 24 h to form a gel. After being thawed, the precursor thus
Accepted 5 August 2012
obtained is desiccated, decomposed all binders and polystyrenes and sintered. The result is high
Available online 15 August 2012
porosity metal foam that maintains the molded shape. When stainless steel powder is used in this
Keywords: simple process, metal foams with a porosity ranging from 90% to 97%, plateau stress from 4 to 0.4 MPa
Metal foam and thermal conductivity from 0.5 to 0.1 W/mK are produced. Pore size can be controlled by using
Stainless steel
different sizes of EPS beads. This simple process can be used to produce metal foams of high porosity at
Space holder
high throughput. Also, this process can be used to reduce production costs using coarser and cheaper
Mechanical property
Thermal conductivity powder. Such coarse powder foam has lower thermal conductivity, and thus can be used as a thermal
insulator. These low cost metal foams can expand these applications such as filters, thermal insulators,
shock absorbers and acoustic insulators.
& 2012 Elsevier B.V. All rights reserved.

1. Introduction porous titanium or micro stainless parts [10–13], it has difficulty

High porosity metal foams are lightweight, have good thermal
and acoustical insulation properties, and can be produced using one
of several processes [1]. For example, one method produces alumi-
num foams from molten metal by using a foaming agent [2].
Another method coats metal onto a template material to produce
nickel foams by CVD [3] or by galvanizing onto polyurethane foam
[4]. Processes using metal powder have also been developed. Using a
foaming agent and a gelatin of a metal powder slurry produces
higher porosity metal foams compared to conventional metal
powder processes [5,6]. A process for making metal foam using
MHS (Metal Hollow Spheres) has been developed by the Fraunhofer
Institute [7]. MHS is produced by coating metal powder onto EPS
beads and then sintering. Also, aluminum-steel composite foam is
produced using MHS [8,9]. Although all of these above-mentioned
methods can produce metal foams with high porosity, each method
has its own inherent problems. Some require expensive foaming
agents or metal powders, some have limited applicable metals, and
some are currently too complicated for mass production.
An easy, commonly used process involves metal powders with
space holders. Numerous organic and inorganic materials can
be used as the space holders, and are decomposed during the
sintering process. Although this method can be used to produce
producing a high porosity metal foam. We therefore developed a
process that combines the space holder method and the whipped
slurry method to produce metal foam. This process, which we call
the SWS (Space holders and Whipped Slurry) process, can
produce metal foams with porosity exceeding 95%. In this process,
relatively coarse powder can be used, thus reducing the cost of
foams to less than that for foams fabricated by processes that use
powder alone to fabricate metal foams.
Here, we developed a new, low cost process to fabricate high
porosity foams. Then, we used this SWS process to fabricate
stainless steel foams, and evaluated their cell structure, mechan-
ical properties, and thermal conductivity.
2. Experiment
2.1. SWS process to fabricate metal foam
Fig. 1 schematically illustrates a general outline of the SWS
process to fabricate metal foam, and this process uses EPS
(Expanded Polystyrene) as space holder. EPS is the polystyrene beads
foamed using foaming agent and steam heating. It is commonly
used as cushioning or buffer material. The SWS process involves
the following five steps.

n
Corresponding author. Tel.: þ81 29 861 7183; fax: þ 81 29 861 7167. (1) Metal powder is mixed with a PVA (Poly Vinyl Alcohol) water
E-mail address: toru-shimizu@aist.go.jp (T. Shimizu). solution binder and a surfactant to form slurry.

0921-5093/$ - see front matter & 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.msea.2012.08.010
344 T. Shimizu et al. / Materials Science & Engineering A 558 (2012) 343–348

PVA 1) 2) 3) 4) 5)
Water
Solution
Surfactant Debinding & Sintering Furnace
Metal Powder -20°C Refrigerator 60°C Oven
EPS
Beads Molding & Defrosting& Sintering
Slurry Freezing Desiccating

Fig. 1. SWS method for fabricating metal foam, (1) preparation of slurry, (2) mixing of the slurry and EPS beads and whipping, (3) molding and freezing, (4) gelation of
slurry and desiccating, (5) debinding and sintering.

Air Bubble
EPS Space Holder

3) Strut

Cell wall

Fig. 2. Photographs of precursor and sintered metal foam. (1) Precursor and (2) sintered metal foam (3) schematic of foam structure.

The slurry is whipped and mixed with EPS space holders to Table 1
(2)
trap air bubbles and prevent these bubbles from bursting. Concentrations of prepared slurries.
(3) The mixture of slurry and space holders is molded, and then
frozen and left for 24 h to allow the PVA binder to become a No. SUS316 Powder weight/ PVA binder Surfactant EPS beads
powder (g) EPS weight (8 wt%) (ml) (ml) (f0.8 mm)
gel [14].
(4) The obtained precursor is defrosted and the molded shape is 1 400 20 90 10 20 g (670 ml)
retained due to the PVA binder gel. Then the precursor is 2 200 10 90 10 20 g (670 ml)
heated at 333 K for 48 h for solidification by desiccation. 3 120 6.0 90 10 20 g (670 ml
(5) The precursor is heated at 773 K for 2 h to decompose the
space holders and binder. The precursor is then sintered.
1773
Fig. 2 shows (1) photograph of the precursor, (2) photograph of PF20F
sintered foam and (3) schematic of the foam structure. The foam 1423K
becomes the structure composed of the pore made by air bubble 1323K
1273
and the EPS space holder. The pores made by air bubble become
Debind process PF3F
the structure composed by struts. The pores made by ESP are
coated surface by metal powder and they have cell walls. 773

2.2. Fabrication of stainless steel foam using the SWS process

273
0 2 4 6 8 10 12
The SWS process was applied to fabricate stainless steel foam Time/ hr
as follows. Two SUS 316L stainless steel powders were used; PF3F
Fig. 3. Debinding and sintering temperature diagrams of the furnace.
and PF20F (Epson Atmix Co.) with an average particle diameter of
about 3 and 10 mm, respectively. The PVA binder was an 8 wt%
water solution of HA-26PVA (Nippon Gohsei Co.), which has a
polymerization degree of about 2600. This PVA binder becomes a and kept at that temperature for 30 min to bond the powder.
gel as a result of the freezing and thawing operations in the SWS The two stainless steel foams are hereafter called PF3F and PF20F
process. The surfactant was a neutral detergent (Saraya Kagaku foams. They have a skin on the sample surface. A skin is a slurry
Co.). The EPS space holder has particle size distribution from film composed on a mold surface, and this part has high density
0.6 to 1.0 mm, average diameter about 0.85 mm, and a tap density compare to inner part of the foam. Therefore, skin part is avoided
of 30 kg/m3. The metal powder, PVA binder, surfactant and EPS for evaluation of foam properties.
beads were mixed at different concentrations as shown in Table 1.
Debinding, decomposing of EPS and sintering were continuously
processed by heating the precursors in vacuum as shown in Fig. 3. 2.3. Evaluation of stainless steel foams
The precursors were sintered in a vacuum. The binder and EPS
beads were decomposed by slowly heating from room tempera- The cell structures of the SST foams were observed using an
ture to 773 K over a period of 4 h, and then kept at that optical microscope (Olympus BX60M, Japan). To prepare the
temperature for 2 h. The temperature was then increased to cross-sections for photography, the pores in foams were infil-
sintering temperatures 1323 K (PF3F) and 1423 K (PF20F) for 2 h trated with resin.
 T. Shimizu et al. / Materials Science & Engineering A 558 (2012) 343–348 345

The mechanical properties of the SST foams were evaluated 100

using compression tests with a autograph (Shimadsu AGS-10kND, 98 PF20F
Japan). Testing specimens with 25 mm high and 15  15 mm in PF3F
cross-section are cut out using band saw. In the compression test, 96

Porosiy [%]
crosshead speed was 10 mm/min, and a laser distance meter
94
(Keyence LK-080, Japan) was used to measure the compression
distance. 92
Thermal conductivities of the SST foams were measured using
90
the hot disk method [15], in which a hot disk sensor was
sandwiched between two 60  60  30 mm samples of the SST 88
0 10 20 30
foam, as shown in Fig. 4. These tensing samples are cut out using
Powder Weight/EPS Weight
band saw. Used measuring system was hot disk thermal analyzer
(Hot Disk AB, TPS2500, Sweden). Diameter of the sensor is f20, Fig. 5. Porosity of PF20F and PF3F stainless steel foams after sintering.
and sample foams have enough largeness. Therefore, convention
heat transfer outside the samples has no effect for the measure- 60

Volumetric Shrinkage Ratio [%]

ment. An electric current was applied to the sensor, and the
resistance of the sensor element was measured based on the 50
current and voltage. The temperature increase was then esti-
40
mated from the resistance, and the heat generated at the sensor
element was estimated from the current and voltage. The thermal 30
conductivity of the foams was evaluated based on the relation
PF20F
between the heat input and temperature increase. 20
PF3F
10
3. Results and discussion
0
0 10 20 30
3.1. Porosity and shrinkage of the stainless steel foams after
Powder Weight/EPS Weight
sintering
Fig. 6. Volumetric shrinkage of PF20F and PF3F stainless steel foams after sintering.
Table 2 shows the porosity and shrinkage ratio from the
precursor size after sintering. Fig. 5 shows the relation between
the mixing ratio (powder weight/EPS weight) and porosity of the
PF3F and PF20F foams. Fig. 6 shows the relation between mixing 3.2. Cell structure of the stainless steel foams
ratio (powder weight/EPS weight) and volumetric shrinkage ratio
of the PF3F and PF20F foams. Porosities are decreasing with Fig. 7 shows photographs of the cross-sections of the PF3F and
decreasing the mixing ratio, and this tendency is strong in the PF20F foams. Hollow spaces formed by the EPS space holders and
PF3 foam. This reason is that PF3F foams shrink higher ratio small bubbles are clearly evident. At low porosity, EPS space
compare to PF20F foams as shown in Fig. 6. PF3F powder is finer holders are covered by powder and foam remaining cell walls.
than PF20F powder and it is sintered at lower temperature. In this When the porosity increased, volume of cell walls decreases,
method, precursor volume shrinks from 1/2 to 1/3 keeping the struts becomes slender and the cell structure becomes more
molded shape. Also, the shrinkage has torsion deformation and permeable structure. The PF3F foam exhibited well-sintered cell,
the foams need machining or forming for actual industrial usage. whereas the PF20F foam exhibited the starting powder particles
in the sintered struts.

3.3. Mechanical properties of stainless steel foams

Figs. 8 and 9 show the results of the compression test. Also

Measured Hot Di sk
shown in these figures are the bulk densities of the specimens.
Specimens Sensor These bulk densities are lower than the densities shown in Table 2
because the produced foams had skins heavier than the interior
1) Hot disk sensor 2) Measuring method
parts, and these skins were removed during preparation of the
foam specimens for the compression tests. Fig. 10 shows the
Fig. 4. Hot disk method to measure thermal conductivity of stainless steel foams. relation between density and stress. In this figure, both axes are
logarithmic scales, and compression stress was evaluated using
this plateau stress, sp, defined as the average stress between the
20 and 30% strain, and using the nominal stress s0.5 at 50% strain
Table 2
Densities and shrinkage of stainless steel foams. [16–17]. For the comparison, strength of SST foams fabricated
using a GF process (gelation of slurry and foaming involving
No. Sintering Volumetric Porosity Bulk density slurry gelation, a foaming agent, and hydro-gel binder) with PF3F
temperature (K) shrinkage (%) (%) (g/cm3)
powder are shown in Fig. 10. The strength curves are drawn using
PF3F-1 1323 42 90.0 0.79 12 samples these densities are from 0.15 to 0.42 g/cm3 [6]. The
PF3F-2 1323 46 94.0 0.47 compression test results showed that the PF3F SST foam was
PF3F-3 1323 49 96.3 0.29 stronger than the PF20F SST foam due to stronger cell walls and
PF20F-1 1423 34 92.4 0.60 struts. The compression stresses were proportional to density to
PF20F-2 1423 34 95.0 0.40
the 2.6–3.0 power. In general, the foam strength was proportional
PF20F-3 1423 37 96.4 0.28
to density to the 1.5–2.0 power [16], but, the strength of foam
346 T. Shimizu et al. / Materials Science & Engineering A 558 (2012) 343–348

Bubble pore

EPS pore EPS pore EPS pore

Bubble
pore
Bubble
pore

Bubble
pore
Bubble
EPS pore
pore EPS pore EPS
Bubble
pore pore

1mm
Fig. 7. Photographs of cross-sections of PF20F and PF3F stainless steel foams.

15 100
PF3F Strain=50%
PF3F-1 ρ=0.66g/cm3 PF3F Plateau Stress
PF20F Strain=50%
Stress σ [M Pa]

10 PF20F Plateau Stress

GF Process Strain=50%
σ [M Pa]

GF Process Plateau Stress

10
PF3F-3 ρ=0.46g/cm3
5
Compression Stress

Compression Stress
of SST Foam by GF
PF3F-5 ρ=0.30g/cm3 Process
0
0 20 40 60 80 σ 0.5
1.0
Nominal Strain [%]
σp
Fig. 8. Stress–strain curve of PF3F stainless steel foams by compression test.
3
2 σ ∝ ρ
σ∝ ρ

15
0.1
0.2 0.4 0.6 0.8
PF20F-1 ρ=0.58g/cm3
Density ρ [g/cm3]
Stress σ [M Pa]

10
Fig. 10. Relations between density and plateau stress, stress when nominal strain
is 50%.

5
3 PF20F-5
PF20F-3 ρ=0.41g/cm
ρ=0.30g/cm3
0
0 20 40 60 80
Nominal Strain [%]
Fig. 9. Stress–strain curve of PF20F stainless steel foams by compression test.
fabricated by the SWS process is more sensitive to the foam
density compared to the sensitivity of other processed foams. This
higher density dependence is due to the difference in cell
structure between high and low density foams. The PF3F and
PF20F SST foams fabricated here using the SWS process have
closed cell type structure at high density, but have open cell type
structure at low density. In comparison, SST foams fabricated
using a GF process have closed type cell structure and have a
strength proportional to density to the 1.9 power as shown in
Fig. 10. Cell structures of GS process foams are shown in Fig. 11,
and confirmed that they have closed cell type structure. When the
density is high (0.6 to 0.7), the strength of the SST foams
fabricated using the SWS process is higher than that for foams
fabricated using only the GF process, whereas at low density (0.2–
0.3), their strength is lower.
The consumption energy Ec of the foams in Fig. 12 was estimated
from the stress–strain curves in Figs. 7 and 8, integrating the stress by
the strain from 0 to 50% [16,17]. Fig. 11 shows the estimated Ec
together with that of foams by GF process. The Ec of the foams
produced by GF process is linearly proportional to the bulk density of
the foam. However, Ec of the foams by SWS process is increasing
proportional to density to the 1.8 power, because compression stress
is proportional to about density to the 2.6–3.0 power.
3.4. Thermal conductivity of stainless steel foams
Fig. 13 shows the relation between densities and thermal
conductivity of the foams. Thermal conductivity of metal foam
 T. Shimizu et al. / Materials Science & Engineering A 558 (2012) 343–348 347

at small pore size, and conduction lg is only about 0.023 W/mK

[16]. In a metal foam, ls depends almost entirely on thermal
conductivity by the cell structure. Also, ls depends on the relative
density of the foam, and this dependency is comparable to the
electric conductivity of the foam [18]. Therefore, we can use
electric conductivity to estimate the thermal conduction through
1) ρ JFP 2) ρ JFP 3) ρ JFP the solid in a metal foam. The relation between relative electric
1mm conductivity of foams and the relative density is expressed by the
Lemlich formula as follows. [19,20] This formula is also known as
Fig. 11. Consumption energy Ec up to 50% strain as a function of bulk density of Ashby–Glicksman model [21].
the stainless steel foam.


ls =lb ¼ sf =sb ¼ 1=3 ðrstrut =rb Þ þ 2=3 ðrwall =rb Þ ð2Þ
6 where lb is the thermal conductivity of the bulk solid, sf is the
PF3F
electric conductivity of foam, sb is the electric conductivity of
E c= 12.1 ρ 1.85
5 bulk solid, rf is the density of foam and expressed as rf
¼ rstrut þ rwall. The foam is composed of cell walls and struts,
ConsumptionEnergy {J/g]

and rstrut is the density proportional to the struts volume, rface is

4 the density proportional to the cell walls volume, rb is the density
of bulk solid. The thermal conductivity of SUS316 stainless steel,
lSST, is 16.7 W/mK at room temperature, and thus, lf can be
3 expressed as follows:
Consumption energy of
SST Foam by GF

Process
lf ¼ lSST ðsf =sb Þ þ lg þ lc þ lr ¼ 16:7ðsf =sb ÞðW=mKÞ þ0:023 W=mK
2



E c= 6.7 ρ PF20F ¼ 16:7f 1=3 ðrstrut =rb Þ þ 2=3 ðrwall =rb Þg ðW=mKÞ þ 0:023 W=mK
E c= 7.03 ρ 1.81 ð3Þ
1
where rstrut ¼ rf, rwall ¼0 when the foam is open cell structure
and rstrut ¼0, rwall ¼ rf when the foam is closed cell structure.
0 Fig. 13 shows lf estimated using Eq. (3). The estimated lf values
0 0.2 0.4 0.6 0.8
for an open cell structure agree relatively well with the measured
Density [g/cm3] lf values. The PF20F SST foams have lower lf compared to the
PF3F foam, and are lower than the estimated lf. The reason for the
Fig. 12. Consumption energy Ec up to 50% strain as a function of bulk density of
the stainless steel foam. lower lf is that the PF20F foam used a coarse powder and had
numerous voids inside the cell walls and struts. These walls and
struts thus had lower thermal conductivity compared with dense
0.8
stainless steel.
Eq.(3) ᧤Closed Cell, ρ = ρ face)
Thermal Conductivity λ f [W/mK]

Eq.(3) ᧤Open Cell, ρ f = ρedge ) 4. Conclusions

0.6
A simple method, called SWS process, was developed to
fabricate high porosity metal foams (90–97%) from metal powder
slurry using EPS beads as space holders. In this process, the
0.4 precursor shrinks, although the molded shape is maintained after
sintering. Foams with high porosity and improved mechanical
properties can be fabricated by using finer metal powders.
A coarser, cheaper powder, such as PF20F powder, can be used
0.2 PF3F to reduce production costs, and still successfully achieve 96%
porosity. Also, PF20F foam has lower thermal conductivity, and
PF20F thus can be used as a thermal insulator. These metal foams can be
used for many applications such as filters, thermal insulators,
0 shock absorbers and acoustic insulators. Low-cost metal foams
0 0.2 0.4 0.6 0.8 1 can expand these applications.
3
Density ρf [g/cm ]
References
Fig. 13. Measured thermal conductivities of PF3F and PF20F stainless steel
foams, and estimated thermal conductivities of open and closed SUS316 foams
using the Lemlich formula (Eq. (3)). [1] J. Banhart, Prog. Mater. Sci. 46 (2001) 599–632.
[2] T. Miyoshi, M. Itoh, S. Akiyama, A. Kitahara, Metal Foams and Porous Metal
Structures, in: J. Banhart, M.F. Ashby, N.A. Fleck (Eds.), Verlag MIT Publishing,
Berlin, 1999, pp. 125–132.
lf is shown in a following equation. [3] V. Paserin, S. Marcuson, J. Shu, Cellular Metals: Manufacturing Properties

Applications, in: J. Banhart, N. Fleck, A. Mortensen (Eds.), Verlag MIT Publish-
lf ¼ ls þ lg þ lc þ lr ¼ ls þ0:023 W=mK ð1Þ ing, Berlin, 2003, pp. 31–38.
[4] M. Honda, Materia 38 (1999) 470–474.
where ls, lg, lc and lr are conduction through a solid, conduction [5] C.S.Y. Jee, N. Ozguven, Z.X. Guo, J.R.G. Evans, Metall. Mater. Trans. B 31B
through a gas, convection within the cells, and radiation through (2000) 1345–1351.
[6] T. Shimizu, K. Matsuzaki, Porous Metals and Metal Forming Technology, in:
the cell walls, respectively. However, effect of conduction lr is H. Nakajima, N. Kanetake (Eds.), The Japan Institute of Metals, Sendai, 2005,
very small at room temperature, conduction lc can be neglected pp. 191–194.
348 T. Shimizu et al. / Materials Science & Engineering A 558 (2012) 343–348
[7] O. Andersen, U. Waag, L. Schneider, G. Stephani, B. Kieback, Metal Foams and
Porous Metal Structures, in: J. Banhart, M.F. Ashby, N.A. Fleck (Eds.), Verlag
MIT Publishing, Berlin, 1999, pp. 183–188.
[8] A. Rabiei, A.T. O’Neill, Mater. Sci. Eng. A 404 (2005) 159–164.
[9] L.J. Vendra, A. Rabiei, Mater. Sci. Eng. A 465 (2007) 59–67.
[10] C. Stiller, H.P. Buchkremer, R. Joos, D. Stover, H. Nelles, Proceedings of 1998
PM World Congress vol. 5, European Powder Metallurgy Association,
Granada, 1988, pp.189–194.
[11] C.E. Wen, M. Mabuchi, Y. Yamada, K. Shimojima, Y. Chino, T. Asahina, Scr.
Mater. 45 (2001) 1147–1153.
[12] T. Shimizu, K. Matsuzaki, Y. Ohara, J. Jpn. Soc. Powder Powder Metall. 53
(2006) 36–41.
[13] K. Nishiyabu, Porous Metals and Metal Forming Technology, in: H. Nakajima,
N. Kanetake (Eds.), The Japan Institute of Metals, Sendai, 2005, pp. 619–622.
[14] M. Watase, K. Nishinari, Macromol. Chem. Phys. 186 (1985) 1081–1086.
[15] H. Nagai, M. Mamiya, T. Okutani, Jpn. J. Appl. Phys. 46 (2007) 7920.
[16] JIS H 7009 (2008) 1–14.
[17] N. Kanetake, J. Jpn. Soc. Technol. Plast. 50 (2009) 1004–1008.
[18] L.J. Gibson, M.F. Ashby, Cellular Solids: Structure and Properties-Second
Edition, Cambridge University Press, Cambridge, 1997.
[19] D. Weaire, S. Hutzler, The Physics of Foam, Oxford University Press, New York,
1999.
[20] R. Lemlich, J. Colloid Interface Sci. 64 (1978) 107–110.
[21] L.R. Glicksman, Low Density Cellular Plastics, in: N.C. Hilyard, A. Cunningham
(Eds.), Chapman & Hall, London, 1994, pp. 104–152.