Materials Science and Engineering A327 (2002) 128 132 www.elsevier.

com/locate/msea

Solid-state diffusion bonding of closed-cell aluminum foams

K. Kitazono a,*, A. Kitajima b, E. Sato a, J. Matsushita b, K. Kuribayashi a
a b

The Institute of Space and Astronautical Science, 3 -1 -1 Yoshinodai, Sagamihara, Kanagawa 229 -8510, Japan Department of Applied Chemistry, Tokai Uni6ersity, 1117 Kitakaname, Hiratsuka, Kanagawa 259 -1292, Japan Received 17 May 2001; received in revised form 28 June 2001

Abstract Solid-state diffusion bonding (DB) was demonstrated for joining closed-cell aluminum foams (ALPORAS). A superplastic 5083 aluminum alloy sheet was inserted between the foams to assist the DB process. Microscopic observation revealed that the cell wall of the foams penetrated into the 5083 alloy sheet and their boundary partly disappeared. Energy dispersion X-ray spectrometer (EDS) conrmed the diffusion of magnesium element from the 5083 alloy to the aluminum foam regions. The bonding strength was evaluated by four-points bending tests. The obtained exure stress was about 50% of the original foam at room temperature and was more than 60% at 423 K. The advantage of the DB process in the high temperature applications was discussed comparing with the adhesive bonding of aluminum foams. 2002 Elsevier Science B.V. All rights reserved.
Keywords: Aluminum foam; 5083 AlMg alloy; Superplasticity; Diffusion bonding; Bending test

1. Introduction Foamed materials have many attractive properties because of their light weight and cell structure. The advantages of metal foams over the conventional polymer foams are the high melting temperature and the large compressive strength. The manufacturing process of metal foams has been improved in the past few years; especially closed-cell aluminum and its alloy foams have been made by casting [1] and powder metallurgy processes [2]. If we use metal foams in engineering structures, we must consider the methods of joining them to other materials. The most popular joining method of massive metals is welding. Welding process for metal foams, however, seems to have much difculty because of the volume reduction during the solidication of the melted cell structures. Bonding by adhesives or solders has been used in most metal honeycomb structures. In this case, the bonding strength of the adhesives dominates the mechanical properties of the honeycomb structures. Though the adhesive bonding process is superior for low temperature applica-

* Corresponding author. Tel.: + 81-42-759-8264; fax: +81-42-7598461. E-mail address: kitazono@materials.isas.ac.jp (K. Kitazono).

tions, the expansion to high temperature region is limited. In the present study, solid-state diffusion bonding (DB) process is proposed and demonstrated experimentally against aluminum foams. The DB process is an attractive manufacturing method for joining dissimilar metals and for making the complex components [3]. Unlike other joining processes, it preserves the base metal microstructure at the interface. The DB process generally requires large compressive stress to break up oxide layers and to contact the surfaces closely enough in the atomic scale. Metal foams, however, will easily deform plastically under such a large stress condition. Inserting a superplastic material [4] into the boundary will decrease the compressive stress necessary for DB process. Large plastic deformation due to the superplastic forming under relatively low stresses assists the oxide layers to break up and the surfaces to contact. This paper consists of the following. First, the DB condition is discussed based on the high temperature properties of aluminum foams and superplastic aluminum alloy. Second, the DB tests are performed under the estimated conditions. Finally, the results of the DB tests are discussed through microscopic observation and mechanical test. The bonded boundary is observed by optical microscope and analyzed by energy dispersion

0921-5093/02/$ - see front matter 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 1 - 5 0 9 3 ( 0 1 ) 0 1 7 6 6 - X

K. Kitazono et al. / Materials Science and Engineering A327 (2002) 128132

129

X-ray spectrometer (EDS). The bonding strength is examined by the four-points bending test

3. Bonding condition

3.1. High temperature property of ALPORAS foam

2. Materials preparation Closed-cell aluminum foam (trade name ALPORAS [5,6]) was supplied from Shinko Wire Co. ALPORAS foam was fabricated by casting process of pure aluminum with titanium hydride as the foaming agent and calcium to increase the viscosity of the melt. The chemical composition was 1.04 wt.% Ti, 1.61 wt.% Ca and the remaining of aluminum. The cell structure of ALPORAS foam was almost equiaxial and the size was within 3 6 mm in diameter. The relative density was about 0.11, which was calculated from the mass and the apparent volume. Superplasticity has been reported in many types of aluminum alloys and composites. Since 5083 Al Mg alloy [79] is a typical aluminum alloy showing superplasticity, a superplastic 5083 alloy sheet with a thickness of 1 mm, which was supplied from Sky Aluminum Co., was chosen for the experiments [9]. The nominal composition was 4.7 wt.% Mg, 0.65 wt.% Mn, 0.13 wt.% Cr, 0.04 wt.% Fe, 0.04 wt.% Si, 0.03 wt.% Ti, and the remaining of aluminum. The sheet was manufactured by hot rolling and cold rolling followed by recrystallization treatment. Observation by optical microscope revealed that the alloy consisted of ne equiaxial crystal grains with an average size of 12 mm. The precise superplastic properties were reported in Ref. [9]. The DB process must of course be carried out in a high temperature condition, which corresponds to a creep condition. Though the creep behavior of aluminum foam was reported in Ref. [10], it was not the ALPORAS foam. In the case of the ALPORAS foam, the data of the mechanical property contain a more pronounced scatter than those of conventional massive metals or alloys. Therefore, the compression behavior of the ALPORAS foam at high temperature is necessary to estimate the DB condition. Uniaxial compression tests were performed in atmosphere using Shimadzu Autograph testing machine. The temperature condition was maintained by an electric furnace at 773 and 823 K. The displacement of the crosshead was used to determine the overall compressive strain and the displacement rate was held at 0.1 mm min 1. The results of compression tests were plotted in Fig. 1 as nominal stressstrain curves. Two curves are similar to the deformation behavior of other closed-cell foams at room temperature [11]. After the initial increase in stress, a plateau region characterized by small slope appeared in the stressstrain curve. After the extended plateau region, the stressstrain curve gradually enters into the densication region, where the cell walls touch each other, accompanied by a steep increase in stress. The plateau stresses were estimated as 0.35 MPa at 773 K and 0.22 MPa at 823 K. The following DB test must be carried out below these plateau stresses of ALPORAS foam to prevent the fracture of the cell structure.

3.2. Estimation of DB condition

When the external stress | ext is applied to the ALPORAS foam during the bonding test, the magnitude of the local stress applied on the surface of the 5083 alloy sheet becomes much higher than | ext, because the cell wall only attaches to the 5083 alloy sheet. The stereological analysis of the dual phase material describes that the area fraction of the second phase is identical to the volume fraction of the second phase [12]. Therefore, the local stress can be estimated as 1/0.11 of | ext. For example, the uniaxial external stress of 0.2 MPa, which is smaller than the plateau stress, corresponds to the local stress of 1.8 MPa. We can obtain the uniaxial tensile creep data of the present 5083 alloy from gure 1 in the previous report [9], which describes that the ow stress of 1.8 MPa at 823 K yields high strain rate sensitivity (0.4B mB 0.7). The high value of m corresponds to an ideal superplastic condition. The above estimation indicates that the external stress of 0.2 MPa will give the ideal super-

Fig. 1. Results of uniaxial compression test of ALPORAS foam at 823 and 773 K. The crosshead speed is 0.1 mm min 1. The compressive strength is observed as 0.24 and 0.41 MPa at 823 and 773 K, respectively.

130

K. Kitazono et al. / Materials Science and Engineering A327 (2002) 128132

Fig. 2. Schematic illustration of DB apparatus. The inserted 5083 alloy sheet between the foams is locally heated by an induction coil. The temperature is measured through a spot-welded thermocouple onto the sheet.

MPa was uniaxially applied to the ALPORAS foam for 30 min. Bonding strengths of the diffusion-bonded specimens were examined by four-points bending tests with a distance of 30 mm between supporting rods. After DB test, the specimen was machined to a rectangular plate with 38 mm length, 10 mm width and 4 mm height. The 5083 alloy sheet was positioned at the center of the plate. The apparatus was set in Shimadzu Autograph testing machine and the plate was bent with the constant crosshead speed of 0.5 mm min 1. The tests were performed in atmosphere under the conditions of room temperature (298 K) and high temperature (423 K). For comparison, similar bending tests were carried out using original ALPORAS foam and adhesive bonded ALPORAS foams. The adhesives used were commercial epoxy and polyester. Epoxy adhesive is mainly used in room temperature region and polyester in a slightly higher temperature region.

4.2. Microstructural obser6ation

After the DB test, a cross-section of the bonded interface was observed using a stereoscopic micrograph as shown in Fig. 3(a). It was found that the 5083 alloy sheet plastically deformed at the interface and the cell shape of the ALPORAS foams was maintained as before except for at the bonding interface. Fig. 3(b) shows an optical micrograph of the bonded interface. The arrows in the photograph indicate the boundary of two materials. It was found that the cell wall penetrated to the 5083 alloy sheet and a part of their boundaries disappeared. These results suggested that the mutual diffusion was achieved between these materials. To examine the evidence of the mutual diffusion in detail, the composition of Mg element in the area around the bonded interface was measured by an EDAX/EDS system attached on a scanning electron microscope. The original ALPORAS foam does not contain Mg element. After chemical polishing of the cross section, the electron beam with the accelerated voltage of 5 kV was irradiated to the specimen. The mass fraction of Mg element obtained by EDAX standardless quantication was plotted in Fig. 4. It was 5.8 wt.% at 5083 alloy region and gradually decreased to zero in ALPORAS foam region. This result indicates that Mg element diffused from 5083 alloy to ALPORAS foam regions.

plastic deformation of 5083 alloy. In the following section, the DB test is carried out under the condition of the external stress of 0.2 MPa. However, it is noted that the deformation mode during the DB test is similar to indentation creep for 5083 alloy, which is somewhat different from the uniaxial tensile creep. We need the data of indentation creep of the inserted superplastic material for more precise estimation of the DB condition.

4. Diffusion bonding

4.1. Experimental procedure

DB tests were carried out in atmosphere at 773 and 823 K. The 5083 alloy sheet was inserted between two ALPORAS foams. The experimental apparatus is schematically illustrated in Fig. 2. The shape of ALPORAS foams was rectangular with 20 mm height, 10 mm width and 10 mm depth. The surfaces of the ALPORAS foams and 5083 alloy sheet were chemically polished by Keller solution (HF: 2 ml, HCl: 3 ml, HNO3: 5 ml, H2O: 190 ml) just before the experiment. The sheet was locally heated using a one-turn high-frequency induction coil. Rapid heating, namely in 5 min up to the bonding temperature, was carried out in order to prevent the grain growth of 5083 alloy and the oxidation of the bonding surfaces. The temperature was measured and controlled by a spot-welded thermocouple onto the sheet. After the temperature was raised to the experimental condition, a constant stress of 0.2

4.3. Bonding strength

Examples of the relation between the load and the exure at 298 K were shown in Fig. 5. The original ALPORAS foam beaded without fracture. After the maximum load, the bending strength sharply dropped due to the collapse of the cell structure. On the other

K. Kitazono et al. / Materials Science and Engineering A327 (2002) 128132

131

hand, the diffusion-bonded specimen at 773 K was broken at the bonded interface. The exure strength was about a half of the original ALPORAS foam. Bending strength was quantitatively evaluated by the maximum exure stress |f, which is calculated as |f = PL , wh 2 (1)

where P is the maximum applied load, L is the distance between the support rods, w is the width of the specimen and h is the height of the specimen. The exure stresses in various bending tests are shown in Fig. 6. At room temperature as shown in Fig. 6(a), the exure stresses of the original ALPORAS foam and the adhesive-bonded foam by epoxy were almost identical. Relatively low strength was measured on the adhesivebonded specimens by polyester and two kinds of diffu-

Fig. 4. Composition of Mg element measured by EDAX/EDS system. Mg element diffused from 5083 alloy to ALPORAS foam region in the bonding process at 773 K for 30 min.

Fig. 5. Typical result of four-points bending test at room temperature. The uniaxial load was applied with the constant speed of 0.5 mm min 1. Original ALPORAS foam was plastically deformed without fracture and the diffusion-bonded specimen was fractured at the bonded boundary.

Fig. 3. (a) Macroscopic photograph of the diffusion-bonded specimen after bonding test at 773 K for 30 min. (b) Optical micrograph around the diffusion-bonded boundary. Cell wall of ALPORAS foam penetrated into the 5083 alloy sheet.

sion-bonded specimens at 773 and 823 K. The effects of the bonding temperature did not appear in the present experimental results. In the case of high temperature bending test at 423 K as shown in Fig. 6(b), the exure stress of original and adhesive bonded foams signicantly decreased; especially adhesive-bonded specimens by epoxy could not maintain the bonding. On the other hand, the exure stresses of the diffusion-bonded specimens were identical to those measured at room temperature. The relative exure strength of the diffusion-bonded specimen was more than 60% of that of the original foam, but did not reach the original strength. The DB process of aluminum alloys without special techniques is essentially difcult because of the insolu-

132

K. Kitazono et al. / Materials Science and Engineering A327 (2002) 128132

tures. In the present paper, the DB process was rst demonstrated using the closed-cell aluminum foam (ALPORAS). Commercial superplastic 5083 aluminum alloy sheet was inserted between the foams to assist the DB process. EDS analysis showed Mg diffusion from the 5083 alloy to the ALPORAS foam. Flexure strength was evaluated by four-points bending test. The maximum exure stress of diffusion-bonded specimen was half of the ALPORAS foam. Though the adhesive bonded ALPORAS foam showed signicant drops of exure stress, the diffusion bonded specimen showed null decrease of the bending strength. This result indicates that DB is effective at high temperature condition. The obtained bonding strength will be improved by controlling the surfaces and atmosphere conditions to minimize the stable oxide layer at the surfaces of ALPORAS foams and 5083 alloy sheet. We can conclude that the present demonstration of the superplastic DB process is easily expanded to other foamed materials.

Acknowledgements This work was partially funded by Kanagawa Academy of Science and Technology. The authors thank Shinko Wire Co. and Sky Aluminum Co. for providing the materials.

References
Fig. 6. Results of four-points bending tests at 298 K (a) and 423 K (b). Maximum exure stresses are plotted for ALPORAS foam, diffusion-bonded and adhesive-bonded specimens. The epoxy adhesive could not maintain the bonding at 423 K. [1] A.E. Simone, L.J. Gibson, Acta Mater. 46 (1998) 3109. [2] C. Korner, F. Berger, M. Arnold, C. Stadelmann, R.F. Singer, Mater. Sci. Tech. 16 (2000) 781. [3] H. Kokawa, T. Tsuzuki, T. Kuwana, ISIJ Int. 35 (1995) 1291. [4] J.W. Edington, K.N. Melton, C.P. Cutler, Prog. Mater. Sci. 21 (1976) 61 overview. [5] S. Akiyama, et al., United States Patent No. 4713277 (1987). [6] T. Miyoshi, M. Ito, S. Akiyama, A. Kitahara, Adv. Eng. Mater. 2 (2000) 179. [7] R. Verma, P.A. Friedman, A.K. Ghosh, S. Kim, C. Kim, Metall. Trans. 27A (1996) 1889. [8] K. Kannan, C.H. Johnson, C.H. Hamilton, Metall. Trans. 29A (1998) 1211. [9] H. Iwasaki, H. Hosokawa, T. Mori, T. Tagata, K. Higashi, Mater. Sci. Eng. A252 (1998) 199. [10] E.W. Andrews, L.J. Gibson, M.F. Ashby, Acta Mater. 47 (1999) 2853. [11] F. Han, Z. Zhu, J. Gao, Metall. Trans. 29A (1998) 2497. [12] ASTM E 1245 (1995). [13] P.-J. Winkler, in: Proceedings of the International Conference on Superplasticity and Superplastic Forming, TMS, 1988, p. 491.

ble oxide layer. One report described that the relative strength of diffusion-bonded aluminum alloy was 50% of that of the original material [13]. The present experimental result agrees well with the previous report. The advanced DB process in aluminum alloy requires the removal of oxide layer followed by the protection of the surface by coating techniques.

5. Summary Until now, adhesive bonding process has been widely used for joining metal foams to make sandwich struc-