1School of Materials Science and Engineering, Tsinghua University, Beijing, 100084, China2Metals & Chemistry Research Institute, China Academy of Railway Sciences, Beijing, 100081, China3Key Laboratory for Advanced Materials Processing Technology, Ministry of Education, Tsinghua University,Beijing, 100084, China
Aluminum foams are new kind of structural-functional composite materials which comprised of aluminum matrix and gas pores. In thispaper, aluminum foams with homogeneous pore structure made by 6063 aluminum alloy were fabricated by melt foaming method. Ascomparison, the aluminum foams were also fabricated by pure aluminum. Studies on process parameter, microstructure and mechanicalproperties have been carried out. Process optimization for aluminum foams made by 6063 alloys has been studied according to the orthogonalexperimental design method. The cell wall of aluminum foam is comprised of Al matrix, Ca-thickening phase and Ti-containing phase. Thedifference between foams made by 6063 aluminum alloy and pure aluminum is that there is element Mg and element Si dissolved in the matrix inthe 6063 aluminum foam, which are positive to the foam strength accordingly. For different base materials, the compressive strength of foamsmade by 6063 aluminum alloy is greater than that made by pure aluminum. [doi:10.2320/matertrans.M2017300]
(Received September 29, 2017; Accepted January 25, 2018; Published March 2, 2018)
Metal foams are a new kind of materials originatedfrom nature, like wood and bones, which composed ofcontinuous metal phase and separated gas phase. For theirunique structures, metal foams are becoming promisingstructural-functional composite materials. Aluminum foam,the most commonly used metal foam, is light-weight, andhas advantages in energy absorption, noise reduction,electromagnetic shielding, heat insulation, etc.15) The brightapplications in aviation, aerospace, building industry andautomobile industry arouse widespread attention fromresearchers.68)
Melt foaming method is one of the most popularmanufacturing techniques.1,2) The fabricating process iscomplicated and controlled by many factors. So single-factorexperiments for aluminum foams are not enough to figureout all the influences. Orthogonal experimental design, alsonamed Taguchi method, is a kind of experimental method tostudy multi-factor and multi-level experiments, which can beused to improve manufacturing process, optimize or improveefficiency.9,10) Surace et al.11) applied Taguchi method tooptimize the gas injection manufacturing parameters foraluminum foams. Surace et al.12) and Solórzano et al.13) usedTaguchi method to find the optimal process parameters foraluminum foam manufactured by powder metallurgy route.In order to investigate the influences of experimentalparameters’ on aluminum foams, orthogonal experimentaldesign is an excellent choice for reducing the experimentworkload and improving the efficiency correspondingly.
As structural-functional integrative material, the porestructure and mechanical properties of aluminum foams arewidely concerned. However, reports on the microstructure ofcell wall are few. The manufacturing process of melt foamingmethod brings in various compounds which created into
sophisticated phases after reaction at high temperatures.Thus, the sophisticated cell wall microstructure results inthe sophisticated mechanical behaviour. When the porestructure is the same, the different cell wall microstructureaffects the micro deformation mechanism significantly.14)
There was evidence indicating that some kind of phasecan induce property failure, such as Al3Ti15) and Al13Fe.16)
V.K. Jeenager et al.17) developed different microstructuresthrough solutionizing and quenching followed by thermalageing treatment, and the uniaxial compression tests showedthat the solutionized sample had the best energy absorptioncapacity (29.3MPa). Yuan et al.18) prepared aluminum foamsby gas injection method and found that the cell wall propertywas impaired by the defects in cell walls and the oxidefilms on the cell wall surface. Taking these into consideration,the relationship between cell wall microstructure andcompressive property is of vital importance.
Commercial 6063 alloy are widely used in constructionfields for the applications of doors and windows, and largeamount of scraps were produced. Preparing foams byaluminum 6063 scraps is environmental friendly forrecycling and can reduce production cost greatly. Meanwhile,the 6063 aluminum alloy is suitable for foaming due to thesmall amount of element Mg existing in 6063 alloy is justappropriate to reduce the surface tension between pores andmelt.19)
In this study, aluminum foams based on 6063 aluminumalloy were fabricated by melt foaming method. As acomparison, aluminum foams based on commercial purealuminum ingot were also produced. Four manufacturingparameters were performed in the orthogonal experimentaldesign: Ca adding temperature, TiH2 adding temperature,amount of TiH2 and holding time after TiH2 stirring. Whatis more, the microstructure and compressive property offoams made by 6063 aluminum alloy and pure aluminum arecompared.
The base materials used were 6063 aluminum alloy(matrix material, Al, Mg: 0.450.9wt.%, Si: 0.20.6wt.%),commercial pure aluminum ingot (matrix material, purity99.7%), calcium granules (thickening agent, average size 34mm, purity 99.7%), titanium hydride powder (foamingagent, TiH2 < 48 µm, purity 99.4%). TiH2 powder waspre-treated at 500°C for 120 minutes in air to delay itsdecomposition in foaming process.
Aluminum foams were prepared by melt foaming route.(1) Melting: the base materials were melt in an electricalresistance furnace at the temperature of 700°C; (2)Thickening: put 2wt.% calcium granules into the melt atthe temperature from 660°C to 700°C and stirred the mixturemechanically for about 10 minutes; (3) Stirring: put TiH2
into the melt and stirred the melt for about 6 minutes; (4)Foaming: the melt was preserved heat in the furnace forfoaming; (5) Cooling: aluminum foam was taken out in theair.
Orthogonal experimental design was applied. Four factorsand three levels were selected to set up a L934 orthogonalexperiments. The factors and levels were shown in Table 1.The primary and secondary of the factors and the bestcombination of them are obtained according to rangeanalysis. The range value is derived from the differencebetween the maximum value and the minimum value ofevery level. The high range value demonstrates the factorinfluences the mearurable variable greatly. Porosity, pore sizeand uniformity were the measurable variables to characterisethe pore structure of aluminum foams. Porosity ª of foamswere calculated by eq. (1), where µr is the relative density, µ*is the foam density, µs is the density of corresponding densematerial, which is the density of pure aluminum, 2.7 g·cm¹3,M is the foam weight, and V is the foam volume. Pore sizewas calculated by Image-Pro Plus 6.0.19) The uniformity ofpore size was represented by the slope value of the porediamter accumulative fraction curves. The specific method isas follows: select the scatters ranged from 5% to 95% ofthe pore diamter accumulative fraction and make a linearfitting, then the slope value of the fitting line is that of theaccumulative fraction curves. The range from 5% to 95% isthe estimation interval of the population parameter consistedof pore size value, which indicates that the probability of thetrue value of this parameter to fall around the measurementresults is 95%. As a result, a high slope value means a goodpore size uniformity.
ª ¼ ð1� µrÞ ¼ 1� µ�
µs
� �¼ 1� M
Vµs
� �ð1Þ
Microstructure was investigated on a scanning electronmicroscope (SEM, Zeiss MERLIN-VP-COMPACT,accelerating voltage = 15 kV) with the energy dispersive
spectrometer (EDS, Oxford INCA). The quasi-staticcompressive tests were carried on a WDW computer-controlled electronic universal testing machine (SinotestEquipment Co., Ltd., Model: WDW-50) with the loadingrate of 5mm/min.
3. Results and Discussions
3.1 Orthogonal experimental design analysisTable 2 shows the L934 orthogonal array and the
measurable variables of the foams accordingly. Figure 1shows the longitudinal sections of the foams. The pore sizedistribution and its accumulative fraction are shown in Fig. 2,which indicates that foam 9 has the smallest pore size andthe best uniformity.
The range analysis value is shown in Table 3. Porosityrange value ranks as RB > RA > RD > RC which demonstratesthat porosity is affected by TiH2 adding temperature most,and the followings are Ca adding temperature, holding timeafter TiH2 stirring, amount of TiH2. The optimal combinationof parameters to get the highest porosity is A2B1C3D2, whichis adding calcium at 680°C, adding 0.6wt.% TiH2 at 650°C,and cooling after keeping foaming in the furnace for 5min.Pore size range value ranks as RB > RC > RA > RD whichdemonstrates that pore size is affected by TiH2 addingtemperature most, and followed by the amount of TiH2, Caadding temperature, holding time after TiH2 stirring. Theoptimal combination of parameters to get the smallest poresize is A2B1C3D2, which is adding calcium at 660°C, adding0.4wt.% TiH2 at 670°C, and cooling after keeping foamingfor 7min. Uniformity value ranks as RB > RA > RD > RC,which demonstrates that uniformity is affected by TiH2
adding temperature most, and the followings are Ca addingtemperature, holding time after TiH2 stirring, amount of TiH2.The optimal combination of parameters to get the bestuniformity is A1B3C2D3, which is adding calcium at 660°C,adding 0.5wt.% TiH2 at 670°C, and cooling after keepingfoaming for 7min.
As illustrated in Fig. 3, the relationship between eachfactor and the measurable variable is obviously presented.Showing in Fig. 3(a), with the increasing of Ca addingtemperature, the porosity and the pore size increase first andthen decrease, and the uniformity of the foams goes theopposite. When adding Ca at 660°C the reaction in the melt isnot much severe than adding Ca at 680°C and 700°C. Thus,the thickening phase is less and the melt viscosity is low,which makes the foaming agent easier to escape out of the
Table 1 Factor-level table of orthogonal experimental design.
Table 2 L934 orthogonal array.
T. Shi et al.626
Fig. 1 Longitudinal sections of aluminum foams.
Table 3 Range analysis value.
Microstructure and Compressive Properties of Aluminum Foams Made by 6063 Aluminum Alloy and Pure Aluminum 627
melt. Then the porosity and the pore size of the foam aresmaller. When adding Ca at 700°C, the thorough mixing ofCalcium and the melt helps produce more thickening agent,which makes the melt move hard and lets the foaming agentget together to burn out. Thus, there is less TiH2 left and thefoam porosity and the pore size are small. When adding Ca at680°C, the temperature is appropriate for thickening, which
makes a high foam porosity and a large pore size. Generallyspeaking, if the pore size is large, the pore uniformity is bad.It is illustrated in the variation tendency of uniformity inFig. 3(a).
As shown in Fig. 3(b), with the increasing of TiH2 addingtemperature, the porosity value and the pore size valueincrease, and the uniformity value goes the opposite. When
Fig. 2 Pore size distribution and its accumulative fraction in orthogonal experimental design.
T. Shi et al.628
adding TiH2 at a temperature higher than the meltingpoint, the decomposition rate goes up with the addingtemperature.20) Then the undecomposed TiH2 left in themelt becomes less, which makes the porosity and pore sizesmaller. So the uniformity of the foams goes bettercorrespondingly.
It can be seen from Fig. 3(c), with the increase of amountof TiH2, the pore size value increases. Less TiH2 makes thepore size smaller and the cell wall thicker. Too much TiH2
prompts pores crack and combine together to form big-sizedpores and thin cell walls. With the increase of TiH2 amount,pore porosity goes down just a little, which indicates that thechange in amount of TiH2 at a thousandth level has littleinfluence on porosity. The uniformity of the foams has a peakvalue with the increase of amount of TiH2, which indicatesthat there is a best stirring condition for different amount offoaming agent in the melt.
As everyone can be seen from Fig. 3(d), with theincreasing of holding time after TiH2 stirring, the porosityvalue and the pore size value increase first and then decrease,and the uniformity value goes the opposite. During theprocess of foaming, the foam experiences rapid expansionstage, slow expansion stage, stabilisation stage, and shrinkingstage in turn. The most appropriate holding time is stoppingheat preservation just before the foam expands to the highestlevel. When the holding time is from 5min to 6min, thehydrogen released from the TiH2 spreads out fully, and thepores can have a long time to grow up. This situationbrings about high-porosity foam with homogeneous pores.Prolonging the holding time after TiH2 stirring continuously,the pores will crack and incur defects as local macro pores,collapse, and crispy cell walls.
3.2 MicrostructureSEM microstructure and EDS energy spectra of 6063
alloy, re-melting 6063 alloy, re-melting 6063 alloy with2wt.% calcium are shown in Fig. 4.
Morphology of as-received 6063 alloy in Fig. 4(a) revealsthat the grains are fine, granular and homogeneous distributedwithout segregation which goes against subsequent process.EDS analysis (Fig. 4(b)) indicates the bright white acicularstructure at Point 1 is AlFeSi phase (¢-Al8Fe2Si), and thedark gray is the matrix phase. The solution of Fe in liquid Al
is relatively high (700°C, 2.5wt.%; 800°C, 5wt.%), so Fe isinevitable in aluminum alloys.21) However, the solution of Fein solid Al is low (655°C, 0.03wt.%), so Fe precipitates outto forming the bright white acicular structure.21)
After re-melting, 6063 alloy has been fully homogenized,and the bright write structure (Point 2 in Fig. 4(c)) gatheredat the crystal boundary which is in the form of network. Themicrostructure of re-melting 6063 alloy is similar to that of6063 alloy. The difference is that there is less bright writestructure in the re-melting alloy. That is because the AlFeSiphase (¢-Al8Fe2Si) is dissolved in the melt gradually after re-melting, which leads to a decrease in amount.22) In analysis,there is no big bulk Mg2Si gathered in, and the Mg element insolid solution is homogeneous distributed. In addition, the Sielement is partially distributed in solid solution of the matrixand partially exists in AlFeSi phase. The Fe element ismainly in AlFeSi phase.
The microstructure of the re-melting 6063 alloy with2wt.% calcium is illustrated in Fig. 4(e). The compoundof Ca (Point 3), which is the thickening phase, precipitatesalong the dendritic network boundary of eutectic Si inthe ¡-Al matrix. Besides, there exists some write potdiscontinuously in the dendritic network, which is provedto be the Fe impurity phase according to the EDS inFig. 4(g), which phenomenon indicates that some hard brittle
660 680 7000
10
20
30
40
50
60
70
80
90
Porosity (%)Pore Size (mm)Uniformity
Ca Adding Temperature (°C)
Poro
sity
(%)
0.00.51.01.52.02.53.03.54.04.55.0
(a)
0
10
20
30
40
50
60
70
80
Uni
form
ity650 655 660 665 670 675
0
10
20
30
40
50
60
70
80
90
Porosity (%)Pore Size (mm)Uniformity
TiH2 Adding Temperature(°C)
Poro
sity
(%)
0.00.51.01.52.02.53.03.54.04.55.0
(b)
0
10
20
30
40
50
60
70
80
Uni
form
ity
0.4 0.5 0.60
10
20
30
40
50
60
70
80
90
Poro
sity
(%)
Porosity (%)Pore Size (mm)Uniformity
Amount of TiH2(%)
0.00.51.01.52.02.53.03.54.04.55.0
(c)
0
10
20
30
40
50
60
70
80
Uni
form
ity
5 6 70
10
20
30
40
50
60
70
80
90
Holding Time (min)
Poro
sity
(%)
Porosity (%)Pore Size (mm)Uniformity
0.00.51.01.52.02.53.03.54.04.55.0
(d)
0
10
20
30
40
50
60
70
80
Uni
form
ity
Fig. 3 Influence of the four factors on the measurable variables.
0 1 2 3 4 5 6 7 8 9
FeFeSiFe
Cnt
Energy(keV)
Al(b)
0 1 2 3 4 5 6 7 8 9
FeFeSiFe
Cnt
Energy(keV)
Al(d)
0 1 2 3 4 5 6 7 8 9
Ca CaSiMg
Cnt
Energy (keV)
Al(f)
0 1 2 3 4 5 6 7 8 9
MgCa CaCa
Al
Cnt
Energy(keV)
FeSi Fe Fe
(g)
0 1 2 3 4 5
Mg Si
Al
Cnt
Energy(keV)
(h)
Fig. 4 Microstructure and EDS spectra of 6063 alloy.
Microstructure and Compressive Properties of Aluminum Foams Made by 6063 Aluminum Alloy and Pure Aluminum 629
plate-like ¢-Al8Fe2Si transformed into Chinese-script ¡-Al5FeSi. When the calcium granules were added into themelt, they will have a fully reaction with the melt and releaseheat during the stirring, which makes the melt temperature goup to about 900°C. Literature 14 reported that the superheatfor the melt helps promote forming Chinese-script ¡-Al5FeSi.23) The plastic processing property of monoclinic¢-Al8Fe2Si is better than that of bcc-structured ¡-Al5FeSi.The EDS analysis of the matrix (Point 5) is shown inFig. 4(h) with a small quantity of Si element and Mgelement, which are dissolved in the Al matrix.
Figure 5 is the microstructure of the cell wall of foamsmade by 6063 aluminum alloy. The bright write acicularstructure (Point 6) is iron-riched phase with the EDSspectrum of Al-Fe-Si. The morphology of iron-riched phaseis shown in Fig. 5(b). Due to the foaming crucible is made ofcarbon steel, some Fe element is dissolved into Al melt toincrease the Fe content in the foam cell wall to some extentafter rapid stirring in high temperature. Another phase in thecell wall is the Ca-containing phase with a gray networkstructure (Point 7). The morphology of the gray phase isAl-Ca-Mg-Si compound which is indicated by the EDSspectrum is shown in Fig. 5(d). There also exists some writepolygonal phase (Point 8) which shown in Fig. 5(f ) in the Almatrix regarded as a kind of oxide Al-O-Ca-Mg according to
the EDS analysis. In addition, for the blowing agent is addedinto the melt to react and release hydrogen, so the Ti elementis left behind to form Al-Ca-Ti compound (Point 9). Suchcompound is Al-Mg-Ti-Ca bearing phase, which tends togather together near the pores in the form of bulk shapeshown in Fig. 5(h) by the EDS spectrum indicated.
SEM microstructure and EDS energy spectra ofcommercial pure aluminum ingot, re-melting pure aluminum,re-melting pure aluminum with 2wt.% calcium are shownin Fig. 6. The gains are homogenously distributed incommercial pure aluminum ingot. The EDS spectrum inFig. 6(b) shows the bright write acicular structure is an iron-riched phase. Microstructure of re-melting pure aluminumand its EDS spectrum are shown in Fig. 6(c), (d). Afteradding the thickening agent, Ca phase is precipitated alongthe eutectic Si network boundary.
Table 4 presents the contents of 6063 aluminum alloy andcommercial pure aluminum. The major elements are Al, Si,Mg and Al, respectively. Compared with 6063 aluminumalloy, there are less Si, Mg and Fe in pure aluminum than6063 alloy. Thus, there is less acicular iron-riched phase inpure aluminum. The network of eutectic Si in pure aluminumis intermittent because of the low Si content.
Figure 7 is the microstructure of the cell wall of purealuminum foam. The bright write phase is iron-riched
0 1 2 3 4 5 6 7 8 9
CaCa
FeFeFe
Cnt
Energy (keV)
Al(c)
0 1 2 3 4 5 6 7 8 9
Ca CaMg Si
Cnt
Energy (keV)
Al(d)
0 1 2 3 4 5 6 7 8 9
OCa CaMg
Cnt
Energy (keV)
Al(f)
0 1 2 3 4 5 6 7 8 9
CaTi TiCaMg
Cnt
Energy (keV)
Al(h)
Fig. 5 Microstructure and EDS spectra of the cell wall of foams made by6063 aluminum alloy.
0 1 2 3 4 5 6 7 8 9
FeFeFe
Cnt
Energy (keV)
Al(b)
0 1 2 3 4 5 6 7 8 9
FeFeFeSi
Cnt
Energy (keV)
Al(d)
0 1 2 3 4 5 6 7 8 9
CaCa
Ca Si
Cnt
Energy (keV)
Al(f)
0 1 2 3 4 5 6 7 8 9
CaCaCa FeFeFe
Cnt
Energy (keV)
Al(g)
0 1 2 3 4 5 6 7 8 9C
ntEnergy (keV)
Al(h)
Fig. 6 Microstructure and EDS spectra of pure aluminum.
T. Shi et al.630
compound (Point 6), the gray network compound is Ca-containing phase (Point 7), and the dark gray structure is thematrix phase (Point 9). After foaming, the remanets of theblowing agent TiH2 are observed in Al matrix (Point 8),which is shown in Fig. 7(b).
The difference between foams made by 6063 aluminumalloy and pure aluminum is that there is element Mg andelement Si dissolved in the matrix in the 6063 aluminumfoam, which are positive to the foam strength accordingly.
3.3 Compression behaviorBoth pore structure and matrix property affect the
compressive behaviour of aluminum foams greatly. Porestructure is determined by the relative density, while thematrix property is determined by the matrix material.According to the eq. (2) that is given by Gibson and Ashby,the relationship between the plateau stress of aluminum foamand the relative density is presented as follows:24)
·�pl
·ys
¼ Cµ�
µs
� �n
ð2Þ
where ·�pl and µ* are the foam’s plateau stress and foam’s
density, ·ys and µs are the strength and density of basematerial, C and n are coefficients which are related to base
material. Generally speaking, the range of coefficient n isfrom 1.5 to 2.25) With the increase of the relative density andthe decrease of porosity, the plateau stress of aluminum foamincreases.
The plateau stress ·�pl is calculated by the following
equation:26)
·�pl ¼
Z ¾cd
¾y
·ð¾Þd¾
¾cd � ¾yð3Þ
where ¾y is the strain of the upper yield stress, ¾cd is the strainof the maximum energy absorption efficiency, which can becalculated by the following two equations:27)
where the solution ¾ from the equations above is ¾cd.In this paper, ·ys refers to the compressive strength of the
solid base material combined with the thickening agent.The adding amount of blowing agent is not too much, soits influence can be ignored comparing with the solid basematerial. The compressive strain-stress curves of the solidbase materials with 2wt.% Ca are shown in Fig. 8, whichimplies that the compressive strengths of the 6063 alloy with2wt.% calcium and pure aluminum with 2wt.% calcium are111MPa and 100MPa, respectively. Due to the hole-defectcomes from the casting process, the real compressive strengthshould be a mixture interaction of initial test value (Fig. 8)and the porosity of the solid material. After correction, thereal compressive strengths of the 6063 alloy with 2wt.%calcium and pure aluminum with 2wt.% calcium are about148MPa and 127MPa, with the porosity of about 25% and21%.
Figure 9 shows the plateau stress of 6063 aluminum foamand pure aluminum foam, and their fitting curves. Whenthe relative density is the same, the plateau stress of foamsmade by 6063 aluminum alloy is bigger than that by purealuminum, which indicates that the compressive strength offoams made by 6063 aluminum alloy is better than that bypure aluminum. The equations of the fitting curves are shownin Table 5.
Table 4 Contents of 6063 aluminum alloy and commercial pure aluminum.
0 1 2 3 4 5 6 7 8 9
FeFeFeCa
Ca
Cnt
Energy (keV)
Al(c)
0 1 2 3 4 5 6 7 8 9
CaCa
Cnt
Energy (keV)
Al(d)
0 1 2 3 4 5 6 7 8 9
TiTiTiCa Ca
Cnt
Energy (keV)
Al(e)
Fig. 7 Microstructure and EDS spectra of the cell wall of foams made bypure aluminum.
0.0 0.2 0.40
100
200
300
Stre
ss (M
Pa)
Strain
6063+2wt.%Capure aluminum+2wt.%Ca
Fig. 8 True compressive stain-stress curves.
Microstructure and Compressive Properties of Aluminum Foams Made by 6063 Aluminum Alloy and Pure Aluminum 631
Actually, there are some differences between the fittingcurves in Fig. 9 and traditional Gibson-Ashby model (n =1.5): The plateau stress by the traditional model is higherin the low-relative density region, and lower in the high-relative density region than the practical fitting curves. Thedifferences are caused by the pore geometrical characteristicand deformation mechanism. In the traditional model, asingle pore is regarded as polygon under ideal conditions,27)
which is shown in Fig. 10(a). However, defects like cracksand holes in the cell wall of the foam with a low relativedensity reduce the strength greatly. The pore tends to be inspherical shape when the relative density of foam aluminumis high (Fig. 1(r)), so that the deformation mechanism ofthe foam in the traditional model is no longer not applicable.The cell walls of the foams are short and squat that they yieldaxially before they bend.24) Model with small pore size28) ismore applicable (Fig. 10(b)) for the aluminum foam arethought of as the solid materials with holes in them. Then theplateau stress-relative density curves should be segment fittedto present the characteristics of foams with different densityin Fig. 11. The equations of the fitting curves are shown inTable 6. When the relative density is lower than 0.2, thecoefficient C in the fitting curve is lower than 1, and the
coefficient n is approximate to 1.5. When the relative densityis higher than 0.2, the coefficient C in the fitting curve ishigher than 1, and the coefficient n is approximate to 2.According to the literature,27) the variation of the coefficientsC and n can be derived from two mainly aspects: thegeometry of the foams, which means the geometry differs indifferent densities; or the value of ·ys is rarely known withprecision, which varies from the contents change broughtabout by the foaming agent, the cell wall oxidation, and otherunknown factors.
According to Lu et al.,29) the plateau stress is governed byone of three different possible failure mechanisms of thefoam cells: buckling, collapse or fracture. The power of theequations of the fitting curves is between 1.5 and 2 for therelative density of the foams lower than 0.2 reflecting thecombined effect of buckling and collapse of the cell walls.While for aluminum foams with relatively thick cell walls,which means that the relative density of the foams is higherthan 0.2, collapse of these walls will be the mechanismgoverning the plateau stress, so the power of the equations ofthe fitting curves is greater than 2.
4. Conclusion
(1) The optimized fabrication process has been studiedaccording to the orthogonal experimental designmethod. For the maximum porosity and the maximumpore size, the optimized process is adding Ca at 680°C,adding TiH2 0.6wt.% at 650°C, foaming in the cruciblefor 6min, and cooling in the air. For the minimumporosity and the best uniformity, the optimized processis adding Ca at 660°C, adding TiH2 0.5wt.% at 670°C,foaming in the crucible for 7min, and cooling in the air.For the minimum pore size, the optimized process is
0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.400
5
10
15
20
25
30foams made by 6063 alloyfoams made by pure aluminum
fitting curves
Plat
eau
Stre
ss (M
Pa)
Relative Density
Fig. 9 Plateau stress-relative density fitting curves of foams made by 6063aluminum alloy and pure aluminum.
Table 5 Fitting equations of Fig. 9.
(b)(a)
Fig. 10 Single-pore force model of aluminum foam.
0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.400
5
10
15
20
25
30
ρ∗/ρs>0.20
ρ∗/ρs<0.20
foams made by 6063 alloyfoams made by 6063 alloyfoams made by pure aluminum fitting curves
Plat
eau
Stre
ss (M
Pa)
Relative Density
Fig. 11 Segmental plateau stress-relative density fitting curves of foamsmade by 6063 aluminum alloy and pure aluminum.
Table 6 Fitting equations of Fig. 11.
T. Shi et al.632
adding Ca at 660°C, adding TiH2 0.4wt.% at 670°C,foaming in the crucible for 7min, and cooling in theair.
(2) According to the range value analysis, porosity, poresize, and uniformity are affected by TiH2 addingtemperature most for the four factors of TiH2 addingtemperature, Ca adding temperature, holding time afterTiH2 stirring, and amount of TiH2.
(3) The cell wall of aluminum foam made by 6063aluminum alloy is comprised of Al matrix, Ca-thickening phase, Ti-containing phase, and some iron-riched phase and oxide. The EDS spectrum shows thatthere are Mg element and Si element existed in most ofthe cell wall phases. The cell wall of aluminum foammade by pure aluminum is comprised of Al matrix,Ca-thickening phase, Ti-containing phase, and someiron-riched phase. The difference between the two kindsof foams is that there is element Mg and element Sidissolved in the matrix in the 6063 aluminum foam,which are positive to the foam strength accordingly.
(4) Quasi-static compressive tests resultes shown that therelationship between the plateau stress and the relativedensity follows the fitting curves of ·�
pl=·ys ¼0:78ðµ�=µsÞ1:63 when µ�=µs < 0:2, ·�
pl=·ys ¼1:49ðµ�=µsÞ2:10 when µ�=µs > 0:2 for foams madeby 6063 alloy, and ·�
pl=·ys ¼ 0:82ðµ�=µsÞ1:73 whenµ�=µs < 0:2 for foams made by pure aluminum. Whenthe relative density is lower than 0.2, the coefficient Cin the fitting curve is lower than 1, and the coefficientn is approximate to 1.5. When the relative density ishigher than 0.2, the coefficient C in the fitting curve ishigher than 1, and the coefficient n is approximate to 2.For different base materials, the compressive strength offoams made by 6063 aluminum alloy is better than thatmade by pure aluminum.
REFERENCES
1) T. Miyoshi, M. Itoh, S. Akiyama and A. Kitahara: Adv. Eng. Mater. 2(2000) 179183.
4) Y. Cheng, Y.X. Li, X. Chen, T. Shi, Z.Y. Liu and N.Z. Wang: Metall.Mater. Trans. B 48 (2017) 754762.
5) J.Y. Yuan, Y.X. Li, N.Z. Wang, Y. Cheng and X. Chen: Metall. Mater.Trans. B 47 (2016) 16491660.
6) H. Nakajima: Prog. Mater. Sci. 52 (2007) 10911173.7) J. Banhart: Prog. Mater. Sci. 46 (2001) 559632.8) A. Soloki and M. Esmailian: Metall. Mater. Trans. B 46 (2015) 1052
1057.9) G. Taguchi and S. Konishi: Taguchi Methods: Orthogonal Arrays and
Linear Graphs, (American Supplier Institute, USA, 1987).10) P.J. Ross: Taguchi Techniques for Quality Engineering, (McGraw-Hill,
USA, 1996) p. 43.11) R. Surace, S. Bruno, L.A.C. De Filippis and A.D. Ludovico: Int. J.
Simul. Model. 8 (2009) 8189.12) R. Surace, L.A.C. De Filippis, A.D. Ludovico and G. Boghetich: Int. J.
Mater. Form. 3 (2010) 15.13) E. Solórzano, J.A. Reglero, J.M.A. Rodríguez-Pérez, J.A. Saja and
M.L. Rodríguez-Mendez: J. Mater. Sci. 42 (2007) 72277238.14) A.E. Markaki and T.W. Clyne: Acta Mater. 49 (2001) 16771686.15) P.R. Onck, R.V. Merkerk, A. Raaijmakers and J.T.M.D. Hosson:
J. Mater. Sci. 40 (2005) 58215828.16) R.E. Raj and B.S.S. Daniel: J. Alloys Compd. 467 (2009) 550556.17) V.K. Jeenager and V. Pancholi: Mater. Des. 56 (2014) 454459.18) J.Y. Yuan and Y.X. Li: T. Nonferr. Metal Soc. 25 (2015) 16191625.19) T. Shi, X. Chen, Y. Cheng and Y.X. Li: Mater. Trans. 58 (2017) 243
248.20) Z.L. Song, J.S. Zhu, L.Q. Ma and D.P. He: Mater. Sci. Eng. A 298
(2001) 137143.21) H.W.L. Phillips: Annotated Equilibrium Diagrams of Some Aluminum
Alloy Systems, (Institute of Metals, London, UK, 1959).22) X.Y. Zheng, F. Jiang, X.L. Jin, W. Chen, Y.Z. Chen and Y.P. Xu: Light
Alloy Fabrication Tech. 40 (2012) 2124.23) Y. Awano and Y. Shimizu: AFS Trans. 98 (1990) 889895.24) L.J. Gibson and M.F. Ashby: Cellular Solids: Structure and Properties,
(Press Syndicate of the University of Cambridge, Cambridge, UK,1997) pp. 175231.
25) M.F. Ashby, A.G. Evans, N.A. Fleck, L.G. Gibson, J.W. Hutchinsonand H.N.G. Wadley: Metal Foams: A Design Guide, (ISBN 0-7506-7219-6, Butterworth-Heinemann, Oxford, UK, 2000) p. 251.
