- Email: [email protected]
Geometry anisotropy and mechanical property isotropy in titanium foam fabricated by replica impregnation method
Materials Science & Engineering A 655 (2016) 388–395
Contents lists available at ScienceDirect
Materials Science & Engineering A journal homepage: www.elsevier.com/locate/msea
Geometry anisotropy and mechanical property isotropy in titanium foam fabricated by replica impregnation method Anchalee Manonukul a,n, Pathompoom Srikudvien a, Makiko Tange b, Chedtha Puncreobutr c a National Metal and Materials Technology Center (MTEC), National Science and Technology Development Agency (NSTDA), 114 Thailand Science Park, Paholyothin Rd., Klong 1, Klong Luang, Pathumthani 12120, Thailand b Taisei Kogyo (Thailand) Co., Ltd., Room INC2d-409, Innovation Cluster 2 Building, Tower D, 141 Thailand Science Park, Paholyothin Rd., Klong 1, Klong Luang, Pathumthani 12120, Thailand c Department of Metallurgical Engineering, Faculty of Engineering, Chulalongkorn University, Pathumwan, Bangkok 10330, Thailand
art ic l e i nf o
a b s t r a c t
Article history: Received 30 November 2015 Received in revised form 6 January 2016 Accepted 6 January 2016 Available online 7 January 2016
Polyurethane (PU) foams have both geometry and mechanical property anisotropy. Metal foams, which are manufacturing by investment casting or melt deposition method and using PU foam as a template, also have mechanical property anisotropy. This work studied the mechanical properties in two directions of titanium foam with four different cell sizes fabricated using the replica impregnation method. The two directions are (1) the loading direction parallel to the foaming direction where the cells are elongated (EL direction) and (2) the loading direction perpendicular to the foaming direction where the cell are equiaxed (EQ direction). The results show that the compression responses for both EL and EQ directions are isotropy. Micrographs and X-ray micro-computed tomography show that the degree of geometry anisotropy is not strong enough to results in mechanical property anisotropy. & 2016 Elsevier B.V. All rights reserved.
Keywords: Powder metallurgy Titanium foam Mechanical properties Isotropy anisotropy
1. Introduction Reticulated polyurethane (PU) foams are processed using reaction of chemicals, which are carried along a conveyer belt for continuous production. This is followed by the peeling of parallelepipedic PU foam blocks to obtain a periodic variation of the open-cell structure [1]. The chemical reaction causes foaming in the vertical direction, hence the cells are elongated in the foam rising direction and thus geometrically anisotropic. Mechanical property anisotropy is also confirmed by experimental and modelling of open-cell PU foam [2]. The compression strength and stiffness of PU foams loading in the foaming direction is higher than in the rolling direction or the transverse direction [3,4]. X-ray micro-computed tomography (μCT) of PU foam shows that cell anisotropy decreases with a higher apparent density [3,4]. In addition, mechanical property anisotropy decreases as the aspect ratio of cell decreases and becomes indifferent as the aspect ratio of cell is less than 1.2 for PU foam [3]. Metal foams can be manufactured by many methods. Open-cell aluminium foam (e.g. Duocels and m.pores) are produced using investment casting with PU foam as a template, while nickel foam n
Corresponding author. E-mail address: [email protected] (A. Manonukul).
http://dx.doi.org/10.1016/j.msea.2016.01.017 0921-5093/& 2016 Elsevier B.V. All rights reserved.
(e.g. Incofoams) uses melt deposition on PU foam [5]. Both metal foams have perfect replica of PU foam template, hence geometry and mechanical property anisotropy are expected to present and have been reported, for example, for aluminium foam [6,7] and for nickel foam [1,4]. The load bearing capability of aluminium and nickel foam loading in the direction parallel to the foaming direction is higher than those loading in the traversed direction [1,4,6,7]. For titanium foam, powder metallurgy is the common manufacturing process. Open-cell titanium foam with low to medium porosity can be produced by powder space holder technique [8– 10]. Although the powder space holder might initially be relatively equiaxed, if the preforms are compacted or hot pressed, pores in metal foam will be elongated perpendicular to the compaction or pressing direction because the powder space holder became flatten [10,11]. This results in geometry anisotropy and effectively mechanical property anisotropy after sintering. For titanium foams fabricated using compaction with powder space holder technique, the titanium foams are slightly stronger perpendicular to the compaction direction and slightly weaker along the compaction axis [9]. The difference in mechanical property anisotropy increases as the degree of geometric anisotropy increases for metal foams [1,12]. On the other hand, if the performs are formed using metal injection moulding method, pores in metal foam will remain equiaxed [13].
A. Manonukul et al. / Materials Science & Engineering A 655 (2016) 388–395
Titanium foam can also be fabricated by replica impregnation method. For this manufacturing method, a PU foam is used as a sacrificial template. PU foams are dipped in titanium slurry and subsequently burnt out. Titanium powders were then sintered. Titanium foams should replicate the structure of corresponding PU foams. There is no previous report on the geometry and mechanical property anisotropy of open-cell titanium foam fabricated by replica impregnation method. Since PU foams have both geometry and mechanical property anisotropy, the objective of this work is to investigate if titanium foams retain both geometry and mechanical property anisotropy. Titanium foams with four different cell sizes were produced and monotonic and interrupted compression tests were carried out using titanium foams in two different loading directions.
389
Table 1 Chemical compositions of commercially pure titanium (wt%). Fe
H
N
C
O
Ti
0.033
0.005
0.009
0.005
0.113
Balance
2. Experimental procedures 2.1. Titanium foam preparation Argon gas atomised commercially pure (CP) titanium powder was used in this work. From SEM observation, the powders were spherical [13]. Table 1, which shows the chemical composition of the CP titanium powder and traces of Fe, H, N, C and O were observed. Manufacturer’s specification of powder size was o45 mm. From a powder size analysis, the measured average powder size was 22.94 mm (D50). The measured density of the powder was 4.49 g/cm3. In this work, a thickening agent was Polyvinyl alcohols (PVA) and a dispersing agent was Dolapix. 4 wt% PVA in aqueous solution has the viscosity of 25–31 cP at 20 °C. Dolapix contains carboxylic acid solution with approximately 65% active matter. To prepare titanium slurry, PVA and Dolapix, were dissolved in water using magnetic stirrer at 90 °C for 90 min and CP titanium powder was subsequently added and mixed at 90 °C for 10 min using an electronic mixer. The mixing ratio of titanium slurry was 75 wt% of titanium powder, 1.5 wt% of PVA, 1 wt% of Dolapix and balanced by water. The effect of each component on the rheology properties was previously studied and the mixing ratio used here is the optimum from previous investigation [14]. Reticulated PU foams as a sacrificial template was submerged into homogenisedly mixed titanium slurry. Four different cell sizes of PU foam, which were 25–40 ppi with an increment of 5 ppi, were used. ‘ppi’ is pore per inch and it is a unit for cell size measurement, which is commonly used in PU foam industry. In this work, polyester-based PU foam was used because it does not react with titanium powder during processing [15]. Subsequently, excess slurry was removed from PU foam coated with titanium slurry by compression. After at least 24 h air drying, PU foam coated with dried titanium powders, which is commonly called ‘green titanium foam’, was debound and sintered. The debinding condition was 600 °C for 2 h with the heating rate of 1 °C/min to remove any organic contents, i.e. PU foam, PVA and Dolapix. The sintering condition was 1150 °C for 2 h under high vacuum of less than 10 3 Pa. 2.2. Compression tests and characterisation Compression samples have cylindrical shape with a diameter of 15 mm and a height of 18 mm. Compression samples were wire cut from sintered titanium foam in two direction perpendicular to each other as shown in Fig. 1, which are (1) the axial direction of a cylinder (the loading direction) parallel to the foaming direction and (2) the axial direction of a cylinder perpendicular to the foaming direction. It is noted that in the commercial production of PU foam, the vertical direction is a non-constraint direction, where the foam can rise due to chemical reaction. This vertical direction
Fig. 1. A schematic diagram of titanium foam sample, which indicates the foaming direction of PU foam, where the compression samples were selected and the cell structure of titanium foam with respect to the compression direction.
is called the foaming direction. As a result, reticulated cells will tend to be elongated in the foaming direction as shown in Fig. 2 (b) and equiaxed perpendicular to the foaming direction as shown in Fig. 2(a). The first group of samples with the axial direction of a cylinder parallel to the forming direction had elongated cells bearing the compression force and this was named EL samples. The second group of samples with the axial direction of a cylinder perpendicular to the foaming direction had equiaxed cells bearing the compression force and this was named EQ samples. In addition to monotonic compression test, interrupted compression tests were also performed. At each 1 mm increment in displacement, the test was stop and the cross head was retreated so the cross head was reloaded for the next increment. EL and EQ samples were subjected to monotonic and interrupted compression tests with a constant speed of 1 mm/min and were repeated 4 times for each set of experiment. Microstructures were investigated using three dimensional optical microscope (3D-OM) and scanning electron microscope (SEM). From 3D-OM micrographs, average cell sizes can be determined using image analysis software. It was done by in a similar way to that which the PU foam manufacturer uses to determine the cell diameter of PU foam. Details of the procedures can be found in [16]. Each average cell size was determined from at least 150 representative cells. In addition, an apparent density of titanium foam was determined as the averaged valued of density measured from mass divided by overall volume for each titanium foam sample. A relative density could then be determined from the ratio of an apparent density to the bulk density. For SEM analysis, the acceleration voltage was 10 kV, the working distance was 5–7 mm, and the emission current was 70 μA. Chemical compositions were also analysed using XRD. Small titanium particles were prepared by grinding the sintered titanium foams together with liquid nitrogen. The analysis parameters for 2θ–θ method XRD were 50 kV and 300 mA. A combustion technique was used to determined carbon and oxygen contents of sintered titanium foam.
390
A. Manonukul et al. / Materials Science & Engineering A 655 (2016) 388–395
Fig. 2. Optical micrographs of (a)–(b) 35 ppi polyurethane foam, (c)–(d) green titanium foam and (e)–(f) sintered titanium foam, where taken [(a), (c) and (e)] observed along the foaming direction and [(b), (d) and (f)] perpendicular to the foaming direction. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
A. Manonukul et al. / Materials Science & Engineering A 655 (2016) 388–395
3. Results and discussion 3.1. Geometry anisotropy of titanium foam Fig. 2 shows the 3D-OM micrographs of polyurethane foam, 35 ppi green titanium foam and sintered titanium foams observed from 2 directions, along the foaming direction and perpendicular to the foaming direction. Since the micrographs in both directions of PU foams with other cell sizes were similar, only the results for 35 ppi cell size are shown here. Fig. 2(a) and (b) confirm that the PU foam is reticulated and the cells were equiaxed when observed along the foaming direction, and elongated when observed perpendicular to the foaming direction respectively. Green titanium foams observed from both directions shown in Fig. 2(c) and (d) replicated the structure of the PU foam. The impregnation was relatively uniform. In addition, PU foam was well covered by titanium powder since red PU foam area cannot be observed. Struts were thicker from the coating of titanium powders. Sintered titanium foam observed from both directions, as shown in Fig. 2 (e) and (f), can retain the foam structure with smaller cell size due to sintering shrinkage. PU foams with four different cell sizes were used in this work and the 3D-OM micrographs are shown in Fig. 3 for 25, 30, 35 and 40 ppi from (a) to (d) respectively. PU foam structure was polygonal ball structure and the average diameter of polygonal balls was considered the average cell size. As the number of ppi increased from left to right in Fig. 3, there are more pores/cells in a given length/area/volume and hence the cell size decreased. Fig. 3 shows the micrograph when observed in the direction along the foaming direction, hence the cells were equiaxed. The corresponding titanium foams producing using PU foams with different cell sizes in Fig. 3 are shown in Fig. 4. The titanium foams appeared to be denser with decreasing cell size from left to right as the number of ppi in PU foam increased. This is also evident from the cell size and apparent density results shown in Fig. 5. In addition, the result shows the non-linear relationship of cell size and apparent density of titanium foam with respect to PU foam cell size. Fig.6 illustrates SEM micrographs of sintered titanium foam using PU foam with varied cell size in both directions of observation. It is interesting to see that the anisotropy is less obvious for sintered parts when comparing to PU foam e.g. Fig. 2(a) and (b). Trace of powder shape was still evident in SEM micrographs. Some imperfections, which were some cracks on some struts especially at thick areas (corners where struts met) and small circular voids, could be observed. These small circular voids were trace of air pockets during the slurry preparation. It is noted that these imperfections was not pronounced in 3D-OM micrographs comparing Fig. 2(e) and (f) with SEM micrographs Fig. 6(b) and (f), for titanium foam producing using 35 ppi PU foam. In addition, triangular voids were sited where PU foam struts were and this replicated the cross-section of PU foam struts.
391
The degree of anisotropy in titanium foams could be quantified by the results from X-ray micro-computed tomography (μCT). X-ray μCT was performed at a spatial resolution of 5.16 μm on titanium foam using a 35 ppi PU foam. Two-dimensional X-ray images acquired during a full 360-degree rotation scan, commonly called radiographs, were used to reconstruct a three-dimensional volumetric image. The reconstructed volumetric image was then segmented to differentiate pore and titanium strut. Subsequently, watershed segmentation algorithm was applied to convert the connected open-cell pores into individual closed pores [11]. The average pore size and pore size distribution were then quantified. Within a considered volume 4.2 4.2 4.2 mm3, there were 82 pores. When those segmented pores located on the boundaries of the representative volume were excluding, there were 11 complete pores remaining and the average pore diameter was 1.40 mm. This is slightly smaller than the average cell size for titanium foam using 35 ppi PU foam in Fig. 5, which is 1.45 mm since the average cell size included one strut thickness. From these complete pores, the aspect ratio of the major axis and the minor axis together with the angle that the major axis inclined to the compression direction can be determined in 3D using principal component analysis technique [17] and the results are shown in Fig. 7. The corresponding average values are shown as solid lines. The average aspect ratio of pores was 1.16 over the range of 1.02 to 1.27. It is noted that if a pore is circular, the aspect ratio is 1. If a pore becomes more elongated (less degree of geometry anisotropy), the aspect ratio will further depart from 1. The aspect ratio of 1.16 was considered having relatively low degree of geometry anisotropy. Because of the impregnation process, the structure of titanium foam is thicker where struts met as can be seen in Fig. 6. This is also observed in aluminium foams manufactured using the same replica impregnation method [18], lumps of aluminium were observed where struts met. This resulted in lower aspect ratio of aluminium foam cell when comparing with PU foam cell using optical microstructure observation [18]. Apart from the aspect ratio of pore/cell, the angle that the major axis inclined to the compression direction can also indicated the degree of geometry anisotropy. The average angle that the major axis inclined to the compression direction was 20.4° over the range of 9.1° to 46.7° as shown in Fig. 7. The angle that the major axis inclined to the compression was relatively low and relatively random, this also contribute to the low degree of geometry anisotropy. The low degree of geometry anisotropy is confirmed in the next section where compression responses were insignificantly different when loading in the directions that elongated cells and equiaxed cells aligned with the compression direction. 3.2. Mechanical property isotropy of titanium foam Fig. 8 shows the representative compression stress–strain curves of titanium foam produced using PU foam with different
Fig. 3. Optical micrographs observed along the foaming direction of polyurethane foam with (a) 25, (b) 30, (c) 35 and (d) 40 ppi as the template for titanium foam [16].
392
A. Manonukul et al. / Materials Science & Engineering A 655 (2016) 388–395
Fig. 4. Photograph of titanium foam fabricated using polyurethane foam with 25, 30, 35 and 40 ppi from left to right respectively. The photograph was observed along the foaming direction [16].
Fig. 5. Cell size and apparent density of titanium foam using PU foam with varied cell size.
pore sizes for both EL and EQ directions. EL and EQ directions means the directions that elongated cells and equiaxed cells aligned with the compression direction respectively. The compression results in both EL and EQ directions were extremely similar for all four cell sizes. The degree of geometry anisotropy in
Fig. 7. Aspect ratio of pores and angle of major axis to compression direction of titanium foam using 35 ppi PU foam generated from X-ray μCT data. Solid lines are average values.
titanium foam structure produced using the replica impregnation technique was not significant enough to produce noticeable difference in compression results in EL and EQ directions, i.e. there is no mechanical property anisotropy. This is in good agreement with
Fig. 6. SEM micrographs of sintered titanium foam with (a) and (e) 25, (b) and (f) 30, (c) and (g) 35, and (d) and (h) 40 ppi polyurethane foam as the template and (a)–(d) observed along the foaming direction and (e)–(f) observed perpendicular to the foaming direction.
A. Manonukul et al. / Materials Science & Engineering A 655 (2016) 388–395
393
Fig. 8. Representative compression responses of titanium foam using (a) 25, (b) 30, (c) 35 and (d) 40 ppi polyurethane foam with the elongated cells (EL) and the equiaxed (EQ) cell in the loading direction.
Fig. 9. Schematic diagram of stress–strain response of foam materials with increasing apparent density.
Ridha and Shim [3], which reported that the mechanical property anisotropy become insignificant when the aspect ratio of cells is less than 1.2 for PU foams. Comparing compression results of titanium foams using PU foams with varied cell sizes, as the number of ppi in PU foams increased, the load bearing capability of titanium foam increased, i.e. higher compression stress (s), higher quasi-elastic modulus (E) and higher limit stress (plateau stress, sp) were required, in response to the increase in apparent density as shown in Fig. 5 [16] and schematically in Fig. 9. Fig. 10 shows the comparison of monotonic and interrupted compression tests in both EL and EQ directions. Only the results for titanium foam using 35 ppi PU foam are shown here since the trends were similar. As expected, the overall stress–strain
responses of monotonic and interrupted tests were indifferent for both EL and EQ directions. The results of interrupted test show unloading and reloading segments. The unloading and reloading segments were slightly different. The unloading segments were linear since the load was suddenly removed, while the reloading segments were controlled constant velocity at 1 mm/s. Quasielastic modulus of each loading/unloading interruption was calculated and shown in the same graph on the secondary axis in Fig. 10. During the plateau stage (2–40% strain) where the stress remained relatively constant while the strain increased, the corresponding interrupted quasi-elastic modulus also remain relatively constant. The interrupted modulus of elasticity increased rapidly as the deformation entering the densification stage where the structure became compacted. Fig. 11 shows the comparison between interrupted compression results in both EL and EQ directions for titanium foams using varied cell sizes. Similar to the monotonic results shown in Fig. 7 for each PU foam size, the stress–strain response for both directions were very similar. As the monotonic compression results were isotropy, it is expected that the corresponding interrupted compression results were also isotropy for all cell sizes. The interrupted compression results reinforce that there is no significant difference in the deformation in EL and EQ directions. Hence, titanium foams fabricated using the replica impregnation method have mechanical property isotropy.
4. Conclusions Titanium foams with four different cell sizes were produced using the replica impregnation method. Compression samples
394
A. Manonukul et al. / Materials Science & Engineering A 655 (2016) 388–395
Fig. 10. Monotonic responses, interrupted compression responses and corresponding modulus of elasticity of titanium foam using 35 ppi polyurethane foam with (a) the elongated cells (EL) and (b) the equiaxed (EQ) cell in the loading direction.
were wire cut from titanium foams in two directions so that (1) elongated cells were in the loading direction (EL direction) and (2) equiaxed cells were in the loading direction (EQ direction). From both monotonic and interrupted compression tests, the compression responses in both directions are similar. There is no mechanical property anisotropy observed in titanium foam fabricated using the replica impregnation method. From the microstructural observations, it could be observed that the geometry anisotropy was less pronounced in sintered titanium foam comparing to the corresponding PU foam. This is the result of the impregnation where titanium slurry tended to accumulate where struts meet. In addition, the X-ray micro-computed tomography results showed that the average aspect ratio of pores was 1.16 and the average angle that the major axis inclined to the compression was 20.4° for titanium foams fabricated using 35 ppi PU foams. As a result, the degree of geometric anisotropy was not high enough
to cause significant mechanical property anisotropy. In addition, the interrupted compression tests suggested that the quasi-elastic modulus also remain relatively constant during the plateau stage where the stress also remained relatively constant while the strain increased.
Acknowledgements The authors would like to thank the National Metal and Materials Technology Center (MTEC), Thailand and Taisei Kogyo (Thailand) Co., Ltd. for co-funding the project (Grant number 430026223) and Assoc. Prof. Yuichi Otsuka (Nagaoka University of Technology, Japan) for supplying two-dimensional X-ray microcomputed tomography images.
Fig. 11. Hysteresis compression responses of titanium foam using (a) 25, (b) 30, (c) 35 and (d) 40 ppi polyurethane foam with the elongated cells (EL) and the equiaxed (EQ) cell in the loading direction.
A. Manonukul et al. / Materials Science & Engineering A 655 (2016) 388–395
References [1] X. Badiche, S. Forest, T. Guibert, Y. Bienvenu, J.-D. Bartout, P. Ienny, M. Croset, H. Bernet, Mater. Sci. Eng. A 289 (2000) 276–288. [2] L. Gong, S. Kyriakides, W.-Y. Jang, Int. J. Solids Struct. 42 (2005) 1355–1379. [3] M. Ridha, V.P.W. Shim, Exp. Mech. 48 (2008) 763–776. [4] A. Burteau, F. N’Guyen, J.D. Bartout, S. Forest, Y. Bienvenu, S. Saberi, D. Naumann, Int. J. Solids Struct. 49 (2012) 2714–2732. [5] M.F. Ashby, A.G. Evans, N.A. Fleck, L.J. Gibson, J.W. Hutchinson, H.N.G. Wadley, Met. Foam.: A Des. Guide Batter.-Heinemann 2000. [6] P.J. Tan, S.R. Reid, J.J. Harigan, Int. J. Solids Struct. 49 (2012) 2744–2753. [7] M. Vesenjak, C. Veyhl, R. Fiedlerm, Mater. Sci. Eng. A 541 (2012) 105–109. [8] N. Wengjuan, B. Chenguang, Q. GuiBao, W. Qiang, Mater. Sci. Eng. A 506 (2009) 148–151.
395
[9] T. Imwinkelried, J. Biomed. Mater. Res. 75A (2007) 964–970. [10] B. Ye, D.C. Dunand, Mater. Sci. Eng. A 528 (2010) 691–697. [11] N. Michailidis, F. Stergioudi, A. Tsouknidas, E. Pavlidou, Mater. Sci. Eng. A 528 (2011) 1661–1667. [12] C. Park, S.R. Nutt, Mater. Sci. Eng. A 299 (2001) 68–74. [13] A. Manonukul, N. Muenya, F. Léaux, S. Amaranan, J. Mater. Process. Technol. 201 (2010) 529–535. [14] A. Manonukul, M. Tange, P. Srikudvien, N. Denmud, P. Wattanapornphan, Powder Technol. 206 (2014) 120–134. [15] M. Tange, A. Manonukul, P. Srikudvien, Mater. Sci. Eng. A 641 (2015) 54–61. [16] C. Puncreobutr, P.D. Lee, R. Hamilton, A. Phillion, JOM 64 (2011) 89–95. [17] A. Bhattacharya, V.V. Calmidi, R.L. Mahajan, Int. J. Heat Mass Transf. 45 (2002) 1017–1031. [18] A. Manonukul, P. Srikudvien, M. Tange, Mater. Sci. Eng. A 650 (2016) 432–437.
Contents lists available at ScienceDirect
Materials Science & Engineering A journal homepage: www.elsevier.com/locate/msea
Geometry anisotropy and mechanical property isotropy in titanium foam fabricated by replica impregnation method Anchalee Manonukul a,n, Pathompoom Srikudvien a, Makiko Tange b, Chedtha Puncreobutr c a National Metal and Materials Technology Center (MTEC), National Science and Technology Development Agency (NSTDA), 114 Thailand Science Park, Paholyothin Rd., Klong 1, Klong Luang, Pathumthani 12120, Thailand b Taisei Kogyo (Thailand) Co., Ltd., Room INC2d-409, Innovation Cluster 2 Building, Tower D, 141 Thailand Science Park, Paholyothin Rd., Klong 1, Klong Luang, Pathumthani 12120, Thailand c Department of Metallurgical Engineering, Faculty of Engineering, Chulalongkorn University, Pathumwan, Bangkok 10330, Thailand
art ic l e i nf o
a b s t r a c t
Article history: Received 30 November 2015 Received in revised form 6 January 2016 Accepted 6 January 2016 Available online 7 January 2016
Polyurethane (PU) foams have both geometry and mechanical property anisotropy. Metal foams, which are manufacturing by investment casting or melt deposition method and using PU foam as a template, also have mechanical property anisotropy. This work studied the mechanical properties in two directions of titanium foam with four different cell sizes fabricated using the replica impregnation method. The two directions are (1) the loading direction parallel to the foaming direction where the cells are elongated (EL direction) and (2) the loading direction perpendicular to the foaming direction where the cell are equiaxed (EQ direction). The results show that the compression responses for both EL and EQ directions are isotropy. Micrographs and X-ray micro-computed tomography show that the degree of geometry anisotropy is not strong enough to results in mechanical property anisotropy. & 2016 Elsevier B.V. All rights reserved.
Keywords: Powder metallurgy Titanium foam Mechanical properties Isotropy anisotropy
1. Introduction Reticulated polyurethane (PU) foams are processed using reaction of chemicals, which are carried along a conveyer belt for continuous production. This is followed by the peeling of parallelepipedic PU foam blocks to obtain a periodic variation of the open-cell structure [1]. The chemical reaction causes foaming in the vertical direction, hence the cells are elongated in the foam rising direction and thus geometrically anisotropic. Mechanical property anisotropy is also confirmed by experimental and modelling of open-cell PU foam [2]. The compression strength and stiffness of PU foams loading in the foaming direction is higher than in the rolling direction or the transverse direction [3,4]. X-ray micro-computed tomography (μCT) of PU foam shows that cell anisotropy decreases with a higher apparent density [3,4]. In addition, mechanical property anisotropy decreases as the aspect ratio of cell decreases and becomes indifferent as the aspect ratio of cell is less than 1.2 for PU foam [3]. Metal foams can be manufactured by many methods. Open-cell aluminium foam (e.g. Duocels and m.pores) are produced using investment casting with PU foam as a template, while nickel foam n
Corresponding author. E-mail address: [email protected] (A. Manonukul).
http://dx.doi.org/10.1016/j.msea.2016.01.017 0921-5093/& 2016 Elsevier B.V. All rights reserved.
(e.g. Incofoams) uses melt deposition on PU foam [5]. Both metal foams have perfect replica of PU foam template, hence geometry and mechanical property anisotropy are expected to present and have been reported, for example, for aluminium foam [6,7] and for nickel foam [1,4]. The load bearing capability of aluminium and nickel foam loading in the direction parallel to the foaming direction is higher than those loading in the traversed direction [1,4,6,7]. For titanium foam, powder metallurgy is the common manufacturing process. Open-cell titanium foam with low to medium porosity can be produced by powder space holder technique [8– 10]. Although the powder space holder might initially be relatively equiaxed, if the preforms are compacted or hot pressed, pores in metal foam will be elongated perpendicular to the compaction or pressing direction because the powder space holder became flatten [10,11]. This results in geometry anisotropy and effectively mechanical property anisotropy after sintering. For titanium foams fabricated using compaction with powder space holder technique, the titanium foams are slightly stronger perpendicular to the compaction direction and slightly weaker along the compaction axis [9]. The difference in mechanical property anisotropy increases as the degree of geometric anisotropy increases for metal foams [1,12]. On the other hand, if the performs are formed using metal injection moulding method, pores in metal foam will remain equiaxed [13].
A. Manonukul et al. / Materials Science & Engineering A 655 (2016) 388–395
Titanium foam can also be fabricated by replica impregnation method. For this manufacturing method, a PU foam is used as a sacrificial template. PU foams are dipped in titanium slurry and subsequently burnt out. Titanium powders were then sintered. Titanium foams should replicate the structure of corresponding PU foams. There is no previous report on the geometry and mechanical property anisotropy of open-cell titanium foam fabricated by replica impregnation method. Since PU foams have both geometry and mechanical property anisotropy, the objective of this work is to investigate if titanium foams retain both geometry and mechanical property anisotropy. Titanium foams with four different cell sizes were produced and monotonic and interrupted compression tests were carried out using titanium foams in two different loading directions.
389
Table 1 Chemical compositions of commercially pure titanium (wt%). Fe
H
N
C
O
Ti
0.033
0.005
0.009
0.005
0.113
Balance
2. Experimental procedures 2.1. Titanium foam preparation Argon gas atomised commercially pure (CP) titanium powder was used in this work. From SEM observation, the powders were spherical [13]. Table 1, which shows the chemical composition of the CP titanium powder and traces of Fe, H, N, C and O were observed. Manufacturer’s specification of powder size was o45 mm. From a powder size analysis, the measured average powder size was 22.94 mm (D50). The measured density of the powder was 4.49 g/cm3. In this work, a thickening agent was Polyvinyl alcohols (PVA) and a dispersing agent was Dolapix. 4 wt% PVA in aqueous solution has the viscosity of 25–31 cP at 20 °C. Dolapix contains carboxylic acid solution with approximately 65% active matter. To prepare titanium slurry, PVA and Dolapix, were dissolved in water using magnetic stirrer at 90 °C for 90 min and CP titanium powder was subsequently added and mixed at 90 °C for 10 min using an electronic mixer. The mixing ratio of titanium slurry was 75 wt% of titanium powder, 1.5 wt% of PVA, 1 wt% of Dolapix and balanced by water. The effect of each component on the rheology properties was previously studied and the mixing ratio used here is the optimum from previous investigation [14]. Reticulated PU foams as a sacrificial template was submerged into homogenisedly mixed titanium slurry. Four different cell sizes of PU foam, which were 25–40 ppi with an increment of 5 ppi, were used. ‘ppi’ is pore per inch and it is a unit for cell size measurement, which is commonly used in PU foam industry. In this work, polyester-based PU foam was used because it does not react with titanium powder during processing [15]. Subsequently, excess slurry was removed from PU foam coated with titanium slurry by compression. After at least 24 h air drying, PU foam coated with dried titanium powders, which is commonly called ‘green titanium foam’, was debound and sintered. The debinding condition was 600 °C for 2 h with the heating rate of 1 °C/min to remove any organic contents, i.e. PU foam, PVA and Dolapix. The sintering condition was 1150 °C for 2 h under high vacuum of less than 10 3 Pa. 2.2. Compression tests and characterisation Compression samples have cylindrical shape with a diameter of 15 mm and a height of 18 mm. Compression samples were wire cut from sintered titanium foam in two direction perpendicular to each other as shown in Fig. 1, which are (1) the axial direction of a cylinder (the loading direction) parallel to the foaming direction and (2) the axial direction of a cylinder perpendicular to the foaming direction. It is noted that in the commercial production of PU foam, the vertical direction is a non-constraint direction, where the foam can rise due to chemical reaction. This vertical direction
Fig. 1. A schematic diagram of titanium foam sample, which indicates the foaming direction of PU foam, where the compression samples were selected and the cell structure of titanium foam with respect to the compression direction.
is called the foaming direction. As a result, reticulated cells will tend to be elongated in the foaming direction as shown in Fig. 2 (b) and equiaxed perpendicular to the foaming direction as shown in Fig. 2(a). The first group of samples with the axial direction of a cylinder parallel to the forming direction had elongated cells bearing the compression force and this was named EL samples. The second group of samples with the axial direction of a cylinder perpendicular to the foaming direction had equiaxed cells bearing the compression force and this was named EQ samples. In addition to monotonic compression test, interrupted compression tests were also performed. At each 1 mm increment in displacement, the test was stop and the cross head was retreated so the cross head was reloaded for the next increment. EL and EQ samples were subjected to monotonic and interrupted compression tests with a constant speed of 1 mm/min and were repeated 4 times for each set of experiment. Microstructures were investigated using three dimensional optical microscope (3D-OM) and scanning electron microscope (SEM). From 3D-OM micrographs, average cell sizes can be determined using image analysis software. It was done by in a similar way to that which the PU foam manufacturer uses to determine the cell diameter of PU foam. Details of the procedures can be found in [16]. Each average cell size was determined from at least 150 representative cells. In addition, an apparent density of titanium foam was determined as the averaged valued of density measured from mass divided by overall volume for each titanium foam sample. A relative density could then be determined from the ratio of an apparent density to the bulk density. For SEM analysis, the acceleration voltage was 10 kV, the working distance was 5–7 mm, and the emission current was 70 μA. Chemical compositions were also analysed using XRD. Small titanium particles were prepared by grinding the sintered titanium foams together with liquid nitrogen. The analysis parameters for 2θ–θ method XRD were 50 kV and 300 mA. A combustion technique was used to determined carbon and oxygen contents of sintered titanium foam.
390
A. Manonukul et al. / Materials Science & Engineering A 655 (2016) 388–395
Fig. 2. Optical micrographs of (a)–(b) 35 ppi polyurethane foam, (c)–(d) green titanium foam and (e)–(f) sintered titanium foam, where taken [(a), (c) and (e)] observed along the foaming direction and [(b), (d) and (f)] perpendicular to the foaming direction. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
A. Manonukul et al. / Materials Science & Engineering A 655 (2016) 388–395
3. Results and discussion 3.1. Geometry anisotropy of titanium foam Fig. 2 shows the 3D-OM micrographs of polyurethane foam, 35 ppi green titanium foam and sintered titanium foams observed from 2 directions, along the foaming direction and perpendicular to the foaming direction. Since the micrographs in both directions of PU foams with other cell sizes were similar, only the results for 35 ppi cell size are shown here. Fig. 2(a) and (b) confirm that the PU foam is reticulated and the cells were equiaxed when observed along the foaming direction, and elongated when observed perpendicular to the foaming direction respectively. Green titanium foams observed from both directions shown in Fig. 2(c) and (d) replicated the structure of the PU foam. The impregnation was relatively uniform. In addition, PU foam was well covered by titanium powder since red PU foam area cannot be observed. Struts were thicker from the coating of titanium powders. Sintered titanium foam observed from both directions, as shown in Fig. 2 (e) and (f), can retain the foam structure with smaller cell size due to sintering shrinkage. PU foams with four different cell sizes were used in this work and the 3D-OM micrographs are shown in Fig. 3 for 25, 30, 35 and 40 ppi from (a) to (d) respectively. PU foam structure was polygonal ball structure and the average diameter of polygonal balls was considered the average cell size. As the number of ppi increased from left to right in Fig. 3, there are more pores/cells in a given length/area/volume and hence the cell size decreased. Fig. 3 shows the micrograph when observed in the direction along the foaming direction, hence the cells were equiaxed. The corresponding titanium foams producing using PU foams with different cell sizes in Fig. 3 are shown in Fig. 4. The titanium foams appeared to be denser with decreasing cell size from left to right as the number of ppi in PU foam increased. This is also evident from the cell size and apparent density results shown in Fig. 5. In addition, the result shows the non-linear relationship of cell size and apparent density of titanium foam with respect to PU foam cell size. Fig.6 illustrates SEM micrographs of sintered titanium foam using PU foam with varied cell size in both directions of observation. It is interesting to see that the anisotropy is less obvious for sintered parts when comparing to PU foam e.g. Fig. 2(a) and (b). Trace of powder shape was still evident in SEM micrographs. Some imperfections, which were some cracks on some struts especially at thick areas (corners where struts met) and small circular voids, could be observed. These small circular voids were trace of air pockets during the slurry preparation. It is noted that these imperfections was not pronounced in 3D-OM micrographs comparing Fig. 2(e) and (f) with SEM micrographs Fig. 6(b) and (f), for titanium foam producing using 35 ppi PU foam. In addition, triangular voids were sited where PU foam struts were and this replicated the cross-section of PU foam struts.
391
The degree of anisotropy in titanium foams could be quantified by the results from X-ray micro-computed tomography (μCT). X-ray μCT was performed at a spatial resolution of 5.16 μm on titanium foam using a 35 ppi PU foam. Two-dimensional X-ray images acquired during a full 360-degree rotation scan, commonly called radiographs, were used to reconstruct a three-dimensional volumetric image. The reconstructed volumetric image was then segmented to differentiate pore and titanium strut. Subsequently, watershed segmentation algorithm was applied to convert the connected open-cell pores into individual closed pores [11]. The average pore size and pore size distribution were then quantified. Within a considered volume 4.2 4.2 4.2 mm3, there were 82 pores. When those segmented pores located on the boundaries of the representative volume were excluding, there were 11 complete pores remaining and the average pore diameter was 1.40 mm. This is slightly smaller than the average cell size for titanium foam using 35 ppi PU foam in Fig. 5, which is 1.45 mm since the average cell size included one strut thickness. From these complete pores, the aspect ratio of the major axis and the minor axis together with the angle that the major axis inclined to the compression direction can be determined in 3D using principal component analysis technique [17] and the results are shown in Fig. 7. The corresponding average values are shown as solid lines. The average aspect ratio of pores was 1.16 over the range of 1.02 to 1.27. It is noted that if a pore is circular, the aspect ratio is 1. If a pore becomes more elongated (less degree of geometry anisotropy), the aspect ratio will further depart from 1. The aspect ratio of 1.16 was considered having relatively low degree of geometry anisotropy. Because of the impregnation process, the structure of titanium foam is thicker where struts met as can be seen in Fig. 6. This is also observed in aluminium foams manufactured using the same replica impregnation method [18], lumps of aluminium were observed where struts met. This resulted in lower aspect ratio of aluminium foam cell when comparing with PU foam cell using optical microstructure observation [18]. Apart from the aspect ratio of pore/cell, the angle that the major axis inclined to the compression direction can also indicated the degree of geometry anisotropy. The average angle that the major axis inclined to the compression direction was 20.4° over the range of 9.1° to 46.7° as shown in Fig. 7. The angle that the major axis inclined to the compression was relatively low and relatively random, this also contribute to the low degree of geometry anisotropy. The low degree of geometry anisotropy is confirmed in the next section where compression responses were insignificantly different when loading in the directions that elongated cells and equiaxed cells aligned with the compression direction. 3.2. Mechanical property isotropy of titanium foam Fig. 8 shows the representative compression stress–strain curves of titanium foam produced using PU foam with different
Fig. 3. Optical micrographs observed along the foaming direction of polyurethane foam with (a) 25, (b) 30, (c) 35 and (d) 40 ppi as the template for titanium foam [16].
392
A. Manonukul et al. / Materials Science & Engineering A 655 (2016) 388–395
Fig. 4. Photograph of titanium foam fabricated using polyurethane foam with 25, 30, 35 and 40 ppi from left to right respectively. The photograph was observed along the foaming direction [16].
Fig. 5. Cell size and apparent density of titanium foam using PU foam with varied cell size.
pore sizes for both EL and EQ directions. EL and EQ directions means the directions that elongated cells and equiaxed cells aligned with the compression direction respectively. The compression results in both EL and EQ directions were extremely similar for all four cell sizes. The degree of geometry anisotropy in
Fig. 7. Aspect ratio of pores and angle of major axis to compression direction of titanium foam using 35 ppi PU foam generated from X-ray μCT data. Solid lines are average values.
titanium foam structure produced using the replica impregnation technique was not significant enough to produce noticeable difference in compression results in EL and EQ directions, i.e. there is no mechanical property anisotropy. This is in good agreement with
Fig. 6. SEM micrographs of sintered titanium foam with (a) and (e) 25, (b) and (f) 30, (c) and (g) 35, and (d) and (h) 40 ppi polyurethane foam as the template and (a)–(d) observed along the foaming direction and (e)–(f) observed perpendicular to the foaming direction.
A. Manonukul et al. / Materials Science & Engineering A 655 (2016) 388–395
393
Fig. 8. Representative compression responses of titanium foam using (a) 25, (b) 30, (c) 35 and (d) 40 ppi polyurethane foam with the elongated cells (EL) and the equiaxed (EQ) cell in the loading direction.
Fig. 9. Schematic diagram of stress–strain response of foam materials with increasing apparent density.
Ridha and Shim [3], which reported that the mechanical property anisotropy become insignificant when the aspect ratio of cells is less than 1.2 for PU foams. Comparing compression results of titanium foams using PU foams with varied cell sizes, as the number of ppi in PU foams increased, the load bearing capability of titanium foam increased, i.e. higher compression stress (s), higher quasi-elastic modulus (E) and higher limit stress (plateau stress, sp) were required, in response to the increase in apparent density as shown in Fig. 5 [16] and schematically in Fig. 9. Fig. 10 shows the comparison of monotonic and interrupted compression tests in both EL and EQ directions. Only the results for titanium foam using 35 ppi PU foam are shown here since the trends were similar. As expected, the overall stress–strain
responses of monotonic and interrupted tests were indifferent for both EL and EQ directions. The results of interrupted test show unloading and reloading segments. The unloading and reloading segments were slightly different. The unloading segments were linear since the load was suddenly removed, while the reloading segments were controlled constant velocity at 1 mm/s. Quasielastic modulus of each loading/unloading interruption was calculated and shown in the same graph on the secondary axis in Fig. 10. During the plateau stage (2–40% strain) where the stress remained relatively constant while the strain increased, the corresponding interrupted quasi-elastic modulus also remain relatively constant. The interrupted modulus of elasticity increased rapidly as the deformation entering the densification stage where the structure became compacted. Fig. 11 shows the comparison between interrupted compression results in both EL and EQ directions for titanium foams using varied cell sizes. Similar to the monotonic results shown in Fig. 7 for each PU foam size, the stress–strain response for both directions were very similar. As the monotonic compression results were isotropy, it is expected that the corresponding interrupted compression results were also isotropy for all cell sizes. The interrupted compression results reinforce that there is no significant difference in the deformation in EL and EQ directions. Hence, titanium foams fabricated using the replica impregnation method have mechanical property isotropy.
4. Conclusions Titanium foams with four different cell sizes were produced using the replica impregnation method. Compression samples
394
A. Manonukul et al. / Materials Science & Engineering A 655 (2016) 388–395
Fig. 10. Monotonic responses, interrupted compression responses and corresponding modulus of elasticity of titanium foam using 35 ppi polyurethane foam with (a) the elongated cells (EL) and (b) the equiaxed (EQ) cell in the loading direction.
were wire cut from titanium foams in two directions so that (1) elongated cells were in the loading direction (EL direction) and (2) equiaxed cells were in the loading direction (EQ direction). From both monotonic and interrupted compression tests, the compression responses in both directions are similar. There is no mechanical property anisotropy observed in titanium foam fabricated using the replica impregnation method. From the microstructural observations, it could be observed that the geometry anisotropy was less pronounced in sintered titanium foam comparing to the corresponding PU foam. This is the result of the impregnation where titanium slurry tended to accumulate where struts meet. In addition, the X-ray micro-computed tomography results showed that the average aspect ratio of pores was 1.16 and the average angle that the major axis inclined to the compression was 20.4° for titanium foams fabricated using 35 ppi PU foams. As a result, the degree of geometric anisotropy was not high enough
to cause significant mechanical property anisotropy. In addition, the interrupted compression tests suggested that the quasi-elastic modulus also remain relatively constant during the plateau stage where the stress also remained relatively constant while the strain increased.
Acknowledgements The authors would like to thank the National Metal and Materials Technology Center (MTEC), Thailand and Taisei Kogyo (Thailand) Co., Ltd. for co-funding the project (Grant number 430026223) and Assoc. Prof. Yuichi Otsuka (Nagaoka University of Technology, Japan) for supplying two-dimensional X-ray microcomputed tomography images.
Fig. 11. Hysteresis compression responses of titanium foam using (a) 25, (b) 30, (c) 35 and (d) 40 ppi polyurethane foam with the elongated cells (EL) and the equiaxed (EQ) cell in the loading direction.
A. Manonukul et al. / Materials Science & Engineering A 655 (2016) 388–395
References [1] X. Badiche, S. Forest, T. Guibert, Y. Bienvenu, J.-D. Bartout, P. Ienny, M. Croset, H. Bernet, Mater. Sci. Eng. A 289 (2000) 276–288. [2] L. Gong, S. Kyriakides, W.-Y. Jang, Int. J. Solids Struct. 42 (2005) 1355–1379. [3] M. Ridha, V.P.W. Shim, Exp. Mech. 48 (2008) 763–776. [4] A. Burteau, F. N’Guyen, J.D. Bartout, S. Forest, Y. Bienvenu, S. Saberi, D. Naumann, Int. J. Solids Struct. 49 (2012) 2714–2732. [5] M.F. Ashby, A.G. Evans, N.A. Fleck, L.J. Gibson, J.W. Hutchinson, H.N.G. Wadley, Met. Foam.: A Des. Guide Batter.-Heinemann 2000. [6] P.J. Tan, S.R. Reid, J.J. Harigan, Int. J. Solids Struct. 49 (2012) 2744–2753. [7] M. Vesenjak, C. Veyhl, R. Fiedlerm, Mater. Sci. Eng. A 541 (2012) 105–109. [8] N. Wengjuan, B. Chenguang, Q. GuiBao, W. Qiang, Mater. Sci. Eng. A 506 (2009) 148–151.
395
[9] T. Imwinkelried, J. Biomed. Mater. Res. 75A (2007) 964–970. [10] B. Ye, D.C. Dunand, Mater. Sci. Eng. A 528 (2010) 691–697. [11] N. Michailidis, F. Stergioudi, A. Tsouknidas, E. Pavlidou, Mater. Sci. Eng. A 528 (2011) 1661–1667. [12] C. Park, S.R. Nutt, Mater. Sci. Eng. A 299 (2001) 68–74. [13] A. Manonukul, N. Muenya, F. Léaux, S. Amaranan, J. Mater. Process. Technol. 201 (2010) 529–535. [14] A. Manonukul, M. Tange, P. Srikudvien, N. Denmud, P. Wattanapornphan, Powder Technol. 206 (2014) 120–134. [15] M. Tange, A. Manonukul, P. Srikudvien, Mater. Sci. Eng. A 641 (2015) 54–61. [16] C. Puncreobutr, P.D. Lee, R. Hamilton, A. Phillion, JOM 64 (2011) 89–95. [17] A. Bhattacharya, V.V. Calmidi, R.L. Mahajan, Int. J. Heat Mass Transf. 45 (2002) 1017–1031. [18] A. Manonukul, P. Srikudvien, M. Tange, Mater. Sci. Eng. A 650 (2016) 432–437.