Development of Nanoporous Copper Foams by Chemical Dealloyingof Mechanically Alloyed Al­Cu Compounds

Seungjin Nam, Hyungyung Jo, Heeman Choe, Donghwan Ahn and Hyunjoo Choi+

School of Advanced Materials Engineering, Kookmin University, Seoul 136-702, Republic of Korea

We developed open-cell copper foams with uniformly distributed nanopores by chemical dealloying of Al2Cu samples that weresynthesized by a powder process. High-energy ball-milling was utilized for mechanical alloying of aluminum and copper powders and either hotrolling or hot pressing was then employed to consolidate the ball-milled powders into sheets or pellets. Subsequent dealloying aluminum fromthe samples produced nanoporous copper foams. X-ray diffractometry and scanning electron microscopy confirmed the uniform formation ofAl2Cu and aluminum phases prior to dealloying, and the introduction of nanoporous structures after dealloying.[doi:10.2320/matertrans.M2014068]

(Received February 25, 2014; Accepted June 2, 2014; Published July 11, 2014)

Keywords: porous material, mechanical alloying, dealloying, powder metallurgy, nanopores

1. Introduction

Use of the large surface area of open-cell foams has beenconsidered one of the most promising solutions to overcomethe inferior catalytic reaction in bulk materials. Reducing thefoam pore size to the nanoscale may maximize the rate of heatand mass transfer, and optimize catalytic reaction in thematerial, enabling these foams to be used for a variety ofapplications such as electrodes,1) sensors,2) actuators,3) andmicrofluidic flow controllers.4) In particular, nanoporousopen-cell metallic foams exhibit the typical advantages ofhighly porous materials, such as light weight and high specificsurface area; in addition, they exhibit the beneficial propertiesof metals, such as high durability and electrical conductivity,which are not observed in their ceramic and organiccounterparts. Recent studies have focused considerableattention on the processing and characterization of high-performance nanoporous noble metals, as well as nanofoamsmade of the more affordable copper, titanium, and nickel.5)

One promising technique to fabricate metallic nanofoamsis chemical dealloying, whereby atoms in one or more phasesof a two-component crystal are selectively dissolved in anelectrolyte. The second component diffuses along the liquid-solid interface to self-assemble into struts.6,7) The pore sizeis controlled by both the etching rate of the less noblecomponent and the rate of surface diffusion of the more noblecomponent.8) In principle, as the rate of etching exceeds therate of diffusion, the strut and pore sizes can be reduced to afew nanometers.9) Experimentally, however, the actual strutand pore sizes are many orders of magnitude larger than thetheoretically expected values.10) The difference between thetheoretical and the actual pore sizes is generally due to theresult of an inhomogeneous microstructure in the parenttwo-component material, because it is extremely difficult toachieve a completely homogeneous two-component parentmaterial.

Both liquid- and solid-state synthesis routes have so farbeen developed to prepare porous open-cell foams bydealloying. Liquid-state synthesis involves the alloying of

two or more metallic species via arc melting,11) meltspinning,12) etc. The molten alloys are solidified into ingotsor ribbons as they are cooled, and the alloy is then etched insolution to form the foam. Disadvantages of this techniquestem from the high processing temperatures that are involved,which may result in contamination, high energy consump-tion, and phase separation during cooling.11,12) Phaseseparation can lead to variations in microstructure becausethe differing potential energies of the individual phasesexhibit different rates of dealloying. This produces aninhomogeneous and coarse pore structure after dealloying.

In contrast, solid-state synthesis involving powder metal-lurgy can overcome these disadvantages because it isperformed at lower temperatures. In powder metallurgy,two different metal powders are mixed and the mixture isthen consolidated at a sufficiently high temperature andload.13) However, the resulting samples may contain micro-scale pores or defects, which may lead to cracks in the finalfoams after dealloying. The difficulty in achieving atomic-level alloying is another technical drawback of thisprocess.14) Although Wang et al. has recently developednanoporous copper foam by dealloying of mechanically-alloyed AlCu powder, the development of bulk-scale copperfoams with a homogeneous nanoporous structure has yet tobe reported.15)

In this study, highly uniform nanoporous copper foams arefabricated by the dealloying of Al from Al­Cu compoundscreated by a solid-state powder metallurgy route fromaluminum and copper powders. We introduce high-energyball milling method, which enables the parent elements to bemixed and alloyed at almost atomic scale. Furthermore, thehot-working process suppresses the generation of micro-scaledefects during consolidation of the ball-milled powder. Weexamine dealloying of aluminum from the consolidatedsample and the subsequent formation of nanoscale pores inthe copper foams.

2. Experimental Procedure

2.1 Sample preparationCopper foams were produced using chemical dealloying of+Corresponding author, E-mail: hyunjoo@kookmin.ac.kr

Materials Transactions, Vol. 55, No. 9 (2014) pp. 1414 to 1418©2014 The Japan Institute of Metals and Materials

mechanically alloyed Al­Cu pellets or sheets. Mechanicalalloying through high-energy ball-milling was utilized toform an Al2Cu intermetallic compound from a mixture ofAl (³150 µm, 99.99mass% in purity, Changsung Co. Ltd.Korea) and Cu (³180 µm, 99.99mass% in purity, KojundoChemical Laboratory Co. Ltd. Japan) powders. Even thoughthe atomic ratio of Al2Cu intermetallic compound is 2 : 1, weintentionally selected the target atomic ratio on 7 : 3 ratherthan 2 : 1, in order to introduce ³10 vol% monolithic Alphase in Al2Cu phase and to increase porosity of samples. Aland Cu powders were charged into a stainless steel chamber,with stainless steel balls as milling media; the ball-to-powderweight ratio was 15 : 1. Prior to milling, 1mass% of stearicacid (CH3(CH2)16CO2H, Sigma Aldrich Korea Co. Ltd.) wasadded to prevent powder agglomeration and excessive coldwelding between the powder and the chamber wall orblades. The outer surface of the chamber was cooled usingcirculating cold water. Attrition milling was carried out at500 rpm for 24 h in an argon atmosphere.

The ball-milled powder was consolidated by one of twodifferent hot-working processes: hot-rolling or hot-pressing.For hot-rolling, ball-milled Al­Cu powders were packed intoan annealed copper tube at room temperature. The coppertube was then sealed by welding, and heated to 480°C at aheating rate of ³15°C/min. The tube was passed betweena pair of rollers 18 times, producing a 12% reduction inthickness each time. After rolling, the copper tube was peeledoff the surface of the resulting sheet, which had a thickness of³1.1mm. For hot-pressing, ball-milled Al­Cu powder waspacked into a stainless steel mold with a diameter of 15mmand compacted under a pressure of 28MPa at roomtemperature. The powder was consolidated under a pressureof approximately 210MPa at 500°C for 1 h. Boron nitridewas used as a lubricant to minimize the effect of frictionduring hot-pressing. A pellet thickness was ³0.9mm.

Chemical dealloying was conducted in a 20mass% NaOHsolution at room temperature to remove aluminum from theAl­Cu pellet. First, specimens were polished using SiC paper(to 2000 grit) and polishing cloths with alumina paste (to3 µm particle size). And then, the thickness of the specimenswas reduced to ³100 µm. The samples were then cleanedwith ethyl alcohol and deionized (D. I.) water. Five differentsamples were immersed in the 20mass% NaOH solution forvaried dealloying times (1, 2, 3, 5, and 9 h). After dealloying,the samples were rinsed with D. I. water and ethyl alcohol.Figure 1 shows optical images of a Al­Cu sheet before(Fig. 1(a)) and after (Fig. 1(b)) chemical dealloying for 9 h.

2.2 CharacterizationX-ray diffraction (XRD, Rigaku Ultima III X-ray diffrac-

tometer) with Cu K¡ radiation was utilized to identify a varietyof bulk Al and Cu phases in the samples at each step. Themicrostructures of the copper foams were observed using ascanning electron microscope (SEM, JEOL JSM 2001F). Thesamples were fractured to observe the cross-section area. Thespecimens were attached to a carbon tape for SEM analysis.

3. Results and Discussions

Figure 2 shows a sequence of XRD pattern of sample

obtained at each fabrication stage: (from top to bottom) amixture of starting Al and Cu powders; Al­Cu alloy powdersfabricated via 24 h attrition milling; an Al­Cu pelletfabricated by hot-pressing of the ball-milled powders; andan Al­Cu sheet fabricated by hot-rolling of the ball-milledpowders. The gray circles, green triangles, and yellowsquares in the XRD spectra correspond to the referencedpeak positions for Al2Cu, Al, and Cu, respectively.

While the XRD pattern of the hand-mixed startingpowders shows only peaks corresponding to single phasesof Al and Cu, that of the ball-milled powders shows intensepeaks corresponding to the Al2Cu phase, with faint peaksindicating traces of Al. This indicates the formation of Al2Cuintermetallic compounds from Al and Cu powder. During themilling, the parent Al and Cu powder mixture is flattened bythe high impact energy of the stainless steel milling balls, andthe flattened powder becomes cold-welded to form an Al­Culamellar structure. The repeated cold-welding and fracturingof the powder, together with high strain and local hightemperature by local impact, reduces the interlamellar spacesof Al and Cu layers.16,17) This results in atomic-level alloyingand the formation of Al2Cu intermetallic compounds. Thefaint XRD peaks due to single-phase Al are resulted from thatthe initial atomic ratio of Al to Cu is intentionally controlledat 7 : 3, causing a small excess of single phase Al once all theCu phase has been consumed in the alloyed compound.Figure 2 also displays XRD patterns of samples consolidatedthrough hot-rolling and hot-pressing. These spectra aresimilar to that of the ball-milled powder. Peaks correspondingto other intermetallic compounds such as Al4Cu9, Al2Cu3,Al3Cu4, and AlCu are not observed in the consolidated

(b)(a)

Fig. 1 Optical images of (a) a Al­Cu sheet and (b) a Cu nanofoamdealloyed in NaOH solution for 9 h.

Fig. 2 XRD patterns of a simple mixture of Al­Cu powder, ball-milled Al­Cu powder, a hot-pressed Al­Cu pellet, and a hot-rolled Al­Cu sheet.

Development of Nanoporous Copper Foams by Chemical Dealloying of Mechanically Alloyed Al­Cu Compounds 1415

samples. Phase separation or transformation is rarely found tooccur during hot-working processes, whose working temper-ature is lower than the melting temperature.

Figure 3 shows XRD patterns of Al­Cu pellets prepared atvarious dealloying times: 0, 0.25, 0.5, 1, 2, 3, 5 and 9 h. Allsamples were dealloyed in 20mass% NaOH solution at roomtemperature. Here, the peaks showing the presence of Al2Cu,Al, and Cu are marked by gray circles, green triangles, andyellow squares, respectively. Al2Cu and Al peaks disappearafter 0.5 h, while the intensity of Cu peaks graduallyincreases. During the chemical dealloying, Al atoms in theAl2Cu intermetallic compound are dissolved into the solu-tion18) and hydrogen (H2) bubbles are released from thespecimen surface. This chemical reaction can be expressed asfollows:

2Alþ 2NaOHþ 6H2O

! 2Naþ þ 2½AlðOHÞ4�� þ 3H2ðgÞ" ð1Þ

Hydrogen gas was rapidly released when dealloying begins,and the rate of release decreased with increasing dealloyingtime. This indicates that the rate of dealloying graduallydecreases. The reaction is complete when all Al is dissolvedin the NaOH solution. The remaining Cu atoms are unstablebecause of their large surface area, so they recrystallize toform a face-centered cubic (FCC) crystal structure.19) Asthe dealloying proceeds, Cu atoms are gathered to reducesurface energy and are then arranged to form FCC crystalstructure to reduce the internal energy. It thus results in boththe increased strut size and intensity of XRD peaks as shownin Fig. 3.

Figure 4 is a schematic depiction to compare the chemicaldealloying processes in our study with those in previousstudies.11,12) Micro-scale Al phase, frequently observed incast samples, may leave micro-scale voids after dealloying.Owing to the atomic-level mixing of Al and Cu atoms,however, the Al phase would be smaller than tens nano-meters. Although surplus Al is firstly removed and aluminumatoms in Al2Cu are then dissolved in electrolyte duringdealloying, remained copper atoms may diffuse toward nano-scale voids and fill the voids.

Figures 5(a)­5(c) show SEM images for samples whosedealloying time varied from 2 to 5 h. Figures 5(d)­5(f ) aremagnified portions of Figs. 5(a)­5(c), respectively. Thepore and strut sizes gradually increase with dealloying time,as the Cu atoms constantly assemble to form struts. Themean pore size is around 12 « 5, 18 « 5, and 21 « 18 nmafter 2, 3, and 5 h of dealloying, respectively. The size wasmeasured on the basis of the standard metallographicmeasurement method, which involves measuring a lengthof pores included on lines that was randomly drawn onthe SEM image of Cu foam. Equation 2 expresses therelationship between the mean strut size and dealloyingtime.20)

r4 ¼ r40 þ Bt ð2ÞFig. 3 XRD patterns of the Al­Cu precursor sample before and after

dealloying in the 20mass% NaOH solution for 0.25, 0.5, 1, 2, 3, 5, and9 h, respectively.

Fig. 4 Schematic description of chemical dealloying procedure of samples in the present study (solid-state technique, upper) and inprevious studies (liquid-state technique, lower).

S. Nam, H. Jo, H. Choe, D. Ahn and H. Choi1416

Here, r is the mean strut size after dealloying for a certaintime (t), from the initial mean strut size (r0), with growthrate (B), which is expressed by the following eq. (3)21)

B ¼ ð2£a4DsÞ=kT ð3ÞThe growth rate of struts is affected by the dealloyingtemperature (T), surface energy under the dealloyingcondition (£), the lattice parameter (a), and the surfacediffusivity (Ds); k is the Boltzmann constant. The latticeconstant (0.361 nm) and temperature (298K) are fixed in thisexperiment; therefore, the growth rate can be reduced bydecreasing the surface diffusivity and surface energy.22,23)

Figure 6 shows SEM images of the Cu foam afterchemical dealloying in 20mass% NaOH solution for 9 h.Figure 6(a) shows the surface microstructure of the sample.As shown in Fig. 6(a), nanopores were uniformly distributedthroughout the sample. The mean pore and strut sizes are25 « 20 nm and 30 « 20 nm, respectively. The color of theCu foam upon dealloying was dark, as shown in Fig. 1(b),perhaps because of the light scattering caused by its highsurface area, and because its pore size is much smaller thanthe mean wavelength of light.24) The cross-section inFig. 6(b) shows that a uniform porous structure, similar tothat of the sample surface, is formed also in the sampleinterior. Right after dealloying of Al atoms, Cu nanoporousstructure is formed by a self-assembly process throughsurface diffusion. Since Al and Cu atoms are mixed atatomic-level, more Cu atoms may move to form strut andit requires longer time for the self-assembly of Cu.25,26)

This results in nanometer-scale pores and struts in the Cufoam.

4. Conclusion

Highly uniform open-cell Cu nanofoams with large surfacearea have been prepared through chemical dealloying of Al­

Cu compounds, which were carefully prepared using a ball-milling process. High-energy ball milling enabled a uniformatomic-scale arrangement of Al and Cu atoms, resulting in ahomogeneous nanoporous microstructure upon dealloying.

Fig. 5 SEM images showing the microstructure of Cu foams, produced by chemical dealloying in 20mass% NaOH solution for (a) 2,(b) 3, and (c) 5 h, respectively; and (d)­(f ) present their magnified images, respectively.

Fig. 6 SEM images (a) on the surface and (b) in the cross-section of a Cufoam produced by dealloying in 20mass% NaOH solution for 9 h.

Development of Nanoporous Copper Foams by Chemical Dealloying of Mechanically Alloyed Al­Cu Compounds 1417

The uniform structure delayed the assembly of Cu atomsduring the removal of Al. Thus the pores and struts in theresulting Cu nanofoam had respective mean sizes ofapproximately 25 and 30 nm.

Acknowledgments

H.C. acknowledges the support from the Priority ResearchCenters Program through the National Research Foundationof Korea (NRF) funded by the Ministry of Education,Science and Technology (2009-0093814) and the supportfrom the Leading Foreign Research Institute RecruitmentProgram through the NRF funded by the Ministry of Science,ICT & Future Planning (MSIP) (2013K1A4A3055679). S.N.acknowledges support from Honors Challenging Researchthrough the Korea Foundation for the Advancement ofScience and Creativity (KOFAC) funded by the Ministry ofEducation, Science and Technology (A2012-0461).

REFERENCES

1) J. S. Wang, P. Liu, E. Sherman and M. H. Tataria: J. Power Source 196(2011) 8714­8718.

2) J.-H. Jeun, D.-H. Kim and S.-H. Hong: Sen. Actuators B: Chemical161 (2012) 784­790.

3) S. Shen, F. Masters, H. Upjohn, II and C. Ferraro: Compo. Struct. 99(2013) 419­432.

4) A. Testouri, L. Arriaga, C. Honorez, M. Ranft, J. Rodrigues, A. van derNet, A. Lecchi, A. Salonen, E. Rio and R.-M. Guillermic: ColloidsSurf. A: Physicochem. Eng. Aspects 413 (2012) 17­24.

5) L. P. Lefebvre, J. Banhart and D. C. Dunand: Adv. Eng. Mater. 10(2008) 775­787.

6) A. M. Hodge, J. R. Hayes, J. A. Caro, J. Biener and A. V. Hamza: Adv.Eng. Mater. 8 (2006) 853­857.

7) J. Erlebacher, M. J. Aziz, A. Karma, N. Dimitrov and K. Sieradzki:Nature 410 (2001) 450­453.

8) Z. Dan, F. Qin, Y. Sugawara, I. Muto and N. Hara: Micro. Meso. Mater.165 (2013) 257­264.

9) F. Kertis, J. Snyder, L. Govada, S. Khurshid, N. Chayen and J.Erlebacher: J. Oper. Manage. 62 (2010) 50­56.

10) J. Erlebacher: Phys. Rev. Lett. 106 (2011) 225504.11) W. Liu, S. Zhang, N. Li, J. Zheng and Y. Xing: J. Electrochem. Soc.

157 (2010) D666.12) C. Zhao, Z. Qi, X. Wang and Z. Zhang: Corros. Sci. 51 (2009) 2120­

2125.13) G. Davies and S. Zhen: J. Mater. Sci. 18 (1983) 1899­1911.14) J. Banhart: Prog. Mater. Sci. 46 (2001) 559­632.15) Y. Wang, Y. Wang, C. Zhang, T. Kou and Z. Zhang: CrystEngComm 14

(2012) 8352­8356.16) H. Fecht, E. Hellstern, Z. Fu and W. Johnson: Metall. Trans. A 21

(1990) 2333­2337.17) J. S. Benjamin and T. E. Volin: Metall. Trans. 5 (1974) 1929­1934.18) W. B. Liu, S. C. Zhang, N. Li, J. W. Zheng and Y. L. Xing: Corros. Sci.

53 (2011) 809­814.19) Z. Qi, C. Zhao, X. Wang, J. Lin, W. Shao, Z. Zhang and X. Bian:

J. Phys. Chem. C 113 (2009) 6694­6698.20) G. Andreasen, M. Nazzarro, J. Ramirez, R. Salvarezza and A. Arivia:

J. Electrochem. Soc. 143 (1996) 466­471.21) B. K. Chakraverty: J. Phys. Chem. Solids 28 (1967) 2401­2412.22) C. Zhang, J. Sun, J. Xu, X. Wang, H. Ji, C. Zhao and Z. Zhang:

Electrochim. Acta 63 (2012) 302­311.23) W. B. Liu, S. C. Zhang, N. Li, J. W. Zheng and Y. L. Xing: Micro.

Meso. Mater. 138 (2011) 1­7.24) F. Jia, C. Yu, K. Deng and L. Zhang: J. Phys. Chem. C 111 (2007)

8424­8431.25) H. Lu, Y. Li and F. Wang: Scr. Mater. 56 (2007) 165­168.26) K. Wagner, S. R. Brankovic, N. Dimitrov and K. Sieradzki:

J. Electrochem. Soc. 144 (1997) 3545­3555.

S. Nam, H. Jo, H. Choe, D. Ahn and H. Choi1418