T. Songa,
M. Yana,
Y. Gaob,
A. Atrensc and
M. Qian*a
aRMIT University, School of Aerospace, Mechanical and Manufacturing Engineering, Centre for Additive Manufacturing, Melbourne, VIC 3001, Australia. E-mail: ma.qian@rmit.edu.au
bShanghai Key Laboratory of Modern Metallurgy and Materials Processing, Shanghai University, 200072 Shanghai, P.R. China
cThe University of Queensland, Division of Materials, School of Mechanical and Mining Engineering, Brisbane, QLD 4072, Australia
First published on 5th January 2015
We report the concurrence of de-alloying and re-alloying in a ternary Al67Cu18Sn15 alloy (at.%) de-alloyed in a 5 wt% hydrochloric acid (HCl) solution at 70 ± 2 °C, and the fabrication of three-dimensional (3D) nanoporous Cu3Sn–Cu–Cu6Sn5 composites in the form of self-supporting foils. Re-alloying occurred in Al67Cu18Sn15 compared to de-alloying alone in binary Al–Cu alloys. Both Cu3Sn and Cu6Sn5 phases formed through an accompanied re-alloying process. This finding further proves the temperature sensitivity of phase formation in the Cu–Sn system established from Cu–Sn diffusion couple studies, and demonstrates the capability of designing and creating nanoporous composite materials via de-alloying a multicomponent alloy.
There are two prerequisites for an alloy to undergo de-alloying:5–7 (i) the constituting elements in the alloy should have different electrochemical activities (e.g. A is less noble whilst B is nobler in a binary AB alloy);5 and (ii) the concentration of the nobler element B is below a critical composition, referred to as parting limit, beyond which de-alloying does not take place due to surface passivation by the nobler element B.6,7 Many binary alloys meet these two requirements, including single solid-solution alloys of Au–Ag,8 Pt–Co,9,10 and Au–Cu,11 and two-phase alloys of Al–Cu,12,13 Al–Ag,14 Zn–Cu15 and Mg–Cu.12,16 They have been de-alloyed to produce nanoporous pure metals, and the mechanisms that control the nanoporosity and pattern formation have been investigated.6,7,9,10,12–19 De-alloying of a binary AB alloy involves the dissolution of A, and the diffusion of B.17 Both the rate of dissolution of A, and the surface diffusivity of B, have a significant influence on the formation of the nanoporous structure, including the size of the ligaments/channels in the nanoporous B.20
Research on de-alloying over the last decade has been largely focused on de-alloying of binary alloys (AB) with a view to producing nanoporous structures of essentially pure metal B, and understanding the de-alloying process. Only a few studies have dealt with de-alloying of alloys containing a third element such as Mg90−xCuxY10,21 Mg77Ag18.4Pd4.6,22 Ag64Au30Pt6,23 Al75Pt15Au1024 and Al66Au27.2X6.8 (X = Pt, Pd, PtPd, Ni, Co and NiCo) (in at.%).25,26 However, it should be pointed out that in these ternary ABC alloys, the third element C introduced was similar to the nobler element B, a slow diffuser. The purpose was to slow down the surface diffusivity of B, in order to decrease the ligament/channel size of the resulting nanoporous B.21–23,26 The small amount of this third element C substituted for the noble element B in the lattice of the precursor alloys.21–23,26 Accordingly, de-alloying of these deceptively ternary alloys was no different from the de-alloying of binary alloys.
For de-alloying of ternary alloys (ABC), which consist of three distinctly different elements, the only research reported to date appears to be that by Feng et al.27 As with de-alloying of binary AB alloys, which produces nanoporous structures of essentially pure metal B, de-alloying of ternary ABC alloys has the potential to produce nanoporous composite structures with constituting elements B and C and with characteristics different from those de-alloyed from binary alloys for various applications. In addition, de-alloying of ternary alloys is expected to show different dissolution and diffusion behaviors due to the involvement of the third distinctly element. Understanding their de-alloying characteristics is necessary for the fabrication of more complex nanoporous metal structures aside from enhancing the knowledge base of de-alloying.
This paper studied the de-alloying of a ternary Al67Cu18Sn15 alloy at 70 ± 2 °C in a selected acid solution. The rationale for the selection of an AlCuSn ternary alloy is given as follows. First, dense nanostructured intermetallic Cu6Sn5 anodes outperform pure Sn anodes in lithium ion batteries, where the inactive Cu matrix can act as a stress buffer to accommodate the large volume change caused by lithiation/delithiation during battery operation.28–30 The use of a nanoporous structure has proved to be effective to absorb the volume change of the anode material in lithium ion batteries, such as nanoporous SnO2 anodes31 and nanoporous Ge–C anodes.32 Nanoporous Cu–Sn based composites, if they can be created by de-alloying of a ternary AlCuSn alloy, may have the potential to show desired strain-accommodation capabilities due to their nanoporous structure. In addition, they may offer some good composite strain buffer ability too, like the Sn–Cu composites.28–30 Second, Al is electrochemically more active than both Cu and Sn.5 This allows the creation of various Cu–Sn based nanostructures via the de-alloying of AlCuSn alloys. Third, both Cu and Sn have fast self-diffusion rates,33 which are in favour of complete de-alloying of the entire sample. Additionally, Cu atoms can diffuse interstitially into Sn even at room temperature to enable intermetallic formation.34,35 These characteristics imply that AlCuSn alloys are promising candidate ternary precursors for the creation of a Cu–Sn based nanoporous structure via de-alloying of the Al. The composition of the ternary Al67Cu18Sn15 alloy was determined as follows. The concentration of 67 at.% Al was chosen to be close to the parting limit of Al6,7 whilst the relative concentrations of Cu and Sn were chosen to match the ratio of 1.2:1 for Cu to Sn in Cu6Sn5.30 The selection of the de-alloying temperature (70 ± 2 °C) is based on the phase formation sequence in the Cu–Sn system established from Cu–Sn diffusion couple studies,35 which identified that below 60 °C only Cu6Sn5 forms while at or above 60 °C both Cu6Sn5 and Cu3Sn can develop.35 Feng et al.27 have recently reported the formation of Cu6Sn5 by de-alloying of an Al10Cu3Sn alloy at 60 °C but no formation of Cu3Sn was observed. The selection of 70 ± 2 °C is expected to lead to the formation of Cu3Sn and this will allow us to compare the phase formation in de-allying with the observations made from diffusion couple studies.
De-alloying of the foil samples was carried out at 70 ± 2 °C in 200 ml of 5 wt% HCl aqueous solution. A hot plate and a mercurial thermometer were used to control the temperature. According to the preliminary tests performed on three disc samples, it was found that complete de-alloying of each sample required about 480 min. On this basis, samples were de-alloyed for various durations from 60 min to 480 min. This permitted a systematic study of the de-alloying process. De-alloyed samples were removed from the acid solution, rinsed in distilled water and dehydrated with alcohol.
The phase constitution and microstructure were characterized using X-ray diffraction (XRD, Bruker D8 instrument, with Cu Kα radiation, with a scanning rate of 1° min−1), scanning electron microscopy (SEM) in both the second electron (SE) imaging and backscattered electron (BSE) imaging modes (JEOL 7001, operated at 15 keV equipped with energy dispersive X-ray spectroscopy (EDX) made by INCA). Transmission electron microscopy (TEM) and selected area electron diffraction (SAED) (from JEOL 2100, operated at 200 kV) were also employed to study the microstructure and phase constitution. Samples for TEM analysis were prepared by grinding the as-dealloyed foils into powder, followed by dispersing the powder in ethanol by ultrasonic, and finally releasing just a few drops of the particles-containing ethanol solution on a 3 mm diameter carbon film supported on a copper grid. After drying in air for 20 min, the powder samples were ready for TEM analysis.
(i) The α-Al(Cu) had disappeared after the first 60 min de-alloying, and there were new phases, identified to be Cu3Sn, Cu and Cu6Sn5 with a preponderance of the Cu3Sn phase. In addition, there was a noticeable decrease in the intensity of the XRD peaks for both Al2Cu and Sn. Fig. 3 shows the microstructures after 60 min, 120 min and 180 min de-alloying. The preferential de-alloying of α-Al(Cu) can be seen from Fig. 3a. Fig. 3b shows remnants of Sn and a surface nanoporous structure (with an average ligament width of 40 ± 8 nm) on the Al2Cu substrate, which is interpreted to be Cu3Sn and/or Cu according to the XRD results in Fig. 2a and c. With the progress of de-alloying, the nanoporous structure on the Al2Cu substrate became increasingly coarser (see Fig. 3c and d) while the amount of Cu3Sn and/or Cu continued to increase (see Fig. 2a).
(ii) No Sn was detected by XRD after 240 min de-alloying, and there was a noticeable decrease in the intensity of the XRD peaks for Al2Cu. The SEM BSE image in Fig. 4a confirmed the absence of Sn. The surface nanoporous structure after the first 180 min de-alloying (Fig. 3b–d) had further developed after 240 min de-alloying as shown in Fig. 4b. TEM selected area electron diffraction (SAED) identified the presence of both Cu3Sn and Cu6Sn5 (Fig. 4d), consistent with the XRD results.
(iii) The Al2Cu phase remained after 300 min de-alloying but had disappeared after 480 min de-alloying (Fig. 2b), indicative of complete de-alloying of the Al2Cu phase. This was confirmed by the microstructures shown in Fig. 5a and b, in which no Al2Cu was observed. The three strongest XRD lines in the 2θ range of 23–47° for each of the Cu3Sn (JSPDS reference no. 03-065-4653), Cu (00-004-0836) and Cu6Sn5 (01-076-2703) phases in the International Centre for Diffraction Data (ICDD) database are shown in Fig. 2c to assist in phase identification.
The microstructure of the final dealloyed product is shown in Fig. 5. A nanoporous microstructure with an average ligament width of 170 ± 50 nm was obtained on the surface (Fig. 5a and b) and throughout the longitudinal section (Fig. 5c and d). The inhomogeneous microstructure observed after 240 min (Fig. 4c) and 300 min de-alloying (Fig. 4e and f) had evolved into a homogenous nanoporous structure (Fig. 5c). TEM examination confirmed the nanoporous nature of the product (Fig. 5e). Also, the existence of Cu3Sn and Cu6Sn5 was identified by SAED and high resolution TEM images (Fig. 5f), consistent with the XRD results. EDX analyses detected no Al but did detect Cu and Sn (as presented in the inset in Fig. 5c). In addition, the atomic ratio of Cu:Sn detected by EDX analyses is around 78:22, which is clearly greater than the ratio of Cu to Sn in either Cu3Sn (3:1) or Cu6Sn5 (1.2:1.0). This supports the detection of free Cu in the de-alloyed sample by XRD. The nanoporous Cu3Sn–Cu–Cu6Sn5 composite is produced in the form of self-supporting foils (0.6 mm thick and 8 mm in diameter).
2α-Al(Cu)(s) + 6HCl(l) → 2Cu(s) + 2AlCl3(l) + 3H2(g) | (1) |
De-alloying of the Al2Cu phase also occurred in this process, as informed by the notable decrease in the intensity of the Al2Cu XRD peaks. Considering both the release of H2 and the presence of free Cu in the final de-alloyed product, the de-alloying process of Al2Cu may be described by eqn (2) below
Al2Cu(s) + 6HCl(l) → 2AlCl3(l) + Cu(s) + 3H2(g) | (2) |
As mentioned previously, the only other study of de-alloying of ternary AlCuSn alloys was that by Feng et al.27 Table 1 lists the distinct differences between this study and Feng et al.'s work.27
De-alloying system | Feng et al.'s paper27 | This study |
---|---|---|
Precursor composition (at.%) | Al10Cu3Sn | Al67Cu18Sn15 |
Electrolyte solution | 20 wt% NaOH | 5 wt% HCl |
De-alloying temperature (°C) | 60 | 70 ± 2 |
De-alloyed product | Cu–Cu6Sn5 | Cu3Sn–Cu–Cu6Sn5 |
As a result of the differences discussed above, Feng et al. obtained nanoporous Cu–Cu6Sn5 (ref. 27) while this study attained nanoporous Cu3Sn–Cu–Cu6Sn5. These two different studies are complementary to each other and together they show the significances of precursor composition, electrolyte type, and de-alloying temperature in the de-alloying process of ternary AlCuSn alloys. Also noticed from the XRD results shown in Fig. 2a and b is a small presence of the SnO phase in the samples de-alloyed for 60–300 min. This can be attributed to oxidation in the de-alloying solution. However, no SnO phase was detected after de-alloying for 300 min (Fig. 2b). The consumption of Sn by oxidation could be one of the reasons responsible for the lower Sn content in the final product of Cu3Sn–Cu–Cu6Sn5 than in the precursor alloy.
3Cu(s) + Sn(s) → Cu3Sn(s) | (3) |
6Cu(s) + 5Sn(s) → Cu6Sn5(s) | (4) |
It has long been established that at room temperature, Cu atoms can diffuse fast (interstitially) into the lattice of Sn, particularly along the c-axis direction in the lattice of Sn,34 and then react with Sn to form Cu6Sn5.35,42,43 This occurs at room temperature and from the formation of many other intermetallic compounds which typically occurs at a much higher temperature. The free Cu atoms released by de-alloying according to eqn (1) and (2) were expected to be chemically reactive, as would also be the newly exposed Sn atoms in the precursor alloy. In addition, the de-alloying or removal of the Al atoms from the ternary precursor alloy Al67Cu18Sn15 naturally left Cu and Sn atoms in contact. This enables the formation of Cu6Sn5 and Cu3Sn during de-alloying at 70 ± 2 °C. The formation of Cu3Sn and Cu6Sn5 is based on the interactions between Cu and Sn atoms, similar to the recent study of dealloying of ternary Al–Au–M (M = Ni, Co or NiCo) alloys by Zhang et al.,25 who discussed the importance of the interactions between Au and other atoms. Our study supports Tu's observation35 that “the ordered ε-phase (Cu3Sn) was found only in those specimens that had been annealed above 60 °C”, although no experimental data were given by Tu. It has been found that there was formation of just Cu6Sn5 during de-alloying of a ternary Al10Cu3Sn alloy in a 20 wt% NaOH solution at 60 °C.27
Table 2 lists the detailed literature data on intermetallic formation in the Cu–Sn system under various conditions together with the experimental observations of this study. Previous Cu–Sn diffusion couple studies have established that Cu6Sn5 can form over a wide range of temperatures3,27,35,42–44,46,47 down to −2 °C (Table 2) while Cu3Sn forms mainly at temperatures above 100 °C.35,42–44,46,47 Dreyer et al.3 reported that Cu3Sn nucleated at the temperature of 87 °C in their Cu/Cu6Sn5 thin film diffusion couple study. It is clear from Table 2 that the formation of Cu3Sn at 70 ± 2 °C observed in this study is the lowest temperature reported to date for the intermetallic formation of Cu3Sn with solid experimental data. This improves the knowledge base of phase formation in the Cu–Sn system. Experimental studies have further established that once nucleated, the apparent activation energy (Ea) for the growth of Cu3Sn (70.28 kJ mol−1), determined over the temperature range of 120–200 °C, is smaller than that for the growth of Cu6Sn5 (84.3 kJ mol−1 determined over the temperature range of 70–200 °C).48 This implies that Cu3Sn tends to grow easier than Cu6Sn5. The preponderance of Cu3Sn over Cu6Sn5 in the de-alloyed product can be attributed to this reason.
Diffusion couple | Temperature (°C) | Duration | Intermetallic compounds | Ref. |
---|---|---|---|---|
a Without experimental data. NA: not available. | ||||
Cu/Sn | −2a | NA | Cu6Sn5 | 42 |
Cu/Sn | Room temperature | 15 days | Cu6Sn5 | 42 |
Cu/Sn | Room temperature | One year | Cu6Sn5 | 42 and 43 |
Cu/Sn | Room temperature | 10 days | Cu6Sn5 | 35 |
Cu/Sn | 60a | NA | Cu6Sn5 and Cu3Sn | 42 |
Cu/Sn (de-alloying of Al10Cu3Sn) | 60 | 8 h | Cu6Sn5 | 27 |
Cu/Sn (de-alloying of Al67Cu18Sn15) | 70 ± 2 | 60 min | Cu6Sn5 and Cu3Sn | This study |
Sn–5Bi–3.5Ag solder/Cu | 70, 100 and 120 | 30 days | Cu6Sn5 | 44 |
Cu/Sn | 87 | NA | Cu3Sn nucleation | 3 |
Cu/Sn | 100 | 36 h | Cu6Sn5 and Cu3Sn | 42 and 43 |
Cu/Sn | 100 | 60 h | Cu6Sn5 and Cu3Sn | 42 |
SnPb solder/Cu6Sn5/Cu3Sn/Cu | 100, 125 and 150 | 80 days | Cu6Sn5 and Cu3Sn | 45 |
Lead-free solder/Cu | 100, 125, 150 and 170 | >50 h | Cu6Sn5 and Cu3Sn | 46 |
Cu/Cu6Sn5 | 115–150 | 10 min | Cu6Sn5 and Cu3Sn | 35 |
Lead-free solder/Cu | 150 | NA | Cu6Sn5 and Cu3Sn | 47 |
Sn–5Bi–3.5Ag solder/Cu | 150, 170 and 200 | 30 days | Cu6Sn5 and Cu3Sn | 44 |
Cu/Sn | 200 | 10 min | Cu3Sn | 35 and 43 |
SnPb solder/Cu | 220 | 3–4 min | Cu6Sn5 and Cu3Sn | 45 |
After the disappearance of Sn, de-alloying of Al2Cu continued to release free Cu atoms by reaction (2). However, because of the absence of Sn, reactions (3) and (4) no longer occurred. The Cu–Sn diffusion couple studies at temperatures above 100 °C have identified another reaction, described by reaction (5) below,35,45,49 for the formation and continued growth of Cu3Sn in Cu–Sn diffusion couples.
(5) |
It is plausible that the same reaction may have also occurred in the de-alloying process studied contributing to the preponderance of Cu3Sn over Cu6Sn5 in the final de-alloyed nanoporous Cu3Sn–Cu–Cu6Sn5 structures. However, the noticeable presence of both Cu and Cu6Sn5 detected by XRD after 480 min de-alloying suggests that reaction (5) may have only occurred to a small extent by the end of the 480 min de-alloying process. The two predominant reasons are: (i) reaction (5) is slow; it has been found that a large amount of residual Cu6Sn5 and Cu still remained after even 80 days of annealing at 150 °C,45 and (ii) the de-alloying temperature (70 ± 2 °C) used is inadequate to completely overcome the large energy barrier (95.5 kJ mol−1, determined over the temperature range of 115–150 °C35) required for reaction (5) to occur. In fact, it is ideal to have a noticeable presence of Cu in the as-dealloyed product as Cu offers better thermal and electrical conductivities than both Cu6Sn5 and Cu3Sn,35,42 in addition to the much needed ductility to hold Cu6Sn5 and Cu3Sn together. In this regard, it is desired that reaction (5) is slow.
This journal is © The Royal Society of Chemistry 2015 |