Dealloying: Atomistic Investigation of Selective Dissolution

The process involves removal of base metals from the surface layers of a cheaper alloy to leave them considerably enriched in precious metal. In a binary alloy, selective dissolution of less noble component in a corrosive environment often results in a porous network of nobler alloy. Porous materials comprises of a solid skeleton interspersed with pores or voids. Such architecture provides a material with high specific surface area and low specific weight, which qualify them for many applications as catalysts, sensors, and actuators. Today, selective dissolution is being utilized to produce materials of potential technological importance, for instance, the most active catalyst particles for the oxygen reduction in fuel cells.

The simplest scenario of dealloying occurs, if a binary alloy of elements with sufficiently different Nernst potentials is exposed to an aqueous electrolyte in which no stable bulk oxide is formed. The fundamental understanding of the selective dissolution process has been largely studied using noble metal binary alloys like Cu-Au. Cu dissolves initially as the surface is polarized above the corresponding equilibrium potential, Eeq1. Beyond a so-called critical potential Ec, and with a sufficiently high content of the more reactive element in the alloy (i.e. parting limit), the entire alloy then transforms into the mentioned nanoporous network of the noble element. Our previous in-situ X-ray diffraction results 2-4 showed that upon increasing potential above Eeq, an epitaxial ultra-thin Au-rich passive layer forms with inverted stacking sequence than that of the substrate (Fig.1b). Inverted Au islands then grow at medium overpotentials (~ 300-400mV). The observation of a stacking reversal in the surface film indicates that the Au atoms arrange themselves in an energetically favourable orientation probably via surface diffusion.

So far it was not clear if these Au film covers the entire surface at that stage. Therefore we carried out scanning Auger microscopy (SAM) measurements to analyze the chemical composition of the dealloyed Cu₃Au (111) surface. Figure 2 shows the scanning Auger maps and SEM images of the Cu₃Au (111) surface dealloyed at 200 and 300 mV, respectively. At 200mV, initial stripping of Cu from the alloy surface already results in Au-rich alloy surface. Quantitative analysis provided a thickness of Au of 1 monolayer.

At 300mV, further dissolution of Cu leads to the formation of Au islands, which is evident from the scanning Auger maps. Also between the islands the surface is enriched in Au. After removing the dealloyed surface (sputtering) we recover the original composition of Cu₃Au (111).

At the interface between the Au layer and the Cu-Au substrate the local coordination across the interface is modulated including very unfavourable positions. In order to gather a deeper and on-atomic-scale understanding of the strain relaxation mechanisms underlying the Cu₃Au/Au system we performed large scale calculations using Embedded Atom Method (EAM) potentials4. For the interatomic interactions of the Cu-Au alloy system the parametrization proposed by Barrera et al. is used5. We performed the atomistic simulations of an ultrathin (3 ML) Au film on a Cu₃Au (111) substrate. We modelled the Cu₃Au/Au (111) surface using a slab geometry consisting of 6 monolayers (MLs) of Cu₃Au in the L12 crystal structure and 3 MLs of a Au adlayer (see Fig. 3a). The Au adlayer is by 5% biaxially compressed while the Cu₃Au substrate is strain free. The atomic positions of the top 6 MLs are relaxed by a simulated annealing procedure while the bottom 3 MLs of Cu₃Au are kept fixed.

Our calculations revealed that the in-plane strain of the Au adlayer is accommodated by the formation of a triangular network of misfit dislocations which separate regions of fcc and hcp stacking sequences. These regions are shown in Fig. 3b and have a triangular shape and are laterally arranged in a hexagonal pattern. The aforementioned disregistry network is clearly illustrated in Fig. 3c where only the atoms which deviate from the ideal fcc 12 times coordination are shown. This network has an average thickness of 3 MLs and extends by 2 MLs in the Cu₃Au substrate. The underlying local strain along the modulated surface layers should also favour the exchange of atoms between the interface and surface layers even at room temperature, although dislocation networks were not observed sofar for the initial dealloying of Cu₃Au (111). One may speculate that if we have interdiffusion from the substrate to the adlayer this may preferentially occur in the disregistry regions.

The influence of the strain at the substrate-overlayer interface is expected to be large especially during the initial formation of the (flat) thin passive layer in the regime below the critical potential. A detailed comparative study between Cu₃Au and e.g. the nearly strain-free system Ag-Au has not been reported to date.