On Constituents of the Periodic Layered Microstructure Developed in the Solid-State Reaction Between Zinc and Co₂Si

The present study originates in the quest for better understanding of the operational mechanism governing periodic layer formation in displacement solid-state reactions. This special class of reaction zone morphology, which has been found in several metal/ceramic (i.e., SiC/Ni, SiC/Pt, and SiO₂/Mg) as well as in a number of metallic (e.g., Fe₃Si/Zn, Co₂Si/Zn, Ni₃Si₂/Zn, and Ni₅₀Co₂₀Fe₃₀/Mg) systems, is now understood as a manifestation of the Kirkendall effect accompanying reactive phase formation in the solid state. The latter induces a shift of in-situ inert inclusions (reaction product) toward the faster diffusing component. In a more general way, the mechanism is operative when the components have widely different mobilities in adjacent reaction product layers.

More details concerning this peculiar reaction phenomenon can be found in our recent compilation,[1] and here, we will concentrate only on the diffusion microstructure developed in the Co₂Si/Zn system. The selection of this system was dictated by three reasons: (1) this is the finest periodic layered structure (with spacing of ~1 μm) found until now in metal silicide systems,[2,3] (2) the apparent influence of crystallographic orientation of the Co₂Si-substrate grains on the band frequency (spacing) within the product periodic morphology (Figure 1(a)), and (3) the striking pattern (appearance of “loops” composed of CoSi particles, resembling the dislocation loops emerging from a Bardeen–Herring source[4]) observed in a section of the periodic layered reaction zone of the Co₂Si/Zn couple perpendicular to the diffusion direction (Figure 1(b)). This peculiar pattern is also reminiscent of the spiral wave activity in the thin-layer homogeneous Belousov–Zhabotincky reaction.[5] (Note: Before embarking upon the present investigation (for the sake of comparison), we have reproduced (repeated) some diffusion couple experiments reported in Reference 2).

The purpose of this communication is twofold:

Precisely weighed amounts of cobalt (of 99.95 pct purity, Alfa Aesar) and silicon (of 99.999 pct purity, Alfa Aesar) were melted in an induction furnace under argon atmosphere to obtain homogenous Co₂Si intermetallic. The Co₂Si rods were made using an arc furnace equipped with an attachment for the suction casting. In this casting technique, a premixed alloy is arc melted and the product melt is subsequently sucked into a copper mold due to the pressure difference prevailing in the chamber (~ 800 mbar) and in the vacuum (~ 2 × 10⁻² mbar) container (“receiver”). After electric arc ignition, the melting current was gradually increased from 80 to approximately 230 A. The melt was heated for ~ 5 seconds, and after that, it was sucked into the mold by opening a valve separating the furnace chamber and the receiver.

Solid-state reactions between Co₂Si and zinc (of 99.994 pct purity, Alfa Aesar) were conducted in a chamber furnace in quartz ampules at 390 °C for 144 hours. In all experiments, the temperature was controlled within ± 2 deg. However, the actual temperature at the place of the sample might deviate ± 5 deg from the value reported here.

After annealing and standard metallographic preparation, cross sections of the annealed diffusion couples were first analyzed by scanning electron microscopy (SEM, Quanta 3D, FEI) and electron-probe microanalysis (EPMA). The EPMA was performed at an accelerating voltage of 20 kV using pure Co, Si, and Zn as standards.

Further examination of the reaction products formed in the diffusion zones was carried out with transmission electron microscopy (TEM, Tecnai G2, FEI). Cross-sectional TEM specimens were prepared using dual beam technology SEM-focused ion beam (FIB). In this case, a so-called “lift-out” technique was used. An FIB (Quanta 2D, FEI) instrument, which allowed probing specimens from well-defined positions within the reaction zone, located closely either to a Co₂Si or Zn end-member of the Co₂Si/Zn diffusion couple was employed.

In Figure 2, a series of backscattered electron images (BEIs) of the reaction zone in the Co₂Si/Zn diffusion couple annealed at 390 °C for 144 hours is presented. The periodic layered morphology found in these experiments is very much similar to that observed in the previous studies.[1‐3,6‐8] The reaction product consists of a continuous main layer and thin bands of particles, exhibiting “dark” contrast in BEIs, embedded in the matrix phase. Some particles can also be seen in between the bands. The microstructure of the reaction zone is formed by cells with remarkably distinct morphologies. Some are regular and consist of parallel alternating layers, while some are irregular, but a certain kind of periodicity can still be observed. From the analysis of the intensity distribution of characteristic X-ray radiation across the product layer, it appears that bands within the periodic morphology contain silicon and cobalt (Figure 2(d)).

Electron-microprobe measurements show that the Co content in the matrix phase of the reaction zone varies from 11.4 to 12.1 at. pct at the Co₂Si side to 7.7 to 9.5 at. pct at the Zn side of the transition zone, which corresponds, respectively, to γ₁ and γ₂ intermediate phases present in the binary Co-Zn system (Figure 3).[9] In addition, a small amount of silicon (up to 0.7 at. pct) has also been detected in the matrix layer.

The solid-state domain of the ternary Co-Zn-Si system has not been extensively investigated, but a cross section at 395 °C was tentatively described by Osinski.[6] No ternary compounds exist on this isotherm. It was found that binary Co-Zn phases, γ₁ and γ₂, are in equilibrium with the CoSi intermetallic, and the terminal solubility of silicon in these Co-Zn binary phases as well as that of zinc in the CoSi is very low (below 1 at. pct).

Later, Wang et al.[10] conducted an experimental study and thermodynamic calculations of this ternary system but at higher temperatures, viz. 450 °C to 1050 °C. No ternary phases were found by these authors. Also, phase relations involving binary Co-Zn phases, γ₁ and γ₂, and the CoSi as well as the terminal solid-state solubilities in the binary compounds were reported to be similar to those determined in Reference 6.

As to the interaction in ternary Co₂Si/Zn diffusion couples, the only publication found in the literature (apart from the already mentioned work of van Loo and co-workers[2,7,8] and by Yongchong et al.[3]) is that by Liu et al.[11] However, it is important to realize that, in the cited article, the reaction phenomena in the Co₂Si/Zn system have been studied at 450 °C, i.e., above the melting point of Zn, and therefore, a direct comparison of these results with the diffusion couple experiments performed in the present investigation is rather difficult. Although a certain kind of “periodicity” in the arrangement of the phases inside the reaction zone of the “solid-liquid” diffusion couples has been observed, the product microstructure formed under these conditions does not lend itself to a straightforward analysis. The presence of “unreacted” zinc within the product layer renders the rationalization of the morphological developments in the reaction zone very cumbersome.

It is also interesting to mention here the results on the reactive phase formation in binary Co/Zn diffusion couples available in the literature. Duchenko et al.[12,13] investigated the reaction kinetics in Co/Zn “solid-solid” diffusion couples over the temperature range 250 °C to 400 °C, but the preparation of the initial couples was accomplished by dipping of Co pieces into molten zinc. The authors reported that, during annealing at 250 °C to 400 °C, two product layers grow between cobalt and zinc. The layer next to the zinc end-member was found to be the γ₂ phase of the Co-Zn binary system, whereas the nature of the second layer formed in the diffusion zone was not unequivocally established. In contrast, Takamatsu et al.,[14] in their study on reaction kinetics in “sandwich” Zn/Co/Zn (solid-solid) couples, demonstrated that, even at relatively low temperatures, 250 °C to 300 °C, three product layers, i.e., of γ, γ₁, and γ₂ phases of the Co-Zn system, can be identified inside the diffusion zone.

Further investigation of the reaction zone was carried out with TEM on the FIB fabricated specimens. A typical example of the TEM specimen examined is shown in Figure 4. Close inspection of the reaction zone in the vicinity of the Co₂Si substrate, where the composition of the Co-Zn reaction product corresponds to the γ₁ phase, revealed that the bands are actually agglomerates of small particles embedded in the intermetallic matrix (Figures 5(a) and (b)). Electron diffraction confirmed the expected presence of the cubic CoSi as inclusions within the bands of the periodic microstructure (Figure 5(c)).

To the best of the present authors’ knowledge, no crystallographic data on the γ₁ phase of the binary Co-Zn system are available. It turned out that the selected area diffraction patterns (SAEDs) taken from the matrix phase (Figures 5(d) and (e)) can be readily indexed as \( \left[ {2\bar{1}\bar{1} 0} \right] \) and \( \left[ {1\bar{2} 1\bar{3}} \right] \) directions in a hexagonal lattice with a set of data reported by Boström and Lidin[16] for binary compound δ-Co₂Zn₁₅. It is, however, important to realize that the δ phase is stable in the Co-Zn binary system only between 675 °C and 746 °C (Figure 3), which is significantly higher than the temperature used in the present investigation.

Even more unexpected was finding that SAEDs taken from the Co-Zn matrix phase within the periodic product microstructure with the composition corresponding to the binary γ₂-phase (i.e., at the location close to the Zn end-member) cannot be indexed using generally accepted information for the monoclinic γ₂-CoZn₁₃ compound (mC28, C2/m).[17,18] Surprisingly, it can be done, as in the previous case, with the data for the same binary phase δ-Co₂Zn₁₅.

In Figure 6, a montage of TEM (bright-field (BF)) images of the reaction zone developed in the annealed (390 °C, 144 hours) Co₂Si/Zn diffusion couple at the location close to the Zn end-member, where composition of the matrix phase corresponds to the γ₂ phase of the Co-Zn system along with the SAEDs obtained from the particle in the band and from the matrix.

It should be mentioned that δ phase in the Co-Zn system is not a stoichiometric compound. It has a rather narrow homogeneity range and, as mentioned previously, is stable in the binary system within a temperature range of 675 °C to 746 °C. The crystal structure for the δ phase was solved using single-crystal X-ray and neutron powder diffraction data for the composition corresponding to the formula Co₂Zn₁₅.[16] According to the cited work, this is the first example of an intermetallic compound that crystallizes in the acentric hexagonal space group P6₂ (a = 1.1292 nm and c = 0.7750 nm). It has a unique chiral structure, which is composed of highly regular icosahedra of zinc sharing the vertices to form a double helix around the translation hexad. A chain of distorted Archimedean antiprisms is present inside these identical helices.

Obviously, results of the present study differ from what might be expected from the available information on the Co-Si-Zn phase diagram. At this point, it is important to stress that crystallographic studies on the (Co, Si, Zn) alloys involving Zn-rich intermetallics are scarce. There have been only two attempts of phase identification in this system with X-ray diffraction (XRD) analysis using either powdered annealed (bulk) alloys[10] or powder samples made by pulverization of the reaction products formed in diffusion couples between Co₂Si and liquid Zn.[11] Unfortunately, the reported results are not particularly convincing. Studying the original work and looking at the diffraction patterns presented in the cited articles, it is easy to conclude that the monoclinic γ₂-CoZn₁₃ in the multiphase powder alloy samples has been identified by using only a very limited number of X-ray lines (often of rather low intensity). Given a severe overlapping of these lines with some of the X-ray lines corresponding to other alloy constituents, such phase identification becomes unwieldy at best and too intricate to be considered as a reliable procedure.

No explanation can be offered for the apparent ambiguities in the correlating results of the present electron microscopic and diffraction studies of the reaction zone in the annealed Co₂Si/Zn diffusion couples with the available crystallographic and phase diagram information. Nevertheless, it can be conjectured that a small amount of silicon, which was invariably detected (up to 0.7 at. pct) in the matrix phase of the reaction zone of the annealed Co₂Si/Zn diffusion couples, can stabilize the hexagonal structure of the δ phase and this intermetallic compound becomes stable in the ternary Co-Si-Zn system at 390 °C.

As to the operational mechanism responsible for the pattering in the diffusion zone of the Co₂Si/Zn couples, it should be mentioned that, in general, the emergence of the periodic layered morphologies during solid-state reactions is still difficult to predict a priori. Nevertheless, based on the phase relations and knowledge of the relative diffusion mobilities of reactive species, several suggestions have been made.[1]

The periodic layered morphology generated in the reaction zone of the Co₂Si/Zn diffusion couples during annealing at 390 °C consists of bands of in situ-formed precipitates of cubic CoSi, embedded in a continuous matrix of hexagonal Co-Zn-based phase containing up to 0.7 at. pct silicon. The band frequency (spacing) within the periodic microstructure is influenced by the crystallographic orientation of the grains in the Co₂Si substrate. The formation of the hexagonal intermetallic of the same crystal structure as the binary δ phase of the Co-Zn system can be attributed to the possible stabilizing effect of silicon: the δ phase becomes stable in the ternary Co-Si-Zn system at 390 °C.

Although the appearance of loops decorated by CoSi inclusions embedded in the Co-Zn-based intermetallic matrix, observed in a section of the periodic layered reaction zone of the Co₂Si/Zn couple perpendicular to the diffusion direction, does not lend itself at this point to a straightforward analysis, some conjectures can still be made. Most likely this conspicuous phenomenon can be linked to the extraordinary crystal structure of the product matrix phase, related to the binary hexagonal compound δ-Co₂Zn₁₅, and the expected anisotropy of diffusion behavior in this intermetallic phase.