Laves phases forming in the system ScCo₂-“InCo₂”-TaCo₂

2.1 Sample preparation

For the study of (Sc,In,Ta)Co₂ Laves phases, scandium pieces (Smart Elements; 99.999%), cobalt powder (Sigma Aldrich; 99.9%), indium ingots (Smart Elements; 99.995%) and tantalum powder (Johnson-Matthey, 60 mesh, m3N8) were mixed to the total weight of ∼0.5 g.

The first samples with the initial compositions of 16.7Sc:66.7Co:16.7In and 9.7Sc:66.7Co:23.6In were synthesized using a high frequency furnace (Hüttinger Elektronik, Freiburg, Typ TIG 1.5/300). The metals were arc-welded [9] in small tantalum ampoules under an argon atmosphere (purified with titanium sponge (T = 870 K), molecular sieves and silica gel) of ca. 800 mbar. The ampoules were placed in the water-cooled sample chamber of the high-frequency furnace [10], slowly heated up to the 1720 K and cooled to 1170 K within 2 h. This temperature was kept for another hour followed by 3 h of annealing at 870 K. Finally, the ampoules were cooled to room temperature by switching off the power supply. These samples already showed signs of reaction with the tantalum crucibles, which was confirmed by EDX analysis. The broken ingots contained lustrous, irregularly shaped crystals of good quality for a diffractometer study.

We then decided to study the quaternary (Sc,In,Ta)Co₂ Laves phases as well as the ternary solid solutions Sc_1−xIn_xCo₂ and Sc_1−xTa_xCo₂. A sample with the starting composition In_0.5Ta_0.5Co₂ did not form a homogenous phase, probably due to mutual insolubility and the large difference of the melting points of indium and tantalum.

To prevent the reaction with tantalum, a different synthesis approach was used for the solid solution Sc_1−xIn_xCo₂. Samples with nominal compositions of 16.7Sc:66.7Co:16.7In, 13.9Sc:66.7Co:19.4In, 8.3Sc:66.7Co:25In and 5Sc:66.7Co:28.3In were directly obtained by arc-melting [9] of the elements (Co powder was pressed into pellets) under 800 mbar of purified argon. The samples were re-melted several times to ensure homogeneity. The weight losses of the final ingots were always smaller than 0.1% (mostly indium evaporation). The ingots were sealed in evacuated quartz tubes and annealed in a high frequency furnace within a water-cooled sample chamber (Hüttinger Elektronik, Freiburg, Typ TIG 5/300) [11] with the following sequence: (i) heating to 1470 K, (ii) 2 h at 1470 K, (iii) a gradual decrease to 1000 K within 2 h, and (iv) cooling to room temperature by radiative heat loss (the furnace was turned off). These samples showed no signs of a reaction with the ampoule.

A slightly different method was used for the (Sc,In,Ta)Co₂ samples to prevent significant losses of indium. The preparation of the samples with the initial compositions of 11.7Sc:66.7Co:5In:16.7Ta and 17Sc:67Co:9In:7Ta had three steps: (i) the cobalt and tantalum powders were mixed, cold-pressed to pellets and melted in the arc furnace, (ii) scandium and indium chunks were melted into an ingot, forming the second precursor, and (iii) both precursor ingots were reacted by arc-melting. The product buttons were re-melted several times to provide homogeneity. The samples were then sealed in evacuated quartz tubes and heated with the annealing program described in the previous paragraph.

Sc_0.5Ta_0.5Co₂ was prepared from the binary Laves phases ScCo₂ and TaCo₂ as precursors (synthesized by arc-melting of the elements). The precursors were weighted in the ideal molar ratio of 1:1 and arc-melted. The product button was sealed in a quartz tube and annealed in the high frequency furnace with the following temperature program: rapid heating to ∼1800 K, (ii) annealing at 1800 K for 2 h and (iii) decreasing the temperature to 1000 K within 5 h followed by quenching.

2.2 X-ray diffraction

The polycrystalline Laves phase samples were studied by X-ray powder diffraction using the Guinier technique (Nonius FR 552 camera, CuKα₁ radiation, image plate detection system (Fujifilm and BAS-READER 1800), α-quartz (a = 491.30, c = 540.46 pm) as internal standard). The cubic and respectively hexagonal lattice parameters (Table 1) were obtained from least-squares refinements. Comparison with calculated patterns (LazyPulverix [12]) ensured correct indexing.

Irregularly-shaped single-crystalline splinters were selected from the carefully crushed samples under a light microscope. The crystals were glued to glass fibres using bees wax and their quality was checked by Laue photographs on a Buerger precession camera with white Mo radiation and an image plate technique. Intensity data of suitable crystals was collected on a STOE IPDS-II image plate system (two-circle diffractometer, graphite-monochromatized MoKα radiation (λ = 71.073 pm), oscillation mode) at ambient temperature. Numerical absorption corrections were applied to the data sets. The Sc_0.51In_0.49Co₂ data set was collected on a STOE StadiVari (Mo micro focus source and a Pilatus detection system) single-crystal diffractometer. The Gaussian-shaped profile of the micro focus X-ray source required scaling along with a numerical absorption correction. Details on the data collections and the structure refinements are listed in Tables 2–4.

2.3 Structure refinements

Ten data sets were collected for crystals that have been selected from the different ternary and quaternary samples. The basic crystallographic data is summarized in Tables 2–4. The data sets of Sc_0.08In_0.19Ta_0.73Co₂, Sc_0.26In_0.5Ta_0.24Co₂, Sc_0.25In_0.34Ta_0.41Co₂ and Sc_0.53Ta_0.47Co₂ showed face-centered cubic lattices, indicating that their structures are derived from the cubic Laves phase MgCu₂, or its non-centrosymmetric, ordered variant MgCu₄Sn [13], [14]. A hexagonal lattice was evident for the Sc_0.5In_0.5Co₂, Sc_0.51In_0.49Co₂, Sc_0.63In_0.37Co₂ and Sc_0.48Ta_0.52Co₂ crystals. The systematic extinctions were compatible with space group type P6₃mc, in agreement with our previous results on the new ordering variant Sc_0.63In_0.37Co₂ [4]. Also the Sc_0.63In_0.15Ta_0.22Co₂ and Sc_0.49In_0.28Ta_0.23Co₂ crystals showed hexagonal lattices, but with much larger c parameters of ∼2380 pm. The systematic extinctions also pointed to space group type P6₃mc and it turned out that these two phases are non-centrosymmetric ordering variants of MgNi_0.9Cu_1.1 (space group P6₃/mmc) [5].

The starting atomic parameters were gathered with the charge-flipping algorithm [15] implemented in Superflip [16], and the structure was refined on F² with the Jana2006 software package [17] with anisotropic displacement parameters for all sites. Refinements of the occupancy parameters of the B sites of all our AB₂ Laves phases revealed full occupancy by cobalt, while most A sites (except Sc_0.5In_0.5Co₂ with complete Sc/In ordering) revealed higher electron density than a pure scandium occupancy. This is the sore point of the current study, since an enhanced occupancy of scandium can result from Sc/In as well as from Sc/Ta mixing: Sc (21 e^–), In (49 e^–) and Ta (73 e^–).

The A site occupancies were refined on the basis of the following considerations: The binary Laves phases ScCo₂ [18] and TaCo₂ [19] are known and the structure refinements of Sc_0.53Ta_0.47Co₂ and Sc_0.48Ta_0.52Co₂ showed solubility of tantalum on the scandium sites. On the other side, no binary Laves phase “InCo₂” [2] is known and the solid solution Sc_1−xIn_xCo₂ is limited to approximately x = 0.5. The remaining phases can be considered as quaternary Laves phases in the triangle ScCo₂–TaCo₂–“InCo₂” (Figure 1). For these crystals, the Sc/In/Ta occupancy on the A sites was finally matched with the EDX data (Table 5) of the studied crystals and the bulk samples. It is clear that single-crystal X-ray diffraction is at its limits for these cases and the refined occupancies still exhibit minor uncertainties which might be around ±3 at.%.

Figure 1:

Schematic presentation of the location of ternary and quaternary Laves phases in the system ScCo₂–TaCo₂–“InCo₂”. Yellow points – MgCu₂ type, red points – MgCu₄Sn type, blue points – Sc_0.5In_0.5Co₂ type, green points – MgNi_0.9Cu_1.1 type. Point 10 (Sc_0.74In_0.26Co₂) from Ref. [23].

Refinement of the correct absolute structures of the non-centrosymmetric structures was ensured through calculation of the Flack x parameter [20], [21], [22]. Indeed, all crystals showed twinning by inversion and the twin ratios are listed in Tables 2–4. The final difference Fourier syntheses revealed no significant residual electron density. The refined atomic positions, displacement parameters, and interatomic distances are given in Tables 6–8.

CCDC 2079070–2079075, 2079112–2079114 and 2079117 contain the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

2.4 EDX data

The single crystals studied on the diffractometers and the bulk samples were semi-quantitatively studied by energy dispersive analyses of X-rays (EDX) using a Zeiss EVO^® MA10 scanning electron microscope in variable pressure mode (60 Pa) with Sc, InAs, Ta, and Co as standards. The EDX analyses are summarized in Table 5. No impurity elements were detected.