New transition metal-rich rare-earth palladium/platinum aluminides with RET₅Al₂ composition: structure, magnetism and ²⁷Al NMR spectroscopy

2.1 Crystal structures

In the standardized body-centered tetragonal unit cell, the rare-earth cations occupy the Wyckoff site 2b (4/mmm; 0, 0, ¹/₂) and are surrounded by 12 T atoms in the shape of a slightly distorted cuboctahedron, a structural motif found in, e.g., Cu₃Au (Pm3̅m, Fig. 1, left). In RET₅Al₂, these cuboctahedra are condensed via four of their square faces to form infinite layers in the ab plane. The [RET₃] slabs are intercalated by layers of Al atoms, which are also 12-fold coordinated by four Al and eight Pd atoms, again in the shape of a cuboctahedron (Fig. 1, right). While the Al layers form an A, A, … stacking in the crystal structure, the [RET₃] layers are stacked in an A, B, A, … sequence along the crystallographic c axis (Fig. 2). Since the whole range of REPd₅Al₂ compounds (with the exception of La) exists, and also the RET₃ series is complete with all rare-earth metals (RE = Sc, Y, La–Nd, Sm–Lu, [9, 10]), one might think that due to the common structural feature an existence of a RET₃ phase also favors a 1-5-2 phase. Starting from the above mentioned concept, one can consider other RET₃ binaries crystallizing in the Cu₃Au-type structure and try to extend the structural regime of the 1-5-2-type structures. For T = Pt, also compounds with the whole range of rare-earth elements exist, with the exception of europium [9, 11–17]. We succeeded in the synthesis of the REPt₅Al₂ series (RE = Y, Gd–Tm, Lu), where again all known members crystallize in the tetragonal ZrNi₂Al₅-type structure (I4/mmm, Z = 2). The lattice parameters range from a = 416–409 and c = 1488–1462 pm for the palladium compounds and from a = 409–405 and c = 1501–1488 pm for the platinum compounds, respectively (Table 1). In Fig. 3, the lattice parameters and the unit cell volumes of the REPd₅Al₂ series are shown and in Fig. 4 for REPt₅Al₂. All compounds show decreasing lattice parameters and unit cell volumina as expected due to the lanthanide contraction. For the large rare-earth elements (RE = La–Sm) and platinum, however, the expected 1-5-2-type structure is not formed, but a new structure type with a complex, most likely modulated, structure has been found (Renner et al., in preparation).

Fig. 1

Coordination polyhedra of the RE (left) and Al (right) atoms in the RET₅Al₂ structures. Both exhibit a 12-fold coordination in the shape of an almost regular and a distorted cuboctahedral environment, respectively.

Fig. 2

Schematic representation of the tetragonal unit cell of the RET₅Al₂ structures. The slabs of [Al] and [RET₃] forming alternating stacks along the crystallographic c axis are emphasized.

Fig. 3

Lattice parameters of the REPd₅Al₂ series obtained from Guinier powder X-ray diffraction measurements and from the literature [5, 6]. All members show the typical lanthanide contraction.

Fig. 4

Lattice parameters of the REPt₅Al₂ (RE = Gd–Tm, Lu) series obtained from Guinier powder X-ray diffraction measurements. All members show the typical lanthanide contraction.

The layered structure of the compounds causes significant problems when searching for single-crystal specimens for X-ray structure analysis. Although the samples contained numerous specimens that have the typical optical appearance of single crystals, the Laue photographs showed low quality. Long time annealing or synthesis under different conditions, e.g., in a high-frequency furnace, did not change the quality of the bulk sample or the single crystal specimens. Inoue et al. used single crystals of CePd₅Al₂ grown by the Czochralski method for the investigations of direction-dependent property measurements [7]. Lacking the facilities of growing single crystals by this method, we were limited to the investigation of the arc-melted samples. All samples reported here exhibit peak broadening in the Guinier X-ray diffraction patterns as well as diffuse intensities for the investigated single-crystal specimens. The synthesized samples, however, are phase-pure as seen by powder X-ray diffraction and SEM/EDX (energy dispersive X-ray) measurements.

2.2 Structural considerations

The presented 1-5-2 compounds show structural features which also exist in the binary 1-3 compounds. We have investigated a potential correlation of the RET₃ and RET₅Al₂ structures with the help of group-subgroup relationships between the Cu₃Au- and ZrNi₂Al₅-prototypic structures. Both structure types exhibit layers of face-sharing [RET₁₂] cuboctahedra (Fig. 1, left), whereas in ZrNi₂Al₅ the Al atoms are also coordinated in cuboctahedral fashion (Fig. 1, right). A different coloring of the atoms in Cu₃Au along with an enlargement of the unit cell might lead to the ZrNi₂Al₅ type, as a superstructure of Cu₃Au.

Cu₃Au and RET₃ (T = Pd, Pt) crystallize in the cubic space group Pm3̅m (Z = 1) with a∼ 400 pm. RET₅Al₂ (ZrNi₂Al₅ type) crystallizes with the tetragonal space group I4/mmm (Z = 2) with a∼ 400 and c∼ 1450 pm. Both Cu₃Au and ZrNi₂Al₅ can be derived from a fcc cell (Cu, Fm3̅m); a direct relationship, however, does not exist. Figure 5 shows the group-subgroup relationship between the two structures, derived from the crystal structure of copper. Since there is no direct way to transform the cubic space group Pm3̅m (Cu₃Au) into the tetragonal space group I4/mmm (ZrNi₂Al₅), the crystal structures cannot be related to each other from a group-subgroup point of view. While ZrNi₂Al₅ is one example for the ordered structure crystallizing in I4/mmm, a Zr₂NiAl₅ sample was found to crystallize in the primitive Cu₃Au type with space group Pm3̅m [18], however, still with a statistical distribution. Therefore, both branches in the symmetry tree are accessible, RET₅Al₂ favoring the tetragonal one. Two Bärnighausen trees [19] for fcc superstructures from Cu to ZrAl₃ have already been published [20, 21]. The transition from CuAu to ZrNi₂Al₅ or ZrAl₃ is based on the removal of different mirror planes. This removal leads to two different superstructures with the same space group type but with different coloring of the atoms in the unit cell on different Wyckoff sites. A similar transition can be found when going from U₃Si₂ (P4/mbm) to HT-IrIn₃ (P4₂/mnm) or Zr₃Al₂ (P4₂/mnm) by two different klassengleiche transitions of index 2 [22]. The Bärnighausen formalism, showing the relationship between the Cu- and ZrNi₂Al₅-type structures, also yields additional interesting information. The translationengleiche (t) transition of index 3 from Cu to In is responsible for the formation of trillings, while the two klassengleiche (k) transitions are accountable for the formation of anti-phase boundaries. These two features might be the reason for the difficulties observed when growing large single crystals of these materials.

Fig. 5

Group-subgroup scheme for the structure types of Cu₃Au and ZrNi₂Al₅, derived from Cu. The indices for the translationengleiche (t), klassengleiche (k) and isomorphic (i) symmetry reductions as well as the unit cell transformations are given.

2.3 ²⁷Al MAS-NMR investigations

For structure validation, the diamagnetic representatives YT₅Al₂ and LuT₅Al₂ (T = Pd, Pt) were studied by ²⁷Al magic-angle spinning nuclear magnetic resonance (MAS-NMR) spectroscopy. All samples were measured at 11.74 and 9.4 T; additional data at 7.05 T was acquired for the yttrium compounds. The experimental data and their line shape simulations are summarized in Figs. 6 and 7, and the interaction parameters extracted from them are listed in Table 2. All recorded spectra exhibited only one main signal, as expected from the proposed prototypic structure with only one Al site. Due to the interaction of the conduction electrons with the local moments of the nucleus, a strong resonance shift, known as Knight shift, is observed. These Knight shifts are obtained from the positions of the central MAS NMR peaks, which originate from the +¹/₂ ↔ –¹/₂ Zeeman transitions. The palladium compounds show significantly larger Knight shifts than the platinum compounds. The values measured are comparable to those reported in the literature for various transition metal trialuminides [23–25] and for the Zintl compound Ca₁₄AlP₁₁ [26], but they are significantly smaller than those determined for metallic aluminium [27], Mg-Al solid solutions [27] or Cu_1–xAl₂ intermetallics [28].

Fig. 6

Experimental (black) and simulated (red) ²⁷Al MAS-NMR spectra of YPt₅Al₂ (left) and LuPt₅Al₂ (right) measured at 11.74 T. Impurity peaks are marked by asterisks.

Fig. 7

Field-dependent ²⁷Al MAS-NMR spectra of YPd₅Al₂ (left) and LuPd₅Al₂ (right). Top spectra were measured at 11.74 T and bottom spectra at 9.4 T. The simulated spectra are shown in red. Signals marked with an asterisk are attributed to impurities.

From Fig. 7, it is evident that both palladium compounds also exhibit severe line broadening effects, suggesting significant local disorder. Sample annealing did not change the appearance of the spectra. To examine the origin of these line broadening effects, the full width at half maximum (FWHM) was carefully studied as a function of applied magnetic field strength. While Knight shift distribution effects would predict the line widths (in Hz) to increase linearly with field strength, second-order quadrupolar broadening effects would result in an inverse field strength dependence. Table 2 reveals that these field dependent line widths do not exhibit any clear trend, suggesting the latter to originate from a composite effect of both broadening mechanisms.

In addition to the dominant central Zeeman transitions, the spectra show wide spinning sideband manifolds, which originate from the outer ±¹/₂ ↔ ±³/₂ and ±³/₂ ↔ ±⁵/₂ Zeeman transitions, which are anisotropically broadened by first-order nuclear electric quadrupolar perturbations. Since the coordination environment of the Al atoms is not spherical, the quadrupole moment of the ²⁷Al nucleus interacts with the local electrical field gradient, which results in the specific form of the NMR signal and the resulting spinning sideband pattern. From the intensity distributions of these patterns, the quadrupolar coupling constants C_Q and the electric field gradient asymmetry parameters η_Q were obtained via simulations using the Dmfit program [29]. Because of resonance offset effects, matching the experimental spinning sideband profiles within a ±200.000 Hz frequency range of the central resonance was emphasized during the fitting process; at higher resonance offsets, bandwidth limitations cause inevitable attenuation of experimental peak intensities.

2.4 Magnetic properties

Magnetic ordering in the palladium compounds has been reported for CePd₅Al₂ from susceptibility measurements (T_N1 = 4.1 K; T_N2 = 2.9 K), and features of ordering have been found for NdPd₅Al₂ (T_N = 1.2 K), SmPd₅Al₂ (T_N = 1.7 K) and GdPd₅Al₂ (T_N = 6 K) using C_P measurements, while for PrPd₅Al₂ no long-range magnetic ordering phenomenon was found down to 0.3 K [8]. Due to the low amount of rare-earth cations in the structure and their large interatomic distances (e.g., d(Ce–Ce) = 416 pm [5]), only few compounds exhibit magnetic ordering phenomena above 2.5 K. We have investigated the REPd₅Al₂ (RE = Tb–Yb) and REPt₅Al₂ (RE = Y, Gd–Tm) series using a quantum design physical property measurement system (PPMS) without a ³He option, thus being able to measure the samples down to only 2.5 K. For the palladium series, TbPd₅Al₂ and DyPd₅Al₂ exhibit antiferromagnetic ordering at T_N = 9.8(1) and T_N = 3.7(1) K, respectively. The smaller rare-earth elements, however, exhibit no magnetic ordering. Figure 8 (top) shows the temperature dependence of the magnetic and inverse magnetic susceptibility (χ and χ^–1 data) of DyPd₅Al₂ measured at 10 kOe (1 kOe = 7.96 × 10⁴ A m^–1). A fit of the χ^–1 data between 50 and 305 K using the Curie–Weiss law resulted in an effective magnetic moment of μ_eff = 10.56(1) μ_B per dysprosium atom and a Weiss constant of θ_p = –17.8(5) K. The effective magnetic moment agrees well with the theoretical value of 10.65 μ_B for a free Dy³⁺ ion. The negative value of the Weiss constant is giving rise to an antiferromagnetic long-range interaction. The saturation magnetization (μ_sm = 4.84 μ_B) observed in the magnetization isotherms is drastically smaller than the expected saturation magnetization [(Dy³⁺) = 10 μ_B] according to g_J × J. One reason might be that the magnetization isotherm was recorded at 3 K, which is close to the ordering temperature of T_N = 3.7(1) K.

Fig. 8

Magnetic properties of DyPd₅Al₂: (top) Temperature dependence of the magnetic susceptibility χ and its reciprocal χ^–1 measured with a magnetic field strength of 10 kOe; (middle) magnetic susceptibility in zero-field- (ZFC) and field-cooled (FC) mode at 100 Oe; (bottom) magnetization isotherms at 3, 10 and 50 K.

Within the platinum series, no sample exhibits magnetic ordering in the observed temperature range. However, all χ^–1 data could be fitted using the Curie–Weiss law χ = C/(T–θ) in the temperature range of 50–300 K. Table 3 summarizes the magnetic data. All compounds exhibit effective magnetic moments μ_eff close to the expected theoretical moments. YPt₅Al₂ exhibits diamagnetic behavior, no evidence for superconductivity being found down to 2 K using an applied field of H = 50 Oe. The zero-field-cooled (ZFC) measurement (H = 10 kOe) exhibits a diamagnetic susceptibility of χ_Dia ∼ 2.6 × 10^–4 emu mol^–1 (Fig. 9). In Fig. 10, the ZFC measurement (H = 10 kOe) from 3–305 K, the ZFC/FC investigations (H = 100 Oe) and the magnetization isotherms (recorded at 5, 25, 50 and 75 K) for TbPt₅Al₂ are shown. As expected, the magnetization isotherms at 50 and 75 K show a linear increase of the magnetization. At 25 K a slight and at 5 K a pronounced curvature of the magnetization can be seen with saturation at high fields. The isotherms appear to be typical Brillouin functions with field-induced saturation of the paramagnet at low temperatures and high fields. However, the saturation magnetization (μ_sm = 6.61 μ_B) is significantly smaller than the expected saturation magnetization [(Tb³⁺) = 9 μ_B] according to g_J × J.

Fig. 9

Temperature dependence of the magnetic susceptibility χ of diamagnetic YPt₅Al₂, measured with a magnetic field strength of 10 kOe.

Fig. 10

Magnetic properties of TbPt₅Al₂: (top) Temperature dependence of the magnetic susceptibility χ and its reciprocal χ^–1 measured with a magnetic field strength of 10 kOe; (middle) magnetic susceptibility in zero-field- (ZFC) and field-cooled (FC) mode at 100 Oe; (bottom) magnetization isotherms at 5, 25, 50 and 75 K.

4.1 Synthesis

The RET₅Al₂ compounds were synthesized from the elements, using rare-earth metal ingots (ChemPur, Smart Elements; all with stated purities higher than 99%), platinum powder (Degussa, 99.9%) palladium pieces (Allgussa, 99.9%) and aluminium turnings (Koch Chemicals, 99.99%). Pieces of the rare-earth ingot were first arc-melted [30] under an argon pressure of 800 mbar in a water-cooled copper hearth. The platinum powder was pressed into pellets with a diameter of 6 mm. The starting materials were arc-melted under argon at 800 mbar. The obtained button was remelted and turned over several times to increase the homogeneity. Fragments of the crushed buttons were sealed in quartz tubes under vacuum and annealed at 800 °C for 7 days. The samples show metallic luster and a pronounced alignment of the crystallites within the button. They are stable in air over months.

4.2 Powder X-ray diffraction

The polycrystalline samples were characterized by Guinier patterns (imaging plate detector, Fujifilm BAS-1800 scanner) with CuK_α1 radiation using α-quartz (a = 491.30, c = 540.46 pm, Riedel-de-Häen) as an internal standard. Correct indexing of the diffraction lines was ensured through intensity calculations [31]. The lattice parameters were obtained through least-squares fits with standard deviations smaller than ±0.1 pm for the a axis and ±0.4 pm for the c axis [31].

4.3 EDX data

Semiquantitative EDX analyses on all bulk samples were carried out on a Leica 420i scanning electron microscope. The polycrystalline pieces from the arc-melted buttons or from the annealed pieces were embedded in a methylmethacrylate matrix and polished with diamond and SiO₂ emulsions of different particle sizes. The experimentally observed compositions were close to the weighed ones, and phase pure samples within the limitations of the instrument were observed after annealing. No impurity elements heavier than sodium (detection limit of the instrument) were observed.

4.4 Magnetic properties

Polycrystalline pieces of the annealed ingots were packed in kapton foil and attached to the sample holder rod of a vibrating sample magnetometer unit for measuring the magnetization M(T,H) in a quantum design PPMS. The samples were investigated in the temperature range of 2.5–305 K and with magnetic flux densities up to 80 kOe (1 kOe = 7.96 × 10⁴ A m^–1).

4.5 ²⁷Al MAS-NMR spectroscopy

The ²⁷Al MAS-NMR spectra were recorded at 130.287, 104.230 and 78.722 MHz on Bruker DSX 500, Bruker AVANCE 400 and Bruker AVANCE III 300 spectrometers using magic-angle spinning (MAS) conditions. The samples synthesized by arc-melting were ground to a fine powder and mixed with an equivalent amount of SiO₂ to reduce the density and the electrical conductivity of the sample. The diluted samples were loaded into a cylindrical ZrO₂ rotor with a diameter of 2.5 mm. Single-pulse experiments were conducted with a 1 molar Al(NO₃)₃ solution in H₂O set to 0 ppm as a reference. The NMR spectra were recorded using the Bruker Topspin software [32], and the analysis was performed with the help of the Dmfit software [29].