Effect of Transition Metal Substitution on the Structure and Properties of a Clathrate-Like Compound Eu₇Cu44As23.

A series of substitutional solid solutions-Eu₇Cu44-xTxAs23 (T = Fe, Co, Ni)-based on a recently discovered clathrate-like compound (Eu₇Cu44As23) were synthesized from the elements at 800 °C. Almost up to 50% of Cu can be substituted by Ni, resulting in a linear decrease of the cubic unit cell parameter from a = 16.6707(1) Å for the ternary compound to a = 16.3719(1) Å for the sample with the nominal composition Eu₇Cu24Ni20As23. In contrast, Co and Fe can only substitute less than 20% of Cu. Crystal structures of six samples of different composition were refined from powder diffraction data. Despite very small differences in scattering powers of Cu, Ni, Co, and Fe, we were able to propose a reasonable model of dopant distribution over copper sites based on the trends in interatomic distances as well as on Mössbauer spectra for the iron-substituted compound Eu₇Cu36Fe₈As23. Ni doping increases the Curie temperature to 25 K with respect to the parent compound, which is ferromagnetically ordered below 17.5 K, whereas Fe doping suppresses the ferromagnetic ordering in the Eu sublattice.

1. Introduction

Clathrates belong to a peculiar class of inclusion compounds with a complete segregation of guests inside large polyhedral cages forming a framework. More than 250 inorganic/intermetallic clathrate compounds are known. They are grouped into 10 structure types and include over 40 chemical elements as constituents of host and guest substructures [1]. Despite such a structural and chemical variety, the nature of guests is typically limited to alkali and alkali-earth metals for anionic clathrates, and to halogens and chalcogens for inverse clathrates. However, there are several compounds that feature rare-earth elements Ce, Pr, or Eu. These examples are relatively scarce and feature type-I and type-VIII clathrates only. Nevertheless, the presence of a rare-earth cation gives rise to various interesting properties, including the enhancement of thermopower [2] and the formation of magnetic order [3,4] that can trigger a giant magnetocaloric effect [5,6].

A combination of europium, copper, and a group 15 metals gives rise to a broad family of ternary compounds with a great variety of crystal structures and properties. The structures range from pseudo-layered—related to the types known for Fe–As based superconductors [7,8]—to truly three-dimensional structures, in which europium occupies large voids and displays high coordination numbers. The vast majority of these compounds contain divalent europium, with the 4f7 ground state configuration giving rise to strong paramagnetic response and, eventually, magnetic ordering. It is worth noting that the pseudo-layered compounds typically feature antiferromagnetic (AFM) ordering [7], whereas ferromagnetic (FM) ordering is rare—Eu2Cu6P5 and EuCu4P3 being the only examples [9]. On the contrary, compounds with 3D structures frequently display FM ordering, as in the clathrate compound EuBa8−Cu16P30, exhibiting a superstructure of the type-I clathrate [10], and in the recently discovered clathrate-like compound Eu7Cu44As23 [11]. Whereas the former is a typical clathrate compound whose structure and properties can be rationalized using the Zintl–Klemm approach, the latter phase is an unbalanced metallic compound. Its resemblance to clathrates is ensured by a high coordination number of the Eu2+cations. In its crystal structure (Figure 1), two types of Eu2+cations alternate, one of which resides in a cubic environment of eight arsenic atoms, whereas the other one sits in the center of a 20-vertex polyhedral void. Low-temperature thermodynamic measurements revealed ferromagnetic behavior of Eu7Cu44As23 below 17.5 K, owing to the interaction between the localized 4f7 Eu2+cations, presumably through the conducting Cu–As framework.

Figure 1

Projection of the Eu7Cu44As23 crystal structure onto the (001) plane.

Given the electronic imbalance and the corresponding metallic behavior of Eu7Cu44As23, we considered that extended solid solutions could be formed by substituting Cu with its neighboring 3d-elements possessing lesser number of valence electrons than copper. In this paper, we present synthesis and the investigation of solid solutions formed by substituting Fe, Co, or Ni for copper in Eu7Cu44As23, and discuss the influence of such substitutions on structural and magnetic properties.

2. Results

2.1. Synthesis and Homogeneity Ranges

An optimal synthetic procedure for the solid solutions is almost the same as for the parent compound. The only difference is that we had to increase the annealing time by 48 h in order to reach equilibrium. The largest homogeneity range was observed for T = Ni, with an almost linear (Figure 2) decrease of the unit cell parameter from a = 16.6707(2) Å [11] for the undoped phase to a = 16.3719(1) Å for the sample with x = 20 (the sample with the nominal composition Eu7Cu24Ni20As23 contains up to 5% EuNi5As3). The homogeneity range for Co and Fe were found to be narrower, with the substitution limit of x = 8.

Figure 2

Cubic unit cell parameters vs. x in solid solutions Eu7Cu44−NiAs23. Markers cover standard deviations.

2.2. Crystal Structure Refinement and Description

Refinement of powder XRD samples with the crystal structure of Eu7Cu44As23 as a starting model resulted in low residuals for all samples (Table 1 and Table 2). This fact indicates that no significant structure distortion occurs during the substitution. In all cases, we observe a gradual decrease of the cubic cell parameter with almost the same increment (about 0.1% per atom) for T = Co and Ni, while for T = Fe, the increment is smaller, about 0.04% per atom.

Table 1

Details of the powder XRD experiment for Eu7Cu44−TAs23 phases (space group) 2.

Composition (T, x)	Ni, 2	Ni, 8	Ni, 12	Ni, 20	Co, 8	Fe, 8
Z Cell parameters	4
a, Å	16.6487(2)	16.5407(1)	16.4830(1)	16.3719(1)	16.5421(1)	16.6251(1)
V, Å3	4614.65(8)	4525.42(3)	4478.25(3)	4388.34(6)	4526.57(3)	4595.11(1)
Calculated density, g/cm3	8.0359	8.1253	8.1451	8.2506	8.1599	7.9947
Radiation	CuKα1,2
2θ range	5–140	5–110	5–110	5–120	5–110	5–110
Data points/reflections	10,282/271	7996/186	7996/184	8758/205	7996/186	1997/187
Overall/structural parameters	44/17	43/17	52/17	62/17	56/17	46/20
Analyzing package R values 1 (%):	Jana 2006 [12]
RB	3.60	1.92	2.03	1.62	1.94	1.76
Rp	1.32	1.13	1.51	1.15	0.81	1.35
Rexp	0.81	0.74	1.44	1.13	0.55	1.23
GOF	2.39	2.26	1.38	1.30	2.47	1.47

1 R—Bragg R-factor; R—profile R-factor; 2 Further details of the crystal structures can be found in Supplementary Materials.

Table 2

Refined atomic parameters for Eu7Cu44−TAs23.

T = Ni, x = 2
Atom	Position	x/a	y/b	z/c	Uiso
Eu1	24e(x, 0, 0)	0.2447(1)	-	-	0.0101(5)
Eu2	4a(0, 0, 0)	-	-	-	0.005(1)
Cu/Ni1	96k(x, x, z)	0.13666(4)	-	0.2544(1)	0.0168(6)
Cu/Ni2	48h(0, y, y)	-	0.1825(1)	-	0.0161(9)
Cu/Ni3	32f(x, x, x)	0.41199(9)	-	-	0.021(1)
As1	48i(½, y, y)	-	0.17097(8)	-	0.0080(6)
As2	32f(x, x, x)	0.11008(7)	-	-	0.0072(7)
As3	8c(¼, ¼, ¼)	-	-	-	0.0034(9)
As4	4b(½, ½, ½)	-	-	-	0.042(3)
T = Ni, x = 8
Atom	Position	x/a	y/b	z/c	Uiso
Eu1	24e(x, 0, 0)	0.24356(9)	-	-	0.0066(3)
Eu2	4a(0, 0, 0)	-	-	-	0.0024(9)
Cu/Ni1	96k(x, x, z)	0.13658(4)	-	0.2543(1)	0.0105(3)
Cu/Ni2	48h(0, y, y)	-	0.18399(7)	-	0.0124(6)
Cu/Ni3	32f(x, x, x)	0.41328(8)	-	-	0.0121(8)
As1	48i(½, y, y)	-	0.16919(6)	-	0.0072(4)
As2	32f(x, x, x)	0.10942(5)	-	-	0.0085(5)
As3	8c(¼, ¼, ¼)	-	-	-	0.0087(7)
As4	4b(½, ½, ½)	-	-	-	0.048(2)
T = Ni, x = 12
Atom	Position	x/a	y/b	z/c	Uiso
Eu1	24e(x, 0, 0)	0.2434(1)	-	-	0.0054(4)
Eu2	4a(0, 0, 0)	-	-	-	0.002(1)
Cu/Ni1	96k(x, x, z)	0.13670(5)	-	0.2544(1)	0.0083(4)
Cu/Ni2	48h(0, y, y)	-	0.18523(9)	-	0.0104(8)
Cu/Ni3	32f(x, x, x)	0.4138(1)	-	-	0.0095(10)
As1	48i(½, y, y)	-	0.16905(7)	-	0.0039(5)
As2	32f(x, x, x)	0.10934(7)	-	-	0.0066(7)
As3	8c(¼, ¼, ¼)	-	-	-	0.0023(9)
As4	4b(½, ½, ½)	-	-	-	0.020(2)
T = Ni, x = 20
Atom	Position	x/a	y/b	z/c	Uiso
Eu1	24e(x, 0, 0)	0.2425(1)	-	-	0.0032(3)
Eu2	4a(0, 0, 0)	-	-	-	0.009(1)
Cu/Ni1	96k(x, x, z)	0.13647(3)	-	0.2543(1)	0.0047(3)
Cu/Ni2	48h(0, y, y)	-	0.18447(8)	-	0.0016(7)
Cu/Ni3	32f(x, x, x)	0.41368(8)	-	-	0.0073(9)
As1	48i(½, y, y)	-	0.16986(6)	-	0.0035(4)
As2	32f(x, x, x)	0.10923(5)	-	-	0.0042(5)
As3	8c(¼, ¼, ¼)	-	-	-	0.0016(7)
As4	4b(½, ½, ½)	-	-	-	0.004(2)
T = Co, x = 8
Atom	Position	x/a	y/b	z/c	Uiso
Eu1	24e(x, 0, 0)	0.24150(9)	-	-	0.0094(3)
Eu2	4a(0, 0, 0)	-	-	-	0.010(1)
Cu/Co1	96k(x, x, z)	0.13653(4)	-	0.25467(1)	0.0161(8)
Cu/Co2	48h(0, y, y)	-	0.18542(8)	-	0.018(7)
Cu/Co3	32f(x, x, x)	0.41383(8)	-	-	0.014(5)
As1	48i(½, y, y)	-	0.16785(6)	-	0.0110(5)
As2	32f(x, x, x)	0.10955(6)	-	-	0.0122(6)
As3	8c(¼, ¼, ¼)	-	-	-	0.0092(8)
As4	4b(½, ½, ½)	-	-	-	0.023(2)
T = Fe, x = 8
Atom	Position	x/a	y/b	z/c	Uiso
Eu1	24e(x, 0, 0)	0.23912(9)	-	-	0.0088(3)
Eu2	4a(0, 0, 0)	-	-	-	0.010(1)
Cu/Fe1	96k(x, x, z)	0.13663(4)	-	0.2547(1)	0.0140(4)
Cu/Fe2	48h(0, y, y)	-	0.18436(8)	-	0.0118(7)
Cu/Fe3	32f(x, x, x)	0.41385(8)	-	-	0.0115(8)
As1	48i(½, y, y)	-	0.16843(6)	-	0.0074(5)
As2	32f(x, x, x)	0.11001(6)	-	-	0.0048(5)
As3	8c(¼, ¼, ¼)	-	-	-	0.0033(8)
As4	4b(½, ½, ½)	-	-	-	0.020(2)

A general view of the crystal structure of Eu7Cu44−TAs23 is shown in Figure 1. The 3D framework built of As and T features large voids occupied by Eu, having a 20-fold coordination composed of eight As + 12T atoms (Figure 3a), with the distances to the neighbors varying from 3.13 to 3.54 Å. The remaining Eu atoms occupy smaller cubic voids formed solely by the As atoms. Within the framework, the T and As atoms occupy three and four crystallographic sites, respectively. The coordination of the framework atoms is quite different; importantly, there are no As-As bonds, which suggests that all As atoms can be considered as As−3 anions. Additionally, each of three independent T atoms has four As neighbors forming slightly distorted tetrahedra, and the coordination is supplemented by five, six, or seven T atoms (Figure 4b–d). In general, the atomic arrangement resembles that found in the crystal structure of BaHg11 [13], with the doubling of a cubic unit cell parameter, a(Eu7Cu44−TAs23) ≈ 2a(BaHg11).

Figure 3

(a–d) Coordination polyhedra of Eu1, Cu1, Cu2, and Cu3, respectively, in the crystal structure of Eu7Cu44As23.

Figure 4

57Fe Mössbauer spectrum of Eu7Cu36Fe8As23 recorder at 15 K.

2.3. Mössbauer Spectra

Our X-ray diffraction data did not allow us to distinguish between Cu and the substituting T element. Therefore, we used other methods in order to shed more light onto the distribution of the T atoms. To this end, we chose the sample with the Eu7Cu36Fe8As23 composition and performed the 57Fe-Mössbauer study at low temperatures. The experimental spectrum presented in Figure 4 consists of a single narrow quadrupole doublet with the hyperfine parameters listed in Table 3. We note that these parameters are not sensitive to temperature in the entire investigated range. The isomer shift of 0.63–0.64 mm·s−1 and quadrupole splitting of 0.17–0.18 mm·s−1 are characteristic of high-spin Fe3+cations in a symmetric coordination environment with a high coordination number. We note, however, that the observed value of the isomer shift is higher than those reported for other compounds exhibiting numerous Fe–Fe bonds. For instance, the isomer shift of 0.30–0.43 mm·s−1 was found for Fe3GeTe2 [14], where the coordination number of iron ranges from eight to ten, including up to four Fe–Fe bonds. We believe that the difference might arise from shorter Fe–As bonds in our compound compared to Fe–Ge and Fe–Te bonds in Fe3GeTe2.

Table 3

Hyperfine parameters of the 57Fe Mössbauer spectra of Eu7Cu36Fe8As23 at different temperatures; δ is the isomer shift, Δ is the quadrupole splitting, and W is the linewidth.

T, K (±1 K)	δ (mm/s)	∆ (mm/s)	W (mm/s)
15	0.636(3)	0.169(7)	0.28(2)
27	0.638(3)	0.176(6)	0.27(2)
41	0.633(2)	0.166(7)	0.29(2)

The constant hyperfine parameters in the temperature range of 14.7–41.1 K rule out the possibility of electron transfer between the transition metal atoms, whereas a very low half-width of the doublet points at a single coordination site occupied by the iron atoms. This is because the difference in the coordination numbers of Fe in the T1, T2, and T3 positions would result in essentially dissimilar hyperfine parameters. However, as long as three metal sites possess similar—though not identical—coordination, the 57Mössbauer spectrum alone cannot distinguish which site is preferred by Fe in the crystal structure of Eu7Cu36Fe8As23.

2.4. Interatomic Distances

Substitution of Ni for Cu provides the most extended solid solution up to x = 20. Upon substitution, the cubic unit cell parameter decreases almost linearly with the composition (Figure 2). The same trend is observed for the majority of bond distances (Table 4, Figure 3).

Table 4

Selected interatomic distances (in Å) for Eu7Cu44−TAs23. Cu/T denotes a mixed site of Cu and T (T = Fe, Co, Ni).

Composition (T, x)		Undoped	Ni, 2	Ni, 8	Ni, 12	Ni, 20	Co, 8	Fe, 8
Eu1	1 × Eu2	4.0863(2)	4.074(2)	4.029(2)	4.012(2)	3.970(2)	3.993(2)	3.975(2)
	4 × Cu/T1	3.2217(3)	3.2217(7)	3.1999(6)	3.1917(8)	3.1657(6)	3.2010(8)	3.2227(7)
	4 × Cu/T2	3.2092(2)	3.210(2)	3.199(1)	3.200(2)	3.166(1)	3.203(2)	3.197(1)
	4 × Cu/T3	3.4702(2)	3.471(2)	3.464(2)	3.453(2)	3.442(2)	3.493(2)	3.541(2)
	4 × As1	3.1814(1)	3.174(2)	3.149(1)	3.138(1)	3.129(1)	3.157(1)	3.194(1)
	4 × As2	3.4326(2)	3.427(2)	3.387(1)	3.373(2)	3.340(1)	3.365(2)	3.361(1)
Eu2	8 × As2	3.1659(3)	3.174(1)	3.1346(9)	3.122(1)	3.0974(9)	3.139(1)	3.1677(9)
	6 × Eu1	4.0863(2)	4.074(2)	4.029(2)	4.012(2)	3.970(2)	3.992(2)	3.975(2)
Cu/T1	2 × Cu/T1	2.5811(5)	2.565(2)	2.553(2)	2.539(2)	2.529(2)	2.544(2)	2.556(2)
	2 × Cu/T1	2.7703(5)	2.773(2)	2.752(2)	2.743(2)	2.728(2)	2.766(2)	2.775(2)
	2 × Cu/T2	2.6830(2)	2.682(1)	2.659(1)	2.649(1)	2.630(1)	2.659(1)	2.675(1)
	1 × Cu/T3	2.8802(5)	2.862(2)	2.878(2)	2.856(2)	2.856(2)	2.881(3)	2.900(2)
	2 × As1	2.6560(2)	2.654(1)	2.646(1)	2.637(1)	2.614(1)	2.647(1)	2.660(1)
	1 × As2	2.4887(4)	2.483(2)	2.479(2)	2.474(2)	2.457(2)	2.483(2)	2.485(2)
	1 × As3	2.6765(3)	2.6696(7)	2.6540(6)	2.6420(7)	2.6295(6)	2.6563(7)	2.6667(7)
	1 × Eu1	3.2217(3)	3.2217(7)	3.1999(6)	3.1917(8)	3.1657(6)	3.2010(8)	3.2227(7)
Cu/T2	2 × As1	2.4462(3)	2.447(2)	2.441(2)	2.417(2)	2.397(2)	2.445(2)	2.462(2)
	2 × As2	2.4968(3)	2.503(2)	2.514(1)	2.525(2)	2.497(1)	2.536(2)	2.530(1)
	4 × Cu/T1	2.6830(2)	2.682(1)	2.659(1)	2.649(1)	2.630(1)	2.659(1)	2.674(1)
	1 × Eu1	3.2092(2)	3.210(2)	3.199(1)	3.200(2)	3.166(1)	3.203(2)	3.197(1)
Cu/T3	3 × As1	2.4643(3)	2.442(2)	2.404(2)	2.398(2)	2.396(2)	2.383(2)	2.407(2)
	1 × As4	2.5305(5)	2.538(1)	2.484(1)	2.462(2)	2.448(1)	2.472(2)	2.481(1)
	3 × Cu/T1	2.8802(5)	2.862(2)	2.878(2)	2.856(2)	2.856(2)	2.881(3)	2.900(2)
	3 × Cu/T3	2.9220(6)	2.931(2)	2.868(2)	2.843(2)	2.827(2)	2.854(2)	2.864(2)
	3 × Eu1	3.4702(2)	3.471(2)	3.464(2)	3.453(2)	3.442(2)	3.493(2)	3.541(2)
As1	4 × Cu/T1	2.6560(2)	2.654(1)	2.646(1)	2.637(1)	2.614(1)	2.647(1)	2.660(1)
	2 × Cu/T2	2.4462(3)	2.447(2)	2.441(2)	2.417(2)	2.397(2)	2.445(2)	2.462(2)
	2 × Cu/T3	2.4643(3)	2.442(2)	2.404(2)	2.398(2)	2.396(2)	2.383(2)	2.407(2)
	2 × Eu1	3.1814(1)	3.174(2)	3.149(1)	3.138(1)	3.129(1)	3.157(1)	3.194(1)
As2	3 × Cu/T1	2.4887(4)	2.483(2)	2.479(2)	2.474(2)	2.457(2)	2.483(2)	2.485(2)
	3 × Cu/T2	2.4968(3)	2.503(2)	2.514(1)	2.525(2)	2.497(1)	2.536(2)	2.530(1)
	3 × Eu1	3.4326(2)	3.427(2)	3.387(1)	3.373(2)	3.340(1)	3.365(2)	3.361(1)
	1 × Eu2	3.1659(3)	3.174(1)	3.1346(9)	3.122(1)	3.0974(9)	3.139(1)	3.1677(9)
As3	12 × Cu/T1	2.6765(3)	2.6696(7)	2.6540(6)	2.6420(7)	2.6295(6)	2.6563(7)	2.6667(7)
As4	8 × Cu/T3	2.5305(5)	2.538(1)	2.484(1)	2.462(2)	2.448(1)	2.472(2)	2.481(1)

However, exceptions are present, most importantly within the [TAs4] tetrahedra. Figure 5a shows that the T3–As1 and T3–As4 distances display the greatest shrinking upon the substitution, decreasing by 0.07 and 0.08 Å, respectively (Table 3). This is in striking contrast with the minor changes in the T1–As bonding distances as well as the T2–As distances, the latter indicating that the T2As4 tetrahedra distort rather than exhibit a linear decrease in the bonding distances. Consequently, it can be assumed that Ni—having a smaller covalent radius than Cu—prefers the T3 position until it is fully filled at x = 8 (32f site, Z = 4), after which Ni starts to occupy other positions. This is further corroborated by the T-T distances presented in Figure 5b, which shows that the T3–T3 intermetallic distance displays the greatest decrease upon the Ni-for-Cu substitution. The Fe-for-Cu substitution stops at x = 8, and we do not have enough data for the similar analysis of interatomic distances. However, we note that only the T3–As and T3–T3 interatomic distances decrease substantially. Taking into account that the Mössbauer data points at a single position of the iron atoms, we believe that Fe most likely occupies the T3 site (Table 3).

Figure 5

The (a) Cu/Ni–As; and (b) Cu/Ni–Cu/Ni distances in the structure of Eu7Cu44−NiAs23 for different x. Cu/Ni denotes mixed sites of Cu and Ni. The standard deviations are below the sizes of experimental datapoints. The lines are drawn to guide the eye.

2.5. Magnetic Properties

Temperature dependences of the magnetic susceptibility for Eu7Cu36Fe8As23and Eu7Cu42Ni2As23 are presented in Figure 6a,b respectively.

Figure 6

Magnetic susceptibility vs. temperature for (a) Eu7Cu36Fe8As23 and (b) Eu7Cu42Ni2As2; inverse magnetic susceptibility vs. temperature for (c) Eu7Cu36Fe8As23 and (d) Eu7Cu42Ni2As2; (e) magnetization vs. field for Eu7Cu36Fe8As23.

The former sample behaves as a typical Curie–Weiss paramagnet (Figure 6c), with the Weiss temperature only slightly exceeding zero, θ = 0.95 K. A deviation from the Curie–Weiss behavior at low temperatures with a visible increase of the magnetic susceptibility likely stems from minor paramagnetic impurities. The calculated magnetic moment Meff = 8.32 μB per Eu-atom is noticeably higher than the expected value for pure Eu2+ (M = 7.94 μB for J = S = 7/2), indicating that some contribution from iron is also present. Assuming additivity of the magnetic moments, where a square of the effective moment is a sum of the squares of individual moments, we obtain per one Fe-atom, which is in the typical range for Fe-based itinerant magnets (compare to 2 μB in FeSi [15]). Note that the paramagnetic effective moments are calculated as μeff = g[S(S + 1)]1/2μB and, for example, μeff of Eu2+ is 7.94 μB assuming g = 2.0 for J = 0 (4f14). On the other hand, Fe only weakly contributes to the ordered moment (Figure 6e) at low temperatures,

The presence of the magnetic contribution from Fe atoms is also consistent with their high-spin state inferred from the Mössbauer spectra.

In contrast to Eu7Cu36Fe8As23, Eu7Cu42Ni2As23 exhibits FM ordering below 25 K, whereas above T it behaves as a Curie–Weiss paramagnet (Figure 6d) with the effective moment of 7.89 μB, which is only slightly lower than the expected value of 7.94 for Eu2+ (4f7). The Curie–Weiss temperature extracted (Figure 6d) from the high-temperature paramagnetic susceptibility (θ = 25 K) coincides with TC, showing that the FM ordering stems from localized Eu2+cations. It is worth noting that the parent compound Eu7Cu44As23 orders ferromagnetically at 17.5 K; above this temperature, it behaves as a Curie–Weiss paramagnet with the effective moment of 7.94 μB and at 2 K the moment saturates with the saturation moment MS = 7.0 μB. These observations suggest that the localized Eu2+ (4f7) cations undergo FM ordering. Taking into account that the shortest Eu-Eu separation exceeds 4 Å and the compound displays metallic conductivity, one can assume that the Eu2+cations interact through the conduction electrons of the Cu–As clathrate-like framework. The partial substitution of iron for copper in the framework drastically changes the magnetic properties. The FM ordering is lost, and, as long as the Eu–Eu separation does not change substantially (by 0.08 Å only), we believe that the change in the charge carrier concentration within the framework is responsible for the modification of magnetic properties upon doping. In the case of Eu7Cu42Ni2As23, a different substitution picture seems to appear because Ni tends to behave as an effectively d10 atom in many intermetallic and related compounds [16,17]. As a result, it shows no magnetic moment that could interfere the FM ordering of Eu2+cations, which is observed at 25 K.

3. Discussion

Recently we have discovered two new arsenides, namely Eu7Cu44As23 and Sr7Cu44As23, which are the first representatives of a new structure type derived from the intermetallic compound BaHg11. As often observed for ternary arsenides of coinage and alkaline earth metals, these two compounds are the only representatives showing that the new crystal structure is sensitive to the radius of the A-cation. This is not surprising as long as the crystal structure is quite complex and demonstrates a clathrate-like environment of 6/7 of Eu(Sr) atoms by 20 distant Cu and As atoms, whereas the rest of the Eu(Sr) atoms reside in the cubic voids built of eight copper atoms. Such a combination of structure elements requires precise matching of atomic sizes. As a result, isostructural compounds with smaller Ca or larger Ba do not form due to an apparent size mismatch.

Since Eu7Cu44As23 is electronically unbalanced and demonstrates a metallic type of conductivity, we supposed that extended solid solutions could be formed by substituting copper with 3d-elements having lower number of valence electrons but similar atomic radius. Indeed, our assumption proved to be correct and we have observed homogeneity ranges of Eu7Cu44−TAs23 (T = Fe, Co, Ni) extended to 50% for Ni and 20% in the cases of Co and Fe. A close location of Cu and the substituting element T in the Periodic Table resulted in a quite challenging task of determining the dopants distribution among copper sites. The Rietveld refinement against powder X-ray diffraction data was not sensitive enough (as expected); nevertheless, it allowed us to analyze changes in interatomic distances with T for Cu substitution and to indicate the most probable position for the T for Cu substitution. This assumption was facilitated by the 57Fe Mössbauer data for the iron-containing sample, which confirmed that Fe substitutes for Cu only at a single site. Interestingly, the substitution of Cu by Fe leads to suppression of ferromagnetic ordering in Eu-sublattice, while small amounts of Ni increase TC with respect to the parent phase.

The obtained results call for further investigation aimed at expanding our knowledge about this structure type. The main challenge is to rationalize the FM-ordering mechanism in Eu7Cu44As23 and to check for possible magnetocaloric effect near the transition temperature. Another important task is to examine the geometrical and electronic limits for the Eu7Cu44As23 structure type by partially substituting Eu by Na, Ca, or Ce, and As by Sb, Ge, and Te. The respective research is currently in progress.

4. Materials and Methods

4.1. Synthesis and Primary Characterization

The starting materials were ingots of Eu, Cu, Fe, Co, and Ni, as well as As powder of at least analytical grade. The procedure was essentially the same as for the previously reported A7Cu44As23 (A = Eu, Sr) [11]. Prior to use, the Fe, Co, and Ni powders were annealed in hydrogen to remove surface oxide. All operations were performed in an Ar-filled glovebox (M’Braun, p(O2, H2O) < 1 ppm). The elements were mixed according to the composition Eu7Cu44−TAs23, pressed into pellets, loaded in carbon-lined silica tubes, evacuated to ~0.05 mTorr and annealed at 200, 400, 600, and 800 °C (ramp 1 °C/min, soak 12 h). The obtained samples were ground, pressed, and annealed at 800 °C for 48 h, three times. The phase composition was checked using a Bruker D8/Advance diffractometer (CuKα1,2 radiation, LynxEye PSD).

4.2. Crystal Structure Determination

Phase-pure or nearly phase-pure powder samples were used for the crystal structure refinement. PXRD data were collected on Powder X’Pert diffractometer (CuKα1,2 radiation, PANalytical, Almelo, The Netherlands) and processed using the Jana2006 package [12] utilizing the Rietveld method with the crystal structure of Eu7Cu44As23 as a starting model. At the first step, profile parameters and atomic coordinates were refined, while atomic displacement parameters for all atoms were fixed at the value of 0.01 Å2, and the distribution of T (T = Fe, Co, Ni) atoms over three copper sites was set to be random. The attempt to refine the atomic displacement of all atoms simultaneously led to unrealistic values for the As atom (close to zero or negative); consequently, they were fixed, and the T/Cu ratio (T = Fe, Co, Ni) was refined at the Cu sites. Then, this ratio was fixed, and atomic displacement parameters were refined. When the satisfactory values of atomic displacement parameters were obtained, we checked the occupancy of Eu and As atoms—which appeared to be close to unity—and were then fixed at their ideal values. Details of the refinement, refined structural parameters, and selected interatomic distances are collected in Table 1, Table 2 and Table 3, respectively. A typical Rietveld plot is presented in Figure 7.

Figure 7

The Rietveld refinement plot for Eu7Cu36Ni8As23. Experimental profile, green; peak positions, black; differential profile, red.

4.3. Magnetic Properties

Magnetic susceptibility measurements were carried out with the vibrating sample magnetometer (VSM) setup of a Physical Property Measurement System (PPMS, Quantum Design, San Diego, CA, USA). The data were collected in external magnetic fields between 0 T and 14 T in the temperature range of 2–380 K.

4.4. Mössbauer Study

57Fe Mössbauer spectra of the sample implemented within the closed-cycle refrigerator system were recorded between 10 and 50 K using a conventional constant-acceleration spectrometer MS-1104Em in the transmission geometry. The radiation source 57Co(Rh) was kept at room temperature. All isomer shifts are referred to α-Fe at 300 K. The experimental spectra were processed and analyzed using the SpectrRelax program [18].

5. Conclusions

In conclusion, we studied possibilities of copper substitution in the recently discovered clathrate-like compound Eu7Cu44As23. We showed that up to nearly 50% of Cu can be substituted by Ni, and almost 20% can be substituted by Fe and Co. Based on the X-ray structure analysis and Mössbauer spectroscopy, we analyzed the distribution of dopants among Cu sites. We showed that the introduction of even a small amount of Ni increases TС, while Fe doping suppresses ferromagnetic ordering in the Eu-sublattice.