High T_c electron doped Ca₁₀(Pt₃As₈)(Fe₂As₂)₅ and Ca₁₀(Pt₄As₈)(Fe₂As₂)₅ superconductors with skutterudite intermediary layers

Abstract

It has been argued that the very high transition temperatures of the highest T_c cuprate superconductors are facilitated by enhanced CuO₂ plane coupling through heavy metal oxide intermediary layers. Whether enhanced coupling through intermediary layers can also influence T_c in the new high T_c iron arsenide superconductors has never been tested due the lack of appropriate systems for study. Here we report the crystal structures and properties of two iron arsenide superconductors, Ca₁₀(Pt₃As₈)(Fe₂As₂)₅ (the “10-3-8 phase”) and Ca₁₀(Pt₄As₈)(Fe₂As₂)₅ (the “10-4-8 phase”). Based on -Ca-(Pt_nAs₈)-Ca-Fe₂As₂- layer stacking, these are very similar compounds for which the most important differences lie in the structural and electronic characteristics of the intermediary platinum arsenide layers. Electron doping through partial substitution of Pt for Fe in the FeAs layers leads to T_c of 11 K in the 10-3-8 phase and 26 K in the 10-4-8 phase. The often-cited empirical rule in the arsenide superconductor literature relating T_c to As-Fe-As bond angles does not explain the observed differences in T_c of the two phases; rather, comparison suggests the presence of stronger FeAs interlayer coupling in the 10-4-8 phase arising from the two-channel interlayer interactions and the metallic nature of its intermediary Pt₄As₈ layer. The interlayer coupling is thus revealed as important in enhancing T_c in the iron pnictide superconductors.

In early 2008, superconductivity near 26 K was reported (1) in LaFeAsO_0.9F_0.1, marking the discovery of a second family of high T_c superconductors in addition to the cuprates. Since then, intensive effort has been devoted to the search for new Fe-based superconductors, not only in the hope that technologically applicable superconductors may exist in this class, but also because comparison between the Fe pnictides and the cuprates may help to specify the essential ingredients that give rise to high T_c superconductivity. Several new families of high T_c pnictide superconductors have been discovered (2–4). All of these materials are based on layers of edge-sharing FeAs₄ or FeSe₄ tetrahedra (5), virtually always stacked with other intermediary layers in their crystal structures. The intermediary layers range from simple alkali or alkaline earth ions (2, 3) to more complex but still chemically trivial perovskite layers such as Sr₄V₂O₆ (4). Recently, Nohara et al. reported the existence of new superconductors with T_c up to 38 K in the Ca-Fe-Pt-As chemical system (6). Although layered -Ca-(Fe, Pt, As)-Ca-(Fe₂As₂)- stacking was suggested for the structures of the superconductors, neither the crystal structures nor the detailed physical properties were characterized. This report is intriguing because Fe 122 arsenides, such as BaFe₂As₂ (2), form in the ThCr₂Si₂ structure type while Pt 122 arsenides, such as SrPt₂As₂ (7), form in the CaBe₂Ge₂ structure type, reflecting fundamental chemical differences between FeAs and PtAs layers. Here we isolate, identify and report the crystal structures and physical properties of two superconductors in the Ca-Pt-Fe-As system. These materials raise the chemical complexity of the superconducting pnictides to a new level. The first, Ca₁₀(Pt₃As₈)(Fe₂As₂)₅ (10-3-8), has triclinic symmetry, which is extremely rare among superconductors. The second, Ca₁₀(Pt₄As₈)(Fe₂As₂)₅ (10-4-8), is a higher symmetry tetragonal phase. These two superconductors have novel structure types containing platinum arsenide intermediary layers, with formulas Pt₃As₈ and Pt₄As₈, respectively, which are based on the skutterudite structure, a common crystal structure type in binary pnictides. As-As dimers are present in these layers and are critical to understanding the electronic state of the superconductors.

For the triclinic 10-3-8 phase, we find that T_c can be tuned from 0 to 11 K through 6(1)% to 13(1)% of Pt substitution on the Fe site in the (Fe₂As₂)₅ layers, which dopes the layers with electrons. Although the ground state of the undoped 10-3-8 phase remains unclear, the Pt doping study provides evidence for similarities between this new superconducting family and the frequently studied 1111 (such as LaFeAs(O_1-xF_x)) (1) and 122 (such as Ba(Fe_1-xCo_x)₂As₂) (8) families of arsenide superconductors. For the tetragonal 10-4-8 phase, superconductivity occurs at a much higher temperature, 26 K.

These two chemically and structurally similar compounds, with significantly different T_cs, provide a particularly interesting platform for studying superconductivity in the pnictides. Comparison between them indicates that simple rigid band filling ideas and the frequently invoked empirical rule relating T_c to the As-Fe-As bond angle in superconducting iron pnictides are not enough to account for the different T_cs here. Rather the data and analysis suggest that the difference is due to the stronger FeAs interlayer coupling in the 10-4-8 phase, which arises from the two-channel interlayer interactions present in the 10-4-8 structure and the metallic nature of its intermediary Pt₄As₈ layer. This stronger interlayer coupling is thus believed to be an important factor in enhancing T_c in the Fe pnictides. What’s more, bearing in mind that doping the FeAs layers in Fe pnictide superconductors leads to lower T_c than doping the intermediary layers, it may be that T_c can be pushed even higher in these systems through doping on the Ca or Pt_nAs₄ (n = 3, 4) layers exclusively.

Results

Crystal Structure.

The crystal structures of the Ca₁₀(Pt₃As₈)(Fe₂As₂)₅ (10-3-8) and Ca₁₀(Pt₄As₈)(Fe₂As₂)₅ (10-4-8) superconducting phases, which have -Ca-(Pt_nAs₈)-Ca-(Fe₂As₂)- stacking, as well as selected single crystal and powder X-ray diffraction patterns, are shown in Fig. 1. The detailed crystallographic data determined from the single crystal structure refinements are summarized in Tables 1 and 2; appropriate twin laws were incorporated into the structure model. Some streaking of diffraction spots along the c^∗ axis, indicative of stacking faults, was always observed in the 10-4-8 phase crystals. The systematic change in reflection intensities caused by stacking faults may introduce some uncertainty in the refined structural parameters, but the good agreement factor obtained between model and data, R₁ = 5.78%, implies that these uncertainties are small.

Fig. 1.

(Upper) The X-ray powder diffraction pattern of the 10-3-8 phase taken at 300 K. This pattern shows that single phase 10-3-8 can be obtained but the powder X-ray diffraction pattern was not employed in the crystal structure determination. Upper black dots, observed pattern; upper red curve, calculated pattern; tic marks, calculated peak positions using the crystal structure in Table 1; green (lower) curve, difference between observed and calculated pattern. (Inset) The crystal structures of the 10-3-8 and 10-4-8 phases. In the 10-3-8 phase, the occupancy of the Pt sites at the corners is 50%. (Lower Panels) The single crystal X-ray diffraction patterns for the (h0l) zone of the 10-3-8 phase and the (hk0) as well as (h0l) zones of the 10-4-8 phase. , , a^∗ and b^∗ are described in the text.

Table 1.

Crystal structure of 10-3-8 phase at 100 K

The 10-3-8 phase *
Crystal system	Triclinic	Sample size		0.034 × 0.063 × 0.067 mm3
Space group	P - 1 (# 2)	Total reflection		2,229
Z	1	Absorption coefficient		39.203/ mm
Unit cell parameters	a = 8.759(4) Å , b = 8.759(4) Å , c = 10.641(5) Å , V = 788.1(6) Å3α = 94.744(5)°, β = 104.335(5)°, γ = 90.044(5)°
Atomic position
site	Wyck	x/a	y/b	z/c
Ca1	2i	0.3655(4)	0.1218(4)	0.2330(4)
Ca2	2i	0.7745(4)	−0.0751(4)	0.2352(4)
Ca3	2i	0.4386(4)	0.4798(4)	0.7947(4)
Ca4	2i	0.0284(4)	0.6810(4)	0.7658(4)
Ca5	2i	0.1688(4)	0.7278(4)	0.2336(4)
Fe1*	2i	0.1521(7)	0.5495(8)	0.5001(6)
Fe2*	2i	0.2521(7)	0.2518(7)	0.4986(6)
Fe3*	2i	0.3494(5)	−0.0501(6)	0.5004(4)
Fe4*	2i	0.0498(7)	−0.1517(7)	0.4977(5)
Fe5*	2i	0.4516(8)	0.6481(9)	0.5012(8)
Pt1	1c	0	1/2	0
Pt2	1d	1/2	0	0
Pt3*	2i	−0.01687(17)	−0.00552(17)	−0.05612(16)
As1	2i	0.11089(19)	0.03715(19)	0.36734(18)
As2	2i	0.50913(19)	−0.16606(19)	0.36329(18)
As3	2i	0.70762(19)	0.23873(19)	0.36320(18)
As4	2i	0.68943(19)	0.56295(19)	0.63737(17)
As5	2i	0.09241(19)	0.36517(19)	0.63661(18)
As6	2i	0.7344(2)	0.40095(19)	0.00012(19)
As7	2i	0.4011(2)	0.2652(2)	−0.00063(19)
As8	2i	0.2432(2)	0.8819(2)	−0.0001(2)
As9	2i	0.1184(2)	0.2435(2)	−0.0001(2)

R1 = 0.0450 and wR2 = 0.1102 for [F₀ > 4σF₀] and R1 = 0.0902 for all reflections.

^*Fe1, Fe2, Fe3, Fe4 and Fe5 sites represent Fe_1-xPt_x with refined x = 0.040 (2); the site occupancy of Pt3 is 0.5.

Table 2.

Crystal structure of 10-4-8 phase at 100 K

The 10-4-8 phase *
Crystal system	Tetragonal	Sample size		0.096 × 0.046 × 0.033 mm3
Space group	P4/n (# 85)	Total reflection		895
Z	1	Absorption coefficient		41.902/ mm
Unit cell parameters	a = 8.733(3) Å , a = 8.733(3) Å , c = 10.481(4) Å , V = 799.3(5) Å3α = 90°, β = 90°, γ = 90°
Atomic position
site	Wyck	x/a	y/b	z/c
Ca1	2c	3/4	3/4	0.2294(8)
Ca2	8g	−0.1544(4)	0.0477(4)	0.7582(4)
Fe1*	2b	3/4	1/4	1/2
Fe2*	8g	0.6502(2)	−0.0505(2)	0.5008(2)
Pt1*	2a	3/4	1/4	0
Pt2	2c	1/4	1/4	0.06845(17)
As1	8g	0.0102(2)	0.14126(18)	−0.01695(19)
As2	2c	3/4	-1/4	0.6378(4)
As3	8g	0.85009(17)	0.04866(17)	0.3641(2)

R1 = 0.0578 and wR2 = 0.1559 for [F₀ > 4σF₀] and R1 = 0.0809 for all reflections.

^*Fe1 and Fe2 sites represent Fe_1-xPt_x with refined x = 0.03(1); the site occupancy of the Pt1 site is 0.877(9). The refined chemical formula is Ca₁₀(Pt_4-δAs₈)((Fe_0.97Pt_0.03)₂As₂)₅ (δ = 0.246).

The top panel of Fig. 1 shows that the powder X-ray pattern for the 10-3-8 phase can be described well by the determined crystal structure. For both compounds, the strongest reflections reveal a simple tetragonal basal plane subcell with a₀ ∼ 3.91 Å as shown in the (hk0) single crystal X-ray diffraction pattern. However, weaker superlattice reflections, which correspond to a square structural supercell in the real space basal plane, oriented in the (210) direction, are also observed and lead to . This unusual superlattice condition is mathematically equivalent to that seen in K_0.8Fe_2-δSe₂ (9) but arises from a completely different chemical mechanism—the commensurability condition of the (Pt₃As₈) or (Pt₄As₈) skutterudite layers with the Fe₂As₂ layers, rather than partial vacancy ordering. Inspection of the single crystal diffraction patterns in the c^∗ direction indicates that 10-4-8 phase has a primitive tetragonal cell, while the 10-3-8 phase shows significant shifting of the stacking between neighboring layers. This layer shift results in a triclinic unit cell, despite the fact that the basal plane cell is essentially square—i.e., with a = b and γ ≈ 90°.

The 10-3-8 and 10-4-8 phases are new structure types that can be classified as derivatives of the SrZnBi₂ and SrZnSb₂ structures, respectively (10), as shown in the top left panel of Fig. 2. As variants of the more common ThCr₂Si₂ structure, which accounts for many of the high T_c Fe pnictide superconductors, the SrZnSb₂ and SrZnBi₂ structure types have every other M₂X₂ tetrahedral layer in AM₂X₂ replaced with a square X₂ layer of the same size, as shown in Fig. 2. The only difference between the SrZnSb₂ and SrZnBi₂ structures is that one is body centered while the other is primitive. This difference in centering is allowed chemically because the X₂ layer is not sensitive to the adjacent A cation arrangement, resulting in the fact that both staggered and nonstaggered stackings can form around this layer. The difference in stacking has a profound effect in the superconducting phases, in which the X₂ layer is replaced by a more complex Pt-As layer where some potential Pt sites are forced to be vacant depending on the A site stacking.

Fig. 2.

(Top Left) The crystal structures of SrZnSb₂ and SrZnBi₂ (10), which can be considered parent phases of the new Ca-Fe-Pt-As superconductors. (Top Right) Schematic illustration of the transformation from the X₂ layer in the parent phases to the Pt_nAs₈ skutterudite-like layers in the superconductors (the center interstitial Pt is only present in the Pt₄As₈ layer). Displacements are exaggerated for clarity. (Bottom) Comparison of the crystal structures of skutterudite-structure IrAs₃ (11), the platinum arsenide layer in BaPt₄As₆ (12), and the Pt₃As₈ intermediary layer in the 10-3-8 superconducting phase. As-As bonds are shown bolded.

The X₂ layers are transformed to Pt₃As₈ and Pt₄As₈ layers in Ca₁₀(Pt₃As₈)(Fe₂As₂)₅ and Ca₁₀(Pt₄As₈)(Fe₂As₂)₅, respectively. This transformation is illustrated schematically in the top left panel in Fig. 2. Starting from the square lattice of As atoms, the substitution of 1/5 of the As with Pt and the insertion of interstitial Pt lead to strong displacements of the As from their ideal positions. This occurs so that intralayer As-As dimers are formed and Pt_nAs₈ (n = 3, 4) skutterudite-like layers emerge. The periodicity of these layers is based on the size of the the Pt sublattice, which matches the FeAs lattice size with . Thus, the Fe₂As₂ and Pt_nAs₈ layers become commensurate in the (210) Fe₂As₂ direction (Fig. 2, upper right). The resulting Pt₃As₈ intermediary layer for the 10-3-8 superconductor is illustrated in the right bottom panel of Fig. 2. The layer consists of a square lattice of corner-sharing PtAs₄ squares with a rotation of approximately 25° about an axis perpendicular to the plane, governed by the formation of intraplanar As-As dimers. This arrangement of atoms is unique in the superconducting iron pnictides but is fairly common in platinum group pnictides. A simple example of a compound where such rotations of corner shared MX₄ squares is dictated by the formation of pnictide-pnictide bonds is skutterudite IrAs₃ (11), shown in the lower panel of Fig. 2. In this compound the arrangement of As atoms has the same projected in-plane structure, but all the Ir atoms are octahedrally coordinated rather than having the lower coordination observed for Pt in the 10-3-8 and 10-4-8 phases. Platinum-based compounds with very similar structures have also been observed, such as BaPt₄As₆ (12), shown in Fig. 2, where 1/2 of the Pt are octahedrally coordinated and the rest are square planes. In both cases the rotations have an out-of-plane component, which causes the spacing between Pt atoms to contract. The As-As bond distance in the dimers is quite similar in these two compounds, 2.47 Å–2.54 Å in IrAs₃ and 2.41 Å–2.42 Å in BaPt₄As₆. These distances are comparable to the ones in the new Ca-Fe-Pt-As superconductors where they are 2.48 Å–2.49 Å in the 10-3-8 phase and 2.50 Å in the 10-4-8 phase.

The limited range of the As-As bond lengths present in the new superconductors indicates that the size of the Pt sublattice is constrained by both the As-As dimer size and the Fe₂As₂ sublattice size. The net result of these strong constraints is that only 1/2 of the total number of square sites in the plane are large enough to contain a Pt atom in a simple square planar coordination with As. The other 1/2 Pt cannot sit exactly in the middle of the remaining squares but instead must be displaced to a position about 0.5 Å above or below the plane. This displacement causes a conflict with the Ca ions that are adjacent to the Pt-As layers, resulting in the difference in symmetry and formula for the primitive and body centered structures. This situation is illustrated in Fig. 3, which compares the Pt₃As₈ and Pt₄As₈ layers in detail. The Pt atoms shown in blue are the ones called substitutional Pt atoms in Fig. 2. The Pt atoms shown in red (the ones called interstitial Pt atoms in Fig. 2) are critical to the Pt stoichiometry of the layers. The differences can be seen by focusing on the parts of the structures that are encircled by the dashed ovals in the figure. In the primitive 10-4-8 case (top panel), for each interstitial Pt site, only one side of the plane is blocked by Ca, so that Pt atoms can sit above the plane on one site and below the plane on the other. In the 10-3-8 case, on the other hand (bottom panel), one of the potential interstitial Pt sites is blocked on both sides of the plane by Ca ions, and thus no Pt atoms can sit on these sites; for the other potential interstitial sites, however, Ca does not interfere, and Pt can occupy a position on either side of the plane. Pt cannot occupy both sides of the plane at once, because then the Pt-Pt separations would be too small; thus the above-plane and below-plane sites are randomly occupied with a 50% probability. We do not have any evidence that there is any long range ordering to lift this disorder. This difference in the blocking of the interstitial Pt sites, due to the different arrangements of neighboring Ca, accounts for the difference in formulas of the two new superconductors because the 10-3-8 structure can only accommodate filling of 1/2 of the interstitial Pt sites while the 10-4-8 structure allows filling of all of them.

Fig. 3.

(Top) The detailed structure of the Pt₄As₈ layer in Ca₁₀(Pt₄As₈)(Fe₂As₂)₅. (Bottom) The detailed structure of the Pt₃As₈ layer in Ca₁₀(Pt₃As₈)(Fe₂As₂)₅. The dashed lines show the out-of-plane Pt-As bonding. The dashed ovals show the regions of Pt-Ca positional conflict. The bolded black lines show the As-As dimers. The green lines show the basal plane edges in one unit cell.

Physical Properties.

In all preparations of the 10-3-8 and 10-4-8 superconductors, we found some fraction of Pt substitution on the Fe site in the FeAs layers. (The determination of the Pt doping concentration is described in detail in Methods. To easily compare the physical properties of the new superconductors with the other Fe pnictide superconductors, which have much simpler formulas, the units of molar susceptibility, magnetization, and heat capacity presented are normalized to one (Fe_1-xPt_x)₂ per formula unit.)

As a representative of the 10-3-8 phase, the picture of the mm-size single crystals and the physical properties of the x = 0.09(1) sample are shown in Fig. 4. Fig. 4A shows the electrical resistivity, ρ(T), from 2 to 300 K. As temperature decreases, the resistivity slowly decreases from 0.7 mΩ-cm at 300 K to 0.61 mΩ-cm near 190 K, indicated with the red arrow, and then increases monotonically to 1.1 mΩ-cm at 12 K, followed by a sharp decrease to zero at the superconducting transition. The magnitude of the normal-state resistivity is similar to that seen in other Fe pnictide superconductors, an indication of the “poor metal” nature of the phase. Fig. 4B presents the normal-state magnetic properties. At 10 K, the magnetization M is linearly proportional to H and extrapolates to 0 at H = 0, indicating that no ferromagnetic impurities are present in the single crystal sample. Thus the temperature-dependent susceptibility, χ(T), was measured at 5 T and calculated as M(T)/H. Magnetic anomalies, such as are often associated with structural or magnetic phase transitions in the pnictide superconductors, were not observed for the 10-3-8 compound, with either H∥ab or H⊥ab. The upturn in χ below 80 K may be attributed to paramagnetic impurities in the sample. From 80 to 300 K, χ(T) increases approximately linearly with temperature. The ratio of χ^∥ab over χ^⊥ab at 300 K is approximately 2. This value is larger than the ratio of 1.6 observed in CaFe₂As₂, indicating higher anisotropy in the 10-3-8 superconductor. The Hall coefficient, R_H, shown in the inset of Fig. 4C, is negative at all temperatures, indicating that electron carriers are dominant in this compound. If a single band model is assumed, then the carrier concentration n(T) can be estimated as -1/eR_H. This is plotted as a function of temperature in Fig. 4C. n decreases from 5.5 × 10²¹ cm^-3 at 300 K (3.2 times of that of LaFeAsO_0.89F_0.11) to 0.74 × 10²¹ cm^-3 at 12 K. A slope change is observed in n(T) at around 190 K, indicated by the red arrow, at the same temperature where the minimum normal-state resistivity is found (Fig. 4A). The Seebeck coefficient, shown in Fig. 4D, is negative throughout the measured temperature range, with a room temperature value of -24.3 μV/K and minimum value of -34.3 μV/K near 150 K; this again indicates the dominant role of electrons in the transport.

Fig. 4.

The characterization of the normal-state properties for the 10-3-8 phase Ca₁₀(Pt₃As₈)((Fe_1-xPt_x)₂As₂)₅ with x = 0.09(1). (A) Temperature-dependent in-plane electrical resistivity ρ. (B) The temperature dependent magnetic susceptibility, M(T)/H, taken at 5 T with H∥ab and H⊥ab. (Inset) The field dependent magnetization data, M(H), taken at 10 K with H∥ab. (Inset) 10-3-8 single crystals and 1 mm scale. (C) The estimated temperature dependent carrier density n. (Inset) The temperature dependent Hall coefficient R_H. (D) The temperature dependent in-plane Seebeck coefficient, S.

Fig. 5 presents the physical properties of the superconducting state for the triclinic 10-3-8 phase. The zero-field-cooled (ZFC) and field-cooled (FC) DC susceptibility measurements were performed at 2 mT with H∥ab so that the demagnetization effect could be minimized. The diamagnetic signal observed below 9 K in both ZFC and FC measurements confirms the bulk superconductivity in this compound and is consistent with the resistivity measurements. The shielding fraction estimated from the ZFC data is around 120%, similar to what is observed in transition metal doped BaFe2As₂ (8). The Meissner fraction inferred from the FC data is only about 6% due to the flux pinning, which is the usual case in the Fe pnictide superconductors (8). The temperature-dependent C_p(T)/T data is presented in Fig. 4B. A feature in C_p/T associated with the superconducting transition is observed near 9 K, confirming the bulk superconductivity. The subtle character of this feature could be due to a distribution of T_cs in the sample or could be intrinsic. An empirical relationship has been observed for unannealed Fe pnictide superconductor families (13). For the 122 superconductors with similar T_c to the 10-3-8 phase, such as Ba(Fe_0.885Co_0.115)₂As₂ (T_c ∼ 9 K) and Ba(Fe_0.965Co_0.035)₂As₂ (T_c ∼ 7 K), the heat capacity feature at T_c is also small (14). The inset shows the low temperature C_p/T vs. T² plot, from 2 to 5 K. The data can be fitted with C_p = γT + βT³, where γ = 4.5(1) mJ/mole-K² and β = 1.06(1) mJ/mole-K⁴. ρ(T)/ρ(300 K,0 T), measured at 9, 7, 5, 3, 2, 1, 0.5, 0.2, 0 T with H∥ab and H⊥ab is presented in Fig. 5C. With applied field, T_c is suppressed to lower temperatures and the resistive transition broadens, indicating the presence of strong thermal fluctuation of the vortices. This is different from what is observed in the Ba(Fe_1-xCo_x)₂As₂ superconductors, where no broadening was observed (8) but is reminiscent of that in RFeAs(O_1-xF_x) and cuprates (15). To determine T_c at each field, 90%, 50% and 10% of the normal-state resistance at 16 K are used as the criteria. At 0 T, , , and . For all three criteria, the curves show roughly linear behavior. The curve, however, changes from a concave shape for the 90% criterion to a convex shape for the 10% criterion. As a compromise, we focus on the H_c2 values inferred from the 50% criterion, shown in Fig. 5D. With 9 T applied field, the T_c was suppressed to 0.9T_c0 with H∥ab and 0.5T_c0 with H⊥ab. The single band WHH theory, without taking into account the effects of spin paramagnetism and spin-orbit scattering is used to fit the H_c2 curves (16). This fit is shown as the solid curve in the panel. The resulting , and . The anisotropy parameter decreases from 100 near T_c0 to 25 at 0.9T_c0, as shown in the inset of Fig. 5D. In accordance with the Ginzburg–Landau theory, and , the coherence lengths are estimated to be ξ_//ab(0) = 50 Å and ξ_⊥ab(0) = 12 Å.

Fig. 5.

The characterization of the superconducting properties for the 10-3-8 phase Ca₁₀(Pt₃As₈)((Fe_1-xPt_x)₂As₂)₅ with x = 0.09(1). (A) Enlargement of the temperature dependent electrical resistivity near T_c and the field-cooled/zero-field-cooled DC susceptibility M(T)/H with H∥ab. (B) C_p(T)/T vs. T. (Inset) C_p(T)/T vs. T². The solid line shows the fit using C_p(T)/T = γ + β T². (C) The normalized resistivity, ρ/ρ(300 K,0 T) taken at 9, 7, 5, 3, 2, 1, 0.5, 0.2 and 0 T with H∥ab and H⊥ab. The 10%, 50% and 90% criteria are shown. (D) The inferred anisotropic H_c2(T) for H∥ab and H⊥ab using the 50% criterion. The solid line is the single band WHH fit (16). (Inset) The anisotropic parameter .

We have been successful in tuning the ground state of the 10-3-8 phase from normal to superconducting by controlling the Pt concentration. The effect of Pt doping in the 10-3-8 phase is summarized in Fig. 6. Because the samples are easily exfoliated, and thus the direct comparison of the resistivities among them are not suitable (8, 17) (the resistivities at 300 K are all in the 1 mΩ-cm range), the normalized resistivity, ρ/ρ(300 K), is employed in the figure. The enlarged ρ/ρ(300 K) near T_c is shown in Fig. 6A. Zero resistivity was not observed in x = 0.06(1) sample but does appear at higher Pt concentrations; superconductivity shows up in the x = 0.07(1) sample with and increases to in the x = 0.09(1) sample, in x = 0.13(1) sample. The nature of the bulk superconductivity in these samples is confirmed by the AC susceptibility data, presented in Fig. 6B, which shows large diamagnetic throws with similar magnitudes. The T_cs inferred from both types of measurements are consistent with each other and summarized in Table 3. Fig. 6C shows the evolution of ρ/ρ(300 K) from 2 to 300 K with doping [from x = 0.06(1) to 0.13(1), each subsequent dataset is shifted upward by 0.3 for clarity]. With decreasing temperature, the resistivity of the x = 0.06(1) sample slowly decreases to a minimum around 170 K and then monotonically increases; a slope break is observed around 60 K. This resistivity shape is reminiscent of the one in underdoped Ba(Fe_1-xCo_x)₂As₂ (8), which is associated with the structural and magnetic phase transits. No magnetic anomaly was observed from 2 to 300 K for this compound, however. For x = 0.07(1) and 0.09(1) compounds, upon cooling, the resistivity decreases slowly first and then increases before dropping to zero; no slope break is observed. The x = 0.13(1) sample shows a quite different resistivity shape, which decreases continuously from 300 to 50 K, followed by a subtle increase and then a drop to zero. This shape is reminiscent of the nearly optimally doped Ba(Fe_1-xCo_x)₂As₂ (8). The Seebeck coefficient data from 40 to 300 K is presented in Fig. 6D. S(T) are all negative in measured temperature, implying the dominant role of the electron carriers. At 300 K, S decreases from -17 μV/K for x = 0.07(1) to -37 μV/K for x = 0.13(1), indicating that the Pt doping is electron doping.

Fig. 6.

The characterization of the doping-dependent properties for the 10-3-8 phase Ca₁₀(Pt₃As₈)((Fe_1-xPt_x)₂As₂)₅. (A) The enlarged normalized resistivity, ρ(T)/ρ(300 K), near T_c for the 10-3-8 series. (B) Temperature-dependent AC susceptibility data, M(T)/H, taken at 0.5 mT with H∥ab in the 10-3-8 phase. (C) The normalized resistivity, ρ(T)/ρ(300 K), from 2 to 300 K. Each subsequent dataset is shifted upward by 0.3 for clarity. (D) The Seebeck coefficient S(T).

Table 3.

Starting material ratio in sample growth

Phase	Starting ratio Ca∶Fe∶Pt∶As	EDS Ca∶Fe∶Pt∶As	x	(K)	(K)
10-3-8	2∶2∶0.4∶4	2∶1.93(3)∶0.70(1)∶3.86(5)	0.06(1)	0	0
	2∶2∶0.5∶4	2∶1.89(3)∶0.73(1)∶3.83(5)	0.07(1)	5.9	4.9
	4.3∶2∶0.7∶6.3	2∶1.86(2)∶0.77(1)∶3.88(2)	0.09(1)	9.6	8.2
	3∶2∶1∶5	2∶1.79(4)∶0.88(2)∶3.7(1)	0.13(1)	9.9	10.6
10-4-8	2∶1.8∶0.9∶3.5 *			26.4	25

The EDS measured ratio of Ca∶Fe∶Pt∶As and the resulting x in Ca₁₀(Pt₃As₈)((Fe_1-xPt_x)₂)₅ for the 10-3-8 phase. is the T_c obtained from the 50% criterion from the resistivity data. is the T_c obtained from the AC susceptibility data.

^*The chemical formula of this compound is Ca₁₀(Pt_4-δAs₈)((Fe_0.97Pt_0.03)₂As₂)₅ (δ = 0.246), determined from the single crystal structure refinement.

The single crystal X-ray refinement shows that there is Pt deficiency on the Pt₄As₈ layer and there is also Pt substitution for Fe in the FeAs layer, resulting in the formula for this sample of Ca₁₀(Pt_4-δAs₈)((Fe_0.97Pt_0.03)₂As₂)₅ (δ = 0.246). ρ(T) from 2 to 300 K is shown in Fig. 7. The resistivity monotonically decreases with decreasing temperature from 0.6 mΩ-cm to 0.24 mΩ-cm followed by a sharp drop to zero at and . The bulk superconductivity is confirmed by the low field DC susceptibility, shown in Fig. 7B. T_c is inferred as 26 K. The estimated shielding fraction is around 100%. The normalized resistivity, ρ(T)/ρ(300 K,0 T), with field along the ab plane and the c axis is shown in Fig. 7C. At 0 T, is 26.4 K. The H_c2 data was inferred using the 50% criterion and is summarized in Fig. 7D. The T_c was only suppressed to 0.9T_c0 with H∥ab and 0.7T_c0 with H⊥ab at 9 T. Comparing with the x = 0.13(1) 10-3-8 phase, less anisotropy was observed. The anisotropic parameter Γ shown in the inset is 32 at 0.9T_c0 and 14 near T_c0 while this value is 25 at 0.9T_c0 and 100 near T_c0 in x = 0.13(1) 10-3-8 phase.

Fig. 7.

The characterization of the superconducting properties for the 10-4-8 phase Ca₁₀(Pt_4-δAs₈)((Fe_0.97Pt_0.03)₂As₂)₅ (δ = 0.246). (A) The resistivity taken at H = 0 T from 2 to 300 K. (B) The field-cooled/zero-field-cooled DC susceptibility M(T)/H with H∥ab. (C) The normalized resistivity, ρ/ρ(300 K,0 T) taken at 9, 7, 5, 3, 1, 0.5, and 0 T with H∥ab and H⊥ab. The 50% criterion is shown. (D) The inferred anisotropic H_c2(T) for H∥ab and H⊥ab using the 50% criterion. The dashed line is the guide for the eyes. (Inset) The anisotropic parameter .

Discussion

Although systematic composition-dependent physical properties measurements of the 10-4-8 phase are currently not available, the lower T_c triclinic 10-3-8 phase can be readily compared to the other Fe pnictides: (i) From 80 to 300 K, a linear temperature dependence of M(T)/H has also been observed in all the 1111 and 122 Fe pnictide superconductors, such as LaFeAs(O_1-xF_x) and Ba(Fe_1-xCo_x)₂As₂ (8, 18). In the latter case, the linear temperature dependence persists up to 700 K, the highest temperature measured. It has been suggested that the linear temperature dependence of M(T)/H is related to antiferromagnetic spin fluctuations (19). (ii) Upon doping, the resistivity changes from a “semiconducting” appearance in the underdoped region to a “metallic” appearance near the optimal doping concentration. This systematic change is reminiscent of what is seen in other Fe pnictides, where the “semiconducting” appearance comes from the formation of an SDW gap (1, 8). However, no magnetic anomalies implying a magnetic phase transition were observed from 2 to 300 K in the 10-3-8 phase. It is possible that the doping level is still too high, even in the lowest Pt-content samples, and that the SDW transition exists but has already been fully suppressed. Further investigation is needed to clarify the ground state of the undoped or very underdoped phase. (iii) The anisotropy parameter ranges from 100 near T_c0 to 25 at 0.9T_c0. This is much larger than the 1 to 4 in (Ba_0.55K_0.45)Fe₂As₂ (20) and even larger than the ones in RFeAsO_0.8F_0.2 (15) and indicates a highly anisotropic 2D nature for the 10-3-8 superconductor.

The T_c in the 10-4-8 phase is almost twice the highest T_c we observed in the 10-3-8 phase. Although we have not yet successfully overdoped the 10-3-8 phase, the almost linear normal-state resistivity of the x = 0.13(1) 10-3-8 compound, indicates that this compound is very close to the optimal doping and thus represents the nearly maximum T_c obtainable in the triclinic 10-3-8 phase in the Ca-Pt-Fe-As system. Although we never observed a T_c higher than 26 K in the 10-4-8 phase, a T_c of 38 K was reported by Nohara et al., for a phase of unreported formula with a crystallographic cell corresponding to our tetragonal 10-4-8 phase (6), suggesting that T_c can be even higher in the 10-4-8 phase when an optimal doping condition is obtained. Simple rigid band filling ideas is not enough to account for the T_c difference in these two compounds. If we assign Ca₁₀(Pt₃As₈)(Fe₂As₂)₅ as the parent compound, the optimally doped T_c = 11 K 10-3-8 sample has 0.26 doped electrons per Fe, while the T_c = 26 K 10-4-8 sample has 0.21 doped electrons per Fe, which is not enough to account for the difference in T_c.

It is thus of interest to compare these two chemically and structurally similar compounds, which show quite different T_cs, in more detail. From the chemical point of view, we can straightforwardly model the effective charges in these compounds. The common assignments of Fe²⁺ and As^3- ions in the FeAs layer gives the chemical formula [(Fe₂As₂)₅]^10-. In the Pt_nAs₈ layer, due to the formation of strong As-As dimers in both compounds, the As in Pt_nAs₈ layer contributes 1 electron to the As-As bond, forming . For reduced compounds such as skutterudites and those studied here, the stability of the d⁸ configuration for Pt rules, and the only oxidation state observed for platinum is +2; this d⁸ configuration is consistent with the observed Pt-As coordination polyhedra in the superconductors. In their undoped form, this yields an intermediary layer of [Pt₃As₈]^10- for the 10-3-8 phase and [Pt₄As₈]^8- for the 10-4-8 phase, leading to a major chemical and electronic difference between the two superconductors. Ca₁₀(Pt₃As₈)(Fe₂As₂)₅ is a valence satisfied compound through the Zintl concept (21)—the [Pt₃As₈]^10- layer is perfectly charge balanced by the [Ca₁₀]²⁰⁺ and [Fe₁₀As₁₀]^10-, leading to a semiconducting nature for the Pt₃As₈ intermediary layer (i.e., it will not contribute density of states at E_f) and thus will result in weak FeAs interlayer coupling through this intermediary layer (22). The 10-4-8 phase on the other hand has one more Pt atom in the Pt₄As₈ intermediary layer, exceeding its valence satisfaction requirements, indicating that this layer is likely to have states at E_f and is therefore metallic in character—thus leading to stronger FeAs interlayer coupling. Enhanced coupling through intermediary layers has been suggested as the origin of the very high transition temperatures in the highest Tc cuprates (22), but in the iron arsenide superconductors this is the only system where this possibility could be tested.

Structural differences may also have an impact on the superconductivity, but the usual considerations in the arsenide superconductors do not explain the observations in the present materials. With a few exceptions, an empirical rule between T_c and α_As-Fe-As, the As-Fe-As bond angle, or Pn height, the distance between the adjacent Fe and As layer, has been observed in the Fe arsenides (23–25). This empirical rule implies that T_c is enhanced for As-Fe-As bond angles near that of an ideal tetrahedron, 109.47°, or for particular values of P_n ∼ 1.38 Å . For our compounds, α_As-Fe-As in the 10-3-8 phase has an average value of 108.99° while α_As-Fe-As in the 10-4-8 phase has an average value of 107.63°. Because the 10-4-8 phase has a higher T_c than the 10-3-8 phase, our compounds do not follow the empirical rule of angles. Similarly, the average Pn is 1.40 Å in the 10-3-8 phase and 1.43 Å in the 10-4-8 phase, which also does not follow the empirical rule of heights. Thus the structural differences usually credited with governing the T_c in the arsenide superconductors cannot be operating in our phases, supporting our argument that it is the metallicity and resulting enhanced intralayer coupling in the 10-4-8 phase that determine its higher T_c.

Significant differences in the Pt-As interactions between Pt₃As₈/Pt₄As₈ intermediary layers and the neighboring FeAs layers are also present, further strengthening our argument about the importance of the interlayer coupling in determining the T_cs. In the 10-3-8 phase there is only one interlayer Pt-As interaction channel per unit cell with the Pt-As distance of 3.20 Å (bottom left of Fig. 3, indicated by the dashed line), and it appears to be random whether the Pt-As bond points “up” or “down.” This not only leads to weaker interlayer coupling from the structural perspective but also helps to electronically isolate the Pt₃As₈ intermediary layers from the FeAs layers, making that layer more electronically blocking and reinforcing its semiconducting nature. In the 10-4-8 compound, on the other hand, due to two interlayer Pt-As interaction channels rather than one (top left of Fig. 3, indicated by the dashed line), a shorter Pt-As distance of 3.08 Å , and their fully structurally ordered character, stronger interlayer coupling is realized and the metallic nature of the Pt₄As₈ layer is reinforced. What’s more, the overall tetragonal symmetry of the phase promotes good orbital overlap between layers. The anisotropic H_c2 measurements in the superconducting state, which show the 10-3-8 phase to be much more anisotropic than the 10-4-8 phase, support the chemical picture. Thus though one can infer that both compounds should show strong 2D character in their electronic structures, the 10-4-8 compound will exhibit more and better hybridized electronic states associated with the stronger interlayer interactions, which we argue leads to its higher T_c. Our study reveals the importance of strong FeAs interlayer coupling in enhancing T_c in the Fe pnictides and suggests that further searches for superconducting iron arsenide phases with metallic intermediary layers rather than the commonly found insulating intermediary layers may be a fruitful path for obtaining higher T_cs in the pnictide superconductor family.

Methods

To prepare the crystals, CaAs, FeAs, Fe, Pt, and As were mixed in an argon-filled glovebox, pressed into pellets, and put into alumina crucibles. The crucibles were then sealed in quartz tubes under 1/3 atmosphere of Ar. For heating temperatures between 700 °C and 950 °C, a polycrystalline mixture of one or both superconducting phases together with a considerable amount of PtAs₂ was obtained. For heating temperatures above 1,100 °C, single crystalline 10-3-8 or 10-4-8 phases intergrown with PtAs₂ were obtained. For the 10-3-8 phase, 3 × 2 × 0.5 mm³ size single crystals with no intergrown impurities could be obtained when excess CaAs was added to the mixture, with the Pt doping concentration manipulated by tuning the Fe to Pt ratio; starting material ratios are summarized in Table 3. The crystal growth tubes were heated to 1,100–1,180 °C, held for one week, furnace-cooled or cooled by 5 °C/h to 975 °C, and then water quenched. The plate-like 10-3-8 single crystals were then separated by washing out the ionic CaAs in distilled water. For the 10-4-8 phase, the tubes were heated to 1,100–1,180 °C, held for one week, furnace-cooled to 900 °C, held for 1 d, and then water quenched; pure 10-4-8 phase single crystals with 0.5 × 0.5 × 0.03 mm³ size can be obtained. Growth details for the 10-4-8 phase are also summarized in Table 3.

Crystal structure determination for both new phases was performed on single crystals at 100 K using a Bruker Apex II single crystal X-ray diffractometer with graphite-monochromated Mo K_α radiation (λ = 0.71073 Å ). The sample used for single crystal structure refinement of the 10-3-8 phase was chosen from the x = 0.07(1) batch. The sample used for single crystal structure refinement of the 10-4-8 phase was cut from the same piece used for the resistivity and anisotropic H_c2 measurement. Unit cell refinement and data integration were performed with Bruker APEX2 software package. Unit cell determination was aided by the CELL NOW program (26). The crystal structures were refined using SHELXL-97 (26) implemented through WinGX (27). X-ray powder diffraction patterns were collected on a Bruker D8 Focus diffractometer employing Cu K_α (λ ∼ 1.5406 Å) radiation and a graphite diffracted beam monochromator. Rietveld refinement was carried out using the FULLPROF program suite (28). The Bragg peaks were refined using the Thompson–Cox–Hastings pseudo-Voigt function convoluted with an axial divergence asymmetric peak shape; [001] preferred orientation was included in the powder refinements.

For the 10-3-8 phase, the Pt doping concentrations were determined using energy dispersive X-ray spectroscopy (EDS) in an FEI Quanta 200 FEG Environmental-SEM through measuring the percentages of elements present. By making several EDS measurements on each sample characterized, the average percentage of each element was obtained. Because for the 10-3-8 phase the single crystal X-ray measurements show there is neither Pt deficiency nor Fe substitution in the Pt₃As₈ layer, resulting in the formula Ca₁₀(Pt₃As₈)((Fe_1-xPt_x)₂As₂)₅, the average Pt percentage measured by the EDS measurement leads to an x₁, and the average Fe percentage leads to an x₂ that were used to determined x as 1/2(x₁ + x₂) and the error in composition as 1/2|x₁ - x₂|. The EDS results are summarized in Table 3. Using the single crystal growth method described in the paper, we have not successfully obtained 10-3-8 crystals with x less than 0.06. For the 10-4-8 phase, because the crystals for single crystal X-ray, resistivity, anisotropic H_c2, and EDS measurements were cut from the same exact piece, the formula of the sample was determined from the single crystal X-ray crystal structure refinement.

DC magnetization, M(H) and M(T), were measured in a Quantum Design (QD) Magnetic Properties Measurement System (MPMS) superconducting quantum interface device (SQUID) magnetometer. For the 10-4-8 phase, due to the smallness of the single crystals, the sample used in the χ(T) measurement was a composite of several single crystals that were carefully aligned so that the applied field was parallel to the ab planes.

Heat capacity (relaxation method), AC susceptibility, resistivity, and Hall coefficient measurements were performed in a QD Physical Properties Measurement System (PPMS). The standard four-probe technique was employed for the resistivity measurements (I = 1 or 0.3 mA). A four-wire geometry was used in the Hall coefficient measurements. To remove the magnetoresistive components, the polarity of the magnetic field (H⊥ab) was switched. In both resistivity and Hall effect measurements, the four thin platinum/gold wires employed were attached to the sample with Epotek H20E silver epoxy. Seebeck coefficient measurements were performed with a modified MMR Technologies SB100 Seebeck measurement system.