EuFe$_2$As$_2$ under high pressure: an antiferromagnetic bulk superconductor

We report the ac magnetic susceptibility $\chi_{ac}$ and resistivity $\rho$ measurements of EuFe$_2$As$_2$ under high pressure $P$. By observing nearly 100% superconducting shielding and zero resistivity at $P$ = 28 kbar, we establish that $P$-induced superconductivity occurs at $T_c \sim$~30 K in EuFe$_2$As$_2$. $\rho$ shows an anomalous nearly linear temperature dependence from room temperature down to $T_c$ at the same $P$. $\chi_{ac}$ indicates that an antiferromagnetic order of Eu$^{2+}$ moments with $T_N \sim$~20 K persists in the superconducting phase. The temperature dependence of the upper critical field is also determined.

The discovery of superconductivity (SC) at a transition temperature T c = 26 K in LaFeAsO 1−x F x by Kamihara et al. 1 has triggered extensive studies of SC in layered iron pnictides and related compounds. Rotter et al. found that BaFe 2 As 2 with a simpler structure can be made superconducting by doping: T c = 38 K in (Ba 1−x K x )Fe 2 As 2 with x = 0.4. 2 Perhaps more importantly, it is reported that 122 compounds of the form AFe 2 As 2 (A = Ca, Sr, Ba, and Eu) can be tuned to SC by the application of high pressure P . 3-10 P tuning can provide opportunities to determine the nature of the iron-pnictide high-temperature SC without being adversely affected by disorder due to doping. However, most of these reports are based only on resistivity ρ measurements and hence cannot establish the bulk nature of P -induced SC. 11 Even when magnetic measurements are reported, results are not conclusive: In ref. 5, magnetic measurements were performed on SrFe 2 As 2 and BaFe 2 As 2 , but the observed volume fraction was expressed in arbitrary units. In ref. 9, the volume fraction of the P -induced superconducting phase of CaFe 2 As 2 was estimated to be at least 50%, while in ref. 12 CaFe 2 As 2 was reported not to exhibit SC under hydrostatic P produced by the use of helium as a pressure-transmitting medium.
EuFe 2 As 2 exhibits two phase transitions, at T o ∼ 190 K and T N ∼ 19 K, at ambient P . 13

Letter
The transition at T o is a combined structural and magnetic transition, similar to those in the other 122 compounds: the crystal structure changes from tetragonal to orthorhombic and the Fe 2+ moments order antiferromagnetically. The transition at T N is due to the antiferromagnetic (AFM) ordering of the Eu 2+ moments. The AFM coupling of the Eu 2+ moments is rather weak: the field-induced paramagnetic state with a saturated moment of ∼7 µ B /Eu is easily reached by the application of ∼1 or 2 T in the ab-plane or along the c-axis, respectively. 17 A temperature (T )-P phase diagram has been determined from ρ measurements: 10 while T o decreases with P and is not detected above P = 23 kbar, T N is nearly P -independent up to 26 kbar (the highest P in ref. 10). The authors of ref. 10 state that P -induced SC at T c ∼30 K occurs above 20 kbar. However, their ρ data (at P = 21.6 kbar) shows only a partial drop and approximately half of the normal-state ρ appears to remain as T → 0. Obviously, further experimental confirmation is necessary.
In this letter, we report measurements of the ac magnetic susceptibility χ ac and ρ of EuFe 2 As 2 single crystals under high P . By observing a nearly 100% shielding volume fraction and a sharp resistive transition to the zero-resistivity state at P = 28 and 29 kbar, we establish that EuFe 2 As 2 is a bulk superconductor at these values of P . We also show evidence that the AFM order of the Eu 2+ moments persists in the superconducting phase.
A single-crystal ingot of EuFe 2 As 2 was grown by the Bridgman method from a stoichiometric mixture of its constituent elements. A 3 He/ 4 He dilution refrigerator or a 3 He refrigerator and a superconducting magnet were used for measurements. For the measurements of χ ac (χ − iχ ), a piece with an ab-plane area of 1.15 x 1.15 mm 2 and a c-axis thickness of 0.5 mm was cut. Note that the use of a thick sample enabled us to reliably estimate the volume fraction for the external field B appl in the c-direction. In the case of a thin sample, a large demagnetization factor for B appl c makes the estimation of the volume fraction very difficult. For measurements along the c-axis (in the ab-plane), the sample was placed in a clamped piston-cylinder pressure cell with the c-axis (the ab-plane) parallel to the cylinder axis. An ac modulation field (f = 67.1 Hz and B ac ∼ 0.04 mT) and external magnetic field B appl were applied along the cylinder axis. In order to estimate the size of the signal corresponding to 100% shielding, a piece of Pb with nearly the same dimensions as the sample was measured with the same pick-up coil for the two orientations (namely, the c-axis and ab-plane orientations). The accuracy of these estimations was estimated to be about ±10%. For ρ measurements, a thin sample with dimensions of ∼1 x 0.2 x 0.03 mm 3 was exfoliated, where 0.03 mm is along the c-axis. After four gold wires were attached to a (001) surface with conducting silver paste, the sample was placed in a clamped piston-cylinder pressure cell with the longest axis, which is the electrical current direction, parallel to the cylinder axis. A standard four-contact method was used with a low-frequency ac current (I = 0.1 mA, f = 17 Hz), and the field B appl was applied parallel to the current. For both χ ac and ρ measurements, Daphne7474 (Idemitsu Kosan 2/9 J. Phys. Soc. Jpn. Co., Ltd., Tokyo) was used as a pressure-transmitting medium. 18 This oil does not solidify up to 37 kbar at room temperature (RT) 19 and hence ensures hydrostatic-pressure generation in the present measurements (highest P ∼30 kbar). Furthermore, the pressure cells were cooled slowly ( 1.5 K/min) from RT down to ∼20 K to prevent the possible development of nonhydrostaticity. Note that we always refer to the applied field B appl , which may be different from the internal field by ∼1 T, because of the large saturation moment of Eu 2+ . fields. At ambient P (P = 0 kbar) and B appl = 0 T, the real part χ increases with decreasing T and reaches a maximum at T N = 20 K, below which the T dependence is weak. This is essentially the same as the T dependence of dc magnetic susceptibility. 17 At P = 24 kbar and B appl = 0 T, χ exhibits a similar T dependence with a maximum at 20 K, although it is enhanced over that at ambient P . This is consistent with the previous report that T N is almost P -independent. 10 We  We define T c ("B c2 ") as the temperature (applied magnetic field) where χ deviates from the normal-state value. The reason why χ , not χ , is used is that χ exhibits T and field dependences in the normal state, which makes the unambiguous determination of the onset 5/9  of SC difficult in some cases [see the B appl = 1 T curve in Fig. 1(a) and the T = 25 K curve in Fig. 1(b), for example]. The quotation marks attached to B c2 indicate that B c2 values are based on the applied field, not on the internal field. The superconducting phase diagram determined from the data in Fig. 1 is shown in Fig. 4.  suppressed. There is evidence from the µSR and NMR measurements of (Ba 1−x K x )Fe 2 As 2 that the apparent coexistence of the orthorhombic AFM phase and superconducting phase in iron pnictide superconductors is not a true coexistence but can be explained by phase separation. 20,21 Also note that the critical pressure at which the structural/magnetic transition disappears differs between ref. 10 and the present work: in ref. 10, no transition was detected above P = 23 kbar, while a transition was detected at P = 25 kbar in our study. A very recent high-P study of SrFe 2 As 2 showed that the critical pressure is very sensitive to the homogeneity of the applied pressure and that it is higher when the pressure is more hydrostatic. 22 This may explain the difference in the critical pressure between ref. 10 and the present work. As shown in Fig. 3(a), the appearance of bulk SC is associated with the anomalous nearly Tlinear ρ. This and the fact that partial (or filamentary) and bulk SC occur below and above the critical pressure of magnetism, respectively, bear resemblance to cases of P -induced SC in some heavy-fermion compounds. 23,24 It is interesting to note that T -linear ρ is also observed in a wide T range in optimally doped BaFe 1.8 Co 0.2 As 2 . 25 The superconducting phase diagram (Fig. 4) indicates that "B c2 " is almost isotropic. The initial slope -d"B c2 "/dT at T = T c can be estimated to be 0.5(1), 0.6(2), and 0.19(4) T/K from the c-axis χ ac , ab-plane χ ac , and ab-plane ρ data for B appl 0.5 T, respectively. Although the slope depends on the type of determination method for "B c2 ", the estimated slopes are much smaller than those in the P -induced SC of other 122 compounds. 3,4,7 It is interesting to note that a very large slope of 3.87 T/K was reported for Eu 0.7 Na 0.3 Fe 2 As 2 . 26 The "B c2 "-T curves show a strong concave curvature above about 1 T. This reminds us of the T dependence of B c2 observed in the ternary molybdenum sulphide Sn 0.2 Eu 0.8 Mo 6.35 S 8 , 27 which can be explained by the Jaccarino-Peter compensation effect arising from the exchange interaction between local moments and conduction carriers. 28 In conclusion, when the structural/magnetic transition is suppressed by high P , EuFe 2 As 2 shows an anomalous nearly T -linear dependence of ρ and becomes a bulk superconductor at T c ∼30 K. The AFM order of the Eu 2+ moments at T N ∼20 K persists in the superconducting phase. The upper critical field exhibits a unique T dependence, which indicates the effect of the exchange interaction between the Eu 2+ moments and conduction carriers.