Preparation and characterization of superconducting Ba_1−xCs_xTi₂Sb₂O, and its pressure dependence of superconductivity

A new superconducting material, Ba_1−xNa_xTi₂Sb₂O, which shows a layered titanium-based pnictide oxide structure [tetragonal structure (space group: P4/mmm, No. 123)], was prepared and characterized by Doan et al.¹⁾ The superconductivity was reported to be observed at x = 0–0.33 in Ba_1−xNa_xTi₂Sb₂O.¹⁾ Subsequently, the superconductivity and physical features of Ba_1−xNa_xTi₂Sb₂O were extensively studied from theoretical and experimental points of view.^2–7) The maximum superconducting transition temperature, T_c, reached 5.5 K at x ∼ 0.15.¹⁾ This superconductivity is observed with the Na doping of BaTi₂Sb₂O in which the magnetic/charge ordering such as a charge density wave (CDW)/spin density wave (SDW) transition is suppressed; non-doped BaTi₂Sb₂O is also a superconductor with T_c of 1.2 K.^8,9) Admittedly, the hole is accumulated for Ti₂Sb₂O layers with Na doping, which lowers the CDW/SDW transition temperature, T_s, from 54 K at x = 0 to 30 K at x = 0.25.¹⁾ The superconductivity is observed together with CDW/SDW state at x = 0–0.33. Such a coexistence of superconducting state and magnetic/charge order state is found in unconventional superconductors such as heavy Fermions¹⁰⁾ and two-dimensional layered chalcogenides.^11,12)

Furthermore, Ba_1−xK_xTi₂Sb₂O was prepared by Pachmayr and Johrendt.¹³⁾ Its crystal structure was the same as that of Ba_1−xNa_xTi₂Sb₂O.¹⁾ The T_c value reached 6.1 K at x = 0.12 and no superconductivity was observed at x > 0.20.¹³⁾ Moreover, the superconductivity with T_c of 5.4 K was observed in Ba_1−xRb_xTi₂Sb₂O (x = 0.2).¹⁴⁾ It is reported that the phase diagrams of T_c–x are similar each other among Ba_1−xNa_xTi₂Sb₂O, Ba_1−xK_xTi₂Sb₂O, and Ba_1−xRb_xTi₂Sb₂O.^1,13,14) The pressure dependence of superconductivity in Ba_1−xNa_xTi₂Sb₂O (x = 0, 0.1 and 0.15) was reported in a pressure range from 0 to 1.6 GPa (=16 kbar).¹⁵⁾ The behavior of T_c in the pressure range is different depending on x, i.e. the T_c for x = 0.15 decreases straightforwardly from 5.2 K (0 GPa) to 4.2 K (1.6 GPa), while the T_c value increases monotonically from 2.1 K (0 GPa) to 2.9 K (1.6 GPa) at x = 0. Moreover, the T_c value for x = 0.1 increases from 3.8 K (0 GPa) to 4.2 K (0.5 GPa) and saturates at pressure of 0.5 to 1.6 GPa. Such a drastic variation of pressure dependence against x is a very interesting and exciting feature in Ba_1−xNa_xTi₂Sb₂O.

Here, we report a new discovery of superconductivity in Ba_1−xCs_xTi₂Sb₂O, and a characterization of its superconducting behavior at ambient and high pressures. The T_c was 4.4 K for Ba_0.75Cs_0.25Ti₂Sb₂O at ambient pressure. The stoichiometry of this sample was Ba_0.59(3)Cs_0.41(3)Ti_1.8(4)Sb_1.7(3)O_y, which was determined by energy dispersive X-ray (EDX) spectroscopy; the y value cannot be determined by EDX. The T_c gradually decreased with an increase in x from x = 0.25. The X-ray diffraction (XRD) pattern of Ba_0.75Cs_0.25Ti₂Sb₂O was well analyzed under the tetragonal structure [space group of P4/mmm (No. 123)]. Pressure-dependent XRD patterns were recorded in a pressure range of 4.08–23.4 GPa, all of which were analyzed in the space group of P4/mmm. The temperature dependence of resistance (R) were measured in a four terminal measurement mode at 0.88–12.3 GPa to draw the T_c–p phase diagram in a wider pressure range than that of Ba_1−xNa_xTi₂Sb₂O (0–1.6 GPa) reported previously.¹⁵⁾

The Ba_1−xCs_xTi₂Sb₂O sample was prepared according to the reaction process shown in Fig. 1(a). The nominal x value in Ba_1−xCs_xTi₂Sb₂O was selected in an x range of 0.1–0.6. The compounds of BaO (99.99%, Sigma-Aldrich), BaO₂ (99%, High Purity Chemicals), Cs₂CO₃ (99.995%, Sigma-Aldrich), Ti (99.98%, Nilaco), Sb (99.999%, Nilaco) were employed for the generation of Ba_1−xCs_xTi₂Sb₂O. The compounds mixed at the selected stoichiometry were introduced into a Nb tube. The Nb tube was welded under Ar atmosphere, and it was sealed in a quartz tube under vacuum. The tube was heated according to the reaction process shown schematically in Fig. 1(a). The obtained Ba_1−xCs_xTi₂Sb₂O sample was treated without an exposure to air, because it is very air-sensitive.

Zoom In Zoom Out Reset image size

The XRD, EDX and magnetic susceptibility (M/H) were recorded for the confirmation of generation of Ba_1−xCs_xTi₂Sb₂O; M and H refer to magnetization and applied magnetic field, respectively. The XRD measurements at ambient and high pressures were performed using synchrotron radiation at BL12B2 of SPring-8, and the wavelength, λ, of X-ray beam was 0.688 80 Å. A diamond anvil cell (DAC) was used for the high-pressure XRD measurement, with the sample being set into the hole of a stainless steel (SUS) plate. Daphne 7373 was employed as the pressure medium for the XRD measurement under high pressure. The pressure was monitored by ruby fluorescence. The M/H measurement was performed using SQUID magnetometer (Quantum Design MPMS2). The EDX spectrum was recorded with an EDX spectrometer equipped with a scanning electron microscope (KEYENCE VE-9800-EDAX Genesis XM2). The EDX spectra were measured for 5–10 different positions of the each sample at room temperature.

For the measurement of temperature dependence of R, the Ba_1−xCs_xTi₂Sb₂O sample (nominal x = 0.25) was introduced into the DAC in Ar-filled glove box because this sample is very air-sensitive. The polycrystalline sample was loaded directly on a Kapton sheet/epoxy resin/rhenium in the DAC; six Cu electrodes were attached to the Kapton sheet. NaCl was used as the pressure medium to obtain the static pressure. The applied pressure was determined by monitoring ruby fluorescence.

The value of R was recorded at 290–1.5 K in a standard four-terminal measurement mode using an Oxford superconducting magnet system. The temperature was controlled using an Oxford Instruments MercuryiTC. The H was controlled using Oxford Instruments MercuryiPS. Electrical current (I) was supplied by a Keithley 220 programmable current source, and the exact value of I was monitored by an Advantest R-8240 digital electrometer. The voltage (V) was measured by an Agilent 34420 digital nanovoltmeter.

The M/H–T plot under pressure was measured using the above SQUID machine; the Ba_1−xCs_xTi₂Sb₂O sample (nominal x = 0.25) was introduced in a piston cylinder cell in Ar-filled glove box, and the daphne 7373 was used as a pressure medium.

Figure 1(b) shows temperature (T) dependence of M/H of Ba_1−xCs_xTi₂Sb₂O (nominal x = 0.25) in zero field cooling (ZFC) and field cooling (FC) modes at 0 GPa. The rapid drop of M/H was recorded below 5 K in M/H–T plot in ZFC mode, showing the onset superconducting transition temperature, T_c^onset, as high as 5.0 K. The T_c was 4.4 K which was determined from the crossing point of the M/H–T plot at normal state and the rapid drop [Inset of Fig. 1(b)]. The H was 10 Oe (= 1.0 mT). The shielding fraction, SF, evaluated at 2.0 K was 32.2%. The M/H–T plot in FC mode showed the drop of M/H–T plot, but the Meissner fraction was low (several %). The T_c^onset of 5.0 K and T_c of 4.4 K were a little lower than the maximum T_c values [T_c = 5.5 K for Ba_1−xNa_xTi₂Sb₂O (x = 0.15),¹⁾ T_c = 6.1 K for Ba_1−xK_xTi₂Sb₂O (x = 0.12)¹³⁾ and T_c = 5.4 K for Ba_1−xRb_xTi₂Sb₂O (x = 0.2)¹⁴⁾] reported previously. As seen from Fig. 1(c), the small peak is observed at around 44 K in M/H–T plot recorded at 1.0 T, which may be assigned to the CDW/SDW transition reported for Ba_1−xNa_xTi₂Sb₂O, Ba_1−xK_xTi₂Sb₂O and Ba_1−xRb_xTi₂Sb₂O.^1,13,14)

The EDX spectrum [Fig. 1(d)] showed that the stoichiometry of the above sample [Ba_1−xCs_xTi₂Sb₂O (nominal x = 0.25)] was Ba_0.59(3)Cs_0.41(3)Ti_1.8(4)Sb_1.7(3)O_y. The EDX spectrum provided only peaks ascribable to Ba, Cs, Ti, Sb and O atoms, but the exact y value (O amount) was not determined owing to the contamination from atmosphere. Through this study, the sample of Ba_0.59(3)Cs_0.41(3)Ti_1.8(4)Sb_1.7(3)O_y was employed for measurements of XRD and R–T under pressure.

The XRD pattern of the Ba_1−xCs_xTi₂Sb₂O (nominal x = 0.25), which corresponds to the stoichiometry (Ba_0.59(3)Cs_0.41(3)Ti_1.8(4)Sb_1.7(3)O_y), recorded at ambient pressure is shown in Fig. 2(a), with being consistent with the XRD pattern reported previously.¹⁾ The XRD pattern was analyzed by Le Bail fitting to determine the lattice constants. The analysis was performed under the space group of P4/mmm (tetragonal structure, No. 123), which is the same as Ba_1−xNa_xTi₂Sb₂O.¹⁾ As seen from Fig. 2(a), the XRD pattern was well fitted by the calculated pattern. The lattice constants, a and c, were determined to be 4.121 04(5) and 8.1394(2) Å, respectively.

Zoom In Zoom Out Reset image size

Moreover, the XRD pattern was analyzed by Rietveld refinement under the above space group, showing a good fit between the experimental XRD and calculated patterns [Fig. 2(b)]. Table I lists the crystallographic data of Ba_1−xCs_xTi₂Sb₂O at the x value determined from EDX (x = 0.41) determined by the Rietveld refinement. The values of a and c were determined to be 4.120 73(5) and 8.1388(2) Å, respectively, in consistent with those determined by Le Bail fitting.

Figure 2(c) shows the schematic representation of crystal structure of Ba_1−xCs_xTi₂Sb₂O prepared using the atomic coordinates listed in Table I. The crystal is called "CeCr₂Si₂C type structure." First, we must comment that the crystal structure of Ba_1−xCs_xTi₂Sb₂O is the same as those of Ba_1−xM_xTi₂Sb₂O (M: Na, Rb and K).^1,13,14) Namely, the crystal structure of Ba_1−xNa_xTi₂Sb₂O is still remained even by an intercalation of Cs atom with the largest ionic radius among alkali metal atoms. If the crystal structure is viewed in more detail, it can be noted that the crystal is formed by the OTi₄ antiperovskite layers, which is similar to CuO₂ layers in the high-T_c cuprate superconductor.^16–23) The layers are capped by Sb at both top and bottom of empty center site surrounded by four Ti atoms, and the Ba/Cs atoms are located at the top and bottom of a part of O site surrounded by four Ti sites. This crystal structure is recognized as a stacking structure of Ti₂Sb₂O layers along c-axis, indicating that two dimensional (2D) Ti₂Sb₂O layer is the conduction layer. This crystal structure is not only interesting from similarity to the structure of cuprate described above, but also fascinating as the target material in the investigation of local structure by fluorescence holography, because of the unique 2D layer-by-layer structure.

We have investigated the local structure around doped element in various types of 2D layered materials, graphite²⁴⁾ and topological insulator,²⁵⁾ by use of X-ray photoelectron and fluorescence holograms thus far, which provided the precise location of doped element; the X-ray photoelectron and fluorescence holography can determine the surface and bulk structures, respectively. Namely, the local structure determined using holography technique has a very high reliability and precision, since the three dimensional real image can be directly constructed from hologram pattern containing phase information. The study on local structure in the bulk crystals of Ba_1−xMxTi₂Sb₂O and Ba_1−xCs_xTi₂Sb₂O using X-ray fluorescence holography technique is the future task.

Figure 3(a) shows the T_c–x plot of Ba_1−xCs_xTi₂Sb₂O, in which x is the nominal value. The maximum T_c of 4.4 K is found at x = 0.25, and it slowly decreases with an increase in x. The x dependence of SF determined from M/H value at 2 K is shown in Fig. 3(b). The bulk superconductivity was not observed below x = 0.25. Actually, the SF at 2 K was ∼1% for x = 0.2. Thus, the bulk superconductivity in Ba_1−xCs_xTi₂Sb₂O was found only at x = 0.25–0.6. The SF (=55.2%) for Ba_1−xCs_xTi₂Sb₂O (nominal x = 0.5) is the highest value realized in Ba_1−xCs_xTi₂Sb₂O. This is a little different from that of Ba_1−xNa_xTi₂Sb₂O and Ba_1−xK_xTi₂Sb₂O which showed the bulk superconductivity even below x = 0.2.^1,13)

Zoom In Zoom Out Reset image size

Figure 4(a) shows the XRD patterns of Ba_1−xCs_xTi₂Sb₂O (nominal x = 0.25) recorded at 4.08–23.4 GPa. All XRD patterns were well analyzed in the space group of P4/mmm which is the same as that at 0 GPa. The values of a and c were determined by Le Bail analysis for whole XRD patterns at 4.08–23.4 GPa. The pressure dependence of a and c are shown in Figs. 4(b) and 4(c), respectively, which provides a smooth decrease in both values against pressure. Significant structural phase transitions were not found at 0–23.4 GPa. In addition, any decomposition of the sample was not found in a whole pressure range.

Zoom In Zoom Out Reset image size

Finally, we recorded temperature dependence of R of Ba_1−xCs_xTi₂Sb₂O (nominal x = 0.25) at different pressures. The plots of R/R(275 K)–T and R/R(10 K)–T at different pressures are shown in Figs. 5(a) and 5(b), respectively. The metallic behavior at 0.88–12.3 GPa is suggested from the R/R(275 K)–T plots [Fig. 5(a)]. A clear CDW/SDW transition was not observed in R/R (275 K)–T plots at 0.88–12.3 GPa, although the small peak at around 44 K was observed in M/H–T plot recorded at 1.0 T under ambient pressure [Fig. 1(c)]. Moreover, the M/H–T plots were recorded at 10 Oe and 1.0 T under pressure of 0.24 GPa. The M/H–T plot at 10 Oe showed a clear superconducting transition at T_c = 3.8 K, while that at 1.0 T never showed any peak which is different from that shown in Fig. 1(c), indicating no CDW/SDW transition at 0.24 GPa. Also, any peak was not observed at 1.0 T under pressure of 0.59 GPa [see Fig. S1 in the online supplementary data, which is available online at stacks.iop.org/JJAP/58/110603/mmedia], and the superconducting transition (T_c = 3.75 K) was also recorded at 10 Oe under pressure of 0.59 GPa. Namely, the CDW/SDW transition disappeared by the application of pressure for Ba_1−xCs_xTi₂Sb₂O.

Zoom In Zoom Out Reset image size

The superconducting transition was observed for all R–T plots measured at 0.88–12.3 GPa, as seen from Fig. 5(b). The value of T_c is plotted against pressure [Fig. 5(c)]. The T_c monotonically decreases with an increase in pressure up to ∼4 GPa; the minimum T_c is 2.8 K at 3.37  GPa. The T_c value increases with further increasing pressure. The T_c became 4.0 K at 9.50 GPa. Such a variation of T_c cannot be explained by the simple BCS theorem which suggests the decrease in T_c owing to the reduction of density of states (DOS). However, we have never seen the structural phase transitions at around 4.0 GPa, suggestive of the presence of electronic transition causing the increase in T_c. One possibility is the Lifshitz transition accompanying the variation of the Fermi surface topology. If this is the case, a drastic change of DOS causing the observed enhancement of T_c may be expected. Such a drastic variation of topology of the Fermi surface caused by doping of carrier^26–31) and/or by applying pressure^32,33) is suggested in various superconductors from theoretical and experimental points of view. For the confirmation of Lifshitz transition under pressure, the Hall effect measurement may be indispensable in a wide pressure range.

The systematic study on the new superconductor, Ba_1−xCs_xTi₂Sb₂O, was achieved in a wide pressure range. The maximum T_c was 4.4 K for x = 0.25, and the T_c–x plot was drawn. The crystal structure was tetragonal [space group of P4/mmm (No. 123)] up to 23.4 GPa. Very interesting non-BCS type T_c–p phase diagram was observed, indicating that this superconductor is very attractive as the research target in superconductor physics. Ba_1−xCs_xTi₂Sb₂O may be very fascinating as the target material for the study on local structure by fluorescence holography, owing to the interesting layer-by-layer stacking structure of Ti₂Sb₂O and a partial substitution of Cs for Ba.

This study was partly supported by Grants-in-Aid (26105004, 17K05500, 18K04940, 18K18736 and 19H02676) from MEXT. The XRD measurements at SPring-8 were carried out under the proposals of 2018A4140 and 2019A4131.