Multiband nodeless superconductivity near the charge-density-wave quantum critical point in ZrTe3−xSex
Cui Shan1, He Lan-Po1, Hong Xiao-Chen1, Zhu Xiang-De2, 4, Petrovic Cedomir4, Li Shi-Yan1, 3, †,
State Key Laboratory of Surface Physics, Department of Physics, and Laboratory of Advanced Materials, Fudan University, Shanghai 200433, China
High Magnetic Field Laboratory, Chinese Academy of Sciences and University of Science and Technology of China, Hefei 230031, China
Collaborative Innovation Center of Advanced Microstructures, Nanjing 210093, China
Condensed Matter Physics and Materials Science Department, Brookhaven National Laboratory, Upton, New York 11973, USA

 

† Corresponding author. E-mail: shiyan_li@fudan.edu.cn

Project supported by the National Basic Research Program of China (Grant Nos. 2012CB821402 and 2015CB921401), the National Natural Science Foundation of China (Grant Nos. 91421101, 11422429, and 11204312), the Program for Professor of Special Appointment (Eastern Scholar) at Shanghai Institutions of Higher Learning, China, and STCSM of China (Grant No. 15XD1500200). Work at Brookhaven National Laboratory was supported by the US DOE under Contract No. DESC00112704.

Abstract
Abstract

It was found that selenium doping can suppress the charge-density-wave (CDW) order and induce bulk superconductivity in ZrTe3. The observed superconducting dome suggests the existence of a CDW quantum critical point (QCP) in ZrTe3−xSex near x ≈ 0.04. To elucidate the superconducting state near the CDW QCP, we measure the thermal conductivity of two ZrTe3−xSex single crystals (x = 0.044 and 0.051) down to 80 mK. For both samples, the residual linear term κ0/T at zero field is negligible, which is a clear evidence for nodeless superconducting gap. Furthermore, the field dependence of κ0/T manifests a multigap behavior. These results demonstrate multiple nodeless superconducting gaps in ZrTe3−xSex, which indicates conventional superconductivity despite of the existence of a CDW QCP.

1. Introduction

Charge-density-wave (CDW) order usually exists in some low-dimensional materials, especially those transition-metal chalcogenides.[14] When the CDW order is suppressed by doping or pressure, a list of them can be tuned to superconductors.[58] In the temperature–doping (Tx) or temperature–pressure (Tp) phase diagram, sometimes a superconducting dome is observed on top of a CDW quantum critical point (QCP).[58] The reminiscent of this kind of phase diagram to that of the heavy-fermion and high-Tc cuprate superconductors raises the possibility of unconventional superconductivity caused by CDW fluctuations.[59]

ZrTe3 is such a compound in which the CDW order and superconductivity compete and coexist.[10] It belongs to a family of trichalcogenides MX3 (M = Ti, Zr, Hf, U, Th, and X = S, Se, Te). The structure consists of infinite XX chains formed by stacking MX3 prisms.[11] The polyhedra are arranged in double sheets and stacked along the monoclinic c axis by van der Waals forces.[11] Pristine ZrTe3 itself harbors filamentary superconductivity with Tc ∼ 2 K.[10] The CDW vector q ≈ (1/14; 0; 1/3) is developed in ZrTe3 below TCDW ∼ 63 K.[12] Like other CDW materials, pressure and doping can melt the CDW order and stabilize its superconductivity to bulk.[1316] Recently, isovalent substitution of Se for Te was also found to cause a superconducting dome in the ZrTe3−xSex system, with maximum Tc = 4.4 K at the optimal doping x = 0.04.[17] It was suggested that this superconductivity may be mediated by quantum critical charge fluctuations.[17] To clarifying the underlying pairing mechanism, it is important to know the superconducting gap symmetry and structure.

Ultra-low-temperature heat transport is an established bulk technique to probe the superconducting gap structure.[18] The existence of a finite residual linear term κ0/T in zero magnetic field is an evidence for gap nodes.[18] The field dependence of κ0/T may further give support for a nodal superconducting state, and provide information on the gap anisotropy, or multiple gaps.[18]

In this paper, we measure the ultra-low-temperature thermal conductivity of ZrTe3−xSex single crystals near optimal doping to investigate whether the superconducting state is unconventional. The negligible κ0/T in zero field and the rapid field dependence of κ0(H)/T in low field strongly suggest multiple nodeless superconducting gaps in ZrTe3−xSex. In this sense, the superconductivity in ZrTe3−xSex is likely conventional.

2. Experiment

The ZrTe3−xSex single crystals were grown by iodine vapor transport method.[13,17] Two single crystals from different batches, both with nominal composition x = 0.04, were used for this study. Their exact compositions were determined by wavelength-dispersive spectroscopy (WDS), utilizing an electron probe microanalyzer (Shimadzu EPMA-1720). The dc magnetization was measured at H = 20 Oe, with zero-field cooling, using a SQUID (MPMS, Quantum Design). The samples were cleaved and cut to rectangular bars, with typical dimensions of 2.12 mm×1.01 mm×0.030 mm. The largest surface is ab-plane. The contacts were made directly on the sample surfaces with silver paint, which were used for both resistivity and thermal conductivity measurements. The contacts are metallic with typical resistance 200 mΩ at 2 K. The in-plane thermal conductivity was measured in a dilution refrigerator, using a standard four-wire steady-state method with two RuO2 chip thermometers, calibrated in situ against a reference RuO2 thermometer. Magnetic fields were applied along the c axis and perpendicular to the heat current. To ensure a homogeneous field distribution in the sample, all fields were applied at temperature above Tc.

3. Results and discussion

According to the WDS results, the actual Se contents of the two ZrTe3−xSex single crystals are x = 0.044 and 0.051, respectively. Below we will use the actual x. Figure 1(a) presents the normalized dc magnetization of ZrTe2.956Se0.044 and ZrTe2.949Se0.051 single crystals. The Tc defined by the onset of diamagnetic transition is 4.0 K for both samples. The significant diamagnetic response confirms that the superconductivity is stabilized to bulk from the filamentary superconductivity in pristine ZrTe3, which is consistent with the previous report.[17] This bulk superconductivity will be further supported by our thermal conductivity data in this study.

Fig. 1. (a) The normalized dc magnetization of ZrTe2.956Se0.044 and ZrTe2.949Se0.051 single crystals, measured in H = 20 Oe with zero-field-cooled (ZFC) process. (b) The in-plane resistivity of ZrTe2.956Se0.044 and ZrTe2.949Se0.051 single crystals. No anomaly is observed in the normal state, suggesting the complete suppression of CDW state. (c) The resistive superconducting transition at low temperature. For clarity, the resistance of the x = 0.044 sample is magnified by five times. The Tc defined by ρ = 0 is 4.06 K and 3.87 K for x = 0.044 and 0.051 samples, respectively.

Figure 1(b) shows the in-plane resistivity ρ(T) of ZrTe2.956Se0.044 and ZrTe2.949Se0.051 single crystals. No anomaly is observed in the normal state, suggesting the complete suppression of CDW state in them.[17] Fitting the normal-state resistivity data below 60 K to ρ(T) = ρ0 + ATn gives residual resistivity ρ0 = 2.82 μΩ·cm and 21.5 μΩ·cm for the x = 0.044 and 0.051 samples, respectively. The resistive superconducting transition at low temperature is plotted in Fig. 1(c). The Tc defined by ρ = 0 is 4.06 K and 3.87 K for the x = 0.044 and 0.051 samples, respectively. Both of them are near the optimal doping in the phase diagram of ZrTe3−xSex, and the x = 0.051 sample is slightly overdoped.[17]

To determine their upper critical fields Hc2, the low-temperature resistivity of these two samples under magnetic fields was also measured. Figures 2(a) and 2(b) show the low temperature ρ(T) curves of ZrTe2.956Se0.044 and ZrTe2.949Se0.051 single crystals under various fields. With increasing field, the superconducting transition is gradually suppressed to lower temperature, and the magnetoresistance in the normal state is very weak. The Hc2(T), defined by ρ = 0 in Figs. 2(a) and 2(b), is plotted in Fig. 2(c) for both x = 0.044 and 0.051 samples. From Fig. 2(c), we roughly estimate Hc2(0) ≈ 1.40 T and 0.85 T for them, respectively.

Fig. 2. Low-temperature resistivity of (a) ZrTe2.956Se0.044 and (b) ZrTe2.949Se0.051 single crystals under various magnetic fields. (c) Temperature dependence of the upper critical field Hc2(T) defined by ρ = 0 in panels (a) and (b). The dashed lines are guide to eye, which point to Hc2(0) ≈ 1.40 T and 0.85 T for x = 0.044 and 0.051 samples, respectively.

The temperature dependence of in-plane thermal conductivity for ZrTe2.949Se0.044 and ZrTe2.956Se0.051 single crystals in zero and applied magnetic fields is shown in Fig. 3, plotted as κ/T vs. T. The thermal conductivity at very low temperature can usually be fitted to κ/T = a + bTα−1.[19,20] The two terms aT and bTα represent contributions from electrons and phonons, respectively. The power α is typically between 2 and 3 due to the specular reflections of phonons at the boundary.[19,20] One can see that all the curves in Fig. 3 are roughly linear, therefore we fix α to 2. In zero field, the fittings give κ0/T = 0.008 ± 0.008 mW·K−2·cm−1 and 0.009 ± 0.002 mW·K−2·cm−1 for the x = 0.044 and 0.051 samples, respectively. Such a tiny κ0/T in zero field is negligible for both samples. As T → 0, since all electrons become Cooper pairs for the s-wave nodeless superconductors, there are no fermionic quasiparticles to conduct heat. Therefore there is no residual linear term of κ0/T, as seen in V3Si.[19] However, for the unconventional superconductors with nodes in the superconducting gap, the nodal quasiparticles will contribute a finite κ0/T in zero field.[18] For example, κ0/T = 1.41 mW·K−2·cm−1 for the overdoped cuprate Tl2Ba2CuO6+δ (Tl-2201), a d-wave superconductor with Tc = 15 K.[21] For the p-wave superconductor Sr2RuO4, κ0/T = 17 mW·K−2·cm−1.[22] Therefore, the negligible κ0/T of the x = 0.044 and 0.051 samples suggests that the superconducting gap of ZrTe3−xSex is nodeless. Note that the negligible κ0/T in zero field also supports the bulk superconductivity in our samples.

Fig. 3. Low-temperature in-plane thermal conductivity of (a) ZrTe2.956Se0.044 and (b) ZrTe2.949Se0.051 single crystals in zero and magnetic fields. The lines are fits of the data to κ/T = a + bTα−1, with α fixed to 2. The dashed lines represent the normal-state Wiedemann–Franz law expectations L0/ρ0 for the x = 0.044 and 0.051 samples, respectively.

When applying field, κ/T gradually increases with increasing field, as seen in Fig. 3. In H = 0.5 T, the fittings give κ0/T = 8.27 ± 0.08 mW·K−2·cm−1 and 1.14 ±0.03 mW·K−2·cm−1 for the x = 0.044 and 0.051 samples, respectively. These values roughly meet their Wiedemann–Franz law expectations L0/ρ0 (L0 is the Lorenz number 2.45 × 10−8 W·Ω·K−2 and ρ0 is the sample’s residual resistivity). The verification of the Wiedemann–Franz law in the normal state shows the reliability of our thermal conductivity measurements. The bulk Hc2(0) ≈ 0.5 T is taken for both samples, which is lower than those determined from the resistivity measurements.

To gain more information of the gap structure in ZrTe2.956Se0.044 and ZrTe2.949Se0.051, we check the field dependence of their κ0/T. The normalized κ0/T as a function of H/Hc2 is plotted in Fig. 4. For comparison, the data of the clean s-wave superconductor Nb,[23] the multiband s-wave superconductor NbSe2,[24] and an overdoped sample of the d-wave superconductor Tl-2201 are also plotted.[21] The slow field dependence of κ0/T in low field for Nb manifests its single isotropic superconducting gap. In Fig. 4, the curves of the x = 0.044 and 0.051 samples are similar to that of NbSe2, a multiband s-wave superconductor with the gap ratio Δl/Δs ≈ 3.[24] This suggests that ZrTe3−xSex also has multiple nodeless superconducting gaps. Previously, an ab initio calculation of the band structure for ZrTe3 at ambient pressure gave a central rounded 2D Fermi surface sheet and two flatter q1D sheets.[8] Therefore, the observation of multiple nodeless superconducting gaps in ZrTe3−xSex system is not surprising.

Fig. 4. Normalized residual linear term κ0/T of ZrTe2.956Se0.044 and ZrTe2.949Se0.051 single crystals as a function of H/Hc2. Similar data of the clean s-wave superconductor Nb,[23] an overdoped d-wave cuprate superconductor Tl-2201,[21] and the multiband s-wave superconductor NbSe2[24] are also plotted for comparison.

Theoretically, it has been shown that unconventional superconductivity with dxy symmetry can appear in close proximity to a charge-ordered phase, and the superconductivity is mediated by charge fluctuations.[25,26] Since the dxy-wave gap has line nodes, our results clear rule out this kind of unconventional superconductivity in ZrTe3−xSex. In this context, the superconductivity in ZrTe3−xSex is likely conventional. Similar situation happens in the CuxTiSe2 system. Thermal conductivity measurements suggested conventional s-wave superconductivity with a single isotropic gap in Cu0.06TiSe2, near where the CDW order vanishes.[27] So far, the evidence for unconventional superconductivity induced by CDW fluctuations in real materials is still lack. The experiments on more systems with superconductivity near a CDW QCP are needed.

4. Conclusion

In summary, we have measured the ultra-low-temperature thermal conductivity of ZrTe2.956Se0.044 and ZrTe2.949Se0.051 single crystals, which are near the optimal doping in the phase diagram of the ZrTe3−xSex system. The absence of κ0/T in zero field for both compounds gives strong evidence for nodeless superconducting gap. The field dependence of κ0(H)/T further suggests multiple nodeless gaps in ZrTe3−xSex. Unconventional superconductivity with line nodes is excluded in this trichalcogenide system although there is a CDW QCP. It is likely that the superconductivity in ZrTe3−xSex is still conventional.

Reference
1Wilson J ADi Salvo F JMahajan S1975Adv. Phys.24117
2Kim S JPark S JJeon I CKim C HPyun C HYee K A 1997 J. Phys. Chem. Solids 58 659
3Di Salvo F JMoncton D EWaszczak J V 1976 Phys. Rev. 14 4321
4Boswell FBennett J C 1996 Mater. Res. Bull. 31 1083
5Morosan EZandbergen H WDennis B SBos J W GOnose YKlimczuk TRamirez A POng N PCava R J 2006 Nat. Phys. 2 544
6Kusmartseva A FSipos BBerger HForro LTutis E 2009 Phys. Rev. Lett. 103 236401
7Sipos BKusmartseva A FAkrap ABerger HForro LTutis E 2008 Nat. Mater. 7 960
8Hoesch MGarbarino GBattaglia CAebi PBerger H 2016 Phys. Rev. 93 125102
9Norman M R 2011 Science 332 196
10Yamaya KTakayanagi STanda S 2012 Phys. Rev. 85 184513
11Furuseth SBrattas LKjekshus A1975Acta Chem. Scand. A29623
12Eaglesham D JSteeds J WWilson J A 1984 J. Phys. C: Solid State Phys. 17 L697
13Zhu X DLei H CPetrovic C 2011 Phys. Rev. Lett. 106 246404
14Lei H CZhu X DPetrovic C 2011 Europhys. Lett. 95 17011
15Yamaya KYoneda MYasuzuka SOkajima YTanda S 2002 J. Phys.: Condens. Matter 14 10767
16Zhu X YLv BWei F YXue Y YLorenz BDeng L ZSun Y YChu C W 2013 Phys. Rev. 87 024508
17Zhu X DNing WLi L JLing L SZhang R RWang K FLiu YPi LMa Y CDu H FTian M LSun Y PPetrovic CZhang Y H2016Sci. Rep.accepted
18Shakeripour HPetrovic CTaillefer L 2009 New J. Phys. 11 055065
19Sutherland MHawthorn D GHill R WRonning FWakimoto SZhang HProust CBoaknin ELupien CTaillefer L 2003 Phys. Rev. 67 174520
20Li S YBonnemaison J BPayeur AFournier PWang C HChen X HTaillefer L 2008 Phys. Rev. 77 134501
21Proust CBoaknin EHill R WTaillefer LMackenzie A P 2002 Phys. Rev. Lett. 89 147003
22Suzuki MTanatar M AKikugawa NMao Z QMaeno YIshiguro T 2002 Phys. Rev. Lett. 88 227004
23Lowell JSousa J B 1970 J. Low. Temp. Phys. 3 65
24Boaknin ETanatar M APaglione JHawthorn DRonning FHill R WSutherland MTaillefer LSonier JHayden S MBrill J W 2003 Phys. Rev. Lett. 90 117003
25Scalapino D JLoh EHirsch J E 1987 Phys. Rev. 35 6694
26Merino JMcKenzie R H 2001 Phys. Rev. Lett. 87 237002
27Li S YWu GChen X HTaillefer L 2007 Phys. Rev. Lett. 99 107001