Superconductor Transition in a Copper-Oxide Superconductor
Total Page:16
File Type:pdf, Size:1020Kb
ARTICLE Received 5 Mar 2013 | Accepted 18 Aug 2013 | Published 20 Sep 2013 DOI: 10.1038/ncomms3459 Disappearance of nodal gap across the insulator–superconductor transition in a copper-oxide superconductor Yingying Peng1, Jianqiao Meng1, Daixiang Mou1, Junfeng He1, Lin Zhao1,YueWu1, Guodong Liu1, Xiaoli Dong1, Shaolong He1, Jun Zhang1, Xiaoyang Wang2, Qinjun Peng2, Zhimin Wang2, Shenjin Zhang2, Feng Yang2, Chuangtian Chen2, Zuyan Xu2, T.K. Lee3 & X.J. Zhou1 The parent compound of the copper-oxide high-temperature superconductors is a Mott insulator. Superconductivity is realized by doping an appropriate amount of charge carriers. How a Mott insulator transforms into a superconductor is crucial in understanding the unusual physical properties of high-temperature superconductors and the superconductivity mechanism. Here we report high-resolution angle-resolved photoemission measurement on heavily underdoped Bi2Sr2–xLaxCuO6 þ d system. The electronic structure of the lightly doped samples exhibit a number of characteristics: existence of an energy gap along the nodal direction, d-wave-like anisotropic energy gap along the underlying Fermi surface, and coex- istence of a coherence peak and a broad hump in the photoemission spectra. Our results reveal a clear insulator–superconductor transition at a critical doping level of B0.10 where the nodal energy gap approaches zero, the three-dimensional antiferromagnetic order dis- appears, and superconductivity starts to emerge. These observations clearly signal a close connection between the nodal gap, antiferromagnetism and superconductivity. 1 National Laboratory for Superconductivity, Beijing National Laboratory for Condensed Matter Physics, Institute of Physics, Chinese Academy of Sciences, Beijing 100190, China. 2 Technical Institute of Physics and Chemistry, Chinese Academy of Sciences, Beijing 100190, China. 3 Institute of Physics, Academia Sinica, Nankang, Taipei 11529, Taiwan. Correspondence and requests for materials should be addressed to X.J.Z. (email: [email protected]). NATURE COMMUNICATIONS | 4:2459 | DOI: 10.1038/ncomms3459 | www.nature.com/naturecommunications 1 & 2013 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3459 he parent compound of the copper-oxide superconductors temperature, these curves show a linear dependence. This is is an antiferromagnetic Mott insulator; doping of charge consistent with the 2D variable-range hopping model, carriers into the parent compound leads to an insulator– where the resistivity–temperature dependence is expected to be: T 1/3 (T0/T) metal transition and the emergence of high-temperature super- r ¼ r0e (refs 8,9). conductivity (SC)1. In the superconducting state, predominantly d-wave superconducting gap is well established with a zero gap Underlying Fermi surface. Figure 2 shows the doping evolution along the nodal direction2. In the normal state of the underdoped of the underlying Fermi surface across the non-superconductor– superconducting region, anisotropic pseudogap appears with a superconductor transition and indications of the nodal energy zero gap along the nodal direction3. A particular, yet important, gap observed for the lightly doped La–Bi2201 samples. As seen region exists between the parent compound and the from Fig. 2a–d, in the lightly doped samples, the spectral weight superconducting region, that is, a lightly doped region where, distribution dominates near the (p/2,p/2) nodal region and then on slight doping, the three-dimensional (3D) antiferromagnetic spreads towards the (0,p) antinodal region with increasing dop- order is rapidly suppressed and an insulator–metal transition ing. As commonly used before, we denote the high-intensity occurs. What is the main reason for the antiferromagnetic phase contour in the spectral weight distribution as the ‘underlying remaining insulating after being doped? What are the electronic Fermi surface’, although in principle a strict Fermi surface refers characteristics of this peculiar lightly doped region? How is the to the locus of momentum without an energy gap10. The electronic structure of this particular region connected to the photoemission spectra (energy distribution curves, EDCs) along pseudogap and superconducting gap in the underdoped super- the underlying Fermi surface are displayed in Fig. 2e,g,i,k for conducting region? Investigation of this lightly doped region is different doping levels, with their corresponding symmetrized critical in understanding how a Mott insulator can transform into EDCs shown in Fig. 2f,h,j,l respectively. The EDC symmetrization a d-wave superconductor, the origin of the pseudogap and its is an empirical and intuitive way to examine the existence of an relationship with the superconducting gap4,5. energy gap by removing the Fermi distribution function from the In this paper, we report angle-resolved photoemission (ARPES) original ARPES data11. As seen from Fig. 2e, with very low measurements on the electronic structure evolution with doping doping of P ¼ 0.03, only a broad hump feature dominates the in Bi (Sr La )CuO (La–Bi2201) system focusing on the 2 2–x x 6 þ d measured spectra. It disperses towards low binding energy, lightly doped region and the insulator–superconductor transition. reaches its minimum B200 meV along the nodal direction and Our super-high-resolution ARPES experiments reveal an then turns back. The spectral weight near the Fermi level E is insulator–superconductor transition at a critical doping level of F strongly suppressed for this lightly doped sample, and there is PB0.10. In the lightly doped region (P ¼ 0 À 0.10), the photo- apparently a gap opening (Fig. 2f). Further increase of the doping emission spectra are characterized by the coexistence of a to P ¼ 0.055 results in an emergence of a coherence peak near the coherence peak and a broad hump consistent with a polaronic Fermi level, which coexists with the broad hump structure behaviour. Moreover, a fully opened d-wave-like anisotropic (Fig. 2g). A close inspection reveals an energy gap for the energy gap is observed, which is offset by an energy gap along the coherence peak along the entire underlying Fermi surface, (0,0)-(p,p) nodal direction. The nodal gap decreases with including the nodal region (Fig. 2h). When the doping level increasing doping and approaches zero at a critical doping level reaches P ¼ 0.08, the coherence peak becomes prominent. In the of PB0.10, where the 3D antiferromagnetic order vanishes and meantime, the broad hump gets weaker but remains visible SC starts to emerge6. In addition, a smooth transition from the (Fig. 2i). The energy gap opening is clear along the entire energy gap in the lightly doped region to the pseudogap in the underlying Fermi surface (Fig. 2j). Eventually, when the doping underdoped superconducting region is observed. These level becomes P ¼ 0.10, the coherence peak gets sharp and the observations indicate a close relationship between the nodal broad hump remains discernable (Fig. 2k). But in this case, no energy gap, antiferromagnetic insulating phase and SC. They will indication of an energy gap opening is observed near the nodal shed important light on the origin of the pseudogap in high- region (Fig. 2l). However, there are clear indications of gap temperature superconductors, and point to the importance of opening in the EDCs far away from the nodal region. This is combining the electron correlation, antiferromagnetism (AF) and consistent with the observation of the Fermi arc in underdoped strong electron–phonon coupling in understanding the electronic superconducting samples12–14. structure of the lightly doped and underdoped regions. Nodal band structure. Figure 3 focuses on the nodal gap and its Results detailed doping evolution. Figure 3a–g shows doping evolution of Transport properties of La–Bi2201. Figure 1 shows the tem- the band structure along the nodal direction. The Fermi dis- perature dependence of the in-plane resistivity rab (Fig. 1a) and tribution function is divided out from these data to facilitate the the magnetization (Fig. 1b) of the La–Bi2201 single crystals. The revelation of the energy gap near the Fermi level. Original EDCs La–Bi2201 samples with P ¼ 0–0.08 doping levels are non- for the La–Bi2201 samples with different doping levels are shown superconducting. The P ¼ 0.10 sample is superconducting with a in Fig. 3h on the underlying Fermi surface along the nodal B Tc at 3 K, which is on the border of insulator-to-super- direction, with their corresponding symmetrized EDCs shown in conductor transition. The P ¼ 0.105 sample has a Tc of 12 K. Fig. 3i. From the raw data (Fig. 3a–g), it is clear that with It is interesting to note that, the insulator–superconductor increasing doping, the spectral weight transfers from high binding transition takes place near the universal two-dimensional energy to low binding energy. For the P ¼ 0.03 sample, the 2 (2D) resistance, h/4e , which corresponds to 3D resistivity spectral weight near EF is strongly depleted (Fig. 3a) and the B0.8 mO Á cm in Bi2201. The insulator–superconductor transi- nodal EDC is characterized by a prominent broad hump near tion can also be driven by Zn doping7 or by changing the 250 meV. A slight increase of the doping level to 0.04 brings a thickness of cuprate thin films4, and it is known that it takes place significant amount of spectral weight towards low binding energy at nearly the same resistance values. and a weak coherence peak emerges near the Fermi level The in-plane resistivity–temperature curves of the La–Bi2201 (Fig. 3b,h). Further doping brings more spectral weight to the single crystals with low doping levels of 0.03, 0.04 and 0.08 are re- Fermi level so that the coherence peak becomes sharper, while the plotted in Fig. 1c as lnrBT À 1/3. Over an extended range of low broad hump structure becomes weaker. The relative spectral 2 NATURE COMMUNICATIONS | 4:2459 | DOI: 10.1038/ncomms3459 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited.