Near a quantum-critical point in a metal a strong fermion-fermion interaction, mediated by a soft boson, destroys fermionic coherence and also gives rise to an attraction in one or more pairing channels. The two tendencies compete with each other, and in a class of large N models, where the tendency to incoherence is parametrically stronger, one would naively expect an incoherent (non-Fermi liquid) normal state behavior to persist down to T=0. However, this is not the case for quantum-critical systems, described by Eliashberg theory. In such systems, the part of the fermionic self-energy Σ(ωm), relevant for spin-singlet pairing, is large for a generic Matsubara frequency ωm=πT(2m+1) but vanishes for fermions with ωm=±πT, while the pairing interaction between fermions with these two frequencies remains strong. It has been shown [Y. Wang Phys. Rev. Lett. 117, 157001 (2016)PRLTAO0031-900710.1103/PhysRevLett.117.157001] that due to this peculiarity, the onset temperature for the pairing Tp is finite even at large N, when the scaling analysis predicts a non-Fermi liquid normal state. We consider the system behavior below Tp and contrast the conventional case, when ωm=±πT are not special, and the case when the pairing is induced by fermions with ωm=±πT. We obtain the solution of the nonlinear gap equations in Matsubara frequencies and then convert to real frequency axis and obtain the spectral function A(k,ω) and the density of states N(ω). In a conventional BCS-type superconductor, A(k,ω) and N(ω) are peaked at the gap value Δ(T), and the peak position shifts to a smaller ω as temperature increases towards Tp, i.e., the gap "closes in." We show that, when the pairing is induced by fermions with ωm=±πT, the situation is qualitatively different from the standard BCS result. Namely, the peak in N(ω) remains at a finite frequency even at T=Tp-0, the gap just "fills in" near this T. The spectral function A(k,ω) either shows almost the same "gap filling" behavior as the density of states, or its peak position shifts to zero frequency already at a finite Δ ("emergent Fermi arc" behavior), depending on the position of k on the Fermi surface. As an example, we compare our results with the data for the cuprates and argue that "gap filling" behavior holds in the antinodal region, while the "emergent Fermi arc" behavior holds in the nodal region.
Bibliographical noteFunding Information:
We thank D. Dessau, A. Kanigel, J. Lesueur, A. Millis, N. Prokofiev, M. Randeria, S. Raghu, G. Torroba, and A. Yazdani for useful discussions. The work by Y.-M.W. and A.V.C. was supported by the NSF DMR-1523036.
The work by Y.-M.W. and A.V.C. was supported by the NSF DMR-1523036.