Conditions for magnetically induced singlet d-wave superconductivity on the square lattice

It is expected that at weak to intermediate coupling, d-wave superconductivity can be induced by antiferromagnetic fluctuations. However, one needs to clarify the role of Fermi surface topology, density of states, pseudogap, and wave vector of the magnetic fluctuations on the nature and strength of the induced d-wave state. To this end, we study the generalized phase diagram of the two-dimensional half-filled Hubbard model as a function of interaction strength U / t, frustration induced by second-order hopping t / t, and temperature T / t. In experiment, U / t and t / t can be controlled by pressure. We use the two-particle self-consistent approach, valid from weak to intermediate coupling. We first calculate as a function of t / t and U / t the temperature and wave vector at which the spin response function begins to grow exponentially. d-wave superconductivity in a half-filled band can be induced by such magnetic fluctuations at weak to intermediate coupling, but only if they are near commensurate wave vectors and not too close to perfect nesting conditions where the pseudogap becomes detrimental to superconductivity. For given U / t, there is thus an optimal value of frustration t / t where the superconducting Tc is maximum. The noninteracting density of states plays little role. The symmetry dx2−y2 vs dxy of the superconducting order parameter depends on the wave vector of the underlying magnetic fluctuations in a way that can be understood qualitatively from simple arguments.