A theoretical proposal for realizing and detecting spin supercurrent in an isotropic antiferromagnetic insulator is reported. Superfluid spin transport is achieved by inserting the antiferromagnet between two metallic reservoirs and establishing a spin accumulation in one reservoir such that a spin bias is applied across the magnet. We consider a class of bipartite antiferromagnets with Néel ground states, and temperatures well below the ordering temperature, where spin transport is mediated essentially by the condensate. Landau-Lifshitz and magnetocircuit theories are used to directly relate spin current in different parts of the heterostructure to the spin-mixing conductances characterizing the $\text{antiferromagnet}|\text{metal}$ interfaces and the antiferromagnet bulk damping parameters, quantities all obtainable from experiments. We study the efficiency of spin angular-momentum transfer at an $\text{antiferromagnet}|\text{metal}$ interface by developing a microscopic scattering theory for the interface and extracting the spin-mixing conductance for a simple model. Within the model, a quantitative comparison between the spin-mixing conductances obtained for the $\text{antiferromagnet}|\text{metal}$ and $\text{ferromagnet}|\text{metal}$ interfaces is made.

• Received 16 April 2014
• Revised 1 August 2014

DOI:https://doi.org/10.1103/PhysRevB.90.094408