Algorithmic Cooling of Nuclear Spin Pairs using a Long-Lived Singlet State.

Algorithmic cooling methods manipulate an open quantum system in order to lower its temperature below that of the environment. We show that significant cooling is achieved on an ensemble of spin-pair systems by exploiting the long-lived nuclear singlet state, which is an antisymmetric quantum superposition of the "up" and "down" qubit states. The effect is demonstrated by nuclear magnetic resonance (NMR) experiments on a molecular system containing a coupled pair of near-equivalent 13C nuclei. The populations of the system are subjected to a repeating sequence of cyclic permutations separated by relaxation intervals. The long-lived nuclear singlet order is pumped well beyond the unitary limit, and the nuclear magnetization is enhanced by 21% relative to its thermal equilibrium value. To our knowledge this is the first demonstration of algorithmic cooling using a quantum superposition state and without making a distinction between rapidly and slowly relaxing qubits.