Decoupling interacting spins in solids
The dominant interaction between nuclear spins in solids is the magnetic dipolar interaction. This interaction is also key in systems of dilute electronic spins in which the distance between spins is greater than than the spatial scale of the electron wave function. These dipolar interactions can significantly impact our ability to control the dynamics of an electronic or nuclear spin system. In NMR spectroscopy these interactions shorten the observed coherence times and broaden the observed resonance lines.
Periodic decoupling techniques based on Average Hamiltonian Theory allow us to engineer the dynamics of an interacting spin system. In the simplest case, the goal is to just refocus interactions completely, resulting in a “time-suspension” sequence. The figure alongside shows the performance of a variety of decoupling sequences including the classic Wahuha (WHH), MREV8 and Cory-48 sequences, as well as two sequences (YXX24 and YXX48) that were designed using reinforcement learning (in collaboration with Paola Cappellaro’s group at MIT) for the proton spins in adamantane. An interesting observation is that the WHH and MREV8 sequences, which attempt to preserve information about DC magnetic field offsets, do not perform as well as those sequence which intentionally average these effects out.
We are currently exploring the use of both conventional Hamiltonian Engineering and reinforcement learning to design sequences for sensing as well as time-suspension. Our goal is to design sequences that are robust to experimental imperfections and can be adapted for use in both electronic and nuclear spin systems. The figure alongside shows the results of a WHH experiment on a system of interacting P1 centers (substitutional nitrogen) in diamond as a function of the pulse spacing.