Shot noise in strongly correlated double quantum spin Hall edges
Andreas Tsantilas, Trithep Devakul, Julian May-Mann
TL;DR
This work analyzes how interactions modify the edges of double quantum spin Hall insulators realized in moiré TMDs. Using bosonization, RG, and a folded Luttinger-liquid framework, it shows two symmetry-preserving edge phases: a weakly correlated phase with two helical edge pairs and a strongly correlated phase with a single pair in which single-electron excitations are gapped but paired excitations remain gapless. A key prediction is that shot-noise measurements at a quantum point contact yield a Fano factor of $\mathcal{F}=1$ for the weakly correlated edge and $\mathcal{F}=2$ for the strongly correlated edge, because the minimal backscattered charge changes from $e$ to $2e$. The results generalize to $N$-fold QSHIs, with the strongly correlated edge exhibiting a factor of $\mathcal{F}=N$, and have direct relevance for moiré TMD platforms where strong correlations are accessible. Overall, the paper provides a concrete, experimentally accessible diagnostic for interaction-driven edge physics in topological moiré materials.
Abstract
We consider the effects of interactions on the edges of ``double" quantum spin Hall insulators (DQSHIs), motivated by recent experiments on moiré twisted metal dichalcogenides. Without interactions, a DQSHI can be understood as two copies of a conventional quantum spin Hall insulator. If interactions are present and $s^z$-spin is conserved, we show that there are two possible phases for the DQSHI edge. First is a weakly correlated edge which has two pairs of helical modes and is adiabatically equivalent to two conventional quantum spin Hall edges. Second is a strongly correlated edge with only one pair of helical modes. The strongly correlated edge also has a gap to single electrons, but is gapless to pairs of electrons. In a quantum point contact geometry, this single-electron gap leads to a Fano factor of $2$ in shot noise measurements, compared to a Fano factor of $1$ for a weakly correlated edge.
