Topological quantum computation (TQC) based on Majorana zero modes (MZM) has been actively pursued in the past decade. Aside from well-recognized challenges in unambiguously identifying MZM in existing effective one-dimensional p-wave superconductors, the next step of demonstrating braiding of MZM is even more formidable. The most prevalent proposal on realizing braiding operation is based on networks of effective p-wave superconductor wires subject to many tiny electric gates that are distributed along the wires and can be sequentially turned on or off, so that the MZM can be adiabatically moved across different parts of the network. It is then a natural question to ask if there exists an alternative, possibly more practical architecture for braiding MZM, that may even be more friendly to near-term MZM systems.
Motivated by this question, as well as the recent experimental breakthrough on demonstrating MZM in a minimal Kitaev chain consisting of only two quantum dots [Dvir et al., Nature 614, 445 (2023)], we propose in [1] a couple of new structures for braiding MZM based on triangular superconducting islands that can either be constructed in a bottom-up manner as a minimal extension of the two-site Kitaev chain, or appear naturally in epitaxial growth of ultrathin films. For the former case, we have shown that a minimal 3-site “Kitaev triangle” can host MZM at different pairs of vertices that can be controlled by a non-uniform vector potential. We have demonstrated braiding two MZM in this minimal model by explicitly calculating the many-body Berry phase. For the latter case, we have extended the 3-site Kitaev triangle to a finite-size triangular island with a hollow interior. We have shown that using a uniform vector potential, one can induce a pair of MZM at different pairs of vertices of the triangle. Moreover, rotating the uniform vector potential can change the positions of the MZM without closing the bulk band gap, i.e., adiabatically. Finally we have given a scalable design, backed by numerical calculations, for braiding two out of four MZM that corresponds to nontrivial logical gate operations, in a network of corner-sharing hollow triangles.
Our work provides a novel platform in parallel with the coupled-wire network design for braiding MZM, and is practical especially considering near-term MZM devices. The Kitaev triangle offers an alternative strategy towards MZM-based TQC that is not based on bulk-boundary correspondence, which may be easier to realize in near-term quantum-dot=based devices than coupled wires. On the other hand, our proposal of a uniform vector potential coupled to hollow triangles explores the utility of geometry rather than the individual control of superconducting nanowires. It highlights that triangles, as a geometry, are unique compared to other quasi-2D structures such as wires, squares, or circles, since they naturally break 2D inversion symmetry and do not present a straightforward strategy for morphing into either 1D or 2D structures with periodic boundary conditions.