We derive an efficient formula for Z 2 topological invariants characterizing the quantum spin Hall effect. It is defined in a lattice Brillouin zone, which enables us to implement numerical calculations for realistic models even in three dimensions. Based on this, we study the quantum spin Hall effect in Bi and Sb in quasi-two and three dimensions using a tight-binding model. Quantum spin Hall (QSH) effect [1][2][3][4][5] has been attracting much current interest as a new device of spintronics. [6][7][8][9] It is a topological insulator [10][11][12] analogous to the quantum Hall (QH) effect, but it is realized in time-reversal (T ) invariant systems. While QH states are specified by Chern numbers, 13,14) QSH states are characterized by Z 2 topological numbers, 2) which suggests that Z 2 invariants would have deep relationship with the Z 2 anomaly of the Majorana fermions. 15,16) Graphene has been expected to be in the QSH phase.
1,2)However, recent calculations have suggested that the spinorbit coupling in graphene is too small to reveal the QSH effect experimentally. 17,18) Recently, it has been pointed out that Bi thin film is another plausible material for QSH effect.19) Also by the idea of adiabatic deformation of the diamond lattice, it has been conjectured that Bi in three dimensions (3D) is in a topological phase.
20)While systems in two dimensions (2D) are characterized by a single Z 2 topological invariant, four independent Z 2 invariants are needed in 3D. [20][21][22][23] This makes it difficult to investigate realistic models, in which complicated many-band structure is involved. Therefore, for the direct study of Bi in 3D as well as for the search for other materials, to establish a simple and efficient computational method of Z 2 invariants in 3D is an urgent issue to be resolved.In this paper, we present a method of computing Z 2 invariants based on the formula derived by Fu and Kane 24) together with the recent development of computing Chern numbers in a lattice Brillouin zone. [25][26][27] This method is based on recent developments in lattice gauge theories [30][31][32][33][34][35] but simple enough to compute Z 2 invariants even for realistic 3D systems. Based on this, we study a tight-binding model for Bi and Sb.First, we derive a lattice version of the Fu-Kane formula.24) To this end, we restrict our discussions, for simplicity, to systems in 2D, where a single Z 2 invariant is relevant. Let T be the time-reversal transformation T ¼ i 2 K, and assume that the Hamiltonian in the momentum space HðkÞ transforms under T as T HðkÞT À1 ¼ HðÀkÞ. Let ðkÞ ¼ ðj1ðkÞi; . . . ; j2MðkÞiÞ denote the 2M dimensional ground state multiplet of the Hamiltonian: HðkÞjnðkÞi ¼ E n ðkÞjnðkÞi. 11,12) Assuming that the many-body energy gap is finite, we focus on topological invariants under the Uð2MÞ transformation ðkÞ ! ðkÞUðkÞ; UðkÞ 2 Uð2MÞ:As discussed, 2,27) the pfaffian defined by pðkÞ ¼ pf É y ðT ÉÞ characterizes the topological phases of T invariant systems. To be precise, the systems belong to topological insu...