Abstract:Very recently, higher-order topological insulators and superconductors have attracted wide attention and aroused the great interest of researchers. Owing to their unconventional bulk-boundary correspondence, higher-order topological insulators and superconductors possess novel boundary modes whose dimensions are always lower than the first-order (or say conventional) topological insulators and superconductors, provided that their bulk dimensions are the same. The essence of higher-order topological phase is the formation of Dirac-mass domain walls on the gapped one-dimensional lower boundary. Roughly speaking, the origins of the formation can be classified as " intrinsic” and " extrinsic” type. For the former case, the formation of domain walls is forced by symmetry, suggesting that the resulting higher-order topological phases can be taken as topological crystalline phases. For this case, the domain walls are quite robust if the corresponding symmetry is preserved. For the latter case, the domain walls are formed simply because the one-dimensional lower boundary modes are gapped in a nontrivial way, however, the nontrivial way is not forced by symmetry. For this case, the domain walls are also stable against perturbations as long as the separations between them are large enough. The domain walls can have various patterns, which indicates that the higher-order topological phases are very rich. In this paper, we first reveal the connection between the higher-order topological phase and the first-order topological phase. Concretely, we show how to realize a higher-order topological phase by breaking some symmetries of a first-order topological phase, as well as stacking lower-dimensional first-order topological systems in an appropriate way. After these, we review the recent progress of theoretical and experimental study of higher-order topological insulators and superconductors. For the higher-order topological insulators, we find that the electronic materials are still laking though a lot of experimental realizations have been achieved. For higher-order topological superconductors, we find that their experimental realization and investigation are still in the very primary stage though quite a lot of relevant theoretical studies have been carried out. In order to comprehensively understand this newly-emerging field there are still many things to be done. Keywords:higher-order topological insulator/ higher-order topological superconductor/ bulk-boundary correspondence
其中, ${{{\sigma}}_{x, y, z}}$和${{{s}}_{x, y, z}}$为泡利矩阵, 分别作用在子能带(subbands)自由度和自旋自由度上; 为了简化表达式, 晶格长度在本文中将自始至终设置为1; A, B和M为常数. 不失一般性, 我们考虑A和B均为正数. 刻画此哈密顿量的${Z_2}$不变量[37]具有下面的形式:
其中$A(\ell ) = A$, 而$m(\ell )$在I, II, III, IV四个边界上的取值分别为$ - \varLambda M/2 B$, $\varLambda M/2 B$, $ - \varLambda M/2 B$, $\varLambda M/2 B$. 由此可以看出, 在角的地方$A(\ell )$保持不变, 而$m(\ell )$的符号将发生改变. 这意味着在角的地方正好形成了一个狄拉克质量的畴壁. 根据Jackiw-Rebbi理论[39], 这样的畴壁将束缚一个零维的零能量束缚态. 由于四个角都形成了畴壁, 因此一共存在四个零能量束缚态. 由于这些零能量束缚态存在于系统的角上, 因此它们也被称为角模(corner modes). 利用边界态理论, 可以清晰地看出如何从一阶拓扑物态转变为高阶拓扑物态. 简言之, 如果从一个d维的一阶拓扑物态出发, 要实现一个d维的高阶拓扑物态, 首先需要破坏保护一阶拓扑物态无能隙边界态的对称性, 如保护一阶拓扑绝缘体的时间反演对称性; 其次, 破坏对称性的方式需要具有各向异性的特点. 虽然利用边界态理论时为了分析上的简单, 我们假设了新引入的项可以当作微扰, 但这并不意味着上面的图像和结论只有当Λ很小时成立. 为了显示这一事实, 直接根据(9)式中的格点哈密顿量数值计算Λ项对边界态的影响. 计算结果确定了即使Λ项不能当作微扰, 上述的图像依然成立, 如图2所示. 图 2 从一阶拓扑绝缘体到二阶拓扑绝缘体 (a) 沿x方向取开放边界条件(${L_x} = 100$), 沿y方向取周期边界条件, 参数为$M = B = A = 1$, $\varLambda = 0$, 对应BHZ模型, 能谱反映出无能隙边界态的存在; (b) 插图中红点对应的能量本征态的波函数分布, 参数同(a), 但沿x和y两个方向均取开放边界条件; 红色的深浅对应波函数分布概率的大小, 可以看出对一阶拓扑绝缘体, 波函数分布在整个边界上; (c) 边界条件和参数同(a), 除了此处$\varLambda = 0.5$, 可看出Λ项的出现让边界态打开了能隙; (d) 零模的波函数分布, 参数同(c), 但沿x和y两个方向均取开放边界条件; 从插图中可发现存在四个零模, 这四个零模的波函数局域在四个角上 Figure2. From first-order topological insulator to second-order topological insulator. (a) Energy spectra for a sample with open boundary condition in the x direction (the system size ${L_x} = 100$) and periodic boundary condition in the y direction. Parameters are $M = B = A = 1$, $\varLambda = 0$, which corresponds to the original BHZ model. The energy spectra reflect the existence of gapless boundary modes. (b) the density profile of a boundary mode. The parameters are the same as in (a), but now open boundary conditions are taken both in the x and y directions. One can see that the density profile of the boundary mode distributes over the whole boundary. (c) the boundary conditions and parameters are the same as in (a), except now $\varLambda = 0.5$. One can see that the presence of the Λ term opens a gap for the boundary modes. (d) the density profiles of zero modes.The parameters are the same as in (c), but now open boundary conditions are taken both in the x and y directions. One can see that there are four zero-energy modes in the inset. Their wave functions are found to be localized around the corners.
其中${t_x}$和${t'_x}$为最近邻格点间的跃迁系数, 如图3(a)所示. ${{{\sigma}} _i}$为泡利矩阵, 作用在$ ({\rm{A}}, {\rm{B}})$子格上. 此模型由于具有子格(或手征)对称性(算符为${{{\sigma}} _z}$), 其拓扑性质由绕数(winding number)刻画, 图 3 从一维一阶拓扑绝缘体到二维二阶拓扑绝缘体 (a) 一维SSH链的示意图; (b) 利用一维SSH链构造二维二阶拓扑绝缘体, 每个单位元胞中有一个${\text{π}}$磁通 Figure3. Constructing two-dimensional second-order topological insulator by using one-dimensional topological insulator: (a) A schematic diagram of the SSH chain; (b) using the one-dimensional SSH chains to construct a two-dimensional second-order topological insulator, within each small square, there is a ${\text{π}}$-flux.