betaFeedback

WorksPhysics and Astronomy

Second Chern Number Protected Gapless Vortex Modes in Superconducting Weyl Semimetals

Abstract

We have studied the vortex bound states of superconducting linear and quadratic Weyl semimetals with time reversal symmetry. For vortex with vorticity nn (vortex line along zz direction) and Weyl semimetals with Weyl point chirality χχ, there are ∣nχ∣\vert nχ\vert gapless vortex modes around the location of each Weyl point in superconducting Weyl semimetals. We show that the appearance of ∣nχ∣\vert nχ\vert gapless vortex modes originates from the imbalanced pumping of flux bound states between Chern insulators obtained at kzk_{z} and −kz-k_{z} in Weyl semimetals. Such a physical picture is verified by numerical calculations on lattice models. Moreover, the gapless modes around each Weyl point can be characterized by the second Chern number defined in a synthetic four-dimensional space based on the effective BdG Hamiltonian. Finally, these gapless vortex modes are found to be robust against rotational symmetry-breaking terms.

Introduction

Majorana zero modes (MZMs) are fascinating objects which obey non-Abelian statistics, and the braiding of MZMs can be used to construct fault-tolerant quantum gates [1, 2, 3, 4, 5, 6, 7]. It is predicted that there are MZMs in the vortex cores of two-dimensional chiral pp-wave superconductors [1]. Subsequently, it is found that there is effective px+ipyp_x + ip_y superconductivity on the interface of the ss-wave superconductor-TI heterostructure [8], and an MZM could be trapped at the vortex core in such a heterostructure [9]. Since then, the vortex bound states (VBSs) of superconducting (SC) topological insulators (TIs) [8] [9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20], iron-based superconductors [21–31], and SC Dirac semimetals [31–45] have attracted a lot of attentions.

Recent years have also witnessed remarkable progress in the study of the VBSs of SC Weyl semimetals [42, 46–65]. It is found that the vortex line can host gapless chiral Majorana modes in the pair density wave phase of a Weyl semimetal breaking time reversal symmetry [61]. For time reversal symmetric Weyl semimetals with BCS pairing, it is predicted that if every constant-kzk_z (the vortex is inserted along zz direction) plane contains zero chirality of Weyl points, the VBSs will be fully gapped, and if certain constant-kzk_z planes contain a net chirality of Weyl points, there will be gapless modes [63]. The topological invariants have also been proposed to characterize the non-trivial fully gapped phase with vortex-end MZMs [42, 63]. However, previous studies have focused on the VBSs of superconducting linear Weyl semimetals, whereas those of superconducting quadratic Weyl semimetals remain less explored. Moreover, topological invariants characterizing the gapless vortex modes in time-reversal-symmetric superconducting linear and quadratic Weyl semimetals have yet to be proposed, and the physical origin of the appearance of these gapless modes hasn’t been thoroughly understood. Apart from the theoretical study, experimental progress about superconducting Dirac semimetals and Weyl semimetals has also been reported in the past decade [38, 66–68]. Some candidate materials of Weyl semimetals are found to be superconductors under high pressure [50, 52, 58], and technologies such as point contact [69] and selective ion sputtering [70] have also been developed to induce superconductivity in Dirac semimetals and Weyl semimetals. The experimental progress facilitates the study of VBSs of SC Weyl semimetals and calls for further theoretical exploration.

In this article, we study the VBSs of SC linear and quadratic Weyl semimetals [71, 72] with time reversal symmetry (TRS). As shown in Figure 1, we consider a flux inserted into a SC Weyl semimetal with chirality χ\chi, and χ=1\chi= 1 (χ=2\chi= 2) corresponds to linear (quadratic) Weyl semimetal. Figure 1(c) shows the schematic figure of the

Figure 1. (a) The SC Weyl semimetal and inserted flux with vorticity nn. (b) The four Weyl points are located on the kzk_z axis, and Weyl points colored in red (blue) have the chirality χ\chi (−χ-\chi). The first Chern numbers are labelled for two-dimensional insulators at different kzk_z. (c) There are ∣nχ∣|n\chi| gapless propagating modes around the kzk_z component of each Weyl point.

Model

In this section, we write down the k⋅pk\cdot p model of linear and quadratic Weyl semimetals. First, the continuous model of the linear Weyl semimetal with TRS, and C4zC_4^z rotational symmetry is written under the basis (∣+,12⟩,∣−,32⟩,∣+,−12⟩,∣−,−32⟩)T(|+,\frac{1}{2}\rangle,|-,\frac{3}{2}\rangle,|+,-\frac{1}{2}\rangle,|-,-\frac{3}{2}\rangle)^T, in which each state is the eigenstate of inversion and C4zC_4^z operations labelled by the corresponding eigenvalues.

HLWSM(k)=(M1(k)Ak+0B∗(k)Ak−−M1(k)B∗(k)00B(k)M2(k)−Ak−B(k)0−Ak+−M2(k)).(1)H_{\mathrm{LWSM}}(\boldsymbol{k})= \begin{pmatrix} M_1(\boldsymbol{k}) & Ak_+ & 0 & B^*(\boldsymbol{k}) \\ Ak_- & -M_1(\boldsymbol{k}) & B^*(\boldsymbol{k}) & 0 \\ 0 & B(\boldsymbol{k}) & M_2(\boldsymbol{k}) & -Ak_- \\ B(\boldsymbol{k}) & 0 & -Ak_+ & -M_2(\boldsymbol{k}) \end{pmatrix}. \tag*{(1)}

Here M1,2(k)=M0−M1kz2−M2(kx2+ky2)±ηkzM_{1,2}(\boldsymbol{k})=M_0-M_1k_z^2-M_2(k_x^2+k_y^2)\pm\eta k_z, and B(k)=βkz(kx2−ky2)+2iαkxkykzB(\boldsymbol{k})=\beta k_z(k_x^2-k_y^2)+2i\alpha k_xk_yk_z. As shown in Figure 1(b), the gap-closing condition of the above Hamiltonian gives four Weyl points with chirality χ=1\chi=1 located on kzk_z axis. The corresponding tight-binding lattice model is given in the supplemental materials [73], in which we also give the correspondence between the parameters in k⋅pk\cdot p model and the parameters in tight-binding lattice model.

The k⋅pk\cdot p model of the quadratic Weyl semimetal is written under the basis (∣+,52⟩,∣−,12⟩,∣+,−52⟩,∣−,−12⟩)T(|+,\frac{5}{2}\rangle,|-,\frac{1}{2}\rangle,|+,-\frac{5}{2}\rangle,|-,-\frac{1}{2}\rangle)^T, in which each state is the eigenstate of inversion and C6zC_6^z operations labelled by the corresponding eigenvalues.

HQWSM=(M1(k)Ak−2kz0B∗(k)Ak+2kz−M1(k)B∗(k)00B(k)M2(k)−Ak−2kzB(k)0−Ak+2kz−M2(k)).(2)H_{\mathrm{QWSM}}= \begin{pmatrix} M_1(\boldsymbol{k}) & Ak_-^2k_z & 0 & B^*(\boldsymbol{k}) \\ Ak_+^2k_z & -M_1(\boldsymbol{k}) & B^*(\boldsymbol{k}) & 0 \\ 0 & B(\boldsymbol{k}) & M_2(\boldsymbol{k}) & -Ak_-^2k_z \\ B(\boldsymbol{k}) & 0 & -Ak_+^2k_z & -M_2(\boldsymbol{k}) \end{pmatrix}. \tag*{(2)}

In the above Hamiltonian Eq. (2), we have M1,2(k)=M0−M1kz2−M2(kx2+ky2)±ηkzM_{1,2}(\boldsymbol{k})=M_0-M_1k_z^2-M_2(k_x^2+k_y^2)\pm\eta k_z, and B(k)=αkx(3ky2−kx2)+iβky(3kx2−ky2)B(\boldsymbol{k})=\alpha k_x(3k_y^2-k_x^2)+i\beta k_y(3k_x^2-k_y^2). Eq. (2) describes a quadratic Weyl semimetal with four quadratic Weyl points with chirality χ=−2\chi=-2 located on the kzk_z axis. The corresponding tight-binding lattice model is given in the supplemental materials [73].

The bases of the k⋅pk\cdot p models of linear and quadratic Weyl semimetals can be written in a unified form ψ(k)=(CA,1(k),CB,m(k),CA,−1(k),CB,−m(k))\psi(\boldsymbol{k})=(C_{A,1}(\boldsymbol{k}),C_{B,m}(\boldsymbol{k}),C_{A,-1}(\boldsymbol{k}),C_{B,-m}(\boldsymbol{k})), in which A/BA/B label the parity, and l/ml/m label the angular momentum LzL_z. When we consider the Hamiltonian describing SC Weyl semimetals, we choose the Nambu basis as Ψ(k)=(ψ(k),ψ†(−k))T\Psi(\boldsymbol{k})=(\psi(\boldsymbol{k}),\psi^\dagger(-\boldsymbol{k}))^T. Under this basis, the BdG Hamiltonian of the SC Weyl semimetal is H=∑kΨ†(k)HBdG(k)Ψ(k)H=\sum_{\boldsymbol{k}}\Psi^\dagger(\boldsymbol{k})H_{\mathrm{BdG}}(\boldsymbol{k})\Psi(\boldsymbol{k}), where

HBdG(k)=(H0(k)−μΔΔ†μ−H0∗(−k)).(3)H_{\mathrm{BdG}}(\boldsymbol{k})= \begin{pmatrix} H_0(\boldsymbol{k})-\mu& \Delta\\ \Delta^\dagger& \mu-H_0^*(-\boldsymbol{k}) \end{pmatrix}. \tag*{(3)}

In the BdG Hamiltonian Eq. (3), H0(k)H_0(\boldsymbol{k}) is the normal state Hamiltonian of linear or quadratic Weyl semimetal. The vortex line adds the phase eiθe^{i\theta} (θ=arctan⁡y/x\theta=\arctan y/x) to the pairing potential and preserves the translation symmetry in zz direction. Thus, kzk_z is a good quantum number, and we can calculate the VBSs for each kzk_z with open boundary condition in xx and yy directions. We consider the ss-wave singlet pairing and assume that the pairing potential at the vortex core is Δ=0\Delta=0, and Δ(r)=Δ0eiθ\Delta(\boldsymbol{r})=\Delta_0e^{i\theta} on other lattice sites. For SC linear Weyl semimetal, the parameters in the lattice model are set as {t1D,t4C,m,t1A,t3A,t4B,Δ0}={−0.25,−0.125,2.2,−0.5,0.5,0.25,0.2}\{t_1^D,t_4^C,m,t_1^A,t_3^A,t_4^B,\Delta_0\}=\{-0.25,-0.125,2.2,-0.5,0.5,0.25,0.2\}, and for SC quadratic Weyl semimetal, the parameters in the lattice model are set as {t1D,t1A,m,t4A,t4B,t3C,Δ0}={−2.5,−1.4,8.4,−1.0,−0.46,0.6,0.4}\{t_1^D,t_1^A,m,t_4^A,t_4^B,t_3^C,\Delta_0\}=\{-2.5,-1.4,8.4,-1.0,-0.46,0.6,0.4\}.

The VBSs Characterized by the Second Chern Number

We have performed numerical calculations of the VBSs of SC linear and quadratic Weyl semimetals, and the results are shown in Figure 2. Concretely, Figure 2(a) shows the VBSs of SC linear Weyl semimetal (χ=1\chi=1) with vorticity n=1n=1. Figure 2(b) shows the VBSs of SC quadratic Weyl semimetal (χ=−2\chi=-2) with vorticity n=1n=1. The red and blue points in the two sub-figures label the $k_z coordinates of Weyl points with opposite chirality. We notice that the gapless modes around each Weyl point can be characterized by the product −nχ-n\chi. To be specific, the number of gapless modes around the kzk_z component of each Weyl point is ∣nχ∣|n\chi|, and the propagating direction of gapless modes around each Weyl point is determined by the sign of the product −nχ-n\chi. For example, there is ∣nχ∣=1|n\chi|=1 gapless mode around each linear Weyl point in Figure 2(a), while there are ∣nχ∣=2|n\chi|=2 gapless modes around each quadratic Weyl point in Figure 2(b). Besides, the propagating direction of gapless modes around each Weyl point in Figure 2(b) is opposite to that in Figure 2(a), which is consistent with the fact that we have χ=−2\chi=-2 in Figure 2(b) and χ=1\chi=1 in Figure 2(a). In fact, in the supplemental materials [73], there are more numerical examples demonstrating that the dependence of propagating direction and number of gapless modes on nn and χ\chi can be characterized by the product −nχ-n\chi.

VBSs of SC Weyl semimetals

Figure 2. The VBSs of SC Weyl semimetals. The red and blue points in the two sub-figures are Weyl points with opposite chirality. (a) The VBSs (E−kzE-k_z relation) of SC linear Weyl semimetal are calculated with μ=0.2\mu=0.2, η=0.3\eta=0.3, and n=1n=1. (b) The VBSs of SC quadratic Weyl semimetal are calculated with μ=0.4\mu=0.4, η=0.7\eta=0.7, and n=1n=1.

We find that the product −nχ-n\chi is nothing but the emergent second Chern number [74–77] defined in a synthetic four-dimensional space that stems from the effective BdG Hamiltonian [78, 79]. For simplicity, we fix μ=0\mu=0 in the following discussions. In this case, the normal-state Fermi surface consists of isolated Weyl points. To derive the effective BdG Hamiltonian, we focus on the low-energy physics around the four Weyl points. The detailed derivations of the effective BdG Hamiltonian are given in the supplementary materials [73]. If we use qiq_i (i=x,y,zi=x,y,z) to denote the momentum relative to a Weyl point and assume q±=qx±iqy=qe±iθq_{\pm}=q_x\pm iq_y=qe^{\pm i\theta}, the final form of the effective BdG Hamiltonian is

HBdG(q)=(vqzAqχΔeinϕ0Aqχ−vqz0ΔeinϕΔe−inϕ0−vqz−Aqχ0Δe−inϕ−Aqχvqz).(4)H_{\mathrm{BdG}}(\mathbf{q})= \begin{pmatrix} vq_z & Aq^{\chi} & \Delta e^{in\phi} & 0 \\ Aq^{\chi} & -vq_z & 0 & \Delta e^{in\phi} \\ \Delta e^{-in\phi} & 0 & -vq_z & -Aq^{\chi} \\ 0 & \Delta e^{-in\phi} & -Aq^{\chi} & vq_z \end{pmatrix}. \tag*{(4)}

For a region far away from the vortex core, we may assume it is a three-dimensional space with approximate constant ϕ\phi. Then combining all such regions together leads to a synthetic four-dimensional (qx,qy,qz,ϕ)(q_x,q_y,q_z,\phi) space with periodic boundary condition in parameter ϕ\phi [43, 74]. Therefore, we can define the second Chern number C2C_2 in this synthetic four-dimensional space.

To calculate C2C_2 of the above effective BdG Hamiltonian, we reformulate Eq. (4) as

HBdG=Aqχcos⁡(χθ)χ3⊗τ1+Aqχsin⁡(χθ)χ3⊗τ2+vqzχ3⊗τ3−Δsin⁡(nϕ)χ2⊗τ0+Δcos⁡(nϕ)χ1⊗τ0.(5)H_{\mathrm{BdG}}=Aq^{\chi}\cos(\chi\theta)\chi_3\otimes\tau_1+Aq^{\chi}\sin(\chi\theta)\chi_3\otimes\tau_2 +vq_z\chi_3\otimes\tau_3-\Delta\sin(n\phi)\chi_2\otimes\tau_0 +\Delta\cos(n\phi)\chi_1\otimes\tau_0. \tag*{(5)}

Then we define d1=Aqχcos⁡χθd_1=Aq^{\chi}\cos\chi\theta, d2=Aqχsin⁡χθd_2=Aq^{\chi}\sin\chi\theta, d3=vqzd_3=vq_z, d4=−Δsin⁡nϕd_4=-\Delta\sin n\phi, d5=Δcos⁡nϕd_5=\Delta\cos n\phi. In this way, we can write the effective BdG Hamiltonian as HBdG(q)=∑idi(q,ϕ)⋅ΓiH_{\mathrm{BdG}}(\mathbf{q})=\sum_i d_i(q,\phi)\cdot\Gamma_i, in which Γi\Gamma_i are five mutually anti-commuting matrices in Eq. (5). The definition of the second Chern number is

C2=38π2∫d3q∫dϕ1d5ϵabcdeda∂qdb∂qdc∂qdd∂qde.(6)C_2=\frac{3}{8\pi^2}\int d^3q\int d\phi\frac{1}{d^5}\epsilon^{abcde}d_a\partial_qd_b\partial_qd_c\partial_qd_d\partial_qd_e. \tag*{(6)}

Here we have defined d=(∑idi2)1/2d=(\sum_i d_i^2)^{1/2}. If we substitute the explicit forms of the coefficients did_i into the definition of the second Chern number, we will obtain C2=−nχC_2=-n\chi after some algebra [73]. This expected topological invariant which describes the dependence of gapless modes on the vorticity nn and chirality χ\chi in SC linear and quadratic Weyl semimetals.

The Pumping Picture

Now we provide a physical picture to understand such dependence of the VBSs on nn and χ\chi. As shown in Figure 1(b), the Weyl semimetals we consider have four Weyl points on kzk_z axis, and Chern numbers can be defined for the two-dimensional insulators obtained by fixing kzk_z in the model of Weyl semimetals. Thus, the Weyl semimetals can be regarded as Chern insulators with different Chern numbers stacked along zz direction. The imbalanced pumping between Chern insulators at kzk_z and −kz-k_z in Weyl semimetals can lead to the appearance of ∣nχ∣|n\chi| gapless vortex modes around the location of each Weyl point in the VBSs of SC Weyl semimetals. Now we take the VBSs of SC linear Weyl semimetal as an example, and we consider inserting a flux into the linear Weyl semimetal along zz direction. Figure 3 shows the pumping of the spectra of Chern insulators at kz0k_{z0} and −kz0-k_{z0}. Here kz0=0.35πk_{z0}=0.35\pi is a typical value of kzk_z, at which the insulator has nonzero Chern number. The red dots are states which localize around the flux core, and they are dubbed flux bound states. Because of the existence of TRS, Chern insulators at kzk_z and −kz-k_z have opposite Chern numbers. As a result, we can find that the flux bound states at kz0k_{z0} cross the Fermi level downwards, while the flux bound states at −kz0-k_{z0} cross the Fermi level upwards. In fact, there is also the pumping of edge state. But to highlight the pumping of flux bound states, we don’t plot the edge states in this figure.

Energy spectrum of flux bound states

Figure 3. Understanding the appearance of the gapless modes from the π\pi-flux pumping of Chern insulators in a linear Weyl semimetal. The red dots are flux bound states localized at the flux core. This figure shows the pumping of flux bound states of two-dimensional insulators at kz0k_{z0} (0.35π)(0.35\pi) and −kz0-k_{z0} as the inserted flux increases from ϕ=0\phi=0 to ϕ=2π\phi=2\pi. The results are calculated with μ=0\mu=0 and η=0.3\eta=0.3.

Although Figure 3 shows the 2π2\pi-flux pumping, we should focus on the π\pi-flux pumping to understand the VBSs of SC linear Weyl semimetals. Combining the π\pi-flux pumping of the flux bound states of two Chern insulators at kz0k_{z0} and −kz0-k_{z0}, we conclude that depending on the value of chemical potential, the flux bound states at −kz0-k_{z0} are pumped from occupied to unoccupied, or the flux bound states at kz0k_{z0} are pumped from unoccupied to occupied. In the BdG Hamiltonian, there are both particle and hole degrees of freedom [62], so we have to consider the hole counterpart of the aforementioned pumping process. It turns out that for the π\pi-flux pumping in the hole part, we also obtain the result that the flux bound states at −kz0-k_{z0} are pumped from occupied to unoccupied, or the flux bound states at kz0k_{z0} are pumped from unoccupied to occupied. We assume that the pumping of flux bound states won’t be qualitatively changed in the weak pairing limit. Consequently, we have Nsc(kz0)−Nsc(−kz0)=2N_{\mathrm{sc}}(k_{z0})-N_{\mathrm{sc}}(-k_{z0})=2, where Nsc(kz)N_{\mathrm{sc}}(k_z) is the number of negative energy vortex bound states at kzk_z. Meanwhile, as shown in Figure 2(a), we notice that around kz=0k_z=0 we have Nsc(kz)−Nsc(−kz)=0N_{\mathrm{sc}}(k_z)-N_{\mathrm{sc}}(-k_z)=0, because for insulators around kz=0k_z=0 with Chern number C=0C=0, there is no the pumping of flux bound states. However, if kzk_z is greater than the kzk_z coordinate of the right blue Weyl point in Figure 2(a), we will have Nsc(kz)−Nsc(−kz)=2N_{\mathrm{sc}}(k_z)-N_{\mathrm{sc}}(-k_z)=2 according to the aforementioned argument. So we expect that there is a positive-energy VBS crossing the Fermi level and becoming a negative-energy VBS, resulting ∣nχ∣=1|n\chi|=1 gapless vortex mode around the right blue Weyl point. Similarly, we can argue that there is ∣nχ∣=1|n\chi|=1 gapless mode around the kzk_z component of each of the four Weyl points.

Analogous to the pumping picture above, we can readily understand the dependence of the number of gapless modes on the chirality χ\chi and the vorticity nn. For instance, for a quadratic Weyl semimetal with chirality χ=−2\chi=-2, we can obtain C=2C=2 Chern insulator at an appropriate value of kzk_z. Similar to Figure 3, after inserting a flux with vorticity n=1n=1, there will be two flux bound states pumped from occupied to unoccupied at −kz0-k_{z0}, or two flux bound states pumped from unoccupied to occupied at kz0k_{z0}. Correspondingly, in the weak pairing limit we have Nsc(kz0)−Nsc(−kz0)=4N_{\mathrm{sc}}(k_{z0})-N_{\mathrm{sc}}(-k_{z0})=4, and there will be ∣nχ∣=2|n\chi|=2 gapless modes around each Weyl point. On the other hand, if we keep χ=1\chi=1 (linear Weyl semimetal) and change the vorticity from 11 to 22, there are effectively two n=1n=1 fluxes. As a result, there will also be two flux bound states pumped from occupied to unoccupied at −kz0-k_{z0}, or two flux bound states pumped from unoccupied to occupied at kz0k_{z0}. Therefore, the number of gapless modes around each Weyl point also doubles. To sum up, we have shown that the number of gapless modes around each Weyl point is proportional to the product ∣nχ∣|n\chi| through the pumping picture.

The Robustness of Gapless Modes

The models of linear and quadratic Weyl semimetals above are written based on models of linear and quadratic Dirac semimetals, and they are rotational symmetric linear (C4C_4) and quadratic Weyl semimetals (C6C_6). However, it is not necessary for a Weyl semimetal to be rotational symmetric. So it is natural to ask whether the multiple gapless modes around the Weyl points are robust against rotational symmetry-breaking perturbations. To answer this question, we add the rotational symmetry breaking term ϵσ3⊗τ2\epsilon\sigma_3\otimes\tau_2 into both the linear and quadratic Weyl semimetals. As a result, the four Weyl points are no longer located on the kzk_z axis. Instead, they are now located on the ky−kzk_y-k_z plane. The corresponding VBSs of rotational symmetry-breaking SC linear and quadratic Weyl semimetals are shown in Figure 4(a) and Figure 4(b), respectively. We find from the inset in Figure 4(a) that there is still one propagating gapless mode around each Weyl point, but the intersection of two counter-propagating modes is gapped by the rotational symmetry-breaking term. Similarly, Figure 4(b) shows that there are still two propagating gapless modes around each Weyl point, while the intersections of four counter-propagating modes are gapped by the rotational symmetry-breaking perturbation.

The robustness of multiple gapless modes against rotational symmetry-breaking perturbations

Figure 4. The robustness of multiple gapless modes against rotational symmetry-breaking perturbations. (a) The VBSs of SC linear Weyl semimetal without uniaxial rotational symmetry. Here we set μ=0.2\mu=0.2, η=0.1\eta=0.1, ϵ=0.1\epsilon=0.1, and n=1n=1. (b) The VBSs of SC quadratic Weyl semimetal without uniaxial rotational symmetry. The parameters are set as μ=0.6\mu=0.6, η=0.7\eta=0.7, ϵ=0.2\epsilon=0.2, and n=1n=1.

Although the intersection of counter-propagating modes are gapped, they are far away from Fermi energy. Since the main properties of a superconductor are determined by the low-energy spectra near the Fermi energy, the spectra around the intersections don’t play the main role in determining the physical properties. In this sense, we claim that the spectra in Figure 4 are still gapless, and the multiple gapless propagating modes around each Weyl point are robust against rotational symmetry-breaking perturbations.

Summary and Discussion

In summary, we have studied the VBSs of SC Weyl semimetals with time reversal symmetry. We show that the imbalanced pumping of flux bound states between Chern insulators obtained at kzk_z and −kz-k_z in Weyl semimetals can lead to the appearance of ∣nχ∣|n_\chi| gapless modes around each Weyl point in the VBSs of SC Weyl semimetals. Such a physical picture is verified by numerical calculations based on lattice models. We also conclude from the numerical results that the gapless modes around each Weyl point can be characterized by the product −nχ-n_\chi, which is the second Chern number defined in a synthetic four-dimensional space spanned by parameters qxq_x, qyq_y, qzq_z, and ϕ\phi. Finally, we show these low-energy gapless modes are robust against rotational symmetry-breaking terms.

In this article, we have also used the pumping picture to understand the origin of gapless modes in the vortex line of SC Weyl semimetals. As a matter of fact, because the pumping of flux bound state is a generic feature for topological bands, it is natural to expect that topological VBSs will appear in not only SC Weyl semimetals but also SC topological insulators and SC Dirac semimetals, which have been extensively studied in recent years [16, 41, 42, 63]. Thus, the flux-pumping picture of the normal-state topological models may provide a unified framework for understanding VBSs in corresponding superconducting topological materials. A systematic exploration of this possibility is left for future work.

ACKNOWLEDGEMENTS

We thank Shengshan Qin and Professor Xiong-Jun Liu sincerely for helpful discussions. J.P. Hu is supported by the Ministry of Science and Technology (Grant No. 2022YFA1403900), the National Natural Science Foundation of China (Grant No. NSFC-12494594), and the New Cornerstone Investigator Program.

Supplemental Materials: Second Chern Number Protected Gapless Vortex Modes in Superconducting Weyl Semimetals

Zhicheng Zhang; Jiangping Hu

02_supplement.pdf

Supplemental Materials: Second Chern Number Protected Gapless Vortex Modes in Superconducting Weyl Semimetals

The Effective BdG Hamiltonians of SC Weyl Semimetals and Emergent Second Chern Number

In this section we will derive the effective BdG Hamiltonian [1, 2] describing the low-energy physics of SC Weyl semimetals, based on which we will define the emergent second Chern number [3, 4, 5, 6] in a synthetic four-dimensional space. We first derive the effective BdG Hamiltonian of SC linear Weyl semimetal with vorticity n=1n=1. Figure S1 shows the distribution of the four Weyl points of linear Weyl semimetals. We now focus on the effective BdG Hamiltonian with the pairing between electrons around Weyl points at k1c=(0,0,kz,1c)\mathbf{k}_1^{\mathrm{c}}=(0,0,k_{z,1}^{\mathrm{c}}) and k4c=(0,0,kz,4c)\mathbf{k}_4^{\mathrm{c}}=(0,0,k_{z,4}^{\mathrm{c}}), which is shown in Figure S1.

Figure S1. The distribution of Weyl points and the schematic figure of the BCS pairing. (a) The distribution of the four Weyl points in our linear Weyl semimetal. (b) The pairing between electrons around two Weyl points with same chirality and opposite momentum.

The low-energy effective Hamiltonians around Weyl points k1c\mathbf{k}_1^{\mathrm{c}} (Eq. (16)) and k4c\mathbf{k}_4^{\mathrm{c}} (Eq. (17)) are derived in section S-3, based on which we can derive the effective BdG Hamiltonian. We consider the Nambu basis Ψq=(c1,q,+,12†,c1,q,−,−32†,c4,−q,+,−12†,c4,−q,−,−32†)T\Psi_{\mathbf{q}}=(c_{1,\mathbf{q},+,\frac{1}{2}}^\dagger,c_{1,\mathbf{q},-,-\frac{3}{2}}^\dagger,c_{4,-\mathbf{q},+,-\frac{1}{2}}^\dagger,c_{4,-\mathbf{q},-,-\frac{3}{2}}^\dagger)^{T} and write the effective BdG Hamiltonians under this basis. Taking the pairing term into consideration, the BdG Hamiltonian can be written as HBdG=∑qΨq†HBdG(q)ΨqH_{\mathrm{BdG}}=\sum_{\mathbf{q}}\Psi_{\mathbf{q}}^\dagger H_{\mathrm{BdG}}(\mathbf{q})\Psi_{\mathbf{q}}, in which

HBdG(q)=(vqzAq+Δ0Aq−−vqz0ΔΔ0−vqz−Aq+0Δ−Aq−vqz).(7)H_{\mathrm{BdG}}(\mathbf{q})= \begin{pmatrix} v q_z & A_{\mathbf{q}+} & \Delta& 0\\ A_{\mathbf{q}-} & -v q_z & 0 & \Delta\\ \Delta& 0 & -v q_z & -A_{\mathbf{q}+}\\ 0 & \Delta& -A_{\mathbf{q}-} & v q_z \end{pmatrix}. \tag*{(7)}

Here we have considered the BCS pairing term HBCS=∑qΔc1,q,+12†c4,−q,+,−12†+Δc1,q,−,−32†c4,−q,−,−32†+H.c.H_{\mathrm{BCS}}=\sum_{\mathbf{q}}\Delta c_{1,\mathbf{q},+\frac{1}{2}}^\dagger c_{4,-\mathbf{q},+,-\frac{1}{2}}^\dagger+\Delta c_{1,\mathbf{q},-,-\frac{3}{2}}^\dagger c_{4,-\mathbf{q},-,-\frac{3}{2}}^\dagger+H.c., in which electrons with opposite momentum and angular momentum are paired together.

If we consider the π\pi flux inserted along zz direction, the effective BdG Hamiltonian will become

HBdG(q)=(vqzAq+Δeiϕ0Aq−−vqz0ΔeiϕΔe−iϕ0−vqz−Aq+0Δe−iϕ−Aq−vqz).(8)H_{\mathrm{BdG}}(\mathbf{q})= \begin{pmatrix} v q_z & A_{\mathbf{q}+} & \Delta e^{i\phi} & 0\\ A_{\mathbf{q}-} & -v q_z & 0 & \Delta e^{i\phi}\\ \Delta e^{-i\phi} & 0 & -v q_z & -A_{\mathbf{q}+}\\ 0 & \Delta e^{-i\phi} & -A_{\mathbf{q}-} & v q_z \end{pmatrix}. \tag*{(8)}

in which the phase eiϕe^{i\phi} attached to Δ\Delta describes the effect of the vortex line. Similarly, we can obtain the effective BdG Hamiltonian for SC Weyl semimetals with chirality χ\chi and vorticity nn

HBdG(q)=(vqzAq−χΔe−inϕ0Aq+χ−vqz0ΔeinϕΔe−inϕ0−vqz−Aq−χ0Δe−inϕ−Aq+χvqz),(9)H_{\mathrm{BdG}}(\mathbf{q}) = \begin{pmatrix} vq_z & Aq_-^\chi& \Delta e^{-in\phi} & 0 \\ Aq_+^\chi& -vq_z & 0 & \Delta e^{in\phi} \\ \Delta e^{-in\phi} & 0 & -vq_z & -Aq_-^\chi\\ 0 & \Delta e^{-in\phi} & -Aq_+^\chi& vq_z \end{pmatrix}, \tag*{(9)}

in which qiq_i (i=x,y,zi=x,y,z) is the momentum relative to a Weyl point, and q±=qx±iqy=qe±iθq_{\pm}=q_x\pm iq_y=qe^{\pm i\theta}. If the Hamiltonian changes slowly as we change the parameter ϕ\phi, the three spatial variables qxq_x, qyq_y, and qzq_z together with the fourth variable ϕ\phi constitute a synthetic four-dimensional space [3, 7], in which we can define the second Chern number C2C_2. To calculate C2C_2 of the above effective BdG Hamiltonian, we reformulate Eq. (9) as

HBdG(q)=Aqχcos⁡(χθ)χ3⊗τ1+Aqχsin⁡(χθ)χ3⊗τ2+vqzχ3⊗τ3−Δsin⁡nϕχ2⊗τ0+Δcos⁡nϕχ1⊗τ0.(10)\begin{aligned} H_{\mathrm{BdG}}(\mathbf{q}) ={}& Aq^\chi\cos(\chi\theta)\chi_3\otimes\tau_1 + Aq^\chi\sin(\chi\theta)\chi_3\otimes\tau_2 \\ &+vq_z\chi_3\otimes\tau_3-\Delta\sin n\phi\chi_2\otimes\tau_0+\Delta\cos n\phi\chi_1\otimes\tau_0. \tag*{(10)} \end{aligned}

Then we define d1=Aqχcos⁡(χθ)d_1=Aq^\chi\cos(\chi\theta), d2=Aqχsin⁡(χθ)d_2=Aq^\chi\sin(\chi\theta), d3=vqzd_3=vq_z, d4=−Δsin⁡nϕd_4=-\Delta\sin n\phi, and d5=Δcos⁡nϕd_5=\Delta\cos n\phi. In this way, we can write the effective BdG Hamiltonian as HBdG=∑idi(q,ϕ)⋅ΓiH_{\mathrm{BdG}}=\sum_i d_i(q,\phi)\cdot\Gamma_i. Here Γi\Gamma_i are five mutually anti-commuting matrices in Eq. (10). In this case, the definition of the second Chern number is

C2=38π2∫d3q∫dϕ1d5ϵabcdeda∂qdb∂θdc∂qzdd∂ϕde.(11)C_2=\frac{3}{8\pi^2}\int d^3q\int d\phi\frac{1}{d^5}\epsilon^{abcde}d_a\partial_qd_b\partial_\theta d_c\partial_{q_z}d_d\partial_\phi d_e. \tag*{(11)}

Here we have defined d=(∑idi2)1/2d=(\sum_i d_i^2)^{1/2}. If we substitute the explicit forms of the coefficients did_i into Eq. (11), we will obtain

C2=38π2∫0∞dq∫02πdθ∫−∞∞dqz∫02πdϕ1d5ϵabcdeda∂qdb∂θdc∂qzdd∂ϕde=38π2∫0∞dq∫02πdθ∫−∞∞dqz∫02πdϕ1d5⋅{−A2χ2vΔ2nq2χ−1cos⁡2(nϕ)cos⁡2(χθ)−A2χ2vΔ2nq2χ−1cos⁡2(nϕ)sin⁡2(χθ)−A2χ2vΔ2nq2χ−1sin⁡2(nϕ)cos⁡2(χθ)−A2χ2vΔ2nq2χ−1sin⁡2(nϕ)sin⁡2(χθ)}.(12)\begin{aligned} C_2={}&\frac{3}{8\pi^2}\int_0^\infty dq\int_0^{2\pi}d\theta\int_{-\infty}^{\infty}dq_z\int_0^{2\pi}d\phi\frac{1}{d^5}\epsilon^{abcde}d_a\partial_qd_b\partial_\theta d_c\partial_{q_z}d_d\partial_\phi d_e \\ ={}&\frac{3}{8\pi^2}\int_0^\infty dq\int_0^{2\pi}d\theta\int_{-\infty}^{\infty}dq_z\int_0^{2\pi}d\phi\frac{1}{d^5}\cdot\{-A^2\chi^2v\Delta^2nq^{2\chi-1}\cos^2(n\phi)\cos^2(\chi\theta)\\ &-A^2\chi^2v\Delta^2nq^{2\chi-1}\cos^2(n\phi)\sin^2(\chi\theta)-A^2\chi^2v\Delta^2nq^{2\chi-1}\sin^2(n\phi)\cos^2(\chi\theta)\\ &-A^2\chi^2v\Delta^2nq^{2\chi-1}\sin^2(n\phi)\sin^2(\chi\theta)\}. \tag*{(12)} \end{aligned}

The integrand can be written in a more compact form

C2=−38π2A2χ2vΔ2n∫0∞dq∫02πdθ∫−∞∞dqz∫02πdϕq2χ−1(A2q2χ+v2qz2+Δ2)5/2=−38π2A2χ2vΔ2n2π2π∫−∞∞dqz∫0∞dqq2χ−1(A2q2χ+v2qz2+Δ2)5/2=−38π2χ2vΔ2n2π2π12χ∫−∞∞dqz{−231(x+v2qz2+Δ2)3/2∣x=0x=∞}=−38π2χ2vΔ2n2π2π12χ∫−∞∞dqz231(v2qz2+Δ2)3/2=−n⋅χ.(13)\begin{aligned} C_2={}&-\frac{3}{8\pi^2}A^2\chi^2v\Delta^2n\int_0^\infty dq\int_0^{2\pi}d\theta\int_{-\infty}^{\infty}dq_z\int_0^{2\pi}d\phi\frac{q^{2\chi-1}}{(A^2q^{2\chi}+v^2q_z^2+\Delta^2)^{5/2}}\\ ={}&-\frac{3}{8\pi^2}A^2\chi^2v\Delta^2n2\pi2\pi\int_{-\infty}^{\infty}dq_z\int_0^\infty dq\frac{q^{2\chi-1}}{(A^2q^{2\chi}+v^2q_z^2+\Delta^2)^{5/2}}\\ ={}&-\frac{3}{8\pi^2}\chi^2v\Delta^2n2\pi2\pi\frac{1}{2\chi}\int_{-\infty}^{\infty}dq_z\left\{-\frac{2}{3}\frac{1}{(x+v^2q_z^2+\Delta^2)^{3/2}}\bigg|_{x=0}^{x=\infty}\right\}\\ ={}&-\frac{3}{8\pi^2}\chi^2v\Delta^2n2\pi2\pi\frac{1}{2\chi}\int_{-\infty}^{\infty}dq_z\frac{2}{3}\frac{1}{(v^2q_z^2+\Delta^2)^{3/2}}\\ ={}&-n\cdot\chi. \tag*{(13)} \end{aligned}

This is the expected topological invariant which describes the dependence of gapless vortex modes on the vorticity nn and chirality χ\chi in SC linear and quadratic Weyl semimetals.

Dependence of VBSs on Chirality and Vorticity nn

In this section we will discuss how the chirality χ\chi and vorticity nn affect the numbers and propagating directions of gapless modes in the vortex line of SC Weyl semimetals. Figure S2 and Figure S2 shows the VBSs (E−kzE-k_z relation) of SC linear Weyl semimetal (χ=1\chi=1) with vorticity n=1n=1 and n=−1n=-1, respectively. Figure S2 and Figure S2 shows the VBSs of SC quadratic Weyl semimetal (χ=−2\chi=-2) with vorticity n=1n=1 and n=−2n=-2, respectively. We find that there exist gapless modes in each of the four sub-figures.

Figure S2. The VBSs of SC Weyl semimetals calculated with various chirality χ\chi and vorticity nn. (a) The VBS of SC linear Weyl semimetal (χ=1\chi= 1) are calculated with μ=0.2\mu= 0.2, η=0.3\eta= 0.3, and n=1n = 1. (b) The VBS of SC linear Weyl semimetal (χ=1\chi= 1) are calculated with μ=0.2\mu= 0.2, η=0.3\eta= 0.3, and n=−1n = -1. (c) The VBS of SC quadratic Weyl semimetal (χ=−2\chi= -2) are calculated with μ=0.4\mu= 0.4, η=0.7\eta= 0.7, and n=1n = 1. (d) The VBS of SC quadratic Weyl semimetal (χ=−2\chi= -2) are calculated with μ=0.4\mu= 0.4, η=0.7\eta= 0.7, and n=−2n = -2.

By comparing the VBSs in two figures with same nn (χ\chi) and different χ\chi (nn), we can find how χ\chi (nn) affects the propagating direction and number of gapless modes around the kzk_z component of one Weyl point. It turns out that the number of gapless modes around (the kzk_z component of) each Weyl point is ∣nχ∣|n\chi|, and the propagating direction of the gapless modes around each Weyl point is determined by the sign of the product −nχ-n\chi. For example, there is one gapless mode around each linear Weyl point in Figure S2, while there are two gapless modes around each quadratic Weyl point in Figure S2, which verifies that the number of gapless modes around each Weyl point is proportional to ∣χ∣|\chi|. There are four gapless modes around each quadratic Weyl point in Figure S2, while there are two gapless modes around each Weyl point in Figure S2, which shows that the number of gapless modes around each Weyl point is proportional to ∣n∣|n|. The propagating directions of gapless modes around each linear Weyl point in Figure S2 are opposite to those in Figure S2, showing that the propagating directions of gapless modes are proportional to the sign of nn. We know that in our models the chirality of two inner Weyl points is opposite to that of two outer Weyl points on the kzk_z axis. As a result, in each of the four sub-figures in Figure S2 the propagating direction of the inner two branches of gapless modes is opposite to that of the outer two branches of gapless modes, which shows that the propagating directions of the gapless modes are proportional to the sign of χ\chi.

The Tight-binding Models of Linear and Quadratic Weyl Semimetals

In the main text, we have given the k⋅p\boldsymbol{k} \cdot\boldsymbol{p} model of linear and quadratic Weyl semimetals. However, we use the lattice models rather than the k⋅p\boldsymbol{k} \cdot\boldsymbol{p} models to calculate the vortex bound states. So in this section, we introduce the tight-binding lattice models of linear and quadratic Weyl semimetals. Since Dirac semimetals are well classified [8], we decide to break inversion symmetry in a Dirac semimetal to obtain a Weyl semimetal.

The Lattice Model and kp Model of a Linear Weyl Semimetal

The tight-binding model of the linear Dirac semimetal with TRS, inversion symmetry, and C4zC_4^z rotational symmetry is written under the basis (∣+,12⟩,∣−,32⟩,∣+,−12⟩,∣−,−32⟩)T(|+,\frac{1}{2}\rangle, |-,\frac{3}{2}\rangle, |+,-\frac{1}{2}\rangle, |-,-\frac{3}{2}\rangle)^T, in which each state is the eigenstate of inversion and C4zC_4^z operations labelled by the corresponding eigenvalues. Adding the inversion symmetry breaking term ηsin⁡kzσ3⊗τ3\eta\sin k_z \sigma_3 \otimes\tau_3 into the model of linear Dirac semimetal leads to the tight-binding lattice model of linear Weyl semimetal

HLWSM(k)=(m+2t1Acos⁡kx+2t1Acos⁡ky+2t1Dcos⁡kz)σ0⊗τ3+2t3Asin⁡kxσ3⊗τ1−2t3Asin⁡kyσ0⊗τ2−8t4Csin⁡kz(cos⁡kx−cos⁡ky)σ1⊗τ1+8t4Bsin⁡kxsin⁡kysin⁡kzσ2⊗τ1+ηsin⁡kzσ3⊗τ3.(14)\begin{aligned} H_{\mathrm{LWSM}}(\mathbf{k})={}&(m+2t_{1}^{A}\cos k_x+2t_{1}^{A}\cos k_y+2t_{1}^{D}\cos k_z)\sigma_0\otimes\tau_3\\ &+2t_{3}^{A}\sin k_x\sigma_3\otimes\tau_1-2t_{3}^{A}\sin k_y\sigma_0\otimes\tau_2\\ &-8t_{4}^{C}\sin k_z(\cos k_x-\cos k_y)\sigma_1\otimes\tau_1\\ &+8t_{4}^{B}\sin k_x\sin k_y\sin k_z\sigma_2\otimes\tau_1+\eta\sin k_z\sigma_3\otimes\tau_3. \tag*{(14)} \end{aligned}

Here we consider the cubic lattice. t1At_{1}^{A} and t3At_{3}^{A} are the nearest-neighbor hopping terms in x−yx-y plane, and t1Dt_{1}^{D} is the nearest neighbor hopping term in zz direction. t4Ct_{4}^{C} is the next nearest neighbor hopping term, and t4Bt_{4}^{B} is the third nearest neighbor hopping term. mm and η\eta are the on-site term and inversion symmetry breaking term, respectively. Here σ\sigma and τ\tau represent spin and orbital degrees of freedom, respectively.

If we substitute sin⁡ki\sin k_i with kik_i and substitute cos⁡ki\cos k_i with 1−12ki21-\frac{1}{2}k_i^2 (i=xi=x, yy, and zz), we will obtain the k⋅p\mathbf{k}\cdot\mathbf{p} model of linear Weyl semimetal

HLWSM(k)=(M(k)+ηkzAk+0B∗(k)Ak−−M(k)−ηkzB∗(k)00B(k)M(k)−ηkz−Ak−B(k)0−Ak+−M(k)+ηkz).(15)H_{\mathrm{LWSM}}(\mathbf{k})= \begin{pmatrix} M(\mathbf{k})+\eta k_z & Ak_{+} & 0 & B^{*}(\mathbf{k})\\ Ak_{-} & -M(\mathbf{k})-\eta k_z & B^{*}(\mathbf{k}) & 0\\ 0 & B(\mathbf{k}) & M(\mathbf{k})-\eta k_z & -Ak_{-}\\ B(\mathbf{k}) & 0 & -Ak_{+} & -M(\mathbf{k})+\eta k_z \end{pmatrix}. \tag*{(15)}

Here M(k)=M0−M1kz2−M2(kx2+ky2)M(\mathbf{k})=M_0-M_1k_z^2-M_2(k_x^2+k_y^2), and B(k)=βkz(kx2−ky2)+2iαkxkykzB(\mathbf{k})=\beta k_z(k_x^2-k_y^2)+2i\alpha k_xk_yk_z. The correspondence between parameters in the k⋅p\mathbf{k}\cdot\mathbf{p} model and those in the lattice model is A=2t3AA=2t_{3}^{A}, α=4t4B\alpha=4t_{4}^{B}, β=4t4C\beta=4t_{4}^{C}, M0=m+4t1A+2t1DM_0=m+4t_{1}^{A}+2t_{1}^{D}, M1=t1DM_1=t_{1}^{D}, and M2=t1AM_2=t_{1}^{A}.

Next we show that the low-energy effective Hamiltonian is a Weyl Hamiltonian. The gap closing condition of above Hamiltonian leads to four Weyl points located on kzk_z axis with kz1,2=(−η±η2+4M0M1)/(−2M1)k_{z1,2}=(-\eta\pm\sqrt{\eta^2+4M_0M_1})/(-2M_1) and kz3,4=(η±η2+4M0M1)/(−2M1)k_{z3,4}=(\eta\pm\sqrt{\eta^2+4M_0M_1})/(-2M_1). When we consider the low-energy physics, we omit the higher-order term B(k)B(\mathbf{k}) in the k⋅p\mathbf{k}\cdot\mathbf{p} model of linear Weyl semimetal. Then the Hamiltonian will be decoupled into the 2×22\times2 upper-left block and the 2×22\times2 lower-right block. The Weyl point at k1c=(0,0,kz,1c)\mathbf{k}_{1}^{c}=(0,0,k_{z,1}^{c}) is obtained from the gap closing condition in the upper-left block. So we expand the upper-left block of the Hamiltonian around the Weyl point located at k1c\mathbf{k}_{1}^{c}

HUL(k1c+δk)=((η−2M1kz,1c)δkzA(δkx+iδky)A(δkx−iδky)−(η−2M1kz,1c)δkz)=((η−2M1kz,1c)qzA(qx+iqy)A(qx−iqy)−(η−2M1kz,1c)qz)=(vqzA(qx+iqy)A(qx−iqy)−vqz).(16)\begin{aligned} H_{\mathrm{UL}}(\mathbf{k}_{1}^{c}+\delta\mathbf{k}) &=\begin{pmatrix} (\eta-2M_1k_{z,1}^{c})\delta k_z & A(\delta k_x+i\delta k_y)\\ A(\delta k_x-i\delta k_y) & -(\eta-2M_1k_{z,1}^{c})\delta k_z \end{pmatrix}\\ &=\begin{pmatrix} (\eta-2M_1k_{z,1}^{c})q_z & A(q_x+iq_y)\\ A(q_x-iq_y) & -(\eta-2M_1k_{z,1}^{c})q_z \end{pmatrix}\\ &=\begin{pmatrix} vq_z & A(q_x+iq_y)\\ A(q_x-iq_y) & -vq_z \end{pmatrix}. \tag*{(16)} \end{aligned}

Here we use the notation q=δk\mathbf{q}=\delta\mathbf{k} and v=η−2M1kz,1c=η2+4M0M1v=\eta-2M_1k_{z,1}^{c}=\sqrt{\eta^2+4M_0M_1}. The Weyl point at k4c=(0,0,kz4c)\mathbf{k}_{4}^{c}=(0,0,k_{z4}^{c}) is obtained from the gap closing condition in the lower-right block. Then we expand the lower-right block around k4c\mathbf{k}_{4}^{c}

HLR(k4c+δk)=((−η−2M1kz,4c)δkz−A(δkx−iδky)−A(δkx+iδky)(η+2M1kz,4c)δkz)=((−η−2M1kz,4c)qz−A(qx−iqy)−A(qx+iqy)(η+2M1kz,4c)qz)=(−vqz−A(qx−iqy)−A(qx+iqy)vqz).(17)\begin{aligned} H_{\mathrm{LR}}(\mathbf{k}_{4}^{c}+\delta\mathbf{k}) &=\begin{pmatrix} (-\eta-2M_1k_{z,4}^{c})\delta k_z & -A(\delta k_x-i\delta k_y)\\ -A(\delta k_x+i\delta k_y) & (\eta+2M_1k_{z,4}^{c})\delta k_z \end{pmatrix}\\ &=\begin{pmatrix} (-\eta-2M_1k_{z,4}^{c})q_z & -A(q_x-iq_y)\\ -A(q_x+iq_y) & (\eta+2M_1k_{z,4}^{c})q_z \end{pmatrix}\\ &=\begin{pmatrix} -vq_z & -A(q_x-iq_y)\\ -A(q_x+iq_y) & vq_z \end{pmatrix}. \tag*{(17)} \end{aligned}

Here we use the notation q=δk\mathbf{q}=\delta\mathbf{k} and v=η+2M1kz,4c=η2+4M0M1v=\eta+2M_1k_{z,4}^{c}=\sqrt{\eta^2+4M_0M_1}. Obviously, Eq. (16) and Eq. (17) describe linear Weyl fermions. In fact, the low-energy Hamiltonians around other two Weyl points k2c\mathbf{k}_{2}^{c} and k3c\mathbf{k}_{3}^{c} are also Weyl Hamiltonians. Thus, Eq. (15) is indeed a Hamiltonian describing a Weyl semimetal.

The Lattice Model and k⋅pk \cdot p Model of Quadratic Weyl Semimetal

According to the classification of Dirac semimetals [8], the appearance of a quadratic Dirac point requires the existence of C6zC_{6}^{z} symmetry. Therefore, the lattice model of the quadratic Dirac semimetal is written under the basis (∣+,52⟩,∣−,12⟩,∣+,−52⟩,∣−,−12⟩)T\left(\left|+,\frac{5}{2}\right\rangle,\left|-,\frac{1}{2}\right\rangle,\left|+,-\frac{5}{2}\right\rangle,\left|-,-\frac{1}{2}\right\rangle\right)^{T}, in which each state is the eigenstate of inversion and C6zC_{6}^{z} operations labelled by the corresponding eigenvalues. After considering an inversion symmetry breaking term ηsin⁡kzσ3⊗τ3\eta\sin k_{z}\sigma_{3}\otimes\tau_{3}, we write down the tight-binding lattice model of a quadratic Weyl semimetal

HQWSM(k)={m+2t1A[cos⁡kx+2cos⁡12kxcos⁡32ky]+2t1Dcos⁡kz}σ0⊗τ3−2t4A[sin⁡kx−2sin⁡12kxcos⁡32ky]σ1⊗τ1−2t4B(2cos⁡32kxsin⁡32ky−sin⁡3ky)σ2⊗τ1−4t3Csin⁡kz(cos⁡kx−cos⁡12kxcos⁡32ky)σ3⊗τ1+43t3Csin⁡kzsin⁡12kxsin⁡32kyσ0⊗τ2+ηsin⁡kzσ3⊗τ3.(18)\begin{aligned} H_{\mathrm{QWSM}}(\mathbf{k})={}&\left\{m+2t_{1}^{A}\left[\cos k_{x}+2\cos\frac{1}{2}k_{x}\cos\frac{\sqrt{3}}{2}k_{y}\right]+2t_{1}^{D}\cos k_{z}\right\}\sigma_{0}\otimes\tau_{3}\\ &-2t_{4}^{A}\left[\sin k_{x}-2\sin\frac{1}{2}k_{x}\cos\frac{\sqrt{3}}{2}k_{y}\right]\sigma_{1}\otimes\tau_{1}\\ &-2t_{4}^{B}\left(2\cos\frac{3}{2}k_{x}\sin\frac{\sqrt{3}}{2}k_{y}-\sin\sqrt{3}k_{y}\right)\sigma_{2}\otimes\tau_{1}\\ &-4t_{3}^{C}\sin k_{z}\left(\cos k_{x}-\cos\frac{1}{2}k_{x}\cos\frac{\sqrt{3}}{2}k_{y}\right)\sigma_{3}\otimes\tau_{1}\\ &+4\sqrt{3}t_{3}^{C}\sin k_{z}\sin\frac{1}{2}k_{x}\sin\frac{\sqrt{3}}{2}k_{y}\sigma_{0}\otimes\tau_{2}\\ &+\eta\sin k_{z}\sigma_{3}\otimes\tau_{3}. \tag*{(18)} \end{aligned}

Here we consider the three-dimensional hexagonal lattice. t1At_{1}^{A} and t4At_{4}^{A} are the nearest-neighbor hopping terms in x−yx-y plane, and t1Dt_{1}^{D} is the nearest neighbor hopping term in zz direction. t4Bt_{4}^{B} is the third nearest neighbor hopping term in x−yx-y plane, and t3Ct_{3}^{C} is the next nearest neighbor hopping term. mm and η\eta are the on-site term and inversion symmetry-breaking term, respectively. Here σ\sigma and τ\tau represent spin and orbital degrees of freedom, respectively.

If we substitute sin⁡ki\sin k_{i} with kik_{i} and substitute cos⁡ki\cos k_{i} with 1−12ki21-\frac{1}{2}k_{i}^{2} (i=x,y,i=x,y, and zz), we will obtain the k⋅pk\cdot p model of quadratic Weyl semimetal

HQWSM(k)=(M(k)+ηkzAk+2kz0B∗(k)Ak+2kz−M(k)−ηkzB∗(k)00B(k)M(k)−ηkz−Ak+2kzB(k)0−Ak+2kz−M(k)+ηkz).(19)H_{\mathrm{QWSM}}(\mathbf{k})= \begin{pmatrix} M(\mathbf{k})+\eta k_{z} & Ak_{+}^{2}k_{z} & 0 & B^{*}(\mathbf{k})\\ Ak_{+}^{2}k_{z} & -M(\mathbf{k})-\eta k_{z} & B^{*}(\mathbf{k}) & 0\\ 0 & B(\mathbf{k}) & M(\mathbf{k})-\eta k_{z} & -Ak_{+}^{2}k_{z}\\ B(\mathbf{k}) & 0 & -Ak_{+}^{2}k_{z} & -M(\mathbf{k})+\eta k_{z} \end{pmatrix}. \tag*{(19)}

Here M(k)=M0−M1kz2−M2(kx2+ky2)M(\mathbf{k})=M_{0}-M_{1}k_{z}^{2}-M_{2}(k_{x}^{2}+k_{y}^{2}), and B(k)=αkx(kx2−3ky2)+iβky(3kx2−ky2)B(\mathbf{k})=\alpha k_{x}(k_{x}^{2}-3k_{y}^{2})+i\beta k_{y}(3k_{x}^{2}-k_{y}^{2}). The correspondence between parameters in the k⋅pk\cdot p model and those in the lattice model is A=32t3CA=\frac{3}{2}t_{3}^{C}, α=14t4A\alpha=\frac{1}{4}t_{4}^{A}, β=2t4B(32)3\beta=2t_{4}^{B}\left(\frac{\sqrt{3}}{2}\right)^{3}, M0=m+6t1A+2t1DM_{0}=m+6t_{1}^{A}+2t_{1}^{D}, M1=t1DM_{1}=t_{1}^{D}, and M2=32t1AM_{2}=\frac{3}{2}t_{1}^{A}.

Next we show that the corresponding low-energy effective Hamiltonian is a quadratic Weyl Hamiltonian. The gap closing condition of above Hamiltonian leads to four Weyl points located on kzk_{z} axis with kz1,2c=(−η±η2+4M0M1)/(−2M1)k_{z1,2}^{c}=(-\eta\pm\sqrt{\eta^{2}+4M_{0}M_{1})/(-2M_{1})} and kz3,4c=(η±η2+4M0M1)/(−2M1)k_{z3,4}^{c}=(\eta\pm\sqrt{\eta^{2}+4M_{0}M_{1})/(-2M_{1})}. When we consider the low-energy physics, we omit the higher-order term B(k)B(\mathbf{k}) in the k⋅pk\cdot p model of quadratic Weyl semimetal. Then the Hamiltonian will be decoupled into the 2×22\times2 upper-left block and the 2×22\times2 lower-right block. The Weyl point at k1c=(0,0,kz1c)\mathbf{k}_{1}^{c}=(0,0,k_{z1}^{c}) is obtained from the gap closing condition in the upper-left block. So we expand the upper-left block of the Hamiltonian around the Weyl point located at k1c\mathbf{k}_{1}^{c}

HUL(k1c+δk)=((η−2M1kz,1c)δkzAkz,1c(δkx−iδky)2Akz,1c(δkx+iδky)2−(η−2M1kz,1c)δkz)=((η−2M1kz,1c)qzAkz,1c(qx−iqy)2Akz,1c(qx+iqy)2−(η−2M1kz,1c)qz)=(νqzAkz,1c(qx−iqy)2Akz,1c(qx+iqy)2−νqz).(20)\begin{aligned} H_{\mathrm{UL}}(\mathbf{k}_{1}^{c}+\delta\mathbf{k}) &=\begin{pmatrix} (\eta-2M_{1}k_{z,1}^{c})\delta k_{z} & Ak_{z,1}^{c}(\delta k_{x}-i\delta k_{y})^{2}\\ Ak_{z,1}^{c}(\delta k_{x}+i\delta k_{y})^{2} & -(\eta-2M_{1}k_{z,1}^{c})\delta k_{z} \end{pmatrix}\\ &=\begin{pmatrix} (\eta-2M_{1}k_{z,1}^{c})q_{z} & Ak_{z,1}^{c}(q_{x}-iq_{y})^{2}\\ Ak_{z,1}^{c}(q_{x}+iq_{y})^{2} & -(\eta-2M_{1}k_{z,1}^{c})q_{z} \end{pmatrix}\\ &=\begin{pmatrix} \nu q_{z} & Ak_{z,1}^{c}(q_{x}-iq_{y})^{2}\\ Ak_{z,1}^{c}(q_{x}+iq_{y})^{2} & -\nu q_{z} \end{pmatrix}. \tag*{(20)} \end{aligned}

Here we use the notation q=δk\mathbf{q}=\delta\mathbf{k} and ν=η−2M1kz,1c=η2+4M0M1\nu=\eta-2M_{1}k_{z,1}^{c}=\sqrt{\eta^{2}+4M_{0}M_{1}}. We find that Eq. (20) describes quadratic Weyl fermions. It is not hard to derive the low-energy Hamiltonians around other three Weyl points, and all of them are quadratic Weyl Hamiltonians. Therefore, the Hamiltonian Eq. (19) is the model describing a quadratic Weyl semimetal.

References

References

  1. [1]G. Y. Cho, J. H. Bardarson, Y.-M. Lu, and J. E. Moore, Phys. Rev. B 86, 214514 (2012).
  2. [2]T. Meng and L. Balents, Phys. Rev. B 86, 054504 (2012).
  3. [3]J. C. Y. Teo and C. L. Kane, Phys. Rev. B 82, 115120 (2010).
  4. [4]B. A. Bernevig, in Topological Insulators and Topological Superconductors (Princeton university press, 2013).
  5. [5]X.-L. Qi, T. L. Hughes, and S.-C. Zhang, Phys. Rev. B 78, 195424 (2008).
  6. [6]C. Chan and X.-J. Liu, Phys. Rev. Lett. 118, 207002 (2017).
  7. [7]C. Chan, L. Zhang, T. F. J. Poon, Y.-P. He, Y.-Q. Wang, and X.-J. Liu, Phys. Rev. Lett. 119, 047001 (2017).
  8. [8]B.-J. Yang and N. Nagaosa, Nature Communications 5, 4898 (2014).

Paper details

vlnc:W_vhvdphvb
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 2

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents