This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Double Andreev reflections and double normal reflections in nodal-line semimetal-superconductor junctions

Qiang Cheng School of Science, Qingdao University of Technology, Qingdao, Shandong 266520, China International Center for Quantum Materials, School of Physics, Peking University, Beijing 100871, China    Zhe Hou International Center for Quantum Materials, School of Physics, Peking University, Beijing 100871, China    Qing-Feng Sun sunqf@pku.edu.cn International Center for Quantum Materials, School of Physics, Peking University, Beijing 100871, China Collaborative Innovation Center of Quantum Matter, Beijing 100871, China CAS Center for Excellence in Topological Quantum Computation, University of Chinese Academy of Sciences, Beijing 100190, China
Abstract

We study systematically the scattering processes and the conductance spectra in nodal-line semimetal-superconductor junctions using the extended Blonder-Tinkham-Klapwijk theory. The coexistence of peculiar quadruple reflections are found, which are the specular normal reflection, the retro-normal reflection, the specular Andreev reflection and the retro-Andreev reflection. The incident angle dependence and the quasiparticle energy dependence of the double normal reflections and the double Andreev reflections are investigated under various of values of parameters such as the interfacial barrier height, the chemical potentials and the orbital coupling strength. It is found that the appearance and the disappearance of the reflections and their magnitudes can be controlled through tuning these parameters. The scattering mechanism for the reflections are analyzed in details from the viewpoint of the band structure. We also investigate the conductance spectra for the junctions, which show distinctive features and strong anisotropy about the orientation relationships of the nodal-line and interface. The unique scattering processes and conductance spectra found in the junctions are helpful in designing new superconducting electronic devices and searching for the nodal-line in new materials experimentally.

I Introduction

As three-dimensional gapless materials with point nodes, Dirac and Weyl semimetals have attracted tremendous research interest due to their nontrivially topological properties, connections with particle physics and novel responses to electric and magnetic fields Armitage . In contrast, nodal-line semimetals (NLSMs) are another new type of three-dimensional topological material with band touching along lines in the momentum space Burkov ; Chiu ; Fang ; Wang . It is important that the nodal-lines in NLSM can evolve into Dirac nodes or Weyl nodes when the protecting symmetry is broken Weng ; Fang2 . Recently, NLSMs of a ring-shaped nodal-line and the drumhead surface states become a hot topic in condensed matter physics due to the emerging new physical properties Yan ; Araujo ; Huh ; Gao ; Moors ; Ma ; Ruiz ; Yang ; Oroszlany ; Liu ; Behrends . For example, the quantum oscillations of NLSMs exhibit the phase shift different from Weyl fermions Li . The orbital susceptibility of NLSMs shows a stronger δ\delta-function singularity than that of Dirac and Weyl semimetals with point nodes Koshino .

The distinctive features of NLSMs with a ring-shaped nodal-line also show themselves in the transport aspects Molina ; Chen ; Chen2 ; Barati ; Hao . The transport in the ballistic regime are usually responsible for or can be used to probe the extraordinary phenomena Lv . The anomalous transverse current in NLSM with small inversion breaking can be induced by an electric field, which may be detected by the dumbbell device with a ballistic constriction Khokhlov . The Klein tunneling in the single-particle ballistic scattering of NLSMs can be realized when the incident angle differing from 9090^{\circ} and the transport properties show strong anisotropy related to the orientation of the crystallographic axis Rui . The unexpected nonuniversal conductance fluctuation is found in NLSM, which amplitude rises as the increase of spin-orbit coupling strength Hu .

Nonetheless, researches on scattering processes and conductance in NLSM in contact with a superconductor (SC) are still blank, which may provide specific signatures for the ring-shaped nodal-line. It is well known that the so-called retro-Andreev reflection (RAR) occurs in the ordinary normal metal-SC junctions Andreev . The Andreev reflection dominates the conductance of the junction when the bias is less than the superconducting gap.addsun1 ; addsun2 After the discovery of the two-dimensional gapless material, the specular Andreev reflection (SAR) is possible in the graphene-SC junctions, which dominates the subgap conductance under the weakly doped situation Beenakker ; addref1 ; addsun3 ; addsun4 ; addref2 . Recently, it is proposed that the type-II Weyl semimetal-SC junctions can host double Andreev reflections, both RAR and SAR, due to the band tilt. The evolution of the double Andreev reflections are also studied when the orientation between interface and the band tilt is changed Hou . It should be noted that the normal reflection in the above junctions is always the specular one.

In this paper, we study the transport properties in the NLSM-SC junction with the ring-shaped nodal-line. We find the quadruple reflections can happen simultaneously, which are RAR, SAR, the specular normal reflection (SNR) and the retro-normal reflection (RNR). The probabilities of the double Andreev reflections and the double normal reflections strongly depend on the incident angle and energy of quasiparticles, the interfacial barrier height, the chemical potentials, the orbital coupling strength and the mutual orientation of interface and the nodal-line. The reflection types and their magnitudes can be regulated by changing one of the parameters. The scattering mechanism are analyzed in details from the point of the band structure. The conductance spectra for the nodal-line parallel to and perpendicular to interface exhibit strong anisotropy although they both have the zeros when the quasiparticle energy equals to the chemical potential in NLSM.

The rest of paper is organized as follows. In Sec.II, we demonstrate the model and numerical results for the junction with the nodal-line perpendicular to interface. The Hamiltonians for NLSM and SC and wave functions in the two regions are given. The reflection properties are defined and the expression of conductance is derived. Their dependences on parameters are discussed detailedly. In Sec.III, we present the formalism and results for the junction with nodal-line parallel to interface. Sec.IV concludes this paper.

II Nodal-line perpendicular to interface

II.1 Model and formalism

We consider the NLSM-SC junctions as shown in Fig.1(a), which consist of the semi-infinite NLSM and SC. The interface is located at x=0x=0, which is parallel to the yy-zz plane. The interfacial barrier is modeled by a delta function V(x)=Vδ(x)V(x)=V\delta(x). The transport along the xx direction is considered.

Refer to caption
Figure 1: (a) Schematic illustration of the NLSM-SC junction with the nodal-line perpendicular to interface. (b) The torus-like isoenergetic surface in NLSM. Two concentric isoenergy-circles will be obtained if the surface is cut by a plane parallel to the kxk_{x}-kyk_{y} plane and with a small value of kzk_{z}. (c) The projection of the scattering processes on the kxk_{x}-kyk_{y} plane. The solid arrows denote the group velocity of ELQs (filled circles) and the dashed ones denote the group velocity of HLQs (empty circles). The isoenergetic circles for ELQs (solid lines) and HLQs (dashed lines) coincide with each other due to μN=0\mu_{N}=0. When μN0\mu_{N}\neq 0 and μN<|Ez|\mu_{N}<|E_{z}|, the isoenergetic circles no longer coincide but the scattering possesses the similar processes. (d) The projection of the scattering processes on the kxk_{x}-kyk_{y} plane for μN>|Ez|\mu_{N}>|E_{z}|. The isoenergetic circles for ELQs and HLQs do not coincide. The retro-Andreev reflected HLQs occupy the big isoenergetic circle while the specular Andreev reflected HLQs occupy the small circle, which are different from the situation with μN<|Ez|\mu_{N}<|E_{z}| in (c). (e) The projection of the scattering processes on the xx-yy plane in the real space for μN=0\mu_{N}=0. The gray solid line denotes the NLSM-SC interface and the symbol θ\theta denotes the incident angle of ELQs. The arrows are parallel to those in (c). (f) The projection of the scattering processes on the xx-yy plane in the real space for μN>|Ez|\mu_{N}>|E_{z}|. The arrows are parallel to those in (d).

We consider the two-orbit effective Hamiltonian for NLSM, which is given by Xie ; Chan

H^N(𝒌)=ϵkσ^zvkzσ^yμNσ^0,\hat{H}_{N}(\bm{k})=\epsilon_{k}\hat{\sigma}_{z}-\hbar vk_{z}\hat{\sigma}_{y}-\mu_{N}\hat{\sigma}_{0}, (1)

with ϵk=22m(kx2+ky2+kz2)E0\epsilon_{k}=\frac{\hbar^{2}}{2m}(k_{x}^{2}+k_{y}^{2}+k_{z}^{2})-E_{0}. Here, the identity matrix σ^0\hat{\sigma}_{0} and the Pauli matrices σ^y\hat{\sigma}_{y} and σ^z\hat{\sigma}_{z} are defined in the orbit space. The three component wave vector is 𝒌=(kx,ky,kz)\bm{k}=(k_{x},k_{y},k_{z}) and μN\mu_{N} is the chemical potential. For μN=0\mu_{N}=0, the Hamiltonian describes the NLSM with a ring-shaped nodal-line in the plane kz=0k_{z}=0, which radius is given by k=2mE0/2k=\sqrt{{2mE_{0}}/{\hbar^{2}}}. For μN0\mu_{N}\neq 0, the nodal-line will evolve into a torus-shaped Fermi surface as shown in Fig.1(b). Since the nodal-line is in the plane kz=0k_{z}=0, the transport properties in the zz direction is different from that in the xx and yy directions, i.e. the NLSM with the Hamiltonian in Eq.(1) has the strong anisotropy. Here we consider the transport along the xx direction with the nodal-line perpendicular to the NLSM-SC interface at x=0x=0 as shown in Fig.1(a). The transport along the zz direction with the nodal-line parallel to the NLSM-SC interface is studied in Sec.III.

The Bogoliubov-de Gennes (BdG) Hamiltonian for NLSM can be written as

HˇN(𝒌)=(H^N(𝒌)00H^N(𝒌)),\check{H}_{N}(\bm{k})=\left(\begin{array}[]{cc}\hat{H}_{N}(\bm{k})&0\\ 0&-\hat{H}^{*}_{N}(-\bm{k})\end{array}\right), (2)

in the orbit\otimesparticle-hole space. Through solving the BdG equation HˇN(i)ψN=EψN\check{H}_{N}(-i\nabla)\psi_{N}=E\psi_{N} with the substitution of i-i\nabla for 𝒌\bm{k} in HˇN(𝒌)\check{H}_{N}(\bm{k}), the energy dispersions for the electron-like quasiparticles (ELQs) and the hole-like quasiparticles (HLQs) can be obtained, which are

Ee±=±ϵk2+Ez2μN,E_{e}^{\pm}=\pm\sqrt{\epsilon_{k}^{2}+E_{z}^{2}}-\mu_{N}, (3)

and

Eh±=±ϵk2+Ez2+μN,E_{h}^{\pm}=\pm\sqrt{\epsilon_{k}^{2}+E_{z}^{2}}+\mu_{N}, (4)

with Ez=vkzE_{z}=\hbar vk_{z} characterizing the orbital coupling strength. From the dispersions, it is found that the gap 2|Ez|2|E_{z}| situated at the nodal-line is opened in the quasiparticle spectrum. The gap spans from |Ez|μN-|E_{z}|-\mu_{N} to |Ez|μN|E_{z}|-\mu_{N} for ELQs. For HLQs, the gap spans from |Ez|+μN-|E_{z}|+\mu_{N} to |Ez|+μN|E_{z}|+\mu_{N}.

For μN=0\mu_{N}=0, the two gaps are both symmetric about the zero energy, which span from |Ez|-|E_{z}| to |Ez||E_{z}|. When the energy EE of the incident ELQs is larger than |Ez||E_{z}|, the conduction band Ee+E_{e}^{+} and the valence band Eh+E_{h}^{+} participate in the scattering processes. The isoenergetic circles with E>|Ez|E>|E_{z}| in the kxk_{x}-kyk_{y} plane are shown in Fig.1(c). The isoenergy-circles for ELQs and for HLQs coincide in this situation. Considering that the incident energy EE and the wave vectors kyk_{y} and kzk_{z} (i.e. EzE_{z}) are conserved in the scattering processes, there are two types of ELQs (HLQs): one occupies the big circle with 𝒌=(kxe(h)+,ky,kz)\bm{k}=(k_{x}^{e(h)+},k_{y},k_{z}) and the other occupies the small circle with 𝒌=(kxe(h),ky,kz)\bm{k}=(k_{x}^{e(h)-},k_{y},k_{z}), in which kxe,h±=2m2(E0±E22v2kz2)ky2kz2k_{x}^{e,h\pm}=\sqrt{\frac{2m}{\hbar^{2}}(E_{0}\pm\sqrt{E^{2}-\hbar^{2}v^{2}k_{z}^{2}})-k_{y}^{2}-k_{z}^{2}}.

If we denote the group velocity as 𝒗=(vx,vy,vz)\bm{v}=(v_{x},v_{y},v_{z}), it can be found that (vx,vy,vz)(v_{x},v_{y},v_{z}) have the same sign with (kxe(h)+,ky,kz)(k_{x}^{e(h)+},k_{y},k_{z}) and (kxe(h),ky,kz)(-k_{x}^{e(h)-},-k_{y},k_{z}). For the injection of ELQs from NLSM, there are two types of the incident ELQs with the wave vectors being 𝒌=(kxe+,ky,kz)\bm{k}=(k_{x}^{e+},k_{y},k_{z}) and 𝒌=(kxe,ky,kz)\bm{k}=(-k_{x}^{e-},k_{y},k_{z}). However, when a beam of ELQs are injected from NLSM, regardless of their wave vector being kxe+k_{x}^{e+} or kxe-k_{x}^{e-}, there will always be four reflection processes (see Figs.1(c) and (e)): double Andreev reflections (SAR and RAR) with the wave vector 𝒌=(kxh+,ky,kz)\bm{k}=(-k_{x}^{h+},k_{y},k_{z}) and (kxh,ky,kz)(k_{x}^{h-},k_{y},k_{z}) and double normal reflections (SNR and RNR) with 𝒌=(kxe+,ky,kz)\bm{k}=(-k_{x}^{e+},k_{y},k_{z}) and (kxe,ky,kz)(k_{x}^{e-},k_{y},k_{z}).

Taking the injection of an ELQ with energy EE and the wave vector kxe+k_{x}^{e+} as an example, the resulting wave function in NLSM can be solved from the BdG equation and written as

ψN(x<0)=(iχ1100)eikxe+x+rn1(iχ1100)eikxe+x+rn2(iχ2100)eikxex+ra1(00iχ21)eikxh+x+ra2(00iχ11)eikxhx,\begin{split}\psi_{N}(x<0)=&\left(\begin{array}[]{c}i\chi_{1}\\ 1\\ 0\\ 0\end{array}\right)e^{ik_{x}^{e+}x}+r_{n1}\left(\begin{array}[]{c}i\chi_{1}\\ 1\\ 0\\ 0\end{array}\right)e^{-ik_{x}^{e+}x}\\ +r_{n2}&\left(\begin{array}[]{c}i\chi_{2}\\ 1\\ 0\\ 0\end{array}\right)e^{ik_{x}^{e-}x}+r_{a1}\left(\begin{array}[]{c}0\\ 0\\ -i\chi_{2}\\ 1\end{array}\right)e^{-ik_{x}^{h+}x}\\ +r_{a2}&\left(\begin{array}[]{c}0\\ 0\\ -i\chi_{1}\\ 1\end{array}\right)e^{ik_{x}^{h-}x},\end{split} (5)

with χ1=(E+ΩN)/Ez\chi_{1}=(E+\Omega_{N})/E_{z}, χ2=(EΩN)/Ez\chi_{2}=(E-\Omega_{N})/E_{z} and ΩN=E2Ez2\Omega_{N}=\sqrt{E^{2}-E_{z}^{2}}. The symbols rn1,rn2,ra1r_{n1},r_{n2},r_{a1} and ra2r_{a2} represent reflection amplitudes for SNR, RNR, SAR and RAR, respectively.

For μN0\mu_{N}\neq 0, neither of the two gaps is symmetric about E=0E=0. We discuss this situation in two aspects. The first aspect is μN<|Ez|\mu_{N}<|E_{z}|. In this case, the gap for ELQs spans from |Ez|μN<0-|E_{z}|-\mu_{N}<0 to |Ez|μN>0|E_{z}|-\mu_{N}>0, while that for HLQs spans from |Ez|+μN<0-|E_{z}|+\mu_{N}<0 to |Ez|+μN>0|E_{z}|+\mu_{N}>0. The isoenergetic circles for ELQs and HLQs split and the big circle for ELQs (HLQs) becomes bigger (smaller) and the small circle for ELQs (HLQs) becomes smaller (bigger). When E>|Ez|+μNE>|E_{z}|+\mu_{N}, the bands involved in the scattering processes are still Ee+E_{e}^{+} and Eh+E_{h}^{+} and the reflections are similar to those given in Figs.1(c) and (e). The specular Andreev reflected HLQs still occupy the big circle and the retro-Andreev reflected HLQs still occupy the small circle.

Consider also the injection of an ELQ with energy EE and the wave vector kxe+k_{x}^{e+}, the resulting wave function in NLSM is

ψN(x<0)=(iχ11100)eikxe+x+rn1(iχ11100)eikxe+x+rn2(iχ12100)eikxex+ra1(00iχ221)eikxh+x+ra2(00iχ211)eikxhx,\begin{split}\psi_{N}(x<0)=&\left(\begin{array}[]{c}i\chi_{11}\\ 1\\ 0\\ 0\end{array}\right)e^{ik_{x}^{e+}x}+r_{n1}\left(\begin{array}[]{c}i\chi_{11}\\ 1\\ 0\\ 0\end{array}\right)e^{-ik_{x}^{e+}x}\\ +r_{n2}&\left(\begin{array}[]{c}i\chi_{12}\\ 1\\ 0\\ 0\end{array}\right)e^{ik_{x}^{e-}x}+r_{a1}\left(\begin{array}[]{c}0\\ 0\\ -i\chi_{22}\\ 1\end{array}\right)e^{-ik_{x}^{h+}x}\\ +r_{a2}&\left(\begin{array}[]{c}0\\ 0\\ -i\chi_{21}\\ 1\end{array}\right)e^{ik_{x}^{h-}x},\end{split} (6)

with χ11(12)=(E+μN+()ΩN+)/Ez\chi_{11(12)}=(E+\mu_{N}+(-)\Omega_{N}^{+})/E_{z}, χ21(22)=(EμN+()ΩN)/Ez\chi_{21(22)}=(E-\mu_{N}+(-)\Omega_{N}^{-})/E_{z}, kxe±=2m2(E0±ΩN+)ky2kz2k_{x}^{e\pm}=\sqrt{\frac{2m}{\hbar^{2}}(E_{0}\pm\Omega_{N}^{+})-k_{y}^{2}-k_{z}^{2}}, kxh±=2m2(E0±ΩN)ky2kz2k_{x}^{h\pm}=\sqrt{\frac{2m}{\hbar^{2}}(E_{0}\pm\Omega_{N}^{-})-k_{y}^{2}-k_{z}^{2}} and ΩN±=(E±μN)2Ez2\Omega_{N}^{\pm}=\sqrt{(E\pm\mu_{N})^{2}-E_{z}^{2}}. The coefficients rn1r_{n1}, rn2r_{n2}, ra1r_{a1} and ra2r_{a2} have the same meanings as those in Eq.(5). As μN0\mu_{N}\rightarrow 0, the wave function in Eq.(6) degenerates into that in Eq.(5).

The second aspect for μN0\mu_{N}\neq 0 is μN>|Ez|\mu_{N}>|E_{z}|. The gap for ELQs spans from |Ez|μN<0-|E_{z}|-\mu_{N}<0 to |Ez|μN<0|E_{z}|-\mu_{N}<0 and that for HLQs spans from μN|Ez|>0\mu_{N}-|E_{z}|>0 to μN+|Ez|>0\mu_{N}+|E_{z}|>0. Therefore, when 0<E<μN|Ez|0<E<\mu_{N}-|E_{z}|, the conduction bands Ee+E_{e}^{+} and EhE_{h}^{-} participate in the scattering processes. In this situation, the isoenergetic circles for ELQs and HLQs are shown in Fig.1(d). The reflections possess a little difference from those with μN<|Ez|\mu_{N}<|E_{z}|. The incident ELQs will be specular Andreev reflected as HLQs with the wave vector (kxh,ky,kz)(-k_{x}^{h-},k_{y},k_{z}) on the small circle and will be retro-Andreev reflected as HLQs with the wave vector (kxh+,ky,kz)(k_{x}^{h+},k_{y},k_{z}) on the big circle (see Figs.1(d) and (f)). The wave function in NLSM is

ψN(x<0)=(iχ11100)eikxe+x+rn1(iχ11100)eikxe+x+rn2(iχ12100)eikxex+ra1(00iχ211)eikxhx+ra2(00iχ221)eikxh+x,\begin{split}\psi_{N}(x<0)=&\left(\begin{array}[]{c}i\chi_{11}\\ 1\\ 0\\ 0\end{array}\right)e^{ik_{x}^{e+}x}+r_{n1}\left(\begin{array}[]{c}i\chi_{11}\\ 1\\ 0\\ 0\end{array}\right)e^{-ik_{x}^{e+}x}\\ +r_{n2}&\left(\begin{array}[]{c}i\chi_{12}\\ 1\\ 0\\ 0\end{array}\right)e^{ik_{x}^{e-}x}+r_{a1}\left(\begin{array}[]{c}0\\ 0\\ -i\chi_{21}\\ 1\end{array}\right)e^{-ik_{x}^{h-}x}\\ +r_{a2}&\left(\begin{array}[]{c}0\\ 0\\ -i\chi_{22}\\ 1\end{array}\right)e^{ik_{x}^{h+}x},\end{split} (7)

where χ11\chi_{11}, χ12\chi_{12}, χ21\chi_{21}, χ22\chi_{22}, kxe±k_{x}^{e\pm} and kxh±k_{x}^{h\pm} are the same as those in Eq.(6). Now, if one increases the energy of the incident ELQs to E>μN+|Ez|E>\mu_{N}+|E_{z}|, the involved bands in the scattering processes become again the conduction band Ee+E_{e}^{+} and the valence band Eh+E_{h}^{+}. In this case, the wave function in NLSM is still given by Eq.(6).

The superconductivity in NLSM can be induced by its high-quality contact with SC. In this paper, we consider the spin-singlet ss-wave pairing realized in the superconducting NLSM. The BdG Hamiltonian for the SC formed in NLSM at the region of x>0x>0 is

HˇS(𝒌)=(H^S(𝒌)Δ^Δ^H^S(𝒌)),\check{H}_{S}(\bm{k})=\left(\begin{array}[]{cc}\hat{H}_{S}(\bm{k})&\hat{\Delta}\\ \hat{\Delta}&-\hat{H}^{*}_{S}(-\bm{k})\end{array}\right), (8)

in which H^S=ϵkσ^zvkzσ^yμSσ^0\hat{H}_{S}=\epsilon_{k}\hat{\sigma}_{z}-\hbar vk_{z}\hat{\sigma}_{y}-\mu_{S}\hat{\sigma}_{0} and Δ^=Δσ^0\hat{\Delta}=\Delta\hat{\sigma}_{0} with μS\mu_{S} the chemical potential in SC and Δ\Delta the superconducting gap magnitude.

By solving the BdG equation for SC, the wave function in the superconducting region can be written as

ψS(x>0)=t1(iuη11uiη111)eipx+x+t2(iuη12uiη121)eipxx+t3(ivη21viη211)eiqx+x+t4(ivη22viη221)eiqxx,\begin{split}\psi_{S}(x>0)=&t_{1}\left(\begin{array}[]{c}iu\eta_{11}\\ u\\ i\eta_{11}\\ 1\end{array}\right)e^{ip_{x}^{+}x}+t_{2}\left(\begin{array}[]{c}iu\eta_{12}\\ u\\ i\eta_{12}\\ 1\end{array}\right)e^{-ip_{x}^{-}x}\\ +&t_{3}\left(\begin{array}[]{c}iv\eta_{21}\\ v\\ i\eta_{21}\\ 1\end{array}\right)e^{-iq_{x}^{+}x}+t_{4}\left(\begin{array}[]{c}iv\eta_{22}\\ v\\ i\eta_{22}\\ 1\end{array}\right)e^{iq_{x}^{-}x},\end{split} (9)

where u=(E+ΩS)/Δu=(E+\Omega_{S})/\Delta, v=(EΩS)/Δv=(E-\Omega_{S})/\Delta, η11(12)=(μS+ΩS+()ΩS+)/Ez\eta_{11(12)}=(\mu_{S}+\Omega_{S}+(-)\Omega_{S}^{+})/E_{z}, η21(22)=(μSΩS+()ΩS)/Ez\eta_{21(22)}=(\mu_{S}-\Omega_{S}+(-)\Omega_{S}^{-})/E_{z} with ΩS=E2Δ2\Omega_{S}=\sqrt{E^{2}-\Delta^{2}} and ΩS±=(ΩS±μS)2Ez2\Omega_{S}^{\pm}=\sqrt{(\Omega_{S}\pm\mu_{S})^{2}-E_{z}^{2}}. The wave vector px±p_{x}^{\pm} and qx±q_{x}^{\pm} are given by px±=2m2(E0±ΩS+)ky2kz2p_{x}^{\pm}=\sqrt{\frac{2m}{\hbar^{2}}(E_{0}\pm\Omega_{S}^{+})-k_{y}^{2}-k_{z}^{2}} and qx±=2m2(E0±ΩS)ky2kz2q_{x}^{\pm}=\sqrt{\frac{2m}{\hbar^{2}}(E_{0}\pm\Omega_{S}^{-})-k_{y}^{2}-k_{z}^{2}}. The symbols t1t_{1}, t2t_{2}, t3t_{3} and t4t_{4} denote the transmission amplitudes of quasiparticles.

The reflection and transmission amplitudes can be obtained using the boundary conditions at x=0x=0, which are

ψN(x=0)=ψS(x=0+),ψS(x=0+)ψN(x=0)=2mV2MˇψN(x=0),\begin{split}\psi_{N}(x=0^{-})&=\psi_{S}(x=0^{+}),\\ \psi^{{}^{\prime}}_{S}(x=0^{+})&-\psi^{{}^{\prime}}_{N}(x=0^{-})=\frac{2mV}{\hbar^{2}}\check{M}\psi_{N}(x=0),\end{split} (10)

in which Mˇ\check{M} is a 4×44\times 4 diagonal matrix with the diagonal elements (1,1,1,1)(1,-1,1,-1).

The probabilities for SNR and RNR can be defined as

Rn1(E)=|rn1|2,Rn2(E)=Re[kxekxe+]|rn2|2|χ1221χ1121|.\begin{split}R_{n1}(E)&=|r_{n1}|^{2},\\ R_{n2}(E)&=\text{Re}\Big{[}\frac{k_{x}^{e-}}{k_{x}^{e+}}\Big{]}|r_{n2}|^{2}\Big{|}\frac{\chi_{12}^{2}-1}{\chi_{11}^{2}-1}\Big{|}.\end{split} (11)

The probabilities for SAR and RAR can be defined as

Ra1(E)=Re[kxh+kxe+]|ra1|2|χ2221χ1121|,Ra2(E)=Re[kxhkxe+]|ra2|2|χ2121χ1121|,\begin{split}R_{a1}(E)&=\text{Re}\Big{[}\frac{k_{x}^{h+}}{k_{x}^{e+}}\Big{]}|r_{a1}|^{2}\Big{|}\frac{\chi_{22}^{2}-1}{\chi_{11}^{2}-1}\Big{|},\\ R_{a2}(E)&=\text{Re}\Big{[}\frac{k_{x}^{h-}}{k_{x}^{e+}}\Big{]}|r_{a2}|^{2}\Big{|}\frac{\chi_{21}^{2}-1}{\chi_{11}^{2}-1}\Big{|},\end{split} (12)

for μN<|Ez|\mu_{N}<|E_{z}| or μN>|Ez|\mu_{N}>|E_{z}| and E>μN+|Ez|E>\mu_{N}+|E_{z}|, and

Ra1(E)=Re[kxhkxe+]|ra1|2|χ2121χ1121|,Ra2(E)=Re[kxh+kxe+]|ra2|2|χ2221χ1121|,\begin{split}R_{a1}(E)&=\text{Re}\Big{[}\frac{k_{x}^{h-}}{k_{x}^{e+}}\Big{]}|r_{a1}|^{2}\Big{|}\frac{\chi_{21}^{2}-1}{\chi_{11}^{2}-1}\Big{|},\\ R_{a2}(E)&=\text{Re}\Big{[}\frac{k_{x}^{h+}}{k_{x}^{e+}}\Big{]}|r_{a2}|^{2}\Big{|}\frac{\chi_{22}^{2}-1}{\chi_{11}^{2}-1}\Big{|},\end{split} (13)

for μN>|Ez|\mu_{N}>|E_{z}| and E<μN|Ez|E<\mu_{N}-|E_{z}|.

The reflection amplitudes r~n1\tilde{r}_{n1}, r~n2\tilde{r}_{n2}, r~a1\tilde{r}_{a1}, r~a2\tilde{r}_{a2} and the corresponding probabilities R~n1\tilde{R}_{n1}, R~n2\tilde{R}_{n2}, R~a1\tilde{R}_{a1}, R~a2\tilde{R}_{a2} for the injection of an ELQ with the wave vector kxe-k_{x}^{e-} can also be calculated by using the same way. The defined probabilities satisfy the conservation conditions,

Rn1+Rn2+Ra1+Ra2=1,R~n1+R~n2+R~a1+R~a2=1,\begin{split}R_{n1}+R_{n2}+R_{a1}+R_{a2}=1,\\ \tilde{R}_{n1}+\tilde{R}_{n2}+\tilde{R}_{a1}+\tilde{R}_{a2}=1,\end{split} (14)

inside the gap, i.e. E<ΔE<\Delta, due to the absence of quasiparticle transmissions.

According to the Blonder-Tinkham-Klapwijk formalism Blonder , the conductance σ\sigma can be expressed as

σ=2e2hS(2π)22mE02vπ2π2eVbeVb(σ1+σ2)cosθdθdEz,\sigma=\frac{2e^{2}}{h}\frac{S}{(2\pi)^{2}}\frac{\sqrt{2mE_{0}}}{\hbar^{2}v}{\int_{-\frac{\pi}{2}}^{\frac{\pi}{2}}\int_{-eV_{b}}^{eV_{b}}(\sigma_{1}+\sigma_{2})\cos{\theta}d\theta dE_{z}}, (15)

in which σ1(eVb)=1+Ra1(eVb)+Ra2(eVb)Rn1(eVb)Rn2(eVb)\sigma_{1}(eV_{b})=1+R_{a1}(eV_{b})+R_{a2}(eV_{b})-R_{n1}(eV_{b})-R_{n2}(eV_{b}), σ2(eVb)=1+R~a1(eVb)+R~a2(eVb)R~n1(eVb)R~n2(eVb)\sigma_{2}(eV_{b})=1+\tilde{R}_{a1}(eV_{b})+\tilde{R}_{a2}(eV_{b})-\tilde{R}_{n1}(eV_{b})-\tilde{R}_{n2}(eV_{b}), VbV_{b} is the bias on the NLSM-SC junction, SS is the cross-sectional area of the junction and θ\theta is the incident angle in the xx-yy plane for ELQs (see Figs.1(e) and (f)). The factor 22 in 2e2/h2e^{2}/h accounts for the spin degeneracy of energy bands in NLSM. The reflection probabilities are even functions of θ\theta and EzE_{z}, so do σ1\sigma_{1} and σ2\sigma_{2}. The normalized conductance can be defined by σ/σ0\sigma/\sigma_{0} with σ0\sigma_{0} the conductance of the NLSM-NLSM junction.

II.2 Results and discussions

We define the wave vector k0=2mΔ/2k_{0}=\sqrt{2m\Delta/\hbar^{2}} and the effective barrier height can be given by V0=2mV2k0V_{0}=\frac{2mV}{\hbar^{2}k_{0}}. Throughout the calculations, the energy E0E_{0} characterizing the size of the nodal-line is taken as 200Δ200\Delta. We also take μN=μS\mu_{N}=\mu_{S} which will eliminate the wave vector mismatch caused by the different chemical potentials in NLSM and SC. If μSμN\mu_{S}\not=\mu_{N} and μS\mu_{S} much larger than Δ\Delta, only the probabilities of the reflections are slightly affected, but double normal-reflections and double Andreev reflections are always present. For clarity, only the numerical results for the scattering processes associated with the incident ELQs having the wave vector kxe+k_{x}^{e+} are presented and discussed. The results for the incident ELQs with the wave vector kxe-k_{x}^{e-} can be obtained and analyzed in a similar way. Next, we will discuss the incident angle dependence and the energy dependence of the reflection probabilities and the conductance spectra in details.

II.2.1 The incident angle dependence of reflections

Refer to caption
Figure 2: The θ\theta dependence of reflection probabilities for μN=μS=0.5Δ\mu_{N}=\mu_{S}=0.5\Delta, Ez=0.3ΔE_{z}=0.3\Delta and V0=10V_{0}=10 with (a) the incident energy E=0.1E=0.1, (b) E=0.5ΔE=0.5\Delta, (c) E=0.9ΔE=0.9\Delta, and (d) E=1.2ΔE=1.2\Delta.

First, we present the reflection probabilities as functions of the incident angle θ\theta, which are always symmetric about θ=0\theta=0. Fig.2 shows the results for different incident energy EE at Ez=0.3ΔE_{z}=0.3\Delta and μN=μS=0.5Δ\mu_{N}=\mu_{S}=0.5\Delta. In this instance, the band gap for ELQs spans from 0.8Δ-0.8\Delta to 0.2Δ-0.2\Delta and that for HLQs spans from 0.2Δ0.2\Delta to 0.8Δ0.8\Delta. When 0<E<0.2Δ0<E<0.2\Delta, ELQs on the conduction band Ee+E_{e}^{+} and HLQs on the conduction band EhE_{h}^{-} participate in the scattering processes which are schematically presented in Figs.1(d) and (f). Double Andreev reflections and double normal reflections exist simultaneously. The four reflection probabilities are all non-zero as shown in Fig.2(a) when E=0.1ΔE=0.1\Delta. As θ\theta increases from 0 to π/2\pi/2, SNR Rn1R_{n1} is enhanced while RNR Rn2R_{n2} and SAR Ra1R_{a1} are weakened. As for RAR Ra2R_{a2}, it exhibits a small oscillation. The curve “sum” gives the summation of the reflection probabilities. Its value is always 11 for E<ΔE<\Delta according to the conservation relations in Eq.(14).

When 0.2Δ<E<0.8Δ0.2\Delta<E<0.8\Delta, only ELQs on the conduction band Ee+E_{e}^{+} take part in the scattering processes. For the incident energy E=0.5ΔE=0.5\Delta in Fig.2 (b), the energy lying in the gap for HLQs leads to the vanishing SAR and RAR and Ra1=Ra2=0R_{a1}=R_{a2}=0 exactly. Double normal reflections still exist as two inverse phase oscillating curves. SNR Rn1R_{n1} and RNR Rn2R_{n2} reach their valley value and peak value at θ=0\theta=0, respectively. When E>0.8ΔE>0.8\Delta, HLQs on the valence band Eh+E_{h}^{+} will be activated and participate in the scattering processes which are similar to Figs.1(c) and (e). Double Andreev reflections and double normal reflections reappear as shown in Fig.2(c). In the four reflections, RAR Ra2R_{a2} becomes the dominant process in a large angle range around θ=0\theta=0. If one continuously increases the value of EE until E>ΔE>\Delta as given in Fig.2 (d) with E=1.2ΔE=1.2\Delta, the reflections will be weakened since quasiparticle transmissions begin to happen. The conservation relations in Eq.(14) should be revised. In this case, the summation of the reflection and transmission probabilities equals to 11. From the above results, it is found that the reflection styles and magnitudes can be controlled by adjusting the incident quasiparticle energy.

Refer to caption
Figure 3: The θ\theta dependence of reflection probabilities for E=0.6ΔE=0.6\Delta, Ez=0.3ΔE_{z}=0.3\Delta and V0=10V_{0}=10 with (a) μN=μS=0\mu_{N}=\mu_{S}=0, (b) μN=μS=0.2Δ\mu_{N}=\mu_{S}=0.2\Delta, (c) μN=μS=0.4Δ\mu_{N}=\mu_{S}=0.4\Delta, and (d) μN=μS=1.4Δ\mu_{N}=\mu_{S}=1.4\Delta.

Fig.3 shows the reflection probabilities as functions of the incident angle θ\theta for different chemical potentials μN/S\mu_{N/S}. The incident quasiparticle energy is taken as E=0.6ΔE=0.6\Delta and EzE_{z} is fixed at 0.3Δ0.3\Delta. For μN=μS=0\mu_{N}=\mu_{S}=0, ELQs and HLQs in NLSM have the same energy dispersions. The isoenergetic circles for ELQs and HLQs in the kxk_{x}-kyk_{y} plane coincide with each other as presented in Fig.1 (c). The energy gap is symmetric about E=0E=0. The ELQs on the conduction band Ee+E_{e}^{+} and HLQs on the valence band Eh+E_{h}^{+} participate in the scattering processes which are shown in Figs.1(c) and (e). When E<EzE<E_{z}, all reflections disappear due to the absence of the incident states in the gap. But, as long as E>EzE>E_{z}, there will be double Andreev reflections and double normal reflections (see Fig.3 (a)). The variations of SNR Rn1R_{n1} and RAR Ra2R_{a2} are complementary. That is to say, SNR will acquire its maximum (minimum) value if RAR gets its minimum (maximum) value. Actually, this is an universal character when four types of reflections coexist. RNR Rn2R_{n2} exhibits oscillating behaviour and obtains its peak value near θ=π/2\theta=\pi/2. In contrast, the magnitude of SAR Ra1R_{a1} is higher near θ=0\theta=0.

For μN=μS>0\mu_{N}=\mu_{S}>0, the energy bands for ELQs and HLQs split. The energy for ELQs is lowered and that for HLQs is raised. The energy gap is asymmetric about E=0E=0. When μN=μS=0.2Δ\mu_{N}=\mu_{S}=0.2\Delta, the gap for ELQs spans from 0.5Δ-0.5\Delta to 0.1Δ0.1\Delta and that for HLQs spans from 0.1Δ-0.1\Delta to 0.5Δ0.5\Delta. The bands involved in the scattering processes are still the conduction band Ee+E_{e}^{+} and valence band Eh+E_{h}^{+} since μN<Ez\mu_{N}<E_{z}. The four reflections are presented in Fig.3(b). Comparing with Fig.3(a), the changes of double Andreev reflections are not obvious while SNR and RNR are slightly elevated and reduced, respectively. The probability for SAR Ra1R_{a1} is obviously larger than RNR Rn2R_{n2} in a wide angle range around θ=0\theta=0.

For μN=μS=0.4Δ\mu_{N}=\mu_{S}=0.4\Delta, the gap for ELQs is from 0.7Δ-0.7\Delta to 0.1Δ-0.1\Delta and that for HLQs is from 0.1Δ0.1\Delta to 0.7Δ0.7\Delta. The incident energy E=0.6ΔE=0.6\Delta is just in the gap of HLQs and crosses the conduction band Ee+E_{e}^{+} of ELQs. Therefore, only ELQs will be reflected and the Andreev reflections vanish with Ra1=Ra2=0R_{a1}=R_{a2}=0 exactly. The probabilities of the double normal reflections are given in Fig.3(c), which exhibit inverse phase oscillations. When we increase the chemical potentials, the lower edge of the HLQs gap moves up while the upper edge of the ELQs gap moves down. If μN=μS=0.9Δ\mu_{N}=\mu_{S}=0.9\Delta, the energy E=0.6ΔE=0.6\Delta still crosses the conduction band Ee+E_{e}^{+} and at the same time it is tangent to the lower edge of the HLQs gap. For the lager values of the chemical potentials such as μN=μS=1.4Δ\mu_{N}=\mu_{S}=1.4\Delta in Fig.3(d), the energy E=0.6ΔE=0.6\Delta will simultaneously intersect the conduction bands Ee+E_{e}^{+} and EhE_{h}^{-} and double Andreev reflections reoccur. In this case, both the incident ELQs and outgoing HLQs are in the conduction bands. Here RNR Rn2R_{n2} reaches its peak value at θ=0\theta=0 and RAR Ra2R_{a2} is suppressed slightly around θ=0\theta=0 comparing to Figs.3(a) and (b). From the above analyses, it is found that the positions of band gaps and hence the reflection processes in NLSM-SC junctions can be tuned by the chemical potential which can be controlled by an gate voltage.

Refer to caption
Figure 4: The θ\theta dependence of reflection probabilities for E=0.7ΔE=0.7\Delta, μN=μS=0.2Δ\mu_{N}=\mu_{S}=0.2\Delta and Ez=0.4ΔE_{z}=0.4\Delta with (a) the barrier potential V0=0V_{0}=0, (b) V0=2V_{0}=2, (c) V0=5V_{0}=5, and (d) V0=10V_{0}=10.

Fig.4 gives the incident angle dependence of reflection probabilities for different interfacial barrier heights. We have taken Ez=0.4ΔE_{z}=0.4\Delta, μN=μS=0.2Δ\mu_{N}=\mu_{S}=0.2\Delta and E=0.7ΔE=0.7\Delta. Under these parameters, the conduction band Ee+E_{e}^{+} and valence band Eh+E_{h}^{+} both participate in the scattering. However, for the transparent interface with V0=0V_{0}=0, there are only retro-reflections (RNR and RAR) as shown in Fig.4(a). In particular, the SNR disappears also, which is similar to the ordinary normal-metal-SC junction.Blonder The probabilities are approximatively independent of the incident angle. In fact, the parameters μN\mu_{N}, μS\mu_{S}, the incident energy EE, and EzE_{z} in Fig.4(a) are much smaller than E0=200ΔE_{0}=200\Delta, leading to the wave vectors kxe±k_{x}^{e\pm} and kxh±k_{x}^{h\pm} in NLSM and px±p_{x}^{\pm} and qx±q_{x}^{\pm} for quasiparticles in SC can well be simplified into 2m2E0ky2=2m2E0cosθ\sqrt{\frac{2m}{\hbar^{2}}E_{0}-k_{y}^{2}}=\frac{2m}{\hbar^{2}}E_{0}\cos{\theta}. Under this simplification and V0=0V_{0}=0, the RNR and RAR coefficients can be analytically obtained as

rn2\displaystyle r_{n2} =\displaystyle= v(η21χ11)(η11+χ21)u(η11χ11)(η21+χ21)v(η21χ12)(η11+χ21)+u(η11χ12)(η21+χ21),\displaystyle\frac{v(\eta_{21}-\chi_{11})(\eta_{11}+\chi_{21})-u(\eta_{11}-\chi_{11})(\eta_{21}+\chi_{21})}{-v(\eta_{21}-\chi_{12})(\eta_{11}+\chi_{21})+u(\eta_{11}-\chi_{12})(\eta_{21}+\chi_{21})}, (16)
ra2\displaystyle r_{a2} =\displaystyle= (η11η21)(χ11χ12)v(η21χ12)(η11+χ21)+u(η11χ12)(η21+χ21),\displaystyle\frac{-(\eta_{11}-\eta_{21})(\chi_{11}-\chi_{12})}{-v(\eta_{21}-\chi_{12})(\eta_{11}+\chi_{21})+u(\eta_{11}-\chi_{12})(\eta_{21}+\chi_{21})}, (17)

which are independent of the incident angle θ\theta, and SNR and SAR coefficients rn1=ra1=0r_{n1}=r_{a1}=0 exactly. As a fact, a finite barrier height is the necessary condition for the formation of the specular reflections. Even for a small barrier height with V0=2V_{0}=2 (see Fig.4(b)), the specular reflections will immediately appear although they are not dominant in a large angle range centered at θ=0\theta=0. When the barrier height is raised, SNR Rn1R_{n1} and SAR Ra1R_{a1} are markedly enhanced and the curve of SAR Ra1R_{a1} exceeds that of RNR Rn2R_{n2} as shown in Fig. 4(c) with V0=5V_{0}=5 in a large range. When V0=10V_{0}=10, the probabilities of SNR and SAR at θ=0\theta=0 can achieve about 34%34\% and 12%12\%, respectively. The presence of finite barrier height is beneficial to the specular reflections and detrimental to the retro-reflections. The reflection styles and magnitudes can be regulated by the interfacial transparency.

Refer to caption
Figure 5: The θ\theta dependence of reflection probabilities for E=0.8ΔE=0.8\Delta, μN=μS=0.2Δ\mu_{N}=\mu_{S}=0.2\Delta and V0=10V_{0}=10 with (a) Ez=0E_{z}=0, (b) Ez=0.3ΔE_{z}=0.3\Delta, (c) Ez=0.5ΔE_{z}=0.5\Delta, and (d) Ez=0.7ΔE_{z}=0.7\Delta.

Fig.5 gives the reflection probabilities as functions of the incident angle θ\theta for different EzE_{z} (i.e., the incident wave vector kzk_{z}). For Ez=0E_{z}=0, the orbital coupling in the Hamiltonian HˇN\check{H}_{N} (see Eq.(1)) and the gaps for ELQs and HLQs vanish. The energy dispersions degenerate to Ee±=±ϵkμNE_{e}^{\pm}=\pm\epsilon_{k}-\mu_{N} and Eh±=±ϵk+μNE_{h}^{\pm}=\pm\epsilon_{k}+\mu_{N}. When ELQs on the conduction band Ee+E_{e}^{+} are injected from NLSM to the NLSM-SC interface, they will be normal reflected as ELQs on Ee+E_{e}^{+} and Andreev reflected as HLQs on the valence EhE_{h}^{-}. In other words, only two reflections, SNR and RAR, can happen as shown in Fig.5(a). In this situation, the scattering processes in NLSM-SC junctions are the same with those in the ordinary normal metal-SC junctions.Blonder For Ez0E_{z}\neq 0, the coupling of two orbits leads to the appearances of RNR and SAR. Double normal reflections and double Andreev reflections are realized as shown in Figs.5(b) and (c). The conduction band Ee+E_{e}^{+} and the valence band Eh+E_{h}^{+} are involved in the scattering. With the increase of EzE_{z}, RNR Rn2R_{n2} and SAR Ra1R_{a1} are enhanced. However for Ez=0.7ΔE_{z}=0.7\Delta and μN=μS=0.2Δ\mu_{N}=\mu_{S}=0.2\Delta, the situation is different. Under these parameters, the energy E=0.8ΔE=0.8\Delta lies in the gap of HLQs and crosses the band Ee+E_{e}^{+} of ELQs. The Andreev reflections (RAR and SAR) disappear and only the normal reflections of ELQs are possible. The curves of SNR and RNR oscillate with the incident angle θ\theta, which have two valley values and two peak values respectively as shown in Fig.5(d).

II.2.2 The energy dependence of reflections

Refer to caption
Figure 6: The incident energy EE dependence of reflection probabilities for Ez=0.4ΔE_{z}=0.4\Delta, μN=μS=0.7Δ\mu_{N}=\mu_{S}=0.7\Delta and V0=10V_{0}=10 with (a) θ=0\theta=0, (b) θ=0.1π\theta=0.1\pi, (c) θ=0.3π\theta=0.3\pi, and θ=0.45π\theta=0.45\pi.

Now, we turn to the incident energy dependence of reflection probabilities. Fig.6 presents the results for different θ\theta at μN=μS=0.7Δ\mu_{N}=\mu_{S}=0.7\Delta and Ez=0.4ΔE_{z}=0.4\Delta. The gap for HLQs spans from 0.3Δ0.3\Delta to 1.1Δ1.1\Delta. As a result, the Andreev reflections, both RAR and SAR, are absent in the energy range 0.3Δ<E<1.1Δ0.3\Delta<E<1.1\Delta. The double Andreev reflections can happen in the other energy range, E<0.3ΔE<0.3\Delta or E>1.1ΔE>1.1\Delta, as shown in Figs.6 (a)-(d). For E<0.3ΔE<0.3\Delta, the double Andreev reflections originate from the involved conduction band EhE_{h}^{-}. On the other hand, while for E>1.1ΔE>1.1\Delta, the double Andreev reflections originate from the involved valence band Eh+E_{h}^{+}. For E<0.3ΔE<0.3\Delta, the changes of RAR and SAR are not dramatic. When the incident energy EE increases from 1.1Δ1.1\Delta, the curves of RAR and SAR rise first and then gradually decrease. Note, the summation of the reflection probabilities for E>ΔE>\Delta will not equal to 11 due to the emergence of the quasiparticle transmissions. As for the normal reflections, SNR Rn1R_{n1} and RNR Rn2R_{n2} oscillate dramatically in the range 0.3Δ<E<Δ0.3\Delta<E<\Delta while the changes in the range E<0.3ΔE<0.3\Delta and E>1.1ΔE>1.1\Delta are not obvious. In addition, the Andreev reflections, both RAR and SAR, will tend to zero as θ\theta approaches to 0.5π0.5\pi as shown in Fig.6 (d). This is consistent with the incident angle dependence of reflections discussed above.

Refer to caption
Figure 7: The incident energy EE dependence of the reflection probabilities for θ=0.2π\theta=0.2\pi, μN=μS=0.4Δ\mu_{N}=\mu_{S}=0.4\Delta and V0=10V_{0}=10 with (a) Ez=0E_{z}=0, (b) Ez=0.2ΔE_{z}=0.2\Delta, (c) Ez=0.4ΔE_{z}=0.4\Delta, and (d) Ez=0.6ΔE_{z}=0.6\Delta.

Fig.7 presents the incident energy dependence of reflections for different EzE_{z} at μN=μS=0.4Δ\mu_{N}=\mu_{S}=0.4\Delta with the fixed incident angle θ=0.2π\theta=0.2\pi. For Ez=0E_{z}=0, there are only SNR and RAR in the scattering processes due to the orbital coupling is absent (see Fig.7(a)). For Ez=0.2ΔE_{z}=0.2\Delta, the gap for HLQs is from 0.2Δ0.2\Delta to 0.6Δ0.6\Delta. Correspondingly, the Andreev reflections are absent in the range 0.2Δ<E<0.6Δ0.2\Delta<E<0.6\Delta (see Fig.7(b)). In this energy range, the behaviours of the normal reflections are similar to the results in Fig.6. In the energy range 0.6Δ<E<Δ0.6\Delta<E<\Delta, the probability of SNR drops sharply and RAR Ra2R_{a2} dominates the scattering processes. For Ez=0.4ΔE_{z}=0.4\Delta, the gap for HLQs is from 0 to 0.8Δ0.8\Delta and there are only normal reflections in the energy range E<0.8ΔE<0.8\Delta (see Fig.7 (c)). For Ez=0.6ΔE_{z}=0.6\Delta, the gap for HLQs spans from 0 to Δ\Delta, the Andreev reflections will be absent inside the whole gap (see Fig.7 (d)).

Refer to caption
Figure 8: The incident energy EE dependence of the reflection probabilities for θ=0.2π\theta=0.2\pi, Ez=0.4ΔE_{z}=0.4\Delta and V0=10V_{0}=10 with (a) μN=μS=0\mu_{N}=\mu_{S}=0, (b) μN=μS=0.2Δ\mu_{N}=\mu_{S}=0.2\Delta, (c) μN=μS=0.4Δ\mu_{N}=\mu_{S}=0.4\Delta, and (d) μN=μS=0.6Δ\mu_{N}=\mu_{S}=0.6\Delta.

Fig.8 gives the reflection probabilities for different μN\mu_{N} and μS\mu_{S} at Ez=0.4ΔE_{z}=0.4\Delta and θ=0.2π\theta=0.2\pi. For μN=μS=0\mu_{N}=\mu_{S}=0, the gaps for ELQs and HLQs are both from 0.4Δ-0.4\Delta to 0.4Δ0.4\Delta which are symmetric about E=0E=0. For E<0.4ΔE<0.4\Delta, the energy of the incident quasiparticles lies in the gaps and there is no traveling wave for ELQs. The reflections, both the normal ones and Andreev ones, are not present as shown in Fig.8 (a). When E>0.4ΔE>0.4\Delta, ELQs on the conduction band Ee+E_{e}^{+} and HLQs on the valence band Eh+E_{h}^{+} participate in the scattering processes. Double normal reflections and double Andreev reflections exist simultaneously. It is worth noting that the summation of reflection probabilities equals to 11 from E=0.4ΔE=0.4\Delta to E1.077ΔE\approx 1.077\Delta not to Δ\Delta (see Fig.8(a)). This indicates no quasiparticle transmissions in the energy range Δ<E<1.077Δ\Delta<E<1.077\Delta and the superconducting gap in SC is larger than Δ\Delta in this case. Actually, the quasiparticle dispersion in SC is E=ϵk2+Ez2+Δ2E=\sqrt{\epsilon_{k}^{2}+E_{z}^{2}+\Delta^{2}} when μS=0\mu_{S}=0. The effective gap in SC is Ez2+Δ2\sqrt{E_{z}^{2}+\Delta^{2}} not Δ\Delta which is consistent with the results in Fig.8(a).

When μS0\mu_{S}\neq 0, the quasiparticle dispersion in SC becomes E=[ϵk2+Ez2μS]2+Δ2E=\sqrt{[\sqrt{\epsilon_{k}^{2}+E_{z}^{2}}-\mu_{S}]^{2}+\Delta^{2}}. When μS<Ez\mu_{S}<E_{z}, the effective gap in SC is [|Ez|μS]2+Δ2\sqrt{[|E_{z}|-\mu_{S}]^{2}+\Delta^{2}}. For example, if Ez=0.4ΔE_{z}=0.4\Delta and μS=0.2Δ\mu_{S}=0.2\Delta as considered in Fig.8(b), the effective gap will approximate to 1.02Δ1.02\Delta (see also Fig.7(d)). At the same time, there will be a gap from 0.6Δ-0.6\Delta to 0.2Δ0.2\Delta for ELQs in NLSM. Hence, there are no reflections in the range 0<E<0.2Δ0<E<0.2\Delta and the sum of probabilities for four reflections equals to 11 in the range 0.2Δ<E<[|Ez|μS]2+Δ21.02Δ0.2\Delta<E<\sqrt{[|E_{z}|-\mu_{S}]^{2}+\Delta^{2}}\simeq 1.02\Delta. When μS\mu_{S} is raised to μS=Ez\mu_{S}=E_{z}, the gap in SC reverts to its intrinsic value Δ\Delta and the gap for ELQs in NLSM completely locates below E=0E=0, which leads to the reflections probabilities shown in Fig.8(c). Here the Andreev reflections disappear at E<0.8ΔE<0.8\Delta due to the gap for HLQs in NLSM. When μS\mu_{S} continues to rise, the gap in SC will not be changed and keep the value Δ\Delta. For μN=μS=0.6Δ\mu_{N}=\mu_{S}=0.6\Delta in Fig.8(d), ELQs and HLQs involved in the scattering are from the conduction bands Ee+E_{e}^{+} and EhE_{h}^{-} in the energy range 0<E<0.2Δ0<E<0.2\Delta while for 0.2Δ<E<Δ0.2\Delta<E<\Delta, only ELQs from Ee+E_{e}^{+} participate in the scattering.

Refer to caption
Figure 9: The incident energy EE dependence of the reflection probabilities for θ=0.2π\theta=0.2\pi, μN=μS=0.7Δ\mu_{N}=\mu_{S}=0.7\Delta and Ez=0.4ΔE_{z}=0.4\Delta with (a) V0=0V_{0}=0, (b) V0=2V_{0}=2, (c) V0=5V_{0}=5, and V0=10V_{0}=10.

Fig.9 shows the incident energy dependence of reflections for different interfacial barriers at μN=μS=0.7Δ\mu_{N}=\mu_{S}=0.7\Delta and Ez=0.4ΔE_{z}=0.4\Delta. Since the relation μN>Ez\mu_{N}>E_{z} is fixed, the conduction band Ee+E_{e}^{+} is always involved in the scattering processes while the dispersions for HLQs have a gap from 0.3Δ0.3\Delta to 1.1Δ1.1\Delta. For 0.3Δ<E<1.1Δ0.3\Delta<E<1.1\Delta, there will be no Andreev reflections as given in Fig.9 (a)-(d). When the interface is transparent, the specular reflections, SNR and SAR, are also absent. Therefore, there will be only RNR in the energy range 0.3Δ<E<1.1Δ0.3\Delta<E<1.1\Delta for V0=0V_{0}=0 with Rn2=1R_{n2}=1, as shown in Fig.9(a). In this case, the incident ELQs are completely reflected back along the retro-reflected direction, which phenomenon is very rare. In the energy range E<0.3ΔE<0.3\Delta, the conduction band EhE_{h}^{-} also participates in the scattering and two retro-reflections coexist. This is consistent with the results in Fig.4(a). When V00V_{0}\neq 0, the specular reflections are activated. For 0.3Δ<E<1.1Δ0.3\Delta<E<1.1\Delta, there will be double normal reflections and for E<0.3ΔE<0.3\Delta or E>1.1ΔE>1.1\Delta, there will be four types of reflections. From Figs.9 (b)-(d), we can find the specular reflections are enhanced as the opacity of the interface is increased. In other words, the presence of finite barrier height is beneficial to the specular reflections and detrimental to the retro-reflections. This conclusion is consistent with the results in Figs.4 (b)-(d).

II.2.3 Conductance spectra

Refer to caption
Figure 10: The normalized conductance spectra for different chemical potentials with (a) the barrier potential V0=0V_{0}=0 and (b) V0=10V_{0}=10.

Here, we discuss the conductance of the NLSM-SC junctions. In the expression of conductance in Eq.(15), there is an integral of EzE_{z} from eVb-eV_{b} to eVbeV_{b} with VbV_{b} the bias. Since the probabilities are even functions of EzE_{z}, it is enough to analyze the conductance supposing Ez>0E_{z}>0. Fig.10(a) gives the normalized conductance for the transparent interface with the barrier potential V0=0V_{0}=0. According to the discussions in the last subsection, the specular reflections in this situation are absent and only RNR and RAR exist. For μN=0\mu_{N}=0, the gaps for ELQs and HLQs in NLSM are symmetric about the energy E=0E=0 and span from Ez-E_{z} to EzE_{z}. The expression of conductance requires eVb>EzeV_{b}>E_{z}, which indicates the conduction band Ee+E_{e}^{+} and the valence band Eh+E_{h}^{+} always participate in the scattering processes. Inside the gap, the total RAR will lead to the normalized conductance being 22. This is the same value with the conductance of the ordinary normal-metal-SC junctions with the transparent interface Blonder .

For μN=μS0\mu_{N}=\mu_{S}\neq 0, HLQs are not always involved in the scattering. The gap for HLQs is no longer symmetric about E=0E=0 and spans from μNEz\mu_{N}-E_{z} to μN+Ez\mu_{N}+E_{z}. The condition for RAR is eVb+Ez<μNeV_{b}+E_{z}<\mu_{N} or eVbEz>μNeV_{b}-E_{z}>\mu_{N} (Note, SAR is absent due to V0=0V_{0}=0.). In the former situation, HLQs from the conduction band EhE_{h}^{-} are responsible for RAR. For eVb<μNeV_{b}<\mu_{N}, when eVbeV_{b} is increased, the value range of EzE_{z} satisfying the former condition becomes narrow. Correspondingly, the EzE_{z} range for RAR also narrows and the conductance decreases. In the latter situation, HLQs from the valence band Eh+E_{h}^{+} are responsible for RAR. For eVb>μNeV_{b}>\mu_{N}, when eVbeV_{b} is increased, the value range of EzE_{z} satisfying the latter condition becomes wide. The EzE_{z} range for RAR also widens and the conductance rises. For the special case of eVb=μNeV_{b}=\mu_{N}, no value of EzE_{z} under the supposed relation Ez>0E_{z}>0 meets the two conditions for RAR. There is only the total RNR in the scattering processes which will cause the zero conductance. Our analysis can be immediately demonstrated by the numerical results with μN=μS=0.1Δ\mu_{N}=\mu_{S}=0.1\Delta, 0.5Δ0.5\Delta and 1.0Δ1.0\Delta in Fig.10(a). For μN=μS>Δ\mu_{N}=\mu_{S}>\Delta, the zero conductance phenomenon will not disappear, since quasiparticle transmissions happen at the incident energy larger than Δ\Delta.

Fig.10(b) gives the normalized conductance for V0=10V_{0}=10. According to the discussions in the last subsection, the specular reflections will be activated. For μN=0\mu_{N}=0, double normal reflections and double Andreev reflections coexist. The presence of the normal reflections will weaken the conductance and its value can not reach 2σ02\sigma_{0} inside the gap. For μN0\mu_{N}\neq 0, the condition for the Andreev reflections, both RAR and SAR, is also eVb+Ez<μNeV_{b}+E_{z}<\mu_{N} or eVbEz>μNeV_{b}-E_{z}>\mu_{N}. For eVb=μNeV_{b}=\mu_{N}, the conductance becomes zero. On the both sides of eVb=μNeV_{b}=\mu_{N}, the variations of conductance are similar to the curves in Fig.10(a). The conductance for μN=μS=5Δ\mu_{N}=\mu_{S}=5\Delta at V0=10V_{0}=10 is also presented in Fig.10(b) and its properties are similar to the curves with μN=μS=0\mu_{N}=\mu_{S}=0. Although the impressive vanishing conductance at eVb=μNeV_{b}=\mu_{N} are reminiscent of graphene-SC junctions Beenakker , the chemical potential dependence of conductance spectra is very different. For the case of graphene, the single Andreev reflection, RAR or SAR, happens depending on the value of the chemical potential.

III Nodal-line parallel to interface

III.1 Model and formalism

Refer to caption
Figure 11: (a) Schematic illustration of the NLSM-SC junction with the nodal-line parallel to the interface. The interface is located at z=0z=0 and parallel to the xyxy plane. (b) The torus-like isoenergetic surface in NLSM. The crystal axis kzk_{z} is perpendicular to the interface of the junction.

In this subsection, we give the formalism for the scattering processes and conductance of the NLSM-SC junction with the nodal-line parallel to the interface as shown in Fig.11. In this case, the NLSM and SC are located in the regions z<0z<0 and z>0z>0, respectively, with the NLSM-SC interface being at z=0z=0. The transport is along the zz direction. The Hamiltonians of the NLSM and SC are the same as the Eqs.(2) and (8) in Sec.II.1. The quasiparticle dispersions of the NLSM for ELQs and HLQs solved from the BdG equation are

Ee±=±ϵk2+2v2kz2μN,E_{e}^{\pm}=\pm\sqrt{\epsilon_{k}^{2}+\hbar^{2}v^{2}k_{z}^{2}}-\mu_{N}, (18)

and

Eh±=±ϵk2+2v2kz2+μN,E_{h}^{\pm}=\pm\sqrt{\epsilon_{k}^{2}+\hbar^{2}v^{2}k_{z}^{2}}+\mu_{N}, (19)

respectively, which are the same as the dispersions in Eqs.(3) and (4). However, for each band, there is only one type of quasiparticle. Taking the conduction band Ee+E_{e}^{+} as an example, when the incident energy EE and wave vectors kxk_{x} and kyk_{y} are given, the equation Ee+(kx,ky,kz)=EE_{e}^{+}(k_{x},k_{y},k_{z})=E of kzk_{z} has only two solutions, kzek_{z}^{e} and kze-k_{z}^{e}. Here the kzek_{z}^{e} and kze-k_{z}^{e} states propagate along the +z+z and z-z directions, respectively. This is different from the case of the nodal-line perpendicular to the interface, in which there are two types of quasiparticle for each band. So for the present case, there are only two reflection processes, the normal reflection and Andreev reflection, when a beam of ELQs are injected from NLSM side. Consider that the incident ELQs are from the conduction band Ee+E_{e}^{+}. They will be specularly reflected as ELQs on the band Ee+E_{e}^{+} and retro-Andreev reflected as HLQs on the conduction band EhE_{h}^{-} if E<μNE<\mu_{N} or specular Andreev reflected as HLQs on the valence band Eh+E_{h}^{+} if E>μNE>\mu_{N}. In this situation, the scattering processes are the same as that in the graphene-SC junction.Beenakker ; addsun3 ; addsun4

Assuming an ELQ with the wave vector kzek_{z}^{e} is injected from NLSM, the wave function in NLSM can be written as

ΨN=(iξ1100)eikzez+rn(iξ1100)eikzez+ra(00αiξ21)eαikzhz,\begin{split}\Psi_{N}&=\left(\begin{array}[]{cc}i\xi_{1}\\ 1\\ 0\\ 0\\ \end{array}\right)e^{ik_{z}^{e}z}+r_{n}\left(\begin{array}[]{cc}-i\xi_{1}\\ 1\\ 0\\ 0\\ \end{array}\right)e^{-ik_{z}^{e}z}\\ &+r_{a}\left(\begin{array}[]{cc}0\\ 0\\ -\alpha i\xi_{2}\\ 1\end{array}\right)e^{\alpha ik_{z}^{h}z},\end{split} (20)

where ξ1=E+μN+ΩN+vkze\xi_{1}=\frac{E+\mu_{N}+\Omega_{N}^{+}}{\hbar vk_{z}^{e}}, ξ2=EμNΩNvkzh\xi_{2}=\frac{E-\mu_{N}-\Omega_{N}^{-}}{\hbar vk_{z}^{h}} and kze(h)=2m2{(ek+mv2)2+[(E+()μN)2ek2](ek+mv2)}k_{z}^{e(h)}=\sqrt{\frac{2m}{\hbar^{2}}\{\sqrt{(e_{k}+mv^{2})^{2}+[(E+(-)\mu_{N})^{2}-e_{k}^{2}]}-(e_{k}+mv^{2})\}} with ΩN+()=(E+()μN)2(vkze(h))2\Omega_{N}^{+(-)}=\sqrt{(E+(-)\mu_{N})^{2}-\left(\hbar vk_{z}^{e(h)}\right)^{2}} and ek=22m(kx2+ky2)E0e_{k}=\frac{\hbar^{2}}{2m}(k_{x}^{2}+k_{y}^{2})-E_{0}. When E<μNE<\mu_{N}, α=1\alpha=1 and rar_{a} denotes the RAR amplitude; when E>μNE>\mu_{N}, α=1\alpha=-1 and rar_{a} denotes the SAR amplitude.

The wave function in SC is

ΨS=c(iuζ1uiζ11)eikz+z+d(ivζ2viζ21)eikzz,\begin{split}\Psi_{S}&=c\left(\begin{array}[]{cc}iu\zeta_{1}\\ u\\ i\zeta_{1}\\ 1\\ \end{array}\right)e^{ik_{z}^{+}z}+d\left(\begin{array}[]{cc}-iv\zeta_{2}\\ v\\ -i\zeta_{2}\\ 1\\ \end{array}\right)e^{-ik_{z}^{-}z},\end{split} (21)

where u=E+ΩSΔu=\frac{E+\Omega_{S}}{\Delta}, v=EΩSΔv=\frac{E-\Omega_{S}}{\Delta}, ζ1=μS+ΩS+ΩS+vkz+\zeta_{1}=\frac{\mu_{S}+\Omega_{S}+\Omega_{S}^{+}}{\hbar vk_{z}^{+}}, ζ2=μSΩS+ΩSvkz\zeta_{2}=\frac{\mu_{S}-\Omega_{S}+\Omega_{S}^{-}}{\hbar vk_{z}^{-}} and kz±=2m2{(ek+mv2)2+(μS±ΩS)2ek2(ek+mv2)}k_{z}^{\pm}=\sqrt{\frac{2m}{\hbar^{2}}\{\sqrt{(e_{k}+mv^{2})^{2}+(\mu_{S}\pm\Omega_{S})^{2}-e_{k}^{2}}-(e_{k}+mv^{2})\}} with ΩS=E2Δ2\Omega_{S}=\sqrt{E^{2}-\Delta^{2}} and ΩS±=(μS±ΩS)22v2kz±2\Omega_{S}^{\pm}=\sqrt{(\mu_{S}\pm\Omega_{S})^{2}-\hbar^{2}v^{2}{k_{z}^{\pm}}^{2}}.

We are concerned with the nearly linear regime of the dispersion E(kz)E(k_{z}) where the parameter mv2mv^{2} is large enough. In this situation, the following boundary condition for the wave functions are suitable,

ψN(z=0)=ψS(z=0+).\psi_{N}(z=0^{-})=\psi_{S}(z=0^{+}). (22)

The reflection amplitudes rr and rar_{a} can be obtained from the boundary condition and the reflection probabilities are given by

Rn(E)=|r|2,Ra(E)=|ξ2ξ1||ra|2.\begin{split}R_{n}(E)&=|r|^{2},\\ R_{a}(E)&=\big{|}\frac{\xi_{2}}{\xi_{1}}\big{|}|r_{a}|^{2}.\end{split} (23)

Inside the gap with E<ΔE<\Delta, the conservation relation Rn+Ra=1R_{n}+R_{a}=1 is satisfied.

Since the traveling wave solution for ELQs is restricted in the interval (eVb+μN)<ek<eVb+μN-(eV_{b}+\mu_{N})<e_{k}<eV_{b}+\mu_{N}, the conductance can be expressed as

σ=2e2hS2πkminkmax[1+Ra(eVb)Rn(eVb)]k𝑑k,\sigma=\frac{2e^{2}}{h}\frac{S}{2\pi}\int_{k_{min}}^{k_{max}}[1+R_{a}(eV_{b})-R_{n}(eV_{b})]kdk, (24)

with kmin=2m2[E0(eVb+μN)]k_{min}=\sqrt{\frac{2m}{\hbar^{2}}[E_{0}-(eV_{b}+\mu_{N})]} and kmax=2m2[E0+(eVb+μN)]k_{max}=\sqrt{\frac{2m}{\hbar^{2}}[E_{0}+(eV_{b}+\mu_{N})]}, VbV_{b} is the bias between the NLSM and SC. The normalized conductance is σ/σ0\sigma/\sigma_{0} with σ0\sigma_{0} the conductance of the NLSM-NLSM junction.

III.2 Results and discussions

Refer to caption
Figure 12: The normalized conductance spectra for the junction with the nodal-line parallel to the interface. The parameters are taken as E0=200ΔE_{0}=200\Delta and μS=500Δ\mu_{S}=500\Delta.

We present the conductance of the NLSM-SC junction with the nodal-line parallel to the NLSM-SC interface for various of μN\mu_{N} in Fig.12. Now, the kzk_{z} axis is perpendicular to the interface and the current flows along the zz axis. The nearly linear dispersion regime of EE about kzk_{z} is concerned, when the parameter mv2mv^{2} is large enough. In this situation, there are only SNR and a single Andreev reflection, RAR or SAR, in the scattering processes. For the Andreev reflection, RAR happens for eVb<μNeV_{b}<\mu_{N} while SAR happens for eVb>μNeV_{b}>\mu_{N}. Inside the gap, RAR dominates the conductance if μNΔ\mu_{N}\gg\Delta, while SAR dominates the conductance if μNΔ\mu_{N}\ll\Delta. The scattering processes are the same with those in the graphene-SC junctionBeenakker . Actually, the conductance spectra in Fig.12 are highly similar to the results for the graphene junction as shown in Ref.[Beenakker, ]. From the results in Figs.10 and 12, it is found that the transport properties of the NLSM-SC junction exhibit anisotropy. The conductance spectra depend on the orientation relationship between the crystal axis (or the nodal-line) and the junction interface.

IV Conclusions

We study the scattering processes and conductance in NLSM-SC junctions. We find the novel quadruple reflections, double normal reflections and double Andreev reflections, can happen simultaneously in NLSM. The probabilities of reflections are systematically studied under different parameters. The occurrence and disappearance of the reflections and their magnitudes can be controlled by changing the incident quasiparticle energy, the incident angle, the interfacial barrier, the chemical potentials and the orbital coupling strength. Distinctive features manifest themselves in the conductance spectra and the spectra show anisotropy depending on the mutual orientation of the nodal-line and the interface. This unique quadruple reflections may be useful for the engineering of the electronic beam splitter based on the ballistic system.

ACKNOWLEDGMENTS

This work was financially supported by National Key R and D Program of China (2017YFA0303301), NSF-China under Grants Nos. 11921005 and 11447175, the Strategic Priority Research Program of Chinese Academy of Sciences (XDB28000000) and the Natural Science Foundation of Shandong Province under Grants No. ZR2017QA009.

REFERENCES

References

  • (1) N. P. Armitage, E. J. Mele, and A. Vishwanath, Rev. Mod. Phys. 90, 015001 (2018).
  • (2) A. A. Burkov, M. D. Hook, and L. Balents, Phys. Rev. B 84, 235126 (2011).
  • (3) C.-K. Chiu, and A. P. Schnyder, Phys. Rev. B 90, 205136 (2014).
  • (4) C. Fang, Y. Chen, H.-Y. Kee, and L. Fu, Phys. Rev. B 92, 081201 (2015).
  • (5) J.-T. Wang, H. Weng, S. Nie, Z. Fang, Y. Kawazoe, and C. Chen, Phys. Rev. Lett. 116, 195501 (2016).
  • (6) H. Weng, C. Fang, Z. Fang, B. A. Bernevig, and X. Dai, Phys. Rev. X 5, 011029 (2015).
  • (7) C. Fang, L. Lu, J. Liu, and L. Fu, Nat. Phys. 12, 936 (2016).
  • (8) Z. Yan, P.-W. Huang, Z. Wang, Phys. Rev. B 93, 085138 (2016).
  • (9) M. A. N. Arau´\acute{\text{u}}jo, and L. Li, Phys. Rev. B 98, 155114 (2018).
  • (10) Y. Huh, E.-G. Moon, and Y. B. Kim, Phys. Rev. B 93, 035138 (2016).
  • (11) Y. Gao, Y. Chen, Y. Xie, P.-Y. Chang, M. L. Cohen, and S. Zhang, Phys. Rev. B 97, 121108(R) (2018).
  • (12) K. Moors, A. A. Zyuzin, A. Y. Zyuzin, R. P. Tiwari, and T. L. Schmidt, Phys. Rev. B 99, 041116(R) (2019).
  • (13) D.-S. Ma, J. Zhou, B. Fu, Z.-M. Yu, C.-C. Liu, and Y. Yao, Phys. Rev. B 98, 201104(R) (2018).
  • (14) A. Marti´\acute{\text{i}}n-Ruiz, and A. Cortijo, Phys. Rev. B 98 155125 (2018).
  • (15) H. Yang, R. Moessner, and L.-K. Lim, Phys. Rev. B 97, 165118 (2018).
  • (16) L. Oroszla´\acute{\text{a}}ny, B. Do´\acute{\text{o}}ra, J. Cserti, A. Cortijo, Phys. Rev. B 97, 205107 (2018).
  • (17) J. Liu, and L. Balents, Phys. Rev. B 95, 075426 (2017).
  • (18) J. Behrends, J.-W. Rhim, S. Liu, A. G. Grushin, and J. H. Bardarson, Phys. Rev. B 96, 245101 (2017).
  • (19) C. Li, C. M. Wang, B. Wan, X. Wan, H.-Z. Lu, and X. C. Xie, Phys. Rev. Lett. 120, 146602 (2018).
  • (20) M. Koshino, and I. F. Hizbullah, Phys. Rev. B 93, 045201 (2016).
  • (21) R. A. Molina, and J. Gonza´\acute{\text{a}}lez, Phys. Rev. Lett. 120, 146601 (2018).
  • (22) W. Chen, H.-Z. Lu, and O. Zilberberg, Phys. Rev. Lett. 122, 196603 (2019).
  • (23) W. Chen, K. Luo, L. Li, and O. Zilberberg, Phys. Rev. Lett. 121, 166802 (2018).
  • (24) S. Barati, and S. H. Abedinpour, Phys. Rev. B 96, 155150 (2017).
  • (25) Y.-R. Hao, L. Wang, and D.-X. Yao, Phys. Rev. B 99, 165406 (2019).
  • (26) P. Lv, N. Dai, and Q.-F. Sun, Phys. Rev. B 97, 235425 (2018).
  • (27) D. A. Khokhlov, A. L. Rakhmanov, and A. V. Rozhkov, Phys. Rev. B 97, 235418 (2018).
  • (28) W. B. Rui, Y. X. Zhao, and A. P. Schnyder, Phys. Rev. B 97, 161113 (2018).
  • (29) Y. Hu, H. Liu, H. Jiang, and X. C. Xie, Phys. Rev. B 96, 134201 (2017).
  • (30) A. F. Andreev, Zh. Eksp. Teor. Fiz. 46, 1823 (1964) [Sov. Phys. JETP 19, 1228 (1964)].
  • (31) Q.-F. Sun, J. Wang, and T.-H. Lin, Phys. Rev. B 59, 3831 (1999).
  • (32) Q.-F. Sun and X. C. Xie, J. Phys.: Condens. Matter 21, 344204 (2009).
  • (33) C. W. J. Beenakker, Phys. Rev. Lett. 97, 067007 (2006).
  • (34) Q. Y. Zhang, D.Y. Fu, B. G. Wang, R. Zhang, D. Y. Xing, Phys. Rev. Lett. 101, 047005 (2008).
  • (35) S.-G. Cheng, Y. Xing, J. Wang, and Q.-F. Sun, Phys. Rev. Lett. 103, 167003 (2009).
  • (36) Y. Xing, J. Wang, and Q.-F. Sun, Phys. Rev. B 83, 205418 (2011).
  • (37) J. Linder, T. Yokoyama, D. Huertas-Hernando, and A. Sudbø, Phys. Rev. Lett. 100, 187004 (2008).
  • (38) Z. Hou, and Q.-F. Sun, Phys. Rev. B 96, 155305 (2017).
  • (39) L. S. Xie, L. M. Schoop, E. M. Seibel, Q. D. Gibson, W. Xie, and R. J. Cava, APL Materials 3, 083602 (2015).
  • (40) Y.-H. Chan, C.-K. Chiu, M. Y. Chou, and A. P. Schnyder, Phys. Rev. B 93, 205132 (2016).
  • (41) G. E. Blonder, M. Tinkham, and T. M. Klapwijk, Phys. Rev. B 25, 4515 (1982).