betaFeedback

WorksPhysics and Astronomy

Prediction of ferroelectric superconductors with reversible superconducting diode effect

Abstract

A noncentrosymmetric superconductor can have a superconducting diode effect, where the critical current in opposite directions is different when time-reversal symmetry is also broken. We theoretically propose that a ferroelectric superconductor with coexisting ferroelectricity and superconductivity can support a ferroelectric reversible superconducting diode effect. Through first-principles calculation, we predict that monolayer CuNb2_2Se4_4 (i.e., bilayer NbSe2_2 intercalated with Cu) is such a ferroelectric superconductor, where ferroelectricity controls the layer polarization as well as the sign of spin-orbit coupling induced spin splittings. Because the nonreciprocal effect of the critical current is proportional to the spin splittings, the superconducting diode effect is reversible upon electric switch of ferroelectricity. While we use CuNb2_2Se4_4 as a model system, the predicted effect can appear in a class of two-dimensional superconducting bilayers with ferroelectricity induced by interlayer sliding. Our work opens the door to studying the interplay between superconductivity and ferroelectricity in two-dimensional materials.

Introduction

A noncentrosymmetric material can support nonreciprocal charge transport, where the electrical resistance becomes different if the direction of the charge current is reversed. Recently, nonreciprocal phenomena in superconductors have become an active research topic [3, 4, 5, 6, 7, 8] [1, 2, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18]. In bulk metals without the inversion symmetry, nonreciprocal charge transport occurs when the time-reversal symmetry is also broken. This nonreciprocity induced by the magnetochiral anisotropy is significantly enhanced for the paraconductivity near the superconducting transition temperature TcT_{\mathrm{c}} because of the superconducting fluctuation [1]. Moreover, in the superconducting state below TcT_{\mathrm{c}}, the critical current along opposite directions differs, i.e., jc(n^)≠jc(−n^)j_{\mathrm{c}}(\hat{n}) \ne j_{\mathrm{c}}(-\hat{n}), where jc(n^)j_{\mathrm{c}}(\hat{n}) represents the magnitude of critical current along direction n^\hat{n}. This nonreciprocity results in the superconducting diode effect (SDE) [7], where the system is superconducting in one direction but resistive in the opposite direction if the applied current has a magnitude between jc(n^)j_{\mathrm{c}}(\hat{n}) and jc(−n^)j_{\mathrm{c}}(-\hat{n}). The SDE has recently been observed experimentally in several systems [7, 8] [9, 10, 11, 12, 19], including an artificial superlattice [Nb/V/Ta]n_n [7] and a heterostructure of twisted trilayer graphene and WSe2_2 [19]. Theory on the SDE has been developed based on Ginzburg–Landau free energy as well as microscopic calculation [14–18].

In this Letter, we introduce a new type of noncentrosymmetric superconductors, i.e., ferroelectric superconductors, where ferroelectricity acts as a new knob in tuning superconductivity. A ferroelectric material breaks the inversion symmetry with a spontaneous electric polarization that can be reversed by an applied electric field. While ferroelectric semiconductors/insulators have been widely studied, ferroelectricity can also exist in metals as exemplified by few-layer WTe2_2 [20]. When a ferroelectric metal (also known as polar metal) becomes superconducting at low temperatures, a ferroelectric superconductor forms and supports the SDE if time-reversal symmetry is further broken. We predict that the superconducting direction of the diode can be reversed upon ferroelectric reversal, which we term as the reversible SDE. This prediction represents an example on the controlling of superconductivity through ferroelectricity.

For material realization, we propose monolayer CuNb2_2Se4_4 to be a ferroelectric superconductor with the reversible SDE. Monolayer CuNb2_2Se4_4 can be viewed as a 2H bilayer NbSe2_2 intercalated by Cu atoms, as illustrated in Figure 1. We establish both ferroelectricity and superconductivity in this material through first-principles calculations. The ferroelectricity in monolayer CuNb2_2Se4_4 controls the layer and spin degrees of freedom of low-energy states across the Fermi energy in ±K\pm K valleys (two corners of the hexagonal Brillouin zone), where both the layer polarization and the valley-dependent spin splittings [induced by spin-orbit coupling (SOC)] are reversed by ferroelectric reversal. In the superconducting state, the nonreciprocal factor η\eta Eq. (8) of the critical current is proportional to the SOC-induced spin splittings. Therefore, η\eta changes sign by the reverse of ferroelectricity, which leads to the reversible SDE. While we study monolayer CuNb2_2Se4_4 in detail to elucidate the physics, reversible SDE is generally expected in ferroelectric superconductors, which can be realized in van der Waals (vdW) superconducting bilayers with ferroelectricity arising from interlayer sliding [21]. Our work not only presents a novel approach to detect ferroelectricity through superconducting transport, but also opens up new opportunities for the construction of electrically controllable and nondissipative diodes.

Top and side views of eight high-symmetry structures and energy landscape of monolayer CuNb$_2$Se$_4$

Figure 1. (a) Top and side views of 8 high-symmetry structures of monolayer CuNb2_2Se4_4. In structure I–III, Nb atoms are aligned vertically. In structure IV and V, Se atoms are aligned vertically. In structure VI–VIII, Nb atoms are aligned with Se atoms vertically. Structure II and VII are inversion symmetric, while other structures are not. (b) The energy per formula unit for different lattice structures of monolayer CuNb2_2Se4_4. The red dashed circles mark the high-symmetry structures shown in (a). The transition structures between two neighboring high-symmetry structures are generated by the nudged-elastic band method.

Ferroelectricity.— The vdW materials, such as graphite and transition metal dichalcogenides (e.g., NbS2_2, NbSe2_2, MoS2_2), can be intercalation host, perconducting at low temperatures, a ferroelectric superconductor forms and supports the SDE if time-reversal symmetry is further broken. We predict that the superconducting direction of the diode can be reversed upon ferroelectric reversal, which we term as the reversible SDE. This prediction represents an example on the controlling of superconductivity through ferroelectricity.

For material realization, we propose monolayer CuNb2_2Se4_4 to be a ferroelectric superconductor with the reversible SDE. Monolayer CuNb2_2Se4_4 can be viewed as a 2H bilayer NbSe2_2 intercalated by Cu atoms, as illustrated in Figure 1. We establish both ferroelectricity and superconductivity in this material through first-principles calculations. The ferroelectricity in monolayer CuNb2_2Se4_4 controls the layer and spin degrees of freedom of low-energy states across the Fermi energy in ±K\pm K valleys (two corners of the hexagonal Brillouin zone), where both the layer polarization and the valley-dependent spin splittings [induced by spin-orbit coupling (SOC)] are reversed by ferroelectric reversal. In the superconducting state, the nonreciprocal factor η\eta Eq. (8) of the critical current is proportional to the SOC-induced spin splittings. Therefore, η\eta changes sign by the reverse of ferroelectricity, which leads to the reversible SDE. While we study monolayer CuNb2_2Se4_4 in detail to elucidate the physics, reversible SDE is generally expected in ferroelectric superconductors, which can be realized in van der Waals (vdW) superconducting bilayers with ferroelectricity arising from interlayer sliding [21]. Our work not only presents a novel approach to detect ferroelectricity through superconducting transport, but also opens up new opportunities for the construction of electrically controllable and nondissipative diodes.

Ferroelectricity.— The vdW materials, such as graphite and transition metal dichalcogenides (e.g., NbS2_2, NbSe2_2, MoS2_2), can be intercalation host, provides a powerful approach to induce a variety of exotic quantum phenomena [22–28] including ferroelectricity [29]. Intercalation has also been achieved in the two dimensional (2D) limit [30–32]. Here we theoretically study monolayer CuNb2_2Se4_4. The strategy is to start from a prototypical 2D superconductor (i.e., bilayer NbSe2_2) [33], which we show to develop ferroelectricity upon Cu intercalation while remaining superconducting.

We demonstrate ferroelectricity in monolayer CuNb2_2Se4_4 by studying 8 high-symmetry structures, as illustrated in Figure 1(a). In the monolayer CuNb2_2Se4_4 under study, the top and bottom NbSe2_2 layers are rotated by 180°. The 8 structures can be distinguished by the in-plane relative positions of the top NbSe2_2 layer, the middle Cu layer and the bottom NbSe2_2 layer. We perform lattice relaxation for each structure using first-principles calculation implemented in the Vienna ab-initio simulation package (VASP) [34] and obtain the corresponding energy. The energy landscape plotted in Figure 1(b) shows that two different structures, i.e., structure VI and VIII, have the same lowest energy. We note that VI and VIII structures are inversion partners, although each of them on its own lacks inversion symmetry. Therefore, VI and VIII structures have the same energy, but opposite layer polarizations, which gives rise to ferroelectricity. We further perform ab-initio molecular dynamics (AIMD) simulations and phonon spectrum calculations for these two structures to verify structural stability. The computational details are presented in the Supplemental Material (SM) [35] (see, also, Refs. 36–39 therein). In AIMD simulations, the energy fluctuates slightly and the structure maintains integrity after 5 ps at 300 K, indicating the thermodynamic stability. Meanwhile, the phonon dispersion calculated using PHONOPY code [40] has no virtual frequency in the whole Brillouin zone, implying the dynamic stability.

We use Bader charge analysis [41] to quantitatively characterize the ferroelectricity. For structure VI, we find that the average number of electrons transferred from one Cu atom to the top and bottom NbSe2_2 layers is 0.23 and 0.1, respectively [35], which are unequal because the Cu atoms occupy noncentrosymmetric sites. Thus, the structure VI poses a downward electric dipole moment. By contrast, the structure VIII has an upward electric dipole moment. Hereafter, we refer to VI and VIII structures as Pdw_{\rm dw} and Pup_{\rm up} structures, respectively. We also confirm the electric polarization by calculating the electrostatic potential difference across the monolayer, as discussed in the SM [35]. We investigate the ferroelectric transition process using the nudged-elastic band method [42]. The transition barrier [Figure 1(b)] between Pdw_{\rm dw} and Pup_{\rm up} is about 0.6 eV per formula unit, which is comparable to that of monolayer In2_2Se3_3 [43]. Therefore, ferroelectric reversal by an applied out-of-plane electric field is feasible in monolayer CuNb2_2Se4_4.

Band Structure.— The band structures including SOC effects are plotted in Figure 2. Because Pup_{\rm up} and Pdw_{\rm dw} structures are inversion partners and time-reversal symmetry is preserved, their band structures have identical energy dispersion. At the Fermi energy, there are 8 Fermi pockets, of which 4 are in Γ\Gamma valley, 2 in +K+K valley, and 2 in −K-K valley [Fig. 2(c)]. Here Γ\Gamma and ±K\pm K represent, respectively, the center and two inequivalent corners of the Brillouin zone.

Layer- and spin-projected band structures and Fermi surfaces of monolayer CuNb$_2$Se$_4$

Figure 2. (a) The layer-projected band structure of monolayer CuNb2_2Se4_4 with SOC effect for Pup_{\rm up} (left panel) and Pdw_{\rm dw} (right panel). (b) The spin-projected band structure. The spin projection is along out-of-plane z^\hat{z} axis with red representing spin up and blue representing spin down. Fermi energy is set to be 0 in (a) and (b). (c) The spin-projected Fermi surfaces.

The opposite electric polarization in Pup_{\rm up} and Pdw_{\rm dw} structures results in differences in electronic states regarding the layer and spin degrees of freedom. The layer-projected band structures in Figure 2(a) show that the two bands crossing the Fermi energy EFE_{\rm F} in ±K\pm K valleys are mainly localized in the top (bottom) NbSe2_2 layer for Pup_{\rm up} (Pdw_{\rm dw}) structure. The opposite layer polarization in combination with the 180° rotation between the two layers leads to ferroelectric reversible spin-valley coupling. To elaborate on this feature, we first focus on the Pup_{\rm up} structure. In Pup_{\rm up}, the higher and lower energy bands across EFE_{\rm F} carry, respectively, up and down spin polarization in +K+K valley, but down and up spin polarization in −K-K valley dictated by time-reversal symmetry [Fig. 2(b)]. Here the spin polarization is along the out-of-plane z^\hat{z} axis. This is the well-known valley-dependent spin splitting effect in transition metal dichalcogenides

[44]. We now turn to the Pdw_{\rm dw} structure, where the spin splitting in a given valley is opposite compared to that in the Pup_{\rm up} structure. Therefore, the spin-valley coupling is controlled by the layer polarization, which is, in turn, controlled by the ferroelectricity.

The spin and valley dependent band structure in ±K\pm K valleys can be described by the following effective Hamiltonian,

H0=−ℏ2k22m∗+λw(kx3−3kxky2)τz+ΔSOC(ℓ)τzσz−EF(1)\mathcal{H}_0=-\frac{\hbar^2 k^2}{2m^*}+\lambda_w(k_x^3-3k_xk_y^2)\tau_z+\Delta_{\mathrm{SOC}}^{(\ell)}\tau_z\sigma_z-E_{\mathrm F} \tag*{(1)}

where τz=±\tau_z=\pm for ±K\pm K valleys, σz=±\sigma_z=\pm for spin up (↑\uparrow) and down (↓\downarrow), k=(kx,ky)\mathbf{k}=(k_x,k_y) is the momentum defined relative to τzK\tau_zK point, m∗m^* is the effective mass, λw\lambda_w is the parameter of the trigonal warping of the Fermi surfaces, and ΔSOC(ℓ)\Delta_{\mathrm{SOC}}^{(\ell)} is the spin splitting. To capture the dependence on layer polarization, we take ΔSOC(ℓ)=ℓΔSOC\Delta_{\mathrm{SOC}}^{(\ell)}=\ell\Delta_{\mathrm{SOC}} where ℓ=+1\ell=+1 in Pup_{\rm up} structure and ℓ=−1\ell=-1 in Pdw_{\rm dw} structure. By fitting to the band structure, we obtain m∗≈0.46m0m^*\approx0.46m_0, λw≈7.5\lambda_w\approx7.5 meV⋅\cdotnm3^3, ΔSOC≈50\Delta_{\mathrm{SOC}}\approx50 meV, and EF≈−0.2E_{\rm F}\approx-0.2 eV, where m0m_0 is the free electron mass. The SOC induced spin splitting is much weaker in the Γ\Gamma valley, which we do not analyze in detail.

Superconductivity.— Since monolayer CuNb2_2Se4_4 is metallic, it can become superconducting at low temperatures. We note that intercalation of Cu atoms into bulk NbS2_2 [45] and NbSe2_2 [46] has been achieved experimentally, and superconductivity persists after the intercalation. Here we consider phonon mediated superconductivity for the monolayer Pup_{\rm up} (Pdw_{\rm dw}) structure and estimate its superconducting transition temperature TcT_{\rm c} based on the McMillan–Allen–Dynes formula [47, 48],

kBTc=ℏωlog⁡1.2exp⁡(−1.04(1+λ)λ−μ∗(1+0.62λ)),(2)k_{\rm B}T_{\rm c}=\frac{\hbar\omega_{\log}}{1.2}\exp\left(-\frac{1.04(1+\lambda)}{\lambda-\mu^*(1+0.62\lambda)}\right), \tag*{(2)}

where

ωlog⁡=exp⁡(2λ∫0∞dωα2F(ω)ωlog⁡ω),λ=2∫0∞α2F(ω)ωdω.(3)\begin{aligned} \omega_{\log} &= \exp\left(\frac{2}{\lambda}\int_{0}^{\infty}d\omega\frac{\alpha^{2}F(\omega)}{\omega}\log\omega\right),\\ \lambda&= 2\int_{0}^{\infty}\frac{\alpha^{2}F(\omega)}{\omega}d\omega. \tag*{(3)} \end{aligned}

Here ω\omega is the phonon frequency, ωlog⁡\omega_{\log} is the logarithmic average of the phonon frequencies, α2F(ω)\alpha^{2}F(\omega) is the Eliashberg function [49], λ\lambda is the electron-phonon coupling strength, and μ∗\mu^{*} is the parameter accounting for the Coulomb repulsion. We recalculate the phonon spectra using the density functional perturbation theory (DFPT) as coded in QUANTUM ESPRESSO [50]. The obtained phonon dispersion shown in Figure 3(a) is consistent with that calculated by PHONOPY code. The electron-phonon coupling is then calculated using “Electron-phonon Wannier” (EPW) code [51, 52], and the results are presented in Figure 3(b), where λ\lambda is found to be 1.22. Taking the empirical parameter μ∗\mu^{*} to be 0.15 [49], we obtain a TcT_{\mathrm{c}} of 3.04 K. This estimation is consistent with the experimental TcT_{\mathrm{c}} measured in electron-doped bilayer NbSe2_2 [53], which provides a strong support for our prediction of superconductivity in monolayer CuNb2_2Se4_4.

Phonon dispersion and electron-phonon interaction strength of monolayer CuNb$_2$Se$_4$

Figure 3. (a) Phonon dispersion of the monolayer CuNb2_2Se4_4 in Pup_{\rm up} (Pdw_{\rm dw}) structure. (b) Eliashberg function α2F(ω)\alpha^2F(\omega) (red line), and cumulative electron-phonon interaction strength λ(ω)\lambda(\omega) (blue line).

Superconducting diode effect

A ferroelectric superconductor lacks spatial inversion symmetry and supports SDE provided that time-reversal symmetry T^\hat{T} is also broken. To break the T^\hat{T} symmetry, we consider a minimal model with a spin-splitting term added to the Hamiltonian, H=H0+Δzσz\mathcal{H}=\mathcal{H}_0+\Delta_z\sigma_z, where H0\mathcal{H}_0 is given by Eq. (1). The Δzσz\Delta_z\sigma_z term can be induced by the proximity effect of a ferromagnetic insulator [54–57]. Here we take Δz\Delta_z as a phenomenological parameter in order to demonstrate the effect. A magnetic-field-free superconducting diode effect has recently been demonstrated in Ref. 56, where the time-reversal symmetry breaking for superconductors is generated through proximitized magnetization. This experiment [56] supports our proposal of using proximitized magnetization to induce superconducting diode effect.

To theoretically analyze the SDE, We focus on states in ±K\pm K valleys, which have strong SOC induced spin splittings and contribute most significantly to the SDE. We introduce an order parameter Δq,τz\Delta_{\mathbf{q},\tau_z} for intralayer pairing between (τz,↑)(\tau_z,\uparrow) and (−τz,↓)(-\tau_z,\downarrow) states, and q\mathbf{q} is the center-of-mass momentum of the Cooper pair. The free energy

per area for Δq,τz\Delta_{\mathbf{q},\tau_z}, derived in the SM [35], is given by

F[Δq,τz]=αq,τz∣Δq,τz∣2+β2∣Δq,τz∣4,αq,τz=α0+γτzq2+κτz(qx3−3qxqy2),α0=νT−TcTc,γτz=7ζ(3)ν(τzΔSOC(ℓ)−EF)ℏ24(πkBT)24m∗,κτz=−τzνλWΔz9ζ(5)(τzΔSOC(ℓ)−EF)216(πkBT)4.(4)\begin{aligned} F[\Delta_{\mathbf{q},\tau_z}] &= \alpha_{\mathbf{q},\tau_z}|\Delta_{\mathbf{q},\tau_z}|^{2}+\frac{\beta}{2}|\Delta_{\mathbf{q},\tau_z}|^{4},\\ \alpha_{\mathbf{q},\tau_z} &= \alpha_0+\gamma_{\tau_z}q^{2}+\kappa_{\tau_z}(q_x^{3}-3q_xq_y^{2}),\\ \alpha_0 &= \nu\frac{T-T_{\mathrm{c}}}{T_{\mathrm{c}}},\\ \gamma_{\tau_z} &= \frac{7\zeta(3)\nu(\tau_z\Delta_{\mathrm{SOC}}^{(\ell)}-E_{\mathrm{F}})\hbar^{2}}{4(\pi k_{\mathrm{B}}T)^{2}4m^{*}},\\ \kappa_{\tau_z} &= -\tau_z\nu\lambda_{\mathrm{W}}\Delta_z\frac{9\zeta(5)(\tau_z\Delta_{\mathrm{SOC}}^{(\ell)}-E_{\mathrm{F}})^{2}}{16(\pi k_{\mathrm{B}}T)^{4}}. \tag*{(4)} \end{aligned}

where TT is the temperature, ν\nu is the density of states per spin and per valley, ζ(n)\zeta(n) is the Riemann zeta function, and β=7ζ(3)ν/8(πkBT)2\beta=7\zeta(3)\nu/8(\pi k_{\mathrm{B}}T)^{2}. We minimize F[Δq,τz]F[\Delta_{\mathbf{q},\tau_z}] with respect to Δq,τz\Delta_{\mathbf{q},\tau_z}, and the free energy becomes

Fq,τz=−αq,τz22β.(5)F_{\mathbf{q},\tau_z}=-\frac{\alpha_{\mathbf{q},\tau_z}^{2}}{2\beta}. \tag*{(5)}

The supercurrent carried by the Cooper pairs with momentum q\mathbf{q} is calculated as

j=2eℏ∑τz∇qFq,τz=2eℏ1β∑τz∣αq,τz∣∇qαq,τz.(6)\mathbf{j}=\frac{2e}{\hbar}\sum_{\tau_z}\nabla_{\mathbf{q}}F_{\mathbf{q},\tau_z}=\frac{2e}{\hbar}\frac{1}{\beta}\sum_{\tau_z}|\alpha_{\mathbf{q},\tau_z}|\nabla_{\mathbf{q}}\alpha_{\mathbf{q},\tau_z}. \tag*{(6)}

where 2e<02e<0 is the charge of a Cooper pair, and therefore, j\mathbf{j} is antiparallel to q\mathbf{q}. We parametrize q\mathbf{q} as −q(cos⁡θ,sin⁡θ)-q(\cos\theta,\sin\theta). The critical current is obtained by maximizing ∣j∣|\mathbf{j}| with respect to q\mathbf{q} for q>0q>0, which leads to the following orientation dependence of the critical current

jc(θ)=j0(1+ηcos⁡3θ),(7)j_{\mathrm{c}}(\theta)=j_0(1+\eta\cos3\theta), \tag*{(7)}

where j0j_0 is the isotropic part of the critical current. In Eq. (7), the nonreciprocal factor η\eta is given by

η=−∣α0∣3∑τzκτz/γτz∑τzγτz≈6731ζ(5)7πζ(3)3/2λWkF3ΔSOC(ℓ)∣EF∣2ΔzkBTTc−TTc≈0.04ℓΔzkBTTc−TTc.(8)\begin{aligned} \eta&= -\sqrt{\frac{|\alpha_0|}{3}}\frac{\sum_{\tau_z}\kappa_{\tau_z}/\gamma_{\tau_z}}{\sum_{\tau_z}\sqrt{\gamma_{\tau_z}}}\\ &\approx\sqrt{\frac{6}{7}}\frac{31\zeta(5)}{7\pi\zeta(3)^{3/2}}\frac{\lambda_{\mathrm{W}}k_{\mathrm{F}}^{3}\Delta_{\mathrm{SOC}}^{(\ell)}}{|E_{\mathrm{F}}|^{2}}\frac{\Delta_z}{k_{\mathrm{B}}T}\sqrt{\frac{T_{\mathrm{c}}-T}{T_{\mathrm{c}}}}\\ &\approx0.04\ell\frac{\Delta_z}{k_{\mathrm{B}}T}\sqrt{\frac{T_{\mathrm{c}}-T}{T_{\mathrm{c}}}}. \tag*{(8)} \end{aligned}

where kFk_F is defined as 2m∗∣EF∣/ℏ\sqrt{2m^{*}|E_F|}/\hbar. The factor η\eta is proportional to ΔSOC(ℓ)\Delta_{\mathrm{SOC}}^{(\ell)}, Δz\Delta_z and λw\lambda_w, which characterize, respectively, SOC induced spin splittings due to inversion symmetry breaking, spin splittings due to time-reversal symmetry breaking, and trigonal warping. The critical current is generally different for opposite directions since jc(θ)≠jc(θ+π)j_c(\theta) \ne j_c(\theta+\pi) for a generic θ\theta, as shown in Figure 4. We assume Δz\Delta_z is positive for definiteness in the following. In the Pup_{\mathrm{up}} state, ℓ=+1\ell=+1, which renders η>0\eta>0 and jc(0)>jc(π)j_c(0)>j_c(\pi); therefore, the system is superconducting for a current with a magnitude in the range of j0(1−∣η∣,1+∣η∣)j_0(1-|\eta|,1+|\eta|) along +x^+\hat{x} direction, but resistive along −x^-\hat{x} direction, which leads to the SDE. By contrast, ℓ=−1\ell=-1 and η<0\eta<0 in the Pdw_{\mathrm{dw}} state; the superconducting direction is changed to −x^-\hat{x} direction for a current with a magnitude in the same range. Therefore, the SDE is reversed upon ferroelectric reversal. As shown by Eq. (8), η\eta increases with decreasing TT. If we take Δz=kBT=0.1kBTc\Delta_z=k_BT=0.1k_BT_c, η≈0.04ℓ\eta\approx0.04\ell, which represents an experimentally measurable effect [7].

Angle dependence of the critical current in the Pup and Pdw states

Figure 4. The angle dependence of the critical current in (a) Pup_{\mathrm{up}} state and (b) Pdw_{\mathrm{dw}} state. The nonreciprocal factor η\eta is taken to be 0.04ℓ0.04\ell, where ℓ\ell is +1+1 for Pup_{\mathrm{up}} and −1-1 for Pdw_{\mathrm{dw}}, respectively.

Conclusions.— In summary, we have proposed a microscopic mechanism for ferroelectric reversible SDE using monolayer CuNb2_2Se4_4 as a model system. In addition to CuNb2_2Se4_4, we expect ferroelectricity can widely exist in 2D superconductors. In particular, sliding ferroelectricity has been shown to be ubiquitous in 2D vdW stacked layers, where the layer polarization can be switched by the in-plane interlayer sliding [21, 58–63]. Thus, a superconducting vdW bilayer can naturally host coexisting superconductivity and ferroelectricity. A promising candidate is bilayer MoTe2_2 in the TdT_d structure, which carries the sliding ferroelectricity and becomes superconducting at Tc∼2T_c\sim2 K [64]. The bilayer TdT_d–MoTe2_2 has Rashba spin splittings [65], and an in-plane magnetic field can effectively break the time-reversal symmetry. As in CuNb2_2Se4_4, ferroelectricity controls the sign of the Rashba spin splittings as well as the SDE in the bilayer TdT_d–MoTe2_2. With the recent rapid developments in the study of 2D ferroelectric materials and nonreciprocal superconducting transport, we anticipate that our proposed ferroelectric reversible SDE should soon be experimentally realizable. In a broader prospective, our work establishes a new type of superconductors, in which ferroelectricity acts as a tuning knob in controlling the superconductivity properties. Further theoretical, computational and experimental works are expected to substantially broaden the material candidates and device functionalities of ferroelectric superconductors.

Acknowledgments.— This work is supported by National Key R&D Program of China (No. 2018YFA0703700, 2021YFA1401300), the National Natural Science Foundation of China (Nos. 91964203, 62104171, 62104172 and 62004142), the Strategic Priority Research Program of Chinese Academy of Sciences (No. XDB44000000), the Natural Science Foundation of Hubei Province, China (No. 2021CFB037), and the Fundamental Research Funds for the Central Universities (No. 2042021kf0067). The numerical calculations in this paper have been done on the supercomputing system in the Supercomputing Center of Wuhan University.

Note added. Controlling of superconductivity through ferroelectricity has recently been demonstrated in twisted bilayer graphene aligned with hBN [66].

Supplemenatry Materials for “Prediction of Ferroelectric Superconductors with Reversible Superconducting Diode Effect”

Baoxing Zhai; Bohao Li; Yao Wen; Fengcheng Wu; Jun He

This Supplemental Material has two sections. Section specifies the first-principles computational methods. Section presents the derivation of the free energy for the superconductivity.

First-Principles Computational Methods

For the studies of total energy and electronic structures, we perform the density functional theory (DFT) calculations using the Vienna ab-initio simulation package (VASP). The Perdew-Burke-Ernzerhof (PBE) exchange correlation function in the generalized gradient approximation (GGA) is used [1]. The electron-ion potential is described by the projected augmented wave [2]. The kinetic energy cutoff is set to be 500 eV for the plane wave expansion. The Brillouin zone integration is carried out using 9×9×19 \times9 \times1 Monkhorst-Pack kk-mesh for geometry optimization of CuNb2Se4\mathrm{CuNb}_{2}\mathrm{Se}_{4} monolayer [3]. All geometric structures are fully relaxed until energy and forces are converged to 10−610^{-6} eV and 0.01 eV/Å, respectively. The spin-orbital coupling effect is included in the calculations. Moreover, the thickness of vacuum region is set to be larger than 20 Å to avoid spurious interlayer interaction. The above calculation procedure generates the energy landscape shown in Figure 1(b) of the main text. To further test the accuracy of the results, we reperform the above calculation in QUANTUM ESPRESSO using a truncated out of plane Coulomb interaction [4] to prevent interaction between periodic images. The obtained results, as shown in Figure S1, are consistent with Figure 1(b) of the main text, indicating that our results are reliable.

Energy per formula unit of the eight high-symmetry structures of monolayer CuNb₂Se₄

Figure S1. The energy (relative to that of structure VI) per formula unit of the eight high-symmetry structures of monolayer CuNb2_2Se4_4. The calculation is done using QUANTUM ESPRESSO assuming isolated 2D layer.

Based on the VASP results, we further calculate the phonon dispersion of monolayer CuNb2Se4\mathrm{CuNb}_{2}\mathrm{Se}_{4} with a 4×4×14 \times4 \times1 supercell using the PHONOPY package. The phonon spectra are shown in Figure S2. The thermal stability of monolayer CuNb2Se4\mathrm{CuNb}_{2}\mathrm{Se}_{4} is examined by ab initio molecular dynamics (AIMD) simulations with a 4×4×14 \times4 \times1 supercell at 300 K with a time step of 1 fs for 5 ps, as shown in Figure S2.

Evolution of total energy and phonon dispersions for monolayer CuNb₂Se₄

Figure S2. (a) Evolution of total energy for a 4×4×14 \times4 \times1 supercell as a function of simulation time in AIMD simulations for monolayer CuNb2_2Se4_4 at 300 K. The top and side views of monolayer CuNb2_2Se4_4 at the end of AIMD simulations are shown in (a). (b) Phonon dispersions for monolayer CuNb2_2Se4_4 calculated using the PHONOPY package.

To study the superconducting properties of monolayer CuNb2Se4\mathrm{CuNb}_{2}\mathrm{Se}_{4}, we start by computing the charge density of ground state with the density functional theory as implemented in the QUANTUM ESPRESSO code. We adopt the generalized gradient approximation of Perdew-Burke-Ernzerhof parametrization for the exchange-correlation functional. The plane wave cutoff energy is 60 Ry. A uniform unshifted 16×16×116 \times16 \times1 kk-point mesh is used in the Brillouin zone integrations and a Gaussian smearing of 0.01 Ry has been adopted to deal with the metallic character of the material. The phonon dispersion is obtained by Fourier interpolation of the dynamical matrices computed on a 4×4×14 \times4 \times1 qq-point mesh. For the electron-phonon coupling (EPC) coefficient, we employ a Wannier interpolation as implemented in the EPW code; Cu dd, Nb dd, and Se pp orbitals are used for projection in the construction of the Wannier function. We solve the Migdal-Eliashberg equations in the anisotropic approximations to obtain the superconducting gap, and we use 128×128×1128 \times128 \times1 for the kk- and qq-point grids. A 0.2 eV cutoff for the Matsubara frequency is chosen (roughly five times the largest phonon frequency). Smearing in the energy-conserving delta functions is 0.05 eV and smearing for sum over qq in the EPC is 0.05 meV. A value of 0.15 is used for the screened Coulomb parameter μ∗\mu^{*}.

Ferroelectricity in monolayer CuNb2Se4\mathrm{CuNb}_{2}\mathrm{Se}_{4} is analyzed in Figure S3 and S4. In Figure S3, we present the Bader charge analysis of the Pdw_{\mathrm{dw}} and Pup_{\mathrm{up}} structure. In Figure S4, the electrostatic potential difference across the monolayer CuNb2Se4\mathrm{CuNb}_{2}\mathrm{Se}_{4} is shown, which further confirms the out-of-plane electric polarization.

Calculated charge transfer for monolayer CuNb$_2$Se$_4$ in (a) $P_{\mathrm{dw}}$ and (b) $P_{\mathrm{up}}$ states

Figure S3. The calculated charge transfer for the monolayer CuNb2_2Se4_4 in (a) PdwP_{\mathrm{dw}} and (b) PupP_{\mathrm{up}} states by Bader charge analysis. The numbers listed on the right hand side of the structure is the actual number of valence electrons of each atom. By contrast, the intrinsic valence electrons of Cu, Nb and Se atoms in our pseudopotential file are 11, 11 and 6, respectively.

Electrostatic potential of monolayer CuNb$_2$Se$_4$

Figure S4. (a) The electrostatic potential of monolayer CuNb2_2Se4_4. The blue and red line represents PdwP_{\mathrm{dw}} and PupP_{\mathrm{up}} state, respectively. (b) A zoom-in view of the blue region of Figure S3 (a).

Derivation of free energy for superconductivity

We present the derivation of the free energy for the superconducting order parameters. The single-particle Hamiltonian including the Zeeman term is

Hk,σz,τz=−ℏ2k22m∗+λw(kx3−3kxky2)τz+ΔSOC(ℓ)τzσz+Δzσz−EF(9)\mathcal{H}_{k,\sigma_z,\tau_z}=-\frac{\hbar^2 k^2}{2m^*}+\lambda_w(k_x^3-3k_xk_y^2)\tau_z+\Delta_{\mathrm{SOC}}^{(\ell)}\tau_z\sigma_z+\Delta_z\sigma_z-E_F \tag*{(9)}

where τz=±\tau_z=\pm is the valley index and σz=±\sigma_z=\pm represents spin up (↑\uparrow) and down (↓\downarrow). We also define ϵk(0)=−ℏ2k22m∗\epsilon_k^{(0)}=-\frac{\hbar^2 k^2}{2m^*}, fk=kx3−3kxky2f_k=k_x^3-3k_xk_y^2, and ϵk=ϵk(0)+λwτzfk−EF\epsilon_k=\epsilon_k^{(0)}+\lambda_w\tau_z f_k-E_F for convenience. We assume a local attractive interaction given by

Hp=−g∑τz∫d2r ψ↑,τz†(r)ψ↓,−τz†(r)ψ↓,−τz(r)ψ↑,τz(r)(10)H_p=-g\sum_{\tau_z}\int d^2r\,\psi_{\uparrow,\tau_z}^{\dagger}(r)\psi_{\downarrow,-\tau_z}^{\dagger}(r)\psi_{\downarrow,-\tau_z}(r)\psi_{\uparrow,\tau_z}(r) \tag*{(10)}

where g>0g>0 characterizes the attractive interaction strength. We introduce an order parameter Δq,τz\Delta_{q,\tau_z} for pairing between (τz,↑)(\tau_z,\uparrow) and (−τz,↓)(-\tau_z,\downarrow) states, and q\boldsymbol{q} is the center-of-mass momentum of the Cooper pair. The free energy per area derived using path integral formalism is

F=∑q,τzF[Δq,τz],\mathcal{F}=\sum_{q,\tau_z}\mathcal{F}[\Delta_{q,\tau_z}],
F[Δq,τz]=αq,τz∣Δq,τz∣2+β2∣Δq,τz∣4,\mathcal{F}[\Delta_{q,\tau_z}]=\alpha_{q,\tau_z}|\Delta_{q,\tau_z}|^2+\frac{\beta}{2}|\Delta_{q,\tau_z}|^4,
αq,τz=1g+kBTA∑k,nGk+q,↑,τz,ωn(e)Gk,↓,−τz,ωn(h),\alpha_{q,\tau_z}=\frac{1}{g}+\frac{k_{\mathrm{B}}T}{\mathcal{A}}\sum_{k,n}G^{(e)}_{k+q,\uparrow,\tau_z,\omega_n}G^{(h)}_{k,\downarrow,-\tau_z,\omega_n},
β≈kBTA∑k,n∣Gk,↑,τz,ωn(e)Gk,↓,−τz,ωn(h)∣2(11)\beta\approx\frac{k_{\mathrm{B}}T}{\mathcal{A}}\sum_{k,n}\left|G^{(e)}_{k,\uparrow,\tau_z,\omega_n}G^{(h)}_{k,\downarrow,-\tau_z,\omega_n}\right|^2 \tag*{(11)}

where TT is the temperature and A\mathcal{A} is the total area of the system. In Eq. (11), the electron and hole Green’s functions are as follows

Gk,↑,τz,ωn(e)=1iωn−Hk,↑,τz,Gk,↓,−τz,ωn(h)=1iωn+H−k,↓,−τz,(12)G_{\mathbf{k},\uparrow,\tau_z,\omega_n}^{(e)}=\frac{1}{i\omega_n-H_{\mathbf{k},\uparrow,\tau_z}}, \qquad G_{\mathbf{k},\downarrow,-\tau_z,\omega_n}^{(h)}=\frac{1}{i\omega_n+H_{-\mathbf{k},\downarrow,-\tau_z}}, \tag*{(12)}

where ωn=(2n+1)πkBT\omega_n=(2n+1)\pi k_{\mathrm{B}}T is the Matsubara frequency.

For the coefficient αq,τz\alpha_{\mathbf{q},\tau_z}, we keep terms up to the third order in q\mathbf{q},

αq,τz≈α0+α2+α3.(13)\alpha_{\mathbf{q},\tau_z}\approx\alpha_0+\alpha_2+\alpha_3. \tag*{(13)}

α0\alpha_0 is zeroth order in q\mathbf{q} and is given by

α0=1g+kBTA∑k,nGk,↑,τz,ωn(e)Gk,↓,−τz,ωn(h)=1g−ν∫−ωDωDdϵtanh⁡[ϵ/(2kBT)]2ϵ≈νT−TcTc(14)\begin{aligned} \alpha_0&=\frac{1}{g}+\frac{k_{\mathrm{B}}T}{\mathcal{A}}\sum_{\mathbf{k},n}G_{\mathbf{k},\uparrow,\tau_z,\omega_n}^{(e)}G_{\mathbf{k},\downarrow,-\tau_z,\omega_n}^{(h)}\\ &=\frac{1}{g}-\nu\int_{-\omega_D}^{\omega_D}d\epsilon\frac{\tanh[\epsilon/(2k_{\mathrm{B}}T)]}{2\epsilon}\\ &\approx\nu\frac{T-T_c}{T_c} \tag*{(14)} \end{aligned}

where ν\nu is the density of states per spin and per valley.

α2\alpha_2 is second order in q\mathbf{q} and is calculated to be

α2=kBTA∑k,n12Gk,↓,−τz,ωn(h)(q⋅∇k)2Gk,↑,τz,ωn(e)=−kBT2A∑k,n(q⋅∇k)Gk,↓,−τz,ωn(h)(q⋅∇k)Gk,↑,τz,ωn(e)=kBT2A∑k,nqiqj∂ϵk∂ki∂ϵk∂kj[Gk,↓,−τz,ωn(h)]2[Gk,↑,τz,ωn(e)]2≈kBTAϵq0∑k,nϵk0[gk,↓,−τz,ωn(h)]2[gk,↑,τz,ωn(e)]2=kBTνϵq0(EF−τzΔSOC(ℓ))∑nπ2∣ωn∣3=7ζ(3)16ℏ2ν(τzΔSOC(ℓ)−EF)m∗(πkBTc)2q2,(15)\begin{aligned} \alpha_2&=\frac{k_{\mathrm{B}}T}{\mathcal{A}}\sum_{\mathbf{k},n}\frac{1}{2}G_{\mathbf{k},\downarrow,-\tau_z,\omega_n}^{(h)}(\mathbf{q}\cdot\nabla_{\mathbf{k}})^2G_{\mathbf{k},\uparrow,\tau_z,\omega_n}^{(e)}\\ &=-\frac{k_{\mathrm{B}}T}{2\mathcal{A}}\sum_{\mathbf{k},n}(\mathbf{q}\cdot\nabla_{\mathbf{k}})G_{\mathbf{k},\downarrow,-\tau_z,\omega_n}^{(h)}(\mathbf{q}\cdot\nabla_{\mathbf{k}})G_{\mathbf{k},\uparrow,\tau_z,\omega_n}^{(e)}\\ &=\frac{k_{\mathrm{B}}T}{2\mathcal{A}}\sum_{\mathbf{k},n}q_iq_j\frac{\partial\epsilon_{\mathbf{k}}}{\partial k_i}\frac{\partial\epsilon_{\mathbf{k}}}{\partial k_j}\left[G_{\mathbf{k},\downarrow,-\tau_z,\omega_n}^{(h)}\right]^2\left[G_{\mathbf{k},\uparrow,\tau_z,\omega_n}^{(e)}\right]^2\\ &\approx\frac{k_{\mathrm{B}}T}{\mathcal{A}}\epsilon_{\mathbf{q}}^0\sum_{\mathbf{k},n}\epsilon_{\mathbf{k}}^0\left[g_{\mathbf{k},\downarrow,-\tau_z,\omega_n}^{(h)}\right]^2\left[g_{\mathbf{k},\uparrow,\tau_z,\omega_n}^{(e)}\right]^2\\ &=k_{\mathrm{B}}T\nu\epsilon_{\mathbf{q}}^0(E_F-\tau_z\Delta_{\mathrm{SOC}}^{(\ell)})\sum_n\frac{\pi}{2\lvert\omega_n\rvert^3}\\ &=\frac{7\zeta(3)}{16}\frac{\hbar^2\nu(\tau_z\Delta_{\mathrm{SOC}}^{(\ell)}-E_F)}{m^*(\pi k_{\mathrm{B}}T_c)^2}q^2, \tag*{(15)} \end{aligned}

where g(e)g^{(e)} and g(h)g^{(h)} are Green’s functions without the trigonal warping term.

α3\alpha_3 is third order in q\mathbf{q} and is found to be proportional to λw\lambda_w and Δz\Delta_z,

α3=kBTA∑k,n16Gk,↓,−τz,ωn(h)(q⋅∇)3Gk,↑,τz,ωn(e)=kBTA∑k,nGk,↓,−τz,ωn(h)qiqjqm{16∂3ϵk∂ki∂kj∂km[Gk,↑,τz,ωn(e)]2+∂2ϵk∂ki∂kj∂ϵk∂km[Gk,↑,τz,ωn(e)]3+∂ϵk∂ki∂ϵk∂kj∂ϵk∂km[Gk,↑,τz,ωn(e)]4}=−πkBTτzλwΔzνfq∑n(12∣ωn∣3+3(EF−τzΔSOC(ℓ))2∣ωn∣5)=−πkBTτzλwΔzνfq[7ζ(3)81(πkBT)3+93ζ(5)16(EF−τzΔSOC(ℓ))2(πkBT)5]≈−τzλwΔzνfq93ζ(5)16(EF−τzΔSOC(ℓ))2(πkBTc)4.\begin{aligned} \alpha_3&=\frac{k_{\mathrm{B}}T}{\mathcal{A}}\sum_{\mathbf{k},n}\frac{1}{6}G_{\mathbf{k},\downarrow,-\tau_z,\omega_n}^{(h)}(\mathbf{q}\cdot\nabla)^3G_{\mathbf{k},\uparrow,\tau_z,\omega_n}^{(e)}\\ &=\frac{k_{\mathrm{B}}T}{\mathcal{A}}\sum_{\mathbf{k},n}G_{\mathbf{k},\downarrow,-\tau_z,\omega_n}^{(h)}q_iq_jq_m\left\{\frac{1}{6}\frac{\partial^3\epsilon_{\mathbf{k}}}{\partial k_i\partial k_j\partial k_m}\left[G_{\mathbf{k},\uparrow,\tau_z,\omega_n}^{(e)}\right]^2+\frac{\partial^2\epsilon_{\mathbf{k}}}{\partial k_i\partial k_j}\frac{\partial\epsilon_{\mathbf{k}}}{\partial k_m}\left[G_{\mathbf{k},\uparrow,\tau_z,\omega_n}^{(e)}\right]^3+\frac{\partial\epsilon_{\mathbf{k}}}{\partial k_i}\frac{\partial\epsilon_{\mathbf{k}}}{\partial k_j}\frac{\partial\epsilon_{\mathbf{k}}}{\partial k_m}\left[G_{\mathbf{k},\uparrow,\tau_z,\omega_n}^{(e)}\right]^4\right\}\\ &=-\pi k_{\mathrm{B}}T\tau_z\lambda_w\Delta_z\nu f_q\sum_n\left(\frac{1}{2\lvert\omega_n\rvert^3}+\frac{3(E_F-\tau_z\Delta_{\mathrm{SOC}}^{(\ell)})^2}{\lvert\omega_n\rvert^5}\right)\\ &=-\pi k_{\mathrm{B}}T\tau_z\lambda_w\Delta_z\nu f_q\left[\frac{7\zeta(3)}{8}\frac{1}{(\pi k_{\mathrm{B}}T)^3}+\frac{93\zeta(5)}{16}\frac{(E_F-\tau_z\Delta_{\mathrm{SOC}}^{(\ell)})^2}{(\pi k_{\mathrm{B}}T)^5}\right]\\ &\approx-\tau_z\lambda_w\Delta_z\nu f_q\frac{93\zeta(5)}{16}\frac{(E_F-\tau_z\Delta_{\mathrm{SOC}}^{(\ell)})^2}{(\pi k_{\mathrm{B}}T_c)^4}. \end{aligned}

Finally, the coefficient β\beta is calculated to be

β≈kBTA∑k,ωn[Gk,↑,τz,ωn(e)Gk,↓,−τz,ωn(h)]2≈νkBT∑n∫−∞+∞dϵ1(iωn−ϵ)2(iωn+ϵ)2=νkBT∑nπ2∣ωn∣3=ν7ζ(3)8(πkBT)2.(16)\begin{aligned} \beta&\approx\frac{k_B T}{A}\sum_{k,\omega_n}\left[G^{(e)}_{k,\uparrow,\tau_z,\omega_n}G^{(h)}_{k,\downarrow,-\tau_z,\omega_n}\right]^2 \\ &\approx\nu k_B T\sum_n\int_{-\infty}^{+\infty}\mathrm{d}\epsilon\frac{1}{(i\omega_n-\epsilon)^2(i\omega_n+\epsilon)^2}\\ &=\nu k_B T\sum_n\frac{\pi}{2\lvert\omega_n\rvert^3}\\ &=\nu\frac{7\zeta(3)}{8(\pi k_B T)^2}. \tag*{(16)} \end{aligned}

In summary, the free energy per area is given by

F[Δq,τz]=αq,τz∣Δq,τz∣2+β2∣Δq,τz∣4,αq,τz=α0+γτzq2+κτz(qx3−3qxqy2),α0=νT−TcTc,γτz=7ζ(3)4ν(τzΔSOC(ℓ)−EF)(πkBT)2ℏ24m∗,κτz=−τzνλwΔz93ζ(5)16(τzΔSOC(ℓ)−EF)2(πkBT)4,β=ν7ζ(3)8(πkBT)2.(17)\begin{aligned} \mathcal{F}[\Delta_{q,\tau_z}]&=\alpha_{q,\tau_z}\lvert\Delta_{q,\tau_z}\rvert^2+\frac{\beta}{2}\lvert\Delta_{q,\tau_z}\rvert^4,\\ \alpha_{q,\tau_z}&=\alpha_0+\gamma_{\tau_z}q^2+\kappa_{\tau_z}(q_x^3-3q_xq_y^2),\\ \alpha_0&=\nu\frac{T-T_c}{T_c},\\ \gamma_{\tau_z}&=\frac{7\zeta(3)}{4}\frac{\nu(\tau_z\Delta_{\mathrm{SOC}}^{(\ell)}-E_F)}{(\pi k_B T)^2}\frac{\hbar^2}{4m^*},\\ \kappa_{\tau_z}&=-\tau_z\nu\lambda_w\Delta_z\frac{93\zeta(5)}{16}\frac{(\tau_z\Delta_{\mathrm{SOC}}^{(\ell)}-E_F)^2}{(\pi k_B T)^4},\\ \beta&=\nu\frac{7\zeta(3)}{8(\pi k_B T)^2}. \tag*{(17)} \end{aligned}

In the calculation of F\mathcal{F}, we keep terms up to the third order of qq, the first order of λw\lambda_w and Δz\Delta_z, and the fourth order of ∣Δq,τz∣\lvert\Delta_{q,\tau_z}\rvert.

In the above analysis, the Zeeman term is used to break the time-reversal symmetry and could be generated from proximity effect of a magnetic insulator. This magnetic proximity effect has been well established in 2D heterostructures. For example, the valley Zeeman term has been demonstrated using first-principles band structure calculations in monolayer MoTe2_2 on a EuO substrate [5]. Such magnetic proximity effect is not due to the magnetic field generated by the magnetic insulator, but due to electron tunneling between the two materials. Namely, the proximitized magnetization arises from band structure effect and therefore, survives even in the superconducting state. We note that signatures of time-reversal symmetry breaking in superconducting NbSe2_2 from proximitized magnetization has been demonstrated in NbSe2_2/CrBr3_3 heterostructure [6], where CrBr3_3 is a magnetic insulator with out-of-plane magnetizations.

A magnetic-field-free superconducting diode effect has recently been demonstrated in Ref. [7] using noncentrosymmetric [Nb/V/Co/V/Ta]20_{20} multilayers, where the magnetic layers generate time-reversal symmetry breaking for the superconducting layers through proximitized magnetization. This experiment [7] supports our proposal of using proximitized magnetization to induce superconducting diode effect.

We also note that an out-of-plane magnetic field can also break the time-reversal symmetry and produce the superconducting diode effect, as demonstrated in few-layer NbSe2_2 in Ref. [8]. The mechanism of the superconducting diode effect induced by an out-of-plane magnetic field is likely not fully captured by the free energy in Eq. (17), since the magnetic field generates not only Zeeman effect but also orbital effect. Nevertheless, we expect that the out-of-plane magnetic field can enable the reversible superconducting diode effect in our case of ferroelectric superconductor CuNb2_2Se4_4.

References

  1. [3]H. J. Monkhorst, and J. D. Pack, Special points for brillouin-zone integrations, Phys. Rev. B 13, 5188 (1976).DOI
  2. [4]T. Sohier, M. Calandra, and F. Mauri, Density functional perturbation theory for gated two-dimensional heterostructures: Theoretical developments and application to flexural phonons in graphene, Phys. Rev. B 96, 075448 (2017).DOI
  3. [5]J. Qi, X. Li, Q. Niu, and J. Feng, Giant and tunable valley degeneracy splitting in MoTe2, Phys. Rev. B 92, 121403 (2015).DOI
  4. [6]S. Kezilebieke, M. N. Huda, V. Vänö, M. Aapro, S. C. Ganguli, O. J. Silveira, S. Glodzik, A. S. Foster, T. Ojanen, and P. Liljeroth, Topological superconductivity in a van der Waals heterostructure, Nature 588, 424 (2020).
  5. [7]H. Narita, J. Ishizuka, R. Kawarazaki, D. Kan, Y. Shiota, T. Moriyama, Y. Shimakawa, A. V. Ognjev, A. S. Samardak, Y. Yanase, and T. Ono, Field-free superconducting diode effect in noncentrosymmetric superconductor/ferromagnet multilayers, Nat. Nanotechnol. 17, 823 (2022).DOI
  6. [8]L. Bauriedl, C. Bäuml, L. Fuchs, C. Baumgartner, N. Paulik, J. M. Bauer, K.-Q. Lin, J. M. Lupton, T. Taniguchi, K. Watanabe, C. Strunk, and N. Paradiso, Supercurrent diode effect and magnetochiral anisotropy in few-layer NbSe2, Nat. Commun. 13, 4266 (2022).DOI

Paper details

vlnc:W_f5tuz3tz
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

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