betaFeedback

WorksPhysics and Astronomy

Equal-spin pairing at finite Cooper-pair momentum in pp-wave magnets

Abstract

The realization of Fulde-Ferrell (FF) superconducting states has long been pursued but remains challenging because of the large Zeeman splitting required. Here, we demonstrate that pp-wave magnets can host equal-spin finite-pair-momentum states with pp-wave symmetry, termed FF states, at arbitrarily weak fields. Our Bogoliubov-de Gennes analysis shows that, owing to the displacement of the Fermi surface, two equal-spin electrons in the FF states pair at the same energy, just as in the opposite-spin zero-pair-momentum states of the same gap symmetry. At zero field, the FF states are therefore exactly degenerate with these zero-pair-momentum counterparts. An arbitrarily weak Zeeman field lifts this degeneracy and makes the FF states the superconducting ground state, since they remain stable owing to their equal-spin nature, whereas their counterparts are suppressed by the formation of Bogoliubov Fermi surfaces. These findings identify the fermiology of pp-wave magnets as a promising route toward long-sought finite-momentum superconductivity.

Introduction

Finite-momentum Cooper pairing—in which the Cooper pair carries a nonzero center-of-mass momentum—has been a longstanding pursuit in the study of unconventional superconductivity [1, 2]. The paradigmatic example is the Fulde–Ferrell–Larkin–Ovchinnikov state, in which pairing of two electrons having nonzero net momentum is stabilized by a Zeeman-split Fermi surface (FS), allowing electrons of opposite spin to pair while conserving energy [3, 4]. To realize such finite-momentum pairing, researchers have conventionally considered an external magnetic field or a ferromagnetic exchange field [5, 6, 7], both of which are accompanied by a finite net magnetization that limits their practical utility, for instance by inducing vortices via orbital coupling to the applied field [8].

Field-free routes to finite-momentum pairing have also been explored, most notably as a pair-density wave (PDW) intertwined with stripe order in doped Mott insulators [9, 10], or as an Amperian pairing mediated by gauge-field fluctuations [11]; however, these mechanisms rely on strong correlations or proximity to a stripe-ordered state, rather than providing a generic route accessible in a clean, weakly correlated metal.

Recently, unconventional antiferromagnets have emerged as a novel platform for realizing finite-momentum pairing states without requiring a net magnetization [12, 13], potentially circumventing the constraints faced by the mechanisms above. This class of collinear magnets combines zero net magnetization with momentum-dependent spin splitting, a combination that has recently been experimentally confirmed in several materials [14, 15, 16, 17, 18, 19, 20, 21]. Such spin splitting can suppress the conventional ss-wave spin-singlet channel, opening a pathway toward finite-momentum superconducting states. On the weak-coupling side, a series of theoretical studies established that the spin splitting alone can stabilize finite-momentum Cooper pairing, whether induced by the proximity [22], intrinsic attractive interactions [23, 24], or an exchange-field-generated Bogoliubov FS [25], with follow-up works [26, 27]. Beyond this field-mimic picture, interaction-driven approaches have also identified finite-momentum pairing, arising from spin fluctuations in a weakly correlated Hubbard model [28] and from strong correlations in a doped Mott insulator [29]. Recently identified odd-parity magnets—pp-wave magnets [30, 31, 32] and ff-wave magnets [33, 34]—are similarly expected to offer a possible route toward finite-momentum pairing via the momentum-dependent spin splitting induced by the exchange field. Despite this potential, existing studies have largely focused on superconducting phenomena within the scope of zero-momentum Cooper pairing [35, 36, 37, 38, 39, 40], and finite-momentum pairing states in odd-parity magnets remain elusive.

In this work, we establish a finite-momentum equal-spin pairing instability intrinsic to a two-dimensional pp-wave magnet on a square lattice. There, the pxp_x-wave exchange field displaces the two spin-split FSs in opposite directions along kxk_x [38, 39]. We introduce equal-spin, odd-parity finite-momentum pairing states whose pairing momentum is set by this displacement, hereafter referred to as “Fulde–Ferrell (FF) states,” although the original FF state denotes pairing between electrons of opposite spin with even-parity form factors [3]. Our Bogoliubov–de Gennes (BdG) analysis shows that two electrons of the same spin are paired at the same energy at every point of the FS Figure 1(a), which maximizes the pairing amplitude and makes the FF states degenerate with the relevant zero-momentum pairing states Figure 1(b). Applying a Zeeman field then lowers the energy of the FF states below that of the zero-momentum pairing states, stabilizing the FF states as the energetically favored superconducting phases. We finally show that the quasiparticle energy dispersions under a Zeeman field provide an experimental probe that distinguishes the FF states from the zero-momentum pairing states, which are identical at zero field.

Schematic illustration of the pairing channels

Figure 1. Schematic illustration of the pairing channels. (a) Equal-spin finite-momentum pairing in the Fulde–Ferrell (FF) states. The red and blue lines represent the up- and down-spin FSs, respectively, each shifted by +Q2+\frac{Q}{2} and −Q2-\frac{Q}{2}, with Q=Qx^\mathbf{Q}=Q\hat{x} and Q=−2tan⁡−1(Jt)Q=-2\tan^{-1}\left(\frac{J}{t}\right). The vectors k+Q2\mathbf{k}+\frac{\mathbf{Q}}{2} and −k+Q2-\mathbf{k}+\frac{\mathbf{Q}}{2} (k−Q2\mathbf{k}-\frac{\mathbf{Q}}{2} and −k−Q2-\mathbf{k}-\frac{\mathbf{Q}}{2}) denote the momenta of the two paired spin-up (spin-down) electrons, with the pair momentum Q\mathbf{Q} (−Q-\mathbf{Q}). (b) Inter-spin zero-momentum pairing, shown on the same FSs as in (a). The parameters used are J=1J=1 and μ=−1\mu=-1.

Model

Extended attractive Hubbard model

We consider an effective two-band model on a square lattice for pp-wave magnets [38, 39],

H^=H^0+H^int,\hat{H}=\hat{H}_0+\hat{H}_{\mathrm{int}},
H^0=−t∑σ∑⟨i,j⟩(ciσ†cjσ+cjσ†ciσ)+J2∑σ∑isσ(ciσ†ci+x,σ−ci+x,σ†ciσ)−h∑σ∑isσniσ−μ∑σ∑iniσ,(1)\hat{H}_0=-t\sum_{\sigma}\sum_{\langle i,j\rangle}\left(c_{i\sigma}^{\dagger}c_{j\sigma}+c_{j\sigma}^{\dagger}c_{i\sigma}\right) +\frac{J}{2}\sum_{\sigma}\sum_i s_{\sigma}\left(c_{i\sigma}^{\dagger}c_{i+x,\sigma}-c_{i+x,\sigma}^{\dagger}c_{i\sigma}\right) -h\sum_{\sigma}\sum_i s_{\sigma}n_{i\sigma}-\mu\sum_{\sigma}\sum_i n_{i\sigma}, \tag*{(1)}
H^int=−U∑ini↑ni↓−V∑⟨i,j⟩∑σ,σ′niσnjσ′,\hat{H}_{\mathrm{int}}=-U\sum_i n_{i\uparrow}n_{i\downarrow}-V\sum_{\langle i,j\rangle}\sum_{\sigma,\sigma'}n_{i\sigma}n_{j\sigma'},

where ciσ†c_{i\sigma}^{\dagger} (ciσc_{i\sigma}) creates (annihilates) an electron at site ii with spin σ\sigma, and niσ=ciσ†ciσn_{i\sigma}=c_{i\sigma}^{\dagger}c_{i\sigma} is the number operator; the summation over σ,σ′∈{↑,↓}\sigma,\sigma'\in\{\uparrow,\downarrow\} is implied throughout. The quadratic Hamiltonian H^0\hat{H}_0 consists of nearest-neighbor

IrrepΔα\Delta_{\alpha}gα(k)g_{\alpha}(\mathbf{k})ParitySource
A1A_1Δs\Delta_sgs=1g_s=1evenUU
A1A_1Δsx\Delta_{s_x}gsx=cos⁡kxg_{s_x}=\cos k_xevenVV
A1A_1Δsy\Delta_{s_y}gsy=cos⁡kyg_{s_y}=\cos k_yevenVV
A1A_1Δpx\Delta_{p_x}gpx=sin⁡kxg_{p_x}=\sin k_xoddVV
B2B_2Δpy\Delta_{p_y}gpy=sin⁡kyg_{p_y}=\sin k_yoddVV

Table 1. Pairing channels are classified according to the C2vC_{2v} irreducible representations (irreps). The broken inversion symmetry merges singlet and triplet states into the same A1A_1 irrep. The exchange field breaks the fourfold rotation, so the bond singlets Δsx\Delta_{s_x} and Δsy\Delta_{s_y} — which would otherwise combine into a single higher-dimensional irrep — instead belong to A1A_1 individually. The table is adapted from Ref. [39].

hopping with amplitude tt, a chemical potential μ\mu, an external Zeeman field hh, and a pp-wave exchange field JJ. The exchange term couples only sites separated along the xx direction, with the +x+x and −x-x hopping processes carrying opposite phase. The resulting spin splitting is odd in momentum and carries opposite sign for spin-up and spin-down electrons, with s↑=+1s_{\uparrow}=+1 and s↓=−1s_{\downarrow}=-1. The interaction part H^int\hat{H}_{\mathrm{int}} contains an on-site attractive interaction UU and a nearest-neighbor density-density interaction VV. All energies are given in units of tt, and we set t=1t=1 throughout.

In momentum space, the quadratic Hamiltonian H^0\hat{H}_0 is written as

H^0=∑k,σϵσ(k)ckσ†ckσ,(2)\hat{H}_0=\sum_{\mathbf{k},\sigma}\epsilon_{\sigma}(\mathbf{k})c_{\mathbf{k}\sigma}^{\dagger}c_{\mathbf{k}\sigma}, \tag*{(2)}

where the kinetic energy dispersion ϵσ(k)\epsilon_{\sigma}(\mathbf{k}) is given by

ϵσ(k)=−2tcos⁡kx−2tcos⁡ky+2Jsσsin⁡kx−hsσ−μ.(3)\epsilon_{\sigma}(\mathbf{k})=-2t\cos k_x-2t\cos k_y+2Js_{\sigma}\sin k_x-hs_{\sigma}-\mu. \tag*{(3)}

This dispersion can be rewritten as

ϵσ(k)=−2t1cos⁡(kx+sσϕJ)−2tcos⁡ky−hsσ−μ,(4)\epsilon_{\sigma}(\mathbf{k})=-2t_1\cos\left(k_x+s_{\sigma}\phi_J\right)-2t\cos k_y-hs_{\sigma}-\mu, \tag*{(4)}

where t1=t2+J2t_1=\sqrt{t^2+J^2} and ϕJ=tan⁡−1(Jt)\phi_J=\tan^{-1}\left(\frac{J}{t}\right). Microscopically, the model in Eq. (1) is an effective single-sublattice description of a parent lattice model, in which coplanar local moments rotating by 90∘90^\circ at each nearest-neighbor step generate the pp-wave exchange field experienced by itinerant electrons hopping between the two square sublattices [38, 39].

Zero-momentum pairing states

To describe the superconducting states, we perform a self-consistent BdG analysis. The interaction term in Eq. (1) can be decomposed into pairing channels, each with its own form factor gα(k)g_{\alpha}(\mathbf{k}). For the inter-spin zero-momentum pairing states—that is, pairing between upand down-spin electrons—these channels are classified according to the irreducible representations (irreps) of the C2vC_{2v} point group, as summarized in Table I. Two symmetry-allowed channels exist: the A1A_1 and B2B_2 states. The A1A_1 state is described by the mean-field BdG Hamiltonian

H^BdGA1=∑kΨk†(ϵ↑(k)ΔA1(k)[ΔA1(k)]∗−ϵ↓(−k))Ψk+C,(5)\hat{H}_{\mathrm{BdG}}^{A_1}=\sum_{\boldsymbol{k}}\Psi_{\boldsymbol{k}}^\dagger\begin{pmatrix}\epsilon_{\uparrow}(\boldsymbol{k}) & \Delta_{A_1}(\boldsymbol{k}) \\ [\Delta_{A_1}(\boldsymbol{k})]^* & -\epsilon_{\downarrow}(-\boldsymbol{k})\end{pmatrix}\Psi_{\boldsymbol{k}}+C, \tag*{(5)}

with the Nambu basis Ψk=(ck↑,c−k↓†)T\Psi_{\boldsymbol{k}}=\left(c_{\boldsymbol{k}\uparrow},c_{-\boldsymbol{k}\downarrow}^{\dagger}\right)^{T}. The constant CC collects the single-particle energy shift generated by normal ordering together with the terms quadratic in the pairing amplitudes, C=∑kϵ↓(k)+N2V∣Δs∣2+N2V∑α=sx,sy,px∣Δα∣2C=\sum_{\boldsymbol{k}}\epsilon_{\downarrow}(\boldsymbol{k})+\frac{N}{2V}|\Delta_s|^2+\frac{N}{2V}\sum_{\alpha=s_x,s_y,p_x}|\Delta_\alpha|^2. The gap function ΔA1(k)\Delta_{A_1}(\boldsymbol{k}) comprises three even-parity components (α=s,sx,sy\alpha=s,s_x,s_y) and one odd-parity component (α=px\alpha=p_x),

ΔA1(k)=∑α=s,sx,sy,pxΔαgα(k).(6)\Delta_{A_1}(\boldsymbol{k})=\sum_{\alpha=s,s_x,s_y,p_x}\Delta_\alpha g_\alpha(\boldsymbol{k}). \tag*{(6)}

The exchange field breaks inversion symmetry and therefore mixes the even-parity spin-singlet and odd-parity spin-triplet channels belonging to the same irrep. Each pairing amplitude satisfies its own self-consistency condition,

Δs=−UN∑k⟨c−k↓ck↑⟩,\Delta_s=-\frac{U}{N}\sum_{\boldsymbol{k}}\langle c_{-\boldsymbol{k}\downarrow}c_{\boldsymbol{k}\uparrow}\rangle,
Δsx=−2VN∑kcos⁡kx ⟨c−k↓ck↑⟩,\Delta_{s_x}=-\frac{2V}{N}\sum_{\boldsymbol{k}}\cos k_x\,\langle c_{-\boldsymbol{k}\downarrow}c_{\boldsymbol{k}\uparrow}\rangle,
Δsy=−2VN∑kcos⁡ky ⟨c−k↓ck↑⟩,\Delta_{s_y}=-\frac{2V}{N}\sum_{\boldsymbol{k}}\cos k_y\,\langle c_{-\boldsymbol{k}\downarrow}c_{\boldsymbol{k}\uparrow}\rangle,
Δpx=−2VN∑ksin⁡kx ⟨c−k↓ck↑⟩.(7)\Delta_{p_x}=-\frac{2V}{N}\sum_{\boldsymbol{k}}\sin k_x\,\langle c_{-\boldsymbol{k}\downarrow}c_{\boldsymbol{k}\uparrow}\rangle. \tag*{(7)}

where NN is the number of sites. The B2B_2 state is described by the BdG Hamiltonian

H^BdGB2=∑kΨk†(ϵ↑(k)ΔB2(k)[ΔB2(k)]∗−ϵ↓(−k))Ψk+C,(8)\hat{H}_{\mathrm{BdG}}^{B_2}=\sum_{\boldsymbol{k}}\Psi_{\boldsymbol{k}}^\dagger\begin{pmatrix}\epsilon_{\uparrow}(\boldsymbol{k}) & \Delta_{B_2}(\boldsymbol{k}) \\ [\Delta_{B_2}(\boldsymbol{k})]^* & -\epsilon_{\downarrow}(-\boldsymbol{k})\end{pmatrix}\Psi_{\boldsymbol{k}}+C, \tag*{(8)}

with the same Nambu spinor as in the A1A_1 case and the constant C=∑kϵ↓(k)+N2V∣Δpy∣2C=\sum_{\boldsymbol{k}}\epsilon_{\downarrow}(\boldsymbol{k})+\frac{N}{2V}|\Delta_{p_y}|^2. The gap function reads

ΔB2(k)=Δpygpy(k),(9)\Delta_{B_2}(\boldsymbol{k})=\Delta_{p_y}g_{p_y}(\boldsymbol{k}), \tag*{(9)}

where Δpy\Delta_{p_y} is determined self-consistently as

Δpy=−2VN∑ksin⁡ky ⟨c−k↓ck↑⟩.(10)\Delta_{p_y}=-\frac{2V}{N}\sum_{\boldsymbol{k}}\sin k_y\,\langle c_{-\boldsymbol{k}\downarrow}c_{\boldsymbol{k}\uparrow}\rangle. \tag*{(10)}

Finite-momentum pairing states

We assume an equal-spin configuration for the FF states, which restricts the pairing to the odd-parity components α∈{px,py}\alpha\in\{p_x,p_y\}. We choose the pair momentum as

Qσ=−2sσφJk^x≡Qσk^x.(11)\boldsymbol{Q}_{\sigma}=-2s_{\sigma}\varphi_J\hat{\boldsymbol{k}}_x\equiv Q_{\sigma}\hat{\boldsymbol{k}}_x. \tag*{(11)}

Equal-spin pairing at a finite pair momentum Qσ∥k^x\boldsymbol{Q}_{\sigma}\parallel\hat{\boldsymbol{k}}_x reduces the relevant symmetry group from C2vC_{2v} to Cs={E,My}C_s=\{E,M_y\}, where MyM_y denotes the mirror operation ky→−kyk_y\to-k_y that leaves Qσ\boldsymbol{Q}_{\sigma} invariant. Under this group, the form factors gpx(k)=sin⁡kxg_{p_x}(k)=\sin k_x and gpy(k)=sin⁡kyg_{p_y}(k)=\sin k_y belong to the irreps A′A' and A′′A'', respectively, so the two associated states, hereafter referred to as the FF-pxp_x and FF-pyp_y states, do not mix and may be treated separately. They are described by the BdG Hamiltonian

H^BdGFF−α=12∑k,σΨk,σ†(ϵ~σ(k)Δα;σFF(k)[Δα;σFF(k)]∗−ϵ~σ(−k))Ψk,σ+C,(12)\hat{H}_{\mathrm{BdG}}^{\mathrm{FF}-\alpha}=\frac{1}{2}\sum_{\boldsymbol{k},\sigma}\Psi_{\boldsymbol{k},\sigma}^{\dagger}\begin{pmatrix}\tilde{\epsilon}_{\sigma}(\boldsymbol{k}) & \Delta_{\alpha;\sigma}^{\mathrm{FF}}(\boldsymbol{k}) \\ [\Delta_{\alpha;\sigma}^{\mathrm{FF}}(\boldsymbol{k})]^* & -\tilde{\epsilon}_{\sigma}(-\boldsymbol{k})\end{pmatrix}\Psi_{\boldsymbol{k},\sigma}+C, \tag*{(12)}

where α=px\alpha=p_x (pyp_y) corresponds to the FF-pxp_x (FF-pyp_y) state. Here, the Nambu basis is Ψk,σ=(ck+Qσ2,σ†,c−k+Qσ2,σ)T\Psi_{\boldsymbol{k},\sigma}=\left(c_{\boldsymbol{k}+\frac{\boldsymbol{Q}_{\sigma}}{2},\sigma}^{\dagger},c_{-\boldsymbol{k}+\frac{\boldsymbol{Q}_{\sigma}}{2},\sigma}\right)^{T} and the constant is C=12∑k,σϵσ(k)+N4V∑σ∣Δα;σFF∣2C=\frac{1}{2}\sum_{\boldsymbol{k},\sigma}\epsilon_{\sigma}(\boldsymbol{k})+\frac{N}{4V}\sum_{\sigma}|\Delta_{\alpha;\sigma}^{\mathrm{FF}}|^2. The gap function is

Δα;σFF(k)=Δα;σFFgα(k),(13)\Delta_{\alpha;\sigma}^{\mathrm{FF}}(\boldsymbol{k})=\Delta_{\alpha;\sigma}^{\mathrm{FF}}g_\alpha(\boldsymbol{k}), \tag*{(13)}

and the shifted dispersion ϵ~σ(k)≡ϵσ(k+Qσ2)\tilde{\epsilon}_{\sigma}(\boldsymbol{k})\equiv\epsilon_{\sigma}\left(\boldsymbol{k}+\frac{\boldsymbol{Q}_{\sigma}}{2}\right) reads

ϵ~σ(k)=−2t1cos⁡kx−2tcos⁡ky−hsσ−μ.(14)\tilde{\epsilon}_{\sigma}(\boldsymbol{k})=-2t_1\cos k_x-2t\cos k_y-hs_{\sigma}-\mu. \tag*{(14)}

The pairing amplitude is determined self-consistently by

Δα;σFF=−2VN∑kgα(k)⟨c−k+Qσ2,σck+Qσ2,σ⟩.(15)\Delta_{\alpha;\sigma}^{\mathrm{FF}}=-\frac{2V}{N}\sum_{\boldsymbol{k}}g_\alpha(\boldsymbol{k})\left\langle c_{-\boldsymbol{k}+\frac{\boldsymbol{Q}_{\sigma}}{2},\sigma}c_{\boldsymbol{k}+\frac{\boldsymbol{Q}_{\sigma}}{2},\sigma}\right\rangle. \tag*{(15)}

We emphasize that the choice of pair momentum in Eq. (11) is essential for the energetic stabilization of the FF states. With this choice, the two electrons forming a Cooper pair have exactly the same energy [see Figure 1(a)],

ϵ~σ(k)=ϵ~σ(−k).(16)\tilde{\epsilon}_{\sigma}(\boldsymbol{k})=\tilde{\epsilon}_{\sigma}(-\boldsymbol{k}). \tag*{(16)}

In general, the denominator of the gap equation [Eq. (41)] involves the symmetrized dispersion ϵ~σS(k)=12[ϵ~σ(k)+ϵ~σ(−k)]\tilde{\epsilon}_{\sigma}^{S}(\boldsymbol{k})=\frac{1}{2}[\tilde{\epsilon}_{\sigma}(\boldsymbol{k})+\tilde{\epsilon}_{\sigma}(-\boldsymbol{k})] through [ϵ~σS(k)]2+∣Δα;σFFgα(k)∣2\sqrt{[\tilde{\epsilon}_{\sigma}^{S}(\boldsymbol{k})]^2+|\Delta_{\alpha;\sigma}^{\mathrm{FF}}g_\alpha(\boldsymbol{k})|^2}. Under the equal-energy condition of Eq. (16), this reduces to ϵ~σ(k)2+∣Δα;σFFgα(k)∣2\sqrt{\tilde{\epsilon}_{\sigma}(\boldsymbol{k})^2+|\Delta_{\alpha;\sigma}^{\mathrm{FF}}g_\alpha(\boldsymbol{k})|^2}, governed by the single dispersion ϵ~σ(k)\tilde{\epsilon}_{\sigma}(\boldsymbol{k}). The zero-energy contour of this dispersion therefore coincides with the entire shifted FS, supporting a substantial pairing amplitude in this channel. By contrast, for a general Qσ\boldsymbol{Q}_{\sigma} the two paired electrons have different energies ϵ~σ(k)\tilde{\epsilon}_{\sigma}(\boldsymbol{k}) and ϵ~σ(−k)\tilde{\epsilon}_{\sigma}(-\boldsymbol{k}), and the zero-energy contour of ϵ~σS\tilde{\epsilon}_{\sigma}^{S} no longer covers a full FS. The pairing amplitude is therefore suppressed whenever Qσ\boldsymbol{Q}_{\sigma} deviates from Eq. (11), indicating that this choice is the optimal pair momentum for stabilizing the FF states.

In total, we investigate four superconducting states: the equal-spin FF-pxp_x and FF-pyp_y states [Figure 1(a)], and the two inter-spin A1_1 and B2_2 states with zero pair momentum [Figure 1(b)]. These four channels cover the pairing configurations in which every electron on the FSs can participate, thereby potentially maximizing the pairing amplitude.

Emergence of FF states

Zero-field phase diagrams

We determine the energetically favored superconducting state among the four candidate pairing states—A1_1, B2_2, FF-pxp_x, and FF-pyp_y—over a range of parameters (μ\mu, JJ, UU, and VV) by numerically solving the self-consistent gap equations at zero temperature. The resulting pairing amplitudes are substituted into the corresponding ground-state energy expressions, and the state with the lowest energy is identified as the superconducting ground state, from which the phase diagrams are constructed. The explicit gap equations, ground-state energy expressions, and the numerical procedure are provided in the Methods section.

Figure 2 displays phase diagrams of the superconducting ground state in the JJ–μ\mu plane for four combinations of (UU, VV). At (UU, VV) = (1, 3), the A1_1 phase occupies most of the parameter space [Figure 2(a)]. The FF-pxp_x phase occupies a region centered at μ=0\mu= 0, which sets in at J≃1.44J \simeq1.44 and widens with increasing JJ, reaching ∣μ∣≃1.5|\mu| \simeq1.5 at J=2J = 2; this region is symmetric under μ↔−μ\mu\leftrightarrow-\mu, as required by the particle-hole symmetry of the bipartite square lattice with nearest-neighbor hopping. By contrast, the FF-pyp_y phase occupies narrow regions at intermediate chemical potential, 1.4≲∣μ∣≲2.01.4 \lesssim|\mu| \lesssim2.0 at J=0J = 0, which drift toward larger ∣μ∣|\mu| as JJ increases and terminate at J≃1.5J \simeq1.5. Both the FF-pxp_x and FF-pyp_y phases are degenerate with their zero-momentum counterparts, namely the A1_1 and B2_2 phases, respectively.

Phase diagrams of the pairing ground state for four interaction strengths

Figure 2. Emergence of FF states. (a)–(d) Phase diagrams of the pairing ground state as functions of the pp-wave exchange field strength JJ and the chemical potential μ\mu. Light blue denotes the even-parity A1A_1 phase, red the degenerate FF-pxp_x/odd-parity A1A_1 phase, and orange the degenerate FF-pyp_y/B2B_2 phase. The interaction strengths are (a) (U,V)=(1,3)(U,V)=(1,3), (b) (1,2)(1,2), (c) (2,3)(2,3), and (d) (2,2)(2,2).

Qualitatively similar phase diagrams are obtained for other (UU, VV) combinations. Reducing VV from 3 to 2 at the same UU shrinks the FF-pyp_y/B2_2 regions, while the FF-pxp_x/A1_1 region keeps its onset at J≃1.4J \simeq1.4 and only its μ\mu range becomes slightly narrower [Figure 2(b)]. Raising UU from 1 to 2 at the same VV suppresses the former far more strongly, whereas the latter again keeps its JJ onset and its μ\mu range narrows further [Figure 2(c)]. When both changes are combined at (UU, VV) = (2, 2), the FF-pyp_y/B2_2 phase disappears altogether, whereas the FF-pxp_x/A1_1 phase survives with a similar onset at J≃1.4J \simeq1.4 and a narrower μ\mu range [Figure 2(d)]. These results indicate that the degenerate FF-pxp_x/A1_1 phase persists at large JJ near μ=0\mu= 0 largely independently of UU and VV over the range examined, whereas its FF-pyp_y/B2_2 counterpart requires a sufficiently large VV and a sufficiently small UU and is suppressed otherwise. Increasing UU further to 3 eliminates either the FF-pyp_y phase alone or both the FF-pyp_y and FF-pxp_x phases, depending on the value of VV. A detailed analysis is given in Supplementary Note 1.

We find that within the A1_1 phase, the pairing components Δsx\Delta_{sx} and Δpx\Delta_{px} of the A1_1 gap function are locked into a vector form (Δsx,Δpx)=Δe(cos⁡ϕJ,−sin⁡ϕJ)≡Δxeven(\Delta_{sx},\Delta_{px}) = \Delta_e(\cos\phi_J,-\sin\phi_J) \equiv\boldsymbol{\Delta}_x^{\mathrm{even}}, whose orientation is determined by the momentum shift ϕJ\phi_J. This locking combines the two form factors gsx(k)g_{sx}(\boldsymbol{k}) and gpx(k)g_{px}(\boldsymbol{k}) into the single form factor cos⁡(kx+ϕJ)\cos(k_x+\phi_J), which carries the same momentum shift as the energy dispersion and is even under the reflection kx→−kx−2ϕJk_x \to-k_x-2\phi_J, giving the even-parity gap function

ΔA1even(k)=Δs+Δsycos⁡ky+Δecos⁡(kx+ϕJ).(17)\Delta_{A_1}^{\mathrm{even}}(\boldsymbol{k})=\Delta_s+\Delta_{sy}\cos k_y+\Delta_e\cos(k_x+\phi_J). \tag*{(17)}

In the degenerate FF-pxp_x/A1_1 phase, by contrast, Δsx\Delta_{sx} and Δpx\Delta_{px} are locked into the vector form (Δsx,Δpx)=Δo(sin⁡ϕJ,cos⁡ϕJ)≡Δxodd(\Delta_{sx},\Delta_{px}) = \Delta_o(\sin\phi_J,\cos\phi_J) \equiv\boldsymbol{\Delta}_x^{\mathrm{odd}}, while Δs=Δsy=0\Delta_s = \Delta_{sy} = 0. The two form factors then combine into sin⁡(kx+ϕJ)\sin(k_x+\phi_J), which is odd under the same reflection, leading to the odd-parity gap function

ΔA1odd(k)=Δosin⁡(kx+ϕJ).(18)\Delta_{A_1}^{\mathrm{odd}}(\boldsymbol{k})=\Delta_o\sin(k_x+\phi_J). \tag*{(18)}

The pairing vectors Δxeven\boldsymbol{\Delta}_x^{\mathrm{even}} and Δxodd\boldsymbol{\Delta}_x^{\mathrm{odd}} are eigenvectors of the reflection matrix

RJ=(cos⁡2ϕJ−sin⁡2ϕJ−sin⁡2ϕJ−cos⁡2ϕJ),(19)R_J=\begin{pmatrix} \cos2\phi_J & -\sin2\phi_J\\ -\sin2\phi_J & -\cos2\phi_J \end{pmatrix}, \tag*{(19)}

with eigenvalues +1+1 and −1-1, respectively, which define their parities. Each is therefore mapped onto itself under RJR_J, up to an overall sign for the odd-parity vector that can be removed by a global phase rotation, and thus has no symmetry-related partner. This is in contrast to a general mixed form, Δmixed=Δxeven+Δxodd\boldsymbol{\Delta}^{\mathrm{mixed}} = \boldsymbol{\Delta}^{\mathrm{even}}_x + \boldsymbol{\Delta}^{\mathrm{odd}}_x, with the gap function ΔA1mixed(k)=Δs+Δsycos⁡ky+Δecos⁡(kx+ϕJ)+Δosin⁡(kx+ϕJ)\Delta_{A_1}^{\mathrm{mixed}}(\boldsymbol{k}) = \Delta_s+\Delta_{sy}\cos k_y+\Delta_e\cos(k_x+\phi_J)+\Delta_o\sin(k_x+\phi_J), which is not invariant under RJR_J and is always accompanied by a degenerate partner obtained by applying RJR_J to it. A detailed analysis of this degeneracy is provided in Supplementary Note 2.

Figure 3 shows the four A1_1 amplitudes Δs\Delta_s, Δsy\Delta_{sy}, Δe\Delta_e, and Δo\Delta_o, together with the two FF amplitudes, along JJ-cuts at fixed μ\mu and cuts fixed JJ, for (UU, VV) = (1, 3) as in Figure 2(a). The B2_2 amplitude is omitted because it always equals that of the FF-pyp_y state (see Supplementary Note 2), and only the spin-up components of the FF states are shown because the spin-down components are identical. Along the JJ-cut at μ=0\mu= 0, Δe\Delta_e and Δsy\Delta_{sy} are nonzero within the A1_1 phase whereas Δo\Delta_o remains zero, consistent with the even-parity A1_1 state [see Figure 3(a)]. In the degenerate FF-pxp_x/A1_1 phase, by contrast, Δo\Delta_o is the only nonzero component, and its magnitude is identical to that of Δpx,↑\Delta_{px,\uparrow}. This is not a coincidence but reflects the degeneracy of the two states within this phase, which will be elucidated in Sec. III B. Similar patterns are observed along the μ\mu-cut at J=2J = 2 [Figure 3(b)]. Within the degenerate FF-pxp_x/A1_1 phase, Δo\Delta_o is the only nonzero component and equals Δpx;↑FF\Delta^{\mathrm{FF}}_{p_x;\uparrow}, whereas within the A1A_1 phase, Δo\Delta_o vanishes and Δs\Delta_s, Δsy\Delta_{s_y}, and Δe\Delta_e are nonzero.

Distribution of pairing amplitude components

Figure 3. Distribution of pairing amplitude components. (a), (c) Pairing amplitudes as a function of JJ for (a) μ=0\mu=0 and (c) μ=1.6\mu=1.6. (b), (d) Pairing amplitudes as a function of μ\mu for (b) J=2J=2 and (d) J=0.5J=0.5. The parameters (U,V)=(1,3)(U,V)=(1,3) are used throughout, as in Figure 2(a). The curves denote the ss-wave (Δs\Delta_s, blue solid), sys_y-wave (Δsy\Delta_{sy}, blue dotted), even-parity (Δe\Delta_e, blue dashed), and odd-parity (Δo\Delta_o, blue dot-dashed) components of the A1A_1 state, together with the FF amplitudes Δpx;↑FF\Delta^{\mathrm{FF}}_{p_x;\uparrow} (red) and Δpy;↑FF\Delta^{\mathrm{FF}}_{p_y;\uparrow} (orange). In (c) and (d), insets display the difference Δpy;↑FF−Δpx;↑FF\Delta^{\mathrm{FF}}_{p_y;\uparrow}-\Delta^{\mathrm{FF}}_{p_x;\uparrow} within the FF-py/B2p_y/B_2 phase. In each panel, the stabilized ground-state phase is annotated within its region, with the regions separated by dashed lines.

The JJ-cut at μ=1.6\mu=1.6 shows that, within the degenerate FF-pyp_y/B2B_2 phase, Δpx;↑FF\Delta^{\mathrm{FF}}_{p_x;\uparrow} exceeds Δpy;↑FF\Delta^{\mathrm{FF}}_{p_y;\uparrow} by a slight margin [Figure 3(c), inset]. Within the same phase, the A1A_1 gap function is of the odd-parity type Δxodd\Delta_x^{\mathrm{odd}} for J<0.63J<0.63, whereas it takes the mixed form Δxmixed\Delta_x^{\mathrm{mixed}} for 0.63≤J<0.760.63\leq J<0.76. For J≥0.78J\geq0.78, where the A1A_1 phase is stabilized, the A1A_1 gap function is of the even-parity type Δxeven\Delta_x^{\mathrm{even}}. These observations indicate that the odd-parity and mixed forms appear only within the degenerate phases, whereas the even-parity form characterizes the A1A_1 phase. Similar behavior is observed along the μ\mu-cut at J=0.5J=0.5 [Figure 3(d)]. Within the degenerate FF-pyp_y/B2B_2 windows at 1.46<∣μ∣<2.101.46<|\mu|<2.10, Δpy;↑FF\Delta^{\mathrm{FF}}_{p_y;\uparrow} exceeds Δpx;↑FF\Delta^{\mathrm{FF}}_{p_x;\uparrow} by a small margin [Figure 3(d), inset], and the A1A_1 gap function is of the odd-parity type, whereas it is of the even-parity type outside these windows. It is worth mentioning that Δs\Delta_s remains the smallest component of the even-parity A1A_1 gap function over most of the JJ–μ\mu plane, which we attribute to the small values of UU (U=1,2U=1,2) considered here, whereas for U=3U=3 it can become comparable to the other components, as detailed in Supplementary Note 1.

Pairing-channel selection mechanism

Having established the phase diagrams of the FF states and their degenerate zero pair-momentum counterparts, we now examine the mechanism that selects the pairing channel in each region of the JJ–μ\mu plane. To this end, we consider reduced gap equations for the even-parity A1A_1, odd-parity A1A_1, and B2B_2 states at h=0h=0,

Δe=VN∑k′[Δsygsy(k′)+Δegsx(k′)]gsx(k′)ϵ~(k′)2+∣Δsygsy(k′)+Δegsx(k′)∣2,(20)\Delta_e=\frac{V}{N}\sum_{k'}\frac{[\Delta_{sy}g_{sy}(k')+\Delta_eg_{sx}(k')]g_{sx}(k')}{\sqrt{\widetilde{\epsilon}(k')^2+\lvert\Delta_{sy}g_{sy}(k')+\Delta_eg_{sx}(k')\rvert^2}}, \tag*{(20)}
Δsy=VN∑k′[Δsygsy(k′)+Δegsx(k′)]gsy(k′)ϵ~(k′)2+∣Δsygsy(k′)+Δegsx(k′)∣2,(21)\Delta_{sy}=\frac{V}{N}\sum_{k'}\frac{[\Delta_{sy}g_{sy}(k')+\Delta_eg_{sx}(k')]g_{sy}(k')}{\sqrt{\widetilde{\epsilon}(k')^2+\lvert\Delta_{sy}g_{sy}(k')+\Delta_eg_{sx}(k')\rvert^2}}, \tag*{(21)}
Δo=VN∑k′Δogpx(k′)2ϵ~(k′)2+∣Δogpx(k′)∣2,(22)\Delta_o=\frac{V}{N}\sum_{k'}\frac{\Delta_og_{px}(k')^2}{\sqrt{\widetilde{\epsilon}(k')^2+\lvert\Delta_og_{px}(k')\rvert^2}}, \tag*{(22)}
Δpy=VN∑k′Δpygpy(k′)2ϵ~(k′)2+∣Δpygpy(k′)∣2,(23)\Delta_{py}=\frac{V}{N}\sum_{k'}\frac{\Delta_{py}g_{py}(k')^2}{\sqrt{\widetilde{\epsilon}(k')^2+\lvert\Delta_{py}g_{py}(k')\rvert^2}}, \tag*{(23)}

where the dispersion is

ϵ~(k′)=−2t1cos⁡kx′−2tcos⁡ky′−μ,(24)\widetilde{\epsilon}(k')=-2t_1\cos k'_x-2t\cos k'_y-\mu, \tag*{(24)}

the form factors are gsx=cos⁡kx′g_{sx}=\cos k'_x, gsy=cos⁡ky′g_{sy}=\cos k'_y, gpx=sin⁡kx′g_{px}=\sin k'_x, and gpy=sin⁡ky′g_{py}=\sin k'_y, and the momentum k′k' represents the shifted momentum coordinates

kx′=kx+ϕJ,ky′=ky.(25)k'_x=k_x+\phi_J,\qquad k'_y=k_y. \tag*{(25)}

The derivation of the reduced gap equations is given in Supplementary Note 3. Equations (20) and (21) govern the even-parity A1A_1 pairing amplitudes, where we have neglected Δs\Delta_s for simplicity. A fully self-consistent treatment would also require the gap equation for Δs\Delta_s, but the omission is supported by the small magnitude of Δs\Delta_s in the self-consistent solutions (Figure 3). Equations (22) and (23) govern the odd-parity A1A_1 and B2B_2 amplitudes, respectively. The momentum summation is dominated by the vicinity of the contour ϵ~(k′)=0\widetilde{\epsilon}(k')=0, where the summand is weighted by the corresponding form factor. The pairing channel selected for given parameters is therefore governed by the overlap between the contour and the form factors.

To visualize this overlap, we plot the zero-energy contour ϵ~(k′)=0\widetilde{\epsilon}(k')=0 together with the momentum-space distributions of the form factors gsx(k′)g_{sx}(k'), gsy(k′)g_{sy}(k'), gpx(k′)g_{px}(k'), and gpy(k′)g_{py}(k'), as illustrated in Figure 4. At μ=0\mu=0 and J=0J=0, the contour passes through the van Hove points (kx′,ky′)=(π,0)(k'_x,k'_y)=(\pi,0) and (0,π)(0,\pi), where the density of states diverges. At these points, gsx(k′)g_{s_x}(\mathbf{k}') and gsy(k′)g_{s_y}(\mathbf{k}') reach their antinodes while gpx(k′)g_{p_x}(\mathbf{k}') and gpy(k′)g_{p_y}(\mathbf{k}') vanish. In addition, the fourfold rotation symmetry ensures that gsx(k′)g_{s_x}(\mathbf{k}') and gsy(k′)g_{s_y}(\mathbf{k}') have identical overlaps with the contour, and likewise for gpx(k′)g_{p_x}(\mathbf{k}') and gpy(k′)g_{p_y}(\mathbf{k}') [Figs. 4(a)–(d)]. The overlaps entering the gap equations (20)–(23) are therefore larger for gsx(k′)g_{s_x}(\mathbf{k}') and gsy(k′)g_{s_y}(\mathbf{k}') than for gpx(k′)g_{p_x}(\mathbf{k}') and gpy(k′)g_{p_y}(\mathbf{k}'). This accounts for the stabilization of the even-parity A1A_1 state, built from the cosine channels, at small JJ [Fig. 3(a)]. In this regime, the two cosine channels have comparable overlaps, and the even-parity gap takes the dd-wave-like form Δecos⁡kx′+Δsycos⁡ky′\Delta_e\cos k'_x+\Delta_{s_y}\cos k'_y with Δsy≈−Δe\Delta_{s_y}\approx-\Delta_e, which maximizes the gap magnitude at the two van Hove points, ∣ΔA1even∣≈2∣Δe∣|\Delta_{A_1}^{\mathrm{even}}|\approx2|\Delta_e|. At larger JJ, the contour moves away from the van Hove points and the fourfold rotation symmetry is broken, so that the four overlaps all become different. Specifically, at J=0.7J=0.7, the contour has greater overlap with gsy(k′)g_{s_y}(\mathbf{k}') than with gsx(k′)g_{s_x}(\mathbf{k}'). This accounts for the observed growth of the sys_y component in the even-parity A1A_1 state [Fig. 3(a)]. As JJ increases further, the exchange field elongates the contour along ky′k'_y into nearly straight segments near kx′=±π/2k'_x=\pm\pi/2, which are the nodes of gsx(k′)g_{s_x}(\mathbf{k}') and the antinodes of gpx(k′)g_{p_x}(\mathbf{k}') [Figs. 4(a) and (c)]. Consequently, at J≥1.44J\geq1.44, the odd-parity A1A_1 state, governed by gpx(k′)g_{p_x}(\mathbf{k}'), becomes energetically favored over the even-parity one.

Selection of pairing channels

Figure 4. Selection of pairing channels. Squared form factors ∣gα(k′)∣2|g_{\alpha}(\mathbf{k}')|^2 (color scale) overlaid with the zero-energy contour ϵ~(k′)=0\tilde{\epsilon}(\mathbf{k}')=0 (lines) at the exchange field strengths J=0J=0, 0.70.7, 1.41.4, and 2.12.1 (black, orange, magenta, and green, respectively). (a)–(d) μ=0\mu=0, for the (a) sxs_x-, (b) sys_y-, (c) pxp_x-, and (d) pyp_y-wave form factors. (e)–(h) Same as (a)–(d), but at μ=1.6\mu=1.6.

At μ=1.6\mu=1.6 and J=0J=0, the contour runs through the antinodes of the sine form factors gpx(k′)g_{p_x}(\mathbf{k}') and gpy(k′)g_{p_y}(\mathbf{k}') [Figs. 4(g) and (h)] and close to the nodes of the cosine form factors gsx(k′)g_{s_x}(\mathbf{k}') and gsy(k′)g_{s_y}(\mathbf{k}') [Figs. 4(e) and (f)]. The sine channels therefore carry the larger overlap, so that the odd-parity A1A_1 and B2B_2 states, governed by the sine channels, are favored over the even-parity A1A_1 state, governed by the cosine channels. Introducing a small JJ enhances the overlap with the antinode of gpy(k′)g_{p_y}(\mathbf{k}') while reducing that of gpx(k′)g_{p_x}(\mathbf{k}') [Figs. 4(g) and (h)], stabilizing the B2B_2 state for J<0.78J<0.78 [Fig. 3(c)]. In the same JJ interval, the overlap with gpx(k′)g_{p_x}(\mathbf{k}') still exceeds those with gsx(k′)g_{s_x}(\mathbf{k}') and gsy(k′)g_{s_y}(\mathbf{k}') [Figs. 4(e)–(g)], so that the odd-parity form is favored over the even-parity one within the A1A_1 state [Fig. 3(c)]. Finally, at J=1.4J=1.4 and 2.12.1, the dominance of gpy(k′)g_{p_y}(\mathbf{k}') is lost and the contour approaches the van Hove points, where both gsx(k′)g_{s_x}(\mathbf{k}') and gsy(k′)g_{s_y}(\mathbf{k}') reach their antinodes [Figs. 4(e) and (f)], so that the even-parity A1A_1 state becomes the energetically favored one [Fig. 3(c)].

We note that the odd-parity A1A_1 gap equation (22) coincides with the FF-pxp_x gap equation (41). The coincidence arises because the global shift kx→kx−ϕJk_x\to k_x-\phi_J removes ϕJ\phi_J from the A1A_1 gap equation and leaves it in the same form as the FF one, in which the pair momentum plays the same role. This equivalence is understood in terms of the position- and spin-dependent gauge transformation,

diσ=eisσϕJxiciσ,dk′σ=ckσ,k′=k+sσϕJx^.(26)d_{i\sigma}=e^{i s_\sigma\phi_J x_i}c_{i\sigma},\quad d_{\mathbf{k}'\sigma}=c_{\mathbf{k}\sigma},\quad\mathbf{k}'=\mathbf{k}+s_\sigma\phi_J\hat{x}. \tag*{(26)}

Under this transformation, the A1A_1 BdG Hamiltonian (5) with the odd-parity gap function (18) becomes

H^BdGA1,odd=∑k′Φk′†(ϵ~(k′)Δosin⁡kx′Δo∗sin⁡kx′−ϵ~(k′))Φk′+C(27)\hat{H}^{A_1,\mathrm{odd}}_{\mathrm{BdG}}=\sum_{\boldsymbol{k}'}\Phi_{\boldsymbol{k}'}^\dagger\begin{pmatrix}\tilde{\epsilon}(\boldsymbol{k}') & \Delta_o\sin k'_x \\ \Delta_o^*\sin k'_x & -\tilde{\epsilon}(\boldsymbol{k}')\end{pmatrix}\Phi_{\boldsymbol{k}'}+C \tag*{(27)}

where Φk′=(dk′↑,d−k′↓†)T\Phi_{\boldsymbol{k}'}=(d_{\boldsymbol{k}'\uparrow},d^\dagger_{-\boldsymbol{k}'\downarrow})^T. Equation (27) is identical to the FF-pxp_x BdG Hamiltonian (12) apart from the spin structure of the Nambu spinor. This difference in the spin structure, however, is immaterial at h=0h=0, where the two spin sectors of the FF-pxp_x BdG Hamiltonian are equivalent, and the two problems therefore reduce to one and the same eigenvalue problem. By the same token, the B2B_2 gap equation (23) coincides with the FF-pyp_y gap equation (41). The derivation given here further implies that the ground-state energies of the odd-parity A1A_1 and B2B_2 states agree with those of the FF-pxp_x and FF-pyp_y states, and a rigorous proof is provided in Supplementary Note 2. The correspondence established here indicates that the odd-parity A1A_1 and B2B_2 states are exactly degenerate with the FF-pxp_x and FF-pyp_y states, as reflected in Figure 2.

Taken together, the overlaps between the energy contour and the form factors examined above indicate that the alignment of the energy contour with the antinodes of gpx(k′)g_{p_x}(\boldsymbol{k}') at large JJ and μ=0\mu=0 is the main driver for the stabilization of the odd-parity A1A_1 state in this parameter regime. By contrast, the overlap with gpy(k′)g_{p_y}(\boldsymbol{k}') is largest at small JJ and μ≈1.6\mu\approx1.6, favoring the B2B_2 state instead. Because the FF-pxp_x and FF-pyp_y states are degenerate with the odd-parity A1A_1 and B2B_2 states at h=0h=0, the same overlaps account for the stabilization of the FF-pxp_x and FF-pyp_y states in their respective regions.

While the discussion in this section has assumed U=0U=0, the mechanism remains valid in the presence of U>0U>0, where a finite UU directly enhances the ss-wave component Δs\Delta_s of the even-parity A1A_1 state. As a result, the even-parity A1A_1 phase region widens at the expense of the FF-pxp_x phase in the phase diagram [compare, e.g., Figure 2(a) and Figure 2(c)]. This microscopic picture is also consistent with the observation that the JJ threshold of the FF-pxp_x/even-parity A1A_1 phase boundary is nearly independent of VV [Figure 2(a) and Figure 2(b)], which suggests that the threshold is set by the overlap between the contour and the form factors rather than by the strength of VV. In contrast, the decrease of the μ\mu threshold with UU can be attributed to the growth of Δs\Delta_s driven by UU, which lies outside the mechanism discussed above.

Stabilization of FF states via Zeeman effect

To demonstrate the stabilization of the FF phases via the Zeeman effect, we recalculate the zero-field superconducting phase diagram of Figure 2(a) with the field hh turned on. At a tiny field h=0.01h=0.01, the states occupy largely the same regions as in the zero-field case, but the energies of the FF-pxp_x and FF-pyp_y states become slightly lower than those of the odd-parity A1A_1 and B2B_2 states, respectively.

As a result, the FF states become the stable superconducting phases in their respective regions [Figure 5(a)]. As the field increases, the FF-pxp_x and FF-pyp_y regions expand at the expense of the even-parity A1A_1 region, as shown for h=0.2h=0.2 in Figure 5(b). At h=0.6h=0.6, the entire examined JJ–μ\mu parameter space (J≤2J\leq2 and ∣μ∣≤3|\mu|\leq3) is occupied by the FF-pxp_x and FF-pyp_y phases, and the even-parity A1A_1 phase no longer appears [Figure 5(c)]. The phase diagram in the JJ–hh plane at fixed μ=0\mu=0 shows the growing tendency of the FF-pxp_x phase. The Zeeman field hh required to stabilize the FF-pxp_x phase decreases as JJ increases, dropping from approximately 0.550.55 at J=0J=0 to nearly 00 at J=1.44J=1.44 [Figure 5(d)]. For instance, taking t=0.5t=0.5 eV, the field of h≈0.55h\approx0.55 corresponds to a magnetic field of approximately 4800 T. This indicates that the FF-pxp_x phase can be stabilized at progressively weaker Zeeman fields as the exchange field JJ grows.

Phase diagrams of the superconducting ground state as a function of $J$, $\mu$, and $h$

Figure 5. Stabilization of the FF states by the Zeeman field. (a)–(c) Phase diagrams of the superconducting ground state as a function of JJ and μ\mu for increasing Zeeman field: (a) h=0.01h=0.01, (b) h=0.2h=0.2, and (c) h=0.6h=0.6. (d) Phase diagram as a function of JJ and hh at μ=0\mu=0. In all panels, the parameters (U,V)=(1,3)(U,V)=(1,3) are used as in Figure 2(a). Color coding follows Figure 2. Dashed lines mark the corresponding h=0h=0 phase boundaries of Figure 2.

The dominance of the FF states over the zero pair-momentum states under the Zeeman field can be understood from how the field changes their Bloch Hamiltonians. In the presence of the Zeeman field hh, the Bloch Hamiltonian for the zero pair-momentum states reads

HBdG(k′)=(ϵ~(k′)−hΔ(k′)Δ∗(k′)−ϵ~(k′)−h)(28)H_{\mathrm{BdG}}(\boldsymbol{k}')=\begin{pmatrix}\tilde{\epsilon}(\boldsymbol{k}')-h & \Delta(\boldsymbol{k}') \\ \Delta^*(\boldsymbol{k}') & -\tilde{\epsilon}(\boldsymbol{k}')-h\end{pmatrix} \tag*{(28)}

where Δ(k′)=ΔA1even(k′),ΔA1odd(k′)\Delta(\boldsymbol{k}')=\Delta^{\mathrm{even}}_{A_1}(\boldsymbol{k}'),\Delta^{\mathrm{odd}}_{A_1}(\boldsymbol{k}'), and ΔB2(k′)\Delta_{B_2}(\boldsymbol{k}') for the even-parity A1A_1, odd-parity A1A_1, and B2B_2 states, respectively. Notably, hh enters equally in the particle and hole sectors and, after the Bogoliubov transformation, shifts the quasiparticle energies uniformly. The odd-parity A1A_1 state has a positive minimum quasiparticle energy at h=0h=0, so its quasiparticle energies remain positive for fields below this minimum, which corresponds to the Pauli limit [1]. Accordingly, in this weak-field regime, neither the pairing amplitude nor the ground-state energy changes. Once hh exceeds the Pauli limit, the occupation of negative-energy quasiparticle states in the superconducting ground state reduces both the pairing amplitude and the energy gain of the odd-parity A1A_1 state [25]. By contrast, the even-parity A1A_1 and B2B_2 states have nodal points where the quasiparticle energy vanishes at h=0h=0, so that negative-energy quasiparticle states are occupied around these points at any finite hh, reducing the pairing amplitude and the energy gain. A systematic analysis of the gap structures in the quasiparticle dispersions of these states is presented in Sec. IV.

For the FF-pxp_x and FF-pyp_y states, the Zeeman field enters the Bloch Hamiltonian of each spin sector with opposite signs in the particle and hole sectors,

HBdGσ(k′)=(ϵ~(k′)−hsσΔσ(k′)Δσ∗(k′)−ϵ~(k′)+hsσ)(29)H^\sigma_{\mathrm{BdG}}(\boldsymbol{k}')=\begin{pmatrix}\tilde{\epsilon}(\boldsymbol{k}')-h s_\sigma& \Delta_\sigma(\boldsymbol{k}') \\ \Delta^*_\sigma(\boldsymbol{k}') & -\tilde{\epsilon}(\boldsymbol{k}')+h s_\sigma\end{pmatrix} \tag*{(29)}

where Δσ(k′)=Δpx;σFFsin⁡kx′\Delta_\sigma(\boldsymbol{k}')=\Delta^{\mathrm{FF}}_{p_x;\sigma}\sin k'_x for the FF-pxp_x state and Δσ(k′)=Δpy;σFFsin⁡ky′\Delta_\sigma(\boldsymbol{k}')=\Delta^{\mathrm{FF}}_{p_y;\sigma}\sin k'_y for the FF-pyp_y state. The field now acts as a spin-dependent shift of the chemical potential by +hsσ+h s_\sigma, which changes the size of each spin-resolved FS and thereby alters the pairing amplitude and lowers the ground-state energy even at small hh. This is analogous to the Pauli paramagnetic response of a normal metal [41], and we therefore describe the energy lowering to leading order in hh by

FαFF(h)=FαFF(0)−12χNh2+O(h4),(30)F_{\alpha}^{\mathrm{FF}}(h)=F_{\alpha}^{\mathrm{FF}}(0)-\frac{1}{2}\chi_N h^2+O(h^4), \tag*{(30)}

where FαFF(0)F_{\alpha}^{\mathrm{FF}}(0) is the ground-state energy at zero field and χN\chi_N is the Pauli paramagnetic spin susceptibility defined from the normal-state ground-state energy FN(h)F^N(h),

χN=−∂2FN∂h2∣h=0=∑k,σδ(ϵσ(k))∣h=0=2N(0),(31)\chi_N=-\left.\frac{\partial^2 F^N}{\partial h^2}\right|_{h=0}=\left.\sum_{\boldsymbol{k},\sigma}\delta\left(\epsilon_\sigma(\boldsymbol{k})\right)\right|_{h=0}=2N(0), \tag*{(31)}

with N(0)=∑kδ(ϵ0(k))N(0)=\sum_{\boldsymbol{k}}\delta\left(\epsilon_0(\boldsymbol{k})\right) the density of states per spin at the Fermi level. The identification of the susceptibility of the FF states with the Pauli paramagnetic spin susceptibility holds up to corrections of order (Δ/EF)2(\Delta/E_{\mathrm F})^2 [41].

Taken together, the ground-state energy of the odd-parity A1A_1 state remains unchanged in the weak-field regime below the Pauli limit and increases above it, while those of the even-parity A1A_1 and B2B_2 states increase already at weak fields. By contrast, the field lowers the ground-state energy of the FF states at any finite hh via the Pauli paramagnetic spin susceptibility. The energy gain of the FF states therefore exceeds that of the zero pair-momentum states at any finite hh, so that the FF states replace the odd-parity A1A_1 and B2B_2 states and expand further at the expense of the even-parity A1A_1 state as hh increases [Fig. 5].

We emphasize the crucial role of the spin-dependent FS shift generated by the pp-wave exchange field. The shift makes the FF states degenerate with the zero pair-momentum states at h=0h=0, and owing to this degeneracy, the FF states are favored over the zero pair-momentum states at any finite hh. This is in contrast to conventional FF states, where the Zeeman splitting of the Fermi surfaces must be sufficiently large, comparable to the pairing gap, for a finite-momentum pair to access a larger portion of the FS as a pairing channel than a zero-momentum pair [3, 4]. Therefore, pp-wave magnets can serve as a platform for realizing the FF states through an intrinsic mechanism, requiring only an arbitrarily weak field.

Experimental signatures

We investigate the quasiparticle dispersions as a potential experimental probe for distinguishing the different superconducting states. We focus on (U,V)=(1,3)(U,V)=(1,3) as in Figure 2(a) and take (μ,J,h)=(0,1,0)(\mu,J,h)=(0,1,0), (0,2,0)(0,2,0), (0,2,0.2)(0,2,0.2), (−2.3,1,0)(-2.3,1,0), and (−2.3,1,0.2)(-2.3,1,0.2) for the even-parity A1A_1, odd-parity A1A_1, FF-pxp_x, B2B_2, and FF-pyp_y states, respectively, so that each state is the stabilized ground state at the chosen parameters [Figure 6(a)–(Figure 6e)]. The dispersions of the even-parity and odd-parity A1A_1 states are plotted using Eq. (36), that of the B2B_2 state using Eq. (39), and those of the FF-pxp_x and FF-pyp_y states using Eq. (44).

Experimental signatures: quasiparticle dispersions and density of states

Figure 6. Experimental signatures. (a)–(e) Quasiparticle dispersions of (a) the even-parity A1A_1, (b) the odd-parity A1A_1, (c) the FF-pxp_x, (d) the B2B_2, and (e) the FF-pyp_y states along the square loop of (kx,ky)(k_x,k_y), (0,0)→(φJ,0)→(φJ,π)→(0,π)→(0,0)(0,0) \to(\varphi_J,0) \to(\varphi_J,\pi) \to(0,\pi) \to(0,0), whose four segments are drawn with equal length. Solid and dashed curves correspond to σ=↑\sigma= \uparrow and ↓\downarrow, respectively, and dotted gray curves show the normal-state dispersions ±ϵ~σ(k)\pm\tilde{\epsilon}_{\sigma}(\boldsymbol{k}). (f) Density of states N(E)N(E) of the states in (a)–(e), computed from their quasiparticle energies with the same parameters as in the corresponding panels, summed over both spins on a 1600×16001600 \times1600 momentum grid and broadened by a Gaussian of width 0.0150.015. The parameters (μ,J,h)(\mu,J,h) are (a) (0,1,0)(0,1,0), (b) (0,2,0)(0,2,0), (c) (0,2,0.2)(0,2,0.2), (d) (−2.3,1,0)(-2.3,1,0), and (e) (−2.3,1,0.2)(-2.3,1,0.2), with (U,V)=(1,3)(U,V)=(1,3) in all panels.

The even-parity A1A_1 dispersion displays nodal points at (kx,ky)=(±π/4,±π/2)(k_x,k_y)=(\pm\pi/4,\pm\pi/2) for each spin, of which only the spin-up node at (kx,ky)=(π/4,π/2)(k_x,k_y)=(\pi/4,\pi/2) is shown in Figure 6(a). The presence of these nodal points can be understood from the condition that the normal-state dispersion and the gap function vanish simultaneously, namely ϵ~(k′)=−2t1cos⁡kx′−2tcos⁡ky′−μ=0\tilde{\epsilon}(\boldsymbol{k}')=-2t_1\cos k'_x-2t\cos k'_y-\mu=0 and ΔA1even(k′)=Δs+Δecos⁡kx′+Δsycos⁡ky′=0\Delta^{\mathrm{even}}_{A_1}(\boldsymbol{k}')=\Delta_s+\Delta_e\cos k'_x+\Delta_{s_y}\cos k'_y=0. Each condition defines lines in the kx′k'_x–ky′k'_y plane, and the quasiparticle energy vanishes only at their intersections.

For a dd-wave-like pattern with Δe≈−Δsy\Delta_e\approx-\Delta_{s_y} and Δs≈0\Delta_s\approx0 [for example, the J<1.44J<1.44 region in Figure 3(a)], the condition ΔA1even(k′)=0\Delta^{\mathrm{even}}_{A_1}(\boldsymbol{k}')=0 reduces to the two diagonals ky′=±kx′k'_y=\pm k'_x. These diagonals always cross the two segments of the contour ϵ~(k′)=0\tilde{\epsilon}(\boldsymbol{k}')=0 at four points related by the mirror symmetries kx′→−kx′k'_x\to-k'_x and ky′→−ky′k'_y\to-k'_y, giving rise to four nodal points for each spin branch of the quasiparticle energies and hence eight in total. When the A1A_1 solution deviates substantially from the dd-wave-like pattern [for example, the ∣μ∣>1.53|\mu|>1.53 region in Figure 3(b)], the two lines may in principle intersect at only two points or not at all. We find that the self-consistent values of Δs\Delta_s, Δe\Delta_e, and Δsy\Delta_{s_y} remain in the range where the two lines intersect at four points over most of the region where the even-parity A1A_1 state is the ground state, including the even-parity A1A_1 region of the μ\mu-cut at J=2J=2 in Figure 3(b), ensuring eight nodal points in total in the Brillouin zone. Only at large ∣μ∣|\mu| for small JJ, beyond ∣μ∣≃2.0|\mu|\simeq2.0 at J≃0J\simeq0, where Δe\Delta_e and Δsy\Delta_{s_y} take the same sign, do the two lines no longer intersect, so that the state is fully gapped there, while a narrow band along the outer edge for J≳1.1J \gtrsim1.1 hosts only four nodal points in total (see Supplementary Note 4 for details).

By contrast, the odd-parity A1A_1 state does not possess nodal points [Figure 6(b)]. This behavior can be understood by noting that the condition ΔA1odd(k′)=Δosin⁡kx′=0\Delta^{\mathrm{odd}}_{A_1}(\mathbf{k}') = \Delta_o \sin k'_x = 0 defines the straight lines kx′=0k'_x = 0 and kx′=πk'_x = \pi. Near μ=0\mu= 0 and J≥1.4J \ge1.4, where this state is stabilized, the exchange field elongates the two segments of the contour ϵ~(k′)=0\tilde{\epsilon}(\mathbf{k}') = 0 along ky′k'_y, so that they run parallel to these lines and avoid them [Figure 4(c)]. For sufficiently small JJ and large ∣μ∣|\mu|, the contour can in principle cross these lines at two points for each spin branch, giving four nodal points in total [Figure 4(g)], but the corresponding parameters lie outside the stabilized region in the JJ–μ\mu plane (see Supplementary Note 4). The same reasoning applies to the FF-pxp_x state at h=0h = 0, whose dispersion reduces to that of the odd-parity A1A_1 state. At finite hh, a sufficiently large field can in principle shift the spin-dependent chemical potential μ+hsσ\mu+ hs_\sigma into the range where the two lines cross, while the FF-pxp_x state remains stable. Such fields are, however, beyond the physically relevant range. For instance, taking t=0.5t = 0.5 eV, we find that the nodal points start to appear only for h≳0.08h \gtrsim0.08 in a limited region of (J,μ)(J,\mu), which corresponds to a magnetic field of approximately 700 T. The FF-pxp_x state is thus fully gapped in practice, similarly to the odd-parity A1A_1 state [Figure 6(c)]. The two states can nevertheless be distinguished by the field-induced spin splitting of the dispersion along the (0,0)(0,0)–(0,π)(0,\pi) line [Figure 6(b) and Figure 6(c)]. Along this line the normal-state dispersion is spin degenerate, because the spin splitting 2Jsσsin⁡kx2Js_\sigma\sin k_x vanishes at kx=0k_x = 0. The odd-parity A1A_1 dispersion therefore remains spin degenerate there, whereas the field shifts the two spin branches of the FF-pxp_x dispersion in opposite directions and splits them.

The dispersion of the B2B_2 state has nodal points at (kx,ky)≃(±0.68,0)(k_x,k_y) \simeq(\pm0.68,0) for each spin, of which only the spin-up node at (kx,ky)≃(0.68,0)(k_x,k_y) \simeq(0.68,0) is shown in Figure 6(d). Near μ=0\mu= 0, the two segments of the contour cross the gap zeros at ky′=0k'_y = 0 and π\pi at four points per spin branch for J>0J > 0 (two for J=0J = 0), giving eight nodal points in total, whereas for sufficiently large ∣μ∣|\mu| they cross at two points per spin branch, giving four in total [Figure 4(d) and Figure 4(h)]. In contrast to the odd-parity A1A_1 state, these nodal points are unavoidable. The gap zeros of the B2B_2 state run perpendicular to the elongation of the contour along ky′k'_y, so the contour necessarily cuts through them, whereas those of the odd-parity A1A_1 state run parallel to it. Since the FF-pyp_y state shares this nodal-point structure, with μ\mu replaced by the spin-dependent chemical potential μ+hsσ\mu+ hs_\sigma [Figure 6(e)], the nodal points alone cannot tell it apart from the B2B_2 state. As with the odd-parity A1A_1 and FF-pxp_x pair, the two can instead be identified from the dispersion along the (0,0)(0,0)–(0,π)(0,\pi) line, where the exchange splitting vanishes [Figure 6(d) and Figure 6(e)]. Here the two spin branches of the B2B_2 dispersion coincide, while those of the FF-pyp_y dispersion are separated by the Zeeman field.

We finally examine the density of states N(E)N(E) computed from the quasiparticle energies of the states in Figure 6(a)–(Figure 6e) [Figure 6(f)]. The odd-parity A1A_1 and FF-pxp_x states show hard gaps at low energy, reflecting their fully gapped dispersions, whereas the even-parity A1A_1, B2B_2, and FF-pyp_y states retain finite spectral weight down to zero energy because of their nodal points. Examining the existence of a hard gap therefore allows the odd-parity A1A_1 and FF-pxp_x states to be distinguished from the even-parity A1A_1, B2B_2, and FF-pyp_y states. Within each of these groups, the field-induced splitting pattern provides a further distinction. The odd-parity A1A_1 and FF-pxp_x states exhibit different satellite peaks in the high-energy region 2≲E≲2.72 \lesssim E \lesssim2.7, which can be attributed to the Zeeman splitting of the FF-pxp_x dispersion. The B2B_2 and FF-pyp_y states, by contrast, display distinct features at lower energies, since the nodal points of the FF-pyp_y state are displaced from those of the B2B_2 state in a spin-dependent manner by the field. Measuring the evolution of N(E)N(E) under an applied field could therefore potentially discriminate among all five states.

Discussion

In this study, we discovered that the FS displacement induced by the pp-wave exchange field provides a mechanism for realizing equal-spin, FF-type pairing states. In contrast to the classical FF states [3], the pairing found here incorporates the entire FS owing to the equal-energy condition, maximizing the pairing amplitude. These FF states become degenerate with the usual zero-momentum pairing states through the FS geometry, which is tuned jointly by the pp-wave exchange field and the chemical potential, and become the leading pairing instability under arbitrarily small Zeeman fields. These findings pave the way toward realizing the long-sought FF states through an intrinsic mechanism, rather than through the extrinsic mechanisms used in previous approaches, such as strong external magnetic fields or proximity coupling to magnetic materials.

For experimental realization, we note the recent discovery of a pp-wave magnet induced by a commensurate spin helix [30], which gives rise to the required pp-wave exchange field. Alternatively, one could consider orbital order [34], or induce the required odd-parity exchange field via light irradiation of conventional collinear antiferromagnets through Floquet engineering [42, 43]; see Ref. [33] for a systematic symmetry classification of pp-wave magnetism. The formation of the FF-pxp_x state could then be detected by tracking the unconventional, spin-split nodal-point structure of the BdG quasiparticle bands identified here, via scanning tunneling spectroscopy and, in particular, Bogoliubov quasiparticle interference imaging, which directly resolves the momentum dependence of the excitation gap [44, 45].

It remains an important task for future work to extend the current effective-model approach to more realistic model setups. Specifically, one can employ two-sublattice models [38, 39] that capture the underlying spin texture responsible for the pp-wave exchange field, or incorporate spin-orbit coupling; the latter might affect the subtle energetic competition between the FF and non-FF states observed here, thereby controlling their instability. One might also consider higher-order exchange fields (dd-, ff-, gg-wave types, etc.) [46] to explore their realization. For this direction, however, we point out that these higher-order exchange fields are not expected to induce an FS displacement, a crucial distinction that limits the direct applicability of the matched-momentum mechanism. In this regard, we argue that the pp-wave exchange field plays a distinctive role in stabilizing finite-momentum pairing states.

Methods

Gap equations of A1 states

Evaluating the anomalous averages in Eq. (7) at zero temperature casts the gap equations in the explicit form [39]

ΔA1=MΔA1,(32)\Delta_{A_1}=\mathcal{M}\Delta_{A_1}, \tag*{(32)}

where ΔA1≡(Δs,Δsx,Δsy,Δpx)T\Delta_{A_1}\equiv(\Delta_s,\Delta_{s_x},\Delta_{s_y},\Delta_{p_x})^{\mathrm{T}} and

M=(UIs,sUIs,sxUIs,syUIs,px2VIsx,s2VIsx,sx2VIsx,sy2VIsx,px2VIsy,s2VIsy,sx2VIsy,sy2VIsy,px2VIpx,s2VIpx,sx2VIpx,sy2VIpx,px),(33)\mathcal{M}= \begin{pmatrix} U\mathcal{I}_{s,s} & U\mathcal{I}_{s,s_x} & U\mathcal{I}_{s,s_y} & U\mathcal{I}_{s,p_x}\\ 2V\mathcal{I}_{s_x,s} & 2V\mathcal{I}_{s_x,s_x} & 2V\mathcal{I}_{s_x,s_y} & 2V\mathcal{I}_{s_x,p_x}\\ 2V\mathcal{I}_{s_y,s} & 2V\mathcal{I}_{s_y,s_x} & 2V\mathcal{I}_{s_y,s_y} & 2V\mathcal{I}_{s_y,p_x}\\ 2V\mathcal{I}_{p_x,s} & 2V\mathcal{I}_{p_x,s_x} & 2V\mathcal{I}_{p_x,s_y} & 2V\mathcal{I}_{p_x,p_x} \end{pmatrix}, \tag*{(33)}

with the overlap integrals

Iα,β=1N∑kgα(k)gβ(k)Θ(E↑A1(k))2ϵ0(k)2+∣ΔA1(k)∣2,(34)\mathcal{I}_{\alpha,\beta}=\frac{1}{N}\sum_{\boldsymbol{k}}\frac{g_{\alpha}(\boldsymbol{k})g_{\beta}(\boldsymbol{k})\Theta(E^{A_1}_{\uparrow}(\boldsymbol{k}))}{2\sqrt{\epsilon_0(\boldsymbol{k})^2+\lvert\Delta_{A_1}(\boldsymbol{k})\rvert^2}}, \tag*{(34)}

for α,β∈{s,sx,sy,px}\alpha,\beta\in\{s,s_x,s_y,p_x\}. Here ϵ0(k)=12[ϵ↑(k)+ϵ↓(−k)]\epsilon_0(\boldsymbol{k})=\frac{1}{2}[\epsilon_{\uparrow}(\boldsymbol{k})+\epsilon_{\downarrow}(-\boldsymbol{k})] denotes the symmetric combination of the dispersions of the two paired electrons, in which the Zeeman term cancels,

ϵ0(k)=−2t1cos⁡(kx+φJ)−2tcos⁡ky−μ,(35)\epsilon_0(\boldsymbol{k})=-2t_1\cos(k_x+\varphi_J)-2t\cos k_y-\mu, \tag*{(35)}

and EσA1(k)E^{A_1}_{\sigma}(\boldsymbol{k}) denotes the quasiparticle energy

EσA1(k)=ϵ0(k)2+∣ΔA1(sσk)∣2−hsσ.(36)E^{A_1}_{\sigma}(\boldsymbol{k})=\sqrt{\epsilon_0(\boldsymbol{k})^2+\lvert\Delta_{A_1}(s_{\sigma}\boldsymbol{k})\rvert^2}-h s_{\sigma}. \tag*{(36)}

The ground-state energy is

FA1=∑k[ϵ↓(k)−E↓A1(k)]+N2V∑α=sx,sy,px∣Δα∣2+NU∣Δs∣2+∑kΘ(−E↑A1(k))E↑A1(k).(37)F_{A_1}=\sum_{\boldsymbol{k}}\left[\epsilon_{\downarrow}(\boldsymbol{k})-E^{A_1}_{\downarrow}(\boldsymbol{k})\right]+\frac{N}{2V}\sum_{\alpha=s_x,s_y,p_x}\lvert\Delta_{\alpha}\rvert^2+\frac{N}{U}\lvert\Delta_s\rvert^2+\sum_{\boldsymbol{k}}\Theta\left(-E^{A_1}_{\uparrow}(\boldsymbol{k})\right)E^{A_1}_{\uparrow}(\boldsymbol{k}). \tag*{(37)}

The derivations of these equations are given in Supplementary Note 5.

Gap equations of B2 states

Evaluating the anomalous average in Eq. (10) at zero temperature casts the B2B_2 gap equation in the explicit form

Δpy=VN∑kΔpygpy(k)2Θ(E↑B2(k))ϵ0(k)2+∣ΔB2(k)∣2,(38)\Delta_{p_y}=\frac{V}{N}\sum_{\boldsymbol{k}}\frac{\Delta_{p_y}g_{p_y}(\boldsymbol{k})^2\Theta(E^{B_2}_{\uparrow}(\boldsymbol{k}))}{\sqrt{\epsilon_0(\boldsymbol{k})^2+\lvert\Delta_{B_2}(\boldsymbol{k})\rvert^2}}, \tag*{(38)}

where the quasiparticle energies of the two paired modes are

EσB2(k)=ϵ0(sσk)2+∣ΔB2(sσk)∣2−hsσ.(39)E^{B_2}_{\sigma}(\boldsymbol{k})=\sqrt{\epsilon_0(s_{\sigma}\boldsymbol{k})^2+\lvert\Delta_{B_2}(s_{\sigma}\boldsymbol{k})\rvert^2}-h s_{\sigma}. \tag*{(39)}

The ground-state energy is

FB2=∑k[ϵ↓(k)−E↓B2(k)]+N2V∣Δpy∣2+∑kΘ(−E↑B2(k))E↑B2(k).(40)F_{B_2}=\sum_{\boldsymbol{k}}\left[\epsilon_{\downarrow}(\boldsymbol{k})-E^{B_2}_{\downarrow}(\boldsymbol{k})\right]+\frac{N}{2V}\lvert\Delta_{p_y}\rvert^2+\sum_{\boldsymbol{k}}\Theta\left(-E^{B_2}_{\uparrow}(\boldsymbol{k})\right)E^{B_2}_{\uparrow}(\boldsymbol{k}). \tag*{(40)}

The derivations of these equations are given in Supplementary Note 6.

Gap equations of FF states

Evaluating the anomalous average in Eq. (15) at zero temperature brings this to the standard form

Δα;σFF=VN∑kΔα;σFF∣gα(k)∣2ϵ~σ(k)ϵ~σ(k)2+∣Δα;σFFgα(k)∣2,(41)\Delta^{\mathrm{FF}}_{\alpha;\sigma}=\frac{V}{N}\sum_{\boldsymbol{k}}\frac{\Delta^{\mathrm{FF}}_{\alpha;\sigma}\lvert g_{\alpha}(\boldsymbol{k})\rvert^2}{\tilde{\epsilon}_{\sigma}(\boldsymbol{k})\sqrt{\tilde{\epsilon}_{\sigma}(\boldsymbol{k})^2+\lvert\Delta^{\mathrm{FF}}_{\alpha;\sigma}g_{\alpha}(\boldsymbol{k})\rvert^2}}, \tag*{(41)}

for α=px,py\alpha=p_x,p_y. The ground-state energy is

FαFF=12∑k,σ[ϵ~σ(k)−E~α;σFF(k)]+N4V∑σ∣Δα;σFF∣2,(42)F^{\mathrm{FF}}_{\alpha}=\frac{1}{2}\sum_{\boldsymbol{k},\sigma}\left[\tilde{\epsilon}_{\sigma}(\boldsymbol{k})-\tilde{E}^{\mathrm{FF}}_{\alpha;\sigma}(\boldsymbol{k})\right]+\frac{N}{4V}\sum_{\sigma}\lvert\Delta^{\mathrm{FF}}_{\alpha;\sigma}\rvert^2, \tag*{(42)}

where the quasiparticle energy is

E~α;σFF(k)=ϵ~σ(k)2+∣Δα;σFFgα(k)∣2.(43)\tilde{E}^{\mathrm{FF}}_{\alpha;\sigma}(\boldsymbol{k})=\sqrt{\tilde{\epsilon}_{\sigma}(\boldsymbol{k})^2+\lvert\Delta^{\mathrm{FF}}_{\alpha;\sigma}g_{\alpha}(\boldsymbol{k})\rvert^2}. \tag*{(43)}

The derivations of these equations are provided in Supplementary Note 7. To plot the quasiparticle dispersions for the FF-pxp_x and FF-pyp_y states Figure 6(c) and (Figure 6e), we used

Eα;σFF(k)=ϵσ(k)2+∣Δα;σFFgα(k−Qσ2)∣2,(44)E^{\mathrm{FF}}_{\alpha;\sigma}(\boldsymbol{k})=\sqrt{\epsilon_{\sigma}(\boldsymbol{k})^2+\left\lvert\Delta^{\mathrm{FF}}_{\alpha;\sigma}g_{\alpha}\left(\boldsymbol{k}-\frac{\boldsymbol{Q}_{\sigma}}{2}\right)\right\rvert^2}, \tag*{(44)}

which follows from Eq. (43) by the substitution k→k−Qσ/2\boldsymbol{k}\to\boldsymbol{k}-\boldsymbol{Q}_{\sigma}/2 for each spin σ\sigma, so that the energies of both spins are expressed as functions of the same physical momentum.

Solving gap equations

The gap equations [Eqs. (32), (38), and (41)] were solved iteratively. All momentum sums were evaluated in the shifted coordinates k′\mathbf{k}' of Eq. (25) on a grid of 300×300300 \times300 points, so that all four pairing states are described by the same dispersion −2t1cos⁡kx′−2tcos⁡ky′−μ-2t_1 \cos k'_x - 2t \cos k'_y - \mu on the same grid. At each step, the right-hand side of the gap equation was evaluated with the current pairing amplitudes and taken as the amplitudes of the next step, an update which typically does not increase the ground-state energy. The iteration was terminated when the change of every amplitude between successive steps fell below 10−810^{-8}, or after 300 steps. When all amplitudes fell below 10−610^{-6}, the solution was identified as the normal state. For the coupled A1A_1 gap equations, the iteration was started from four initial configurations, (Δs,Δsx,Δsy,Δpx)=(1,1,1,0)(\Delta_s, \Delta_{s_x}, \Delta_{s_y}, \Delta_{p_x}) = (1,1,1,0), (0,−1,1,0)(0,-1,1,0), (0,0,0,1)(0,0,0,1), and (0,−1,1,1)(0,-1,1,1), corresponding to ss-wave-like, dd-wave-like, odd-parity, and mixed-parity configurations, respectively, and the solution with the lowest ground-state energy was retained. The B2B_2 and FF

gap equations, each involving a single amplitude, were iterated from Δpy=1\Delta_{p_y} = 1 and Δα;σFF=1\Delta^{\mathrm{FF}}_{\alpha;\sigma} = 1, with the FF gap equation solved separately for each spin σ\sigma. The resulting pairing amplitudes were substituted into the ground-state energy expressions [Eqs. (37), (40), and (42)], and the state with the lowest energy was identified as the superconducting ground state, from which the phase diagrams in Figure 2 and Figure 5 were constructed. Energies differing by less than 10−710^{-7} were regarded as degenerate, and the normal state was assigned unless a superconducting state lowered the energy by more than 10−710^{-7}.

Acknowledgments

This research was supported by an appointment to the JRG Program at the APCTP through the Science and Technology Promotion Fund and Lottery Fund of the Korean Government, the Korean Local Governments (Gyeongsangbuk-do Province and Pohang City), and the National Research Foundation of Korea (NRF) funded by the Korean government (Ministry of Science and ICT, MSIT) (No. RS-2026-25499525).

References

  1. [1]R. Casalbuoni and G. Nardulli, Inhomogeneous superconductivity in condensed matter and qcd, Rev. Mod. Phys. 76, 263 (2004).DOI
  2. [2]D. F. Agterberg, J. S. Davis, S. D. Edkins, E. Fradkin, D. J. Van Harlingen, S. A. Kivelson, P. A. Lee, L. Radzihovsky, J. M. Tranquada, and Y. Wang, The physics of pair-density waves: Cuprate superconductors and beyond, Annu. Rev. Condens. Matter Phys. 11, 231 (2020).DOI
  3. [3]P. Fulde and R. A. Ferrell, Superconductivity in a strong spin-exchange field, Phys. Rev. 135, A550 (1964).DOI
  4. [4]A. I. Larkin and Y. N. Ovchinnikov, Nonuniform state of superconductors, Sov. Phys. JETP 20, 762 (1965), [Zh. Eksp. Teor. Fiz. 47, 1136 (1964)].
  5. [5]E. A. Demler, G. B. Arnold, and M. R. Beasley, Superconducting proximity effects in magnetic metals, Phys. Rev. B 55, 15174 (1997).DOI
  6. [6]V. V. Ryazanov, V. A. Obnoznov, A. Y. Rusanov, A. V. Veretennikov, A. A. Golubov, and J. Aarts, Coupling of two superconductors through a ferromagnet: Evidence for a π junction, Phys. Rev. Lett. 86, 2427 (2001).pubmed.ncbi.nlm.nih.gov/11289946
  7. [7]T. Kontos, M. Aprili, J. Lesueur, F. Genêt, B. Stephanidis, and R. Boursier, Josephson junction through a thin ferromagnetic layer: Negative coupling, Phys. Rev. Lett. 89, 137007 (2002).pubmed.ncbi.nlm.nih.gov/12225057
  8. [8]Y. Matsuda and H. Shimahara, Fulde–ferrell–larkin–ovchinnikov state in heavy fermion superconductors, J. Phys. Soc. Jpn 76, 051005 (2007).DOI
  9. [9]E. Berg, E. Fradkin, and S. A. Kivelson, Theory of the striped superconductor, Phys. Rev. B 79, 064515 (2009).DOI
  10. [10]E. Fradkin, S. A. Kivelson, and J. M. Tranquada, Colloquium: Theory of intertwined orders in high temperature superconductors, Rev. Mod. Phys. 87, 457 (2015).DOI
  11. [11]P. A. Lee, Amperean pairing and the pseudogap phase of cuprate superconductors, Phys. Rev. X 4, 031017 (2014).arxiv.org/abs/1401.0519
  12. [12]I. I. Mazin, Notes on antiferromagnetism and superconductivity, AAPPS Bulletin 35, 18 (2025).
  13. [13]Y. Fukaya, B. Lu, K. Yada, Y. Tanaka, and J. Cayao, Superconducting phenomena in systems with unconventional magnets, J. Phys. Condens. Matter 37, 313003 (2025).DOI
  14. [14]S. Lee, S. Lee, S. Jung, et al., Broken kramers degeneracy in altermagnetic mnte, Phys. Rev. Lett. 132, 036702 (2024).DOI
  15. [15]J. Krempaský, L. Šmejkal, S. W. D’Souza, et al., Altermagnetic lifting of kramers spin degeneracy, Nature 626, 517 (2024).DOI
  16. [16]J. Ding, Z. Jiang, X. Chen, et al., Large band splitting in g-wave altermagnet crsb, Phys. Rev. Lett. 133, 206401 (2024).DOI
  17. [17]S. Reimers, L. Odenbreit, L. Šmejkal, et al., Direct observation of altermagnetic band splitting in crsb thin films, Nat. Commun. 15, 2116 (2024).
  18. [18]G. Yang, Z. Li, S. Yang, J. Li, H. Zheng, W. Zhu, Z. Pan, Y. Xu, S. Cao, W. Zhao, A. Jana, J. Zhang, M. Ye, Y. Song, L.-H. Hu, L. Yang, J. Fujii, I. Vobornik, M. Shi, H. Yuan, Y. Zhang, Y. Xu, and Y. Liu, Three-dimensional mapping of the altermagnetic spin splitting in crsb, Nat. Commun. 16, 1442 (2025).DOI
  19. [19]S. Santosh, P. Corbae, W. J. Yáñez-Parreño, S. Ghosh, C. J. Jensen, A. V. Fedorov, M. Hashimoto, D. Lu, J. A. Borchers, A. J. Grutter, T. R. Charlton, S. Islam, D. Golovanova, Y. Zhao, A. Tauraso, A. Richardella, B. Yan, K. A. Mkhoyan, C. J. Palmstrøm, Y. Ou, and N. Samarth, Altermagnetic band splitting in 10 nm epitaxial crsb thin films, Adv. Mater. 37, e08977 (2025).
  20. [20]F. Zhang, X. Cheng, Z. Yin, C. Liu, L. Deng, Y. Qiao, Z. Shi, S. Zhang, J. Lin, Z. Liu, M. Ye, Y. Huang, X. Meng, C. Zhang, T. Okuda, K. Shimada, S. Cui, Y. Zhao, G.-H. Cao, S. Qiao, J. Liu, and C. Chen, Crystal-symmetry-paired spin–valley locking in a layered room-temperature metallic altermagnet candidate, Nat. Phys. 21, 760 (2025).DOI
  21. [21]B. Jiang, M. Hu, J. Bai, Z. Song, C. Mu, G. Qu, W. Li, W. Zhu, H. Pi, Z. Wei, Y.-J. Sun, Y. Huang, X. Zheng, Y. Peng, L. He, S. Li, J. Luo, Z. Li, G. Chen, H. Li, H. Weng, and T. Qian, A metallic room-temperature d-wave altermagnet, Nat. Phys. 21, 754 (2025).
  22. [22]S.-B. Zhang, L.-H. Hu, and T. Neupert, Finite-momentum Cooper pairing in proximitized altermagnets, Nat. Commun. 15, 1801 (2024).
  23. [23]D. Chakraborty and A. M. Black-Schaffer, Zero-field finite-momentum and field-induced superconductivity in altermagnets, Phys. Rev. B 110, L060508 (2024).arxiv.org/abs/2309.14427
  24. [24]G. Sim and J. Knolle, Pair density waves and supercurrent diode effect in altermagnets, Phys. Rev. B 112, L020502 (2025).DOI
  25. [25]S. Hong, M. J. Park, and K.-M. Kim, Unconventional p-wave and finite-momentum superconductivity induced by altermagnetism through the formation of bogoliubov fermi surface, Phys. Rev. B 111, 054501 (2025).DOI
  26. [26]H. Hu, Z. Liu, and X.-J. Liu, Unconventional superconductivity of an altermagnetic metal: Polarized bcs and inhomogeneous \u↔llo states, Phys. Rev. B 112, 184501 (2025).DOI
  27. [27]Z. Liu, H. Hu, and X.-J. Liu, Fulde-ferrell-larkin-ovchinnikov states and topological bogoliubov fermi surfaces in altermagnets: An analytical study, Phys. Rev. B 113, 024518 (2026).DOI
  28. [28]N. Parthenios, P. M. Bonetti, R. González-Hernández, W. H. Campos, L. Šmejkal, and L. Classen, Spin and pair density waves in two-dimensional altermagnetic metals, Phys. Rev. B 112, 214410 (2025).DOI
  29. [29]S. A. Chen, Q. Chen, G. Yu, and Z. Zhu, Pair-density wave from doping an altermagnetic mott insulator (2026), arXiv:2607.23654 [cond-mat.str-el].arxiv.org/abs/2607.23654
  30. [30]R. Yamada, M. T. Birch, P. R. Baral, S. Okumura, R. Nakano, S. Gao, M. Ezawa, T. Nomoto, J. Masell, Y. Ishihara, K. K. Kolincio, I. Belopolски, H. Sagayama, H. Nakao, K. Ohishi, T. Ohhara, R. Kiyanagi, T. Nakajima, Y. Tokura, T.-h. Arima, Y. Motome, M. M. Hirschmann, and M. Hirschberger, A metallic p-wave magnet with commensurate spin helix, Nature 646, 837 (2025).
  31. [31]A. B. Hellenes, T. Jungwirth, R. Jaeschke-Ubiergo, A. Chakraborty, J. Sinova, and L. Šmejkal, p-wave magnets, arXiv:2309.01607 (2023).arxiv.org/abs/2309.01607
  32. [32]G. Sim and S. Rachel, Quantum spin models of commensurate p-wave magnets, Phys. Rev. Lett. 137, 026503 (2025).
  33. [33]M. Zeng, Z. Qin, L. Qin, S. Feng, L. Wu, D.-H. Xu, and R. Wang, Odd-parity altermagnetism: A spin group study, Phys. Rev. B 113, L220412 (2026).DOI
  34. [34]Z.-Y. Zhuang, D. Zhu, D. Liu, Z. Wu, and Z. Yan, Odd-parity altermagnetism originated from orbital orders (2026).
  35. [35]Y. Nagae, L. Katayama, and S. Ikegaya, Flat-band zero-energy states and anomalous proximity effects in p-wave magnet-superconductor hybrid systems, Phys. Rev. B 111, 174519 (2025).DOI
  36. [36]Z.-T. Sun, X. Feng, Y.-M. Xie, B. T. Zhou, J.-X. Hu, and K. T. Law, Pseudo-ising superconductivity induced by p-wave magnetism, Phys. Rev. B 112, 214504 (2025).DOI
  37. [37]P. Sukhachov, H. G. Gili, B. Brekke, and J. Linder, Coexistence of p-wave magnetism and superconductivity, Phys. Rev. B 111, L220403 (2025).DOI
  38. [38]M. Khodas, L. Šmejkal, and I. I. Mazin, Nonrelativistic ising superconductivity in p-wave magnets, Phys. Rev. B 114, L080505 (2026).
  39. [39]K.-M. Kim, G. Sim, and M. J. Park, Topological ising superconductivity in two-dimensional p-wave magnet (2026), 2605.01686.arxiv.org/abs/2605.01686
  40. [40]A. Pal, P. Dutta, and A. Saha, Emergent superconducting phases in unconventional p-wave magnets (2026), 2603.03221.arxiv.org/abs/2603.03221
  41. [41]M. Sigrist and K. Ueda, Phenomenological theory of unconventional superconductivity, Rev. Mod. Phys. 63, 239 (1991).DOI
  42. [42]S. Huang, Z. Qin, F. Zhan, D.-H. Xu, D.-S. Ma, and R. Wang, Light-induced odd-parity magnetism in conventional antiferromagnetism, Phys. Rev. Lett. 136, 126703 (2026).DOI
  43. [43]T. Zhu, D. Zhou, H. Wang, S.-H. Wei, and J. Ruan, Floquet odd-parity collinear magnets, Phys. Rev. Lett. 136, 126704 (2026).DOI
  44. [44]T. Hanaguri, S. Niitaka, K. Kuroki, and H. Takagi, Unconventional s-wave superconductivity in Fe(Se,Te), Science 328, 474 (2010).DOI
  45. [45]M. P. Allan, A. W. Rost, A. P. Mackenzie, Y. Xie, J. C. Davis, K. Kihou, C. H. Lee, A. Iyo, H. Eisaki, and T.-M. Chuang, Anisotropic energy gaps of iron-based superconductivity from intraband quasiparticle interference in LiFeAs, Science 336, 563 (2012).
  46. [46]L. Šmejkal, J. Sinova, and T. Jungwirth, Beyond conventional ferromagnetism and antiferromagnetism: A phase with nonrelativistic spin and crystal rotation symmetry, Phys. Rev. X 12, 031042 (2022).DOI

Paper details

vlnc:W_xnsfgm26
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 1

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