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

Spin Susceptibility of a J=3/2J=3/2 Superconductor

Dakyeong Kim1    Takumi Sato1    Shingo Kobayashi2    Yasuhiro Asano1 1Department of Applied Physics, Hokkaido University, Sapporo 060-8628, Japan.
2RIKEN Center for Emergent Matter Science, Wako, Saitama 351-0198, Japan.
Abstract

We discuss the spin susceptibility of superconductors in which a Cooper pair consists of two electrons having the angular momentum J=3/2J=3/2 due to strong spin-orbit interactions. The susceptibility is calculated analytically for several pseudospin quintet states in a cubic superconductor within the linear response to a Zeeman field. The susceptibility for A1gA_{1g} symmetry states is isotropic in real space. For EgE_{g} and T2gT_{2g} symmetry cases, the results depend sensitively on choices of order parameter. The susceptibility is isotropic for a T2gT_{2g} symmetry state, whereas it becomes anisotropic for an EgE_{g} symmetry state. We also find in a T2gT_{2g} state that the susceptibility tensor has off-diagonal elements.

I Introduction

Spin-orbit interaction is a source of exotic electronic states realized in topological semimetals [1, 2], topological insulators [3, 4], and topological superconductors [5, 6, 7, 8, 9, 9]. In the presence of strong spin-orbit interactions, spin S=1/2S=1/2 and orbital angular momentum L=1L=1 of an electron are inseparable degrees of freedom. Electronic properties of such materials are characterized by an electron with pseudospin J=L+S=3/2J=L+S=3/2. Recent studies have suggested a possibility of superconductivity due to Cooper pairing between two electrons with J=3/2J=3/2 [10, 11]. The large angular momentum of an electron enriches the symmetry of the order parameter such as pseudospin-quintet even-parity and pseudospin-septet odd-parity [12, 13] in addition to conventional spin-singlet even-parity and spin-triplet odd-parity. Such high angular-momentum pairing states would feature superconducting phenomena of J=3/2J=3/2 superconductors [14, 15, 16, 17, 18, 19]. In particular, a large angular momentum of a Cooper pair would qualitatively change the magnetic response of a superconductor to an external magnetic field.

The spin susceptibility reflects well the internal spin structures of a Cooper pair. It is well known in spin-singlet superconductors that the spin susceptibility decreases monotonically with the decrease of temperature below TcT_{c} and vanishes at zero temperature [20]. This phenomenon occurs independently of the direction of a Zeeman field 𝑯\bm{H} because a Cooper pair has no spin. In spin-triplet superconductors, on the other hand, the susceptibility can be anisotropic depending on the relative alignment between a Zeeman field and a 𝒅\bm{d} vector in the order parameter. For 𝒅𝑯\bm{d}\perp\bm{H}, the spin susceptibility is constant independent of temperature. Thus, the unchanged Knight shift across TcT_{c} in experiments could be a strong evidence of spin-triplet superconductivity. For 𝒅𝑯\bm{d}\parallel\bm{H}, the susceptibility decreases with decreasing temperature below TcT_{c}. Such anisotropy is more remarkable when the number of components in a 𝒅\bm{d} vector is smaller. For J=3/2J=3/2 superconductors, however, our knowledge of the spin susceptibility is very limited to a theoretical paper that reported vanishing the spin susceptibility at zero temperature for a singlet-quintet mixed state in a centrosymmetric superconductor [21].

In this paper, we study theoretically the response of pseudospin-quintet even-parity superconductors to an external Zeeman field. The angular momentum of a Cooper pair in such superconductors is J=2J=2. Since the pairing symmetries of the pseudospin-quintet states are not well understood, we decided to calculate the spin susceptibility for the plausible pair potentials at a cubic superconductor preserving time-reversal symmetry. The spin susceptibility is analytically calculated based on the linear response formula [22]. The pair potential of pseudospin-quintet states is described by a five-component vector that couples to five 4×44\times 4 matrices in pseudospin space. Such complicated internal structures of the pair potential enrich the magnetic response of J=3/2J=3/2 superconductors. When the symmetry of the pair potential is high in pseudospin space (A1gA_{1g} state), the magnetic response is isotropic in real space and the spin susceptibility decreases monotonically with the decrease of temperature. The results are similar to those of 3He B-phase. When the pair potential is independent of wavenumber (T2gT_{2g} and EgE_{g} states), the spin susceptibility shows unique features to pseudospin-quintet states. The magnetic response becomes anisotropic in real space in an EgE_{g} state and the susceptibility tensor has finite off-diagonal elements in a T2gT_{2g} state.

This paper is organized as follows. In Sec. II, we described the electronic structure and the pair potential of a J=3/2J=3/2 superconductor in terms of five 4×44\times 4 matrices. The characteristic behaviors of the spin-susceptibility are discussed in Sec. III. The conclusion is given in Sec.IV. Algebras of 4×44\times 4 matrices and a number of mathematical relationships used in the paper are summarized in Appendices. Throughout this paper, we use the system of units =kB=c=1\hbar=k_{B}=c=1, where kBk_{B} is the Boltzmann constant and cc is the speed of light.

II J=3/2 Superconductor

We begin our analysis with the normal state Hamiltonian adopted in Ref. 13. The electronic states have four degrees of freedom consisting of two orbitals of equal parity and spin 1/2. In the presence of strong spin-orbit interactions, the effective Hamiltonian for a J=3/2J=3/2 electron is given by  [23]

N=\displaystyle\mathcal{H}_{\mathrm{N}}= 𝒌Ψ𝒌HN(𝒌)Ψ𝒌,\displaystyle\sum_{\bm{k}}\Psi_{\bm{k}}^{\dagger}\,H_{\mathrm{N}}(\bm{k})\,\Psi_{\bm{k}}, (1)
Ψ𝒌=\displaystyle\Psi_{\bm{k}}= [c𝒌,3/2,c𝒌,1/2,c𝒌,1/2,c𝒌,3/2]T,\displaystyle\left[c_{\bm{k},3/2},c_{\bm{k},1/2},c_{\bm{k},-1/2},c_{\bm{k},-3/2}\right]^{\mathrm{T}}, (2)

where T\mathrm{T} means the transpose of a matrix and c𝒌,jzc_{\bm{k},j_{z}} is the annihilation operator of an electron at 𝒌\bm{k} with the zz-component of angular momentum being jzj_{z}. The normal state Hamiltonian is represented by

HN(𝒌)=\displaystyle H_{\mathrm{N}}(\bm{k})= α𝒌2+β(𝒌𝑱)2μ=ξ𝒌 14×4+ϵ𝒌γ\displaystyle\alpha\bm{k}^{2}+\beta\left(\bm{k}\cdot\bm{J}\right)^{2}-\mu=\xi_{\bm{k}}\,1_{4\times 4}+\vec{\epsilon}_{\bm{k}}\cdot\vec{\gamma} (3)

with ξ𝒌=ϵ𝒌,0μ\xi_{\bm{k}}=\epsilon_{\bm{k},0}-\mu and

ϵ𝒌,0=\displaystyle\epsilon_{\bm{k},0}= (α+54β)𝒌2,ϵ𝒌,j=β𝒌2ejcj(𝒌^),\displaystyle\left(\alpha+\frac{5}{4}\beta\right)\bm{k}^{2},\quad\epsilon_{\bm{k},j}=\beta\,\bm{k}^{2}\,e_{j}\,c_{j}(\hat{\bm{k}}), (4)
c1(𝒌^)=\displaystyle c_{1}(\hat{\bm{k}})= 15k^xk^y,c2(𝒌^)=15k^yk^z,\displaystyle\sqrt{15}\,\hat{k}_{x}\,\hat{k}_{y},\quad c_{2}(\hat{\bm{k}})=\sqrt{15}\,\hat{k}_{y}\,\hat{k}_{z}, (5)
c3(𝒌^)=\displaystyle c_{3}(\hat{\bm{k}})= 15k^zk^x,c4(𝒌^)=152(k^x2k^y2),\displaystyle\sqrt{15}\,\hat{k}_{z}\,\hat{k}_{x},\quad c_{4}(\hat{\bm{k}})=\frac{\sqrt{15}}{2}\,(\hat{k}_{x}^{2}-\hat{k}_{y}^{2}), (6)
c5(𝒌^)=\displaystyle c_{5}(\hat{\bm{k}})= 52(2k^z2k^x2k^y2),\displaystyle\frac{\sqrt{5}}{2}\,(2\hat{k}_{z}^{2}-\hat{k}_{x}^{2}-\hat{k}_{y}^{2}), (7)

where k^j=kj/|𝒌|\hat{k}_{j}=k_{j}/|\bm{k}| for j=x,yj=x,y and zz represents the direction of wavenumber on the Fermi surface. The constants α>0\alpha>0 and β\beta determine the normal state property. The spin-orbit interactions increase with the increase of β>0\beta>0. The coefficients cjc_{j} are normalized as

ci(𝒌^)cj(𝒌^)𝒌^d𝒌^4πci(𝒌^)cj(𝒌^)=δi,j,\displaystyle\langle c_{i}(\hat{\bm{k}})\;c_{j}(\hat{\bm{k}})\rangle_{\hat{\bm{k}}}\equiv\int\frac{d\hat{\bm{k}}}{4\pi}\,c_{i}(\hat{\bm{k}})c_{j}(\hat{\bm{k}})=\delta_{i,j}, (8)

where 𝒌^\langle\cdots\rangle_{\hat{\bm{k}}} means the integral over the solid angle on the Fermi surface. The normalized five-component vector e=(e1,e2,e3,e4,e5)/|e|\vec{e}=(e_{1},e_{2},e_{3},e_{4},e_{5})/|\vec{e}| determines the dependence of the normal state dispersions on pseudospins. The spinors for the angular momenum of J=3/2J=3/2 are described by,

Jx=\displaystyle J_{x}= 12[0300302002030030],\displaystyle\frac{1}{2}\left[\begin{array}[]{cccc}0&\sqrt{3}&0&0\\ \sqrt{3}&0&2&0\\ 0&2&0&\sqrt{3}\\ 0&0&\sqrt{3}&0\end{array}\right], (13)
Jy=\displaystyle J_{y}= 12[0i300i302i002i0i300i30],\displaystyle\frac{1}{2}\left[\begin{array}[]{cccc}0&-i\sqrt{3}&0&0\\ i\sqrt{3}&0&-2i&0\\ 0&2i&0&-i\sqrt{3}\\ 0&0&i\sqrt{3}&0\end{array}\right], (18)
Jz=\displaystyle J_{z}= 12[3000010000100003].\displaystyle\frac{1}{2}\left[\begin{array}[]{cccc}3&0&0&0\\ 0&1&0&0\\ 0&0&-1&0\\ 0&0&0&-3\end{array}\right]. (23)

The 4×44\times 4 matrices in pseudospin space are defined as

γ1=\displaystyle\gamma^{1}= 13(JxJy+JyJx),γ2=13(JyJz+JzJy),\displaystyle\frac{1}{\sqrt{3}}(J_{x}J_{y}+J_{y}J_{x}),\quad\gamma^{2}=\frac{1}{\sqrt{3}}(J_{y}J_{z}+J_{z}J_{y}), (24)
γ3=\displaystyle\gamma^{3}= 13(JzJx+JxJz),γ4=13(Jx2Jy2),\displaystyle\frac{1}{\sqrt{3}}(J_{z}J_{x}+J_{x}J_{z}),\quad\gamma^{4}=\frac{1}{\sqrt{3}}(J_{x}^{2}-J_{y}^{2}), (25)
γ5=\displaystyle\gamma^{5}= 13(2Jz2Jx2Jy2),\displaystyle\frac{1}{3}(2J_{z}^{2}-J_{x}^{2}-J_{y}^{2}), (26)

and 14×41_{4\times 4} is the identity matrix. They satisfy the following relations

γνγλ+γλγν=2×14×4δν,λ,\displaystyle\gamma^{\nu}\,\gamma^{\lambda}+\gamma^{\lambda}\,\gamma^{\nu}=2\times 1_{4\times 4}\delta_{\nu,\lambda}, (27)
γ1γ2γ3γ4γ5=14×4,\displaystyle\gamma^{1}\,\gamma^{2}\,\gamma^{3}\,\gamma^{4}\,\gamma^{5}=-1_{4\times 4}, (28)
{γν}={γν}T=UTγνUT1,UT=γ1γ2,\displaystyle\{\gamma^{\nu}\}^{\ast}=\{\gamma^{\nu}\}^{\mathrm{T}}=U_{T}\,\gamma^{\nu}\,U_{T}^{-1},\quad U_{T}=\gamma^{1}\,\gamma^{2}, (29)

where UTU_{T} is the unitary part of the time-reversal operation 𝒯=UT𝒦\mathcal{T}=U_{T}\,\mathcal{K} with 𝒦\mathcal{K} meaning complex conjugation. The superconducting pair potential is represented as

Δ(𝒌)=η𝒌γUT,\displaystyle\Delta(\bm{k})=\vec{\eta}_{\bm{k}}\cdot\vec{\gamma}\,U_{T}, (30)

where a five-component vector η𝒌\vec{\eta}_{\bm{k}} represents an even-parity pseudospin-quintet state. Throughout this paper, we assume that all components of η𝒌\vec{\eta}_{\bm{k}} are real values. As a result of the Fermi-Dirac statistics of electrons, the pair potential is antisymmetric under the permutation of two pseudospins, (i.e., ΔT(𝒌)=Δ(𝒌)\Delta^{\mathrm{T}}(\bm{k})=-\Delta(\bm{k})). The Bogoliubov-de Gennes Hamiltonian reads,

HBdG(𝒌)=[HN(𝒌)Δ(𝒌)Δ~(𝒌)H~N(𝒌)],\displaystyle H_{\mathrm{BdG}}(\bm{k})=\left[\begin{array}[]{cc}H_{\mathrm{N}}(\bm{k})&\Delta(\bm{k})\\ -\undertilde{\Delta}(\bm{k})&-\undertilde{H}_{\mathrm{N}}(\bm{k})\end{array}\right], (33)

where X~(𝒌,iω)X(𝒌,iω)\undertilde{X}(\bm{k},i\omega)\equiv X^{\ast}(-\bm{k},i\omega) represents the particle-hole conjugation of X(𝒌,iω)X(\bm{k},i\omega).

The interaction with a uniform Zeeman field 𝑯\bm{H} is described by [23]

HZ=\displaystyle H_{\mathrm{Z}}= μB𝑱~𝑯,\displaystyle-\mu_{B}\tilde{\bm{J}}\cdot\bm{H}, (34)
J~j=\displaystyle\tilde{J}_{j}= g1Jj+g3Jj3,\displaystyle g_{1}{J}_{j}+g_{3}{J}_{j}^{3}, (35)

for j=x,y,j=x,y, and zz, where μB\mu_{B} is the Bohr’s magneton, and g1g_{1} and g3g_{3} are the coupling constants. The matrix structures of JjJ_{j} and Jj3J_{j}^{3} are displayed in Appendix A. The angular momenta in the Zeeman Hamiltonian are then given by

J~x=\displaystyle\tilde{J}_{x}= i2(3p1γ2γ5+p2γ1γ3+p1γ2γ4),\displaystyle-\frac{i}{2}\left(\sqrt{3}p_{1}\gamma^{2}\,\gamma^{5}+p_{2}\gamma^{1}\,\gamma^{3}+p_{1}\gamma^{2}\,\gamma^{4}\right), (36)
J~y=\displaystyle\tilde{J}_{y}= i2(3p1γ3γ5+p2γ1γ2p1γ3γ4),\displaystyle\frac{i}{2}\left(\sqrt{3}p_{1}\gamma^{3}\,\gamma^{5}+p_{2}\gamma^{1}\,\gamma^{2}-p_{1}\gamma^{3}\,\gamma^{4}\right), (37)
J~z=\displaystyle\tilde{J}_{z}= i2(p2γ2γ3+2p1γ1γ4),\displaystyle\frac{i}{2}\left(p_{2}\gamma^{2}\,\gamma^{3}+2p_{1}\gamma^{1}\,\gamma^{4}\right), (38)
p1=\displaystyle p_{1}= g1+74g3,p2=g1+134g3.\displaystyle g_{1}+\frac{7}{4}g_{3},\quad p_{2}=g_{1}+\frac{13}{4}g_{3}. (39)

In the linear response theory, the spin susceptibility is calculated by using the formula [22]

χμν=\displaystyle\chi_{\mu\nu}= χNδμ,ν(μB2)2Tωnd𝒌(2π)3\displaystyle\chi_{\mathrm{N}}\,\delta_{\mu,\nu}-\left(\frac{\mu_{B}}{2}\right)^{2}T\sum_{\omega_{n}}\int\frac{d\bm{k}}{(2\pi)^{3}}
×Tr\displaystyle\times\mathrm{Tr} [G(𝒌,iωn)J~μG(𝒌,iωn)J~ν\displaystyle\left[{G}(\bm{k},i\omega_{n})\,\tilde{J}_{\mu}\,G(\bm{k},i\omega_{n})\,\tilde{J}_{\nu}\right.
+F~(𝒌,iωl)J~μF(𝒌,iωn)(J~ν)\displaystyle+\undertilde{F}(\bm{k},i\omega_{l})\,\tilde{J}_{\mu}\,{F}(\bm{k},i\omega_{n})\,(\tilde{J}_{\nu})^{\ast}
GN(𝒌,iωn)J~μGN(𝒌,iωn)J~ν].\displaystyle\left.-{G}_{\mathrm{N}}(\bm{k},i\omega_{n})\,\tilde{J}_{\mu}\,G_{\mathrm{N}}(\bm{k},i\omega_{n})\,\tilde{J}_{\nu}\right]. (40)

The summation over the Matsubara frequency and that over the wavenumber are regularized by introducing the Green’s functions in the normal state GN{G}_{\mathrm{N}} and the spin susceptibility χN\chi_{\mathrm{N}} in the normal state [24].

The Green’s function for a superconducting state can be obtained by solving the Gor’kov equation

[iωnHBdG(𝒌)][G(𝒌,iωn)F(𝒌,iωn)F~(𝒌,iωn)G~(𝒌,iωn)]\displaystyle\left[\begin{array}[]{cc}i\omega_{n}-H_{\mathrm{BdG}}(\bm{k})\end{array}\right]\left[\begin{array}[]{cc}G(\bm{k},i\omega_{n})&F(\bm{k},i\omega_{n})\\ -\undertilde{F}(\bm{k},i\omega_{n})&-\undertilde{G}(\bm{k},i\omega_{n})\end{array}\right] (44)
=18×8.\displaystyle=1_{8\times 8}. (45)

The anomalous Green’s function results in

F1\displaystyle F^{-1} (𝒌,iωn)=UTη𝒌2[(ωn2+ξ𝒌2+η𝒌2)η𝒌γ\displaystyle(\bm{k},i\omega_{n})=\frac{U_{T}}{{\vec{\eta}_{\bm{k}}}^{2}}\left[(\omega_{n}^{2}+{\xi_{\bm{k}}}^{2}+{\vec{\eta}_{\bm{k}}}^{2})\vec{\eta}_{\bm{k}}\cdot\vec{\gamma}\right.
+iωn[η𝒌γ,ϵ𝒌γ]\displaystyle+i\omega_{n}[\vec{\eta}_{\bm{k}}\cdot\vec{\gamma},\vec{\epsilon}_{\bm{k}}\cdot\vec{\gamma}]_{-}
+2ξ𝒌η𝒌ϵ𝒌+(ϵ𝒌γ)(η𝒌γ)(ϵ𝒌γ)].\displaystyle\left.+2\,\xi_{\bm{k}}\,\vec{\eta}_{\bm{k}}\cdot\vec{\epsilon}_{\bm{k}}+(\vec{\epsilon}_{\bm{k}}\cdot\vec{\gamma})\,(\vec{\eta}_{\bm{k}}\cdot\vec{\gamma})\,(\vec{\epsilon}_{\bm{k}}\cdot\vec{\gamma})\right]. (46)

Generally speaking, it is not easy to calculate analytically the inversion of 4×44\times 4 matrices.

To proceed analytic calculation, we consider a cubic symmetric superconductor [14], in which even-parity pair potentials are classified into A1gA_{1g}, EgE_{g}, and T2gT_{2g} states according to irreducible representations (irreps) of cubic symmetry. With focusing on pseudospin-quintet Cooper pairs, their pairing states are explicitly represented as

A1g:\displaystyle A_{1g}:\quad η𝒌γ=Δj=15hjcj(𝒌^)γj,\displaystyle\vec{\eta}_{\bm{k}}\cdot\vec{\gamma}=\Delta\,\sum_{j=1}^{5}h_{j}\,c_{j}(\hat{\bm{k}})\,\gamma^{j}, (47)
Eg:\displaystyle E_{g}:\quad η𝒌γ=Δ(l4γ4+l5γ5),\displaystyle\vec{\eta}_{\bm{k}}\cdot\vec{\gamma}=\Delta\,(l_{4}\gamma^{4}+l_{5}\gamma^{5}), (48)
T2g:\displaystyle T_{2g}:\quad η𝒌γ=Δ(l1γ1+l2γ2+l3γ3),\displaystyle\vec{\eta}_{\bm{k}}\cdot\vec{\gamma}=\Delta\,(l_{1}\gamma^{1}+l_{2}\gamma^{2}+l_{3}\gamma^{3}), (49)

where the unit vector h\vec{h} determines the pseudospin structure of A1gA_{1g} state and lil_{i}\in\mathbb{C} (i=15i=1-5). A1gA_{1g} states involve momentum-dependent coefficients, which comes from the fact that γ4\gamma^{4} and γ5\gamma^{5} (γ1,γ2(\gamma^{1},\gamma^{2}, and γ3)\gamma^{3}) themselves belong to EgE_{g} (T2gT_{2g}) irreps of cubic symmetry.

Generally speaking, the coefficients lil_{i} for i=15i=1-5 are determined from the steady states of the free energy [25]. For EgE_{g} state, three distinct steady states exist: (l4,l5)=(1,0)(l_{4},l_{5})=(1,0), (0,1)(0,1), and (1,i)/2(1,i)/\sqrt{2}. The first two states preserve time-reversal symmetry, while the last state breaks time-reversal symmetry. For T2gT_{2g} state, there are four distinct steady states: (l1,l2,l3)=(1,1,1)/3(l_{1},l_{2},l_{3})=(1,1,1)/\sqrt{3}, (1,0,0)(1,0,0), (1,ei2π/3,ei4π/3)/3(1,e^{i2\pi/3},e^{i4\pi/3})/\sqrt{3}, (1,i,0)/2(1,i,0)/\sqrt{2}. The last two states break time-reversal symmetry. Time-reversal symmetry-breaking superconducting states have a problem specific to them: the formation of Bogoliubov-Fermi surfaces [13, 14, 19]. Since the relations between the pseudospin structures of a Cooper pair and the magnetic response of a superconductor is the main issue in this paper, we focus on time-reversal symmetry respecting superconducting states. In addition to the steady states, we also consider another pseudospin states described by (l4,l5)=(1,1)/2(l_{4},l_{5})=(1,1)/\sqrt{2}, (l1,l2,l3)=(1,1,0)/2(l_{1},l_{2},l_{3})=(1,1,0)/\sqrt{2}, and (l1,l2,l3,l4,l5)=(1,1,1,1,1)/5(l_{1},l_{2},l_{3},l_{4},l_{5})=(1,1,1,1,1)/\sqrt{5}. The last one is the admixture of EgE_{g} and T2gT_{2g} states. These states complement possible combinations of γi\gamma^{i} (i=15i=1-5). The comparison between the calculated results for such states and those for the steady states helps us to understand the relations between the pseudospin structures of a Cooper pair and the spin susceptibility. Note that (l1,l2,l3)=(1,1,0)/2(l_{1},l_{2},l_{3})=(1,1,0)/\sqrt{2} represents a possible order parameter in tetragonal symmetric superconductors with the high symmetry axis being the yy direction [25].

III Spin susceptibility

Refer to caption
Figure 1: The spin susceptibility for A1gA_{1g} states in the pseudospin-quintet superconductors is plotted as a function of temperature in (a), where we consider g3=0g_{3}=0 and ϵ=h\vec{\epsilon}=\vec{h}. The diagonal elements in the spin susceptibility tensor are isotropic in real space and the off-diagonal elements are zero. In (b), the susceptibility of spin-triplet superconductors are shown for a helical spin-triplet superconductor with two solid lines and for a 3He B-phase with a broken line. In (c), we consider the effects of Jj3J_{j}^{3} terms in the Zeeman Hamiltonian by choosing g3=g1g_{3}=g_{1}. Although the amplitudes of susceptibility deviate slightly from those in (a), the characteristic features of the susceptibility retain.

As shown in the second term in Eq. (46), the anomalous Green’s function contains the pairing correlation belonging to odd-frequency symmetry class. The stable superconducting states can be described by

[η𝒌γ,ϵ𝒌γ]=0,\displaystyle[\vec{\eta}_{\bm{k}}\cdot\vec{\gamma},\vec{\epsilon}_{\bm{k}}\cdot\vec{\gamma}]_{-}=0, (50)

which means the absence of odd-frequency pairs. Odd-frequency pairs increase the free-energy of a uniform superconducting state [26] because they indicate the paramagnetic response to a magnetic field [27, 28, 29]. A phenomenological argument on the paramagnetic response of odd-frequency Cooper pairs is given in Appendix A of Ref. [30]. Eq. (50) gives a guide that relates the stable pair potential η\vec{\eta} to the electronic structures ϵ\vec{\epsilon}. To understand this situation, we briefly summarize the case of spin-triplet superconductors in the presence of a strong Rashba spin-orbit interaction 𝝀𝝈\bm{\lambda}\cdot\bm{\sigma} with

𝝀=λso(k^y𝒆xk^x𝒆y),\displaystyle\bm{\lambda}=\lambda_{\mathrm{so}}(\hat{k}_{y}\bm{e}_{x}-\hat{k}_{x}\bm{e}_{y}), (51)

where λso\lambda_{\mathrm{so}} represents the amplitude of spin-orbit interaction, σj\sigma_{j} and 𝒆j\bm{e}_{j} for j=x,yj=x,y and zz are the Pauli matrix and the unit vector in spin space, respectively. The stable order parameter i𝒅𝝈σyi\bm{d}\cdot\bm{\sigma}\,\sigma_{y} is determined as

[𝒅𝝈,𝝀𝝈]=0or𝒅𝝀,\displaystyle\left[\bm{d}\cdot\bm{\sigma},\bm{\lambda}\cdot\bm{\sigma}\right]_{-}=0\quad\text{or}\quad\bm{d}\parallel\bm{\lambda}, (52)

so that odd-frequency pairs are absent and the transition temperature is optimal [26, 31]. Namely, the order parameter of a spin helical state is stable in this case. The pair potentials other than the helical state would be realized when the Rashba spin-orbit interaction is sufficiently weak. The choice in Eq. (52) and that in Eq. (50) are equivalent to each other. We choose the normal state dispersion ϵ𝒌\vec{\epsilon}_{\bm{k}} so that Eq. (50) is satisfied for the pair potentials in Eqs. (47)-(49). The Green’s function in the superconducting state can be expressed simply and analytically under Eq. (50). The results of the Green’s function in such a case are shown in Appedix B.

The spin susceptibility results in

χμνχN\displaystyle\frac{\chi_{\mu\nu}}{\chi_{\mathrm{N}}} =δμ,ν\displaystyle=\delta_{\mu,\nu}
\displaystyle- πTωn12Ω3(η2δμ,ν+Lμ,ν(η)P+)𝒌^,\displaystyle\pi T\sum_{\omega_{n}}\left\langle\frac{1}{2\Omega^{3}}\left({\vec{\eta}}^{2}\,\delta_{\mu,\nu}+\frac{{L}_{\mu,\nu}(\vec{\eta})}{P_{+}}\right)\right\rangle_{\hat{\bm{k}}}, (53)

with Ω=ωn2+η2\Omega=\sqrt{\omega_{n}^{2}+{\vec{\eta}}^{2}}. The tensor is defined by

Lμ,ν(η)\displaystyle L_{\mu,\nu}(\vec{\eta})\equiv Tr[ηγJ~μηγJ~ν],\displaystyle\mathrm{Tr}\left[\vec{\eta}\cdot\vec{\gamma}\,\tilde{J}_{\mu}\,\vec{\eta}\cdot\vec{\gamma}\tilde{J}_{\nu}\right], (54)

and its elements are calculated to be

Lxx=\displaystyle{L}_{xx}= P(η12η22+η32)+R+η42Rη52\displaystyle P_{-}(\eta_{1}^{2}-\eta_{2}^{2}+\eta_{3}^{2})+R_{+}\eta_{4}^{2}-R_{-}\eta_{5}^{2}
43p12η4η5,\displaystyle-4\sqrt{3}p_{1}^{2}\,\eta_{4}\,\eta_{5}, (55)
Lyy=\displaystyle{L}_{yy}= P(η12+η22η32)+R+η42Rη52\displaystyle P_{-}(\eta_{1}^{2}+\eta_{2}^{2}-\eta_{3}^{2})+R_{+}\eta_{4}^{2}-R_{-}\eta_{5}^{2}
+43p12η4η5,\displaystyle+4\sqrt{3}p_{1}^{2}\,\eta_{4}\,\eta_{5}, (56)
Lzz=\displaystyle{L}_{zz}= P(η12+η22+η32η42)+P+η52,\displaystyle P_{-}(-\eta_{1}^{2}+\eta_{2}^{2}+\eta_{3}^{2}-\eta_{4}^{2})+P_{+}\eta_{5}^{2}, (57)
Lxy=\displaystyle{L}_{xy}= 2[P+η2η323p1p2η1η5],\displaystyle 2\left[P_{+}\,\eta_{2}\,\eta_{3}-2\sqrt{3}\,p_{1}p_{2}\eta_{1}\,\eta_{5}\right], (58)
Lyz=\displaystyle{L}_{yz}= 2[P+η1η3+3p1p2(η2η53η2η4)],\displaystyle 2\left[P_{+}\,\eta_{1}\,\eta_{3}+\sqrt{3}\,p_{1}p_{2}(\eta_{2}\,\eta_{5}-\sqrt{3}\,\eta_{2}\eta_{4})\right], (59)
Lzx=\displaystyle{L}_{zx}= 2[P+η1η2+3p1p2(η3η5+3η3η4)],\displaystyle 2\left[P_{+}\,\eta_{1}\,\eta_{2}+\sqrt{3}\,p_{1}\,p_{2}(\eta_{3}\,\eta_{5}+\sqrt{3}\,\eta_{3}\eta_{4})\right], (60)
P±\displaystyle P_{\pm}\equiv 4p12±p22,R±=2p12±p22.\displaystyle 4p_{1}^{2}\pm p_{2}^{2},\quad R_{\pm}=2p_{1}^{2}\pm p_{2}^{2}. (61)

The results in Eq. (55)-(60) describe the characteristic features of the susceptibility of J=3/2J=3/2 superconductors.

III.1 A1gA_{1g} state

In the presence of attractive interactions in an A1gA_{1g} channel, the pair potential is represented by Eqs. (30) and47). We always set e=h\vec{e}=\vec{h} so that Eq. (50) is satisfied. Different from the ss-wave spin-singlet pairing that also belongs to the A1gA_{1g} irrep, an A1gA_{1g} state in the pseudo-spin quintet pairing has the pseudospin degrees of freedom, which allows us to choose a variety of pseudospin structures. We first choose a pseudospin structure of a T2gT_{2g} irrep characterized by h=hT2g=(1,1,1,0,0)/3\vec{h}=\vec{h}_{T_{2g}}=(1,1,1,0,0)/\sqrt{3}. The spin susceptibility is calculated as

χμν=\displaystyle\chi_{\mu\nu}= χNδμ,ν[112(1+P3P+)𝒬],\displaystyle\chi_{\mathrm{N}}\,\delta_{\mu,\nu}\left[1-\frac{1}{2}\left(1+\frac{P_{-}}{3P_{+}}\right)\mathcal{Q}\right], (62)
𝒬(T)=\displaystyle\mathcal{Q}(T)= πTωnΔ2Ω3.\displaystyle\pi T\sum_{\omega_{n}}\frac{{\Delta}^{2}}{\Omega^{3}}. (63)

In BCS theory, 𝒬\mathcal{Q} represents the fraction of Cooper pairs to quasiparticles on the Fermi surface. Indeed 𝒬\mathcal{Q} is zero at T=TcT=T_{c}, increases monotonically with the decrease of TT, and becomes unity at T=0T=0. The susceptibility tensor is diagonal and isotropic in real space. The susceptibility for a pseudospin structure of an EgE_{g} irrep characterized by h=hEg=(0,0,0,1,1)/2\vec{h}=\vec{h}_{E_{g}}=(0,0,0,1,1)/\sqrt{2} results in,

χμν=\displaystyle\chi_{\mu\nu}= χNδμ,ν[112(1+p22P+)𝒬].\displaystyle\chi_{\mathrm{N}}\,\delta_{\mu,\nu}\left[1-\frac{1}{2}\left(1+\frac{p_{2}^{2}}{P_{+}}\right)\mathcal{Q}\right]. (64)

The susceptibility for an admixture of T2gT_{2g} and EgE_{g} irreps ( hT2g+Eg=(1,1,1,1,1)/5\vec{h}_{T_{2g}+E_{g}}=(1,1,1,1,1)/\sqrt{5}) is also calculated as

χμν=\displaystyle\chi_{\mu\nu}= χNδμ,ν[112(1+15)𝒬].\displaystyle\chi_{\mathrm{N}}\,\delta_{\mu,\nu}\left[1-\frac{1}{2}\left(1+\frac{1}{5}\right)\mathcal{Q}\right]. (65)

The off-diagonal elements in the susceptibility tensor are always zero in these cases, (χμν=0\chi_{\mu\nu}=0 for μν\mu\neq\nu).

Refer to caption
Figure 2: The spin susceptibility is plotted as a function of temperature for a T2gT_{2g} state with (l1,l2,l3)=(1,1,1)/3(l_{1},l_{2},l_{3})=(1,1,1)/\sqrt{3} in (a). The susceptibility tensor in a T2gT_{2g} state has off-diagonal elements. In (b), the results for the state with (1,1,0)/2(1,1,0)/\sqrt{2} are displayed, where we delete γ3\gamma^{3} component from the T2gT_{2g} pair potential. The results for a single component pair potential with (1,0,0)/2(1,0,0)/\sqrt{2} are shown in (c). Although we put g3=0g_{3}=0 in these figures, Jj3J_{j}^{3} terms in the Zeeman potential do not change the characteristic features.

In Fig. 1(a), we plot the susceptibility at g3=0g_{3}=0 as a function of temperature. The dependence of the pair potential on temperature is calculated by solving the gap equation in the weak coupling limit. As shown in Appedix C, the gap equation is common for all the superconducting states considered in the present paper and is identical to that in BCS theory. The results show that the susceptibility decreases with the decrease of temperature below TcT_{c}. The results are independent of such choices of the pseudospin structures as T2gT_{2g}, EgE_{g}, and T2g+EgT_{2g}+E_{g} at g3=0g_{3}=0. Although a Cooper pair has the angular momentum of J=2J=2 in the quintet states, the susceptibility is isotropic for all the pseudospin structures. The characteristic behaviors are independent of the strength of spin-orbit interaction β/α\beta/\alpha. The isotropic feature of the spin susceptibility is considered as a result of high symmetry of the pair potential. The pseudospin structures in the pair potential are characterized by γj\gamma^{j} which can be described by the linear combination of JμJνJ_{\mu}J_{\nu} as shown in Eqs. (24)-(26). For a T2gT_{2g} irrep, γj\gamma^{j} for j=13j=1-3 characterize both the normal state dispersion and the pair potential. The pair potential with hT2g\vec{h}_{T_{2g}} is symmetric under the cyclic permutation among Jx,JyJ_{x},J_{y} and JzJ_{z}. For an EgE_{g} irrep, the pair potential is symmetric under interchanging JxJyJ_{x}\leftrightarrow J_{y}. The structure of hEg\vec{h}_{E_{g}} results in the isotropic susceptibility including the zz direction. Thus, the direction of a Zeeman field 𝑯\bm{H} in real space does not point a specific direction in pseudospin space. For comparison, we briefly mention the spin susceptibility in superfluid 3He B-phase described by

𝒅=Δ(k^x𝒆x+k^y𝒆y+k^z𝒆z).\displaystyle\bm{d}=\Delta(\hat{k}_{x}\bm{e}_{x}+\hat{k}_{y}\bm{e}_{y}+\hat{k}_{z}\bm{e}_{z}). (66)

The pair potential is symmetric under the cyclic permutation among x,yx,y and zz. As a result, the susceptibility plotted with a broken line in Fig. 1(b) is isotropic in real space.

At the end of the subsection, we briefly discuss the effects of Jμ3J_{\mu}^{3} term on the spin susceptibility by choosing g3=g1g_{3}=g_{1}. The results are shown in Fig. 1(c). The isotropic nature of the susceptibility remains unchanged even for g3=g1g_{3}=g_{1}. The amplitude of the susceptibility depends on the pseudospin structure of the pair potential; the amplitude for a T2gT_{2g} irrep becomes slightly larger than that for an EgE_{g} irrep.

III.2 T2gT_{2g} state

When attractive interactions between two electrons work in a T2gT_{2g} or an EgE_{g} channel, the order parameters in Eqs. (48) and (49) are isotropic in momentum space. To satisfy Eq. (50), we switch off ϵ=0\vec{\epsilon}=0 and consider a simple pseudospin quintet superconductor in the following subsections. In other words, the pair potentials independent of momenta are stable when |ϵ||\vec{\epsilon}| is sufficiently smaller than the Fermi energy μ\mu. Even if we choose ϵ=0\vec{\epsilon}=0, superconductors show the rich magnetic response depending on the pseudospin structures of the pair potential. The effects of the pseudospin-dependent dispersions ϵ0\vec{\epsilon}\neq 0 on the magnetic response will be discussed later.

We first discuss a T2gT_{2g} state with (l1,l2,l3)=(1,1,1)/3(l_{1},l_{2},l_{3})=(1,1,1)/\sqrt{3}. In addition to the diagonal element given by Eq.(62), the susceptibility has the off-diagonal elements as

χxy=\displaystyle\chi_{xy}= χyz=χzx=χN13𝒬.\displaystyle\chi_{yz}=\chi_{zx}=-\chi_{\mathrm{N}}\frac{1}{3}\mathcal{Q}. (67)

The results for g3=0g_{3}=0 are displayed in Fig. 2(a). The pair potential includes off-diagonal terms JμJνJ_{\mu}\,J_{\nu} with μν\mu\neq\nu through γj\gamma^{j} for j=13j=1-3 as shown in Eqs. (24)-(26). The first term of Lμν(η)L_{\mu\nu}(\vec{\eta}) in Eqs. (58)-(60) is the direct results of such pseudospin structure. Since the pair potential is independent of wavenumber, these off-diagonal terms remain nonzero values even after averaging over directions in momentum space on the Fermi surface. Thus, the appearance of the off-diagonal elements in the susceptibility tensor is a characteristic feature of a T2gT_{2g} state.

Secondly, we display the susceptibility for a T2gT_{2g} state with (l1,l2,l3)=(1,1,0)/2(l_{1},l_{2},l_{3})=(1,1,0)/\sqrt{2} in Fig. 2(b), where we delete γ3\gamma^{3} component from Eq. (49). As a result, JyJ_{y} is no longer equivalent to JxJ_{x} and JzJ_{z}, which explains the anisotropy of the diagonal elements in Fig. 2(b). As shown in Eqs. (58)-(60), the off-diagonal elements are finite only for the multi-component pair potentials. At the present case, only the χzx\chi_{zx} element remains finite because of l3=0l_{3}=0.

Finally, we display the susceptibility for a T2gT_{2g} state with (l1,l2,l3)=(1,0,0)/2(l_{1},l_{2},l_{3})=(1,0,0)/\sqrt{2} in Fig. 2(c), which has only γ1\gamma^{1} component. The diagonal elements are anisotropic as they are in Fig. 2(b). All the off-diagonal elements vanish because the pair potential has only one pseudospin component. Thus, the off-diagonal elements emerge if γi\gamma^{i} and γj\gamma^{j} (i,j=1,2,3;iji,j=1,2,3;\,i\neq j) coexist in the pair potential. In particular, the nonzero off-diagonal elements are determined from the direction of high symmetry axis in tetragonal symmetric superconductors, e.g., χxy=χyz=0\chi_{xy}=\chi_{yz}=0 and χzx0\chi_{zx}\neq 0 for the pair potential with (l1,l2,l3)=(1,1,0)/2(l_{1},l_{2},l_{3})=(1,1,0)/\sqrt{2} since the high symmetry axis is the yy direction.

III.3 EgE_{g} state

The susceptibility for an EgE_{g} state with (l4,l5)=(1,1)/2(l_{4},l_{5})=(1,1)/\sqrt{2} is calculated as

χxx=\displaystyle\chi_{xx}= χN[112(1+p223p12P+)𝒬],\displaystyle\chi_{\mathrm{N}}\left[1-\frac{1}{2}\left(1+\frac{p_{2}^{2}-\sqrt{3}p_{1}^{2}}{P_{+}}\right)\mathcal{Q}\right], (68)
χyy=\displaystyle\chi_{yy}= χN[112(1+p22+3p12P+)𝒬],\displaystyle\chi_{\mathrm{N}}\left[1-\frac{1}{2}\left(1+\frac{p_{2}^{2}+\sqrt{3}p_{1}^{2}}{P_{+}}\right)\mathcal{Q}\right], (69)
χzz=\displaystyle\chi_{zz}= χN[112(1+p22P+)𝒬],\displaystyle\chi_{\mathrm{N}}\left[1-\frac{1}{2}\left(1+\frac{p_{2}^{2}}{P_{+}}\right)\mathcal{Q}\right], (70)
χxy=\displaystyle\chi_{xy}= χyz=χzx=0.\displaystyle\chi_{yz}=\chi_{zx}=0. (71)

The results are shown in Fig. 3(a). All the off-diagonal elements vanish as shown in Eqs. (58)-(60). The product of γ4γ5\gamma^{4}\,\gamma^{5} does not include such off-diagonal terms as JμJνJ_{\mu}J_{\nu} with μν\mu\neq\nu, which explains the absence of the off-diagonal elements in the susceptibility tensor. The diagonal elements of the susceptibility becomes anisotropic due to the last term in Eqs. (55) and (56). The product of γ4γ5\gamma^{4}\,\gamma^{5} includes Jx4+Jy4-J_{x}^{4}+J_{y}^{4}, which breaks symmetry between JxJ_{x} and JyJ_{y}. As a result, the last term in Eq. (55) and that in Eq. (56) have the opposite signs to each other. Thus the anisotropy in the susceptibility is a characteristic feature of an EgE_{g} state. The degree of the anisotropy in an EgE_{g} state is rather weaker than that in a spin-triplet superconductor. For comparison, in Fig. 1(b), we plot the susceptibility of a spin-triplet helical state characterized by 𝒅𝝀\bm{d}\parallel\bm{\lambda} in Eq. (51) with two solid lines. When a Zeeman field is perpendicular to 𝒅\bm{d}, the susceptibility is a constant across TcT_{c}. The results for 𝑯𝒅\bm{H}\parallel\bm{d}, on the other hand, the susceptibility decreases down to (1/2)χN(1/2)\chi_{\mathrm{N}}. These behaviors are independent of the amplitudes of λso>0\lambda_{\mathrm{so}}>0. In spin-triplet superconductors, both the dimension in spin space and that in real space are three. Therefore, it is possible to define two different directions relatively to the direction of a Zeeman field: (𝒅𝑯\bm{d}\parallel\bm{H} and 𝒅𝑯\bm{d}\perp\bm{H}). The clear anisotropy of the susceptibility in Fig. 1(b) is a result of the dimensional consistency between in spin space and in real space.

In Fig. 3(b) and (c), we display the results for the single component states with (l4,l5)=(1,0)(l_{4},l_{5})=(1,0) and (0,1)(0,1), respectively. They are possible order parameters of EgE_{g} states in the presence of time-reversal symmetry [25]. We find the relation

χxx=χyyχzz,\displaystyle\chi_{xx}=\chi_{yy}\neq\chi_{zz}, (72)

because the last term in Eqs. (55) and (56) is absent. Including the results in Fig. 2(c), Eqs. (55)-(60) suggest that the anisotropic response like Eq. (72) and the absense of off-diagonal elements are the common feature of the single component order parameter.

Refer to caption
Figure 3: The spin susceptibility is plotted as a function of temperature for a EgE_{g} state in (a).

III.4 Admixture of T2gT_{2g} and EgE_{g} states

The results for a the admixture of T2gT_{2g} and EgE_{g} states, (i. e., (l1,l2,l3,l4,l5)=(1,1,1,1,1)/5(l_{1},l_{2},l_{3},l_{4},l_{5})=(1,1,1,1,1)/\sqrt{5}), are calculated as

χxx=\displaystyle\chi_{xx}= χN[112(1+P+43p125P+)𝒬],\displaystyle\chi_{\mathrm{N}}\left[1-\frac{1}{2}\left(1+\frac{P_{+}-4\sqrt{3}p_{1}^{2}}{5P_{+}}\right)\mathcal{Q}\right], (73)
χyy=\displaystyle\chi_{yy}= χN[112(1+P++43p125P+)𝒬],\displaystyle\chi_{\mathrm{N}}\left[1-\frac{1}{2}\left(1+\frac{P_{+}+4\sqrt{3}p_{1}^{2}}{5P_{+}}\right)\mathcal{Q}\right], (74)
χzz=\displaystyle\chi_{zz}= χN[112(1+15)𝒬].\displaystyle\chi_{\mathrm{N}}\left[1-\frac{1}{2}\left(1+\frac{1}{5}\right)\mathcal{Q}\right]. (75)

The off-diagonal elements are calculated in the similar way,

χxy=\displaystyle\chi_{xy}= χN𝒬P+23p1p25P+,\displaystyle-\chi_{\mathrm{N}}\,\mathcal{Q}\,\frac{P_{+}-2\sqrt{3}\,p_{1}\,p_{2}}{5P_{+}}, (76)
χyz=\displaystyle\chi_{yz}= χN𝒬P++3p1p2(13)5P+,\displaystyle-\chi_{\mathrm{N}}\mathcal{Q}\,\frac{P_{+}+\sqrt{3}\,p_{1}\,p_{2}(1-\sqrt{3})}{5P_{+}}, (77)
χzx=\displaystyle\chi_{zx}= χN𝒬P++3p1p2(1+3)5P+.\displaystyle-\chi_{\mathrm{N}}\mathcal{Q}\,\frac{P_{+}+\sqrt{3}\,p_{1}\,p_{2}(1+\sqrt{3})}{5P_{+}}. (78)

The calculated results for g3=0g_{3}=0 are plotted in Fig. 3(d). Not only the diagonal elements but also the off-diagonal elements are anisotropic. All the elements in Eqs. (55)-(60) are finite and different from one another. The degree of the anisotropy in the diagonal elements are weaker than that in EgE_{g} state and stronger than that in T2gT_{2g} state. The characteristic features of the susceptibility displayed in Figs. 2 and 3 retain even if we consider Jμ3J_{\mu}^{3} term in the Zeeman Hamiltonian.

Finally, we briefly discuss the effects of the pseudospin-dependent dispersion ϵ\vec{\epsilon} on the characteristic behaviors of the susceptibility. When we switch on ϵ\vec{\epsilon} in T2gT_{2g} and EgE_{g} states, Eq. (50) is no longer holds. As a result, the additional terms such as

Cf2iωn[η𝒌γ,ϵ𝒌γ]=Cfiωnijϵiηjγiγj,\displaystyle\frac{C_{f}}{2}\,i\omega_{n}\,[\vec{\eta}_{\bm{k}}\cdot\vec{\gamma},\vec{\epsilon}_{\bm{k}}\cdot\vec{\gamma}]_{-}=C_{f}\,i\omega_{n}\,\sum_{i\neq j}\epsilon_{i}\,\eta_{j}\,\gamma^{i}\,\gamma^{j}, (79)

appear at the numerator of the anomalous Green’s function in Eq. (108), where CfC_{f} is a constant. Such components represent the admixture of pseudospin-triplet and the pseudospin-septet pairing correlations [19]. Their contribution to the susceptibility tensor is proportional to

Cf2ωn2Tr[ijϵiηjγiγjJ~μklϵkηlγkγlJ~ν],\displaystyle C_{f}^{2}\,\omega_{n}^{2}\,\mathrm{Tr}\left[\sum_{i\neq j}\epsilon_{i}\,\eta_{j}\gamma^{i}\gamma^{j}\;\tilde{J}_{\mu}\;\sum_{k\neq l}\epsilon_{k}\,\eta_{l}\gamma^{k}\gamma^{l}\;\tilde{J}_{\nu}\right], (80)

which modify the susceptibility tensor, However, they do not always cancel the off-diagonal elements in Fig. 2(a) in a T2gT_{2g} state. They do not wash out the anisotropy of the diagonal elements in Fig. 3 in EgE_{g} states.

The results in Figs. 1-3 suggest that the behaviors of the susceptibility depends sensitively on the orbital symmetry and the pseudospin structures of the pair potential. In particular, the appearance of the off-diagonal elements in the susceptibility tensor is a unique feature to J=3/2J=3/2 superconductors.

IV Conclusion

We have studied theoretically the spin susceptibility of pseudospin-quintet pairing states in a J=3/2J=3/2 superconductor that preserves cubic lattice symmetry and time-reversal symmetry. Within the linear response to a Zeeman field, we calculate the spin susceptibility by using the Green’s function that is obtained by solving the Gor’kov equation analytically. The pair potentials are chosen so that a superconducting state is stable under the pseudospin structures in the normal state. The calculated results indicat that the magnetic response of pseudospin-quintet states depends sensitively on the pseudospin structures of the pair potential. The susceptibility tensor in A1gA_{1g} states is isotropic in real space as that in the B-phase of superfluid 3He. For EgE_{g} states, the susceptibility tensor becomes anisotropic in real space. We found in a T2gT_{2g} state that the susceptibility tensor has the off-diagonal elements.

Acknowledgements.
The authors are grateful to R. Nomura for useful discussion. This work was supported by JSPS KAKENHI (Nos. JP19K14612, JP20H01857, JP22K03478) and JSPS Core-to-Core Program (No. JPJSCCA20170002). T. S. is supported in part by the establishment of university fellowships towards the creation of science technology innovation from the Ministry of Education, Culture, Sports, Science, and Technology (MEXT) of Japan. S.K. was supported by the CREST project (Grants No. JPMJCR19T2) from Japan Science and Technology Agency (JST).

Appendix A Normal state

The spin susceptibility in the normal state is give by

χN=\displaystyle\chi_{\mathrm{N}}= (μB2)2Tωnd𝒌(2π)dTr[GN(𝒌,iωn)JμGN(𝒌,iωn)Jν],\displaystyle-\left(\frac{\mu_{B}}{2}\right)^{2}T\sum_{\omega_{n}}\int\frac{d\bm{k}}{(2\pi)^{d}}\mathrm{Tr}\left[{G}_{\mathrm{N}}(\bm{k},i\omega_{n})\,J_{\mu}\,G_{\mathrm{N}}(\bm{k},i\omega_{n})\,J_{\nu}\right], (81)
GN=\displaystyle G_{\mathrm{N}}= iωnξ+ϵ𝒌γ(iωnξ)2ϵ𝒌2=αN+βNϵ𝒌γ,αN=12[1zN++1zN],βN=12|ϵ𝒌|[1zN+1zN],\displaystyle\frac{i\omega_{n}-\xi+\vec{\epsilon}_{\bm{k}}\cdot\vec{\gamma}}{(i\omega_{n}-\xi)^{2}-{\vec{\epsilon}_{\bm{k}}}^{2}}=\alpha_{N}+\beta_{N}\vec{\epsilon}_{\bm{k}}\cdot\vec{\gamma},\quad\alpha_{N}=\frac{1}{2}\left[\frac{1}{z_{N+}}+\frac{1}{z_{N-}}\right],\quad\beta_{N}=\frac{1}{2|\vec{\epsilon_{\bm{k}}}|}\left[\frac{1}{z_{N+}}-\frac{1}{z_{N-}}\right], (82)
zN±=\displaystyle z_{N\pm}= iωnξ±,ξ±=ξ±|ϵ𝒌|.\displaystyle i\omega_{n}-\xi_{\pm},\quad\xi_{\pm}=\xi\pm|\vec{\epsilon_{\bm{k}}}|. (83)

The the Green’s function is the solution of

[iωnHN]GN(𝒌,ωn)=1,HN(𝒌)=ξ𝒌+ϵ𝒌γ.\displaystyle\left[i\omega_{n}-H_{\mathrm{N}}\right]G_{\mathrm{N}}(\bm{k},\omega_{n})=1,\quad H_{\mathrm{N}}(\bm{k})=\xi_{\bm{k}}+\vec{\epsilon}_{\bm{k}}\cdot\vec{\gamma}. (84)

The trace of the Green’s function is calculated as

Tr[GN(𝒌,iωn)J~μGN(𝒌,iωn)J~ν]=δμ,νP+αN2+αNβNMμ,ν(ϵ𝒌)+βN2Lμ,ν(ϵ𝒌),\displaystyle\mathrm{Tr}\left[{G}_{\mathrm{N}}(\bm{k},i\omega_{n})\,\tilde{J}_{\mu}\,G_{\mathrm{N}}(\bm{k},i\omega_{n})\,\tilde{J}_{\nu}\right]=\delta_{\mu,\nu}\,P_{+}\,\alpha_{N}^{2}+\alpha_{N}\,\beta_{N}M_{\mu,\nu}(\vec{\epsilon}_{\bm{k}})+\beta_{N}^{2}L_{\mu,\nu}(\vec{\epsilon}_{\bm{k}}), (85)

where we use Tr(J~μJ~ν)=P+δμ,ν\mathrm{Tr}(\tilde{J}_{\mu}\tilde{J}_{\nu})=P_{+}\,\delta_{\mu,\nu} and define the tensor

Mμ,ν(ϵ)Tr[ϵγ(J~μJ~ν+J~νJ~μ)].\displaystyle M_{\mu,\nu}(\vec{\epsilon})\equiv\mathrm{Tr}\left[\vec{\epsilon}\cdot\vec{\gamma}(\tilde{J}_{\mu}\,\tilde{J}_{\nu}+\tilde{J}_{\nu}\tilde{J}_{\mu})\right]. (86)

The angular momenta JνJ_{\nu} are expressed in terms of γν\gamma^{\nu}

Jx=\displaystyle J_{x}= i2[3γ2γ5+γ1γ3+γ2γ4],Jy=i2[3γ3γ5+γ1γ2γ3γ4],Jz=i2[γ2γ3+2γ1γ4].\displaystyle\frac{-i}{2}\left[\sqrt{3}\gamma^{2}\,\gamma^{5}+\gamma^{1}\,\gamma^{3}+\gamma^{2}\,\gamma^{4}\right],\quad J_{y}=\frac{i}{2}\left[\sqrt{3}\gamma^{3}\,\gamma^{5}+\gamma^{1}\,\gamma^{2}-\gamma^{3}\,\gamma^{4}\right],\quad J_{z}=\frac{i}{2}\left[\gamma^{2}\,\gamma^{3}+2\,\gamma^{1}\,\gamma^{4}\right]. (87)

They obey the relation UT(Jν)UT1=JνU_{T}\,(J^{\nu})^{\ast}U_{T}^{-1}=-J^{\nu}, which simply means that the angular momenta are antisymmetric under the time-reversal operation. The expression of Jν3J_{\nu}^{3}

Jx3=\displaystyle J_{x}^{3}= i8(73γ2γ5+13γ1γ3+7γ2γ4),Jy3=i8(73γ3γ5+13γ1γ27γ3γ4),\displaystyle-\frac{i}{8}\left(7\sqrt{3}\gamma^{2}\,\gamma^{5}+13\gamma^{1}\,\gamma^{3}+7\gamma^{2}\,\gamma^{4}\right),\;J_{y}^{3}=\frac{i}{8}\left(7\sqrt{3}\gamma^{3}\,\gamma^{5}+13\gamma^{1}\,\gamma^{2}-7\gamma^{3}\,\gamma^{4}\right), (88)
Jz3=\displaystyle J_{z}^{3}= i8[13γ2γ3+14γ1γ4],\displaystyle\frac{i}{8}\left[13\gamma^{2}\,\gamma^{3}+14\gamma^{1}\,\gamma^{4}\right], (89)

suggests that JνJ_{\nu} and Jν3J_{\nu}^{3} share the common matrix structures. The elements of the tensor are calculated as

Mxx(ϵ)=\displaystyle M_{xx}(\vec{\epsilon})= 4p1p2,(3ϵ4ϵ5),Myy(ϵ)=4p1p2,(3ϵ4+ϵ5),Mzz(ϵ)=8p1p2,ϵ5,\displaystyle 4\,p_{1}\,p_{2},(\sqrt{3}\epsilon_{4}-\epsilon_{5}),\quad M_{yy}(\vec{\epsilon})=-4\,p_{1}\,p_{2},(\sqrt{3}\epsilon_{4}+\epsilon_{5}),\quad M_{zz}(\vec{\epsilon})=8\,p_{1}\,p_{2},\epsilon_{5}, (90)
Mxy(ϵ)=Myx(ϵ)=\displaystyle M_{xy}(\vec{\epsilon})=M_{yx}(\vec{\epsilon})= 43p12ϵ1,Mxz(ϵ)=Mzx(ϵ)=43p12ϵ3,Myz(ϵ)=Mzy(ϵ)=43p12ϵ2,\displaystyle 4\sqrt{3}p_{1}^{2}\epsilon_{1},\quad M_{xz}(\vec{\epsilon})=M_{zx}(\vec{\epsilon})=4\sqrt{3}p_{1}^{2}\epsilon_{3},\quad M_{yz}(\vec{\epsilon})=M_{zy}(\vec{\epsilon})=4\sqrt{3}p_{1}^{2}\epsilon_{2}, (91)

where ϵj=β𝒌2ejcj(𝒌^)\epsilon_{j}=\beta\bm{k}^{2}e_{j}\,c_{j}(\hat{\bm{k}}) as defined in Eq. (4) and we have used the relations

Tr[γj]=\displaystyle\mathrm{Tr}[\gamma^{j}]= 0,Tr[γiγj]=4δi,j,Tr[γiγjγk]=0,Tr[γiγjγkγl]=4[δi,jδk,lδi,kδj,l+δi,lδj,k].\displaystyle 0,\quad\mathrm{Tr}[\gamma^{i}\,\gamma^{j}]=4\delta_{i,j},\quad\mathrm{Tr}[\gamma^{i}\,\gamma^{j}\,\gamma^{k}]=0,\quad\mathrm{Tr}[\gamma^{i}\,\gamma^{j}\,\gamma^{k}\,\gamma^{l}]=4\left[\delta_{i,j}\delta_{k,l}-\delta_{i,k}\delta_{j,l}+\delta_{i,l}\delta_{j,k}\right]. (92)

The another tensor Lμ,νL_{\mu,\nu} is defined by Eq. (54).

The summation over the wavenumber is replaced by the integration as

d𝒌(2π)dF(𝒌)N0𝑑ξF(ξ,𝒌^)𝒌^,F(ξ,𝒌^)𝒌^=d𝒌^4πF(ξ,𝒌^),\displaystyle\int\frac{d\bm{k}}{(2\pi)^{d}}F(\bm{k})\to N_{0}\int_{-\infty}^{\infty}d\xi\,\langle F(\xi,\hat{\bm{k}})\rangle_{\hat{\bm{k}}},\quad\langle F(\xi,\hat{\bm{k}})\rangle_{\hat{\bm{k}}}=\int\frac{d\hat{\bm{k}}}{4\pi}F(\xi,\hat{\bm{k}}), (93)

where 𝒌^\hat{\bm{k}} is the unit vector on the Fermi surface. By using the relations

Mμ,ν(ϵ𝒌)𝒌^=0,Lμ,ν(ϵ𝒌)𝒌^=Lμ,μ𝒌^δμ,ν.\displaystyle\langle M_{\mu,\nu}(\vec{\epsilon}_{\bm{k}})\rangle_{\hat{\bm{k}}}=0,\quad\langle L_{\mu,\nu}(\vec{\epsilon}_{\bm{k}})\rangle_{\hat{\bm{k}}}=\langle L_{\mu,\mu}\rangle_{\hat{\bm{k}}}\,\delta_{\mu,\nu}. (94)

the susceptibility in the normal state becomes

χN=\displaystyle\chi_{\mathrm{N}}= (μB2)2Tωnd𝒌(2π)d[P+αN2+βN2Lμ,μ𝒌^]δμ,ν.\displaystyle-\left(\frac{\mu_{B}}{2}\right)^{2}T\sum_{\omega_{n}}\int\frac{d\bm{k}}{(2\pi)^{d}}\left[P_{+}\alpha_{N}^{2}\,+\beta_{N}^{2}\langle L_{\mu,\mu}\rangle_{\hat{\bm{k}}}\right]\,\delta_{\mu,\nu}. (95)

The summation over the Matsubara frequency is carried out as

TωnαN2=\displaystyle T\sum_{\omega_{n}}\alpha_{N}^{2}= 14Tωn[1zN+2+1|ϵ𝒌|(1zN+1zN)+1zN2],\displaystyle\frac{1}{4}T\sum_{\omega_{n}}\left[\frac{1}{z_{N+}^{2}}+\frac{1}{|\vec{\epsilon_{\bm{k}}}|}\left(\frac{1}{z_{N+}}-\frac{1}{z_{N-}}\right)+\frac{1}{z_{N-}^{2}}\right], (96)
=\displaystyle= 14[14Tcosh2ξ+2T12|ϵ𝒌|(tanhξ+2Ttanhξ2T)14Tcosh2ξ2T].\displaystyle\frac{1}{4}\left[-\frac{1}{4T}\cosh^{-2}\frac{\xi_{+}}{2T}-\frac{1}{2|\vec{\epsilon_{\bm{k}}}|}\left(\tanh\frac{\xi_{+}}{2T}-\tanh\frac{\xi_{-}}{2T}\right)-\frac{1}{4T}\cosh^{-2}\frac{\xi_{-}}{2T}\right]. (97)

The integration over ξ\xi after the summation over the frequency can be calculated exactly as

𝑑ξTωnαN2=1,𝑑ξTωnβN2=0.\displaystyle\int_{-\infty}^{\infty}d\xi\,T\sum_{\omega_{n}}\alpha_{N}^{2}=-1,\quad\int_{-\infty}^{\infty}d\xi\,T\sum_{\omega_{n}}\beta_{N}^{2}=0. (98)

The resulting spin susceptibility

χN=\displaystyle\chi_{\mathrm{N}}= (μB2)2P+N0,\displaystyle\left(\frac{\mu_{B}}{2}\right)^{2}\,P_{+}\,N_{0}, (99)

is diagonal and isotropic independent of the direction of a Zeeman field.

The normal Green’s function can be described alternatively as

GN=\displaystyle G_{\mathrm{N}}= 1ZN(ωn)[AN+BNϵ𝒌γ],ZN=ξ4+2ξ2(ωn2ϵ𝒌2)+(ωn2+ϵ𝒌2)2\displaystyle\frac{-1}{Z_{N}(\omega_{n})}\left[A_{N}+B_{N}\vec{\epsilon}_{\bm{k}}\cdot\vec{\gamma}\right],\quad Z_{N}=\xi^{4}+2\xi^{2}(\omega_{n}^{2}-{\vec{\epsilon}_{\bm{k}}}^{2})+(\omega_{n}^{2}+{\vec{\epsilon}_{\bm{k}}}^{2})^{2} (100)
AN=\displaystyle A_{N}= (ωn2+ξ2+ϵ𝒌2)iωn+(ωn2+ξ2ϵ𝒌2)ξ,BN={(iωnξ)2ϵ𝒌2}.\displaystyle(\omega_{n}^{2}+\xi^{2}+{\vec{\epsilon}_{\bm{k}}}^{2})i\omega_{n}+(\omega_{n}^{2}+\xi^{2}-{\vec{\epsilon}_{\bm{k}}}^{2})\xi,\quad B_{N}=-\left\{(i\omega_{n}-\xi)^{2}-{\vec{\epsilon}_{\bm{k}}}^{2}\right\}. (101)

When we carry out the summation over the wavenumber first as

N0𝑑ξ\displaystyle N_{0}\int_{-\infty}^{\infty}d\xi Tr[GN(𝒌,iωn)J~μGN(𝒌,iωn)J~ν]𝒌^=N0𝑑ξ1ZN2[δμ,νP+AN2+BN2Lμ,ν𝒌^],\displaystyle\langle\mathrm{Tr}\left[{G}_{\mathrm{N}}(\bm{k},i\omega_{n})\,\tilde{J}_{\mu}\,G_{\mathrm{N}}(\bm{k},i\omega_{n})\,\tilde{J}_{\nu}\right]\rangle_{\hat{\bm{k}}}=N_{0}\int_{-\infty}^{\infty}d\xi\frac{1}{Z_{N}^{2}}\left[\delta_{\mu,\nu}\,P_{+}\,A_{N}^{2}+B_{N}^{2}\langle L_{\mu,\nu}\rangle_{\hat{\bm{k}}}\right], (102)

we find χN=0\chi_{\mathrm{N}}=0 because of

𝑑ξAN2ZN2=𝑑ξBN2ZN2=0.\displaystyle\int_{-\infty}^{\infty}d\xi\,\frac{A_{N}^{2}}{Z_{N}^{2}}=\int_{-\infty}^{\infty}d\xi\,\frac{B_{N}^{2}}{Z_{N}^{2}}=0. (103)

The discrepancy is derived from the fact that the integration over the wavenumber and the summation over the frequency do not converge.[24] On the way to Eq. (103), we have used the following relations

I0=\displaystyle I_{0}= 𝑑ξ1ZN=π2|ωn|(ωn2+ε2),Jn=𝑑ξξnZN2,\displaystyle\int_{-\infty}^{\infty}d\xi\,\frac{1}{Z_{N}}=\frac{\pi}{2|\omega_{n}|(\omega_{n}^{2}+\varepsilon^{2})},\quad J_{n}=\int_{-\infty}^{\infty}d\xi\,\frac{\xi^{n}}{Z_{N}^{2}}, (104)
J0=\displaystyle J_{0}= I08ωn25ωn2+ε2(ωn2+ε2)2,J2=I08ωn2,J4=I08ωn2(ωn2+ε2),J6=I08ωn2(ωn2+ε2)(5ωn2+ε2).\displaystyle\frac{I_{0}}{8\omega_{n}^{2}}\frac{5\omega_{n}^{2}+\varepsilon^{2}}{(\omega_{n}^{2}+\varepsilon^{2})^{2}},\quad J_{2}=\frac{I_{0}}{8\omega_{n}^{2}},\quad J_{4}=\frac{I_{0}}{8\omega_{n}^{2}}(\omega_{n}^{2}+\varepsilon^{2}),\quad J_{6}=\frac{I_{0}}{8\omega_{n}^{2}}(\omega_{n}^{2}+\varepsilon^{2})(5\omega_{n}^{2}+\varepsilon^{2}). (105)

We approximately replace ϵ𝒌^2>0{\vec{\epsilon}_{\hat{\bm{k}}}}^{2}>0 by ε2=(α/β+5/4)2μ2\varepsilon^{2}=(\alpha/\beta+5/4)^{-2}\mu^{2}.

Appendix B Superconducting state

The Green’s function under Eq. (50) is calculated to be

G(𝒌,ωn)=\displaystyle G(\bm{k},\omega_{n})= 1ZS[Ag+Bgϵ𝒌γ],\displaystyle\frac{1}{Z_{\mathrm{S}}}\left[A_{g}+{B_{g}}\vec{\epsilon}_{\bm{k}}\cdot{\vec{\gamma}}\right], (106)
Ag=\displaystyle A_{g}= (ωn2+ξ𝒌2+η𝒌2+ϵ𝒌2)iωn(ωn2+ξ𝒌2+η𝒌2ϵ𝒌2)ξ𝒌,Bg=ωn2ξ22iωnξ𝒌+η𝒌2+ϵ𝒌2,\displaystyle-(\omega_{n}^{2}+{\xi_{\bm{k}}}^{2}\ +{\vec{\eta}_{\bm{k}}}^{2}+{\vec{\epsilon}_{\bm{k}}}^{2})i\omega_{n}-(\omega_{n}^{2}+{\xi_{\bm{k}}}^{2}+{\vec{\eta}_{\bm{k}}}^{2}-{\vec{\epsilon}_{\bm{k}}}^{2})\,\xi_{\bm{k}},\quad B_{g}=\omega_{n}^{2}-\xi^{2}-2\,i\,\omega_{n}\,\xi_{\bm{k}}+{\vec{\eta}_{\bm{k}}}^{2}+{\vec{\epsilon}_{\bm{k}}}^{2}, (107)
F(𝒌,ωn)=\displaystyle{F}(\bm{k},\omega_{n})= 1ZS[Af+Bfη𝒌γ]UT,F~(𝒌,ωn)=UTZS[Af+Bfη𝒌γ],\displaystyle\frac{1}{Z_{\mathrm{S}}}\left[A_{f}+{B_{f}}\vec{\eta}_{\bm{k}}\cdot{\vec{\gamma}}\right]U_{T},\quad\undertilde{F}(\bm{k},\omega_{n})=\frac{U_{T}}{Z_{\mathrm{S}}}\left[A_{f}+{B_{f}}\vec{\eta}_{\bm{k}}\cdot{\vec{\gamma}}\right], (108)
Af=\displaystyle A_{f}= 2ξ𝒌ϵ𝒌η𝒌,Bf=(ωn2+ξ𝒌2+η𝒌2+ϵ𝒌2),ZS=(ωn2+ξ𝒌2+η𝒌2+ϵ𝒌2)24ξ𝒌2ϵ𝒌2=ZN(Ω),\displaystyle-2\xi_{\bm{k}}\,\vec{\epsilon}_{\bm{k}}\cdot\vec{\eta}_{\bm{k}},\quad B_{f}=(\omega_{n}^{2}+{\xi_{\bm{k}}}^{2}+{\vec{\eta}_{\bm{k}}}^{2}+{\vec{\epsilon}_{\bm{k}}}^{2}),\quad Z_{\mathrm{S}}=(\omega_{n}^{2}+{\xi_{\bm{k}}}^{2}+{\vec{\eta}_{\bm{k}}}^{2}+{\vec{\epsilon}_{\bm{k}}}^{2})^{2}-4{\xi_{\bm{k}}}^{2}{\vec{\epsilon}_{\bm{k}}}^{2}=Z_{N}(\Omega), (109)

with Ω=ωn2+η𝒌2\Omega=\sqrt{\omega_{n}^{2}+{\vec{\eta}_{\bm{k}}}^{2}}. When we carry out the summation over the wavenumber, we find

N0\displaystyle N_{0} 𝑑ξTr[GS(𝒌,iωn)J~μGS(𝒌,iωn)J~ν]𝒌^,\displaystyle\int_{-\infty}^{\infty}d\xi\langle\mathrm{Tr}\left[{G}_{\mathrm{S}}(\bm{k},i\omega_{n})\,\tilde{J}^{\mu}\,G_{\mathrm{S}}(\bm{k},i\omega_{n})\,\tilde{J}^{\nu}\right]\rangle_{\hat{\bm{k}}},
=\displaystyle= N0𝑑ξ[{AN2(Ω)ZN2(Ω)+η𝒌2(Ω2+ϵ𝒌2+ξ2)2ZS2}J~μJ~ν+{BN2(Ω)ZN2(Ω)+4ϵ𝒌2ξ2ZS2}Lμ,ν(η𝒌)]𝒌^,\displaystyle N_{0}\int_{-\infty}^{\infty}d\xi\left\langle\left[\left\{\frac{A_{N}^{2}(\Omega)}{Z_{N}^{2}(\Omega)}+\frac{{\vec{\eta}_{\bm{k}}}^{2}(\Omega^{2}+{\vec{\epsilon}_{\bm{k}}}^{2}+\xi^{2})^{2}}{Z_{\mathrm{S}}^{2}}\right\}\tilde{J}_{\mu}\,\tilde{J}_{\nu}+\left\{\frac{B_{N}^{2}(\Omega)}{Z_{N}^{2}(\Omega)}+\frac{4\,{\vec{\epsilon}_{\bm{k}}}^{2}\,\xi^{2}}{Z_{\mathrm{S}}^{2}}\right\}L_{\mu,\nu}(\vec{\eta}_{\bm{k}})\right]\right\rangle_{\hat{\bm{k}}},
=\displaystyle= πN04Ω3(Ω2+ε2)[P+η𝒌2(2Ω2+ε2)δμ,ν+ε2Lμ,ν(η𝒌)]𝒌^,\displaystyle\left\langle\frac{\pi\,N_{0}}{4\Omega^{3}(\Omega^{2}+\varepsilon^{2})}\left[P_{+}\,{\vec{\eta}_{\bm{k}}}^{2}\,(2\Omega^{2}+\varepsilon^{2})\delta_{\mu,\nu}+\varepsilon^{2}\,{L}_{\mu,\nu}(\vec{\eta}_{\bm{k}})\right]\right\rangle_{\hat{\bm{k}}}, (110)
N0\displaystyle N_{0} 𝑑ξTr[F~S(𝒌,iωn)J~μFS(𝒌,iωn)(J~ν)]𝒌^,\displaystyle\int_{-\infty}^{\infty}d\xi\langle\mathrm{Tr}\left[\undertilde{F}_{\mathrm{S}}(\bm{k},i\omega_{n})\,\tilde{J}^{\mu}\,F_{\mathrm{S}}(\bm{k},i\omega_{n})\,(\tilde{J}^{\nu})^{\ast}\right]\rangle_{\hat{\bm{k}}},
=\displaystyle= N0𝑑ξ[{4η𝒌2ϵ𝒌2ξ2)ZS2}J~μJ~ν+{(ξ2+Ω2+ϵ𝒌2)2ZS2}Lμ,ν(η𝒌)]𝒌^,\displaystyle N_{0}\int_{-\infty}^{\infty}d\xi\left\langle\left[\left\{\frac{4{\vec{\eta}_{\bm{k}}}^{2}\,{\vec{\epsilon}_{\bm{k}}}^{2}\,\xi^{2})}{Z_{\mathrm{S}}^{2}}\right\}\tilde{J}_{\mu}\,\tilde{J}_{\nu}+\left\{\frac{(\xi^{2}+\Omega^{2}+{\vec{\epsilon}_{\bm{k}}}^{2})^{2}}{Z_{\mathrm{S}}^{2}}\right\}L_{\mu,\nu}(\vec{\eta}_{\bm{k}})\right]\right\rangle_{\hat{\bm{k}}},
=\displaystyle= πN04Ω3(Ω2+ε2)[P+η𝒌2ε2δμ,ν+(2Ω2+ε2)Lμ,ν(η𝒌)]𝒌^.\displaystyle\left\langle\frac{\pi N_{0}}{4\Omega^{3}(\Omega^{2}+\varepsilon^{2})}\left[P_{+}\,{\vec{\eta}_{\bm{k}}}^{2}\,\varepsilon^{2}\delta_{\mu,\nu}+(2\Omega^{2}+\varepsilon^{2})\,{L}_{\mu,\nu}(\vec{\eta}_{\bm{k}})\right]\right\rangle_{\hat{\bm{k}}}. (111)

The average Lμ,ν𝒌^\langle L_{\mu,\nu}\rangle_{\hat{\bm{k}}} describes the anisotropy and the off-diagonal response of the spin susceptibility.

Appendix C Gap equation

The attractive interactions between two electrons are necessary for Cooper pairing. Some bosonic excitation usually mediates the attractive interactions. In this Appendix, we assume the attractive interaction phenomenologically and derive the gap equation for superconducting states discussed in this paper. The pair potential of the superconducting states is defined by

Δα,β(𝒌^)=1Vvol𝒌λ,τVα,β;λ,γ(𝒌𝒌)c𝒌,λc𝒌,τ=1Vvol𝒌Tωnλ,τVα,β;λ,τ(𝒌𝒌)Fλ,τ(𝒌,ωn),\displaystyle\Delta_{\alpha,\beta}(\hat{\bm{k}})=\frac{1}{V_{\mathrm{vol}}}\sum_{\bm{k}^{\prime}}\sum_{\lambda,\tau}V_{\alpha,\beta;\lambda,\gamma}({\bm{k}}-{\bm{k}}^{\prime})\left\langle c_{\bm{k}^{\prime},\lambda}\,c_{-\bm{k}^{\prime},\tau}\right\rangle=-\frac{1}{V_{\mathrm{vol}}}\sum_{\bm{k}^{\prime}}T\sum_{\omega_{n}}\sum_{\lambda,\tau}V_{\alpha,\beta;\lambda,\tau}({\bm{k}}-{\bm{k}}^{\prime})\,F_{\lambda,\tau}(\bm{k}^{\prime},\omega_{n}), (112)

where α\alpha, β\beta, λ\lambda, and τ\tau are the indices of pseudospin of an electron. The attractive interaction Vα,β;λ,τV_{\alpha,\beta;\lambda,\tau} works on two electrons with λ\lambda and τ\tau and generats the pair potential between two electrons with α\alpha and β\beta. The attractive interaction can be decomposed as

Vα,β;λ,τ(𝒌𝒌)=ν=15gν(𝒌𝒌)(γνUT)α,β(γνUT)λ,τ.\displaystyle V_{\alpha,\beta;\lambda,\tau}({\bm{k}}-{\bm{k}}^{\prime})=\sum_{\nu=1-5}g_{\nu}({\bm{k}}-{\bm{k}}^{\prime})\,(\gamma_{\nu}\,U_{T})_{\alpha,\beta}\,(\gamma_{\nu}\,U_{T})^{\ast}_{\lambda,\tau}. (113)

For A1gA_{1g} states in Sec. III A, we choose

gν(𝒌𝒌)=\displaystyle g_{\nu}(\bm{k}-\bm{k}^{\prime})= {gcν(𝒌^)cν(𝒌^)ν=130ν=4,5T2girreps\displaystyle\left\{\begin{array}[]{ll}g\,c_{\nu}(\hat{\bm{k}})\,c_{\nu}(\hat{\bm{k}^{\prime}})&\nu=1-3\\ 0&\nu=4,5\end{array}\right.\quad T_{2g}\,\textrm{irreps} (116)
gν(𝒌𝒌)=\displaystyle g_{\nu}(\bm{k}-\bm{k}^{\prime})= {0ν=13gcν(𝒌^)cν(𝒌^)ν=4,5Egirreps\displaystyle\left\{\begin{array}[]{ll}0&\nu=1-3\\ g\,c_{\nu}(\hat{\bm{k}})\,c_{\nu}(\hat{\bm{k}^{\prime}})&\nu=4,5\end{array}\right.\quad E_{g}\,\textrm{irreps} (119)
gν(𝒌𝒌)=\displaystyle g_{\nu}(\bm{k}-\bm{k}^{\prime})= gcν(𝒌^)cν(𝒌^),ν=15T2g+Egirreps.\displaystyle g\,c_{\nu}(\hat{\bm{k}})\,c_{\nu}(\hat{\bm{k}^{\prime}}),\;\nu=1-5\quad T_{2g}+E_{g}\,\textrm{irreps}. (120)

By substituting the anomalous Green’s function in Eq. (108) into Eq. (112), we obtain the gap equation

Δ=TωngN0𝑑ξBfΔZS,\displaystyle\Delta=T\sum_{\omega_{n}}gN_{0}\int d\xi\frac{B_{f}\,\Delta}{Z_{\mathrm{S}}}, (121)

where we have used Eq. (8). After integrating over ξ\xi, we obtain

1=gN0Tωn1ωn2+Δ2.\displaystyle 1=gN_{0}T\sum_{\omega_{n}}\frac{1}{\sqrt{\omega_{n}^{2}+\Delta^{2}}}. (122)

The results coinsides with the gap equation in BCS theory. For T2gT_{2g}, EgE_{g}, and an admixture state of them in Sec. III B and C, we replace cν(𝒌^)c_{\nu}(\hat{\bm{k}}) by 1 for all ν\nu in Eqs. (116)-(120). The gap equation for such states is identical to Eq. (122).

References