Prepared for submission to JHEP
Introduction
The origin of the nucleon mass is one of the most fundamental open problems in the strong-interaction sector of the Standard Model. The Higgs mechanism fixes the current masses of the light quarks, but these account for only a few percent of the nucleon mass; the overwhelming remainder is generated dynamically by the nonperturbative dynamics of quantum chromodynamics (QCD). In the conventional picture this dynamics is governed by the spontaneous breaking of chiral symmetry: a nonvanishing quark condensate ⟨ q ˉ q ⟩ \langle\bar{q}q\rangle ⟨ q ˉ q ⟩ dresses the nearly massless current quarks into massive constituent quarks and sets the scale of the low-lying hadron spectrum [1 , 2 , 3 , 4 , 5 , 6 ] . Taken literally, this mechanism ties essentially the entire nucleon mass to the chiral order parameter, so that the mass is expected to become almost zero as chiral symmetry is restored at high temperature or baryon density.
The parity doublet model (PDM) offers a physically distinct viewpoint [7 , 8 , 9 ] . In this framework, a baryon and its chiral partner are embedded in a single chiral multiplet, which naturally accommodates a chiral invariant mass m 0 m_0 m 0 that does not originate from the quark condensate. As a consequence, only part of the baryon mass is generated by the chiral condensate: the chiral invariant mass m 0 m_0 m 0 persists even in the chiral restored phase, where the two chiral partners become degenerate. The existence of chiral invariant masss is supported by the lattice QCD calculations at finite temperature [10 , 11 , 12 ] along with the sum rule calculations [13 , 14 , 15 ] and the model has furthermore been successfully applied to the studies of nuclear matter and neutron stars [16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 ] .
Recent observations from neutron star provide strong constraints and a relatively large value of m 0 m_0 m 0 is favored [27 , 28 , 29 , 30 , 31 , 32 ] . More recently, the framework has been extended from two to three flavors, placing the octet baryons and their parity partners on an equal footing [33 , 34 , 35 , 36 , 37 , 38 , 39 , 40 ] . In all of these realizations of the parity doublet model, the analysis is carried out from a purely hadronic perspective, and the chiral invariant mass m 0 m_0 m 0 enters as a phenomenological parameter: it is fixed by fitting the baryon spectrum or nuclear and neutron-star data, rather than being derived from more fundamental degrees of freedom. Which component of the QCD dynamics generates a baryon mass that is insensitive to the chiral condensate thus remains an open question, and clarifying the microscopic origin of the chiral invariant mass is one of the most important problems in this context.
A natural microscopic perspective on this question is provided by the diquark correlations inside the baryon. Lattice QCD, Dyson–Schwinger studies, and phenomenology consistently support a picture in which two quarks form a compact color-antitriplet diquark correlation that subsequently combines with the remaining quark to form a color-singlet baryon, with the scalar “good” diquark (J P = 0 + J^P = 0^+ J P = 0 + ) being the most strongly correlated channel [41 , 42 , 43 ] . Since the diquark itself is a colored object, it should not be regarded as an asymptotic hadron, but rather as a confined substructure inside the baryon. Its binding and mass scale are therefore generated by the non-Abelian gluonic dynamics that is responsible for confinement in the full color-singlet baryon. In this sense, a substantial part of the diquark mass can be attributed to chromodynamic interactions, rather than to the quark condensate associated with spontaneous chiral symmetry breaking. In addition, the gluon-mediated interaction that binds the diquark and the remaining quark into the baryon is of chromodynamic origin and insensitive to the chiral condensate. This observation suggests that the chiral invariant interaction together with the chiral invariant mass of the diquark, both of which survive even when chiral symmetry is restored, generates the chiral invariant mass of baryons. The diquark sector, moreover, carries a parity doubling structure of its own [44 ] : alongside the scalar (0 + 0^{+} 0 + ) diquark there exists a pseudoscalar (0 − 0^{-} 0 − ) diquark, and the two transform into one another under chiral S U ( 3 ) R × S U ( 3 ) L \mathrm{SU}(3)_{R} \times\mathrm{SU}(3)_{L} SU ( 3 ) R × SU ( 3 ) L and form a diquark parity doublet. Furthermore, two diquarks are degenerated with each other to have the chiral invariant mass when the chiral symmetry is restored. Coupling the remaining quark to this diquark doublet transfers the doubling structure to the baryons and naturally realizes the mirror assignment of the parity doublet model. In this picture, both characteristic ingredients of the parity doublet model—the chiral invariant baryon mass and the parity doubling of baryons—emerge from the same underlying diquark dynamics driven predominantly by gluonic interactions.
Motivated by these considerations, we construct a three-flavor chiral quark–diquark model to describe the baryons and their excited states starting from the underlying quark and diquark degrees of freedom, in the spirit of NJL-type quark–diquark descriptions of baryons [45 , 46 ] . The scalar (0 + 0^{+} 0 + ) and pseudoscalar (0 − 0^{-} 0 − ) diquarks are organized into a parity doublet characterized by a chiral invariant diquark mass, while the quark and diquark are bound into a baryon through an effective one-gluon-exchange interaction. This interaction is formulated as a chiral symmetric four-body quark–diquark coupling together with its crossed and U ( 1 ) A \mathrm{U}(1)_{A} U ( 1 ) A -breaking counterparts. Employing an auxiliary-field hadronization reformulation, we introduce color-singlet composite baryon fields and evaluate the one- and two-loop quark–diquark self-energies, from which the baryon mass matrix—including the mixing induced by the chiral order parameter Σ \Sigma Σ —is derived analytically. A notable feature of this construction is that the chiral representations of the baryons are not assigned by hand but dictated by the quark–diquark structure: pairing a right-handed or left-handed quark with a diquark of the same chirality produces composites in ( 8 , 1 ) ⊕ ( 1 , 8 ) (8,1)\oplus(1,8) ( 8 , 1 ) ⊕ ( 1 , 8 ) representation under the S U ( 3 ) R × S U ( 3 ) L \mathrm{SU}(3)_{R}\times\mathrm{SU}(3)_{L} SU ( 3 ) R × SU ( 3 ) L symmetry, while the crossed pairing produces composites in ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) (3,\bar{3})\oplus(\bar{3},3) ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) representation, and each comes automatically with its mirror partner. The existence of the mirror assignment and of the two distinct chiral representations—inputs of the conventional hadronic parity doublet model—is thus naturally explained, and all four multiplets are fully mixed by the chiral order parameter. The resulting spectrum is organized by these two chiral representations, whose mixing determines both the physical masses and the chiral invariant mass content of each state. Within this framework, the baryon chiral invariant masses are no longer phenomenological input parameters of a hadronic Lagrangian but emerge dynamically from the chiral invariant diquark mass in combination with the four-body quark–diquark interaction: as we will show explicitly, the composite baryon acquires a finite mass from the quark–diquark loop dynamics even for massless quarks, with both the diquark mass and the interaction strength setting its scale. When chiral symmetry is restored, the constituent quark mass generated by spontaneous chiral symmetry breaking vanishes, and the remaining baryon mass is inherited from the chiral invariant diquark sector—the invariant diquark mass and the four-body quark–diquark interaction—thereby providing a microscopic realization of the parity-doublet mass parameter.
We then confront the model with experimental data. Scanning the two chiral invariant masses and fitting the remaining seven couplings, we find that the model reproduces the masses of the octet ground states and their excited partners as well as the nucleon sigma terms σ π N \sigma_{\pi N} σ π N and σ s N \sigma_{s N} σ s N remarkably well, with the correct ordering m N < m Σ < m Ξ m_N < m_\Sigma< m_\Xi m N < m Σ < m Ξ in both parity sectors—an economy of description that was not achieved in the purely hadronic three-flavor parity doublet models to the best of our knowledge.
The remainder of the paper is organized as follows. In § 2 we fix our conventions for the chiral and discrete transformations of the quark, meson, and diquark fields. § 3 constructs the chiral invariant four-body quark–diquark interaction, defines the color-singlet composites together with their mirror and crossed partners, and recasts the interaction through the auxiliary-field method. In Sec. 4 we evaluate the one-loop quark–diquark contribution to the two-point function of the composite baryon. We derive the chiral invariant mass of baryons even in the absence of the chiral order parameter, exhibiting explicitly how the baryon mass is generated in the model. In Sec. 5 we couple the quarks and diquarks to the meson field which induces the mixing among the composite baryons. We then assemble the N N N , Σ \Sigma Σ , and Ξ \Xi Ξ mass matrices, including the explicit chiral symmetry breaking. § 7 presents the numerical analysis: fitting to the octet spectrum and the nucleon sigma terms; determination of chiral-invariant mass components in eigenstates and parity-doubling patterns under the restoration of chiral symmetry. We summarize and discuss our results in § 8 . Technical details regarding the proof of discrete symmetries and of the loop integrals are collected in Appendices § 10 and § 11 .
Fields, chiral representations, and transformations
The model is built on the chiral symmetry group G = S U ( 3 ) R × S U ( 3 ) L G = \mathrm{SU}(3)_R \times\mathrm{SU}(3)_L G = SU ( 3 ) R × SU ( 3 ) L acting on three light quark flavors. We use the chiral projectors P R , L = 1 2 ( 1 ± γ 5 ) P_{R,L} = \frac{1}{2}(1 \pm\gamma_5) P R , L = 2 1 ( 1 ± γ 5 ) and write q R , L = P R , L q q_{R,L} = P_{R,L}q q R , L = P R , L q for the right- and left-handed quark fields. The dynamical degrees of freedom are the quarks, color-antitriplet scalar (0 + 0^+ 0 + ) and pseudoscalar (0 − 0^- 0 − ) diquarks, and treat the color-singlet meson field as an external field. In this section we fix our conventions for the chiral and discrete-symmetry transformations of these fields. Throughout, the first entry of a representation label ( R , L ) (\mathbf{R},\mathbf{L}) ( R , L ) refers to S U ( 3 ) R \mathrm{SU}(3)_R SU ( 3 ) R and the second to S U ( 3 ) L \mathrm{SU}(3)_L SU ( 3 ) L .
Quarks
The right- and left-handed quarks carry a color index a a a (fundamental 3 \mathbf{3} 3 ) and flavor indices α \alpha α and i i i , and belong to the following representations:
q R a , α ∼ ( 3 , 1 ) , q L a , i ∼ ( 1 , 3 ) , q ˉ R , a , α ∼ ( 3 ˉ , 1 ) , q ˉ L , a , i ∼ ( 1 , 3 ˉ ) , (1) q_R^{a,\alpha} \sim(3,1), \qquad q_L^{a,i} \sim(1,3), \qquad\bar q_{R,a,\alpha} \sim(\bar{3},1), \qquad\bar q_{L,a,i} \sim(1,\bar{3}), \tag*{(1)} q R a , α ∼ ( 3 , 1 ) , q L a , i ∼ ( 1 , 3 ) , q ˉ R , a , α ∼ ( 3 ˉ , 1 ) , q ˉ L , a , i ∼ ( 1 , 3 ˉ ) , (1)
which transform as
q R a , α → ( U R ) α β q R a , β , q L a , i → ( U L ) i j q L a , j , (2) q_R^{a,\alpha} \to(U_R)^\alpha{}_\beta q_R^{a,\beta}, \qquad q_L^{a,i} \to(U_L)^i{}_j q_L^{a,j}, \tag*{(2)} q R a , α → ( U R ) α β q R a , β , q L a , i → ( U L ) i j q L a , j , (2)
q ˉ R , a , α → q ˉ R , a , β ( U R † ) β α , q ˉ L , a , i → q ˉ L , a , j ( U L † ) j i . (3) \bar q_{R,a,\alpha} \to\bar q_{R,a,\beta}(U_R^\dagger)^\beta{}_\alpha, \qquad\bar q_{L,a,i} \to\bar q_{L,a,j}(U_L^\dagger)^j{}_i. \tag*{(3)} q ˉ R , a , α → q ˉ R , a , β ( U R † ) β α , q ˉ L , a , i → q ˉ L , a , j ( U L † ) j i . (3)
with U R , L ∈ S U ( 3 ) R , L U_{R,L} \in\mathrm{SU}(3)_{R,L} U R , L ∈ SU ( 3 ) R , L . Here and throughout we use indices ( α , β , … ) (\alpha,\beta,\ldots) ( α , β , … ) for S U ( 3 ) R \mathrm{SU}(3)_{R} SU ( 3 ) R and ( i , j , … ) (i,j,\ldots) ( i , j , … ) for S U ( 3 ) L \mathrm{SU}(3)_{L} SU ( 3 ) L . Under the parity transformation, the chirality of the quark is flipped,
P : q R a ( t , x ⃗ ) → γ 0 q L a ( t , − x ⃗ ) . (4) P:\quad q_{R}^{a}(t,\vec{x}) \to\gamma^{0}q_{L}^{a}(t,-\vec{x}). \tag*{(4)} P : q R a ( t , x ) → γ 0 q L a ( t , − x ) . (4)
Meson field
The chiral order parameter is expressed by the complex 3 × 3 3 \times3 3 × 3 matrix as
Σ α i = σ α i + i π α i ∼ ( 3 ˉ , 3 ) , Σ → U L Σ U R † , P : Σ → Σ † , (5) \Sigma_{\alpha}^{i} = \sigma_{\alpha}^{i} + i\pi_{\alpha}^{i} \sim(\bar{3},3),\qquad\Sigma\to U_{L}\Sigma U_{R}^{\dagger},\qquad P:\ \Sigma\to\Sigma^{\dagger}, \tag*{(5)} Σ α i = σ α i + i π α i ∼ ( 3 ˉ , 3 ) , Σ → U L Σ U R † , P : Σ → Σ † , (5)
whose scalar (σ α i \sigma_{\alpha}^{i} σ α i ) and pseudoscalar (π α i \pi_{\alpha}^{i} π α i ) components describe the scalar and pseudoscalar meson nonets. Spontaneous chiral symmetry breaking is encoded in the vacuum expectation value ⟨ Σ α i ⟩ = f δ α i \langle\Sigma_{\alpha}^{i}\rangle= f\delta_{\alpha}^{i} ⟨ Σ α i ⟩ = f δ α i with f ≃ f π f \simeq f_{\pi} f ≃ f π , where f π f_{\pi} f π is the pion decay constant; once S U ( 3 ) \mathrm{SU}(3) SU ( 3 ) flavor breaking is included this generalizes to ⟨ Σ ⟩ = diag ( f π , f π , f s ) \langle\Sigma\rangle= \operatorname{diag}(f_{\pi},f_{\pi},f_{s}) ⟨ Σ ⟩ = diag ( f π , f π , f s ) with f s = 2 f K − f π f_{s}=2f_{K}-f_{\pi} f s = 2 f K − f π , where f K f_{K} f K is the kaon decay constant.
Scalar and pseudoscalar diquarks
The spin-0, color-antitriplet diquarks are expressed by the local quark bilinears as
( d R ) a , α = ϵ a b c ϵ α β γ [ ( q R T ) b , β C ( q R ) c , γ ] , ( d L ) a , i = ϵ a b c ϵ i j k [ ( q L T ) b , j C ( q L ) c , k ] , (6) (d_{R})_{a,\alpha}=\epsilon_{abc}\epsilon_{\alpha\beta\gamma}\left[(q_{R}^{T})^{b,\beta}C(q_{R})^{c,\gamma}\right],\qquad(d_{L})_{a,i}=\epsilon_{abc}\epsilon_{ijk}\left[(q_{L}^{T})^{b,j}C(q_{L})^{c,k}\right], \tag*{(6)} ( d R ) a , α = ϵ ab c ϵ α β γ [ ( q R T ) b , β C ( q R ) c , γ ] , ( d L ) a , i = ϵ ab c ϵ ij k [ ( q L T ) b , j C ( q L ) c , k ] , (6)
with the charge-conjugation matrix C = i γ 0 γ 2 C=i\gamma^{0}\gamma^{2} C = i γ 0 γ 2 . The antisymmetric color (ϵ a b c \epsilon_{abc} ϵ ab c ) and flavor (ϵ α β γ , ϵ i j k \epsilon_{\alpha\beta\gamma},\epsilon_{ijk} ϵ α β γ , ϵ ij k ) contractions project onto the color and flavor antitriplets, so that the diquarks form chiral antitriplets,
( d R ) a , α ∼ ( 3 ˉ , 1 ) , ( d L ) a , i ∼ ( 1 , 3 ˉ ) , (7) (d_{R})_{a,\alpha}\sim(\bar{3},1),\qquad(d_{L})_{a,i}\sim(1,\bar{3}), \tag*{(7)} ( d R ) a , α ∼ ( 3 ˉ , 1 ) , ( d L ) a , i ∼ ( 1 , 3 ˉ ) , (7)
transforming as
( d R ) a , α → ( d R ) a , β ( U R † ) α β , ( d L ) a , i → ( d L ) a , j ( U L † ) i j . (8) (d_{R})_{a,\alpha}\to(d_{R})_{a,\beta}(U_{R}^{\dagger})^{\beta}_{\alpha},\qquad(d_{L})_{a,i}\to(d_{L})_{a,j}(U_{L}^{\dagger})^{j}_{i}. \tag*{(8)} ( d R ) a , α → ( d R ) a , β ( U R † ) α β , ( d L ) a , i → ( d L ) a , j ( U L † ) i j . (8)
Under the parity transformation, two chiralities are interchanged with flipping the sign as
P : d R ↔ − d L , (9) P:\quad d_{R}\leftrightarrow-d_{L}, \tag*{(9)} P : d R ↔ − d L , (9)
which follows directly from the quark parity transformation together with γ 0 C γ 0 = − C \gamma^{0}C\gamma^{0}=-C γ 0 C γ 0 = − C . The parity eigenstates are the scalar (0 + 0^{+} 0 + ) and pseudoscalar (0 − 0^{-} 0 − ) combinations given by
S a , i = 1 2 [ ( d R ) a , i − ( d L ) a , i ] ( 0 + ) , P a , i = 1 2 [ ( d R ) a , i + ( d L ) a , i ] ( 0 − ) , (10) S_{a,i}=\frac{1}{\sqrt{2}}\left[(d_{R})_{a,i}-(d_{L})_{a,i}\right]\ (0^{+}),\qquad P_{a,i}=\frac{1}{\sqrt{2}}\left[(d_{R})_{a,i}+(d_{L})_{a,i}\right]\ (0^{-}), \tag*{(10)} S a , i = 2 1 [ ( d R ) a , i − ( d L ) a , i ] ( 0 + ) , P a , i = 2 1 [ ( d R ) a , i + ( d L ) a , i ] ( 0 − ) , (10)
or equivalently ( d R ) a , i = 1 2 ( P a , i + S a , i ) (d_{R})_{a,i}=\frac{1}{\sqrt{2}}(P_{a,i}+S_{a,i}) ( d R ) a , i = 2 1 ( P a , i + S a , i ) and ( d L ) a , i = 1 2 ( P a , i − S a , i ) (d_{L})_{a,i}=\frac{1}{\sqrt{2}}(P_{a,i}-S_{a,i}) ( d L ) a , i = 2 1 ( P a , i − S a , i ) . We summarize the quantum numbers and chiral representations of all the fields in Table 1. The scalar and pseudoscalar diquarks thus constitute a parity doublet; their coupling to Σ \Sigma Σ , introduced in the next stage of the construction, together with ⟨ Σ ⟩ \langle\Sigma\rangle ⟨ Σ ⟩ will split their masses.
caption: Quantum numbers and chiral representations of the quark, meson, and diquark fields.
Effective Lagrangian of the quark-diquark model without the Σ \Sigma Σ field
It is instructive to first analyze the quark–diquark dynamics in the absence of the chiral order parameter Σ \Sigma Σ ; the coupling to Σ \Sigma Σ is introduced in the next section. In this section we write down a chiral symmetric four-body quark–diquark interaction generated by one-gluon exchange, and define the color-singlet quark–diquark (baryon) composites together with their mirror and crossed partners.
The Lagrangian of the quark-diquark model without meson field Σ \Sigma Σ is expressed as
L = q ˉ i ∂̸ q + ∂ μ d R , α ( ∂ μ d R , α ) † + ∂ μ d L , i ( ∂ μ d L , i ) † − m d 2 ( d R , α d R , α † + d L , i d L , i † ) + L 4 , (11) \mathcal{L}=\bar{q}i\not{\partial}q+\partial_{\mu}d_{R,\alpha}(\partial^{\mu}d_{R,\alpha})^{\dagger}+\partial_{\mu}d_{L,i}(\partial^{\mu}d_{L,i})^{\dagger}-m_{d}^{2}\left(d_{R,\alpha}d_{R,\alpha}^{\dagger}+d_{L,i}d_{L,i}^{\dagger}\right)+\mathcal{L}_{4}, \tag*{(11)} L = q ˉ i ∂ q + ∂ μ d R , α ( ∂ μ d R , α ) † + ∂ μ d L , i ( ∂ μ d L , i ) † − m d 2 ( d R , α d R , α † + d L , i d L , i † ) + L 4 , (11)
where m d m_d m d is the chiral invariant diquark mass and L 4 \mathcal{L}_4 L 4 collects all the four-body interactions discussed below.
Four-body quark–diquark interactions
The chiral invariant four-body interaction, which pairs a quark and a diquark of the same chirality is given by
L q d = − G q d [ { ( q ˉ L ) c , i ( d L † ) c , j } { ( i ∂ μ d L ) a , j γ μ ( q L ) a , i } + { ( q ˉ L ) c , i γ μ ( − i ∂ μ d L † ) c , j } { ( d L ) a , j ( q L ) a , i } + { ( q ˉ R ) c , α ( d R † ) c , β } { ( i ∂ μ d R ) a , β γ μ ( q R ) a , α } + { ( q ˉ R ) c , α γ μ ( − i ∂ μ d R † ) c , β } { ( d R ) a , β ( q R ) a , α } ] . (12) \begin{aligned}
\mathcal{L}_{qd}=-G_{qd}\Big[&
\left\{(\bar{q}_{L})_{c,i}(d_{L}^{\dagger})^{c,j}\right\}
\left\{(i\partial_{\mu}d_{L})_{a,j}\gamma^{\mu}(q_{L})^{a,i}\right\}\\
&+\left\{(\bar{q}_{L})_{c,i}\gamma^{\mu}(-i\partial_{\mu}d_{L}^{\dagger})^{c,j}\right\}
\left\{(d_{L})_{a,j}(q_{L})^{a,i}\right\}\\
&+\left\{(\bar{q}_{R})_{c,\alpha}(d_{R}^{\dagger})^{c,\beta}\right\}
\left\{(i\partial_{\mu}d_{R})_{a,\beta}\gamma^{\mu}(q_{R})^{a,\alpha}\right\}\\
&+\left\{(\bar{q}_{R})_{c,\alpha}\gamma^{\mu}(-i\partial_{\mu}d_{R}^{\dagger})^{c,\beta}\right\}
\left\{(d_{R})_{a,\beta}(q_{R})^{a,\alpha}\right\}
\Big]. \tag*{(12)}
\end{aligned} L q d = − G q d [ { ( q ˉ L ) c , i ( d L † ) c , j } { ( i ∂ μ d L ) a , j γ μ ( q L ) a , i } + { ( q ˉ L ) c , i γ μ ( − i ∂ μ d L † ) c , j } { ( d L ) a , j ( q L ) a , i } + { ( q ˉ R ) c , α ( d R † ) c , β } { ( i ∂ μ d R ) a , β γ μ ( q R ) a , α } + { ( q ˉ R ) c , α γ μ ( − i ∂ μ d R † ) c , β } { ( d R ) a , β ( q R ) a , α } ] . (12)
We also allow the crossed interaction
L q d ( 2 ) = − G q d ( 2 ) [ { ( q ˉ L ) c , i ( d R † ) c , β } { ( i ∂ μ d R ) a , β γ μ ( q L ) a , i } + { ( q ˉ L ) c , i γ μ ( − i ∂ μ d R † ) c , β } { ( d R ) a , β ( q L ) a , i } + { ( q ˉ R ) c , α ( d L † ) c , j } { ( i ∂ μ d L ) a , j γ μ ( q R ) a , α } + { ( q ˉ R ) c , α γ μ ( − i ∂ μ d L † ) c , j } { ( d L ) a , j ( q R ) a , α } ] . (13) \mathcal{L}_{qd}^{(2)}=-G_{qd}^{(2)}\left[\left\{\left(\bar{q}_{L}\right)_{c,i}\left(d_{R}^{\dagger}\right)^{c,\beta}\right\}\left\{\left(i\partial_{\mu}d_{R}\right)_{a,\beta}\gamma^{\mu}\left(q_{L}\right)^{a,i}\right\}+\left\{\left(\bar{q}_{L}\right)_{c,i}\gamma^{\mu}\left(-i\partial_{\mu}d_{R}^{\dagger}\right)^{c,\beta}\right\}\left\{\left(d_{R}\right)_{a,\beta}\left(q_{L}\right)^{a,i}\right\}
+\left\{\left(\bar{q}_{R}\right)_{c,\alpha}\left(d_{L}^{\dagger}\right)^{c,j}\right\}\left\{\left(i\partial_{\mu}d_{L}\right)_{a,j}\gamma^{\mu}\left(q_{R}\right)^{a,\alpha}\right\}
+\left\{\left(\bar{q}_{R}\right)_{c,\alpha}\gamma^{\mu}\left(-i\partial_{\mu}d_{L}^{\dagger}\right)^{c,j}\right\}\left\{\left(d_{L}\right)_{a,j}\left(q_{R}\right)^{a,\alpha}\right\}\right]. \tag*{(13)} L q d ( 2 ) = − G q d ( 2 ) [ { ( q ˉ L ) c , i ( d R † ) c , β } { ( i ∂ μ d R ) a , β γ μ ( q L ) a , i } + { ( q ˉ L ) c , i γ μ ( − i ∂ μ d R † ) c , β } { ( d R ) a , β ( q L ) a , i } + { ( q ˉ R ) c , α ( d L † ) c , j } { ( i ∂ μ d L ) a , j γ μ ( q R ) a , α } + { ( q ˉ R ) c , α γ μ ( − i ∂ μ d L † ) c , j } { ( d L ) a , j ( q R ) a , α } ] . (13)
which pairs a quark and a diquark of opposite chirality.
We would like to stress that both interactions may be motivated by one-gluon exchange between the quark and the diquark, as illustrated in Figure 1 (a). The gluon couples to the quark through the color current q ˉ γ μ T A q \bar{q}\gamma^{\mu}T^{A}q q ˉ γ μ T A q , which conserves the quark chirality, and to the spin-0 color-antitriplet diquark through the scalar-QCD current
Figure 1. Microscopic origin of the four-body quark–diquark interactions. (a) One-gluon exchange between the quark (solid line) and the spin-0, color-antitriplet diquark (dashed line). The upper vertex is the chirality-conserving quark–gluon coupling q ˉ γ μ T A q \bar{q}\gamma^\mu T_A q q ˉ γ μ T A q , and the lower vertex is the scalar-QCD diquark–gluon coupling i d † T ˉ A ∂ ↔ μ d i d^\dagger\bar{T}_A\overleftrightarrow{\partial}_\mu d i d † T ˉ A ∂ μ d , which supplies the single derivative acting on the diquark. (b) In the contact approximation for the gluon propagator, a color Fierz rearrangement into the attractive color-singlet quark–diquark channel reduces (a) to the four-body vertices of Eqs. (12) and (3.3) (filled square), which inherit the γ μ \gamma^\mu γ μ structure and the derivative from the two currents in (a).
i d † T ˉ A ∂ μ ↔ d = i d † T ˉ A ( ∂ μ d ) − i ( ∂ μ d † ) T ˉ A d , i d^{\dagger}\bar{T}^{A}\overleftrightarrow{\partial^{\mu}}d=id^{\dagger}\bar{T}^{A}\left(\partial^{\mu}d\right)-i\left(\partial^{\mu}d^{\dagger}\right)\bar{T}^{A}d, i d † T ˉ A ∂ μ d = i d † T ˉ A ( ∂ μ d ) − i ( ∂ μ d † ) T ˉ A d ,
where T ˉ A = − ( T A ) ∗ \bar{T}^{A}=-(T^{A})^{*} T ˉ A = − ( T A ) ∗ is the generator in the 3 ˉ \bar{3} 3 ˉ representation. The diquark–gluon vertex therefore contains a single derivative acting on the diquark line. In the contact approximation for the gluon propagator, as is standard in NJL-type models, the exchange reduces to the current–current interaction
L O G E ∝ − g e f f 2 ( q ˉ γ μ T A q ) ( i d † T ˉ A ∂ μ ↔ d ) . (14) \mathcal{L}_{\mathrm{OGE}}\propto-g_{\mathrm{eff}}^{2}\left(\bar{q}\gamma_{\mu}T^{A}q\right)\left(id^{\dagger}\bar{T}^{A}\overleftrightarrow{\partial^{\mu}}d\right). \tag*{(14)} L OGE ∝ − g eff 2 ( q ˉ γ μ T A q ) ( i d † T ˉ A ∂ μ d ) . (14)
Projecting this interaction onto the color-singlet quark–diquark channel, 3 ⊗ 3 ˉ ⊃ 1 3\otimes\bar{3}\supset1 3 ⊗ 3 ˉ ⊃ 1 , gives the attractive color interaction. Then, after decomposing the quark and diquark fields into their chiral components we obtain the structures in Eqs. (12) and (13) : a chirality-conserving quark current contracted with a single derivative acting on the spin-0 diquark, with the flavor indices connected between the two color-singlet quark–diquark brackets. At the pointlike one-gluon-exchange level the gluon is blind to the chiral labels of both the quark and the diquark, so one naturally expects $G_{qd}=G_{qd}^{(2)}$. In the low-energy effective theory, however, compositeness effects, higher-dimensional gluonic operators, and U(1)$_A$-breaking interactions can renormalize the two channels differently. We therefore treat $G_{qd}$ and $G_{qd}^{(2)}$ as independent phenomenological couplings. By imposing hermiticity of the Lagrangian together with parity and charge-conjugation invariance, one can show that the couplings $G_{qd}$ and $G_{qd}^{(2)}$ are real (see Appendix A`).
In addition to the U(1)A _A A -symmetric interactions (3.2) and (3.3), we may include a four-body interaction term that breaks the U(1)A _A A symmetry,
L A = G A [ { ( q ˉ R ) c , i ( d R † ) c , i } { ( d L ) a , j ( q L ) a , j } + { ( q ˉ L ) c , i ( d L † ) c , i } { ( d R ) a , j ( q R ) a , j } ] . (15) \mathcal{L}_{A}=G_{A}\left[\left\{\left(\bar{q}_{R}\right)_{c,i}\left(d_{R}^{\dagger}\right)^{c,i}\right\}\left\{\left(d_{L}\right)_{a,j}\left(q_{L}\right)^{a,j}\right\}+\left\{\left(\bar{q}_{L}\right)_{c,i}\left(d_{L}^{\dagger}\right)^{c,i}\right\}\left\{\left(d_{R}\right)_{a,j}\left(q_{R}\right)^{a,j}\right\}\right]. \tag*{(15)} L A = G A [ { ( q ˉ R ) c , i ( d R † ) c , i } { ( d L ) a , j ( q L ) a , j } + { ( q ˉ L ) c , i ( d L † ) c , i } { ( d R ) a , j ( q R ) a , j } ] . (15)
which can be derived from the Kobayashi-Maskawa-’t Hooft (KMT) type six-quark interaction. Some comments on the interaction basis are in order. Since the diquarks carry spin zero, a local four-body interaction in the quark–diquark channel has the schematic form
{ q ˉ Γ 1 d † } { d Γ 2 q } \{\bar{q}\Gamma_1 d^\dagger\}\{d\Gamma_2 q\} { q ˉ Γ 1 d † } { d Γ 2 q }
where Γ 1 \Gamma_1 Γ 1 and Γ 2 \Gamma_2 Γ 2 denote certain combinations of the gamma matrices, which act only on the quark fields. The Dirac structure therefore reduces to an ordinary quark bilinear q ˉ Γ 1 Γ 2 q \bar{q}\Gamma_1\Gamma_2q q ˉ Γ 1 Γ 2 q , possibly accompanied by derivatives acting on the diquark fields. Chirality imposes a useful selection rule. With q L , R = P L , R q q_{L,R}=P_{L,R}q q L , R = P L , R q and q ˉ L , R = q ˉ P R , L \bar{q}_{L,R}=\bar{q}P_{R,L} q ˉ L , R = q ˉ P R , L , scalar bilinears connect opposite chiral labels, whereas vector bilinears connect equal chiral labels:
q ˉ L q L = 0 , q ˉ R q R = 0 , q ˉ L γ μ q R = 0 , q ˉ R γ μ q L = 0. (16) \bar{q}_Lq_L=0,\qquad\bar{q}_Rq_R=0,\qquad\bar{q}_L\gamma^\mu q_R=0,\qquad\bar{q}_R\gamma^\mu q_L=0. \tag*{(16)} q ˉ L q L = 0 , q ˉ R q R = 0 , q ˉ L γ μ q R = 0 , q ˉ R γ μ q L = 0. (16)
and similarly
q ˉ L σ μ ν q L = 0 , q ˉ R σ μ ν q R = 0. (17) \bar{q}_L\sigma^{\mu\nu}q_L=0,\qquad\bar{q}_R\sigma^{\mu\nu}q_R=0. \tag*{(17)} q ˉ L σ μν q L = 0 , q ˉ R σ μν q R = 0. (17)
Consequently, at zeroth order in derivatives the nonvanishing quark–diquark structures are chirality-flipping scalar structures. The S U ( 3 ) L × S U ( 3 ) R \mathrm{SU}(3)_L \times\mathrm{SU}(3)_R SU ( 3 ) L × SU ( 3 ) R invariant single-trace combination of this type is the U ( 1 ) A \mathrm{U}(1)_A U ( 1 ) A -breaking interaction in Eq. (15) . In terms of the underlying quarks it has the schematic six-quark structure q ˉ L 3 q R 3 + q ˉ R 3 q L 3 \bar{q}_L^3q_R^3+\bar{q}_R^3q_L^3 q ˉ L 3 q R 3 + q ˉ R 3 q L 3 of the KMT-type. At first order in derivatives, the leading vector structure is instead the chirality-conserving current q ˉ L , R γ μ q L , R \bar{q}_{L,R}\gamma^\mu q_{L,R} q ˉ L , R γ μ q L , R contracted with the scalar-diquark current i d † ∂ ↔ μ d i d^\dagger\overleftrightarrow{\partial}_\mu d i d † ∂ μ d . The resulting U ( 1 ) A \mathrm{U}(1)_A U ( 1 ) A -symmetric single-trace operators are precisely the same-chirality interaction (3.2) and the crossed interaction (3.3). Operators in which the derivative acts on the quark field can be shifted, up to total derivatives, into the above basis plus terms proportional to the free quark equation of motion. Tensor structures involving σ μ ν \sigma^{\mu\nu} σ μν require additional Lorentz indices and therefore enter only at higher order in the derivative expansion.
Within the single-trace form in flavor space for the quark–diquark exchange channel used to generate baryons, Eqs. (12) –(15) therefore form the minimal leading-order interaction basis adopted in this work. Additional local operators are in principle allowed, such as double-flavor-trace terms, in which the flavor indices are closed within each bracket rather than between the two brackets, e.g.
L q d ( 2 t r ) = − G ~ q d { ( q ˉ L ) c , i ( d L † ) c , i } { ( i ∂ μ d L ) a , j γ μ ( q L ) a , j } + ( L → R ) , (18) \mathcal{L}_{qd}^{(2\mathrm{tr})}=-\widetilde{G}_{qd}\left\{(\bar{q}_{L})_{c,i}(d_{L}^{\dagger})^{c,i}\right\}\left\{(i\partial_{\mu}d_{L})_{a,j}\gamma^{\mu}(q_{L})^{a,j}\right\}+(L\to R), \tag*{(18)} L q d ( 2 tr ) = − G q d { ( q ˉ L ) c , i ( d L † ) c , i } { ( i ∂ μ d L ) a , j γ μ ( q L ) a , j } + ( L → R ) , (18)
and direct current–current interactions, in which the quark and diquark fields are contracted into separately color-singlet bilinears, e.g.
L ( c c ) = − G V [ ( q ˉ L ) c , i γ μ ( q L ) c , i ] [ i ( d L † ) a , j ∂ ↔ μ ( d L ) a , j ] + ⋯ , (19) \mathcal{L}^{(\mathrm{cc})}=-G_{V}\left[(\bar{q}_{L})_{c,i}\gamma_{\mu}(q_{L})^{c,i}\right]\left[i(d_{L}^{\dagger})^{a,j}\overleftrightarrow{\partial}^{\mu}(d_{L})_{a,j}\right]+\cdots, \tag*{(19)} L ( cc ) = − G V [ ( q ˉ L ) c , i γ μ ( q L ) c , i ] [ i ( d L † ) a , j ∂ μ ( d L ) a , j ] + ⋯ , (19)
where the ellipsis stands for the flavor-octet, ( L → R ) (L\to R) ( L → R ) , and mixed-chirality combinations such as ( q ˉ L γ μ q L ) ( i d R † ∂ ↔ μ d R ) (\bar{q}_{L}\gamma_{\mu}q_{L})(i d_{R}^{\dagger}\overleftrightarrow{\partial}^{\mu}d_{R}) ( q ˉ L γ μ q L ) ( i d R † ∂ μ d R ) . Their effects are beyond the minimal basis considered here and may be absorbed into higher-order contact terms in the effective theory.
Quark–diquark composites and their mirror and crossed partners
Contracting the color indices while leaving the flavor indices free, we define the color-singlet quark–diquark composites as
( Z 1 R ) β α ∼ ( q R ) a , α ( d R ) a , β , ( Z 1 L ) j i ∼ ( q L ) a , i ( d L ) a , j , (20) (Z_{1R})^{\alpha}_{\beta}\sim(q_{R})^{a,\alpha}(d_{R})_{a,\beta},\qquad(Z_{1L})^{i}_{j}\sim(q_{L})^{a,i}(d_{L})_{a,j}, \tag*{(20)} ( Z 1 R ) β α ∼ ( q R ) a , α ( d R ) a , β , ( Z 1 L ) j i ∼ ( q L ) a , i ( d L ) a , j , (20)
each of which is a spin-1 2 \frac{1}{2} 2 1 Dirac field. Under S U ( 3 ) R × S U ( 3 ) L \mathrm{SU}(3)_{R}\times\mathrm{SU}(3)_{L} SU ( 3 ) R × SU ( 3 ) L they are decomposed as
Z 1 R ∼ ( 1 , 1 ) ⊕ ( 8 , 1 ) , Z 1 L ∼ ( 1 , 1 ) ⊕ ( 1 , 8 ) . (21) Z_{1R}\sim(1,1)\oplus(8,1),\qquad Z_{1L}\sim(1,1)\oplus(1,8). \tag*{(21)} Z 1 R ∼ ( 1 , 1 ) ⊕ ( 8 , 1 ) , Z 1 L ∼ ( 1 , 1 ) ⊕ ( 1 , 8 ) . (21)
It is convenient to introduce the baryon fields with mirror assignment as
( Z 2 L ) β α ∼ ( γ μ q R ) a , α ( i ∂ μ d R ) a , β , ( Z 2 R ) j i ∼ ( γ μ q L ) a , i ( i ∂ μ d L ) a , j , (22) (Z_{2L})^{\alpha}_{\beta}\sim(\gamma^{\mu}q_{R})^{a,\alpha}(i\partial_{\mu}d_{R})_{a,\beta},\qquad(Z_{2R})^{i}_{j}\sim(\gamma^{\mu}q_{L})^{a,i}(i\partial_{\mu}d_{L})_{a,j}, \tag*{(22)} ( Z 2 L ) β α ∼ ( γ μ q R ) a , α ( i ∂ μ d R ) a , β , ( Z 2 R ) j i ∼ ( γ μ q L ) a , i ( i ∂ μ d L ) a , j , (22)
which belong to
Z 2 L ∼ ( 1 , 1 ) ⊕ ( 8 , 1 ) , Z 2 R ∼ ( 1 , 1 ) ⊕ ( 1 , 8 ) . (23) Z_{2L}\sim(1,1)\oplus(8,1),\qquad Z_{2R}\sim(1,1)\oplus(1,8). \tag*{(23)} Z 2 L ∼ ( 1 , 1 ) ⊕ ( 8 , 1 ) , Z 2 R ∼ ( 1 , 1 ) ⊕ ( 1 , 8 ) . (23)
We further decompose the Z \mathcal{Z} Z field into the singlet and the octet part such as
( Z 1 R ) β α = B 1 R s i n δ β α + ( B 1 R ) β α . (24) (Z_{1R})^{\alpha}_{\beta}=B_{1R}^{\mathrm{sin}}\delta^{\alpha}_{\beta}+(B_{1R})^{\alpha}_{\beta}. \tag*{(24)} ( Z 1 R ) β α = B 1 R sin δ β α + ( B 1 R ) β α . (24)
Since the four-body interaction L A \mathcal{L}_{A} L A in Eq. (15) contributes to the masses of B 1 s i n \mathcal{B}_{1}^{\mathrm{sin}} B 1 sin and B 2 s i n \mathcal{B}_{2}^{\mathrm{sin}} B 2 sin , and we expect that the flavor-singlet baryons are heavier than the octet baryons, in the following content, we will only focus on the octet part B \mathcal{B} B . We should note that, at this level, both B 1 \mathcal{B}_{1} B 1 and B 2 \mathcal{B}_{2} B 2 carry positive parity, with parity acting as
B 1 L ↔ P B 1 R , B 2 L ↔ P B 2 R . (25) \mathcal{B}_{1L}\xleftrightarrow{P}\mathcal{B}_{1R},\qquad\mathcal{B}_{2L}\xleftrightarrow{P}\mathcal{B}_{2R}. \tag*{(25)} B 1 L P B 1 R , B 2 L P B 2 R . (25)
Finally, we introduce the crossed composites, built from a quark and a diquark of opposite chirality,
( ψ 1 R ) i α ∼ ( q R ) a , α ( d L ) a , i , ( ψ 1 L ) α i ∼ ( q L ) a , i ( d R ) a , α , (\psi_{1R})^{\alpha}_{i}\sim(q_{R})^{a,\alpha}(d_{L})_{a,i},\qquad(\psi_{1L})^{i}_{\alpha}\sim(q_{L})^{a,i}(d_{R})_{a,\alpha}, ( ψ 1 R ) i α ∼ ( q R ) a , α ( d L ) a , i , ( ψ 1 L ) α i ∼ ( q L ) a , i ( d R ) a , α ,
( ψ 2 L ) i α ∼ ( γ μ q R ) a , α ( i ∂ μ d L ) a , i , ( ψ 2 R ) α i ∼ ( γ μ q L ) a , i ( i ∂ μ d R ) a , α , (26) (\psi_{2L})^{\alpha}_{i}\sim(\gamma^{\mu}q_{R})^{a,\alpha}(i\partial_{\mu}d_{L})_{a,i},\qquad(\psi_{2R})^{i}_{\alpha}\sim(\gamma^{\mu}q_{L})^{a,i}(i\partial_{\mu}d_{R})_{a,\alpha}, \tag*{(26)} ( ψ 2 L ) i α ∼ ( γ μ q R ) a , α ( i ∂ μ d L ) a , i , ( ψ 2 R ) α i ∼ ( γ μ q L ) a , i ( i ∂ μ d R ) a , α , (26)
which correspond to the ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) (\mathbf{3},\bar{\mathbf{3}})\oplus(\bar{\mathbf{3}},\mathbf{3}) ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) chiral representations.
To prepare for the loop calculation, we adopt the auxiliary-field method. For the same-chirality sector we add
L q d A F = C q d [ ( ( B ˉ 1 L ) i j − 1 C q d { ( q ˉ L ) c , i ( d L † ) c , j } ) ( ( B 2 R ) j i − G q d { ( i ∂ μ d L ) a , j γ μ ( q L ) a , i } ) + ( ( B ˉ 2 R ) i j − G q d { ( q ˉ L ) c , i γ μ ( − i ∂ μ d L † ) c , j } ) ( ( B 1 L ) j i − 1 C q d { ( d L ) a , j ( q L ) a , i } ) + ( ( B ˉ 1 R ) i j − 1 C q d { ( q ˉ R ) c , i ( d R † ) c , j } ) ( ( B 2 L ) j i − G q d { ( i ∂ μ d R ) a , j γ μ ( q R ) a , i } ) + ( ( B ˉ 2 L ) i j − G q d { ( q ˉ R ) c , i γ μ ( − i ∂ μ d R † ) c , j } ) ( ( B 1 R ) j i − 1 C q d { ( d R ) a , j ( q R ) a , i } ) ] , (27) \begin{aligned}
\mathcal{L}_{qdAF} = C_{qd} \Biggl[&
\left((\bar{B}_{1L})_{i}^{j} - \frac{1}{C_{qd}}\left\{(\bar{q}_{L})_{c,i}(d_{L}^{\dagger})^{c,j}\right\}\right)
\left((B_{2R})_{j}^{i} - G_{qd}\left\{(i\partial_{\mu}d_{L})_{a,j}\gamma^{\mu}(q_{L})^{a,i}\right\}\right) \\
&+ \left((\bar{B}_{2R})_{i}^{j} - G_{qd}\left\{(\bar{q}_{L})_{c,i}\gamma^{\mu}(-i\partial_{\mu}d_{L}^{\dagger})^{c,j}\right\}\right)
\left((B_{1L})_{j}^{i} - \frac{1}{C_{qd}}\left\{(d_{L})_{a,j}(q_{L})^{a,i}\right\}\right) \\
&+ \left((\bar{B}_{1R})_{i}^{j} - \frac{1}{C_{qd}}\left\{(\bar{q}_{R})_{c,i}(d_{R}^{\dagger})^{c,j}\right\}\right)
\left((B_{2L})_{j}^{i} - G_{qd}\left\{(i\partial_{\mu}d_{R})_{a,j}\gamma^{\mu}(q_{R})^{a,i}\right\}\right) \\
&+ \left((\bar{B}_{2L})_{i}^{j} - G_{qd}\left\{(\bar{q}_{R})_{c,i}\gamma^{\mu}(-i\partial_{\mu}d_{R}^{\dagger})^{c,j}\right\}\right)
\left((B_{1R})_{j}^{i} - \frac{1}{C_{qd}}\left\{(d_{R})_{a,j}(q_{R})^{a,i}\right\}\right)
\Biggr], \tag*{(27)}
\end{aligned} L q d A F = C q d [ ( ( B ˉ 1 L ) i j − C q d 1 { ( q ˉ L ) c , i ( d L † ) c , j } ) ( ( B 2 R ) j i − G q d { ( i ∂ μ d L ) a , j γ μ ( q L ) a , i } ) + ( ( B ˉ 2 R ) i j − G q d { ( q ˉ L ) c , i γ μ ( − i ∂ μ d L † ) c , j } ) ( ( B 1 L ) j i − C q d 1 { ( d L ) a , j ( q L ) a , i } ) + ( ( B ˉ 1 R ) i j − C q d 1 { ( q ˉ R ) c , i ( d R † ) c , j } ) ( ( B 2 L ) j i − G q d { ( i ∂ μ d R ) a , j γ μ ( q R ) a , i } ) + ( ( B ˉ 2 L ) i j − G q d { ( q ˉ R ) c , i γ μ ( − i ∂ μ d R † ) c , j } ) ( ( B 1 R ) j i − C q d 1 { ( d R ) a , j ( q R ) a , i } ) ] , (27)
and, for the crossed sector,
L q d A F ( 2 ) = C q d ( 2 ) [ ( ( ψ ˉ 1 L ) i β − 1 C q d ( 2 ) { ( q ˉ L ) c , i ( d R † ) c , β } ) ( ( ψ 2 R ) β i − G q d ( 2 ) { ( i ∂ μ d R ) a , β γ μ ( q L ) a , i } ) + ( ( ψ ˉ 2 R ) i β − G q d ( 2 ) { ( q ˉ L ) c , i γ μ ( − i ∂ μ d R † ) c , β } ) ( ( ψ 1 L ) β i − 1 C q d ( 2 ) { ( d R ) a , β ( q L ) a , i } ) + ( ( ψ ˉ 1 R ) i j − 1 C q d ( 2 ) { ( q ˉ R ) c , α ( d L † ) c , j } ) ( ( ψ 2 L ) j α − G q d ( 2 ) { ( i ∂ μ d L ) a , j γ μ ( q R ) a , α } ) + ( ( ψ ˉ 2 L ) α j − G q d ( 2 ) { ( q ˉ R ) c , α γ μ ( − i ∂ μ d L † ) c , j } ) ( ( ψ 1 R ) j α − 1 C q d ( 2 ) { ( d L ) a , j ( q R ) a , α } ) ] , (28) \begin{aligned}
\mathcal{L}_{qdAF}^{(2)} = C_{qd}^{(2)} \Biggl[&
\left((\bar{\psi}_{1L})_{i}^{\beta} - \frac{1}{C_{qd}^{(2)}}\left\{(\bar{q}_{L})_{c,i}(d_{R}^{\dagger})^{c,\beta}\right\}\right)
\left((\psi_{2R})_{\beta}^{i} - G_{qd}^{(2)}\left\{(i\partial_{\mu}d_{R})_{a,\beta}\gamma^{\mu}(q_{L})^{a,i}\right\}\right) \\
&+ \left((\bar{\psi}_{2R})_{i}^{\beta} - G_{qd}^{(2)}\left\{(\bar{q}_{L})_{c,i}\gamma^{\mu}(-i\partial_{\mu}d_{R}^{\dagger})^{c,\beta}\right\}\right)
\left((\psi_{1L})_{\beta}^{i} - \frac{1}{C_{qd}^{(2)}}\left\{(d_{R})_{a,\beta}(q_{L})^{a,i}\right\}\right) \\
&+ \left((\bar{\psi}_{1R})_{i}^{j} - \frac{1}{C_{qd}^{(2)}}\left\{(\bar{q}_{R})_{c,\alpha}(d_{L}^{\dagger})^{c,j}\right\}\right)
\left((\psi_{2L})_{j}^{\alpha} - G_{qd}^{(2)}\left\{(i\partial_{\mu}d_{L})_{a,j}\gamma^{\mu}(q_{R})^{a,\alpha}\right\}\right) \\
&+ \left((\bar{\psi}_{2L})_{\alpha}^{j} - G_{qd}^{(2)}\left\{(\bar{q}_{R})_{c,\alpha}\gamma^{\mu}(-i\partial_{\mu}d_{L}^{\dagger})^{c,j}\right\}\right)
\left((\psi_{1R})_{j}^{\alpha} - \frac{1}{C_{qd}^{(2)}}\left\{(d_{L})_{a,j}(q_{R})^{a,\alpha}\right\}\right)
\Biggr], \tag*{(28)}
\end{aligned} L q d A F ( 2 ) = C q d ( 2 ) [ ( ( ψ ˉ 1 L ) i β − C q d ( 2 ) 1 { ( q ˉ L ) c , i ( d R † ) c , β } ) ( ( ψ 2 R ) β i − G q d ( 2 ) { ( i ∂ μ d R ) a , β γ μ ( q L ) a , i } ) + ( ( ψ ˉ 2 R ) i β − G q d ( 2 ) { ( q ˉ L ) c , i γ μ ( − i ∂ μ d R † ) c , β } ) ( ( ψ 1 L ) β i − C q d ( 2 ) 1 { ( d R ) a , β ( q L ) a , i } ) + ( ( ψ ˉ 1 R ) i j − C q d ( 2 ) 1 { ( q ˉ R ) c , α ( d L † ) c , j } ) ( ( ψ 2 L ) j α − G q d ( 2 ) { ( i ∂ μ d L ) a , j γ μ ( q R ) a , α } ) + ( ( ψ ˉ 2 L ) α j − G q d ( 2 ) { ( q ˉ R ) c , α γ μ ( − i ∂ μ d L † ) c , j } ) ( ( ψ 1 R ) j α − C q d ( 2 ) 1 { ( d L ) a , j ( q R ) a , α } ) ] , (28)
where C q d C_{qd} C q d and C q d ( 2 ) C_{qd}^{(2)} C q d ( 2 ) are constants of mass dimension one that ensure the composite fields carry the canonical mass dimension 3/2. Other choices for the normalization of the auxiliary fields are possible; the fields can be normalized canonically at a later stage.
By construction, the equations of motion of the auxiliary fields reproduce the original four-body interactions, so that adding (27) to (12) yields
L q d ′ = L q d + L q d A F = C q d tr [ B ˉ 1 L B 2 R + B ˉ 2 R B 1 L + B ˉ 1 R B 2 L + B ˉ 2 L B 1 R ] − C q d G q d { ( B ˉ 1 L ) i j { ( i ∂ μ d L ) a , j γ μ ( q L ) a , i } − { ( q ˉ L ) c , i ( d L † ) c , j } ( B 2 R ) j i } − ( B ˉ 2 R ) j i { ( d L ) a , j ( q L ) a , i } − C q d G q d { ( q ˉ L ) c , i γ μ ( − i ∂ μ d L † ) c , j } ( B 1 L ) j i − C q d G q d ( B ˉ 1 R ) j i { ( i ∂ μ d R ) a , j γ μ ( q R ) a , i } − { ( q ˉ R ) c , i ( d R † ) c , j } ( B 2 L ) j i − ( B ˉ 2 L ) j i { ( d R ) a , j ( q R ) a , i } − C q d G q d { ( q ˉ R ) c , i γ μ ( − i ∂ μ d R † ) c , j } ( B 1 R ) j i , (29) \begin{aligned}
\mathcal{L}'_{qd} &= \mathcal{L}_{qd}+\mathcal{L}_{qdAF}\\
&= C_{qd}\operatorname{tr}\left[\bar{B}_{1L}B_{2R}+\bar{B}_{2R}B_{1L}+\bar{B}_{1R}B_{2L}+\bar{B}_{2L}B_{1R}\right]\\
&\quad-C_{qd}G_{qd}\left\{(\bar{B}_{1L})_{i}^{j}\left\{(i\partial_{\mu}d_{L})_{a,j}\gamma^{\mu}(q_{L})^{a,i}\right\}-\left\{(\bar{q}_{L})_{c,i}(d_{L}^{\dagger})^{c,j}\right\}(B_{2R})^{i}_{j}\right\}\\
&\quad-\left(\bar{B}_{2R}\right)^{i}_{j}\left\{(d_{L})_{a,j}(q_{L})^{a,i}\right\}-C_{qd}G_{qd}\left\{(\bar{q}_{L})_{c,i}\gamma^{\mu}\left(-i\partial_{\mu}d_{L}^{\dagger}\right)^{c,j}\right\}(B_{1L})^{i}_{j}\\
&\quad-C_{qd}G_{qd}\left(\bar{B}_{1R}\right)^{i}_{j}\left\{(i\partial_{\mu}d_{R})_{a,j}\gamma^{\mu}(q_{R})^{a,i}\right\}-\left\{(\bar{q}_{R})_{c,i}(d_{R}^{\dagger})^{c,j}\right\}(B_{2L})^{i}_{j}\\
&\quad-\left(\bar{B}_{2L}\right)^{i}_{j}\left\{(d_{R})_{a,j}(q_{R})^{a,i}\right\}-C_{qd}G_{qd}\left\{(\bar{q}_{R})_{c,i}\gamma^{\mu}\left(-i\partial_{\mu}d_{R}^{\dagger}\right)^{c,j}\right\}(B_{1R})^{i}_{j}, \tag*{(29)}
\end{aligned} L q d ′ = L q d + L q d A F = C q d tr [ B ˉ 1 L B 2 R + B ˉ 2 R B 1 L + B ˉ 1 R B 2 L + B ˉ 2 L B 1 R ] − C q d G q d { ( B ˉ 1 L ) i j { ( i ∂ μ d L ) a , j γ μ ( q L ) a , i } − { ( q ˉ L ) c , i ( d L † ) c , j } ( B 2 R ) j i } − ( B ˉ 2 R ) j i { ( d L ) a , j ( q L ) a , i } − C q d G q d { ( q ˉ L ) c , i γ μ ( − i ∂ μ d L † ) c , j } ( B 1 L ) j i − C q d G q d ( B ˉ 1 R ) j i { ( i ∂ μ d R ) a , j γ μ ( q R ) a , i } − { ( q ˉ R ) c , i ( d R † ) c , j } ( B 2 L ) j i − ( B ˉ 2 L ) j i { ( d R ) a , j ( q R ) a , i } − C q d G q d { ( q ˉ R ) c , i γ μ ( − i ∂ μ d R † ) c , j } ( B 1 R ) j i , (29)
and, analogously for the crossed sector,
L q d ′ ( 2 ) = L q d ( 2 ) + L q d A F ( 2 ) = C q d ( 2 ) tr [ ψ ˉ 1 L ψ 2 R + ψ ˉ 2 R ψ 1 L + ψ ˉ 1 R ψ 2 L + ψ ˉ 2 L ψ 1 R ] − C q d ( 2 ) G q d ( 2 ) ( ψ ˉ 1 L ) β i { ( i ∂ μ d R ) a , β γ μ ( q L ) a , i } − { ( q ˉ L ) c , i ( d R † ) c , β } ( ψ 2 R ) β i − ( ψ ˉ 2 R ) β i { ( d R ) a , β ( q L ) a , i } − C q d ( 2 ) G q d ( 2 ) { ( q ˉ L ) c , i γ μ ( − i ∂ μ d R † ) c , β } ( ψ 1 L ) β i − C q d ( 2 ) G q d ( 2 ) ( ψ ˉ 1 R ) j α { ( i ∂ μ d L ) a , j γ μ ( q R ) a , α } − { ( q ˉ R ) c , α ( d L † ) c , j } ( ψ 2 L ) j α − ( ψ ˉ 2 L ) j α { ( d L ) a , j ( q R ) a , α } − C q d ( 2 ) G q d ( 2 ) { ( q ˉ R ) c , α γ μ ( − i ∂ μ d L † ) c , j } ( ψ 1 R ) j α . (30) \begin{aligned}
\mathcal{L}_{qd}^{\prime(2)} &= \mathcal{L}_{qd}^{(2)}+\mathcal{L}_{qdAF}^{(2)}\\
&= C_{qd}^{(2)}\operatorname{tr}\left[\bar{\psi}_{1L}\psi_{2R}+\bar{\psi}_{2R}\psi_{1L}+\bar{\psi}_{1R}\psi_{2L}+\bar{\psi}_{2L}\psi_{1R}\right]\\
&\quad-C_{qd}^{(2)}G_{qd}^{(2)}(\bar{\psi}_{1L})^{i}_{\beta}\left\{(i\partial_{\mu}d_{R})_{a,\beta}\gamma^{\mu}(q_{L})^{a,i}\right\}-\left\{(\bar{q}_{L})_{c,i}(d_{R}^{\dagger})^{c,\beta}\right\}(\psi_{2R})^{i}_{\beta}\\
&\quad-(\bar{\psi}_{2R})^{i}_{\beta}\left\{(d_{R})_{a,\beta}(q_{L})^{a,i}\right\}-C_{qd}^{(2)}G_{qd}^{(2)}\left\{(\bar{q}_{L})_{c,i}\gamma^{\mu}\left(-i\partial_{\mu}d_{R}^{\dagger}\right)^{c,\beta}\right\}(\psi_{1L})^{i}_{\beta}\\
&\quad-C_{qd}^{(2)}G_{qd}^{(2)}(\bar{\psi}_{1R})^{\alpha}_{j}\left\{(i\partial_{\mu}d_{L})_{a,j}\gamma^{\mu}(q_{R})^{a,\alpha}\right\}-\left\{(\bar{q}_{R})_{c,\alpha}(d_{L}^{\dagger})^{c,j}\right\}(\psi_{2L})^{\alpha}_{j}\\
&\quad-(\bar{\psi}_{2L})^{\alpha}_{j}\left\{(d_{L})_{a,j}(q_{R})^{a,\alpha}\right\}-C_{qd}^{(2)}G_{qd}^{(2)}\left\{(\bar{q}_{R})_{c,\alpha}\gamma^{\mu}\left(-i\partial_{\mu}d_{L}^{\dagger}\right)^{c,j}\right\}(\psi_{1R})^{\alpha}_{j}. \tag*{(30)}
\end{aligned} L q d ′ ( 2 ) = L q d ( 2 ) + L q d A F ( 2 ) = C q d ( 2 ) tr [ ψ ˉ 1 L ψ 2 R + ψ ˉ 2 R ψ 1 L + ψ ˉ 1 R ψ 2 L + ψ ˉ 2 L ψ 1 R ] − C q d ( 2 ) G q d ( 2 ) ( ψ ˉ 1 L ) β i { ( i ∂ μ d R ) a , β γ μ ( q L ) a , i } − { ( q ˉ L ) c , i ( d R † ) c , β } ( ψ 2 R ) β i − ( ψ ˉ 2 R ) β i { ( d R ) a , β ( q L ) a , i } − C q d ( 2 ) G q d ( 2 ) { ( q ˉ L ) c , i γ μ ( − i ∂ μ d R † ) c , β } ( ψ 1 L ) β i − C q d ( 2 ) G q d ( 2 ) ( ψ ˉ 1 R ) j α { ( i ∂ μ d L ) a , j γ μ ( q R ) a , α } − { ( q ˉ R ) c , α ( d L † ) c , j } ( ψ 2 L ) j α − ( ψ ˉ 2 L ) j α { ( d L ) a , j ( q R ) a , α } − C q d ( 2 ) G q d ( 2 ) { ( q ˉ R ) c , α γ μ ( − i ∂ μ d L † ) c , j } ( ψ 1 R ) j α . (30)
Equations (29) and (30) are the forms used in the subsequent one-loop analysis of the baryon masses.
Baryon mass without the Σ \Sigma Σ field
In this section, we study the baryon mass when the chiral symmetry is restored. In such a case, the vanishing of the chiral order parameter removes the dynamically generated constituent quark mass. Here, we treat the current quark mass is zero, so that the quark propagator is S q ( k ) = k̸ / k 2 S_q(k)=\not{k}/k^2 S q ( k ) = k / k 2 , while the diquark propagator is D ( P − k ) = 1 / [ ( P − k ) 2 − m d 2 ] D(P-k)=1/[(P-k)^2-m_d^2] D ( P − k ) = 1/ [( P − k ) 2 − m d 2 ] with the chiral invariant diquark mass m d m_d m d of Eq. (11) . Our aim is to construct the effective Lagrangian for the composite baryons at leading order in the derivative expansion: the mass terms are read off from the two-point functions at vanishing external momentum P P P , and the kinetic terms from their first order in P̸ \not{P} P .
The reformulated Lagrangian L q d ′ \mathcal{L}'_{qd} L q d ′ of Eq. (29) contains two distinct quark–diquark vertices, so that the one-loop quark–diquark two-point function becomes a 2 × 2 2\times2 2 × 2 matrix in vertex space as shown in Figure 2 (b)-(d). For a quark line of momentum k k k and a diquark line of momentum q d = P − k q_d=P-k q d = P − k , the two vertex factors are
Figure 2. Two-point functions of the composite baryon fields in the absence of the Σ \Sigma Σ field. Solid internal lines denote the massless quark propagator S_q(k) = \slashed{k}/k^2 and dashed lines the diquark propagator D ( P − k ) = 1 / [ ( P − k ) 2 − m d 2 ] D(P-k) = 1/[(P-k)^2-m_d^2] D ( P − k ) = 1/ [( P − k ) 2 − m d 2 ] . Filled squares denote the derivative (f 1 f_1 f 1 ) vertex of Eq. (31) with coupling C q d G q d C_{qd}G_{qd} C q d G q d , filled circles the trivial (f 2 f_2 f 2 ) vertex with unit coupling, and the open circle in (a) the tree-level B ˉ 1 B 2 \bar{B}_1B_2 B ˉ 1 B 2 mixing C q d C_{qd} C q d of Eq. (29) . Panels (b)–(d) give the loop components Π 11 \Pi_{11} Π 11 , Π 12 \Pi_{12} Π 12 , and Π 22 \Pi_{22} Π 22 of Eqs. (40) , (39) , and (38) : at vanishing external momentum the diagonal components (b) and (d) generate the kinetic normalizations Z 1 Z_1 Z 1 and Z 3 Z_3 Z 3 , while the off-diagonal component (c), combined with the tree-level mixing (a), yields the B ˉ 1 B 2 \bar{B}_1B_2 B ˉ 1 B 2 chiral-invariant mass term that becomes the physical baryon mass M B M_B M B of Eq. (45) after canonical normalization. The identical diagrams with C q d → C q d ( 2 ) C_{qd} \to C_{qd}^{(2)} C q d → C q d ( 2 ) , G q d → G q d ( 2 ) G_{qd} \to G_{qd}^{(2)} G q d → G q d ( 2 ) apply to the crossed (ψ \psi ψ ) sector.
f 1 = ̸ q d = P̸ − k̸ (coupling C q d G q d ) , (31) f_1=\not{q}_d=\not{P}-\not{k}\qquad\text{(coupling }C_{qd}G_{qd}\text{)}, \tag*{(31)} f 1 = q d = P − k (coupling C q d G q d ) , (31)
shown by filled square in Figure 2 , and
f 2 = 1 (coupling 1), (32) f_2 = 1 \quad\text{(coupling 1),} \tag*{(32)} f 2 = 1 (coupling 1), (32)
shown by filled circle, each accompanied by a chiral projector P ^ R , L = 1 2 ( 1 ± γ 5 ) \widehat{P}_{R,L} = \frac{1}{2}(1 \pm\gamma_5) P R , L = 2 1 ( 1 ± γ 5 ) . Since the massless quark propagator anticommutes with the projectors, P ^ χ k̸ = k̸ P ^ − χ \widehat{P}_{\chi}\not{k} = \not{k}\widehat{P}_{-\chi} P χ k = k P − χ , the projectors merely route the left and right components of the composite fields and can be stripped from the loop integral. The two-point function then reduces to
Π a b ( P ) = i ∫ d 4 k ( 2 π ) 4 f a g a k̸ f b g b k 2 [ ( P − k ) 2 − m d 2 ] , a , b = 1 , 2 , (33) \Pi_{ab}(P) = i \int\frac{d^4 k}{(2\pi)^4} \frac{f_a g_a \not{k} f_b g_b}{k^2\left[(P-k)^2-m_d^2\right]}, \qquad a,b=1,2, \tag*{(33)} Π ab ( P ) = i ∫ ( 2 π ) 4 d 4 k k 2 [ ( P − k ) 2 − m d 2 ] f a g a k f b g b , a , b = 1 , 2 , (33)
with g a , g b g_a,g_b g a , g b the couplings at the two vertices, g 1 = C q d G q d g_1 = C_{qd}G_{qd} g 1 = C q d G q d and g 2 = 1 g_2 = 1 g 2 = 1 . Only two loop functions appear in the evaluation: the diquark tadpole T d T_d T d and the vector function Π V \Pi_V Π V ,
T d = i ∫ d 4 k ( 2 π ) 4 1 ( P − k ) 2 − m d 2 (34) T_d = i \int\frac{d^4 k}{(2\pi)^4} \frac{1}{(P-k)^2-m_d^2} \tag*{(34)} T d = i ∫ ( 2 π ) 4 d 4 k ( P − k ) 2 − m d 2 1 (34)
P̸ Π V = i ∫ d 4 k ( 2 π ) 4 k̸ k 2 [ ( P − k ) 2 − m d 2 ] . (35) \not{P}\Pi_V = i \int\frac{d^4 k}{(2\pi)^4} \frac{\not{k}}{k^2\left[(P-k)^2-m_d^2\right]}. \tag*{(35)} P Π V = i ∫ ( 2 π ) 4 d 4 k k 2 [ ( P − k ) 2 − m d 2 ] k . (35)
By taking the rest frame P μ = ( W , 0 ⃗ ) P^\mu=(W,\vec{0}) P μ = ( W , 0 ) and introducing a three-momentum cutoff Λ \Lambda Λ , these are evaluated as
T d = 1 2 π 2 ∫ 0 Λ d k k 2 2 E d , (36) T_d = \frac{1}{2\pi^2}\int_0^\Lambda dk\,\frac{k^2}{2E_d}, \tag*{(36)} T d = 2 π 2 1 ∫ 0 Λ d k 2 E d k 2 , (36)
Π V = − 1 2 π 2 ∫ 0 Λ d k k 2 2 E d [ ( k + E d ) 2 − W 2 ] , (37) \Pi_V = -\frac{1}{2\pi^2}\int_0^\Lambda dk\,\frac{k^2}{2E_d\left[(k+E_d)^2-W^2\right]}, \tag*{(37)} Π V = − 2 π 2 1 ∫ 0 Λ d k 2 E d [ ( k + E d ) 2 − W 2 ] k 2 , (37)
with E d = k 2 + m d 2 E_d=\sqrt{k^2+m_d^2} E d = k 2 + m d 2 (see Appendix § 11 for detail.) Note that T d T_d T d is independent of the external momentum, and that Π V \Pi_V Π V is finite and negative at W 2 = 0 W^2=0 W 2 = 0 .
Π 22 \Pi_{22} Π 22 component. With the trivial vertex f 2 g 2 = 1 f_2g_2=1 f 2 g 2 = 1 on both sides, Eq. (33) directly gives
Π 22 = P̸ Π V . (38) \Pi_{22}=\not{P}\Pi_V. \tag*{(38)} Π 22 = P Π V . (38)
Π 12 = Π 21 \Pi_{12}=\Pi_{21} Π 12 = Π 21 component. With the derivative vertex f 1 f_1 f 1 on one side and the trivial vertex f 2 f_2 f 2 on the other, the numerator reduces via ̸ q d k̸ = ( P̸ − k̸ ) k̸ = P̸ k̸ − k 2 \not{q}_d\not{k}=(\not{P}-\not{k})\not{k}=\not{P}\not{k}-k^2 q d k = ( P − k ) k = P k − k 2 , so that
Π 12 = C q d G q d [ P̸ P̸ Π V − T d ] = C q d G q d ( P 2 Π V − T d ) . (39) \Pi_{12}=C_{qd}G_{qd}\left[\not{P}\not{P}\Pi_V-T_d\right]=C_{qd}G_{qd}\left(P^2\Pi_V-T_d\right). \tag*{(39)} Π 12 = C q d G q d [ P P Π V − T d ] = C q d G q d ( P 2 Π V − T d ) . (39)
Π 11 \Pi_{11} Π 11 component. With the derivative vertex f 1 f_1 f 1 on both sides, we write ̸ q d k̸ ̸ q d = ( ̸ q d k̸ ) P̸ − k 2 ̸ q d \not{q}_d\not{k}\not{q}_d=(\not{q}_d\not{k})\not{P}-k^2\not{q}_d q d k q d = ( q d k ) P − k 2 q d . The first piece reproduces the integrand of Π 12 \Pi_{12} Π 12 , while the second cancels the quark propagator and leaves i ∫ d 4 k / ( 2 π ) 4 ̸ q d / [ ̸ q d 2 − m d 2 ] i\int d^4k/(2\pi)^4\,\not{q}_d/[\not{q}_d^2-m_d^2] i ∫ d 4 k / ( 2 π ) 4 q d / [ q d 2 − m d 2 ] , which vanishes: in the rest frame the shift k 0 → k 0 + W k^0\to k^0+W k 0 → k 0 + W leaves the three-momentum cutoff untouched, and the resulting integrand is odd in the loop momentum. Hence
Π 11 = C q d G q d P̸ Π 12 = ( C q d G q d ) 2 P̸ ( P 2 Π V − T d ) . (40) \Pi_{11}=C_{qd}G_{qd}\not{P}\Pi_{12}=(C_{qd}G_{qd})^2\not{P}\left(P^2\Pi_V-T_d\right). \tag*{(40)} Π 11 = C q d G q d P Π 12 = ( C q d G q d ) 2 P ( P 2 Π V − T d ) . (40)
Effective Lagrangian and baryon mass
We now take the vanishing external momentum. Since Π V \Pi_V Π V is finite at P 2 = 0 P^2=0 P 2 = 0 , the terms P 2 Π V P^2\Pi_V P 2 Π V in Eqs. (39) and (40) drop out, and
\Pi_{11} \to-(C_{qd}G_{qd})^2 T_d \slashed{P}, \qquad\Pi_{12} \to-C_{qd}G_{qd}T_d, \qquad\Pi_{22} \to\Pi_V \slashed{P}. \tag*{(41)}
The diagonal components contain no mass term. This is dictated by chiral symmetry: with massless quarks a B ˉ 1 B 1 \bar{B}_1B_1 B ˉ 1 B 1 or B ˉ 2 B 2 \bar{B}_2B_2 B ˉ 2 B 2 mass term would require a chirality flip along the quark line, whereas the off-diagonal B ˉ 1 B 2 \bar{B}_1B_2 B ˉ 1 B 2 channel connects opposite Dirac chiralities through the γ μ \gamma^\mu γ μ of the B 2 B_2 B 2 interpolating field and is therefore chiral invariant. The loop thus generates the kinetic terms
Z_1 \bar{B}_1 i\slashed{\partial}B_1,\qquad Z_3 \bar{B}_2 i\slashed{\partial}B_2,\qquad Z_1=(C_{qd}G_{qd})^2T_d,\qquad Z_3=-\Pi_V>0, \tag*{(42)}
together with an off-diagonal mass mixing between B 1 B_1 B 1 and B 2 B_2 B 2 ,
Z 2 T d ( B ˉ 1 B 2 + B ˉ 2 B 1 ) , Z 2 = C q d G q d , (43) Z_2T_d(\bar{B}_1B_2+\bar{B}_2B_1),\qquad Z_2=C_{qd}G_{qd}, \tag*{(43)} Z 2 T d ( B ˉ 1 B 2 + B ˉ 2 B 1 ) , Z 2 = C q d G q d , (43)
where Π V \Pi_V Π V is understood at P 2 = 0 P^2=0 P 2 = 0 hereafter. Collecting these together with the tree-level mixing C q d C_{qd} C q d of Eq. (29) ,
\mathcal{L}_{\mathrm{eff}}=Z_1\bar{B}_1i\slashed{\partial}B_1+Z_3\bar{B}_2i\slashed{\partial}B_2-(C_{qd}+Z_2T_d)(\bar{B}_1B_2+\bar{B}_2B_1). \tag*{(44)}
Canonical normalization of the kinetic terms turns the off-diagonal mass into the physical baryon mass,
M B = C q d + Z 2 T d Z 1 Z 3 = C q d ( 1 + G q d T d ) ( C q d G q d ) 2 T d ( − Π V ) = 1 + G q d T d G q d − T d Π V , (45) M_B = \frac{C_{qd}+Z_2T_d}{\sqrt{Z_1Z_3}} = \frac{C_{qd}(1+G_{qd}T_d)}{\sqrt{(C_{qd}G_{qd})^2T_d(-\Pi_V)}} = \frac{1+G_{qd}T_d}{G_{qd}\sqrt{-T_d\Pi_V}}, \tag*{(45)} M B = Z 1 Z 3 C q d + Z 2 T d = ( C q d G q d ) 2 T d ( − Π V ) C q d ( 1 + G q d T d ) = G q d − T d Π V 1 + G q d T d , (45)
which is independent of the normalization C q d C_{qd} C q d and finite even though the quarks are massless. It should be noted that the baryon mass is not simply proportional to the diquark mass: M B M_B M B remains nonzero even in the m d → 0 m_d \to0 m d → 0 limit, where the loop functions are saturated by the cutoff, and it grows as 1 / G q d 1/G_{qd} 1/ G q d at weak coupling, signaling that a weakly bound composite is pushed up in mass. Rather, the mass scale is generated jointly by the invariant diquark mass m d m_d m d , entering through T d T_d T d and Π V \Pi_V Π V , and the gluon-mediated quark–diquark coupling G q d G_{qd} G q d , with the cutoff Λ \Lambda Λ setting the scale of the underlying interactions. The essential point is that all of these ingredients are independent of the chiral order parameter: the baryon mass (45) is therefore chiral invariant by construction, originating from the gluonic dynamics of the diquark sector rather than from the quark condensate. The identical analysis in the crossed sector, Eq. (30) , yields the same expression with G q d → G q d ( 2 ) G_{qd} \to G_{qd}^{(2)} G q d → G q d ( 2 ) and C q d → C q d ( 2 ) C_{qd} \to C_{qd}^{(2)} C q d → C q d ( 2 ) , these two masses are the chiral invariant masses m B 0 m_{B0} m B 0 and m ψ 0 m_{\psi0} m ψ 0 of the composite sectors introduced in the next section.
Effective Lagrangian including the Σ \Sigma Σ field
We now couple the quarks and diquarks to the meson field Σ \Sigma Σ . By keeping the leading order of Σ \Sigma Σ , the relevant Yukawa-type interactions are
L Σ = L q + L d , (46) \mathcal{L}_{\Sigma}=\mathcal{L}_q+\mathcal{L}_d, \tag*{(46)} L Σ = L q + L d , (46)
with
L q = − g q [ q ˉ L , a , i Σ α i q R a , α + q ˉ R , a , α ( Σ † ) i α q L a , i ] , (47) \mathcal{L}_q=-g_q\left[\bar{q}_{L,a,i}\Sigma^i_{\ \alpha}q_R^{a,\alpha}+\bar{q}_{R,a,\alpha}(\Sigma^\dagger)^\alpha_{\ i}q_L^{a,i}\right], \tag*{(47)} L q = − g q [ q ˉ L , a , i Σ α i q R a , α + q ˉ R , a , α ( Σ † ) i α q L a , i ] , (47)
L d = − Y d [ ( d L ) a , i Σ α i ( d R ) a , α + ( d R ) a , α ( Σ † ) i α ( d L ) a , i ] . (48) \mathcal{L}_d=-Y_d\left[(d_L)_{a,i}\Sigma^i_{\ \alpha}(d_R)^{a,\alpha}+(d_R)_{a,\alpha}(\Sigma^\dagger)^\alpha_{\ i}(d_L)^{a,i}\right]. \tag*{(48)} L d = − Y d [ ( d L ) a , i Σ α i ( d R ) a , α + ( d R ) a , α ( Σ † ) i α ( d L ) a , i ] . (48)
The quark term L q \mathcal{L}_q L q is the standard chiral invariant Yukawa coupling and it generates the constituent quark mass M q = g q f M_q=g_qf M q = g q f . The diquark term L d \mathcal{L}_d L d couples the scalar and pseudoscalar diquarks through Σ \Sigma Σ and, with ⟨ Σ ⟩ = diag ( α , α , γ ) \langle\Sigma\rangle=\operatorname{diag}(\alpha,\alpha,\gamma) ⟨ Σ ⟩ = diag ( α , α , γ ) , splits the 0 + 0^+ 0 + and 0 − 0^- 0 − diquark masses; it originates from a Kobayashi–Maskawa–’t Hooft-like six-quark interaction and therefore breaks the U(1)A _A A symmetry.
Effective Lagrangian and baryon mass matrix
Once Σ \Sigma Σ acquires its vacuum expectation value, the composites B 1 , 2 B_{1,2} B 1 , 2 mix with the crossed composites ψ 1 , 2 \psi_{1,2} ψ 1 , 2 through a chirality flip along either the quark or the diquark line. There are four such one-loop mixing diagrams, shown in Figure 3 –Figure 6 ; in each case the Σ \Sigma Σ insertion is marked by a crossed vertex.
Figure 3. One-loop diagram mixing B 1 L B_{1L} B 1 L and ψ 1 R \psi_{1R} ψ 1 R . Both quark–diquark vertices are of the derivative (f 1 f_1 f 1 ) type, carrying C q d G q d C_{qd}G_{qd} C q d G q d and C q d ( 2 ) G q d ( 2 ) C_{qd}^{(2)}G_{qd}^{(2)} C q d ( 2 ) G q d ( 2 ) , respectively. This diagram generates the induced interaction of Eq. (50) .
Figure 6. One-loop diagram mixing B 2 L \mathcal{B}_{2L} B 2 L and ψ 1 R \psi_{1R} ψ 1 R . The ψ 1 R \psi_{1R} ψ 1 R vertex is of the derivative (f 1 f_1 f 1 ) type and the B 2 L \mathcal{B}_{2L} B 2 L vertex is of the trivial (f 2 f_2 f 2 ) type. This diagram generates the induced interaction of Eq. (59) .
B 1 L B_{1L} B 1 L –ψ 1 R \psi_{1R} ψ 1 R mixing (Figure 3 ). The chirality flips along the quark line, q L → q R q_L \to q_R q L → q R . In the massless-quark limit,
Π ( 1 ) = i ∫ d 4 k ( 2 π ) 4 [ C q d ( 2 ) G q d ( 2 ) k̸ 1 + γ 5 2 ] k̸ k 2 [ C q d G q d k̸ 1 − γ 5 2 ] 1 k 2 − m d 2 = ( C q d ( 2 ) G q d ( 2 ) C q d G q d ) T d 1 − γ 5 2 , (49) \begin{aligned}
\Pi^{(1)} = i \int\frac{d^4 k}{(2\pi)^4} \left[C_{qd}^{(2)}G_{qd}^{(2)} \not{k} \frac{1+\gamma_5}{2}\right]\frac{\not{k}}{k^2}\left[C_{qd}G_{qd}\not{k}\frac{1-\gamma_5}{2}\right]\frac{1}{k^2-m_d^2} \\
&= \left(C_{qd}^{(2)}G_{qd}^{(2)}C_{qd}G_{qd}\right)T_d\frac{1-\gamma_5}{2}, \tag*{(49)}
\end{aligned} Π ( 1 ) = i ∫ ( 2 π ) 4 d 4 k [ C q d ( 2 ) G q d ( 2 ) k 2 1 + γ 5 ] k 2 k [ C q d G q d k 2 1 − γ 5 ] k 2 − m d 2 1 = ( C q d ( 2 ) G q d ( 2 ) C q d G q d ) T d 2 1 − γ 5 , (49)
which corresponds to the induced interaction
L i n t e r ( 1 ) = − g q ( Z 1 B ) 1 / 2 ( Z 1 ψ ) 1 / 2 ( ψ ˉ 1 R ) α j ( Σ † ) i α ( B 1 L ) j i , (50) \mathcal{L}_{\mathrm{inter}}^{(1)}=-g_q\left(Z_1^B\right)^{1/2}\left(Z_1^\psi\right)^{1/2}\left(\bar{\psi}_{1R}\right)_{\alpha}^{j}\left(\Sigma^\dagger\right)^\alpha_i\left(B_{1L}\right)^i_j, \tag*{(50)} L inter ( 1 ) = − g q ( Z 1 B ) 1/2 ( Z 1 ψ ) 1/2 ( ψ ˉ 1 R ) α j ( Σ † ) i α ( B 1 L ) j i , (50)
with
Z 1 B = ( C q d G q d ) 2 T d , Z 1 ψ = ( C q d ( 2 ) G q d ( 2 ) ) 2 T d . (51) Z_1^B=\left(C_{qd}G_{qd}\right)^2T_d,\qquad Z_1^\psi=\left(C_{qd}^{(2)}G_{qd}^{(2)}\right)^2T_d. \tag*{(51)} Z 1 B = ( C q d G q d ) 2 T d , Z 1 ψ = ( C q d ( 2 ) G q d ( 2 ) ) 2 T d . (51)
B 1 L B_{1L} B 1 L –ψ 2 R \psi_{2R} ψ 2 R mixing (Figure 4 ). Here the chirality flips along the diquark line, d L → d R d_L \to d_R d L → d R , so that
Figure 4. One-loop diagram mixing B 1 L B_{1L} B 1 L and ψ 2 R \psi_{2R} ψ 2 R . The B 1 L B_{1L} B 1 L vertex is of the derivative (f 1 f_1 f 1 ) type and the ψ 2 R \psi_{2R} ψ 2 R vertex is of the trivial (f 2 f_2 f 2 ) type. This diagram generates the induced interaction of Eq. (54) .
Π ( 2 ) = i ∫ d 4 k ( 2 π ) 4 1 − γ 5 2 k̸ k 2 ( C q d G q d k̸ 1 − γ 5 2 ) ( 1 k 2 − m d 2 ) 2 = ( C q d G q d ) L d 1 − γ 5 2 , (52) \Pi^{(2)}=i\int\frac{d^4k}{(2\pi)^4}\frac{1-\gamma_5}{2}\frac{\not{k}}{k^2}\left(C_{qd}G_{qd}\not{k}\frac{1-\gamma_5}{2}\right)\left(\frac{1}{k^2-m_d^2}\right)^2=(C_{qd}G_{qd})L_d\frac{1-\gamma_5}{2}, \tag*{(52)} Π ( 2 ) = i ∫ ( 2 π ) 4 d 4 k 2 1 − γ 5 k 2 k ( C q d G q d k 2 1 − γ 5 ) ( k 2 − m d 2 1 ) 2 = ( C q d G q d ) L d 2 1 − γ 5 , (52)
with
L d = i ∫ d 4 k ( 2 π ) 4 ( 1 k 2 − m d 2 ) 2 . (53) L_d=i\int\frac{d^4k}{(2\pi)^4}\left(\frac{1}{k^2-m_d^2}\right)^2. \tag*{(53)} L d = i ∫ ( 2 π ) 4 d 4 k ( k 2 − m d 2 1 ) 2 . (53)
The induced interaction is
L i n t e r ( 2 ) = − Y d ( C q d G q d ) L d ( ψ ˉ 2 R ) i α Σ α i ( B 1 L ) i j . (54) \mathcal{L}_{\mathrm{inter}}^{(2)}=-Y_d(C_{qd}G_{qd})L_d\left(\bar{\psi}_{2R}\right)^\alpha_i\Sigma^i_\alpha\left(B_{1L}\right)^j_i. \tag*{(54)} L inter ( 2 ) = − Y d ( C q d G q d ) L d ( ψ ˉ 2 R ) i α Σ α i ( B 1 L ) i j . (54)
B 2 L B_{2L} B 2 L –ψ 2 R \psi_{2R} ψ 2 R mixing (Figure 5 ). The chirality flips along the quark line, q R → q L q_R \to q_L q R → q L , and both vertices are of the trivial (f 2 f_2 f 2 ) type:
Figure 5. One-loop diagram mixing B 2 L B_{2L} B 2 L and ψ 2 R \psi_{2R} ψ 2 R . Both quark–diquark vertices are of the trivial (f 2 f_2 f 2 ) type. This diagram generates the induced interaction of Eq. (57) .
Π ( 3 ) = i ∫ d 4 k ( 2 π ) 4 1 − γ 5 2 k̸ k 2 k̸ k 2 1 − γ 5 2 1 k 2 − m d 2 = T d − T q m d 2 1 − γ 5 2 , (55) \Pi^{(3)}=i\int\frac{d^4k}{(2\pi)^4}\frac{1-\gamma_5}{2}\frac{\not{k}}{k^2}\frac{\not{k}}{k^2}\frac{1-\gamma_5}{2}\frac{1}{k^2-m_d^2}=\frac{T_d-T_q}{m_d^2}\frac{1-\gamma_5}{2}, \tag*{(55)} Π ( 3 ) = i ∫ ( 2 π ) 4 d 4 k 2 1 − γ 5 k 2 k k 2 k 2 1 − γ 5 k 2 − m d 2 1 = m d 2 T d − T q 2 1 − γ 5 , (55)
where
T q = 1 4 π 2 ∫ 0 Λ d k k = Λ 2 8 π 2 . (56) T_q = \frac{1}{4\pi^2}\int_0^\Lambda dk k = \frac{\Lambda^2}{8\pi^2}. \tag*{(56)} T q = 4 π 2 1 ∫ 0 Λ d k k = 8 π 2 Λ 2 . (56)
The induced interaction is
L i n t e r ( 3 ) = − g q T d − T q m d 2 ( ψ ˉ 2 R ) i β Σ α i ( B 2 L ) β α . (57) \mathcal{L}_{\mathrm{inter}}^{(3)}=-g_q\frac{T_d-T_q}{m_d^2}(\bar{\psi}_{2R})^\beta_i\Sigma^i_{\alpha}(B_{2L})^\alpha_{\beta}. \tag*{(57)} L inter ( 3 ) = − g q m d 2 T d − T q ( ψ ˉ 2 R ) i β Σ α i ( B 2 L ) β α . (57)
B 2 L B_{2L} B 2 L –ψ 1 R \psi_{1R} ψ 1 R mixing (Figure 6 ). The chirality flips along the diquark line, d R → d L d_R \to d_L d R → d L , giving
\Pi^{(4)}=i\int\frac{d^4k}{(2\pi)^4}\left(C_{qd}^{(2)}G_{qd}^{(2)}\slashed{k}\frac{1+\gamma_5}{2}\right)\frac{\slashed{k}}{k^2}\frac{1-\gamma_5}{2}\left(\frac{1}{k^2-m_d^2}\right)^2=C_{qd}^{(2)}G_{qd}^{(2)}L_d\frac{1-\gamma_5}{2}. \tag*{(58)}
The induced interaction is
L i n t e r ( 4 ) = − Y d C q d ( 2 ) G q d ( 2 ) L d ( ψ ˉ 1 R ) α j ( Σ ) j β ( B 2 L ) β α . (59) \mathcal{L}_{\mathrm{inter}}^{(4)}=-Y_d C_{qd}^{(2)}G_{qd}^{(2)}L_d(\bar{\psi}_{1R})^j_\alpha(\Sigma)^\beta_j(B_{2L})^\alpha_\beta. \tag*{(59)} L inter ( 4 ) = − Y d C q d ( 2 ) G q d ( 2 ) L d ( ψ ˉ 1 R ) α j ( Σ ) j β ( B 2 L ) β α . (59)
We note that the two diagrams in which the chirality flips along the diquark line, Eqs. (54) and (59) , share the same coefficient form ∝ Y d L d \propto Y_dL_d ∝ Y d L d , whereas the two diagrams flipping the chirality along the quark line, Eqs. (50) and (57) , carry different coefficients (∝ g q T d \propto g_qT_d ∝ g q T d and ∝ g q ( T d − T q ) / m d 2 \propto g_q(T_d-T_q)/m_d^2 ∝ g q ( T d − T q ) / m d 2 , respectively). After canonical normalization of the B B B and ψ \psi ψ fields the former two reduce to a single mixing coupling, while the latter two yield two distinct couplings.
Collecting the tree-level mixings and the induced terms, the mass matrix for the nucleons in the basis ( B 1 , B 2 , ψ 1 , ψ 2 ) (\mathcal{B}_1,\mathcal{B}_2,\psi_1,\psi_2) ( B 1 , B 2 , ψ 1 , ψ 2 ) takes the form
( B ‾ 1 B ‾ 2 ψ ‾ 1 ψ ‾ 2 ) ( 0 m B 0 g 1 α y 1 γ m B 0 0 y 1 γ g 2 α g 1 α y 1 γ 0 m ψ 0 y 1 γ g 2 α m ψ 0 0 ) ( B 1 B 2 ψ 1 ψ 2 ) , (60) \left(\overline{\mathcal{B}}_1\ \overline{\mathcal{B}}_2\ \overline{\psi}_1\ \overline{\psi}_2\right)
\begin{pmatrix}
0 & m_{\mathcal{B}0} & g_1\alpha& y_1\gamma\\
m_{\mathcal{B}0} & 0 & y_1\gamma& g_2\alpha\\
g_1\alpha& y_1\gamma& 0 & m_{\psi0} \\
y_1\gamma& g_2\alpha& m_{\psi0} & 0
\end{pmatrix}
\begin{pmatrix}
\mathcal{B}_1 \\
\mathcal{B}_2 \\
\psi_1 \\
\psi_2
\end{pmatrix}, \tag*{(60)} ( B 1 B 2 ψ 1 ψ 2 ) 0 m B 0 g 1 α y 1 γ m B 0 0 y 1 γ g 2 α g 1 α y 1 γ 0 m ψ 0 y 1 γ g 2 α m ψ 0 0 B 1 B 2 ψ 1 ψ 2 , (60)
where m B 0 m_{\mathcal{B}0} m B 0 and m ψ 0 m_{\psi0} m ψ 0 are the chiral invariant masses of the B \mathcal{B} B and ψ \psi ψ sectors, α \alpha α and γ \gamma γ are the condensate parameters associated with the light (f π f_\pi f π ) and strange (f s f_s f s ) components of ⟨ Σ ⟩ \langle\Sigma\rangle ⟨ Σ ⟩ , and g 1 , 2 g_{1,2} g 1 , 2 , y 1 y_1 y 1 are the normalized mixing couplings induced by the diagrams above.
A complete evaluation must also include the two-loop diagrams in which the chirality flips on both the quark and the diquark lines, connecting ψ 1 L , 2 L \psi_{1L,2L} ψ 1 L , 2 L to ψ 1 R , 2 R \psi_{1R,2R} ψ 1 R , 2 R and thereby generating diagonal entries in the ψ \psi ψ sector. The relevant chiral invariant quark–quark–diquark Yukawa coupling is
L q q d = G d [ ( d L † ) a , i ϵ a b c ϵ i j k ( q L T ) b , j C ( q L ) c , k + ( d R † ) a , α ϵ a b c ϵ α β γ ( q R T ) b , β C ( q R ) c , γ + h . c . ] . (61) \mathcal{L}_{qqd}=G_d\left[(d_L^\dagger)^{a,i}\epsilon_{abc}\epsilon_{ijk}(q_L^T)^{b,j}C(q_L)^{c,k}+(d_R^\dagger)^{a,\alpha}\epsilon_{abc}\epsilon_{\alpha\beta\gamma}(q_R^T)^{b,\beta}C(q_R)^{c,\gamma}+\mathrm{h.c.}\right]. \tag*{(61)} L q q d = G d [ ( d L † ) a , i ϵ ab c ϵ ij k ( q L T ) b , j C ( q L ) c , k + ( d R † ) a , α ϵ ab c ϵ α β γ ( q R T ) b , β C ( q R ) c , γ + h.c. ] . (61)
The antisymmetric color and flavor contractions match the quantum numbers of the diquark fields of § 2 , and the same argument as in Appendix § 10 shows that parity and charge-conjugation invariance force G d G_d G d to be real. Since L q q d \mathcal{L}_{qqd} L q q d is chiral invariant, its Σ \Sigma Σ -independent loop effects only renormalize the diquark mass and the wave-function normalizations already parametrized in Sec. 4, and we absorb them accordingly; its role here is to enable the double chirality flip. A representative diagram is shown in Figure 7 : the diquark dissolves through one G d G_d G d vertex, the Σ \Sigma Σ insertion flips one of the constituent quarks with strength g q g_q g q , and the flipped quark recombines with the spectator into the opposite-chirality diquark through a second G d G_d G d vertex. With derivative (f 1 f_1 f 1 ) composite vertices at both ends the diagram generates the ψ 1 \psi_1 ψ 1 -sector diagonal entry, and with trivial (f 2 f_2 f 2 ) vertices the ψ 2 \psi_2 ψ 2 -sector one; rather than evaluating the two-loop integrals explicitly, we parametrize the canonically normalized strengths of these entries by the couplings g 1 a g_1^a g 1 a and g 2 a g_2^a g 2 a , treated as independent parameters in the fit below. In addition, mixed diagrams with one derivative (f 1 f_1 f 1 ) and one trivial (f 2 f_2 f 2 ) composite vertex connect ψ 1 \psi_1 ψ 1 and ψ 2 \psi_2 ψ 2 ; as discussed in § 8 , they generate no mass term but only a chiral invariant kinetic mixing, which we neglect as a higher-order correction.
Figure 7. Two-loop diagram generating the diagonal entry g 1 a g_1^a g 1 a in the ψ \psi ψ sector (ψ 1 L \psi_{1L} ψ 1 L –ψ 1 R \psi_{1R} ψ 1 R mixing). The diquark is resolved into its constituent quarks, and a single Σ \Sigma Σ insertion (crossed vertex), combined with the quark rearrangement, flips the chirality of both the quark and the diquark lines, q L d R → q R d L q_L d_R \to q_R d_L q L d R → q R d L . The same topology with the derivative-type (f 1 f_1 f 1 ) vertices yields g 2 a g_2^a g 2 a .
Assembling the results obtained so far, the total effective Lagrangian for the normalized composite baryon fields reads
L e f f = ∑ n = 1 , 2 [ B ‾ n i ∂̸ B n + ψ ‾ n i ∂̸ ψ n ] − m B 0 ( B ‾ 1 B 2 + B ‾ 2 B 1 ) − m ψ 0 ( ψ ‾ 1 ψ 2 + ψ ‾ 2 ψ 1 ) − g 1 s ( ψ ‾ 1 R ) α j ( Σ † ) i α ( B 1 L ) j i − g 2 s ( ψ ‾ 2 R ) i β Σ α i ( B 2 L ) β α − y 1 [ ( ψ ‾ 2 R ) i α Σ α i ( B 1 L ) j j + ( ψ ‾ 1 R ) α j Σ j β ( B 2 L ) β α ] − g 1 a ( ψ ‾ 1 R ) α i ( Σ ) β j ( ψ 1 L ) γ k ϵ i j k ϵ α β γ − g 2 a ( ψ ‾ 2 L ) α i ( Σ ) β j ( ψ 2 R ) γ k ϵ i j k ϵ α β γ + ( L ↔ R , h . c . ) , (62) \begin{aligned}
\mathcal{L}_{\mathrm{eff}}={}&\sum_{n=1,2}\left[\overline{B}_n i\not\partial B_n+\overline{\psi}_n i\not\partial\psi_n\right]-m_{B0}\left(\overline{B}_1B_2+\overline{B}_2B_1\right)-m_{\psi0}\left(\overline{\psi}_1\psi_2+\overline{\psi}_2\psi_1\right)\\
&-g_1^s(\overline{\psi}_{1R})^j_{\alpha}(\Sigma^\dagger)^\alpha_i(B_{1L})^i_j-g_2^s(\overline{\psi}_{2R})^\beta_i\Sigma^i_{\alpha}(B_{2L})^\alpha_{\beta}\\
&-y_1\left[(\overline{\psi}_{2R})^\alpha_i\Sigma^i_{\alpha}(B_{1L})^j_j+(\overline{\psi}_{1R})^j_{\alpha}\Sigma^\beta_j(B_{2L})^\alpha_{\beta}\right]\\
&-g_1^a(\overline{\psi}_{1R})^i_{\alpha}(\Sigma)^j_{\beta}(\psi_{1L})^k_{\gamma}\epsilon_{ijk}\epsilon^{\alpha\beta\gamma}-g_2^a(\overline{\psi}_{2L})^i_{\alpha}(\Sigma)^j_{\beta}(\psi_{2R})^k_{\gamma}\epsilon_{ijk}\epsilon^{\alpha\beta\gamma}\\
&+\left(L\leftrightarrow R,\ \mathrm{h.c.}\right), \tag*{(62)}
\end{aligned} L eff = n = 1 , 2 ∑ [ B n i ∂ B n + ψ n i ∂ ψ n ] − m B 0 ( B 1 B 2 + B 2 B 1 ) − m ψ 0 ( ψ 1 ψ 2 + ψ 2 ψ 1 ) − g 1 s ( ψ 1 R ) α j ( Σ † ) i α ( B 1 L ) j i − g 2 s ( ψ 2 R ) i β Σ α i ( B 2 L ) β α − y 1 [ ( ψ 2 R ) i α Σ α i ( B 1 L ) j j + ( ψ 1 R ) α j Σ j β ( B 2 L ) β α ] − g 1 a ( ψ 1 R ) α i ( Σ ) β j ( ψ 1 L ) γ k ϵ ij k ϵ α β γ − g 2 a ( ψ 2 L ) α i ( Σ ) β j ( ψ 2 R ) γ k ϵ ij k ϵ α β γ + ( L ↔ R , h.c. ) , (62)
where the terms proportional to g 1 a g_1^a g 1 a or g 2 a g_2^a g 2 a are obtained from two-loop contribution as shown above. Distinguishing these diagonal couplings from the off-diagonal mixing couplings (g s g^s g s ), and resolving the flavor (strangeness) content of each octet member through the diagonal structure of ⟨ Σ ⟩ \langle\Sigma\rangle ⟨ Σ ⟩ , the mass matrices in the N N N , Σ \Sigma Σ , and Ξ \Xi Ξ sectors read
M N = ( 0 m B 0 g 1 s α y 1 γ m B 0 0 y 1 γ g 2 s α g 1 s α y 1 γ g 1 a α m ψ 0 y 1 γ g 2 s α m ψ 0 g 2 a α ) (63) \mathcal{M}_N =
\begin{pmatrix}
0 & m_{B0} & g_1^s\alpha& y_1\gamma\\
m_{B0} & 0 & y_1\gamma& g_2^s\alpha\\
g_1^s\alpha& y_1\gamma& g_1^a\alpha& m_{\psi0} \\
y_1\gamma& g_2^s\alpha& m_{\psi0} & g_2^a\alpha
\end{pmatrix} \tag*{(63)} M N = 0 m B 0 g 1 s α y 1 γ m B 0 0 y 1 γ g 2 s α g 1 s α y 1 γ g 1 a α m ψ 0 y 1 γ g 2 s α m ψ 0 g 2 a α (63)
M Σ = ( 0 m B 0 g 1 s α y 1 α m B 0 0 y 1 α g 2 s α g 1 s α y 1 α g 1 a γ m ψ 0 y 1 α g 2 s α m ψ 0 g 2 a γ ) (64) \mathcal{M}_{\Sigma} =
\begin{pmatrix}
0 & m_{B0} & g_1^s\alpha& y_1\alpha\\
m_{B0} & 0 & y_1\alpha& g_2^s\alpha\\
g_1^s\alpha& y_1\alpha& g_1^a\gamma& m_{\psi0} \\
y_1\alpha& g_2^s\alpha& m_{\psi0} & g_2^a\gamma
\end{pmatrix} \tag*{(64)} M Σ = 0 m B 0 g 1 s α y 1 α m B 0 0 y 1 α g 2 s α g 1 s α y 1 α g 1 a γ m ψ 0 y 1 α g 2 s α m ψ 0 g 2 a γ (64)
M Ξ = ( 0 m B 0 g 1 s γ y 1 α m B 0 0 y 1 α g 2 s γ g 1 s γ y 1 α g 1 a α m ψ 0 y 1 α g 2 s γ m ψ 0 g 2 a α ) (65) \mathcal{M}_{\Xi} =
\begin{pmatrix}
0 & m_{B0} & g_1^s\gamma& y_1\alpha\\
m_{B0} & 0 & y_1\alpha& g_2^s\gamma\\
g_1^s\gamma& y_1\alpha& g_1^a\alpha& m_{\psi0} \\
y_1\alpha& g_2^s\gamma& m_{\psi0} & g_2^a\alpha
\end{pmatrix} \tag*{(65)} M Ξ = 0 m B 0 g 1 s γ y 1 α m B 0 0 y 1 α g 2 s γ g 1 s γ y 1 α g 1 a α m ψ 0 y 1 α g 2 s γ m ψ 0 g 2 a α (65)
We should note that, since we include the flavor-symmetry breaking contribution generated by the difference between α \alpha α and γ \gamma γ , the iso-singlet members (Λ \Lambda Λ baryons) of the octet baryons mix with the flavor-singlet baryons, which are omitted in the present analysis. We therefore concentrate on the mass spectra of the nucleons, Σ \Sigma Σ , and Ξ \Xi Ξ baryons in the following analysis.
Explicit chiral symmetry breaking
Explicit chiral symmetry breaking from the current quark masses is incorporated through the quark Yukawa term (47) with bare quark mass matrix,
L q = − g q [ q ˉ L , a , i Σ j i q R a , j + h . c . ] − [ q ˉ L , a , i ( M q ) j i q R a , j + q ˉ R , a , j ( M q † ) i j q L a , i ] (66) \mathcal{L}_q=-g_q\left[\bar{q}_{L,a,i}\Sigma^i_j q_R^{a,j}+\mathrm{h.c.}\right]-\left[\bar{q}_{L,a,i}(\mathcal{M}_q)^i_jq_R^{a,j}+\bar{q}_{R,a,j}(\mathcal{M}_q^\dagger)^j_iq_L^{a,i}\right] \tag*{(66)} L q = − g q [ q ˉ L , a , i Σ j i q R a , j + h.c. ] − [ q ˉ L , a , i ( M q ) j i q R a , j + q ˉ R , a , j ( M q † ) i j q L a , i ] (66)
with
M q = diag ( m u , m d , m s ) (67) \mathcal{M}_q=\operatorname{diag}(m_u,m_d,m_s) \tag*{(67)} M q = diag ( m u , m d , m s ) (67)
where m u m_u m u , m d m_d m d , m s m_s m s are the current quark masses. In the following analysis, we take the isospin limit m d = m u m_d=m_u m d = m u and use m u m_u m u only. Equation (66) is rewritten as
L q = − g q [ q ˉ L , a , i ( Σ + c 1 M q ) j i q R a , j + q ˉ R , a , j ( Σ † + c 1 M q † ) i j q L a , i ] , c 1 = 1 g q (68) \mathcal{L}_q=-g_q\left[\bar{q}_{L,a,i}(\Sigma+c_1\mathcal{M}_q)^i_jq_R^{a,j}+\bar{q}_{R,a,j}(\Sigma^\dagger+c_1\mathcal{M}_q^\dagger)^j_iq_L^{a,i}\right],\qquad c_1=\frac{1}{g_q} \tag*{(68)} L q = − g q [ q ˉ L , a , i ( Σ + c 1 M q ) j i q R a , j + q ˉ R , a , j ( Σ † + c 1 M q † ) i j q L a , i ] , c 1 = g q 1 (68)
so that, in the mass matrices (63) –(65) , the condensate parameters entering the g 1 s g_1^s g 1 s -, g 2 s g_2^s g 2 s -, g 1 a g_1^a g 1 a -, and g 2 a g_2^a g 2 a -terms are replaced by
α → α + c 1 m u , γ → γ + c 1 m s . (69) \alpha\to\alpha+c_1m_u,\qquad\gamma\to\gamma+c_1m_s. \tag*{(69)} α → α + c 1 m u , γ → γ + c 1 m s . (69)
An analogous bare-mass term in L d \mathcal{L}_d L d produces, in the y 1 y_1 y 1 entries,1
α → α + c 2 m u , γ → γ + c 2 m s . (70) \alpha\to\alpha+c_2m_u,\qquad\gamma\to\gamma+c_2m_s. \tag*{(70)} α → α + c 2 m u , γ → γ + c 2 m s . (70)
Before proceeding, we point out an exact relation among the mass matrices (63) –(65) . The chiral invariant masses m B 0 m_{B0} m B 0 , m ψ 0 m_{\psi0} m ψ 0 and the mixing couplings g 1 , 2 s g^{s}_{1,2} g 1 , 2 s , g 1 , 2 a g^{a}_{1,2} g 1 , 2 a , y 1 y_1 y 1 enter only off-diagonal entries, so the trace of each matrix is only the function of g 1 , 2 a g^{a}_{1,2} g 1 , 2 a , yielding
Tr M N = Tr M Ξ = ( g 1 a + g 2 a ) α (71) \operatorname{Tr}\mathcal{M}_N=\operatorname{Tr}\mathcal{M}_{\Xi}=(g^a_1+g^a_2)\alpha \tag*{(71)} Tr M N = Tr M Ξ = ( g 1 a + g 2 a ) α (71)
and the equality Tr M N = Tr M Ξ \operatorname{Tr}\mathcal{M}_N=\operatorname{Tr}\mathcal{M}_{\Xi} Tr M N = Tr M Ξ holds for arbitrary values of the condensates; it is unaffected by the explicit breaking introduced in Sec. 5.2, which replaces α → α + c m u \alpha\to\alpha+c m_u α → α + c m u and γ → γ + c m s \gamma\to\gamma+c m_s γ → γ + c m s in Eq. (71) .
Mass spectra and sigma terms
Having fixed the structure of the baryon mass matrices in Sec. 5, we now define the nucleon sigma terms used in our analysis and confront the model with the octet baryon spectrum. We first derive the sigma terms from the quark-mass dependence of the nucleon mass matrix, and then describe the fitting procedure and the corresponding numerical results.
Nucleon sigma terms
The nucleon sigma terms are defined by
σ s N = m s ⟨ N ∣ s ˉ s ∣ N ⟩ = m s ∂ m N ∂ m s , σ π N = m u ⟨ N ∣ u ˉ u ∣ N ⟩ = m u ∂ m N ∂ m u (72) \sigma_{sN}=m_s\langle N|\bar{s}s|N\rangle=m_s\frac{\partial m_N}{\partial m_s},\qquad\sigma_{\pi N}=m_u\langle N|\bar{u}u|N\rangle=m_u\frac{\partial m_N}{\partial m_u} \tag*{(72)} σ s N = m s ⟨ N ∣ s ˉ s ∣ N ⟩ = m s ∂ m s ∂ m N , σ π N = m u ⟨ N ∣ u ˉ u ∣ N ⟩ = m u ∂ m u ∂ m N (72)
and are evaluated in the present model using the Feynman–Hellmann theorem. The nucleon mass matrix, obtained from Eq. (63) together with the explicit-breaking replacements of Sec. 5.2, is expressed as
M N = ( 0 m B 0 g 1 s ( α + c 1 m u ) y 1 ( γ + c 2 m s ) m B 0 0 y 1 ( γ + c 2 m s ) g 2 s ( α + c 1 m u ) g 1 s ( α + c 1 m u ) y 1 ( γ + c 2 m s ) g 1 a ( α + c 1 m u ) m ψ 0 y 1 ( γ + c 2 m s ) g 2 s ( α + c 1 m u ) m ψ 0 g 2 a ( α + c 1 m u ) ) . (73) \mathcal{M}_N=
\begin{pmatrix}
0 & m_{B0} & g^s_1(\alpha+c_1m_u) & y_1(\gamma+c_2m_s)\\
m_{B0} & 0 & y_1(\gamma+c_2m_s) & g^s_2(\alpha+c_1m_u)\\
g^s_1(\alpha+c_1m_u) & y_1(\gamma+c_2m_s) & g^a_1(\alpha+c_1m_u) & m_{\psi0}\\
y_1(\gamma+c_2m_s) & g^s_2(\alpha+c_1m_u) & m_{\psi0} & g^a_2(\alpha+c_1m_u)
\end{pmatrix}. \tag*{(73)} M N = 0 m B 0 g 1 s ( α + c 1 m u ) y 1 ( γ + c 2 m s ) m B 0 0 y 1 ( γ + c 2 m s ) g 2 s ( α + c 1 m u ) g 1 s ( α + c 1 m u ) y 1 ( γ + c 2 m s ) g 1 a ( α + c 1 m u ) m ψ 0 y 1 ( γ + c 2 m s ) g 2 s ( α + c 1 m u ) m ψ 0 g 2 a ( α + c 1 m u ) . (73)
It is diagonalized by an orthogonal matrix O ^ \hat{O} O ^ according to
O ^ M N O ^ T = diag ( m N , m N ( 1440 ) , m N ( 1535 ) , m N ( 1650 ) ) . (74) \hat{O}\mathcal{M}_N\hat{O}^{T}=\operatorname{diag}\left(m_N,m_{N(1440)},m_{N(1535)},m_{N(1650)}\right). \tag*{(74)} O ^ M N O ^ T = diag ( m N , m N ( 1440 ) , m N ( 1535 ) , m N ( 1650 ) ) . (74)
The Feynman–Hellmann theorem then gives
σ s N = [ O ^ m s ∂ M N ∂ m s O ^ T ] 11 , σ π N = [ O ^ m u ∂ M N ∂ m u O ^ T ] 11 (75) \sigma_{sN}=\left[\hat{O}m_s\frac{\partial\mathcal{M}_N}{\partial m_s}\hat{O}^{T}\right]_{11},\qquad\sigma_{\pi N}=\left[\hat{O}m_u\frac{\partial\mathcal{M}_N}{\partial m_u}\hat{O}^{T}\right]_{11} \tag*{(75)} σ s N = [ O ^ m s ∂ m s ∂ M N O ^ T ] 11 , σ π N = [ O ^ m u ∂ m u ∂ M N O ^ T ] 11 (75)
where [ X ] 11 [X]_{11} [ X ] 11 denotes the ( 1 , 1 ) (1,1) ( 1 , 1 ) component associated with the nucleon ground state. To evaluate the derivatives, we assume that the couplings g 1 , 2 s g^s_{1,2} g 1 , 2 s , g 1 , 2 a g^a_{1,2} g 1 , 2 a , y 1 y_1 y 1 , c 1 c_1 c 1 , and c 2 c_2 c 2 are independent of the current quark masses, and that the quark-mass dependence of the condensate sector is carried entirely by α \alpha α and γ \gamma γ . We take the SU(3)-breaking splitting between the two condensates to be linear in m s − m u m_s-m_u m s − m u ,
γ − α = ( m s − m u ) C (76) \gamma-\alpha=(m_s-m_u)C \tag*{(76)} γ − α = ( m s − m u ) C (76)
where C C C is a constant. The chiral-limit value of α \alpha α is fixed by the pion decay constant in the chiral limit [47 ] ,
α = f 0 + m u C u , f 0 = 88 M e V , (77) \alpha= f_0 + m_u C_u, \qquad f_0 = 88\ \mathrm{MeV}, \tag*{(77)} α = f 0 + m u C u , f 0 = 88 MeV , (77)
where C u C_u C u is another constant.
Strangeness sigma term. At fixed m u m_u m u , only the entries proportional to γ + c 2 m s \gamma+c_2m_s γ + c 2 m s depend on m s m_s m s . Equation (76) gives
m s ∂ γ ∂ m s = m s ∂ ( γ − α ) ∂ m s = m s m s − m u ( γ − α ) , (78) m_s\frac{\partial\gamma}{\partial m_s}
= m_s\frac{\partial(\gamma-\alpha)}{\partial m_s}
= \frac{m_s}{m_s-m_u}(\gamma-\alpha), \tag*{(78)} m s ∂ m s ∂ γ = m s ∂ m s ∂ ( γ − α ) = m s − m u m s ( γ − α ) , (78)
and hence
m s ∂ M N ∂ m s = y 1 γ ~ ( 0 0 0 1 0 0 1 0 0 1 0 0 1 0 0 0 ) , m_s\frac{\partial\mathcal{M}_N}{\partial m_s}
= y_1\widetilde{\gamma}
\begin{pmatrix}
0 & 0 & 0 & 1 \\
0 & 0 & 1 & 0 \\
0 & 1 & 0 & 0 \\
1 & 0 & 0 & 0
\end{pmatrix}, m s ∂ m s ∂ M N = y 1 γ 0 0 0 1 0 0 1 0 0 1 0 0 1 0 0 0 ,
γ ~ = m s m s − m u ( γ − α ) + c 2 m s . (79) \widetilde{\gamma}
= \frac{m_s}{m_s-m_u}(\gamma-\alpha)+c_2m_s. \tag*{(79)} γ = m s − m u m s ( γ − α ) + c 2 m s . (79)
Pion–nucleon sigma term. At fixed m s m_s m s , both the explicit terms proportional to c 1 m u c_1m_u c 1 m u and the implicit quark-mass dependence of the condensates contribute. From Eq. (77) , one obtains
m u ∂ α ∂ m u = α − f 0 . (80) m_u\frac{\partial\alpha}{\partial m_u}=\alpha-f_0. \tag*{(80)} m u ∂ m u ∂ α = α − f 0 . (80)
Combining this relation with Eq. (76) gives
m u ∂ γ ∂ m u = ( α − f 0 ) − m u m s − m u ( γ − α ) . (81) m_u\frac{\partial\gamma}{\partial m_u}
=(\alpha-f_0)-\frac{m_u}{m_s-m_u}(\gamma-\alpha). \tag*{(81)} m u ∂ m u ∂ γ = ( α − f 0 ) − m s − m u m u ( γ − α ) . (81)
Therefore,
m u ∂ M N ∂ m u = ( 0 0 g 1 s α ~ y 1 γ ^ 0 0 y 1 γ ^ g 2 s α ~ g 1 s α ~ y 1 γ ^ g 1 q α ~ 0 y 1 γ ^ g 2 s α ~ 0 g 2 q α ~ ) , (82) m_u\frac{\partial\mathcal{M}_N}{\partial m_u}
=
\begin{pmatrix}
0 & 0 & g_1^s\widetilde{\alpha} & y_1\widehat{\gamma} \\
0 & 0 & y_1\widehat{\gamma} & g_2^s\widetilde{\alpha} \\
g_1^s\widetilde{\alpha} & y_1\widehat{\gamma} & g_1^q\widetilde{\alpha} & 0 \\
y_1\widehat{\gamma} & g_2^s\widetilde{\alpha} & 0 & g_2^q\widetilde{\alpha}
\end{pmatrix}, \tag*{(82)} m u ∂ m u ∂ M N = 0 0 g 1 s α y 1 γ 0 0 y 1 γ g 2 s α g 1 s α y 1 γ g 1 q α 0 y 1 γ g 2 s α 0 g 2 q α , (82)
where
α ~ = ( α − f 0 ) + c 1 m u , γ ^ = ( α − f 0 ) − m u m s − m u ( γ − α ) . (83) \widetilde{\alpha}=(\alpha-f_0)+c_1m_u,\qquad
\widehat{\gamma}=(\alpha-f_0)-\frac{m_u}{m_s-m_u}(\gamma-\alpha). \tag*{(83)} α = ( α − f 0 ) + c 1 m u , γ = ( α − f 0 ) − m s − m u m u ( γ − α ) . (83)
The terms proportional to α − f 0 \alpha-f_0 α − f 0 describe the response of the chiral order parameter, equivalently the pion decay constant, to m u m_u m u . They thus represent the meson-sector, or condensate, contribution to σ π N \sigma_{\pi N} σ π N . The numerical importance of this contribution will be discussed below.
Fitting procedure
The model contains seven coupling constants,
g 1 s , g 2 s , g 1 a , g 2 a , y 1 , c 1 , c 2 , (84) g_1^s,\quad g_2^s,\quad g_1^a,\quad g_2^a,\quad y_1,\quad c_1,\quad c_2, \tag*{(84)} g 1 s , g 2 s , g 1 a , g 2 a , y 1 , c 1 , c 2 , (84)
in addition to the two chiral invariant mass parameters m B 0 m_{B0} m B 0 and m ψ 0 m_{\psi0} m ψ 0 . We treat ( m B 0 , m ψ 0 ) (m_{B0},m_{\psi0}) ( m B 0 , m ψ 0 ) as external parameters and scan them over a two-dimensional grid. At each grid point, we minimize the reduced chi-squared,
χ 2 = 1 N d o f [ ∑ i ( m i f i t − m i e x p δ m i ) 2 + ( σ s N f i t − σ s N e x p δ σ s N ) 2 + ( σ π N f i t − σ π N e x p δ σ π N ) 2 ] , (85) \chi^2 = \frac{1}{N_{\mathrm{dof}}}\left[\sum_i \left(\frac{m_i^{\mathrm{fit}}-m_i^{\mathrm{exp}}}{\delta m_i}\right)^2+\left(\frac{\sigma_{sN}^{\mathrm{fit}}-\sigma_{sN}^{\mathrm{exp}}}{\delta\sigma_{sN}}\right)^2+\left(\frac{\sigma_{\pi N}^{\mathrm{fit}}-\sigma_{\pi N}^{\mathrm{exp}}}{\delta\sigma_{\pi N}}\right)^2\right], \tag*{(85)} χ 2 = N dof 1 [ i ∑ ( δ m i m i fit − m i exp ) 2 + ( δ σ s N σ s N fit − σ s N exp ) 2 + ( δ σ π N σ π N fit − σ π N exp ) 2 ] , (85)
to determine the seven couplings. The mass inputs are listed in Table 2 . We assign δ m i = 10 \delta m_i=10 δ m i = 10 MeV to the ground states and δ m i = 100 \delta m_i=100 δ m i = 100 MeV to the excited states. For the strangeness sigma term, we use the FLAG 2024 average [48 ] ,
Table 2. Baryon mass inputs used in the fit. Each row corresponds to one SU(3)-flavor octet, of which only the N N N , Σ \Sigma Σ , and Ξ \Xi Ξ members are used here. Masses are given in MeV; entries marked (?) are not included in the fit.
Table 2. Mass inputs for octet members [MeV]
σ s N e x p = 44.9 ± 6.4 M e V . (86) \sigma_{sN}^{\mathrm{exp}}=44.9\pm6.4\ \mathrm{MeV}. \tag*{(86)} σ s N exp = 44.9 ± 6.4 MeV . (86)
Lattice-QCD calculations of the pion–nucleon sigma term [49 , 50 , 51 , 52 , 53 , 54 ] typically give values in the range σ π N ≃ 30 \sigma_{\pi N}\simeq30 σ π N ≃ 30 –46 46 46 MeV, whereas an analysis of π N \pi N π N scattering data gives σ π N = 59.1 ± 3.5 \sigma_{\pi N}=59.1\pm3.5 σ π N = 59.1 ± 3.5 MeV [55 ] . To accommodate most of these determinations, we use
σ π N e x p = 45 ± 15 M e V . (87) \sigma_{\pi N}^{\mathrm{exp}}=45\pm15\ \mathrm{MeV}. \tag*{(87)} σ π N exp = 45 ± 15 MeV . (87)
The current quark masses are fixed to ( m u , m s ) = ( 3.5 , 95 ) (m_u,m_s)=(3.5,95) ( m u , m s ) = ( 3.5 , 95 ) MeV. We note that the fit contains eight mass inputs and two sigma-term inputs. With seven couplings determined at each grid point, the number of degrees of freedom is therefore
N d o f = 10 − 7 = 3. (88) N_{\mathrm{dof}}=10-7=3. \tag*{(88)} N dof = 10 − 7 = 3. (88)
Mass-spectrum and sigma-term results
The result of the scan is shown in Figure 8 . The global minimum,
Figure 8. Reduced chi-squared, χ 2 / N d o f \chi^2/N_{\mathrm{dof}} χ 2 / N dof , from the seven-coupling fit in the ( m ψ 0 , m B 0 ) (m_{\psi0},m_{B0}) ( m ψ 0 , m B 0 ) plane. At each point, χ 2 \chi^2 χ 2 is minimized over the seven couplings. The uncolored region corresponds to χ 2 / N d o f > 10 \chi^2/N_{\mathrm{dof}} > 10 χ 2 / N dof > 10 . The star marks the global minimum at ( m B 0 , m ψ 0 ) = ( 500 , 200 ) (m_{B0},m_{\psi0}) = (500, 200) ( m B 0 , m ψ 0 ) = ( 500 , 200 ) MeV.
χ 2 N d o f = 0.46 , (89) \frac{\chi^2}{N_{\mathrm{dof}}}=0.46, \tag*{(89)} N dof χ 2 = 0.46 , (89)
is located at ( m B 0 , m ψ 0 ) = ( 500 , 200 ) (m_{B0},m_{\psi0})=(500,200) ( m B 0 , m ψ 0 ) = ( 500 , 200 ) MeV. In the following, we adopt this best-fit point as our reference. The corresponding couplings, condensate parameters, and sigma terms are collected in Table 3 .
Table 3. Fitted couplings, condensate parameters, and predicted sigma terms at the best-fit point ( m B 0 , m ψ 0 ) = ( 500 , 200 ) (m_{B0},m_{\psi0}) = (500, 200) ( m B 0 , m ψ 0 ) = ( 500 , 200 ) MeV.
The corresponding mass spectra are displayed in Figure 9 . With only seven couplings, the model reproduces all eight mass inputs and correctly yields the hierarchy m N < m Σ < m Ξ m_N<m_\Sigma<m_\Xi m N < m Σ < m Ξ for all the ground and excited states. Achieving this with such a minimal parameter set was not possible in the purely hadronic description of our previous analyses [37 , 38 , 39 ] . We attribute the improvement to the additional dynamical structure supplied by the quark–diquark construction. The predicted sigma terms at the best-fit point are σ s N = 45 \sigma_{sN}=45 σ s N = 45 MeV and σ π N = 38 \sigma_{\pi N}=38 σ π N = 38 MeV, and are compared with lattice-QCD results in Figure 10 . Both values are consistent with the corresponding fit inputs and with most lattice determinations. The condensate contribution identified in Eq. (83) is essential for the pion–nucleon sigma term: omitting the terms proportional to α − f 0 \alpha-f_0 α − f 0 gives the unphysical value σ π N = − 11.4 \sigma_{\pi N}=-11.4 σ π N = − 11.4 MeV, whereas including them yields σ π N ≃ 38 \sigma_{\pi N}\simeq38 σ π N ≃ 38 MeV.
Figure 9. Octet baryon mass spectra at the best-fit point ( m B 0 , m ψ 0 ) = ( 500 , 200 ) (m_{B0},m_{\psi0})=(500,200) ( m B 0 , m ψ 0 ) = ( 500 , 200 ) MeV. Solid black and dashed blue bars denote the model 1 / 2 + 1/2^+ 1/ 2 + and 1 / 2 − 1/2^- 1/ 2 − states, respectively, while red bars show the experimental inputs with their assigned uncertainties. The shaded bands are the 1 σ 1\sigma 1 σ uncertainties propagated from the coupling fit.
Figure 10. Nucleon sigma terms at the best-fit point ( m B 0 , m ψ 0 ) = ( 500 , 200 ) (m_{B0},m_{\psi0})=(500,200) ( m B 0 , m ψ 0 ) = ( 500 , 200 ) MeV. The model predictions, shown by their central values and 1 σ 1\sigma 1 σ bands from the coupling fit, are compared with lattice-QCD calculations represented by the colored error bars: red [49 ] , green [50 ] , purple [51 ] , brown [52 ] , and pink [53 ] .
Mixing structure and effective chiral invariant mass
Each physical baryon eigenstate is a linear combination of the four basis fields. For the nucleon,
∣ N ⟩ = a B 1 ∣ B 1 ⟩ + a B 2 ∣ B 2 ⟩ + a ψ 1 ∣ ψ 1 ⟩ + a ψ 2 ∣ ψ 2 ⟩ , (90) \lvert N\rangle=a_{B1}\lvert B_1\rangle+a_{B2}\lvert B_2\rangle+a_{\psi1}\lvert\psi_1\rangle+a_{\psi2}\lvert\psi_2\rangle, \tag*{(90)} ∣ N ⟩ = a B 1 ∣ B 1 ⟩ + a B 2 ∣ B 2 ⟩ + a ψ 1 ∣ ψ 1 ⟩ + a ψ 2 ∣ ψ 2 ⟩ , (90)
with the normalization a B 1 2 + a B 2 2 + a ψ 1 2 + a ψ 2 2 = 1 a_{B1}^2+a_{B2}^2+a_{\psi1}^2+a_{\psi2}^2=1 a B 1 2 + a B 2 2 + a ψ 1 2 + a ψ 2 2 = 1 , and analogously for the Σ \Sigma Σ and Ξ \Xi Ξ channels. Since the B B B fields are assigned to the ( 8 , 1 ) ⊕ ( 1 , 8 ) (8,1)\oplus(1,8) ( 8 , 1 ) ⊕ ( 1 , 8 ) chiral representation and the ψ \psi ψ fields to ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) (3,\bar{3})\oplus(\bar{3},3) ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) , it is natural to group the amplitudes into the total field contents
R B = a B 1 2 + a B 2 2 , R ψ = a ψ 1 2 + a ψ 2 2 , R B + R ψ = 1. (91) R_B=a_{B1}^{2}+a_{B2}^{2},\qquad R_\psi=a_{\psi1}^{2}+a_{\psi2}^{2},\qquad R_B+R_\psi=1. \tag*{(91)} R B = a B 1 2 + a B 2 2 , R ψ = a ψ 1 2 + a ψ 2 2 , R B + R ψ = 1. (91)
These quantities measure the probability with which each chiral representation is realized in a given state. The mixing at the best-fit point is collected in Table 4 , and its evolution across the spectrum is shown in Figure 11 .
Figure 11. Field content of the octet-baryon eigenstates at the best-fit point ( m B 0 , m ψ 0 ) = ( 500 , 200 ) M e V (m_{B0},m_{\psi0})=(500,200)\ {\rm MeV} ( m B 0 , m ψ 0 ) = ( 500 , 200 ) MeV . For each state, we plot the weights of the two interpolating fields, ∣ ⟨ B ∣ i ⟩ ∣ 2 |\langle\mathcal{B}|i\rangle|^2 ∣ ⟨ B ∣ i ⟩ ∣ 2 and ∣ ⟨ ψ ∣ i ⟩ ∣ 2 |\langle\psi|i\rangle|^2 ∣ ⟨ ψ ∣ i ⟩ ∣ 2 , which sum to unity and give the fractions of the naive B \mathcal{B} B -type and partner ψ \psi ψ -type components in the physical baryon. States are ordered by parity and mass as 1 / 2 1 + , 1 / 2 2 + , 1 / 2 1 − , 1 / 2 2 − 1/2^+_1,1/2^+_2,1/2^-_1,1/2^-_2 1/ 2 1 + , 1/ 2 2 + , 1/ 2 1 − , 1/ 2 2 − . The shaded bands are the 1 σ 1\sigma 1 σ uncertainties propagated from the coupling fit.
Table 4. Mass spectrum and eigenvector mixing structure of the octet baryons at ( m B 0 , m ψ 0 ) = ( 500 , 200 ) (m_{B0},m_{\psi0})=(500,200) ( m B 0 , m ψ 0 ) = ( 500 , 200 ) MeV. The last two columns give the total B B B - and ψ \psi ψ -field contents, R B ≡ a B 1 2 + a B 2 2 R_B \equiv a_{B_1}^2+a_{B_2}^2 R B ≡ a B 1 2 + a B 2 2 and R ψ ≡ a ψ 1 2 + a ψ 2 2 R_\psi\equiv a_{\psi_1}^2+a_{\psi_2}^2 R ψ ≡ a ψ 1 2 + a ψ 2 2 .
For every octet ground state, we find R B ≃ R ψ R_B\simeq R_\psi R B ≃ R ψ : the ( 8 , 1 ) ⊕ ( 1 , 8 ) (8,1)\oplus(1,8) ( 8 , 1 ) ⊕ ( 1 , 8 ) and ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) (3,\bar{3})\oplus(\bar{3},3) ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) components enter with almost equal weight. This mixing pattern agrees with the purely hadronic analysis of Ref. [37 ] and shows that the ground-state octet is not dominated by a single chiral representation. A direct consequence is that the chiral invariant mass of each ground state is shared between the two assignments. In particular, the effective invariant mass
m 0 e f f ≃ R B m B 0 + R ψ m ψ 0 (92) m_{0}^{\mathrm{eff}}\simeq R_Bm_{B0}+R_\psi m_{\psi0} \tag*{(92)} m 0 eff ≃ R B m B 0 + R ψ m ψ 0 (92)
lies close to the mean ( m B 0 + m ψ 0 ) / 2 (m_{B0}+m_{\psi0})/2 ( m B 0 + m ψ 0 ) /2 , so neither m B 0 m_{B0} m B 0 nor m ψ 0 m_{\psi0} m ψ 0 alone characterizes the origin of the nucleon mass. The excited states are more polarized and follow a clear pattern. The first 1 / 2 + 1/2^{+} 1/ 2 + excitation is B B B dominated (R B > R ψ R_B>R_\psi R B > R ψ ), the lowest 1 / 2 − 1/2^{-} 1/ 2 − state is ψ \psi ψ dominated (R ψ > R B R_\psi>R_B R ψ > R B ), and the second 1 / 2 − 1/2^{-} 1/ 2 − state returns to a nearly equal mixture (R B ≃ R ψ R_B\simeq R_\psi R B ≃ R ψ ). The excited spectrum therefore carries a more definite chiral-representation character than the ground state. This distinguishes these levels dynamically rather than merely by their masses. We stress that this field content is a vacuum quantity. Because the mixing is generated by the same symmetry-breaking condensates that enter M N \mathcal{M}_N M N , it depends on the condensate, and the eigenstates reorganize into the pure B \mathcal{B} B and ψ \psi ψ doublets as f / f 0 → 0 f/f_0 \to0 f / f 0 → 0 .
Chiral restoration and parity doubling
A distinctive property of the parity-doublet framework is that the chiral invariant masses m B 0 m_{B0} m B 0 and m ψ 0 m_{\psi0} m ψ 0 survive in the chiral-restored phase, where the positive- and negative-parity baryons become degenerate. To exhibit this explicitly, we keep the best-fit point ( m B 0 , m ψ 0 ) = ( 500 , 200 ) M e V (m_{B0},m_{\psi0})=(500,200)\ {\rm MeV} ( m B 0 , m ψ 0 ) = ( 500 , 200 ) MeV and the seven couplings fixed, and follow the four eigenstates of each of M N \mathcal{M}_N M N , M Σ \mathcal{M}_\Sigma M Σ , and M Ξ \mathcal{M}_\Xi M Ξ in Eqs. (63) –(65) as the chiral order parameter is reduced from its vacuum value to zero.
Every entry of the mass matrices is proportional to one of the four symmetry-breaking weights α + c i m u \alpha+c_i m_u α + c i m u , γ + c i m s \gamma+c_i m_s γ + c i m s (i = 1 , 2 i=1,2 i = 1 , 2 ). In the chiral limit (m u = m s = 0 m_u=m_s=0 m u = m s = 0 ), all decay constants collapse to the common value f 0 f_0 f 0 , so each weight separates into a spontaneous (condensate) piece f 0 f_0 f 0 and an explicit-breaking remainder,
α + c i m u = f 0 + [ ( α − f 0 ) + c i m u ] ≡ f 0 + α ˉ i , (93) \alpha+c_i m_u=f_0+\left[(\alpha-f_0)+c_i m_u\right]\equiv f_0+\bar{\alpha}_i, \tag*{(93)} α + c i m u = f 0 + [ ( α − f 0 ) + c i m u ] ≡ f 0 + α ˉ i , (93)
γ + c i m s = f 0 + [ ( γ − f 0 ) + c i m s ] ≡ f 0 + γ ˉ i , i ∈ { 1 , 2 } , (94) \gamma+c_i m_s=f_0+\left[(\gamma-f_0)+c_i m_s\right]\equiv f_0+\bar{\gamma}_i,\qquad i\in\{1,2\}, \tag*{(94)} γ + c i m s = f 0 + [ ( γ − f 0 ) + c i m s ] ≡ f 0 + γ ˉ i , i ∈ { 1 , 2 } , (94)
where the strange remainders γ ˉ i \bar{\gamma}_i γ ˉ i are enhanced by m s ≫ m u m_s\gg m_u m s ≫ m u [cf. Eqs. (76) and (77) ]. Numerically, from Table 3 ,
α ˉ 1 ≃ 3.3 M e V , α ˉ 2 ≃ − 0.9 M e V , γ ˉ 1 ≃ 12.1 M e V , γ ˉ 2 ≃ − 103.9 M e V . (95) \bar{\alpha}_1\simeq3.3\ \mathrm{MeV},\qquad\bar{\alpha}_2\simeq-0.9\ \mathrm{MeV},\qquad\bar{\gamma}_1\simeq12.1\ \mathrm{MeV},\qquad\bar{\gamma}_2\simeq-103.9\ \mathrm{MeV}. \tag*{(95)} α ˉ 1 ≃ 3.3 MeV , α ˉ 2 ≃ − 0.9 MeV , γ ˉ 1 ≃ 12.1 MeV , γ ˉ 2 ≃ − 103.9 MeV . (95)
In the following we model chiral restoration by scaling only the spontaneous part with the fraction x ≡ f / f 0 ∈ [ 0 , 1 ] x\equiv f/f_0\in[0,1] x ≡ f / f 0 ∈ [ 0 , 1 ] , keeping the current quark masses—and hence the explicit remainders—at their physical values,
α + c i m u ⟶ x f 0 + α ˉ i , \alpha+c_i m_u\longrightarrow x f_0+\bar{\alpha}_i, α + c i m u ⟶ x f 0 + α ˉ i ,
γ + c i m s ⟶ x f 0 + γ ˉ i , (96) \gamma+c_i m_s\longrightarrow x f_0+\bar{\gamma}_i, \tag*{(96)} γ + c i m s ⟶ x f 0 + γ ˉ i , (96)
so that x = 1 x=1 x = 1 corresponds to the vacuum and x = 0 x=0 x = 0 to the chiral-restored phase.
The key structural point is that the symmetry breaking effects are differently distributed in the mass matrices for three baryons N N N , Σ \Sigma Σ and Ξ \Xi Ξ , as summarized in Table 5 : the effect from the current strange quark γ ˉ i \bar{\gamma}_i γ ˉ i enters the N N N channel through the B B B –ψ \psi ψ mixing coupling y 1 y_1 y 1 , the Σ \Sigma Σ channel only through the diagonal g 1 , 2 a g^a_{1,2} g 1 , 2 a entries of the ψ \psi ψ sector, and the Ξ \Xi Ξ channel through the g 1 , 2 s g^s_{1,2} g 1 , 2 s mixing entries. Since only the remainders α ˉ i \bar{\alpha}_i α ˉ i and γ ˉ i \bar{\gamma}_i γ ˉ i in Eq. (95)
Table 5. Remnants of the explicit symmetry breaking existing in the N N N , Σ \Sigma Σ , and Ξ \Xi Ξ mass matrices, Eqs. (63) –(65) . The numbers in parentheses give the explicit remainders (in MeV) surviving in the chiral-restored phase, x = 0 x=0 x = 0 .
survive at x = 0 x = 0 x = 0 , with ∣ γ ˉ 2 ∣ ≫ ∣ γ ˉ 1 ∣ ≫ ∣ α ˉ 1 ∣ , ∣ α ˉ 2 ∣ |\bar{\gamma}_{2}| \gg|\bar{\gamma}_{1}| \gg|\bar{\alpha}_{1}|, |\bar{\alpha}_{2}| ∣ γ ˉ 2 ∣ ≫ ∣ γ ˉ 1 ∣ ≫ ∣ α ˉ 1 ∣ , ∣ α ˉ 2 ∣ , this assignment controls both where the restored doublets sit and how strongly they remain mixed.
Figure 12 shows the evolution of the twelve eigenvalues with x x x . For all baryon species, the parity splittings decrease monotonically, and parity doubling becomes nearly exact at x = 0 x = 0 x = 0 even at physical quark masses. The restored masses, however, differ markedly among the different baryons, following directly from Table 5 . In the N N N channel, the restored doublets sit at ≃ 830 \simeq830 ≃ 830 and ≃ 1530 \simeq1530 ≃ 1530 MeV, far from the invariant masses m ψ 0 m_{\psi0} m ψ 0 and m B 0 m_{B0} m B 0 . The reason is that the y 1 y_{1} y 1 entry carries the explicit symmetry-breaking effect from the strange quark, so the mixing element Y 1 ≡ ∣ y 1 γ ˉ 2 ∣ ≃ 1.17 Y_{1} \equiv|y_{1}\bar{\gamma}_{2}| \simeq1.17 Y 1 ≡ ∣ y 1 γ ˉ 2 ∣ ≃ 1.17 GeV survives in the restored phase. Keeping only this entry among the symmetry-breaking terms, the 4 × 4 4 \times4 4 × 4 matrix M N \mathcal{M}_{N} M N is partitioned into two 2 × 2 2 \times2 2 × 2 blocks based on the basis { ( B 1 + B 2 ) / 2 , ( ψ 1 + ψ 2 ) / 2 } \{(B_{1} + B_{2})/\sqrt{2},(\psi_{1} + \psi_{2})/\sqrt{2}\} {( B 1 + B 2 ) / 2 , ( ψ 1 + ψ 2 ) / 2 } and { ( B 1 − B 2 ) / 2 , ( ψ 1 − ψ 2 ) / 2 } \{(B_{1} - B_{2})/\sqrt{2},(\psi_{1} - \psi_{2})/\sqrt{2}\} {( B 1 − B 2 ) / 2 , ( ψ 1 − ψ 2 ) / 2 } . As a result, there are two exactly degenerate parity doublet with masses given by
Figure 12. Masses of the four N N N (top left), Σ \Sigma Σ (top right), and Ξ \Xi Ξ (bottom) eigenstates at the best-fit point ( m B 0 , m ψ 0 ) = ( 500 , 200 ) (m_{B0},m_{\psi0})=(500,200) ( m B 0 , m ψ 0 ) = ( 500 , 200 ) MeV as functions of the condensate fraction x = f / f 0 x=f/f_0 x = f / f 0 , obtained from the scaling (96) with the current quark masses held at their physical values; x = 1 x=1 x = 1 is the vacuum and x = 0 x=0 x = 0 the chiral-restored phase, and the upper axes give f = x f 0 f=xf_0 f = x f 0 in MeV. Solid (dashed) curves denote the positive- (negative-) parity states, and the blue (red) bands show the parity splitting Δ m = ∣ m + − m − ∣ \Delta m=\lvert m_+-m_-\rvert Δ m = ∣ m + − m − ∣ of the ground-state (excited-state) doublet; the quoted numbers give Δ m \Delta m Δ m at x = 0 x=0 x = 0 in MeV.
m N r e s t , ∓ = Y 1 2 + 1 4 ( m B 0 − m ψ 0 ) 2 ∓ 1 2 ( m B 0 + m ψ 0 ) ≃ 0.83 and 1.53 GeV , (97) m_{N}^{\mathrm{rest},\mp} = \sqrt{Y_{1}^{2} + \frac{1}{4}(m_{B0} - m_{\psi0})^{2}} \mp\frac{1}{2}(m_{B0} + m_{\psi0}) \simeq0.83\ \text{and}\ 1.53\ \text{GeV}, \tag*{(97)} m N rest , ∓ = Y 1 2 + 4 1 ( m B 0 − m ψ 0 ) 2 ∓ 2 1 ( m B 0 + m ψ 0 ) ≃ 0.83 and 1.53 GeV , (97)
which are separated by exactly m B 0 + m ψ 0 = 700 m_{B0} + m_{\psi0} = 700 m B 0 + m ψ 0 = 700 MeV. Since Y 1 Y_{1} Y 1 connects the B \mathcal{B} B and ψ \psi ψ sectors symmetrically in parity, it shifts the doublets without splitting them; the small residual splittings, Δ m ≃ 21 \Delta m \simeq21 Δ m ≃ 21 and 14 14 14 MeV, are generated by the light entries ∝ α ˉ 1 \propto\bar{\alpha}_{1} ∝ α ˉ 1 . The restored ground-state doublet, built from N ( 939 ) N(939) N ( 939 ) and N ( 1535 ) N(1535) N ( 1535 ) , thus remains at ≃ 830 \simeq830 ≃ 830 MeV, consistent with the chiral invariant masses favored by neutron-star observations [28 , 29 , 30 , 31 , 32 ] and with the gravitational-form-factor perspective of Ref. [56 ] .
In the Σ \Sigma Σ channel, on the other hand, all the entries generating B \mathcal{B} B –ψ \psi ψ mixing are proportional to α ˉ i \bar{\alpha}_{i} α ˉ i which effectively vanishes at x = 0 x = 0 x = 0 , so M Σ \mathcal{M}_{\Sigma} M Σ becomes block diagonal. The B \mathcal{B} B doublet settles at the invariant mass, m B 0 = 500 m_{B0} = 500 m B 0 = 500 MeV, with Δ m ≃ 1 \Delta m \simeq1 Δ m ≃ 1 MeV. The ψ \psi ψ block retains the strange-quark contribution on its diagonal and yields masses ≃ m ψ 0 ± 1 2 ( g 1 a + g 2 a ) γ ˉ 1 ≃ 137 \simeq m_{\psi0} \pm\frac{1}{2}(g_{1}^{a} + g_{2}^{a})\bar{\gamma}_{1} \simeq137 ≃ m ψ 0 ± 2 1 ( g 1 a + g 2 a ) γ ˉ 1 ≃ 137 and 262 262 262 MeV: a doublet centered at m ψ 0 = 200 m_{\psi0} = 200 m ψ 0 = 200 MeV with splitting Δ m ≃ ( g 1 a + g 2 a ) γ ˉ 1 ≃ 125 \Delta m \simeq(g_{1}^{a} + g_{2}^{a})\bar{\gamma}_{1} \simeq125 Δ m ≃ ( g 1 a + g 2 a ) γ ˉ 1 ≃ 125 MeV. The restored Σ \Sigma Σ spectrum thus exposes both chiral invariant masses directly. In Ξ \Xi Ξ channel, only the mixing entries g 1 , 2 s γ ˉ 1 ≃ ∓ 0.17 g_{1,2}^{s}\bar{\gamma}_{1} \simeq\mp0.17 g 1 , 2 s γ ˉ 1 ≃ ∓ 0.17 GeV survive. Their level repulsion lifts both doublets by ≃ ( g s γ ˉ 1 ) 2 / ( m B 0 + m ψ 0 ) ≃ 40 \simeq(g^{s}\bar{\gamma}_{1})^{2}/(m_{B0} + m_{\psi0}) \simeq40 ≃ ( g s γ ˉ 1 ) 2 / ( m B 0 + m ψ 0 ) ≃ 40 MeV, so the restored doublets settle slightly above the invariant masses, at ≃ 540 \simeq540 ≃ 540 and ≃ 240 \simeq240 ≃ 240 MeV. The excited doublet is again almost degenerate (Δ m ≃ 1 \Delta m \simeq1 Δ m ≃ 1 MeV), while the ground-state splitting, Δ m ≃ ( g 1 a + g 2 a ) α ˉ 1 ≃ 33 \Delta m \simeq(g_{1}^{a} + g_{2}^{a})\bar{\alpha}_{1} \simeq33 Δ m ≃ ( g 1 a + g 2 a ) α ˉ 1 ≃ 33 MeV, comes from the diagonal entries.
The flavor hierarchy is therefore inverted in the restored phase: the hyperon doublets drop to the vicinity of the invariant masses, while the nucleon doublets are held up by the strange explicit breaking. This counterintuitive pattern—an m s m_{s} m s effect largest in the non-strange channel—reflects the U(1)A _{A} A breaking coupling of the diquarks to the condensate as well as the current quark masses, similarly to the inverted hierarchy pointed in Ref. [44 ] : in the y 1 y_{1} y 1 coupling, the ϵ \epsilon ϵ -tensor contraction ties the light [ u d ] [ud] [ u d ] diquark inside the nucleon to the strange component of ⟨ Σ ⟩ \langle\Sigma\rangle ⟨ Σ ⟩ , whereas for the Σ \Sigma Σ and Ξ \Xi Ξ channels it picks out the light components.
To clarify the internal structure of the restored states, we also examine the chiral representation content of the eigenvectors at x = 0 x = 0 x = 0 , shown in Figure 13 . The approach to parity doubling is accompanied by a reorganization of the eigenvectors: in every flavor channel, the two members of a doublet acquire almost identical chiral-representation fractions,
Figure 13. Chiral representation content of the N N N , Σ \Sigma Σ , and Ξ \Xi Ξ eigenstates in the chiral-restored phase (x = 0 x = 0 x = 0 ) at the best-fit point ( m B 0 , m ψ 0 ) = ( 500 , 200 ) (m_{B0},m_{\psi0}) = (500,200) ( m B 0 , m ψ 0 ) = ( 500 , 200 ) MeV. For each state, the fractions R B R_B R B and R ψ = 1 − R B R_{\psi} = 1 - R_B R ψ = 1 − R B give the weights of the ( 8 , 1 ) ⊕ ( 1 , 8 ) (8,1) \oplus(1,8) ( 8 , 1 ) ⊕ ( 1 , 8 ) and ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) (3,\bar{3}) \oplus(\bar{3},3) ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) components.
R B ( F 1 + ) ≃ R B ( F 3 − ) , R ψ ( F 2 + ) ≃ R ψ ( F 4 − ) , ( F = N , Σ , Ξ ) , (98) R_B(F_1^+) \simeq R_B(F_3^-), \qquad R_{\psi}(F_2^+) \simeq R_{\psi}(F_4^-), \qquad(F = N, \Sigma, \Xi), \tag*{(98)} R B ( F 1 + ) ≃ R B ( F 3 − ) , R ψ ( F 2 + ) ≃ R ψ ( F 4 − ) , ( F = N , Σ , Ξ ) , (98)
so that the restored partners share not only nearly degenerate masses but also a common internal chiral structure.
The degree of residual mixing again mirrors Table 5 . For the nucleon, the surviving mixing Y 1 ∝ m s Y_1 \propto m_s Y 1 ∝ m s is non-negligible, and both doublets remain close to maximal mixing between ( 8 , 1 ) ⊕ ( 1 , 8 ) (8,1) \oplus(1,8) ( 8 , 1 ) ⊕ ( 1 , 8 ) and ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) (3,\bar{3}) \oplus(\bar{3},3) ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) ,
R B ( N 1 + ) ≃ 0.44 , R B ( N 3 − ) ≃ 0.43 ; R B ( N 2 + ) ≃ 0.57 , R B ( N 4 − ) ≃ 0.56. (99) R_B(N_1^+) \simeq0.44,\qquad R_B(N_3^-) \simeq0.43;\qquad R_B(N_2^+) \simeq0.57,\qquad R_B(N_4^-) \simeq0.56. \tag*{(99)} R B ( N 1 + ) ≃ 0.44 , R B ( N 3 − ) ≃ 0.43 ; R B ( N 2 + ) ≃ 0.57 , R B ( N 4 − ) ≃ 0.56. (99)
For the Σ \Sigma Σ , all mixing entries are proportional to the light remainders ∝ m u \propto m_u ∝ m u , and the restored eigenstates separate cleanly into an almost pure ψ \psi ψ ground-state doublet (R ψ ≃ 0.99 R_{\psi} \simeq0.99 R ψ ≃ 0.99 ) and an almost pure B \mathcal{B} B excited doublet (R B ≃ 0.99 R_{\mathcal{B}} \simeq0.99 R B ≃ 0.99 ). The Ξ \Xi Ξ lies in between: the surviving mixing g 1 Ξ γ ˉ 1 ∝ c 1 m s g_{1}^{\Xi}\bar{\gamma}_{1} \propto c_{1}m_{s} g 1 Ξ γ ˉ 1 ∝ c 1 m s is numerically modest because the small c 1 c_{1} c 1 compensates the large m s m_{s} m s , leaving a residual admixture of only about 5% in each doublet.
Finally, in the exact chiral limit m u = m s = 0 m_{u}=m_{s}=0 m u = m s = 0 , all remainders in Eq. (95) vanish, and at x = 0 x=0 x = 0 the mass matrix of every flavor channel collapses to the common block-diagonal form
M F ( x = 0 ) ∣ m u = m s = 0 = ( 0 m B 0 0 0 m B 0 0 0 0 0 0 0 m ψ 0 0 0 m ψ 0 0 ) , F = N , Σ , Ξ , (100) \left.\mathcal{M}_{F}(x=0)\right|_{m_{u}=m_{s}=0} =
\begin{pmatrix}
0 & m_{\mathcal{B}0} & 0 & 0 \\
m_{\mathcal{B}0} & 0 & 0 & 0 \\
0 & 0 & 0 & m_{\psi0} \\
0 & 0 & m_{\psi0} & 0
\end{pmatrix},
\qquad F=N,\ \Sigma,\ \Xi, \tag*{(100)} M F ( x = 0 ) ∣ m u = m s = 0 = 0 m B 0 0 0 m B 0 0 0 0 0 0 0 m ψ 0 0 0 m ψ 0 0 , F = N , Σ , Ξ , (100)
with eigenvalues { ± m B 0 , ± m ψ 0 } \{\pm m_{\mathcal{B}0},\pm m_{\psi0}\} { ± m B 0 , ± m ψ 0 } . All twelve baryon states then collapse onto two universal parity doublets located exactly at the chiral invariant masses, one purely of B \mathcal{B} B type and one purely of ψ \psi ψ type. The channel-dependent offsets, splittings, and mixings found above are therefore entirely a footprint of the explicit chiral symmetry breaking—dominated by the strange quark mass—superimposed on this universal parity-doubled structure.
Summary and discussion
In this paper we have proposed a microscopic realization of the parity-doublet structure of baryons, in which the chiral invariant mass—an input parameter of the conventional hadronic parity doublet model—is traced to the chiral invariant quark-diquark interaction generated by the gluon dynamics combined with the chiral invariant mass carried by the diquark. Furthermore, the existence of scalar and pseudoscalar diquarks naturally leads to the emergence of a parity-doublet structure in the resulting baryons. We would like to stress that, because the binding of the colored diquark is controlled by non-Abelian gluon dynamics rather than by the spontaneous breaking of chiral symmetry, this invariant mass is naturally insensitive to the quark condensate and persists in the chiral restored phase.
To turn this picture into a quantitative model, we constructed a three-flavor chiral quark–diquark model from the underlying quark and diquark degrees of freedom in § 3 . The model includes two independent chiral symmetric four-body quark–diquark interactions which are obtainable from one-gluon exchange, and U(1)A _{A} A breaking one. Using the auxiliary-field method we introduced the color-singlet composite baryons: the direct composites B \mathcal{B} B , belonging to ( 8 , 1 ) ⊕ ( 1 , 8 ) (8,1) \oplus(1,8) ( 8 , 1 ) ⊕ ( 1 , 8 ) , and their crossed partners ψ \psi ψ , belonging to ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) (3,\bar{3}) \oplus(\bar{3},3) ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) . Evaluating the one- and two-loop quark–diquark self-energies, we derived the baryon mass matrices analytically and showed in Sec. 4. We showed that a finite baryon mass, Eq. (45) , is generated even for massless quarks, jointly by the invariant diquark mass and the four-body quark–diquark interaction. The chiral invariant masses m B 0 m_{\mathcal{B}0} m B 0 and m ψ 0 m_{\psi0} m ψ 0 of the two composite sectors are therefore no longer phenomenological inputs of a hadronic Lagrangian but outputs of the quark–diquark dynamics. Coupling the quarks and diquarks to the chiral order parameter Σ \Sigma Σ then induces the mixing between B \mathcal{B} B and ψ \psi ψ and assembles the N N N , Σ \Sigma Σ , and Ξ \Xi Ξ mass matrices.
Confronting the model with data, we treated ( m B 0 , m ψ 0 ) (m_{\mathcal{B}0},m_{\psi0}) ( m B 0 , m ψ 0 ) as external parameters and fixed the seven remaining couplings at each grid point by fitting the mass spectra of N N N , Σ \Sigma Σ and Ξ \Xi Ξ baryons together with the two nucleon sigma terms in § 7 . The best fit, χ 2 / N d o f = 0.46 \chi^2/N_{\mathrm{dof}} = 0.46 χ 2 / N dof = 0.46 , is reached at ( m B 0 , m ψ 0 ) = ( 500 , 200 ) (m_{B0},m_{\psi0}) = (500, 200) ( m B 0 , m ψ 0 ) = ( 500 , 200 ) MeV, where the model reproduces all eight mass inputs and correctly orders the ground and excited states as m N < m Σ < m Ξ m_N < m_\Sigma< m_\Xi m N < m Σ < m Ξ . Achieving this within such a minimal parameter set was not possible in our earlier purely hadronic analyses [37 , 38 , 39 ] , and we attribute the improvement to the additional dynamical structure of the quark–diquark construction. At the same point the predicted sigma terms, σ s N ≃ 45 \sigma_{sN} \simeq45 σ s N ≃ 45 MeV and σ π N ≃ 38 \sigma_{\pi N} \simeq38 σ π N ≃ 38 MeV, are consistent with recent lattice determinations; we emphasized that a physical σ π N \sigma_{\pi N} σ π N requires the response of the chiral condensate itself (the α − f 0 \alpha-f_0 α − f 0 term in Eq. (77) ), without which the value is strongly underestimated.
The eigenvector analysis reveals a clear organizing pattern. Each ground state carries almost equal weight in the two chiral representations, R B ≃ R ψ ≃ 1 / 2 R_B \simeq R_\psi\simeq1/2 R B ≃ R ψ ≃ 1/2 , so that its chiral invariant mass is shared between the two assignments: the effective invariant mass sits close to the mean ( m B 0 + m ψ 0 ) / 2 (m_{B0}+m_{\psi0})/2 ( m B 0 + m ψ 0 ) /2 , and neither m B 0 m_{B0} m B 0 nor m ψ 0 m_{\psi0} m ψ 0 alone characterizes the origin of the nucleon mass. The excited states are more polarized in chiral-representation content and thereby acquire a definite character beyond their masses.
Dialing the chiral order parameter to zero with the current quark masses held at their physical values, we found in § 7.5 that parity doubling becomes nearly exact in all three flavor channels, while the explicit breaking leaves a characteristic, flavor-resolved imprint on the restored spectrum. The pattern is fixed by where the current mass of the strange quark enters into the mass matrices (Table 5 ). In the nucleon channel it enters the B \mathcal{B} B –ψ \psi ψ mixing, so the restored doublets are lifted to ≃ 830 \simeq830 ≃ 830 and ≃ 1530 \simeq1530 ≃ 1530 MeV—separated by exactly m B 0 + m ψ 0 m_{B0}+m_{\psi0} m B 0 + m ψ 0 , Eq. (97) —and remain almost maximally mixed between the two chiral representations. In the Σ \Sigma Σ channel, on the other hand, all mixing entries do not include the strange-quark mass and thus small, so the restored spectrum exposes the two invariant masses directly: an almost pure B \mathcal{B} B doublet at m B 0 = 500 m_{B0}=500 m B 0 = 500 MeV and an almost pure ψ \psi ψ doublet centered at m ψ 0 = 200 m_{\psi0}=200 m ψ 0 = 200 MeV. The Ξ \Xi Ξ channel is intermediate, with nearly pure doublets settling slightly above the invariant masses. The flavor hierarchy is thus inverted in the restored phase—all of Σ \Sigma Σ and Ξ \Xi Ξ baryons become lighter than four nucleons—a pattern that traces back to the inverse hierarchy of the diquark, which ties the [ u d ] [ud] [ u d ] diquark inside the nucleon to the current strange-quark mass. Only in the exact chiral limit do all twelve octet states collapse onto two universal parity doublets located precisely at m ψ 0 = 200 m_{\psi0}=200 m ψ 0 = 200 and m B 0 = 500 m_{B0}=500 m B 0 = 500 MeV, isolating the two chiral invariant masses [Eq. (6.29)]. The restored nucleon ground-state doublet at ≃ 830 \simeq830 ≃ 830 MeV is compatible with the chiral invariant masses favored by recent neutron-star observations [28 , 29 , 30 , 31 , 32 ] , while the distinct doubling patterns of the N N N , Σ \Sigma Σ , and Ξ \Xi Ξ channels constitute a prediction that can in principle be confronted with lattice QCD studies of nucleon and hyperon parity doubling at finite temperature [10 , 11 , 12 ] .
Before closing, we comment on an approximation made in the two-loop analysis of Sec. 5. There, we retained only the diagrams carrying the same composite vertex type—derivative (f 1 f_1 f 1 ) or trivial (f 2 f_2 f 2 )—at both ends, which generate the diagonal mass entries g 1 a g_1^a g 1 a and g 2 a g_2^a g 2 a in the ψ \psi ψ sector. Mixed diagrams, with one f 1 f_1 f 1 and one f 2 f_2 f 2 composite vertex, are also allowed and connect ψ 1 \psi_1 ψ 1 to ψ 2 \psi_2 ψ 2 . A counting of Dirac matrices shows, however, that they cannot generate a mass term: with massless quarks each internal quark propagator supplies one γ \gamma γ matrix and the charge-conjugation matrices of the G d G_d G d vertices drop out of the counting, so the mixed diagrams carry an odd number of γ \gamma γ matrices between the external Dirac indices. The resulting structure is proportional to P̸ \not{P} P , vanishes at zero external momentum, and pairs composites of the same Dirac chirality; its leading term in the derivative expansion is the chiral invariant kinetic mixing
L m i x k i n = ε k i n ϵ i j k ϵ α β γ ( Σ ) γ k ( ψ ˉ 2 L ) α j i ∂̸ ( ψ 1 L ) β i + ( L ↔ R , Σ → Σ † ) + h . c . , (101) \mathcal{L}_{\mathrm{mix}}^{\mathrm{kin}} = \varepsilon_{\mathrm{kin}} \epsilon_{ijk} \epsilon^{\alpha\beta\gamma} \left(\Sigma\right)^{k}_{\gamma} \left(\bar{\psi}_{2L}\right)^{j}_{\alpha} i\not{\partial} \left(\psi_{1L}\right)^{i}_{\beta} + \left(L \leftrightarrow R,\ \Sigma\to\Sigma^{\dagger}\right) + \mathrm{h.c.}, \tag*{(101)} L mix kin = ε kin ϵ ij k ϵ α β γ ( Σ ) γ k ( ψ ˉ 2 L ) α j i ∂ ( ψ 1 L ) β i + ( L ↔ R , Σ → Σ † ) + h.c. , (101)
with ε k i n \varepsilon_{\mathrm{kin}} ε kin a two-loop coefficient. Once Σ \Sigma Σ acquires its expectation value, Eq. (101) induces an off-diagonal kinetic mixing between ψ 1 \psi_1 ψ 1 and ψ 2 \psi_2 ψ 2 that cannot be absorbed into the wave-function normalizations of the individual fields: a consistent treatment would require diagonalizing the kinetic matrix first and only then rediagonalizing the mass matrix in the resulting basis. Moreover, owing to the ϵ Σ ϵ \epsilon\Sigma\epsilon ϵ Σ ϵ flavor structure, the induced kinetic mixing takes different values in the N N N , Σ \Sigma Σ , and Ξ \Xi Ξ channels, so that this procedure would have to be carried out channel by channel, substantially complicating the analysis. Since Eq. (101) arises only at the two-loop level and constitutes an O ( ⟨ Σ ⟩ ) \mathcal{O}(\langle\Sigma\rangle) O (⟨ Σ ⟩) correction to the one-loop kinetic terms of Sec. 4, we neglect it in the present analysis as a higher-order correction to the kinetic sector.
As we stated at the end of § 3.2 , we also omit the iso-singlet Λ \Lambda Λ baryons included in the octets in the present analysis since they mix with the flavor singlet baryons. It is interesting to study the mass spectra of Λ \Lambda Λ baryons by including the effective interaction given in Eq. (15) , which we leave as a future work.
Taken together, these results support a picture in which the part of the baryon mass that is insensitive to the chiral condensate is inherited from the diquark sector—the invariant diquark mass together with the four-body quark–diquark interaction, both generated by gluon dynamics and chiral invariant—thereby providing a microscopic origin for the chiral invariant mass. A distinctive feature of the quark–diquark construction is that this invariant mass is not a single number but is resolved into two contributions, m B 0 m_{B0} m B 0 and m ψ 0 m_{\psi0} m ψ 0 , tied to the ( 8 , 1 ) ⊕ ( 1 , 8 ) (8,1)\oplus(1,8) ( 8 , 1 ) ⊕ ( 1 , 8 ) and ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) (3,\bar{3})\oplus(\bar{3},3) ( 3 , 3 ˉ ) ⊕ ( 3 ˉ , 3 ) representations, whose mixing shapes the physical spectrum in the vacuum and whose separation is exposed in the restored phase.
Several directions deserve further study. The chiral invariant mass generated here from the diquark makes the model a natural starting point for the equation of state of dense matter and its application to neutron stars, where a relatively large m 0 m_0 m 0 is favored; the lightness of the restored hyperon doublets found above suggests, in addition, that hyperonic degrees of freedom may play an enhanced role in the vicinity of chiral restoration. On the spectroscopic side, the framework can be extended to the flavor-singlet and decuplet baryons, and to spin-1 (“bad”) diquark correlations, which would test the diquark picture more stringently. A systematic treatment of the Σ \Sigma Σ -induced kinetic mixing of Eq. (101) , including the channel-dependent rediagonalization that it entails, is another natural refinement. Finally, it would be valuable to compute the invariant diquark mass directly from the underlying gluon dynamics, rather than taking it as input, thereby closing the link between the chiral invariant baryon mass and the color interactions from which we have argued that it originates. We leave these developments to future work.
Acknowledgement
The work of B.G. is supported in part by JSPS KAKENHI Grant No. 26K17147. The work of M.H. is supported in part by JSPS KAKENHI Grant Nos. 23H05439 and 24K07045.
Discrete symmetries of the four-body interaction
In this appendix we show that the four-body interaction (12) is invariant under parity (P P P ) and charge conjugation (C C C ), and that these symmetries force the coupling constant to be real. We write G ≡ G q d G \equiv G_{qd} G ≡ G q d throughout; the same argument applies verbatim to L q d ( 2 ) \mathcal{L}_{qd}^{(2)} L q d ( 2 ) and its coupling G q d ( 2 ) G_{qd}^{(2)} G q d ( 2 ) .
Discrete transformations of the quark and diquark fields
Parity. Under parity the chirality of a quark field is flipped,
P : q L , R → γ 0 q R , L , q ˉ L , R → q ˉ R , L γ 0 . (102) P:\quad q_{L,R} \to\gamma^{0}q_{R,L},\qquad\bar{q}_{L,R} \to\bar{q}_{R,L}\gamma^{0}. \tag*{(102)} P : q L , R → γ 0 q R , L , q ˉ L , R → q ˉ R , L γ 0 . (102)
Applying this to the diquark and using ( γ 0 ) T = γ 0 (\gamma^{0})^{T}=\gamma^{0} ( γ 0 ) T = γ 0 together with γ 0 C γ 0 = − C \gamma^{0}C\gamma^{0}=-C γ 0 C γ 0 = − C ,
P : ( d R ) a , i = ϵ a b c ϵ i j k ( q R b , j ) T C q R c , k → ϵ a b c ϵ i j k ( q L b , j ) T γ 0 C γ 0 q L c , k = − ( d L ) a , i , (103) P:\quad(d_{R})_{a,i}=\epsilon_{abc}\epsilon_{ijk}(q_{R}^{b,j})^{T}Cq_{R}^{c,k}
\to\epsilon_{abc}\epsilon_{ijk}(q_{L}^{b,j})^{T}\gamma^{0}C\gamma^{0}q_{L}^{c,k}=-(d_{L})_{a,i}, \tag*{(103)} P : ( d R ) a , i = ϵ ab c ϵ ij k ( q R b , j ) T C q R c , k → ϵ ab c ϵ ij k ( q L b , j ) T γ 0 C γ 0 q L c , k = − ( d L ) a , i , (103)
and, in exactly the same way, ( d L ) a , i → − ( d R ) a , i (d_{L})_{a,i}\to-(d_{R})_{a,i} ( d L ) a , i → − ( d R ) a , i . Parity therefore interchanges the two chiral diquarks up to a sign,
P : ( d L ) a , i → − ( d R ) a , i , ( d R ) a , i → − ( d L ) a , i . (104) P:\quad(d_{L})_{a,i}\to-(d_{R})_{a,i},\qquad(d_{R})_{a,i}\to-(d_{L})_{a,i}. \tag*{(104)} P : ( d L ) a , i → − ( d R ) a , i , ( d R ) a , i → − ( d L ) a , i . (104)
Charge conjugation. Under charge conjugation the quark fields transform as
C : q → C q ˉ T , q ˉ → q T C . (105) C:\quad q\to C\bar{q}^{T},\qquad\bar{q}\to q^{T}C. \tag*{(105)} C : q → C q ˉ T , q ˉ → q T C . (105)
Acting on the left-handed component and using γ 5 C = C γ 5 T \gamma_{5}C=C\gamma_{5}^{T} γ 5 C = C γ 5 T together with q ˉ T = γ 0 q ∗ \bar{q}^{T}=\gamma^{0}q^{*} q ˉ T = γ 0 q ∗ ,
q L = 1 − γ 5 2 q → 1 − γ 5 2 C q ˉ T = C 1 − γ 5 T 2 γ 0 q ∗ = C ( 1 − γ 5 2 ) T q ˉ T = C q ˉ R T , (106) q_{L}=\frac{1-\gamma_{5}}{2}q\to\frac{1-\gamma_{5}}{2}C\bar{q}^{T}
=C\frac{1-\gamma_{5}^{T}}{2}\gamma^{0}q^{*}
=C\left(\frac{1-\gamma_{5}}{2}\right)^{T}\bar{q}^{T}
=C\bar{q}_{R}^{T}, \tag*{(106)} q L = 2 1 − γ 5 q → 2 1 − γ 5 C q ˉ T = C 2 1 − γ 5 T γ 0 q ∗ = C ( 2 1 − γ 5 ) T q ˉ T = C q ˉ R T , (106)
so that charge conjugation exchanges the chiralities. The full set of rules is
C : q L a , i → C ( q ˉ R T ) a , i , q R a , i → C ( q ˉ L T ) a , i , ( q ˉ L ) a , i → ( q R a , i ) T C , ( q ˉ R ) a , i → ( q L a , i ) T C . (107) \begin{aligned}
C:\quad q_{L}^{a,i}&\to C(\bar{q}_{R}^{T})_{a,i},\qquad q_{R}^{a,i}\to C(\bar{q}_{L}^{T})_{a,i},\\
(\bar{q}_{L})_{a,i}&\to(q_{R}^{a,i})^{T}C,\qquad(\bar{q}_{R})_{a,i}\to(q_{L}^{a,i})^{T}C. \tag*{(107)}
\end{aligned} C : q L a , i ( q ˉ L ) a , i → C ( q ˉ R T ) a , i , q R a , i → C ( q ˉ L T ) a , i , → ( q R a , i ) T C , ( q ˉ R ) a , i → ( q L a , i ) T C . (107)
Applying these to the diquark,
( d L ) a , i = ϵ a b c ϵ i j k ( q L b , j ) T C q L c , k → C ϵ a b c ϵ i j k ( C q ˉ R T ) b , j T C ( C q ˉ R T ) c , k = ϵ a b c ϵ i j k ( q ˉ R ) b , j C T C C ( q ˉ R T ) c , k = ϵ a b c ϵ i j k ( q ˉ R ) b , j C ( q ˉ R T ) c , k , (108) \begin{aligned}
(d_{L})_{a,i}
&=\epsilon_{abc}\epsilon_{ijk}(q_{L}^{b,j})^{T}Cq_{L}^{c,k}
\xrightarrow{C}\epsilon_{abc}\epsilon_{ijk}(C\bar{q}_{R}^{T})_{b,j}^{T}C(C\bar{q}_{R}^{T})_{c,k}\\
&=\epsilon^{abc}\epsilon^{ijk}(\bar{q}_{R})_{b,j}C^{T}CC(\bar{q}_{R}^{T})_{c,k}\\
&=\epsilon^{abc}\epsilon^{ijk}(\bar{q}_{R})_{b,j}C(\bar{q}_{R}^{T})_{c,k}, \tag*{(108)}
\end{aligned} ( d L ) a , i = ϵ ab c ϵ ij k ( q L b , j ) T C q L c , k C ϵ ab c ϵ ij k ( C q ˉ R T ) b , j T C ( C q ˉ R T ) c , k = ϵ ab c ϵ ij k ( q ˉ R ) b , j C T C C ( q ˉ R T ) c , k = ϵ ab c ϵ ij k ( q ˉ R ) b , j C ( q ˉ R T ) c , k , (108)
where C T C C = C C^{T}CC=C C T C C = C . On the other hand, the hermitian conjugate of d R d_{R} d R is
( d R † ) a , i = [ ϵ a b c ϵ i j k ( q R b , j ) T C q R c , k ] † = ϵ a b c ϵ i j k ( q ˉ R ) c , k γ 0 C † γ 0 ( q ˉ R T ) b , j = ϵ a b c ϵ i j k ( q ˉ R ) b , j C ( q ˉ R T ) c , k , (109) \left(d_{R}^{\dagger}\right)^{a,i}=\left[\epsilon_{abc}\epsilon_{ijk}\left(q_{R}^{b,j}\right)^{T}Cq_{R}^{c,k}\right]^{\dagger}=\epsilon^{abc}\epsilon^{ijk}\left(\bar{q}_{R}\right)_{c,k}\gamma^{0}C^{\dagger}\gamma^{0}\left(\bar{q}_{R}^{T}\right)_{b,j}
=\epsilon^{abc}\epsilon^{ijk}\left(\bar{q}_{R}\right)_{b,j}C\left(\bar{q}_{R}^{T}\right)_{c,k}, \tag*{(109)} ( d R † ) a , i = [ ϵ ab c ϵ ij k ( q R b , j ) T C q R c , k ] † = ϵ ab c ϵ ij k ( q ˉ R ) c , k γ 0 C † γ 0 ( q ˉ R T ) b , j = ϵ ab c ϵ ij k ( q ˉ R ) b , j C ( q ˉ R T ) c , k , (109)
where we used γ 0 C † γ 0 = C \gamma^{0}C^{\dagger}\gamma^{0}=C γ 0 C † γ 0 = C and, in the last step, relabeled the dummy indices ( b , j ) ↔ ( c , k ) (b,j)\leftrightarrow(c,k) ( b , j ) ↔ ( c , k ) , under which the product of the two Levi-Civita symbols is unchanged. Comparing (A.7) with (A.8),
C : ( d L ) a , i → ( d R † ) a , i , ( d R ) a , i → ( d L † ) a , i . (110) C:\quad(d_{L})_{a,i}\to\left(d_{R}^{\dagger}\right)^{a,i},\qquad(d_{R})_{a,i}\to\left(d_{L}^{\dagger}\right)^{a,i}. \tag*{(110)} C : ( d L ) a , i → ( d R † ) a , i , ( d R ) a , i → ( d L † ) a , i . (110)
Invariance of the four-body interaction and reality of G G G
It is convenient to define the operators
O L , R = { ( q ˉ L , R ) c , i ( d L , R † ) c , j } { ( i ∂ μ d L , R ) a , j γ μ ( q L , R ) a , i } . (111) \mathcal{O}_{L,R}=\left\{\left(\bar{q}_{L,R}\right)_{c,i}\left(d_{L,R}^{\dagger}\right)^{c,j}\right\}\left\{\left(i\partial_{\mu}d_{L,R}\right)_{a,j}\gamma^{\mu}\left(q_{L,R}\right)^{a,i}\right\}. \tag*{(111)} O L , R = { ( q ˉ L , R ) c , i ( d L , R † ) c , j } { ( i ∂ μ d L , R ) a , j γ μ ( q L , R ) a , i } . (111)
in terms of which the four-body interaction (12) takes the compact form
L 4 = − ∑ χ = L , R ( G O χ + G ∗ O χ † ) . (112) \mathcal{L}_{4}=-\sum_{\chi=L,R}\left(G\mathcal{O}_{\chi}+G^{*}\mathcal{O}_{\chi}^{\dagger}\right). \tag*{(112)} L 4 = − χ = L , R ∑ ( G O χ + G ∗ O χ † ) . (112)
We allow G G G to be complex a priori; hermiticity of the Lagrangian fixes only the coefficient of O χ † \mathcal{O}_{\chi}^{\dagger} O χ † to be G ∗ G^{*} G ∗ . We now show that parity invariance holds for any G G G , whereas charge-conjugation invariance forces G G G to be real.
Parity. Using the transformations of Appendix A.1,
O L → P { ( q ˉ R ) c , i γ 0 ( − i d R † ) c , j } { ( − i ∂ μ d R ) a , j γ μ γ 0 ( q R ) a , i } = { ( q ˉ R ) c , i ( d R † ) c , j } { ( i ∂ μ d R ) a , j γ μ ( q R ) a , i } ( x ~ ) = O R ( x ~ ) , (113) \mathcal{O}_{L}\xrightarrow{P}\left\{\left(\bar{q}_{R}\right)_{c,i}\gamma^{0}\left(-id_{R}^{\dagger}\right)^{c,j}\right\}\left\{\left(-i\partial_{\mu}d_{R}\right)_{a,j}\gamma^{\mu}\gamma^{0}\left(q_{R}\right)^{a,i}\right\}
=\left\{\left(\bar{q}_{R}\right)_{c,i}\left(d_{R}^{\dagger}\right)^{c,j}\right\}\left\{\left(i\partial^{\mu}d_{R}\right)_{a,j}\gamma_{\mu}\left(q_{R}\right)^{a,i}\right\}(\tilde{x})=\mathcal{O}_{R}(\tilde{x}), \tag*{(113)} O L P { ( q ˉ R ) c , i γ 0 ( − i d R † ) c , j } { ( − i ∂ μ d R ) a , j γ μ γ 0 ( q R ) a , i } = { ( q ˉ R ) c , i ( d R † ) c , j } { ( i ∂ μ d R ) a , j γ μ ( q R ) a , i } ( x ~ ) = O R ( x ~ ) , (113)
where the two minus signs from the diquark transformations cancel, and the sign flip of the spatial derivatives, ∂ μ → ∂ μ \partial_{\mu}\to\partial^{\mu} ∂ μ → ∂ μ , is compensated by γ 0 γ μ γ 0 = γ μ \gamma^{0}\gamma^{\mu}\gamma^{0}=\gamma_{\mu} γ 0 γ μ γ 0 = γ μ (x ~ \tilde{x} x ~ denotes the parity-reflected argument). An identical computation gives O R → O L ( x ~ ) \mathcal{O}_{R}\to\mathcal{O}_{L}(\tilde{x}) O R → O L ( x ~ ) , so parity simply interchanges the two operators,
P : O L → O R , O R → O L . (114) P:\quad\mathcal{O}_{L}\to\mathcal{O}_{R},\qquad\mathcal{O}_{R}\to\mathcal{O}_{L}. \tag*{(114)} P : O L → O R , O R → O L . (114)
Since O L \mathcal{O}_{L} O L and O R \mathcal{O}_{R} O R enter (A.11) symmetrically and with the same coefficient, L 4 \mathcal{L}_{4} L 4 is parity invariant for any value of G G G .
Charge conjugation. Similarly,
O L → C { ( q R c , i ) T C ( d R ) c , j } { ( i ∂ μ d R † ) a , j γ μ C ( q ˉ R T ) a , i } = { ( q R c , i ) T ( γ μ ) T ( q ˉ R ) a , i } { ( i ∂ μ d R † ) a , j ( d R ) c , j } = − { ( q ˉ R ) a , i γ μ ( q R ) c , i } { ( i ∂ μ d R † ) a , j ( d R ) c , j } = { ( q ˉ R ) a , i γ μ ( − i ∂ μ d R † ) a , j } { ( d R ) c , j ( q R ) c , i } = O R † . (115) \mathcal{O}_{L}\xrightarrow{C}\left\{\left(q_{R}^{c,i}\right)^{T}C\left(d_{R}\right)_{c,j}\right\}\left\{\left(i\partial_{\mu}d_{R}^{\dagger}\right)^{a,j}\gamma^{\mu}C\left(\bar{q}_{R}^{T}\right)_{a,i}\right\}
=\left\{\left(q_{R}^{c,i}\right)^{T}\left(\gamma^{\mu}\right)^{T}\left(\bar{q}_{R}\right)_{a,i}\right\}\left\{\left(i\partial_{\mu}d_{R}^{\dagger}\right)^{a,j}\left(d_{R}\right)_{c,j}\right\}
=-\left\{\left(\bar{q}_{R}\right)_{a,i}\gamma^{\mu}\left(q_{R}\right)^{c,i}\right\}\left\{\left(i\partial_{\mu}d_{R}^{\dagger}\right)^{a,j}\left(d_{R}\right)_{c,j}\right\}
=\left\{\left(\bar{q}_{R}\right)_{a,i}\gamma^{\mu}\left(-i\partial_{\mu}d_{R}^{\dagger}\right)^{a,j}\right\}\left\{\left(d_{R}\right)_{c,j}\left(q_{R}\right)^{c,i}\right\}=\mathcal{O}_{R}^{\dagger}. \tag*{(115)} O L C { ( q R c , i ) T C ( d R ) c , j } { ( i ∂ μ d R † ) a , j γ μ C ( q ˉ R T ) a , i } = { ( q R c , i ) T ( γ μ ) T ( q ˉ R ) a , i } { ( i ∂ μ d R † ) a , j ( d R ) c , j } = − { ( q ˉ R ) a , i γ μ ( q R ) c , i } { ( i ∂ μ d R † ) a , j ( d R ) c , j } = { ( q ˉ R ) a , i γ μ ( − i ∂ μ d R † ) a , j } { ( d R ) c , j ( q R ) c , i } = O R † . (115)
In the second line we used C γ μ C = ( γ μ ) T C\gamma^\mu C = (\gamma^\mu)^{T} C γ μ C = ( γ μ ) T and regrouped the (bosonic) diquark fields; in the third line we transposed the spinor bilinear, which produces a minus sign because the quark fields anticommute. The final expression is the hermitian conjugate of O R \mathcal{O}_{R} O R , as follows from (A.10) using γ 0 ( γ μ ) † γ 0 = γ μ \gamma^0(\gamma^\mu)^\dagger\gamma^0=\gamma^\mu γ 0 ( γ μ ) † γ 0 = γ μ . The analogous computation for O R \mathcal{O}_{R} O R then yields
C : O L → O R † , O R → O L † . (116) C:\quad\mathcal{O}_{L}\to\mathcal{O}_{R}^{\dagger},\qquad\mathcal{O}_{R}\to\mathcal{O}_{L}^{\dagger}. \tag*{(116)} C : O L → O R † , O R → O L † . (116)
Reality of G G G . Applying (A.15) to the Lagrangian (A.11),
G O L + G ∗ O L † + G O R + G ∗ O R † → C G O R † + G ∗ O R + G O L † + G ∗ O L = ∑ χ = L , R ( G ∗ O χ + G O χ † ) . (117) G\mathcal{O}_{L}+G^{*}\mathcal{O}_{L}^{\dagger}+G\mathcal{O}_{R}+G^{*}\mathcal{O}_{R}^{\dagger}\xrightarrow{C}G\mathcal{O}_{R}^{\dagger}+G^{*}\mathcal{O}_{R}+G\mathcal{O}_{L}^{\dagger}+G^{*}\mathcal{O}_{L}
=\sum_{\chi=L,R}\left(G^{*}\mathcal{O}_{\chi}+G\mathcal{O}_{\chi}^{\dagger}\right). \tag*{(117)} G O L + G ∗ O L † + G O R + G ∗ O R † C G O R † + G ∗ O R + G O L † + G ∗ O L = χ = L , R ∑ ( G ∗ O χ + G O χ † ) . (117)
Since O L \mathcal{O}_{L} O L , O R \mathcal{O}_{R} O R , O L † \mathcal{O}_{L}^{\dagger} O L † , and O R † \mathcal{O}_{R}^{\dagger} O R † are independent operator structures, invariance of L 4 \mathcal{L}_{4} L 4 under charge conjugation requires
G = G ∗ . (118) G=G^{*}. \tag*{(118)} G = G ∗ . (118)
The coupling constant G G G is therefore real, and with real G G G the four-body interaction (12) is invariant under both P P P and C C C , and hence under C P CP C P as well.
Evaluation of the two-point functions
In this appendix we evaluate the one-loop quark–diquark two-point functions for a general quark mass M q M_{q} M q and external momentum P P P , and then take the massless-quark and zero-momentum limits used in the matching of Sec. 4. The general expressions are also needed once the quarks acquire the constituent mass M q = g q ⟨ σ ⟩ M_{q}=g_{q}\langle\sigma\rangle M q = g q ⟨ σ ⟩ through the coupling to Σ \Sigma Σ , and may be of use in future in-medium applications of the model.
Loop functions
In the rest frame P = ( W , 0 ⃗ ) P=(W,\vec{0}) P = ( W , 0 ) the denominators factorize in k 0 k^{0} k 0 as
k 2 − M q 2 + i ϵ = ( k 0 − E q + i ϵ ) ( k 0 + E q − i ϵ ) , k^{2}-M_{q}^{2}+i\epsilon=(k^{0}-E_{q}+i\epsilon)(k^{0}+E_{q}-i\epsilon), k 2 − M q 2 + i ϵ = ( k 0 − E q + i ϵ ) ( k 0 + E q − i ϵ ) ,
( P − k ) 2 − m d 2 + i ϵ = ( k 0 − W + E d − i ϵ ) ( k 0 − W − E d + i ϵ ) . (119) (P-k)^{2}-m_{d}^{2}+i\epsilon=(k^{0}-W+E_{d}-i\epsilon)(k^{0}-W-E_{d}+i\epsilon). \tag*{(119)} ( P − k ) 2 − m d 2 + i ϵ = ( k 0 − W + E d − i ϵ ) ( k 0 − W − E d + i ϵ ) . (119)
with E q = k ⃗ 2 + M q 2 E_{q}=\sqrt{\vec{k}^{2}+M_{q}^{2}} E q = k 2 + M q 2 and E d = k ⃗ 2 + m d 2 E_{d}=\sqrt{\vec{k}^{2}+m_{d}^{2}} E d = k 2 + m d 2 . Closing the k 0 k^{0} k 0 contour in the lower half plane picks up the poles k 0 = E q − i ϵ k^{0}=E_{q}-i\epsilon k 0 = E q − i ϵ and k 0 = W + E d − i ϵ k^{0}=W+E_{d}-i\epsilon k 0 = W + E d − i ϵ , and ∫ d 3 k ( 2 π ) 3 = 1 2 π 2 ∫ 0 Λ d k k 2 \int\frac{d^{3}k}{(2\pi)^{3}}=\frac{1}{2\pi^{2}}\int_{0}^{\Lambda}dk\,k^{2} ∫ ( 2 π ) 3 d 3 k = 2 π 2 1 ∫ 0 Λ d k k 2 with the three-momentum cutoff Λ \Lambda Λ . Integrands that fall off only linearly in k 0 k^{0} k 0 are defined by the symmetric (principal-value) k 0 k^{0} k 0 integration; with this prescription odd integrals vanish,
i\int\frac{d^{4}k}{(2\pi)^{4}}\frac{\slashed{k}}{k^{2}-m^{2}}=0,
and equivalently
i\int\frac{d^{4}k}{(2\pi)^{4}}\frac{\slashed{k}}{(P-k)^{2}-m_{d}^{2}}=\slashed{P}\,T_{d}. \tag*{(120)}
as follows from the shift k → P − k k\to P-k k → P − k , which leaves the three-momentum cutoff unchanged in the rest frame.
Tadpoles.
T d = i ∫ d 4 k ( 2 π ) 4 1 ( P − k ) 2 − m d 2 + i ϵ = i ∫ d 3 k ( 2 π ) 3 ∫ d k 0 2 π 1 ( W − k 0 ) 2 − E d 2 + i ϵ = ∫ d 3 k ( 2 π ) 3 1 2 E d = 1 2 π 2 ∫ 0 Λ d k k 2 2 E d 2 , (121) \begin{aligned}
T_d &= i \int\frac{d^4 k}{(2\pi)^4}\frac{1}{(P-k)^2-m_d^2+i\epsilon} \\
&= i \int\frac{d^3 k}{(2\pi)^3}\int\frac{dk^0}{2\pi}\frac{1}{(W-k^0)^2-E_d^2+i\epsilon} \\
&= \int\frac{d^3 k}{(2\pi)^3}\frac{1}{2E_d}
= \frac{1}{2\pi^2}\int_0^\Lambda dk\,\frac{k^2}{2E_d^2}, \tag*{(121)}
\end{aligned} T d = i ∫ ( 2 π ) 4 d 4 k ( P − k ) 2 − m d 2 + i ϵ 1 = i ∫ ( 2 π ) 3 d 3 k ∫ 2 π d k 0 ( W − k 0 ) 2 − E d 2 + i ϵ 1 = ∫ ( 2 π ) 3 d 3 k 2 E d 1 = 2 π 2 1 ∫ 0 Λ d k 2 E d 2 k 2 , (121)
independently of W W W , and analogously
T q M q = i ∫ d 4 k ( 2 π ) 4 1 k 2 − M q 2 + i ϵ = 1 2 π 2 ∫ 0 Λ d k k 2 2 E q . (122) T_q^{M^q}=i\int\frac{d^4 k}{(2\pi)^4}\frac{1}{k^2-M_q^2+i\epsilon}
=\frac{1}{2\pi^2}\int_0^\Lambda dk\,\frac{k^2}{2E_q}. \tag*{(122)} T q M q = i ∫ ( 2 π ) 4 d 4 k k 2 − M q 2 + i ϵ 1 = 2 π 2 1 ∫ 0 Λ d k 2 E q k 2 . (122)
Scalar function Π S \Pi_S Π S . The two residues give
Π S = i ∫ d 4 k ( 2 π ) 4 1 [ k 2 − M q 2 ] [ ( P − k ) 2 − m d 2 ] = ∫ d 3 k ( 2 π ) 3 [ 1 2 E q [ ( W − E q ) 2 − E d 2 ] + 1 2 E d [ ( W + E d ) 2 − E q 2 ] ] = − 1 2 π 2 ∫ 0 Λ d k k 2 ( E q + E d ) 2 E q E d [ ( E q + E d ) 2 − W 2 ] . (123) \begin{aligned}
\Pi_S &= i\int\frac{d^4 k}{(2\pi)^4}\frac{1}{[k^2-M_q^2][(P-k)^2-m_d^2]} \\
&= \int\frac{d^3 k}{(2\pi)^3}\left[\frac{1}{2E_q[(W-E_q)^2-E_d^2]}+\frac{1}{2E_d[(W+E_d)^2-E_q^2]}\right] \\
&= -\frac{1}{2\pi^2}\int_0^\Lambda dk\,\frac{k^2(E_q+E_d)}{2E_qE_d[(E_q+E_d)^2-W^2]}. \tag*{(123)}
\end{aligned} Π S = i ∫ ( 2 π ) 4 d 4 k [ k 2 − M q 2 ] [( P − k ) 2 − m d 2 ] 1 = ∫ ( 2 π ) 3 d 3 k [ 2 E q [( W − E q ) 2 − E d 2 ] 1 + 2 E d [( W + E d ) 2 − E q 2 ] 1 ] = − 2 π 2 1 ∫ 0 Λ d k 2 E q E d [( E q + E d ) 2 − W 2 ] k 2 ( E q + E d ) . (123)
The last line uses, with a ≡ E q a\equiv E_q a ≡ E q , b ≡ E d b\equiv E_d b ≡ E d , and the common factor ( W + b − a ) (W+b-a) ( W + b − a ) ,
1 2 a [ ( W − a ) 2 − b 2 ] + 1 2 b [ ( W + b ) 2 − a 2 ] = 1 2 ( W + b − a ) [ − 1 a ( a + b − W ) + 1 b ( a + b + W ) ] = 1 2 ( W + b − a ) − b ( a + b + W ) + a ( a + b − W ) a b ( a + b − W ) ( a + b + W ) = 1 2 ( W + b − a ) ( a + b ) ( a − b − W ) a b [ ( a + b ) 2 − W 2 ] = − a + b 2 a b [ ( a + b ) 2 − W 2 ] , (124) \begin{aligned}
&\frac{1}{2a[(W-a)^2-b^2]}+\frac{1}{2b[(W+b)^2-a^2]} \\
&=\frac{1}{2(W+b-a)}\left[\frac{-1}{a(a+b-W)}+\frac{1}{b(a+b+W)}\right] \\
&=\frac{1}{2(W+b-a)}\frac{-b(a+b+W)+a(a+b-W)}{ab(a+b-W)(a+b+W)} \\
&=\frac{1}{2(W+b-a)}\frac{(a+b)(a-b-W)}{ab[(a+b)^2-W^2]}
=-\frac{a+b}{2ab[(a+b)^2-W^2]}, \tag*{(124)}
\end{aligned} 2 a [( W − a ) 2 − b 2 ] 1 + 2 b [( W + b ) 2 − a 2 ] 1 = 2 ( W + b − a ) 1 [ a ( a + b − W ) − 1 + b ( a + b + W ) 1 ] = 2 ( W + b − a ) 1 ab ( a + b − W ) ( a + b + W ) − b ( a + b + W ) + a ( a + b − W ) = 2 ( W + b − a ) 1 ab [( a + b ) 2 − W 2 ] ( a + b ) ( a − b − W ) = − 2 ab [( a + b ) 2 − W 2 ] a + b , (124)
where the last step follows from ( a − b − W ) / ( W + b − a ) = − 1 (a-b-W)/(W+b-a)=-1 ( a − b − W ) / ( W + b − a ) = − 1 .
Vector function Π V \Pi_V Π V . The vector loop function is defined by
P̸ Π V = i ∫ d 4 k ( 2 π ) 4 k̸ [ k 2 − M q 2 ] [ ( P − k ) 2 − m d 2 ] . (125) \not{P}\Pi_V=i\int\frac{d^4 k}{(2\pi)^4}\frac{\not{k}}{[k^2-M_q^2][(P-k)^2-m_d^2]}. \tag*{(125)} P Π V = i ∫ ( 2 π ) 4 d 4 k [ k 2 − M q 2 ] [( P − k ) 2 − m d 2 ] k . (125)
Contracting with P̸ \not{P} P and taking the Dirac trace, tr [ P̸ k̸ ] = 4 P ⋅ k \operatorname{tr}[\not{P}\not{k}]=4P\cdot k tr [ P k ] = 4 P ⋅ k , and using 2 P ⋅ k = k 2 − ( P − k ) 2 + P 2 2P\cdot k=k^2-(P-k)^2+P^2 2 P ⋅ k = k 2 − ( P − k ) 2 + P 2 ,
Π V = 1 4 P 2 i ∫ d 4 k ( 2 π ) 4 tr [ P̸ k̸ ] [ k 2 − M q 2 ] [ ( P − k ) 2 − m d 2 ] = 1 2 P 2 [ T d − T q + ( P 2 + M q 2 − m d 2 ) Π S ] . (126) \begin{aligned}
\Pi_V&=\frac{1}{4P^2}i\int\frac{d^4 k}{(2\pi)^4}\frac{\operatorname{tr}[\not{P}\not{k}]}{[k^2-M_q^2][(P-k)^2-m_d^2]} \\
&=\frac{1}{2P^2}\left[T_d-T_q+(P^2+M_q^2-m_d^2)\Pi_S\right]. \tag*{(126)}
\end{aligned} Π V = 4 P 2 1 i ∫ ( 2 π ) 4 d 4 k [ k 2 − M q 2 ] [( P − k ) 2 − m d 2 ] tr [ P k ] = 2 P 2 1 [ T d − T q + ( P 2 + M q 2 − m d 2 ) Π S ] . (126)
where we used
i ∫ d 4 k ( 2 π ) 4 k 2 [ k 2 − M q 2 ] [ ( P − k ) 2 − m d 2 ] = T d + M q 2 Π S , i ∫ d 4 k ( 2 π ) 4 ( P − k ) 2 [ k 2 − M q 2 ] [ ( P − k ) 2 − m d 2 ] = T q + m d 2 Π S . (127) i\int\frac{d^4k}{(2\pi)^4}\frac{k^2}{[k^2-M_q^2][(P-k)^2-m_d^2]}=T_d+M_q^2\Pi_S,
\\
i\int\frac{d^4k}{(2\pi)^4}\frac{(P-k)^2}{[k^2-M_q^2][(P-k)^2-m_d^2]}=T_q+m_d^2\Pi_S. \tag*{(127)} i ∫ ( 2 π ) 4 d 4 k [ k 2 − M q 2 ] [( P − k ) 2 − m d 2 ] k 2 = T d + M q 2 Π S , i ∫ ( 2 π ) 4 d 4 k [ k 2 − M q 2 ] [( P − k ) 2 − m d 2 ] ( P − k ) 2 = T q + m d 2 Π S . (127)
Equivalently, the direct rest-frame evaluation of the γ 0 \gamma^0 γ 0 component of Eq. (B.7) gives
Π V = 1 W ∫ d 3 k ( 2 π ) 3 [ 1 2 [ ( W − E q ) 2 − E d 2 ] + W + E d 2 E d [ ( W + E d ) 2 − E q 2 ] ] = − 1 2 π 2 ∫ 0 Λ d k k 2 2 E d [ ( E q + E d ) 2 − W 2 ] . (128) \begin{aligned}
\Pi_V=\frac{1}{W}\int\frac{d^3k}{(2\pi)^3}\left[\frac{1}{2[(W-E_q)^2-E_d^2]}+\frac{W+E_d}{2E_d[(W+E_d)^2-E_q^2]}\right] \\
&=-\frac{1}{2\pi^2}\int_0^\Lambda dk\frac{k^2}{2E_d[(E_q+E_d)^2-W^2]}. \tag*{(128)}
\end{aligned} Π V = W 1 ∫ ( 2 π ) 3 d 3 k [ 2 [( W − E q ) 2 − E d 2 ] 1 + 2 E d [( W + E d ) 2 − E q 2 ] W + E d ] = − 2 π 2 1 ∫ 0 Λ d k 2 E d [( E q + E d ) 2 − W 2 ] k 2 . (128)
where the bracket is reduced by the same manipulation as in Eq. (B.6), now with numerators 1 2 \frac{1}{2} 2 1 and W + b W+b W + b , which produce the numerator W ( a − b − W ) W(a-b-W) W ( a − b − W ) and hence the factor W W W that cancels the 1 / W 1/W 1/ W :
1 2 [ ( W − a ) 2 − b 2 ] + W + b 2 b [ ( W + b ) 2 − a 2 ] = − W 2 b [ ( a + b ) 2 − W 2 ] . (129) \frac{1}{2[(W-a)^2-b^2]}+\frac{W+b}{2b[(W+b)^2-a^2]}=-\frac{W}{2b[(a+b)^2-W^2]}. \tag*{(129)} 2 [( W − a ) 2 − b 2 ] 1 + 2 b [( W + b ) 2 − a 2 ] W + b = − 2 b [( a + b ) 2 − W 2 ] W . (129)
Note that Π V \Pi_V Π V is finite and negative at W 2 = 0 W^2=0 W 2 = 0 .
General two-point functions
For a quark line of momentum k k k and mass M q M_q M q and a diquark line of momentum q d = P − k q_d=P-k q d = P − k , the one-loop two-point function in vertex space reads
Π a b ( P ) = i ∑ χ a , χ b = L , R ∫ d 4 k ( 2 π ) 4 f a g a P ^ χ a k̸ + M q k 2 − M q 2 P ^ χ b f b g b 1 ( P − k ) 2 − m d 2 , a , b = 1 , 2. (130) \Pi_{ab}(P)=i\sum_{\chi_a,\chi_b=L,R}\int\frac{d^4k}{(2\pi)^4}f_a g_a\widehat{P}_{\chi_a}\frac{\not{k}+M_q}{k^2-M_q^2}\widehat{P}_{\chi_b}f_b g_b\frac{1}{(P-k)^2-m_d^2},\qquad a,b=1,2. \tag*{(130)} Π ab ( P ) = i χ a , χ b = L , R ∑ ∫ ( 2 π ) 4 d 4 k f a g a P χ a k 2 − M q 2 k + M q P χ b f b g b ( P − k ) 2 − m d 2 1 , a , b = 1 , 2. (130)
with the vertex factors f 1 , 2 f_{1,2} f 1 , 2 and couplings of Eq. (??). In the chirality-conserving components evaluated below, the two chiral projectors enforce a chirality flip through the propagator, so that the M q M_q M q piece of the numerator drops out; the chirality-violating components, proportional to M q M_q M q , are the origin of the Σ \Sigma Σ -induced mixings of Sec. 5.
Π 22 \Pi_{22} Π 22 component. With the trivial vertex f 2 f_2 f 2 on both sides, Eq. (B.7) directly gives
Π 22 = P̸ Π V . (131) \Pi_{22}=\not{P}\Pi_V. \tag*{(131)} Π 22 = P Π V . (131)
Π 12 = Π 21 \Pi_{12}=\Pi_{21} Π 12 = Π 21 component. With the derivative vertex f 1 f_1 f 1 on one side, the numerator reduces via ̸ q d k̸ = ( P̸ − k̸ ) k̸ = P̸ k̸ − k 2 \not{q}_d\not{k}=(\not{P}-\not{k})\not{k}=\not{P}\not{k}-k^2 q d k = ( P − k ) k = P k − k 2 , and Eqs. (B.7) and (B.9) yield
Π 12 = C q d G q d ( P 2 Π V − T d − M q 2 Π S ) . (132) \Pi_{12}=C_{q_d}G_{q_d}\left(P^2\Pi_V-T_d-M_q^2\Pi_S\right). \tag*{(132)} Π 12 = C q d G q d ( P 2 Π V − T d − M q 2 Π S ) . (132)
Π 11 \Pi_{11} Π 11 component. With f 1 f_1 f 1 on both sides, we use ̸ q d k̸ ̸ q d = 2 ( a ⋅ b ) ̸ q d − a 2 k̸ \not q_d\not k\not q_d=2(a\cdot b)\not q_d-a^2\not k q d k q d = 2 ( a ⋅ b ) q d − a 2 k with a = q d a=q_d a = q d , b = k b=k b = k , together with q d 2 = P 2 − 2 P ⋅ k + k 2 q_d^2=P^2-2P\cdot k+k^2 q d 2 = P 2 − 2 P ⋅ k + k 2 and q d ⋅ k = P ⋅ k − k 2 q_d\cdot k=P\cdot k-k^2 q d ⋅ k = P ⋅ k − k 2 , to obtain
̸ q d k̸ ̸ q d = 2 ( P ⋅ k − k 2 ) P̸ + ( k 2 − P 2 ) k̸ . (133) \not q_d\not k\not q_d=2(P\cdot k-k^2)\not P+(k^2-P^2)\not k. \tag*{(133)} q d k q d = 2 ( P ⋅ k − k 2 ) P + ( k 2 − P 2 ) k . (133)
The first structure integrates, using i ∫ d 4 k ( 2 π ) 4 ( P ⋅ k ) / [ ⋯ ] = P 2 Π V i\int\frac{d^4k}{(2\pi)^4}(P\cdot k)/[\cdots]=P^2\Pi_V i ∫ ( 2 π ) 4 d 4 k ( P ⋅ k ) / [ ⋯ ] = P 2 Π V [which follows from Eq. (B.8)] and Eq. (B.9)], to
2 P̸ i ∫ d 4 k ( 2 π ) 4 P ⋅ k − k 2 [ k 2 − M q 2 ] [ ( P − k ) 2 − m d 2 ] = 2 P̸ [ P 2 Π V − T d − M q 2 Π S ] . (134) 2\not P i\int\frac{d^4k}{(2\pi)^4}\frac{P\cdot k-k^2}{[k^2-M_q^2][(P-k)^2-m_d^2]}=2\not P\left[P^2\Pi_V-T_d-M_q^2\Pi_S\right]. \tag*{(134)} 2 P i ∫ ( 2 π ) 4 d 4 k [ k 2 − M q 2 ] [( P − k ) 2 − m d 2 ] P ⋅ k − k 2 = 2 P [ P 2 Π V − T d − M q 2 Π S ] . (134)
For the second structure, splitting k 2 − P 2 = [ ( P − k ) 2 − m d 2 ] + M q 2 − P 2 k^2-P^2=[(P-k)^2-m_d^2]+M_q^2-P^2 k 2 − P 2 = [( P − k ) 2 − m d 2 ] + M q 2 − P 2 [valid inside the 1 / ( k 2 − M q 2 ) 1/(k^2-M_q^2) 1/ ( k 2 − M q 2 ) integrand after adding and subtracting M q 2 M_q^2 M q 2 ] and using Eq. (B.2),
i ∫ d 4 k ( 2 π ) 4 ( k 2 − P 2 ) k̸ [ k 2 − M q 2 ] [ ( P − k ) 2 − m d 2 ] = P̸ T d + ( M q 2 − P 2 ) P̸ Π V . (135) i\int\frac{d^4k}{(2\pi)^4}\frac{(k^2-P^2)\not k}{[k^2-M_q^2][(P-k)^2-m_d^2]}=\not P T_d+(M_q^2-P^2)\not P\Pi_V. \tag*{(135)} i ∫ ( 2 π ) 4 d 4 k [ k 2 − M q 2 ] [( P − k ) 2 − m d 2 ] ( k 2 − P 2 ) k = P T d + ( M q 2 − P 2 ) P Π V . (135)
Adding the two contributions,
Π 11 = ( C q d G q d ) 2 P̸ [ ( P 2 + M q 2 ) Π V − 2 M q 2 Π S − T d ] . (136) \Pi_{11}=(C_{qd}G_{qd})^2\not P\left[(P^2+M_q^2)\Pi_V-2M_q^2\Pi_S-T_d\right]. \tag*{(136)} Π 11 = ( C q d G q d ) 2 P [ ( P 2 + M q 2 ) Π V − 2 M q 2 Π S − T d ] . (136)
Massless-quark limit and zero-momentum matching
For M q = 0 M_q=0 M q = 0 one has E q = k E_q=k E q = k , and the loop functions reduce to
T q = T q M q ∣ M q = 0 = 1 2 π 2 ∫ 0 Λ d k k 2 = Λ 2 8 π 2 , Π V ∣ M q = 0 = − 1 2 π 2 ∫ 0 Λ d k k 2 2 E d [ ( k + E d ) 2 − W 2 ] . (137) T_q=T_q^{M_q}\big|_{M_q=0}=\frac{1}{2\pi^2}\int_0^\Lambda dk\,\frac{k}{2}=\frac{\Lambda^2}{8\pi^2},\qquad\Pi_V\big|_{M_q=0}=-\frac{1}{2\pi^2}\int_0^\Lambda dk\,\frac{k^2}{2E_d[(k+E_d)^2-W^2]}. \tag*{(137)} T q = T q M q M q = 0 = 2 π 2 1 ∫ 0 Λ d k 2 k = 8 π 2 Λ 2 , Π V M q = 0 = − 2 π 2 1 ∫ 0 Λ d k 2 E d [( k + E d ) 2 − W 2 ] k 2 . (137)
while T d T_d T d is unchanged. Since Π S \Pi_S Π S remains finite as M q → 0 M_q\to0 M q → 0 , all terms proportional to M q 2 M_q^2 M q 2 in Eqs. (B.14) and (B.18) drop out,
Π 11 → ( C q d G q d ) 2 P̸ ( P 2 Π V − T d ) , Π 12 → C q d G q d ( P 2 Π V − T d ) , Π 22 = P̸ Π V , (138) \Pi_{11}\to(C_{qd}G_{qd})^2\not P(P^2\Pi_V-T_d),\qquad\Pi_{12}\to C_{qd}G_{qd}(P^2\Pi_V-T_d),\qquad\Pi_{22}=\not P\Pi_V, \tag*{(138)} Π 11 → ( C q d G q d ) 2 P ( P 2 Π V − T d ) , Π 12 → C q d G q d ( P 2 Π V − T d ) , Π 22 = P Π V , (138)
and the further limit P 2 → 0 P^2\to0 P 2 → 0 reproduces the values used in the matching of Sec. 4,
Π 11 → − ( C q d G q d ) 2 T d P̸ , Π 12 → − C q d G q d T d , Π 22 → P̸ Π V ∣ W 2 = 0 . (139) \Pi_{11}\to-(C_{qd}G_{qd})^2T_d\not P,\qquad\Pi_{12}\to-C_{qd}G_{qd}T_d,\qquad\Pi_{22}\to\not P\Pi_V\big|_{W^2=0}. \tag*{(139)} Π 11 → − ( C q d G q d ) 2 T d P , Π 12 → − C q d G q d T d , Π 22 → P Π V W 2 = 0 . (139)
Finally, we collect the auxiliary integrals appearing in the Σ \Sigma Σ -induced mixings of Sec. 5. The diquark bubble L d L_d L d of Eq. (53) follows from the tadpole by differentiation,
L d = i ∫ d 4 k ( 2 π ) 4 ( 1 k 2 − m d 2 ) 2 = ∂ T d ∂ m d 2 = − 1 2 π 2 ∫ 0 Λ d k k 2 4 E d 3 , (140) L_d=i\int\frac{d^4k}{(2\pi)^4}\left(\frac{1}{k^2-m_d^2}\right)^2=\frac{\partial T_d}{\partial m_d^2}=-\frac{1}{2\pi^2}\int_0^\Lambda dk\,\frac{k^2}{4E_d^3}, \tag*{(140)} L d = i ∫ ( 2 π ) 4 d 4 k ( k 2 − m d 2 1 ) 2 = ∂ m d 2 ∂ T d = − 2 π 2 1 ∫ 0 Λ d k 4 E d 3 k 2 , (140)
and the combination entering Eq. (57) is
i ∫ d 4 k ( 2 π ) 4 1 k 2 [ k 2 − m d 2 ] = T d − T q m d 2 = 1 2 π 2 ∫ 0 Λ d k m d 2 ( k 2 2 E d − k 2 ) , (141) i\int\frac{d^4k}{(2\pi)^4}\frac{1}{k^2[k^2-m_d^2]}=\frac{T_d-T_q}{m_d^2}=\frac{1}{2\pi^2}\int_0^\Lambda\frac{dk}{m_d^2}\left(\frac{k^2}{2E_d}-\frac{k}{2}\right), \tag*{(141)} i ∫ ( 2 π ) 4 d 4 k k 2 [ k 2 − m d 2 ] 1 = m d 2 T d − T q = 2 π 2 1 ∫ 0 Λ m d 2 d k ( 2 E d k 2 − 2 k ) , (141)
which is only logarithmically divergent, the quadratic divergences of the two tadpoles cancelling in the difference.