Drastic magnetic-field-induced chiral current order and emergent current-bond-field interplay in kagome metal AV3Sb5 (A=Cs,Rb,K)

Rina Tazai1, Youichi Yamakawa2, and Hiroshi Kontani2 000βˆ—These authors made equal contribution to the work. 1Yukawa Institute for Theoretical Physics, Kyoto University, Kyoto 606-8502, Japan
2 Department of Physics, Nagoya University, Furo-cho, Nagoya 464-8602, Japan.
(May 2, 2024)
Abstract

In kagome metals, the chiral current order parameter 𝜼𝜼{\bm{\eta}}bold_italic_Ξ· with time-reversal-symmetry-breaking is the source of various exotic electronic states, while the method of controlling the current order and its interplay with the star-of-David bond order Ο•bold-italic-Ο•{\bm{\phi}}bold_italic_Ο• are still unsolved. Here, we reveal that tiny uniform orbital magnetization Morb⁒[𝜼,Ο•]subscript𝑀orb𝜼bold-italic-Ο•M_{\rm orb}[{\bm{\eta}},{\bm{\phi}}]italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT [ bold_italic_Ξ· , bold_italic_Ο• ] is induced by the chiral current order, and its magnitude is prominently enlarged under the presence of the bond order. Importantly, we derive the magnetic-field (hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT)-induced Ginzburg-Landau (GL) free energy expression Δ⁒F⁒[hz,𝜼,Ο•]βˆβˆ’hz⁒Morb⁒[𝜼,Ο•]proportional-toΔ𝐹subscriptβ„Žπ‘§πœΌbold-italic-Ο•subscriptβ„Žπ‘§subscript𝑀orb𝜼bold-italic-Ο•\Delta F[h_{z},{\bm{\eta}},{\bm{\phi}}]\propto-h_{z}M_{\rm orb}[{\bm{\eta}},{% \bm{\phi}}]roman_Ξ” italic_F [ italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT , bold_italic_Ξ· , bold_italic_Ο• ] ∝ - italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT [ bold_italic_Ξ· , bold_italic_Ο• ], which enables us to elucidate the field-induced current-bond phase transitions in kagome metals. The emergent current-bond-hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT trilinear coupling term in the free energy, βˆ’3⁒m1⁒hzβ’πœΌβ‹…Ο•β‹…3subscriptπ‘š1subscriptβ„Žπ‘§πœΌbold-italic-Ο•-3m_{1}h_{z}{\bm{\eta}}\cdot{\bm{\phi}}- 3 italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT bold_italic_Ξ· β‹… bold_italic_Ο•, naturally explains the characteristic magnetic field sensitive electronic states in kagome metals, such as the field-induced current order and the strong interplay between the bond and current orders. The GL coefficients of Δ⁒F⁒[hz,𝜼,Ο•]Δ𝐹subscriptβ„Žπ‘§πœΌbold-italic-Ο•\Delta F[h_{z},{\bm{\eta}},{\bm{\phi}}]roman_Ξ” italic_F [ italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT , bold_italic_Ξ· , bold_italic_Ο• ] derived from the realistic multiorbital model are appropriate to explain various experiments. Furthermore, we present a natural explanation for the drastic strain-induced increment of the current order transition temperature Tcsubscript𝑇cT_{\rm c}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT reported by a recent experiment.

Introduction

Recent discovery of unconventional quantum phases in metals has led to a new trend of condensed matter physics. Exotic charge-density-wave orders and unconventional superconductivity in geometrically frustrated kagome metal AV3Sb5 (A=Cs,Rb,K) have been attracting increasing attention kagome-exp1 ; kagome-exp2 . The 2Γ—\timesΓ—2 (inverse) star-of-David order, which is presumably the triple-𝒒𝒒{{\bm{q}}}bold_italic_q (3⁒Q3𝑄3Q3 italic_Q) bond order (BO), occurs at TBOβ‰ˆ100subscript𝑇BO100T_{\rm BO}\approx 100italic_T start_POSTSUBSCRIPT roman_BO end_POSTSUBSCRIPT β‰ˆ 100 K at ambient pressure STM1 ; STM2 . The BO is the time-reversal-symmetry (TRS) preserving modulation in the hopping integral, δ⁒ti⁒jb=real𝛿superscriptsubscript𝑑𝑖𝑗breal\delta t_{ij}^{\rm b}={\rm real}italic_Ξ΄ italic_t start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_b end_POSTSUPERSCRIPT = roman_real. Thomale2013 ; SMFRG ; Thomale2021 ; Neupert2021 ; Balents2021 ; Nandkishore ; Tazai-kagome . Below TBOsubscript𝑇BOT_{\rm BO}italic_T start_POSTSUBSCRIPT roman_BO end_POSTSUBSCRIPT, nodeless superconductivity occurs for A=Cs Roppongi ; SC2 , which is naturally explained based on the BO fluctuation mechanism proposed in Ref. Tazai-kagome .

In kagome metals, unusual TRS breaking (TRSB) phase without long-range spin orders has been reported by ΞΌπœ‡\muitalic_ΞΌ-SR muSR3-Cs ; muSR2-K ; muSR4-Cs ; muSR5-Rb , Kerr rotation birefringence-kagome ; Yonezawa , field-tuned chiral transport eMChA measurements and STM studies under magnetic field STM1 ; eMChA . The transition temperature TTRSBsubscript𝑇TRSBT_{\rm TRSB}italic_T start_POSTSUBSCRIPT roman_TRSB end_POSTSUBSCRIPT is still under debate. Although it is close to TBOsubscript𝑇BOT_{\rm BO}italic_T start_POSTSUBSCRIPT roman_BO end_POSTSUBSCRIPT in many experiments, the TRSB order parameter is strongly magnified at Tβˆ—β‰ˆ35superscript𝑇35T^{*}\approx 35italic_T start_POSTSUPERSCRIPT βˆ— end_POSTSUPERSCRIPT β‰ˆ 35K for A=Cs eMChA ; muSR3-Cs ; muSR4-Cs and Tβˆ—β‰ˆ50superscript𝑇50T^{*}\approx 50italic_T start_POSTSUPERSCRIPT βˆ— end_POSTSUPERSCRIPT β‰ˆ 50K for A=Rb muSR5-Rb . Recently, magnetic torque measurement reveals the nematic order with TRSB at Tβˆ—βˆ—β‰ˆ130superscript𝑇absent130T^{**}\approx 130italic_T start_POSTSUPERSCRIPT βˆ— βˆ— end_POSTSUPERSCRIPT β‰ˆ 130K Asaba , while no TRSB was reported by recent Kerr rotation study Kapitulnik . A natural candidate is the correlation driven TRSB hopping integral modulation: δ⁒ti⁒jc=imaginary𝛿superscriptsubscript𝑑𝑖𝑗cimaginary\delta t_{ij}^{\rm c}={\rm imaginary}italic_Ξ΄ italic_t start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_c end_POSTSUPERSCRIPT = roman_imaginary. This order accompanies topological charge-current Haldane that gives the giant anomalous Hall effect (AHE) AHE1 ; AHE2 .

Theoretically, the BO and the current order emerge in the presence of sizable off-site electron correlations in Fe-based and cuprate superconductors Onari-SCVC ; Yamakawa-FeSe ; Onari-TBG ; Tsuchiizu1 ; Tsuchiizu4 ; Chubukov-PRX2016 ; Fernandes-rev2018 ; Kontani-AdvPhys ; Davis-rev2013 and in kagome metals Thomale2021 ; Neupert2021 ; Balents2021 ; Tazai-kagome ; Tazai-kagome2 ; PDW-kagome ; Fernandes-GL ; Thomale-GL ; Nat-g-ology . Notably, strong off-site interaction (due to the off-site Coulomb repulsion or the BO fluctuations) gives rise to the charge current order Neupert2021 ; Balents2021 ; Nandkishore ; Tazai-kagome2 . Based on the GL free-energy analysis, interesting bond+current nematic (C2subscript𝐢2C_{2}italic_C start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT) coexisting phases have been discussed in two-dimensional (2D) and three-dimensional (3D) models Balents2021 ; Tazai-kagome2 ; Fernandes-GL ; Thomale-GL ; Nat-g-ology . Experimentally, the nematic state is actually observed by the elastoresistance elastoresistance-kagome , the scanning birefringence birefringence-kagome , and the STM STM2 measurements.

In kagome metals, outer magnetic field hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT drastically modifies the electronic states. The chirality of the charge-current is aligned under very tiny |hz|∼1similar-tosubscriptβ„Žπ‘§1|h_{z}|\sim 1| italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT | ∼ 1 Tesla according to the measurements of AHE AHE1 ; AHE2 and field-tuned chiral transport eMChA . In addition, the amplitude of the loop current is strongly magnified by applying small hz(≳1⁒T⁒e⁒s⁒l⁒a)annotatedsubscriptβ„Žπ‘§greater-than-or-equivalent-toabsent1Teslah_{z}\ (\gtrsim 1{\rm Tesla})italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT ( ≳ 1 roman_T roman_e roman_s roman_l roman_a ) muSR4-Cs ; muSR2-K ; muSR5-Rb . Very recent transport measurement of highly symmetric fabricated CsV3Sb5 micro sample Moll-hz reveals that current-order state is drastically enlarged by the small hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT. These drastic hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-dependences are the hallmarks of the TRSB state, and it is important to understand the coupling between the current order, chirality and the magnetic field in kagome metals.

Refer to caption
Figure 1: (a) Kagome lattice tight-binding model. The 3⁒Q3𝑄3Q3 italic_Q BO Ο•=(Ο•,Ο•,Ο•)/3bold-italic-Ο•italic-Ο•italic-Ο•italic-Ο•3{\bm{\phi}}=(\phi,\phi,\phi)/\sqrt{3}bold_italic_Ο• = ( italic_Ο• , italic_Ο• , italic_Ο• ) / square-root start_ARG 3 end_ARG (Ο•>0)italic-Ο•0(\phi>0)( italic_Ο• > 0 ) is depicted. The dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital on each sublattice is drawn. (b) FS at the vHS filling nvHSsubscript𝑛vHSn_{\rm vHS}italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT. Three vHS points π’ŒAsubscriptπ’ŒA{{\bm{k}}}_{\rm A}bold_italic_k start_POSTSUBSCRIPT roman_A end_POSTSUBSCRIPT, π’ŒBsubscriptπ’ŒB{{\bm{k}}}_{\rm B}bold_italic_k start_POSTSUBSCRIPT roman_B end_POSTSUBSCRIPT and π’ŒCsubscriptπ’ŒC{{\bm{k}}}_{\rm C}bold_italic_k start_POSTSUBSCRIPT roman_C end_POSTSUBSCRIPT are respectively composed of A𝐴Aitalic_A (red), B𝐡Bitalic_B (blue), and C𝐢Citalic_C (green) orbitals. The inter-vHS nesting vectors 𝒒nsubscript𝒒𝑛{{\bm{q}}}_{n}bold_italic_q start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT (n=1βˆ’3𝑛13n=1-3italic_n = 1 - 3) are shown. (c) 3⁒Q3𝑄3Q3 italic_Q current order 𝜼=(Ξ·1,Ξ·2,Ξ·3)𝜼subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3{\bm{\eta}}=(\eta_{1},\eta_{2},\eta_{3})bold_italic_Ξ· = ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ) is depicted. Note that (Ξ·1,Ξ·2,Ξ·3)subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3(\eta_{1},\eta_{2},\eta_{3})( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ) and (Ξ·1,βˆ’Ξ·2,βˆ’Ξ·3)subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3(\eta_{1},-\eta_{2},-\eta_{3})( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , - italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , - italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ) are the same bulk states.

In this paper, we reveal that the 3⁒Q3𝑄3Q3 italic_Q loop-current order parameters accompany tiny orbital magnetization, Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT, and its magnitude is drastically enlarged under the presence of the bond order. Importantly, we derive the hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-induced Ginzburg-Landau (GL) free energy expression Δ⁒FΔ𝐹\Delta Froman_Ξ” italic_F, which is useful to study nontrivial phase transitions under the magnetic field. The emergent current-bond-hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT trilinear coupling term in Δ⁒FΔ𝐹\Delta Froman_Ξ” italic_F not only explains the origin of novel field-induced chiral symmetry breaking muSR4-Cs ; muSR2-K ; muSR5-Rb but also provides useful hints to control the charge current in kagome metals. In addition, the β€œstrain-induced increment of TTRSBsubscript𝑇TRSBT_{\rm TRSB}italic_T start_POSTSUBSCRIPT roman_TRSB end_POSTSUBSCRIPT reported in Ref. Moll-hz is naturally understood.

In the present study of Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT, we mainly focus on the current order in the 3⁒dX⁒Z3subscript𝑑𝑋𝑍3d_{XZ}3 italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital (in b3⁒gsubscript𝑏3𝑔b_{3g}italic_b start_POSTSUBSCRIPT 3 italic_g end_POSTSUBSCRIPT-representation) of V ion, which has been intensively studied previously Thomale2021 ; Neupert2021 ; Balents2021 ; Nandkishore ; Tazai-kagome2 . However, other 3⁒d3𝑑3d3 italic_d-orbitals (especially 3⁒dY⁒Z3subscriptπ‘‘π‘Œπ‘3d_{YZ}3 italic_d start_POSTSUBSCRIPT italic_Y italic_Z end_POSTSUBSCRIPT-, dX2βˆ’Y2subscript𝑑superscript𝑋2superscriptπ‘Œ2d_{X^{2}-Y^{2}}italic_d start_POSTSUBSCRIPT italic_X start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Y start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT-, d3⁒Z2βˆ’R2subscript𝑑3superscript𝑍2superscript𝑅2d_{3Z^{2}-R^{2}}italic_d start_POSTSUBSCRIPT 3 italic_Z start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_R start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT-orbitals) also form the large Fermi surfaces (FSs) with the van-Hove singularity (vHS) points near the Fermi level. The impact of these non-b3⁒gsubscript𝑏3𝑔b_{3g}italic_b start_POSTSUBSCRIPT 3 italic_g end_POSTSUBSCRIPT orbitals on the current order has also been studied in Refs. Fernandes-GL ; Nat-g-ology . The present field-induced GL theory does not depend on the d𝑑ditalic_d-orbital character of the current order parameter. We calculate the GL coefficients mnsubscriptπ‘šπ‘›m_{n}italic_m start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT (n=1βˆ’3𝑛13n=1-3italic_n = 1 - 3) for various d𝑑ditalic_d-orbital current order states based on the first-principles kagome metal models. For a fixed order parameter, mnsubscriptπ‘šπ‘›m_{n}italic_m start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT is large when the FS reconstruction due to the current order parameters occurs near the vHS points.

Model Hamiltonian with current and bond orders

Here, we study the kagome-lattice tight-binding model with b3⁒gsubscript𝑏3𝑔b_{3g}italic_b start_POSTSUBSCRIPT 3 italic_g end_POSTSUBSCRIPT (or dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT) orbitals shown in Fig. 1 (a). Each unit-cell is composed of three sublattices A, B, C. We set the nearest-neighbor hopping integral t=βˆ’0.5⁒[eV]𝑑0.5delimited-[]eVt=-0.5[{\rm eV}]italic_t = - 0.5 [ roman_eV ]. In addition, we introduce the nearest intra-sublattice hopping tβ€²=βˆ’0.02⁒[eV]superscript𝑑′0.02delimited-[]eVt^{\prime}=-0.02[{\rm eV}]italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = - 0.02 [ roman_eV ] shown in Fig. 1 (a) to avoid the perfect nesting. (Hereafter, the unit of the energy is eV.) The FS at the vHS filling, n=nvHS=2.55𝑛subscript𝑛vHS2.55n=n_{\rm vHS}=2.55italic_n = italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT = 2.55 per 3-site unit cell, is shown in Fig. 1 (b). Then, the FS lies on the three vHS points (π’ŒAsubscriptπ’ŒA{{\bm{k}}}_{{{\rm A}}}bold_italic_k start_POSTSUBSCRIPT roman_A end_POSTSUBSCRIPT, π’ŒBsubscriptπ’ŒB{{\bm{k}}}_{{{\rm B}}}bold_italic_k start_POSTSUBSCRIPT roman_B end_POSTSUBSCRIPT, π’ŒCsubscriptπ’ŒC{{\bm{k}}}_{{{\rm C}}}bold_italic_k start_POSTSUBSCRIPT roman_C end_POSTSUBSCRIPT), each of which is composed of a single sublattice A, B, or C. This simple three-orbital model well captures the main pure-type FS in kagome metals STM1 ; ARPES-VHS ; ARPES-band ; ARPES-Lifshitz ; Sato-ARPES .

The bond/current order is the modulation of the hopping integral between i𝑖iitalic_i and j𝑗jitalic_j atmos due to the electron correlation, δ⁒ti⁒jb/c𝛿superscriptsubscript𝑑𝑖𝑗bc\delta t_{ij}^{\rm b/c}italic_Ξ΄ italic_t start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_b / roman_c end_POSTSUPERSCRIPT. Theoretically, it is the symmetry breaking in the self-energy, and it is derived from the density-wave (DW) equation Tazai-kagome ; Tazai-kagome2 . The wavevectors of the bond and current orders correspond to the inter-sublattice nesting vectors 𝒒nsubscript𝒒𝑛{{\bm{q}}}_{n}bold_italic_q start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT (n=1βˆ’3𝑛13n=1-3italic_n = 1 - 3) in Fig. 1 (b) according to previous theoretical studies Thomale2021 ; Neupert2021 ; Balents2021 ; Tazai-kagome ; Tazai-kagome2 ; PDW-kagome . The triple Q𝑄Qitalic_Q (3⁒Q3𝑄3Q3 italic_Q) current order between the nearest atoms is depicted in Fig. 1 (c). The form factor (=normalized δ⁒ti⁒jc𝛿superscriptsubscript𝑑𝑖𝑗c\delta t_{ij}^{\rm c}italic_Ξ΄ italic_t start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_c end_POSTSUPERSCRIPT) with 𝒒=𝒒1𝒒subscript𝒒1{\bm{q}}={\bm{q}}_{1}bold_italic_q = bold_italic_q start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT, fi⁒j(1)superscriptsubscript𝑓𝑖𝑗1f_{ij}^{(1)}italic_f start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 1 ) end_POSTSUPERSCRIPT, is +i𝑖+i+ italic_i for (i,j)𝑖𝑗(i,j)( italic_i , italic_j ) belongs to the sites (l,m)=(1,2),(2,4),(4,5),(5,1)π‘™π‘š12244551(l,m)=(1,2),(2,4),(4,5),(5,1)( italic_l , italic_m ) = ( 1 , 2 ) , ( 2 , 4 ) , ( 4 , 5 ) , ( 5 , 1 ) in Fig. 1 (c), and βˆ’i𝑖-i- italic_i for (7,8),(8,10),(10,11),(11,7)788101011117(7,8),(8,10),(10,11),(11,7)( 7 , 8 ) , ( 8 , 10 ) , ( 10 , 11 ) , ( 11 , 7 ). Odd parity relation fi⁒j(1)=βˆ’fj⁒i(1)superscriptsubscript𝑓𝑖𝑗1superscriptsubscript𝑓𝑗𝑖1f_{ij}^{(1)}=-f_{ji}^{(1)}italic_f start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 1 ) end_POSTSUPERSCRIPT = - italic_f start_POSTSUBSCRIPT italic_j italic_i end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 1 ) end_POSTSUPERSCRIPT holds. Other form factors with 𝒒2subscript𝒒2{\bm{q}}_{2}bold_italic_q start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT and 𝒒3subscript𝒒3{\bm{q}}_{3}bold_italic_q start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT, fi⁒j(2)superscriptsubscript𝑓𝑖𝑗2f_{ij}^{(2)}italic_f start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 2 ) end_POSTSUPERSCRIPT and fi⁒j(3)superscriptsubscript𝑓𝑖𝑗3f_{ij}^{(3)}italic_f start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 3 ) end_POSTSUPERSCRIPT, are also derived from Fig. 1 (c). Using 𝒇i⁒j=(fi⁒j(1),fi⁒j(2),fi⁒j(3))subscript𝒇𝑖𝑗superscriptsubscript𝑓𝑖𝑗1superscriptsubscript𝑓𝑖𝑗2superscriptsubscript𝑓𝑖𝑗3{\bm{f}}_{ij}=(f_{ij}^{(1)},f_{ij}^{(2)},f_{ij}^{(3)})bold_italic_f start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT = ( italic_f start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 1 ) end_POSTSUPERSCRIPT , italic_f start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 2 ) end_POSTSUPERSCRIPT , italic_f start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 3 ) end_POSTSUPERSCRIPT ), the current order is

δ⁒ti⁒jc𝛿superscriptsubscript𝑑𝑖𝑗c\displaystyle\delta t_{ij}^{\rm c}italic_Ξ΄ italic_t start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_c end_POSTSUPERSCRIPT =\displaystyle== πœΌβ‹…π’‡i⁒j,β‹…πœΌsubscript𝒇𝑖𝑗\displaystyle{\bm{\eta}}\cdot{\bm{f}}_{ij},bold_italic_Ξ· β‹… bold_italic_f start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT , (1)

where πœΌβ‰‘(Ξ·1,Ξ·2,Ξ·3)𝜼subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3{\bm{\eta}}\equiv(\eta_{1},\eta_{2},\eta_{3})bold_italic_Ξ· ≑ ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ) is the set of current order parameters with the wavevector 𝒒msubscriptπ’’π‘š{{\bm{q}}}_{m}bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT. Also, the BO is given as

δ⁒ti⁒jb𝛿superscriptsubscript𝑑𝑖𝑗b\displaystyle\delta t_{ij}^{\rm b}italic_Ξ΄ italic_t start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_b end_POSTSUPERSCRIPT =\displaystyle== Ο•β‹…π’ˆi⁒j,β‹…bold-italic-Ο•subscriptπ’ˆπ‘–π‘—\displaystyle{\bm{\phi}}\cdot{\bm{g}}_{ij},bold_italic_Ο• β‹… bold_italic_g start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT , (2)

where ϕ≑(Ο•1,Ο•2,Ο•3)bold-italic-Ο•subscriptitalic-Ο•1subscriptitalic-Ο•2subscriptitalic-Ο•3{\bm{\phi}}\equiv(\phi_{1},\phi_{2},\phi_{3})bold_italic_Ο• ≑ ( italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ) is the set of BO parameters with the wavevector 𝒒msubscriptπ’’π‘š{{\bm{q}}}_{m}bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT, and gi⁒j(m)=gj⁒i(m)=Β±1superscriptsubscriptπ‘”π‘–π‘—π‘šsuperscriptsubscriptπ‘”π‘—π‘–π‘šplus-or-minus1g_{ij}^{(m)}=g_{ji}^{(m)}=\pm 1italic_g start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT = italic_g start_POSTSUBSCRIPT italic_j italic_i end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT = Β± 1 is the even-parity form factor for the BO shown in Fig. 1 (a). For example, gi⁒j(1)=+1⁒[βˆ’1]superscriptsubscript𝑔𝑖𝑗11delimited-[]1g_{ij}^{(1)}=+1\ [-1]italic_g start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 1 ) end_POSTSUPERSCRIPT = + 1 [ - 1 ] for (i,j)𝑖𝑗(i,j)( italic_i , italic_j ) belongs to (1,2),(4,5),(8,10),(11,7)1245810117(1,2),(4,5),(8,10),(11,7)( 1 , 2 ) , ( 4 , 5 ) , ( 8 , 10 ) , ( 11 , 7 ) [(2,4),(5,1),(7,8),(10,11)2451781011(2,4),(5,1),(7,8),(10,11)( 2 , 4 ) , ( 5 , 1 ) , ( 7 , 8 ) , ( 10 , 11 )].

The unit cell under the 3⁒Q3𝑄3Q3 italic_Q bond/current order is magnified by 2Γ—2222\times 22 Γ— 2 times. Thus, we analyze the electronic states with the current order based on the 4Γ—3434\times 34 Γ— 3-site kagome lattice model. The Hamiltonian with the bond+current order is H^=βˆ‘π’Œ,l,m,Οƒhl⁒m⁒(π’Œ)⁒cπ’Œ,l,σ†⁒cπ’Œ,m,Οƒ^𝐻subscriptπ’Œπ‘™π‘šπœŽsubscriptβ„Žπ‘™π‘šπ’Œsubscriptsuperscriptπ‘β€ π’Œπ‘™πœŽsubscriptπ‘π’Œπ‘šπœŽ\hat{H}=\sum_{{{\bm{k}}},l,m,\sigma}h_{lm}({{\bm{k}}})c^{\dagger}_{{{\bm{k}}},% l,\sigma}c_{{{\bm{k}}},m,\sigma}over^ start_ARG italic_H end_ARG = βˆ‘ start_POSTSUBSCRIPT bold_italic_k , italic_l , italic_m , italic_Οƒ end_POSTSUBSCRIPT italic_h start_POSTSUBSCRIPT italic_l italic_m end_POSTSUBSCRIPT ( bold_italic_k ) italic_c start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT start_POSTSUBSCRIPT bold_italic_k , italic_l , italic_Οƒ end_POSTSUBSCRIPT italic_c start_POSTSUBSCRIPT bold_italic_k , italic_m , italic_Οƒ end_POSTSUBSCRIPT, where l,m=1∼12π‘™π‘š1similar-to12l,m=1\sim 12italic_l , italic_m = 1 ∼ 12 and hl⁒m⁒(π’Œ)(=hm⁒l⁒(π’Œ)βˆ—)annotatedsubscriptβ„Žπ‘™π‘šπ’Œabsentsubscriptβ„Žπ‘šπ‘™superscriptπ’Œh_{lm}({{\bm{k}}})\ (=h_{ml}({{\bm{k}}})^{*})italic_h start_POSTSUBSCRIPT italic_l italic_m end_POSTSUBSCRIPT ( bold_italic_k ) ( = italic_h start_POSTSUBSCRIPT italic_m italic_l end_POSTSUBSCRIPT ( bold_italic_k ) start_POSTSUPERSCRIPT βˆ— end_POSTSUPERSCRIPT ) is the Fourier transform of the hopping integral t~i⁒j=ti⁒j+δ⁒ti⁒jb+δ⁒ti⁒jcsubscript~𝑑𝑖𝑗subscript𝑑𝑖𝑗𝛿superscriptsubscript𝑑𝑖𝑗b𝛿superscriptsubscript𝑑𝑖𝑗c{\tilde{t}}_{ij}=t_{ij}+\delta t_{ij}^{\rm b}+\delta t_{ij}^{\rm c}over~ start_ARG italic_t end_ARG start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT = italic_t start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT + italic_Ξ΄ italic_t start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_b end_POSTSUPERSCRIPT + italic_Ξ΄ italic_t start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_c end_POSTSUPERSCRIPT. Here, ti⁒jsubscript𝑑𝑖𝑗t_{ij}italic_t start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT is the hopping integral of the original model, and δ⁒ti⁒jb⁒(c)𝛿superscriptsubscript𝑑𝑖𝑗bc\delta t_{ij}^{\rm b(c)}italic_Ξ΄ italic_t start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_b ( roman_c ) end_POSTSUPERSCRIPT is the hopping integral due to the bond (current) order in Eq. [2] (Eq. [1]).

Uniform orbital magnetization

The TRS is broken in the presence of the current order δ⁒ti⁒jc𝛿superscriptsubscript𝑑𝑖𝑗c\delta t_{ij}^{\rm c}italic_Ξ΄ italic_t start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_c end_POSTSUPERSCRIPT. In this case, the uniform orbital magnetization Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT may appear due to the finite Berry curvature as pointed out in Ref. Balents2021 . Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT per V atom in the unit of Bohr magneton ΞΌB=e⁒ℏ/2⁒me⁒csubscriptπœ‡π΅π‘’Planck-constant-over-2-pi2subscriptπ‘šπ‘’π‘\mu_{B}=e\hbar/2m_{e}citalic_ΞΌ start_POSTSUBSCRIPT italic_B end_POSTSUBSCRIPT = italic_e roman_ℏ / 2 italic_m start_POSTSUBSCRIPT italic_e end_POSTSUBSCRIPT italic_c (mesubscriptπ‘šπ‘’m_{e}italic_m start_POSTSUBSCRIPT italic_e end_POSTSUBSCRIPT= free electron mass) is given as Morb-paper1 ; Morb-paper2

Morbsubscript𝑀orb\displaystyle M_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT =\displaystyle== ΞΌBE0⁒Nuc⁒Nβ’βˆ‘π’Œ,Οƒm⁒(π’Œ),subscriptπœ‡Bsubscript𝐸0subscript𝑁uc𝑁subscriptπ’ŒπœŽπ‘šπ’Œ\displaystyle\frac{\mu_{\rm B}}{E_{0}N_{\rm uc}N}\sum_{{{\bm{k}}},\sigma}m({{% \bm{k}}}),divide start_ARG italic_ΞΌ start_POSTSUBSCRIPT roman_B end_POSTSUBSCRIPT end_ARG start_ARG italic_E start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT italic_N start_POSTSUBSCRIPT roman_uc end_POSTSUBSCRIPT italic_N end_ARG βˆ‘ start_POSTSUBSCRIPT bold_italic_k , italic_Οƒ end_POSTSUBSCRIPT italic_m ( bold_italic_k ) , (3)
m⁒(π’Œ)π‘šπ’Œ\displaystyle m({{\bm{k}}})italic_m ( bold_italic_k ) =\displaystyle== βˆ‘Ξ±β‰ Ξ²Im⁒(π‘½Ξ²β’Ξ±β’π’Œβˆ—Γ—π‘½Ξ²β’Ξ±β’π’Œ)zsubscript𝛼𝛽Imsubscriptsubscriptsuperscriptπ‘½π›½π›Όπ’Œsubscriptπ‘½π›½π›Όπ’Œπ‘§\displaystyle\sum_{{\alpha}\neq{\beta}}{\rm Im}({\bm{V}}^{*}_{{\beta}{\alpha}{% {\bm{k}}}}\times{\bm{V}}_{{\beta}{\alpha}{{\bm{k}}}})_{z}βˆ‘ start_POSTSUBSCRIPT italic_Ξ± β‰  italic_Ξ² end_POSTSUBSCRIPT roman_Im ( bold_italic_V start_POSTSUPERSCRIPT βˆ— end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_Ξ² italic_Ξ± bold_italic_k end_POSTSUBSCRIPT Γ— bold_italic_V start_POSTSUBSCRIPT italic_Ξ² italic_Ξ± bold_italic_k end_POSTSUBSCRIPT ) start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT (5)
Γ—((Ο΅Ξ²β’π’Œβˆ’Ο΅Ξ±β’π’Œ)⁒n⁒(Ο΅Ξ±β’π’Œ)βˆ’2⁒T⁒ln⁑[1+eβˆ’(Ο΅Ξ±β’π’Œβˆ’ΞΌ)/T]),absentsubscriptitalic-Ο΅π›½π’Œsubscriptitalic-Ο΅π›Όπ’Œπ‘›subscriptitalic-Ο΅π›Όπ’Œ2𝑇1superscript𝑒subscriptitalic-Ο΅π›Όπ’Œπœ‡π‘‡\displaystyle\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!% \times\left(({\epsilon}_{{\beta}{{\bm{k}}}}-{\epsilon}_{{\alpha}{{\bm{k}}}})n(% {\epsilon}_{{\alpha}{{\bm{k}}}})-2T\ln[1+e^{-({\epsilon}_{{\alpha}{{\bm{k}}}}-% \mu)/T}]\right),Γ— ( ( italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ² bold_italic_k end_POSTSUBSCRIPT - italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ± bold_italic_k end_POSTSUBSCRIPT ) italic_n ( italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ± bold_italic_k end_POSTSUBSCRIPT ) - 2 italic_T roman_ln [ 1 + italic_e start_POSTSUPERSCRIPT - ( italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ± bold_italic_k end_POSTSUBSCRIPT - italic_ΞΌ ) / italic_T end_POSTSUPERSCRIPT ] ) ,
π‘½Ξ±β’Ξ²β’π’Œ=βŸ¨Ξ±β’π’Œ|βˆ‡π’Œhπ’Œ|Ξ²β’π’ŒβŸ©/(Ο΅Ξ±β’π’Œβˆ’Ο΅Ξ²β’π’Œ),subscriptπ‘½π›Όπ›½π’Œquantum-operator-productπ›Όπ’Œsubscriptbold-βˆ‡π’Œsubscriptβ„Žπ’Œπ›½π’Œsubscriptitalic-Ο΅π›Όπ’Œsubscriptitalic-Ο΅π›½π’Œ\displaystyle\!\!\!\!\!\!\!\!{\bm{V}}_{{\alpha}{\beta}{{\bm{k}}}}={\langle{% \alpha}{{\bm{k}}}|{\bm{\nabla}}_{{\bm{k}}}h_{{\bm{k}}}|{\beta}{{\bm{k}}}% \rangle}/{({\epsilon}_{{\alpha}{{\bm{k}}}}-{\epsilon}_{{\beta}{{\bm{k}}}})},bold_italic_V start_POSTSUBSCRIPT italic_Ξ± italic_Ξ² bold_italic_k end_POSTSUBSCRIPT = ⟨ italic_Ξ± bold_italic_k | bold_βˆ‡ start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT italic_h start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT | italic_Ξ² bold_italic_k ⟩ / ( italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ± bold_italic_k end_POSTSUBSCRIPT - italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ² bold_italic_k end_POSTSUBSCRIPT ) ,

where Ο΅Ξ±β’π’Œsubscriptitalic-Ο΅π›Όπ’Œ{\epsilon}_{{\alpha}{{\bm{k}}}}italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ± bold_italic_k end_POSTSUBSCRIPT is α𝛼{\alpha}italic_Ξ±-th eigenenergy of 4Γ—3434\times 34 Γ— 3-site kagome lattice model in the folded BZ. n⁒(Ο΅)𝑛italic-Ο΅n({\epsilon})italic_n ( italic_Ο΅ ) is Fermi distribution function, Nuc=12subscript𝑁uc12N_{\rm uc}=12italic_N start_POSTSUBSCRIPT roman_uc end_POSTSUBSCRIPT = 12 is the site number of 2Γ—2222\times 22 Γ— 2 unit cell, N𝑁Nitalic_N is the π’Œπ’Œ{{\bm{k}}}bold_italic_k-mesh number, and E0=ℏ2/a2⁒mesubscript𝐸0superscriptPlanck-constant-over-2-pi2superscriptπ‘Ž2subscriptπ‘šπ‘’E_{0}=\hbar^{2}/a^{2}m_{e}italic_E start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT = roman_ℏ start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT / italic_a start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_m start_POSTSUBSCRIPT italic_e end_POSTSUBSCRIPT. aπ‘Žaitalic_a is the unit length in the numerical study, and we set a=|𝒂AB|π‘Žsubscript𝒂ABa=|{\bm{a}}_{\rm AB}|italic_a = | bold_italic_a start_POSTSUBSCRIPT roman_AB end_POSTSUBSCRIPT | (=0.275absent0.275=0.275= 0.275 nm in kagome metals). E0=1.0subscript𝐸01.0E_{0}=1.0italic_E start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT = 1.0 eV when a=0.275π‘Ž0.275a=0.275italic_a = 0.275 nm.

At zero temperature, Eq. [3] is rewritten as

MorbT=0superscriptsubscript𝑀orb𝑇0\displaystyle M_{\rm orb}^{T=0}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_T = 0 end_POSTSUPERSCRIPT =\displaystyle== ΞΌBE0⁒Nuc⁒Nβˆ‘π’Œ,Ξ±<Ξ²Im{VΞ²β’Ξ±β’π’ŒxVΞ±β’Ξ²β’π’Œyβˆ’(α↔β)}\displaystyle\frac{\mu_{\rm B}}{E_{0}N_{\rm uc}N}\sum_{{{\bm{k}}},{\alpha}<{% \beta}}{\rm Im}\{V^{x}_{{\beta}{\alpha}{{\bm{k}}}}V^{y}_{{\alpha}{\beta}{{\bm{% k}}}}-({\alpha}\leftrightarrow{\beta})\}divide start_ARG italic_ΞΌ start_POSTSUBSCRIPT roman_B end_POSTSUBSCRIPT end_ARG start_ARG italic_E start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT italic_N start_POSTSUBSCRIPT roman_uc end_POSTSUBSCRIPT italic_N end_ARG βˆ‘ start_POSTSUBSCRIPT bold_italic_k , italic_Ξ± < italic_Ξ² end_POSTSUBSCRIPT roman_Im { italic_V start_POSTSUPERSCRIPT italic_x end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_Ξ² italic_Ξ± bold_italic_k end_POSTSUBSCRIPT italic_V start_POSTSUPERSCRIPT italic_y end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_Ξ± italic_Ξ² bold_italic_k end_POSTSUBSCRIPT - ( italic_Ξ± ↔ italic_Ξ² ) } (6)
Γ—(Ο΅Ξ±β’π’Œ+Ο΅Ξ²β’π’Œβˆ’2⁒μ)⁒(n0⁒(Ο΅Ξ±β’π’Œ)βˆ’n0⁒(Ο΅Ξ²β’π’Œ)),absentsubscriptitalic-Ο΅π›Όπ’Œsubscriptitalic-Ο΅π›½π’Œ2πœ‡superscript𝑛0subscriptitalic-Ο΅π›Όπ’Œsuperscript𝑛0subscriptitalic-Ο΅π›½π’Œ\displaystyle\times({\epsilon}_{{\alpha}{{\bm{k}}}}+{\epsilon}_{{\beta}{{\bm{k% }}}}-2\mu)(n^{0}({\epsilon}_{{\alpha}{{\bm{k}}}})-n^{0}({\epsilon}_{{\beta}{{% \bm{k}}}})),Γ— ( italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ± bold_italic_k end_POSTSUBSCRIPT + italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ² bold_italic_k end_POSTSUBSCRIPT - 2 italic_ΞΌ ) ( italic_n start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ± bold_italic_k end_POSTSUBSCRIPT ) - italic_n start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ² bold_italic_k end_POSTSUBSCRIPT ) ) ,

where n0⁒(Ο΅)superscript𝑛0italic-Ο΅n^{0}({\epsilon})italic_n start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_Ο΅ ) is the Fermi distribution function at T=0𝑇0T=0italic_T = 0. Considering the factor (n0⁒(Ο΅Ξ±β’π’Œ)βˆ’n0⁒(Ο΅Ξ²β’π’Œ))superscript𝑛0subscriptitalic-Ο΅π›Όπ’Œsuperscript𝑛0subscriptitalic-Ο΅π›½π’Œ(n^{0}({\epsilon}_{{\alpha}{{\bm{k}}}})-n^{0}({\epsilon}_{{\beta}{{\bm{k}}}}))( italic_n start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ± bold_italic_k end_POSTSUBSCRIPT ) - italic_n start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ² bold_italic_k end_POSTSUBSCRIPT ) ) with Ξ±<β𝛼𝛽{\alpha}<{\beta}italic_Ξ± < italic_Ξ², Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT originates from the β€œvertical particle-hole (p-h) excitation”, from Ο΅Ξ±β’π’Œ<0subscriptitalic-Ο΅π›Όπ’Œ0{\epsilon}_{{\alpha}{{\bm{k}}}}<0italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ± bold_italic_k end_POSTSUBSCRIPT < 0 to Ο΅Ξ²β’π’Œ>0subscriptitalic-Ο΅π›½π’Œ0{\epsilon}_{{\beta}{{\bm{k}}}}>0italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ² bold_italic_k end_POSTSUBSCRIPT > 0, at the same π’Œπ’Œ{{\bm{k}}}bold_italic_k in the folded BZ.

The folded FS (Ξ·=Ο•=0πœ‚italic-Ο•0\eta=\phi=0italic_Ξ· = italic_Ο• = 0) and bandstructure in the folded Brillouin zone (BZ) at n=nvHS𝑛subscript𝑛vHSn=n_{\rm vHS}italic_n = italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT are shown in Figs. S1 (a)-(c) in the SI A SM . Here, all vHS points A, B, C in the original BZ in Fig. 1 (b) move to ΓΓ\Gammaroman_Ξ“ point, and they will hybridize each other due to the current and/or BO parameter. When nβ‰ˆnvHS𝑛subscript𝑛vHSn\approx n_{\rm vHS}italic_n β‰ˆ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT, prominent band hybridization occurs near the ΓΓ\Gammaroman_Ξ“ point, as understood in Fig. S1 (a). Then, the bonding (antibonding) band energy at ΓΓ\Gammaroman_Ξ“ point is below (above) the Fermi level, as shown in Fig. S1 (b). Because of the factor Im⁒{β‹―}∝|Ο΅Ξ±β’π’Œβˆ’Ο΅Ξ²β’π’Œ|βˆ’2proportional-toImβ‹―superscriptsubscriptitalic-Ο΅π›Όπ’Œsubscriptitalic-Ο΅π›½π’Œ2{\rm Im}\{\cdots\}\propto|{\epsilon}_{{\alpha}{{\bm{k}}}}-{\epsilon}_{{\beta}{% {\bm{k}}}}|^{-2}roman_Im { β‹― } ∝ | italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ± bold_italic_k end_POSTSUBSCRIPT - italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ² bold_italic_k end_POSTSUBSCRIPT | start_POSTSUPERSCRIPT - 2 end_POSTSUPERSCRIPT in Eq. [6], Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT becomes large when nβ‰ˆnvHS𝑛subscript𝑛vHSn\approx n_{\rm vHS}italic_n β‰ˆ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT.

Refer to caption
Figure 2: (a) Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT [ΞΌBsubscriptπœ‡B\mu_{\rm B}italic_ΞΌ start_POSTSUBSCRIPT roman_B end_POSTSUBSCRIPT] per V𝑉Vitalic_V atom due to the 3⁒Q3𝑄3Q3 italic_Q current order 𝜼=(Ξ·,Ξ·,Ξ·)/3πœΌπœ‚πœ‚πœ‚3{\bm{\eta}}=(\eta,\eta,\eta)/\sqrt{3}bold_italic_Ξ· = ( italic_Ξ· , italic_Ξ· , italic_Ξ· ) / square-root start_ARG 3 end_ARG at T=1𝑇1T=1italic_T = 1 meV for n=2.47∼2.63𝑛2.47similar-to2.63n=2.47\sim 2.63italic_n = 2.47 ∼ 2.63. Morb(∝η3)annotatedsubscript𝑀orbproportional-toabsentsuperscriptπœ‚3M_{\rm orb}\ (\propto\eta^{3})italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT ( ∝ italic_Ξ· start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT ) is large when nβ‰ˆnvHS𝑛subscript𝑛vHSn\approx n_{\rm vHS}italic_n β‰ˆ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT. (b) FS around nvHS=2.55subscript𝑛vHS2.55n_{\rm vHS}=2.55italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT = 2.55. (c) Charge-current pattern in real space. Jβ‰ Jβ€²β‰ J′′𝐽superscript𝐽′superscript𝐽′′J\neq J^{\prime}\neq J^{\prime\prime}italic_J β‰  italic_J start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT β‰  italic_J start_POSTSUPERSCRIPT β€² β€² end_POSTSUPERSCRIPT in spite of the same order parameter Ξ·1=Ξ·2=Ξ·3subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3\eta_{1}=\eta_{2}=\eta_{3}italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT. (d) Morb(∝η)annotatedsubscript𝑀orbproportional-toabsentπœ‚M_{\rm orb}\ (\propto\eta)italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT ( ∝ italic_Ξ· ) due to the coexistence of 3⁒Q3𝑄3Q3 italic_Q current order and 3⁒Q3𝑄3Q3 italic_Q BO at n=2.47𝑛2.47n=2.47italic_n = 2.47. (e) 1⁒Q1𝑄1Q1 italic_Q bond+current order with finite Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT. Its time-reversal and translation by 2⁒𝒂AB2subscript𝒂AB2{\bm{a}}_{\rm AB}2 bold_italic_a start_POSTSUBSCRIPT roman_AB end_POSTSUBSCRIPT causes a different state.

Here, we calculate Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT [ΞΌBsubscriptπœ‡B\mu_{\rm B}italic_ΞΌ start_POSTSUBSCRIPT roman_B end_POSTSUBSCRIPT] near the vHS filling nvHS=2.55subscript𝑛vHS2.55n_{\rm vHS}=2.55italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT = 2.55 at T=1𝑇1T=1italic_T = 1 meV, in the case of E0=1.0subscript𝐸01.0E_{0}=1.0italic_E start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT = 1.0 eV. Figure 2 (a) represents the obtained Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT under the 3⁒Q3𝑄3Q3 italic_Q current order 𝜼=(Ξ·,Ξ·,Ξ·)/3πœΌπœ‚πœ‚πœ‚3{\bm{\eta}}=(\eta,\eta,\eta)/\sqrt{3}bold_italic_Ξ· = ( italic_Ξ· , italic_Ξ· , italic_Ξ· ) / square-root start_ARG 3 end_ARG for n=2.47∼2.63𝑛2.47similar-to2.63n=2.47\sim 2.63italic_n = 2.47 ∼ 2.63. (The FS without the current order is shown in Fig. 2 (b).) We obtain the relation Morb∝η3proportional-tosubscript𝑀orbsuperscriptπœ‚3M_{\rm orb}\propto\eta^{3}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT ∝ italic_Ξ· start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT, and it becomes large when the filling is close to nvHSsubscript𝑛vHSn_{\rm vHS}italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT. Therefore, the 3⁒Q3𝑄3Q3 italic_Q current order state is a weak-ferromagnetic (or ferrimagnetic) state. Because the additional free energy under the magnetic field hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT per 3-site unit cell is

Δ⁒F=βˆ’3⁒hz⁒Morb,Δ𝐹3subscriptβ„Žπ‘§subscript𝑀orb\displaystyle\Delta F=-3h_{z}M_{\rm orb},roman_Ξ” italic_F = - 3 italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT , (7)

the chirality of the current order is aligned under tiny hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT. In other words, the 3⁒Q3𝑄3Q3 italic_Q current order is stabilized under hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT. In contrast, Morb=0subscript𝑀orb0M_{\rm orb}=0italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT = 0 for 1⁒Q1𝑄1Q1 italic_Q and 2⁒Q2𝑄2Q2 italic_Q current orders. hz=10βˆ’4subscriptβ„Žπ‘§superscript104h_{z}=10^{-4}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 10 start_POSTSUPERSCRIPT - 4 end_POSTSUPERSCRIPT corresponds to ∼1similar-toabsent1\sim 1∼ 1 Tesla.

Here, we present a symmetry argument to understand Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT induced by the current order. Both 1⁒Q1𝑄1Q1 italic_Q and 2⁒Q2𝑄2Q2 italic_Q current orders are invariant under the time-reversal and the translational operations successively. Therefore, Morb=0subscript𝑀orb0M_{\rm orb}=0italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT = 0 due to the TRS in the bulk state. In contrast, Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT is nonzero in the 3⁒Q3𝑄3Q3 italic_Q current order because this state breaks the TRS in the bulk state Balents2021 . To find an intuitive reason, we calculated the charge-current along the nearest bonds Ji⁒jsubscript𝐽𝑖𝑗J_{ij}italic_J start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT in the 3⁒Q3𝑄3Q3 italic_Q state using the method in Refs. Tazai-cLC ; Kontani-AdvPhys , and found that |Ji⁒j|subscript𝐽𝑖𝑗|J_{ij}|| italic_J start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT | is bond-dependent in spite of the same order parameter Ξ·1=Ξ·2=Ξ·3subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3\eta_{1}=\eta_{2}=\eta_{3}italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT. The obtained relation Jβ‰ Jβ€²β‰ J′′𝐽superscript𝐽′superscript𝐽′′J\neq J^{\prime}\neq J^{\prime\prime}italic_J β‰  italic_J start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT β‰  italic_J start_POSTSUPERSCRIPT β€² β€² end_POSTSUPERSCRIPT shown in Fig. 2 (c) indicates that Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT becomes finite because the magnetic fluxes through triangles and hexagons do not cancel perfectly.

For general order parameter 𝜼=(Ξ·1,Ξ·2,Ξ·3)𝜼subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3{\bm{\eta}}=(\eta_{1},\eta_{2},\eta_{3})bold_italic_Ξ· = ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ), we verified the relation Morb∝η1⁒η2⁒η3proportional-tosubscript𝑀orbsubscriptπœ‚1subscriptπœ‚2subscriptπœ‚3M_{\rm orb}\propto\eta_{1}\eta_{2}\eta_{3}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT ∝ italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT up to the third-order. In fact, based on the perturbation theory with respect to Eq. [1]; Ξ·m⁒f^(m)subscriptπœ‚π‘šsuperscript^π‘“π‘š\eta_{m}{\hat{f}}^{(m)}italic_Ξ· start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT over^ start_ARG italic_f end_ARG start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT (m=1βˆ’3π‘š13m=1-3italic_m = 1 - 3). Morb⁒[𝜼]subscript𝑀orbdelimited-[]𝜼M_{\rm orb}[{\bm{\eta}}]italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT [ bold_italic_Ξ· ] is expanded as βˆ‘p⁒q⁒rbp⁒q⁒r⁒(Ξ·1)p⁒(Ξ·2)q⁒(Ξ·3)rsubscriptπ‘π‘žπ‘Ÿsubscriptπ‘π‘π‘žπ‘Ÿsuperscriptsubscriptπœ‚1𝑝superscriptsubscriptπœ‚2π‘žsuperscriptsubscriptπœ‚3π‘Ÿ\sum_{pqr}b_{pqr}(\eta_{1})^{p}(\eta_{2})^{q}(\eta_{3})^{r}βˆ‘ start_POSTSUBSCRIPT italic_p italic_q italic_r end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT italic_p italic_q italic_r end_POSTSUBSCRIPT ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ) start_POSTSUPERSCRIPT italic_p end_POSTSUPERSCRIPT ( italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) start_POSTSUPERSCRIPT italic_q end_POSTSUPERSCRIPT ( italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ) start_POSTSUPERSCRIPT italic_r end_POSTSUPERSCRIPT with p+q+r=oddπ‘π‘žπ‘Ÿoddp+q+r={\rm odd}italic_p + italic_q + italic_r = roman_odd because Morb⁒[𝜼]subscript𝑀orbdelimited-[]𝜼M_{\rm orb}[{\bm{\eta}}]italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT [ bold_italic_Ξ· ] is an odd function of 𝜼𝜼{\bm{\eta}}bold_italic_Ξ·. In addition, bp⁒q⁒rsubscriptπ‘π‘π‘žπ‘Ÿb_{pqr}italic_b start_POSTSUBSCRIPT italic_p italic_q italic_r end_POSTSUBSCRIPT can be nonzero only when p⁒𝒒1+q⁒𝒒2+r⁒𝒒3=πŸŽπ‘subscript𝒒1π‘žsubscript𝒒2π‘Ÿsubscript𝒒30p{{\bm{q}}}_{1}+q{{\bm{q}}}_{2}+r{{\bm{q}}}_{3}={\bm{0}}italic_p bold_italic_q start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT + italic_q bold_italic_q start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT + italic_r bold_italic_q start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT = bold_0 (modulo original reciprocal vectors), because we study the nonlinear response of the uniform (𝒒=πŸŽπ’’0{{\bm{q}}}={\bm{0}}bold_italic_q = bold_0) magnetization due to the potential Ξ·m⁒f(m)subscriptπœ‚π‘šsuperscriptπ‘“π‘š\eta_{m}f^{(m)}italic_Ξ· start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT italic_f start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT with wavevector 𝒒msubscriptπ’’π‘š{{\bm{q}}}_{m}bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT in the original unit cell. (See the SI B for a more detailed explanation SM .) In Fig. 2 (a), Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT originates from the third-order term b111subscript𝑏111b_{111}italic_b start_POSTSUBSCRIPT 111 end_POSTSUBSCRIPT, which is allowed because of the momentum conservation 𝒒1+𝒒2+𝒒3=𝟎subscript𝒒1subscript𝒒2subscript𝒒30{{\bm{q}}}_{1}+{{\bm{q}}}_{2}+{{\bm{q}}}_{3}={\bm{0}}bold_italic_q start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT + bold_italic_q start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT + bold_italic_q start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT = bold_0.

Refer to caption
Figure 3: Obtained coefficients of MΒ―orbsubscript¯𝑀orb{\bar{M}}_{\rm orb}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT, (a) m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and (b) m2(β‰ˆβˆ’m3)annotatedsubscriptπ‘š2absentsubscriptπ‘š3m_{2}\ (\approx-m_{3})italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ( β‰ˆ - italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ), per V𝑉Vitalic_V atom as a function of n𝑛nitalic_n. Both |m1|subscriptπ‘š1|m_{1}|| italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT | and |m2|subscriptπ‘š2|m_{2}|| italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT | are large for n∼nvHSsimilar-to𝑛subscript𝑛vHSn\sim n_{\rm vHS}italic_n ∼ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT. (c) Folded FS and (d) folded bandstructure in the folded BZ when 𝜼=Ο•β†’πŸŽπœΌbold-italic-Ο•β†’0{\bm{\eta}}={\bm{\phi}}\rightarrow{\bm{0}}bold_italic_Ξ· = bold_italic_Ο• β†’ bold_0. The π’Œπ’Œ{{\bm{k}}}bold_italic_k-path is shown in the (c). We set the origin of the energy at the vHS energy. (e) m1⁒(π’Œ)β‰‘βˆ‚2m⁒(π’Œ)/βˆ‚Ξ·1β’βˆ‚Ο•1|𝜼=Ο•=0subscriptπ‘š1π’Œevaluated-atsuperscript2π‘šπ’Œsubscriptπœ‚1subscriptitalic-Ο•1𝜼bold-italic-Ο•0m_{1}({{\bm{k}}})\equiv{\partial}^{2}m({{\bm{k}}})/{\partial}\eta_{1}{\partial% }\phi_{1}|_{{\bm{\eta}}={\bm{\phi}}=0}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ( bold_italic_k ) ≑ βˆ‚ start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_m ( bold_italic_k ) / βˆ‚ italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT βˆ‚ italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT | start_POSTSUBSCRIPT bold_italic_Ξ· = bold_italic_Ο• = 0 end_POSTSUBSCRIPT and (f) m2⁒(π’Œ)β‰‘βˆ‚3m⁒(π’Œ)/βˆ‚Ξ·1β’βˆ‚Ξ·2β’βˆ‚Ξ·3|𝜼=Ο•=0subscriptπ‘š2π’Œevaluated-atsuperscript3π‘šπ’Œsubscriptπœ‚1subscriptπœ‚2subscriptπœ‚3𝜼bold-italic-Ο•0m_{2}({{\bm{k}}})\equiv{\partial}^{3}m({{\bm{k}}})/{\partial}\eta_{1}{\partial% }\eta_{2}{\partial}\eta_{3}|_{{\bm{\eta}}={\bm{\phi}}=0}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ( bold_italic_k ) ≑ βˆ‚ start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT italic_m ( bold_italic_k ) / βˆ‚ italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT βˆ‚ italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT βˆ‚ italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT | start_POSTSUBSCRIPT bold_italic_Ξ· = bold_italic_Ο• = 0 end_POSTSUBSCRIPT for n=2.58𝑛2.58n=2.58italic_n = 2.58 (nvHS=2.55subscript𝑛vHS2.55n_{\rm vHS}=2.55italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT = 2.55). Note that mnβˆβˆ‘π’Œmn⁒(π’Œ)proportional-tosubscriptπ‘šπ‘›subscriptπ’Œsubscriptπ‘šπ‘›π’Œm_{n}\propto\sum_{{\bm{k}}}m_{n}({{\bm{k}}})italic_m start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT ∝ βˆ‘ start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT italic_m start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT ( bold_italic_k ).

Next stage, we calculate Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT under the coexistence of the 3⁒Q3𝑄3Q3 italic_Q current order 𝜼=(Ξ·,Ξ·,Ξ·)/3πœΌπœ‚πœ‚πœ‚3{\bm{\eta}}=(\eta,\eta,\eta)/\sqrt{3}bold_italic_Ξ· = ( italic_Ξ· , italic_Ξ· , italic_Ξ· ) / square-root start_ARG 3 end_ARG and the 3⁒Q3𝑄3Q3 italic_Q bond order Ο•=(Ο•,Ο•,Ο•)/3bold-italic-Ο•italic-Ο•italic-Ο•italic-Ο•3{\bm{\phi}}=(\phi,\phi,\phi)/\sqrt{3}bold_italic_Ο• = ( italic_Ο• , italic_Ο• , italic_Ο• ) / square-root start_ARG 3 end_ARG. Figure 2 (d) represents Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT at Ο•=0,Β±0.01italic-Ο•0plus-or-minus0.01\phi=0,\pm 0.01italic_Ο• = 0 , Β± 0.01 as a function of Ξ·πœ‚\etaitalic_Ξ·, for n=2.47𝑛2.47n=2.47italic_n = 2.47 and T=1𝑇1T=1italic_T = 1 meV. Interesting relation Morb∝ηproportional-tosubscript𝑀orbπœ‚M_{\rm orb}\propto\etaitalic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT ∝ italic_Ξ· is obtained when Ο•β‰ 0italic-Ο•0\phi\neq 0italic_Ο• β‰  0. Then, the field-induced Δ⁒FΔ𝐹\Delta Froman_Ξ” italic_F contains a non-analytic Ξ·πœ‚\etaitalic_Ξ·-linear term that always produces βŸ¨Ξ·βŸ©β‰ 0delimited-βŸ¨βŸ©πœ‚0\langle\eta\rangle\neq 0⟨ italic_Ξ· ⟩ β‰  0. This fact causes significant field-induced change in the phase diagram, as we will explain in the SI C SM .

To understand the Ξ·πœ‚\etaitalic_Ξ·-linear term in Fig. 2 (d), we expand Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT with respect to the current order Ξ·m⁒f^(m)subscriptπœ‚π‘šsuperscript^π‘“π‘š\eta_{m}{\hat{f}}^{(m)}italic_Ξ· start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT over^ start_ARG italic_f end_ARG start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT and bond order Ο•m⁒g^(m)subscriptitalic-Ο•π‘šsuperscript^π‘”π‘š\phi_{m}{\hat{g}}^{(m)}italic_Ο• start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT over^ start_ARG italic_g end_ARG start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT. Its Taylor expansion is βˆ‘p⁒q⁒r⁒p′⁒q′⁒rβ€²bp⁒q⁒rp′⁒q′⁒r′⁒(Ξ·1)p⁒(Ξ·2)q⁒(Ξ·3)r⁒(Ο•1)p′⁒(Ο•2)q′⁒(Ο•3)rβ€²subscriptπ‘π‘žπ‘Ÿsuperscript𝑝′superscriptπ‘žβ€²superscriptπ‘Ÿβ€²superscriptsubscriptπ‘π‘π‘žπ‘Ÿsuperscript𝑝′superscriptπ‘žβ€²superscriptπ‘Ÿβ€²superscriptsubscriptπœ‚1𝑝superscriptsubscriptπœ‚2π‘žsuperscriptsubscriptπœ‚3π‘Ÿsuperscriptsubscriptitalic-Ο•1superscript𝑝′superscriptsubscriptitalic-Ο•2superscriptπ‘žβ€²superscriptsubscriptitalic-Ο•3superscriptπ‘Ÿβ€²\sum_{pqrp^{\prime}q^{\prime}r^{\prime}}b_{pqr}^{p^{\prime}q^{\prime}r^{\prime% }}(\eta_{1})^{p}(\eta_{2})^{q}(\eta_{3})^{r}(\phi_{1})^{p^{\prime}}(\phi_{2})^% {q^{\prime}}(\phi_{3})^{r^{\prime}}βˆ‘ start_POSTSUBSCRIPT italic_p italic_q italic_r italic_p start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_q start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_r start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT italic_p italic_q italic_r end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_p start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_q start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_r start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ) start_POSTSUPERSCRIPT italic_p end_POSTSUPERSCRIPT ( italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) start_POSTSUPERSCRIPT italic_q end_POSTSUPERSCRIPT ( italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ) start_POSTSUPERSCRIPT italic_r end_POSTSUPERSCRIPT ( italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ) start_POSTSUPERSCRIPT italic_p start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT ( italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) start_POSTSUPERSCRIPT italic_q start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT ( italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ) start_POSTSUPERSCRIPT italic_r start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT with p+q+r=oddπ‘π‘žπ‘Ÿoddp+q+r={\rm odd}italic_p + italic_q + italic_r = roman_odd and (p+pβ€²)⁒𝒒1+(q+qβ€²)⁒𝒒2+(r+rβ€²)⁒𝒒3=πŸŽπ‘superscript𝑝′subscript𝒒1π‘žsuperscriptπ‘žβ€²subscript𝒒2π‘Ÿsuperscriptπ‘Ÿβ€²subscript𝒒30(p+p^{\prime}){{\bm{q}}}_{1}+(q+q^{\prime}){{\bm{q}}}_{2}+(r+r^{\prime}){{\bm{% q}}}_{3}={\bm{0}}( italic_p + italic_p start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ) bold_italic_q start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT + ( italic_q + italic_q start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ) bold_italic_q start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT + ( italic_r + italic_r start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ) bold_italic_q start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT = bold_0 (modulo original reciprocal vectors). The Ξ·πœ‚\etaitalic_Ξ·-linear term in Fig. 2 (c) mainly originates from the second-order term b100100superscriptsubscript𝑏100100b_{100}^{100}italic_b start_POSTSUBSCRIPT 100 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 100 end_POSTSUPERSCRIPT in addition to the third-order term b100011superscriptsubscript𝑏100011b_{100}^{011}italic_b start_POSTSUBSCRIPT 100 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 011 end_POSTSUPERSCRIPT. In fact, the 1⁒Q1𝑄1Q1 italic_Q current + 1⁒Q1𝑄1Q1 italic_Q BO state shown in Fig. 2 (e) violates the TRS in the bulk state.

Field-induced GL free energy expression

From the above discussions, we obtain the following convenient expression of the orbital magnetization and the β€œfield-induced GL free energy formula” up to the third-order:

MΒ―orbsubscript¯𝑀orb\displaystyle{\bar{M}}_{\rm orb}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT =\displaystyle== m1β’Ο•β‹…πœΌ+m2⁒η1⁒η2⁒η3β‹…subscriptπ‘š1bold-italic-Ο•πœΌsubscriptπ‘š2subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3\displaystyle m_{1}{\bm{\phi}}\cdot{\bm{\eta}}+m_{2}\eta_{1}\eta_{2}\eta_{3}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT bold_italic_Ο• β‹… bold_italic_Ξ· + italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT (8)
+m3⁒(Ξ·1⁒ϕ2⁒ϕ3+Ο•1⁒η2⁒ϕ3+Ο•1⁒ϕ2⁒η3),subscriptπ‘š3subscriptπœ‚1subscriptitalic-Ο•2subscriptitalic-Ο•3subscriptitalic-Ο•1subscriptπœ‚2subscriptitalic-Ο•3subscriptitalic-Ο•1subscriptitalic-Ο•2subscriptπœ‚3\displaystyle+m_{3}(\eta_{1}\phi_{2}\phi_{3}+\phi_{1}\eta_{2}\phi_{3}+\phi_{1}% \phi_{2}\eta_{3}),+ italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT + italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT + italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ) ,
Δ⁒F¯Δ¯𝐹\displaystyle\Delta{\bar{F}}roman_Ξ” overΒ― start_ARG italic_F end_ARG =\displaystyle== βˆ’3⁒hz⁒MΒ―orb,3subscriptβ„Žπ‘§subscript¯𝑀orb\displaystyle-3h_{z}{\bar{M}}_{\rm orb},- 3 italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT , (9)

which enables us to elucidate the field-induced bond+current phase transitions in kagome metals. Figures 3 (a) and (b) show the coefficients m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT derived from Eq. [3] numerically, respectively. Interestingly, |m1|subscriptπ‘š1|m_{1}|| italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT | and |m2|subscriptπ‘š2|m_{2}|| italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT | become large for n∼nvHSsimilar-to𝑛subscript𝑛vHSn\sim n_{\rm vHS}italic_n ∼ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT.

In the SI D SM , we show that MΒ―orbsubscript¯𝑀orb{\bar{M}}_{\rm orb}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT in Eq. [8] well reproduces the original Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT when |𝜼|,|Ο•|≲0.02less-than-or-similar-to𝜼bold-italic-Ο•0.02|{\bm{\eta}}|,|{\bm{\phi}}|\lesssim 0.02| bold_italic_Ξ· | , | bold_italic_Ο• | ≲ 0.02. The expression of MΒ―orbsubscript¯𝑀orb{\bar{M}}_{\rm orb}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT is also justified based on the first principles model for |𝜼|,|Ο•|≲0.02less-than-or-similar-to𝜼bold-italic-Ο•0.02|{\bm{\eta}}|,|{\bm{\phi}}|\lesssim 0.02| bold_italic_Ξ· | , | bold_italic_Ο• | ≲ 0.02 in the SI D SM .

To understand the n𝑛nitalic_n-dependences of m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT, we discuss the folded FS and bandstructure given in Figs. 3 (c) and (d), respectively, for 𝜼=Ο•β†’πŸŽπœΌbold-italic-Ο•β†’0{\bm{\eta}}={\bm{\phi}}\rightarrow{\bm{0}}bold_italic_Ξ· = bold_italic_Ο• β†’ bold_0. Here, band aπ‘Žaitalic_a [b𝑏bitalic_b, and c𝑐citalic_c] originates from the bandstructure around vHS-A [B,C] in Fig. 1 (b). As shown in Fig. 3 (d), band aπ‘Žaitalic_a and band b𝑏bitalic_b degenerate along kxsubscriptπ‘˜π‘₯k_{x}italic_k start_POSTSUBSCRIPT italic_x end_POSTSUBSCRIPT-axis and kysubscriptπ‘˜π‘¦k_{y}italic_k start_POSTSUBSCRIPT italic_y end_POSTSUBSCRIPT-axis. Hereafter, we set ΞΌ=0πœ‡0\mu=0italic_ΞΌ = 0.

First, we consider the origin of m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT, which is caused by the band-folding induced by Ξ·1⁒f^(1)subscriptπœ‚1superscript^𝑓1\eta_{1}{\hat{f}}^{(1)}italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT over^ start_ARG italic_f end_ARG start_POSTSUPERSCRIPT ( 1 ) end_POSTSUPERSCRIPT and Ο•1⁒g^(1)subscriptitalic-Ο•1superscript^𝑔1\phi_{1}{\hat{g}}^{(1)}italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT over^ start_ARG italic_g end_ARG start_POSTSUPERSCRIPT ( 1 ) end_POSTSUPERSCRIPT, both of which convey the wavevector 𝒒=𝒒1𝒒subscript𝒒1{{\bm{q}}}={{\bm{q}}}_{1}bold_italic_q = bold_italic_q start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT. Therefore, m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT is induced by the β€œvertical p-h excitation between band aπ‘Žaitalic_a and band b𝑏bitalic_b” in Fig. 3 (d) in the folded BZ. (Band c𝑐citalic_c gives no contribution for m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT.) To verify this consideration, we examine the function m⁒(π’Œ)π‘šπ’Œm({{\bm{k}}})italic_m ( bold_italic_k ) in Eq. [5]. Figure 1 (e) shows the obtained m1⁒(π’Œ)β‰‘βˆ‚2m⁒(π’Œ)/βˆ‚Ξ·1β’βˆ‚Ο•1|𝜼=Ο•=0subscriptπ‘š1π’Œevaluated-atsuperscript2π‘šπ’Œsubscriptπœ‚1subscriptitalic-Ο•1𝜼bold-italic-Ο•0m_{1}({{\bm{k}}})\equiv{\partial}^{2}m({{\bm{k}}})/{\partial}\eta_{1}{\partial% }\phi_{1}|_{{\bm{\eta}}={\bm{\phi}}=0}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ( bold_italic_k ) ≑ βˆ‚ start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_m ( bold_italic_k ) / βˆ‚ italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT βˆ‚ italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT | start_POSTSUBSCRIPT bold_italic_Ξ· = bold_italic_Ο• = 0 end_POSTSUBSCRIPT. (Note that m1βˆβˆ‘π’Œm1⁒(π’Œ)proportional-tosubscriptπ‘š1subscriptπ’Œsubscriptπ‘š1π’Œm_{1}\propto\sum_{{{\bm{k}}}}m_{1}({{\bm{k}}})italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ∝ βˆ‘ start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ( bold_italic_k ).) It is verified that large m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT originates from the FS aπ‘Žaitalic_a and FS b𝑏bitalic_b near the ΓΓ\Gammaroman_Ξ“-M line in Fig. 3 (c). This mechanism occurs for both n>nvHS𝑛subscript𝑛vHSn>n_{\rm vHS}italic_n > italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT and n<nvHS𝑛subscript𝑛vHSn<n_{\rm vHS}italic_n < italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT, so |m1|subscriptπ‘š1|m_{1}|| italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT | takes large value for a wide range of n𝑛nitalic_n. Note that m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT in Fig. 3 (a) changes its sign at n=2.44𝑛2.44n=2.44italic_n = 2.44, when π’ŒABsubscriptπ’ŒAB{{\bm{k}}}_{\rm AB}bold_italic_k start_POSTSUBSCRIPT roman_AB end_POSTSUBSCRIPT and π’ŒABβ€²subscriptsuperscriptπ’Œβ€²AB{{\bm{k}}}^{\prime}_{\rm AB}bold_italic_k start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT start_POSTSUBSCRIPT roman_AB end_POSTSUBSCRIPT in Fig.1 (b) touch in the folded BZ.

Next, we consider the origin of m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT, which is caused by the band-foldings caused by f^(1)superscript^𝑓1{\hat{f}}^{(1)}over^ start_ARG italic_f end_ARG start_POSTSUPERSCRIPT ( 1 ) end_POSTSUPERSCRIPT (at 𝒒=𝒒1𝒒subscript𝒒1{{\bm{q}}}={{\bm{q}}}_{1}bold_italic_q = bold_italic_q start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT), f^(2)superscript^𝑓2{\hat{f}}^{(2)}over^ start_ARG italic_f end_ARG start_POSTSUPERSCRIPT ( 2 ) end_POSTSUPERSCRIPT (at 𝒒=𝒒2𝒒subscript𝒒2{{\bm{q}}}={{\bm{q}}}_{2}bold_italic_q = bold_italic_q start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT) and f^(3)superscript^𝑓3{\hat{f}}^{(3)}over^ start_ARG italic_f end_ARG start_POSTSUPERSCRIPT ( 3 ) end_POSTSUPERSCRIPT (at 𝒒=𝒒3𝒒subscript𝒒3{{\bm{q}}}={{\bm{q}}}_{3}bold_italic_q = bold_italic_q start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT). This situation allows the β€œvertical p-h excitation between band a+bπ‘Žπ‘a+bitalic_a + italic_b and band c𝑐citalic_c” in Fig. 3 (d), which is significant because the band splitting between Ο΅cβ’π’Œ(>0)annotatedsubscriptitalic-Ο΅π‘π’Œabsent0{\epsilon}_{c{{\bm{k}}}}\ (>0)italic_Ο΅ start_POSTSUBSCRIPT italic_c bold_italic_k end_POSTSUBSCRIPT ( > 0 ) and Ο΅a,bβ’π’Œ(<0)annotatedsubscriptitalic-Ο΅π‘Žπ‘π’Œabsent0{\epsilon}_{a,b{{\bm{k}}}}\ (<0)italic_Ο΅ start_POSTSUBSCRIPT italic_a , italic_b bold_italic_k end_POSTSUBSCRIPT ( < 0 ) is very small. This process gives huge |m2|subscriptπ‘š2|m_{2}|| italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT | at n=nvHS𝑛subscript𝑛vHSn=n_{\rm vHS}italic_n = italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT obtained in Fig. 3 (b). Figure 3 (f) shows the obtained m2⁒(π’Œ)β‰‘βˆ‚3m⁒(π’Œ)/βˆ‚Ξ·1β’βˆ‚Ξ·2β’βˆ‚Ξ·3|𝜼=Ο•=0subscriptπ‘š2π’Œevaluated-atsuperscript3π‘šπ’Œsubscriptπœ‚1subscriptπœ‚2subscriptπœ‚3𝜼bold-italic-Ο•0m_{2}({{\bm{k}}})\equiv{\partial}^{3}m({{\bm{k}}})/{\partial}\eta_{1}{\partial% }\eta_{2}{\partial}\eta_{3}|_{{\bm{\eta}}={\bm{\phi}}=0}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ( bold_italic_k ) ≑ βˆ‚ start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT italic_m ( bold_italic_k ) / βˆ‚ italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT βˆ‚ italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT βˆ‚ italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT | start_POSTSUBSCRIPT bold_italic_Ξ· = bold_italic_Ο• = 0 end_POSTSUBSCRIPT. (Note that m2βˆβˆ‘π’Œm2⁒(π’Œ)proportional-tosubscriptπ‘š2subscriptπ’Œsubscriptπ‘š2π’Œm_{2}\propto\sum_{{\bm{k}}}m_{2}({{\bm{k}}})italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ∝ βˆ‘ start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ( bold_italic_k ).) The large m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT originates from the six band-crossing points (π’Œcrosssubscriptπ’Œcross{{\bm{k}}}_{\rm cross}bold_italic_k start_POSTSUBSCRIPT roman_cross end_POSTSUBSCRIPT) near the ΓΓ\Gammaroman_Ξ“ point, due to the vertical p-h excitation among three bands aπ‘Žaitalic_a, b𝑏bitalic_b, c𝑐citalic_c.

To summarize, large |m1|subscriptπ‘š1|m_{1}|| italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT | and |m2|subscriptπ‘š2|m_{2}|| italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT | are caused by the FS crossing near the vHS points and the M-M’ line for n∼nvHSsimilar-to𝑛subscript𝑛vHSn\sim n_{\rm vHS}italic_n ∼ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT. Therefore, the field-induced free energy Δ⁒F¯Δ¯𝐹\Delta{\bar{F}}roman_Ξ” overΒ― start_ARG italic_F end_ARG should be significant in real kagome metals. In the SI B SM , we verify that the relation m3β‰ˆβˆ’m2subscriptπ‘š3subscriptπ‘š2m_{3}\approx-m_{2}italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT β‰ˆ - italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT holds very well. This relation originates from the relation fi⁒j(m)=i⁒gi⁒j(m)superscriptsubscriptπ‘“π‘–π‘—π‘šπ‘–superscriptsubscriptπ‘”π‘–π‘—π‘šf_{ij}^{(m)}=ig_{ij}^{(m)}italic_f start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT = italic_i italic_g start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT or βˆ’i⁒gi⁒j(m)𝑖superscriptsubscriptπ‘”π‘–π‘—π‘š-ig_{ij}^{(m)}- italic_i italic_g start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT when both order parameters are composed of only the nearest bonds.

We verified in the SI E SM that the magnitudes of m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT at tβ€²=βˆ’0.08superscript𝑑′0.08t^{\prime}=-0.08italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = - 0.08 is comparable to those in Fig. 3. When tβ€²=βˆ’0.02superscript𝑑′0.02t^{\prime}=-0.02italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = - 0.02 in Fig. 3 (b), the obtained m2⁒(nvHS)subscriptπ‘š2subscript𝑛vHSm_{2}(n_{\rm vHS})italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ( italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT ) takes a large value because the p-h asymmetry around E=EvHS𝐸subscript𝐸vHSE=E_{\rm vHS}italic_E = italic_E start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT is significant. When tβ€²=βˆ’0.08superscript𝑑′0.08t^{\prime}=-0.08italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = - 0.08 in Fig. S6 (e), in contrast, m2⁒(nvHS)subscriptπ‘š2subscript𝑛vHSm_{2}(n_{\rm vHS})italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ( italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT ) is small and m2⁒(n)subscriptπ‘š2𝑛m_{2}(n)italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ( italic_n ) tends to be an odd function of nβˆ’nvHS𝑛subscript𝑛vHSn-n_{\rm vHS}italic_n - italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT because of the approximate p-h symmetry and the factor (Ο΅Ξ±β’π’Œ+Ο΅Ξ²β’π’Œβˆ’2⁒μ)subscriptitalic-Ο΅π›Όπ’Œsubscriptitalic-Ο΅π›½π’Œ2πœ‡({\epsilon}_{{\alpha}{{\bm{k}}}}+{\epsilon}_{{\beta}{{\bm{k}}}}-2\mu)( italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ± bold_italic_k end_POSTSUBSCRIPT + italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ² bold_italic_k end_POSTSUBSCRIPT - 2 italic_ΞΌ ) in Eq. [6]. (The numerical results of m2⁒(n)subscriptπ‘š2𝑛m_{2}(n)italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ( italic_n ) in the 30-orbital model in Figs. 5 (g)-(h) are closer to the results at tβ€²=βˆ’0.02superscript𝑑′0.02t^{\prime}=-0.02italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = - 0.02 near the vHS filling.)

Below, we derive the order parameters under hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT by minimizing the GL free energy F=F0+Δ⁒F¯𝐹superscript𝐹0Δ¯𝐹F=F^{0}+\Delta{\bar{F}}italic_F = italic_F start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT + roman_Ξ” overΒ― start_ARG italic_F end_ARG, where F0=FΞ·0+FΟ•0+FΞ·,Ο•0superscript𝐹0subscriptsuperscript𝐹0πœ‚subscriptsuperscript𝐹0italic-Ο•subscriptsuperscript𝐹0πœ‚italic-Ο•F^{0}=F^{0}_{\eta}+F^{0}_{\phi}+F^{0}_{\eta,\phi}italic_F start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = italic_F start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_Ξ· end_POSTSUBSCRIPT + italic_F start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_Ο• end_POSTSUBSCRIPT + italic_F start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_Ξ· , italic_Ο• end_POSTSUBSCRIPT is the free energy at hz=0subscriptβ„Žπ‘§0h_{z}=0italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 0 Balents2021 ; Tazai-kagome2 :

FΟ•0=ab⁒|Ο•|2+b1⁒ϕ1⁒ϕ2⁒ϕ3subscriptsuperscript𝐹0italic-Ο•subscriptπ‘Žbsuperscriptbold-italic-Ο•2subscript𝑏1subscriptitalic-Ο•1subscriptitalic-Ο•2subscriptitalic-Ο•3\displaystyle\!\!\!\!\!\!\!\!\!F^{0}_{\phi}=a_{\rm b}|{\bm{\phi}}|^{2}+b_{1}% \phi_{1}\phi_{2}\phi_{3}italic_F start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_Ο• end_POSTSUBSCRIPT = italic_a start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT | bold_italic_Ο• | start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_b start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT
+d1(Ο•14+Ο•24+Ο•34)+d2(Ο•12Ο•22+(cycl.)),\displaystyle+d_{1}(\phi_{1}^{4}+\phi_{2}^{4}+\phi_{3}^{4})+d_{2}(\phi_{1}^{2}% \phi_{2}^{2}+({\rm cycl.})),+ italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ( italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT + italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT + italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT ) + italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ( italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + ( roman_cycl . ) ) , (10)
Fη0=ac|𝜼|2+d3(η14+η24+η34)+d4(η12η22+(cycl.)),\displaystyle\!\!\!\!\!\!\!\!\!F^{0}_{\eta}=a_{\rm c}|{\bm{\eta}}|^{2}+d_{3}(% \eta_{1}^{4}+\eta_{2}^{4}+\eta_{3}^{4})+d_{4}(\eta_{1}^{2}\eta_{2}^{2}+({\rm cycl% .})),italic_F start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_η end_POSTSUBSCRIPT = italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT | bold_italic_η | start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ( italic_η start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT + italic_η start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT + italic_η start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT ) + italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT ( italic_η start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_η start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + ( roman_cycl . ) ) , (11)

and FΞ·,Ο•0subscriptsuperscript𝐹0πœ‚italic-Ο•F^{0}_{\eta,\phi}italic_F start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_Ξ· , italic_Ο• end_POSTSUBSCRIPT contains the current-bond cross terms proportional to Ξ·2⁒ϕ1superscriptπœ‚2superscriptitalic-Ο•1\eta^{2}\phi^{1}italic_Ξ· start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ο• start_POSTSUPERSCRIPT 1 end_POSTSUPERSCRIPT and Ξ·2⁒ϕ2superscriptπœ‚2superscriptitalic-Ο•2\eta^{2}\phi^{2}italic_Ξ· start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ο• start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT:

FΞ·,Ο•0=b2(Ο•1Ξ·2Ξ·3+(cycl.))\displaystyle\!\!\!\!\!\!\!\!\!F^{0}_{\eta,\phi}=b_{2}(\phi_{1}\eta_{2}\eta_{3% }+({\rm cycl.}))italic_F start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_Ξ· , italic_Ο• end_POSTSUBSCRIPT = italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ( italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT + ( roman_cycl . ) )
+2d5(Ο•12Ξ·12+Ο•22Ξ·22+Ο•32Ξ·32)+d6(Ο•12Ξ·22+(cycl.)).\displaystyle+2d_{5}(\phi_{1}^{2}\eta_{1}^{2}+\phi_{2}^{2}\eta_{2}^{2}+\phi_{3% }^{2}\eta_{3}^{2})+d_{6}(\phi_{1}^{2}\eta_{2}^{2}+({\rm cycl.})).+ 2 italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT ( italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ) + italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ( italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + ( roman_cycl . ) ) . (12)

Here, ac⁒(b)=rc⁒(b)⁒(Tβˆ’Tc⁒(b)0)subscriptπ‘Žcbsubscriptπ‘Ÿcb𝑇superscriptsubscript𝑇cb0a_{\rm c(b)}=r_{\rm c(b)}(T-T_{\rm c(b)}^{0})italic_a start_POSTSUBSCRIPT roman_c ( roman_b ) end_POSTSUBSCRIPT = italic_r start_POSTSUBSCRIPT roman_c ( roman_b ) end_POSTSUBSCRIPT ( italic_T - italic_T start_POSTSUBSCRIPT roman_c ( roman_b ) end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ), where Tc⁒(b)0superscriptsubscript𝑇cb0T_{\rm c(b)}^{0}italic_T start_POSTSUBSCRIPT roman_c ( roman_b ) end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT is the current-order (BO) transition temperature without other orders. Theoretically, ac⁒(b)∼N⁒(0)⁒(βˆ’1+λη⁒(Ο•)βˆ’1)similar-tosubscriptπ‘Žcb𝑁01superscriptsubscriptπœ†πœ‚italic-Ο•1a_{\rm c(b)}\sim N(0)(-1+\lambda_{\eta(\phi)}^{-1})italic_a start_POSTSUBSCRIPT roman_c ( roman_b ) end_POSTSUBSCRIPT ∼ italic_N ( 0 ) ( - 1 + italic_Ξ» start_POSTSUBSCRIPT italic_Ξ· ( italic_Ο• ) end_POSTSUBSCRIPT start_POSTSUPERSCRIPT - 1 end_POSTSUPERSCRIPT ), where N⁒(0)𝑁0N(0)italic_N ( 0 ) is the density-of-states (∼1⁒eVβˆ’1similar-toabsent1superscripteV1\sim 1\ {\rm eV^{-1}}∼ 1 roman_eV start_POSTSUPERSCRIPT - 1 end_POSTSUPERSCRIPT) and λη⁒(Ο•)subscriptπœ†πœ‚italic-Ο•\lambda_{\eta(\phi)}italic_Ξ» start_POSTSUBSCRIPT italic_Ξ· ( italic_Ο• ) end_POSTSUBSCRIPT is the DW equation eigenvalue of the current order (bond order) Tazai-LW . λη⁒(Ο•)=1subscriptπœ†πœ‚italic-Ο•1\lambda_{\eta(\phi)}=1italic_Ξ» start_POSTSUBSCRIPT italic_Ξ· ( italic_Ο• ) end_POSTSUBSCRIPT = 1 at T=Tc⁒(b)0𝑇superscriptsubscript𝑇cb0T=T_{\rm c(b)}^{0}italic_T = italic_T start_POSTSUBSCRIPT roman_c ( roman_b ) end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT, while λη⁒(Ο•)=0subscriptπœ†πœ‚italic-Ο•0\lambda_{\eta(\phi)}=0italic_Ξ» start_POSTSUBSCRIPT italic_Ξ· ( italic_Ο• ) end_POSTSUBSCRIPT = 0 (i.e., ac⁒(b)=∞subscriptπ‘Žcba_{\rm c(b)}=\inftyitalic_a start_POSTSUBSCRIPT roman_c ( roman_b ) end_POSTSUBSCRIPT = ∞) in the absence of interaction. According to Ref. Tazai-LW , rb⁒Tb0∼0.1⁒N⁒(0)similar-tosubscriptπ‘Ÿbsuperscriptsubscript𝑇b00.1𝑁0r_{\rm b}T_{\rm b}^{0}\sim 0.1N(0)italic_r start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ∼ 0.1 italic_N ( 0 ) for the nematic BO in FeSe, while the corresponding value for BCS superconductors is ∼N⁒(0)similar-toabsent𝑁0\sim N(0)∼ italic_N ( 0 ).

To discuss the phase diagram qualitatively, we set m1=5subscriptπ‘š15m_{1}=5italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = 5, m2=βˆ’m3=βˆ’1000subscriptπ‘š2subscriptπ‘š31000m_{2}=-m_{3}=-1000italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = - italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT = - 1000, both of which are moderate compared with the values in Fig. 3. We also put di=150subscript𝑑𝑖150d_{i}=150italic_d start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT = 150 (i=1βˆ’4)𝑖14(i=1-4)( italic_i = 1 - 4 ) by seeing the numerical results in the SI F SM and set rc=rb=30subscriptπ‘Ÿcsubscriptπ‘Ÿb30r_{\rm c}=r_{\rm b}=30italic_r start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT = italic_r start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT = 30. Then, both current and bond orders become 3⁒Q3𝑄3Q3 italic_Q states because of the relations 2⁒d1/d2>12subscript𝑑1subscript𝑑212d_{1}/d_{2}>12 italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT > 1 and 2⁒d3/d4>12subscript𝑑3subscript𝑑412d_{3}/d_{4}>12 italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT > 1. A more detailed explanation for the GL parameters is presented in the SI F SM .

hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-effect on 3⁒Q3𝑄3Q3 italic_Q current order state

In this section, we demonstrate that the 3⁒Q3𝑄3Q3 italic_Q current order state is drastically modified by hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT. Figure 4 (a) shows the obtained 3⁒Q3𝑄3Q3 italic_Q current order 𝜼=(Ξ·1,Ξ·2,Ξ·3)𝜼subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3{\bm{\eta}}=(\eta_{1},\eta_{2},\eta_{3})bold_italic_Ξ· = ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ) with Ξ·1=Ξ·2=Ξ·3subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3\eta_{1}=\eta_{2}=\eta_{3}italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT, when Tc0=0.01superscriptsubscript𝑇c00.01T_{\rm c}^{0}=0.01italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = 0.01 in the absence of BO (Tb0=βˆ’βˆžsuperscriptsubscript𝑇b0T_{\rm b}^{0}=-\inftyitalic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = - ∞). Here, we set hz=0∼2Γ—10βˆ’3subscriptβ„Žπ‘§0similar-to2superscript103h_{z}=0\sim 2\times 10^{-3}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 0 ∼ 2 Γ— 10 start_POSTSUPERSCRIPT - 3 end_POSTSUPERSCRIPT. Because of the Ξ·3superscriptπœ‚3\eta^{3}italic_Ξ· start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT-term by m2=βˆ’1000subscriptπ‘š21000m_{2}=-1000italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = - 1000, 3⁒Q3𝑄3Q3 italic_Q current order 𝜼=(Ξ·,Ξ·,Ξ·)/3πœΌπœ‚πœ‚πœ‚3{\bm{\eta}}=(\eta,\eta,\eta)/\sqrt{3}bold_italic_Ξ· = ( italic_Ξ· , italic_Ξ· , italic_Ξ· ) / square-root start_ARG 3 end_ARG with negative Ξ·πœ‚\etaitalic_Ξ· is stabilized in the presence of hz>0subscriptβ„Žπ‘§0h_{z}>0italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT > 0. In addition, the field-induced first-order transition occurs at T=Tc0+Δ⁒Tc𝑇superscriptsubscript𝑇c0Ξ”subscript𝑇cT=T_{\rm c}^{0}+\Delta T_{\rm c}italic_T = italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT + roman_Ξ” italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT, where Δ⁒Tc=(hz⁒m2)2/4⁒rc⁒(d3+d4)Ξ”subscript𝑇csuperscriptsubscriptβ„Žπ‘§subscriptπ‘š224subscriptπ‘Ÿcsubscript𝑑3subscript𝑑4\Delta T_{\rm c}=(h_{z}m_{2})^{2}/4r_{\rm c}(d_{3}+d_{4})roman_Ξ” italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT = ( italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT / 4 italic_r start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT + italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT ) is quite small.

Next, we demonstrate the hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-induced 3⁒Q3𝑄3Q3 italic_Q current order state above Tc0superscriptsubscript𝑇c0T_{\rm c}^{0}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT inside the BO state. Under the 3⁒Q3𝑄3Q3 italic_Q BO phase Ο•=(Ο•,Ο•,Ο•)/3bold-italic-Ο•italic-Ο•italic-Ο•italic-Ο•3{\bm{\phi}}=(\phi,\phi,\phi)/\sqrt{3}bold_italic_Ο• = ( italic_Ο• , italic_Ο• , italic_Ο• ) / square-root start_ARG 3 end_ARG, the 2nd order GL coefficient acsubscriptπ‘Žca_{\rm c}italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT in Eq. [11] is renormalized as aΒ―c=ac+(d5+d6)⁒(2⁒ϕ2/3)subscriptΒ―π‘Žcsubscriptπ‘Žcsubscript𝑑5subscript𝑑62superscriptitalic-Ο•23{\bar{a}}_{\rm c}=a_{\rm c}+(d_{5}+d_{6})(2\phi^{2}/3)overΒ― start_ARG italic_a end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT = italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT + ( italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT + italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ) ( 2 italic_Ο• start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT / 3 ) due to the d5subscript𝑑5d_{5}italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT, d6subscript𝑑6d_{6}italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT terms. It is given as aΒ―c=rΒ―c⁒(Tβˆ’TΒ―c0)subscriptΒ―π‘ŽcsubscriptΒ―π‘Ÿc𝑇superscriptsubscript¯𝑇c0{\bar{a}}_{\rm c}={\bar{r}}_{\rm c}(T-{\bar{T}}_{\rm c}^{0})overΒ― start_ARG italic_a end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT = overΒ― start_ARG italic_r end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( italic_T - overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ), where TΒ―c0<Tc0superscriptsubscript¯𝑇c0superscriptsubscript𝑇c0{\bar{T}}_{\rm c}^{0}<T_{\rm c}^{0}overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT < italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT and rΒ―c<rcsubscriptΒ―π‘Ÿcsubscriptπ‘Ÿc{\bar{r}}_{\rm c}<r_{\rm c}overΒ― start_ARG italic_r end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT < italic_r start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT as we explain in the SI F SM . The smallness of rΒ―csubscriptΒ―π‘Ÿc{\bar{r}}_{\rm c}overΒ― start_ARG italic_r end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT is favorable for the hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-induced current order. In this subsection, we simply denote TΒ―c0superscriptsubscript¯𝑇c0{\bar{T}}_{\rm c}^{0}overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT and rΒ―csubscriptΒ―π‘Ÿc{\bar{r}}_{\rm c}overΒ― start_ARG italic_r end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT as Tc0superscriptsubscript𝑇c0T_{\rm c}^{0}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT and rcsubscriptπ‘Ÿcr_{\rm c}italic_r start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT, respectively.

First, we drop the 3rd order GL terms bisubscript𝑏𝑖b_{i}italic_b start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT (i=1,2𝑖12i=1,2italic_i = 1 , 2) to concentrate on the field-induced novel phenomena. Figure 4 (b) exhibits the 3⁒Q3𝑄3Q3 italic_Q current order with Tc0=0.005superscriptsubscript𝑇c00.005T_{\rm c}^{0}=0.005italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = 0.005, in the presence of the BO phase below Tb0=0.01superscriptsubscript𝑇b00.01T_{\rm b}^{0}=0.01italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = 0.01. For hz=1Γ—10βˆ’4subscriptβ„Žπ‘§1superscript104h_{z}=1\times 10^{-4}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 1 Γ— 10 start_POSTSUPERSCRIPT - 4 end_POSTSUPERSCRIPT and 2Γ—10βˆ’42superscript1042\times 10^{-4}2 Γ— 10 start_POSTSUPERSCRIPT - 4 end_POSTSUPERSCRIPT, Ξ·πœ‚\etaitalic_Ξ· starts to emerge just below Tb0superscriptsubscript𝑇b0T_{\rm b}^{0}italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT thanks to the Ξ·πœ‚\etaitalic_Ξ·-linear term Δ⁒FΒ―=βˆ’3⁒hz⁒(m1⁒ϕ+m3⁒ϕ2)⁒ηΔ¯𝐹3subscriptβ„Žπ‘§subscriptπ‘š1italic-Ο•subscriptπ‘š3superscriptitalic-Ο•2πœ‚\Delta{\bar{F}}=-3h_{z}(m_{1}\phi+m_{3}\phi^{2})\etaroman_Ξ” overΒ― start_ARG italic_F end_ARG = - 3 italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT ( italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ο• + italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT italic_Ο• start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ) italic_Ξ·, and therefore Tc=Tb0subscript𝑇csuperscriptsubscript𝑇b0T_{\rm c}=T_{\rm b}^{0}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT = italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT. When hzβ‰ 0subscriptβ„Žπ‘§0h_{z}\neq 0italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT β‰  0, Tc0superscriptsubscript𝑇c0T_{\rm c}^{0}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT is just a crossover temperature. To understand the role of the non-analytic term qualitatively, we analyze a simple GL free energy with a Ξ·πœ‚\etaitalic_Ξ·-linear term in the SI C SM : It is found that the field-induced current order |𝜼ind|subscript𝜼ind|{\bm{\eta}}_{{\rm ind}}|| bold_italic_Ξ· start_POSTSUBSCRIPT roman_ind end_POSTSUBSCRIPT | under the BO Ο•bold-italic-Ο•{\bm{\phi}}bold_italic_Ο• is approximately given as

|𝜼ind|β‰ˆ9⁒|m1⁒hz⁒ϕ|/2⁒ac.subscript𝜼ind9subscriptπ‘š1subscriptβ„Žπ‘§bold-italic-Ο•2subscriptπ‘Žc\displaystyle|{\bm{\eta}}_{{\rm ind}}|\approx 9|m_{1}h_{z}{\bm{\phi}}|/2a_{\rm c}.| bold_italic_Ξ· start_POSTSUBSCRIPT roman_ind end_POSTSUBSCRIPT | β‰ˆ 9 | italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT bold_italic_Ο• | / 2 italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT . (13)

Thus, the field-induced |𝜼ind|subscript𝜼ind|{\bm{\eta}}_{{\rm ind}}|| bold_italic_Ξ· start_POSTSUBSCRIPT roman_ind end_POSTSUBSCRIPT | is prominent when the system at hz=0subscriptβ„Žπ‘§0h_{z}=0italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 0 is close to the current order state (i.e., ac≳0greater-than-or-equivalent-tosubscriptπ‘Žc0a_{\rm c}\gtrsim 0italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ≳ 0). The expression of |𝜼ind|subscript𝜼ind|{\bm{\eta}}_{{\rm ind}}|| bold_italic_Ξ· start_POSTSUBSCRIPT roman_ind end_POSTSUBSCRIPT | in Eq. [13] is proportional to Ο•bold-italic-Ο•{\bm{\phi}}bold_italic_Ο• in contrast to Eq. (31) of Ref. Balents2021 .

A schematic picture for the field-induced current order is shown in Fig. 4 (c). In the 3⁒Q3𝑄3Q3 italic_Q BO phase Ο•βˆ(1,1,1)proportional-tobold-italic-Ο•111{\bm{\phi}}\propto(1,1,1)bold_italic_Ο• ∝ ( 1 , 1 , 1 ), the field-induced 𝜼𝜼{\bm{\eta}}bold_italic_Ξ· above Tc0superscriptsubscript𝑇c0T_{\rm c}^{0}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT is proportional to Ο•bold-italic-Ο•{\bm{\phi}}bold_italic_Ο• to maximize MΒ―orbβˆΟ•β‹…πœΌproportional-tosubscript¯𝑀orbβ‹…bold-italic-Ο•πœΌ{\bar{M}}_{\rm orb}\propto{\bm{\phi}}\cdot{\bm{\eta}}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT ∝ bold_italic_Ο• β‹… bold_italic_Ξ·. This coexisting state (πœΌβˆΟ•proportional-to𝜼bold-italic-Ο•{\bm{\eta}}\propto{\bm{\phi}}bold_italic_Ξ· ∝ bold_italic_Ο•) has C6subscript𝐢6C_{6}italic_C start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT symmetry; see Ref. Tazai-kagome2 .

Refer to caption
Figure 4: (a)(b) Obtained 3⁒Q3𝑄3Q3 italic_Q current order 𝜼=(Ξ·1,Ξ·2,Ξ·3)𝜼subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3{\bm{\eta}}=(\eta_{1},\eta_{2},\eta_{3})bold_italic_Ξ· = ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ) with Ξ·1=Ξ·2=Ξ·3subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3\eta_{1}=\eta_{2}=\eta_{3}italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT for 2⁒d1/d2=22subscript𝑑1subscript𝑑222d_{1}/d_{2}=22 italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = 2, 2⁒d3/d4=22subscript𝑑3subscript𝑑422d_{3}/d_{4}=22 italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT = 2 and b1=b2=0subscript𝑏1subscript𝑏20b_{1}=b_{2}=0italic_b start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = 0 under hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT. hz=10βˆ’4subscriptβ„Žπ‘§superscript104h_{z}=10^{-4}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 10 start_POSTSUPERSCRIPT - 4 end_POSTSUPERSCRIPT corresponds to 1 Tesla. (a) 3⁒Q3𝑄3Q3 italic_Q current order for Tc0=0.01superscriptsubscript𝑇c00.01T_{\rm c}^{0}=0.01italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = 0.01 and Tb0=βˆ’βˆžsuperscriptsubscript𝑇b0T_{\rm b}^{0}=-\inftyitalic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = - ∞. The first-order transition is induced by hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT due to m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT-term. (b) 3⁒Q3𝑄3Q3 italic_Q current order for Tc0=0.005superscriptsubscript𝑇c00.005T_{\rm c}^{0}=0.005italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = 0.005 under the 3⁒Q3𝑄3Q3 italic_Q BO Ο•βˆ(1,1,1)proportional-tobold-italic-Ο•111{\bm{\phi}}\propto(1,1,1)bold_italic_Ο• ∝ ( 1 , 1 , 1 ) with Tb0=0.01superscriptsubscript𝑇b00.01T_{\rm b}^{0}=0.01italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = 0.01. For hzβ‰ 0subscriptβ„Žπ‘§0h_{z}\neq 0italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT β‰  0, Ξ·πœ‚\etaitalic_Ξ· starts to increase below Tbsubscript𝑇bT_{\rm b}italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT thanks to m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT-term, i.e., Tc=Tb0subscript𝑇csuperscriptsubscript𝑇b0T_{\rm c}=T_{\rm b}^{0}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT = italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT. (c) hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-induced C6subscript𝐢6C_{6}italic_C start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT-symmetry bond+current order above Tc0superscriptsubscript𝑇c0T_{\rm c}^{0}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT the shaded area in (b). (d) 3⁒Q3𝑄3Q3 italic_Q current order for βˆ’b1=b2=1.0subscript𝑏1subscript𝑏21.0-b_{1}=b_{2}=1.0- italic_b start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = 1.0. Other model parameters are equal to (b). The symmetry of the bond+current coexisting state is C6subscript𝐢6C_{6}italic_C start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT (Ξ·1=Ξ·2=Ξ·3subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3\eta_{1}=\eta_{2}=\eta_{3}italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT) for T>Tc0𝑇superscriptsubscript𝑇c0T>T_{\rm c}^{0}italic_T > italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT, while Ccsubscript𝐢𝑐C_{c}italic_C start_POSTSUBSCRIPT italic_c end_POSTSUBSCRIPT (Ξ·1=Ξ·2β‰ˆβˆ’Ξ·3subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3\eta_{1}=\eta_{2}\approx-\eta_{3}italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT β‰ˆ - italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT) for <Tc0absentsuperscriptsubscript𝑇c0<T_{\rm c}^{0}< italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT. (e) Bond+current coexisting state below Tb0superscriptsubscript𝑇b0T_{\rm b}^{0}italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT. It has C2subscript𝐢2C_{2}italic_C start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT symmetry to gain the 3rd order GL terms. (f) Obtained current order Ξ·msubscriptπœ‚π‘š\eta_{m}italic_Ξ· start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT for 2⁒d3/d4=1/1.22subscript𝑑3subscript𝑑411.22d_{3}/d_{4}=1/1.22 italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT = 1 / 1.2 at Tc0=0.01superscriptsubscript𝑇c00.01T_{\rm c}^{0}=0.01italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = 0.01 and hz=5Γ—10βˆ’4subscriptβ„Žπ‘§5superscript104h_{z}=5\times 10^{-4}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 5 Γ— 10 start_POSTSUPERSCRIPT - 4 end_POSTSUPERSCRIPT, for Tb0=βˆ’βˆžsuperscriptsubscript𝑇b0T_{\rm b}^{0}=-\inftyitalic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = - ∞. The 3⁒Q3𝑄3Q3 italic_Q current order appears due to the m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT-term, while it changes to 1⁒Q1𝑄1Q1 italic_Q order at Tβ‰ˆ0.008𝑇0.008T\approx 0.008italic_T β‰ˆ 0.008.

Next, we consider the effect of 3rd order GL terms in Fig. 4 (d), by setting βˆ’b1=b2=1.0subscript𝑏1subscript𝑏21.0-b_{1}=b_{2}=1.0- italic_b start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = 1.0. (The relation b1⁒b2<0subscript𝑏1subscript𝑏20b_{1}b_{2}<0italic_b start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT < 0 is general Tazai-kagome2 .) Due to the energy gain from the b1subscript𝑏1b_{1}italic_b start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT-term, 3Q BO Ο•βˆ(1,1,1)proportional-tobold-italic-Ο•111{\bm{\phi}}\propto(1,1,1)bold_italic_Ο• ∝ ( 1 , 1 , 1 ) appears at Tβ‰ˆTb0=0.01𝑇superscriptsubscript𝑇b00.01T\approx T_{\rm b}^{0}=0.01italic_T β‰ˆ italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = 0.01 as the first order transition Balents2021 . At hz=0subscriptβ„Žπ‘§0h_{z}=0italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 0, the 3Q current order 𝜼=(Ξ·,Ξ·,βˆ’Ξ·β€²)πœΌπœ‚πœ‚superscriptπœ‚β€²{\bm{\eta}}=(\eta,\eta,-\eta^{\prime})bold_italic_Ξ· = ( italic_Ξ· , italic_Ξ· , - italic_Ξ· start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ) appears below Tc0=0.005superscriptsubscript𝑇c00.005T_{\rm c}^{0}=0.005italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = 0.005 to maximize the energy gain from the b2subscript𝑏2b_{2}italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT term (b1⁒b2<0subscript𝑏1subscript𝑏20b_{1}b_{2}<0italic_b start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT < 0), as explained in Ref. Tazai-kagome2 . Just below Tc0superscriptsubscript𝑇c0T_{\rm c}^{0}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT (|Ο•|≫|𝜼|much-greater-thanbold-italic-Ο•πœΌ|{\bm{\phi}}|\gg|{\bm{\eta}}|| bold_italic_Ο• | ≫ | bold_italic_Ξ· |), Ξ·β€²=2⁒ηsuperscriptπœ‚β€²2πœ‚\eta^{\prime}=2\etaitalic_Ξ· start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 2 italic_Ξ· is satisfied. (More generally, 𝜼∝(Ξ·1,Ξ·2,βˆ’Ξ·1βˆ’Ξ·2)proportional-to𝜼subscriptπœ‚1subscriptπœ‚2subscriptπœ‚1subscriptπœ‚2{\bm{\eta}}\propto(\eta_{1},\eta_{2},-\eta_{1}-\eta_{2})bold_italic_Ξ· ∝ ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , - italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT - italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ).) Figure 4 (e) depicts the β€œnematic” bond+current coexisting state below Tc0superscriptsubscript𝑇c0T_{\rm c}^{0}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT with 𝜼∼(1,1,βˆ’1)similar-to𝜼111{\bm{\eta}}\sim(1,1,-1)bold_italic_Ξ· ∼ ( 1 , 1 , - 1 ) and Ο•βˆΌ(1,1,1)similar-tobold-italic-Ο•111{\bm{\phi}}\sim(1,1,1)bold_italic_Ο• ∼ ( 1 , 1 , 1 ); see Ref. Tazai-kagome2 for detail.

For hz=4Γ—10βˆ’4subscriptβ„Žπ‘§4superscript104h_{z}=4\times 10^{-4}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 4 Γ— 10 start_POSTSUPERSCRIPT - 4 end_POSTSUPERSCRIPT in Fig. 4 (d), the field-induced 3⁒Q3𝑄3Q3 italic_Q current order 𝜼∝(1,1,1)proportional-to𝜼111{\bm{\eta}}\propto(1,1,1)bold_italic_Ξ· ∝ ( 1 , 1 , 1 ) start to emerge just below Tb0superscriptsubscript𝑇b0T_{\rm b}^{0}italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT, similarly to Fig. 4 (b). The realized C6subscript𝐢6C_{6}italic_C start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT symmetry coexisting state above Tc0superscriptsubscript𝑇c0T_{\rm c}^{0}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT is shown in Fig. 4 (c). Below Tc0superscriptsubscript𝑇c0T_{\rm c}^{0}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT, the coexisting state changes to the nematic (C2subscript𝐢2C_{2}italic_C start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT) bond+current state shown in Figure 4 (e). We stress that sgn⁒(Ξ·1⁒η2⁒η3)sgnsubscriptπœ‚1subscriptπœ‚2subscriptπœ‚3{\rm sgn}(\eta_{1}\eta_{2}\eta_{3})roman_sgn ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ) changes at Tβ‰ˆTc0𝑇superscriptsubscript𝑇c0T\approx T_{\rm c}^{0}italic_T β‰ˆ italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT. To summarize, the hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-induced coexisting state changes its symmetry from C6subscript𝐢6C_{6}italic_C start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT (T>Tc0𝑇superscriptsubscript𝑇c0T>T_{\rm c}^{0}italic_T > italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT) to C2subscript𝐢2C_{2}italic_C start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT (T<Tc0𝑇superscriptsubscript𝑇c0T<T_{\rm c}^{0}italic_T < italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT) with decreasing T𝑇Titalic_T. The field-induced first-order phase transition occurs at Tβ‰ˆTc0)T\approx T_{\rm c}^{0})italic_T β‰ˆ italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ).

hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-effect on 1⁒Q1𝑄1Q1 italic_Q current order state

In this scetion, we explain that the 1⁒Q1𝑄1Q1 italic_Q current order state is also drastically modified by hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT. Here, we increase d4=150subscript𝑑4150d_{4}=150italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT = 150 to 360360360360 (i.e., 2⁒d3/d4=1/1.22subscript𝑑3subscript𝑑411.22d_{3}/d_{4}=1/1.22 italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT = 1 / 1.2) in the previous GL parameters. A clear evidence of the 1⁒Q1𝑄1Q1 italic_Q current order has been reported recently in Ref. Asaba . We set Tc0=0.01superscriptsubscript𝑇c00.01T_{\rm c}^{0}=0.01italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = 0.01. Figure 4 (f) shows the obtained order parameters without BO (Tb0=βˆ’βˆžsuperscriptsubscript𝑇b0T_{\rm b}^{0}=-\inftyitalic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = - ∞) at hz=5Γ—10βˆ’4subscriptβ„Žπ‘§5superscript104h_{z}=5\times 10^{-4}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 5 Γ— 10 start_POSTSUPERSCRIPT - 4 end_POSTSUPERSCRIPT. We obtain the field-induced 3⁒Q3𝑄3Q3 italic_Q-current order due to the m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT-term, while it changes to 1⁒Q1𝑄1Q1 italic_Q order at Tβ‰ˆ0.008𝑇0.008T\approx 0.008italic_T β‰ˆ 0.008. The field-induced first-order transitions reported in Ref. Asaba would originate from the 1⁒Q1𝑄1Q1 italic_Q current order at Tβˆ—βˆ—(>Tb0)annotatedsuperscript𝑇absentabsentsuperscriptsubscript𝑇b0T^{**}\ (>T_{\rm b}^{0})italic_T start_POSTSUPERSCRIPT βˆ— βˆ— end_POSTSUPERSCRIPT ( > italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ) together with Δ⁒F¯Δ¯𝐹\Delta{\bar{F}}roman_Ξ” overΒ― start_ARG italic_F end_ARG. In Fig. 4 (f), we set 2⁒d3/d4=1.22subscript𝑑3subscript𝑑41.22d_{3}/d_{4}=1.22 italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT = 1.2 and m2=1000subscriptπ‘š21000m_{2}=1000italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = 1000. The obtained hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-induced 3⁒Q3𝑄3Q3 italic_Q current order is realized even when m2β‰ˆ100subscriptπ‘š2100m_{2}\approx 100italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT β‰ˆ 100 when 2⁒d3/d4≳1greater-than-or-equivalent-to2subscript𝑑3subscript𝑑412d_{3}/d_{4}\gtrsim 12 italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT ≳ 1.

In Fig. 4 (f), we set Tb0=βˆ’βˆžsuperscriptsubscript𝑇b0T_{\rm b}^{0}=-\inftyitalic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = - ∞ for simplicity. However, Tb0β‰ˆ0.01(β‰ˆ100⁒K)superscriptsubscript𝑇b0annotated0.01absent100KT_{\rm b}^{0}\approx 0.01\ (\approx 100\ {\rm K})italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT β‰ˆ 0.01 ( β‰ˆ 100 roman_K ) experimentally. In this case, as revealed in Ref. Balents2021 , the 3⁒Q3𝑄3Q3 italic_Q current order induces the finite 3⁒Q3𝑄3Q3 italic_Q BO even above Tb0superscriptsubscript𝑇b0T_{\rm b}^{0}italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT via the b2subscript𝑏2b_{2}italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT term. This fact means that the 3⁒Q3𝑄3Q3 italic_Q current order is energetically favorable when b2β‰ 0subscript𝑏20b_{2}\neq 0italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT β‰  0. Therefore, hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-induced transition from 1⁒Q1𝑄1Q1 italic_Q to 3⁒Q3𝑄3Q3 italic_Q current order shown in Fig. 4 (f) can emerge even if 2⁒d3/d4≲1less-than-or-similar-to2subscript𝑑3subscript𝑑412d_{3}/d_{4}\lesssim 12 italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT ≲ 1 for b2β‰ 0subscript𝑏20b_{2}\neq 0italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT β‰  0.

For reference, we studied the case of the intra-original-unit-cell current order (𝒒=πŸŽπ’’0{{\bm{q}}}={\bm{0}}bold_italic_q = bold_0) in kagome lattice in the SI G SM . In this case, Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT is Ξ·πœ‚\etaitalic_Ξ·-linear even if the BO is absent.

Derivation of GL coefficients based on the first-priciples kagome metal model

In the next stage, we calculate the GL coefficients based on the first-principles tight-binding model for kagome metals. We reveal that |m1|subscriptπ‘š1|m_{1}|| italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT | and |m2|subscriptπ‘š2|m_{2}|| italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT | becomes large due to the β€œinter-orbital (dX⁒Z+dY⁒Zsubscript𝑑𝑋𝑍subscriptπ‘‘π‘Œπ‘d_{XZ}+d_{YZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT + italic_d start_POSTSUBSCRIPT italic_Y italic_Z end_POSTSUBSCRIPT) mixture” even if the current order emerges only in the dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital.

First, we derive 30-orbital (15 d𝑑ditalic_d-orbital and 15 p𝑝pitalic_p-orbitals) tight-binding model for CsV3Sb5 by using WIEN2k and Wannier90 softwares. Figure 5 (a) shows the FS of the obtained model in the kz=0subscriptπ‘˜π‘§0k_{z}=0italic_k start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 0 plane. (Its bandstructure is shown in Fig. S9 (a).) The dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital β€œpure-type” FS corresponds to the FS of the present three-orbital model. Its vHS energy is located at EvHSX⁒Zβ‰ˆβˆ’0.1superscriptsubscript𝐸vHS𝑋𝑍0.1E_{\rm vHS}^{XZ}\approx-0.1italic_E start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_X italic_Z end_POSTSUPERSCRIPT β‰ˆ - 0.1. Also, the dY⁒Zsubscriptπ‘‘π‘Œπ‘d_{YZ}italic_d start_POSTSUBSCRIPT italic_Y italic_Z end_POSTSUBSCRIPT-orbital forms the β€œmix-type” FS whose vHS energy is EvHSY⁒Zβ‰ˆ+0.1superscriptsubscript𝐸vHSπ‘Œπ‘0.1E_{\rm vHS}^{YZ}\approx+0.1italic_E start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_Y italic_Z end_POSTSUPERSCRIPT β‰ˆ + 0.1. In addition, both dX2βˆ’Y2subscript𝑑superscript𝑋2superscriptπ‘Œ2d_{X^{2}-Y^{2}}italic_d start_POSTSUBSCRIPT italic_X start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Y start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT and d3⁒Z2βˆ’R2subscript𝑑3superscript𝑍2superscript𝑅2d_{3Z^{2}-R^{2}}italic_d start_POSTSUBSCRIPT 3 italic_Z start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_R start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT orbitals form one pure-type band with the vHS energy EvHSX2βˆ’Y2β‰ˆ0superscriptsubscript𝐸vHSsuperscript𝑋2superscriptπ‘Œ20E_{\rm vHS}^{X^{2}-Y^{2}}\approx 0italic_E start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_X start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Y start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT β‰ˆ 0.

Figure 5 (b) shows the FS with introducing the p𝑝pitalic_p-orbital shift Δ⁒Ep=βˆ’0.2Ξ”subscript𝐸𝑝0.2\Delta E_{p}=-0.2roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2 eV. Its bandstructure is shown in Fig. S9 (b). Here, the dY⁒Zsubscriptπ‘‘π‘Œπ‘d_{YZ}italic_d start_POSTSUBSCRIPT italic_Y italic_Z end_POSTSUBSCRIPT-orbital (mix-type) FS approaches the M point, and its vHS point shifts to the Fermi level, consistently with the ARPES measurement in Ref. ARPES-VHS . Figure 5 (c) shows the change in the FS topology due to the pure-mix hybridization in the Ep=βˆ’0.2subscript𝐸𝑝0.2E_{p}=-0.2italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2 model for n=30.48∼31.00𝑛30.48similar-to31.00n=30.48\sim 31.00italic_n = 30.48 ∼ 31.00. When n=nvHSX⁒Z(=30.48)𝑛annotatedsuperscriptsubscript𝑛vHS𝑋𝑍absent30.48n=n_{\rm vHS}^{XZ}\ (=30.48)italic_n = italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_X italic_Z end_POSTSUPERSCRIPT ( = 30.48 ), we obtain two hole-pockets that attach the M point. With increasing n𝑛nitalic_n, it changes to the electron-like Dirac pockets at nβ‰ˆnD(=30.66)𝑛annotatedsubscript𝑛Dabsent30.66n\approx n_{\rm D}\ (=30.66)italic_n β‰ˆ italic_n start_POSTSUBSCRIPT roman_D end_POSTSUBSCRIPT ( = 30.66 ), composed of both dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT and dY⁒Zsubscriptπ‘‘π‘Œπ‘d_{YZ}italic_d start_POSTSUBSCRIPT italic_Y italic_Z end_POSTSUBSCRIPT orbitals. At n=nL(=30.76)𝑛annotatedsubscript𝑛Labsent30.76n=n_{\rm L}\ (=30.76)italic_n = italic_n start_POSTSUBSCRIPT roman_L end_POSTSUBSCRIPT ( = 30.76 ), large hole-like pure-type FS and large electron-like mix-type FS are formed. At n=31𝑛31n=31italic_n = 31, the mix-type FS lines on the M-M’ line. We will see that large |m1|subscriptπ‘š1|m_{1}|| italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT | and |m2|subscriptπ‘š2|m_{2}|| italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT | appear at nβ‰ˆnvHS𝑛subscript𝑛vHSn\approx n_{\rm vHS}italic_n β‰ˆ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT, nDsubscript𝑛Dn_{\rm D}italic_n start_POSTSUBSCRIPT roman_D end_POSTSUBSCRIPT and 31313131.

Figure 5 (d) shows the obtained m2⁒(π’Œ)subscriptπ‘š2π’Œm_{2}({{\bm{k}}})italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ( bold_italic_k ) with the folded FSs. For both n=nD𝑛subscript𝑛Dn=n_{\rm D}italic_n = italic_n start_POSTSUBSCRIPT roman_D end_POSTSUBSCRIPT and 31313131, the pure-mix hybridization contributes to the large GL coefficient in the 30-orbital model. This mechanism is absent in the simple single orbital model.

We note that EvHSX⁒Zβ‰ˆ0superscriptsubscript𝐸vHS𝑋𝑍0E_{\rm vHS}^{XZ}\approx 0italic_E start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_X italic_Z end_POSTSUPERSCRIPT β‰ˆ 0 is reported in an ARPES study for pristine CsV3Sb5 in Ref. Sato-ARPES . In this case, both m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT become very large theoretically.

Now, we calculate m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT in MΒ―orbsubscript¯𝑀orb{\bar{M}}_{\rm orb}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT by introducing the current and bond orders on the dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT orbitals in the 30-orbital kagome lattice model given in Figs. 5 (a) (Δ⁒Ep=0Ξ”subscript𝐸𝑝0\Delta E_{p}=0roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = 0) and (b) (Δ⁒Ep=βˆ’0.2Ξ”subscript𝐸𝑝0.2\Delta E_{p}=-0.2roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2). We derive the coefficients m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT defined as MΒ―orb=m1⁒ϕ~β‹…πœΌ~+m2⁒η1~⁒η2~⁒η3~subscript¯𝑀orbβ‹…subscriptπ‘š1~bold-italic-Ο•~𝜼subscriptπ‘š2~subscriptπœ‚1~subscriptπœ‚2~subscriptπœ‚3{\bar{M}}_{\rm orb}=m_{1}{\tilde{\bm{\phi}}}\cdot{\tilde{\bm{\eta}}}+m_{2}{% \tilde{\eta_{1}}}{\tilde{\eta_{2}}}{\tilde{\eta_{3}}}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT = italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT over~ start_ARG bold_italic_Ο• end_ARG β‹… over~ start_ARG bold_italic_Ξ· end_ARG + italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT over~ start_ARG italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT end_ARG over~ start_ARG italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT end_ARG over~ start_ARG italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT end_ARG, where Ξ·i~~subscriptπœ‚π‘–\tilde{\eta_{i}}over~ start_ARG italic_Ξ· start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT end_ARG and Ο•i~~subscriptitalic-ϕ𝑖\tilde{\phi_{i}}over~ start_ARG italic_Ο• start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT end_ARG are the dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital order parameters projected on the pure-type band. (We set Ξ·i~=Ξ·i⁒WX⁒Z~subscriptπœ‚π‘–subscriptπœ‚π‘–subscriptπ‘Šπ‘‹π‘\tilde{\eta_{i}}=\eta_{i}W_{XZ}over~ start_ARG italic_Ξ· start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT end_ARG = italic_Ξ· start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT italic_W start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT and Ο•i~=Ο•i⁒WX⁒Z~subscriptitalic-ϕ𝑖subscriptitalic-ϕ𝑖subscriptπ‘Šπ‘‹π‘\tilde{\phi_{i}}=\phi_{i}W_{XZ}over~ start_ARG italic_Ο• start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT end_ARG = italic_Ο• start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT italic_W start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT with WX⁒Z=0.7subscriptπ‘Šπ‘‹π‘0.7W_{XZ}=0.7italic_W start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT = 0.7, which is the dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-weight on the pure-type band.)

The obtained m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT is shown in Figs. 5 (e) (Δ⁒Ep=0Ξ”subscript𝐸𝑝0\Delta E_{p}=0roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = 0) and (f) (Δ⁒Ep=βˆ’0.2Ξ”subscript𝐸𝑝0.2\Delta E_{p}=-0.2roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2), as function of the electron filling n𝑛nitalic_n. Here, m1⁒(2)subscriptπ‘š12m_{1(2)}italic_m start_POSTSUBSCRIPT 1 ( 2 ) end_POSTSUBSCRIPT is derived from the kz=0subscriptπ‘˜π‘§0k_{z}=0italic_k start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 0 plane electronic structure. First, we discuss the case n=31𝑛31n=31italic_n = 31 that corresponds to undoped CsV3Sb5. In both Figs. 5 (e) and (f), the obtained m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT is very larger for nβ‰ˆnvHSX⁒Z(<31)𝑛annotatedsuperscriptsubscript𝑛vHS𝑋𝑍absent31n\approx n_{\rm vHS}^{XZ}\ (<31)italic_n β‰ˆ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_X italic_Z end_POSTSUPERSCRIPT ( < 31 ). At n=31𝑛31n=31italic_n = 31, m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT in Fig. 5 (e) is relatively small (βˆΌβˆ’0.5similar-toabsent0.5\sim-0.5∼ - 0.5) accidentally, while its magnitude becomes large (m1βˆΌβˆ’2similar-tosubscriptπ‘š12m_{1}\sim-2italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ∼ - 2) in Fig. 5 (f), where the mix-type FS is closer to the M-M’ line. In fact, the mix-type FS contains finite dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-weight (∼10similar-toabsent10\sim 10∼ 10%) owing to the inter-orbital mixture in kagome metal bandstructure. For this reason, the mix-type band can cause large Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT even if the current order parameter occurs only in dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital. Thus, the current-bond-field trillinear coupling due to m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT term will cause the drastic field-induced chiral current order shown in Fig. 4 (b). This is the main result of this study.

The obtained m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT is shown in Figs. 5 (g) (Δ⁒Ep=0Ξ”subscript𝐸𝑝0\Delta E_{p}=0roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = 0) and (h) (Δ⁒Ep=βˆ’0.2Ξ”subscript𝐸𝑝0.2\Delta E_{p}=-0.2roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2), at T=0.01𝑇0.01T=0.01italic_T = 0.01 and 0.0050.0050.0050.005. The obtained m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT is extremely larger for nβ‰ˆnvHSX⁒Z(<31)𝑛annotatedsuperscriptsubscript𝑛vHS𝑋𝑍absent31n\approx n_{\rm vHS}^{XZ}\ (<31)italic_n β‰ˆ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_X italic_Z end_POSTSUPERSCRIPT ( < 31 ). At n=31𝑛31n=31italic_n = 31, m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT becomes relatively small (βˆΌβˆ’50similar-toabsent50\sim-50∼ - 50) in Fig. 5 (g). However, m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT increases to ∼400similar-toabsent400\sim 400∼ 400 in Fig. 5 (h) because the mix-type FS is closer to the M-M’ line. Thus, large |m2|subscriptπ‘š2|m_{2}|| italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT | can be realized by the mix-type band with finite dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital weight even when |EvHSX⁒Zβˆ’ΞΌ|∼0.1similar-tosuperscriptsubscript𝐸vHSπ‘‹π‘πœ‡0.1|E_{\rm vHS}^{XZ}-\mu|\sim 0.1| italic_E start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_X italic_Z end_POSTSUPERSCRIPT - italic_ΞΌ | ∼ 0.1. We stress that the relation m2β‰ˆβˆ’m3subscriptπ‘š2subscriptπ‘š3m_{2}\approx-m_{3}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT β‰ˆ - italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT is well satisfied; see the SI B SM .

Refer to caption
Figure 5: (a) FS of CsV3Sb5 model in the kz=0subscriptπ‘˜π‘§0k_{z}=0italic_k start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 0 plane. The dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital band gives the pure-type FS, and the dY⁒Zsubscriptπ‘‘π‘Œπ‘d_{YZ}italic_d start_POSTSUBSCRIPT italic_Y italic_Z end_POSTSUBSCRIPT-orbital band gives the mix-type FS. The dX2βˆ’Y2subscript𝑑superscript𝑋2superscriptπ‘Œ2d_{X^{2}-Y^{2}}italic_d start_POSTSUBSCRIPT italic_X start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Y start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT+d3⁒Z2βˆ’R2subscript𝑑3superscript𝑍2superscript𝑅2d_{3Z^{2}-R^{2}}italic_d start_POSTSUBSCRIPT 3 italic_Z start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_R start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT-orbital band gives another pure-type FS. (b) FS with the p𝑝pitalic_p-orbital shift Δ⁒Ep=βˆ’0.2Ξ”subscript𝐸𝑝0.2\Delta E_{p}=-0.2roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2 eV. Here, the mix-type FS is closer to the M points. (c) Change in the FS topology with n𝑛nitalic_n in Ep=βˆ’0.2subscript𝐸𝑝0.2E_{p}=-0.2italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2 model inside the dashed line square in (b) (d) m2⁒(π’Œ)subscriptπ‘š2π’Œm_{2}({{\bm{k}}})italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ( bold_italic_k ) inside the dotted line square in (c). with the folded FSs at n=nD𝑛subscript𝑛Dn=n_{\rm D}italic_n = italic_n start_POSTSUBSCRIPT roman_D end_POSTSUBSCRIPT and n=31𝑛31n=31italic_n = 31. (e-h) Obtained mnsubscriptπ‘šπ‘›m_{n}italic_m start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT at T=0.01𝑇0.01T=0.01italic_T = 0.01 and 0.0050.0050.0050.005: (e) m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT in Δ⁒Ep=0Ξ”subscript𝐸𝑝0\Delta E_{p}=0roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = 0 model, where m1βˆΌβˆ’0.5similar-tosubscriptπ‘š10.5m_{1}\sim-0.5italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ∼ - 0.5 at n=31𝑛31n=31italic_n = 31. (f) m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT in Δ⁒Ep=βˆ’0.2Ξ”subscript𝐸𝑝0.2\Delta E_{p}=-0.2roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2 model, where m1βˆΌβˆ’2similar-tosubscriptπ‘š12m_{1}\sim-2italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ∼ - 2 at n=31𝑛31n=31italic_n = 31. (g) m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT in Δ⁒Ep=0Ξ”subscript𝐸𝑝0\Delta E_{p}=0roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = 0 model, where m2βˆΌβˆ’50similar-tosubscriptπ‘š250m_{2}\sim-50italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ∼ - 50 at n=31𝑛31n=31italic_n = 31. (h) m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT in Δ⁒Ep=βˆ’0.2Ξ”subscript𝐸𝑝0.2\Delta E_{p}=-0.2roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2 model, where m2∼300similar-tosubscriptπ‘š2300m_{2}\sim 300italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ∼ 300 at n=31𝑛31n=31italic_n = 31.

Next, we discuss the n𝑛nitalic_n-dependences of m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT in more detail. In Figs. 5 (e)-(h), both coefficients take very large values at nβ‰ˆnvHSX⁒Z𝑛superscriptsubscript𝑛vHS𝑋𝑍n\approx n_{\rm vHS}^{XZ}italic_n β‰ˆ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_X italic_Z end_POSTSUPERSCRIPT due to the hybridization between dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital bands, as we explained in Fig. 3. With increasing n𝑛nitalic_n, these coefficients exhibit drastic n𝑛nitalic_n-dependences when the FS changes its topology due to the pure-mix hybridization. When nβ‰ˆnD𝑛subscript𝑛Dn\approx n_{\rm D}italic_n β‰ˆ italic_n start_POSTSUBSCRIPT roman_D end_POSTSUBSCRIPT, the electron-like Dirac pockets made of dX⁒Z+dY⁒Zsubscript𝑑𝑋𝑍subscriptπ‘‘π‘Œπ‘d_{XZ}+d_{YZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT + italic_d start_POSTSUBSCRIPT italic_Y italic_Z end_POSTSUBSCRIPT orbitals cause large |m1|subscriptπ‘š1|m_{1}|| italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT | and |m2|subscriptπ‘š2|m_{2}|| italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT |. |m1|subscriptπ‘š1|m_{1}|| italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT | and |m2|subscriptπ‘š2|m_{2}|| italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT | are also enlarged when the mix-band (with finite dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-weight) is close to the M-M’ line at nβ‰ˆ31𝑛31n\approx 31italic_n β‰ˆ 31 in Δ⁒Ep=βˆ’0.2Ξ”subscript𝐸𝑝0.2\Delta E_{p}=-0.2roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2 model [at nβ‰ˆ31.2𝑛31.2n\approx 31.2italic_n β‰ˆ 31.2 in Δ⁒Ep=0Ξ”subscript𝐸𝑝0\Delta E_{p}=0roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = 0 model].

To summarize, both m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT exhibit interesting sensitivity to the multiorbital character and the nesting condition of the bandstructure around M points. Unexpectedly, both dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital (pure-type) band and dY⁒Zsubscriptπ‘‘π‘Œπ‘d_{YZ}italic_d start_POSTSUBSCRIPT italic_Y italic_Z end_POSTSUBSCRIPT-orbital (mix-type) band play important role for m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT even when the current order emerges only on the dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital. When Δ⁒Ep=βˆ’0.2Ξ”subscript𝐸𝑝0.2\Delta E_{p}=-0.2roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2, the mix-type FS approaches the M point, consistently with the ARPES bandstructure ARPES-VHS . Then, we obtain large |m1|(∼2)annotatedsubscriptπ‘š1similar-toabsent2|m_{1}|(\sim 2)| italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT | ( ∼ 2 ) and |m2|(≲400)annotatedsubscriptπ‘š2less-than-or-similar-toabsent400|m_{2}|(\lesssim 400)| italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT | ( ≲ 400 ) at n=31𝑛31n=31italic_n = 31 as shown in Figs. 5 (f) and (h). Thus, the hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-induced change in Fig. 4 can be obtained by changing the parameters slightly. (β€œNearly commensurate” current order may appear in Δ⁒Ep=0Ξ”subscript𝐸𝑝0\Delta E_{p}=0roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = 0 model, in which the nesting of the FS at 𝒒msubscriptπ’’π‘š{{\bm{q}}}_{m}bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT is not perfect Tazai-kagome . The calculated mnsubscriptπ‘šπ‘›m_{n}italic_m start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT will be larger by using the correct incommensurate nesting vector of Δ⁒Ep=0Ξ”subscript𝐸𝑝0\Delta E_{p}=0roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = 0 model. ) In the SI H SM , we calculate the GL coefficients when the current and bond orders emerge on other d𝑑ditalic_d-orbitals. It is found that large m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT are obtained in various cases. Thus, the present study is valid for various types of current order mechanisms, not restricted to Ref. Tazai-kagome2 .

Summary and Discussions

In summary, the chiral current order in kagome metal exhibits weak ferromagnetism, and its magnitude are enlarged in the bond order state. Importantly, we derived the hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-induced GL free energy expression Δ⁒F¯Δ¯𝐹\Delta{\bar{F}}roman_Ξ” overΒ― start_ARG italic_F end_ARG in Eq. [9], which provides an important basis for future research in kagome metals. The emergent Ξ·β‹…Ο•β‹…hzβ‹…πœ‚italic-Ο•subscriptβ„Žπ‘§\eta\cdot\phi\cdot h_{z}italic_Ξ· β‹… italic_Ο• β‹… italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT trilinear term in Δ⁒F¯Δ¯𝐹\Delta{\bar{F}}roman_Ξ” overΒ― start_ARG italic_F end_ARG explains prominent field-induced chiral symmetry breaking in kagome metals. We revealed that mnsubscriptπ‘šπ‘›m_{n}italic_m start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT becomes large when the vHS energy is close to the Fermi level because the prominent FS reconstruction occurs due to the current order parameters. Furthermore, the multi-orbital mixing in the bandstructure of real kagome metals makes mnsubscriptπ‘šπ‘›m_{n}italic_m start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT larger for a wide filling range. The finding that mnsubscriptπ‘šπ‘›m_{n}italic_m start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT sensitively depends on the multiorbital bandstructure in Fig. 5 provides a useful hint to control the charge current or to understand the difference between Cs-based and K-based compounds Moll-K . Interestingly, we obtain large mnsubscriptπ‘šπ‘›m_{n}italic_m start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT when the current order parameter emerges in not only dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital, but also other d𝑑ditalic_d-orbitals. It is interesting to study mnsubscriptπ‘šπ‘›m_{n}italic_m start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT for theoretically proposed exotic TRS breaking states, such as the inter-orbital exciton order Nat-g-ology and multipolar and toroidal magnetic orders Fernandes-GL .

Below, we discuss several important issues.

.1 Comparison with experiments: hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT–induced current order

In Ref. Moll-hz , it was proposed that CsV3Sb5 is located at the quantum critical point of the current order (TΒ―c0β‰ˆ0superscriptsubscript¯𝑇c00{\bar{T}}_{\rm c}^{0}\approx 0overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT β‰ˆ 0) in the absence of the uniaxial strain. (As we discussed above, Tc0superscriptsubscript𝑇c0T_{\rm c}^{0}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT is renormalized to TΒ―c0superscriptsubscript¯𝑇c0{\bar{T}}_{\rm c}^{0}overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT under the BO phase.) The field-induced (hz∼9similar-tosubscriptβ„Žπ‘§9h_{z}\sim 9italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT ∼ 9T) current order at T∼20similar-to𝑇20T\sim 20italic_T ∼ 20K is naturally understood based on the GL free energy analysis with the current-bond-hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT trillinear term. Figures 4 (b) and (d) corresponds to the experimental report in Ref. Moll-hz by considering that TΒ―c0β‰ˆ0superscriptsubscript¯𝑇c00{\bar{T}}_{\rm c}^{0}\approx 0overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT β‰ˆ 0. In addition, the present GL theroy explains the field-induced enhancement of the local magnetic flux (∝η1proportional-toabsentsuperscriptπœ‚1\propto\eta^{1}∝ italic_Ξ· start_POSTSUPERSCRIPT 1 end_POSTSUPERSCRIPT) observed by ΞΌπœ‡\muitalic_ΞΌSR measurements in AV3Sb5 muSR4-Cs ; muSR2-K ; muSR5-Rb and field-tuned chiral transport study eMChA . Interestingly, the obtained mnsubscriptπ‘šπ‘›m_{n}italic_m start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT sensitively depends on the multiorbital bandstructure and the filling in Fig. 5 (and Fig. S9). The present discovery provides a useful hint to control the charge current in kagome metals. The present finding will also be essential to understand the significant difference between Cs-based and K-based compounds reported in Ref. Moll-K . In addition, the discovery of CDW state in double-layer kagome metal ScV6Sn6 166-CDW has attracted increasing theoretical interest Thomale-GL ; Bernevig-166-exp ; Bernevig-166 . The TRS breaking state and its increment under hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT discovered in ScV6Sn6 166-mSR may be understood by developing the present GL theory.

We note that the trilinear term in Δ⁒F¯Δ¯𝐹\Delta{\bar{F}}roman_Ξ” overΒ― start_ARG italic_F end_ARG is also caused by the spin magnetization in the presence of spin-orbit interaction Fernandes-GL . Future quantitative calculations would be important.

.2 Comparison with experiments: Strain-induced current order

Reference Moll-hz also reports interesting prominent strain induced increment of TΒ―c0superscriptsubscript¯𝑇c0{\bar{T}}_{\rm c}^{0}overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT. Under the uniaxial strain, the degeneracy of the current order transition temperature at 𝒒=𝒒m𝒒subscriptπ’’π‘š{{\bm{q}}}={{\bm{q}}}_{m}bold_italic_q = bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT (m=1∼3π‘š1similar-to3m=1\sim 3italic_m = 1 ∼ 3), TΒ―c(m)superscriptsubscript¯𝑇cπ‘š{\bar{T}}_{\rm c}^{(m)}overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT, is lifted by the change in the 2nd order GL term as discussed in Ref. Moll-hz . Then, TΒ―c=maxm⁑TΒ―c(m)subscript¯𝑇csubscriptπ‘šsuperscriptsubscript¯𝑇cπ‘š{\bar{T}}_{\rm c}=\max_{m}{\bar{T}}_{\rm c}^{(m)}overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT = roman_max start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT will be larger than the original TΒ―c0superscriptsubscript¯𝑇c0{\bar{T}}_{\rm c}^{0}overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT. In the SI I SM , we find that additional significant contribution to the increment of TΒ―c0superscriptsubscript¯𝑇c0{\bar{T}}_{\rm c}^{0}overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT originates from the strain induced change in the 4th order GL terms (disubscript𝑑𝑖d_{i}italic_d start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT). Here, the β€œBO-induced suppression of the current order by d5subscript𝑑5d_{5}italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT and d6subscript𝑑6d_{6}italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT GL terms is found to be drastically reduced by the strain. In fact, we find that disubscript𝑑𝑖d_{i}italic_d start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT is sensitive to the strain because the vHS energy is close to the Fermi level; see Fig. S10 in the SI I SM .

Below, we explain the discussion in the SI I SM . Considering the D6⁒hsubscript𝐷6β„ŽD_{6h}italic_D start_POSTSUBSCRIPT 6 italic_h end_POSTSUBSCRIPT symmetry, we assume that the E1⁒gsubscript𝐸1𝑔E_{1g}italic_E start_POSTSUBSCRIPT 1 italic_g end_POSTSUBSCRIPT symmetry strain (Ο΅,Ο΅β€²)italic-Ο΅superscriptitalic-Ο΅β€²({\epsilon},{\epsilon}^{\prime})( italic_Ο΅ , italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ) induces the shift of the vHS energy levels as Δ⁒𝑬≑(Δ⁒EA,Δ⁒EB,Δ⁒EC)=α⁒(Ο΅,0,βˆ’Ο΅)Δ𝑬Δsubscript𝐸𝐴Δsubscript𝐸𝐡Δsubscript𝐸𝐢𝛼italic-Ο΅0italic-Ο΅\Delta{\bm{E}}\equiv(\Delta E_{A},\Delta E_{B},\Delta E_{C})={\alpha}({% \epsilon},0,-{\epsilon})roman_Ξ” bold_italic_E ≑ ( roman_Ξ” italic_E start_POSTSUBSCRIPT italic_A end_POSTSUBSCRIPT , roman_Ξ” italic_E start_POSTSUBSCRIPT italic_B end_POSTSUBSCRIPT , roman_Ξ” italic_E start_POSTSUBSCRIPT italic_C end_POSTSUBSCRIPT ) = italic_Ξ± ( italic_Ο΅ , 0 , - italic_Ο΅ ). (Δ⁒𝑬′Δsuperscript𝑬′\Delta{\bm{E}}^{\prime}roman_Ξ” bold_italic_E start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT by Ο΅β€²superscriptitalic-Ο΅β€²{\epsilon}^{\prime}italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT is presented in the SI I SM .) Hereafter, we set Ξ±=1𝛼1{\alpha}=1italic_Ξ± = 1. Under the 3⁒Q3𝑄3Q3 italic_Q BO phase Ο•0=(Ο•0,Ο•0,Ο•0)/3superscriptbold-italic-Ο•0superscriptitalic-Ο•0superscriptitalic-Ο•0superscriptitalic-Ο•03{\bm{\phi}}^{0}=(\phi^{0},\phi^{0},\phi^{0})/\sqrt{3}bold_italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = ( italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT , italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT , italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ) / square-root start_ARG 3 end_ARG, the 2nd order GL coefficient acsubscriptπ‘Žca_{\rm c}italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT in Eq. [11] in the absence of the strain (Ο΅=0italic-Ο΅0{\epsilon}=0italic_Ο΅ = 0) is changed to

aΒ―c⁒(0)subscriptΒ―π‘Žc0\displaystyle{\bar{a}}_{\rm c}(0)overΒ― start_ARG italic_a end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( 0 ) =\displaystyle== ac+2⁒(d5+d6)⁒((Ο•0)2/3)subscriptπ‘Žc2superscript𝑑5superscript𝑑6superscriptsuperscriptitalic-Ο•023\displaystyle a_{\rm c}+2(d^{5}+d^{6})((\phi^{0})^{2}/3)italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT + 2 ( italic_d start_POSTSUPERSCRIPT 5 end_POSTSUPERSCRIPT + italic_d start_POSTSUPERSCRIPT 6 end_POSTSUPERSCRIPT ) ( ( italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ) start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT / 3 ) (14)
∝proportional-to\displaystyle\propto∝ Tβˆ’TΒ―c0,𝑇superscriptsubscript¯𝑇c0\displaystyle T-{\bar{T}}_{\rm c}^{0},italic_T - overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ,

where |Ο•0|≳Tbgreater-than-or-equivalent-tosuperscriptitalic-Ο•0subscript𝑇b|\phi^{0}|\gtrsim T_{\rm b}| italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT | ≳ italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT when Tβ‰ͺTbmuch-less-than𝑇subscript𝑇bT\ll T_{\rm b}italic_T β‰ͺ italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT. Here, we assume that Ο•0superscriptitalic-Ο•0\phi^{0}italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT is T𝑇Titalic_T-independent because we study the strain-induced current order for Tβ‰ͺTb0much-less-than𝑇superscriptsubscript𝑇b0T\ll T_{\rm b}^{0}italic_T β‰ͺ italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT. For finite Ο΅italic-Ο΅{\epsilon}italic_Ο΅, it is changed by Δ⁒F⁒(Ο΅)Δ𝐹italic-Ο΅\Delta F({\epsilon})roman_Ξ” italic_F ( italic_Ο΅ ) as

aΒ―c⁒(Ο΅)subscriptΒ―π‘Žcitalic-Ο΅\displaystyle{\bar{a}}_{\rm c}({\epsilon})overΒ― start_ARG italic_a end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( italic_Ο΅ ) =\displaystyle== aΒ―c0βˆ’[(2⁒g5+g6)+D⁒(2⁒g1+g2)]⁒((Ο•0)2/6)superscriptsubscriptΒ―π‘Žc0delimited-[]2subscript𝑔5subscript𝑔6𝐷2subscript𝑔1subscript𝑔2superscriptsuperscriptitalic-Ο•026\displaystyle{\bar{a}}_{\rm c}^{0}-[(2g_{5}+g_{6})+D(2g_{1}+g_{2})]((\phi^{0})% ^{2}/6)overΒ― start_ARG italic_a end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT - [ ( 2 italic_g start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT + italic_g start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ) + italic_D ( 2 italic_g start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT + italic_g start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) ] ( ( italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ) start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT / 6 ) (15)
∝proportional-to\displaystyle\propto∝ Tβˆ’TΒ―c⁒(Ο΅),𝑇subscript¯𝑇citalic-Ο΅\displaystyle T-{\bar{T}}_{\rm c}({\epsilon}),italic_T - overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( italic_Ο΅ ) ,

where glsubscript𝑔𝑙g_{l}italic_g start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT are Ο΅italic-Ο΅{\epsilon}italic_Ο΅-linear. Here, TΒ―c⁒(0)=TΒ―c0subscript¯𝑇c0superscriptsubscript¯𝑇c0{\bar{T}}_{\rm c}(0)={\bar{T}}_{\rm c}^{0}overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( 0 ) = overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT. glsubscript𝑔𝑙g_{l}italic_g start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT and dl(m)superscriptsubscriptπ‘‘π‘™π‘šd_{l}^{(m)}italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT are related as Ο΅β‹…(βˆ‚Ο΅dl(1),βˆ‚Ο΅dl(2),βˆ‚Ο΅dl(3))=(βˆ’1)lβ‹…gl⁒(1,βˆ’1,0)β‹…italic-Ο΅subscriptitalic-Ο΅superscriptsubscript𝑑𝑙1subscriptitalic-Ο΅superscriptsubscript𝑑𝑙2subscriptitalic-Ο΅superscriptsubscript𝑑𝑙3β‹…superscript1𝑙subscript𝑔𝑙110{\epsilon}\cdot({\partial}_{{\epsilon}}d_{l}^{(1)},{\partial}_{{\epsilon}}d_{l% }^{(2)},{\partial}_{{\epsilon}}d_{l}^{(3)})=(-1)^{l}\cdot g_{l}(1,-1,0)italic_Ο΅ β‹… ( βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 1 ) end_POSTSUPERSCRIPT , βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 2 ) end_POSTSUPERSCRIPT , βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 3 ) end_POSTSUPERSCRIPT ) = ( - 1 ) start_POSTSUPERSCRIPT italic_l end_POSTSUPERSCRIPT β‹… italic_g start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT ( 1 , - 1 , 0 ). The the current order appears in the BO phase when Eq. [15] becomes negative. Therefore, TΒ―c⁒(Ο΅)subscript¯𝑇citalic-Ο΅{\bar{T}}_{\rm c}({\epsilon})overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( italic_Ο΅ ) will increase in proportion to Ο΅italic-Ο΅{\epsilon}italic_Ο΅ if (2⁒g5+g6)>02subscript𝑔5subscript𝑔60(2g_{5}+g_{6})>0( 2 italic_g start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT + italic_g start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ) > 0. Therefore, we obtain TΒ―c⁒(Ο΅)∼(Ο΅/Ο΅0)⁒(Tc0βˆ’TΒ―c0)similar-tosubscript¯𝑇citalic-Ο΅italic-Ο΅subscriptitalic-Ο΅0superscriptsubscript𝑇c0superscriptsubscript¯𝑇c0{\bar{T}}_{\rm c}({\epsilon})\sim({\epsilon}/{\epsilon}_{0})(T_{\rm c}^{0}-{% \bar{T}}_{\rm c}^{0})overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( italic_Ο΅ ) ∼ ( italic_Ο΅ / italic_Ο΅ start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT ) ( italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT - overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ) with Ο΅0=0.025subscriptitalic-Ο΅00.025{\epsilon}_{0}=0.025italic_Ο΅ start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT = 0.025 (0.0290.0290.0290.029) for n=2.56𝑛2.56n=2.56italic_n = 2.56 (2.482.482.482.48), according to the numerical study in Fig. S10. This increment of TΒ―c⁒(Ο΅)subscript¯𝑇citalic-Ο΅{\bar{T}}_{\rm c}({\epsilon})overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( italic_Ο΅ ) originates from the reduction of the BO-induced suppression of the current order. Therefore, considering that Tc0∼0.01similar-tosuperscriptsubscript𝑇c00.01T_{\rm c}^{0}\sim 0.01italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ∼ 0.01 and TΒ―c0=0superscriptsubscript¯𝑇c00{\bar{T}}_{\rm c}^{0}=0overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = 0 (= at current order QCP), we obtain TΒ―c⁒(Ο΅)∼0.002similar-tosubscript¯𝑇citalic-Ο΅0.002{\bar{T}}_{\rm c}({\epsilon})\sim 0.002overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( italic_Ο΅ ) ∼ 0.002 (β‰ˆ20absent20\approx 20β‰ˆ 20 K) at ϡ∼0.005similar-toitalic-Ο΅0.005{\epsilon}\sim 0.005italic_Ο΅ ∼ 0.005 (β‰ˆ50⁒K)absent50K(\approx 50\ {\rm K})( β‰ˆ 50 roman_K ). Based on this extrinsic strain scenario, one may understand the difference of Tcsubscript𝑇cT_{\rm c}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT by experiments, such as the absence of the TRS breaking at hz=0subscriptβ„Žπ‘§0h_{z}=0italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 0 reported by Kerr rotation study Kapitulnik .

.3 Nematic 3D stacking 3⁒Q3𝑄3Q3 italic_Q BO and current order

As revealed in Ref. Balents2021 , the nematic state can be realized by the Ο€πœ‹\piitalic_Ο€-shift 3D stacking of the 3⁒Q3𝑄3Q3 italic_Q BO thanks to the 3rd order GL term b1subscript𝑏1b_{1}italic_b start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT. This state is the combination of three BO states at wavevectors 𝒒l3⁒D=(𝒒l2⁒D,qlz)superscriptsubscript𝒒𝑙3Dsuperscriptsubscript𝒒𝑙2Dsuperscriptsubscriptπ‘žπ‘™π‘§{{\bm{q}}}_{l}^{\rm 3D}=({{\bm{q}}}_{l}^{\rm 2D},q_{l}^{z})bold_italic_q start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 3 roman_D end_POSTSUPERSCRIPT = ( bold_italic_q start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 roman_D end_POSTSUPERSCRIPT , italic_q start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_z end_POSTSUPERSCRIPT ) for l=1,2,3𝑙123l=1,2,3italic_l = 1 , 2 , 3 (𝒒l2⁒Dsuperscriptsubscript𝒒𝑙2D{\bm{{{\bm{q}}}}}_{l}^{\rm 2D}bold_italic_q start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 roman_D end_POSTSUPERSCRIPT is shown in Fig. 1 (a)) when {q1z,q2z,q3z}={Ο€,Ο€,0}superscriptsubscriptπ‘ž1𝑧superscriptsubscriptπ‘ž2𝑧superscriptsubscriptπ‘ž3π‘§πœ‹πœ‹0\{q_{1}^{z},q_{2}^{z},q_{3}^{z}\}=\{\pi,\pi,0\}{ italic_q start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_z end_POSTSUPERSCRIPT , italic_q start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_z end_POSTSUPERSCRIPT , italic_q start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_z end_POSTSUPERSCRIPT } = { italic_Ο€ , italic_Ο€ , 0 }, {Ο€,0,Ο€}πœ‹0πœ‹\{\pi,0,\pi\}{ italic_Ο€ , 0 , italic_Ο€ } or {0,Ο€,Ο€}0πœ‹πœ‹\{0,\pi,\pi\}{ 0 , italic_Ο€ , italic_Ο€ }. Because of the relation βˆ‘l1,2,3𝒒l3⁒D=𝟎superscriptsubscript𝑙123superscriptsubscript𝒒𝑙3D0\sum_{l}^{1,2,3}{{\bm{q}}}_{l}^{\rm 3D}={\bm{0}}βˆ‘ start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 1 , 2 , 3 end_POSTSUPERSCRIPT bold_italic_q start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 3 roman_D end_POSTSUPERSCRIPT = bold_0, the Ο€πœ‹\piitalic_Ο€-shift stacking gains the 3rd order GL free energy due to the b1subscript𝑏1b_{1}italic_b start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT term. Its necessary condition is that the 2nd order GL term absubscriptπ‘Žba_{\rm b}italic_a start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT is almost qzsubscriptπ‘žπ‘§q_{z}italic_q start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-independent. Consistently, we recently find that the DW equation eigenvalue for the BO, λϕ⁒(𝒒l2⁒D,qz)subscriptπœ†italic-Ο•subscriptsuperscript𝒒2D𝑙superscriptπ‘žπ‘§\lambda_{\phi}({{\bm{q}}}^{\rm 2D}_{l},q^{z})italic_Ξ» start_POSTSUBSCRIPT italic_Ο• end_POSTSUBSCRIPT ( bold_italic_q start_POSTSUPERSCRIPT 2 roman_D end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT , italic_q start_POSTSUPERSCRIPT italic_z end_POSTSUPERSCRIPT ), is almost qzsubscriptπ‘žπ‘§q_{z}italic_q start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-independent by reflecting almost 2D character of 3D 30-orbital CsV3Sb5 model Onari-future . (Note that ab∝(βˆ’1+Ξ»Ο•βˆ’1)proportional-tosubscriptπ‘Žb1superscriptsubscriptπœ†italic-Ο•1a_{\rm b}\propto(-1+\lambda_{\phi}^{-1})italic_a start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT ∝ ( - 1 + italic_Ξ» start_POSTSUBSCRIPT italic_Ο• end_POSTSUBSCRIPT start_POSTSUPERSCRIPT - 1 end_POSTSUPERSCRIPT ).) In contrast, the qzsubscriptπ‘žπ‘§q_{z}italic_q start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-dependence of b1subscript𝑏1b_{1}italic_b start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT is rather prominent Onari-future . Thus, we expect that the namatic 3Q BO state discussed in Ref. Balents2021 is realistic.

Next, we discuss the 3D structure of the field-induced 3⁒Q3𝑄3Q3 italic_Q current order. In the absence of the BO, by the same argument as above, the 3⁒Q3𝑄3Q3 italic_Q current order will form the Ο€πœ‹\piitalic_Ο€-shift stacking to gain the m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT-term contribution in Δ⁒F¯Δ¯𝐹\Delta{\bar{F}}roman_Ξ” overΒ― start_ARG italic_F end_ARG. When the 3⁒Q3𝑄3Q3 italic_Q current order appears inside the 3⁒Q3𝑄3Q3 italic_Q BO state, the 3D stacking of the current order would be mainly determined by the the 3rd order GL term b2subscript𝑏2b_{2}italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT that describes the bond-current coupling energy.

.4 Derivation of E0subscript𝐸0E_{0}italic_E start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT

The expressions of orbital magnetization derived by Refs. Morb-paper1 ; Morb-paper2 is given as

Morb=e⁒a22⁒c⁒ℏ⁒Nuc⁒Nβ’βˆ‘π’Œβ’Οƒm⁒(π’Œ),subscript𝑀orb𝑒superscriptπ‘Ž22𝑐Planck-constant-over-2-pisubscript𝑁uc𝑁subscriptπ’ŒπœŽπ‘šπ’Œ\displaystyle\!\!\!\!\!\!\!\!M_{\rm orb}=\frac{ea^{2}}{2c\hbar N_{\rm uc}N}% \sum_{{{\bm{k}}}\sigma}m({{\bm{k}}}),italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT = divide start_ARG italic_e italic_a start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_ARG start_ARG 2 italic_c roman_ℏ italic_N start_POSTSUBSCRIPT roman_uc end_POSTSUBSCRIPT italic_N end_ARG βˆ‘ start_POSTSUBSCRIPT bold_italic_k italic_Οƒ end_POSTSUBSCRIPT italic_m ( bold_italic_k ) ,

where m⁒(π’Œ)π‘šπ’Œm({{\bm{k}}})italic_m ( bold_italic_k ) is given in Eq. [5]. aπ‘Žaitalic_a is the unit length in this numerical calculation. Here, we set a=|𝒂AB|π‘Žsubscript𝒂ABa=|\bm{a}_{\rm AB}|italic_a = | bold_italic_a start_POSTSUBSCRIPT roman_AB end_POSTSUBSCRIPT |. Nucsubscript𝑁ucN_{\rm uc}italic_N start_POSTSUBSCRIPT roman_uc end_POSTSUBSCRIPT is the site number of the unit cell, N𝑁Nitalic_N is the π’Œπ’Œ{{\bm{k}}}bold_italic_k-mesh number. By using the ΞΌB=e⁒ℏ2⁒me⁒csubscriptπœ‡π΅π‘’Planck-constant-over-2-pi2subscriptπ‘šπ‘’π‘\mu_{B}=\frac{e\hbar}{2m_{e}c}italic_ΞΌ start_POSTSUBSCRIPT italic_B end_POSTSUBSCRIPT = divide start_ARG italic_e roman_ℏ end_ARG start_ARG 2 italic_m start_POSTSUBSCRIPT italic_e end_POSTSUBSCRIPT italic_c end_ARG and E0≑ℏ2me⁒a2subscript𝐸0superscriptPlanck-constant-over-2-pi2subscriptπ‘šπ‘’superscriptπ‘Ž2E_{0}\equiv\frac{\hbar^{2}}{m_{e}a^{2}}italic_E start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT ≑ divide start_ARG roman_ℏ start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_ARG start_ARG italic_m start_POSTSUBSCRIPT italic_e end_POSTSUBSCRIPT italic_a start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_ARG, we obtain Eq. [3]. By using the me⁒c2=0.511Γ—106subscriptπ‘šπ‘’superscript𝑐20.511superscript106m_{e}c^{2}=0.511\times 10^{6}italic_m start_POSTSUBSCRIPT italic_e end_POSTSUBSCRIPT italic_c start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT = 0.511 Γ— 10 start_POSTSUPERSCRIPT 6 end_POSTSUPERSCRIPT [eV] and ℏme⁒c=3.86Γ—10βˆ’13Planck-constant-over-2-pisubscriptπ‘šπ‘’π‘3.86superscript1013\frac{\hbar}{m_{e}c}=3.86\times 10^{-13}divide start_ARG roman_ℏ end_ARG start_ARG italic_m start_POSTSUBSCRIPT italic_e end_POSTSUBSCRIPT italic_c end_ARG = 3.86 Γ— 10 start_POSTSUPERSCRIPT - 13 end_POSTSUPERSCRIPT [m], we obtain E0=0.5subscript𝐸00.5E_{0}=0.5italic_E start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT = 0.5 [eV] for a=0.4π‘Ž0.4a=0.4italic_a = 0.4 [nm]. In Kagome metals (a=0.275π‘Ž0.275a=0.275italic_a = 0.275 [nm]), E0=1.0subscript𝐸01.0E_{0}=1.0italic_E start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT = 1.0 [eV]. In the numerical calculation, large number of N𝑁Nitalic_N is required to obtain a reliable result at low temperatures. Here, we set N=(2400)2𝑁superscript24002N=(2400)^{2}italic_N = ( 2400 ) start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT at T=1𝑇1T=1italic_T = 1 [meV].

.5 Acknowledgments

We are grateful to Y. Matsuda, T. Shibauchi, K. Hashimoto, T. Asaba, S. Onari, A. Ogawa and K. Shimura for fruitful discussions. This study has been supported by Grants-in-Aid for Scientific Research from MEXT of Japan (JP20K03858, JP20K22328, JP22K14003), and by the Quantum Liquid Crystal No. JP19H05825 KAKENHI on Innovative Areas from JSPS of Japan.

References

  • (1) B. R. Ortiz, L. C. Gomes, J. R. Morey, M. Winiarski, M. Bordelon, J. S. Mangum, I. W. H. Oswald, J. A. Rodriguez-Rivera, J. R. Neilson, S. D. Wilson, E. Ertekin, T. M. McQueen, and E. S. Toberer, New kagome prototype materials: discovery of KV3⁒Sb5,RbV3⁒Sb5subscriptKV3subscriptSb5subscriptRbV3subscriptSb5{\mathrm{KV}}_{3}{\mathrm{Sb}}_{5},{\mathrm{RbV}}_{3}{\mathrm{Sb}}_{5}roman_KV start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT roman_Sb start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT , roman_RbV start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT roman_Sb start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT, and CsV3⁒Sb5subscriptCsV3subscriptSb5{\mathrm{CsV}}_{3}{\mathrm{Sb}}_{5}roman_CsV start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT roman_Sb start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT, Phys. Rev. Materials 3, 094407 (2019).
  • (2) B. R. Ortiz, S. M. L. Teicher, Y. Hu, J. L. Zuo, P. M. Sarte, E. C. Schueller, A. M. M. Abeykoon, M. J. Krogstad, S. Rosenkranz, R. Osborn, R. Seshadri, L. Balents, J. He, and S. D. Wilson, CsV3⁒Sb5subscriptCsV3subscriptSb5\mathrm{Cs}{\mathrm{V}}_{3}{\mathrm{Sb}}_{5}roman_CsV start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT roman_Sb start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT: A β„€2subscriptβ„€2{\mathbb{Z}}_{2}blackboard_Z start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT Topological Kagome Metal with a Superconducting Ground State, Phys. Rev. Lett. 125, 247002 (2020).
  • (3) Y.-X. Jiang, J.-X. Yin, M. M. Denner, N. Shumiya, B. R. Ortiz, G. Xu, Z. Guguchia, J. He, M. S. Hossain, X. Liu, J. Ruff, L. Kautzsch, S. S. Zhang, G. Chang, I. Belopolski, Q. Zhang, T. A. Cochran, D. Multer, M. Litskevich, Z.-J. Cheng, X. P. Yang, Z. Wang, R. Thomale, T. Neupert, S. D. Wilson, and M. Z. Hasan, Unconventional chiral charge order in kagome superconductor KV3Sb5, Nat. Mater. 20, 1353 (2021).
  • (4) H. Li, H. Zhao, B. R. Ortiz, T. Park, M. Ye, L. Balents, Z. Wang, S. D. Wilson, and I. Zeljkovic, Rotation symmetry breaking in the normal state of a kagome superconductor KV3Sb5, Nat. Phys. 18, 265 (2022).
  • (5) M. L. Kiesel, C. Platt, and R. Thomale, Unconventional Fermi Surface Instabilities in the Kagome Hubbard Model, Phys. Rev. Lett. 110, 126405 (2013).
  • (6) W.-S. Wang, Z.-Z. Li, Y.-Y. Xiang, and Q.-H. Wang, Competing electronic orders on kagome lattices at van Hove filling, Phys. Rev. B 87, 115135 (2013).
  • (7) X. Wu, T. Schwemmer, T. MΓΌller, A. Consiglio, G. Sangiovanni, D. Di Sante, Y. Iqbal, W. Hanke, A. P. Schnyder, M. M. Denner, M. H. Fischer, T. Neupert, and R. Thomale, Nature of Unconventional Pairing in the Kagome Superconductors A⁒V3⁒Sb5𝐴subscriptV3subscriptSb5A{\mathrm{V}}_{3}{\mathrm{Sb}}_{5}italic_A roman_V start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT roman_Sb start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT (A=K,Rb,Cs𝐴KRbCsA=\mathrm{K},\mathrm{Rb},\mathrm{Cs}italic_A = roman_K , roman_Rb , roman_Cs), Phys. Rev. Lett. 127, 177001 (2021).
  • (8) M. M. Denner, R. Thomale, and T. Neupert, Analysis of Charge Order in the Kagome Metal A⁒V3⁒Sb5𝐴subscriptV3subscriptSb5A{\mathrm{V}}_{3}{\mathrm{Sb}}_{5}italic_A roman_V start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT roman_Sb start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT (A=K,Rb,Cs𝐴KRbCsA=\mathrm{K},\mathrm{Rb},\mathrm{Cs}italic_A = roman_K , roman_Rb , roman_Cs), Phys. Rev. Lett. 127, 217601 (2021).
  • (9) T. Park, M. Ye, and L. Balents, Electronic instabilities of kagome metals: Saddle points and Landau theory, Phys. Rev. B 104, 035142 (2021).
  • (10) Y.-P. Lin and R. M. Nandkishore, Complex charge density waves at Van Hove singularity on hexagonal lattices: Haldane-model phase diagram and potential realization in the kagome metals AV3Sb5 (A = K, Rb, Cs), Phys. Rev. B 104, 045122 (2021).
  • (11) R. Tazai, Y. Yamakawa, S. Onari, and H. Kontani, Mechanism of exotic density-wave and beyond-Migdal unconventional superconductivity in kagome metal AV3Sb5 (A = K, Rb, Cs), Sci. Adv. 8, eabl4108 (2022).
  • (12) M. Roppongi, K. Ishihara, Y. Tanaka, K. Ogawa, K. Okada, S. Liu, K. Mukasa, Y. Mizukami, Y. Uwatoko, R. Grasset, M. Konczykowski, B. R. Ortiz, S. D. Wilson, K. Hashimoto, and T. Shibauchi, Bulk evidence of anisotropic s𝑠sitalic_s-wave pairing with no sign change in the kagome superconductor CsV3Sb5, Nat. Commun. 14, 667 (2023).
  • (13) W. Zhang, X. Liu, L. Wang, C. W. Tsang, Z. Wang, S. T. Lam, W. Wang, J. Xie, X. Zhou, Y. Zhao, S. Wang, J. Tallon, K. T. Lai, and S. K. Goh, Nodeless superconductivity in kagome metal CsV3Sb5 with and without time reversal symmetry breaking, Nano Lett., 23, 872 (2023).
  • (14) L. Yu, C. Wang, Y. Zhang, M. Sander, S. Ni, Z. Lu, S. Ma, Z. Wang, Z. Zhao, H. Chen, K. Jiang, Y. Zhang, H. Yang, F. Zhou, X. Dong, S. L. Johnson, M. J. Graf, J. Hu, H.-J. Gao, and Z. Zhao, Evidence of a hidden flux phase in the topological kagome metal CsV3Sb5, arXiv:2107.10714 (avalable at https://arxiv.org/abs/2107.10714).
  • (15) C. Mielke, D. Das, J.-X. Yin, H. Liu, R. Gupta, Y.-X. Jiang, M. Medarde, X. Wu, H. C. Lei, J. Chang, P. Dai, Q. Si, H. Miao, R. Thomale, T. Neupert, Y. Shi, R. Khasanov, M. Z. Hasan, H. Luetkens, and Z. Guguchia, Time-reversal symmetry-breaking charge order in a kagome superconductor, Nature 602, 245 (2022).
  • (16) R. Khasanov, D. Das, R. Gupta, C. Mielke, M. Elender, Q. Yin, Z. Tu, C. Gong, H. Lei, E. T. Ritz, R. M. Fernandes, T. Birol, Z. Guguchia, and H. Luetkens, Time-reversal symmetry broken by charge order in CsV3⁒Sb5subscriptCsV3subscriptSb5{\mathrm{CsV}}_{3}{\mathrm{Sb}}_{5}roman_CsV start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT roman_Sb start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT, Phys. Rev. Research 4, 023244 (2022).
  • (17) Z. Guguchia, C. Mielke, D. Das, R. Gupta, J.-X. Yin, H. Liu, Q. Yin, M. H. Christensen, Z. Tu, C. Gong, N. Shumiya, M. S. Hossain, T. Gamsakhurdashvili, M. Elender, P. Dai, A. Amato, Y. Shi, H. C. Lei, R. M. Fernandes, M. Z. Hasan, H. Luetkens, and R. Khasanov, Tunable unconventional kagome superconductivity in charge ordered RbV3Sb5 and KV3Sb5, Nat. Commun. 14, 153 (2023).
  • (18) Y. Xu, Z. Ni, Y. Liu, B. R. Ortiz, Q. Deng, S. D. Wilson, B. Yan, L. Balents, and L. Wu, Three-state nematicity and magneto-optical Kerr effect in the charge density waves in kagome superconductors, Nat. Phys. 18, 1470 (2022).
  • (19) Y. Hu, S. Yamane, G. Mattoni, K. Yada, K. Obata, Y. Li, Y. Yao, Z. Wang, J. Wang, C. Farhang, J. Xia, Y. Maeno, and S. Yonezawa, Time-reversal symmetry breaking in charge density wave of CsV3Sb5 detected by polar Kerr effect, arXiv:2208.08036 (avalable at https://arxiv.org/abs/2208.08036).
  • (20) C. Guo, C. Putzke, S. Konyzheva, X. Huang, M. Gutierrez-Amigo, I. Errea, D. Chen, M. G. Vergniory, C. Felser, M. H. Fischer, T. Neupert, and P. J. W. Moll, Switchable chiral transport in charge-ordered kagome metal CsV3Sb5, Nature 611, 461 (2022).
  • (21) T. Asaba et al., unpublished.
  • (22) D. R. Saykin, C. Farhang, E. D. Kountz, D. Chen, B. R. Ortiz, C. Shekhar, C. Felser, S. D. Wilson, R. Thomale, J. Xia, and A. Kapitulnik, High Resolution Polar Kerr Effect Studies of CsV3Sb5: Tests for Time Reversal Symmetry Breaking Below the Charge Order Transition, Phys. Rev. Lett. 131, 016901 (2023).
  • (23) F. D. M. Haldane, Model for a Quantum Hall Effect without Landau Levels: Condensed-Matter Realization of the ”Parity Anomaly”, Phys. Rev. Lett. 61, 2015 (1988).
  • (24) S.-Y. Yang, Y. Wang, B. R. Ortiz, D. Liu, J. Gayles, E. Derunova, R. Gonzalez-Hernandez, L. Sˇˇ𝑆\check{S}overroman_Λ‡ start_ARG italic_S end_ARGmejkal, Y. Chen, S. S. P. Parkin, S. D. Wilson, E. S. Toberer, T. McQueen, and M. N. Ali, Giant, unconventional anomalous Hall effect in the metallic frustrated magnet candidate, KV3Sb5, Sci. Adv. 6, eabb6003 (2020).
  • (25) F. H. Yu, T. Wu, Z. Y. Wang, B. Lei, W. Z. Zhuo, J. J. Ying, and X. H. Chen, Concurrence of anomalous Hall effect and charge density wave in a superconducting topological kagome metal, Phys. Rev. B 104, L041103 (2021).
  • (26) S. Onari and H. Kontani, Self-consistent Vertex Correction Analysis for Iron-based Superconductors: Mechanism of Coulomb Interaction-Driven Orbital Fluctuations, Phys. Rev. Lett. 109, 137001 (2012).
  • (27) Y. Yamakawa, S. Onari, and H. Kontani, Nematicity and Magnetism in FeSe and Other Families of Fe-Based Superconductors, Phys. Rev. X 6, 021032 (2016).
  • (28) S. Onari and H. Kontani, SU(4) Valley+SpinValleySpin\text{Valley}+\text{Spin}Valley + Spin Fluctuation Interference Mechanism for Nematic Order in Magic-Angle Twisted Bilayer Graphene: The Impact of Vertex Corrections, Phys. Rev. Lett. 128, 066401 (2022).
  • (29) M. Tsuchiizu, Y. Ohno, S. Onari, and H. Kontani, Orbital Nematic Instability in the Two-Orbital Hubbard Model: Renormalization-Group + Constrained RPA Analysis, Phys. Rev. Lett. 111, 057003 (2013).
  • (30) M. Tsuchiizu, K. Kawaguchi, Y. Yamakawa, and H. Kontani, Multistage electronic nematic transitions in cuprate superconductors: A functional-renormalization-group analysis, Phys. Rev. B 97, 165131 (2018).
  • (31) A. V. Chubukov, M. Khodas, and R. M. Fernandes, Magnetism, Superconductivity, and Spontaneous Orbital Order in Iron-Based Superconductors: Which Comes First and Why?, Phys. Rev. X 6, 041045 (2016).
  • (32) R. M. Fernandes, P. P. Orth, and J. Schmalian, Intertwined Vestigial Order in Quantum Materials: Nematicity and Beyond, Annu. Rev. Condens. Matter Phys. 10, 133 (2019).
  • (33) H. Kontani, R. Tazai, Y. Yamakawa, and S. Onari, Unconventional density waves and superconductivities in Fe-based superconductors and other strongly correlated electron systems, Adv. Phys. 70, 355 (2021).
  • (34) J. C. S. Davis and D.-H. Lee, Concepts relating magnetic interactions, intertwined electronic orders, and strongly correlated superconductivity, Proc. Natl. Acad. Sci. U.S.A. 110, 17623 (2013).
  • (35) R. Tazai, Y. Yamakawa, and H. Kontani, Charge-loop current order and Z3 nematicity mediated by bond-order fluctuations in kagome metal AV3Sb5 (A=Cs,Rb,K), arXiv:2207.08068 (avalable at https://arxiv.org/abs/2207.08068).
  • (36) S. Zhou and Z. Wang, Chern Fermi pocket, topological pair density wave, and charge-4e and charge-6e superconductivity in kagomΓ© superconductors, Nat. Commun. 13, 7288 (2022).
  • (37) M. H. Christensen, T. Biro, B. M. Andersen, and R. M. Fernandes, Loop currents in AV3Sb5 kagome metals: Multipolar and toroidal magnetic orders, Phys. Rev. B 106, 144504 (2022)
  • (38) F. Grandi, A. Consiglio, M. A. Sentef, R. Thomale, and D. M. Kennes, Theory of nematic charge orders in kagome metals, Phys. Rev. B 107, 155131 (2023)
  • (39) H. D. Scammell, J. Ingham, T. Li, and O. P. Sushkov, Chiral excitonic order from twofold van Hove singularities in kagome metals, Nat. Commun. 14, 605 (2023)
  • (40) L. Nie, K. Sun, W. Ma, D. Song, L. Zheng, Z. Liang, P. Wu, F. Yu, J. Li, M. Shan, D. Zhao, S. Li, B. Kang, Z. Wu, Y. Zhou, K. Liu, Z. Xiang, J. Ying, Z. Wang, T. Wu, and X. Chen, Charge-density-wave-driven electronic nematicity in a kagome superconductor, Nature 604, 59 (2022).
  • (41) C. Guo, G. Wagner, C. Putzke, D. Chen, K. Wang, L. Zhang, M. G. Amigo, I. Errea, M. G. Vergniory, C. Felser, M. H. Fischer, T. Neupert, and P. J. W. Moll, Correlated order at the tipping point in the kagome metal CsV3Sb5, arXiv:2304.00972.
  • (42) Y. Hu, X. Wu, B. R. Ortiz, S. Ju, X. Han, J. Ma, N. C. Plumb, M. Radovic, R. Thomale, S. D. Wilson, A. P. Schnyder, and M. Shi, Rich nature of Van Hove singularities in Kagome superconductor CsV3Sb5, Nat. Commun. 13, 2220 (2022).
  • (43) Y. Luo, S. Peng, S. M. L. Teicher, L. Huai, Y. Hu, B. R. Ortiz, Z. Wei, J. Shen, Z. Ou, B. Wang, Y. Miao, M. Guo, M. Shi, S. D. Wilson, and J.-F. He, Distinct band reconstructions in kagome superconductor CsV3Sb5, Phys. Rev. B 105, L241111 (2022).
  • (44) Z. Liu, N. Zhao, Q. Yin, C. Gong, Z. Tu, M. Li, W. Song, Z. Liu, D. Shen, Y. Huang, K. Liu, H. Lei, and S. Wang, Charge-Density-Wave-Induced Bands Renormalization and Energy Gaps in a Kagome Superconductor RbV3⁒Sb5subscriptRbV3subscriptSb5{\mathrm{RbV}}_{3}{\mathrm{Sb}}_{5}roman_RbV start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT roman_Sb start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT, Phys. Rev. X 11, 041010 (2021).
  • (45) K. Nakayama, Y. Li, T. Kato, M. Liu, Z. Wang, T. Takahashi, Y. Yao, and T. Sato Carrier Injection and Manipulation of Charge-Density Wave in Kagome Superconductor CsV3Sb5, Phys. Rev. X 12, 011001 (2022).
  • (46) D. Ceresoli, T. Thonhauser, D. Vanderbilt, and R. Resta, Orbital magnetization in crystalline solids: Multi-band insulators, Chern insulators, and metals, Phys. Rev. B 74, 024408 (2006).
  • (47) J. Shi, G. Vignale, D. Xiao, and Q. Niu, Quantum Theory of Orbital Magnetization and Its Generalization to Interacting Systems, Phys. Rev. Lett. 99, 197202 (2007).
  • (48) R. Tazai, Y. Yamakawa, and H. Kontani, Emergence of charge loop current in the geometrically frustrated Hubbard model: A functional renormalization group study, Phys. Rev. B 103, L161112 (2021).
  • (49) Supplementary Information.
  • (50) R. Tazai, S. Matsubara, Y. Yamakawa, S. Onari, and H. Kontani, Rigorous formalism for unconventional symmetry breaking in Fermi liquid theory and its application to nematicity in FeSe, Phys. Rev. B 107, 035137 (2023).
  • (51) C. Guo, M. R. Delft, M. G. Amigo, D. Chen, C. Putzke, G. Wagner, M. H. Fischer, T. Neupert, I. Errea, M. G. Vergniory, S. Wiedmann, C. Felser, P. J. W. Moll, Distinct switching of chiral transport in the kagome metals KV3Sb5 and CsV3Sb5, arXiv:2306.00593.
  • (52) H. W. S. Arachchige, W. R. Meier, M. Marshall, T. Matsuoka, R. Xue, M. A. McGuire, R. P. Hermann, H. Cao, and D. Mandrus, Charge Density Wave in Kagome Lattice Intermetallic ScV6Sn6, Phys. Rev. Lett. 129, 216402 (2022).
  • (53) A. Korshunov, H. Hu, D. Subires, Y. Jiang, D. Calugaru, X. Feng, A. Rajapitamahuni, C. Yi, S. Roychowdhury, M. G. Vergniory, J. Strempfer, C. Shekhar, E. Vescovo, D. Chernyshov, A. H. Said, A. Bosak, C. Felser, B. A. Bernevig, and S. Blanco-Canosa, Softening of a flat phonon mode in the kagome ScV6Sn6, arXiv: 2304.09173.
  • (54) H. Hu, Y. Jiang, D. Calug, X. Feng, D. Subires, M. G. Vergniory, C. Felser, S. Blanco-Canosa, and B. A. Bernevig, Kagome Materials I: SG 191, ScV6Sn6. Flat Phonon Soft Modes and Unconventional CDW Formation: Microscopic and Effective Theory, arXiv:2305.15469.
  • (55) Z. Guguchia, D.J. Gawryluk, Soohyeon Shin, Z. Hao, C. Mielke III, D. Das, I. Plokhikh, L. Liborio, K. Shenton, Y. Hu, V. Sazgari, M. Medarde, H. Deng, Y. Cai, C. Chen, Y. Jiang, A. Amato, M. Shi, M.Z. Hasan, J.-X. Yin, R. Khasanov, E. Pomjakushina, and H. Luetkens, Hidden magnetism uncovered in charge ordered bilayer kagome material ScV6Sn6, arXiv:2304.06436.
  • (56) S. Onari, R. Tazai, Y. Yamakawa, and H. Kontani, unpublished.

[Supplementary Information]
Drastic magnetic-field-induced chiral current order and emergent current-bond-field interplay in kagome metal AV3Sb5 (A=Cs,Rb,K)

Rina Tazai1βˆ—, Youichi Yamakawa2βˆ—, and Hiroshi Kontani2

1Yukawa Institute for Theoretical Physics, Kyoto University, Kyoto 606-8502, Japan
2Department of Physics, Nagoya University, Nagoya 464-8602, Japan

A: Bandstructure with bond+current orders

In the main text, we study the orbital magnetization Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT in the presence of the current and bond orders based on Eqs. [3] and [6] in the main text Morb-paper1S ; Morb-paper2S . Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT originates from the vertical p-h excitation between the occupied bands (Ο΅Ξ±β’π’Œ<ΞΌsubscriptitalic-Ο΅π›Όπ’Œπœ‡{\epsilon}_{{\alpha}{{\bm{k}}}}<\muitalic_Ο΅ start_POSTSUBSCRIPT italic_Ξ± bold_italic_k end_POSTSUBSCRIPT < italic_ΞΌ) and unoccupied bands (Ο΅Ξ²β’π’Œ>ΞΌsubscriptitalic-Ο΅π›½π’Œπœ‡{\epsilon}_{{\beta}{{\bm{k}}}}>\muitalic_Ο΅ start_POSTSUBSCRIPT italic_Ξ² bold_italic_k end_POSTSUBSCRIPT > italic_ΞΌ). In the present study, Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT becomes finite in the presence of the TRSB order current order parameters. Thus, it is important to understand the change in the bandstructure due to the order parameters. Hereafter, the unit of energy is eV unless otherwise noted.

Refer to caption
Figure S1: (a) Folded FS at vHS filling for Ο•=Ξ·=0italic-Ο•πœ‚0\phi=\eta=0italic_Ο• = italic_Ξ· = 0. (b)(c) Folded bandstructure in the presence of (b) the 3⁒Q3𝑄3Q3 italic_Q current order (Ξ·=0πœ‚0\eta=0italic_Ξ· = 0, 0.010.010.010.01) and (c) the 3⁒Q3𝑄3Q3 italic_Q BO (Ο•=0italic-Ο•0\phi=0italic_Ο• = 0, 0.010.010.010.01). Here, we set the origin of the energy at the vHS energy.

The folded FS at the vHS filling for Ο•=Ξ·=0italic-Ο•πœ‚0\phi=\eta=0italic_Ο• = italic_Ξ· = 0 is shown in Fig. S1 (a). Figure S1 (b) shows the folded bandstructure at n=nvHS𝑛subscript𝑛vHSn=n_{\rm vHS}italic_n = italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT around ΓΓ\Gammaroman_Ξ“ point in the presence of 3⁒Q3𝑄3Q3 italic_Q current order 𝜼=(Ξ·,Ξ·,Ξ·)πœΌπœ‚πœ‚πœ‚{\bm{\eta}}=(\eta,\eta,\eta)bold_italic_Ξ· = ( italic_Ξ· , italic_Ξ· , italic_Ξ· ) for Ο•=0italic-Ο•0\phi=0italic_Ο• = 0 and 0.010.010.010.01. Here, all vHS points A, B, C in Fig. 1 (b) move to ΓΓ\Gammaroman_Ξ“ point, and they split into one bonding (δ⁒E=βˆ’3⁒η𝛿𝐸3πœ‚\delta E=-\sqrt{3}\etaitalic_Ξ΄ italic_E = - square-root start_ARG 3 end_ARG italic_Ξ·), one antibonding (δ⁒E=3⁒η𝛿𝐸3πœ‚\delta E=\sqrt{3}\etaitalic_Ξ΄ italic_E = square-root start_ARG 3 end_ARG italic_Ξ·), and one unhybridized (δ⁒E=0𝛿𝐸0\delta E=0italic_Ξ΄ italic_E = 0) states for finite Ξ·πœ‚\etaitalic_Ξ·. (The bandstructure for 3⁒Q3𝑄3Q3 italic_Q BO Ο•=(Ο•,Ο•,Ο•)bold-italic-Ο•italic-Ο•italic-Ο•italic-Ο•{\bm{\phi}}=(\phi,\phi,\phi)bold_italic_Ο• = ( italic_Ο• , italic_Ο• , italic_Ο• ) is shown in Fig. S1 (c). The three vHS points split into two bonding (δ⁒E=βˆ’Ο•π›ΏπΈitalic-Ο•\delta E=-\phiitalic_Ξ΄ italic_E = - italic_Ο•) and one antibonding (δ⁒E=2⁒ϕ𝛿𝐸2italic-Ο•\delta E=2\phiitalic_Ξ΄ italic_E = 2 italic_Ο•).)

In the bandstructure shown in Fig. S1 (a), the β€œvertical p-h excitations” (from Ο΅Ξ±β’π’Œ<0subscriptitalic-Ο΅π›Όπ’Œ0{\epsilon}_{{\alpha}{{\bm{k}}}}<0italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ± bold_italic_k end_POSTSUBSCRIPT < 0 to Ο΅Ξ²β’π’Œ>0subscriptitalic-Ο΅π›½π’Œ0{\epsilon}_{{\beta}{{\bm{k}}}}>0italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ² bold_italic_k end_POSTSUBSCRIPT > 0 for ΞΌ=0πœ‡0\mu=0italic_ΞΌ = 0) in the expression of Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT are allowed in a wide π’Œπ’Œ{{\bm{k}}}bold_italic_k-space for μ∼0similar-toπœ‡0\mu\sim 0italic_ΞΌ ∼ 0; around ΓΓ\Gammaroman_Ξ“-M and M-M’ lines and around ΓΓ\Gammaroman_Ξ“ and M points. Then, the factor (Ο΅Ξ±β’π’Œ+Ο΅Ξ²β’π’Œβˆ’2⁒μ)βˆ’2superscriptsubscriptitalic-Ο΅π›Όπ’Œsubscriptitalic-Ο΅π›½π’Œ2πœ‡2({\epsilon}_{{\alpha}{{\bm{k}}}}+{\epsilon}_{{\beta}{{\bm{k}}}}-2\mu)^{-2}( italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ± bold_italic_k end_POSTSUBSCRIPT + italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ² bold_italic_k end_POSTSUBSCRIPT - 2 italic_ΞΌ ) start_POSTSUPERSCRIPT - 2 end_POSTSUPERSCRIPT in the integrand of Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT is O⁒(Ξ·βˆ’2)𝑂superscriptπœ‚2O(\eta^{-2})italic_O ( italic_Ξ· start_POSTSUPERSCRIPT - 2 end_POSTSUPERSCRIPT ), and the cancellation due to the factor (Ο΅Ξ±β’π’Œ+Ο΅Ξ²β’π’Œβˆ’2⁒μ)subscriptitalic-Ο΅π›Όπ’Œsubscriptitalic-Ο΅π›½π’Œ2πœ‡({\epsilon}_{{\alpha}{{\bm{k}}}}+{\epsilon}_{{\beta}{{\bm{k}}}}-2\mu)( italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ± bold_italic_k end_POSTSUBSCRIPT + italic_Ο΅ start_POSTSUBSCRIPT italic_Ξ² bold_italic_k end_POSTSUBSCRIPT - 2 italic_ΞΌ ) is imperfect due to the p-h asymmetry. For this reason, large |Morb|subscript𝑀orb|M_{\rm orb}|| italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT | is realized at n∼nvHSsimilar-to𝑛subscript𝑛vHSn\sim n_{\rm vHS}italic_n ∼ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT.

B: Relation m2β‰ˆβˆ’m3subscriptπ‘š2subscriptπ‘š3m_{2}\approx-m_{3}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT β‰ˆ - italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT, Expansion of Morb⁒[Ξ·,Ο•]subscript𝑀orbπœ‚italic-Ο•M_{\rm orb}[{\eta},{\phi}]italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT [ italic_Ξ· , italic_Ο• ] based on the Green function method

In Fig. S2, we show the obtained m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT and βˆ’m3subscriptπ‘š3-m_{3}- italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT in the (a) three-orbital kagome lattice model and (b) 30-orbital first-principles model. In both models, the relation m2β‰ˆβˆ’m3subscriptπ‘š2subscriptπ‘š3m_{2}\approx-m_{3}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT β‰ˆ - italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT is well satisfied. For this reason, we present only m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT in the main text.

Refer to caption
Figure S2: Obtained m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT and βˆ’m3subscriptπ‘š3-m_{3}- italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT in the (a) three-orbital kagome lattice model and (b) 30-orbital first-principles model with Δ⁒Ep=βˆ’0.2Ξ”subscript𝐸𝑝0.2\Delta E_{p}=-0.2roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2. (c) Diagrammatic expression of Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT given in Eq. [S1].

To understand the approximate relation m2β‰ˆβˆ’m3subscriptπ‘š2subscriptπ‘š3m_{2}\approx-m_{3}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT β‰ˆ - italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT, we consider the expansion of Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT with respect to 𝜼𝜼{\bm{\eta}}bold_italic_Ξ· and Ο•bold-italic-Ο•{\bm{\phi}}bold_italic_Ο•; Morb=βˆ‘bp⁒r⁒qp′⁒q′⁒r′⁒η1p⁒η2q⁒η3r⁒ϕ1p′⁒ϕ2q′⁒ϕ3rβ€²subscript𝑀orbsuperscriptsubscriptπ‘π‘π‘Ÿπ‘žsuperscript𝑝′superscriptπ‘žβ€²superscriptπ‘Ÿβ€²superscriptsubscriptπœ‚1𝑝superscriptsubscriptπœ‚2π‘žsuperscriptsubscriptπœ‚3π‘Ÿsuperscriptsubscriptitalic-Ο•1superscript𝑝′superscriptsubscriptitalic-Ο•2superscriptπ‘žβ€²superscriptsubscriptitalic-Ο•3superscriptπ‘Ÿβ€²M_{\rm orb}=\sum b_{prq}^{p^{\prime}q^{\prime}r^{\prime}}\eta_{1}^{p}\eta_{2}^% {q}\eta_{3}^{r}\phi_{1}^{p^{\prime}}\phi_{2}^{q^{\prime}}\phi_{3}^{r^{\prime}}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT = βˆ‘ italic_b start_POSTSUBSCRIPT italic_p italic_r italic_q end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_p start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_q start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_r start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_p end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_q end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_r end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_p start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_q start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_r start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT. In this notation, m2=b111000subscriptπ‘š2subscriptsuperscript𝑏000111m_{2}=b^{000}_{111}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = italic_b start_POSTSUPERSCRIPT 000 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT 111 end_POSTSUBSCRIPT and m3=b001110subscriptπ‘š3subscriptsuperscript𝑏110001m_{3}=b^{110}_{001}italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT = italic_b start_POSTSUPERSCRIPT 110 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT 001 end_POSTSUBSCRIPT. To discuss the nature of the coefficient bp⁒r⁒qp′⁒q′⁒rβ€²superscriptsubscriptπ‘π‘π‘Ÿπ‘žsuperscript𝑝′superscriptπ‘žβ€²superscriptπ‘Ÿβ€²b_{prq}^{p^{\prime}q^{\prime}r^{\prime}}italic_b start_POSTSUBSCRIPT italic_p italic_r italic_q end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_p start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_q start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_r start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT, we analyze the expression of Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT based on the thermal Green function method derived in Ref. KotliarS :

Morbsubscript𝑀orb\displaystyle M_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT =\displaystyle== ΞΌBE0⁒Nuc⁒Nβ’βˆ‘a⁒bx,y,zΟ΅z⁒a⁒bsubscriptπœ‡Bsubscript𝐸0subscript𝑁uc𝑁superscriptsubscriptπ‘Žπ‘π‘₯𝑦𝑧subscriptitalic-Ο΅π‘§π‘Žπ‘\displaystyle\frac{\mu_{\rm B}}{E_{0}N_{\rm uc}N}\sum_{ab}^{x,y,z}{\epsilon}_{zab}divide start_ARG italic_ΞΌ start_POSTSUBSCRIPT roman_B end_POSTSUBSCRIPT end_ARG start_ARG italic_E start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT italic_N start_POSTSUBSCRIPT roman_uc end_POSTSUBSCRIPT italic_N end_ARG βˆ‘ start_POSTSUBSCRIPT italic_a italic_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_x , italic_y , italic_z end_POSTSUPERSCRIPT italic_Ο΅ start_POSTSUBSCRIPT italic_z italic_a italic_b end_POSTSUBSCRIPT (S1)
Γ—Tβ’βˆ‘π’Œ,nImTr⁒(Ξ»^π’Œβ’G^⁒(k)⁒v^π’Œb⁒G^⁒(k)⁒v^π’Œc⁒G^⁒(k)),absent𝑇subscriptπ’Œπ‘›ImTrsubscript^πœ†π’Œ^πΊπ‘˜superscriptsubscript^π‘£π’Œπ‘^πΊπ‘˜superscriptsubscript^π‘£π’Œπ‘^πΊπ‘˜\displaystyle\times T\sum_{{{\bm{k}}},n}{\rm Im}{\rm Tr}\large({\hat{\lambda}}% _{{\bm{k}}}{\hat{G}}(k){\hat{v}}_{{\bm{k}}}^{b}{\hat{G}}(k){\hat{v}}_{{\bm{k}}% }^{c}{\hat{G}}(k)\large),Γ— italic_T βˆ‘ start_POSTSUBSCRIPT bold_italic_k , italic_n end_POSTSUBSCRIPT roman_ImTr ( over^ start_ARG italic_Ξ» end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT over^ start_ARG italic_G end_ARG ( italic_k ) over^ start_ARG italic_v end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_b end_POSTSUPERSCRIPT over^ start_ARG italic_G end_ARG ( italic_k ) over^ start_ARG italic_v end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_c end_POSTSUPERSCRIPT over^ start_ARG italic_G end_ARG ( italic_k ) ) ,

where h^π’Œsubscript^β„Žπ’Œ{\hat{h}}_{{\bm{k}}}over^ start_ARG italic_h end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT is the tight-binding Hamiltonian with finite 𝜼,Ο•πœΌbold-italic-Ο•{\bm{\eta}},{\bm{\phi}}bold_italic_Ξ· , bold_italic_Ο•, Ξ»^π’Œ=h^π’Œβˆ’ΞΌβ’1^subscript^πœ†π’Œsubscript^β„Žπ’Œπœ‡^1{\hat{\lambda}}_{{\bm{k}}}={\hat{h}}_{{\bm{k}}}-\mu{\hat{1}}over^ start_ARG italic_Ξ» end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT = over^ start_ARG italic_h end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT - italic_ΞΌ over^ start_ARG 1 end_ARG, vπ’Œa=βˆ‚h^π’Œ/βˆ‚kasuperscriptsubscriptπ‘£π’Œπ‘Žsubscript^β„Žπ’Œsubscriptπ‘˜π‘Žv_{{\bm{k}}}^{a}={\partial}{\hat{h}}_{{\bm{k}}}/{\partial}k_{a}italic_v start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_a end_POSTSUPERSCRIPT = βˆ‚ over^ start_ARG italic_h end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT / βˆ‚ italic_k start_POSTSUBSCRIPT italic_a end_POSTSUBSCRIPT is the velocity, and G^⁒(k)^πΊπ‘˜{\hat{G}}(k)over^ start_ARG italic_G end_ARG ( italic_k ) is the Green function. Ο΅a⁒b⁒csubscriptitalic-Ο΅π‘Žπ‘π‘{\epsilon}_{abc}italic_Ο΅ start_POSTSUBSCRIPT italic_a italic_b italic_c end_POSTSUBSCRIPT is the antisymmetric tensor, and k≑(π’Œ,Ο΅n)π‘˜π’Œsubscriptitalic-ϡ𝑛k\equiv({{\bm{k}}},{\epsilon}_{n})italic_k ≑ ( bold_italic_k , italic_Ο΅ start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT ) with Ο΅n=(2⁒n+1)⁒π⁒Tsubscriptitalic-ϡ𝑛2𝑛1πœ‹π‘‡{\epsilon}_{n}=(2n+1)\pi Titalic_Ο΅ start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT = ( 2 italic_n + 1 ) italic_Ο€ italic_T. The diagrammatic expression of Eq. [S1] is shown in Fig. S2 (c). The Hamiltonian of the 4Γ—3434\times 34 Γ— 3 site model,

h^π’Œ=h^π’Œ0+Ξ£^π’Œ,subscript^β„Žπ’Œsuperscriptsubscript^β„Žπ’Œ0subscript^Ξ£π’Œ\displaystyle{\hat{h}}_{{\bm{k}}}={\hat{h}}_{{\bm{k}}}^{0}+{\hat{\Sigma}}_{{% \bm{k}}},over^ start_ARG italic_h end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT = over^ start_ARG italic_h end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT + over^ start_ARG roman_Ξ£ end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT , (S2)

where h^π’Œ0superscriptsubscript^β„Žπ’Œ0{\hat{h}}_{{\bm{k}}}^{0}over^ start_ARG italic_h end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT is the 12Γ—12121212\times 1212 Γ— 12 Hamiltonian for 𝜼=Ο•=𝟎𝜼bold-italic-Ο•0{\bm{\eta}}={\bm{\phi}}={\bm{0}}bold_italic_Ξ· = bold_italic_Ο• = bold_0 and Ξ£^π’Œl,m=Ξ·l,m⁒fπ’Œl,m+Ο•l,m⁒gπ’Œl,msuperscriptsubscript^Ξ£π’Œπ‘™π‘šsubscriptπœ‚π‘™π‘šsuperscriptsubscriptπ‘“π’Œπ‘™π‘šsubscriptitalic-Ο•π‘™π‘šsuperscriptsubscriptπ‘”π’Œπ‘™π‘š{\hat{\Sigma}}_{{\bm{k}}}^{l,m}=\eta_{l,m}f_{{\bm{k}}}^{l,m}+\phi_{l,m}g_{{\bm% {k}}}^{l,m}over^ start_ARG roman_Ξ£ end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l , italic_m end_POSTSUPERSCRIPT = italic_Ξ· start_POSTSUBSCRIPT italic_l , italic_m end_POSTSUBSCRIPT italic_f start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l , italic_m end_POSTSUPERSCRIPT + italic_Ο• start_POSTSUBSCRIPT italic_l , italic_m end_POSTSUBSCRIPT italic_g start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l , italic_m end_POSTSUPERSCRIPT, where fπ’Œl,msuperscriptsubscriptπ‘“π’Œπ‘™π‘šf_{{\bm{k}}}^{l,m}italic_f start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l , italic_m end_POSTSUPERSCRIPT [gπ’Œl,msuperscriptsubscriptπ‘”π’Œπ‘™π‘šg_{{\bm{k}}}^{l,m}italic_g start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l , italic_m end_POSTSUPERSCRIPT] is the form factor of the current order [BO], which is given by the Fourier transform of fi⁒jsubscript𝑓𝑖𝑗f_{ij}italic_f start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT [gi⁒jsubscript𝑔𝑖𝑗g_{ij}italic_g start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT] introduced in the main text. Ξ·l,msubscriptπœ‚π‘™π‘š\eta_{l,m}italic_Ξ· start_POSTSUBSCRIPT italic_l , italic_m end_POSTSUBSCRIPT [Ο•l,msubscriptitalic-Ο•π‘™π‘š\phi_{l,m}italic_Ο• start_POSTSUBSCRIPT italic_l , italic_m end_POSTSUBSCRIPT] (l,m=1∼12π‘™π‘š1similar-to12l,m=1\sim 12italic_l , italic_m = 1 ∼ 12) is the current order [BO] parameter. Then, the Green function is expanded as

G^⁒(k)^πΊπ‘˜\displaystyle{\hat{G}}(k)over^ start_ARG italic_G end_ARG ( italic_k ) =\displaystyle== G^0⁒(k)+G^0⁒(k)⁒Σ^π’Œβ’G^0⁒(k)superscript^𝐺0π‘˜superscript^𝐺0π‘˜subscript^Ξ£π’Œsuperscript^𝐺0π‘˜\displaystyle{\hat{G}}^{0}(k)+{\hat{G}}^{0}(k){\hat{\Sigma}}_{{\bm{k}}}{\hat{G% }}^{0}(k)over^ start_ARG italic_G end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_k ) + over^ start_ARG italic_G end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_k ) over^ start_ARG roman_Ξ£ end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT over^ start_ARG italic_G end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_k ) (S3)
+G^0⁒(k)⁒Σ^π’Œβ’G^0⁒(k)⁒Σ^π’Œβ’G^0⁒(k)superscript^𝐺0π‘˜subscript^Ξ£π’Œsuperscript^𝐺0π‘˜subscript^Ξ£π’Œsuperscript^𝐺0π‘˜\displaystyle+{\hat{G}}^{0}(k){\hat{\Sigma}}_{{\bm{k}}}{\hat{G}}^{0}(k){\hat{% \Sigma}}_{{\bm{k}}}{\hat{G}}^{0}(k)+ over^ start_ARG italic_G end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_k ) over^ start_ARG roman_Ξ£ end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT over^ start_ARG italic_G end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_k ) over^ start_ARG roman_Ξ£ end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT over^ start_ARG italic_G end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_k )
+β‹―,β‹―\displaystyle+\cdots,+ β‹― ,

where G^0⁒(k)=((i⁒ϡnβˆ’ΞΌ)⁒1^βˆ’h^π’Œ0)βˆ’1superscript^𝐺0π‘˜superscript𝑖subscriptitalic-Ο΅π‘›πœ‡^1subscriptsuperscript^β„Ž0π’Œ1{\hat{G}}^{0}(k)=((i{\epsilon}_{n}-\mu){\hat{1}}-{\hat{h}}^{0}_{{\bm{k}}})^{-1}over^ start_ARG italic_G end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_k ) = ( ( italic_i italic_Ο΅ start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT - italic_ΞΌ ) over^ start_ARG 1 end_ARG - over^ start_ARG italic_h end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT ) start_POSTSUPERSCRIPT - 1 end_POSTSUPERSCRIPT. Thus, the coefficient m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT (m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT) can be derived from Eq. [S1], by expanding it with respect to Ξ·1⁒ϕ1subscriptπœ‚1subscriptitalic-Ο•1\eta_{1}\phi_{1}italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT (Ξ·1⁒η2⁒η3subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3\eta_{1}\eta_{2}\eta_{3}italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT) by using Eqs. [S2]-[S3].

Because of the relations

hπ’Œ0,l⁒m=hβˆ’π’Œ0,m⁒l,superscriptsubscriptβ„Žπ’Œ0π‘™π‘šsuperscriptsubscriptβ„Žπ’Œ0π‘šπ‘™\displaystyle h_{{\bm{k}}}^{0,lm}=h_{-{{\bm{k}}}}^{0,ml},italic_h start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 , italic_l italic_m end_POSTSUPERSCRIPT = italic_h start_POSTSUBSCRIPT - bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 , italic_m italic_l end_POSTSUPERSCRIPT , (S4)
G0,l⁒m⁒(π’Œ,Ο΅m)=G0,m⁒l⁒(βˆ’π’Œ,Ο΅m),superscript𝐺0π‘™π‘šπ’Œsubscriptitalic-Ο΅π‘šsuperscript𝐺0π‘šπ‘™π’Œsubscriptitalic-Ο΅π‘š\displaystyle G^{0,lm}({{\bm{k}}},{\epsilon}_{m})=G^{0,ml}(-{{\bm{k}}},{% \epsilon}_{m}),italic_G start_POSTSUPERSCRIPT 0 , italic_l italic_m end_POSTSUPERSCRIPT ( bold_italic_k , italic_Ο΅ start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT ) = italic_G start_POSTSUPERSCRIPT 0 , italic_m italic_l end_POSTSUPERSCRIPT ( - bold_italic_k , italic_Ο΅ start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT ) , (S5)

Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT given by Eq. [S1] is zero for Ο•=𝜼=𝟎bold-italic-Ο•πœΌ0{\bm{\phi}}={\bm{\eta}}={\bm{0}}bold_italic_Ο• = bold_italic_Ξ· = bold_0. Equations [S4] and [S5] are violated for πœΌβ‰ πŸŽπœΌ0{\bm{\eta}}\neq{\bm{0}}bold_italic_Ξ· β‰  bold_0 because the current order form factor is odd-parity: fπ’Œl,m=βˆ’fβˆ’π’Œm,l=βˆ’(fβˆ’π’Œl,m)βˆ—superscriptsubscriptπ‘“π’Œπ‘™π‘šsuperscriptsubscriptπ‘“π’Œπ‘šπ‘™superscriptsuperscriptsubscriptπ‘“π’Œπ‘™π‘šf_{{\bm{k}}}^{l,m}=-f_{-{{\bm{k}}}}^{m,l}=-(f_{-{{\bm{k}}}}^{l,m})^{*}italic_f start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l , italic_m end_POSTSUPERSCRIPT = - italic_f start_POSTSUBSCRIPT - bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_m , italic_l end_POSTSUPERSCRIPT = - ( italic_f start_POSTSUBSCRIPT - bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l , italic_m end_POSTSUPERSCRIPT ) start_POSTSUPERSCRIPT βˆ— end_POSTSUPERSCRIPT. (Note that gπ’Œl,m=fβˆ’π’Œm,l=(gβˆ’π’Œl,m)βˆ—superscriptsubscriptπ‘”π’Œπ‘™π‘šsuperscriptsubscriptπ‘“π’Œπ‘šπ‘™superscriptsuperscriptsubscriptπ‘”π’Œπ‘™π‘šg_{{\bm{k}}}^{l,m}=f_{-{{\bm{k}}}}^{m,l}=(g_{-{{\bm{k}}}}^{l,m})^{*}italic_g start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l , italic_m end_POSTSUPERSCRIPT = italic_f start_POSTSUBSCRIPT - bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_m , italic_l end_POSTSUPERSCRIPT = ( italic_g start_POSTSUBSCRIPT - bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l , italic_m end_POSTSUPERSCRIPT ) start_POSTSUPERSCRIPT βˆ— end_POSTSUPERSCRIPT for the BO. Any Hermitian orders with the wavevector 𝒒=πŸŽπ’’0{{\bm{q}}}={\bm{0}}bold_italic_q = bold_0 satisfy fπ’Œl,m=(fπ’Œm,l)βˆ—superscriptsubscriptπ‘“π’Œπ‘™π‘šsuperscriptsuperscriptsubscriptπ‘“π’Œπ‘šπ‘™f_{{\bm{k}}}^{l,m}=(f_{{{\bm{k}}}}^{m,l})^{*}italic_f start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l , italic_m end_POSTSUPERSCRIPT = ( italic_f start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_m , italic_l end_POSTSUPERSCRIPT ) start_POSTSUPERSCRIPT βˆ— end_POSTSUPERSCRIPT.) Thus, only odd-order terms with respect to Ξ·πœ‚\etaitalic_Ξ· can give nonzero Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT.

Here, we consider Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT in Eq. [S1] in the original 3-site unit cell picture. Then, the Green function in Eq. [S1] is expressed as G^⁒(k+q;k)=G^0⁒(k)⋅δ𝒒,𝟎+βˆ‘mG^0⁒(k+qm)⁒Σ^k(m)⁒G^0⁒(k)⋅δ𝒒,𝒒m+βˆ‘m,nG^0⁒(k+qm+qn)⁒Σ^k+qn(m)⁒G^0⁒(k+qn)⁒Σ^k(n)⁒G^0⁒(k)⋅δ𝒒,𝒒m+𝒒n+β‹―^πΊπ‘˜π‘žπ‘˜β‹…superscript^𝐺0π‘˜subscript𝛿𝒒0subscriptπ‘šβ‹…superscript^𝐺0π‘˜subscriptπ‘žπ‘šsuperscriptsubscript^Ξ£π‘˜π‘šsuperscript^𝐺0π‘˜subscript𝛿𝒒subscriptπ’’π‘šsubscriptπ‘šπ‘›β‹…superscript^𝐺0π‘˜subscriptπ‘žπ‘šsubscriptπ‘žπ‘›subscriptsuperscript^Ξ£π‘šπ‘˜subscriptπ‘žπ‘›superscript^𝐺0π‘˜subscriptπ‘žπ‘›subscriptsuperscript^Ξ£π‘›π‘˜superscript^𝐺0π‘˜subscript𝛿𝒒subscriptπ’’π‘šsubscript𝒒𝑛⋯{\hat{G}}(k+q;k)={\hat{G}}^{0}(k)\cdot\delta_{{{\bm{q}}},{\bm{0}}}+\sum_{m}{% \hat{G}}^{0}(k+q_{m}){\hat{\Sigma}}_{k}^{(m)}{\hat{G}}^{0}(k)\cdot\delta_{{{% \bm{q}}},{{\bm{q}}}_{m}}+\sum_{m,n}{\hat{G}}^{0}(k+q_{m}+q_{n}){\hat{\Sigma}}^% {(m)}_{k+q_{n}}{\hat{G}}^{0}(k+q_{n}){\hat{\Sigma}}^{(n)}_{k}{\hat{G}}^{0}(k)% \cdot\delta_{{{\bm{q}}},{{\bm{q}}}_{m}+{{\bm{q}}}_{n}}+\cdotsover^ start_ARG italic_G end_ARG ( italic_k + italic_q ; italic_k ) = over^ start_ARG italic_G end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_k ) β‹… italic_Ξ΄ start_POSTSUBSCRIPT bold_italic_q , bold_0 end_POSTSUBSCRIPT + βˆ‘ start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT over^ start_ARG italic_G end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_k + italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT ) over^ start_ARG roman_Ξ£ end_ARG start_POSTSUBSCRIPT italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT over^ start_ARG italic_G end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_k ) β‹… italic_Ξ΄ start_POSTSUBSCRIPT bold_italic_q , bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT end_POSTSUBSCRIPT + βˆ‘ start_POSTSUBSCRIPT italic_m , italic_n end_POSTSUBSCRIPT over^ start_ARG italic_G end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_k + italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT + italic_q start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT ) over^ start_ARG roman_Ξ£ end_ARG start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_k + italic_q start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT end_POSTSUBSCRIPT over^ start_ARG italic_G end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_k + italic_q start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT ) over^ start_ARG roman_Ξ£ end_ARG start_POSTSUPERSCRIPT ( italic_n ) end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_k end_POSTSUBSCRIPT over^ start_ARG italic_G end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_k ) β‹… italic_Ξ΄ start_POSTSUBSCRIPT bold_italic_q , bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT + bold_italic_q start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT end_POSTSUBSCRIPT + β‹―, where G^^𝐺{\hat{G}}over^ start_ARG italic_G end_ARG and Ξ£^(m)superscript^Ξ£π‘š{\hat{\Sigma}}^{(m)}over^ start_ARG roman_Ξ£ end_ARG start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT are 3Γ—3333\times 33 Γ— 3 matrices. The momentum 𝒒msubscriptπ’’π‘š{{\bm{q}}}_{m}bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT is introduced by Ξ£^π’Œ(m)subscriptsuperscript^Ξ£π‘šπ’Œ{\hat{\Sigma}}^{(m)}_{{\bm{k}}}over^ start_ARG roman_Ξ£ end_ARG start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT. To obtain finite Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT, the total momenta introduced by Ξ£^(mi)superscript^Ξ£subscriptπ‘šπ‘–{\hat{\Sigma}}^{(m_{i})}over^ start_ARG roman_Ξ£ end_ARG start_POSTSUPERSCRIPT ( italic_m start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) end_POSTSUPERSCRIPT, βˆ‘i1,2,⋯𝒒misuperscriptsubscript𝑖12β‹―subscript𝒒subscriptπ‘šπ‘–\sum_{i}^{1,2,\cdots}{{\bm{q}}}_{m_{i}}βˆ‘ start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 1 , 2 , β‹― end_POSTSUPERSCRIPT bold_italic_q start_POSTSUBSCRIPT italic_m start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT end_POSTSUBSCRIPT, should be 𝟎0{\bm{0}}bold_0 (modulo original reciprocal vectors).

Hereafter, we discuss the reason for the relation m2β‰ˆβˆ’m3subscriptπ‘š2subscriptπ‘š3m_{2}\approx-m_{3}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT β‰ˆ - italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT. In this study, the nearest-neighbor order parameters are given in Fig. 1 (c) for the current order and Fig. 1 (a) for the BO. In the original 3-site unit cell picture, the relation fπ’Œβ€²l⁒l′⁒fπ’Œl′⁒lβ€²β€²=βˆ’gπ’Œβ€²l⁒l′⁒gπ’Œl′⁒lβ€²β€²superscriptsubscript𝑓superscriptπ’Œβ€²π‘™superscript𝑙′superscriptsubscriptπ‘“π’Œsuperscript𝑙′superscript𝑙′′superscriptsubscript𝑔superscriptπ’Œβ€²π‘™superscript𝑙′superscriptsubscriptπ‘”π’Œsuperscript𝑙′superscript𝑙′′f_{{{\bm{k}}}^{\prime}}^{ll^{\prime}}f_{{\bm{k}}}^{l^{\prime}l^{\prime\prime}}% =-g_{{{\bm{k}}}^{\prime}}^{ll^{\prime}}g_{{\bm{k}}}^{l^{\prime}l^{\prime\prime}}italic_f start_POSTSUBSCRIPT bold_italic_k start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT italic_f start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_l start_POSTSUPERSCRIPT β€² β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT = - italic_g start_POSTSUBSCRIPT bold_italic_k start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT italic_g start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_l start_POSTSUPERSCRIPT β€² β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT (π’Œβ€²=π’Œ+(𝒒lβ€²βˆ’π’’lβ€²β€²)superscriptπ’Œβ€²π’Œsubscript𝒒superscript𝑙′subscript𝒒superscript𝑙′′{{\bm{k}}}^{\prime}={{\bm{k}}}+({{\bm{q}}}_{l^{\prime}}-{{\bm{q}}}_{l^{\prime% \prime}})bold_italic_k start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = bold_italic_k + ( bold_italic_q start_POSTSUBSCRIPT italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT - bold_italic_q start_POSTSUBSCRIPT italic_l start_POSTSUPERSCRIPT β€² β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT )) holds, where lβ‰ lβ€²β‰ l′′𝑙superscript𝑙′superscript𝑙′′l\neq l^{\prime}\neq l^{\prime\prime}italic_l β‰  italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT β‰  italic_l start_POSTSUPERSCRIPT β€² β€² end_POSTSUPERSCRIPT. For instance, we consider a m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT term given by replacing v^x⁒(y)β’π’Œsubscript^𝑣π‘₯π‘¦π’Œ{\hat{v}}_{x(y){{\bm{k}}}}over^ start_ARG italic_v end_ARG start_POSTSUBSCRIPT italic_x ( italic_y ) bold_italic_k end_POSTSUBSCRIPT, Ξ»^π’Œsubscript^πœ†π’Œ{\hat{\lambda}}_{{\bm{k}}}over^ start_ARG italic_Ξ» end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT and G^^𝐺{\hat{G}}over^ start_ARG italic_G end_ARG with (βˆ‚/βˆ‚kx⁒(y))⁒f^π’Œsubscriptπ‘˜π‘₯𝑦subscript^π‘“π’Œ({\partial}/{\partial}k_{x(y)}){\hat{f}}_{{\bm{k}}}( βˆ‚ / βˆ‚ italic_k start_POSTSUBSCRIPT italic_x ( italic_y ) end_POSTSUBSCRIPT ) over^ start_ARG italic_f end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT, f^π’Œsubscript^π‘“π’Œ{\hat{f}}_{{\bm{k}}}over^ start_ARG italic_f end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT and G^0superscript^𝐺0{\hat{G}}^{0}over^ start_ARG italic_G end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT, respectively. The corresponding m3subscriptπ‘š3m_{3}italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT term is given by replacing (βˆ‚/βˆ‚kx⁒(y))⁒f^π’Œsubscriptπ‘˜π‘₯𝑦subscript^π‘“π’Œ({\partial}/{\partial}k_{x(y)}){\hat{f}}_{{\bm{k}}}( βˆ‚ / βˆ‚ italic_k start_POSTSUBSCRIPT italic_x ( italic_y ) end_POSTSUBSCRIPT ) over^ start_ARG italic_f end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT with (βˆ‚/βˆ‚kx⁒(y))⁒g^π’Œsubscriptπ‘˜π‘₯𝑦subscript^π‘”π’Œ({\partial}/{\partial}k_{x(y)}){\hat{g}}_{{\bm{k}}}( βˆ‚ / βˆ‚ italic_k start_POSTSUBSCRIPT italic_x ( italic_y ) end_POSTSUBSCRIPT ) over^ start_ARG italic_g end_ARG start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT in m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT. Due to the absence of sublattice hybridization around the vHS points Thomale2021S , G^0⁒(k)superscript^𝐺0π‘˜{\hat{G}}^{0}(k)over^ start_ARG italic_G end_ARG start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( italic_k ) is sublattice diagonal around the vHS points, that is, G0,m⁒m′⁒(k)=0superscript𝐺0π‘šsuperscriptπ‘šβ€²π‘˜0G^{0,mm^{\prime}}(k)=0italic_G start_POSTSUPERSCRIPT 0 , italic_m italic_m start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT ( italic_k ) = 0 for mβ‰ mβ€²π‘šsuperscriptπ‘šβ€²m\neq m^{\prime}italic_m β‰  italic_m start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT. Then, the relation m2=βˆ’m3subscriptπ‘š2subscriptπ‘š3m_{2}=-m_{3}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = - italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT is satisfied approximately due to the relation fπ’Œβ€²l⁒l′⁒fπ’Œl′⁒lβ€²β€²=βˆ’gπ’Œβ€²l⁒l′⁒gπ’Œl′⁒lβ€²β€²superscriptsubscript𝑓superscriptπ’Œβ€²π‘™superscript𝑙′superscriptsubscriptπ‘“π’Œsuperscript𝑙′superscript𝑙′′superscriptsubscript𝑔superscriptπ’Œβ€²π‘™superscript𝑙′superscriptsubscriptπ‘”π’Œsuperscript𝑙′superscript𝑙′′f_{{{\bm{k}}}^{\prime}}^{ll^{\prime}}f_{{\bm{k}}}^{l^{\prime}l^{\prime\prime}}% =-g_{{{\bm{k}}}^{\prime}}^{ll^{\prime}}g_{{\bm{k}}}^{l^{\prime}l^{\prime\prime}}italic_f start_POSTSUBSCRIPT bold_italic_k start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT italic_f start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_l start_POSTSUPERSCRIPT β€² β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT = - italic_g start_POSTSUBSCRIPT bold_italic_k start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT italic_g start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_l start_POSTSUPERSCRIPT β€² β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT with π’Œβ€²=π’Œ+(𝒒lβ€²βˆ’π’’lβ€²β€²)superscriptπ’Œβ€²π’Œsubscript𝒒superscript𝑙′subscript𝒒superscript𝑙′′{{\bm{k}}}^{\prime}={{\bm{k}}}+({{\bm{q}}}_{l^{\prime}}-{{\bm{q}}}_{l^{\prime% \prime}})bold_italic_k start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = bold_italic_k + ( bold_italic_q start_POSTSUBSCRIPT italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT - bold_italic_q start_POSTSUBSCRIPT italic_l start_POSTSUPERSCRIPT β€² β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT ).

We have just started the analysis based on the Green function method. This is our important future issue.

C: Analysis of GL free energy with non-analytic Ξ·πœ‚\etaitalic_Ξ·-linear term

In the main text, we studied the strong interplay between the bond and current orders under the magnetic field in kagome metals. In Fig. 4, we studied the situation where bond order transition temperature Tb0superscriptsubscript𝑇b0T_{\rm b}^{0}italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT is higher than the chiral current order one Tc0superscriptsubscript𝑇c0T_{\rm c}^{0}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT at hz=0subscriptβ„Žπ‘§0h_{z}=0italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 0. We revealed that chiral current order emerges at T=Tb0𝑇superscriptsubscript𝑇b0T=T_{\rm b}^{0}italic_T = italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT under hz∼10βˆ’4(∼1⁒T)similar-tosubscriptβ„Žπ‘§annotatedsuperscript104similar-toabsent1Th_{z}\sim 10^{-4}\ (\sim 1{\rm T})italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT ∼ 10 start_POSTSUPERSCRIPT - 4 end_POSTSUPERSCRIPT ( ∼ 1 roman_T ). That is, the current-order transition temperature is enlarged to Tb0superscriptsubscript𝑇b0T_{\rm b}^{0}italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT under small hzsubscriptβ„Žπ‘§h_{z}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT, as shown in Fig. 3 (b). The drastic field-induced 3⁒Q3𝑄3Q3 italic_Q current order originates from the non-analytic Ξ·πœ‚\etaitalic_Ξ·-linear terms in Δ⁒F¯Δ¯𝐹\Delta{\bar{F}}roman_Ξ” overΒ― start_ARG italic_F end_ARG, that is, Ξ”FΒ―=βˆ’3hz[m1Ο•β‹…πœΌ+m3(Ξ·1Ο•2Ο•3+cycl.)+O(Ξ·3)]\Delta{\bar{F}}=-3h_{z}[m_{1}{\bm{\phi}}\cdot{\bm{\eta}}+m_{3}(\eta_{1}\phi_{2% }\phi_{3}+{\rm cycl.})+O(\eta^{3})]roman_Ξ” overΒ― start_ARG italic_F end_ARG = - 3 italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT [ italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT bold_italic_Ο• β‹… bold_italic_Ξ· + italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT + roman_cycl . ) + italic_O ( italic_Ξ· start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT ) ].

Refer to caption
Figure S3: (a) Induced Ξ·gsubscriptπœ‚π‘”\eta_{g}italic_Ξ· start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT due to the Ξ·πœ‚\etaitalic_Ξ·-linear term for as a function of a(>0)annotatedπ‘Žabsent0a\ (>0)italic_a ( > 0 ). Here, g=0.9⁒hz𝑔0.9subscriptβ„Žπ‘§g=0.9h_{z}italic_g = 0.9 italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT and hz=1∼3Γ—10βˆ’4subscriptβ„Žπ‘§1similar-to3superscript104h_{z}=1\sim 3\times 10^{-4}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 1 ∼ 3 Γ— 10 start_POSTSUPERSCRIPT - 4 end_POSTSUPERSCRIPT. This result is not sensitive to the choice of d𝑑ditalic_d when g𝑔gitalic_g is small. (b) Δ⁒F¯Δ¯𝐹\Delta{\bar{F}}roman_Ξ” overΒ― start_ARG italic_F end_ARG as a function of Ξ·πœ‚\etaitalic_Ξ· for a=0.06π‘Ž0.06a=0.06italic_a = 0.06. (Inset) Enlarged plots around the bottoms.

To understand the effect of the field-induced non-analytic free energy qualitatively, we analyze the following simple GL free energy with a Ξ·πœ‚\etaitalic_Ξ·-linear term:

F⁒(Ξ·)=a⁒η2+d⁒η4βˆ’g⁒η.πΉπœ‚π‘Žsuperscriptπœ‚2𝑑superscriptπœ‚4π‘”πœ‚\displaystyle F(\eta)=a\eta^{2}+d\eta^{4}-g\eta.italic_F ( italic_Ξ· ) = italic_a italic_Ξ· start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_d italic_Ξ· start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT - italic_g italic_Ξ· . (S6)

Here, we assume aπ‘Žaitalic_a and d𝑑ditalic_d are positive. When gβ‰ 0𝑔0g\neq 0italic_g β‰  0, F⁒(Ξ·)πΉπœ‚F(\eta)italic_F ( italic_Ξ· ) is minimized at finite Ξ·πœ‚\etaitalic_Ξ· even if a>0π‘Ž0a>0italic_a > 0. The solution is given as

Ξ·gsubscriptπœ‚π‘”\displaystyle\eta_{g}italic_Ξ· start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT =\displaystyle== X1/3⁒(dβˆ’1βˆ’2β‹…31/3⁒a⁒Xβˆ’2/3)/(2β‹…32/3),superscript𝑋13superscript𝑑1β‹…2superscript313π‘Žsuperscript𝑋23β‹…2superscript323\displaystyle X^{1/3}(d^{-1}-2\cdot 3^{1/3}aX^{-2/3})/(2\cdot 3^{2/3}),italic_X start_POSTSUPERSCRIPT 1 / 3 end_POSTSUPERSCRIPT ( italic_d start_POSTSUPERSCRIPT - 1 end_POSTSUPERSCRIPT - 2 β‹… 3 start_POSTSUPERSCRIPT 1 / 3 end_POSTSUPERSCRIPT italic_a italic_X start_POSTSUPERSCRIPT - 2 / 3 end_POSTSUPERSCRIPT ) / ( 2 β‹… 3 start_POSTSUPERSCRIPT 2 / 3 end_POSTSUPERSCRIPT ) , (S7)
X𝑋\displaystyle Xitalic_X =\displaystyle== 9⁒g⁒d2+3⁒d3⁒(8⁒a3+27⁒d⁒g2).9𝑔superscript𝑑23superscript𝑑38superscriptπ‘Ž327𝑑superscript𝑔2\displaystyle 9gd^{2}+\sqrt{3}\sqrt{d^{3}(8a^{3}+27dg^{2})}.9 italic_g italic_d start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + square-root start_ARG 3 end_ARG square-root start_ARG italic_d start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT ( 8 italic_a start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT + 27 italic_d italic_g start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ) end_ARG . (S8)

Note that

Ξ·gβ‰ˆg/2⁒a,subscriptπœ‚π‘”π‘”2π‘Ž\displaystyle\eta_{g}\approx g/2a,italic_Ξ· start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT β‰ˆ italic_g / 2 italic_a , (S9)

for small g𝑔gitalic_g. Therefore, Ξ·gsubscriptπœ‚π‘”\eta_{g}italic_Ξ· start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT is finite even when a>0π‘Ž0a>0italic_a > 0 due to the Ξ·πœ‚\etaitalic_Ξ·-linear term.

Here, we derive d𝑑ditalic_d and g𝑔gitalic_g from the GL free energy of kagome metal, Eqs. [11] and [10] in the main text. By setting Ο•=(Ο•,Ο•,Ο•)bold-italic-Ο•italic-Ο•italic-Ο•italic-Ο•{\bm{\phi}}=(\phi,\phi,\phi)bold_italic_Ο• = ( italic_Ο• , italic_Ο• , italic_Ο• ) and 𝜼=(Ξ·,Ξ·,Ξ·)πœΌπœ‚πœ‚πœ‚{\bm{\eta}}=(\eta,\eta,\eta)bold_italic_Ξ· = ( italic_Ξ· , italic_Ξ· , italic_Ξ· ) in Eqs. [11] and [10], we obtain d=6⁒di𝑑6subscript𝑑𝑖d=6d_{i}italic_d = 6 italic_d start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT and g=3⁒(3⁒m1⁒ϕ+3⁒m3⁒ϕ2)⁒hz𝑔33subscriptπ‘š1italic-Ο•3subscriptπ‘š3superscriptitalic-Ο•2subscriptβ„Žπ‘§g=3(3m_{1}\phi+3m_{3}\phi^{2})h_{z}italic_g = 3 ( 3 italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ο• + 3 italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT italic_Ο• start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ) italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT. When di=150subscript𝑑𝑖150d_{i}=150italic_d start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT = 150, m1=5subscriptπ‘š15m_{1}=5italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = 5, m3=1000subscriptπ‘š31000m_{3}=1000italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT = 1000 and Ο•=0.01italic-Ο•0.01\phi=0.01italic_Ο• = 0.01, we get d=900𝑑900d=900italic_d = 900 and g=1.35⁒hz𝑔1.35subscriptβ„Žπ‘§g=1.35h_{z}italic_g = 1.35 italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT. aπ‘Žaitalic_a in Eq. [S6] corresponds to 3⁒ac3subscriptπ‘Žc3a_{\rm c}3 italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT in the main text. In Fig. 3 (b) in the main text, we set ac=rc⁒(Tβˆ’Tc0)subscriptπ‘Žcsubscriptπ‘Ÿc𝑇superscriptsubscript𝑇c0a_{\rm c}=r_{\rm c}(T-T_{\rm c}^{0})italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT = italic_r start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( italic_T - italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ) with rc=30subscriptπ‘Ÿc30r_{\rm c}=30italic_r start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT = 30 and Tc0=0.005superscriptsubscript𝑇c00.005T_{\rm c}^{0}=0.005italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = 0.005, so T=Tc0⁒(2⁒Tc0=Tb0)𝑇superscriptsubscript𝑇c02superscriptsubscript𝑇c0superscriptsubscript𝑇b0T=T_{\rm c}^{0}\ (2T_{\rm c}^{0}=T_{\rm b}^{0})italic_T = italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( 2 italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ) in Fig. 4 (b) corresponds to a=0⁒(0.45)π‘Ž00.45a=0\ (0.45)italic_a = 0 ( 0.45 ) in Eq. [S6].

Figure S3 (a) shows Ξ·gsubscriptπœ‚π‘”\eta_{g}italic_Ξ· start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT given in Eq. [S8] as a function of aπ‘Žaitalic_a, in the case of d=1000𝑑1000d=1000italic_d = 1000 and g=1.35⁒hz𝑔1.35subscriptβ„Žπ‘§g=1.35h_{z}italic_g = 1.35 italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT. We see that hz∼10βˆ’4similar-tosubscriptβ„Žπ‘§superscript104h_{z}\sim 10^{-4}italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT ∼ 10 start_POSTSUPERSCRIPT - 4 end_POSTSUPERSCRIPT, which corresponds to ∼1similar-toabsent1\sim 1∼ 1T, induces sizable current order even above Tc0superscriptsubscript𝑇c0T_{\rm c}^{0}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT if a≳0greater-than-or-equivalent-toπ‘Ž0a\gtrsim 0italic_a ≳ 0. The obtained value of Ξ·gsubscriptπœ‚π‘”\eta_{g}italic_Ξ· start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT in Fig S3 (a) at aβ‰ˆ0π‘Ž0a\approx 0italic_a β‰ˆ 0 is comparable to the field-induced order in Fig. 4 (b). The field-induced Ξ·πœ‚\etaitalic_Ξ· is prominent only when the system at hz=0subscriptβ„Žπ‘§0h_{z}=0italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 0 is close to the current order state (i.e., a≳0greater-than-or-equivalent-toπ‘Ž0a\gtrsim 0italic_a ≳ 0). Asymmetric Δ⁒F¯Δ¯𝐹\Delta{\bar{F}}roman_Ξ” overΒ― start_ARG italic_F end_ARG as a function of Ξ·πœ‚\etaitalic_Ξ· is shown in Fig S3 (b).

D: Comparison between MΒ―orbsubscript¯𝑀orb{\bar{M}}_{\rm orb}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT and Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT

In the main text, we calculated the orbital magnetization Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT in kagome metal with current order 𝜼𝜼{\bm{\eta}}bold_italic_Ξ· and bond-order Ο•bold-italic-Ο•{\bm{\phi}}bold_italic_Ο• using Eq. [3]. Next, we derived its expression up to the third-order of 𝜼𝜼{\bm{\eta}}bold_italic_Ξ· and Ο•bold-italic-Ο•{\bm{\phi}}bold_italic_Ο•: MΒ―orb=m1Ο•β‹…πœΌ+m2Ξ·1Ξ·2Ξ·3+m3(Ξ·1Ο•2Ο•3+(cycl.)){\bar{M}}_{\rm orb}=m_{1}{\bm{\phi}}\cdot{\bm{\eta}}+m_{2}\eta_{1}\eta_{2}\eta% _{3}+m_{3}(\eta_{1}\phi_{2}\phi_{3}+({\rm cycl.}))overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT = italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT bold_italic_Ο• β‹… bold_italic_Ξ· + italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT + italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT + ( roman_cycl . ) ). To obtained the coefficients m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT, we calculate Morb⁒[𝜼,Ο•]subscript𝑀orb𝜼bold-italic-Ο•M_{\rm orb}[{\bm{\eta}},{\bm{\phi}}]italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT [ bold_italic_Ξ· , bold_italic_Ο• ] very accurately and expand it around 𝜼=Ο•=0𝜼bold-italic-Ο•0{\bm{\eta}}={\bm{\phi}}=0bold_italic_Ξ· = bold_italic_Ο• = 0 numerically. For instance, we derive m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT as m2=Morb⁒[(Ξ·1,Ξ·2,Ξ·3),𝟎]/Ξ·1⁒η2⁒η3subscriptπ‘š2subscript𝑀orbsubscriptπœ‚1subscriptπœ‚2subscriptπœ‚30subscriptπœ‚1subscriptπœ‚2subscriptπœ‚3m_{2}=M_{\rm orb}[(\eta_{1},\eta_{2},\eta_{3}),{\bm{0}}]/\eta_{1}\eta_{2}\eta_% {3}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT [ ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ) , bold_0 ] / italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT with Ξ·m=0.001subscriptπœ‚π‘š0.001\eta_{m}=0.001italic_Ξ· start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT = 0.001. (Note that Morb⁒[𝜼,𝟎]=0subscript𝑀orb𝜼00M_{\rm orb}[{\bm{\eta}},{\bm{0}}]=0italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT [ bold_italic_Ξ· , bold_0 ] = 0 if one of Ξ·msubscriptπœ‚π‘š\eta_{m}italic_Ξ· start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT is zero.)

The expression MΒ―orbsubscript¯𝑀orb{\bar{M}}_{\rm orb}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT is very useful to understand the strong interplay between current and bond orders in kagome metal. By considering the field-induced free energy Δ⁒FΒ―=βˆ’3⁒hz⁒MΒ―orbΔ¯𝐹3subscriptβ„Žπ‘§subscript¯𝑀orb\Delta{\bar{F}}=-3h_{z}{\bar{M}}_{\rm orb}roman_Ξ” overΒ― start_ARG italic_F end_ARG = - 3 italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT, we understand the characteristic phase diagram of kagome metals under the magnetic field.

Here, we verify that MΒ―orbsubscript¯𝑀orb{\bar{M}}_{\rm orb}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT in Eq. [8] well reproduces Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT in Eq. [3]. Figures S4 (a)-(c) show the obtained results at n=2.47𝑛2.47n=2.47italic_n = 2.47 under 𝜼=(Ξ·,Ξ·,Ξ·)/3πœΌπœ‚πœ‚πœ‚3{\bm{\eta}}=(\eta,\eta,\eta)/\sqrt{3}bold_italic_Ξ· = ( italic_Ξ· , italic_Ξ· , italic_Ξ· ) / square-root start_ARG 3 end_ARG and Ο•=(Ο•,Ο•,Ο•)/3bold-italic-Ο•italic-Ο•italic-Ο•italic-Ο•3{\bm{\phi}}=(\phi,\phi,\phi)/\sqrt{3}bold_italic_Ο• = ( italic_Ο• , italic_Ο• , italic_Ο• ) / square-root start_ARG 3 end_ARG as functions of Ξ·πœ‚\etaitalic_Ξ·, in the case of (a) Ο•=0italic-Ο•0\phi=0italic_Ο• = 0, (b) Ο•=+0.01italic-Ο•0.01\phi=+0.01italic_Ο• = + 0.01 and (b) Ο•=βˆ’0.01italic-Ο•0.01\phi=-0.01italic_Ο• = - 0.01. It is found that MΒ―orbsubscript¯𝑀orb{\bar{M}}_{\rm orb}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT well reproduce the original Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT when |𝜼|,|Ο•|≲0.02less-than-or-similar-to𝜼bold-italic-Ο•0.02|{\bm{\eta}}|,|{\bm{\phi}}|\lesssim 0.02| bold_italic_Ξ· | , | bold_italic_Ο• | ≲ 0.02, unless the shape of the Fermi surface is drastically changed by order parameters.

Refer to caption
Figure S4: Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT and MΒ―orbsubscript¯𝑀orb{\bar{M}}_{\rm orb}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT for 𝜼=(Ξ·,Ξ·,Ξ·)/3πœΌπœ‚πœ‚πœ‚3{\bm{\eta}}=(\eta,\eta,\eta)/\sqrt{3}bold_italic_Ξ· = ( italic_Ξ· , italic_Ξ· , italic_Ξ· ) / square-root start_ARG 3 end_ARG in the cases of (a) Ο•=𝟎bold-italic-Ο•0{\bm{\phi}}={\bm{0}}bold_italic_Ο• = bold_0 and (b) Ο•=(Ο•,Ο•,Ο•)/3bold-italic-Ο•italic-Ο•italic-Ο•italic-Ο•3{\bm{\phi}}=(\phi,\phi,\phi)/\sqrt{3}bold_italic_Ο• = ( italic_Ο• , italic_Ο• , italic_Ο• ) / square-root start_ARG 3 end_ARG at Ο•=+0.01italic-Ο•0.01\phi=+0.01italic_Ο• = + 0.01 and (c) βˆ’0.010.01-0.01- 0.01.

Next, we examine the validity of the expansion expression MΒ―orbsubscript¯𝑀orb{\bar{M}}_{\rm orb}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT in the realistic 30-orbital model for kagome metal. Figure S5 shows the coefficients m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT given by Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT in Eq. [3] in the main text, derived from the region |Ξ·i|,|Ο•i|<Ξ›subscriptπœ‚π‘–subscriptitalic-ϕ𝑖Λ|\eta_{i}|,|\phi_{i}|<\Lambda| italic_Ξ· start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT | , | italic_Ο• start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT | < roman_Ξ›. The convergence of the obtained results for both m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT is good for Ξ›<0.01Ξ›0.01\Lambda<0.01roman_Ξ› < 0.01 for wide range of n𝑛nitalic_n, except at the close vicinity of the vHS filling. Therefore, the GL free energy expression [9] in the main text is valid for real kagome metals.

Refer to caption
Figure S5: Obtained coefficients (a) m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and (b) m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT at T=0.005𝑇0.005T=0.005italic_T = 0.005 derived from the region |Ξ·i|,|Ο•i|<Ξ›subscriptπœ‚π‘–subscriptitalic-ϕ𝑖Λ|\eta_{i}|,|\phi_{i}|<\Lambda| italic_Ξ· start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT | , | italic_Ο• start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT | < roman_Ξ›.

E: Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT in kagome lattice model with (t,tβ€²)=(βˆ’0.5,βˆ’0.08)𝑑superscript𝑑′0.50.08(t,t^{\prime})=(-0.5,-0.08)( italic_t , italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ) = ( - 0.5 , - 0.08 )

In the main text, we studied Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT in kagome lattice model with the bare hopping integrals t=βˆ’0.5𝑑0.5t=-0.5italic_t = - 0.5 and tβ€²=βˆ’0.02superscript𝑑′0.02t^{\prime}=-0.02italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = - 0.02. The obtained coefficients m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT in MΒ―orbsubscript¯𝑀orb{\bar{M}}_{\rm orb}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT take large values for nβ‰ˆnvHS𝑛subscript𝑛vHSn\approx n_{\rm vHS}italic_n β‰ˆ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT, where nvHSsubscript𝑛vHSn_{\rm vHS}italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT is the van-Hove filling. Using the obtained coefficients m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT, we discovered the mechanism of the field-induced chiral current order in kagome metals.

To verify the robustness of the obtained results, here we analyze Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT in kagome lattice model with large tβ€²superscript𝑑′t^{\prime}italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT; (t,tβ€²)=(βˆ’0.5,βˆ’0.08)𝑑superscript𝑑′0.50.08(t,t^{\prime})=(-0.5,-0.08)( italic_t , italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ) = ( - 0.5 , - 0.08 ). The obtained FS shown in Fig. S6 (a) has large curvature due to large |tβ€²|superscript𝑑′|t^{\prime}|| italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT |. Here, we have introduce the 3⁒Q3𝑄3Q3 italic_Q BO Ο•=(Ο•,Ο•,Ο•)/3bold-italic-Ο•italic-Ο•italic-Ο•italic-Ο•3{\bm{\phi}}=(\phi,\phi,\phi)/\sqrt{3}bold_italic_Ο• = ( italic_Ο• , italic_Ο• , italic_Ο• ) / square-root start_ARG 3 end_ARG and the 3⁒Q3𝑄3Q3 italic_Q current order 𝜼=(Ξ·,Ξ·,Ξ·)/3πœΌπœ‚πœ‚πœ‚3{\bm{\eta}}=(\eta,\eta,\eta)/\sqrt{3}bold_italic_Ξ· = ( italic_Ξ· , italic_Ξ· , italic_Ξ· ) / square-root start_ARG 3 end_ARG. Figure S6 (b) shows the obtained Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT [ΞΌBsubscriptπœ‡B\mu_{\rm B}italic_ΞΌ start_POSTSUBSCRIPT roman_B end_POSTSUBSCRIPT] per V𝑉Vitalic_V atom due to the 3⁒Q3𝑄3Q3 italic_Q current order with Ο•=0italic-Ο•0\phi=0italic_Ο• = 0. The relation Morb∝η3proportional-tosubscript𝑀orbsuperscriptπœ‚3M_{\rm orb}\propto\eta^{3}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT ∝ italic_Ξ· start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT is satisfied, and its magnitude is enlarged when nβ‰ˆnvHS𝑛subscript𝑛vHSn\approx n_{\rm vHS}italic_n β‰ˆ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT. Figure S6 (c) shows Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT [ΞΌBsubscriptπœ‡B\mu_{\rm B}italic_ΞΌ start_POSTSUBSCRIPT roman_B end_POSTSUBSCRIPT] due to the coexistence of 3⁒Q3𝑄3Q3 italic_Q current order and 3⁒Q3𝑄3Q3 italic_Q BO. The relation Morb∝η1proportional-tosubscript𝑀orbsuperscriptπœ‚1M_{\rm orb}\propto\eta^{1}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT ∝ italic_Ξ· start_POSTSUPERSCRIPT 1 end_POSTSUPERSCRIPT is satisfied when Ο•β‰ 0italic-Ο•0\phi\neq 0italic_Ο• β‰  0. Figures S6 (d) and (e) represent the obtained coefficients m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT in MΒ―orbsubscript¯𝑀orb{\bar{M}}_{\rm orb}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT. The magnitudes of m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT for tβ€²=βˆ’0.08superscript𝑑′0.08t^{\prime}=-0.08italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = - 0.08 are comparable to those for tβ€²=βˆ’0.02superscript𝑑′0.02t^{\prime}=-0.02italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = - 0.02 given in the main text.

Refer to caption
Figure S6: (a) FS around nvHS=2.75subscript𝑛vHS2.75n_{\rm vHS}=2.75italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT = 2.75 at tβ€²=βˆ’0.08superscript𝑑′0.08t^{\prime}=-0.08italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = - 0.08. (b) Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT [ΞΌBsubscriptπœ‡B\mu_{\rm B}italic_ΞΌ start_POSTSUBSCRIPT roman_B end_POSTSUBSCRIPT] per V𝑉Vitalic_V atom due to the 3⁒Q3𝑄3Q3 italic_Q current order 𝜼=(Ξ·,Ξ·,Ξ·)/3πœΌπœ‚πœ‚πœ‚3{\bm{\eta}}=(\eta,\eta,\eta)/\sqrt{3}bold_italic_Ξ· = ( italic_Ξ· , italic_Ξ· , italic_Ξ· ) / square-root start_ARG 3 end_ARG at T=1𝑇1T=1italic_T = 1 meV for n=2.67∼2.83𝑛2.67similar-to2.83n=2.67\sim 2.83italic_n = 2.67 ∼ 2.83. (c) Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT due to the coexistence of 3⁒Q3𝑄3Q3 italic_Q current order and 3⁒Q3𝑄3Q3 italic_Q BO at n=2.67𝑛2.67n=2.67italic_n = 2.67. Obtained (d) m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and (e) m2(β‰ˆβˆ’m3)annotatedsubscriptπ‘š2absentsubscriptπ‘š3m_{2}\ (\approx-m_{3})italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ( β‰ˆ - italic_m start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ), per V𝑉Vitalic_V atom as a function of n𝑛nitalic_n. Both |m1|subscriptπ‘š1|m_{1}|| italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT | and |m2|subscriptπ‘š2|m_{2}|| italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT | are large for n∼nvHSsimilar-to𝑛subscript𝑛vHSn\sim n_{\rm vHS}italic_n ∼ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT.

F: Calculation of GL parameters, Renormalization of acsubscriptπ‘Žca_{\rm c}italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT

To verify that the GL free energy coefficients assumed in the main text are qualitatively reasonable, we calculate GL coefficients based on the diagrammatic method. The 4rh order GL parameters per unit cell are given as

d1subscript𝑑1\displaystyle d_{1}italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT =\displaystyle== I1111g,superscriptsubscript𝐼1111𝑔\displaystyle I_{1111}^{g},italic_I start_POSTSUBSCRIPT 1111 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_g end_POSTSUPERSCRIPT , (S10)
d2subscript𝑑2\displaystyle d_{2}italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT =\displaystyle== 2⁒I1212g+4⁒I1122g,2superscriptsubscript𝐼1212𝑔4superscriptsubscript𝐼1122𝑔\displaystyle 2I_{1212}^{g}+4I_{1122}^{g},2 italic_I start_POSTSUBSCRIPT 1212 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_g end_POSTSUPERSCRIPT + 4 italic_I start_POSTSUBSCRIPT 1122 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_g end_POSTSUPERSCRIPT , (S11)
d3subscript𝑑3\displaystyle d_{3}italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT =\displaystyle== I1111f,superscriptsubscript𝐼1111𝑓\displaystyle I_{1111}^{f},italic_I start_POSTSUBSCRIPT 1111 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_f end_POSTSUPERSCRIPT , (S12)
d4subscript𝑑4\displaystyle d_{4}italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT =\displaystyle== 2⁒I1212f+4⁒I1122f,2superscriptsubscript𝐼1212𝑓4superscriptsubscript𝐼1122𝑓\displaystyle 2I_{1212}^{f}+4I_{1122}^{f},2 italic_I start_POSTSUBSCRIPT 1212 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_f end_POSTSUPERSCRIPT + 4 italic_I start_POSTSUBSCRIPT 1122 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_f end_POSTSUPERSCRIPT , (S13)

where

Ih⁒l⁒m⁒nwsuperscriptsubscriptπΌβ„Žπ‘™π‘šπ‘›π‘€\displaystyle I_{hlmn}^{w}italic_I start_POSTSUBSCRIPT italic_h italic_l italic_m italic_n end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_w end_POSTSUPERSCRIPT =\displaystyle== T4β’βˆ‘k,ΟƒTr⁒w^𝒒h⁒(k+𝒒n+𝒒m+𝒒l)𝑇4subscriptπ‘˜πœŽTrsubscript^𝑀subscriptπ’’β„Žπ‘˜subscript𝒒𝑛subscriptπ’’π‘šsubscript𝒒𝑙\displaystyle\frac{T}{4}\sum_{k,{\sigma}}{\rm Tr}{\hat{w}}_{{{\bm{q}}}_{h}}(k+% {{\bm{q}}}_{n}+{{\bm{q}}}_{m}+{{\bm{q}}}_{l})divide start_ARG italic_T end_ARG start_ARG 4 end_ARG βˆ‘ start_POSTSUBSCRIPT italic_k , italic_Οƒ end_POSTSUBSCRIPT roman_Tr over^ start_ARG italic_w end_ARG start_POSTSUBSCRIPT bold_italic_q start_POSTSUBSCRIPT italic_h end_POSTSUBSCRIPT end_POSTSUBSCRIPT ( italic_k + bold_italic_q start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT + bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT + bold_italic_q start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT ) (S14)
Γ—\displaystyle\timesΓ— G^⁒(k+𝒒n+𝒒m+𝒒l)^πΊπ‘˜subscript𝒒𝑛subscriptπ’’π‘šsubscript𝒒𝑙\displaystyle{\hat{G}}(k+{{\bm{q}}}_{n}+{{\bm{q}}}_{m}+{{\bm{q}}}_{l})over^ start_ARG italic_G end_ARG ( italic_k + bold_italic_q start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT + bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT + bold_italic_q start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT )
Γ—\displaystyle\timesΓ— w^𝒒l⁒(k+𝒒n+𝒒m)⁒G^⁒(k+𝒒n+𝒒m)subscript^𝑀subscriptπ’’π‘™π‘˜subscript𝒒𝑛subscriptπ’’π‘š^πΊπ‘˜subscript𝒒𝑛subscriptπ’’π‘š\displaystyle{\hat{w}}_{{{\bm{q}}}_{l}}(k+{{\bm{q}}}_{n}+{{\bm{q}}}_{m}){\hat{% G}}(k+{{\bm{q}}}_{n}+{{\bm{q}}}_{m})over^ start_ARG italic_w end_ARG start_POSTSUBSCRIPT bold_italic_q start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT end_POSTSUBSCRIPT ( italic_k + bold_italic_q start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT + bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT ) over^ start_ARG italic_G end_ARG ( italic_k + bold_italic_q start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT + bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT )
Γ—\displaystyle\timesΓ— w^𝒒m⁒(k+𝒒n)⁒G^⁒(k+𝒒n)⁒w^𝒒n⁒(k)⁒G^⁒(k),subscript^𝑀subscriptπ’’π‘šπ‘˜subscript𝒒𝑛^πΊπ‘˜subscript𝒒𝑛subscript^𝑀subscriptπ’’π‘›π‘˜^πΊπ‘˜\displaystyle{\hat{w}}_{{{\bm{q}}}_{m}}(k+{{\bm{q}}}_{n}){\hat{G}}(k+{{\bm{q}}% }_{n}){\hat{w}}_{{{\bm{q}}}_{n}}(k){\hat{G}}(k),over^ start_ARG italic_w end_ARG start_POSTSUBSCRIPT bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT end_POSTSUBSCRIPT ( italic_k + bold_italic_q start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT ) over^ start_ARG italic_G end_ARG ( italic_k + bold_italic_q start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT ) over^ start_ARG italic_w end_ARG start_POSTSUBSCRIPT bold_italic_q start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT end_POSTSUBSCRIPT ( italic_k ) over^ start_ARG italic_G end_ARG ( italic_k ) ,

where G^⁒(k)^πΊπ‘˜{\hat{G}}(k)over^ start_ARG italic_G end_ARG ( italic_k ) is 3Γ—3333\times 33 Γ— 3 Green function for the original 3-site kagome lattice model and k≑(π’Œ,i⁒ϡn)π‘˜π’Œπ‘–subscriptitalic-ϡ𝑛k\equiv({{\bm{k}}},i{\epsilon}_{n})italic_k ≑ ( bold_italic_k , italic_i italic_Ο΅ start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT ). The diagrammatic expression Ih⁒l⁒m⁒nwsuperscriptsubscriptπΌβ„Žπ‘™π‘šπ‘›π‘€I_{hlmn}^{w}italic_I start_POSTSUBSCRIPT italic_h italic_l italic_m italic_n end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_w end_POSTSUPERSCRIPT is depicted in Fig. S7 (a). w^𝒒⁒(π’Œ)subscript^π‘€π’’π’Œ{\hat{w}}_{{\bm{q}}}({{\bm{k}}})over^ start_ARG italic_w end_ARG start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT ( bold_italic_k ) is the form factor in the momentum space in the original BZ, f^𝒒⁒(π’Œ)subscript^π‘“π’’π’Œ{\hat{f}}_{{\bm{q}}}({{\bm{k}}})over^ start_ARG italic_f end_ARG start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT ( bold_italic_k ) or g^𝒒⁒(π’Œ)subscript^π‘”π’’π’Œ{\hat{g}}_{{\bm{q}}}({{\bm{k}}})over^ start_ARG italic_g end_ARG start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT ( bold_italic_k ), and h,l,m,nβ„Žπ‘™π‘šπ‘›h,l,m,nitalic_h , italic_l , italic_m , italic_n is 1, 2, or 3. Note that f𝒒ml⁒m⁒(π’Œ)subscriptsuperscriptπ‘“π‘™π‘šsubscriptπ’’π‘šπ’Œf^{lm}_{{{\bm{q}}}_{m}}({{\bm{k}}})italic_f start_POSTSUPERSCRIPT italic_l italic_m end_POSTSUPERSCRIPT start_POSTSUBSCRIPT bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT end_POSTSUBSCRIPT ( bold_italic_k ) [f𝒒ml⁒m⁒(π’Œ)subscriptsuperscriptπ‘“π‘™π‘šsubscriptπ’’π‘šπ’Œf^{lm}_{{{\bm{q}}}_{m}}({{\bm{k}}})italic_f start_POSTSUPERSCRIPT italic_l italic_m end_POSTSUPERSCRIPT start_POSTSUBSCRIPT bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT end_POSTSUBSCRIPT ( bold_italic_k )] (l,mπ‘™π‘šl,mitalic_l , italic_m=1-3) is given by the Fourier transform of fi,j(m)superscriptsubscriptπ‘“π‘–π‘—π‘šf_{i,j}^{(m)}italic_f start_POSTSUBSCRIPT italic_i , italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT [gi,j(m)superscriptsubscriptπ‘”π‘–π‘—π‘šg_{i,j}^{(m)}italic_g start_POSTSUBSCRIPT italic_i , italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT] introduced in the main text. The relation 𝒒h+𝒒l+𝒒m+𝒒n=0subscriptπ’’β„Žsubscript𝒒𝑙subscriptπ’’π‘šsubscript𝒒𝑛0{{\bm{q}}}_{h}+{{\bm{q}}}_{l}+{{\bm{q}}}_{m}+{{\bm{q}}}_{n}=0bold_italic_q start_POSTSUBSCRIPT italic_h end_POSTSUBSCRIPT + bold_italic_q start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT + bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT + bold_italic_q start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT = 0 should be satisfied. Hereafter, we use f^𝒒⁒(π’Œ)subscript^π‘“π’’π’Œ{\hat{f}}_{{\bm{q}}}({{\bm{k}}})over^ start_ARG italic_f end_ARG start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT ( bold_italic_k ) derived from the density-wave equation in Ref. Tazai-kagome2S , in which distant-atom components are included. On the other and, g^𝒒⁒(π’Œ)subscript^π‘”π’’π’Œ{\hat{g}}_{{\bm{q}}}({{\bm{k}}})over^ start_ARG italic_g end_ARG start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT ( bold_italic_k ) is derived from the nearest-neighbor BO in Fig. 1 (a). In the same way, the analytic expressions of d5subscript𝑑5d_{5}italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT and d6subscript𝑑6d_{6}italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT are obtained in Ref. Tazai-kagome2S .

Refer to caption
Figure S7: (a) Diagrammatic expressions of 4th order GL parameters. (b) Numerical results of the 4th order GL parameters as functions of T𝑇Titalic_T. (c) Ratio r=2⁒d1/d2π‘Ÿ2subscript𝑑1subscript𝑑2r=2d_{1}/d_{2}italic_r = 2 italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT and rβ€²=2⁒d3/d4superscriptπ‘Ÿβ€²2subscript𝑑3subscript𝑑4r^{\prime}=2d_{3}/d_{4}italic_r start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 2 italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT in the case of tβ€²=0superscript𝑑′0t^{\prime}=0italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 0 and tβ€²=βˆ’0.02superscript𝑑′0.02t^{\prime}=-0.02italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = - 0.02. (d) Diagrammatic expression of the self-energy driven by the BO fluctuations. (e) Expression of BO susceptibility.

The obtained numerical results are given in Fig. S7 (b) for tβ€²=0superscript𝑑′0t^{\prime}=0italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 0 and tβ€²=βˆ’0.02superscript𝑑′0.02t^{\prime}=-0.02italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = - 0.02. Here, the dimensionless form factors are normalized as maxπ’Œβ‘|g𝒒1BA⁒(π’Œ)|=|g𝒒1BA⁒(π’ŒA)|=1subscriptπ’Œsuperscriptsubscript𝑔subscript𝒒1BAπ’Œsuperscriptsubscript𝑔subscript𝒒1BAsubscriptπ’ŒA1\max_{{{\bm{k}}}}|g_{{{\bm{q}}}_{1}}^{\rm BA}({{\bm{k}}})|=|g_{{{\bm{q}}}_{1}}% ^{\rm BA}({{\bm{k}}}_{\rm A})|=1roman_max start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT | italic_g start_POSTSUBSCRIPT bold_italic_q start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_BA end_POSTSUPERSCRIPT ( bold_italic_k ) | = | italic_g start_POSTSUBSCRIPT bold_italic_q start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_BA end_POSTSUPERSCRIPT ( bold_italic_k start_POSTSUBSCRIPT roman_A end_POSTSUBSCRIPT ) | = 1. (This normalization corresponds to |δ⁒ti⁒jb|=1/2𝛿superscriptsubscript𝑑𝑖𝑗b12|\delta t_{ij}^{\rm b}|=1/2| italic_Ξ΄ italic_t start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_b end_POSTSUPERSCRIPT | = 1 / 2.) In the same way, we set maxπ’Œβ‘|f𝒒1BA⁒(π’Œ)|=|f𝒒1BA⁒(π’ŒA)|=1subscriptπ’Œsuperscriptsubscript𝑓subscript𝒒1BAπ’Œsuperscriptsubscript𝑓subscript𝒒1BAsubscriptπ’ŒA1\max_{{{\bm{k}}}}|f_{{{\bm{q}}}_{1}}^{\rm BA}({{\bm{k}}})|=|f_{{{\bm{q}}}_{1}}% ^{\rm BA}({{\bm{k}}}_{\rm A})|=1roman_max start_POSTSUBSCRIPT bold_italic_k end_POSTSUBSCRIPT | italic_f start_POSTSUBSCRIPT bold_italic_q start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_BA end_POSTSUPERSCRIPT ( bold_italic_k ) | = | italic_f start_POSTSUBSCRIPT bold_italic_q start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_BA end_POSTSUPERSCRIPT ( bold_italic_k start_POSTSUBSCRIPT roman_A end_POSTSUBSCRIPT ) | = 1. Thus, the parameter di=150⁒[eVβˆ’3]subscript𝑑𝑖150delimited-[]superscripteV3d_{i}=150\ [{\rm eV^{-3}}]italic_d start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT = 150 [ roman_eV start_POSTSUPERSCRIPT - 3 end_POSTSUPERSCRIPT ] is consistent with Fig. S7 (b) for T∼0.01similar-to𝑇0.01T\sim 0.01italic_T ∼ 0.01. For Ο•=(Ο•,Ο•,Ο•)bold-italic-Ο•italic-Ο•italic-Ο•italic-Ο•{\bm{\phi}}=(\phi,\phi,\phi)bold_italic_Ο• = ( italic_Ο• , italic_Ο• , italic_Ο• ) with Ο•=0.01italic-Ο•0.01\phi=0.01italic_Ο• = 0.01, the 4th order term for the BO is F4b∼150β‹…6⁒(10βˆ’2)4∼10βˆ’5similar-tosuperscriptsubscript𝐹4bβ‹…1506superscriptsuperscript1024similar-tosuperscript105F_{4}^{\rm b}\sim 150\cdot 6(10^{-2})^{4}\sim 10^{-5}italic_F start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_b end_POSTSUPERSCRIPT ∼ 150 β‹… 6 ( 10 start_POSTSUPERSCRIPT - 2 end_POSTSUPERSCRIPT ) start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT ∼ 10 start_POSTSUPERSCRIPT - 5 end_POSTSUPERSCRIPT [eV] for di=150subscript𝑑𝑖150d_{i}=150italic_d start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT = 150. The obtained ratios r=2⁒d1/d2π‘Ÿ2subscript𝑑1subscript𝑑2r=2d_{1}/d_{2}italic_r = 2 italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT and rβ€²=2⁒d3/d4superscriptπ‘Ÿβ€²2subscript𝑑3subscript𝑑4r^{\prime}=2d_{3}/d_{4}italic_r start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 2 italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT are shown in Fig. S7 (c): For tβ€²=0superscript𝑑′0t^{\prime}=0italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 0, we obtain 2⁒d1/d2≫1much-greater-than2subscript𝑑1subscript𝑑212d_{1}/d_{2}\gg 12 italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ≫ 1 and 2⁒d3/d4≳1greater-than-or-equivalent-to2subscript𝑑3subscript𝑑412d_{3}/d_{4}\gtrsim 12 italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT ≳ 1. For tβ€²=βˆ’0.02superscript𝑑′0.02t^{\prime}=-0.02italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = - 0.02, 2⁒d1/d2∼1similar-to2subscript𝑑1subscript𝑑212d_{1}/d_{2}\sim 12 italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ∼ 1 and 2⁒d3/d4β‰ˆ12subscript𝑑3subscript𝑑412d_{3}/d_{4}\approx 12 italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT / italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT β‰ˆ 1. Both ratios tend to become smaller than unity for larger |tβ€²|superscript𝑑′|t^{\prime}|| italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT |. (Note that d1β‰ˆd3β‰ˆd5subscript𝑑1subscript𝑑3subscript𝑑5d_{1}\approx d_{3}\approx d_{5}italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT β‰ˆ italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT β‰ˆ italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT and d2β‰ˆd4β‰ˆd6subscript𝑑2subscript𝑑4subscript𝑑6d_{2}\approx d_{4}\approx d_{6}italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT β‰ˆ italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT β‰ˆ italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT when the form factors satisfy fi⁒j(m)=Β±i⁒gi⁒j(m)superscriptsubscriptπ‘“π‘–π‘—π‘šplus-or-minus𝑖superscriptsubscriptπ‘”π‘–π‘—π‘šf_{ij}^{(m)}=\pm ig_{ij}^{(m)}italic_f start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT = Β± italic_i italic_g start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT.)

In deriving d1∼d6similar-tosubscript𝑑1subscript𝑑6d_{1}\sim d_{6}italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ∼ italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT in Fig. S7 (b), we included the self-energy due to the BO fluctuations Tazai-kagome2S , because the self-energy reduces unrealistic behaviors of disubscript𝑑𝑖d_{i}italic_d start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT at low temperatures when the inter-sublattice nesting vector is not exactly commensurate Balents2021S . (Note that the present method works well only for |tβ€²/t|β‰ͺ1much-less-thansuperscript𝑑′𝑑1|t^{\prime}/t|\ll 1| italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT / italic_t | β‰ͺ 1.) To calculate the self-energy, we introduce the following effective BO interaction Tazai-kagome2S :

H^intsubscript^𝐻int\displaystyle\hat{H}_{\rm int}over^ start_ARG italic_H end_ARG start_POSTSUBSCRIPT roman_int end_POSTSUBSCRIPT =\displaystyle== βˆ’1Nβ’βˆ‘π’’v2⁒O^𝒒g⁒O^βˆ’π’’g,1𝑁subscript𝒒𝑣2subscriptsuperscript^𝑂𝑔𝒒subscriptsuperscript^𝑂𝑔𝒒\displaystyle-\frac{1}{N}\sum_{{{\bm{q}}}}\frac{v}{2}\,\,\hat{O}^{g}_{{{\bm{q}% }}}\,\,\hat{O}^{g}_{-{{\bm{q}}}},- divide start_ARG 1 end_ARG start_ARG italic_N end_ARG βˆ‘ start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT divide start_ARG italic_v end_ARG start_ARG 2 end_ARG over^ start_ARG italic_O end_ARG start_POSTSUPERSCRIPT italic_g end_POSTSUPERSCRIPT start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT over^ start_ARG italic_O end_ARG start_POSTSUPERSCRIPT italic_g end_POSTSUPERSCRIPT start_POSTSUBSCRIPT - bold_italic_q end_POSTSUBSCRIPT , (S15)

where O^𝒒gβ‰‘βˆ‘π’Œ,l,m,Οƒg𝒒l⁒m⁒(π’Œ)⁒cπ’Œ+𝒒,l,σ†⁒cπ’Œ,m,Οƒsubscriptsuperscript^𝑂𝑔𝒒subscriptπ’Œπ‘™π‘šπœŽsuperscriptsubscriptπ‘”π’’π‘™π‘šπ’Œsubscriptsuperscriptπ‘β€ π’Œπ’’π‘™πœŽsubscriptπ‘π’Œπ‘šπœŽ\hat{O}^{g}_{{{\bm{q}}}}\equiv\sum_{{{\bm{k}}},l,m,\sigma}g_{{{\bm{q}}}}^{lm}(% {{\bm{k}}})c^{\dagger}_{{{\bm{k}}}+{{\bm{q}}},l,\sigma}c_{{{\bm{k}}},m,\sigma}over^ start_ARG italic_O end_ARG start_POSTSUPERSCRIPT italic_g end_POSTSUPERSCRIPT start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT ≑ βˆ‘ start_POSTSUBSCRIPT bold_italic_k , italic_l , italic_m , italic_Οƒ end_POSTSUBSCRIPT italic_g start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l italic_m end_POSTSUPERSCRIPT ( bold_italic_k ) italic_c start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT start_POSTSUBSCRIPT bold_italic_k + bold_italic_q , italic_l , italic_Οƒ end_POSTSUBSCRIPT italic_c start_POSTSUBSCRIPT bold_italic_k , italic_m , italic_Οƒ end_POSTSUBSCRIPT is the BO operator Tazai-rev2021S ; Kontani-AdvPhysS ; Kontani-RPAS . and v𝑣vitalic_v is the effective interaction. Here, the BO form factor g𝒒l⁒m⁒(π’Œ)superscriptsubscriptπ‘”π’’π‘™π‘šπ’Œg_{{{\bm{q}}}}^{lm}({{\bm{k}}})italic_g start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l italic_m end_POSTSUPERSCRIPT ( bold_italic_k ) is normalized as maxπ’Œ,l,m⁑|g𝒒l⁒m⁒(π’Œ)|=1subscriptπ’Œπ‘™π‘šsuperscriptsubscriptπ‘”π’’π‘™π‘šπ’Œ1\max_{{{\bm{k}}},l,m}|g_{{{\bm{q}}}}^{lm}({{\bm{k}}})|=1roman_max start_POSTSUBSCRIPT bold_italic_k , italic_l , italic_m end_POSTSUBSCRIPT | italic_g start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l italic_m end_POSTSUPERSCRIPT ( bold_italic_k ) | = 1 at each 𝒒𝒒{{\bm{q}}}bold_italic_q, Δ±.e., |δ⁒ti⁒jb|≑1/2𝛿superscriptsubscript𝑑𝑖𝑗b12|\delta t_{ij}^{\rm b}|\equiv 1/2| italic_Ξ΄ italic_t start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_b end_POSTSUPERSCRIPT | ≑ 1 / 2 for the nearest sites. We next calculate the on-site self-energy due to BO fluctuations as Tazai-kagome2S

Ξ£m⁒(Ο΅n)subscriptΞ£π‘šsubscriptitalic-ϡ𝑛\displaystyle\Sigma_{m}({\epsilon}_{n})roman_Ξ£ start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT ( italic_Ο΅ start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT ) =\displaystyle== TN2β’βˆ‘π’Œ,q,mβ€²β€²,mβ€²β€²β€²Gm′⁒m′′⁒(π’Œ+𝒒,Ο΅n+Ο‰l)𝑇superscript𝑁2subscriptπ’Œπ‘žsuperscriptπ‘šβ€²β€²superscriptπ‘šβ€²β€²β€²subscript𝐺superscriptπ‘šβ€²superscriptπ‘šβ€²β€²π’Œπ’’subscriptitalic-ϡ𝑛subscriptπœ”π‘™\displaystyle\frac{T}{N^{2}}\sum_{{{\bm{k}}},q,m^{\prime\prime},m^{\prime% \prime\prime}}G_{m^{\prime}m^{\prime\prime}}({{\bm{k}}}+{{\bm{q}}},{\epsilon}_% {n}+{\omega}_{l})divide start_ARG italic_T end_ARG start_ARG italic_N start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_ARG βˆ‘ start_POSTSUBSCRIPT bold_italic_k , italic_q , italic_m start_POSTSUPERSCRIPT β€² β€² end_POSTSUPERSCRIPT , italic_m start_POSTSUPERSCRIPT β€² β€² β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT italic_G start_POSTSUBSCRIPT italic_m start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_m start_POSTSUPERSCRIPT β€² β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT ( bold_italic_k + bold_italic_q , italic_Ο΅ start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT + italic_Ο‰ start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT ) (S16)
Γ—Bm⁒mβ€²,m⁒m′′⁒(k,q),absentsubscriptπ΅π‘šsuperscriptπ‘šβ€²π‘šsuperscriptπ‘šβ€²β€²π‘˜π‘ž\displaystyle\times B_{mm^{\prime},mm^{\prime\prime}}(k,q),Γ— italic_B start_POSTSUBSCRIPT italic_m italic_m start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT , italic_m italic_m start_POSTSUPERSCRIPT β€² β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT ( italic_k , italic_q ) ,
Bm⁒l,m′⁒l′⁒(k,q)subscriptπ΅π‘šπ‘™superscriptπ‘šβ€²superscriptπ‘™β€²π‘˜π‘ž\displaystyle B_{ml,m^{\prime}l^{\prime}}(k,q)italic_B start_POSTSUBSCRIPT italic_m italic_l , italic_m start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT ( italic_k , italic_q ) =\displaystyle== g𝒒l⁒m⁒(π’Œ)βˆ—β’g𝒒l′⁒m′⁒(π’Œ)β‹…v⁒(1+v⁒χg⁒(q)),β‹…superscriptsubscriptπ‘”π’’π‘™π‘šsuperscriptπ’Œsuperscriptsubscript𝑔𝒒superscript𝑙′superscriptπ‘šβ€²π’Œπ‘£1𝑣subscriptπœ’π‘”π‘ž\displaystyle g_{{\bm{q}}}^{lm}({{\bm{k}}})^{*}g_{{{\bm{q}}}}^{l^{\prime}m^{% \prime}}({{\bm{k}}})\cdot v(1+v\chi_{g}(q)),italic_g start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l italic_m end_POSTSUPERSCRIPT ( bold_italic_k ) start_POSTSUPERSCRIPT βˆ— end_POSTSUPERSCRIPT italic_g start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_m start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT ( bold_italic_k ) β‹… italic_v ( 1 + italic_v italic_Ο‡ start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT ( italic_q ) ) , (S17)

which is shown in Fig. S7 (d). Here, the BO susceptibility Ο‡g⁒(q)subscriptπœ’π‘”π‘ž\chi_{g}(q)italic_Ο‡ start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT ( italic_q ) is Tazai-kagome2S

Ο‡g⁒(q)subscriptπœ’π‘”π‘ž\displaystyle\chi_{g}(q)italic_Ο‡ start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT ( italic_q ) =\displaystyle== Ο‡g0⁒(q)/(1βˆ’v⁒χg0⁒(q)),,subscriptsuperscriptπœ’0π‘”π‘ž1𝑣subscriptsuperscriptπœ’0π‘”π‘ž\displaystyle\chi^{0}_{g}(q)/(1-v\chi^{0}_{g}(q)),,italic_Ο‡ start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT ( italic_q ) / ( 1 - italic_v italic_Ο‡ start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT ( italic_q ) ) , , (S18)
Ο‡g0⁒(q)subscriptsuperscriptπœ’0π‘”π‘ž\displaystyle\chi^{0}_{g}(q)italic_Ο‡ start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT ( italic_q ) =\displaystyle== βˆ‘l⁒m⁒m′⁒lβ€²Ο‡g0,l⁒m⁒m′⁒l′⁒(q),subscriptπ‘™π‘šsuperscriptπ‘šβ€²superscript𝑙′subscriptsuperscriptπœ’0π‘™π‘šsuperscriptπ‘šβ€²superscriptπ‘™β€²π‘”π‘ž\displaystyle\sum_{lmm^{\prime}l^{\prime}}\chi^{0,lmm^{\prime}l^{\prime}}_{g}(% q),βˆ‘ start_POSTSUBSCRIPT italic_l italic_m italic_m start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT italic_Ο‡ start_POSTSUPERSCRIPT 0 , italic_l italic_m italic_m start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT ( italic_q ) , (S19)
Ο‡g0,l⁒m⁒m′⁒l′⁒(𝒒,Ο‰l)subscriptsuperscriptπœ’0π‘™π‘šsuperscriptπ‘šβ€²superscript𝑙′𝑔𝒒subscriptπœ”π‘™\displaystyle\chi^{0,lmm^{\prime}l^{\prime}}_{g}({{\bm{q}}},\omega_{l})italic_Ο‡ start_POSTSUPERSCRIPT 0 , italic_l italic_m italic_m start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT ( bold_italic_q , italic_Ο‰ start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT ) =\displaystyle== TNβ’βˆ‘π’Œ,Ο΅ng𝒒l⁒m⁒(π’Œ)βˆ—β’Gl⁒m′⁒(π’Œ+𝒒,Ο΅n+Ο‰l)𝑇𝑁subscriptπ’Œsubscriptitalic-ϡ𝑛subscriptsuperscriptπ‘”π‘™π‘šπ’’superscriptπ’Œsubscript𝐺𝑙superscriptπ‘šβ€²π’Œπ’’subscriptitalic-ϡ𝑛subscriptπœ”π‘™\displaystyle\frac{T}{N}\sum_{{{\bm{k}}},{\epsilon}_{n}}g^{lm}_{{{\bm{q}}}}({{% \bm{k}}})^{*}G_{lm^{\prime}}({{\bm{k}}}+{{\bm{q}}},{\epsilon}_{n}+\omega_{l})divide start_ARG italic_T end_ARG start_ARG italic_N end_ARG βˆ‘ start_POSTSUBSCRIPT bold_italic_k , italic_Ο΅ start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT end_POSTSUBSCRIPT italic_g start_POSTSUPERSCRIPT italic_l italic_m end_POSTSUPERSCRIPT start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT ( bold_italic_k ) start_POSTSUPERSCRIPT βˆ— end_POSTSUPERSCRIPT italic_G start_POSTSUBSCRIPT italic_l italic_m start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT ( bold_italic_k + bold_italic_q , italic_Ο΅ start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT + italic_Ο‰ start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT ) (S20)
Γ—Gl′⁒m⁒(π’Œ,Ο΅n)⁒g𝒒m′⁒l′⁒(π’Œ),absentsubscript𝐺superscriptπ‘™β€²π‘šπ’Œsubscriptitalic-ϡ𝑛subscriptsuperscript𝑔superscriptπ‘šβ€²superscriptπ‘™β€²π’’π’Œ\displaystyle\times G_{l^{\prime}m}({{\bm{k}}},{\epsilon}_{n})g^{m^{\prime}l^{% \prime}}_{{{\bm{q}}}}({{\bm{k}}}),Γ— italic_G start_POSTSUBSCRIPT italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_m end_POSTSUBSCRIPT ( bold_italic_k , italic_Ο΅ start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT ) italic_g start_POSTSUPERSCRIPT italic_m start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT italic_l start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT ( bold_italic_k ) ,

which is shown in Fig. S7 (e). Then, the Green function is given as G^⁒(k)=(i⁒ϡn+ΞΌβˆ’h^⁒(π’Œ)βˆ’Ξ£^⁒(Ο΅n))βˆ’1^πΊπ‘˜superscript𝑖subscriptitalic-Ο΅π‘›πœ‡^β„Žπ’Œ^Ξ£subscriptitalic-ϡ𝑛1{\hat{G}}(k)=(i{\epsilon}_{n}+\mu-{\hat{h}}({{\bm{k}}})-{\hat{\Sigma}}({% \epsilon}_{n}))^{-1}over^ start_ARG italic_G end_ARG ( italic_k ) = ( italic_i italic_Ο΅ start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT + italic_ΞΌ - over^ start_ARG italic_h end_ARG ( bold_italic_k ) - over^ start_ARG roman_Ξ£ end_ARG ( italic_Ο΅ start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT ) ) start_POSTSUPERSCRIPT - 1 end_POSTSUPERSCRIPT. The effect of thermal fluctuations described by the self-energy is essential to reproduce the T𝑇Titalic_T-dependence of various physical quantities. In the present numerical study, we calculate Eqs. [S16]-[S20] self-consistently.

We also study the 2nd order GL term, which is derived as F2bβˆΌβˆ’Ο‡g0⁒(𝒒1)⁒R⁒|Ο•|2similar-tosuperscriptsubscript𝐹2bsuperscriptsubscriptπœ’π‘”0subscript𝒒1𝑅superscriptbold-italic-Ο•2F_{2}^{\rm b}\sim-\chi_{g}^{0}({{\bm{q}}}_{1})R|{\bm{\phi}}|^{2}italic_F start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_b end_POSTSUPERSCRIPT ∼ - italic_Ο‡ start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( bold_italic_q start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ) italic_R | bold_italic_Ο• | start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT according to Ref. Tazai-LWS . Here, Ο‡g0⁒(𝒒1)superscriptsubscriptπœ’π‘”0subscript𝒒1\chi_{g}^{0}({{\bm{q}}}_{1})italic_Ο‡ start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( bold_italic_q start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ) is the irreducible BO susceptibility, and R≑(d⁒λ/d⁒T)β‹…(βˆ’Tb0)π‘…β‹…π‘‘πœ†π‘‘π‘‡superscriptsubscript𝑇b0R\equiv(d\lambda/dT)\cdot(-T_{\rm b}^{0})italic_R ≑ ( italic_d italic_Ξ» / italic_d italic_T ) β‹… ( - italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ), where Ξ»πœ†\lambdaitalic_Ξ» is the eigenvalue of the density-wave equation, which is similar to the eigenvalue of the BCS gap equation. R∼O⁒(0.1)similar-to𝑅𝑂0.1R\sim O(0.1)italic_R ∼ italic_O ( 0.1 ) in usual BO phase transitions Tazai-LWS , while R∼1similar-to𝑅1R\sim 1italic_R ∼ 1 in BCS superconductivity because of large log⁒Tlog𝑇{\rm log}Troman_log italic_T singularity of the p-p channel. As a result, F2bβˆΌβˆ’3Γ—10βˆ’5⁒[eV]similar-tosuperscriptsubscript𝐹2b3superscript105delimited-[]eVF_{2}^{\rm b}\sim-3\times 10^{-5}\ [{\rm eV}]italic_F start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_b end_POSTSUPERSCRIPT ∼ - 3 Γ— 10 start_POSTSUPERSCRIPT - 5 end_POSTSUPERSCRIPT [ roman_eV ] for Ο‡g0⁒(𝒒1)∼3similar-tosuperscriptsubscriptπœ’π‘”0subscript𝒒13\chi_{g}^{0}({{\bm{q}}}_{1})\sim 3italic_Ο‡ start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( bold_italic_q start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ) ∼ 3 and R=0.1𝑅0.1R=0.1italic_R = 0.1. Note that Ο‡g0⁒(𝒒1)⁒R∼0.1similar-tosuperscriptsubscriptπœ’π‘”0subscript𝒒1𝑅0.1\chi_{g}^{0}({{\bm{q}}}_{1})R\sim 0.1italic_Ο‡ start_POSTSUBSCRIPT italic_g end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( bold_italic_q start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ) italic_R ∼ 0.1 corresponds to r⁒Tb0∼0.1similar-toπ‘Ÿsuperscriptsubscript𝑇b00.1rT_{\rm b}^{0}\sim 0.1italic_r italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ∼ 0.1 in the main text.

Therefore, the BO total free energy is FtotbβˆΌβˆ’10βˆ’5⁒[eV]similar-tosuperscriptsubscript𝐹totbsuperscript105delimited-[]eVF_{\rm tot}^{\rm b}\sim-10^{-5}\ [{\rm eV}]italic_F start_POSTSUBSCRIPT roman_tot end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_b end_POSTSUPERSCRIPT ∼ - 10 start_POSTSUPERSCRIPT - 5 end_POSTSUPERSCRIPT [ roman_eV ] for Tb0βˆΌΟ•βˆΌ0.01similar-tosuperscriptsubscript𝑇b0italic-Ο•similar-to0.01T_{\rm b}^{0}\sim\phi\sim 0.01italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ∼ italic_Ο• ∼ 0.01 and N⁒(0)∼1similar-to𝑁01N(0)\sim 1italic_N ( 0 ) ∼ 1. The current-order total free energy Ftotcsuperscriptsubscript𝐹totcF_{\rm tot}^{\rm c}italic_F start_POSTSUBSCRIPT roman_tot end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_c end_POSTSUPERSCRIPT will be comparable to Ftotbsuperscriptsubscript𝐹totbF_{\rm tot}^{\rm b}italic_F start_POSTSUBSCRIPT roman_tot end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_b end_POSTSUPERSCRIPT. Thus, the GL parameters assumed in the main text are qualitatively reasonable. (In BCS superconductors, Ftot=βˆ’Ξ”2⁒N⁒(0)/2βˆΌβˆ’10βˆ’4subscript𝐹totsuperscriptΞ”2𝑁02similar-tosuperscript104F_{\rm tot}=-\Delta^{2}N(0)/2\sim-10^{-4}italic_F start_POSTSUBSCRIPT roman_tot end_POSTSUBSCRIPT = - roman_Ξ” start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_N ( 0 ) / 2 ∼ - 10 start_POSTSUPERSCRIPT - 4 end_POSTSUPERSCRIPT when Ξ”=0.01Ξ”0.01\Delta=0.01roman_Ξ” = 0.01 and N⁒(0)∼1similar-to𝑁01N(0)\sim 1italic_N ( 0 ) ∼ 1.) As discussed in Ref. Tazai-LWS , the specific heat jump Δ⁒C/TΔ𝐢𝑇\Delta C/Troman_Ξ” italic_C / italic_T at the BO or current-order phase transition is much smaller than the BCS value Δ⁒C/TSC=1.43⁒N⁒(0)Δ𝐢subscript𝑇SC1.43𝑁0\Delta C/T_{\rm SC}=1.43N(0)roman_Ξ” italic_C / italic_T start_POSTSUBSCRIPT roman_SC end_POSTSUBSCRIPT = 1.43 italic_N ( 0 ).

Finally, we discuss the renormalization of the 2nd order GL coefficient for Ξ·πœ‚\etaitalic_Ξ·, acsubscriptπ‘Žca_{\rm c}italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT, in the BO phase. Under the 3⁒Q3𝑄3Q3 italic_Q BO phase Ο•=(Ο•,Ο•,Ο•)/3bold-italic-Ο•italic-Ο•italic-Ο•italic-Ο•3{\bm{\phi}}=(\phi,\phi,\phi)/\sqrt{3}bold_italic_Ο• = ( italic_Ο• , italic_Ο• , italic_Ο• ) / square-root start_ARG 3 end_ARG, acsubscriptπ‘Žca_{\rm c}italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT in Eq. [11] is renormalized as aΒ―c=ac+(d5+d6)⁒(2⁒ϕ2/3)subscriptΒ―π‘Žcsubscriptπ‘Žcsubscript𝑑5subscript𝑑62superscriptitalic-Ο•23{\bar{a}}_{\rm c}=a_{\rm c}+(d_{5}+d_{6})(2\phi^{2}/3)overΒ― start_ARG italic_a end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT = italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT + ( italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT + italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ) ( 2 italic_Ο• start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT / 3 ) due to the d5subscript𝑑5d_{5}italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT, d6subscript𝑑6d_{6}italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT terms. When 𝜼=𝟎𝜼0{\bm{\eta}}={\bm{0}}bold_italic_Ξ· = bold_0, we obtain Ο•2β‰ˆβˆ’3⁒ab/2⁒(d1+d2)superscriptitalic-Ο•23subscriptπ‘Žπ‘2subscript𝑑1subscript𝑑2\phi^{2}\approx-3a_{b}/2(d_{1}+d_{2})italic_Ο• start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT β‰ˆ - 3 italic_a start_POSTSUBSCRIPT italic_b end_POSTSUBSCRIPT / 2 ( italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT + italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) by neglecting b1subscript𝑏1b_{1}italic_b start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT term, which is allowed except for Tβ‰ˆTb0𝑇superscriptsubscript𝑇b0T\approx T_{\rm b}^{0}italic_T β‰ˆ italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT. Therefore, we obtain aΒ―c=rΒ―c⁒(Tβˆ’TΒ―c0)subscriptΒ―π‘ŽcsubscriptΒ―π‘Ÿc𝑇superscriptsubscript¯𝑇c0{\bar{a}}_{\rm c}={\bar{r}}_{\rm c}(T-{\bar{T}}_{\rm c}^{0})overΒ― start_ARG italic_a end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT = overΒ― start_ARG italic_r end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( italic_T - overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ), where rΒ―c=rc⁒(1βˆ’C)subscriptΒ―π‘Ÿcsubscriptπ‘Ÿc1𝐢{\bar{r}}_{\rm c}=r_{\rm c}(1-C)overΒ― start_ARG italic_r end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT = italic_r start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( 1 - italic_C ), TΒ―c0=(Tc0βˆ’C⁒Tb0)/(1βˆ’C)superscriptsubscript¯𝑇c0superscriptsubscript𝑇c0𝐢superscriptsubscript𝑇b01𝐢{\bar{T}}_{\rm c}^{0}=(T_{\rm c}^{0}-CT_{\rm b}^{0})/(1-C)overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = ( italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT - italic_C italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ) / ( 1 - italic_C ), and C≑(rb/rc)⁒[(d5+d6)/(d1+d2)]𝐢subscriptπ‘Ÿbsubscriptπ‘Ÿcdelimited-[]subscript𝑑5subscript𝑑6subscript𝑑1subscript𝑑2C\equiv(r_{\rm b}/r_{\rm c})[(d_{5}+d_{6})/(d_{1}+d_{2})]italic_C ≑ ( italic_r start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT / italic_r start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ) [ ( italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT + italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ) / ( italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT + italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) ]. Here, we assume C<1𝐢1C<1italic_C < 1 by referring to the relation d5+d6<d1+d2subscript𝑑5subscript𝑑6subscript𝑑1subscript𝑑2d_{5}+d_{6}<d_{1}+d_{2}italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT + italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT < italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT + italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT in Fig. S7, owing to the difference between the BO and current order form factors. For detail, see Ref. Tazai-kagome2 .

G: Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT by intra-original-unit-cell current order in kagome lattice

Here, we calculate Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT in the case of the intra-original-unit-cell (𝒒=πŸŽπ’’0{{\bm{q}}}={\bm{0}}bold_italic_q = bold_0) current order in Fig. S8 (a). In this case, the translational symmetry is preserved. The obtained Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT is shown in Fig. S8 (b). We find that Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT is Ξ·πœ‚\etaitalic_Ξ·-linear even in the absence of the BO, while its coefficient is small for n∼nvHSsimilar-to𝑛subscript𝑛vHSn\sim n_{\rm vHS}italic_n ∼ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT that is realized in kagome metals. Regardless of the presence of the Ξ·πœ‚\etaitalic_Ξ·-linear term in Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT, field-induced intra-original-unit-cell order will be quite small in kagome metals. In fact, the field-induced cLC order at 𝒒=πŸŽπ’’0{{\bm{q}}}={\bm{0}}bold_italic_q = bold_0 becomes sizable only when its second-order GL coefficient, ac∼(βˆ’1+λ𝒒=πŸŽβˆ’1)similar-tosubscriptπ‘Žc1superscriptsubscriptπœ†π’’01a_{\rm c}\sim(-1+\lambda_{{{\bm{q}}}={\bm{0}}}^{-1})italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ∼ ( - 1 + italic_Ξ» start_POSTSUBSCRIPT bold_italic_q = bold_0 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT - 1 end_POSTSUPERSCRIPT ), is very small. Here, λ𝒒=𝟎subscriptπœ†π’’0\lambda_{{{\bm{q}}}={\bm{0}}}italic_Ξ» start_POSTSUBSCRIPT bold_italic_q = bold_0 end_POSTSUBSCRIPT is the eigenvalue of the current order solution at 𝒒=πŸŽπ’’0{{\bm{q}}}={\bm{0}}bold_italic_q = bold_0. However, the relation λ𝒒=𝟎β‰ͺ1much-less-thansubscriptπœ†π’’01\lambda_{{{\bm{q}}}={\bm{0}}}\ll 1italic_Ξ» start_POSTSUBSCRIPT bold_italic_q = bold_0 end_POSTSUBSCRIPT β‰ͺ 1 is obtained in the DW equation analysis in Ref. Tazai-kagome2S .

Refer to caption
Figure S8: (a) Intra-original-unit-cell (𝒒=πŸŽπ’’0{{\bm{q}}}={\bm{0}}bold_italic_q = bold_0) current order. (b) Morbsubscript𝑀orbM_{\rm orb}italic_M start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT for the intra-original-unit-cell current order 𝜼=(Ξ·,Ξ·,Ξ·)/3πœΌπœ‚πœ‚πœ‚3{\bm{\eta}}=(\eta,\eta,\eta)/\sqrt{3}bold_italic_Ξ· = ( italic_Ξ· , italic_Ξ· , italic_Ξ· ) / square-root start_ARG 3 end_ARG. We set (t,tβ€²)=(βˆ’0.5,βˆ’0.02)𝑑superscript𝑑′0.50.02(t,t^{\prime})=(-0.5,-0.02)( italic_t , italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ) = ( - 0.5 , - 0.02 ), and no BO is introduced.

H: First principles 30303030-orbital model for kagome metal

In the main text, we analyzed the GL coefficients based on the first-principles 30-orbital model for kagome metals. Figure S9 (a) shows the obtained bandstructure in the kz=0subscriptπ‘˜π‘§0k_{z}=0italic_k start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 0 plane. Its FS is shown in Fig. 5 (a). The dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital β€œpure-type” band corresponds to the present three-orbital model. Its vHS energy is located at EvHSX⁒Zβ‰ˆβˆ’0.1superscriptsubscript𝐸vHS𝑋𝑍0.1E_{\rm vHS}^{XZ}\approx-0.1italic_E start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_X italic_Z end_POSTSUPERSCRIPT β‰ˆ - 0.1. Also, the dY⁒Zsubscriptπ‘‘π‘Œπ‘d_{YZ}italic_d start_POSTSUBSCRIPT italic_Y italic_Z end_POSTSUBSCRIPT-orbital forms the β€œmix-type” band, whose vHS energy is EvHSY⁒Zβ‰ˆ+0.1superscriptsubscript𝐸vHSπ‘Œπ‘0.1E_{\rm vHS}^{YZ}\approx+0.1italic_E start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_Y italic_Z end_POSTSUPERSCRIPT β‰ˆ + 0.1. In addition, the dX2βˆ’Y2subscript𝑑superscript𝑋2superscriptπ‘Œ2d_{X^{2}-Y^{2}}italic_d start_POSTSUBSCRIPT italic_X start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Y start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT+d3⁒Z2βˆ’R2subscript𝑑3superscript𝑍2superscript𝑅2d_{3Z^{2}-R^{2}}italic_d start_POSTSUBSCRIPT 3 italic_Z start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_R start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT-orbital forms a pure-type band with the vHS energy EvHSβ€²β‰ˆβˆ’0.05superscriptsubscript𝐸vHSβ€²0.05E_{\rm vHS}^{\prime}\approx-0.05italic_E start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT β‰ˆ - 0.05. Around M point, the dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital band near M point is almost kzsubscriptπ‘˜π‘§k_{z}italic_k start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-independent, while other orbital bands exhibits small kzsubscriptπ‘˜π‘§k_{z}italic_k start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT-independences (about 0.1∼0.2similar-to0.10.20.1\sim 0.20.1 ∼ 0.2) in the band calculation.

Figure S9 (b) shows the bandstructure with introducing the p𝑝pitalic_p-orbital shift Δ⁒Ep=βˆ’0.2Ξ”subscript𝐸𝑝0.2\Delta E_{p}=-0.2roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2 eV. Its FS is shown in Fig. 5 (b). Here, the dY⁒Zsubscriptπ‘‘π‘Œπ‘d_{YZ}italic_d start_POSTSUBSCRIPT italic_Y italic_Z end_POSTSUBSCRIPT-orbital (mix-type) band approaches EFsubscript𝐸FE_{\rm F}italic_E start_POSTSUBSCRIPT roman_F end_POSTSUBSCRIPT along the M-M’ line, consistently with the APRES measurement in Ref. ARPES-VHSS .

Refer to caption
Figure S9: (a) Bandstructure of CsV3Sb5 model in the kz=0subscriptπ‘˜π‘§0k_{z}=0italic_k start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 0 plane. Its FS is shown in Fig. 5 (a). The dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital band gives the pure-type FS, and the dY⁒Zsubscriptπ‘‘π‘Œπ‘d_{YZ}italic_d start_POSTSUBSCRIPT italic_Y italic_Z end_POSTSUBSCRIPT-orbital band gives the mix-type FS. The dX2βˆ’Y2subscript𝑑superscript𝑋2superscriptπ‘Œ2d_{X^{2}-Y^{2}}italic_d start_POSTSUBSCRIPT italic_X start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Y start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT+d3⁒Z2βˆ’R2subscript𝑑3superscript𝑍2superscript𝑅2d_{3Z^{2}-R^{2}}italic_d start_POSTSUBSCRIPT 3 italic_Z start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_R start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT-orbital band gives another pure-type FS. ΓΓ\Gammaroman_Ξ“, M, M’ are introduced in Fig. 1 (b) in the main text. (b) Bandstructure with the p𝑝pitalic_p-orbital shift Δ⁒Ep=βˆ’0.2Ξ”subscript𝐸𝑝0.2\Delta E_{p}=-0.2roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2 eV. Its FS is shown in Fig. 5 (a). (c)(d) Obtained m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT for the order parameters in the dY⁒Zsubscriptπ‘‘π‘Œπ‘d_{YZ}italic_d start_POSTSUBSCRIPT italic_Y italic_Z end_POSTSUBSCRIPT, dX2βˆ’Y2subscript𝑑superscript𝑋2superscriptπ‘Œ2d_{X^{2}-Y^{2}}italic_d start_POSTSUBSCRIPT italic_X start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Y start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT, and d3⁒Z2βˆ’R2subscript𝑑3superscript𝑍2superscript𝑅2d_{3Z^{2}-R^{2}}italic_d start_POSTSUBSCRIPT 3 italic_Z start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_R start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT orbitals for (c) Δ⁒Ep=0Ξ”subscript𝐸𝑝0\Delta E_{p}=0roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = 0 model and (d) Δ⁒Ep=βˆ’0.2Ξ”subscript𝐸𝑝0.2\Delta E_{p}=-0.2roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2 model. (e)(f) Obtained m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT for the order parameters in the dY⁒Zsubscriptπ‘‘π‘Œπ‘d_{YZ}italic_d start_POSTSUBSCRIPT italic_Y italic_Z end_POSTSUBSCRIPT, dX2βˆ’Y2subscript𝑑superscript𝑋2superscriptπ‘Œ2d_{X^{2}-Y^{2}}italic_d start_POSTSUBSCRIPT italic_X start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Y start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT, and d3⁒Z2βˆ’R2subscript𝑑3superscript𝑍2superscript𝑅2d_{3Z^{2}-R^{2}}italic_d start_POSTSUBSCRIPT 3 italic_Z start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_R start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT orbitals for (e) Δ⁒Ep=0Ξ”subscript𝐸𝑝0\Delta E_{p}=0roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = 0 model and (f) Δ⁒Ep=βˆ’0.2Ξ”subscript𝐸𝑝0.2\Delta E_{p}=-0.2roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2 model. In (c)-(f), the vHS filling nvHSΞ“superscriptsubscript𝑛vHSΞ“n_{\rm vHS}^{\Gamma}italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT start_POSTSUPERSCRIPT roman_Ξ“ end_POSTSUPERSCRIPT (Ξ“=X⁒Z,Y⁒Z,X2βˆ’Y2Ξ“π‘‹π‘π‘Œπ‘superscript𝑋2superscriptπ‘Œ2\Gamma=XZ,YZ,X^{2}-Y^{2}roman_Ξ“ = italic_X italic_Z , italic_Y italic_Z , italic_X start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Y start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT) is shown. (Note that nvHSX2βˆ’Y2=nvHS3⁒Z2βˆ’R2superscriptsubscript𝑛vHSsuperscript𝑋2superscriptπ‘Œ2superscriptsubscript𝑛vHS3superscript𝑍2superscript𝑅2n_{\rm vHS}^{X^{2}-Y^{2}}=n_{\rm vHS}^{3Z^{2}-R^{2}}italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_X start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Y start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT = italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 3 italic_Z start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_R start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT.)

Here, we derive the coefficients m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT defined as MΒ―orb=m1⁒ϕ~β‹…πœΌ~+m2⁒η~1⁒η~2⁒η~3subscript¯𝑀orbβ‹…subscriptπ‘š1~bold-italic-Ο•~𝜼subscriptπ‘š2subscript~πœ‚1subscript~πœ‚2subscript~πœ‚3{\bar{M}}_{\rm orb}=m_{1}{\tilde{\bm{\phi}}}\cdot{\tilde{\bm{\eta}}}+m_{2}{% \tilde{\eta}_{1}}{\tilde{\eta}_{2}}{\tilde{\eta}_{3}}overΒ― start_ARG italic_M end_ARG start_POSTSUBSCRIPT roman_orb end_POSTSUBSCRIPT = italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT over~ start_ARG bold_italic_Ο• end_ARG β‹… over~ start_ARG bold_italic_Ξ· end_ARG + italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT over~ start_ARG italic_Ξ· end_ARG start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT over~ start_ARG italic_Ξ· end_ARG start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT over~ start_ARG italic_Ξ· end_ARG start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT, where Ξ·i~~subscriptπœ‚π‘–\tilde{\eta_{i}}over~ start_ARG italic_Ξ· start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT end_ARG and Ο•i~~subscriptitalic-ϕ𝑖\tilde{\phi_{i}}over~ start_ARG italic_Ο• start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT end_ARG are the order parameter of a specific d𝑑ditalic_d-orbital projected on the conduction band. The results for dX⁒Zsubscript𝑑𝑋𝑍d_{XZ}italic_d start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT-orbital are shown in Figs. 5 (e)-(h) in the main text. (The weight of dΞ“subscript𝑑Γd_{\Gamma}italic_d start_POSTSUBSCRIPT roman_Ξ“ end_POSTSUBSCRIPT-orbital (Ξ“=X⁒Z,Y⁒ZΞ“π‘‹π‘π‘Œπ‘\Gamma=XZ,YZroman_Ξ“ = italic_X italic_Z , italic_Y italic_Z, etc) on the conduction band is WX⁒Zβ‰ˆ0.7subscriptπ‘Šπ‘‹π‘0.7W_{XZ}\approx 0.7italic_W start_POSTSUBSCRIPT italic_X italic_Z end_POSTSUBSCRIPT β‰ˆ 0.7, WY⁒Zβ‰ˆ0.3subscriptπ‘Šπ‘Œπ‘0.3W_{YZ}\approx 0.3italic_W start_POSTSUBSCRIPT italic_Y italic_Z end_POSTSUBSCRIPT β‰ˆ 0.3, WX2βˆ’Y2β‰ˆW3⁒Z2βˆ’R2β‰ˆ0.4subscriptπ‘Šsuperscript𝑋2superscriptπ‘Œ2subscriptπ‘Š3superscript𝑍2superscript𝑅20.4W_{X^{2}-Y^{2}}\approx W_{3Z^{2}-R^{2}}\approx 0.4italic_W start_POSTSUBSCRIPT italic_X start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Y start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT β‰ˆ italic_W start_POSTSUBSCRIPT 3 italic_Z start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_R start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT β‰ˆ 0.4.) We also calculate m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT when the order parameters emerge on the dY⁒Zsubscriptπ‘‘π‘Œπ‘d_{YZ}italic_d start_POSTSUBSCRIPT italic_Y italic_Z end_POSTSUBSCRIPT, dX2βˆ’Y2subscript𝑑superscript𝑋2superscriptπ‘Œ2d_{X^{2}-Y^{2}}italic_d start_POSTSUBSCRIPT italic_X start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Y start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT, and d3⁒Z2βˆ’R2subscript𝑑3superscript𝑍2superscript𝑅2d_{3Z^{2}-R^{2}}italic_d start_POSTSUBSCRIPT 3 italic_Z start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_R start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_POSTSUBSCRIPT orbitals. The obtained results are shown in Figs. S9 (a) (Δ⁒Ep=0Ξ”subscript𝐸𝑝0\Delta E_{p}=0roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = 0) and (b) (Δ⁒Ep=βˆ’0.2Ξ”subscript𝐸𝑝0.2\Delta E_{p}=-0.2roman_Ξ” italic_E start_POSTSUBSCRIPT italic_p end_POSTSUBSCRIPT = - 0.2), as function of the electron filling n𝑛nitalic_n. (n=31𝑛31n=31italic_n = 31 corresponds to undoped CsV3Sb5.) It is found that large m1subscriptπ‘š1m_{1}italic_m start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and m2subscriptπ‘š2m_{2}italic_m start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT are obtained when the current and bond orders emerge on various d𝑑ditalic_d-orbitals. Thus, the present study is valid for various types of current order mechanisms, not restricted to Ref. Tazai-kagome2S .

I: Increment of Tcsubscript𝑇cT_{\rm c}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT under uniaxial strain in 3⁒Q3𝑄3Q3 italic_Q BO phase

In Ref. Moll-hzS , it was proposed that CsV3Sb5 is located at the quantum critical point of the current order (Tc0β‰ˆ0superscriptsubscript𝑇c00T_{\rm c}^{0}\approx 0italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT β‰ˆ 0 in the absence of the uniaxial strain. The field-induced (hz∼9similar-tosubscriptβ„Žπ‘§9h_{z}\sim 9italic_h start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT ∼ 9T) current order at Tc∼20similar-tosubscript𝑇c20T_{\rm c}\sim 20italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ∼ 20K would be explained by the present theory. In addition, Ref. Moll-hzS reports the strain induced increment of Tcsubscript𝑇cT_{\rm c}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT in the absence of the magnetic field. It was explained in Ref. Moll-hzS based on the GL free energy analysis. Under the uniaxial strain, the degeneracy of the current order transition temperature at 𝒒=𝒒m𝒒subscriptπ’’π‘š{{\bm{q}}}={{\bm{q}}}_{m}bold_italic_q = bold_italic_q start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT (m=1∼3π‘š1similar-to3m=1\sim 3italic_m = 1 ∼ 3), Tc(m)superscriptsubscript𝑇cπ‘šT_{\rm c}^{(m)}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT, is lifted. Then, Tc=maxm⁑Tc(m)subscript𝑇csubscriptπ‘šsuperscriptsubscript𝑇cπ‘šT_{\rm c}=\max_{m}T_{\rm c}^{(m)}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT = roman_max start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT will be larger than the original Tc0superscriptsubscript𝑇c0T_{\rm c}^{0}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT.

Here, we present that the strain induced change in the 4th order GL terms causes additional significant contribution to the increment of Tcsubscript𝑇cT_{\rm c}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT. That is, the BO-induced suppression of the current order is drastically reduced by the strain. Considering the D6⁒hsubscript𝐷6β„ŽD_{6h}italic_D start_POSTSUBSCRIPT 6 italic_h end_POSTSUBSCRIPT symmetry, we assume that the E1⁒gsubscript𝐸1𝑔E_{1g}italic_E start_POSTSUBSCRIPT 1 italic_g end_POSTSUBSCRIPT symmetry strain (Ο΅,Ο΅β€²)italic-Ο΅superscriptitalic-Ο΅β€²({\epsilon},{\epsilon}^{\prime})( italic_Ο΅ , italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ) induces the shift of the vHS energy levels as Δ⁒𝑬≑(Δ⁒EA,Δ⁒EB,Δ⁒EC)=α⁒(Ο΅,0,βˆ’Ο΅)Δ𝑬Δsubscript𝐸𝐴Δsubscript𝐸𝐡Δsubscript𝐸𝐢𝛼italic-Ο΅0italic-Ο΅\Delta{\bm{E}}\equiv(\Delta E_{A},\Delta E_{B},\Delta E_{C})={\alpha}({% \epsilon},0,-{\epsilon})roman_Ξ” bold_italic_E ≑ ( roman_Ξ” italic_E start_POSTSUBSCRIPT italic_A end_POSTSUBSCRIPT , roman_Ξ” italic_E start_POSTSUBSCRIPT italic_B end_POSTSUBSCRIPT , roman_Ξ” italic_E start_POSTSUBSCRIPT italic_C end_POSTSUBSCRIPT ) = italic_Ξ± ( italic_Ο΅ , 0 , - italic_Ο΅ ) and Δ⁒𝑬′≑(Δ⁒EAβ€²,Δ⁒EBβ€²,Δ⁒ECβ€²)=(Ξ±β€²/2)⁒(Ο΅β€²,βˆ’2⁒ϡ′,Ο΅β€²)Ξ”superscript𝑬′Δsuperscriptsubscript𝐸𝐴′Δsuperscriptsubscript𝐸𝐡′Δsuperscriptsubscript𝐸𝐢′superscript𝛼′2superscriptitalic-Ο΅β€²2superscriptitalic-Ο΅β€²superscriptitalic-Ο΅β€²\Delta{\bm{E}}^{\prime}\equiv(\Delta E_{A}^{\prime},\Delta E_{B}^{\prime},% \Delta E_{C}^{\prime})=({\alpha}^{\prime}/2)({\epsilon}^{\prime},-2{\epsilon}^% {\prime},{\epsilon}^{\prime})roman_Ξ” bold_italic_E start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ≑ ( roman_Ξ” italic_E start_POSTSUBSCRIPT italic_A end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT , roman_Ξ” italic_E start_POSTSUBSCRIPT italic_B end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT , roman_Ξ” italic_E start_POSTSUBSCRIPT italic_C end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ) = ( italic_Ξ± start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT / 2 ) ( italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT , - 2 italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT , italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ). Then, the 4th order GL terms are given as

βˆ‘m1,2,3{d1(m)⁒ϕm4+d3(m)⁒ηm4+2⁒d5(m)⁒ϕm2⁒ηm2}superscriptsubscriptπ‘š123superscriptsubscript𝑑1π‘šsuperscriptsubscriptitalic-Ο•π‘š4superscriptsubscript𝑑3π‘šsuperscriptsubscriptπœ‚π‘š42superscriptsubscript𝑑5π‘šsuperscriptsubscriptitalic-Ο•π‘š2superscriptsubscriptπœ‚π‘š2\displaystyle\sum_{m}^{1,2,3}\{d_{1}^{(m)}\phi_{m}^{4}+d_{3}^{(m)}\eta_{m}^{4}% +2d_{5}^{(m)}\phi_{m}^{2}\eta_{m}^{2}\}βˆ‘ start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 1 , 2 , 3 end_POSTSUPERSCRIPT { italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT + italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT + 2 italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT }
+βˆ‘m1,2,3{d2(o)⁒ϕm2⁒ϕn2+d4(o)⁒ηm2⁒ηn2+d6(o)⁒(Ο•m2⁒ηn2+Ο•n2⁒ηm2)},superscriptsubscriptπ‘š123superscriptsubscript𝑑2π‘œsuperscriptsubscriptitalic-Ο•π‘š2superscriptsubscriptitalic-ϕ𝑛2superscriptsubscript𝑑4π‘œsuperscriptsubscriptπœ‚π‘š2superscriptsubscriptπœ‚π‘›2superscriptsubscript𝑑6π‘œsuperscriptsubscriptitalic-Ο•π‘š2superscriptsubscriptπœ‚π‘›2superscriptsubscriptitalic-ϕ𝑛2superscriptsubscriptπœ‚π‘š2\displaystyle+\sum_{m}^{1,2,3}\{d_{2}^{(o)}\phi_{m}^{2}\phi_{n}^{2}+d_{4}^{(o)% }\eta_{m}^{2}\eta_{n}^{2}+d_{6}^{(o)}(\phi_{m}^{2}\eta_{n}^{2}+\phi_{n}^{2}% \eta_{m}^{2})\},+ βˆ‘ start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 1 , 2 , 3 end_POSTSUPERSCRIPT { italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_o ) end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_o ) end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_o ) end_POSTSUPERSCRIPT ( italic_Ο• start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_Ο• start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT italic_m end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ) } ,

where n=m+1π‘›π‘š1n=m+1italic_n = italic_m + 1 and o=mβˆ’1π‘œπ‘š1o=m-1italic_o = italic_m - 1 mod 3. When Ο΅=Ο΅β€²=0italic-Ο΅superscriptitalic-Ο΅β€²0{\epsilon}={\epsilon}^{\prime}=0italic_Ο΅ = italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 0, dl(m)=dlsuperscriptsubscriptπ‘‘π‘™π‘šsubscript𝑑𝑙d_{l}^{(m)}=d_{l}italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT = italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT for any mπ‘šmitalic_m and l𝑙litalic_l. Below, we show that each dl(m)superscriptsubscriptπ‘‘π‘™π‘šd_{l}^{(m)}italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT has large Ο΅italic-Ο΅{\epsilon}italic_Ο΅- and Ο΅β€²superscriptitalic-Ο΅β€²{\epsilon}^{\prime}italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT-linear terms when n∼nvHSsimilar-to𝑛subscript𝑛vHSn\sim n_{\rm vHS}italic_n ∼ italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT.

Refer to caption
Figure S10: (a) FS for n=2.56𝑛2.56n=2.56italic_n = 2.56 with Ο΅β€²=0.005superscriptitalic-Ο΅β€²0.005{\epsilon}^{\prime}=0.005italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 0.005 and (b) normalized dl(m)superscriptsubscriptπ‘‘π‘™π‘šd_{l}^{(m)}italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT (m=1βˆ’3π‘š13m=1-3italic_m = 1 - 3; l=1βˆ’6𝑙16l=1-6italic_l = 1 - 6) as functions of the strain Ο΅β€²superscriptitalic-Ο΅β€²{\epsilon}^{\prime}italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT for Δ𝑬′=Ξ±β€²(Ο΅β€²/2,βˆ’Ο΅β€²,Ο΅β€²/2\Delta{\bm{E}}^{\prime}={\alpha}^{\prime}({\epsilon}^{\prime}/2,-{\epsilon}^{% \prime},{\epsilon}^{\prime}/2roman_Ξ” bold_italic_E start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = italic_Ξ± start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ( italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT / 2 , - italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT , italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT / 2) with Ξ±β€²=1superscript𝛼′1{\alpha}^{\prime}=1italic_Ξ± start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 1. Their Ο΅β€²superscriptitalic-Ο΅β€²{\epsilon}^{\prime}italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT-linear terms give glβ€²superscriptsubscript𝑔𝑙′g_{l}^{\prime}italic_g start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT in Eq. [S23]. At Ο΅=0italic-Ο΅0{\epsilon}=0italic_Ο΅ = 0, (d1,d2,d3,d4,d5,d6)=(26.09,62.52,11.09,33.97,14.09,42.23)subscript𝑑1subscript𝑑2subscript𝑑3subscript𝑑4subscript𝑑5subscript𝑑626.0962.5211.0933.9714.0942.23(d_{1},d_{2},d_{3},d_{4},d_{5},d_{6})=(26.09,62.52,11.09,33.97,14.09,42.23)( italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT , italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT , italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT , italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ) = ( 26.09 , 62.52 , 11.09 , 33.97 , 14.09 , 42.23 ). (c) FS for n=2.48𝑛2.48n=2.48italic_n = 2.48 with Ο΅β€²=0.005superscriptitalic-Ο΅β€²0.005{\epsilon}^{\prime}=0.005italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 0.005 and (d) obtained dl(m)superscriptsubscriptπ‘‘π‘™π‘šd_{l}^{(m)}italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT as functions of the strain Ο΅β€²superscriptitalic-Ο΅β€²{\epsilon}^{\prime}italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT. At Ο΅β€²=0superscriptitalic-Ο΅β€²0{\epsilon}^{\prime}=0italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 0, (d1,d2,d3,d4,d5,d6)=(52.40,19.41,15.09,16.60,21.02,16.87)subscript𝑑1subscript𝑑2subscript𝑑3subscript𝑑4subscript𝑑5subscript𝑑652.4019.4115.0916.6021.0216.87(d_{1},d_{2},d_{3},d_{4},d_{5},d_{6})=(52.40,19.41,15.09,16.60,21.02,16.87)( italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , italic_d start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT , italic_d start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT , italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT , italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ) = ( 52.40 , 19.41 , 15.09 , 16.60 , 21.02 , 16.87 ). In both (b) and (d), the relations dl(1)=dl(2)superscriptsubscript𝑑𝑙1superscriptsubscript𝑑𝑙2d_{l}^{(1)}=d_{l}^{(2)}italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 1 ) end_POSTSUPERSCRIPT = italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 2 ) end_POSTSUPERSCRIPT (l=1βˆ’6𝑙16l=1-6italic_l = 1 - 6) and βˆ‚Ο΅β€²dl(3)=βˆ’2β’βˆ‚Ο΅β€²dl(1)subscriptsuperscriptitalic-Ο΅β€²superscriptsubscript𝑑𝑙32subscriptsuperscriptitalic-Ο΅β€²superscriptsubscript𝑑𝑙1{\partial}_{{\epsilon}^{\prime}}d_{l}^{(3)}=-2{\partial}_{{\epsilon}^{\prime}}% d_{l}^{(1)}βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 3 ) end_POSTSUPERSCRIPT = - 2 βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 1 ) end_POSTSUPERSCRIPT holds. The relation βˆ‚Ο΅β€²d5(m)βˆΌβˆ’βˆ‚Ο΅β€²d6(m)similar-tosubscriptsuperscriptitalic-Ο΅β€²superscriptsubscript𝑑5π‘šsubscriptsuperscriptitalic-Ο΅β€²superscriptsubscript𝑑6π‘š{\partial}_{{\epsilon}^{\prime}}d_{5}^{(m)}\sim-{\partial}_{{\epsilon}^{\prime% }}d_{6}^{(m)}βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT ∼ - βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT means that g5β€²superscriptsubscript𝑔5β€²g_{5}^{\prime}italic_g start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT and g6β€²superscriptsubscript𝑔6β€²g_{6}^{\prime}italic_g start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT have the same sign. Interestingly, βˆ‚Ο΅β€²dl(m)subscriptsuperscriptitalic-Ο΅β€²superscriptsubscriptπ‘‘π‘™π‘š{\partial}_{{\epsilon}^{\prime}}d_{l}^{(m)}βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT for n>nvHS𝑛subscript𝑛vHSn>n_{\rm vHS}italic_n > italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT in (b) and that for n<nvHS𝑛subscript𝑛vHSn<n_{\rm vHS}italic_n < italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT in (d) have opposite signs. The used parameters are T=0.01𝑇0.01T=0.01italic_T = 0.01 and v=0.6𝑣0.6v=0.6italic_v = 0.6.

Here, we calculate the coefficients dl(m)superscriptsubscriptπ‘‘π‘™π‘šd_{l}^{(m)}italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT in the three-orbital model based on the Green function methods explained in the SI F. The used model parameters are T=0.01𝑇0.01T=0.01italic_T = 0.01, g=0.6𝑔0.6g=0.6italic_g = 0.6, tβ€²=0superscript𝑑′0t^{\prime}=0italic_t start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 0, and Ξ±=Ξ±β€²=1𝛼superscript𝛼′1{\alpha}={\alpha}^{\prime}=1italic_Ξ± = italic_Ξ± start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 1. We first study the case of n=2.56(>nvHS)𝑛annotated2.56absentsubscript𝑛vHSn=2.56\ (>n_{\rm vHS})italic_n = 2.56 ( > italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT ), whose FS at Ο΅β€²=0.005superscriptitalic-Ο΅β€²0.005{\epsilon}^{\prime}=0.005italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 0.005 is shown in Fig. S10 (a). Figure S10 (b) shows the obtained dl(m)superscriptsubscriptπ‘‘π‘™π‘šd_{l}^{(m)}italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT normalized by its Ο΅β€²=0superscriptitalic-Ο΅β€²0{\epsilon}^{\prime}=0italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 0 value, as functions of Ο΅β€²superscriptitalic-Ο΅β€²{\epsilon}^{\prime}italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT for |Ο΅β€²|≀0.005superscriptitalic-Ο΅β€²0.005|{\epsilon}^{\prime}|\leq 0.005| italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT | ≀ 0.005. We also study the case of n=2.48(<nvHS)𝑛annotated2.48absentsubscript𝑛vHSn=2.48\ (<n_{\rm vHS})italic_n = 2.48 ( < italic_n start_POSTSUBSCRIPT roman_vHS end_POSTSUBSCRIPT ), whose FS at Ο΅β€²=0.005superscriptitalic-Ο΅β€²0.005{\epsilon}^{\prime}=0.005italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 0.005 is shown in Fig. S10 (c). Figure S10 (d) shows the normalized dl(m)superscriptsubscriptπ‘‘π‘™π‘šd_{l}^{(m)}italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT as functions of Ο΅β€²superscriptitalic-Ο΅β€²{\epsilon}^{\prime}italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT. In both cases, the coefficients exhibits sizable Ο΅β€²superscriptitalic-Ο΅β€²{\epsilon}^{\prime}italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT-linear terms, by reflecting the large p-h asymmetry of the kagome lattice model.

In contrast, the irreducible susceptibility with the current form factor Ο‡l⁒m,l⁒m0⁒(𝒒)=βˆ’Tβ’βˆ‘π’Œ,nf𝒒l⁒m⁒(k)βˆ—β’Gl⁒l⁒(π’Œ+𝒒,Ο΅n)⁒Gm⁒m⁒(π’Œ,Ο΅n)⁒f𝒒l⁒m⁒(k)superscriptsubscriptπœ’π‘™π‘šπ‘™π‘š0𝒒𝑇subscriptπ’Œπ‘›superscriptsubscriptπ‘“π’’π‘™π‘šsuperscriptπ‘˜subscriptπΊπ‘™π‘™π’Œπ’’subscriptitalic-ϡ𝑛subscriptπΊπ‘šπ‘šπ’Œsubscriptitalic-ϡ𝑛superscriptsubscriptπ‘“π’’π‘™π‘šπ‘˜\chi_{lm,lm}^{0}({{\bm{q}}})=-T\sum_{{{\bm{k}}},n}f_{{\bm{q}}}^{lm}(k)^{*}G_{% ll}({{\bm{k}}}+{{\bm{q}}},{\epsilon}_{n})G_{mm}({{\bm{k}}},{\epsilon}_{n})f_{{% \bm{q}}}^{lm}(k)italic_Ο‡ start_POSTSUBSCRIPT italic_l italic_m , italic_l italic_m end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( bold_italic_q ) = - italic_T βˆ‘ start_POSTSUBSCRIPT bold_italic_k , italic_n end_POSTSUBSCRIPT italic_f start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l italic_m end_POSTSUPERSCRIPT ( italic_k ) start_POSTSUPERSCRIPT βˆ— end_POSTSUPERSCRIPT italic_G start_POSTSUBSCRIPT italic_l italic_l end_POSTSUBSCRIPT ( bold_italic_k + bold_italic_q , italic_Ο΅ start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT ) italic_G start_POSTSUBSCRIPT italic_m italic_m end_POSTSUBSCRIPT ( bold_italic_k , italic_Ο΅ start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT ) italic_f start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_l italic_m end_POSTSUPERSCRIPT ( italic_k ) exhibit much smaller relative change in the same range of Ο΅β€²superscriptitalic-Ο΅β€²{\epsilon}^{\prime}italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT; ∼3similar-toabsent3\sim 3∼ 3% for n=2.56𝑛2.56n=2.56italic_n = 2.56 and ∼5similar-toabsent5\sim 5∼ 5% for n=2.48𝑛2.48n=2.48italic_n = 2.48. The form factor f^𝒒⁒(k)subscript^π‘“π’’π‘˜{\hat{f}}_{{\bm{q}}}(k)over^ start_ARG italic_f end_ARG start_POSTSUBSCRIPT bold_italic_q end_POSTSUBSCRIPT ( italic_k ) is introduced in the SI F. (The change in Ο‡l⁒m,l⁒m0⁒(𝒒i)superscriptsubscriptπœ’π‘™π‘šπ‘™π‘š0subscript𝒒𝑖\chi_{lm,lm}^{0}({{\bm{q}}}_{i})italic_Ο‡ start_POSTSUBSCRIPT italic_l italic_m , italic_l italic_m end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ( bold_italic_q start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) modifies the second order coefficient acsubscriptπ‘Žca_{\rm c}italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT.) The change in the 3rd order term is very small, which is O⁒(Ο΅2)𝑂superscriptitalic-Ο΅2O({\epsilon}^{2})italic_O ( italic_Ο΅ start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ) when fi⁒j(m)=Β±i⁒gi⁒j(m)superscriptsubscriptπ‘“π‘–π‘—π‘šplus-or-minus𝑖superscriptsubscriptπ‘”π‘–π‘—π‘šf_{ij}^{(m)}=\pm ig_{ij}^{(m)}italic_f start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT = Β± italic_i italic_g start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT.

Now, we consider the strain-induced change in the 4th order free energy term. By considering the symmetry property of the Feynman diagrams, Δ⁒FΔ𝐹\Delta Froman_Ξ” italic_F due to Ξ”β’π‘¬βˆΟ΅proportional-toΔ𝑬italic-Ο΅\Delta{\bm{E}}\propto{\epsilon}roman_Ξ” bold_italic_E ∝ italic_Ο΅ is given as

Δ⁒F⁒(Ο΅)=g1⁒(Ο•14βˆ’Ο•24)+g3⁒(Ξ·14βˆ’Ξ·24)Δ𝐹italic-Ο΅subscript𝑔1superscriptsubscriptitalic-Ο•14superscriptsubscriptitalic-Ο•24subscript𝑔3superscriptsubscriptπœ‚14superscriptsubscriptπœ‚24\displaystyle\Delta F({\epsilon})=g_{1}(\phi_{1}^{4}-\phi_{2}^{4})+g_{3}(\eta_% {1}^{4}-\eta_{2}^{4})roman_Ξ” italic_F ( italic_Ο΅ ) = italic_g start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ( italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT - italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT ) + italic_g start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT - italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT )
+2⁒g5⁒(Ο•12⁒η12βˆ’Ο•12⁒η22)2subscript𝑔5superscriptsubscriptitalic-Ο•12superscriptsubscriptπœ‚12superscriptsubscriptitalic-Ο•12superscriptsubscriptπœ‚22\displaystyle\ +2g_{5}(\phi_{1}^{2}\eta_{1}^{2}-\phi_{1}^{2}\eta_{2}^{2})+ 2 italic_g start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT ( italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT )
+g2⁒(Ο•32⁒ϕ12βˆ’Ο•22⁒ϕ32)+g4⁒(Ξ·34⁒η12βˆ’Ξ·22⁒η32)subscript𝑔2superscriptsubscriptitalic-Ο•32superscriptsubscriptitalic-Ο•12superscriptsubscriptitalic-Ο•22superscriptsubscriptitalic-Ο•32subscript𝑔4superscriptsubscriptπœ‚34superscriptsubscriptπœ‚12superscriptsubscriptπœ‚22superscriptsubscriptπœ‚32\displaystyle\ +g_{2}(\phi_{3}^{2}\phi_{1}^{2}-\phi_{2}^{2}\phi_{3}^{2})+g_{4}% (\eta_{3}^{4}\eta_{1}^{2}-\eta_{2}^{2}\eta_{3}^{2})+ italic_g start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ( italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ) + italic_g start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT ( italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT )
+g6⁒(Ξ·34⁒ϕ12+Ξ·14⁒ϕ32βˆ’Ξ·22⁒ϕ32βˆ’Ξ·32⁒ϕ22),subscript𝑔6superscriptsubscriptπœ‚34superscriptsubscriptitalic-Ο•12superscriptsubscriptπœ‚14superscriptsubscriptitalic-Ο•32superscriptsubscriptπœ‚22superscriptsubscriptitalic-Ο•32superscriptsubscriptπœ‚32superscriptsubscriptitalic-Ο•22\displaystyle\ +g_{6}(\eta_{3}^{4}\phi_{1}^{2}+\eta_{1}^{4}\phi_{3}^{2}-\eta_{% 2}^{2}\phi_{3}^{2}-\eta_{3}^{2}\phi_{2}^{2}),+ italic_g start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ( italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ) , (S22)

where glsubscript𝑔𝑙g_{l}italic_g start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT are Ο΅italic-Ο΅{\epsilon}italic_Ο΅-linear. glsubscript𝑔𝑙g_{l}italic_g start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT and dl(m)superscriptsubscriptπ‘‘π‘™π‘šd_{l}^{(m)}italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT are related as Ο΅β‹…(βˆ‚Ο΅dl(1),βˆ‚Ο΅dl(2),βˆ‚Ο΅dl(3))=gl⁒(1,βˆ’1,0)β‹…italic-Ο΅subscriptitalic-Ο΅superscriptsubscript𝑑𝑙1subscriptitalic-Ο΅superscriptsubscript𝑑𝑙2subscriptitalic-Ο΅superscriptsubscript𝑑𝑙3subscript𝑔𝑙110{\epsilon}\cdot({\partial}_{\epsilon}d_{l}^{(1)},{\partial}_{\epsilon}d_{l}^{(% 2)},{\partial}_{\epsilon}d_{l}^{(3)})=g_{l}(1,-1,0)italic_Ο΅ β‹… ( βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 1 ) end_POSTSUPERSCRIPT , βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 2 ) end_POSTSUPERSCRIPT , βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 3 ) end_POSTSUPERSCRIPT ) = italic_g start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT ( 1 , - 1 , 0 ) for l𝑙litalic_l=odd, and =βˆ’gl⁒(1,βˆ’1,0)absentsubscript𝑔𝑙110=-g_{l}(1,-1,0)= - italic_g start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT ( 1 , - 1 , 0 ) for l𝑙litalic_l=even. Also, Δ⁒Fβ€²Ξ”superscript𝐹′\Delta F^{\prime}roman_Ξ” italic_F start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT due to Ξ”β’π‘¬β€²βˆΟ΅β€²proportional-toΞ”superscript𝑬′superscriptitalic-Ο΅β€²\Delta{\bm{E}}^{\prime}\propto{\epsilon}^{\prime}roman_Ξ” bold_italic_E start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ∝ italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT is given as

Δ⁒F′⁒(Ο΅β€²)=g1′⁒(Ο•14+Ο•24βˆ’2⁒ϕ32)+g3′⁒(Ξ·14+Ξ·24βˆ’2⁒η34)Ξ”superscript𝐹′superscriptitalic-Ο΅β€²superscriptsubscript𝑔1β€²superscriptsubscriptitalic-Ο•14superscriptsubscriptitalic-Ο•242superscriptsubscriptitalic-Ο•32superscriptsubscript𝑔3β€²superscriptsubscriptπœ‚14superscriptsubscriptπœ‚242superscriptsubscriptπœ‚34\displaystyle\Delta F^{\prime}({\epsilon}^{\prime})=g_{1}^{\prime}(\phi_{1}^{4% }+\phi_{2}^{4}-2\phi_{3}^{2})+g_{3}^{\prime}(\eta_{1}^{4}+\eta_{2}^{4}-2\eta_{% 3}^{4})roman_Ξ” italic_F start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ( italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ) = italic_g start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ( italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT + italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT - 2 italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ) + italic_g start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT + italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT - 2 italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 4 end_POSTSUPERSCRIPT )
+2⁒g5′⁒(Ο•12⁒η12+Ο•22⁒η22βˆ’2⁒ϕ32⁒η32)2superscriptsubscript𝑔5β€²superscriptsubscriptitalic-Ο•12superscriptsubscriptπœ‚12superscriptsubscriptitalic-Ο•22superscriptsubscriptπœ‚222superscriptsubscriptitalic-Ο•32superscriptsubscriptπœ‚32\displaystyle\ +2g_{5}^{\prime}(\phi_{1}^{2}\eta_{1}^{2}+\phi_{2}^{2}\eta_{2}^% {2}-2\phi_{3}^{2}\eta_{3}^{2})+ 2 italic_g start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ( italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - 2 italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT )
+g2′⁒(Ο•22⁒ϕ32+Ο•32⁒ϕ12βˆ’2⁒ϕ12⁒ϕ22)superscriptsubscript𝑔2β€²superscriptsubscriptitalic-Ο•22superscriptsubscriptitalic-Ο•32superscriptsubscriptitalic-Ο•32superscriptsubscriptitalic-Ο•122superscriptsubscriptitalic-Ο•12superscriptsubscriptitalic-Ο•22\displaystyle\ +g_{2}^{\prime}(\phi_{2}^{2}\phi_{3}^{2}+\phi_{3}^{2}\phi_{1}^{% 2}-2\phi_{1}^{2}\phi_{2}^{2})+ italic_g start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ( italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - 2 italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT )
+g4′⁒(Ξ·22⁒η32+Ξ·32⁒η12βˆ’2⁒η12⁒η22)superscriptsubscript𝑔4β€²superscriptsubscriptπœ‚22superscriptsubscriptπœ‚32superscriptsubscriptπœ‚32superscriptsubscriptπœ‚122superscriptsubscriptπœ‚12superscriptsubscriptπœ‚22\displaystyle\ +g_{4}^{\prime}(\eta_{2}^{2}\eta_{3}^{2}+\eta_{3}^{2}\eta_{1}^{% 2}-2\eta_{1}^{2}\eta_{2}^{2})+ italic_g start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ( italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - 2 italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT )
+g6′⁒(Ο•22⁒η32+Ο•32⁒η22+Ο•32⁒η12+Ο•12⁒η32βˆ’2⁒ϕ12⁒η22βˆ’2⁒ϕ22⁒η12),superscriptsubscript𝑔6β€²superscriptsubscriptitalic-Ο•22superscriptsubscriptπœ‚32superscriptsubscriptitalic-Ο•32superscriptsubscriptπœ‚22superscriptsubscriptitalic-Ο•32superscriptsubscriptπœ‚12superscriptsubscriptitalic-Ο•12superscriptsubscriptπœ‚322superscriptsubscriptitalic-Ο•12superscriptsubscriptπœ‚222superscriptsubscriptitalic-Ο•22superscriptsubscriptπœ‚12\displaystyle\ +g_{6}^{\prime}(\phi_{2}^{2}\eta_{3}^{2}+\phi_{3}^{2}\eta_{2}^{% 2}+\phi_{3}^{2}\eta_{1}^{2}+\phi_{1}^{2}\eta_{3}^{2}-2\phi_{1}^{2}\eta_{2}^{2}% -2\phi_{2}^{2}\eta_{1}^{2}),+ italic_g start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ( italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_Ο• start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - 2 italic_Ο• start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT - 2 italic_Ο• start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ) ,
(S23)

where glβ€²superscriptsubscript𝑔𝑙′g_{l}^{\prime}italic_g start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT are Ο΅β€²superscriptitalic-Ο΅β€²{\epsilon}^{\prime}italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT-linear. glβ€²superscriptsubscript𝑔𝑙′g_{l}^{\prime}italic_g start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT and dl(m)superscriptsubscriptπ‘‘π‘™π‘šd_{l}^{(m)}italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT are related as Ο΅β€²β‹…(βˆ‚Ο΅β€²dl(1),βˆ‚Ο΅β€²dl(2),βˆ‚Ο΅β€²dl(3))=gl′⁒(βˆ’1,βˆ’1,2)β‹…superscriptitalic-Ο΅β€²subscriptsuperscriptitalic-Ο΅β€²superscriptsubscript𝑑𝑙1subscriptsuperscriptitalic-Ο΅β€²superscriptsubscript𝑑𝑙2subscriptsuperscriptitalic-Ο΅β€²superscriptsubscript𝑑𝑙3superscriptsubscript𝑔𝑙′112{\epsilon}^{\prime}\cdot({\partial}_{{\epsilon}^{\prime}}d_{l}^{(1)},{\partial% }_{{\epsilon}^{\prime}}d_{l}^{(2)},{\partial}_{{\epsilon}^{\prime}}d_{l}^{(3)}% )=g_{l}^{\prime}(-1,-1,2)italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT β‹… ( βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 1 ) end_POSTSUPERSCRIPT , βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 2 ) end_POSTSUPERSCRIPT , βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 3 ) end_POSTSUPERSCRIPT ) = italic_g start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ( - 1 , - 1 , 2 ) for l𝑙litalic_l=odd, and =βˆ’gl′⁒(βˆ’1,βˆ’1,2)absentsuperscriptsubscript𝑔𝑙′112=-g_{l}^{\prime}(-1,-1,2)= - italic_g start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ( - 1 , - 1 , 2 ) for l𝑙litalic_l=even.

Here, we consider the possible 3⁒Q3𝑄3Q3 italic_Q BO-current coexisting state motivated by the experiment in Ref. Moll-hzS . As explained in Ref. Tazai-kagome2S , in the case of |Ο•|≫|𝜼|much-greater-thanbold-italic-Ο•πœΌ|{\bm{\phi}}|\gg|{\bm{\eta}}|| bold_italic_Ο• | ≫ | bold_italic_Ξ· | at Ο΅=Ο΅β€²=0italic-Ο΅superscriptitalic-Ο΅β€²0{\epsilon}={\epsilon}^{\prime}=0italic_Ο΅ = italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = 0, the 3Q BO Ο•0=(Ο•0,Ο•0,Ο•0)/3superscriptbold-italic-Ο•0superscriptitalic-Ο•0superscriptitalic-Ο•0superscriptitalic-Ο•03{\bm{\phi}}^{0}=(\phi^{0},\phi^{0},\phi^{0})/\sqrt{3}bold_italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT = ( italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT , italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT , italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ) / square-root start_ARG 3 end_ARG coexists with the current order 𝜼∝(Ξ·1,Ξ·2,βˆ’Ξ·1βˆ’Ξ·2)proportional-to𝜼subscriptπœ‚1subscriptπœ‚2subscriptπœ‚1subscriptπœ‚2{\bm{\eta}}\propto(\eta_{1},\eta_{2},-\eta_{1}-\eta_{2})bold_italic_Ξ· ∝ ( italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , - italic_Ξ· start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT - italic_Ξ· start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) due to the energy gain by the 3rd GL terms with b1⁒b2<0subscript𝑏1subscript𝑏20b_{1}b_{2}<0italic_b start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT < 0. (The relation b1⁒b2<0subscript𝑏1subscript𝑏20b_{1}b_{2}<0italic_b start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT < 0 is general Tazai-kagome2S .) Then, the BO-current coexisting state is nematic. Here, we set 𝜼=(Ξ·,0,βˆ’Ξ·)/2πœΌπœ‚0πœ‚2{\bm{\eta}}=(\eta,0,-\eta)/\sqrt{2}bold_italic_Ξ· = ( italic_Ξ· , 0 , - italic_Ξ· ) / square-root start_ARG 2 end_ARG (or 𝜼=(Ξ·,βˆ’2⁒η,Ξ·)/6πœΌπœ‚2πœ‚πœ‚6{\bm{\eta}}=(\eta,-2\eta,\eta)/\sqrt{6}bold_italic_Ξ· = ( italic_Ξ· , - 2 italic_Ξ· , italic_Ξ· ) / square-root start_ARG 6 end_ARG) without loss of generality.

Hereafter, we calculate the change in the current order transition temperature Δ⁒TΒ―c⁒(Ο΅)(∝ϡ)annotatedΞ”subscript¯𝑇citalic-Ο΅proportional-toabsentitalic-Ο΅\Delta{\bar{T}}_{\rm c}({\epsilon})\ (\propto{\epsilon})roman_Ξ” overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( italic_Ο΅ ) ( ∝ italic_Ο΅ ) under the 3⁒Q3𝑄3Q3 italic_Q BO phase. (Below, we set b1=βˆ’b2subscript𝑏1subscript𝑏2b_{1}=-b_{2}italic_b start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = - italic_b start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT for simplicity to obtain the analytic expression.) When 𝜼=𝟎𝜼0{\bm{\eta}}={\bm{0}}bold_italic_Ξ· = bold_0, we obtain the square of the i𝑖iitalic_i-th component of Ο•bold-italic-Ο•{\bm{\phi}}bold_italic_Ο• as (Ο•i)2=(Ο•0)2/3+ui⁒ψ2superscriptsubscriptitalic-ϕ𝑖2superscriptsuperscriptitalic-Ο•023subscript𝑒𝑖superscriptπœ“2(\phi_{i})^{2}=(\phi^{0})^{2}/3+u_{i}\psi^{2}( italic_Ο• start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT = ( italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ) start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT / 3 + italic_u start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT italic_ψ start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT (u1=u2=βˆ’1subscript𝑒1subscript𝑒21u_{1}=u_{2}=-1italic_u start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = italic_u start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT = - 1 and u3=2subscript𝑒32u_{3}=2italic_u start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT = 2), where ψ2=[(2⁒g1+g2)/(2⁒d1βˆ’d2)]⁒(Ο•0)2(∝ϡ)superscriptπœ“2annotateddelimited-[]2subscript𝑔1subscript𝑔22subscript𝑑1subscript𝑑2superscriptsuperscriptitalic-Ο•02proportional-toabsentitalic-Ο΅\psi^{2}=[(2g_{1}+g_{2})/(2d_{1}-d_{2})](\phi^{0})^{2}\ (\propto{\epsilon})italic_ψ start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT = [ ( 2 italic_g start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT + italic_g start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) / ( 2 italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT - italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) ] ( italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ) start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ( ∝ italic_Ο΅ ). We consider |𝜼|2∼O⁒(Ο΅)similar-tosuperscript𝜼2𝑂italic-Ο΅|{\bm{\eta}}|^{2}\sim O({\epsilon})| bold_italic_Ξ· | start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ∼ italic_O ( italic_Ο΅ ) by assuming T∼Tcsimilar-to𝑇subscript𝑇cT\sim T_{\rm c}italic_T ∼ italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT. Then, the change in the free energy by Δ⁒𝑬Δ𝑬\Delta{\bm{E}}roman_Ξ” bold_italic_E of order O⁒(Ξ·2)𝑂superscriptπœ‚2O(\eta^{2})italic_O ( italic_Ξ· start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ) is obtained as Δ⁒F=βˆ’[(2⁒g5+g6)+D⁒(2⁒g1+g2)]⁒((Ο•0)2/6)β‹…Ξ·2Δ𝐹⋅delimited-[]2subscript𝑔5subscript𝑔6𝐷2subscript𝑔1subscript𝑔2superscriptsuperscriptitalic-Ο•026superscriptπœ‚2\Delta F=-[(2g_{5}+g_{6})+D(2g_{1}+g_{2})]((\phi^{0})^{2}/6)\cdot\eta^{2}roman_Ξ” italic_F = - [ ( 2 italic_g start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT + italic_g start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ) + italic_D ( 2 italic_g start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT + italic_g start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) ] ( ( italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ) start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT / 6 ) β‹… italic_Ξ· start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT, where D≑(2⁒d5βˆ’d6)/(2⁒d1βˆ’d2)𝐷2subscript𝑑5subscript𝑑62subscript𝑑1subscript𝑑2D\equiv(2d_{5}-d_{6})/(2d_{1}-d_{2})italic_D ≑ ( 2 italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT - italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ) / ( 2 italic_d start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT - italic_d start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ). In the present numerical study, D∼1similar-to𝐷1D\sim 1italic_D ∼ 1 and (2⁒g5+g6)β‰ˆ(2⁒g1+g2)2subscript𝑔5subscript𝑔62subscript𝑔1subscript𝑔2(2g_{5}+g_{6})\approx(2g_{1}+g_{2})( 2 italic_g start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT + italic_g start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ) β‰ˆ ( 2 italic_g start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT + italic_g start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ). (Exactly speaking, D=1.36𝐷1.36D=1.36italic_D = 1.36 (0.29) for n=2.56𝑛2.56n=2.56italic_n = 2.56 (2.48).)

In the BO phase at Ο΅=0italic-Ο΅0{\epsilon}=0italic_Ο΅ = 0, the original 2nd order GL coefficient acsubscriptπ‘Žca_{\rm c}italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT is changed by Ο•0β‰ 0superscriptitalic-Ο•00\phi^{0}\neq 0italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT β‰  0 as

aΒ―c⁒(0)subscriptΒ―π‘Žc0\displaystyle{\bar{a}}_{\rm c}(0)overΒ― start_ARG italic_a end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( 0 ) =\displaystyle== ac+2⁒(d5+d6)⁒((Ο•0)2/3)subscriptπ‘Žc2superscript𝑑5superscript𝑑6superscriptsuperscriptitalic-Ο•023\displaystyle a_{\rm c}+2(d^{5}+d^{6})((\phi^{0})^{2}/3)italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT + 2 ( italic_d start_POSTSUPERSCRIPT 5 end_POSTSUPERSCRIPT + italic_d start_POSTSUPERSCRIPT 6 end_POSTSUPERSCRIPT ) ( ( italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ) start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT / 3 ) (S24)
∝proportional-to\displaystyle\propto∝ Tβˆ’TΒ―c⁒(0),𝑇subscript¯𝑇c0\displaystyle T-{\bar{T}}_{\rm c}(0),italic_T - overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( 0 ) ,

where |Ο•0|≳Tbgreater-than-or-equivalent-tosuperscriptitalic-Ο•0subscript𝑇b|\phi^{0}|\gtrsim T_{\rm b}| italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT | ≳ italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT when Tβ‰ͺTbmuch-less-than𝑇subscript𝑇bT\ll T_{\rm b}italic_T β‰ͺ italic_T start_POSTSUBSCRIPT roman_b end_POSTSUBSCRIPT. For finite Ο΅italic-Ο΅{\epsilon}italic_Ο΅, it is changed by Δ⁒F⁒(Ο΅)Δ𝐹italic-Ο΅\Delta F({\epsilon})roman_Ξ” italic_F ( italic_Ο΅ ) as

aΒ―c⁒(Ο΅)subscriptΒ―π‘Žcitalic-Ο΅\displaystyle{\bar{a}}_{\rm c}({\epsilon})overΒ― start_ARG italic_a end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( italic_Ο΅ ) =\displaystyle== aΒ―c⁒(0)βˆ’[(2⁒g5+g6)+D⁒(2⁒g1+g2)]⁒((Ο•0)2/6)subscriptΒ―π‘Žc0delimited-[]2subscript𝑔5subscript𝑔6𝐷2subscript𝑔1subscript𝑔2superscriptsuperscriptitalic-Ο•026\displaystyle{\bar{a}}_{\rm c}(0)-[(2g_{5}+g_{6})+D(2g_{1}+g_{2})]((\phi^{0})^% {2}/6)overΒ― start_ARG italic_a end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( 0 ) - [ ( 2 italic_g start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT + italic_g start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ) + italic_D ( 2 italic_g start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT + italic_g start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) ] ( ( italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ) start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT / 6 ) (S25)
∝proportional-to\displaystyle\propto∝ Tβˆ’TΒ―c⁒(Ο΅),𝑇subscript¯𝑇citalic-Ο΅\displaystyle T-{\bar{T}}_{\rm c}({\epsilon}),italic_T - overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( italic_Ο΅ ) ,

The current order appears in the BO phase when Eq. [S26] becomes negative. Therefore, TΒ―csubscript¯𝑇c{\bar{T}}_{\rm c}overΒ― start_ARG italic_T end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT will increase in proportion to Ο΅italic-Ο΅{\epsilon}italic_Ο΅ if (2⁒g5+g6)>02subscript𝑔5subscript𝑔60(2g_{5}+g_{6})>0( 2 italic_g start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT + italic_g start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ) > 0. (If (2⁒g5+g6)<02subscript𝑔5subscript𝑔60(2g_{5}+g_{6})<0( 2 italic_g start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT + italic_g start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ) < 0, Tcsubscript𝑇cT_{\rm c}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT increases when 𝜼∝(0,Ξ·,βˆ’Ξ·)proportional-to𝜼0πœ‚πœ‚{\bm{\eta}}\propto(0,\eta,-\eta)bold_italic_Ξ· ∝ ( 0 , italic_Ξ· , - italic_Ξ· ).)

In the same way, aΒ―c⁒(Ο΅)subscriptΒ―π‘Žcitalic-Ο΅{\bar{a}}_{\rm c}({\epsilon})overΒ― start_ARG italic_a end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( italic_Ο΅ ) is modified by Δ⁒F′⁒(Ο΅β€²)Ξ”superscript𝐹′superscriptitalic-Ο΅β€²\Delta F^{\prime}({\epsilon}^{\prime})roman_Ξ” italic_F start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ( italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ) as

aΒ―c⁒(0)βˆ’[(2⁒g5β€²+g6β€²)+D⁒(2⁒g1β€²+g2β€²)]⁒((Ο•0)2/6).subscriptΒ―π‘Žc0delimited-[]2superscriptsubscript𝑔5β€²superscriptsubscript𝑔6′𝐷2superscriptsubscript𝑔1β€²superscriptsubscript𝑔2β€²superscriptsuperscriptitalic-Ο•026\displaystyle{\bar{a}}_{\rm c}(0)-[(2g_{5}^{\prime}+g_{6}^{\prime})+D(2g_{1}^{% \prime}+g_{2}^{\prime})]((\phi^{0})^{2}/6).overΒ― start_ARG italic_a end_ARG start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT ( 0 ) - [ ( 2 italic_g start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT + italic_g start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ) + italic_D ( 2 italic_g start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT + italic_g start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT ) ] ( ( italic_Ο• start_POSTSUPERSCRIPT 0 end_POSTSUPERSCRIPT ) start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT / 6 ) . (S26)

Note that g5β€²=βˆ’Ο΅β€²β’βˆ‚Ο΅β€²d5(1)superscriptsubscript𝑔5β€²superscriptitalic-Ο΅β€²subscriptsuperscriptitalic-Ο΅β€²superscriptsubscript𝑑51g_{5}^{\prime}=-{\epsilon}^{\prime}{\partial}_{{\epsilon}^{\prime}}d_{5}^{(1)}italic_g start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = - italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 1 ) end_POSTSUPERSCRIPT and g6β€²=Ο΅β€²β’βˆ‚Ο΅β€²d6(1)superscriptsubscript𝑔6β€²superscriptitalic-Ο΅β€²subscriptsuperscriptitalic-Ο΅β€²superscriptsubscript𝑑61g_{6}^{\prime}={\epsilon}^{\prime}{\partial}_{{\epsilon}^{\prime}}d_{6}^{(1)}italic_g start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT = italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT βˆ‚ start_POSTSUBSCRIPT italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT end_POSTSUBSCRIPT italic_d start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( 1 ) end_POSTSUPERSCRIPT has the same sign according to Fig. S10, which is naturally expected analytically.

Considering the drastic Ο΅β€²superscriptitalic-Ο΅β€²{\epsilon}^{\prime}italic_Ο΅ start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT-dependence of dl(m)superscriptsubscriptπ‘‘π‘™π‘šd_{l}^{(m)}italic_d start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT obtained in Fig. S10 (a), in collaboration with the change in acsubscriptπ‘Žca_{\rm c}italic_a start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT already discussed in Ref. Moll-hzS , sizable strain-induced increment of the current order transition temperature Tcsubscript𝑇cT_{\rm c}italic_T start_POSTSUBSCRIPT roman_c end_POSTSUBSCRIPT reported in Ref. Moll-hzS will be realized in the present mechanism.

In the case of fi⁒j(m)=Β±i⁒gi⁒j(m)superscriptsubscriptπ‘“π‘–π‘—π‘šplus-or-minus𝑖superscriptsubscriptπ‘”π‘–π‘—π‘šf_{ij}^{(m)}=\pm ig_{ij}^{(m)}italic_f start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT = Β± italic_i italic_g start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ( italic_m ) end_POSTSUPERSCRIPT, glsubscript𝑔𝑙g_{l}italic_g start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT and glβ€²superscriptsubscript𝑔𝑙′g_{l}^{\prime}italic_g start_POSTSUBSCRIPT italic_l end_POSTSUBSCRIPT start_POSTSUPERSCRIPT β€² end_POSTSUPERSCRIPT coincide with l𝑙litalic_l=even and l𝑙litalic_l=odd, respectively. Then, D=1𝐷1D=1italic_D = 1 and (2⁒g5+g6)=(2⁒g1+g2)2subscript𝑔5subscript𝑔62subscript𝑔1subscript𝑔2(2g_{5}+g_{6})=(2g_{1}+g_{2})( 2 italic_g start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT + italic_g start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT ) = ( 2 italic_g start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT + italic_g start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ). The main results are valid even in this case.

References

  • (1) D. Ceresoli, T. Thonhauser, D. Vanderbilt, and R. Resta, Orbital magnetization in crystalline solids: Multi-band insulators, Chern insulators, and metals, Phys. Rev. B 74, 024408 (2006).
  • (2) J. Shi, G. Vignale, D. Xiao, and Q. Niu, Quantum Theory of Orbital Magnetization and Its Generalization to Interacting Systems, Phys. Rev. Lett. 99, 197202 (2007).
  • (3) R. Nourafkan, G. Kotliar, and A.-M. S. Tremblay, Orbital magnetization of correlated electrons with arbitrary band topology, Phys. Rev. B 90, 125132 (2014).
  • (4) X. Wu, T. Schwemmer, T. MΓΌller, A. Consiglio, G. Sangiovanni, D. Di Sante, Y. Iqbal, W. Hanke, A. P. Schnyder, M. M. Denner, M. H. Fischer, T. Neupert, and R. Thomale, Nature of Unconventional Pairing in the Kagome Superconductors A⁒V3⁒Sb5𝐴subscriptV3subscriptSb5A{\mathrm{V}}_{3}{\mathrm{Sb}}_{5}italic_A roman_V start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT roman_Sb start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT (A=K,Rb,Cs𝐴KRbCsA=\mathrm{K},\mathrm{Rb},\mathrm{Cs}italic_A = roman_K , roman_Rb , roman_Cs), Phys. Rev. Lett. 127, 177001 (2021).
  • (5) R. Tazai, Y. Yamakawa,and H. Kontani, Charge-loop current order and Z3subscript𝑍3Z_{3}italic_Z start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT nematicity mediated by bond-order Fluctuations in kagome metal AV3Sb5 (A= Cs,Rb,K), arXiv:2207.08068 (2022).
  • (6) R. Tazai, Y. Yamakawa, M. Tsuchiizu, and H. Kontani, d𝑑ditalic_d- and p𝑝pitalic_p-wave quantum liquid crystal orders in cuprate superconductors, ΞΊπœ…\kappaitalic_ΞΊ-(BEDT-TTF)2X, and coupled chain Hubbard models: functional-renormalization-group analysis, J. Phys. Soc. Jpn. 90, 111012 (2021).
  • (7) T. Park, M. Ye, and L. Balents, Electronic instabilities of kagome metals: Saddle points and Landau theory, Phys. Rev. B 104, 035142 (2021).
  • (8) H. Kontani, R. Tazai, Y. Yamakawa, and S. Onari, Unconventional density waves and superconductivities in Fe-based superconductors and other strongly correlated electron systems, Adv. Phys. 70, 355 (2021).
  • (9) H. Kontani and S. Onari, Orbital-Fluctuation-Mediated Superconductivity in Iron Pnictides: Analysis of the Five-Orbital Hubbard-Holstein Model,P hys. Rev. Lett. 104, 157001 (2010).
  • (10) R. Tazai, S. Matsubara, Y. Yamakawa, S. Onari, and H. Kontani, Rigorous formalism for unconventional symmetry breaking in Fermi liquid theory and its application to nematicity in FeSe, Phys. Rev. B 107, 035137 (2023).
  • (11) Y. Hu, X. Wu, B. R. Ortiz, S. Ju, X. Han, J. Ma, N. C. Plumb, M. Radovic, R. Thomale, S. D. Wilson, A. P. Schnyder, and M. Shi, Rich nature of Van Hove singularities in Kagome superconductor CsV3Sb5, Nat. Commun. 13, 2220 (2022).
  • (12) C. Guo, G. Wagner, C. Putzke, D. Chen, K. Wang, L. Zhang, M. G. Amigo, I. Errea, M. G. Vergniory, C. Felser, M. H. Fischer, T. Neupert, and P. J. W. Moll, Correlated order at the tipping point in the kagome metal CsV3Sb5, arXiv:2304.00972.