Paramagnon-Interference Mechanism for Three-Dimensional Bond Order in Kagome Metals AV$_3$Sb$_5$ (A=Cs, Rb, K): Analysis by the Density-Wave Equation

The mechanism of CDW and its 3D structure are important fundamental issues in kagome metals. We have previously shown that, based on a 2D model, $2\times 2$ bond order (BO) emerges due to the paramagnon-interference (PMI) mechanism and that its fluctuations lead to $s$-wave superconductivity. This paper studies these issues based on realistic 3D models of kagome metals AV$3$Sb$5$ (A=Cs, Rb, K). We reveal that a commensurate 3D $2\times 2\times 2$ BO is caused by the PMI mechanism, by performing the 3D density-wave (DW) equation analysis for all A=Cs, Rb, K models in detail. Our results indicate a BO transition temperature $T{\rm BO}\sim 100$K within the regime of moderate electron correlation. The 3D structure of BO is attributed to the three-dimensionality of the Fermi surface, while the 3D structure of BO is sensitively changed, since the Fermi surface is quasi-2D. Based on the analysis of the DW equation, by taking into account a finite third-order Ginzburg-Landau (GL) term, (i) shift stacking $2\times 2\times 2$ BO can be realized via a first-order transition below $T{\rm BO}$. Here, the in-plane BO pattern (tri-hexagonal or star-of-David) is determined by the sign of the third-order GL term, with hole doping tending to favor the tri-hexagonal state. On the other hand, if the third-order GL term is very small, (ii) alternating vertical stacking BO may instead be realized via a second-order transition. The present study enhances our understanding of the rich variety of BOs observed experimentally. It is confirmed that the PMI mechanism is the essential origin of the 3D CDW of kagome metals.

💡 Research Summary

In this work the authors address the long‑standing puzzle of the three‑dimensional charge‑density‑wave (CDW) observed in the kagome metals AV₃Sb₅ (A = Cs, Rb, K). Building on their previous two‑dimensional study, they extend the paramagnon‑interference (PMI) mechanism to realistic three‑dimensional multi‑orbital tight‑binding models derived from first‑principles calculations. Using WIEN2k and Wannier90 they construct a 30‑orbital Hamiltonian (V 3d + Sb 5p) for each compound and fine‑tune the Sb p‑level by ΔEₚ = −0.2 eV to reproduce ARPES band dispersions. The resulting Fermi surfaces are quasi‑two‑dimensional but display a clear k_z dependence and a Lifshitz transition near the M point when the electron count is changed from n = 31 to n = 30.8 (hole doping).

Electron correlations are introduced solely as an on‑site Hubbard U (≈ 1 eV) on the b₃g (dₓz) orbitals. Within the random‑phase approximation (RPA) the spin Stoner factor α_S always exceeds the charge Stoner factor α_C, implying that a pure on‑site interaction cannot generate a non‑magnetic bond order (BO) without additional non‑local terms. The authors therefore solve the density‑wave (DW) equation, which includes Hartree, Maki‑Thompson, and Aslamazov‑Larkin (AL) vertex corrections. The AL‑V_C term embodies the PMI mechanism: two particle‑hole bubbles interfere, producing an effective non‑local interaction I_Q(k,p) that strongly enhances the charge‑channel susceptibility at the three nesting vectors q₁, q₂, q₃ of the kagome lattice.

The eigenvalue λ_Q of the DW equation measures the instability toward a BO with wave vector Q = (q, q_z). The largest λ_Q is found for Q‑vectors (π, π, π) and (π, π, 0), corresponding respectively to an alternating vertical stacking (v‑BO) and a shift‑stacking (s‑BO) of the 2 × 2 in‑plane pattern. λ_Q reaches unity at a temperature T_BO ≈ 100 K for moderate U, in excellent agreement with the experimentally observed CDW transition (≈ 90–100 K). This demonstrates that the PMI mechanism alone can generate the three‑dimensional 2 × 2 × 2 BO without invoking large nearest‑neighbor Coulomb V.

To determine the internal structure of the BO, the authors perform a Ginzburg‑Landau (GL) expansion up to third order. The sign of the cubic coefficient β₃ decides whether the in‑plane BO adopts the tri‑hexagonal (TrH, φ ∝ (1,1,1)) or the star‑of‑David (SoD, φ ∝ (−1,1,1)) pattern. Moreover, if |β₃| is sizable, the transition to s‑BO is first order; if β₃ is nearly zero, a continuous second‑order transition to v‑BO occurs. Hole doping shifts β₃ toward positive values, favoring the TrH configuration. Importantly, these conclusions hold for all three alkali metals, indicating that the 3D CDW phenomenology is robust against the modest changes in lattice parameters and electronic filling among Cs, Rb, and K compounds.

The paper thus establishes a unified microscopic picture: (i) the quasi‑2D Fermi surface determines the allowed q_z components; (ii) the AL‑V_C vertex (PMI) supplies an effective non‑local interaction that drives a 3Q bond order even with only on‑site U; (iii) the third‑order GL term selects the in‑plane pattern and the stacking sequence. This framework naturally explains the observed diversity of CDW superstructures (2 × 2 × 1, 2 × 2 × 2, 2 × 2 × 4) and their sensitivity to pressure, doping, and experimental probe. By linking the BO fluctuations to the previously reported s‑wave superconductivity, the study also provides a coherent route to understand the intertwined CDW‑SC phase diagram of kagome metals. The work sets a solid theoretical foundation for future explorations of topological states, chiral current order, and tunable quantum phases in AV₃Sb₅ and related kagome systems.