CsCr6Sb6: Kagome Kondo Lattice

• CsCr6Sb6 is a quasi-2D kagome bilayer material that hosts a Kondo lattice with unique d-electron correlations, flat bands, and heavy-fermion characteristics.
• Its bilayer structure and van der Waals nature create flat and dispersive band intersections, influencing Dirac points, van Hove singularities, and dimensional crossover effects.
• Advanced BN-capping synthesis enables stable, high-quality crystals, facilitating detailed ARPES, STM/STS, and transport studies of Kondo hybridization and frustrated magnetism.

CsCr$_6$Sb$_6$ is a van der Waals-like, quasi-two-dimensional kagome bilayer compound in the 166 family, crystallizing in space group $R\bar{3}m$. It has been reported as the first experimentally realized Kagome Kondo lattice and as a $d$-electron kagome Kondo lattice in which flat-band physics, heavy-fermion behavior, ultra-low carrier density, frustrated magnetism, and dimensionality-tuned electronic reconstruction coexist. Across recent studies, it is described as a platform where Cr $3d$-derived kagome bands near $E_F$ generate both local-moment and itinerant-electron physics, enabling transport, ARPES, STM/STS, and thin-flake measurements to be interpreted within a correlated kagome Kondo-lattice framework (Song et al., 2024, Zhang et al., 19 Mar 2026, Zhang et al., 3 Apr 2026).

1. Structural identity and kagome platform

CsCr$_6$Sb$_6$ is described as a double kagome or kagome bilayer material built from bilayer $(\mathrm{Cr}_3\mathrm{Sb})_2$ networks stacked along $c$ and separated by Cs atoms. The crystallographic symmetry is $_6$0, and one spectroscopic study reports lattice parameters $_6$1 and $_6$2. The structure contains a Cr kagome sublattice and an Sb1 hexagonal sublattice within a prototypical Cr$_6$3Sb plane, with Sb2 honeycomb layers and Cs spacer layers. Two natural cleavage surfaces are reported: Cs-terminated and Sb2-terminated (Song et al., 2024, Zhang et al., 3 Apr 2026).

This layered construction is central to the material’s electronic phenomenology. The large spacer separation and bilayer geometry are reported to reduce interlayer hopping and produce strong quasi-2D character. That geometry also matters because kagome lattices are associated with Dirac points, van Hove singularities, flat bands, and geometric frustration, while the bilayer structure specifically produces kagome doublet bands. In the current literature, the bilayer degree of freedom is not a minor structural detail: it is the basis for the reported intersection between flat and dispersive bands and for the flat-band-resonance interpretation advanced in later ARPES work (Song et al., 2024, Zhang et al., 19 Mar 2026).

Mechanically, the compound is described as van der Waals-like and exfoliable to few-layer and near-2D limits. That exfoliability is not only a sample-preparation convenience. It is directly tied to observations of dimensionality-induced Kondo breakdown and to the emergence of hidden A-type antiferromagnetic order in thin layers, making dimensional reduction an intrinsic control parameter rather than an external perturbation (Song et al., 2024).

2. Growth chemistry and boron nitride capping

A recent crystal-growth study uses CsCr$_6$4Sb$_6$5 as a principal demonstration of a boron nitride capping and sealing method developed for highly reactive syntheses. The motivation is explicit: conventional quartz ampule sealing is inadequate for Cs-rich Sb-containing growths because cesium reacts with quartz at elevated temperatures, depleting the oversaturated Cs vapor environment required for crystal growth and creating risks of tube corrosion, rupture, compositional loss, and experimental inconsistency. The BN-cap method is presented as a low-cost alternative to sealed Ta/Nb crucibles and related specialized containment approaches, especially for alkali-, alkaline-earth-, and rare-earth-containing systems (Song et al., 29 May 2026).

For CsCr$_6$6Sb$_6$7, the growth is deliberately nonstoichiometric because polycrystalline CsCr$_6$8Sb$_6$9 cannot be obtained through stoichiometric reactions. Instead, the starting elements are Cs metal, Cr powder, and Sb chunks in a molar ratio of $R\bar{3}m$0, loaded into an alumina crucible, BN-capped, and sealed in a vacuum quartz tube. The thermal profile is $R\bar{3}m$1 for 24 h, rapid cooling to $R\bar{3}m$2 over 2 h, and then slow cooling to $R\bar{3}m$3 at $R\bar{3}m$4. The tube is then quenched in air and carefully tilted so that the molten flux is decanted and the crystals are exposed; a centrifugal sieve is also noted as a way to simplify sample collection (Song et al., 29 May 2026).

Parameter	Reported value	Context
Growth composition	Cs:Cr:Sb = 10:3:30	Deliberate excess-Cs flux
High-temperature step	$R\bar{3}m$5 for 24 h	Reaction and homogenization
Intermediate cooling	$R\bar{3}m$6 over 2 h	Rapid cool
Final cooling	$R\bar{3}m$7 at $R\bar{3}m$8	Crystal growth
Crucible inner diameter	7 mm	Alumina crucible
BN cap geometry	6.8 mm bottom, 7.3 mm top	Truncated-cone press fit

The cap itself is fabricated from commercial BN rods described as soft and machinable. In the reported implementation, the BN cap is a truncated cone that press-fits onto a standard alumina crucible; slight deformation of the BN against the rigid crucible produces the seal. Relative to ordinary Al$R\bar{3}m$9O$d$0 crucible caps or sieves, which yield extremely low amounts of crystals too small for most measurements, BN capping produces much larger and thicker crystals. The thicker crystals are reported to be about $d$1 in thickness, and one crystal exceeds 3 mm in the $d$2-plane. The stated interpretation is that BN sealing preserves the superstoichiometric, oversaturated Cs environment that conventional quartz-based containment fails to maintain (Song et al., 29 May 2026).

3. Electronic structure and heavy-fermion phenomenology

The discovery report identifies CsCr$d$3Sb$d$4 as a Kagome Kondo lattice because it combines kagome-derived flat-band physics with local magnetic moments coupled to itinerant carriers. ARPES at 15 K shows a nearly dispersionless band along $d$5-M, $d$6-K, and K-M-K, including a flat band about 40 meV below $d$7, together with a small pocket crossing $d$8. The same work reports a Fermi surface containing a small $d$9-centered pocket, a triangular pocket near $3d$0, and spot-like features near $3d$1, as well as a separation between the flat bands and deeper valence bands of roughly 50 meV to 0.5 eV. Orbital-projected calculations assign the relevant low-energy states primarily to Cr $3d$2 orbitals, supporting a kagome electronic structure with local-moment and itinerant components on the same Cr network (Song et al., 2024).

The correlated character is quantified by several renormalization indicators. Comparison between ARPES and LDA yields a reported band renormalization factor of about 10, while DMFT restores a flat band at $3d$3 consistent with experiment. Thermodynamically, the Sommerfeld coefficient is reported as $3d$4, about three times larger than in isostructural CsV$3d$5Sb$3d$6. The quasiparticle mass enhancement is estimated as $3d$7, stated to be more than 100 times larger than the vanadium counterpart. These values place the system, by the authors’ interpretation, in a heavy-fermion regime more commonly associated with $3d$8-electron materials even though CsCr$3d$9Sb$E_F$0 is a $E_F$1-electron kagome system (Song et al., 2024).

Charge transport adds a dilute-carrier dimension to this picture. One study reports insulating behavior at an ultra-low carrier density of about $E_F$2, contrasting it with the higher-carrier-density metallic behavior of CsV$E_F$3Sb$E_F$4. The low-temperature resistivity is described by a Kondo-like minimum and $E_F$5 upturn, written in the form

$E_F$6

In that interpretation, the low carrier density and incomplete screening shift the material away from a conventional heavy-fermion metal toward Kondo-insulating-like behavior. This suggests that the Kondo lattice in CsCr$E_F$7Sb$E_F$8 is realized in an unusually dilute-carrier limit rather than in the metallic regime more typical of canonical heavy-fermion compounds (Song et al., 2024).

4. Spectroscopic hierarchy: Mottness, Kondo hybridization, and flat-band resonance

Two 2026 studies refine the microscopic interpretation beyond the initial phenomenology. STM/STS and ARPES measurements present a two-stage hierarchy in which strong correlations first split a kagome flat band into lower and upper Hubbard bands and only subsequently generate low-temperature Kondo hybridization. In STS, low-temperature spectra on both Cs and Sb2 terminations show an asymmetric gap-like suppression of the density of states near $E_F$9 with substantial residual DOS, fitted by a Fano-type lineshape,

$_6$0

Reported fit parameters include $_6$1 on the Cs surface and $_6$2 on the Sb2 surface, with $_6$3 meV on Cs and $_6$4 meV on Sb2. The width $_6$5 is converted, within a single-impurity Kondo estimate, to $_6$6, while the near-$_6$7 suppression fills in upon warming and closes around $_6$8 K (Zhang et al., 3 Apr 2026).

At higher energy, the same STS work resolves symmetric humps at approximately $_6$9 mV, interpreted as the lower and upper Hubbard bands of a Mott-split kagome flat band. Laser ARPES with $_6$0 eV identifies a weakly dispersive $_6$1 feature around 50 meV below $_6$2, together with near-$_6$3 quasiparticle peaks at $_6$4 and near $_6$5. The $_6$6 meV feature persists to substantially higher temperature, remaining visible above 95 K, whereas the near-$_6$7 quasiparticle peak disappears by about 65 K. The proposed sequence is therefore

$_6$8

with the occupied lower Hubbard band supplying the localized spins that later hybridize with itinerant carriers (Zhang et al., 3 Apr 2026).

A complementary ARPES-centered study emphasizes “flat band resonance” rather than only Kondo hybridization. In that account, CsCr$_6$9Sb$(\mathrm{Cr}_3\mathrm{Sb})_2$0 hosts coexisting flat and dispersive bands near $(\mathrm{Cr}_3\mathrm{Sb})_2$1, and cooling produces a pronounced enhancement of spectral weight and hybridization. The spectral-weight evolution is written as

$(\mathrm{Cr}_3\mathrm{Sb})_2$2

with increasing low-temperature $(\mathrm{Cr}_3\mathrm{Sb})_2$3 interpreted as the onset of coherent resonance between localized flat-band states and itinerant dispersive states. ARPES resolves dispersive electron pockets labeled $(\mathrm{Cr}_3\mathrm{Sb})_2$4 and $(\mathrm{Cr}_3\mathrm{Sb})_2$5 near $(\mathrm{Cr}_3\mathrm{Sb})_2$6 and $(\mathrm{Cr}_3\mathrm{Sb})_2$7, a hole pocket near $(\mathrm{Cr}_3\mathrm{Sb})_2$8, and flat-band-like features near $(\mathrm{Cr}_3\mathrm{Sb})_2$9 and around $c$0 eV. The key claim is that the bilayer kagome doublet permits intersections between flat and dispersive states that generate a temperature-dependent resonance, analogous in spirit to Kondo hybridization but synchronized with short-range antiferromagnetism rather than cleanly separated from it (Zhang et al., 19 Mar 2026).

5. Magnetic response, transport scales, and dimensional crossover

The bulk magnetic response is consistently described as frustrated rather than conventionally ordered. One study reports a kink around 72 K in resistivity and heat capacity, a broad hump in susceptibility, and no sharp $c$1-type anomaly expected for conventional long-range order. Another refers to short-range antiferromagnetic correlations with a resistivity kink near $c$2 and Curie–Weiss deviation in susceptibility. A spectroscopic context places the in-plane resistivity upturn below $c$3 K and the derivative kink at $c$4, interpreted as a frustrated magnetic transition. Taken together, these reports place the higher-temperature magnetic or frustration scale near $c$5–$c$6 K, with resistivity anomalies beginning below $c$7 K, while still distinguishing those scales from the lower-energy Kondo signatures (Song et al., 2024, Zhang et al., 19 Mar 2026, Zhang et al., 3 Apr 2026).

At lower temperature, transport reveals Kondo-lattice behavior. The initial report describes metallic behavior at high $c$8, a resistivity minimum, a canonical $c$9 upturn, and then a weaker increase below about $_6$00 K, interpreted as Kondo singlet formation and heavy-fermion coherence. In BN-grown bulk crystals, resistivity instead shows Kondo insulating behavior with a $_6$01 dependence at low temperature, and the deviation temperature is defined as $_6$02. In the same BN-grown samples, a small kink attributed to frustrated magnetic interactions appears at $_6$03, and these $_6$04 and $_6$05 values are noted to be slightly smaller than previously reported values. This indicates that the reported low-temperature scales are sample- and measurement-dependent within a shared Kondo-frustrated framework rather than numerically identical across all studies (Song et al., 2024, Song et al., 29 May 2026).

Dimensional reduction changes the balance between Kondo screening and magnetism. In thin flakes, the Kondo temperature decreases monotonically with thickness, the Kondo exchange weakens, and in the 2L limit the Kondo signature disappears; this is described as dimensionality-induced Kondo breakdown. At the same time, few-layer samples below about 20 layers display a hidden A-type antiferromagnetic state, with odd-layer samples showing ferromagnetic-like behavior and even-layer samples showing antiferromagnetic behavior. A BN-grown 3-layer thin-flake device exhibits anomalous Hall resistance $_6$06 with a very sharp and symmetric spin flip upon magnetic-field sweep and no additional steps compared with previous results, interpreted as evidence of well-defined magnetic order, reduced exchange bias, and a homogeneous magnetic domain structure. This combination of Kondo suppression and emergent layered magnetism is one of the defining experimental motifs of CsCr$_6$07Sb$_6$08 (Song et al., 2024, Song et al., 29 May 2026).

6. Interpretive status within kagome and Kondo-lattice research

The principal interpretive claim surrounding CsCr$_6$09Sb$_6$10 is that it realizes Kondo-lattice physics in a kagome $_6$11-electron material without relying on atomic $_6$12 or $_6$13 local moments. Instead, local moments are argued to arise from correlation-driven Mottness in a Cr $_6$14-derived kagome flat band. That position is used to distinguish the material from canonical heavy-fermion compounds, where localized orbitals are already atomic in character, and from simpler frustrated magnets, where low-energy hybridization signatures are not expected to follow the same spectroscopic hierarchy (Zhang et al., 3 Apr 2026).

Several alternative readings are explicitly disfavored in the current literature. The spectroscopic work argues against interpreting the system as a straightforward impurity-Kondo problem from excess Cr, against treating the low-energy reconstruction as merely the direct consequence of the higher-temperature magnetic anomaly, and against assigning the observed behavior to conventional long-range-order-driven gap formation. Likewise, the flat-band-resonance study emphasizes that the emergence of coherence in CsCr$_6$15Sb$_6$16 occurs together with short-range antiferromagnetic correlations, contrasting with conventional Kondo lattices in which Kondo resonance is usually not synchronized so sharply with magnetic correlations. This suggests that the compound may sit at an intermediate conceptual position: it is Kondo-like in its low-energy coherence and hybridization phenomenology, but its resonance physics is also described as magnetically intertwined flat-band resonance rather than a simple translation of $_6$17-electron Kondo-lattice behavior into a kagome setting (Zhang et al., 19 Mar 2026, Zhang et al., 3 Apr 2026).

In that sense, CsCr$_6$18Sb$_6$19 functions as a convergence point for several research programs: kagome flat bands near $_6$20, ultra-low-carrier correlated transport, frustrated magnetism without conventional bulk long-range order, thin-flake magnetic reconstruction, and synthesis methodology for reactive alkali-metal systems. A plausible implication is that continued work on this material will be shaped as much by advances in crystal-growth control—especially preservation of the oversaturated Cs environment—as by further refinement of the microscopic relation among Hubbard-band formation, quasiparticle coherence, and short-range antiferromagnetism (Song et al., 2024, Song et al., 29 May 2026).