Si Intercalation Beneath Epitaxial Graphene: Modulating Mott States at the SiC(0001) Interface

TL;DR

This work demonstrates three approaches to intercalate Si beneath epitaxial graphene on SiC(0001) to decouple the buffer layer while revealing a Mott-Hubbard type insulating state associated with Si dangling bonds at the interface. Graphene’s itinerant electrons screen the on-site repulsion, reducing the Mott gap and enabling tunable correlation strength without evident hybridization with the graphene Dirac band. The intercalated system exhibits a lower Hubbard band near 0.6 eV and dispersive dangling-bond states, with the Hubbard parameter satisfying $U/b\ge2.0$ due to graphene screening; interface reconstructions (e.g., $(3\times3)$ and $(\surd{3}\times\surd{3})$R30°) indicate spatial inhomogeneity. Overall, the results establish Si intercalation as a platform to explore Mott physics in graphene–SiC heterostructures and to probe band-gap-tunable correlated states via interface engineering and doping.

Abstract

Intercalation has proven to be a powerful tool for tailoring the electronic properties of freestanding graphene layers as well as for stabilizing the intercalated material in a two-dimensional configuration. This work examines Si intercalation of epitaxial graphene on SiC(0001) using three preparation methods. Dangling bond states at the interface were found to undergo a Mott-Hubbard metal-insulator transition as a result of a significant on-site repulsion. Comparing this heterostructure consisting of graphene and a Mott insulator with a similar system without graphene, reveals the screening ability of graphene's conduction electrons on the on-site repulsion. The system presented here can serve as a template for further research on Mott insulators with variable band gap.

Figures (3)

• Figure 1: X-ray photoelectron spectra of the Si 2p core level recorded after different preparation steps, all starting from an epitaxially grown zero layer graphene (ZLG) sample. A representative spectrum for the ZLG is shown in (b1), featuring contributions from bulk Si (SiC) and topmost Si (SiC$_{\text{ZLG}}$) atoms coupled to the ZLG. Method A: Sequential. (a1) Si was deposited onto the ZLG, resulting in an additional component labeled Si. (a2) Annealing the sample to 800 caused significant spectral changes, modeled with a new component (SiC'), indicating modification of the interface. Method B: Simultaneous. (b2) Si deposition and annealing were performed simultaneously, producing a spectrum similar to (a2), with slightly different contributions. Method C: Exchange. (c1) The ZLG was decoupled via H intercalation by annealing in a hydrogen environment, yielding a spectrum with a shifted bulk contribution (SiC$^{*}$) and a new interface component (SiC$_{\text{H}}$), reflecting hydrogenation of the graphene-SiC interface. (c2) Subsequent Si deposition and annealing led to an energy shift in the spectrum. The insets show ball-and-stick models illustrating the individual preparation steps, with Si, C, and H atoms color-coded in blue, black, and red, respectively.
• Figure 2: (a)-(c) Energy-momentum cuts in the vicinity of the graphene $\text{K}_\text{G}$ point perpendicular to $\overline{\text{\textGamma}\text{K}_\text{G}}$ after different preparation steps. A Dirac cone can be observed in every panel, demonstrating the effectiveness of each intercalation approach. The straight lines depict the fitted band dispersions. The dashed lines indicate the Dirac energies. (d) Corresponding energy distribution curves (EDC's) at $\text{\textGamma}$. The strong contributions observed from 1.0eV towards higher binding energies are ascribed to the SiC valence bands. Only the preparation techniques A and B lead to the additional appearance of the surface state denoted $\text{D}_\text{Si}$ at 0.6eV, which is attributed to the lower Hubbard band of the Mott insulator that formed at the interface.
• Figure 3: Electronic and geometric structure of Si at the interface between graphene and SiC. (a)-(b) Energy distribution curves along the $\overline{\text{K}_{\surd{3}}\text{\textGamma}\text{M}_{\surd{3}}}$ direction at different $k_\|$ values for Si intercalated graphene (G/Si/SiC) (a) and the Si rich ($\surd{3} \times \surd{3}$)$R30^{\circ}$ reconstruction without graphene (Si($\surd{3}$)/SiC) (b), both on 4H-SiC(0001). Contributions from the substrate valence band ($\text{VB}_{\text{SiC}}$) are highlighted with orange arrows. The features marked by black lines ($\text{P}_{\text{1,2}}$) are associated with interactions between Si adatoms and the substrate. The green (blue) line traces the dispersion of the lower Hubbard band $\text{D}_{\text{Si}}$ ($\text{D}_{\surd{3}}$), attributed to dangling bonds of Si adatoms at the interface. (c) Dispersion of the lower Hubbard bands $\text{D}_{\text{Si}}$ and $\text{D}_{\surd{3}}$. The lower binding energy for the Si intercalated sample is attributed to screening by the conduction electrons in graphene. Vertical lines indicate high-symmetry points of the ($\surd{3} \times \surd{3}$)$R30^{\circ}$ (blue) and ($3\times3$) (red) Brillouin zones, as depicted in (d). Refer to the main text for detailed discussion. (e) SPA-LEED measurement ($E=$150eV) of a Si intercalated graphene sample prepared with technique B. Selected reciprocal lattice vectors for graphene (G), SiC, and the ($\surd{3} \times \surd{3}$)$R30^{\circ}$ and ($3\times3$) superstructures are depicted as black arrows. (f) Line scans along the [11$\overline{\text{2}}$0] direction of SiC for G/Si/SiC (solid, green) and Si($\surd{3}$)/SiC (dashed, blue). (g) Line scans along the [1$\overline{\text{1}}$00] direction of SiC for G/Si/SiC (solid, green) and the Si rich ($3\times3$) reconstruction without graphene (Si($3\times3$)/SiC) (dashed, red).