Imprint of transition metal $d$ orbitals on a graphene Dirac cone

We investigate the influence of SiO${}_2$, Au, Ag, Cu, and Pt substrates on the Raman spectrum of graphene. Experiments reveal particularly strong modifications to the Raman signal of graphene on platinum, compared to that of suspended graphene. The modifications strongly depend on the relative orientation of the graphene and platinum lattices. These observations are theoretically investigated and shown to originate basically from hybridization of electronic states in graphene and $d$ orbitals in platinum. It is expected that, quite generally, hybridization between graphene and any material with $d$ orbitals near the Fermi level will result in an imprint on the graphene Dirac cone, which depends sensitively on the relative orientation of the respective lattices.

Figure 1

The Raman signal for graphene on platinum is strongly suppressed compared to suspended graphene, and graphene on SiO${}_2$, Au, Ag, and Cu (514 nm laser, 10 s integration). The data on Pt are from 5000 s integration time and amplified by 100 times for better viewing. The small peak at ∼1554 cm${}^{-1}$ near the $G$ peak on Pt is from environment oxygen.[38] The numbers (black) below the substrate labels are the predicted peak intensity based only on screening [Eq. (1)]. The numbers above each curve are the experimentally measured peak intensities. The intensity is defined as the area covered by the peak and all the intensity values are normalized to the ones from suspended graphene.

Figure 2

The graphene Dirac cone is strongly affected by a Pt substrate, and varies substantially between the three most common misorientations of the graphene and Pt lattices (orientations α, β, and γ[29] correspond to 2 × 2, 3 × 3, and 4 × 4 graphene lattice supercells). A comparison of the Au substrate where the cone structure is preserved in this energy range can be found in Supplemental Material.[17] The thickness of each line in the plot is proportional to the graphenelike character of the state. The color of each line segment is proportional to the mixture of graphene states with different metallic $d$ orbitals in the topmost layer of the Pt slab substrate. Red, green, and blue color components correspond to three different projections of the angular momentum perpendicular to the platinum surface ($m$ = 0, +/−1, or +/−2, respectively). Graphene states with no $d$ character are colored white. The path in reciprocal space for all three misorientations is along the Γ-$K$-$M$ lineof the primitive graphene Brillouin zone.

Figure 3

Reduction of the graphene Raman (laser energy: 1.96 eV) 2D signal intensity upon hybridization with a metallic flat band (with only one metallic orbital per graphene unit cell). (a)−(c) Reduction of the Raman 2D signal depending on the position of the metallic orbital with respect to the graphene lattice (different panels), orbital character (different line colors and styles), and energy of the metallic band relative to the Dirac cone (horizontal axis). Reduction of the Raman 2D signal is almost negligible for the $s$ orbital, even though its head-to-head matrix element is of the same order of magnitude as for the $d$ orbital [both in σ and πorientation; see (d)]. The effect of $d$ orbital hybridization is strongly dependent on the $d$ orbital character.

Figure 4

Left: Raman spectra of graphene measured at different locations on a polycrystalline Pt foil. Laser wavelength: 488 nm. The intensity maps of a 30 × 25 μm region are shown at the bottom. Right: Raman spectra for graphene on single-crystal Pt(111) with 2 × 2 graphene supercell (also shown in Fig. 2). In this case, the intensity for both the 2D and $G$ peaks is found to be uniform across the single-domain sample.