Epitaxial growth of graphene on Ir(111) by liquid precursor deposition

The epitaxial growth of graphene on the surface of an Ir/YSZ/Si(111) multilayer substrate via the deposition of a liquid carbon precursor (acetone) was investigated by x-ray photoelectron spectroscopy, x-ray photoelectron diffraction, low-energy electron diffraction (LEED), and Fermi surface mapping. It is shown that the onset of graphene formation starts in a low temperature range around 600 K and that subsequent annealing up to 1000 K finally results in well-ordered graphene monolayers. Comparison of temperature-dependent LEED data with model calculations suggests that the growth of graphene takes place via a backbonelike growth by the formation of a hexagonal network connecting the hcp and fcc configuration sites within the $\sim$10 × 10/9 × 9 supercell. In LEED, the low intensities of the superstructure related satellite spots give evidence for only small corrugations of the graphene layer due to weak interaction with the Ir(111) surface, making graphene on Ir(111) similar to free-standing graphene with the Fermi surface providing distinct spots at the K points.

Figure 1

(a) Schematic drawing of the experimental setup. For angular resolved experiments, such as XPD, LEED, and FSM, the sample can be rotated by the polar angle ϑ and the azimuthal angle φ for mapping the whole 2π hemisphere above the surface of the sample by a fixed energy analyzer. (b) Schematic drawing of the Ir/YSZ/Si(111) multilayer substrates. (c) Unit cells of the Ir(111) topmost layer and of graphene.

Figure 2

XPS data (Al-K${}_{\alpha }$, $\hslash$ω = 1486.6 eV, normal emission ϑ = 0°) for several stages of graphene formation on the Ir/YSZ/Si(111) substrate. (a) Survey spectrum, (b) O-1s spectra, and (c) C-1s spectra. (From bottom to top) Clean Ir surface, after deposition of acetone precursor, after first annealing for 15 h at 483 K, after several steps of annealing (cf. temperature ramping in Fig. 7) with final annealing at 978 K. Satellites from Ir-4$d_{5/2}$ (∼295 eV) by Al-K${}_{\alpha 3}$ and Al-K${}_{\alpha 4}$ contributions (due to non-monochromatic x rays) in the range of C-1s (∼284.5 eV) have been eliminated numerically by shifting/weighting the spectra by 9.8 eV/6.4% and 11.8 eV/3.2%, respectively, and subtracting them from the raw data. In case of the O-1s binding energy range, the intensities displayed in the spectra for the clean sample and after graphene formation represent rather contributions from the energy loss background (e.g., plasmons) of the Ir-4$p_{3/2}$ line than contributions from O-1s. In the survey spectra, the absence of oxygen was also cross-checked by the missing O-KVV Auger intensities.

Figure 3

Estimation of the thickness of the carbon layer by attenuation of the Ir-4f photoelectron intensity in XPD. (a) Angular Ir-4f intensity distribution for clean Ir/YSZ/Si(111) substrate (Al-K${}_{\alpha }$ radiation, $\hslash$ω = 1486.6 eV); (b) same as (a) after deposition of acetone precursor; (c) same as (a) after graphene formation at 978 K; (d) azimuthally averaged Ir-4f polar intensity distribution ⟨I(ϑ,d)⟩ for the clean Ir/YSZ/Si(111) surface (black, ⟨I(ϑ)⟩${}_0$), after deposition of acetone precursor (blue) and after graphene formation (red), as extracted from (a)–(c); (e) intensity ratio ⟨I(ϑ,d)⟩/⟨I(ϑ)⟩${}_0$ as extracted from (d) after deposition (blue) and after graphene formation (red). The gray solid lines represent the ratio ∼exp(-d/λcosϑ) for d/λ = 0 to d/λ = 1 (from outside to inside in steps of d/λ = 0.1). After graphene formation, the experimental data are very close to the red line that represents the intensity ratio for one monolayer graphene with d/λ = 0.137 for electron mean free path λ = 2.44 nm[30] and a monolayer thickness of d = 0.335 nm, as given by the interlayer spacing of graphite.

Figure 4

LEED patterns ($E_0$ = 45 eV) for (a) a clean Ir/YSZ/Si(111) surface and (b) after graphene formation. In (b), satellite spots just appear around the (0,0) spot. The weak satellite intensities around the (1,0) spots are not visible in this representation (see text).

Figure 5

Polar intensity plots (LEED, $E_0$ = 45 eV) along the [$\overline{1}\overline{1}2$] symmetry axis of the substrate for one monolayer graphene on (a) a weakly interacting Ir/YSZ/Si(111) surface and on (b) a strongly interacting Rh/YSZ/Si(111) surface from Ref. 17. Peaks are labeled in fractions of reciprocal lattice units according to the size of the supercell. Satellite intensities depend on the corrugation of the epitaxial layer (cf. Appendix). The insets schematically show for a linear 4/3 chain the local variation of orbital overlap for large and small interface spacing. For large distances, as in (a), variation of overlap is small, resulting in small corrugation and small satellite intensities, and vice versa in (b).

Figure 6

The surface structure of graphene on Ir(111) can be approximated by a commensurate 10 × 10/9 × 9 supercell with a corrugation of about 0.45 Å,[12] as caused by the enhanced bonding strength at fcc and hcp sites and the reduced bonding strength at top sites [note that the terms “fcc,” “hcp,” and “top,” (cf. red circles), refer to the positions of the centers of the honeycomb, not to the positions of the carbon atoms].

Figure 7

Temperature-dependent LEED profiles ($E_0$ = 45 eV) along the [$\overline{1}\overline{1}2$] direction. The first graphene-related intensities are observed at 583 K. For details, see text.

Figure 8

Schematic representation of the growth models for graphene on Ir(111). In (a), disorder is uniformly distributed throughout the supercell. The displacement of the carbon atoms from the positions of the ideal corrugated graphene layer[12] is described by a displacement amplitude |δ$\underline{r}$| that is weighted with a random number in the range (−1 … 1). The displacement amplitude does not depend on the position within the supercell. In (b), the displacement amplitude depends on the position within the supercell, i.e., |δ$\underline{r}$(x,y)|, and decreases when going from the less stable top sites to the stable fcc or hcp sites. In both cases, the displacement amplitudes decrease with increasing order (i.e., increasing temperature in experiment) and drop to zero for the perfectly ordered graphene layer with a corrugation according to Ref. 12.

Figure 9

Simulation of LEED profiles along [$\overline{1}\overline{1}2$] for different stages of order according to the growth model from Fig. 8a with disorder distributed uniformly throughout the supercell. With increasing order (i.e., temperature) the displacement amplitude decreases from (a)–(e). The surface models also represent the finite clusters with which the simulation was calculated. Intensities are normalized to (0,0) intensity.

Figure 10

Intensity ratios I${}_{(10/9,0)}$ : I${}_{(1/9,0)}$ from (a) the experimental LEED data (Fig. 7) and from the LEED simulations for (b) homogeneous ordering (Fig. 9) and for (c) backbonelike ordering (Fig. 11). The experimental intensity ratios in (a) follow the data in (c) rather than the data in (b).

Figure 11

Simulation of LEED profiles along [$\overline{1}\overline{1}2$] for different stages of order according to the growth model from Fig. 8b with disorder depending on the position within the supercell. Ordering starts with a narrow network along the fcc and hcp sites, as in (a), and with increasing order or temperature, [(b)–(e)], this network is expanded over the whole supercell. Here, disordered areas are represented by white circles. The surface models also represent the finite clusters with which the simulation was calculated. Intensities are normalized to (0,0) intensity.

Figure 12

LEED pattern of (a) an in situ prepared (978 K) monolayer of graphene on Ir(111), (b) after storage for several weeks under ambient conditions (with diffuse LEED spots still visible), and (c) after regeneration in UHV by annealing at 973 K for several hours.

Figure 13

(a) Band structure of graphene as derived from the most simple tight-binding ansatz[1] with the Dirac cones meeting at the K points of the Brillouin zone; (b) Fermi surface mapping (He I radiation) for one monolayer graphene on Ir(111) with distinct intensities at k${}^{\parallel }$ ∼ 1.7 Å${}^{-1}$ at the corners of the Brillouin zone [cf. K points in (a) are at 4π/3a = 1.703 Å${}^{-1}$ for a = 2.46 Å]; (c) Fermi surface mapping of the clean Ir/YSZ/Si(111) surface as a reference. The smaller hexagon corresponds to the surface Brilluoin zone of the substrate.

Figure 14

LEED simulation for a linear chain with a commensurate (blue) and an incommensurate (red) epitaxial layer for different corrugations. In the model structure, the full dots and open circles in the corrugated layer represent the positions for the commensurate and incommensurate case, respectively.

Figure 15

LEED simulation for a linear chain with a commensurate (blue) and an incommensurate (red) flat epitaxial layer for spatial variation of the scattering factors, as represented by the size of the atoms in the model structure.