C. Coletti, C. Riedl, D.S. Lee, B. Krauss, K. von Klitzing, J.H. Smet, U. Starke and L. Patthey1
Epitaxial graphene grown on silicon carbide (SiC) offers realistic prospects for large scale graphene samples. However, the as-grown graphene layers are electron doped as a result of the graphene/SiC interface properties. Ac-cordingly the Fermi energy, EF, is displaced away from the Dirac point energyED. This sit-uation is visualized by a schematic sketch of the π-band dispersion around the Dirac point in Fig. 66(a). In order to exploit many of the unique properties of graphene as semimetal also in SiC based epitaxial graphene, the intrin-sic doping has to be reversed, which means that the electrons have to be extracted out of the graphene layer. In order to reach charge neutrality (ED≈EF) for epitaxial graphene, p-doping must compensate the intrinsic n-doping. This situation is sketched in Fig. 66(b), whereas panel (c) exemplifies the true p-type regime in order to round up the pic-ture. In the present report we show that the excess negative charge in epitaxial graphene can be fully compensated by functionalizing
the graphene surface with the strong elec-tron acceptor (elecelec-tron affinity Eea= 5.24 eV) molecule tetrafluoro-tetracyanoquinodimethane (F4-TCNQ).
Figure 66: Position of Dirac point and Fermi level for monolayer epitaxial graphene on SiC(0001) as a function of doping. Panel (a) stands for the intrinsic n-doping of an as grown monolayer, panel (b) for a charge neutral monolayer, and panel (c) visualizes truly p-doped monolayer graphene.
Epitaxial graphene was grown in ultrahigh vac-uum (UHV) by thermal Si sublimation on hy-drogen etched, atomically-flat 6H-SiC(0001) crystals. The quality of the graphene layers was characterized using low-energy electron diffraction (LEED). The doping level of the
graphene layers was monitored with angle re-solved photoemission spectroscopy (ARPES) measurements of theπ-band dispersion around the K-point of the graphene Brillouin zone.
In-house measurements were carried out using monochromatic He II radiation (hν= 40.8 eV) from a UV discharge source. As shown on the left side of Fig. 67 for an as-grown mono-layer of graphene on SiC(0001) the Fermi level EF is located about 0.42 eV above the Dirac point ED. This corresponds to the well-established charge carrier concentration value ofn≈1⋅1013cm−2for as grown graphene. Sub-sequently, F4-TCNQ molecules were deposited on the graphene/SiC substrates by thermal evaporation from a resistively heated crucible.
For increasing amounts of deposited F4-TCNQ EFmoves back towardsED indicating electron transfer from the graphene to the molecular layer. After deposition of a 0.8 nm thick layer of molecules, charge neutrality is reached, i.e., EF=ED (right side of Fig. 67). For a nominal thickness of the molecular film above 0.8 nm no additional shift of the Fermi energy is ob-served which indicates that the charge transfer saturates. Additional core level photoemission and integrated valence band spectroscopy mea-surements indicated that only two of the four
C≡N groups of the molecule are involved in the charge transfer process [1]. This points to a model of upright standing molecules on the sur-face as indicated in the center sketch of Fig. 67.
We note, however, in addition, that the fluori-nation of the molecule is essential for reaching charge neutrality, since with normal TCNQ the Dirac energy is shifted only by 170 meV [1].
For a detailed quantitative determination of the carrier concentrations, Fermi surface maps were extracted from high-resolution ARPES data acquired using synchrotron radiation at the Swiss Light Source (SLS) of the Paul Scherrer Institut (PSI), Switzerland, at the Surface and Interface Spectroscopy beamline (SIS). Fig-ure 68 compares the π-band dispersion (a)–(c) and constant energy maps (d)–(f) at EF for a clean graphene monolayer (a),(d), an interme-diate F4-TCNQ coverage (b),(e) and charge transfer saturation at full coverage (c),(f). The charge carrier concentration can be derived pre-cisely from the size of the Fermi surface pock-ets as n=(kF–kK)2/π, where kK denotes the wavevector at the boundary of the graphene Brillouin zone. The Fermi surface pocket radius is extracted by using Lorentzian fits of the max-ima of the momentum distribution curves of the electronic dispersion spectra in panels (a)–(c).
Figure 67: Dispersion of theπ-bands measured by UV excited ARPES around the K point of the graphene Brillouin zone for an as-grown graphene monolayer on SiC(0001) (left side) and for a charge neutral same sample after deposition of a F4-TCNQ molecule film of 0.8 nm thickness (right side). The momentum scans are taken perpendicular to theΓK-direction in reciprocal space. The schematic structure of F4-TCNQ molecules deposited on top of a graphene layer grown on SiC is shown in the central sketch.
Figure 68: Dispersion of theπ-band around the K point of the graphene Brillouin zone measured by ARPES with synchrotron light in scans oriented par-allel to theΓK-direction for (a) a pristine epitaxial graphene monolayer, (b) an intermediate F4-TCNQ coverage and (c) the F4-TCNQ coverage leading to charge neutrality. Panels (d) through (f) show the corresponding constant energy maps at EF. From these Fermi surface maps we extract a charge carrier concentration of (7.3±0.2)⋅1012cm−2for the pris-tine graphene, (9±2)⋅1011cm−2for the intermedi-ate coverage and (1.5±2)⋅1011cm−2for full cover-age. All the spectra shown were acquired with cir-cular polarized light with a photon energy of 30 eV and at a sample temperature of 80 K.
The corresponding carrier concentrations are 7.3⋅1012cm−2, 9⋅1011cm−2 and 1.5⋅1011cm−2 for the clean graphene monolayer, the
interme-diate and the higher coverage, respectively. The error bar for the resulting carrier concentrations is±2⋅1011cm−2 and was determined from the variance of the Lorentzian fits.
Graphene bilayers exhibit a more complex π -band structure. Two branches are visible for both conduction and valence band. The band shift caused by the intrinsic n-doping of epitax-ial graphene bilayers on SiC is slightly lower than in the case of monolayers, namely about 0.3 eV. In addition, the electric dipole present at the graphene/SiC interface imposes an elec-trostatic asymmetry between the layers which causes a bandgap to open by roughly 0.1 eV as seen from the ARPES data in Fig. 69(a). In the figure bands obtained from tight-binding calculations are superimposed to the dispersion plot. This facilitates an analytical evaluation of the Dirac energy position and the size of the bandgap. The calculations are based on a sym-metric bilayer Hamiltonian as described by Mc-Cann and Fal’ko [2]. Similar to the monolayer case, F4-TCNQ deposition onto this sample causes a progressive shift of the bilayer bands, i.e., a reduction of the intrinsic n-type doping.
This is illustrated in the measured and calcu-lated dispersion plots in Fig. 69(b)–(d). Concur-rent with the drop of EF–ED, the size of the bandgap increases as seen from the bands fitted with the tight binding simulations.
Figure 69: ARPES band structure plots measured perpendicular to the ΓK-direction for an epitaxially grown graphene bilayer on SiC(0001) (a) without F4-TCNQ coverage and (b)–(d) with increasing amounts of F4-TCNQ. Bands calculated within a tight binding model are superimposed to the experimental data. (e) Evolution of the energy gapEg, the gap midpoint or Dirac pointED, the minimum of the lowest conduction bandEcondand the maximum of the uppermost valence bandEvalas a function of molecular coverage. The definition of the energies is included in panel (c).
The band fitting retrieves the energy at the bot-tom of the lowest conduction band Econd and at the top of the uppermost valence bandEval. From these values the energy gap Eg and the mid gap or Dirac energy ED are derived. The corresponding energies are marked in panel (c). The evolution of the characteristic ener-gies of these fitted bands with the amount of deposited molecules is plotted in Fig. 69(e).
The bandgap Eg increases from 116 meV for a clean as-grown bilayer to 275 meV when a 1.5 nm thick layer of F4-TCNQ molecules has been deposited. We verified that no further charge transfer occurs for higher amounts of deposited molecules. The Fermi energy moves into the bandgap for a molecular layer thickness of 0.4 nm. Hence the bilayer is turned from a conducting system into a truly semiconducting layer. The increase of the bandgap indicates that the molecular deposition increases the on-site Coulomb potential difference between both lay-ers. From the tight binding calculations we get an increase in the on-site Coulomb interaction from 0.12 eV for a clean bilayer to 0.29 eV for a bilayer with a molecular coverage of 1.5 nm.
This increase can be attributed to an increased electrostatic field due to the additional dipole developing at the graphene/F4-TCNQ interface.
The charge neutralization in the graphene lay-ers by F4-TCNQ deposition was corroborated by Raman spectroscopy analysis of the shift of the G-phonon peak position. By monitor-ing peaks in the Raman spectra related to the molecule itself we noted that during laser illu-mination the deposited molecules are gradually removed by evaporation accompanied by a si-multaneous shift of the G-peak. Laser heating can therefore be used to trim the molecule cov-erage and hence tune the charge carrier concen-tration in graphene. In a confocal arrangement it should therefore be possible to spatially mod-ulate the doping level [1]. The notable fact, that the molecules remain stable under ambient con-ditions and at elevated temperatures, as well as the possibility to applied the molecular layer via wet chemistry, make this doping method very attractive since its incorporation into existing technological processes appears feasible [1]
References:
[1] Coletti, C., C. Riedl, D.S. Lee, B. Krauss, L. Patthey, K. von Klitzing, J.H. Smet and U. Starke.Physical Review B81, 235401 (2010).
[2] McCann, E. and V.I. Fal’ko.Physical Review Letters 96, 086805 (2006).
1Paul Scherrer Institute, Villigen PSI, Switzerland