Electronic decoupling of an epitaxial graphene monolayer by gold intercalation
Isabella Gierz, Takayuki Suzuki, Dong Su Lee, Benjamin Krauss, Christian Riedl, Ulrich Starke, Hartmut Höchst, Jurgen H. Smet, Christian R. Ast, Klaus Kern
aa r X i v : . [ c ond - m a t . m t r l - s c i ] M a r Electronic decoupling of an epitaxial graphene monolayer by gold intercalation
Isabella Gierz , ∗ Takayuki Suzuki , Dong Su Lee , Benjamin Krauss , Christian Riedl ,Ulrich Starke , Hartmut H¨ochst , Jurgen H. Smet , Christian R. Ast , and Klaus Kern , Max-Planck-Institut f¨ur Festk¨orperforschung, D-70569 Stuttgart, Germany Synchrotron Radiation Center, University of Wisconsin-Madison, Stoughton, WI, 53589, USA Institut de Physique de la Mati`ere Condens´ee, Ecole Polytechnique F´ed´erale de Lausanne, CH-1015 Lausanne, Switzerland (Dated: October 29, 2018)The application of graphene in electronic devices requires large scale epitaxial growth. The presence of thesubstrate, however, usually reduces the charge carrier mobility considerably. We show that it is possible todecouple the partially sp -hybridized first graphitic layer formed on the Si-terminated face of silicon carbidefrom the substrate by gold intercalation, leading to a completely sp -hybridized graphene layer with improvedelectronic properties. Electrons in graphene — sp -bonded carbon atoms ar-ranged in a honeycomb lattice — behave like massless Diracparticles and exhibit an extremely high carrier mobility [1].So far, the only feasible route towards large scale productionof graphene is epitaxial growth on a substrate. The presenceof the substrate will, however, influence the electronic prop-erties of the graphene layer. To preserve its unique proper-ties it is desirable to decouple the graphene layer from thesubstrate. Here we present a new approach for the growth ofhighly decoupled epitaxial graphene on a silicon carbide sub-strate. By decoupling the strongly interacting, partially sp -hybridized first graphitic layer (commonly referred to as zerolayer (ZL) [2]) from the SiC(0001) substrate by gold interca-lation, we obtain a completely sp -hybridized graphene layerwith improved electronic properties as confirmed by angle-resolved photoemission spectroscopy (ARPES), scanning tun-neling microscopy (STM) and Raman spectroscopy.There are essentially two ways for large scale epitaxialgrowth of graphene on a substrate: by cracking organicmolecules on catalytic metal surfaces [3–7] or by thermalgraphitization of SiC [2, 8–11]. Unfortunately, the presenceof the substrate alters the electronic properties of the graphenelayer on the surface and reduces the carrier mobility. Eventhough it has been shown that the graphene layer can be de-coupled from a metallic substrate [6, 12–14] the system re-mains unsuitable for device applications. This problem canbe solved by decoupling the graphene layer from a semicon-ducting SiC substrate [15].On both the silicon and the carbon terminated face of a SiCsubstrate, graphene is commonly grown by thermal graphi-tization in ultra high vacuum (UHV). When annealing thesubstrate at elevated temperatures Si atoms leave the sur-face whereas the C atoms remain and form carbon layers.On SiC(000 ), the so-called C-face, the weak graphene-to-substrate interaction results in the growth of rotationally dis-ordered multilayer graphene and a precise thickness controlbecomes difficult [16]. On the other hand, the rotational disor-der decouples the graphene layers so that the transport prop-erties resemble those of isolated graphene sheets with roomtemperature mobilities in excess of 200,000 cm /Vs [17].On SiC(0001), i. e. the Si-face, the comparatively stronggraphene-to-substrate interaction results in uniform, long- range ordered layer-by-layer growth. The first carbon layer(=ZL) grown on the Si-face is partially sp -hybridized to thesubstrate, which means that on a ZL no π -bands can developand it has no graphene properties. This can be seen in thefirst panel of Fig. 1 a), where the experimental band structureof the ZL (black) measured by ARPES near the K-point ofthe surface Brillouin zone is shown. The ZL lacks the lin-ear dispersion typical for graphene π -bands. Its band struc-ture consists of two non-dispersing bands at about − . eVand − . eV initial state energy. In addition, the ZL formsa ( √ × √ )R30 ◦ reconstruction with respect to the SiCsubstrate [2, 16, 18].Further graphitization leads to the growth of a completelysp -hybridized graphene layer, for which the ZL acts as abuffer layer. The band structure of this “conventionally”grown graphene monolayer (cML) near the K-point is shownin the second panel of Fig. 1 a). The cML is influenced con-siderably by the underlying SiC substrate. It is n-doped withthe crossing point of the two linear bands (Dirac point) at E D = − meV due to charge transfer from the substrate[8–10, 15]. Furthermore, the possibility of a band gap openinghas been suggested [10] and explained theoretically in con-nection with the formation of midgap states [19]. In additionto that, the strong substrate influence reduces the carrier mo-bility considerably [20]. The ( √ × √ )R30 ◦ reconstruc-tion of the ZL diffracts the outgoing photoelectrons giving riseto the formation of replica bands [2]. This is nicely seen in themeasured Fermi surface of the cML around K in the left panelof Fig. 1 b). The size of the Fermi surface is determined by thecharge carrier density n = k F /π , where k F is the Fermi wavevector with respect to the K-point. The values are summarizedin Table I.To reduce the influence of the substrate we developed a newmethod for the epitaxial growth of graphene on the Si-face ofSiC. We start with the preparation of the ZL exploiting thestrong substrate influence for uniform growth. On top of theZL, we deposit Au atoms at room temperature. After subse-quent annealing of the sample at 800 ◦ C the linear dispersiontypical for graphene appears. Depending on the gold cover-age (about one third or one monolayer, respectively), eithera strongly n-doped (nML Au ) or a p-doped (pML Au ) graphenelayer is formed. The band structures for the pML Au and the -0.2 0.0 0.2nML Au -0.2 0.0 0.2pML Au -0.2 0.0 0.2cML -0.4 -0.2 0.0 0.2 0.4nML Au -2.0-1.5 W a v e V e c t o r k y ( Å - ) -0.4 -0.2 0.0 0.2cML ab Wave Vector k x (Å -1 ) Wave Vector k x (Å -1 ) -1.0-0.50.0 I n i t i a l S t a t e E ne r g y ( e V ) -0.2 0.0 0.2ZL -0.4 -0.2 0.0 0.2 0.4pML Au FIG. 1: (color online) Comparison of ARPES data for conventional graphene on SiC and graphene intercalated with Au: panel a) shows theband structure measured in the direction perpendicular to the Γ K direction near the K-point of the surface Brillouin zone of the zero layer(black), the conventional graphene monolayer (red), the p-doped graphene monolayer intercalated with gold (blue) and the n-doped graphenemonolayer intercalated with gold (green) together with the corresponding Fermi surfaces in panel b). The Fermi surfaces are plotted on alogarithmic color scale to enhance weak features. k x is perpendicular to the Γ K direction, k y is along the Γ K direction. The Fermi surface forthe p-doped graphene monolayer shows a weak contribution of the n-doped phase due to an inhomogeneous Au coverage on the sample. nML Au are compared in Fig. 1 a). In contrast to the ZL, boththe pML Au (blue) and the nML Au (green) clearly show twolinearly dispersing π -bands. The Dirac point for the pML Au is about 100 meV above the Fermi level. This band structurelooks similar to the one reported in Ref. [21]. However, therethe graphene monolayer was prepared by depositing Au di-rectly on a cML and not on a ZL as in this work. For thenML Au the bands cross at about − meV. The band struc-ture of the cML is a superposition of the band structure of theunderlying ZL and the graphene monolayer. Both pML Au andnML Au , however, are formed directly from the ZL. There isno additional carbon layer between the graphene layer and thesubstrate. Therefore, the band structure around the K-pointis given by the pML Au and nML Au alone. The charge carrierdensities deduced from the size of the Fermi surface (see mid-dle and right panel of Fig. 1 b) are listed in Table I.Comparing the Fermi surfaces for the cML (red), the pML Au (blue), and the nML Au (green) in Fig. 1 b), the moststriking difference is the absence of replica bands for thepML Au and nML Au . Even on the logarithmic color scaleof Fig. 1 b) the replica bands are invisible, indicating a re-duced influence of the ( √ × √ )R30 ◦ reconstruction.Low energy electron diffraction (LEED) images (shown in theEPAPS) reveal a strong decrease of the intensity for spots re-lated to the ( √ × √ )R30 ◦ reconstruction for the pML Au as compared to the graphene-related spots. For the nML Au ,however, the ( √ × √ )R30 ◦ spots have a similar intensityas for the cML. We conclude that only the pML Au is less in-fluenced by the underlying substrate. We attribute this to anincreased graphene-to-substrate distance as will be discussedlater in this paper.To analyze the band structure in more detail and gain ac-cess to the relevant scattering mechanisms we determined thefull width at half maximum (FWHM) of the bands by fitting F W H M M DC ( Å - ) -2.0 -1.5 -1.0 -0.5 0.0Initial State Energy (eV)nML Au cMLpML Au E D E D I n t en s i t y ( a r b . un i t s )
92 90 88 86 84 82Binding Energy (eV) pML Au nML Au Au-Si Au-Au Au Au a b c SiCAu-SiML grapheneAu cluster
FIG. 2: (color online) Linewidth analysis, Au core level spectra and schematic: Panel a) shows the full width at half maximum (FWHM)of momentum distribution curves obtained from Fig. 1 a) for the conventional graphene monolayer (red), the p-doped graphene monolayerintercalated with Au (blue) and the n-doped graphene monolayer intercalated with Au (green). A constant background was subtracted from thedata so that the plotted linewidth is determined by electron-phonon, electron-plasmon and electron-electron scattering alone. Panel b) showsthe Au f core level spectra recorded with an incident photon energy of 150 eV for the p-doped monolayer (blue) and the n-doped monolayer(green). The core level spectra indicate the presence of Au-Si bonds (black lines) for both the p- and the n-doped monolayer which is consistentwith the structural model shown in panel c).TABLE I: Characteristic parameters for cML, pML Au andnML Au determined from the photoemission experiments.cML pML Au nML Au Au coverage (ML) 0 1 1/3Au-Si 4f / (eV) 88.20 89.05Au-Si 4f / (eV) 84.54 85.41Au-Au 4f / (eV) 87.82 88.32Au-Au 4f / (eV) 84.15 84.68charge carrier 1 × × × density (cm − ) electrons holes electronsDirac point (meV) −
420 +100 − momentum distribution curves (MDCs) along the Γ K direc-tion with Lorentzian lineshapes and a constant background.The FWHM as a function of the initial state energy for thecML (red), the pML Au (blue) and the nML Au (green) areshown in Fig. 2 a). From the data in Fig. 2 a) a constant off-set of 0.023 ˚A − (cML), 0.027 ˚A − (pML Au ), and 0.041 ˚A − (nML Au ) has been subtracted. For both cML and pML Au thisoffset is mainly determined by the experimental resolution(see EPAPS). For the nML Au , however, the linewidth offsetis significantly larger than the limit set by the experimentalresolution. In this case the offset is determined by impurityscattering which gives a constant contribution to the linewidthat all energies.There are three main contributions to the quasiparticle life-time in graphene [2, 8]. The increase in linewidth around200 meV is caused by electron-phonon coupling which de- pends on the size of the Fermi surface. Therefore, its influ-ence is largest for strongly n-doped graphene (nML Au ). Thepronounced maximum near the Dirac point is attributed toelectron-plasmon scattering. The third contribution to the linewidth is electron-electron scattering, which has been found tobe proportional to | E − E F | α , where < α < [8]. TheFWHM for our cML is in good agreement with the data re-ported in Ref. [2, 8]. Also, the cML and the nML Au havea similar linewidth. The main difference between the two isthe position of the plasmon peak which is determined by theposition of the Dirac point and hence the doping level. ThepML Au , however, has a much lower linewidth over the wholerange of energies indicating a reduced electron-electron scat-tering. As the Fermi surface for the pML Au is rather small (seeFig. 1 b) the electron-phonon contribution to the linewidth isnegligible. The local maximum in linewidth around − eVinitial state energy for the pML Au is not located at the Diracpoint. Therefore, we do not interpret this as originating fromplasmons within the graphene layer according to [2, 8]. Wetentatively attribute this peak to plasmons localized mainlyin the Au clusters on top of the pML Au that interact withthe electrons in the graphene layer. Varykhalov et al. [6]found a similar feature for graphene/Au/Ni(111) which theyattributed to an interaction between Au and graphene. Theoverall much smaller linewidth for the pML Au corroboratesthe conclusion from LEED that the pML Au is decoupled fromthe substrate. As mentioned before, the measured linewidthfor the pML Au near the Fermi level is mainly determined bythe experimental momentum resolution of ∆ k = 0 . ˚A − .This allows us to estimate a lower limit for the carrier lifetimeusing τ = ~ / ( ~ v F ∆ k ) . With ~ v F = 7 . eV ˚A, we find that τ > fs which is the same order of magnitude as the valuereported for multilayer graphene on the C-face of SiC [17].To gain a deeper insight into the structure of the pML Au and nML Au , we measured the Au f core level spectra us-ing a photon energy of 150 eV. The spectra in Fig. 2 b) forthe pML Au (blue) and the nML Au (green) show two differentcontributions to the Au f core level. The doublet at higherbinding energy was attributed to Au-Si bonds before [22, 23].The doublet at lower binding energy belongs to Au-Au bonds[24]. The peak positions are summarized in Table I.Combining these observations with the band structures inFig. 1, we can deduce a schematic as depicted in Fig. 2 c).The appearance of a linear dispersion typical for grapheneimplies that the C-Si bonds between ZL and substrate breakand a completely sp -hybridized carbon monolayer is created.The core level spectra show the existence of Au-Si bonds forboth the nML Au and the pML Au . We conclude that the Auatoms intercalate between the ZL and the substrate replacingthe C-Si bonds by Au-Si bonds. From the core level peak in-tensity for the nML Au , we find about one third monolayer ofAu intercalated (one monolayer corresponds to two Au atomsper graphene unit cell). This is consistent with the observationthat every third carbon atom in the ZL forms a C-Si bond [16].For the pML Au , about one monolayer of gold is intercalated.From atomic force microscopy (AFM - not shown here) andSTM measurements, we find that additional Au atoms are notintercalated, but form Au clusters on top of the graphene layer.Despite the fact that a complete monolayer of gold is interca-lated for the pML Au the substrate does not become metallic.Apart from the graphene bands, there are no other states visi-ble at the Fermi energy.The doping behavior for different Au coverages has beenaddressed by the theoretical work of Giovannetti et al. [25]who predicted p-type doping for graphene on a Au substrate.Reducing the Au-graphene distance to d AuG < . ˚A, how-ever, will lead to n-type doping. The larger amount of in-tercalated Au for the pML Au should increase the distance be-tween graphene and substrate. This is consistent with the ob-served doping behavior as well as the reduced influence ofthe ( √ × √ )R30 ◦ interface reconstruction on the Fermisurface and the LEED images of the pML Au .The peak position for the Au f doublet associated withAu-Si bonds shifts by about 860 meV from nML Au to pML Au .This can be related to the observed difference in the dopingand a small change of the work function. The Au-Au compo-nent, on the other hand, shifts only by about 520 meV. We at-tribute the Au-Au bonds to Au clusters on top of the graphenelayer. These clusters have an average size of a few nanome-ters (see Fig. 3 a). For such nanoparticles the position of thecore levels depends rather sensitively on the size of the parti-cle [26, 27]. Thus, the shift of the Au-Au component of the Au f core level is most likely related to the size of the particularAu clusters.As both LEED and ARPES average over a rather large areaon the sample surface, we used STM to gain access to thestructure of the surface on an atomic scale. Fig. 3 a) showstopographic images of the cML and the pML Au . The cML I n t en s i t y ( a r b . un i t s ) D /I G =0.27I D /I G =0.35 pML Au cML 280026002D 2685cm -1 -1 Raman Shift (cm -1 ) ab cML (6x6)nm pML Au (12x12)nm FIG. 3: (color online) STM images and Raman spectra reveal im-proved crystalline quality of the p-doped Au-intercalated graphene:panel a) shows topographic STM images for the conventionalgraphene monolayer (left) and the p-doped graphene monolayer in-tercalated with Au (right). The images for the conventional mono-layer and p-doped monolayer were recorded at a tunneling currentof 0.2 nA and a bias of − . V and − . V, respectively. Panel b)compares the Raman scattering results for the conventional (red) andthe p-doped (blue) graphene monolayer. shows a honeycomb lattice with a ( √ × √ )R30 ◦ modula-tion imposed by the ZL. The graphene lattice of the pML Au iswell ordered with single defects (black) and some gold clus-ters (white). The pML Au shows a superstructure of parallelstripes with a width of about 3 nm as marked by blue arrowsin the right panel of Fig. 3 a). This superstructure could beof similar origin as the one reported in [28] despite the factthat the samples in [28] were prepared by depositing Au on acML. The change of the lattice constant of the superstructurebetween cML and pML Au is also visible in LEED measure-ments (see EPAPS).To further investigate the degree of decoupling of thepML Au , Fig. 3 b) shows Raman scattering data measured forthe cML and the pML Au . The substrate contribution to the Ra-man data was subtracted from the spectra so that the graphenepeaks are clearly visible [29]. The Raman spectrum for theZL (not shown here) does not show any graphene related fea-tures. The 2D peak of the pML Au (blue) appears at 2685cm − . It is red-shifted by 50 cm − as compared to the 2Dpeak of the cML. As the 2D peak position is only weakly de-pendent on charge doping [30], we attribute the shift of the 2Dpeak to an increase of the lattice constant in agreement withthe LEED data (see EPAPS). The compressive strain presentin the cML is apparently released in the pML Au . This con-firms the strongly reduced interactions observed in the analy-sis of the ARPES linewidth. The data in Fig. 3 b) also suggestthat the D:G peak intensity ratio has decreased for the pML Au (blue). The D peak only exists in the presence of defects in thegraphene lattice. A reduced D:G peak intensity ratio thereforeindicates an improved crystalline quality.We have shown that it is possible to decouple the grapheneZL formed on the Si-face of SiC from the substrate by Auintercalation. This new slightly p-doped graphene has an im-proved quality and is only weakly influenced by the underly-ing substrate. Our ARPES measurements for the pML Au re-veal a considerable reduction in linewidth. Our estimation forthe carrier lifetime is of the same order of magnitude as thevalue for multilayer graphene on the C-face of SiC. There-fore, we expect a considerable increase in carrier mobility forthe pML Au and correspondingly the transport properties of ourpML Au to be closer to those for multilayer graphene on the C-face of SiC.The authors thank C. L. Frewin, C. Locke and S. E. Sad-dow of the University of South Florida for hydrogen etchingof the SiC substrates. C. R. A. acknowledges funding by theEmmy-Noether-Program of the Deutsche Forschungsgemein-schaft (DFG). This work is based in part upon research con-ducted at the Synchrotron Radiation Center of the Universityof Wisconsin-Madison which is funded by the National Sci-ence Foundation under Award No DMR-0537588. ∗ Corresponding author; electronic address: [email protected][1] A. K. Geim and K. S. Novoselov, Nature Mater. , 183 (2007)[2] A. Bostwick, T. Ohta, J. L. McChesney, K. V. Emtsev, T.Seyller, K. Horn and E. Rotenberg, New J. Phys. ,385 (2007)[3] A. B. Preobrajenski, M. L. Ng, A. S. Vinogradov and N.M˚artensson, Phys. Rev. B , 073401 (2008)[4] M. Sasaki, Y. Yamada, Y. Ogiwara, S. Yagyu and S. Yamamoto,Phys. Rev. B , 155653 (2000)[5] I. Pletikosi´c, M. Kralj, P. Pervan, R. Brako, J. Coraux, A.T. N’Diaye, C. Busse and T. Michely, Phys. Rev. Lett. ,056808 (2009)[6] A. Varykhalov, J. S´anchez-Barriga, A. M. Shikin, E. Vescovo,A. Rybkin, D. Marchenko and O. Rader, Phys. Rev. Lett. ,157601 (2008)[7] P. W. Sutter, J.-I. Flege and E. Sutter, Nature Mater. , 406(2008)[8] A. Bostwick, T. Ohta, T. Seyller, K. Horn and E. Rotenberg,Nature Phys. , 36 (2006) and references therein[9] T. Ohta, A. Bostwick, T. Seyller, K. Horn, K. and E. Rotenberg, Science , 951 (2006)[10] S. Y. Zhou, G.-H. Gweon, A. V. Fedorov, P. N. First, W. A. deHeer, D.-H. Lee, F. Guinea, A. H. Castro Neto and A. Lanzara,Nature Mater. , 770 (2007)[11] C. Berger, Z. Song, X. Li, X. Wu, N. Brown, C. Naud, D.Mayou,T. Li, J. Hass, A. N. Marchenkov, E. H. Conrad, P. N.First and W. A. de Heer, Science , 1191 (2006)[12] A. M. Shikin, G. V. Prudnikova, V. K. Adamchuk, F. Morescoand K.-H. Rieder, Phys. Rev. B , 13202 (2000)[13] D. Far´ıas, A. M. Shikin, K.-H. Rieder and Y. S. Dedkov, J. Phys.Condens. Matter , 8453 (1999)[14] Y. S. Dedkov, A. M. Shikin, V. K. Adamchuk, S. L. Molodtsov,C. Laubschat, A. Bauer and G. Kaindl, Phys. Rev. B , 035405(2001)[15] C. Riedl, C. Coletti,T. Iwasaki, A. A. Zakharov, and U. Starke,Phys. Rev. Lett. , 246804 (2009)[16] K. V. Emtsev, F. Speck, T. Seyller, L. Ley, and J. D. Riley, Phys.Rev. B ,155303 (2008)[17] M. Sprinkle, D. Siegel, J. Hicks, A. Tejeda, A. Taleb-Ibrahimi,P. Le F`evre, F. Bertran, S. Vizzini, H. Enriquez, S. Chiang, P.Soukiassian, C. Berger, W. A. de Heer, A. Lanzara and E. H.Conrad, Phys. Rev. Lett. , 226803 (2009)[18] F. Varchon, R. Feng, X. Li, B. Ngoc Nguyen, C. Naud, P. Mal-let, J.-Y. Veuillen, C. Berger, E. H. Conrad and L. Magaud,Phys. Rev. Lett. , 126805 (2007)[19] S. Kim, J. Ihm, H. J. Choi and Y.-W. Son, Phys. Rev. Lett. ,176802 (2008)[20] J. A. Robinson, M. Wetherington, J. L. Tedesco, P. M. Camp-bell, X. Weng, J. Stitt, M. A. Fanton, E. Frantz, D. Snyder, B.L. VanMil, G. G. Jernigan, R. L. Myers-Ward, C. R. Eddy andD. K. Gaskill, Nano Lett. , 2873 (2009)[21] I. Gierz, C. Riedl, U. Starke, C. R. Ast and K. Kern, Nano Lett. , 4603 (2008)[22] C. Virojanadara and L. I. Johansson, Surf. Sci. , 436 (2006)[23] D. Stoltz, S. E. Stoltz and L. S. O. Johansson, J. Phys. Condens.Matter , 266006 (2007)[24] J. F. Moulder, W. F. Stickle, P. E. Sobol and K. D. Bomben,Handbook of X-ray Photoelectron Spectroscopy (PhysicalElectronics, Inc., Minnesota, 1995)[25] G. Giovannetti, P. A. Khomyakov, V. M. Karpan,J. van denBrink and P. J. Kelly, Phys. Rev. Lett. , 026803 (2008)[26] H.-G. Boyen, A. Ethirajan, G. K¨astle, F. Weigl and P. Ziemann,Phys. Rev. Lett. , 016804 (2005)[27] P. Torelli, L. Giordano, S. Benedetti, P. Luches, E. Annese, S.Valeri and G. Pacchioni, J. Phys. Chem. C , 19957 (2009)[28] B. Premlal, M. Cranney, F. Vonau, D. Aubel, D. Casterman,M. M. De Souza and L. Simon, Appl. Phys. Lett. , 263115(2009)[29] D. S. Lee, C. Riedl, B. Krauss, K. von Klitzing, U. Starke, andJ. H. Smet, Nano Lett. , 4320 (2008)[30] J. Yan, Y. B. Zhang, P. Kim and A. Pinczuk, Phys. Rev. Lett. , 166802 (2007) EPAPS Available:
Details about the sample preparation,the different experimental techniques and the data analysis areavailable as EPAPS.
EPAPSSample preparation
We have grown graphene on the Si-face of SiC. Our 4H SiCwafers were hydrogen-etched before insertion into ultra highvacuum (UHV). To remove residual oxygen impurities we de-posited Si from a commercial electron beam evaporator at asubstrate temperature of 800 ◦ C until a sharp (3 ×
3) low energyelectron diffraction (LEED) pattern was observed. We graphi-tized the samples by direct current heating at elevated temper-ature. The sample temperature was measured with an opticalpyrometer at an emissivity of 63 % . An annealing temperatureof 1100 ◦ C for five minutes is sufficient for the formation of thezero layer (ZL), a pure carbon layer, where every third C-atomforms a chemical bond to a Si-atom in the layer below. ThisZL has no graphene properties, in particular, instead of thelinear dispersion at the K-point of the surface Brillouin zonethere are two non-dispersing bands at − . eV and − . eVinitial state energy. Upon further annealing at 1150 ◦ C for fiveminutes a purely sp -hybridized carbon layer forms on top ofthe ZL which shows the linear band structure characteristic ofmassless charge carriers in graphene. This graphene layer isreferred to as the conventional monolayer (cML) in the fol-lowing.We deposited gold from a commercial Knudsen cell at roomtemperature on a graphene ZL and annealed the sample at800 ◦ C for five minutes. After this annealing step the lineardispersion characteristic for graphene is clearly visible aroundthe K-point of the surface Brillouin zone.
Photoemission experiments
The angle-resolved photoemission spectroscopy (ARPES)measurements in Fig. 1 a) of the manuscript were done witha SPECS HSA 3500 hemispherical analyzer with an energyresolution of 10 meV and monochromatized He II radiationat room temperature. The Fermi surfaces in Fig. 1 b) of themanuscript were measured at the Synchrotron Radiation Cen-ter (SRC) in Madison/Wisconsin using a Scienta analyzer withan energy resolution of better than 10 meV, a photon energyof ~ ω = 52 eV at a sample temperature of 100K. The angularresolution of 0.4 ◦ offers a momentum resolution of 0.023 ˚A − at the Fermi level. This is comparable to the offset that wassubtracted in Fig. 2 a) of the manuscript for the cML and thepML Au . The Fermi surfaces for the cML and nML Au weremeasured with a step size of 0.25 ◦ along the Γ K direction. Asthe linewidth for the pML Au is narrower than for the cML andthe nML Au we had to reduce the stepsize to 0.1 ◦ to allow forreasonable data fitting.The core level spectra in Fig. 2 b) were also measured at theSRC using a photon energy of ~ ω = 150 eV. They were fittedwith Lorentzian peaks including a Shirley background. conventional ML nML Au cML ZLpML Au FIG. 4:
LEED images taken at 126 eV for the conventional graphenemonolayer (cML), the zero layer (ZL), the p-doped graphene mono-layer (pML Au ) and the n-doped graphene monolayer (nML Au ). Therelative intensity between graphene spot and satellite spots is a mea-sure for the strength of the substrate influence on the graphene layer. Scanning tunneling experiments
The images in Fig. 3 a) of the manuscript were mea-sured with a room temperature scanning tunneling microscope(STM). The SiC samples with a ZL or cML on top were trans-ferred to the STM chamber in air. Annealing of the samples at800 ◦ C was sufficient to remove any adsorbates from the sur-face. Au was deposited in situ from a commercial electronbeam evaporator. The images for the cML and pML Au wererecorded at a tunneling current of 0.2 nA and a bias voltage of − . V and − . V, respectively.
Low energy electron diffraction measurements
Fig. 4 shows LEED images recorded at 126 eV electron en-ergy. This energy is particularly sensitive to the graphene cov-erage [1]. The image for the cML shows the graphene (10)spot surrounded by satellite peaks from the ( √ × √ )R30 ◦ reconstruction. The graphene (10) spot and the two left lowersatellite spots have roughly the same intensity. For the ZLthere is no graphene spot visible at 126 eV, only the satellitespots are there. The pML Au has a very bright graphene spot,whereas the satellite peaks are considerably reduced in inten-sity. Furthermore, the distance between the satellite peaksand the graphene peak is smaller than for the cML indicat-ing a larger lattice constant of the superstructure. This can berelated to the strain release in the pML Au that was revealedin the Raman measurements in Fig. 3 b) of the manuscript.The strain release results in a new commensurate periodicityin agreement with the STM measurements in Fig. 3 a) of themanuscript that show an increase of the superlattice constantby about a factor of two when comparing cML and pML Au .The LEED image for the nML Au is very similar to that of thecML indicating a similar influence of the underlying substratein both cases. Raman measurements
The Raman spectra shown in the manuscript were mea-sured under ambient conditions using an Argon ion laser witha wavelength of 488 nm. The laser spot size was 400 nmin diameter and the laser power was 4 mW. The measuredgraphene signal is rather weak and superposed by the signalfrom the SiC substrate. For the Raman data shown in Fig. 3b) of the manuscript we subtracted the substrate contributionso that the graphene peaks become clearly visible [2].The Raman spectra are characterized by three maingraphene contributions: The G peak corresponds to an in-plane vibration of the two sublattices with respect to eachother. The D and 2D peak come from a double resonance scat-tering process [3]. The 2D peak is always visible, whereas theD peak only appears in the presence of defects. Both G and2D peaks shift as a function of doping [4-6] and strain [7,8].Therefore, it is difficult to determine charge carrier concen-tration and strain directly from the Raman data. However, thedoping induced shift is strongest for the G peak [5,6], whereasthe effect of strain is more pronounced for the 2D peak [7]. If the effect of the charge carrier concentration can be deter-mined by another procedure (in this case ARPES data), theRaman data provide useful information about strain.[1] C. Riedl, A. A. Zakharov and U. Starke, Appl. Phys.Lett. , 033106 (2008)[2] D. S. Lee, C. Riedl, B. Krauss, K. von Klitzing, U.Starke and J. H. Smet, Nano Lett. , 4320 (2008)[3] S. Reich and C. Thomsen, Phil. Trans. R. Soc. Lond. A , 2271 (2004)[4] S. Pisana, M. Lazzeri, C. Casiraghi, K. S. Novoselov,A. K. Geim, A. C. Ferrari and F. Mauri, Nature Mater. , 198(2007)[5] J. Yan, Y. B. Zhang, P. Kim and A. Pinczuk, Phys. Rev.Lett. , 166802 (2007)[6] C. Stampfer, F. Molitor, D. Graf, K. Ensslin, A. Jungen,C. Hierold and L. Wirtz, Appl. Phys. Lett. , 241907 (2007)[7] M. Huang, H. Yan, C. Chen, D. Song, T. F. Heinz and J.Hone, PNAS , 7304 (2009)[8] Z. H. Ni, H. M. Wang, Y. Ma, J. Kasim, Y. H. Wu andZ. X. Shen, ACS Nano2