Graphene Heterostructures Studied by

Nov 9, 2010 - Comparative analysis for these two different regions (SLG and MLG) ... In the metal/SLG heterostructures, significant differences from t...
3 downloads 0 Views 2MB Size
Download PDF
20042

J. Phys. Chem. C 2010, 114, 20042–20048

Interface Properties of Metal/Graphene Heterostructures Studied by Micro-Raman Spectroscopy Shiro Entani,*,† Seiji Sakai,† Yoshihiro Matsumoto,† Hiroshi Naramoto,† Ting Hao,† and Yoshihito Maeda†,‡ AdVanced Science Research Center, Japan Atomic Energy Agency, Tokai, Ibaraki 319-1195, Japan and Department of Energy Science and Technology, Kyoto UniVersity, Sakyo-ku, Kyoto 606-8501, Japan ReceiVed: July 5, 2010; ReVised Manuscript ReceiVed: September 30, 2010

Studies are conducted for the influence of the interface formation of graphene with various transition metals on its vibrational properties by using confocal micro-Raman spectroscopy. Micrometer-scale heterostructures consisting of patterned regions of the single layer and multilayer graphene (SLG and MLG, hereafter) covered with and without metals on the same graphene sheet were fabricated by thin-film deposition on the graphene surface through a shadow mask. Comparative analysis for these two different regions (SLG and MLG) fabricated within an identical graphene sheet enables us to investigate the interactions at and the doping effect from the metal/graphene interface as a function of the layers number of graphene without the influence of the unintentional doping. Confirmed dependences of the peaks shifts of the Raman bands (D, G, and 2D bands) on the graphene layers number and metal species (Co, Ni, and Au) reveal that the interfacial interactions are dramatically different between single layer and multilayer graphenes. In the metal/MLG heterostructures, the Raman band shifts are reasonably attributed to carrier doping from metals. It is found that the type of the doped carriers (electrons or holes) is different between Co/MLG and Au/MLG, irrespective of almost the same work functions of Co and Au. These analyses also provide the effective thickness of carrier doping (2-3 graphene layers) from the interfaces. In the metal/SLG heterostructures, significant differences from the metal/MLG heterostructures were observed for the Raman parameters of the G and 2D bands. It is suggested that there exist strong interactions at the metal/SLG interfaces different from those at the metal/MLG interfaces. Introduction Graphene has attracted worldwide attention as one of the most promising materials for realizing nanoelectronic and spintronic devices in recent years, due to the novel electronic and electric properties (e.g., quantum electronic transport, characteristic charge carriers which behave as massless Dirac fermions, a tunable band gap, and long spin-diffusion length).1-4 In the design of graphene-based devices, the control of the charge injection process and the electronic structure at the interface between the graphene sheet and a metal electrode is essential.5,6 Plenty of studies have been reported on the interactions at the interfaces in the metal/graphene heterostructures both theoretically and experimentally.7-12 It has been demonstrated that the interactions are strong enough to modify the π bands of graphene in the case of ferromagnetic metals (Co and Ni) and, on the other hand, negligibly weak in the case of noble metals (Au, Ag, and Cu). In most of the analytical studies, the metal/ graphene heterostructures have been synthesized by the epitaxial growth of graphene on the metal surface employing ultrahigh vacuum chemical vapor deposition (UHV-CVD) method.13 In UHV-CVD, it can be considered that graphene grows through the dissociation and polymerization of hydrocarbon molecules on the catalytic metal surfaces. Meanwhile, in the experimental studies on graphene devices, the metal/graphene heterostructures have been fabricated by metal deposition on the graphene sheet prepared with micromechanical cleavage.14 Less information is * To whom correspondence should be addressed. Tel/Fax: +81-29-2843802. E-mail: [email protected]. † Advanced Science Research Center, Japan Atomic Energy Agency. ‡ Department of Energy Science and Technology, Kyoto University.

known on the interfacial interactions and the electronic structures of graphene in the metal deposited heterostructures as in the devices and on whether the metal/graphene interfaces prepared by UHV-CVD and metal deposition are equivalent or not. Furthermore, it is difficult to evaluate the dependences on the layers number of graphene in the UHV-CVD heterostructures from the limitation of the controlled growth of the layers number. In the present work, interface-related phenomena and electronic structure of graphene in the metal/graphene heterostructures are investigated for the samples fabricated by metal deposition (Co, Ni, and Au) and with various graphene layer numbers, n (n ) 1-7 and more), by employing confocal microRaman spectroscopy considering the following two issues. The first issue is related to the limited sizes in the graphene sheet fabricated with micromechanical cleavage (less than 10’s of µm) and also in uniform graphene layers (around several µm typically). Micro-Raman spectroscopy is one of the most powerful techniques to investigate the chemical interactions and electronic structures of graphene by probing the vibrational property at a specified spot with less than about 1 µm in diameter.15-17 It is known that three prominent bands appear in the Raman spectrum of graphene: the G band around 1580 cm-1, the D band around 1360 cm-1 and the 2D (G′) band around 2700 cm-1. The G band is ascribed to the doubly degenerate (TO and LO) phonon mode at the Brillouin zone center, which corresponds to the in-plane vibration of carbon atoms with E2g symmetry. The D band is due to the TO phonons branches around the K point and requires the symmetry lowering for its activation, and is usually related to defects and structural

10.1021/jp106188w  2010 American Chemical Society Published on Web 11/09/2010

Properties of Metal/Graphene Heterostructures

Figure 1. (a) Schematic representation of specimen structure and (b) an optical micrograph of the Co/graphene sample with the structure of SiO2/Co/graphene/glass in the heterostructure region as an example. The graphene sheet contains several areas with different layers numbers.

distortions.18 Thus, the D band provides the information on the perfection of the graphitic structure, and is not generally observed in the as-prepared graphene sheets by micromechanical cleavage. The 2D band is the overtone of the D band and is Raman-active without the distortions different from the D band. The 2D band provides the information on the electronic structure of graphene through the double resonance process. These Raman bands also provide the information on the carrier (electron/hole) doping in graphene. It has been demonstrated that the carrier doping induced by the field-effect can be detected from the upshift and down-shift behaviors of the G and 2D bands.19-21 The 2D band is up-/down-shifted by the hole/electron doping, respectively. In contrast, the G band is up-shifted by both electron and hole doping due to the nonadiabatic removal of the Kohn anomaly at the zone center.22,23 One can thus evaluate the amount of the charge transfer semiquantitatively from the changes in the Raman parameters at the specified spot using a confocal Raman system. The second issue is associated with the unintentional doping to the graphene sheet. Casiraghi et al. have pointed out that an unintentional hole doping to graphene is induced inhomogeneously even within the same graphene sheet, judging from the large variations of the Raman parameters depending on the position of probing laser spot.24 Calizo et al. have reported that the peaks of the G and 2D bands show the shift of the peak position and the modification of the spectral features when the graphene sheet is transferred onto the insulator substrates.25,26 They have pointed out that these spectral changes are closely related to changes in the charge accumulation, the nature and density of the defects, different strength of the graphene-substrate bonding, and etc. It is, therefore, necessary to consider the unintentional doping and/or the changes in the graphene nature by graphene/substrate formation, in order to discuss the spectroscopic changes induced even by metal deposition properly. For this purpose, we fabricated the specially arranged specimen composed of the two different regions; the metal/graphene heterostructure region and the pristine graphene region on the same graphene sheet with the same graphene layers numbers, by using the micrometer patterned metal deposition with a shadow mask. Experimental Methods Figure 1(a),(b) shows a schematic representation of the metal/ graphene sample composed of the two regions of the metal/ graphene heterostructure and pristine graphene, together with an optical micrograph of the fabricated Co/graphene sample as an example. The fabrication procedures and the microstructures of the metal/graphene samples are mentioned in the following. Graphene sheet was transferred onto a glass substrate by microcleaving highly oriented pyrolitic graphite (HOPG).14 In

J. Phys. Chem. C, Vol. 114, No. 47, 2010 20043

Figure 2. AFM images of 3 nm-thick (a) Co and (b) Au films grown on the graphene surfaces.

this work, the graphene layers number, n, in the cleaved graphene sheet is determined through the following procedures: In the case of thin graphene (n ) 1-3), the layers number can be estimated by the component analysis of the 2D band.17 In the thick graphene (n ) 4-7), the evaluation was made successfully by counting the graphene layers number from the stacks of the patch-like thin graphene with n ) 1-3, as can be seen in Figure 1(b). The validity of this evaluation was confirmed by the increase of the G band intensity with the layers number in the confocal Raman spectra. The prepared substrate (graphene/glass) was introduced into an UHV chamber with a base pressure of 3 × 10-7 Pa and was annealed at 423 K for 1 h for degassing, and then the metal/graphene heterostructure was fabricated on a part of the graphene sheet by depositing a thin film (3-10 nm thick) of metals (Co, Ni, and Au) and SiO2 through a micropatterned shadow mask. The metals and SiO2 were evaporated from the electron beam evaporator (Co, Ni, and SiO2) and alumina coated tungsten basket (Au), respectively. The temperature of the glass substrate was kept at ambient temperature during the meal deposition. The surface of the metal film grown on the graphene sheet was capped with a 10 nmthick SiO2 layer successively after the metal deposition, which assured the Raman analysis free from the influence of oxidation. According to these procedures, we can obtain the metal/graphene sample as shown in Figure 1 (b), which is composed of the two kinds of regions; the heterostructure regions (bright regions in the image) with a square shape of 7.5 × 7.5 µm2 arranged at a separate distance of 5 µm and the intermediate region of pristine graphene (dark regions in the image). Here, the Co/ graphene sample in Figure 1(b) has the structure of 10 nmSiO2/5 nm-Co/graphene/glass in the heterostructure regions. Figure 2(a),(b) show the AFM images of Co/graphene and Au/ graphene, respectively. The deposition thicknesses of Co and Au were commonly 3 nm. The layer numbers of graphene were not identified in the AFM measurements. The Co and Ni (not shown here) films have smooth and continuous surface. In contrast, the Au film consists of coarse and flat grains with diameters of 20-100 nm distributed discontinuously on the graphene surface. The grown metal films were characterized with θ-2θ scan X-ray diffraction technique (XRD). Since the amount of the cleaved graphene sheets is insufficient for the XRD assessment of the crystallographic relationship of evaporated layers, the respective films with comparable thickness were deposited on the HOPG surfaces, which were freshly cleaved in the atmosphere and were annealed at 423 K for 1 h in UHV, at room temperature. The XRD results (not shown) confirms the growth of polycrystalline films of the hexagonally closepacked (hcp) Co and the face centered cubic (fcc) Ni and Au, with the (0001)- and (111)-preferred orientations, respectively. The Raman measurements were carried out with a micro-Raman system (NANO-FINDER, Tokyo Instruments, Inc.) in a confocal backscattering geometry. Raman spectra were obtained by focusing Ar-ion laser (488 nm, 1 mW) to the selected spots

20044

J. Phys. Chem. C, Vol. 114, No. 47, 2010

Figure 3. A pair of Raman spectra in the G and 2D bands regions obtained from the Co/graphene heterostructure regions and the pristine graphene regions in the samples with different graphene layers number (n), respectively. Dotted and broken curves denote the resolved and fitted spectra, respectively.

with less than 1 µm in diameter on the sample surface under ambient condition. It has been reported that the G and 2D peaks are shifted, depending on sample temperature. The temperature coefficients for the G and 2D band in SLG are reported to be -(1.6 ( 0.2) × 10-2 and -(3.4 ( 0.4) × 10-2 cm-1/K, respectively.27,28 In the present study, all of the Raman measurements were carried out with the laser power of 1 mW. Taking into account that the Raman measurements with the lower laser power (0.7 mW) at the same measuring position cause no significant peak shift (especially upshift) nor modification of the spectral feature, the influence of the local heating by the laser irradaition can be excluded during the Raman measurements in this study. The spectral resolution of the spectrometer employed was about 3 cm-1. For the measurements of the heterostructure regions, Raman spectra were obtained through the metal film layer covered with SiO2 as depicted in Figure 1(a). The deposition thicknesses of metals in the samples for the Raman measurements are 3 nm for Co and Au and 5 nm for Ni, respectively, and are thin enough for laser penetration and detection of the Raman signals through the metal and SiO2 layers. The microscale patterned structure of the metal/graphene specimen enables us to investigate the graphene layers number (n) dependences of the Raman parameters, since the graphene sheet contains several regions with different layers numbers as seen as the laminar contrast variation in the image of the graphene sheet in Figure 1(b). Changes in the Raman parameters, peak positions and the full width at the half-maximum (fwhm) of the D, G, and 2D bands, due to the formation of the metal/graphene heterostructure are examined by comparing a pair of the Raman spectra obtained at the neighboring two spots in the heterostructure and pristine graphene regions which are separated by only a few to several µm. Results and Discussion Figures 3 and 4 show the evolutions of the Raman spectra in the Co/graphene and Au/graphene samples as a function of the number of graphene layers. In these figures, the data consist of pairs of two spectra obtained at the heterostructure region (black

Entani et al.

Figure 4. A pair of Raman spectra of the Au/graphene samples in the G and 2D bands regions. Black curves are from the Au/graphene heterostructure regions and gray curves are from the pristine graphene regions in the samples with different graphene layers number (n), respectively. The components from the heterostructure and pristine graphene regions (see text) are drawn with dotted and broken curves, respectively.

line) and at the pristine graphene region (gray line) within the same graphene sheet with the identical graphene layers number (n ) 1-7) or thick multilayer (n > 7), as mentioned above. It is found that, in the heterostructure regions, the respective peaks exhibit shifts of the peak positions and broadenings compared to those in the corresponding pristine graphene regions. These changes are more pronounced in the heterostructures composed of the single-several layer graphenes, and therefore can be attributed to the effect of the metal/graphene interface-formation. In the following, discussions will be made on the nature of the layers number dependent changes of the Raman parameters (especially about the G and 2D bands) found in the metal/ graphene heterostructures. A. Characteristics of Carrier Doping into Graphene Layer. Figure 5 illustrates, for the Co/graphene and Au/graphene samples, the shifts of the peak positions of the G and 2D bands (G and 2D band shifts, hereafter) in the heterostructurte region from the positions in the adjacent pristine graphene region as a function of the number of graphene layers. In the case of the Co/graphene samples, the 2D band shift was estimated from the shift of the center of gravity (COG) of each 2D band between the two regions for convenience, since the 2D band in MLG (n g 2) is composed of the several (2-4) components17 (see Figure 3) and the peak separation becomes difficult especially in the heterostructure region due to the band broadening. In the case of the Au/graphene samples, the G and 2D bands show more complicated features. As shown in Figure 4, the G and 2D bands in the Au/graphene heterostructure region can be separated into two band components with larger and smaller shifts with respect to the positions of these bands in the pristine graphene, aside from the several peak components originating each 2D band as mentioned above. The AFM image in Figure 2(b) indicates that a part of the graphene surface is not covered with Au due to the discontinuous distribution of coarse Au grains. Accordingly, the band components with larger and smaller shifts observed in the heterostructure region are considered to be coming from the “true” heterostructure covered with Au and the bare

Properties of Metal/Graphene Heterostructures

J. Phys. Chem. C, Vol. 114, No. 47, 2010 20045

Figure 5. Graphene layers number dependences of the Raman G and 2D band shifts by metal depositions; Posmetal(G) - Pospri(G) (upper) and Posmetal(2D) - Pospri(2D) (lower), evaluated for (a) the Co/graphene samples and (b) the Au/graphene samples.

graphene, respectively. The G band shift in the Au/graphene heterostructure is therefore estimated for the band component with larger shift. However, the 2D band shift is estimated from the COG of the whole 2D band because of the difficult to separate the two band components precisely due to a large number of the peak components included therein. It should be noted that the 2D band shift in the Au/graphene heterostructure would be underestimated since the COG contains the contribution from the area of the bare graphene. From the graphene layers number dependences of the G and 2D band shifts in the Co/graphene and Au/graphene heterostructures the following two significant features can be seen, (i) the metal-dependent systematic shifts of the G and 2D bands by the number of graphene layers in the MLG region, and (ii) the remarkable difference in the G band shifts between the metal/SLG and metal/MLG heterostructures. In the metal/MLG heterostructures, the G band is up-shifted both in the Co/MLG and Au/MLG samples compared to the pristine graphene. Meanwhile, the 2D band is down-shifted in Co/MLG, whereas it is up-shifted in Au/MLG. It has been reported that the doping of electrons/holes in a field effect geometry causes peak shifts of the G and 2D bands,19,20 and has demonstrated that the G peak is up-shifted for both electron and hole doping, whereas the 2D peak is down-shifted and upshifted for the respective cases. Taking account of these correlations, the observed peak shift of the G and 2D bands in the MLG region are attributed to hole doping in the Au/MLG heterostructure and electron doping in the Co/MLG heterostructure, respectively. It is worthy to note that the type of doped carriers cannot simply predict from the work functions of metals and graphene, since the quantities are almost the same between Co (5.1 eV) and Au (5.0 eV),29 and hence one can expect that holes tend to be doped in both heterostructures taking account of the work function of graphene (4.6 eV).13 The present results indicate that the type of the doped carriers is affected by the chemical/electronic interactions at the metal/graphene interface. In Figure 5, the G and 2D band shifts decrease with increasing the number of graphene layers. This indicates that the doped carriers are accumulated within certain layers from the metal/ graphene interface. It therefore would be useful to estimate the accumulation thickness t of the doped carriers from the fits of

Figure 6. A pair of Raman spectra in G and 2D band in the regions of pristine SLG and the heterostructures with Co, Ni, Au, and SiO2, respectively. The intensity of the Ni/graphene spectrum was magnified by a factor of 5.

the layers number dependences of the G and 2D band shifts with an exponential function of a exp(-n/t), where a is the constant and n is the number of graphene layers. The data of the metal/SLG heterostructure (n ) 1) were excluded from the fitting, due to the remarkably different behaviors as represented by the down-shift of the G band peak, which will be discussed later. The accumulation thickness obtained from the fits are t ) 2-3 both for the Co/MLG and Au/MLG heterostrcutures. The similar extent of t irrespective of the different metal species and doped carriers suggests that the accumulation thickness is governed by the inherent electronic property of MLG, and metal diffusion or penetrating into the graphene layer is negligible. B. Interactions at Metal/SLG and Metal/MLG Interfaces. In the following, the interactions at the metal/graphene interface in the heterostructures will be examined. Figure 6 shows the Raman spectra of SLG in the heterostructure regions (black lines) with various metals (Co, Ni, and Au) and SiO2 and also

20046

J. Phys. Chem. C, Vol. 114, No. 47, 2010

Figure 7. (a) Pos(2D) and (b) fwhm(G) as a function of Pos(G). The symbols O, × and 2 denote the Raman parameters evaluated for the as-cleaved SLG and from the regions of pristine SLG and heterostructures in the samples of SLG and Co, Ni, Au, and SiO2, respectively.

in the pristine graphene regions (gray lines). In the metal/SLG heterostructures, one can observe the definite changes in the G and 2D band shapes which show specific characteristics depending on the metal species, whereas it is not the case in SiO2/SLG. It was also found that the D band which indicates the symmetry lowering of graphene appears at ∼1360 cm-1 in the heterostructure regions (not shown). The peak intensity ratio of the D band to the G band provides roughly the degree of the disorder in graphene. The estimated values of the D/G intensity ratio are small and are in the range of 0.04-0.1 in the Co/ SLG, Ni/SLG and Au/SLG heterostructures. The Raman spectra in the heterostructures returned back to the almost similar ones to that in the pristine SLG after the removal of the metal films by soaking the samples in a solution of 20 vol% HNO3. This suggests that these spectral changes are not caused by the irreversible structural changes, such as destruction and damaging of the graphitic structure. Figure 7 shows the changes of the Raman parameters as a function of the peak position of the G band (Pos(G)); (a) the peak position of the 2D band (Pos(2D)) and (b) the fwhm of the G band peak, analyzed for a series of the pairs of the heterostructure and pristine graphene regions with SLG. The parameters for the as-cleaved SLG samples obtained immediately after cleavage are plotted together. As described in the introduction, it has been reported that the Raman parameters are changeable according to the unintentional hole doping in the graphene sheets prepared by micromechanical cleavage.24 The linear increasing of Pos(2D) with Pos(G) for the as-cleaved SLG as seen in Figure 7 (a) is ascribed to the increase of the carrier concentration by hole doping.24 The fwhm(G) decrease with increasing Pos(G) in the as-cleaved SLG is also interpreted by the effect of electron/hole doping.19,20 When the Fermi level is moved by the electron/hole doping, the phonon decay is suppressed reflecting the activation condition of |EF| > pωG/2 between the EF and the pωG (G band phonon energy), which causes the fwhm(G) decrease by carrier doping. One can expect that the relationships between the Raman parameters would be in accordance with the as-cleaved SLG, as far as the effect of hole doping is predominant. Such coincidences with the ascleaved SLG are actually recognized for the pristine graphene regions in the metal/SLG samples. In the case of the SiO2/ graphene sample, the Raman parameters show good coincidence with the ones in the as-cleaved graphene even with or without

Entani et al. the SiO2 layer. This indicates a negligibly weak chemical interaction at the SiO2/SLG interfaces except for hole doping as suggested from the shift of the data points in Figure 6 (a). It is found that the Raman parameters of the metal/SLG heterostructures exhibit the contradictory behaviors to the ascleaved SLG. These deviations from the systematic relationships in the as-cleaved SLG suggest the different mechanism from hole and/or electron (not shown) doping from metallic elements. The most distinct point is the down-shift of the G band induced by the formation of the metal/SLG heterostructures (see Figure 7(a) and also Figure 5), which is expected not to occur either by electron or by hole doping except the heavy electron doping. First, the discussions will be made on the Co/SLG and Ni/ SLG heterostructures. It has been reported that the covalent interactions at the Ni(111)/graphene and transition metal oxide/ graphene interfaces induce the softening the phonon modes of graphene.7,10 Plenty of investigations have shown that the interfacial interactions at the Co/graphene and Ni/graphene interfaces are covalent-like. Taking this into account, the downshift of the G band in the Co/SLG and Ni/SLG heterostructures is considered to be due to the weakening of the graphene C-C bonds by the occupation of the antibonding π*-derived states induced by the π-d hybridization at the interfaces. As represented by the downward deviations from the relationships for the as-cleaved SLG in Figure 7(a), the peak positions of the 2D bands, Pos(2D), in the Co/SLG and Ni/SLG heterostructures cannot explain by carrier doping. Since the 2D band is due to the double-resonant Raman scattering process with the finite phonon wave vector q (* 0), the shifts of Pos(2D) ) 2pωq (ωq: phonon frequency) can be affected by the change of the phonon dispersion related to the covalent interaction at the interface (resulting in the weakening of the graphene C-C bonds). High resolution electron energy loss spectroscopy (HREELS) measurements of graphene/Ni(111) have demonstrated that the graphene phonon dispersion relation changes and shifts to low wavenumber region.30,31 The distortion of the graphitic structure accompanied with the strong covalent interaction might be responsible for the increases of the fwhm(G)s in these heterostructures (Figure 7(b)), considering the small but non-negligible lattice mismatches of 1.3% and 1.9% between the Ni(111) and Co(0001) surfaces with graphene. In Figure 7(a),(b), surprisingly, the down-shift of the G band and the increase of the fwhm(G) are also identified in the Au/ SLG heterostructure although the extents are smaller than those in Co/SLG and Ni/SLG. This implies that the interactions between Au and SLG also have strong chemical nature. The activation of the D band commonly observed for each heterostructure as mentioned above confirms this viewpoint. In the case of the Au/SLG heterostructure, it is also necessary to take into account the upshift of the 2D band different from the downshift in the Co/SLG and Ni/SLG heterostructures. As a possible reason, it is considered the small modification of the electronic structure of graphene around the K point is related to the upshift of the 2D band through the following process. Changes in the π and π* band dispersions can affect the wave vector k which satisfies the relation pωL ) ε(π*, k) - ε(π, k) for the Raman activation, where pωL is the photon energy of the excitation laser, and ε(π*, k) and ε(π, k) are the electron energies of the π* and π bands at k, respectively. As the 2D band is caused by the double-resonance process, the wave vectors k and q of electrons and phonons measured from the K point satisfy the relationship q ≈ 2k. In pristine SLG, the phonon energy pωq increases with increasing q near the K point. Thus, the change in the π band can influence Pos(2D) () 2pωq). It has been

Properties of Metal/Graphene Heterostructures reported that a kink appears near Fermi energy (EF) (0.95 eV) in the π band due to the graphene-Au interaction.9 This π band modification would raise the corresponding wave vector k, since the pωL/2 adopted in the present study is 1.27 eV where the π band would be still away from the K point compared with the former π band structure. As described above, pωq in turn shifts to higher wavenumber without the significant modification of the phonon dispersion around the K point. It is reasonably proposed that the 2D band upshift in the Au/SLG heterostructure mainly comes from the small but non-negligible π-band modification induced by the formation of the Au/graphene interface. In order to explain the different directions of the G band shifts (up-/down-shifts) between the metal/SLG and metal/MLG heterostructures as shown in Figure 5, it is necessary to consider different contributions to the shifts of the Raman parameters depending on single layer or multilayer (n g 2), or the negligibly small signals from the first graphene layer at the interface compared with the signals from the other layers. The latter possibility, however, can be excluded at least for the Au/ graphene heterostrucrures, because of the surface enhancement effect of Raman scattering expected to occur at and near the interface.32 Considering that the quantities of the G and 2D band shifts are well fitted with simple exponential decay as a function of the number of graphene layers within the range of n g 2 both for the Co/graphene and Au/graphene heterostructures, it is surmised that the chemical interactions between metal and graphene in the metal/MLG interfaces are so weak as not to induce changes in the Raman band shapes and hence the π-band structure, in contrast to the strong chemical interactions at the metal/SLG interfaces. Some previous investigations on the interface interaction at the graphene heterostructures have also reported the similar results: Ohta et al. have demonstrated that, for the graphene sheets grown on SiC, the interlayer interaction alters the character of the π wave function from two-dimensional in SLG to a bulk like one in MLG.33 In the graphene sheets grown on TaC(111), it has been reported that the cohesive energy of graphene to the TaC(111) surface changes between single layer and bilayer graphenes.34 It was pointed out that the cohesive energy is possibly associated with the incommensurate relation between graphene and TaC(111) under that condition. In contrast, both the present study and ref 29 would support that the change of the cohesive energy at the interface arises also in the commensurate interfaces such as metal/graphene and graphene/SiC. As for the metal/SLG interfaces, the changes of the Raman parameters in the heterostructures of SLG with different metals reveal the strong interactions enough to induce obvious modifications in the electronic structure of graphene. In the case of Co and Ni, the observed significant phonon softening attributed to the strong covalent interactions at the interfaces is consistent with the previous studies on the comparable heterostructures synthesized by UHV-CVD,8,9 which indicates a similar interface structure in the heterostructures synthesized by the metal deposition and UHV-CVD. In the case of Au, our observations suggest non-negligible modification of the graphene π-band. This contradicts from the reported results on the weak interfacial interactions such as van der Waals-like ones both theoretically and experimentally.10-12 In ref 10, the phonon dispersion of graphene has been analyzed for the heterstructures of SLG/Ni(111) and SLG/Au/Ni(111) by HREELS.10 Definite the differences, however, were not detected between the phonon dispersions in SLG/Au/Ni(111) and pristine graphite, different from the significant phonon softening identified in SLG/Ni(111). We speculate that this contradiction would

J. Phys. Chem. C, Vol. 114, No. 47, 2010 20047 come from the insufficient energy resolution of HREELS (∼10 cm-1). The present results indicate that the strong chemical interactions arisen at the metal/SLG interface can be mostly reduced at the metal/MLG interface, which suggests a possibility that one can prevent the modification of the π-band structure and also adjust the carrier injection efficiency at the interfacial regions of graphene with metal electrodes by using MLG with controlled layers as transport media. Conclusions Interactions of single/multilayer graphene with ferromagnetic and noble metals were investigated for the metal/graphene heterostructures fabricated by metal deposition onto the microcleaved graphene surface, by using confocal micro-Raman spectroscopy. Changes of the Raman parameters in the heterostructures with the metal species (Co, Ni, and Au) and graphene layers number reveal the followings: (i) Characteristics of charge carrier doping into MLG; carrier doping within the critical thickness of a few graphene layers from the interface and the type of the doped carriers (electron/hole doping) governed by the interfacial interactions but not by the vacuum level shift. (ii) Different interfacial interactions between the metal/SLG and metal/MLG heterostructures; strong chemical interactions at the metal/SLG interfaces even in the case of Au and their significant reduction at the metal/MLG interfaces. We believe that the present study provides the important information for improving the performance of graphene-based devices. Acknowledgment. This work was partly supported by Grantin-Aid for Scientific Research (B) (Grant No. 19360290), for Yong Scientists (Start-up) (Grant No. 21860089) and for Yong Scientists (B) (Grant Nos. 22740206 and 22760033) from the Japan Society for the Promotion of Science. References and Notes (1) Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Zhang, Y.; Dubonos, S. V.; Grigorieva, I. V.; Firsov, A. A. Science 2004, 306, 666–669. (2) Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Katsnelson, M. I.; Grigorieva, I. V.; Dubonos, S. V.; Firsov, A. A. Nature 2005, 438, 197– 200. (3) Zhang, Y.; Tan, Y.-W.; Stormer, H. L.; Kim, P. Nature 2005, 438, 201–204. (4) Heersche, H. B.; Jarillo-Herrero, P.; Oostinga, J. B.; Vandersypen, L. M. K.; Morpurgo, A. F. Nature 2007, 446, 56–59. (5) Sakai, S.; Sugai, I.; Mitani, S.; Takanashi, K.; Matsumoto, Y.; Naramoto, H.; Avramov, P. V.; Okayasu, S.; Maeda, Y. Appl. Phys. Lett. 2007, 91, 242104. (6) Matsumoto, Y.; Sakai, S.; Takagi, Y.; Nakagawa, T.; Yokoyama, T.; Shimada, T.; Mitani, S.; Takanashi, K.; Naramoto, H.; Maeda, Y. Chem. Phys. Lett. 2009, 470, 244–248. (7) Nagashima, A.; Tejima, N.; Oshima, C. Phys. ReV. B 1994, 50, 17487–17495. (8) Yamamoto, K.; Fukushima, M.; Osaka, T.; Oshima, C. Phys. ReV. B 1992, 45, 11358–11361. (9) Varykhalov, A.; Sa´nchez-Barriga, J.; Shikin, A. M.; Biswas, C.; Vescovo, E.; Rybkin, A.; Marchenko, D.; Rader, O. Phys. ReV. Lett. 2008, 101, 57601. (10) Shikin, A. M.; Prudnikova, G. V.; Adamchuk, V. K.; Moresco, F.; Rieder, K.-H. Phys. ReV. B 2000, 62, 13202–13208. (11) Farı´as, D.; Shinkin, A. M.; Rieder, K.-H.; Dedkov, Y. S. J. Phys.: Condens. Matter 1999, 11, 8453–8458. (12) Giovannetti, G.; Khomyakov, P. A.; Brocks, G.; Karpan, V. M.; van den Brink, J.; Kelly, P. J. Phys. ReV. Lett. 2008, 101, 026803. (13) Oshima, C.; Nagashima, A J. Phys.: Condens. Matter 1997, 9, 1– 20. (14) Novoselov, K. S.; Jiang, D.; Schedin, F.; Booth, T. J.; Khotkevich, V. V.; Morozov, S. V.; Geim, A. K. Proc. Natl. Acad. Sci. 2005, 102, 10451–10453. (15) Graf, D.; Molitor, F.; Ensslin, K.; Stampfer, C.; Jungen, A.; Hierold, C.; Wirtz, L. Nano Lett. 2007, 7, 238–242.

20048

J. Phys. Chem. C, Vol. 114, No. 47, 2010

(16) Wang, Y. Y.; Ni, Z. H.; Yu, T.; Shen, Z. X.; Wang, H. M.; Wu, Y. H.; Chen, W.; Wee, A. T. S. J. Phys. Chem. C 2008, 112, 10637–10640. (17) Ferrari, A. C.; Meyer, J. C.; Scardaci, V.; Casiraghi, C.; Lazzeri, M.; Mauri, F.; Piscanec, S.; Jiang, D.; Novoselov, K. S.; Roth, S.; Geim, A. K. Phys. ReV. Lett. 2006, 97, 187401. (18) Ferrari, A. C.; Robertson, J. Phys. ReV. B 2000, 61, 14095–14107. (19) Yan, J.; Zhang, Y.; Kim, P.; Pinczuk, A. Phys. ReV. Lett. 2007, 98, 166802. (20) Pisana, S.; Lazzeri, M.; Casiraghi, C.; Novoselov, K. S.; Geim, A. K.; Ferrari, A. C.; Mauri, F. Nat. Mater. 2007, 6, 198–201. (21) Das, A.; Chakraborty, B.; Piscanec, S.; Pisana, S.; Sood, A. K.; Ferrari, A. C. Phys. ReV. B 2009, 79, 155417. (22) Piscanec, S.; Lazzeri, M.; Mauri, F.; Ferrari, A. C.; Robertson, J. Phys. ReV. Lett. 2004, 93, 185503. (23) Lazzeri, M.; Mauri, F. Phys. ReV. Lett. 2006, 97, 266407. (24) Casiraghi, C.; Pisana, S.; Novoselov, K. S.; Geim, A. K.; Ferrari, A. C. Appl. Phys. Lett. 2007, 91, 233108. (25) Calizo, L; Bao, W; Mia, F; Lau, C. N.; Balandin, A. A. Appl. Phys. Lett. 2007, 91, 201904.

Entani et al. (26) Calizo, L; Teweldebrhan, D; Bao, W; Maio, F; Lau, C. N.; Balandin, A. A. J. Phys. Conf. Ser. 2008, 109, 012008. (27) Calizo, L; Balandin, A. A.; Bao, W; Miao, F; Lau, C. N. Nano Lett. 2007, 7, 2645. (28) Calizo, L; Mia, F; Bao, W; Lau, C. N.; Balandin, A. A. Appl. Phys. Lett. 2007, 91, 071913. (29) Michaelson, H. B. J. Appl. Phys. 1977, 48, 4729–4733. (30) Rokuta, E.; Hasegawa, Y.; Suzuki, K.; Gamou, Y.; Oshima, C.; Nagashima, A. Phys. ReV. Lett. 1997, 79, 4609–4612. (31) Shikin, A. M.; Farı´as, D.; Adamchunk, V. K.; Rieder, K.-H. Surf. Sci. 1999, 424, 155–167. (32) Kovacs, G. J.; Loutfy, R. O.; Vincett, P. S.; Jennings, C.; Aroca, R. Langumuir 1986, 2, 689–694. (33) Ohta, T.; Bostwick, A.; McChesney, J. L.; Seyller, T.; Horn, K.; Rotenberg, E. Phys. ReV. Lett. 2007, 98, 206802. (34) Nagashima, A.; Itoh, H.; Ichinokawa, T.; Oshima, C. Phys. ReV. B 1994, 50, 4756–4763.

JP106188W