Direct observation of charge transfer region at interfaces in graphene devices

Naoka Nagamura; Koji Horiba; Satoshi Toyoda; Shodai Kurosumi; Toshihiro Shinohara; Masaharu Oshima; Hirokazu Fukidome; Maki Suemitsu; Kosuke Nagashio; Akira Toriumi

doi:10.1063/1.4808083

Direct observation of charge transfer region at interfaces in graphene devices

Naoka Nagamura, Koji Horiba, Satoshi Toyoda, Shodai Kurosumi, Toshihiro Shinohara, Masaharu Oshima, Hirokazu Fukidome, Maki Suemitsu, Kosuke Nagashio, Akira Toriumi

RIEC - Computing System Platforms Division

Research output: Contribution to journal › Article › peer-review

25 Citations (Scopus)

Abstract

Nanoscale spectromicroscopic characterizing technique is indispensable for realization of future high-speed graphene transistors. Highly spatially resolved soft X-ray photoelectron microscopy measurements have been performed using our "3D nano-ESCA" (three-dimensional nanoscale electron spectroscopy for chemical analysis) equipment in order to investigate the local electronic states at interfaces in a graphene device structure. We have succeeded in detecting a charge transfer region at the graphene/metal-electrode interface, which extends over ∼500 nm with the energy difference of 60 meV. Moreover, a nondestructive depth profiling reveals the chemical properties of the graphene/SiO₂-substrate interface.

Original language	English
Article number	241604
Journal	Applied Physics Letters
Volume	102
Issue number	24
DOIs	https://doi.org/10.1063/1.4808083
Publication status	Published - 2013 Jun 17

Access to Document

10.1063/1.4808083

Cite this

@article{d59be07b07934beca71560ad5a9a5796,

title = "Direct observation of charge transfer region at interfaces in graphene devices",

abstract = "Nanoscale spectromicroscopic characterizing technique is indispensable for realization of future high-speed graphene transistors. Highly spatially resolved soft X-ray photoelectron microscopy measurements have been performed using our {"}3D nano-ESCA{"} (three-dimensional nanoscale electron spectroscopy for chemical analysis) equipment in order to investigate the local electronic states at interfaces in a graphene device structure. We have succeeded in detecting a charge transfer region at the graphene/metal-electrode interface, which extends over ∼500 nm with the energy difference of 60 meV. Moreover, a nondestructive depth profiling reveals the chemical properties of the graphene/SiO2-substrate interface.",

author = "Naoka Nagamura and Koji Horiba and Satoshi Toyoda and Shodai Kurosumi and Toshihiro Shinohara and Masaharu Oshima and Hirokazu Fukidome and Maki Suemitsu and Kosuke Nagashio and Akira Toriumi",

note = "Funding Information: Nagamura Naoka 1,2 a) Horiba Koji 1,2,3 Toyoda Satoshi 2,3 Kurosumi Shodai 1 Shinohara Toshihiro 1 Oshima Masaharu 1,2,3 Fukidome Hirokazu 4 Suemitsu Maki 4 Nagashio Kosuke 5 Toriumi Akira 5 1 Department of Applied Chemistry, Graduate School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656 , Japan 2 Synchrotron Radiation Research Organization, The University of Tokyo , 1-1-1 Kouto, Sayo-cho, Sayo-gun, Hyogo 679-5198, Japan 3 Core Research for Evolutional Science and Technology, Japan Science and Technology Agency , 5-7, Goban-cho, Chiyoda-ku, Tokyo, 102-0076, Japan 4 Research Institute of Electrical Communication, Tohoku University , 2-1-1 Katahira, Aoba-ku, Sendai 980-8577, Japan 5 Department of Materials Engineering, Graduate School of Engineering, The University of Tokyo , 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan a) Present address: Institute of Multidisciplinary Research for Advanced Materials, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai 980-8577, Japan. Electronic mail: nagamura@tagen.tohoku.ac.jp . 17 06 2013 102 24 241604 03 02 2013 07 05 2013 17 06 2013 2013-06-17T11:36:58 2013 AIP Publishing LLC 0003-6951/2013/102(24)/241604/5/ $30.00 Nanoscale spectromicroscopic characterizing technique is indispensable for realization of future high-speed graphene transistors. Highly spatially resolved soft X-ray photoelectron microscopy measurements have been performed using our “3D nano-ESCA” (three-dimensional nanoscale electron spectroscopy for chemical analysis) equipment in order to investigate the local electronic states at interfaces in a graphene device structure. We have succeeded in detecting a charge transfer region at the graphene/metal-electrode interface, which extends over ∼500 nm with the energy difference of 60 meV. Moreover, a nondestructive depth profiling reveals the chemical properties of the graphene/SiO 2 -substrate interface. 7402 2009-2012 crossmark 2013-06-17T11:36:58 Graphene has attracted great interest since the experimental realization of high-quality two-dimensional graphene is obtained easily by mechanical peeling from graphite 1 or thermal decomposition of SiC surfaces. 2,3 Graphene-based materials are actively studied for post-silicon electronics due to its exceptional carrier mobility, mechanical superiority which realize an atomic layer transistor channel without short channel effects, and intriguing band structures which can be modified. 4 However, the mobility of the present graphene field effect transistor (FET) on SiO 2 /Si substrates is far restrained from the desired value, and this results in a deterioration of gate-voltage response of the device. With respect to this limitation, a graphene/metal-electrode contact is considered to be important because a very small density of states near Fermi level in graphene may suppress the current injection from metal to graphene. Therefore, the key to improve the device performance is to understand properties of the interface between graphene and other materials in graphene nanostructures. Indeed, graphene/metal-electrode interfaces have been intensively studied. 5–9 The Fermi level of graphene attached to the metal electrode shifts from the conical point due to the charge transfer near the graphene/metal interface. Several groups have observed the graphene/metal boundary regions by scanning probe microscopy (SPM) techniques, such as scanning photocurrent microscopy 10,11 and scanning gate microscopy. 12 Since SPM techniques can only reveal electronic states in a narrow energy region near the Fermi level so they are not good at spectroscopic identification of elements. Additionally, these measurements are limited to lateral observations without depth-profile analysis. Moreover, the effect of surface contaminants can be hardly avoided especially for the device samples which may be broken during typical sample cleaning processes, e.g., flashing at elevated temperatures (1273 K). Innovative techniques which can be applied to in situ local state analysis of actual device structures have been eagerly needed. In understanding device properties, one of the most powerful tools is photoelectron spectroscopy or electron spectroscopy for chemical analysis (ESCA) which can reveal core-level electronic states for chemical bonding states, chemical-potential shifts, the work function at the surface, and electronic structures. Recently, we have developed a unique ESCA system,“3D nano-ESCA,” 13 equipped with a focused X-ray probe and an angle-resolved photoelectron spectrometer (ARPES) to realize nanoscale three-dimensional spatial distribution analysis. Its extremely high spatial resolution as a scanning photoelectron microscope (SPEM) system helps us to perform quantitative analysis of the target area of interest. In the present study, we have performed nondestructive 3D highly spatial and energy resolved ESCA study of graphene nanodevice structures using the 3D nano-ESCA in order to investigate the charge transfer region (CTR) at the graphene/metal-electrode interface and the chemical states at the graphene/substrate interface. Instead of average information acquired from the previous transport measurements and spectroscopic studies of graphene, pinpoint direct analysis of a functional part by chemical-state-selective information obtained from core level spectra is available by our technique. 14 We use exfoliated monolayer graphene sheets from Kish graphite transferred to a SiO 2 /p + -Si substrate. O 2 plasma treatment was carried out to remove hydrocarbon contaminants on the SiO 2 substrate. 15 This process made the substrate surface hydrophilic. Ni electrodes were fabricated by vacuum deposition after standard electron-beam lithography and lift-off techniques. For pinpoint depth-profile analysis, we use the 3D nano-ESCA system which has an X-ray focusing optical system with a Fresnel zone plate and a vibration isolation system for much better spatial resolution. The equipment has been installed at the University-of-Tokyo outstation beamline, BL07LSU at SPring-8. 16 The spatial resolution better than 70 nm has been achieved in this system. In order to obtain angular dependence of the photoelectron spectra for the depth-profiling, a modified VG Scienta R3000 analyzer with an acceptance angle of 60° is adopted. Figure 1(a) shows an optical microscope image of the monolayer graphene sheet attached to Ni electrodes on a hydrophilic SiO 2 thin film (90 nm thickness) on a p + -Si(100) substrate. 15 Figure 1(b) shows a SPEM image of the sample taken at the photon energies of 1000 eV. The SPEM contrast reflects the X-ray photoelectron spectroscopy (XPS) intensity of C 1 s peak (red in Fig. 1(b) ) distinguished from surface contamination in a way described below and Si 2 p peak (green in Fig. 1(b) ). We can thus distinguish the graphene sheet clearly from the metal electrodes and the SiO 2 substrate region by using the 3D nano-ESCA. XPS spectra of C 1 s taken at the monolayer graphene sheet are represented in Figure 2(a) . The following results of SR-XPS measurements (Figs. 2–4 ) are all taken at excitation photon energy of 1000 eV. The C 1 s spectrum is clearly decomposed into two peaks: one which has larger intensity at the lower binding energy (BE) of 284.24 eV (peak I in Fig. 2(a) ) and the other which has smaller intensity at the higher BE of 285.40 eV (peak II in Fig. 2(a) ). The former is considered to be attributed to the graphene sp 2 component 17,18 and the latter to the surface contaminants from polymer residue in device fabrication process and naturally involved amorphous carbon such as absorbed carbon dioxide, aliphatic hydrocarbon, and defects. 18–20 An asymmetric Doniach-Sunjic 21 line shape is used as a fitting function of the graphene peak. In order to confirm the origin of each component, the depth profiling of electronic states using the angular distribution of core-level photoemission spectra has been measured. Fig. 2(b) is the emission angle ( θ e ) dependence of the C 1 s spectra taken in the middle of the monolayer graphene sheet on the hydrophilic SiO 2 film (Fig. 1(c) ). Here, the intensity of photoelectrons from a given depth, z , at the emission angle from a surface normal, θ e , is expressed by I = I 0 exp ( − z λ cos θ e ) , (1) where λ is the inelastic mean free paths of the solids. As θ e becomes larger, the attenuation factor of the photoemission intensity that is λ cos θ e decreases, which results in surface sensitive measurements by decreasing information from a deeper (larger z ) region. As shown in Figs. 2(b) and 2(c) , the intensity of the peak II increases compared to that of the peak I as θ e becomes larger, suggesting that the component of peak II originates from surface contaminants on the graphene which yield the peak I, that is the graphene sp 2 component. Therefore, we have succeeded in determining the BE and the peak shape of the component purely from a graphene sp 2 bond configuration precisely excluding the effect of contaminations. In the next step, we have focused on the graphene/metal-electrode interface. When a metal electrode and a graphene sheet are brought into contact, charges are transferred through the graphene/metal interface, which leads to lineup of the Fermi levels on both sides and form a dipole layer at the graphene/metal interface. The amount of charge transfer gradually decreases from the graphene/metal interface. The density of states of graphene at the Fermi level is very small so that a long screening length is needed to cancel the potential difference. In order to investigate the spatial variation of the electronic state of monolayer graphene in the vicinity of the graphene/metal-electrode boundary, we have performed nanoscale XPS measurements to plot the line-profile for the BE of the graphene peak (C 1 s peak at the lower BE in Fig. 2(a) ) around the graphene/metal-electrode boundary. The focused light probe with SR is scanned along the dashed line represented in the inset of Figure 2 . The energy shift of the graphene peak by ∼60 meV is clearly detected in the monolayer graphene on the hydrophilic SiO 2 substrate (Fig. 1(c) ), as shown in Fig. 3 . 14 The peak position moves gradually from the graphene/metal-electrode contact toward the middle of the graphene sheet by about 500 nm. This result can be regarded as the direct observation of “a charge transfer region” at the graphene/metal-electrode boundary. 5 To confirm the above-mentioned interpretation of experimental results, we have conducted a theoretical approach in terms of the charge screening in a graphene sheet at the contact region with a metal strip suggested by Khomyakov et al. 22 The band bending in the CTR has been described quantitatively by calculating the screening potential using density-functional theory within the Thomas-Fermi approximation. According to them, the spatial variation of electrostatic potential induced in graphene (x ≥ 0) can be expressed as V ( x ) ∼ Δ E ( x l s + 0.016 + 0.787 ) 1 2 ( x l s + 1.194 ) 1 4 , (2) for undoped graphene, where Δ E is the boundary potential constant derived from the work functions calculated for the close-packed surface of the metal electrode and for the graphene sheet. l s = 8.10 × 10 − 2 × κ / Δ E (nm) indicates a scaling length, where κ is a background dielectric constant in a graphene device structure. Curve fitting has been performed to the experimental data using Eq. (2) with two fitting parameters: Δ E (eV) and κ. The comparison between the calculated screening potential and the peak shift in XPS measurements is presented in Fig. 3 . In the fitting results, the coefficients are obtained as Δ E ∼ 0.29 eV , κ ∼ 14 , and l s ∼ 3.9 nm . The deduced Δ E is consistent with the boundary potential in case of metal electrodes. 22 However, the effective dielectric constant is estimated far larger than the reported typical value, κ ∼ 2.5 , which is given by the average of the static dielectric constant of SiO 2 and that of the vacuum due to the image effect. 23 This large κ estimated in the experimental results should be affected by the substrate. So we have examined the spatial distribution of chemical bonding states for silicon and oxygen at the graphene/substrate interface using the depth profiling analysis. Figure 4(a) indicates the Si 2 p spectrum which consists of a single peak at 103.5 eV (peak III), while O 1 s spectrum is deconvolved as three components centered at 531.3 (peak IV), 532.8 (peak V), and 534.3 eV (peak VI) with Voigt functions as shown in Fig. 4(b) . The peak in the Si 2 p spectrum and the main peak at 532.8 eV in the O 1 s spectrum correspond to Si-O bonds in the SiO 2 film under the graphene sheet. 24 In order to identify the cause of each component in O 1 s core level, the emission angle dependence of the O 1 s spectrum on the monolayer graphene sheet has been investigated as presented in Fig. 4(c) . Figure 4(d) displays the relative core-level intensity ratio of the subpeak IV at the lowest BE and the subpeak VI at the highest BE to the bulk SiO 2 peak (peak V) in O 1 s spectrum as a function of θ e . Both intensity ratios of peak IV/peak V and peak VI/peak V increase monotonically with the increase of θ e . Moreover, the ratio of peak IV/peak V changes more steeper than that of peak VI/peak V. It means that the origin of each component is laid in the order of: IV, VI, V, from the top. The depth profiling of O 1 s core level revealed two different chemical bonding states, peak IV and peak VI, on top of the SiO 2 film (peak V). The peak IV at the topmost surface is assigned to the O 1 s for oxygen-containing carbon contaminants caused by exposure to ambient air 25 and polymer residue in device fabrication process. 17 The peak VI is considered to be due to Si-OH bonds of silanol groups 26 with highly polarization at the graphene/substrate interface. The static dielectric constant of the SiO 2 can be raised by hydrophilic silanol groups and water molecules on SiO 2 film. The polarization inside a graphene sheet depends on the graphene/substrate interface. It means that the state of the interface change the CTR through the effective dielectric constant and then affect transport properties of the graphene device structure. The graphene sheet and the SiO 2 film substrate are not chemically bonded so that the graphene sheet is easily released from the substrate by hydrofluoric acid (HF) etching to fabricate a suspended graphene device. 27,28 Other atoms or molecules could be introduced between the graphene sheet and the substrate. As the background dielectric constant κ which depends on the condition of the graphene/substrate interface increases, the range of the CTR at the graphene/metal-electrode interface become longer according to Eq. (2) . Furthermore, κ is related to the polarization of electrons in the valence band due to virtual interband transitions into the conduction band, as U ∼ e 2 2 π κ γ , (3) where U is a renormalized Coulomb interaction parameter and γ is a band parameter, respectively. 29 The electrical conductivity is inversely proportional to U − 2 , and then proportional to κ 2 . Although further studies are needed, our results strongly suggest a new possibility of controlling the CTR at the graphene/metal-electrode interface and the transport properties of graphene devices by adopting various graphene/substrate interface layers in graphene device structures. The CTR's behavior is considered to be varied under a finite bias voltage and on a hydrophobic substrate surface without polarization. Our results help us to understand asymmetric transfer characteristics (gate-voltage dependence of the drain current) of graphene FETs. The long CTR we observed works as a p-n junction. When a positive gate voltage is applied to a graphene FET, a p-n-p junction which causes an additional contact resistance is formed in a graphene channel, while no p-n junction is formed for a negative gate voltage as explained in the previous works. 5,9 If a channel length is shorter than twice the width of the CTR, the intrinsic region in the channel disappears, and then larger distortion of transfer characteristics and contact resistance are induced. The quantitative estimation of the CTR predicts the anomalous distortion of transport properties and the proper channel length for graphene FETs. In summary, the spatially resolved investigation of electronic and chemical states in a graphene device structure has been performed using the 3D nano-ESCA system with high spatial and energy resolutions without the effect of surface contaminants. We have undoubtedly demonstrated the CTR at the graphene/metal-electrode interface by detecting a core level shift as a good probe. The CTR extends into the graphene for about 500 nm with the electrostatic variation of 60 meV. Furthermore, the chemical bonding states at the graphene/SiO 2 film substrates interface have been investigated. It is revealed that the graphene/substrate interface can affect the CTR at the graphene/electrode interface. Our results clarify the electronic and chemical states of the interfaces which are essential for the device performance. We thank H. Kumigashira for helpful discussions. This work was supported by “Core Research for Evolutional Science and Technology (CREST)” of the Japan Science and Technology Agency (JST) and “Funding Program for World-Leading Innovative R&D on Science and Technology (FIRST Program)” of the Japan Society for the Promotion of Science (JSPS). This work is carried out by the joint research in the Synchrotron Radiation Research Organization, the University of Tokyo (Proposal No. 7402 for 2009-2012). FIG. 1. (a) Optical microscope image and (b) photoelectron intensity map of graphene device structures, which is a monolayer graphene sheet on a SiO 2 film attached to Ni electrodes. FIG. 2. (a) C 1 s core-level photoemission spectrum measured in the middle of the monolayer graphene sheet. Open circles represent the raw data and a gray solid curve is fitted data. The spectrum is well fitted by two components, peak I and peak II. (b) ARPES raw spectra of C 1 s taken at various emission angles for depth profiling. The intensity of angle-resolved spectra is normalized at the peak I intensity. (c) Emission angle dependence of the relative intensity ratio between two components in C 1 s spectrum: peak II/peak I. FIG. 3. Line profile for the binding energy peak position of the graphene component (peak I in Figure 2(a) ) taken along the dashed white line shown in the inset mapping image. The energy shift of ∼60 meV occurs at the graphene/metal-electrode interface. Red circles indicate the results of peak fitting. A gray curve represents the results of calculation using Eq. (2) . FIG. 4. (a) Si 2 p and (b) O 1 s core-level photoemission spectra measured at the same region in Fig. 2 . (c) ARPES raw spectra of O 1 s taken at various emission angles for depth profiling. (d) Emission angle dependence of the relative intensity ratio among three components in O 1 s spectrum: peak IV/peak V (filled circle) and peak VI/peak V (open circle). ",

year = "2013",

month = jun,

day = "17",

doi = "10.1063/1.4808083",

language = "English",

volume = "102",

journal = "Applied Physics Letters",

issn = "0003-6951",

publisher = "American Institute of Physics",

number = "24",

}

TY - JOUR

T1 - Direct observation of charge transfer region at interfaces in graphene devices

AU - Nagamura, Naoka

AU - Horiba, Koji

AU - Toyoda, Satoshi

AU - Kurosumi, Shodai

AU - Shinohara, Toshihiro

AU - Oshima, Masaharu

AU - Fukidome, Hirokazu

AU - Suemitsu, Maki

AU - Nagashio, Kosuke

AU - Toriumi, Akira

N1 - Funding Information: Nagamura Naoka 1,2 a) Horiba Koji 1,2,3 Toyoda Satoshi 2,3 Kurosumi Shodai 1 Shinohara Toshihiro 1 Oshima Masaharu 1,2,3 Fukidome Hirokazu 4 Suemitsu Maki 4 Nagashio Kosuke 5 Toriumi Akira 5 1 Department of Applied Chemistry, Graduate School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656 , Japan 2 Synchrotron Radiation Research Organization, The University of Tokyo , 1-1-1 Kouto, Sayo-cho, Sayo-gun, Hyogo 679-5198, Japan 3 Core Research for Evolutional Science and Technology, Japan Science and Technology Agency , 5-7, Goban-cho, Chiyoda-ku, Tokyo, 102-0076, Japan 4 Research Institute of Electrical Communication, Tohoku University , 2-1-1 Katahira, Aoba-ku, Sendai 980-8577, Japan 5 Department of Materials Engineering, Graduate School of Engineering, The University of Tokyo , 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan a) Present address: Institute of Multidisciplinary Research for Advanced Materials, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai 980-8577, Japan. Electronic mail: nagamura@tagen.tohoku.ac.jp . 17 06 2013 102 24 241604 03 02 2013 07 05 2013 17 06 2013 2013-06-17T11:36:58 2013 AIP Publishing LLC 0003-6951/2013/102(24)/241604/5/ $30.00 Nanoscale spectromicroscopic characterizing technique is indispensable for realization of future high-speed graphene transistors. Highly spatially resolved soft X-ray photoelectron microscopy measurements have been performed using our “3D nano-ESCA” (three-dimensional nanoscale electron spectroscopy for chemical analysis) equipment in order to investigate the local electronic states at interfaces in a graphene device structure. We have succeeded in detecting a charge transfer region at the graphene/metal-electrode interface, which extends over ∼500 nm with the energy difference of 60 meV. Moreover, a nondestructive depth profiling reveals the chemical properties of the graphene/SiO 2 -substrate interface. 7402 2009-2012 crossmark 2013-06-17T11:36:58 Graphene has attracted great interest since the experimental realization of high-quality two-dimensional graphene is obtained easily by mechanical peeling from graphite 1 or thermal decomposition of SiC surfaces. 2,3 Graphene-based materials are actively studied for post-silicon electronics due to its exceptional carrier mobility, mechanical superiority which realize an atomic layer transistor channel without short channel effects, and intriguing band structures which can be modified. 4 However, the mobility of the present graphene field effect transistor (FET) on SiO 2 /Si substrates is far restrained from the desired value, and this results in a deterioration of gate-voltage response of the device. With respect to this limitation, a graphene/metal-electrode contact is considered to be important because a very small density of states near Fermi level in graphene may suppress the current injection from metal to graphene. Therefore, the key to improve the device performance is to understand properties of the interface between graphene and other materials in graphene nanostructures. Indeed, graphene/metal-electrode interfaces have been intensively studied. 5–9 The Fermi level of graphene attached to the metal electrode shifts from the conical point due to the charge transfer near the graphene/metal interface. Several groups have observed the graphene/metal boundary regions by scanning probe microscopy (SPM) techniques, such as scanning photocurrent microscopy 10,11 and scanning gate microscopy. 12 Since SPM techniques can only reveal electronic states in a narrow energy region near the Fermi level so they are not good at spectroscopic identification of elements. Additionally, these measurements are limited to lateral observations without depth-profile analysis. Moreover, the effect of surface contaminants can be hardly avoided especially for the device samples which may be broken during typical sample cleaning processes, e.g., flashing at elevated temperatures (1273 K). Innovative techniques which can be applied to in situ local state analysis of actual device structures have been eagerly needed. In understanding device properties, one of the most powerful tools is photoelectron spectroscopy or electron spectroscopy for chemical analysis (ESCA) which can reveal core-level electronic states for chemical bonding states, chemical-potential shifts, the work function at the surface, and electronic structures. Recently, we have developed a unique ESCA system,“3D nano-ESCA,” 13 equipped with a focused X-ray probe and an angle-resolved photoelectron spectrometer (ARPES) to realize nanoscale three-dimensional spatial distribution analysis. Its extremely high spatial resolution as a scanning photoelectron microscope (SPEM) system helps us to perform quantitative analysis of the target area of interest. In the present study, we have performed nondestructive 3D highly spatial and energy resolved ESCA study of graphene nanodevice structures using the 3D nano-ESCA in order to investigate the charge transfer region (CTR) at the graphene/metal-electrode interface and the chemical states at the graphene/substrate interface. Instead of average information acquired from the previous transport measurements and spectroscopic studies of graphene, pinpoint direct analysis of a functional part by chemical-state-selective information obtained from core level spectra is available by our technique. 14 We use exfoliated monolayer graphene sheets from Kish graphite transferred to a SiO 2 /p + -Si substrate. O 2 plasma treatment was carried out to remove hydrocarbon contaminants on the SiO 2 substrate. 15 This process made the substrate surface hydrophilic. Ni electrodes were fabricated by vacuum deposition after standard electron-beam lithography and lift-off techniques. For pinpoint depth-profile analysis, we use the 3D nano-ESCA system which has an X-ray focusing optical system with a Fresnel zone plate and a vibration isolation system for much better spatial resolution. The equipment has been installed at the University-of-Tokyo outstation beamline, BL07LSU at SPring-8. 16 The spatial resolution better than 70 nm has been achieved in this system. In order to obtain angular dependence of the photoelectron spectra for the depth-profiling, a modified VG Scienta R3000 analyzer with an acceptance angle of 60° is adopted. Figure 1(a) shows an optical microscope image of the monolayer graphene sheet attached to Ni electrodes on a hydrophilic SiO 2 thin film (90 nm thickness) on a p + -Si(100) substrate. 15 Figure 1(b) shows a SPEM image of the sample taken at the photon energies of 1000 eV. The SPEM contrast reflects the X-ray photoelectron spectroscopy (XPS) intensity of C 1 s peak (red in Fig. 1(b) ) distinguished from surface contamination in a way described below and Si 2 p peak (green in Fig. 1(b) ). We can thus distinguish the graphene sheet clearly from the metal electrodes and the SiO 2 substrate region by using the 3D nano-ESCA. XPS spectra of C 1 s taken at the monolayer graphene sheet are represented in Figure 2(a) . The following results of SR-XPS measurements (Figs. 2–4 ) are all taken at excitation photon energy of 1000 eV. The C 1 s spectrum is clearly decomposed into two peaks: one which has larger intensity at the lower binding energy (BE) of 284.24 eV (peak I in Fig. 2(a) ) and the other which has smaller intensity at the higher BE of 285.40 eV (peak II in Fig. 2(a) ). The former is considered to be attributed to the graphene sp 2 component 17,18 and the latter to the surface contaminants from polymer residue in device fabrication process and naturally involved amorphous carbon such as absorbed carbon dioxide, aliphatic hydrocarbon, and defects. 18–20 An asymmetric Doniach-Sunjic 21 line shape is used as a fitting function of the graphene peak. In order to confirm the origin of each component, the depth profiling of electronic states using the angular distribution of core-level photoemission spectra has been measured. Fig. 2(b) is the emission angle ( θ e ) dependence of the C 1 s spectra taken in the middle of the monolayer graphene sheet on the hydrophilic SiO 2 film (Fig. 1(c) ). Here, the intensity of photoelectrons from a given depth, z , at the emission angle from a surface normal, θ e , is expressed by I = I 0 exp ( − z λ cos θ e ) , (1) where λ is the inelastic mean free paths of the solids. As θ e becomes larger, the attenuation factor of the photoemission intensity that is λ cos θ e decreases, which results in surface sensitive measurements by decreasing information from a deeper (larger z ) region. As shown in Figs. 2(b) and 2(c) , the intensity of the peak II increases compared to that of the peak I as θ e becomes larger, suggesting that the component of peak II originates from surface contaminants on the graphene which yield the peak I, that is the graphene sp 2 component. Therefore, we have succeeded in determining the BE and the peak shape of the component purely from a graphene sp 2 bond configuration precisely excluding the effect of contaminations. In the next step, we have focused on the graphene/metal-electrode interface. When a metal electrode and a graphene sheet are brought into contact, charges are transferred through the graphene/metal interface, which leads to lineup of the Fermi levels on both sides and form a dipole layer at the graphene/metal interface. The amount of charge transfer gradually decreases from the graphene/metal interface. The density of states of graphene at the Fermi level is very small so that a long screening length is needed to cancel the potential difference. In order to investigate the spatial variation of the electronic state of monolayer graphene in the vicinity of the graphene/metal-electrode boundary, we have performed nanoscale XPS measurements to plot the line-profile for the BE of the graphene peak (C 1 s peak at the lower BE in Fig. 2(a) ) around the graphene/metal-electrode boundary. The focused light probe with SR is scanned along the dashed line represented in the inset of Figure 2 . The energy shift of the graphene peak by ∼60 meV is clearly detected in the monolayer graphene on the hydrophilic SiO 2 substrate (Fig. 1(c) ), as shown in Fig. 3 . 14 The peak position moves gradually from the graphene/metal-electrode contact toward the middle of the graphene sheet by about 500 nm. This result can be regarded as the direct observation of “a charge transfer region” at the graphene/metal-electrode boundary. 5 To confirm the above-mentioned interpretation of experimental results, we have conducted a theoretical approach in terms of the charge screening in a graphene sheet at the contact region with a metal strip suggested by Khomyakov et al. 22 The band bending in the CTR has been described quantitatively by calculating the screening potential using density-functional theory within the Thomas-Fermi approximation. According to them, the spatial variation of electrostatic potential induced in graphene (x ≥ 0) can be expressed as V ( x ) ∼ Δ E ( x l s + 0.016 + 0.787 ) 1 2 ( x l s + 1.194 ) 1 4 , (2) for undoped graphene, where Δ E is the boundary potential constant derived from the work functions calculated for the close-packed surface of the metal electrode and for the graphene sheet. l s = 8.10 × 10 − 2 × κ / Δ E (nm) indicates a scaling length, where κ is a background dielectric constant in a graphene device structure. Curve fitting has been performed to the experimental data using Eq. (2) with two fitting parameters: Δ E (eV) and κ. The comparison between the calculated screening potential and the peak shift in XPS measurements is presented in Fig. 3 . In the fitting results, the coefficients are obtained as Δ E ∼ 0.29 eV , κ ∼ 14 , and l s ∼ 3.9 nm . The deduced Δ E is consistent with the boundary potential in case of metal electrodes. 22 However, the effective dielectric constant is estimated far larger than the reported typical value, κ ∼ 2.5 , which is given by the average of the static dielectric constant of SiO 2 and that of the vacuum due to the image effect. 23 This large κ estimated in the experimental results should be affected by the substrate. So we have examined the spatial distribution of chemical bonding states for silicon and oxygen at the graphene/substrate interface using the depth profiling analysis. Figure 4(a) indicates the Si 2 p spectrum which consists of a single peak at 103.5 eV (peak III), while O 1 s spectrum is deconvolved as three components centered at 531.3 (peak IV), 532.8 (peak V), and 534.3 eV (peak VI) with Voigt functions as shown in Fig. 4(b) . The peak in the Si 2 p spectrum and the main peak at 532.8 eV in the O 1 s spectrum correspond to Si-O bonds in the SiO 2 film under the graphene sheet. 24 In order to identify the cause of each component in O 1 s core level, the emission angle dependence of the O 1 s spectrum on the monolayer graphene sheet has been investigated as presented in Fig. 4(c) . Figure 4(d) displays the relative core-level intensity ratio of the subpeak IV at the lowest BE and the subpeak VI at the highest BE to the bulk SiO 2 peak (peak V) in O 1 s spectrum as a function of θ e . Both intensity ratios of peak IV/peak V and peak VI/peak V increase monotonically with the increase of θ e . Moreover, the ratio of peak IV/peak V changes more steeper than that of peak VI/peak V. It means that the origin of each component is laid in the order of: IV, VI, V, from the top. The depth profiling of O 1 s core level revealed two different chemical bonding states, peak IV and peak VI, on top of the SiO 2 film (peak V). The peak IV at the topmost surface is assigned to the O 1 s for oxygen-containing carbon contaminants caused by exposure to ambient air 25 and polymer residue in device fabrication process. 17 The peak VI is considered to be due to Si-OH bonds of silanol groups 26 with highly polarization at the graphene/substrate interface. The static dielectric constant of the SiO 2 can be raised by hydrophilic silanol groups and water molecules on SiO 2 film. The polarization inside a graphene sheet depends on the graphene/substrate interface. It means that the state of the interface change the CTR through the effective dielectric constant and then affect transport properties of the graphene device structure. The graphene sheet and the SiO 2 film substrate are not chemically bonded so that the graphene sheet is easily released from the substrate by hydrofluoric acid (HF) etching to fabricate a suspended graphene device. 27,28 Other atoms or molecules could be introduced between the graphene sheet and the substrate. As the background dielectric constant κ which depends on the condition of the graphene/substrate interface increases, the range of the CTR at the graphene/metal-electrode interface become longer according to Eq. (2) . Furthermore, κ is related to the polarization of electrons in the valence band due to virtual interband transitions into the conduction band, as U ∼ e 2 2 π κ γ , (3) where U is a renormalized Coulomb interaction parameter and γ is a band parameter, respectively. 29 The electrical conductivity is inversely proportional to U − 2 , and then proportional to κ 2 . Although further studies are needed, our results strongly suggest a new possibility of controlling the CTR at the graphene/metal-electrode interface and the transport properties of graphene devices by adopting various graphene/substrate interface layers in graphene device structures. The CTR's behavior is considered to be varied under a finite bias voltage and on a hydrophobic substrate surface without polarization. Our results help us to understand asymmetric transfer characteristics (gate-voltage dependence of the drain current) of graphene FETs. The long CTR we observed works as a p-n junction. When a positive gate voltage is applied to a graphene FET, a p-n-p junction which causes an additional contact resistance is formed in a graphene channel, while no p-n junction is formed for a negative gate voltage as explained in the previous works. 5,9 If a channel length is shorter than twice the width of the CTR, the intrinsic region in the channel disappears, and then larger distortion of transfer characteristics and contact resistance are induced. The quantitative estimation of the CTR predicts the anomalous distortion of transport properties and the proper channel length for graphene FETs. In summary, the spatially resolved investigation of electronic and chemical states in a graphene device structure has been performed using the 3D nano-ESCA system with high spatial and energy resolutions without the effect of surface contaminants. We have undoubtedly demonstrated the CTR at the graphene/metal-electrode interface by detecting a core level shift as a good probe. The CTR extends into the graphene for about 500 nm with the electrostatic variation of 60 meV. Furthermore, the chemical bonding states at the graphene/SiO 2 film substrates interface have been investigated. It is revealed that the graphene/substrate interface can affect the CTR at the graphene/electrode interface. Our results clarify the electronic and chemical states of the interfaces which are essential for the device performance. We thank H. Kumigashira for helpful discussions. This work was supported by “Core Research for Evolutional Science and Technology (CREST)” of the Japan Science and Technology Agency (JST) and “Funding Program for World-Leading Innovative R&D on Science and Technology (FIRST Program)” of the Japan Society for the Promotion of Science (JSPS). This work is carried out by the joint research in the Synchrotron Radiation Research Organization, the University of Tokyo (Proposal No. 7402 for 2009-2012). FIG. 1. (a) Optical microscope image and (b) photoelectron intensity map of graphene device structures, which is a monolayer graphene sheet on a SiO 2 film attached to Ni electrodes. FIG. 2. (a) C 1 s core-level photoemission spectrum measured in the middle of the monolayer graphene sheet. Open circles represent the raw data and a gray solid curve is fitted data. The spectrum is well fitted by two components, peak I and peak II. (b) ARPES raw spectra of C 1 s taken at various emission angles for depth profiling. The intensity of angle-resolved spectra is normalized at the peak I intensity. (c) Emission angle dependence of the relative intensity ratio between two components in C 1 s spectrum: peak II/peak I. FIG. 3. Line profile for the binding energy peak position of the graphene component (peak I in Figure 2(a) ) taken along the dashed white line shown in the inset mapping image. The energy shift of ∼60 meV occurs at the graphene/metal-electrode interface. Red circles indicate the results of peak fitting. A gray curve represents the results of calculation using Eq. (2) . FIG. 4. (a) Si 2 p and (b) O 1 s core-level photoemission spectra measured at the same region in Fig. 2 . (c) ARPES raw spectra of O 1 s taken at various emission angles for depth profiling. (d) Emission angle dependence of the relative intensity ratio among three components in O 1 s spectrum: peak IV/peak V (filled circle) and peak VI/peak V (open circle).

PY - 2013/6/17

Y1 - 2013/6/17

N2 - Nanoscale spectromicroscopic characterizing technique is indispensable for realization of future high-speed graphene transistors. Highly spatially resolved soft X-ray photoelectron microscopy measurements have been performed using our "3D nano-ESCA" (three-dimensional nanoscale electron spectroscopy for chemical analysis) equipment in order to investigate the local electronic states at interfaces in a graphene device structure. We have succeeded in detecting a charge transfer region at the graphene/metal-electrode interface, which extends over ∼500 nm with the energy difference of 60 meV. Moreover, a nondestructive depth profiling reveals the chemical properties of the graphene/SiO2-substrate interface.

AB - Nanoscale spectromicroscopic characterizing technique is indispensable for realization of future high-speed graphene transistors. Highly spatially resolved soft X-ray photoelectron microscopy measurements have been performed using our "3D nano-ESCA" (three-dimensional nanoscale electron spectroscopy for chemical analysis) equipment in order to investigate the local electronic states at interfaces in a graphene device structure. We have succeeded in detecting a charge transfer region at the graphene/metal-electrode interface, which extends over ∼500 nm with the energy difference of 60 meV. Moreover, a nondestructive depth profiling reveals the chemical properties of the graphene/SiO2-substrate interface.

UR - http://www.scopus.com/inward/record.url?scp=84879832850&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84879832850&partnerID=8YFLogxK

U2 - 10.1063/1.4808083

DO - 10.1063/1.4808083

M3 - Article

AN - SCOPUS:84879832850

SN - 0003-6951

VL - 102

JO - Applied Physics Letters

JF - Applied Physics Letters

IS - 24

M1 - 241604

ER -

Direct observation of charge transfer region at interfaces in graphene devices

Abstract

Access to Document

Other files and links

Fingerprint

Cite this