You are on page 1of 7



Synthesis of Graphene on Silicon Dioxide by a Solid Carbon Source

Jens Hofrichter,*, Bartholomaus N. Szafranek, Martin Otto, Tim J. Echtermeyer, Matthias Baus, Anne Majerus, Viktor Geringer, Manfred Ramsteiner, and Heinrich Kurz
Advanced Microelectronic Center Aachen (AMICA), AMO GmbH, Otto-Blumenthal-Strasse 25, D-52074 Aachen, Germany, II. Physikalisches Institut B, RWTH Aachen and JARA FIT, D-52056 Aachen, Germany, and Paul-Drude-Institut fur Festkorperelektronik, Hausvogteiplatz 5-7, D-10117 Berlin, Germany
ABSTRACT We report on a method for the fabrication of graphene on a silicon dioxide substrate by solid-state dissolution of an overlying stack of a silicon carbide and a nickel thin lm. The carbon dissolves in the nickel by rapid thermal annealing. Upon cooling, the carbon segregates to the nickel surface forming a graphene layer over the entire nickel surface. By wet etching of the nickel layer, the graphene layer was allowed to settle on the original substrate. Scanning tunneling microscopy (STM) as well as Raman spectroscopy has been performed for characterization of the layers. Further insight into the morphology of the layers has been gained by Raman mapping indicating micrometer-size graphene grains. Devices for electrical measurement have been manufactured exhibiting a modulation of the transfer current by backgate electric elds. The presented approach allows for mass fabrication of polycrystalline graphene without transfer steps while using only CMOS compatible process steps. KEYWORDS Graphene, phase segregation, carbon electronics, Raman spectroscopy, scanning tunneling microscopy

ince the discovery of graphene by Geim and Novoselov in 2004,1 the research on graphene has gained enormous momentum.2 In their experiments, graphene was obtained by mechanical cleaving of natural graphite with adhesive tapes and the successive transfer to the desired substrate leading to a random deposition of graphene akes with both low surface coverage and reproducibility. This method is widely used in research and has led to the observation of a variety of new physical effects in graphene,3,4 but it does not lend itself to mass production. There has been considerable effort in the synthesis of largearea graphene sheets in recent years. High-quality singleand few-layer graphene sheets have been grown by decomposition of single-crystal silicon carbide5 and chemical vapor deposition on single-crystal transition metals such as iridium6 and ruthenium.7 These methods are expensive due to the high cost of the substrates and the ultrahigh vacuum (UHV) process conditions used. Furthermore, the size of the available substrates is relatively small compared to, e.g., silicon wafers. An approach better suited for large-scale application has been ambient-pressure chemical vapor deposition (CVD) of hydrocarbon gases on polycrystalline transition metals. Large-area graphene sheets have been fabricated in this manner on nickel8-10 and copper thin lms.11 With recent advances in this method, single-layer and bilayer graphene coverage of up to 87% of the lm area on nickel12 and
* Corresponding author. Present address: IBM Research GmbH, Zurich Research Laboratory, Saumerstrasse 4, CH-8803 Switzerland. E-mail: Telephone: + 41 44 724 8560. Received for review: 08/6/2009 Published on Web: 12/22/2009
2010 American Chemical Society

95% single-layer coverage on copper foils13 has been achieved. The growth mechanism on nickel is based on gas decomposition and diffusion of atomic carbon into the nickel with subsequent carbon precipitation to the nickel surface upon cooling. With copper, on the other hand, the growth of graphene is self-limiting and appears to be based on a surface-catalyzed process rather than a precipitation process as with nickel.13 The graphene lm can be transferred to arbitrary substrates by wet etching of the underlying metal lm. To support the thin graphene lm during the transfer, a poly(dimethylsiloxane) (PDMS) stamp9 or a supportive coating, e.g., a poly(methyl methacrylate) (PMMA) lm,12 on the graphene can be used. A slightly different approach to graphene synthesis on metal surfaces has been recently published, in which the carbon is supplied by a solid-phase method using a silicon carbide substrate as carbon source.14 In this process a 200 nm nickel thin lm is deposited on SiC via electron beam evaporation. Two kinds of substrates were used: singlecrystal 6H-SiC (0001) and 3C-SiC lms on Si substrates. With both substrates, the samples are rapidly heated in vacuum (10-7 mbar) to 750 C and immediately cooled to room temperature. During the heating process both silicon and carbon dissolve in the nickel, forming nickel silicide saturated with atomic carbon. Upon cooling the carbon segregates to the surface of the nickel/nickel silicide lm, forming single- to few-layer graphene. Here, we present an approach for the synthesis of graphene on an oxidized silicon substrate. Instead of singlecrystal SiC substrates or multicrystalline SiC lms,14 amorphous SiC thin lms were used as a carbon source deposited on a thermally oxidized silicon substrate. In order to reduce
DOI: 10.1021/nl902558x | Nano Lett. 2010, 10, 36-42

FIGURE 1. Surface morphology of the produced layers. (a) Optical microscope image. SEM images of the fabricated layers: (b) side view, (c) top view. AFM images of polygraphene: (d) phase contrast imaging, (e) topography imaging, (f) close-up of one graphene grain.

the carbon concentration in the nickel lm deposited on the SiC lm, we chose a thin SiC layer (50 nm) and a 500 nm thick nickel lm. The graphene lm is formed by rapid thermal annealing (RTA) of the sample at ambient pressure. We found that vacuum is not necessary for graphene formation. During the process, high temperatures of 1100 C are required to completely dissolve the SiC layer in the nickel, which is important for controlling the amount of carbon dissolved in the nickel. A detailed analysis of the fabricated graphene layers was performed, including atomic force microscopy (AFM), scanning tunneling microscopy (STM), Raman spectroscopy, Raman mapping, and a comparison to mechanically exfoliated graphene. Additionally, electrical characterization of micrometer-scaled devices made from the produced graphene layers was performed. We chose thermally grown SiO2 on Si with a thickness of 300 nm as the starting material. In our experiment, 50 nm of amorphous silicon carbide (SiC) is deposited onto the silicon dioxide layer by plasma-enhanced chemical vapor deposition (PECVD). After deposition of a 500 nm thick nickel layer by dc magnetron sputtering, the sample is annealed at 1100 C for 30 s in ambient pressure and
2010 American Chemical Society

nitrogen environment in a rapid thermal annealing (RTA) oven. During this time both the silicon and the carbon of the SiC are dissolved in the nickel layer. By choosing a nickel layer much thicker than the SiC layer, the silicon from the SiC is only dissolved in nickel or forms substoichiometric silicides. The SiC thin lm is completely consumed by this process. Thus the carbon content in the nickel layer can be controlled by the ratio of SiC to Ni layer thickness. However, experiments lowering the SiC thickness by inverse sputtering indicate that carboneous lms are formed but are not of a closed shape thus hindering a wet chemical processing needed for device fabrication and characterization. A certain amount of a solid carbon source has to be present for the formation of a closed layer. The SiO2 substrate has a chemically and thermally stable interface and does not release any Si atoms into the Ni. When the sample is cooled to room temperature, the carbon segregates to the surface of the Ni layer forming a graphene layer. Successively, the nickel layer with the dissolved silicon was wet chemically removed. The lower chemical stability of substoichiometric silicides allows for the usage of hydrogen uoride free etchants. In our experiment, we removed the silicon-nickel layer by a nitric
DOI: 10.1021/nl902558x | Nano Lett. 2010, 10, 36-42

acid and hydrogen peroxide containing etchant. It is possible to transfer the graphene to arbitrary substrates by methods described in refs 11 and 12. However, we chose to let the graphene settle on the original SiO2/Si substrate instead of using a transfer process. Optical microscopy revealed the marmorated nature of the produced layers as shown in Figure 1a. Thin graphene regions are present in the purple areas of the image, whereas the light blue areas indicate thicker graphitic regions. Therefore the produced material will be referred to as polygraphene. Analysis has also been performed by scanning electron microscopy (SEM). As illustrated in Figure 1b, freestanding regions of a thin lm, through which imaging could be performed, could be observed at the borders of the cleaved-facet samples. Again, the marmorated nature of the produced lms becomes evident when taking top-view SEM images as in Figure 1c. Further investigation of the produced layers has been performed by AFM. Topography imaging reveals a mean lm thickness of 6.5 nm and 15 nm in the thickest regions. There are also several areas in the image where no discernible height difference to the substrate could be measured. These could be either regions with very thin thickness or voids in the lm. However, different materials can be distinguished by phase contrast imaging. The phase imaging, presented in Figure 1d, shows that the layer is a closed lm except for a small void in the top left corner of the image. The darker regions are the carbon-containing lm while the lighter yellow region represents a different material, i.e., silicon dioxide. Hence, the lm is either corrugated or of varying thickness having thin regions. Figure 1f displays also a topography AFM image, here with a higher resolution. The wrinkles are elevated and can be attributed to folding of the graphite induced by the different expansion coefcients of nickel and graphite during cooling down after the high-temperature process. STM analysis was used to gain further insight into the microscopic structure of the produced layers and to distinguish between single- and few-layer graphene by comparison with exfoliated natural graphite. Figure 2 displays the atomic structure of the several types of graphene. An artists view of a single graphene layer on top of an appropriate transition metal surface is illustrated in Figure 2a. E.g., the nickel (111) surface corresponds to this image. STM images have been taken in an UHV chamber at room temperature for a bilayer of mechanically exfoliated graphene (Figure 2b) and a monolayer of graphene,15 which is depicted in Figure 2c. Single-layer graphene exhibits the typical honeycomb one expects for a perfect hexagonal lattice. In contrast, several layers of graphene present a triangular lattice structure by a modication of the surface density of states by the underlying layers.16 Also, STM imaging has been performed on the layers fabricated by the method presented in this paper. Figure 2d illustrates such an image which contains a perfect hexagonal lattice with no visible lattice defects. The lattice constant of the synthetic graphene
2010 American Chemical Society

FIGURE 2. The atomic structure of graphene. (a) Artist view of a single layer of graphene on top of a nickel (111) surface, the other gures present STM images of different types of graphene. (b) Few layer exfoliated graphene. (c) Single layer of exfoliated graphene. (d) Graphene single layer produced by the process described in this work.

layers has been determined to 2.49 ( 0.07 , perfectly matching the theoretical value of graphene of 2.46 and the lattice constant of nickel being 2.49 . The band structure is the electronic ngerprint of materials. Taking advantage of resonance effects, Raman spectroscopy is an appropriate method to extract characteristic features of the band structure for different materials.17 In particular, graphite and graphene can be distinguished easily and Raman ngerprints for single layers, bilayers, and few layers reect the changes in the band structure (ref 18 and the references therein). The most intense Raman features in graphite and graphene are the G-peak at about 1600 cm-1 and the 2D-peak at about 2700 cm-1. The G-peak is due to rst-order Raman scattering by the doubly degenerate zonecenter optical phonon mode and the 2D-peak is associated with second-order scattering by zone-boundary phonons.18 In rst-order Raman scattering, the observation of zoneboundery phonons is inhibited by selection rules. However, in defective graphite, the so-called D-peak at about 1350 cm-1 due to rst-order scattering by zone-boundary phonons can be observed. The relatively low intensity or the absence of the D-peak in the spectra of graphene layers reects the small concentration of defects. The Raman ngerprint for single-layer graphene, however, is connected with the position and width of the 2D-peak, which reects the electronic band structure characteristics via a double-resonant Raman scattering process.18 The position of the 2D-peak depends on the incident laser wavelength and for excitation at 632.8 nm a single, relatively narrow and intense 2D-peak around 2700 cm-1 is the signature of single-layer graphene. Strain
DOI: 10.1021/nl902558x | Nano Lett. 2010, 10, 36-42

FIGURE 3. Raman spectra excited at 632.8 nm for (a) rst-order and (b) second-order scattering. The D- and G-peaks are observed around 1350 and 1600 cm-1, respectively, and the 2D-peak at about 2700 cm-1. The width of the 2D-peak for exfoliated single-layer graphene and the minimum 2D-peak width for graphene layers presented in this work are 19 and 30 cm-1, respectively.

induced by the substrate, however, leads to a shift of the 2Dpeak, ruling out the peak position as a measure for the number of layers in our case. Lee et al. have shown that the peak width is decisive for the number of layers of graphene produced by the decomposition of silicon carbide (SiC).19 Hence, their approach of analyzing the produced layers has been used as it allows for the identication of the number of layers while being insensitive to induced stress.19 According to ref 19, the lowest observed full width at half-maximum (fwhm) has been assigned to single-layer graphene while the fwhm given in their paper for bulk graphite has been associated to lms with more than 10 layers. Spectra of rst- and second-order Raman scattering from our lms are shown in parts a and b of Figure 3, respectively. For excitation, a laser wavelength of 632.8 nm has been used. As a reference, the sprectrum of a mechanically exfoliated monolayer of graphene is shown in addition. Figure 3c displays the positions on the sample from which the spectra have been aquired. The position Poly A is a monolayer region whereas Poly B is a multilayer region. Neither the exfoliated monolayer nor our lms show a signicant D-peak around 1350 cm-1, but a strong G-peak around 1600 cm-1 indicating excellent crystallinity. In Figure 3b, the fwhm of the 2D-peak has been determined to be 30 cm-1 for synthetic graphene and 19 cm-1 for the exfoliated monolayer of graphene. The larger fwhm of the 2D-peak of the synthetic graphene can be explained by inhomogeneous stress induced by the high-temperature process, which does
2010 American Chemical Society

not relax upon cooling. This phenomenon is also observed for graphene grown on SiC (ref 20). To obtain further insight in the nature of our lms, Raman mapping has been performed with the scattered light collected by a confocal microscope. Figure 4 illustrates for an area of 46 36 m2 the Raman intensities of (a) the D-peak around 1350 cm-1, (b) the G-peak around 1600 cm-1, and (c) the 2D-peak around 2700 cm-1. The peak width (FWHM) of the 2D-peak around 2700 cm-1 is displayed in Figure 4d and the corresponding number of layers according to ref 19 is shown in Figure 4e, where the dark blue regions indicate mono- or bilayer graphene. This nding is also consistent with the ratio of the 2D-peak intensity to the G-peak intensity displayed in Figure 4f. The AFM results, Raman mapping, and AFM consistently indicate a produced graphene crystallite size in our lms of about 5-10 m. Besides regions consisting of multiple layers of graphene, single layer areas are clearly evident. Electrical characterization of the produced layers has been performed by contacting a 2.0 m wide polygraphene ribbon with metal ngers. First, 50 nm thick nickel contacts have been fabricated on top of the polygraphene layers by a lift-off process. Subsequently, a 1.5 m thick AZ 5214A resist has been deposited and developed serving as a mask for the graphene structuring. The patterning has been performed in an anisotropic oxygen-argon plasma. After the transfer of the structures, the resist has been removed and the samples baked out for 12 h at 300 C in a vacuum
DOI: 10.1021/nl902558x | Nano Lett. 2010, 10, 36-42

FIGURE 4. Raman mapping data for an area of 46 36 m2: (a) D-peak (1350 cm-1) intensity, (b) G-peak (1600 cm-1) intensity, (c) 2D-peak (2700 cm-1) intensity, (d) width of the 2D-peak, and (e) the assigned number of layers according to ref 18. (f) Intensity ratio of the 2D-peak to the G-peak. Note the defect in the center, which can be been in (b) and (c) as a minimum in the G-peak and 2D-peak, respectively.

chamber. The nal device is displayed in Figure 5a. Electrical measurements have been performed with an HP 4156B semiconductor parameter analyzer. Figure 5b displays the transfer current characteristics obtained by applying a dc voltage between two adjacent contact ngers and sweeping the backgate voltage applied to the Si substrate as depicted in Figure 5d. The curve is similar to transfer current measurements of single-layer graphene, also showing a mini 2010 American Chemical Society

mum corresponding to the carge neutrality point at which the charge carrier density is minimal and switches from nto p-conducting state.1,21 On the other hand, the current modulation is much weaker than in single-layer graphene. In single-layer graphene the current should have a maximum to minimum ratio of about two in the measured range21 while the polygraphene curve shows a modulation value of up to 17.5% for the best device and typically of a few
DOI: 10.1021/nl902558x | Nano Lett. 2010, 10, 36-42

FIGURE 5. Electrical characterization of the presented polygraphene layers (a) Graphene eld effect transistor (GFET), on which the measurements have been performed. The width of the ribbon is 2.0 m. (b) Transfer current modulation by backgate elds (c) Modulation of the resistance slope for different backgate elds.

percent as in Figure 5b. This is due to the fact that the contacted material consists of a few graphene layers as in graphite. Thus, the carrier density is only controlled in the lowest graphene layers, while in the upper graphene layers it remains constant because of electronic shielding by the lower layers. Also, transport along grain boundaries is likely to contribute to the residual conductivity. Furthermore, contact resistances also lower the modulation value; however, their exctraction is diffucult due to the inhomogeneous nature of the layers. The charge neutrality point is shifted to positive voltages which is attributed to doping effects from the contacting metal, since Ni and adsorbed water both lead to p-doping of graphene. Figure 5c shows the Ohmic behavior of a typical device at constant gate voltages. In conclusion, we have developed a simple method to grow graphene on polycrystalline nickel surfaces using an amorphous SiC thin lm underneath the nickel layer as carbon source. The growth process is based on phase segregation of a carbon containing layer (SiC) and a solid solvent (Ni). The SiC dissolves in the nickel, completely consuming the SiC thin lm. By wet-etching the Ni layer, the graphene can either be transferred to arbitrary substrates or, as was done here, can be allowed to settle on the SiO2
2010 American Chemical Society

surface of the original substrate. The graphene layers were characterized by optical microscope imaging, SEM, AFM, STM, Raman spectroscopy, and Raman mapping. The analysis consistently shows that polygraphene layers have been produced with regions consisting of monolayers of graphene. Electricalcharacterizationofthelayersdemonstratedgraphenes electric eld effect. Further optimization of the growth process to obtain more uniform layers with larger monolayer coverage should be possible by controlling the amount of available carbon via carefully tuning the ratio of SiC to Ni lm thickness in conjunction with an optimized cooling rate,12 which cannot be varied in the present setup. Also the use of other transition metals, such as ruthenium, might lead to a reduced thickness variation within the graphene layers. Controlling the carbon concentration, the cooling rate, and the crystal orientation of the metal layer appear to be the most important parameters for highly uniform graphene lms on arbitrary substrates. Acknowledgment. The authors acknowledge the funding by the BMBF project ALEGRA and the European project GRAND as well as the technical support by D. Stegers and T. Welter.
DOI: 10.1021/nl902558x | Nano Lett. 2010, 10, 36-42


Novoselov, K. S.; Geim, A. K.; Morozov, V.; Jiang, D.; Zhang, Y.; Dubonos, S. V; Grigorieva, I. V.; Firsov, A. A. Electric eldeffect in atomically thin carbon lms. Science 2004, 306, 666669. (2) Geim, A. K.; Novoselov, K. S. The rise of graphene. Nat. Mater. 2007, 6, 183191. (3) Novoselov, K. S.; et al. Two-dimensional gas of massless Dirac fermions in graphene. Nature 2005, 438, 197200. (4) Zhang, Y.; Tan, J. W.; Stormer, H. L.; Kim, P. Experimental observation of the quantum Hall effect and Berrys phase in graphene. Nature 2005, 438, 201204. (5) Berger, C.; Song, Z.; Li, X.; Wu, X.; Brown, N.; Naud, C.; Mayou, D.; Li, T.; Hass, J.; Marchenkov, A. N.; Conrad, E. H.; First, P. N.; de Heer, W. A. Electronic Connement and Coherence in Patterned Epitaxial Graphene. Science 2006, 312 (5777), 11911196. (6) NDiaye, A. T.; Coraux, J.; Plasa, T. N.; Busse, C.; Michely, T. Structure of epitaxial graphene on Ir (111). New J. Phys. 2008, 10, 043033043049. (7) Sutter, P. W.; Fliege, J. I.; Sutter, E. L. Epitaxial graphene on ruthenium. Nat. Mater. 2008, 7, 406411. (8) Yu, Q.; Lian, J.; Siriponglert, S.; Li, H.; Chen, Y. P.; Pei, S. S. Graphene segregated on Ni surfaces and transferred to insulators. Appl. Phys. Lett. 2008, 93 (11), 3103131033. (9) Kim, K. S.; Zhao, Y.; Jang, H.; Lee, S. Y.; Kim, J. M.; Kim, K. S.; Ahn, J.-H.; Kim, P.; Choi, J. Y.; Hong, B. H. Large-scale pattern growth of graphene lms for stretchable transparent electrodes. Nature 2009, 457, 706710. (10) Reina, Alfonso; Jia, Xiaoting; Ho, John; Nezich, Daniel; Son, Hyungbin; Bulovic, Vladimir; Dresselhaus, Mildred S.; Kong, Jing. Large Area, Few-Layer Graphene Films on Arbitrary Substrates by Chemical Vapor Deposition. Nano Lett. 2009, 9 (1), 3035. (11) Wei, D.; Liu, Y.; Wang, Y.; Zhang, H.; Huang, L.; Yu, G. Synthesis of N-Doped Graphene by Chemical Vapor Deposition and Its Electrical Properties. Nano Lett. 2009, 9 (5), 17521758. (1)

(12) Reina, A.; Thiele, S.; Jia, X.; Bhaviripudi, S.; Dresselhaus, M. S.; Schaefer, J. A.; Kong, J. Growth of Large-Area Single- and Bi-Layer Graphene by Controlled Carbon Precipitation on Polycrystalline Ni Surfaces. Nano Res. 2009, 2, 509516. (13) Li, X.; Cai, W.; An, J.; Kim, S.; Nah, J.; Yang, D.; Piner, R.; Velamakanni, A.; Jung, I.; Tutuc, E.; Banerjee, S. K.; Colombo, L.; Ruoff, R. S. Large-Area Synthesis of High-Quality and Uniform Graphene Films on Copper Foils. Science 2009, 324, 13121314. (14) Juang, Z.-Y.; Wu, C.-Y.; Lo, C.-W.; Chen, W.-Y.; Huang, C.-F.; Hwang, J.-C.; Chen, F.-R.; Leou, K.-C.; Tsai, C.-H. Synthesis of graphene on silicon carbide substrates at low temperature. Carbon 2009, 47, 20262031. (15) Geringer, V.; Liebmann, M.; Echtermeyer, T.; Runte, S.; Schmidt, M.; Ruckamp, R.; Lemme, M.; Morgenstern, M. Intrinsic and extrinsic corrugation of monolayer graphene deposited on SiO2. Phys. Rev. Lett. 2009, (7), 102. (16) Mallet, P.; Varchon, F.; Naud, C.; Magaud, L.; Berger, C.; Veuillen, J. Y. Electron states of mono- and bilayer graphene on SiC probed by scanning-tunneling microscopy. Phys. Rev. B 2007, 76, 04140310414034. (17) Reich, S.; Thomsen, C. Raman spectroscopy of graphite. Philos. Trans. R. Soc. London 2004, 362, 22712288. (18) Ferrari, A. C.; Meyer, J. C.; Scardaci, V.; Casiraghi, C.; Lazzeri, M.; Mauri, F.; Pis-canec, S.; Jiang, D.; Kovoselov, K. S.; Roth, S.; Geim, A. K. Raman spectrum of graphene and graphene layers. Phys. Rev. Lett. 2006, 97, 18740111874014. (19) Lee, D. S.; Riedl, C.; Krauss, B.; von Klitzing, K.; Starke, U.; Smet, J. H. Raman spectra of epitaxial graphene on SiC and of epitaxial graphene transferred to SiO2. Nano Lett. 2008, 5 (12), 43204325. (20) 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, 1063710640. (21) Lemme, M. C.; Echtermeyer, T. J.; Baus, M.; Kurz, H. A graphene eld-effect device. IEEE Electron Device Lett. 2007, 28 (4), 282 284.

2010 American Chemical Society


DOI: 10.1021/nl902558x | Nano Lett. 2010, 10, 36-42