LETTER pubs.acs.org/NanoLett
Suppression of Inhomogeneous Segregation in Graphene Growth on Epitaxial Metal Films Shigeo Yoshii,*,§ Katsuya Nozawa,§ Kenji Toyoda, Nozomu Matsukawa, Akihiro Odagawa, and Ayumu Tsujimura Advanced Technology Research Laboratories, Panasonic Corporation, 3-4 Hikaridai, Seika, Kyoto 619-0237, Japan
bS Supporting Information ABSTRACT: Large-scale uniform graphene growth was achieved by suppressing inhomogeneous carbon segregation using a single domain Ru film epitaxially grown on a sapphire substrate. An investigation of how the metal thickness affected growth and a comparative study on metals with different crystal structures have revealed that locally enhanced carbon segregation at stacking domain boundaries of metal is the origin of inhomogeneous graphene growth. Single domain Ru film has no stacking domain boundary, and the graphene growth on it is mainly caused not by segregation but by a surface catalytic reaction. Suppression of local segregation is essential for uniform graphene growth on epitaxial metal films. KEYWORDS: Graphene, epitaxy, chemical vapor deposition, ruthenium, nickel, cobalt
G
raphene has attracted a great deal of attention due to its distinctive physical properties.1,2 Many efforts have been made to achieve large-scale and high-quality graphene growth. Chemical vapor deposition (CVD) on metal surfaces is becoming a popular method because of its scalability and cost effectiveness.313 However, the quality of CVD graphene on polycrystalline substrates seems to be limited by the size and uniformity of the underlying grains.5,9,11 Arbitrarily oriented metal crystal grains can produce differently oriented graphene grains. They form line defects at their boundaries, which degrade the electrical properties.10 Uniform graphene growth on single crystal substrates, compared to that on polycrystalline metal, has been reported.1114 Nevertheless, the limited size and the high cost of bulk crystal substrates diminish the scalability of CVD growth. Utilizing epitaxial metal films grown on substrates (e.g., sapphire) as an underlayer is a promising technique that satisfies both scalability and surface quality requirements for graphene growth.1518 Large-scale epi-ready sapphire substrates are commercially available. Some metals have crystal surfaces that are approximately lattice-matched both to graphene and c-sapphire (Supporting Information, Figure S1).1922 Ago et al. reported on an epitaxial Co underlayer,16 while Sutter et al. reported on Ru.17 Although there are many potential materials for the underlayer, the difference in graphene growth between metal species has not been clarified. It is essential to study the growth mechanism on epitaxial metals in order to assess the suitability for graphene growth. Two main processes in graphene CVD have been reported: one is a surface catalytic reaction at high temperature, and the other is carbon segregation in cooling.47,11,12 The surface catalytic r 2011 American Chemical Society
reaction is preferable for uniform graphene growth because of its self-terminating aspect. Precursors for graphene synthesis are provided by the catalytic decomposition of the carbon source molecule, which is suppressed in areas already covered by graphene.12 In contrast, in the segregation, carbon is provided from the underlying metal. This sometimes produces multilayered graphene (MLG) locally, which causes inhomogeneity.5,11 Both of these processes play a role in CVD, and it has been difficult to distinguish the contribution of each process. This paper reports our investigation of the dependence of graphene growth on metal thickness and the crystal domain structures of epitaxial metals with different crystal structures. The results revealed differences in the growth processes on metals, and the origin of inhomogeneity was found to be locally enhanced segregation at stacking domain boundaries. Large-scale uniform graphene growth was achieved with a single domain Ru film, on which the major growth process was attributed to the surface catalytic reaction. We used three kinds of metal with different crystal structures: Ni, Co, and Ru, for the epitaxial underlayers. Ni and Ru have facecentered cubic (fcc) and hexagonal close-packed (hcp) crystal structures, respectively. Co is hcp at room temperature, while it transforms to fcc at temperatures above 422 °C.23 The metals were deposited on c-sapphire substrates at room temperature. Ni and Co were deposited by electron-beam evaporation, and Ru was deposited by sputtering. Thicknesses of the Ni, Co, and Ru films were 70, 70, and 30 nm unless mentioned otherwise. The Received: February 21, 2011 Revised: May 13, 2011 Published: June 07, 2011 2628
dx.doi.org/10.1021/nl200604g | Nano Lett. 2011, 11, 2628–2633
Nano Letters
LETTER
Figure 1. XRD patterns of epitaxial metal films on c-sapphire substrates. The θ-2 θ scans of: (a) Ni, (b) Co, and (c) Ru films. Black and red lines in (ac) show the θ-2 θ spectra before and after the solid-phase epitaxy (SPE), respectively. The * represents peaks from sapphire (0006). Pole figures for: (d) Ni {200}, (e) Co {1013}, and (f) Ru {1013} reflections. Upright dashed triangles and inverted solid triangles in (d) indicate that the two sets of poles attributed to the two stacking domains of fcc coexist in the epitaxial Ni film.
Figure 2. Typical Raman spectra measured for graphene layers on epitaxial: (a) Ni, (b) Co, and (c) Ru films. The peaks at ∼2330 cm1 are due to nitrogen in ambient air. HRTEM images of: (d) bilayer graphene on Ni film, (e) Ni/sapphire interface, (f) trilayer graphene on Co film growing over single and double steps of metal surface, and (g) bilayer graphene on Ru film.
metals were then annealed in hydrogen typically at 1000 °C for 5 min for solid-phase epitaxial recrystallization (SPE). X-ray diffraction (XRD) analysis revealed that the Ni, Co, and Ru metals all formed epitaxial films on the c-sapphire in the SPE process. The θ-2 θ scans in Figure 1ac shows a peak from the (111) plane for fcc Ni and a peak from the (0002) plane for hcp Co and Ru. They were enhanced by the SPE process in hydrogen, but no peaks appeared from other crystal planes. Figure 1df shows pole figures of {200} fcc planes and {1013} hcp planes. Although an fcc crystal is expected to have only three {200} planes, six poles are observed for the Ni film. This pseudo six-fold symmetry shows a two-domain structure due to the two types of fcc stacking sequences, ABCABC ... and ACBACB ...18 Each of the two domains generates three poles, but they are separated by 60°. On the other hand, the hcp structure has six-fold {1013} planes; therefore, the six intense poles observed for the Ru film prove its single domain nature. The
Co film also shows an hcp structure, but it has 12 poles with broader peaks, supposedly due to its fcchcp phase transition in the cooling process. Graphene CVD growth was performed using the abovedescribed epitaxial metal films. After the aforementioned SPE, samples were cooled to 800 °C, at which no graphene growth was observed. A mixture gas of 15% methane and hydrogen was introduced into the furnace to replace the atmosphere. The samples were heated again to the CVD temperature (typically 1000 °C), kept for 5 min, and then cooled at a rate of 810 °C/sec, unless mentioned otherwise. XRD results of Ni and Ru films after CVD show no clear differences with those before CVD. In contrast, the Co film after CVD shows poles of fcc {200} planes that were absent before CVD (Supporting Information, Figure S2). Dissolved carbon atoms in the Co film would partially suppress the phase transition from fcc to hcp in the cooling process. 2629
dx.doi.org/10.1021/nl200604g |Nano Lett. 2011, 11, 2628–2633
Nano Letters
LETTER
Figure 3. Optical images and Raman results of epitaxial metal films after CVD. (a) A thin 40 nm Ni film was damaged by CVD growth, and G-peaks are rarely found on it. (b) 70 nm and (c) 150 nm Ni films show rather flat surfaces, and FLG Raman spectra are dominant on them. (d) 300 nm Ni film has many dark spots where MLG Raman spectra are observed. (e) 30 nm Ru film shows a flat surface, and an FLG Raman spectrum is measured on it. The sloping lines show the maximum number of graphene layers that can be generated by segregation in cooling for Ni and Ru. The blue horizontal line shows the segregation limit for monolayer growth. Below it (in light-blue area), the amount of carbon supplied by segregation is not sufficient to form a graphene sheet.
Raman spectra after CVD show graphene-derived G (∼1580 cm 1 ) and 2D (∼2700 cm1) peaks on all of the Ni, Ru, and Co films (Figure 2ac). The intensity ratio of the defectinduced D (∼1350 cm1) peak to the G peak (ID/IG) is comparable to that reported for CVD graphenes,5,11,1518 indicating that the growth here does not induce a higher defect density in graphene. Typical Raman spectra have the characteristics of few-layer graphene (FLG), i.e., a single-Lorentzian 2D peak with sharp line width ( Ni > Co, and it agreed with the order of simplicity of their domain structures in the EBSD and Nomarski images. The uniform growth on Ru film can be attributed to the surface flatness or the absence of boundaries. The inhomogeneous segregation, which was clearly observed on Ni, has been significantly suppressed on Ru film. The absence of boundaries due to the single domain nature of Ru prevented the local enhancement of segregation. The absence of local enhancement may also contribute to the reduction of the total amount of segregation, while there can be other factors affecting why the surface reaction played a major role on Ru. The self-terminating nature of the surface catalytic reaction could also contribute to the uniformity in growth. Cu is another material that is attracting attention as an underlayer for CVD.8,9 The carbon solubility in Cu is lower than that in Ru; hence, Cu is advantageous in terms of suppressing segregation. However, its crystal structure is fcc, the same as Ni, and a similar two-domain structure with boundaries appears when it is epitaxially grown on sapphire. Further study of the epitaxial Cu underlayer is required, but the difficulty in highquality graphene growth over such boundaries is anticipated even if segregation is suppressed. Our findings suggest that metals that have a hexagonal crystal structure, like Ru which readily form a boundary-free single domain hexagonal surface lattice, are promising for high-quality graphene growth. We confirmed the difference in uniformity on Ni and Ru films by transferring the graphene onto a 300 nm SiO2/Si substrate, which makes it possible to visualize the difference in the number of graphene layers (Figure 7).1,28 The transfer process was similar to that described in previous reports.5,17 Many dark spots are observed on the graphene grown on a Ni surface. Spots tend to lie in lines, which corresponds to the Raman mapping image before transfer (Figure 5g). Raman analysis confirmed that the dark spots consist of MLG, whereas the bright area consists of FLG. On the other hand, none of the MLG spots were found in the graphene grown on Ru. A millimeter-scale uniform graphene 2632
dx.doi.org/10.1021/nl200604g |Nano Lett. 2011, 11, 2628–2633
Nano Letters sheet was confirmed, as shown in Figure 7c, even though our transfer process is not yet optimized. In summary, we investigated graphene growth on epitaxial metals with varied metal thicknesses and different crystal structures. We found that locally enhanced carbon segregation takes place at stacking domain boundaries resulting in inhomogeneous graphene. The degree of uniformity of the graphene layer is consistent with the simplicity in the domain structure of the underlying epitaxial metals. A large-scale uniform graphene growth was achieved with single domain Ru film, on which domain boundaries were eliminated, and the graphene growth was mainly caused not by segregation but by a surface catalytic reaction. Suppression of local segregation is shown to be essential for uniform graphene growth on epitaxial metal films.
’ ASSOCIATED CONTENT
bS
Supporting Information. In-plane relationship of crystal surfaces, XRD pole figures of Co film before and after CVD, electron-beam diffraction patterns, and maximum number of graphene layers by segregation. This material is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected]. Telephone: þ81-77498-2580. Author Contributions §
These authors contributed equally.
LETTER
(11) Zhang, Y.; Gomez, L.; Ishikawa, F. N.; Madaria, A.; Ryu, K.; Wang, C.; Badmaev, A.; Zhou, C. J. Phys. Chem. Lett. 2010, 1 (20), 3101–3107. (12) Wintterlin, J.; Bocquet, M. L. Surf. Sci. 2009, 603 (1012), 1841–1852. (13) Sutter, P. W.; Flege, J. I.; Sutter, E. A. Nat. Mat. 2008, 7 (5), 406–411. (14) Loginova, E.; Bartelt, N. C.; Feibelman, P. J.; McCarty, K. F. New J. Phys. 2009, 11, 063046. (15) Ago, H.; Tanaka, I.; Orofeo, C. M.; Tsuji, M.; Ikeda, K. Small 2010, 6 (11), 1226–1233. (16) Ago, H.; Ito, Y.; Mizuta, N.; Yoshida, K.; Hu, B.; Orofeo, C. M.; Tsuji, M.; Ikeda, K.; Mizuno, S. ACS Nano 2010, 4 (12), 7407–7414. (17) Sutter, P.; Albrecht, P.; Sutter, E. Appl. Phys. Lett. 2010, 97 (21), 213101. (18) Iwasaki, T.; Park, H. J.; Konuma, M.; Lee, D. S.; Smet, J. H.; Starke, U. Nano Lett. 2011, 11 (1), 79–84. (19) Bialas, H.; Heneka, K. Vacuum 1994, 45 (1), 79–87. (20) Torndahl, T.; Lu, J.; Ottosson, M.; Carlsson, J. O. J. Cryst. Growth 2005, 276 (12), 102–110. (21) Yamada, S.; Nishibe, Y.; Saizaki, M.; Kitajima, H.; Ohtsubo, S.; Morimoto, A.; Shimizu, T.; Ishida, K.; Masaki, Y. Jpn. J. Appl. Phys. 2002, 41 (2B), L206–L208. (22) Dehm, G.; Inkson, B. J.; Wagner, T. Acta Mater. 2002, 50 (20), 5021–5032. (23) Erbudak, M.; Wetli, E.; Hochstrasser, M.; Pescia, D.; Vvedensky, D. D. Phys. Rev. Lett. 1997, 79 (10), 1893–1896. (24) Sirois, E.; Birnbaum, H. K. Acta Metall. Mater. 1992, 40 (6), 1377–1385. (25) Lander, J. J.; Kern, H. E.; Beach, A. L. J. Appl. Phys. 1952, 23 (12), 1305–1309. (26) Arnoult, W. J.; McLellan, R. B. Scr. Metall. 1972, 6, 1013–1018. (27) Hayat, S. S.; Choudhry, M. A.; Ahmad, S. A. J. Mater. Sci. 2008, 43 (14), 4915–4920. (28) Gray, A.; Balooch, M.; Allegret, S.; De Gendt, S.; Wang, W. E. J. Appl. Phys. 2008, 104 (5), 053109.
’ ACKNOWLEDGMENT The authors thank Dr. Daisuke Ueda and Dr. Mirco Cantoro for their valuable contributions to our discussions. ’ REFERENCES (1) Geim, A. K.; Novoselov, K. S. Nat. Mater. 2007, 6 (3), 183–191. (2) 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 (5696), 666–669. (3) Somani, P. R.; Somani, S. P.; Umeno, M. Chem. Phys. Lett. 2006, 430 (13), 56–59. (4) Yu, Q. K.; Lian, J.; Siriponglert, S.; Li, H.; Chen, Y. P.; Pei, S. S. Appl. Phys. Lett. 2008, 93 (11), 113103. (5) Reina, A.; Jia, X. T.; Ho, J.; Nezich, D.; Son, H. B.; Bulovic, V.; Dresselhaus, M. S.; Kong, J. Nano Lett. 2009, 9 (1), 30–35. (6) 1.Liu, W.; Chung, C. H.; Miao, C. Q.; Wang, Y. J.; Li, B. Y.; Ruan, L. Y.; Patel, K.; Park, Y. J.; Woo, J.; Xie, Y. H. Thin Solid Films 2010, 518, S128–S132. (7) Sutter, E.; Albrecht, P.; Sutter, P. Appl. Phys. Lett. 2009, 95 (13), 133109. (8) Li, X. S.; Cai, W. W.; An, J. H.; Kim, S.; Nah, J.; Yang, D. X.; Piner, R.; Velamakanni, A.; Jung, I.; Tutuc, E.; Banerjee, S. K.; Colombo, L.; Ruoff, R. S. Science 2009, 324 (5932), 1312–1314. (9) Bae, S.; Kim, H.; Lee, Y.; Xu, X. F.; Park, J. S.; Zheng, Y.; Balakrishnan, J.; Lei, T.; Kim, H. R.; Song, Y. I.; Kim, Y. J.; Kim, K. S.; Ozyilmaz, B.; Ahn, J. H.; Hong, B. H.; Iijima, S. Nat. Nanotech. 2010, 5 (8), 574–578. (10) Hwang, J. Y.; Kuo, C. C.; Chen, L. C.; Chen, K. H. Nanotechnology 2010, 21 (46), 465705. 2633
dx.doi.org/10.1021/nl200604g |Nano Lett. 2011, 11, 2628–2633
