Growth of carbon with vertically aligned nanoscale flake structure in capacitively coupled rf glow discharge

ARTICLE IN PRESS

Vacuum 82 (2008) 754–759 www.elsevier.com/locate/vacuum

Growth of carbon with vertically aligned nanoscale flake structure in capacitively coupled rf glow discharge Hui Zhanga,1, Naoto Kikuchia,2, Toshihiro Kogureb, Eiji Kusanoa, a

Advanced Materials Science R&D Center, Kanazawa Institute of Technology, 3-1 Yatsukaho, Hakusan, Ishikawa 924-0838, Japan Department of Earth and Planetary Science, Graduate School of Science, University of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan

b

Received 15 August 2007; received in revised form 27 October 2007; accepted 3 November 2007

Abstract Carbon nanoflakes (CNFs) have been deposited on Si (1 0 0) wafer substrates at a substrate temperature of 670 1C from a glassy carbon target by capacitively coupled rf (13.56 MHz) glow discharge using a mixture discharge gas of Ar and CH4 with a total pressure of 14.5 Pa. Microstructures of deposited carbon were investigated by a field emission scanning electron microscope (FESEM) and a highresolution transmission electron microscope (HRTEM). Under the given conditions, vertically aligned CNFs with a flake length of about 1 mm and thickness of about 20 nm were grown. High intensity and symmetry of electron diffraction pattern indicate that the CNFs deposited by capacitively coupled rf glow discharge have three-dimensionally perfect crystallinity with a graphene interlayer spacing of 335 pm. In particular, there is little disorder in stacking of the layer structure. It was further found that the thickness of the flakes was less dependent of deposition time while the length of the flakes increases to about 1 mm with increasing deposition time to 3 h. The growth rate of a graphite sheet parallel to (0 0 1) stacking layers was much higher than that perpendicular to (0 0 1) stacking layer, resulting in anisotropic growth of a flake-like structure. The formation mechanisms of CNFs are discussed from the viewpoint of the difference in residence time of carbon atoms on CNF surfaces parallel to and perpendicular to (0 0 1) direction and anisotropic heat conductivity of graphite. r 2007 Elsevier Ltd. All rights reserved. Keywords: Carbon; Nanostructure; Nanoflake; rf Glow discharge

1. Introduction Nanostructure materials are known to exhibit novel and technologically attractive properties due to quantum size effects, surface effects, small grain size effects, etc. [1]. Among nanomaterials, as a versatile element constructing the various allotropes such as diamond, graphite, fullerenes and nanotubes, carbon nanoscaled materials have been paid much more attention recently. The high Young’s modulus of about 1.8 MPa, diameter-dependent electric conductivity, and a very low onset of electric field for electron emission have made the carbon nanotubes (CNTs) Corresponding author. Tel.: +81 76 274 9257; fax: +81 76 274 9251.

E-mail address: [email protected] (E. Kusano). On leave from Xi’an Jiaotong University, PR China. 2 Current address: Superconducting Materials Group, Nanoelectronics Res. Inst., National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Ibaraki 3058568, Japan. 1

0042-207X/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.vacuum.2007.11.001

to be one of the most studied nanomaterials in the past decade [2]. Carbon can also constitute other interesting types of the nanoscale allotrope: fullerene and carbon nanoflakes (CNFs). Between these two nanoscale allotropes, CNF has captured more attention because of its interesting chemical, electrical, and mechanical properties. One of the interesting chemical properties is a high ability as catalysis matrix. In graphite, the weak van-der-Waals bonding between two graphene layers allows atoms or molecules to be intercalated into them. In CNFs, a high density of open edges of (1 0 1) and (1 1 2) surfaces and a high specific surface area further increase chemical reactivity [3]. Another interesting application of CNF is for field emission devices. Shang et al. [4] reported that aligned nanoflake films showed an electron emission turn-on field of about 17 MV/m and might have a potential application in vacuum electronic devices. CNFs can be introduced into other application fields such as electrode [5,6], hydrogen

ARTICLE IN PRESS H. Zhang et al. / Vacuum 82 (2008) 754–759

storage materials [7], lubricant [8]. In addition, CNFs have the remarkable advantage over CNTs in fabrication because in the growth of CNFs metal catalysts, high substrate temperature and catalyst passivation needed for CNTs growth [9] are not required. Practical applications can be more expected on the CNFs in the field of nanostructured materials. To date, several methods have been proposed to synthesize the CNFs or nanowalls: arc discharge evaporation [10], laser ablation [11], microwave plasma enhanced chemical vapor deposition [12], and the hot-filament chemical vapor deposition (HFCVD) [4,13]. Regarding the carbon electrode arc-discharging and laser ablation, CNFs usually coexist with nanotubes and other soot, in addition to metal catalyst dopants. The yield CNFs was low when the above methods were applied. Tungsten filaments have to be heated to 2100 1C in HFCVD, resulting in around 3% tungsten contamination from filaments contained in the CNF films [4]. In the present paper, we have deposited CNFs by capacitively coupled rf magnetron discharge using a glassy carbon target as an electrode. By this method, a large area growth of aligned and relatively pure CNFs grown perpendicular to the substrate could be achieved, meanwhile, the deposition method using capacitively coupled rf magnetron glow discharge can overcome the shortcoming of the previously referred methods. Microstructures of the CNFs will be investigated by field emission scanning electron microscope (FESEM) and high-resolution transmission electron microscope (HRTEM). Growth mechanisms of the nanoflakes will be also discussed. 2. Experimental procedure 2.1. Deposition process The apparatus used in this study was a load lock-type sputtering machine (SPC-350UHV: ANELVA corporation), schematically shown in Fig. 1. A glassy carbon disk with 75 mm diameter was set to one of the five cathodes equipped in the high vacuum deposition chamber. The distance between the target surface and the substrate is about 60 mm. The deposition chamber was evacuated to a base pressure lower than 1.0  105 Pa prior to deposition runs by a turbo molecular pump. As substrates, silicon (1 0 0) wafers ultrasonically cleaned were used. rf power applied to the target was 100 W. The mixture of Ar (99.998% in purity) and CH4 (99.9% in purity) was used as a discharge gas. Their flow rates were controlled to be at 14 and 28 sccm, respectively. The total pressure was controlled to be approximately 14.5 Pa. The substrate was heated to 670 1C by lamp heaters set behind the substrate holder. 2.2. Film analysis A FESEM (S-4500: Hitachi) was used to observe the morphology of deposited films and to measure thickness,

755

Shutter

No.4

No.5

Carbon Target Heater

Load-Lock No.3 C

No.1

Pressure Gauge

Gas inlet

Substrate

Fig. 1. Schematic of the sputtering apparatus used in the experiment.

length, and width of grown nanoflakes. A transmission electron microscope (TEM, HF-2000: Hitachi) with a fieldemission electron gun operated at 200 kV was used to observe microstructure of nanoflakes. Cross-sectional TEM specimens were prepared with argon ion-thinning. 3. Results Fig. 2 shows FESEM images of CNFs grown on silicon substrates at 670 1C for various discharge durations. Vertically well-aligned CNFs are observed for all the samples. A typical size of the nanoflakes grown for 3 h is a length of 1 mm, a width of 600 nm, and a thickness of 30 nm. In Fig. 3, SEM top view images of nanoflakes are shown. It is clear that the flakes are not flat but in ‘‘S’’ or ‘‘C’’ shapes. It was found by SEM observation that the condition needed to grow the flake-like structure is a substrate temperature of 4600 1C and a CH4 partial pressure of 46.8 Pa. In Fig. 4(a), an HRTEM image of a sheet of CNFs is shown. A layered structure in a sheet of flakes is clear. The number of layers in one piece is about 30. In Figs. 4(b) and (c), TEM diffraction patterns are shown for the flake shown in Fig. 4(a), with the electron beam normal to and parallel to the layers. The diffraction pattern shown in Fig. 4(b) is six-fold and of hexagonal closed packed structure (HCP). Two bright spots in Fig. 4(c) assigned to those from (0 0 2) planes and concentric. From these two spots, layer spacing was calculated to be 335 pm, which well agrees to the value obtained for graphite crystal without disorder of stacking [14]. Furthermore, the weak spots of (1 0 1) groups also appear. It is clearly shown from the two diffraction patterns in Figs. 4(b) and (c) that a

ARTICLE IN PRESS 756

H. Zhang et al. / Vacuum 82 (2008) 754–759

Fig. 2. SEM images of carbon nanoflakes grown on silicon substrates at 670 1C for various growth durations of (a) 15 min, (b) 30 min, (c) 2 h and (d) 3 h.

sheet consists of three-dimensionally nearly perfect single crystal graphite. In Fig. 5, length and thickness of flakes are shown as a function of growth time. The length and thickness were determined by averaging values of length or thickness obtained for 10 pieces of flakes randomly selected in each SEM images. Both length and thickness of CNFs increase with growth time. Growth rate of a flake parallel to (0 0 1) is about 0.10 nm/s and that normal to (0 0 1) is 1.8 pm/s. The rate perpendicular to (0 0 1) is about 1/56 compared to that parallel to (0 0 1). The difference in the two growth rates results in the growth of a flake shape of graphite. 4. Discussion 4.1. Growth mechanisms In this section, mechanisms of flake-like structure growth are discussed based on the difference in residence time of particles on surfaces of (0 0 1) and on those of (1 0 0) or (0 1 0) and anisotropy of thermal conductivity of /0 0 1S direction and those of /1 0 0S or /0 1 0S direction. It should be first noted that graphene layer growth results from a low-sticking coefficient of carbon on to surface of a high-temperature substrate and from a anisotropy in sticking coefficient of C atoms on (1 0 0) and (0 0 1) surfaces. The intensity of CH4 flux arriving to

the substrate surface is estimated to be 3.7  1019 molecules/(cm2 s) from the partial pressure of CH4. The growth rate perpendicular to (0 0 1) given in the unit of atoms/ (cm2 s) is calculated to be 1.7  1013 atoms/(cm2 s). The sticking coefficient therefore is estimated to be 4.6  107. On the other hand, the growth rate parallel to (1 1 1) is 9.4  1014 atoms/(cm2 s): giving a sticking coefficient of 2.5  104. The sticking coefficient obtained for atoms on (1 0 0) well agrees to a reported sticking coefficient of about 106 for graphene layer growth on highly oriented pyrolytic graphite substrate [15]. These sticking coefficient values ranging from 104 to 107 are small compared to those observed or calculated in metal or metal compound deposition by sputtering [16,17]. Furthermore, it is clear that the difference in the sticking coefficient on (1 0 0) and (0 0 1) surfaces results in the anisotropic flake-like structure growth. One possible mechanism of the anisotropic sticking coefficient is the difference in the residence time of C atoms on the surfaces of (1 0 0) and (0 0 1), resulting from the difference in bond strength within and perpendicular to (0 0 1) surface; i.e. the difference in the bond strength in s and p bondings in graphite. Graphite has two different types of bonds within it: one is sp2 covalent bond: the other is van der Waals bond perpendicular to the stacking structure of graphite. The difference in the strength in the two bonds results in the difference in residential time of particles on the surfaces. The residence time t of an atom

ARTICLE IN PRESS H. Zhang et al. / Vacuum 82 (2008) 754–759

Fig. 3. SEM top view images of nanoflakes grown for 3 h with a magnification of (a)  20 k and (b)  35 k.

arriving on a surface can be given by the following equation:   1 Ea t ¼ exp , (1) n kT where n, k, Ea and T are atom vibration frequency, Boltzmann’s constant, activation energy for atoms to move among positions, and the substrate temperature, respectively. The bond energy of sp2 bond in graphite is given to be 524 kJ/mol (about 5.4 eV/atom) and interlayer van der Waals energy is 7 kJ/mol (about 0.08 eV/atom) [18]. Residence time calculated using the Eq. (1) by assuming n ¼ 1014 s1 is shown in Table 1. The residence time of atoms on (0 0 1) surface ranges from 1013 to 1014 s. On the other hand, those on (1 0 0) surface range from 1077 to 1016 s. Therefore, while an atom arriving on (1 0 0) surface can bond to already existing atoms, one arriving on (0 0 1) surface cannot find out a stable site to bond during the residence time on the surface and revalorize. This is a possible mechanism of anisotropic growth of the graphene sheets. However, the dependence of the growth of the flakelike structure on temperature cannot be explained by this mechanism. If this is the only mechanism of anisotropic graphene layer growth, flake structured carbon can be

757

deposited even at a low temperature. Further the fact that the ratio of the residence time between the two surfaces is in the order of 90 at room temperature and in the order of 28 at 943 K (670 1C), i.e., the fact that, even at room temperature, the ratio is larger, means that anisotropic growth is accelerated at a low temperature. This completely disagrees to the results obtained in this study. The other mechanism proposed here is one based on the anisotropy of thermal conductivity in graphite structure. The anisotropy of the thermal conductivity affects the process of the dissipation of thermal energy during the condensation of C atoms arriving to the surface. The thermal conductivity parallel to /0 0 1S direction at 1000 K is about 530 W/(m K) and that perpendicular to the /0 0 1S direction is 1.6 W/(m K) [19]. Under the deposition conditions that the flake (two-dimensional) structure is formed, the vapor of carbon or CHX (X ¼ 1–3, most possibly X ¼ 3) radicals is in supersaturation. In the case of the supersaturation, the heat generated by condensation of arriving atoms must be emitted effectively. If the bulk like graphite grows, the heat transferred to the surface by C flux from vapor phase cannot dissipate, increasing the probability of reevaporation and suppressing the growth of graphene layers on the (0 0 1) surface. From the above discussions, it is emphasized that for the CNF formation reported in this study the supersaturation of C atoms or CHX radicals and the energy dispersion efficiency on the surface of a growing nanoflake play a key role. However, the difference in the mean stay time of C atoms or CHX radicals on (1 0 0) and (0 0 1) could not be ignored, judging from the fact that the difference calculated from the two binding energies is persuasive enough to explain the anisotropic flake growth. The combination of the two mechanisms is thought to result in the formation of CNF. The role of carbon sputtered from the target and CH4 in the discharge atmosphere should be discussed. As described in the section of results, the condition needed to form CNFs is a high substrate temperature of 4600 1C and a high CH4 partial pressure of 44.3 Pa. At the partial pressure in the given condition, the CH4 flux arriving to the substrate surface is estimated to be 3.7  1019 cm2 s1. On the other hand, the flux of the sputtered C atoms arriving to the substrate is estimated to be in the order of 1016–1017 cm2 s1, judging from a deposition rate for amorphous carbon film deposited at room temperature. This value is, of course, 2–3 orders smaller than the intensity of the flux of CH4 molecule; most of C atoms growing on the substrate were sourced from CH4 gas. In this point, the process employed in this paper is sputteringCVD combined growth. The difference in atomic C and CH2 or CH3 molecules decomposed from CH4 gas is energy to reevaporate and to form radicals; the bond energies are C–C: 618 kJ/mol, C–CH2(singlet): 428 kJ/mol, C–CH3: 147 kJ/mol [20], meaning that a CH2 or CH3 weakly bonded to C on growing CNF could be easily reevaporated. Although conditions to grow CNF should be

ARTICLE IN PRESS H. Zhang et al. / Vacuum 82 (2008) 754–759

758

Fig. 4. HRTEM (a) bright image of a sheet of carbon nanoflakes and its electron diffraction patterns (b) B//(0 0 2) and (c) B ¼ [0 0 2] of carbon nanoflakes.

1200

50 Length of flake Thickness of flake

40 35

800

30 600

25 20

400

15 200

10

Table 1 Residence time of carbon atoms arriving onto different planes

Thickness of flake (nm)

Length of flake (nm)

1000

45

Substrate temperature (K)

298 600 943

Residence time, t(s) Perpendicular to (0 0 2) plane (bond energy: 5.4 eV)

On (0 0 2) plane (bond energy: 0.08 eV)

1077 1031 1014

1013 1014 1014

5

0 0

20

40

0 60 80 100 120 140 160 180 200 Growth time (min)

Fig. 5. Length and thickness of carbon nanoflakes as a function of growth duration.

further investigated, in this experiment no CNFs were grown in pure Ar atmosphere by sputtering using a glassy carbon target, indicating that the existence of volatile CH2 or CH3 radicals in the condition of supersaturation are important to grow nanoflakes. 4.2. Structure of carbon nanoflakes The feature of the CNFs is the three-dimensionally perfect structure. Graphite formed by thermal decomposi-

tion generally shows mis-stacking of layers, resulting in broad or ring patterns in electron diffraction patterns [21]. The reason for the three dimensionally perfect stacking is a slow deposition rate and high growth temperature. As discussed in Section 4.1, an atom arrived on (0 0 1) surface has a very short residence time. This means that only atoms reached at the stable site can reside and form a bond to an underneath carbon atom. Other atoms with misfit cannot form any bonds, resulting in reevaporation. S or C shapes of flakes result from the introduction of pentagon ring or heptagon ring in hexagonal normal ring plane, which is well known in a CNT growth model [22]. Both cause a curvature in a hexagonal network. However, the amount of the pentagon or heptagon rings in the

ARTICLE IN PRESS H. Zhang et al. / Vacuum 82 (2008) 754–759

hexagonal network is very small, causing a very large curvature of grapheme sheet. 5. Conclusions Vertically aligned CNFs have been grown on silicon wafer substrate at a substrate temperature of 670 1C by capacitively coupled rf glow discharge. It was found by HRTEM that the CNFs deposited had a three-dimensionally nearly perfect graphite structure. The length of CNF after 3 h deposition was about 3 mm, while the thickness reached only 30 nm. The growth rate of the flakes was anisotropic; the rate parallel to (0 0 1) was more than one and half order faster than that perpendicular to (0 0 1). The anisotropy in bond strength and thermal conductivity are discussed as mechanisms for anisotropic flake growth. Further works on electrical and mechanical properties are needed. References

[6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]

[17] [18]

[19] [20]

[1] [2] [3] [4] [5]

Iijima S. Nature 1991;354:56. Qin LC, Zhou D, Krrauss AR. Appl Phys Lett 1998;72:3437. Kang ZC, Wang ZL. J Mole Catal A 1997;118:215. Shang NG, Ao FCK, Meng SM. Chem Phys Lett 2002;358:187. Chen MY, Yeh CM, Syu JS, Hwang J, Kou CS. Nanotechnology 2007;18:185706.

[21]

[22]

759

Frysz C, Shui XP, Chung PDL. Carbon 1997;35:893. Hwang JY, Lee SH, Kim JW. Synthetic Metals 2002;126:81. Shaji S, Radhakrishana V. Inter J Mach Tools Manufac 2002;42:733. Jung M, Eun KY, Lee JK. Diamond Related Mater 2001;10:1235. Ando Y, Zhao X, Ohkohchi M. Carbon 1997;35:893. Iijima S, Wakabayashi T, Achiba Y. J Phys Chem 1996;100:5839. Srivastava SK, Shukla AK, Vankar VD, Kumar V. Thin Solid Films 2005;492:124. Dikominos T, Giorgi L, Giorgi R, Lisi N, Salernitano E, Rossi R. Diamond Related Mater 2007;16:1240. The International Centre for Diffraction Data (JCPDS) file no. 411847. Wellmann R, Boettcher A, Kappes M, Kohl U, Niehus H. Surf Sci 2003;542:81. Rossnagel SM, Cuomo JJ, Westwood WD. Handbook of plasma processing technology. Park Ridge, USA: Noyes Publication; 1990. p. 173. Rossnagel SM. J Vac Sci Technol A 1988;6:3049. Pierson HO. Handbook of carbon, graphite, diamond and fullerenesproperties, processing and applications. Park Ridge, USA: Noyes; 1993. p. 45. Lide DR. CRC handbook of chemistry and physics. 88th ed. Boca Raton USA: CRC Press; 2007. p. 12–202. Lide DR. CRC handbook of chemistry and physics. 88th ed. Boca Raton, USA: CRC Press; 2007. p. 9–69. Pierson HO. Handbook of carbon, graphite, diamond and fullerenesproperties, processing and applications. Park Ridge, USA: Noyes; 1993. p. 157. Dresselhaus MS, Dresselhaus G, Avouris P. Carbon nanotubes, in topics in applied physics. Berlin, Germany: Springer; 2000. p. 58.