4 9 ( 2 0 1 1 ) 2 1 3 4 –2 1 4 0
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/carbon
Carbon nanowall growth on carbon paper by hot filament chemical vapour deposition and its microstructure N. Lisi a,*, R. Giorgi a, M. Re b, T. Dikonimos a, L. Giorgi a, E. Salernitano a, S. Gagliardi a, F. Tatti c a b c
ENEA, Casaccia Research Centre, Via Anguillarese 301, 00123 S. Maria di Galeria, Roma, Italy ENEA, Brindisi Research Centre, Strada Statale 7 Appia, km 706, 72100 Brindisi (BR), Italy FEI Company, Achtseweg-noord 5, 5651 GG Eindhoven, The Netherlands
A R T I C L E I N F O
A B S T R A C T
Carbon nanowall films were synthesized by plasma enhanced hot filament chemical
Received 6 July 2010
vapour deposition on carbon paper, a three dimensionally structured material. The
Accepted 24 January 2011
micro-structured nature of the carbon paper, which is composed of an irregular and open
Available online 31 January 2011
mesh of carbon fibres, allowed one to determine the microstructure of the carbon nanowalls both at the tip and at the fibre–nanowall base interface. The number of graphenes which pile up to form the structure of a single nanowall ranges from 1 to 2 at the tip up to several 10s at the base, making this material suitable to study and eventually exploit the electronic properties of graphenes on a macroscopic scale. Such material is promising for electrochemical applications. Ó 2011 Elsevier Ltd. All rights reserved.
Since their first observations [1,2], carbon nanowalls (CNWs) have attracted a considerable interest since they allow one to deposit carbon films with mainly 2D coordination. They have been grown using several chemical vapour deposition (CVD) techniques, hot filament CVD , plasma enhanced hot filament CVD  and many other plasma enhanced CVD techniques [1,5–8]. CNW growth does not require a catalyst, although there is a report  that it might help. The reaction path and the precursor specie that is directly responsible for the CNW growth have not been determined. The presence of hydrogen in the gas mixture has been widely reported but it is not essential, as growth has been reported in hydrogen free environments . Some hydrogen resulting from the decomposition of the hydrocarbon gas precursor is nevertheless always present. When examined with secondary emission microscopy (SEM) they consist of stacks of thin elongated planar
structures eventually building thick ‘‘sponge like’’ superstructures, arranged either perpendicularly to the deposition substrate or at an angle. In the first case they are referred to as carbon nanowalls in the latter as ‘‘petal like’’ or other. Transmission electron microscopy (TEM) observation demonstrated  that such thin walls consist mainly of stacked ‘‘graphenes’’ oriented along the ‘‘wall’’ direction, while a Raman spectroscopy study  showed that the crystal coordination within such stacks is maintained over a spatial scale much smaller than the ‘‘length of the wall’’ as observed with electronic microscopy techniques. While one hopes that nanowall structures could in the future offer one practical way for stabilising and using graphene, the large number of carbon planes makes them a rather classical carbon based material. Still a material based on CNW carpets can be extremely interesting from a practical point of view, due to the chemical stability and the electrical conductivity of carbon together with its nanostructure and overall morphology.
* Corresponding author: Fax: +39 0630484729. E-mail address: [email protected]
(N. Lisi). 0008-6223/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.carbon.2011.01.056
4 9 ( 20 1 1 ) 2 1 3 4–21 4 0
CNWs can be grown over a large class of substrate geometries and materials, provided that they are thermally and chemically compatible with the growth environment. CNWs have also been grown over carbon nanotube templates demonstrating a material with a high surface area . CNWs can be used as catalyst support material, since they appear to enhance the activity of catalysts and electro-catalysts . In this work, we report the growth of carbon nanowalls on carbon paper (Toray TGP-H-90 Carbon Fibre Paper, 280 lm thick), a commercial material based on a 3D matrix of PAN carbon fibres commonly used as the substrate for diffusive layer for low temperature fuel cell applications. Such material has a large open porosity  and it is made of only a graphitic carbon phase, making it a promising candidate for CNW growth being both thermally and chemically compatible with the growth environment.
The hot filament CVD reactor  consists of a vacuum vessel, a vacuum system, a gas feeding apparatus, a substrate heater including a bias system and the filament assembly, reported in Fig. 1. The filaments consist of two 100 mm long, 1.5 mm diameter straight carburised tantalum filaments. The growth environment consists of a 30 kPa (30 mbar) CH4/He atmosphere, where the methane content must be lower than 4% in order to prevent filament ‘‘poisoning’’ and their subsequent mechanical failure . The samples presented in this work were grown at 2% CH4. The gas mixture was fed at 100 sccm (standard cm3 per minute) by means of digitally controlled flow metres. After filling the chamber at the desired pressure, the substrate was heated up to the growth temperature (700 °C) and the filament temperature was raised up to 2200 °C, as measured with two colour pyrometer (Land system 4, type R1 1000/2600C). Finally a DC discharge was initiated between the filaments (ground) and the substrate
Fig. 1 – Scheme of the filament grid assembly of the plasma enhanced HFCVD.
Table 1 – Summary of the experimental parameters. Pressure (mBar) CH4/He (sccm) Substrate temperature (°C) Filament temperature (°C) Grid voltage (V) Grid current (A) Initial substrate voltage (Vs) Substrate current (A)
30 2/100 700 2200 30 1 280 0.05
(negative) with the help of an additional grid electrode (positive). The system consists of a plasma triode resembling closely those utilised for diamond bias enhanced nucleation . The grid electrode collects and multiplies the thermionic electrons emitted by the filaments thus allowing to initiate and regulate also the discharge current between the substrate and the filaments. The experimental parameters utilised for this work are summarised in Table 1. The deposition process was performed at constant substrate current: due to the enhanced secondary electron emission of the growing nanowalls the bias substrate voltage decreased during the process by as much as the 30%. CNWs still grow, but at a lower rate, when no voltage is applied to the substrate, provided that the grid is powered and the substrate is grounded. The present system power and geometry allows one to deposit with good uniformity over areas up to 50 mm by 100 mm. The scaling up of the substrate dimensions is feasible in principle by an adequate system engineering. The carbon paper substrates were utilised for growth as received. SEM micrographs were obtained using a field emission gun SEM LEO 1530, while TEM analysis were performed using a TECNAI G2 F30, operating at 300 kV and with a point resolution of 0.205 nm. The samples were prepared for TEM observations in two ways in order to characterise completely the walls and to observe both the ‘‘tips’’ and the ‘‘bases’’ without artefacts. The sample preparation for the TEM observation of the CNW tips on the as grown fibres was possible by simply scratching the carbon paper surface, collecting some of the fibres, dispersing them in ethanol and finally collecting them on a C coated Cu TEM grid. Relying on the fibre curvature (carbon paper fibres have an average diameter of 6 lm) the CNW tips were directly visible by observing the edge profile of each fibre. In order to observe the carbon nanowall–carbon fibre interface, a this sample section was prepared with the focused ion beam (FIB) method. By using a Dual Beam Helios 600Nanolab working as a Scanning Electron Microscope the interesting region of the sample was examined, while the Focused Ion Beam column (Ga ions) was used to cut and to mill a very thin section of the selected fibre with CNW. In order to preserve the structure during the cutting and the milling, a Pt protection layer had to be deposited: the first 100 nm were deposited with electron beam assisted deposition (where the low energy Pt ions preserve the delicate fibre–CNW interface), then a second layer of about 1.5 lm was deposited on top of the first layer by means of ion beam assisted deposition. The sections of interest, parallel to the fibre axis, were first cut and thinned
4 9 ( 2 0 1 1 ) 2 1 3 4 –2 1 4 0
Fig. 2 – The fibre with the Pt protective layer (a) and the final thinned section (b). The section is parallel to the fibre axis.
(Fig. 2a) using the FIB (Ga ions) at 30 kV and finally polished in two steps, lowering the acceleration voltage to 5 kV for the first one, and to 2 kV for the second. In Fig. 2b, there is a SEM micrograph of the final thinned section.
Results and discussion
Two of the samples were grown with the substrate bias for a different time, 40 0 and 180 0 , respectively, while the third was grown for 120 0 without substrate bias. The film thickness at the surface facing the filaments is about 750 nm for the first, 3 lm for the second and 900 nm for the third.
The CNW film was deposited all way around the carbon fibres which constitute the carbon paper; in Fig. 3 a typical CNW coated fibre of the carbon paper is shown at different magnifications. The CNW film thickness could be estimated consistently using SEM both by measuring the average fibre diameter before and after the growth and by direct observation on the film cross section on broken fibres, as seen in the insert of Fig. 3. By increasing the deposition time a film thickness up to 3 lm could be achieved in a 3 h process, the film thickness was observed to increase linearly with time. Interestingly the CNW growth proceeds all the way around each carbon fibre into the carbon paper and even reaches the back of the carbon paper substrate, which is 280 lm thick, thus realising an ‘‘infiltration’’ process of the CNW film into 3D structures. In Fig. 4, the CNW film thickness depth profile along the sample section is reported for three of the grown samples.
Published TEM observations of the growing edges reveal that the CNWs are wedge shaped near the tip and can be very thin, down to a single graphene layer [16,17], and high magnification SEM is not an adequate technique to resolve the outer thin edges of the CNW. According to recent theoretical and modelling work  on CNW growth, the bases of the CNWs are expected to be much wider than the edges, both in ‘‘plasma’’ and ‘‘neutral flux’’ growth environments. Such hypothesis is confirmed by several independent experimental facts reported in the literature. The first is the direct TEM evidence that the tip of the nanowalls is wedged , the second being their specific surface , compatible with an average number of graphenes in the 10s range. Also the Raman studies evidence that the crystallite size is much smaller than the average wall height and length as measured using SEM  suggests that the walls are not simply vertical stacks of few parallel graphenes. TEM observations were performed on the same sample (CPCNW1, 40 0 growth) examined in the previous section, in
Fig. 3 – The surface of a typical fibre of the carbon paper after the deposition of CNW (40 0 growth). In the insert the section of a broken fibre (3 h growth).
4 9 ( 20 1 1 ) 2 1 3 4–21 4 0
Fig. 4 – Measured CNW film thickness variation with depth inside the carbon paper. The three samples were grown for a different times. order to assess its microstructure. We first focused on the CNW tip edges, observed on fibres scratched from the outer surface of the carbon paper sample, then we studied the nanowall base and the fibre-nanowall interface on the thinned longitudinal fibre’s sections prepared with FIB technique, as described in the Section 2. In Fig. 5, two bright field images of the single fibre are reported, the lower magnification being shown in the insert, a dense carpet of elongated nanostructures mainly oriented perpendicular to the fibre surface is visible. Such structures correspond to those observed from the top with SEM and shown in Fig. 3. High resolution TEM (HRTEM) images showing the tips of some nanowalls, whose graphene planes were well aligned with the TEM electron beam, were taken.
Fig. 5 – TEM bright field images of a fibre scratched from CPCNW1.
The tip edges are consistently very thin, down to a very few graphene planes, which are oriented along the wall direction. This is well evidenced in Fig. 6 where two of the several HRTEM micrographs are reported. The estimated interplanar distance near the tip is about 0.35 nm, close to the value of (0 0 2) ‘‘turbostratic’’ graphite planes .The TEM analysis of the longitudinal fibre section allowed to study both the CNW bases and their interface with the carbon paper fibres. In Fig. 7, the bright field image of a CNW base close to the fibre surface is reported on the left, while on the right it is shown the HRTEM image of the area indicated with the white arrow. The lower magnification image (on the left) can be compared with the TEM bright field image shown in Fig. 5. It appears that the CNW base can be as wide as 100 nm, and that the interplane distance at the base is close to the tip value of 0.35 nm. The Pt clusters that were formed during the electron beam deposition of the protection layer are visible on the sides of the nanowall. It is due to their presence that the CNW tips cannot be observed in the fibre section slices as accurately as in the whole scratched fibres, on the contrary the fibre-CNW interface (along the dashed black line) could be directly observed. Fig. 8 shows a small region of the fibre–nanowall interface. In the left side there is the fibre (the white arrow on the left indicates the fibre axis direction). In spite of the detrimental effect of curved fibre surface, in the white rectangles there are two areas where it is more evident how the graphitic planes of the nanowalls start to bend vertically (white arrows) away from the fibre, until they eventually become perpendicular to the fibre near the tip. It is not clear how this morphology relates to the growth dynamic; it appears that the CNW initiate their growth parallel to the surface, then the crests emerge and start to form and grow, in agreement with previous experimental and theoretical work [17,18]. An additional observation is that the CNW film still grows when the substrate is grounded (Vs = 0), provided that the grid
4 9 ( 2 0 1 1 ) 2 1 3 4 –2 1 4 0
Fig. 6 – HRTEM images of two CNW tips from CPCNW1.
Fig. 7 – Bright field (on the left) and HRTEM (on the right, at the white arrow of the left image) images of sample CPCNW1 near the nanowall base. The dashed black line indicates the interface. Pt clusters are visible at the sides of the wall.
is powered (i.e. Vg and Ig are above or equal to the values reported in Table 1). Under this unbiased growth regime, where no external electric field is applied on the substrate surface, the CNW film growth rate resulted diminished by about 50%, whilst a similar film thickness depth profile was observed (Fig. 4). Even in this case, despite the absence of applied electric field to the substrate, the plasma discharge initiated between the filaments (cathode) and the grid (anode) will presumably extend to reach the substrate (normal glow discharge). Under the biased growth regime (Table 1), we observed that for a fixed substrate current the bias substrate voltage (Vs) dropped during the growth by about 30%. Such drop can only be attributed to a substrate change during growth with an increased emission of secondary electron from the fibre surface, i.e. to the growth of the carbon nanowalls. The increase can have two causes: the first is the change of structure of the fibre surface which increases the local electric field near the nanowall tips and thus the secondary electron emission from the fibres inside the plasma sheath, the second is the change of orientation of the graphenes from parallel to perpendicular to the fibres, which also can increase the secondary electron emission yield . The coupling of the two effects also provides some insight on the reasons for the bending of the graphenes, since the
Fig. 8 – HRTEM images of the CNW–fibre interface of 40 0 grown sample. The white arrow on the left indicates the fibre, the two white boxes show the regions of the CNW bases where the bending occurs (white arrows in the boxes).
4 9 ( 20 1 1 ) 2 1 3 4–21 4 0
plasma current density will flow stronger for thin vertical graphitic structures composed of vertical graphenes, thus selecting and enhancing their growth. In both grounded and biased substrate cases, although one could expect that the electric field inside the carbon paper should be screened by its electrically conductive envelope, both the carbon paper thickness and the average inter-fibre opening size (100 lm in average) are too close to the Debye length  expected in a plasma with parameters close to ours to draw definite conclusions. The Debye length calculated according to the parameters reported in the literature  for a 20 torr, 10 mA/cm2 He glow discharge can be estimated to be around 70 lm, thus it cannot be excluded that the plasma and its potential could reach the inside of the carbon paper. We do not wish to enter into a detailed discussion on the growth mechanism but two alternative mechanism could be at play during the growth, involving either ionic or neutral precursors. The growth could result from the acceleration of a ionic carbon precursor, such as C1+ or C2+ ionic radicals , towards the cathodic substrate potential. The ionic precursor would then be driven towards the growing nanowalls by the presence of a potential on the surface and inside the carbon paper or extracted from the plasma at the sheath surface and then diffused inside the carbon paper. Alternatively the plasma generated by the filament-grid discharge and on the substrate surface could create the neutral radical which might be the precursor of the CNW growth, such neutrals would then diffuse inside the carbon paper. According to the literature, the thermal dissociation of methane in the proximity of heat sources and/or inside plasmas can efficiently generate C2 compounds (such as C2 dimer and acetylene) and higher C (n > 2) radicals [26–28]. Further work is currently in progress in order to distinguish between the two possible channels. We believe that the use of a HFCVD plasma triode and carbon paper based growth process, together with the characterisations reported in this work are well suitable for such investigation.
A method for growing CNW films of controlled thickness on the surface and into the bulk of carbon paper is presented. The method offers a viable solution for coating 3D microscopically-structured substrates with a mainly 2D carbon film, known in the literature as carbon nanowalls. The geometrical properties of the carbon fibres from which the carbon paper is made allowed the complete microstructural characterisation of the CNWs, from the few graphene layers tip to their bases and their interface with the carbon fibres of the substrate. The CNW tips, as thin as a few graphene layers were observed using TEM from fibres scratched from the carbon paper. The CNW bases and their interface between the CNW and the growth substrates were observed using TEM techniques in thinned fibre sections, suitably prepared with focus ion beam (FIB) milling in order to select and preserve the interesting area of fibre-nanowall interface. Such observation constitutes the first TEM evidence reported in the literature, to our knowledge, of the bending of the initially horizontal graphene planes into the vertical nanowalls during growth.
R E F E R E N C E S
 Ando Y, Zhao X, Ohkohchi M. Production of petal like graphite sheets by hydrogen arc discharge. Carbon 1997;35:153–8.  Obraztsov AN, Pavlovsky IY, Volkov AP, Petrov AS, Petrov VI, Rakova EV, et al. Electron field emission and structural properties of carbon chemically vapor-deposited films. Diamond Relat Mater 1999;8:814–9.  Shang NG, Au FCK, Meng XM, Lee CS, Bello I, Lee ST. Uniform carbon nanoflake films and their field emissions. Chem Phys Lett 2002;358:187–91.  Dikonimos TM, Giorgi L, Giorgi R, Lisi N, Salernitano E, Rossi R. DC plasma enhanced growth of oriented carbon nanowall films by HFCVD. Diamond Relat Mater 2007;16:1240–3.  Obraztsov AN, Zolotukhin AA, Ustinov AO, Volkov AP, Svirko Yu, Jefimovs K. DC discharge plasma studies for nanostructured carbon CVD. Diamond Relat Mater 2003;12:917–20.  Wang J, Zhu M, Outlaw RA, Zhao X, Manos DM, Holloway BC. Synthesis of carbon nanosheets by inductively coupled radiofrequency plasma enhanced chemical vapor deposition. Carbon 2004;42:2867–72.  Hiramatsu M, Shiji K, Amano H, Hori M. Fabrication of vertically aligned carbon nanowalls using capacitively coupled plasma-enhanced chemical vapor deposition assisted by hydrogen radical injection. Appl Phys Lett 2004;84:4708.  Sato G, Morio T, Kato T, Hatakeyama R. Fast growth of carbon nanowalls from pure methane using helicon plasmaenhanced chemical vapor deposition. Jpn J Appl Phys 2006;45:5210–2.  Itoh T, Shimabukuro S, Kawamura S, Nonomura S. Preparation and electron field emission of carbon nanowall by Cat-CVD. Thin Solid Films 2006;501:314–7.  Kurita S, Yoshimura A, Kawamoto H, Uchida T, Kojima K, Tachibana M, et al. Raman spectra of carbon nanowalls grown by plasma-enhanced chemical vapor deposition. J Appl Phys 2005;97:104320-1–5.  Chen CC, Chen CF, Lee IH, Lin CL. Fabrication of high surface area graphitic nanoflakes on carbon nanotubes templates. Diamond Relat Mater 2005;14:1897–900.  Giorgi L, Dikonimos TM, Giorgi R, Lisi N, Salernitano E. Electrochemical properties of carbon nanowalls synthesized by HF-CVD. Sensors Actuators B 2007;126:144–52.  Heinzel A, Hebling C, Muller M, Zedda M, Muller C. Fuel cells for low power applications. J Power Sources 2002;105:250–5.  Dandy DS, Coltrin ME. Effects of temperature and filament poisoning on diamond growth in hot-filament reactors. J Appl Phys 1994;76(5):3102–13.  Zhou XT, Lai HL, Peng HY, Sun C, Zhang WJ, Wang N, et al. Heteroepitaxial nucleation of diamond on Si(1 0 0) via double bias-assisted hot filament chemical vapor deposition. Diamond Relat Mater 2000;9:134–9.  Zhu M, Wang J, Outlaw RA, Hou K, Manos DM, Holloway BC. Synthesis of carbon nanosheets and carbon nanotubes by radio frequency plasma enhanced chemical vapor deposition. Diamond Relat Mater 2007;16:196–201.  Malesevic A, Vitchev R, Schouteden K, Volodin A, Zhang L, Van Tendeloo G, et al. Nanotechnology 2008;19:305604-1–6.  Levchenko I, Ostrikov K, Rider AE, Tam E, Vladimirov SV, Xu S. Growth kinetics of carbon nanowall-like structures in lowtemperature plasmas. Phys Plasmas 2007;14:063502.  Tanaka K, Yoshimura M, Okamoto A, Kazayuki U. Growth of carbon nanowalls on a SiO2 substrate by microwave plasmaenhanced chemical vapor deposition. Jpn J Appl Phys 2005; 44 (4A): 2074–2076.
4 9 ( 2 0 1 1 ) 2 1 3 4 –2 1 4 0
 Chuang ATH, Boskovic B, Robertson J. Freestanding carbon nanowalls by microwave plasma-enhanced chemical vapour deposition. Diamond Relat Mater 2006;15:1103–6.  Inagaki M. New carbons – control of structure and functions. Elsevier; 2000. p. 15–17.  Whetten NR. Secondary electron emission of pyrolytic graphite cleaved in a high vacuum. J Appl Phys 1963;4(1):771–3.  Glow Discharge Processes. B. Chapman. New York: John Wiley & Sons; 1980.  Dothan F, Kagan YuM. Level population densities and line intensities in helium discharges at intermediate pressures. J Phys D Appl Phys 1981;15:183–97.
 Vizireanu S, Stoica SD, Luculescu C, Nistor LC, Mitu B, Dinescu G. Plasma techniques for nanostructured carbon materials synthesis. A case study: carbon nanowall growth by low pressure expanding RF plasma. Plasma Sources Sci Technol 2010;19:034016-1–034016-10.  Obraztsov AN, Zolotukhina AA, Ustinova AO, Volkova AP, Svirko Y. Chemical vapor deposition of carbon films: in-situ plasma diagnostics. Carbon 2003;41:836–9.  Obraztsov AN, Zolotukhin AA, Ustinov AO, Volkov AP. Plasma CVD characterization of nanocarbon film growth. Surf Interface Anal 2004;36:481–4.  Appel J, Kramer L. Acetylene and hydrogen from the pyrolysis of methane. Ind Eng Chem 1967;59(1):39–50.