- Email: [email protected]
Synthesis and characterisation of carbon nanotubes grown on silica fibres by injection CVD
CARBON
4 8 ( 2 0 1 0 ) 2 7 7 –2 8 6
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/carbon
Synthesis and characterisation of carbon nanotubes grown on silica fibres by injection CVD Hui Qian a b c
a,b
, Alexander Bismarck b, Emile S. Greenhalgh c, Milo S.P. Shaffer
a,*
Department of Chemistry, Imperial College London, London SW7 2AZ, UK Polymer and Composite Engineering (PaCE) Group, Department of Chemical Engineering, Imperial College London, London SW7 2AZ, UK The Composites Centre, Imperial College London, London SW7 2AZ, UK
A R T I C L E I N F O
A B S T R A C T
Article history:
Carbon nanotube grafted primary reinforcing fibres are under development for a new gen-
Received 1 June 2009
eration of hierarchical composites. Pure and nitrogen-doped multi-walled carbon nano-
Accepted 7 September 2009
tubes (MWCNTs) were grown on silica fibres using the injection chemical vapour
Available online 11 September 2009
deposition method. The morphology and size of the nanotubes were controlled by varying the growth time. The surface structure of the silica fibres after the grafting process was studied by electron microscopy following focused ion beam sectioning; the images confirmed a base growth mechanism and a shallow iron-rich layer. Thermogravimetric analysis indicated a shorter induction period and a faster growth rate for pure rather than nitrogen-doped MWCNTs. Raman characterisation of the grafted MWCNTs showed a decreasing intensity ratio of the D to G modes, moving from the tip to base in both cases; a detailed comparison of different characteristic Raman ratios is provided, using both the peak intensity and the area for the D, G, and G 0 signals. Ó 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
Carbon nanotubes (CNTs) continue to receive huge interest due to their extraordinary mechanical, thermal and electronic properties [1,2] and potential applications in a wide range of fields, such as composites [3] and electronic devices [4]. Amongst a variety of methods to produce CNTs, catalytic chemical vapour deposition (CVD) is widely accepted as the most efficient for large scale synthesis of CNTs [5,6], tending to produce relatively pure, but defective and entangled materials. Alternatively, aligned CNT arrays can be grown by CVD, often using an injection method (ICVD) which involves the simultaneous pyrolysis of solutions containing both the catalyst precursor and the hydrocarbon source [7–10], without the time-consuming preparation of catalyst pre-coated substrates. This method has been used to grow well-aligned nitrogen-doped carbon nanotubes (CNx–CNTs) by injecting
nitrogen-containing organic precursors [11–16], such as pyridine, diazine, acetonitrile and so on. The addition of diazine or triazine precursors has been shown to improve the internal crystalline order, straightness and packing of ICVD-grown nanotubes [11,17]. CNx–CNTs are either metals or narrow energy gap semiconductors, thus possibly offering better electrical conductivity than pure CNTs [16,18]. In addition, the nitrogen sites in the graphene network can also improve the nanotube-matrix interactions in composites [19]. Over the past decade, many researchers have explored the use of CNTs in composites, as a replacement for conventional microscale reinforcing fillers, such as carbon or glass fibres [20]. However, the direct use of CNTs remains limited by the quality of the materials available in bulk and the difficulties associated with dispersion, alignment, and the CNT/matrix interface. Meanwhile, interest is growing in the opportunity to combine CNTs with conventional fibres to create hierarchical
* Corresponding author: Fax: +44 (0)20 7594 5801. E-mail address: [email protected] (M.S.P. Shaffer). 0008-6223/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.carbon.2009.09.029
278
CARBON
4 8 ( 2 0 1 0 ) 2 7 7 –2 8 6
composites. Of the available processing strategies, one of the most appealing is to grow CNTs directly onto fibre surfaces [21–24]; this approach has the potential to provide high loadings of well-spaced CNTs, with a possibly optimal, radial orientation around the fibres. It is reported that the grafting of CNTs on fibres improves the interfacial properties between fibres and matrix [21,23]. In addition, the presence of CNTs in matrix-rich regions could alleviate some of the drawbacks of conventional fibre reinforced polymer composites, especially longitudinal compression performance, interlaminar properties, and potentially damage tolerance [22,24]. Ci et al. [25] first reported the synthesis of CNTs on the surface of ceramic fibres by using the ICVD method and deduced that the base growth mechanism applies. Since then, CNTs have been grown on various substrates, including carbon [26], SiC [22,27], alumina [24], quartz and aluminium silicate [28] fibres or woven fabrics. The present study systematically investigated the growth of pure and N-doped MWCNTs on silica fibres using the ICVD method. The morphology, length and diameter of nanotubes grown at different growth times were characterised using scanning electron microscopy (SEM). The consistency of the ICVD-grown MWCNTs along the nanotube array and fibres was also investigated. The interface structure between nanotubes and fibres is of particular relevance to composite performance; the local structure of the interface was explored for the first time by using the focused ion beam (FIB) technique to prepare cross-sections, from which the growth mechanism of nanotubes on fibres can be deduced. Moreover, the effects of the nitrogenincorporation on the growth rate, oxidation resistance and crystallinity of MWCNTs grown on fibres were studied using thermogravimetric analysis (TGA) and Raman characterisation.
2.
Experimental
2.1.
Materials
Commercially available Silfa silica fibres (9 lm in diameter, Albert Hellhake GmbH) were used for this work. A thermal desizing treatment of the fibres was carried out prior to CNT grafting at 600 °C for 1 h. All chemicals were used as received.
2.2.
Growth of carbon nanotubes on fibres
Pure and N-doped MWCNTs were synthesised on silica fibres using the injection chemical vapour deposition method [8,11,13]. The reactions were carried out in a tubular quartz reactor (50 mm in diameter) equipped with an electrical furnace (PTF 12/50/610, Lenton). The silica fibres were held by an alumina frame and placed in the middle of the furnace to expose them to the hydrocarbon source. For pure MWCNTs, a feed solution of 3 wt.% ferrocene (98%, Sigma–Aldrich) in toluene (BDH AnalaR grade, VWR International Ltd.) was injected continuously into the reaction tube at a rate of 5 ml/h using a syringe pump (KDS100, Linton). For CNx– MWCNTs, a feed solution consisting mixture of 3 wt.% ferrocene, 58.5 wt.% toluene and 38.5 wt.% pyrazine (Kosher grade, Sigma–Aldrich) was used. The liquid feed was preheated to
200 °C, immediately volatilised, and swept into the furnace by a flow of argon (Ar) as carrier gas. The reaction temperature was 760 °C. The growth times for pure MWCNTs were 15, 30 and 120 min. Longer growth times, of 60, 120 and 240 min, were used for the synthesis of CNx–MWCNTs, due to their slower growth rate.
2.3.
Electron microscopy characterisation
The morphology, alignment and size of the grafted nanotubes on the silica fibres were characterised using a field emission gun scanning electron microscope (Gemini LEO 1525 FEGSEM, Carl Zeiss NTS GmbH). The surface of modified fibres was studied using a transmission electron microscope (TEM, JEM-2000FXII Electron Microscope, JEOL, Inc.) operating at 200 kV. Energy dispersive X-ray (EDX) analysis was performed using the INCA X-ray microanalysis system (4.08 suite version, Oxford Instruments). FIB sectioning [29] was carried out using FIB 200 TEM focused ion beam instrument (FEI, Inc.). After deliberate removal of the CNT layer, a fibre was glued onto a copper TEM grid, which was partly cut for better handling. The specimen was pre-coated with gold using a sputter coater (K-550X, Emitech Ltd.). During the FIB sectioning process, a gallium liquid metal ion source (Ga+) was used as a primary ion beam with an energy of 30 keV. The area of interest on the fibre surface was protected by coating with a thin layer of platinum prior to etching. The ultimate thickness of the defined area on the specimen was approximately 100 nm (Supplementary information, Fig. S1).
2.4.
Thermogravimetric analysis
TGA measurements were carried out using a Perkin–Elmer Pyris 1 thermogravimetric analyser (Perkin–Elmer, Inc.). Samples were analysed in a platinum pan at a heating rate of 2 °C/ min up to 900 °C, in an atmosphere of air flowing at 20 mL/ min. Sample masses ranged from 1 to 2 mg.
2.5.
Raman spectroscopy
The Raman spectra presented in this report were collected in the backscattered geometry using a LabRam infinity analytical Raman spectrometer (HORIBA Jobin Yvon Ltd.), at room temperature, using a linearly polarised Helium–Neon laser (k = 632.8 nm). The diameter of the laser spot was a few micrometres and the resolution of the spectrometer was about 1 cm1. All the spectra were fitted with a mixed Gaussian–Lorentzian function using GRAMS software.
3.
Results and discussion
3.1.
Pure MWCNT-grafted silica fibres
3.1.1.
Growth of MWCNTs on silica fibres
SEM images (Fig. 1) showed that well-aligned MWCNTs were successfully grown onto the silica fibres at all growth times between 15 and 120 min; however, different morphologies were obtained. Radial growth in the early stages resulted in a ‘brush-like structure’ (Fig. 1b), which later ‘parted’
CARBON
279
4 8 (2 0 1 0) 2 7 7–28 6
Fig. 1 – SEM images of a silica fibre (a) before and after (b) 15 min, (c) 30 min and (d) 120 min MWCNT growth reaction. The locations of the silica fibres are indicated by arrows. Strong contrast in (a) is caused by the electron charging due to the low electrical conductivity of the as-received silica fibres; the arrow indicates the diameter of one fibre.
asymmetrically. A random orientation observed during nucleation [9], rapidly gives way to radial growth, due to steric interactions between MWCNTs, once the density of catalytic sites is sufficiently high [8], producing the symmetric ‘brush-like’ structure. As the MWCNTs continue to elongate via the base growth mechanism (see detailed discussion below), the circumference described by the tips expands. Eventually, the circumferential tension exceeds the inherent lateral strength of the growing nanotube layer, which has been mostly attributed to the inter-tube entanglements [30]. The layer then splits, or ‘parts’, breaking the cylindrical symmetry, and causing growth to continue in one direction (Fig. 1c and d). Similar behaviour has been reported for MWCNTs grown on other fibres [27]. In addition, gravity, steric
hindrance within the fibre tow, and gas flow effects may also contribute to this splitting phenomenon [31]. In Fig. 1d, regular bands of high contrast are visible, most likely due to variable iron content, as the iron-containing precursor was dripfed into the vapourisation furnace [32]. A statistical analysis of the apparent length and the outer diameter of the MWCNTs was carried out using high resolution SEM images taken from different areas (tip, middle and base of the nanotube array). More than 600 tube diameters per sample were measured using the open-source Java software ImageJ [33]. The average lengths and the outer diameters at different growth times are summarised in Table 1 and the outer diameter distributions are shown in Fig. 2a. As expected, the MWCNT lengths increased with growth time
Table 1 – Comparison of length (l) and outer diameter (D) of pure and N-doped MWCNTs grown on silica fibres at different growth times via ICVD. The standard deviations are shown in the brackets. t (min)
l (lm)
D (nm)
Dtip (nm)
Dmid (nm)
Dbase (nm)
MWCNTs
15 30 120
61 (5) 167 (12) 654 (23)
48 (13) 49 (13) 69 (31)
36 (7) 38 (7) 36 (7)
48 (8) 49 (7) 64 (17)
58 (10) 62 (13) 113 (12)
CNx–MWCNTs
60 120 240
33 (6) 91 (11) 132 (31)
27 (7) 60 (20) 65 (28)
19 (3) 42 (7) 38 (6)
28 (3) 63 (15) 61 (14)
33 (5) 75 (20) 93 (19)
280
CARBON
4 8 ( 2 0 1 0 ) 2 7 7 –2 8 6
Fig. 2 – The outer diameter distributions of (a) pure and (b) N-doped MWCNTs produced at different growth times.
Fig. 3 – (a–c) SEM and (d and e) TEM images of silica fibre surfaces after deliberate removal of the grafted MWCNTs grown in a 120 min reaction. (b) An enlarged view of the boxed area indicated in (a). (f) EDX analysis on the surface area marked in (e).
CARBON
4 8 (2 0 1 0) 2 7 7–28 6
from a few tens of micrometres to around half a millimetre. It was also found that both the average diameter of the MWCNTs, and the breadth of the diameter distributions increased with growth time. The measurements along the length of the MWCNTs clearly show the cause; namely that the diameter gradually thickened at the base of the MWCNTs. This trend could be due to either continued pyrolytic carbon deposition during tip growth or catalyst ripening during base growth. However, the literature suggests that injection CVD usually proceeds by base growth [8,9]; this conclusion is supported by the presence of iron particles at the MWCNT roots observed in this study (see detailed discussion below). Thus, the increasing MWCNT outer diameters at the base can be mainly attributed to catalyst ripening. In order to explore the nature of the MWCNT–fibre interaction further, the surfaces of the silica fibres after the reaction were characterised after deliberately removing the grafted MWCNTs using a razor blade. As shown in Fig. 3a and b, many particles and pits with diameters around 115 ± 13 nm were observed, which were a similar size to the grafted MWCNTs (refer to Dbase in Table 1), again suggesting a base growth mechanism (and catalyst ripening). The pitting also showed that the MWCNTs were attached to the fibre surface via etching or chemical reactions between the catalyst and the substrate. In some cases, the catalyst particles remained visible, embedded in the surface, in other cases, they had been pulled out with the MWCNTs (Fig. 3c); either way, the outlines of the carbon walls were visible as a ring. Note that the MWCNTs shown in Fig. 3c are at the larger end of the size distributions
281
shown in Fig. 2a, as they are located at the base of the longest growth experiment, and have been subjected to the greatest ripening; the majority of MWCNTs in the sample are significantly smaller. At high temperature, iron may diffuse into silica, forming iron oxide or an iron–silicon alloy [34,35]. Qu et al. [36] has reported similar surface damage caused by the hightemperature pyrolysis of iron phthalocyanine (FePc) on silicacoated carbon fibres. Cross-sections of the damaged surface were prepared by FIB sectioning and studied using TEM (Fig. 3d and e). The pits caused by the catalyst etching were clearly observed, with depths ranging between 20 and 55 nm (Fig. 3d); however, few catalyst particles remained attached to the surface after the FIB sectioning, which may be due to the relatively small thickness of the specimen, i.e., around 100 nm. Close-up images of the pits (Fig. 3e) displayed a number of dark features (around 15 nm), which are likely to be iron residue, as suggested by the EDX analysis (Fig. 3f). Other signals shown in the EDX, specifically C, Au, Si and Cu, came from the carbon residue, metal coating, fibre and TEM grid, respectively. The oxidation state of the iron is unclear, but the catalyst remaining after growth could either be iron carbide (Fe3C), a-iron, or c-iron [8,37].
3.1.2.
TGA characterisation
TGA was applied to assess the oxidation temperature and weight fraction of the MWCNTs grafted onto the silica fibres (Fig. 4a). The onset oxidation temperatures for weight loss, corresponding to the combustion of the MWCNTs, were 582, 620 and 654 °C for the samples grown for 15, 30 and
Fig. 4 – TGA curves of silica fibres grafted with (a) pure and (c) N-doped MWCNTs produced at different growth times. Calculated yields of (b) pure and (d) N-doped MWCNTs as a function of growth time.
282
CARBON
4 8 ( 2 0 1 0 ) 2 7 7 –2 8 6
120 min, respectively, suggesting that the degree of crystalline perfection of the MWCNTs was improved as the reaction continued. Assuming that the mass of iron was negligible (measured <3 wt.%), the mass ratios of grafted MWCNTs to silica fibres, Rm ¼ mCNT =mf , were calculated from the mass of residue, i.e. the mass of silica fibres, determined by TGA. Thus, the mass yield of the MWCNTs per unit area, gCNT , can be evaluated from the following equations: mCNT mCNT ¼ Sf 2nprf lf mf Vf ¼ ¼ npr2f lf qf gCNT ¼
ð1Þ ð2Þ
By substituting Eq. (2) into Eq. (1), the yield can be obtained using the equation:
gCNT ¼
qf rf mCNT qf rf ¼ Rm mf 2 2
ð3Þ
where mCNT and mf are the masses of the grafted MWCNTs and silica fibres, respectively, and Rm is their ratio, Sf , n, rf , lf , qf , and Vf are the surface area, number, radius, length, density and volume of the silica fibres, respectively. qf and rf are specified as 2.15 g/cm3 and 9 lm, respectively by the manufacturer. The average yields of MWCNTs grown in the 15, 30 and 120 min reactions were 0.09, 0.74 and 4.93 mg/cm2, respectively. The MWCNT yield is plotted as a function of the growth time in Fig. 4b. The growth rate of MWCNTs on silica fibres, under the experimental conditions used in this work, was linear at around 0.044 mg/(cm2 min). The linear fit crossed the
Fig. 5 – Raman spectra of silica fibres (a) before and (b) after the 15 and 30 min MWCNT grafting reactions. (c) SEM image of a MWCNT-grafted silica fibre produced in a 120 min reaction. (d) Raman spectra taken from different positions identified in (c) along the aligned MWCNT array. Relative Raman (e) intensity and (f) area ratios as a function of position along the MWCNT array.
CARBON
4 8 (2 0 1 0) 2 7 7–28 6
x axis, indicating that the induction period for the growth of MWCNTs was around 13 min. This result is consistent with the previous data shown in the literature [8], suggesting that the initial deposition of ferrocene took about 13.75 min, during which time the nucleation sites for MWCNT growth developed. Singh et al. [8] reported that the growth rate of MWCNTs on quartz materials is initially linear, but eventually reduces, probably due to the diffusion limitation of the hydrocarbon gas through the increasing thickness of the growing films. For the cylindrical geometry of grafting MWCNTs on fibres, diffusion limitations are less likely to be significant.
3.1.3.
Raman characterisation
Raman spectra of the silica fibres before and after MWCNT grafting are shown in Fig. 5a and b. Before the growth, only the silica signal at around 440 cm1 was detected [38]. The spectra taken from the silica fibres after 15 and 30 min reactions showed three main bands, which are the disorder-induced D mode (1321 cm1), the tangential G mode (1570 cm1) and second order G 0 mode (2642 cm1), consistent with the synthesis of MWCNTs during the reactions; no signal was observed for the radial breathing mode (RBM), range between 120 and 250 cm1 [39]. It is widely accepted that the D mode acts as a diagnostic for disruptions in the hexagonal framework of MWCNTs and the intensity ratio of the D mode to G mode can be used qualitatively to compare the crystallinity of CNTs [40]. Fig. 5b shows that the intensity of the D mode was approximately half that of the G mode. The
283
value is typical for ICVD-grown MWCNTs [8,25], but higher than that for most commercial CVD MWCNTs, suggesting a relatively high degree of crystallinity. A more detailed Raman study was carried out on the MWCNTs grafted in the 120 min reaction (Figs. 1d and 5c). The Raman spectra (Fig. 5d), taken at four different positions through the thickness of the layer, confirmed that only MWCNTs were synthesised during the reaction. The laser power and integration times used to record these spectra were identical. At least three spectra were taken from each position within the MWCNT array (Fig. 5c). The D:G intensity ratio, ID/IG, was about 0.62 at the MWCNT tips and fell to approximately 0.23 at the base (see Fig. 5e), close to the silica fibre, indicating a very high degree of crystallinity, particularly for CVD-grown materials. Assuming a base growth mechanism, the data once again show that the crystallinity increased during the growth, as the nanotubes thickened, in agreement with the TGA data. Several studies [41–43] have reported that normalising the D mode intensity with respect to the intensity of the G 0 mode is a more reliable method for estimating the defect concentration in nanotube samples [42]. A comparison of the various possible ratios of the Raman signals as a function of position, calculated based on both peak intensity (Fig. 5e) and area (Fig. 5f) showed similar trends in all cases. Of the three ratios tested, D:G 0 showed the greatest relative variation from tip to base. This effect is more apparent while plotting using the inverse ratios (Supplementary information, Fig. S2).
Fig. 6 – SEM images of silica fibres after (a) 60 min, (b) 120 min and (c) 240 min CNx–MWCNT growth reaction. (d) A close-up of silica fibre surfaces after deliberate removal of the CNx–MWCNT grafting. The positions of silica fibres are indicated by dashed lines and the catalyst particles are indicated by arrows.
284
CARBON
3.2.
N-doped MWCNT-grafted silica fibres
3.2.1.
Growth of CNx–MWCNTs on silica fibres
4 8 ( 2 0 1 0 ) 2 7 7 –2 8 6
Aligned CNx–MWCNTs were grown on silica fibres for all reactions, with growth times ranging from 60 to 240 min (Fig. 6a– c), showing the radial ‘brush-like’ morphology, that was observed in the growth of pure CNTs at short reaction times (Fig. 1b). Fig. 6d shows a close-up image of the fibre surface after the growth reaction, with a few CNx–MWCNTs having been pulled out from the fibres; it is clear that the catalyst sites were very densely distributed on the surface. However, relatively few embedded catalyst particles remained on the fibre surface, suggesting a different balance of fibre-catalyst– CNT interactions than for the pure MWCNTs (refer to Fig. 3b). A change in catalyst behaviour was also evidenced by the elongated iron particles (indicated by arrows) found at the root of the nanotubes. The growth of CNx–MWCNTs also appears to have followed a base growth mechanism, in accordance with previous work [14,44]; the elongated catalyst shape is also characteristic of CNx–MWCNTs [11,14,17]. The statistical analysis of the apparent length and the outer diameter of the CNx–MWCNTs (Table 1) showed the same trends as that observed for pure MWCNTs; the average length and outer diameter, as well as the outer diameter distribution (see Fig. 2b) increased with increasing growth time. Again, the
effect can be attributed to catalyst ripening during the reaction, particularly with the continued supply of iron. After the same reaction time (120 min), the length of CNx–MWCNTs was 91 ± 11 lm, only 14% of that of pure MWCNTs, 654 ± 23 lm. The reduced growth rate is thought to be related to the altered nature of the catalyst–nanotube interaction [15,45].
3.2.2.
TGA characterisation
The onset temperatures for the TGA of CNx–MWCNTs (Fig. 4c) were 475, 484 and 487 °C after the 60, 120 and 240 min reactions, respectively. The trend is similar to the pure CNTs, indicating that the degree of crystalline perfection of the nanotubes was slightly improved as the reaction continued. The lower combustion temperature of CNx–MWCNTs, compared to that of pure MWCNTs, can be attributed to the increasing defect concentration and reactivity on incorporation of nitrogen [15]. Based on the analysis of TGA data, the average yields of CNx–MWCNTs grown in the 60, 120 and 240 min reactions were 0.04, 0.80 and 2.33 mg/cm2, respectively. By plotting the yield as a function of growth time (Fig. 4d), a linear growth rate of 0.013 mg/(cm2 min) was identified, with an induction period of around 55 min. The lower growth rate and longer induction period of CNx–MWCNTs suggested that the incorporation of nitrogen inhibited the
Fig. 7 – (a) SEM image of a silica fibre grafted with CNx–MWCNTs produced in a 240 min reaction. (b) Raman spectra taken from different positions identified in (a) along the aligned nanotube array. (c) Relative Raman ratios and (d) FWHM (cm1) of the D and G modes as a function of position along the nanotube array.
CARBON
4 8 (2 0 1 0) 2 7 7–28 6
growth of nanotubes. Theoretical calculations [45] have indicated that nitrogen prefers to segregate to the nanotube edge, inhibiting growth.
3.2.3.
Raman characterisation
The Raman spectra taken from three different regions along the CNx–MWCNT arrays are displayed in Fig. 7a and b. The D:G ratio, was used to describe the degree of disorder in this case, since the G 0 mode was relatively weak. For CNx– MWCNTs, the disorder introduced by the nitrogen, both in local bonding and overall microstructure [12], tended to increase the D:G ratio. In addition, both intensity (ID/IG) and area (AD/AG) ratios decreased from the tip to the base of the arrays, with the effect more evident in the case of the area ratio (see Fig. 7c), again indicating increasing crystalline quality as growth proceeded, consistent with the TGA data. As an alternative approach, Fig. 7d shows the changes of the full width at half maximum (FWHM) of the D and G modes as a function of position. The values of the FWHM for both D and G modes dropped from tip to base, with a stronger effect on the D mode. The narrowing Raman signals indicated a larger graphitic crystal domain size and consequently a lower defect concentration [12,46].
4.
Conclusions
Aligned, pure and N-doped carbon nanotubes were grown on silica fibres simply by using the ICVD method; the thickness of the grafted nanotube layer can be controlled from a few tens of micrometres to over half a millimetre by varying the growth time. SEM and TEM observation on fibre surfaces after the growth reaction indicated that the nanotubes were bonded to fibres through the catalyst etching into the fibre surfaces; the location of the catalyst implied a base growth mechanism. The average diameter of the nanotubes increased with prolonged growth time in both cases due to the catalyst ripening during the reaction; at the same time, the crystalline quality increased, as shown by both TGA and spatially-resolved Raman. The increase in crystallinity could be a size effect and/or due to the increasing packing density of the MWCNTs. The CNx–MWCNTs have a higher intrinsic packing density, associated with a slower growth rate and longer induction time. However, they appear more defective in both TGA and Raman, due to the disorder introduced by the nitrogen. This paper demonstrates an effective method for producing (both pure and N-doped) nanotube grafted silica fibres and provides a further understanding of the growth mechanism. CNx–MWCNTs may offer advantages in the context of hierarchical composites due to the greater packing density and potential for chemical interaction with a matrix. Any improved electrical conductivity associated with nitrogen doping may also be an advantage, for example, for lightening strike protection, although the high defect concentration may be limiting. Further investigations of mechanical and electrical properties of these nanotube grafted fibres and the CNT–fibre interaction are required to establish their potential in a new generation of composite materials.
285
Acknowledgements The authors would like to thank DSTL and QinetiQ for the financial support of this work. The authors also thank Mahmoud G. Ardakani and Richard J. Chater (Imperial College London) for their assistance on TEM and FIB.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.carbon.2009.09.029.
R E F E R E N C E
[1] Dresselhaus MS, Dresselhaus G, Avouris P. Carbon nanotubes: synthesis, structure, properties and applications. Berlin: Springer; 2001. p. 81–172. [2] Popov VN. Carbon nanotubes: properties and application. Mater Sci Eng R Rep 2004;43(3):61–102. [3] Coleman JN, Khan U, Blau WJ, Gun’ko YK. Small but strong: a review of the mechanical properties of carbon nanotube– polymer composites. Carbon 2006;44:1624–52. [4] Terrones M. Carbon nanotubes: synthesis and properties, electronic devices and other emerging applications. Int Mater Rev 2004;49(6):325–77. [5] Dai LM, Patil A, Gong XY, Guo ZX, Liu LQ, Liu Y, et al. Aligned nanotubes. ChemPhysChem 2003;4(11):1150–69. [6] Hata K, Futaba DN, Mizuno K, Namai T, Yumura M, Iijima S. Water-assisted highly efficient synthesis of impurity-free single-waited carbon nanotubes. Science 2004;306(5700):1362–4. [7] Andrews R, Jacques D, Rao AM, Derbyshire F, Qian D, Fan X, et al. Continuous production of aligned carbon nanotubes: a step closer to commercial realization. Chem Phys Lett 1999;303:467–74. [8] Singh C, Shaffer MSP, Windle AH. Production of controlled architectures of aligned carbon nanotubes by an injection chemical vapour deposition method. Carbon 2003;41:359–68. [9] Pinault M, Pichot V, Khodja H, Launois P, Reynaud C, MayneL’Hermite M. Evidence of sequential lift in growth of aligned multiwalled carbon nanotube multilayers. Nano Lett 2005;5:2394–8. [10] Mayne M, Grobert N, Terrones M, Kamalakaran R, Ruhle M, Kroto HW, et al. Pyrolytic production of aligned carbon nanotubes from homogeneously dispersed benzene-based aerosols. Chem Phys Lett 2001;338(2–3):101–7. [11] Koziol K, Shaffer M, Windle A. Three-dimensional internal order in multiwalled carbon nanotubes grown by chemical vapor deposition. Adv Mater 2005;17:760–3. [12] Liang EJ, Ding P, Zhang HR, Guo XY, Du ZL. Synthesis and correlation study on the morphology and Raman spectra of CNx nanotubes by thermal decomposition of ferrocene/ ethylenediamine. Diamond Relat Mater 2004;13:69–73. [13] Liu J, Czerw R, Carroll DL. Large-scale synthesis of highly aligned nitrogen doped carbon nanotubes by injection chemical vapor deposition methods. J Mater Res 2005;20(2):538–43. [14] Reyes-Reyes M, Grobert N, Kamalakaran R, Seeger T, Golberg D, Ruhle M, et al. Efficient encapsulation of gaseous nitrogen inside carbon nanotubes with bamboo-like structure using aerosol thermolysis. Chem Phys Lett 2004;396(1–3):167–73. [15] Koos AA, Dowling M, Jurkschat K, Crossley A, Grobert N. Effect of the experimental parameters on the structure of
286
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25] [26]
[27]
[28]
[29]
[30]
[31]
CARBON
4 8 ( 2 0 1 0 ) 2 7 7 –2 8 6
nitrogen-doped carbon nanotubes produced by aerosol chemical vapour deposition. Carbon 2009;47(1):30–7. He MS, Zhou S, Zhang J, Liu ZF, Robinson C. CVD growth of Ndoped carbon nanotubes on silicon substrates and its mechanism. J Phys Chem B 2005;109:9275–9. Ducati C, Koziol K, Friedrichs S, Yates TJV, Shaffer MS, Midgley PA, et al. Crystallographic order in multi-walled carbon nanotubes synthesized in the presence of nitrogen. Small 2006;2(6):774–84. Huang YH, Gao JP, Liu RZ. Structure and electronic properties of nitrogen-containing carbon nanotubes. Synth Met 2000;113(3):251–5. Terrones M, Jorio A, Endo M, Rao AM, Kim YA, Hayashi T, et al. New direction in nanotube science. Mater Today 2004;7:30–45. Shaffer M, Sandler J. Carbon nanotube/nanofibre polymer composites. In: Advani SG, editor. Processing and properties of nanocomposites. New Jersey, NJ [etc.]: World Scientific; 2007. p. 1–59. Thostenson ET, Li WZ, Wang DZ, Ren ZF, Chou TW. Carbon nanotubes/carbon fiber hybrid multiscale composites. J Appl Phys 2002;91:6034–7. Veedu VP, Cao A, Li X, Ma K, Soldano C, Kar S, et al. Multifunctional composites using reinforced laminae with carbon-nanotube forests. Nat Mater 2006;5:457–62. Qian H, Bismarck A, Greenhalgh ES, Kalinka G, Shaffer MSP. Hierarchical composites reinforced with carbon nanotube grafted fibers: the potential assessed at the single fiber level. Chem Mater 2008;20(5):1862–9. Garcia EJ, Wardle BL, John Hart A, Yamamoto N. Fabrication and multifunctional properties of a hybrid laminate with aligned carbon nanotubes grown in situ. Compos Sci Technol 2008;68:2034–41. Ci LJ, Zhao ZG, Bai JB. Direct growth of carbon nanotubes on the surface of ceramic fibers. Carbon 2005;43:883–6. Zhu S, Su CH, Lehoczky SL, Muntele I, Ila D. Carbon nanotube growth on carbon fibers. Diamond Relat Mater 2003;12:1825–8. Cao AY, Veedu VP, Li XS, Yao ZL, Ghasemi-Nejhad MN, Ajayan PM. Multifunctional brushes made from carbon nanotubes. Nat Mater 2005;4:540–5. Zhang Q, Qian WZ, Xiang R, Yang Z, Luo GH, Wang Y, et al. In situ growth of carbon nanotubes on inorganic fibers with different surface properties. Mater Chem Phys 2008;107(2– 3):317–21. Nakahara S. Recent development in a TEM specimen preparation technique using FIB for semiconductor devices. Surf Coat Technol 2003;169:721–7. Zhang M, Fang S, Zakhidov AA, Lee SB, Aliev AE, Williams CD, et al. Strong, transparent, multifunctional, carbon nanotube sheets. Science 2005;309:1215–9. Zhu S, Su CH, Cochrane JC, Lehoczky S, Cui Y, Burger A. Growth orientation of carbon nanotubes by thermal chemical vapor deposition. J Cryst Growth 2002;234:584–8.
[32] Cao AY, Ci LJ, Wu GW, Wei BQ, Xu CL, Liang J, et al. An effective way to lower catalyst content in well-aligned carbon nanotube films. Carbon 2001;39:152–5. [33] Rasband WS. ImageJ: US National Institutes of Health, Bethesda, Maryland, USA,
4 8 ( 2 0 1 0 ) 2 7 7 –2 8 6
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/carbon
Synthesis and characterisation of carbon nanotubes grown on silica fibres by injection CVD Hui Qian a b c
a,b
, Alexander Bismarck b, Emile S. Greenhalgh c, Milo S.P. Shaffer
a,*
Department of Chemistry, Imperial College London, London SW7 2AZ, UK Polymer and Composite Engineering (PaCE) Group, Department of Chemical Engineering, Imperial College London, London SW7 2AZ, UK The Composites Centre, Imperial College London, London SW7 2AZ, UK
A R T I C L E I N F O
A B S T R A C T
Article history:
Carbon nanotube grafted primary reinforcing fibres are under development for a new gen-
Received 1 June 2009
eration of hierarchical composites. Pure and nitrogen-doped multi-walled carbon nano-
Accepted 7 September 2009
tubes (MWCNTs) were grown on silica fibres using the injection chemical vapour
Available online 11 September 2009
deposition method. The morphology and size of the nanotubes were controlled by varying the growth time. The surface structure of the silica fibres after the grafting process was studied by electron microscopy following focused ion beam sectioning; the images confirmed a base growth mechanism and a shallow iron-rich layer. Thermogravimetric analysis indicated a shorter induction period and a faster growth rate for pure rather than nitrogen-doped MWCNTs. Raman characterisation of the grafted MWCNTs showed a decreasing intensity ratio of the D to G modes, moving from the tip to base in both cases; a detailed comparison of different characteristic Raman ratios is provided, using both the peak intensity and the area for the D, G, and G 0 signals. Ó 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
Carbon nanotubes (CNTs) continue to receive huge interest due to their extraordinary mechanical, thermal and electronic properties [1,2] and potential applications in a wide range of fields, such as composites [3] and electronic devices [4]. Amongst a variety of methods to produce CNTs, catalytic chemical vapour deposition (CVD) is widely accepted as the most efficient for large scale synthesis of CNTs [5,6], tending to produce relatively pure, but defective and entangled materials. Alternatively, aligned CNT arrays can be grown by CVD, often using an injection method (ICVD) which involves the simultaneous pyrolysis of solutions containing both the catalyst precursor and the hydrocarbon source [7–10], without the time-consuming preparation of catalyst pre-coated substrates. This method has been used to grow well-aligned nitrogen-doped carbon nanotubes (CNx–CNTs) by injecting
nitrogen-containing organic precursors [11–16], such as pyridine, diazine, acetonitrile and so on. The addition of diazine or triazine precursors has been shown to improve the internal crystalline order, straightness and packing of ICVD-grown nanotubes [11,17]. CNx–CNTs are either metals or narrow energy gap semiconductors, thus possibly offering better electrical conductivity than pure CNTs [16,18]. In addition, the nitrogen sites in the graphene network can also improve the nanotube-matrix interactions in composites [19]. Over the past decade, many researchers have explored the use of CNTs in composites, as a replacement for conventional microscale reinforcing fillers, such as carbon or glass fibres [20]. However, the direct use of CNTs remains limited by the quality of the materials available in bulk and the difficulties associated with dispersion, alignment, and the CNT/matrix interface. Meanwhile, interest is growing in the opportunity to combine CNTs with conventional fibres to create hierarchical
* Corresponding author: Fax: +44 (0)20 7594 5801. E-mail address: [email protected] (M.S.P. Shaffer). 0008-6223/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.carbon.2009.09.029
278
CARBON
4 8 ( 2 0 1 0 ) 2 7 7 –2 8 6
composites. Of the available processing strategies, one of the most appealing is to grow CNTs directly onto fibre surfaces [21–24]; this approach has the potential to provide high loadings of well-spaced CNTs, with a possibly optimal, radial orientation around the fibres. It is reported that the grafting of CNTs on fibres improves the interfacial properties between fibres and matrix [21,23]. In addition, the presence of CNTs in matrix-rich regions could alleviate some of the drawbacks of conventional fibre reinforced polymer composites, especially longitudinal compression performance, interlaminar properties, and potentially damage tolerance [22,24]. Ci et al. [25] first reported the synthesis of CNTs on the surface of ceramic fibres by using the ICVD method and deduced that the base growth mechanism applies. Since then, CNTs have been grown on various substrates, including carbon [26], SiC [22,27], alumina [24], quartz and aluminium silicate [28] fibres or woven fabrics. The present study systematically investigated the growth of pure and N-doped MWCNTs on silica fibres using the ICVD method. The morphology, length and diameter of nanotubes grown at different growth times were characterised using scanning electron microscopy (SEM). The consistency of the ICVD-grown MWCNTs along the nanotube array and fibres was also investigated. The interface structure between nanotubes and fibres is of particular relevance to composite performance; the local structure of the interface was explored for the first time by using the focused ion beam (FIB) technique to prepare cross-sections, from which the growth mechanism of nanotubes on fibres can be deduced. Moreover, the effects of the nitrogenincorporation on the growth rate, oxidation resistance and crystallinity of MWCNTs grown on fibres were studied using thermogravimetric analysis (TGA) and Raman characterisation.
2.
Experimental
2.1.
Materials
Commercially available Silfa silica fibres (9 lm in diameter, Albert Hellhake GmbH) were used for this work. A thermal desizing treatment of the fibres was carried out prior to CNT grafting at 600 °C for 1 h. All chemicals were used as received.
2.2.
Growth of carbon nanotubes on fibres
Pure and N-doped MWCNTs were synthesised on silica fibres using the injection chemical vapour deposition method [8,11,13]. The reactions were carried out in a tubular quartz reactor (50 mm in diameter) equipped with an electrical furnace (PTF 12/50/610, Lenton). The silica fibres were held by an alumina frame and placed in the middle of the furnace to expose them to the hydrocarbon source. For pure MWCNTs, a feed solution of 3 wt.% ferrocene (98%, Sigma–Aldrich) in toluene (BDH AnalaR grade, VWR International Ltd.) was injected continuously into the reaction tube at a rate of 5 ml/h using a syringe pump (KDS100, Linton). For CNx– MWCNTs, a feed solution consisting mixture of 3 wt.% ferrocene, 58.5 wt.% toluene and 38.5 wt.% pyrazine (Kosher grade, Sigma–Aldrich) was used. The liquid feed was preheated to
200 °C, immediately volatilised, and swept into the furnace by a flow of argon (Ar) as carrier gas. The reaction temperature was 760 °C. The growth times for pure MWCNTs were 15, 30 and 120 min. Longer growth times, of 60, 120 and 240 min, were used for the synthesis of CNx–MWCNTs, due to their slower growth rate.
2.3.
Electron microscopy characterisation
The morphology, alignment and size of the grafted nanotubes on the silica fibres were characterised using a field emission gun scanning electron microscope (Gemini LEO 1525 FEGSEM, Carl Zeiss NTS GmbH). The surface of modified fibres was studied using a transmission electron microscope (TEM, JEM-2000FXII Electron Microscope, JEOL, Inc.) operating at 200 kV. Energy dispersive X-ray (EDX) analysis was performed using the INCA X-ray microanalysis system (4.08 suite version, Oxford Instruments). FIB sectioning [29] was carried out using FIB 200 TEM focused ion beam instrument (FEI, Inc.). After deliberate removal of the CNT layer, a fibre was glued onto a copper TEM grid, which was partly cut for better handling. The specimen was pre-coated with gold using a sputter coater (K-550X, Emitech Ltd.). During the FIB sectioning process, a gallium liquid metal ion source (Ga+) was used as a primary ion beam with an energy of 30 keV. The area of interest on the fibre surface was protected by coating with a thin layer of platinum prior to etching. The ultimate thickness of the defined area on the specimen was approximately 100 nm (Supplementary information, Fig. S1).
2.4.
Thermogravimetric analysis
TGA measurements were carried out using a Perkin–Elmer Pyris 1 thermogravimetric analyser (Perkin–Elmer, Inc.). Samples were analysed in a platinum pan at a heating rate of 2 °C/ min up to 900 °C, in an atmosphere of air flowing at 20 mL/ min. Sample masses ranged from 1 to 2 mg.
2.5.
Raman spectroscopy
The Raman spectra presented in this report were collected in the backscattered geometry using a LabRam infinity analytical Raman spectrometer (HORIBA Jobin Yvon Ltd.), at room temperature, using a linearly polarised Helium–Neon laser (k = 632.8 nm). The diameter of the laser spot was a few micrometres and the resolution of the spectrometer was about 1 cm1. All the spectra were fitted with a mixed Gaussian–Lorentzian function using GRAMS software.
3.
Results and discussion
3.1.
Pure MWCNT-grafted silica fibres
3.1.1.
Growth of MWCNTs on silica fibres
SEM images (Fig. 1) showed that well-aligned MWCNTs were successfully grown onto the silica fibres at all growth times between 15 and 120 min; however, different morphologies were obtained. Radial growth in the early stages resulted in a ‘brush-like structure’ (Fig. 1b), which later ‘parted’
CARBON
279
4 8 (2 0 1 0) 2 7 7–28 6
Fig. 1 – SEM images of a silica fibre (a) before and after (b) 15 min, (c) 30 min and (d) 120 min MWCNT growth reaction. The locations of the silica fibres are indicated by arrows. Strong contrast in (a) is caused by the electron charging due to the low electrical conductivity of the as-received silica fibres; the arrow indicates the diameter of one fibre.
asymmetrically. A random orientation observed during nucleation [9], rapidly gives way to radial growth, due to steric interactions between MWCNTs, once the density of catalytic sites is sufficiently high [8], producing the symmetric ‘brush-like’ structure. As the MWCNTs continue to elongate via the base growth mechanism (see detailed discussion below), the circumference described by the tips expands. Eventually, the circumferential tension exceeds the inherent lateral strength of the growing nanotube layer, which has been mostly attributed to the inter-tube entanglements [30]. The layer then splits, or ‘parts’, breaking the cylindrical symmetry, and causing growth to continue in one direction (Fig. 1c and d). Similar behaviour has been reported for MWCNTs grown on other fibres [27]. In addition, gravity, steric
hindrance within the fibre tow, and gas flow effects may also contribute to this splitting phenomenon [31]. In Fig. 1d, regular bands of high contrast are visible, most likely due to variable iron content, as the iron-containing precursor was dripfed into the vapourisation furnace [32]. A statistical analysis of the apparent length and the outer diameter of the MWCNTs was carried out using high resolution SEM images taken from different areas (tip, middle and base of the nanotube array). More than 600 tube diameters per sample were measured using the open-source Java software ImageJ [33]. The average lengths and the outer diameters at different growth times are summarised in Table 1 and the outer diameter distributions are shown in Fig. 2a. As expected, the MWCNT lengths increased with growth time
Table 1 – Comparison of length (l) and outer diameter (D) of pure and N-doped MWCNTs grown on silica fibres at different growth times via ICVD. The standard deviations are shown in the brackets. t (min)
l (lm)
D (nm)
Dtip (nm)
Dmid (nm)
Dbase (nm)
MWCNTs
15 30 120
61 (5) 167 (12) 654 (23)
48 (13) 49 (13) 69 (31)
36 (7) 38 (7) 36 (7)
48 (8) 49 (7) 64 (17)
58 (10) 62 (13) 113 (12)
CNx–MWCNTs
60 120 240
33 (6) 91 (11) 132 (31)
27 (7) 60 (20) 65 (28)
19 (3) 42 (7) 38 (6)
28 (3) 63 (15) 61 (14)
33 (5) 75 (20) 93 (19)
280
CARBON
4 8 ( 2 0 1 0 ) 2 7 7 –2 8 6
Fig. 2 – The outer diameter distributions of (a) pure and (b) N-doped MWCNTs produced at different growth times.
Fig. 3 – (a–c) SEM and (d and e) TEM images of silica fibre surfaces after deliberate removal of the grafted MWCNTs grown in a 120 min reaction. (b) An enlarged view of the boxed area indicated in (a). (f) EDX analysis on the surface area marked in (e).
CARBON
4 8 (2 0 1 0) 2 7 7–28 6
from a few tens of micrometres to around half a millimetre. It was also found that both the average diameter of the MWCNTs, and the breadth of the diameter distributions increased with growth time. The measurements along the length of the MWCNTs clearly show the cause; namely that the diameter gradually thickened at the base of the MWCNTs. This trend could be due to either continued pyrolytic carbon deposition during tip growth or catalyst ripening during base growth. However, the literature suggests that injection CVD usually proceeds by base growth [8,9]; this conclusion is supported by the presence of iron particles at the MWCNT roots observed in this study (see detailed discussion below). Thus, the increasing MWCNT outer diameters at the base can be mainly attributed to catalyst ripening. In order to explore the nature of the MWCNT–fibre interaction further, the surfaces of the silica fibres after the reaction were characterised after deliberately removing the grafted MWCNTs using a razor blade. As shown in Fig. 3a and b, many particles and pits with diameters around 115 ± 13 nm were observed, which were a similar size to the grafted MWCNTs (refer to Dbase in Table 1), again suggesting a base growth mechanism (and catalyst ripening). The pitting also showed that the MWCNTs were attached to the fibre surface via etching or chemical reactions between the catalyst and the substrate. In some cases, the catalyst particles remained visible, embedded in the surface, in other cases, they had been pulled out with the MWCNTs (Fig. 3c); either way, the outlines of the carbon walls were visible as a ring. Note that the MWCNTs shown in Fig. 3c are at the larger end of the size distributions
281
shown in Fig. 2a, as they are located at the base of the longest growth experiment, and have been subjected to the greatest ripening; the majority of MWCNTs in the sample are significantly smaller. At high temperature, iron may diffuse into silica, forming iron oxide or an iron–silicon alloy [34,35]. Qu et al. [36] has reported similar surface damage caused by the hightemperature pyrolysis of iron phthalocyanine (FePc) on silicacoated carbon fibres. Cross-sections of the damaged surface were prepared by FIB sectioning and studied using TEM (Fig. 3d and e). The pits caused by the catalyst etching were clearly observed, with depths ranging between 20 and 55 nm (Fig. 3d); however, few catalyst particles remained attached to the surface after the FIB sectioning, which may be due to the relatively small thickness of the specimen, i.e., around 100 nm. Close-up images of the pits (Fig. 3e) displayed a number of dark features (around 15 nm), which are likely to be iron residue, as suggested by the EDX analysis (Fig. 3f). Other signals shown in the EDX, specifically C, Au, Si and Cu, came from the carbon residue, metal coating, fibre and TEM grid, respectively. The oxidation state of the iron is unclear, but the catalyst remaining after growth could either be iron carbide (Fe3C), a-iron, or c-iron [8,37].
3.1.2.
TGA characterisation
TGA was applied to assess the oxidation temperature and weight fraction of the MWCNTs grafted onto the silica fibres (Fig. 4a). The onset oxidation temperatures for weight loss, corresponding to the combustion of the MWCNTs, were 582, 620 and 654 °C for the samples grown for 15, 30 and
Fig. 4 – TGA curves of silica fibres grafted with (a) pure and (c) N-doped MWCNTs produced at different growth times. Calculated yields of (b) pure and (d) N-doped MWCNTs as a function of growth time.
282
CARBON
4 8 ( 2 0 1 0 ) 2 7 7 –2 8 6
120 min, respectively, suggesting that the degree of crystalline perfection of the MWCNTs was improved as the reaction continued. Assuming that the mass of iron was negligible (measured <3 wt.%), the mass ratios of grafted MWCNTs to silica fibres, Rm ¼ mCNT =mf , were calculated from the mass of residue, i.e. the mass of silica fibres, determined by TGA. Thus, the mass yield of the MWCNTs per unit area, gCNT , can be evaluated from the following equations: mCNT mCNT ¼ Sf 2nprf lf mf Vf ¼ ¼ npr2f lf qf gCNT ¼
ð1Þ ð2Þ
By substituting Eq. (2) into Eq. (1), the yield can be obtained using the equation:
gCNT ¼
qf rf mCNT qf rf ¼ Rm mf 2 2
ð3Þ
where mCNT and mf are the masses of the grafted MWCNTs and silica fibres, respectively, and Rm is their ratio, Sf , n, rf , lf , qf , and Vf are the surface area, number, radius, length, density and volume of the silica fibres, respectively. qf and rf are specified as 2.15 g/cm3 and 9 lm, respectively by the manufacturer. The average yields of MWCNTs grown in the 15, 30 and 120 min reactions were 0.09, 0.74 and 4.93 mg/cm2, respectively. The MWCNT yield is plotted as a function of the growth time in Fig. 4b. The growth rate of MWCNTs on silica fibres, under the experimental conditions used in this work, was linear at around 0.044 mg/(cm2 min). The linear fit crossed the
Fig. 5 – Raman spectra of silica fibres (a) before and (b) after the 15 and 30 min MWCNT grafting reactions. (c) SEM image of a MWCNT-grafted silica fibre produced in a 120 min reaction. (d) Raman spectra taken from different positions identified in (c) along the aligned MWCNT array. Relative Raman (e) intensity and (f) area ratios as a function of position along the MWCNT array.
CARBON
4 8 (2 0 1 0) 2 7 7–28 6
x axis, indicating that the induction period for the growth of MWCNTs was around 13 min. This result is consistent with the previous data shown in the literature [8], suggesting that the initial deposition of ferrocene took about 13.75 min, during which time the nucleation sites for MWCNT growth developed. Singh et al. [8] reported that the growth rate of MWCNTs on quartz materials is initially linear, but eventually reduces, probably due to the diffusion limitation of the hydrocarbon gas through the increasing thickness of the growing films. For the cylindrical geometry of grafting MWCNTs on fibres, diffusion limitations are less likely to be significant.
3.1.3.
Raman characterisation
Raman spectra of the silica fibres before and after MWCNT grafting are shown in Fig. 5a and b. Before the growth, only the silica signal at around 440 cm1 was detected [38]. The spectra taken from the silica fibres after 15 and 30 min reactions showed three main bands, which are the disorder-induced D mode (1321 cm1), the tangential G mode (1570 cm1) and second order G 0 mode (2642 cm1), consistent with the synthesis of MWCNTs during the reactions; no signal was observed for the radial breathing mode (RBM), range between 120 and 250 cm1 [39]. It is widely accepted that the D mode acts as a diagnostic for disruptions in the hexagonal framework of MWCNTs and the intensity ratio of the D mode to G mode can be used qualitatively to compare the crystallinity of CNTs [40]. Fig. 5b shows that the intensity of the D mode was approximately half that of the G mode. The
283
value is typical for ICVD-grown MWCNTs [8,25], but higher than that for most commercial CVD MWCNTs, suggesting a relatively high degree of crystallinity. A more detailed Raman study was carried out on the MWCNTs grafted in the 120 min reaction (Figs. 1d and 5c). The Raman spectra (Fig. 5d), taken at four different positions through the thickness of the layer, confirmed that only MWCNTs were synthesised during the reaction. The laser power and integration times used to record these spectra were identical. At least three spectra were taken from each position within the MWCNT array (Fig. 5c). The D:G intensity ratio, ID/IG, was about 0.62 at the MWCNT tips and fell to approximately 0.23 at the base (see Fig. 5e), close to the silica fibre, indicating a very high degree of crystallinity, particularly for CVD-grown materials. Assuming a base growth mechanism, the data once again show that the crystallinity increased during the growth, as the nanotubes thickened, in agreement with the TGA data. Several studies [41–43] have reported that normalising the D mode intensity with respect to the intensity of the G 0 mode is a more reliable method for estimating the defect concentration in nanotube samples [42]. A comparison of the various possible ratios of the Raman signals as a function of position, calculated based on both peak intensity (Fig. 5e) and area (Fig. 5f) showed similar trends in all cases. Of the three ratios tested, D:G 0 showed the greatest relative variation from tip to base. This effect is more apparent while plotting using the inverse ratios (Supplementary information, Fig. S2).
Fig. 6 – SEM images of silica fibres after (a) 60 min, (b) 120 min and (c) 240 min CNx–MWCNT growth reaction. (d) A close-up of silica fibre surfaces after deliberate removal of the CNx–MWCNT grafting. The positions of silica fibres are indicated by dashed lines and the catalyst particles are indicated by arrows.
284
CARBON
3.2.
N-doped MWCNT-grafted silica fibres
3.2.1.
Growth of CNx–MWCNTs on silica fibres
4 8 ( 2 0 1 0 ) 2 7 7 –2 8 6
Aligned CNx–MWCNTs were grown on silica fibres for all reactions, with growth times ranging from 60 to 240 min (Fig. 6a– c), showing the radial ‘brush-like’ morphology, that was observed in the growth of pure CNTs at short reaction times (Fig. 1b). Fig. 6d shows a close-up image of the fibre surface after the growth reaction, with a few CNx–MWCNTs having been pulled out from the fibres; it is clear that the catalyst sites were very densely distributed on the surface. However, relatively few embedded catalyst particles remained on the fibre surface, suggesting a different balance of fibre-catalyst– CNT interactions than for the pure MWCNTs (refer to Fig. 3b). A change in catalyst behaviour was also evidenced by the elongated iron particles (indicated by arrows) found at the root of the nanotubes. The growth of CNx–MWCNTs also appears to have followed a base growth mechanism, in accordance with previous work [14,44]; the elongated catalyst shape is also characteristic of CNx–MWCNTs [11,14,17]. The statistical analysis of the apparent length and the outer diameter of the CNx–MWCNTs (Table 1) showed the same trends as that observed for pure MWCNTs; the average length and outer diameter, as well as the outer diameter distribution (see Fig. 2b) increased with increasing growth time. Again, the
effect can be attributed to catalyst ripening during the reaction, particularly with the continued supply of iron. After the same reaction time (120 min), the length of CNx–MWCNTs was 91 ± 11 lm, only 14% of that of pure MWCNTs, 654 ± 23 lm. The reduced growth rate is thought to be related to the altered nature of the catalyst–nanotube interaction [15,45].
3.2.2.
TGA characterisation
The onset temperatures for the TGA of CNx–MWCNTs (Fig. 4c) were 475, 484 and 487 °C after the 60, 120 and 240 min reactions, respectively. The trend is similar to the pure CNTs, indicating that the degree of crystalline perfection of the nanotubes was slightly improved as the reaction continued. The lower combustion temperature of CNx–MWCNTs, compared to that of pure MWCNTs, can be attributed to the increasing defect concentration and reactivity on incorporation of nitrogen [15]. Based on the analysis of TGA data, the average yields of CNx–MWCNTs grown in the 60, 120 and 240 min reactions were 0.04, 0.80 and 2.33 mg/cm2, respectively. By plotting the yield as a function of growth time (Fig. 4d), a linear growth rate of 0.013 mg/(cm2 min) was identified, with an induction period of around 55 min. The lower growth rate and longer induction period of CNx–MWCNTs suggested that the incorporation of nitrogen inhibited the
Fig. 7 – (a) SEM image of a silica fibre grafted with CNx–MWCNTs produced in a 240 min reaction. (b) Raman spectra taken from different positions identified in (a) along the aligned nanotube array. (c) Relative Raman ratios and (d) FWHM (cm1) of the D and G modes as a function of position along the nanotube array.
CARBON
4 8 (2 0 1 0) 2 7 7–28 6
growth of nanotubes. Theoretical calculations [45] have indicated that nitrogen prefers to segregate to the nanotube edge, inhibiting growth.
3.2.3.
Raman characterisation
The Raman spectra taken from three different regions along the CNx–MWCNT arrays are displayed in Fig. 7a and b. The D:G ratio, was used to describe the degree of disorder in this case, since the G 0 mode was relatively weak. For CNx– MWCNTs, the disorder introduced by the nitrogen, both in local bonding and overall microstructure [12], tended to increase the D:G ratio. In addition, both intensity (ID/IG) and area (AD/AG) ratios decreased from the tip to the base of the arrays, with the effect more evident in the case of the area ratio (see Fig. 7c), again indicating increasing crystalline quality as growth proceeded, consistent with the TGA data. As an alternative approach, Fig. 7d shows the changes of the full width at half maximum (FWHM) of the D and G modes as a function of position. The values of the FWHM for both D and G modes dropped from tip to base, with a stronger effect on the D mode. The narrowing Raman signals indicated a larger graphitic crystal domain size and consequently a lower defect concentration [12,46].
4.
Conclusions
Aligned, pure and N-doped carbon nanotubes were grown on silica fibres simply by using the ICVD method; the thickness of the grafted nanotube layer can be controlled from a few tens of micrometres to over half a millimetre by varying the growth time. SEM and TEM observation on fibre surfaces after the growth reaction indicated that the nanotubes were bonded to fibres through the catalyst etching into the fibre surfaces; the location of the catalyst implied a base growth mechanism. The average diameter of the nanotubes increased with prolonged growth time in both cases due to the catalyst ripening during the reaction; at the same time, the crystalline quality increased, as shown by both TGA and spatially-resolved Raman. The increase in crystallinity could be a size effect and/or due to the increasing packing density of the MWCNTs. The CNx–MWCNTs have a higher intrinsic packing density, associated with a slower growth rate and longer induction time. However, they appear more defective in both TGA and Raman, due to the disorder introduced by the nitrogen. This paper demonstrates an effective method for producing (both pure and N-doped) nanotube grafted silica fibres and provides a further understanding of the growth mechanism. CNx–MWCNTs may offer advantages in the context of hierarchical composites due to the greater packing density and potential for chemical interaction with a matrix. Any improved electrical conductivity associated with nitrogen doping may also be an advantage, for example, for lightening strike protection, although the high defect concentration may be limiting. Further investigations of mechanical and electrical properties of these nanotube grafted fibres and the CNT–fibre interaction are required to establish their potential in a new generation of composite materials.
285
Acknowledgements The authors would like to thank DSTL and QinetiQ for the financial support of this work. The authors also thank Mahmoud G. Ardakani and Richard J. Chater (Imperial College London) for their assistance on TEM and FIB.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.carbon.2009.09.029.
R E F E R E N C E
[1] Dresselhaus MS, Dresselhaus G, Avouris P. Carbon nanotubes: synthesis, structure, properties and applications. Berlin: Springer; 2001. p. 81–172. [2] Popov VN. Carbon nanotubes: properties and application. Mater Sci Eng R Rep 2004;43(3):61–102. [3] Coleman JN, Khan U, Blau WJ, Gun’ko YK. Small but strong: a review of the mechanical properties of carbon nanotube– polymer composites. Carbon 2006;44:1624–52. [4] Terrones M. Carbon nanotubes: synthesis and properties, electronic devices and other emerging applications. Int Mater Rev 2004;49(6):325–77. [5] Dai LM, Patil A, Gong XY, Guo ZX, Liu LQ, Liu Y, et al. Aligned nanotubes. ChemPhysChem 2003;4(11):1150–69. [6] Hata K, Futaba DN, Mizuno K, Namai T, Yumura M, Iijima S. Water-assisted highly efficient synthesis of impurity-free single-waited carbon nanotubes. Science 2004;306(5700):1362–4. [7] Andrews R, Jacques D, Rao AM, Derbyshire F, Qian D, Fan X, et al. Continuous production of aligned carbon nanotubes: a step closer to commercial realization. Chem Phys Lett 1999;303:467–74. [8] Singh C, Shaffer MSP, Windle AH. Production of controlled architectures of aligned carbon nanotubes by an injection chemical vapour deposition method. Carbon 2003;41:359–68. [9] Pinault M, Pichot V, Khodja H, Launois P, Reynaud C, MayneL’Hermite M. Evidence of sequential lift in growth of aligned multiwalled carbon nanotube multilayers. Nano Lett 2005;5:2394–8. [10] Mayne M, Grobert N, Terrones M, Kamalakaran R, Ruhle M, Kroto HW, et al. Pyrolytic production of aligned carbon nanotubes from homogeneously dispersed benzene-based aerosols. Chem Phys Lett 2001;338(2–3):101–7. [11] Koziol K, Shaffer M, Windle A. Three-dimensional internal order in multiwalled carbon nanotubes grown by chemical vapor deposition. Adv Mater 2005;17:760–3. [12] Liang EJ, Ding P, Zhang HR, Guo XY, Du ZL. Synthesis and correlation study on the morphology and Raman spectra of CNx nanotubes by thermal decomposition of ferrocene/ ethylenediamine. Diamond Relat Mater 2004;13:69–73. [13] Liu J, Czerw R, Carroll DL. Large-scale synthesis of highly aligned nitrogen doped carbon nanotubes by injection chemical vapor deposition methods. J Mater Res 2005;20(2):538–43. [14] Reyes-Reyes M, Grobert N, Kamalakaran R, Seeger T, Golberg D, Ruhle M, et al. Efficient encapsulation of gaseous nitrogen inside carbon nanotubes with bamboo-like structure using aerosol thermolysis. Chem Phys Lett 2004;396(1–3):167–73. [15] Koos AA, Dowling M, Jurkschat K, Crossley A, Grobert N. Effect of the experimental parameters on the structure of
286
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25] [26]
[27]
[28]
[29]
[30]
[31]
CARBON
4 8 ( 2 0 1 0 ) 2 7 7 –2 8 6
nitrogen-doped carbon nanotubes produced by aerosol chemical vapour deposition. Carbon 2009;47(1):30–7. He MS, Zhou S, Zhang J, Liu ZF, Robinson C. CVD growth of Ndoped carbon nanotubes on silicon substrates and its mechanism. J Phys Chem B 2005;109:9275–9. Ducati C, Koziol K, Friedrichs S, Yates TJV, Shaffer MS, Midgley PA, et al. Crystallographic order in multi-walled carbon nanotubes synthesized in the presence of nitrogen. Small 2006;2(6):774–84. Huang YH, Gao JP, Liu RZ. Structure and electronic properties of nitrogen-containing carbon nanotubes. Synth Met 2000;113(3):251–5. Terrones M, Jorio A, Endo M, Rao AM, Kim YA, Hayashi T, et al. New direction in nanotube science. Mater Today 2004;7:30–45. Shaffer M, Sandler J. Carbon nanotube/nanofibre polymer composites. In: Advani SG, editor. Processing and properties of nanocomposites. New Jersey, NJ [etc.]: World Scientific; 2007. p. 1–59. Thostenson ET, Li WZ, Wang DZ, Ren ZF, Chou TW. Carbon nanotubes/carbon fiber hybrid multiscale composites. J Appl Phys 2002;91:6034–7. Veedu VP, Cao A, Li X, Ma K, Soldano C, Kar S, et al. Multifunctional composites using reinforced laminae with carbon-nanotube forests. Nat Mater 2006;5:457–62. Qian H, Bismarck A, Greenhalgh ES, Kalinka G, Shaffer MSP. Hierarchical composites reinforced with carbon nanotube grafted fibers: the potential assessed at the single fiber level. Chem Mater 2008;20(5):1862–9. Garcia EJ, Wardle BL, John Hart A, Yamamoto N. Fabrication and multifunctional properties of a hybrid laminate with aligned carbon nanotubes grown in situ. Compos Sci Technol 2008;68:2034–41. Ci LJ, Zhao ZG, Bai JB. Direct growth of carbon nanotubes on the surface of ceramic fibers. Carbon 2005;43:883–6. Zhu S, Su CH, Lehoczky SL, Muntele I, Ila D. Carbon nanotube growth on carbon fibers. Diamond Relat Mater 2003;12:1825–8. Cao AY, Veedu VP, Li XS, Yao ZL, Ghasemi-Nejhad MN, Ajayan PM. Multifunctional brushes made from carbon nanotubes. Nat Mater 2005;4:540–5. Zhang Q, Qian WZ, Xiang R, Yang Z, Luo GH, Wang Y, et al. In situ growth of carbon nanotubes on inorganic fibers with different surface properties. Mater Chem Phys 2008;107(2– 3):317–21. Nakahara S. Recent development in a TEM specimen preparation technique using FIB for semiconductor devices. Surf Coat Technol 2003;169:721–7. Zhang M, Fang S, Zakhidov AA, Lee SB, Aliev AE, Williams CD, et al. Strong, transparent, multifunctional, carbon nanotube sheets. Science 2005;309:1215–9. Zhu S, Su CH, Cochrane JC, Lehoczky S, Cui Y, Burger A. Growth orientation of carbon nanotubes by thermal chemical vapor deposition. J Cryst Growth 2002;234:584–8.
[32] Cao AY, Ci LJ, Wu GW, Wei BQ, Xu CL, Liang J, et al. An effective way to lower catalyst content in well-aligned carbon nanotube films. Carbon 2001;39:152–5. [33] Rasband WS. ImageJ: US National Institutes of Health, Bethesda, Maryland, USA,