Carbon 39 (2001) 2163–2172

Tailoring the diameter of decorated C–N nanotubes bytemperature variations using HF-CVD

*Ralph Kurt , Christian Klinke, Jean-Marc Bonard, Klaus Kern, Ayatollah KarimiDepartment of Physics, Swiss Federal Institute of Technology Lausanne (EPFL), CH-1015 Lausanne, Switzerland

Received 21 November 2000; accepted 7 February 2001

Abstract

Patterned films of decorated nitrogenated carbon (C–N) nanotubes were catalytically synthesised by hot filament chemicalvapour deposition (HF-CVD) in a nitrogen–methane–ammonia environment. The systematic study of a transition betweendifferent kinds of C–N nanostructures as a function of the local substrate temperature ranging from 700 up to 8208C ispresented. The morphology, the diameter as well as the properties of the generated tubular structures showed strongdependence on this parameter. By means of electron microscopy a new type of decoration covering all tubular structures wasobserved. Buckled lattice fringes revealed the disordered graphitic-like character of the hollow C–N nanotubes. Ramanspectroscopy confirmed a transition in the microscopic order as a function of temperature. Furthermore field emission invacuum was studied and showed a spectacular correlation to the deposition temperature and therefore the diameter of theC–N tubes. For arrays of tubes thinner than 50 nm an onset field below 4 V/mm was observed. 2001 Elsevier ScienceLtd. All rights reserved.

Keywords: A. Carbon composites, Carbon nanotubes; B. Chemical vapor deposition; D. Field emission, Microstructure

1. Introduction solid films [13,14]. In that case, the strengthening effecthas been attributed to a cross-linked fullerene-like structure

The discovery of fullerene C by Kroto et al. in 1985 with curved and buckled basal planes [13].60

[1] has sparked a lot of exciting research in the field of Furthermore, doping of carbon nanotubes with nitrogencarbon based nanostructures. Numerous new structures like or boron is expected to induce exciting electronic prop-carbon nanotubes [2], nano-onions [3], nano-horns [4] or erties [15]. Miyamoto et al. [16] showed that tubular formsmicro-trees [5] were reported. Most of these materials hold of graphitic-like carbon nitrides can be stabilised bya variety of interesting properties including electronic [6,7] displacing the C atoms out of the honeycomb network ofand mechanical properties [8,9], which promise to be not graphite. Detailed transmission electron microscopyjust of academic interest but also to be sources for new (TEM) and electron energy loss spectroscopy (EELS) ofapplications. As an example single and multiwalled carbon carbon nitride nanotubulites prepared by chemical vapournanotubes have been used as cold field emitters (e.g., Ref. deposition (CVD) and magnetron sputtering, respectively,[10]) and as hydrogen storage materials [11]. was reported by Suenaga et al. [17,18]. Several other

On the other hand, carbon nitride (CN ) coatings and groups have also succeeded in synthesising nitrogen-con-x

diamond-like-carbon (DLC) films have been under intense taining nanotubes [19–23]. Formation of arrays of alignedinvestigation for a longer time because they can serve as CN nanofibers were reported by Terrones et al. [24]. Alsox

hard layers for tools (e.g., Ref. [12]). Interesting me- field emission properties of such nitrogen doped nanotubeschanical behaviour, in particular an exceptionally high were briefly characterised by Ma et al. [25]. Furthermore,elasticity, has also been observed for sputtered CN thin nanotubes within the boron nitride (BN) [23,26] andx

ternary B–C–N systems [22,27,28] have been reported.Apart from the structure, both the electronic and me-chanical properties are dependent on the chemistry of*Corresponding author. Tel.: 139-4027-448-91; fax: 139-hetero-atomic nanotubes (e.g., Ref. [29]). However, the4027-443-35.

E-mail address: ralph.kurt@philips.com (R. Kurt). correlation between the morphology, crystalline order and

0008-6223/01/$ – see front matter 2001 Elsevier Science Ltd. All rights reserved.PI I : S0008-6223( 01 )00034-3

2164 R. Kurt et al. / Carbon 39 (2001) 2163 –2172

properties of such tubular CN structures is not yet typical flow ratio of N –CH –NH of 100:1:1 wasx 2 4 3

completely understood. applied.Recently we described the autocatalytic growth of a new For all the experiments described in this work, only one

type of carbon structures [30] produced by hot filament resistively heated tungsten carbide (WC) filament ([5500chemical vapour deposition (HF-CVD). They consist of mm) was used in order to study the resulting non-uni-tubular nitrogenated carbon (C–N) nanostructures covered formity in deposition conditions described by the lateralwith graphitic-like sheets growing perpendicular to the distance, d, at the sample surface measured perpendiculartube axis. Furthermore, the essential role of nitrogen on to the filament. The inset in Fig. 1 shows schematically thestructure formation of decorated nanotubes was analysed in experimental setup. A molybdenum (Mo) substrate holderdetail by TEM and EELS [31]. At this stage various placed at r53 mm (fixed value for all further reporteddecorated nanotube-like structures were grown randomly results) underneath the filament was heated independentlydistributed at the surface of a pure Si substrate, and factors to 6508C. Positive and negative dc-bias potential up to 200that influence nucleation were not understood. Vapour- V were applied independently to the filament and to thephase deposition of CN nanotubes with controlled struc- substrate, respectively. The background pressure typicallyx

25ture and chemical composition can provide a tool for reached in the deposition chamber was 10 Pa, whereassynthesis of materials with tailored electronic [16] and the working pressure, p, was kept constant at 300 Pamechanical properties [32]. during deposition. A fixed filament temperature T 5fil

We present here a systematic study of the transition 21008C as measured by means of a two-colour pyrometerbetween different kinds of C–N nanostructures as a was used. The temperature profile at the sample surfacefunction of temperature (given by the lateral distance from (T versus d) as measured by a W-5%Re–W-26%Resub

the hot filament), which determines the tube diameter. thermocouple is given in Fig. 1. Due to the thermalStrong correlation to properties such as Raman signal and radiation a strong temperature gradient was observedfield emission were found. The fundament of these in- depending on the distance between filament and substratevestigations was the deposition of patterned films ofdecorated nanotubes, which were realised by applying acatalyst on the substrate using the microcontact (mCP)printing method.

2. Experimental methods

2.1. Synthesis of nanostructured material

Microcontact printing (mCP) of catalysts [33] is becom-ing a popular method because it allows us to formchemical patterns on the surface of a variety of substrates.Structured stamps were realised by curing the elastomerpoly(dimthyl)siloxane (PDMS) over a master. They werethen hydrophilised by an oxygen plasma treatment andstored under water to keep them hydrophilic. The stampswere dried under nitrogen flow for 10 s before use. Thecatalyst ink solution [Fe(NO ) ?9H O and Ni(NO ) ?3 3 2 3 2

6H O dissolved in ethanol] was applied to the stamps after2

a period of 12 h of ageing, which was found to be ideal forthe catalytic growth of nanotubes [34]. The printing wasdone by bringing the inked stamp into contact with thesurface of a Si wafer for 3 s.

Fig. 1. The measured temperature profiles (filled symbols) at theDecomposition of methane (CH ) using the bias-en-4 sample surface versus the lateral distance, d, from the filamenthanced hot filament chemical vapour deposition techniquewere fitted (calculated lines) assuming a Gaussian temperature

results in a new type of carbon based nanostructures. Pure distribution. Due to thermal radiation from the filament (T 5filnitrogen was used as a carrier gas and a small flow of 21008C) a stronger gradient is observed for smaller distances rammonia (NH ) was introduced during the experiments in between filament and substrate. The inset gives a schematic3

order to increase the amount of nitrogen in the deposits. drawing of the deposition configuration consisting of a resistively3The total gas flow was kept constant at 500 cm and a heated filament placed close to the substrate holder.

R. Kurt et al. / Carbon 39 (2001) 2163 –2172 2165

(r). The measured data were fitted assuming a Gaussian approximately 10 mm was achieved with a 1003 Olympusdistribution. microscope objective. The spectra were calibrated using a

natural diamond single crystal.2.2. Characterisation techniques Additionally, field emission measurements were per-

formed using the examined samples as cathodes with aScanning electron microscopy (SEM) was performed to lateral resolution of |1 mm. The emitted electrons were

analyse the microstructures in plan view. A Philips XL 30 collected on a highly polished stainless steel sphericalmicroscope equipped with a field emission gun (FEG) counterelectrode of 1 cm diameter, which corresponds to

2operating at an acceleration voltage between 2 and 5 kV, a an emission area of |0.007 cm . The distance was initiallyworking distance of typically 10 mm, and in secondary adjusted to 125 mm and was reduced to 50 or 25 mm whenelectron (SE) image mode was used. no emission was observed below 1000 V. A Keithley 237

The growth morphology of the tubular structures and source-measure unit was used for sourcing the voltage (uptheir crystallinity were controlled by analytical transmis- to 1000 V) and measuring the current with pA sensitivity,sion electron microscopy (TEM). For this purpose a allowing the characterisation of current–voltage (I–V)Hitachi HF-2000 field emission microscope equipped with behaviour.a Gatan image plate operating at 200 kV (point resolution0.23 nm) was used.

The micro Raman technique was used to obtain local 3. Resultsinformation about the vibrational properties of the nano-structures. The Raman spectra were recorded in backscat- 3.1. Morphology

1tering configuration using the 514.5 nm line of an Ar ionlaser and a DILOR XY 800 spectrometer. An incident We used mCP to study the specific effect of the catalystmaximal laser power of 20 mW was applied in order to on both the quality of the printed pattern and the tubeavoid peak shifts due to thermal heating during data growth. Fig. 2 shows SEM images of the substrate afteracquisition or structure transformations. A spot size of deposition using various catalyst mixtures. The structure

Fig. 2. Scanning electron microscopy images of patterned films as grown by HF-CVD at a lateral distance d55 mm within 30 min (r53mm). Growth rate and printing behaviour strongly depend on the type and concentration of the used catalyst.

2166 R. Kurt et al. / Carbon 39 (2001) 2163 –2172

consists of lines and dots with a typical structure width of distribution on the surface of a pure Si wafer. Clearly, a10 mm. Best printing was achieved with an Fe catalyst as well-controlled film growth requires the use of a catalyst.shown in Fig. 2a whereas the pure Ni solution showed The above mentioned temperature gradient T versussub

poor printing behaviour. The Ni ink produced small drops the lateral distance from the filament d (Fig. 1) stronglyon the stamp of |10 mm diameter, which were transferred affects the growth conditions for the nitrogenated carbonto the substrate as seen in Fig. 2c. In contrast, a higher tubes. The evolution of the morphology of various tubularefficiency regarding the growth of nanotubes was obtained structures grown onto the substrate as a function of localfor Ni than for pure Fe. The best results in terms of both temperature is represented in Fig. 3 by a series of SEMtube growth and patterns uniformity were achieved using images. For Fig. 3a–d the lateral distance d increases from

31an ethanolic ink containing 60 mMol Fe and 20 mMol 0 to 2, 5, and 10 mm, respectively. From d50 to 5 mm21Ni . This ink was used for the following experiments. (Fig. 3a–c) the structures become finer, i.e., the tube

Note that the same type of decorated tubes also grows diameter decreases from approximately 1 mm to 50 nm.auto-catalytically [30] with the disadvantage of a random Above d510 mm the local substrate temperature remains

Fig. 3. SEM plan view micrographs showing the evolution of shape and diameter of winded nanotubes as a function of the lateral distance d(r53 mm). The surface of each tubule prepared by HF-CVD is always covered with thin lamellas, resulting in a coral-like structure. For allimages the same magnification is used.

R. Kurt et al. / Carbon 39 (2001) 2163 –2172 2167

constant at T ¯7008C (Fig. 1). Simultaneously a signifi- observed data of the outer tube diameter as a function ofsub

cant change in the morphology of the grown tubes is the lateral distance, d. It confirms the strong dependenceobserved. As one can conclude from Fig. 3d nanotubes on d and therefore on the local substrate temperature. Thewith a strong dispersion in diameter (20–200 nm) were diameter was found to vary between 15 and 950 nm. Tubesformed at the same location, i.e., under the same con- with maximum diameters of approximately 1 mm wereditions. In contrast, for higher T (d,7 mm) a sharper observed directly underneath the filament. For d,5 mmsub

local distribution of tubular C–N species in terms of tube we found that as d increased, the tubes were thinner anddiameter was obtained (Fig. 3a–c). longer. Coiled tubes with a length up to 50 mm were

As a rule, all structures are of winded tubular character. observed with typical diameters of approximately 50 nm.It is interesting to note that a decoration is formed over the For d.7 mm the distribution of the tube diametersurface of all studied tubes, which can be clearly seen in becomes broader with increasing d. In Fig. 4 the thinnestFig. 3a. The large dendritic-like surface of the decorated tubes were marked by filled symbols whereas open sym-C–N tubes results from thin foils or lamellae growing bols were used for the thicker species. The positions whereperpendicular to the tube axis. The shape of these nano- Fig. 3a–d was taken are also marked in Fig. 4. The insetaggregates varies with the applied filament temperature, shows the tube diameter ( Ø ) versus the local substratetube

ranging from twisted lamellae to straight sheets showing temperature (T ). In the analysed region one can fit ansub

ramifications [35]. Recently we could show [31] that exponential curve of Ø 5 50 1 exp [S (T 2 7008C)]tube fit sub

coiling as well as decoration depends on the amount of nm to the obtained data, where S is a fit parameterfit

nitrogen incorporated in the tubes. Performing electron related to the slope of the curve. The data were fitted forenergy loss spectroscopy (EELS) a maximum N/C atomic values of tube diameter and local temperature above 50 nmratio of 0.04360.014 was obtained on samples produced and 7008C, respectively. S was determined to be 0.057/fit

under the same conditions but without catalyst [36]. 8C. Considering the nice fit of the calculated curve to theA quantitative analysis of the tube diameter ( Ø ) was experimental data, it is obvious that the nanotube growthtube

done using high-magnification SEM. Fig. 4 shows the depends exponentially on the local temperature. The higherthe deposition temperature the thicker the decorated C–Nnanotubes. The same behaviour was observed using othercatalyst concentrations and even for the autocatalyticgrowth.

TEM investigations were performed to study the micro-structure of decorated nanotubes in more detail. Fig. 5shows typical micrographs of such C–N nanotubes withdecreasing outer diameter obtained by increasing d. Thesame tendency as given in Fig. 4 was confirmed by theTEM analysis. Furthermore Fig. 5a–c reveals again struc-tures radiating from the surface of the nanotubes. Highresolution TEM images suggest that the needle-like struc-tures are in fact thin bent or rolled up graphene sheets.

In the case of thick nanotubes the inner structure isdifficult to study due to the decoration that reduces theelectron transparency. The contrast in Fig. 5d and e clearlyrepresents the tubular structure of very thin C–Nnanotubes and suggests that they are probably hollowinside. Tubes aligned parallel to the electron optical axis(Fig. 5e) indicate concentric wall structures and an innerdiameter in that case of less than 5 nm. The two highmagnification micrographs of individual tubes in plan view(Fig. 5d) and cross section (Fig. 5e) reveal the occurrenceof a crystalline structure. The lattice fringes visible in Fig.5d and e correspond to the h002j basal planes of graphite.

Fig. 4. The diameter of the tubular structures as obtained from Locally the planes are roughly parallel to each other but onSEM shows a strong dependence on the lateral distance d (r53

a larger scale they show bending and interlinking betweenmm). Maximum diameters were observed directly underneath theplanes. Such a buckling effect of the basal planes reflectingfilament. For d.5 mm a large dispersion of the tube diametersthe disorder even on an atomic scale has been alsowas found (open and filled symbols). The positions where Fig.observed in previously reported CN structures [14,17,18]3a–d was taken from are marked. The inset shows the tube x

diameter versus the local substrate temperature. but also in pure catalytically grown carbon nanotubes [37].

2168 R. Kurt et al. / Carbon 39 (2001) 2163 –2172

Fig. 5. TEM analysis of nitrogenated carbon nanotubes with different tube diameter and degree of coverage. The decoration consists of thingraphitic-like foils growing perpendicular to the tube axis. High resolution plan view (d) and cross section (e) reveals disorderedgraphitic-like structure typically characterised by buckling of the graphene planes due to the incorporation of nitrogen. The contrast indicateshollow tubes and a rough tube surface.

Note, that in the current work the term nanotube is used in nanotubes grown at different lateral distances, d, increasingan extended manner meaning hollow tubular structure. The from 0 to 20 mm as indicated.common understanding of concentrical cylinders does not All spectra contain at least two strong peaks confirmingapply in our case. the graphitic-like structure of the material. It is known that

pyrolythic graphite leads to a sharp vibration mode at 158821 23.2. Raman spectroscopy cm [38] due to the presence of C-sp domains (first

order G band). Structural modifications result in a shift ofThe influence of the local substrate temperature T this peak to lower wave numbers as seen in curves (a–c).sub

during deposition of the various types of C–N tubular A shift to higher wave numbers, i.e., higher activationstructures was investigated by Raman spectroscopy. The energy of the vibration mode, as seen in curve (e) might bebonding states in carbon based materials depend on the caused by internal stress in the material. A second strong

21nature of polymorphs, which is influenced by the deposi- peak at approximately 1360 cm is considered to repre-tion parameter. These effects are illustrated in Fig. 6, sent a more disordered structure labelled as D (disordered)which compares the micro-Raman spectra generated from band [39]. Decreasing particle size, bending of the lattice

R. Kurt et al. / Carbon 39 (2001) 2163 –2172 2169

graphite clusters in the C–N tubes, which decreases withincreasing d. According to Ref. [41] one can estimate thelength of the graphite cluster L from the I /I ratiographite D G

using the following equation: L 54.4 nm3(I /graphite D21I ) . In the case of our decorated C–N nanotubes,G

graphite cluster sizes of approximately 9 and 2 nm,respectively, were obtained for d50 and 20 mm.

The half width of the peaks varies only slightly in thepresented curves and reflects either the degree of crys-talline perfection in the case of sharp peaks or the moreamorphous-like character of the material (broader overlap-ping peaks as in curve (e)). An additional band located at

21approximately 1080 cm was detected at the tubularstructures deposited at small lateral distances but it couldnot yet be identified.

3.3. Field emission

Field emission is one of the most promising applicationsfor carbon-based films, and recent studies have stressed thesignificant influence of the film morphology on the emis-sion properties [10,42–45]. The ability to tailor the tube

Fig. 6. Comparison of Raman spectra obtained from decorated diameter by HF-CVD could thus prove to be a decisiveC–N nanotubes grown at different lateral distances d increasing

advantage to produce emitting structures of high quality.from 0 to 20 mm. Beside the graphite G peak and the disordered21 The effect of the distance to the filament, and hence ofgraphitic D peak a shoulder at 1614 cm and an unidentified

21 the tube diameter, is displayed in Fig. 7. The field emissionpeak at 1080 cm were detected. In dependence on the tubewas scanned at 20 locations on one continuous samplediameter as well the peak ratio I /I changes as peak shiftsG D

between d50 and 20 mm as indicated, and Fig. 7 showsregarding to the nominal vibration frequencies (vertical lines) wereobserved. clearly a strong variation in the field emission properties.

Directly underneath the filament (d51 mm), no fieldemission below 1100 V applied voltage (which corresponds

fringes as well as incorporation of N into the structure may to the limit of our acquisition set-up) was observed unlessactivate this band and cause furthermore the observed the interelectrode distance was decreased to 10 mm. Thepeakshifts. Note that in a perfect graphitic crystal this first onset field (applied field necessary to obtain a current

22order vibrational mode is forbidden due to the selection density of 10 nA cm ) was 36 V/mm, and the threshold22rules. current density of 10 mA cm was not obtained below 95

Analysing the G band in more detail, a shoulder located V/mm. As d increased, the field emission properties21at approximately 1614 cm becomes clearly visible improved rapidly (see Fig. 7), and reached a minimum for

(marked by an arrow in Fig. 6). This additional vibration d55 mm. The interelectrode distance was 125 mm at thatmode appears at high T , i.e., d between 0 and 4 mm and point, and the onset and threshold fields were 3.7 and 7.2sub

it is less pronounced below 7508C. Beside that, a signifi- V/mm, respectively. Beyond that distance, the emissioncant shift of the measured G peak position was observed as fields increased again, and levelled off for distances larger

21a function of d. The largest deviation D515 cm was than d58 mm. In that region, the interelectrode distancedetected at d55 mm (curve b) corresponding to decorated was 50 mm, and the onset and threshold fields amounted toC–N nanotubes with an outer diameter less than 50 nm. |10 and |20 V/mm, respectively.

It is worth studying the relative Raman peak area Field emission results are usually analysed with theintensities of the D and G band (I /I ). It was found that Fowlerc–Nordheim theory. The Fowler–Nordheim modelD G

the I /I peak ratio increases from |0.5 for spectrum (a) describes the electron emission from a flat surface under aD G

to |2.5 (spectrum e). This behaviour can be visibly high applied field by tunnelling through the triangularverified in Fig. 6 since the D band becomes dominant for surface potential barrier. The emitted current, I, is propor-

2 3 / 2large d. In line with the discussion above, the D band tional to I~F exp (Bf /F ), where F is the applied fieldarises from the finite size of the graphite domains and just above the emitting surface, f is the workfunction and

9 3 / 2 21originates from grain boundaries or imperfections, such as B is a constant (B56.83310 V eV m ) [46]. In the3substitutional N atoms, sp carbon, or other impurities case of sharp point emitters such as the ones considered

[40]. Therefore, the relative Raman peak area intensities of here, F is generally not known and is therefore usuallythe D and G band are a measure of the degree of order of taken as F 5 bE 5 bV/d , with V the applied voltage, d0 0

2170 R. Kurt et al. / Carbon 39 (2001) 2163 –2172

hypothesis that small flake-like aggregates are alreadyformed in the plasma close to the filament and are attachedto the outer surface of the tubes, where they can act asnuclei for a growth of the needle-like graphene sheets. Theobserved splitting of the Raman G peak at high T maysub

probably also originated by one more ordered and anotherdisordered graphitic-like structure corresponding to the twodifferent growth mechanisms.

In the temperature range from T 5700–8208C asub

continuous increase of the tube diameter (50–1000 nm)with temperature was observed. This strong correlationseems to be a behaviour of universal character due to itsindependence on the major deposition parameter such ascatalyst, kind and concentration of hydrocarbon gases, T ,fil

r, bias potential and working pressure (over large ranges).Pure carbon fibres have been grown for several years

from the decomposition of hydrocarbons (e.g., methane,acetylene, benzene, natural gas) at lower temperatures(e.g., |300–7008C) in the presence of a metallic catalyst(e.g., Fe, Ni, Co) [49]. The diameter of such carbonfilaments range from several hundred micrometers to about

Fig. 7. Current density, J, versus applied electric field, E, for 100 nm and decreases with increasing temperature.patterned films of C–N nanotubes. The onset E and thresholdi Sporadic works also showed synthesis of hollow tubes in

22fields E required to extract a current density of 10 nA cm andthr this temperature region [50]. Moreover thermal decompo-2210 mA cm , respectively, first decrease before increase again as sition of acetylene in an oven at temperatures above 7208C

a function of the lateral distance, d (as indicated). Lowest valuesresults in catalytic growth of nanotubes whose diameterE 53.7 V/mm and E 57.2 V/mm were observed at approxi-i thr increases with temperature [51]. However, carbon basedmately d55 mm. From a mathematical fit using the Fowler–nanotubes seem to exhibit minimum diameter in theNordheim model, the field amplification factor b of 750 wastemperature range from 720 to 7508C confirmed by thedetermined in that case.present work.

The dispersion of the tube diameters for d values abovethe interelectrode distance and E 5V/d the macroscopic 7 mm may be attributed to non-equilibrium growth con-0

applied field. The field amplification factor b is determined ditions in this region far from the filament. Due to thesolely by the geometrical shape of the emitter [47]. small deposition rate an effect of various nucleation times

In the case of Fig. 7, all emission curves followed the and an influence of a flux gradient should probably beFowler–Nordheim model at low emitted currents (up to considered here as well.

22|10 mA cm ), and the field amplification factor could The field emission properties (turn-on fields, fieldthus be easily determined by fitting the emission charac- amplification) were clearly correlated to the diameter of theteristics with the Fowler–Nordheim formula, using a tubular C–N structures. The minimal fields and highestworkfunction of f 55 eV [48]. The field amplification amplifications were found around d54–6 mm, i.e., wherefactor b follows a behaviour which is reciprocal to that of the lowest overall values for the tube diameter are obtainedthe emission fields: the field amplification is very low just (|50 nm). Field emission is a highly selective process, andbelow the filament (b 580 at d50 mm). It increases only structures showing the highest field amplificationrapidly with the distance and reaches a maximal value of (e.g., the smallest tube diameter) will participate in ab 5750 for the most efficient tubes, and decreases then significant proportion to the electron emission. The factagain to b |300 far away from the filament. that the emission is not as efficient for larger diameters

shows however that the field emission is not governedsolely by the diameter of the smallest tubes. In fact, if this

4. Discussion and conclusions were the case the emission fields should decrease furtherfor larger d, as the diameter of the smaller species

The fact that all our C–N nanotubes prepared by HF- decreases further to |30 nm far away from the filament.CVD are always decorated may probably be interpreted by One observes however on the SEM micrographs (e.g., Fig.at least two simultaneous growth mechanisms. The results 3d) that the smaller tubes are surrounded by thicker andallow one to suppose a tubular growth (along the tube axis) higher tubular structures. In such a case, the field amplifi-on the one hand and the formation of graphitic foils on the cation is not only determined by the diameter, but also byother hand. In the later case one can assume as a the distance to the neighbouring tubes [44,45,52]. The field

R. Kurt et al. / Carbon 39 (2001) 2163 –2172 2171

amplification on a film is therefore influenced both by the Referencesgeometrical shape of the emitter and by its surroundings,as the presence of nearby objects may screen the applied [1] Kroto HW, Heath JR, O’Brian SC, Curl RF, Smalley RE.

C : Buckminsterfullerene. Nature 1985;381(6042):162–3.electric field and provoke a decrease in the effective field 60

[2] Iijima S. Helical microtubules of graphitic carbon. Natureamplification.1991;354(6348):56–8.The C–N nanotubes obtained by HF-CVD compare

[3] Ugarte D. Curling and closure of graphitic networks underquite well with pure carbon nanotubes obtained by theelectron-beam irradiation. Nature 1992;359(6397):707–9.catalytic decomposition of acetylene in a tube flow reactor

[4] Iijima S, Yudasaka M, Yamada R, Bandow S, Suenaga K,at 7208C [45]. The latter samples were also structured by

Kokai F, Takahashi K. Nano-aggregates of single-walledmicrocontact printing using catalyst Fe:Ni inks of different graphitic carbon nano-horns. Chem Phys Lett 1999;309(3-densities with patterns identical to the ones used here. The 4):165–70.best nanotube sample had onset and threshold fields of 1.8 [5] Ajayan PM, Nugent JM, Siegel RW, Wei B, Kohler-Redlichand 3.3 V/mm, with a field amplification of b 51208. Ph. Growth of carbon micro-trees. Nature 2000;404:243.Carbon nanotube films with slightly inferior properties, [6] Odom TW, Huang JL, Kim P, Lieber CM. Atomic structure

and electronic properties of single-walled carbon nanotubes.e.g, a film prepared with an ink of lower density (resultingNature 1998;391(6662):62–4.in less dense nanotube patterns), showed properties nearly

[7] Wildoer JWG,Venema LC, Rinzler AG, Smalley RE, Dekkeridentical to the best C–N nanotubes observed here. Actual-C. Electronic structure of atomically resolved carbonly, one carbon nanotube film had onset and threshold fieldsnanotubes. Nature 1998;391(6662):59–62.of 3.6 and 7.0 V/mm for a field amplification of 850, to be

[8] Yu MF, Lourie O, Dyer MJ, Moloni K, Kelly TF, Ruoff RS.compared with 3.7 and 7.2 V/mm and b 5750. It is hence

Strength and breaking mechanism of multiwalled carbonvery probable that the factors governing the field emission nanotubes under tensile load. Science 2000;287(5453):637–are very similar for the pure carbon and C–N nanotubes, 40.and that the incorporated nitrogen plays very little, if any, [9] Salvetat JP, Kulik AJ, Bonard J-M, Andrew G, Briggs D,role in the field emission. ¨ ´ ´ ´Stockli T, Metenier K, Bonnamy S, Beguin F, Burnham NA,

Finally, we can also compare the patterned C–N tubes Forro L. Elastic modulus of ordered and disordered mul-tiwalled carbon nanotubes. Adv Mater 1999;11(2):161–5.with continuous films obtained by the same method [35].

¨ ´[10] Bonard JM, Salvetat JP, Stockli T, Forro L, Kern K,The best emission obtained on a continuous film wasˆChatelain A. Field emission from carbon nanotubes: perspec-characterised by onset and threshold fields of 3.8 and 7.8

tives for applications and clues to the emission mechanism.V/mm and a b 5675, i.e., slightly inferior to the structuredAppl Phys A 1999;69(3):245–54.sample at d55 mm. The continuous films show however

[11] Liu C, Fan YY, Liu M, Cong HT, Cheng HM, Dresselhauspoor homogeneity, and the structured deposition of the

MS. Hydrogen storage in single-walled carbon nanotubes atcatalyst allows a far better control over the emission room temperature. Science 1999;286:1127–9.properties of the sample. [12] Kaltofen R, Sebald T, Weise G, Low-energy ion bombard-

Due to the single filament configuration a strong tem- ment effects in reactive RF. magnetron sputtering of carbonperature gradient appears, which gives rise to the observed nitride films. Thin Solid Films 1997;308–309:118–25.variations in tube diameter. The most promising range for [13] Sjostrom H, Stafstrom S, Boman M, Sundgren JE. Superhard

and elastic carbon nitride thin films having fullerene-likeapplication regarding the field emission properties wasmicrostructure. Phys Rev Lett 1995;75(7):1136–9.found to be for d values between 4 and 6 mm corre-

[14] Hellgren N, Johansson MP, Broitman E, Hultman L,sponding to T 5720–7608C. However, once the optimalsubSundgren JE. Role of nitrogen in the formation of hard andconditions have been determined, homogeneous and con-elastic CN thin films by reactive magnetron sputtering. Physxtrolled film growth can be obtained by adjusting r and Tfil Rev B 1999;59(7):5162–9.

in a multiple parallel filament arrangement as described [15] Blase X, Charlier JC, De Vita A, Car R. Frustration effectsrecently [36]. and microscopic growth mechanisms for BN nanotubes.

Appl Phys Lett 1997;70(8):197–9.[16] Miyamoto Y, Cohen ML, Louie SG. Theoretical investiga-

tion of graphitic carbon nitride and possible tubule forms.Solid State Commun 1997;102(8):605–8.Acknowledgements

[17] Suenaga K, Yudasaka M, Colliex C, Iijima S. Radiallymodulated nitrogen distribution in CN nanotubular struc-xThe Swiss National Science Foundation (SNSF) istures prepared by CVD using Ni phthalocyanine. Chem Physacknowledged for the financial support. The electronLett 2000;316(5–6):365–72.´microscopy was performed at the Centre Interdepartmental

[18] Suenaga K, Johansson MP, Hellgren N, Broitman E, Wallen-de Microscopie Electronique of EPFL. The authors are berg LR, Colliex C, Sundgren JE, Hultman L. Carbon nitride

´grateful to Yann von Kaenel from Departement des nanotubulite–densely-packed and well aligned tubular nano-Materiaux of EPFL for technical assistance in Raman structures. Chem Phys Lett 1999;300(5–6):695–700.

´ ´ ´ ´spectroscopy and to Laszlo Forro from Department de [19] Stephan O, Ajayan PM, Colliex C, Redlich Ph, Lambert JM,Physique of EPFL for fruitful discussions. Bernier P, Lefin P. Doping graphitic and carbon nanotube

2172 R. Kurt et al. / Carbon 39 (2001) 2163 –2172

structures with boron and nitrogen. Science carbon films – morphology and field emission. Diamond1994;266(5191):1683–5. Relat Mater (in press).

[20] Yudasaka M, Kikuchi R, Ohki Y, Yoshimura S. Nitrogen- [36] Kurt R, Bonard JM, Karimi A. Morphology and fieldcontaining carbon nanotube growth from Ni phthalocyanine emission properties of nano-structured nirogenated carbonby chemical vapor deposition. Carbon 1997;35(2):195–201. films produced by plasma enhanced hot filament CVD.

[21] Terrones M, Grobert N, Zhang JP, Terrones H, Olivares J, Carbon (in press).Hsu WK, Hare JP, Cheetham AK, Kroto HW, Walton DRM. [37] Hernadi K, Piedigrosso P, Fonseca A, Nagy JB, Kiricsi I.Preparation of aligned carbon nanotubes catalysed by laser- Carbon nanotube formation over supported catalysts. Moletched cobalt thin films. Chem Phys Lett 1998;285(5- Cryst Liq Cryst 1998;310:179–84.6):299–305. [38] Tuinstra F, Koenig JL. Raman Spectrum of Graphite. J Chem

[22] Satishkumar BC, Govindaraj A, Sen R, Rao CNR. Single- Phys 1970;53(3):1126–30.walled nanotubes by the pyrolysis of acetylene–organometal- [39] Dillon RO, Woollam JA, Katkanant V. Use of Ramanlic mixtures. Chem Phys Lett 1998;293(1–2):47–52. scattering to investigate disorder and crystallite formation in

[23] Goldberg D, Bando Y, Bourgeois L, Kurashima K, Sato T. as-deposited and annealed carbon films. Phys Rev BLarge-scale synthesis and HRTEM analysis of single-walled 1984;29(6):3482–9.B- and N-doped nanotube bundles. Carbon 2000;38:2017– [40] Kaufman JH, Metin S, Saperstein DD. Symmetry breaking in27. nitrogen-doped amorphous carbon: infrared observation of

[24] Terrones M, Redlich P, Grobert N, Trasobares S, Hsu W, the Raman-active G and D bands. Phys Rev BTerrones H, Zhu Y, Hare J, Reeves C, Cheetham A, Ruehle 1989;39:13053–60.M, Kroto H, Walton D. Carbon nitride nanocomposites: [41] Knight DS, White WB. Characterization of diamond films byformation of aligned C N nanofibers. Adv Mater Raman spectroscopy. J Mater Res 1989;4(2):385–93.x y

1999;11(8):655–8. [42] Davidov DN, Sattari PA, AlMawlawi D, Osika A, Haslett[25] Ma X, Wang E, Zhou W, Jefferson D, Chen J, Deng S, Xu N, TL, Moskovits M. Field emitters based on porous aluminum

Yuan J. Polymerized carbon nanobells and their field-emis- oxide templates. J Appl Phys 1999;86:3983–7.¨sion properties. Appl Phys Lett 1999;75(20):3105–7. [43] Bonard JM, Kind H, Stockli T, Nilsson LO. Field emission

[26] Loiseau A, Willaime F, Demoncy N, Shramchenko N, Hug from carbon nanotubes: the first 5 years. Sol State Elec (inG, Colliex C, Pascard H. Boron nitride nanotubes. Carbon press).

¨ ¨1998;36(5–6):743–52. [44] Nilsson L, Groning O, Emmenegger C, Kuttel O, Schaller E,[27] Redlich Ph, Loeffler J, Ajayan PM, Bill J, Aldinger F, Ruhle Schlapbach L, Kind H, Bonard JM, Kern K. Scanning field

M. B–C–N nanotubes and boron doping of carbon emission from structured carbon nanotube films. Appl Physnanotubes. Chem Phys Lett 1996;260(3–4):465–70. Lett 2000;76:2071–3.

¨ ´[28] Terrones M, Benito AM, Manteca-Diego C, Hsu WK, [45] Bonard JM, Weiss N, Kind H, Stockli T, Forro L, Kern K,Osman OI, Hare JP, Reid DG, Terrones H, Cheetham AK, Chatelain A. Tuning the field emission properties of pat-Prassides K, Kroto HW, Walton DRM. Metal particle cata- terned carbon nanotube films. Adv Mater 2001;13(3):184–8.lysed production of nanoscale BN structures. Chem Phys [46] Gadzuk JW, Plummer EW. Field emission energy distributionLett 1996;257(5–6):573–6. (FEED). Rev Mod Phys 1973;45(3):487–548.

[29] Hernandez E, Goze C, Bernier P, Rubio A. Elastic properties [47] Dyke WP, Dolan WW. Field emission. In: Marton L, editor,of C and B C M composite nanotubes. Phys Rev Lett Advances in electronics and electron physics, vol. 8, Newx y z

1998;80(20):4502–5. York: Academic, 1956, pp. 89–185.[30] Kurt R, Karimi A, Hoffmann V. Growth of decorated carbon [48] Dean KA, Groening O, Kuttel OM, Schlapbach L. Nanotube

nanotubes. Chem Phys Lett (in press). electronic states observed with thermal field emission elec-[31] Kurt R, Karimi A. Influence of nitrogen on growth mecha- tron spectroscopy. Appl Phys Lett 1999;75(18):2773–5.

nism of decorated C–N nanotubes. Chem Phys Chem (in [49] Dresselhaus MS, Dresselhaus G, Eklund PC. Science ofpress). fullerenes and carbon nanotubes, San Diego: Academic,

[32] Lowther JE. Defective and amorphous structure of carbon 1996.nitride. Phys Rev B 1998;57(10):5724–7. [50] Ivanov V, Fonseca A, Nagy JB, Lucas A, Lambin P,

[33] Kind H, Bonard JM, Emmenegger C, Nilsson LO, Hernadi Bernaerts D, Zhang XB. Catalytic production and purifica-K, Maillard-Schaller E, Schlapbach L, Forro L, Kern K. tion of nanotubules having fullerene-scale diameters. CarbonPatterned films of nanotubes using microcontact printing of 1995;33:1727–35.catalysts. Adv Mater 1999;11(15):1285–9. [51] Klinke C, Bonard JM, Kurt R, Kern K. Temperature

[34] Kind H, Bonard JM, Forro L, Kern K, Hernadi K, Nilson dependence of the growth of patterned nanotube films (inLO, Schlapbach L. Printing gel-like catalysts for the directed preparation).

¨ ¨ ¨growth of multiwall carbon nanotubes. Langmuir [52] Groning O, Kuttel OM, Emmenegger C, Groning P, Schlap-2000;16:6877–83. bach L. Field emission properties of carbon nanotubes. J Vac

[35] Kurt R, Bonard JM, Karimi A. Nano-structured nitrogenated Sci Technol B 2000;18(2):665–78.