buck minster fuller en discovery and consequences

8/2/2019 Buck Minster Fuller En Discovery and Consequences

1/23

Review

Synthesis of aligned carbon nanotubes

Choon-Ming Seah a, Siang-Piao Chai b, Abdul Rahman Mohamed a,*

a School of Chemical Engineering, Engineering Campus, Universiti Sains Malaysia, Seri Ampangan, 14300 Nibong Tebal, S.P.S. Pulau Pinang,

Malaysiab School of Engineering, Monash University, Jalan Lagoon Selatan, 46150 Bandar Sunway, Selangor, Malaysia

A R T I C L E I N F O

Article history:

Received 27 December 2010

Accepted 25 June 2011

Available online 30 June 2011

A B S T R A C T

Vertically aligned carbon nanotubes (ACNTs) are bundles of carbon nanotubes oriented per-

pendicular to a substrate, and horizontally aligned CNTs are parallel to the substrate. Their

dense and orderly arrangement, along with outstanding physical and chemical properties,

enables ACNTs to be used in various fields. The methods of synthesising ACNTs can be

classified into single-step and double-step techniques. Thermal pyrolysis and flame syn-

thesis are the common single-step methods, and both are relatively simple. The double-

step methods, including catalyst coating and chemical vapour deposition, provide more

control over the catalyst morphology. This review explores different methods used for -

ACNT growth, the process parameters that determine the morphology of ACNTs and the

applications of structured ACNTs.

2011 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4614

2. Single-step methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4615

2.1. Thermal pyrolysis method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4615

2.1.1. Parameters determining the properties of ACNTs synthesised by thermal pyrolysis . . . . . . . . . . . . . . 4616

2.2. Flame synthesis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4617

3. Double-step methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4619

3.1. Physical vapour deposition (PVD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4619

3.1.1. Pretreatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46203.2. Solution-based catalyst precursors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4621

3.3. Catalyst . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4623

3.4. Chemical vapour deposition (CVD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4624

3.5. Supergrowth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4625

4. Substrates and buffer layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4625

5. Alignment of CNTs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4627

6. Horizontally aligned carbon nanotubes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4627

7. Towards mass production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4628

0008-6223/$ - see front matter

2011 Elsevier Ltd. All rights reserved.doi:10.1016/j.carbon.2011.06.090

* Corresponding author: Fax: +60 594 1013.E-mail address: [email protected] (A.R. Mohamed).

C A R B O N 4 9 ( 2 0 1 1 ) 4 6 1 34 6 3 5

a v a i l a b l e a t w w w . s c i e n c e d i r e c t . c o m

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c a r b o n
http://dx.doi.org/10.1016/j.carbon.2011.06.090mailto:[email protected]://dx.doi.org/10.1016/j.carbon.2011.06.090http://dx.doi.org/10.1016/j.carbon.2011.06.090http://dx.doi.org/10.1016/j.carbon.2011.06.090http://www.sciencedirect.com/http://www.elsevier.com/locate/carbonhttp://www.elsevier.com/locate/carbonhttp://www.sciencedirect.com/http://dx.doi.org/10.1016/j.carbon.2011.06.090http://dx.doi.org/10.1016/j.carbon.2011.06.090http://dx.doi.org/10.1016/j.carbon.2011.06.090mailto:[email protected]://dx.doi.org/10.1016/j.carbon.2011.06.090


2/23

8. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4629

Acknowledgements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4629

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4629

1. Introduction

The discovery of buckminsterfullerene (C60) in 1985 by Krotoet al. [1] has led to an entirely new branch of carbon chemis-

try. In the early 1990s, another significant breakthrough was

achieved. Carbon filaments with diameters in the nanometre

range were observed by Iijima using transmission electron

microscopy (TEM) [2]. These carbon filaments were called

carbon nanotubes (CNTs). Two years later, single-walled car-

bon nanotubes (SWCNTs) were synthesised by Iijima and

Ichihashi [3] and Bethune et al. [4]. The research on CNTs

subsequently began in earnest worldwide. Aligned carbon

nanotubes (ACNTs) were first reported by Thess et al. [5],

who were able to bundle 70% of the volume of nanotubes into

crystalline ropes in 1996. In the same year, the Chinese Acad-

emy of Science reported that a 50 lm-thick film of highlyaligned nanotubes had successfully been grown by chemical

vapour deposition (CVD) [6]. Ren and Huang[7] first used Plas-

ma-Enhanced Hot Filament CVD (PE-HF-CVD) to lower the

growth temperature to below 666 C and used electric field

as an external force to provoke the alignment. Meanwhile,

Fan et al. [8] introduced position controlled growth of VACNT

on porous and plain silicon substrate. They also reported the

detailed growth and mechanism of alignment of ACNTs in

their work.

Vertically aligned CNTs (Fig. 1) are quasi-dimensional car-

bon cylinders that align perpendicular to a substrate [9]. Ver-

tically aligned with high aspect ratios [10] and uniform tube

length made it easy spinning into macroscopic fibres [11].The arrays of ACNT arrays are typically grown from a catalyst

that is pinned to a substrate, which produces long, high-pur-

ity nanotubes with sidewalls that are free of catalysts [12]. Be-

cause of these properties, arrays of ACNT are widely used in

nanoelectronics or in composite materials as reinforcing

agents. Large arrays of ACNTs with a high degree of unifor-

mity in terms of tip radius and height provide excellent field

emission properties [13,14]. Furthermore, vertically aligned

CNTs also exhibit a high capability to produce high current

densities under low operating voltages [15]. CNTs are formed

from a sheet of graphene, which possesses a strongly aniso-

tropic structure. Chiral nanotubes could be envisioned in

which the current-carrying state may have an angularmomentum about the tube axis, making them appropriate

for flat panel displays. ACNTs can also be reinforced with

other matrix composites to form anisotropic conductive

materials [16]. ACNTs synthesised on substrates with a pat-

terned trench structure enables them to be applied in ad-

vanced triode-type field emitters [17]. ACNTs have a very

large surface area and a high thermal conductivity, both of

which facilitate rapid heat transfer to the surrounding, mak-

ing them important materials in the construction of solar

cells [18]. ACNTs have also been used in hydrogen storage,

as the interior and interstitial surfaces of open-ended CNTs

have a strong binding energy for adsorbing hydrogen gas mol-

ecules compared with planar carbon surfaces [19,20]. Alignedmulti-walled carbon nanotubes (MWCNTs) were found to

possess a higher adsorption rate of hydrogen than non-

aligned CNTs because of the large inter-nanotube space in be-

tween the parallel nanotubes [20]. The subnanometre pores of

ACNTs are suitable for separation of gases and other small

molecules, such as hydrogen and water.

ACNTs can be potentially used in self-cleaning applica-

tions because of their surface property to be hydrophobic

[21]. Furthermore, ACNTs, possessing larger surface area

and higher electrical conductivity over entangled CNTs, are

ideal electrode material for DNA biosensor [22], sensors for

glucose [23], pH [24] as well as NO2 [25]. A research group from

Tsinghua University [26] had grown super-aligned CNT arrayswhich are greater in nucleation density, lower CNT diameter

distribution and better alignment compared to ordinary ACNT

arrays [26]. These super-aligned CNTs were successfully spun

into continuous yarns with excellent mechanical and electri-

cal properties [2628], which can be further developed to a

touch panel [29], liquid crystal display [30] and transparent

loudspeaker [31]. The property study shows that the yarns

possessed mechanical strength greater that 460 MPa. The

flexibility and strength retained even though the yarns were

exposed to very high or very low temperatures [32]. Instead

of yarns, a transparent CNTs sheet also can be drawn in par-

allel from ACNTS arrays and can be employed to make organ-

ic light emitting diode [33]. Furthermore, it was shown thatACNTs used in a complementary metal-oxide semiconductor

(CMOS) integrated circuit may overcome the problem of large

device-to-device variation when normal CNTs are used [34].

The ACNT array with super-compressible foam-like behav-

ior and quick recovery properties are suitable for use as en-

ergy absorbing coating [35]. In addition, the outstanding

mechanical properties along with the adhesive strength pos-

sessed by the entangle structures at the top of arrays can be

developed to dry adhesive which could withstand up to

100 N cm1 shear force. The normal adhesion force is low

and that makes them easy for lifting off [36].

Since the first ACNT array was reported in 1996 [6], numer-

ous papers describing the growth of ACNTs have beenFig. 1 A scanning electron microscopy (SEM) image

showing ACNTs [19].

4614 C A R B O N 4 9 ( 2 0 1 1 ) 4 6 1 34 6 3 5
http://www.elsevier.com/locate/carbonhttp://www.elsevier.com/locate/carbonhttp://www.elsevier.com/locate/carbonhttp://www.elsevier.com/locate/carbonhttp://www.elsevier.com/locate/carbonhttp://www.elsevier.com/locate/carbonhttp://www.elsevier.com/locate/carbonhttp://www.elsevier.com/locate/carbon


3/23

published. Over more than a decade, various modifications

have been made and new techniques have been discovered

that make it possible to grow high-quality ACNTs with scal-

able production. Large CNT arrays have successfully been

grown on different substrates, such as mesoporous silica

[37], planar silicon substrates [38] and quartz glass plate [16]

by using transition metals such as Fe, Co, or Ni as the catalyst.

Because transition metals have non-filled d shells, it enables

them to adsorb and interact with hydrocarbons. Of all of the

synthesis methods, CVD has been recognised as the most

promising method for producing ACNT arrays. Various types

of CVD have been developed, such as thermal CVD (T-CVD),

plasma-enhanced CVD (PE-CVD) and floating catalyst CVD

(FC-CVD).

2. Single-step methods

2.1. Thermal pyrolysis method

Thermal pyrolysis is also known as metalorganic CVD or FC-

CVD [39]. It is one of the most popular methods for synthes-

ising dense and aligned forms of CNTs. FC-CVD involves the

pyrolysis of organometallic precursors such as ferrocene

[13,38,4053], iron (II) phthalocyanine (FePc) [5458], iron

pentacarbonyl [59], nickelocene and cobaltocene [60] to nucle-

ate the growth of nanotubes. Non-carbonaceous compounds,

such as FeCl3, are also reported to be promising catalyst pre-

cursors for growing ACNTs [61]. In most cases, a carbon

source must be added in excess to increase the carbon-

to-catalyst ratio and prevent high levels of metal impurity

in the CNTs [52]. Aromatic hydrocarbons such as xylene

[38,4043], toluene [44], benzene [45,46] and naphthalene

[47] are often used along with ferrocene because of their

chemical structure similarities [47] and the fact that most of

the aromatic hydrocarbons can dissolve ferrocene easily.

However, the use of heavy hydrocarbons such as aromatic

hydrocarbons and cyclohexane is not suitable because heavy

hydrocarbons will deposit on the reactor wall in a low-tem-

perature zone [53]. As a result, lighter hydrocarbons such as

acetonitrile [48], ethylene [49,53], acetylene [50] and alkanes

[51] are commonly used. In addition, tree products such as

turpentine oil [13] and camphor [52] are also used as carbon

sources for synthesising ACNTs.

After many years of research, FC-CVD has been modified

with the goal of growing ACNTs of better quality and align-

ment. In general, there are two major types of reactors used

for FC-CVD, namely double-furnace and single-furnace

reactors. In the double-furnace setup (Fig. 2), the first furnace

is responsible for the vaporisation and sublimation of the cat-

alyst precursor, while the second furnace is kept at a high

temperature for the catalysts to assemble and nucleate the

growth of CNTs. One of the drawbacks of this process is the

steep temperature gradient that exists between the two

furnaces, which makes it difficult to maintain the same evap-

oration rate throughout the entire process. As for the conven-

tional single-furnace setup, only one high-temperature

furnace is required. The mixture of the catalyst and carbon

feedstock (liquid phase) is first evaporated using a heater be-

fore it is introduced into the reactor. The problem with this

approach lies in controlling the uniformity of the catalyst par-

ticles inside the reactor. Aerosol-assisted carbon deposition

has been developed to overcome this shortcoming. Jeong

et al. [40] used an ultrasonic evaporator to atomise a mixture

of ferrocene and xylene. The mixture was then carried into a

single-furnace reactor by a carrier gas. Spray pyrolysis with

the use of a spray nozzle to atomise the mixture supply com-

ing into the reactor has also been reported [13,46]. Both meth-

ods can continuously generate quantitatively controlled

aerosols in large amounts, ensuring that the carbon source

and the metal particles are distributed evenly in the reactor.

The growth of CNTs from FC-CVD has been suggested to

occur in two different ways. First, the active metals must de-

posit on the substrate before the growth of CNTs can take

place. Chen and Yu [55] supported this point with the obser-

vation of the existence of metal at both ends of the tube, sug-

gesting the co-existence of the tip-growth and base-growth

mechanisms. Li et al. [56] found that CNTs could be grown

from an iron film deposited on substrate that contained dif-

ferent sizes of iron nanoparticles. The larger iron particles

were responsible for producing the carbon atomistic species,

which were required for subsequent growth of the CNTs.

Meanwhile, the smaller iron particles were more catalytically

active because of their higher surface energy. The graphite

layers formed could encapsulate the iron particles of both

sizes and form a concentric graphitic shell or a semi-spherical

graphitic shell. The continuous generation of carbon atoms

by the larger iron particles increased the length of the graph-

ite layers by forcing the large iron particles up while the small

iron particles remained on the substrate, which is why both

ends of the tube contained iron particles. However, the study

of Huang et al. [62] found that the growth of nanotubes

started on the active catalyst at the floating stage. These

authors provided more convincing proof than did the previ-

ously mentioned authors. In their study, four substrates were

High temperature furnaceLow temperature furnace

Carrier gas/ vector gas

Organometallic and hydrocarbon

Fig. 2 Scheme showing the double-furnace setup used in the organometallic/hydrocarbon co-pyrolysis process.

C A R B O N 4 9 ( 2 0 1 1 ) 4 6 1 34 6 3 5 4615


4/23

placed at different locations in the reactor with temperatures

ranging from 200 C to 900 C. Nanotubes were found on the

substrate in the temperature range of 200 C to 300 C. It is

known that this range of temperature is too low for growing

nanotubes. Furthermore, nanotubes were found on Au-coated

wafers, but Au is not a good substrate for growing CNTs. Be-

sides, the U-shaped nanotubes (Fig. 3) are believed to grow

in the floating phase and then become hooked on the sub-

strate, forming a U shape as a result of the flowing gas.

The thermal pyrolysis method is attracting considerable

attention for the synthesis of ACNTs because of its exception-

ally low cost and ease of scaling up to mass production

[47,49,62]. The ACNTs can be grown on flat substrates and

also on cylindrical substrates such as the wall of a tube reac-

tor. Moreover, ACNTs are also reported to grow on spherical

substrates consisting of 50% SiO2, 30% Al2O3, and 20% ZrO2with a diameter of 700 lm (Fig. 4) [53].

2.1.1. Parameters determining the properties of ACNTs

synthesised by thermal pyrolysis

The density and alignment of the CNTs are intimately related

to various process parameters. The parameters that are

widely studied by researchers in order to produce

high-quality, well-aligned CNTs are the growth temperature,

the catalyst concentration, the feed rate and the period of

growth [45,63,64].

The growth temperature is the most crucial parameter for

determining the properties of CNTs. From our review, the

common temperatures for the growth of ACNTs are in the

range of 700950 C. At a relative low temperature, alignment

could not be obtained because of the incomplete dissociation

of catalysts precursors and hydrocarbon species [65]. The coa-

lescence kinetics between the catalyst and the hydrocarbon

species are not sufficient to guarantee the alignment and

crystallinity of the nanotubes [13]. Carbon nanofibres (CNFs)

consisting of defective graphitic qualities such as herringbone

morphology are always found at low reaction temperatures.

At a relatively high temperature, the catalyst can lose its cat-

alytic activity. In addition, the dissolution rate of carbon

atoms in the catalyst is higher than that of the diffusion

and precipitation rates at high reaction temperature, which

causes carbon atoms to accumulate on the surface of cata-

lysts, forming multi-shelled carbon nanocapsules (MS-CNCs)

[42].

The diameter of the ACNTs is strongly correlated to the

diameter of the catalyst cluster [64]. At higher temperatures,

the mobility of the catalyst on the substrate is relatively high,

promoting the coalescence and formation of larger catalyst

nanoparticles. Hence, a higher temperature encourages the

growth of CNTs of larger diameter [44]. However, other

authors have reported that the average diameter of CNTs in-

creases with temperature in a lower temperature range and

decreases with temperature in a higher temperature range,

which is common in the continuous catalyst feeding process.

Normally, the size and mobility of the catalyst particles in-

crease with increasing temperature. However, when the CNTs

start to grow, the floating catalyst will have a more difficult

time reaching the substrate because it becomes blocked by

the growth of CNT arrays [63]. The catalyst particles intro-

duced after this process has occurred will deposit on the tube

wall and nucleate the formation of smaller diameter nano-

tubes, resulting in a bimodal diameter distribution. However,

the bimodal diameter distribution does not exist in the low-

temperature range [44]. There is another possible explanation

for this phenomenon, which contrasts with the previous one.

In the high-temperature range, the energy supply is high,

which enhances the full decomposition of the catalyst precur-

sor to form fine catalyst particles that grow CNTs of smaller

diameter [66,67].

The CNT growth rate is also linearly proportional to the

temperature, a feature caused by the increased mobility of

the floating carbon species. In addition, ACNTs become

straighter with increasing temperature. Lee et al. [68] studied

the quality of ACNTs synthesised at different temperatures

using Raman spectroscopy and high-resolution TEM. They

found that the degree of crystalline perfection increased line-

arly with temperature. It is speculated that high temperatures

will increase the diffusion rate of carbon and create graphitic

sheets with fewer defects. However, after the optimum tem-

perature is reached, the growth rate of CNTs decreases with

increasing temperature. As a result, the carbon deposits, cov-

ering the entire catalyst surface and deactivating the catalyst.

The size of the catalyst nanoparticles is the most influen-

tial factor for controlling the diameter of the CNTs produced

Fig. 3 U-shaped CNTs (the arrow indicates the flow

direction) [62].

Fig. 4 Vertically aligned CNTs grown on a spherical

substrate approximately 700 lm in diameter [53].

4616 C A R B O N 4 9 ( 2 0 1 1 ) 4 6 1 34 6 3 5


5/23

[64,69]. The diameter and the yield of CNTs are always

strongly correlated with the catalyst concentration [40,63]. A

high catalyst/hydrocarbon ratio will boost the agglomeration

of the catalyst particles, resulting in the formation of larger

catalyst particles. When the catalyst/hydrocarbon ratio is

too high, adverse effects will be observed. The catalyst parti-

cles agglomerate at a high rate until they becomes too large

and are no longer active for growing CNTs (low surface-to-vol-

ume ratio of the catalyst nanoparticles). Nanolumps and

branched CNTs are favoured in this condition [66]. The CNTs

found in this condition are generally grown on the smaller-

sized catalyst particles that escaped from the rapid coales-

cence [45]. Furthermore, carbon deficiency inhibits the growth

of new tube walls [70]. If the reaction temperature is high, a

bimodal diameter distribution will also appear. A higher cat-

alyst/hydrocarbon ratio results in the metal catalyst sticking

either on the wall or inside the nanotubes, which affects

the structural perfection of the tube wall. Longer and thinner

tubes are reported in a low Fe/C environment because of the

abundance of carbon and additional active catalyst [71]. The

smaller catalyst particles are much more active and last long-

er. At a very low introduction rate of catalyst precursor, the

yield is less and it is impossible for alignment to occur. Con-

versely, at a high introduction rate, the catalyst particles coa-

lesce together and ultimately yield MS-CNCs [46], or the

catalyst particles may be retained inside the tube and pro-

mote the growth rate and length of the CNTs [41].

Tapaszto et al. [63] used a spray nozzle to carry the feed

(low active solutiontovector gas ratio) at very high flow rate

into the reactor. This approach contributes to the formation

of ACNTs with a narrower distribution of diameters. The

droplets of the liquid reactant involved are small if the liquid

reactant is sprayed at a very high rate, producing longer tubes

with a smaller diameter. The droplets decompose more easily

and promote the growth of ACNTs. In fact, a low retention

time of the feed causes the decomposition of hydrocarbon

to be less effective, and pyrolytic coating occurs, which might

cause the reactor wall to be covered with a sticky layer of con-

tamination if decomposition of heavy hydrocarbons is

involved.

A very high vector gastocarbon source ratio in the reac-

tor will dilute the concentration of the catalyst precursor

and carbon sources, producing a low yield of CNTs with poor

alignment [72]. Meanwhile, Li et al. [56] and Huang et al.[73]

used a low vector flowrate (


6/23

rod-like probes [7579], grids [8083] or plates [8486]. The use

of metalorganics such as iron pentacarbonyl [87,88], ferro-

cene [89] or metal nitrate dissolved in fuel [90,91] has been re-

ported. So far, ACNTs have only been reported to grow on

alloy substrates.

Flame synthesis is a very energy-efficient process because

the fuel itself is a source of heat and carbon. The temperature

achieved can be as high as 1600 K, which is hard to achieve

with CVD in a conventional furnace. For ACNT synthesis over

a large area, it is more economical to use flame rastering and

multiple flames to achieve a controllable residence time and

the desired flame region [84,92]. The production efficiency

and yield per energy input is lower than that of CVD, making

this method suitable for industrial production [82]. Addition-

ally, flame synthesis is a simple one-step method that with-

out substrate preparation. However, complex substrate

preparation has also been reported for flame synthesis

[93,94]. The growth mechanism of ACNTs in the flame synthe-

sis of hydrocarbon can be divided into three main steps

[75,82,93,94]. First, the hydrocarbon fuel is pyrolysed in the

preheated zone to form hydrocarbon species, which will be

the carbon source for the CNTs, while the metal particles

form on the surface of the alloy. The hydrocarbon species dif-

fuses onto the catalyst at an appropriate temperature and is

then absorbed by the catalyst to nucleate the growth of CNTs.

The choice of the catalyst is the main consideration for the

production of the ACNTs. An alloy is always selected because

of its lower melting point and higher solubility for carbon

compared with pure metal. From our review, a majority of

the ACNTs are reported to be grown on alloys containing Ni,

along with Fe and Cr [77,78,81,84]. It is believed that the for-

mation of nickel oxide contributes to the growth of CNTs

[62]. Nevertheless, the alignment still depends on the density

of the metal nanoparticles, which serve as catalysts. Pan et al.

[84] studied different kind of alloys using an ethanol flame

and the authors found that pure Ni and pure Fe only produced

CNTs and CNFs, respectively. They proposed a hollow-core

mechanism for Ni. The diffusivity of carbon at the exterior

surface of Ni is more rapid than in the interior of Ni nanopar-

ticles, resulting in the growth of CNTs with a hollow core in

the ethanol flame synthesis. In contrast, a solid-core mech-

anism applies to Fe because carbon can easily diffuse

through the Fe particle to form CNFs. Arana et al. [75] also

found the same outcome. The solubility of carbon in Ni parti-

cles was higher compared with Fe particles, so carbon would

precipitate more rapidly in Ni. Hence, long, dense and well-

aligned CNTs are obtained in the presence of Ni. Xu et al.

[77,78] studied the growth of CNTs on different alloys for both

co-flow and counterflow methane diffusion flames. They

found that ACNTs could only be grown on Ni/Fe/Cr alloys.

The authors suggested that both Fe and Ni were necessary

for the growth of ACNTs. ACNTs are also reported to grow

on Co-coated stainless steel [95]. However, aerosol metal

organic catalysts show a completely different outcome to that

mentioned earlier [89,90].

The regions of the flame in which ACNTs can be grown are

very limited [83]. Different locations have different carbon

species concentrations and temperature profiles, which

determine the morphology of the CNTs formed. The forma-

tion of CNTs usually happens in the visible orange soot zone

of a normal diffusion flame [96]. Yuan et al. [83] found that the

yellowish flame was the best place for the CNTs to grow. They

suggested that the temperature distinguished the carbon rad-

ical species that contributed to the formation of CNTs. Higher

temperatures were promoting the formation of CNTs over

soot. In another study by Yuan et al. [82], the authors found

that the yield, diameter and height of CNTs increased with

the sampling height from the nozzle and the temperature.

Woo Lee et al. [96] reported that the formation of CNTs on

a catalytic substrate occurred at a location outside of the soot-

ing zone and the flame front of inverses diffusion flame. Xu

et al. [78] confirmed this point by growing CNTs at the tip of

the flame. However, the CNTs were shorter in length than

those produced inside the flame. CNTs are commonly found

in the sooting region of different kinds catalytic probes used

because of the higher temperatures, which promote the for-

mation of active catalytic nanoparticles for CNT nucleation.

The studies of non-premixed diffusion flames show that the

CNTs grow profusely in the areas near to the centreline of

the flames but not at the centre because the centre of the

non-premixed feed is lacking in unsaturated carbon species

and CO that will contribute to the formation of CNTs [78].

Counterflow diffusion flames may provide a stable one-

dimensional reaction zone. There are studies [76,79] based

on the methane flame model (Fig 6(a)) proposed by Beltrame

et al. [97] that predict the temperature profile and major car-

bon species, as shown in Fig 6(b). CNTs are grown 89.5 mm

from the fuel nozzle. ACNTs are only found when an electric

field is introduced [76,79]. Merchan-Merchan et al. [76] found

that highly ordered vertical ACNTs were grown in the region

8.510.0 mm from the fuel nozzle. Xu et al. [77] successfully

produced ACNTs with a methane flame seeded with

acetylene without the electric field. The breakup of the alloy

surface induced by the carbide formed the catalyst nanopar-

ticles that were responsible for the growth of the CNTs. The

high density of the catalytic nanoparticles formed facilitated

the growth of denser CNTs and provided vertical support for

the nanotubes. In some other studies [76,79], the counterflow

diffusion flame was also seeded with acetylene, but no ACNTs

were synthesised, which might be the reason why Xu et al.

[77] conducted the synthesis process at a higher temperature

and why this condition enhanced the formation of catalytic

nanoparticles that could grow CNTs, as compared with those

reported in [76,79].

Flame synthesis is less popular than CVD. There are sev-

eral shortcomings to this method. The apparatus for the

flame synthesis, especially the burner, is complicated so that

the morphology of the flame can be controlled. The gaseous

fuels must be injected safely and carefully. The main draw-

back to this method is the poor quality of the ACNTs obtained.

From our review, the majority of the produced ACNTs are not

straight, except for those synthesised with the aid of an elec-

tric field. The bean-sproutlike bundles containing encapsu-

lated particles at the tips of the nanotubes, as shown in

Fig. 7, are always found. In addition, a certain amount of CNFs

is present in the array of ACNTs, which limits the application

in certain fields. The formation of CNTs in the sooting region

also enables the deposition of carbon other than graphitic car-

bon on the wall. The lengths of ACNTs that are synthesised in

the flame synthesis method are relatively short compared

4618 C A R B O N 4 9 ( 2 0 1 1 ) 4 6 1 34 6 3 5


7/23

with those synthesised from CVD. The study, optimisationand application of ACNTs synthesised by the flame synthesis

method is limited. From our review, no articles addressed the

mechanism of the formation of catalyst nanoparticles, and

characterisation techniques to determine the crystal phases

of the catalyst have not been reported. The actual growth

mechanism of the CNTs synthesised from the flame synthe-

sis method is still unclear. The poor structure and alignment

of ACNTs synthesised by this method have seriously limited

the application of ACNTs. However, the low operating cost

and scalable production enable ACNTs to be applied as a cat-

alyst support and composite reinforcing material, which do

not require ACNTs with perfect alignment and high

crystallinity.

3. Double-step methods

The double-step methods involve coating the active catalyst

onto a substrate, which has the advantage of controlling the

morphology and distribution of the catalyst particles. In addi-

tion, the CNT growth location can be controlled with this ap-

proach. In general, higher-purity ACNTs are produced by

double-step methods compared with single-step methods.

There are two general approaches used to coat the catalyst

onto the substrates, i.e., through physical vapour deposition

(PVD) and by a solution-based precursor.

3.1. Physical vapour deposition (PVD)

PVD is a vacuum deposition process used to coat a thin film of

catalyst on a substrate by vaporising the catalyst precursor

and then condensing it onto the substrate surface. Thin-film

deposition onto a substrate uses the following sequence:

First, the catalyst to be deposited is vaporised by physical

means. The vapour is transported in a low-pressure environ-

ment to the targeted substrate and condensed to form a thin

catalyst film [98,99]. This process is one of the most popular

and most efficient methods for preparing nucleation sites

for growing ACNTs.

PVD is commonly used for the thin-film coating in ACNT

synthesis. The size of the catalyst particles can be controlled

effortlessly by adjusting the film thickness through the depo-

sition time [100]. However, PVD is a relatively costly method.

It requires sophisticated equipment. High vacuum and high

temperature are required for the catalyst deposition, which

dramatically increases the energy consumption. However,

PVD still remains the most popular coating method in the

field of ACNT synthesis.

The thickness of the catalyst film is closely related to the

morphology and alignment of the subsequence growth of

ACNTs. ACNTs will not grow on a continuous catalyst film.

The film must be broken down through annealing to create

nanoparticles, better known as islands. The size of the nano-

particles determines the density, diameter and number of lay-

ers in the tube wall of the ACNTs [101]. If the thickness of the

Fig. 7 An SEM image of bean-sproutlike bundles of well-

aligned CNTs with catalyst nanoparticles lifting off at the tip

of the CNTs [82].

Fig. 6 (a) A schematic of the experiment set-up from Merchan-Merchan et al. [76] using a counterflow diffusion flame

proposed by Beltrame et al. and (b) the numerical predictions of temperature and major chemical species [79,97].

C A R B O N 4 9 ( 2 0 1 1 ) 4 6 1 34 6 3 5 4619


8/23

catalyst film is too thin, a sparse distribution of metal islands

will be produced that results in thin ACNTs of low density

[102]. Conversely, a thick catalyst film will result in sintering

of the metal catalyst, which prevents the catalyst film from

breaking into isolated nanoparticles or islands for growing

ACNTs [103]. Ho et al. [104] studied the effect of 10, 15 and

25 nm Ni films on quartz glass. As shown in Fig. 8, the thinner

film provides a more uniform distribution of small-sized cat-

alyst nanoparticles. As the thickness increased to 15 nm, the

nanoparticles started to coalesce and formed elongated

shapes. For the 25 nm Ni film, the coalescence of the nanopar-

ticles was more extensive and the voids between the nano-

particles were covered, just like before the annealing step. It

was reported that dense and vertical ACNTs can only be

grown on a substrate with catalyst nanoparticles distributed

in the pattern shown in Fig. 8(a). The optimum thickness of

the film required is interdependent with the type of substrate,

the buffer layer, and the annealing environment. The size of

the metal islands can be controlled through temperature

and the time of the annealing process. A thicker film requires

a longer annealing time at high temperature to break the film

into smaller islands and vice versa [101].

Chiu et al. [105] and Wu and Chang [106] studied Fe films

with thicknesses between 0.3 and 3 nm and showed that

the diameter of CNTs was directly proportional to the film

thickness. In addition, CNTs with almost identical diameters

were grown when the same thickness of Fe was used in both

studies, although the experiments were conducted under dif-

ferent conditions. The trend did not change even though the

range of the film thickness was increased to 50 nm [107]. A

high percentage of SWCNT arrays were reported for an Fe film

of 0.6 nm, while double-walled carbon nanotubes (DWCNTs)

were observed on a film of 5 nm [108]. The height of the

CNT arrays and the growth rate of the CNTs were also affected

by the film thickness. The growth rate of the CNTs was more

rapid when thinner films were used because the reactivity of

smaller-sized catalyst particles was higher and they are more

difficult to deactivate [105]. It is well known that the align-

ment of CNTs is induced by van der Waals forces. The larger

catalyst nanoparticles will grow CNTs with larger diameters.

Additionally, the van der Waals force is stronger for a tube

with a larger diameter. This force will restrain the upward

growth of the CNTs [106]. A thicker film will result in denser

ACNT arrays, which inhibits the diffusion of carbon to the cat-

alyst, resulting in a lower CNT growth rate.

3.1.1. Pretreatment

In the pretreatment step, the widely studied parameters are

the use of reducing gas, the time and the temperature [109

113]. Ammonia and hydrogen are the common reducing gases

that have always been applied in dry etching processes. The

metal film will break into small and more uniform nanoparti-

cles in the presence of ammonia or hydrogen gas [109,110].

Ammonia decomposes to hydrogen and nitrogen during the

pretreatment. Hydrogen reduces the metal oxide that

provides a nucleation site for growing nanotubes. It is well

documented that in the initial stage, the average size of nano-

particles decreases but the density increases with etching

time [109,110]. Prolonged pretreatment at high temperatures

enhances the possibility of coalescence with neighbouring

particles to form larger nanoparticles [110]. The inverse effect

is shown when the etching time is too long; the catalyst is

etched by the excited hydrogen generated by either hydrogen

or ammonia. The catalyst nanoparticles are more crystallised

when the hydrogen flow rate is high [112]. It has also been re-

ported that metal nitride will be formed in the presence of

ammonia or nitrogen during the pretreatment stage, which

enhances the growth of CNTs [114,115]. However, the nitro-

gen-to-hydrogen ratio must not be too high, otherwise nitro-

gen will etch away the catalyst [113].

Microwave or other sources of power can be applied when

synthesising ACNTs [116]. The processing pressure in this

treatment must be high enough to provide sufficient flux im-

pact from the plasma, so that the film will receive more en-

ergy and momentum transfer to break the catalyst film into

nanoparticles [113]. The nanoparticle size increases with the

microwave power, as does the disorder of the subsequence

CNTs grown. From Fig. 9, it can be seen that hydrogen plasma

did a better job than ambient hydrogen at breaking the film

into fine catalyst nanoparticles.

Diluted hydrofluoric acid (HF) is widely used in wet etching

to obtain an uneven surface morphology. Lee et al. [118120]

used both HF dipping and NH3 pretreatment for Ni and Ni

Co films on SiO2 substrates. The roughness increased with

dipping time in HF, while NH3 further etched the surface to

form small domains inside the metal cluster. Choi et al.

[121] reported that different surface morphologies were

obtained at different durations of HF dipping. First, HF etched

away the metal catalyst and increased the surface roughness.

Microcracks were formed later. The roughness on the sub-

strate caused by the etching increased the grip of the catalyst

Fig. 8 SEM micrographs of deposited Ni films after annealing, with thicknesses of (a)10 nm, (b) 15 nm and (c) 25 nm [104].

4620 C A R B O N 4 9 ( 2 0 1 1 ) 4 6 1 34 6 3 5


9/23

on the surface. This condition provided a better platform for

the formation of nanoparticles than did a smooth surface.

3.2. Solution-based catalyst precursors

To overcome the shortcoming of PVD, solution-based cata-

lyst precursors have been developed as an alternative way

to coat active catalysts on the substrate for growing ACNTs.

The main techniques used to coat the catalyst on the sub-

strate with solutions are dip coating, spin coating, spray

coating and microcontact printing. All of these techniques

are able to distribute the catalyst nanoparticles evenly on

the substrate, which is the main criterion for obtaining

ACNTs. In the dip-coating method, the substrate is dipped

in a solution containing catalyst precursor at a constant

speed in order to prevent any judder and rippling on the

surface of the solution. The substrate is immersed in the

solution and withdrawn at a slow, uniform speed to obtain

a uniform coating. The volatile solvent is evaporated, leav-

ing the metal catalyst on the substrate [122]. A thick cata-

lyst film will be formed if the withdrawal speed is too

rapid [123]. Vibration damping tools are sometimes required

to ensure that the liquid surface remains ripple-free to ob-

tain a homogeneous thickness across the entire substrate

[124]. In spin coating, a puddle of the solution is placed

on the substrate at the axis of rotation. The substrate is ro-

tated at a very high speed to spin away the excess amount

of solution through centrifugal force. The inertia of the

solution is the reason that the solution is ejected radially

outward. The thickness of the solution decreases at high

spin speeds. A thick film will result in the formation of lar-

ger-sized catalyst particles, and a thin film leads to the for-

mation of smaller-sized but denser particles [125]. The

solvent used in spin coating is almost the same as the solu-

tion used in dip coating. In the spray-coating system, the

precursor is atomised at the nozzle by pressure and then

directed towards the substrate. Microcontact printing, a soft

lithography technique, has also been applied in the synthe-

sis of ACNTs. The stamps for microcontact printing are pre-

pared by curing poly(dimethyl)siloxane. The printing is

carried out by placing the stamp on the surface of the sub-

strate, followed by transferring the catalytic materials onto

the tops of pillars [126]. Usually, microcontact printing is

applied in the patterned growth of ACNTs.

The catalyst solution is crucial for determining the mor-

phology and topology of ACNTs grown from different coating

techniques. Various types of solutions are prepared for this

purpose. The most common solution is an alcoholic solution

containing metal salts. Alcohol is used because of its high vol-

atility, and the metal salts are those that are able to be diluted

easily in alcohol. Metal nitrate and metal acetate, which have

high solubilities in alcohol, are always selected as the catalyst

precursors [127]. Murakami et al. [128130], Maruyama et al.

[131] and Hu et al. [132] used 0.01 wt.% cobalt acetate and

molybdate acetate in ethanol to grow aligned SWCNTs. Metal

acetate is also used in spin coating[133,134]. One of the weak-

nesses of the alcoholic solution is that alcohol has a low

vapour pressure and a low viscosity, which causes the recrys-

tallisation of salts and forms non-uniform dispersion of the

catalyst nanoparticles after the drying process. However,

ethylene glycol can be added to surmount this obstacle.

Liquid nitrogen has been utilised to freeze-dry the solution

and prevent the agglomeration of salts that results from rapid

evaporation of the alcohol, which maintains the uniform

dispersion of the catalyst [134]. Mauron et al. [125] found that

an increase in the concentration of the solution increased the

diameter of the catalyst nanoparticles but decreased the

density of the catalyst nanoparticles [125]. However, if the

concentration is too low, no CNTs are formed or just low-

density entangled ACNTs will be obtained [135].

Cho et al. [136] and Choi et al. [127] used a magnetic fluid to

disperse the catalyst nanoparticles on the substrate. Magnetic

fluids are stable colloidal suspensions composed of single-

domain magnetic nanoparticles dispersed in proper solvents

[137], as shown in Fig. 10. The authors [127,136] used decanoic

acid as the surfactant that dissolved in the acetone. It was

slowly added to an iron chlorideammonia solution to ensure

that the surfactant was able to completely cover the surface

of the Fe3O4 particles so that the repulsion between particles

was effective. In this process, the polar heads of the primary

surfactant adsorb onto the catalyst particles, while the non-

polar tails are exposed to the solvent. The nonpolar tails of

the secondary surfactants mount onto the tails of the primary

surfactants through the van der Waals force, leaving the polar

heads bound to the ammonia and acetone solvent. The steric

repulsion prevents them from agglomerating and maintains

the uniform dispersion after coating. Small catalyst islands

are formed after heat treatment, and these islands are

Fig. 9 SEM images of Ni/Si substrate after 10 min (a) annealing in H2 ambient (b) etching by H2 plasma at 700 C followed by

cooling to room temperature in vacuum [117].

C A R B O N 4 9 ( 2 0 1 1 ) 4 6 1 34 6 3 5 4621


10/23

responsible for growing ACNTs. Polyvinyl alcohol is some-

times added to increase the viscosity of the solution in order

to control the density of the nanoparticles and the film thick-

ness during spin coating[127].

The solgel method is applied for admixtures of tetraeth-oxysilane, iron nitrate aqueous solution and ethanol, which

is used for spray coating [138]. Pan et al. [37] mixed tetra-

ethyl-ortho-silicate, iron nitrate, ethanol and Pluronic P-

123 triblock copolymer to develop mesoporous silica with

the catalyst particles distributed evenly on the substrate.

The addition of the triblock copolymer improved the wet-

ting ability of the solution on the substrate. The substrate

was dip-coated into the mixed solution. Liu et al. [139] used

block copolymers as micelle catalyst templates. Poly(sty-

rene-block-acrylic acid) (PS-PAA) was dissolved in toluene,

stirred and heated to convert all of the polymer material

to the spherical micelle phase. FeCl3 was then added to

the solution as the catalyst precursor. PS-PAA is an amphi-philic block copolymer that forms micelles in solution,

which are capable of self-organising into partially ordered

structures. Fe ions diffused through the thin PS layer before

exchanging with the H+ ions of the carboxyl group and

being effectively bound into the PAA core. A quasi-hexago-

nal monolayer array was obtained within the PS matrix be-

cause of the self-assembly of the PS-PAA micelles, as shown

in Fig. 11(a). The catalyst was uniformly dispersed on the

substrate for the synthesis of ACNTs in a subsequent CVD

process. NaOH was added to facilitate metal loading. The

carboxyl acid group in the PAA core underwent hydration

during neutralisation, causing swelling of carboxyl-contain-

ing latex. The volume expansion of the PAA domains led to

the ruptures illustrated in Fig. 11(b) [140]. The catalyst ions

exchanged directly with the Na+ on the carboxyl group.

Various parameters can be varied to enhance the disper-

sion of the catalyst nanoparticles. The concentration of the

catalyst will affect the diameter of the nanoparticles formed

on the substrate. After the catalyst is saturated in the solu-

tion, the size of the catalyst nanoparticles is determined bythe micelle dimensions. The molecular weight of the PS af-

fected the spacing between the nanoparticles, while PAA

dominated the size of the catalyst nanoparticles. Diluting

the solution with PS homopolymer decreased the density of

the catalyst nanoparticles. Other colloidal solutions such as

Co nanoparticles have been prepared by dispersing AOT[-

bis(2-ethylhexyl)-sulfosuccinate]-stabilised Co nanoparticles

in toluene, and this approach has been used in spin coating

[142].

Ryu et al. [143] used polystyrene nanospheres for shadow

masks, one of the nanosphere lithography techniques used

for the fabrication of nano-pitched metallic arrays. An or-

dered monolayer of nanospheres was spin-coated on the sub-strate, and catalyst solution was coated on top of the

nanopsheres. Catalyst spots were formed on the substrate

through the triangle voids between the spheres. The size of

the nanospheres was used to control the size and density of

the nanoparticles that formed.

The main drawback of the solution-based precursor ap-

proach is that preparing the catalyst precursor is cumber-

some. It may take hours or days to prepare the solution. In

addition, the solution tends to accumulate in the notch on

the substrate. In the spin-coating method, the effect of sur-

face tension will oppose the uniformity and topography of

coating, while contact printing is only suitable for coating

small areas. The catalyst accumulates in recessed areas in

dip coating [144]. The distribution of catalyst nanoparticles

is not as uniform as that created by the PVD process.

Aqueous solution of

FeCl2 and FeCl3

NH4OH + primary

surfactant + acetone

Secondary

surfactant

Secondary surfactantprimary surfactant

Fig. 10 A schematic representation of the synthesis of surfactant bilayer-stabilised magnetic fluids, using fatty acids as

primary and secondary surfactants to obtain stable aqueous magnetic fluids. The black and hollow dots represent the polar

heads of the fatty acids [108].

PS matrixPS matrix

PAA domainsCavitated PAA domains

Fig. 11 A diagram of (a) a non-cavitated thin film on a substrate with metal loaded and (b) the cavitation PAA domain with

NaOH added [140142] (black dots represent iron chloride).

4622 C A R B O N 4 9 ( 2 0 1 1 ) 4 6 1 34 6 3 5


11/23

Millimetre-scale growth of ACNTs is seldom reported by this

method. Although different types of coating methods have

been proposed, extensive studies have not yet been carried

out.

3.3. Catalyst

The most commonly used active catalysts for growing CNTs

are magnetic elements such as Fe, Co or Ni. It is reported that

Co is more appropriate than Fe and Ni for producing better-

quality CNTs [145]. Yuan et al. [146] found that long and

straight ACNTs can also be grown using Pt, Pd, Mn, Mo, Cr,

Sn or Au.

Magnetic elements have a strong tendency to agglomerate

on the substrate at high temperatures because of their mag-

netic properties and their high specific surface energy [147].

Other metal elements areaddedas stabilisers to prevent exces-

sive agglomeration of the active elements. Molybdenum is one

of the proven catalyst stabilisers for the growth of CNTs. Mar-

uyamas research group usedCoMobimetallic catalyst in syn-

thesising aligned SWCNTs [129132]. According to the authors,

the affinity of Mo for oxygen is stronger than that of Co, and

thus Mo tends to form an oxide at the interface of the catalyst

and the substrate during the calcination process. It is reported

that Co will diffuse into MoO3 and form a CoMoOx underlayer.

During the catalyst reduction, CoO and MoO3 will be reduced

to Co and MoOy (y 6 2). CoMoOx remains unchanged because

of its stability. The gooddispersionof Co nanoparticles is attrib-

uted to the strong interaction between metallic Co and Co-

MoOx. The chemical state and the morphology of CoMo

catalysts are shown in Fig. 12. Noda et al. [148] found that the

yield of CNTs was relatively high when the concentration of

Co was slightly higher than the concentration of Mo. This find-

ing supports the mechanism proposed by Maruyamas group.

We have also reported that the presence of CoMoO4 after calci-

nation of CoOMoO/Al2O3 catalyst plays the same role as previ-

ously mentioned [149]. The optimum ratio of CoOx to MoOx is

8:2 (w/w) [150], which deviated from the findings of Maruyama

and Noda. In addition, CoAl2O4 and Co3O4 were formed for the

CoOMoO/Al2O3 catalyst after calcination. It was reported that

the formation of CoO and Co after partial reduction by hydro-

gen was more effective for growing CNTs [151]. Meanwhile

MgMoO4 and CoMoO4 were formed after calcination for Co

Mo/Mg [152]. However in other studies, when CO was intro-

duced in CoMo, molybdate dissociated to form molybdenum

carbide and Co particles of smaller size that could grow

SWCNTs [153155]. Co ions were embedded in the molybdate-

like cluster after calcination, and it remained unchanged after

reduction. Continuously introducing CO reduced the Co parti-

clesto metallic Co, which subsequently aggregated to form lar-

ger particles that led to the formation of MWCNTs and CNFs.

Aligned SWCNTs were also grown on FeMo bimetallic cata-

lysts. The order of deposition of different types of metals has

been shown to have an influence on the morphology of the

ACNTs formed. With Mo deposited on Fe, the catalyst resisted

poisoning at a high hydrocarbonflux duringCVD. Mowas pres-

ent in the outer portion of the particle, protecting Fe. Mean-

while for the case in which Fe was deposited on Mo, Fe easier

to get poisoned [156].

The capability of a TiCo bimetallic hybrid catalyst to pro-

duce aligned SWCNTs was reported in [157]. Ti prevented the

formation of Co-silicate. Sato et al. [158] proposed a mecha-

nism for growing CNTs with TiCo catalyst. It was shown that

TiCoincorporated more carbon, in theform of TiCx, compared

with Co alone.Furthermore,the meltingtemperature of TiCo

C was lower than that of CoC, which enhancedthe carbonsol-

ubility in the catalyst. This feature makes it possible to grow

ACNTs at lower temperatures. The hybrid bimetallic coating

(co-sputtering of Ti and Co) outclassed the layer coating of

the catalyst in yield performance, which is attributable to a

more uniform dispersion of catalyst [159]. CNTs were also pro-

duced with Ni/TiO2 [160] and Mn/Ni/TiO2 catalysts [161]. Gunji-

shima et al. [162] used FeV for the production of aligned

DWCNTs. The authors claimedthat the incorporation of vana-

dium was able to increase the activity of Fe and that it initiated

the growth of CNTs before the catalyst particles started to

aggregate. Cr is another choice of metal that helps in the syn-

thesis of ACNTs. The role of Cr is to disperse the active metal

catalyst uniformly, leading to a better alignment of the grown

CNTs [147]. A large amount of Cr is required for effective pre-

vention of metal catalyst aggregation. CNTs can be grown on

CoV, CoFe, CoNi, CoPt, and CoY. However, ACNTs were

only found on CoV and CoFe [163].

In the nucleation stage, the carbon needs to dissolve in the

catalyst particle until it reaches supersaturation to initiate the

growth of CNTs [74]. Liu et al. [103] co-sputtered iron and

graphite on the substrate with the goal of growing ACNTs,

and they found that the alignment and density of ACNTs

was better than when using pure iron as a catalyst. The pre-

saturation of carbon allows CNTs to grow with a high density

Fig. 12 Schematic representation of the changes in the chemical state and the morphology of CoMo catalysts on quartz

substrates after (a) calcination and (b) reduction [132].

C A R B O N 4 9 ( 2 0 1 1 ) 4 6 1 34 6 3 5 4623


12/23

and uniform in the initial phase of growth, which causes

crowding of the ACNTs and yields ACNTs with a better align-

ment and orientation.

3.4. Chemical vapour deposition (CVD)

Besides FC-CVD, which was discussed in Section 2, there

are three other conventional CVD methods: thermal CVD

(T-CVD), hot-filament CVD (HF-CVD) and plasma-enhanced

CVD (PE-CVD). These methods have been proven to be very

successful in growing ACNTs. PE-CVD outclasses both

T-CVD and HF-CVD in terms of the alignment of the synthes-

ised CNTs because of the applied bias to the substrate [164].

The use of plasma can significantly reduce the activation en-

ergy of the growth. The lowest temperature reported for the

successful growth of ACNTs using PE-CVD was 120 C [165].

However, the ACNTs were poor crystallinity, short and with

cone-liked shape.

Self-bias potential is influenced on the surface of the sub-

strate under plasma conditions, inducing an electric field dur-

ing the growth of CNTs. The field forces the CNTs to align to

the direction of the electric field as they grow [166]. The field

emission of CNT films may be enhanced by treatment with

hydrogen plasma. The presence of CH bonds in ACNT sam-

ples is expected in a hydrogen plasma atmosphere [167]. Be-

sides that, the period of growth of ACNTs can be controlled

precisely. The growth can be quenched immediately by

switching off the power supply for plasma generation [168].

Another advantage of PE-CVD over T-CVD is that PE-CVD is

highly efficient at gas decomposition and the concentration

of reactive species can be controlled [101]. Normally, direct

current (dc), radio frequency (rf) or microwave excitation are

used to generate plasma. Plasma deposition is stable and

highly controllable, leading to reproducible growth condi-

tions, thus lowering the growth temperature [14]. In dc-PE-

CVD, the degree of alignment is improved with the plasma

voltage. Below the plasma excitation potential, no alignment

is observed [169]. The plasma is ignited if the voltage is high

enough, and intensive ion bombardment is applied to the

structure, which initiates the growth of ACNTs. Temperature

control is another crucial factor for successfully using PE-

CVD. Low temperatures could limit the decomposition of

hydrocarbon gas, leading to the formation of amorphous car-

bon. However, if the temperature is too high, CNTs appear in a

bundle form, and most of the CNTs possess disordered walls

[169]. Microwave excitationPE-CVD [116] and rf-PE-CVD [170]

both show the same outcome under high temperature and

power. The crystallinity of the CNTs could be seriously dam-

aged because of the ion bombardment [170]. The negatively

biased substrate tends to attract and accelerate the positively

charged hydrogen and hydrocarbon ions generated within the

plasma. The ions become etchants and cause damage to the

walls of the CNTs [171]. Carbon soot will deposit at the tips

of the CNTs if the power applied is too high. To solve this

problem, a metal plate is used to cover the substrate so as

to prevent the ions from bombarding the substrate [170]. An-

other drawback of PE-CVD is that this approach involves com-

plex equipment setup [172]. Nozaki et al. [173] pointed out

that two issues need to be addressed to improve the growth

of CNTs with PE-CVD. One of them is the preparation of

catalyst nanoparticles that do not coagulate extensively while

maintaining their catalytic function during PE-CVD. The other

one is the use of remote plasma, which would restrict the ion

damage to both the catalyst nanoparticles and the CNTs.

They proposed atmospheric pressure PE-CVD. Under higher

gas pressure, ions that accelerated towards a substrate would

undergo collisions with neutral molecules in the sheath and

the bombardment energy was lower toward the substrate

and prevents the damage on the ACNTs grown [174]. Besides,

Nozaki et al.[175] found that SWCNTs can only be grown in

the atmospheric PE-CVD.

In general, HF-CVD and T-CVD are more desirable tech-

niques than PE-CVD for producing ACNTs because the later

involves high operating costs and sophisticated equipment

setup. Furthermore, HF-CVD and T-CVD are suitable for the

irregular-shaped and multiple substrate coatings that cannot

be used in PE-CVD [67]. HF-CVD and T-CVD are also free from

the complications resulting from the large amount of undesir-

able and uncontrollable radicals created in the plasma during

the growth of CNTs [78]. A combination of HF-CVD and PE-

CVD has been studied to capitalise on the advantages of both

methods [7]. However, it was reported that the CNTs were

quite similar to the CNTs produced from PE-CVD alone [172].

Low temperature CVD becomes important when the sub-

strate involves metallic components, such as in CMOS and

large-scale-integration (LSI) interconnects, in which these

components will deform at temperatures above 550 C

[176]. PE-CVD is not suitable due to the damage caused to

the CNT wall and that increases the electrical resistivity

and constrains in the applications in electronic industry.

Normally, hot filament or feedstock preheating are required

in a low temperature CVD. The temperature of hot filament

has to be high enough to decompose carbon source into ac-

tive components such as radicals to initiate the growth of

ACNTs [177,178]. Higher preheating temperature will pro-

duce ACNT arrays of better crystallinity and higher arrays

[179]. Herringbone-liked structure and CNFs are dominant

if the preheating temperature [180] or reaction temperature

[176] is too low. The graphite sheet of CNTs from low tem-

perature CVD either by HF-CVD or feedstock preheating

method contains more defects and the yield is low as com-

pared to CNTs grown in a high temperature condition

[181,182].

ACNTs can be grown under low temperature CVD with-

out the use of feedstock preheating or hot filament. The

lowest temperature had been reported so far was 350 C

in a simple T-CVD [183]. The author found that the credit

belongs to the small catalyst nanoparticles, where nucle-

ation and growth of CNTs can be initiated on the surface

of the metal particle with proper shape which avoids the li-

quid catalyst-carbon eutectic phase that requires higher

temperature. Meanwhile, Mora et al. [184] found that as

long as the feedstock is decomposed, CNTs will be grown.

Using carbon feedstock associated with exothermic decom-

position will lower the growth temperature. However, the

main drawback of low temperature synthesis is low quality

ACNTs produced [158,159,184,185]. The room of study in low

temperature CVD remains wide and efforts have been put

for improving its efficiency so that it is comparable with

conventional CVD.

4624 C A R B O N 4 9 ( 2 0 1 1 ) 4 6 1 34 6 3 5


13/23

3.5. Supergrowth

The growth of ACNTs in CVD is always limited by the wrap

of amorphous carbon on the surface of the catalyst. It is dif-

ficult to obtain millimetre-scale ACNTs by CVD. Hata and

co-workers [186] achieved a massive breakthrough by inject-

ing a small amount of weak oxidiser, i.e., water vapour, dur-

ing the CVD reaction. The growth rate of ACNTs was

stunning: an ACNT array 2.5 mm in height was grown within

10 min, and an ACNT/catalyst weight ratio of above 50,000%

was achieved. This new CVD method is denoted as super-

growth. Li et al. [187] almost doubled the height of the ACNT

forest to 4.7 mm in 2 h of reaction time. The main purpose of

using a weak oxidant in CVD is to prevent deposition of amor-

phous carbon over the catalyst nanoparticles and to prolong

the catalytic lifetime [186188]. The work of Yamada et al.

[189] shows that the deposition of amorphous carbon was

the dominant factor of terminating CNTs growth in CVD,

water vapour and oxidizer played an important role in sus-

taining supergrowth. Introducing water vapour also increased

the selectivity towards SWCNTs. According to Amama et al.

[190], injecting water vapour in the reaction may hold back

the Ostwald ripening causing large catalyst nanoparticles to

grow larger due to the sintering effect of smaller catalyst par-

ticles during annealing and CVD. In addition, the wall struc-

ture is significantly improved by introducing a small

amount of water vapour during CVD [191194]. It has been re-

ported that with the assistance of water vapour, the aspect ra-

tio of CNTs was greatly increased and a purity of 99.9% and

above was successfully achieved [186,191,195]. At that level

of purity, the purification process can be omitted, as it is well

understood that purification may change the morphology and

cause defects to the ACNT arrays.

The effect of water vapour is interdependent on other

parameters involved in the CVD process. The optimum

amount of water vapour injected is closely related to the

hydrocarbon flowrate [196]. Higher flowrates of hydrocarbon

will deactivate the catalyst by forming encapsulated carbon

over the active catalyst. Water vapour inhibits this process,

and thus a very high initial growth rate (IGR) can be achieved.

From the study reported in [192,194,197], an increase in the

amount of water vapour introduced increased the height of

the CNT arrays, increased the CNT growth rate and improved

the CNT alignment. However, too much water vapour would

oxidise the CNTs. Balancing the rate of amorphous carbon

deposition and the rate of amorphous carbon removal is the

key factor to producing ultralong CNT arrays [198]. The study

of Futaba et al. [196] showed that the optimum water/ethyl-

ene ratio is 1/1000. Our review also shows that the majority

of millimetre-scale ACNT arrays are produced by introducing

a small quantity of water vapour into the reactor during CVD.

However, water vapour also has an etching effect on the cat-

alyst at high temperatures, as reported elsewhere [199].

The main contribution of the supergrowth method is to

provide a solid platform to grow millimetre-scale ACNTs.

There are two different explanations of how the supergrowth

happens. The first explanation is that the water vapour in-

creases the IGR. Futaba et al. [196] derived an equation to cal-

culate the IGR, and they found that the rate of 207 lm/min

was obtained in their study. Patole et al. [200] also obtained

an IGR of more than 200 lm/min. However, Li et al. [187] re-

ported that water vapour did not increase the IGR. Rather, it

increased the lifetime of the catalyst. From both phenomena,

one could speculate that a high IGR is usually found in the

presence of a relatively high water/hydrocarbon ratio. The

water acts as an oxidising agent to prevent the wrap up of

the catalyst nanoparticles by amorphous carbon at the initial

growth stage, maintaining the catalyst in a fresh condition to

give very high CNT growth rates. When ACNT arrays form,

ACNTs keep hydrocarbons and water from reaching the cata-

lyst and water keeps etching away CNTs. Eventually, both

rates become equal, and the growth stops. For low water/

hydrocarbon ratios, the IGR is lower, but the growth can last

longer.

Supergrowth is no longer limited to water vapour. Other

growth enhancers such as alcohols, ethers, esters, ketones,

aldehydes, and even carbon dioxide have been used [188].

Air had also been used to enhance the growth rate of CNTs.

With an optimum amount of air introduced in the reaction,

the CNT growth duration can be extended to more than

15 h [201]. Ammonia and hydrogen also possess the ability

to promote the growth of CNTs, but their efficiency is much

lower than the oxygen-containing compounds. However, it

has been reported that water vapour has no significant effect

on the growth of CNTs with MgO single crystal substrates

[202].

The research has recently been focused on obtaining

aligned SWCNT arrays. Almost all of the reported super-

growth methods use Fe as the active catalyst and alumina

as the support layer. The optimal thickness of the layers of

Fe and alumina during PVD has been widely studied. Super-

growth cannot be achieved if the thickness of Fe on alumina

is more than 5 nm [203]. Those thicknesses also determine

the inter-tube spacing between CNTs, which is an important

factor for supergrowth to take place. Futaba et al. [204] found

the ACNTs only occupied 3.6% of the array space when a

1.2 nm Fe film was used as the catalyst.

The growth of catalyst-free ACNT arrays obtained by the

supergrowth method solves the purification problem. The

extreme high purity of ACNTs enables them to be applied in

various fields such as biology, chemistry and magnetic appli-

cations. The controllable production of thick DWCNT arrays

can be applied to create field-emitting materials that possess

low threshold voltages. The open tip feature and the sparse-

ness shown in [205] that was created with water to oxidise

away the cap make them suitable for gas storage and mem-

brane applications. Our review shows that Fe is the most

widely used catalyst for supergrowth. Although Co and Ni

are also used [206,207], no extensive study or optimisation

has been carried out. Supergrowth seems to be a promising

method to achieve mass production of ACNTs.

4. Substrates and buffer layers

Substrates provide a solid foundation for growing ACNTs. The

substrate must be able to inhibit the mobility of the catalyst

particles in order to prevent agglomeration. The lattice

matching also decides the morphology of the ACNTs formed.

Silicon wafers are one of the most popular substrates studied

C A R B O N 4 9 ( 2 0 1 1 ) 4 6 1 34 6 3 5 4625


14/23

for synthesising ACNTs. However, silicon is not an ideal

substrate. Jung et al. [38] studied the interaction of Fe on plain

silicon wafers during CVD, and they found that Fe incorpo-

rated with silicon to form iron silicide (FeSi2) and iron silicate

(Fe2SiO4) at high temperatures, which are known for their

non-catalytic activity for the growth of CNTs (Fig. 13(b) and

(d)). The same problem affects Ni and Co as well. To overcome

this problem, a buffer layer, also known as the underlayer or

adhesion layer, is always used to isolate silicon from active

catalysts and preserve its activity. Buffer layer also promote

the dispersion of catalyst nanoparticles and increase the sur-

face roughness for better adhesion of catalyst nanoparticles

on the substrate [199]. The TEM images in Fig. 13(a) and (c)

show that the SiO2 layer blocks the catalyst from interacting

with silicon. A SiO2 layer has been proven to be an ideal sup-

port for growing ACNTs because of its high surface roughness

[208].

Alumina and Al can be used to prevent extensive sintering

of Fe particles [209]. By increasing the thickness of the Al

underlayer, the growth of ACNTs can be promoted [210]. Occa-

sionally, Al or alumina is used together with SiO2. The pres-

ence of Al enhanced the morphology of CNTs and increased

the growth rate, as reported by Liu et al. [49]. It was shown

that catalyst oxidation on the catalyst surface can be reduced

by the presence of an Al layer, which helps maintain the cat-

alytic activity [211]. An alloy of Al and catalyst can be formed,

which increases the activity of the catalyst [49]. Alumina is

crucial for the dispersion of smaller NiO crystallites [212].

The role of SiO2 is to prevent the reaction between Al with

Si before alumina is formed. However, if the aluminium oxide

layer is too thick, it will bury the catalyst and inhibit the

growth of CNTs. TungstenTi bimetals and tantalum are suit-

able adhesion layers as they limit the diffusion of Ni into the

substrate [213]. Titanium nitride is also a suitable buffer layer

[112]. Quartz is another substrate often used to replace sili-

con. In FC-CVD, a quartz tube can be directly used as the sub-

strate for growing ACNTs. MgO, sapphire [104], alumina

mullite mixtures, machinable ceramic [49] and Al2O3 fibre

cloth [214] are also well-known substrates for growing ACNTs.

Diameter, density and alignment can be controlled by

manipulating the structural morphology of the substrate.

Lee et al. [215] anodised the surface of a silicon wafer to make

it porous in order to control the density and diameter of the

catalyst particles. Handuja et al. [43] coated a layer of amor-

phous hydrogenated silicon nitride (a-SiNx:H) on silicon. It

was then heated in oxygen to form a crystalline silicon oxide

(SiOx) within the matrix of a-SiNx, which aided the growth of

ACNTs in terms of length and alignment. The size of the

SiNx/SiOx clusters and the orientation of the initial catalytic

centres determine the alignment and diameter of CNTs [43].

In the solgel method, mesoporous silica thin films can be

used. Murakami et al. [128] used tetraethyl-ortho-silicate

(TEOS), ethanol, H2O, HCl and amphiphilic triblock copolymer

[(C2H2O)106(C3H4O)70(C2H2O)106], which served as a struc-

ture-directing agent. The catalyst was loaded after the

film was formed. Xie et al. [216] used tetraethoxysilane

((C2H5O)4Si) hydrolysis in an aqueous solution of iron nitrate

to create a thin film. ACNTs with a specific diameter and dis-

tribution were controlled through the preparation conditions

of the iron/silica substrate [128,216].

Fig. 13 TEM images of a cross section of the substrates: (a) SiO2; (b) Si after CVD; (c) an enlarged picture from the CNT/SiO2interface in (a) showing the presence of gamma-iron particles on the silicon oxide surface and the growth of CNTs from the

particles formed; and (d) an enlarged area from (b) showing the formation of FeSi2 and Fe2SiO4crystals during CVD processing

[38].

4626 C A R B O N 4 9 ( 2 0 1 1 ) 4 6 1 34 6 3 5


15/23

A flat substrate has a low surface area for the growth of

CNT arrays. Ceramic spherical substrates have been used to

mass produce ACNTs [53,217]. Lamellar Fe/Mo/vermiculite

was successfully applied in fluidised bedCVD for mass

production [10]. Li et al. [93,94] grew ACNTs successfully on

1D (Fig. 14) and 2D Si substrates with porous AAO

nanotemplates.

5. Alignment of CNTs

Majorityof thearticles agree with the mechanismof alignment

elucidated by Fan et al. [8] which is simply caused by the van

der Waals force. The strong interaction of the van der Waals

force enables the CNTs bound together to form dense ordered

packing. During the initial stage of CVD, the lack of van der

Waals forces results in the formation of entangled CNTs. As

the catalyst film becomes thicker, the interaction between

the nanotube walls induces the growth of CNTs with a straight

form that is parallel to the substrate. The steric impediment

from neighbouring nanotubesdue to the dense arraypromotes

the aligned growth [119]. The overcrowding of the CNTs in the

array forces the tubes to grow only vertically [71]. The work of

Zhang et al. [218] showed the detailed mechanism of align-

ment from the beginning to the end of the ACNT growth. The

van der Waals interactions can be utilised as binding energy

between adjacent CNTs and the CNT yarns can be drawn for

various application [26,27].

Magnetic fields, electric fields and voltage biasing are

widely applied to provoke the alignment of the CNT array.

PE-CVD is one of the methods that use an electric field to

force CNTs to grow parallel to the electric field. The high

anisotropy of the polarisability of CNTs with an elongated

shape is responsible for the aligned growth in the electric

field. The polarisability is stronger for short CNTs [219]. The

open end of the tubes containing charged dangling bonds

inhibits the closing of the tube end. The interaction between

charged tube ends and the electric field contributes to the

alignment as well [220]. Under the high DC plasma and bias

voltage, ACNTs are grown by pull up force in the electric field

of the sheath on a substrate. Another reason that this ap-

proach works is that the tip of the tubes possess a constant

orientation of catalytic particles. The polarised catalyst parti-

cles and the induced dipole moment align the CNTs parallel

to the electric field all the way through the growth of the CNTs

[76]. The alignment and the thickness of the array increase

with the applied voltage. In flame synthesis, further increases

in the voltage form helically coiled, spiral-like CNTs. When a

strong electric field is employed, it interacts with the induced

charges at the tips of the CNTs, which generates sprouting,

thus splitting catalytic droplets to grow CNTs with L, T

and Y branches [79]. The polarity of the bias or field and

the substrate are factors that influence the appearance of

ACNTs. Otherwise, no alignment could be obtained.

6. Horizontally aligned carbon nanotubes

Horizontally aligned CNTs are another form of CNTs in which

the growth is parallel to the substrate. This orientation is very

important for applications in the electronics fields. Liu and

co-workers [62,221,222] first proposed a flow-directed growth

mechanism for a rapid-heating CVD method. A kite mecha-

nism was proposed. In rapid-heating CVD, the substrate and

the surrounding gas are heated at different rates. As a result,

a convection flow is formed by this temperature difference.

This flow will lift up the nanotubes with catalyst at the tips.

A quartz tube reactor with a smaller diameter is preferred

for CVD, which enhances the laminar flow and thus assists

the floating of the nanotubes [223]. Cu has been used as a

replacement for Fe because of its low level of interaction with

Si, which helps the active tip to float [224]. However, there is

disagreement about this mechanism. Jin et al. [225] pointed

out a drawback of Lius method. The hotter gas near to the

wall of the reactor would levitate along the wall, while in cen-

tre, the cooler gas at the centre of the tube descend down-

ward, causing a symmetrical gas circulation as a secondary

flow. This lateral vortex flow distorted the laminar flow and

decreased the buoyancy effect that uplifts the growing CNTs,

altering the alignment. They used an ultralow gas feeding rate

so that the gas heated up gently to prevent secondary flow.

However, Li et al. [226] found that horizontally aligned CNTs

were not parallel to the flow of the gas feed. The study of

Yu et al. [227] also found that the flow of gas had no effect,

which means that there are other factors that control the

alignment of CNTs instead of gas flow.

The highly anisotropic polarisability of CNTs makes it pos-

sible to use an electric field as an aligning force during

growth. Normally, an electric field is applied through a pair

of electrodes. Dai and co-workers [228] grew highly aligned

suspended SWCNTs under electric fields in the range of 0.5

2 V/mm. High voltage would break down the SiO2 dielectric

layer if Si/SiO2 were used. The short metallic CNTs are more

readily aligned in an electric field than semiconducting CNTs

because of their considerably higher polarisability. As a result,

the angular distribution of short SWCNTs is bimodal. Long

semiconducting CNTs (>1.5 lm) can also be aligned through

electric fields [229]. Although electric fields are capable of

aligning the growth of CNTs, Dai [228,230] and Joselevich

[229,231] found that electric fields are not the dominant factor

affecting the growth direction. A magnetic field may also

assist in the alignment of the CNTs. However, magnetic fields

will interact with the ferromagnetic catalyst instead of the

CNTs [232,233]. The CNTs appear to arch over the substrate,

and it is believed that weak magnetic fields are unable to sup-

port the weight of the catalyst.

Nanotubes grown in both a flow-directed and field-direc-

ted manner have a low density and a low degree of alignment.

Fig. 14 1D sandwich-structured Si-substrate with an AAO

template. [94].

C A R B O N 4 9 ( 2 0 1 1 ) 4 6 1 34 6 3 5 4627


16/23

Introducing strong electric fields is not easy [234]. For this rea-

son, surface-directed growth was developed to address this

problem. So far, various mechanisms have been proposed.

Ismach et al. [235] was the first to report atomic-step template

growth. This is a method that utilises the miscut of a crystal

to grow aligned SWCNTs. This technique takes advantage of

stronger van der Waals interactions caused by the large con-

tact area at the step edge. SWCNTs have been grown on mis-

cut C-plane sapphire wafers; they grew along the 0.2 nm high

atomic steps of the vicinal a-Al2O3 [235]. A wake-growth

mechanism has been proposed in which the catalyst nano-

particle slides along the atomic step and leaves the growing

SWCNTs behind. Aligned SWCNTs were also reported to grow

on miscut single-crystal quartz substrates [236]. However, the

authors themselves could not confirm whether the step-edge

or the lattice directs the growth. Artificial step structures can

be created to guide the growth direction. The surface geome-

try can be modified to choose the nanotube orientation.

Yoshihara et al. [237] grew SWCNTs along trenches they cre-

ated on Si/SiO2 wafers. However, Maehashi et al. [238] found

that the SWCNTs grew along the edge of terraces and claimed

that CasimirPolder interactions direct the growth.

Lattice-oriented growth is the most popular method that

has been widely discussed. Su et al. [239] observed that

SWCNTs only oriented in certain directions on the Si(1 0 0)

and Si(1 1 1) surface. A base-growth mechanism was pro-

posed in which the catalysts

buck minster fuller en discovery and consequences

Documents