Fahlman B.D. Materials Chemistry

Подождите немного. Документ загружается.

Figure 6.76. SEM images and XRD pattern of ZnO nanorod ﬁlms, and photographs of water droplet

shapes on the aligned ZnO ﬁlms before (left) and after (right) UV illumination. Bottom: the reversible

superhydrophobicity/superhydrophilicity character of the ﬁlms is demonstrated by cycling with UV

irradiation/dark storage. Reproduced with permission from J. Am. Chem. Soc. 2004, 126, 62. Copyright

2004 American Chemical Society.

548 6 Nanomaterials

carbon growth initiates at approximately same time from multiple metal sites.

Subsequent growth is self-directed into a parallel array by interactions amongst

neighboring CNTs.

For some applications, a main drawback of CVD is that MWNTs are often

generated alongside SWNTs. Since the size of the catalyst governs the diameter of

resultant tubes, surface-immobilized nanoclusters of iron oxide@PAMAM dendri-

mer, with well-deﬁned sizes, has been shown to yield only SWNTs with an

extremely narrow diameter distribution.

[201]

Two advanced CVD proce sses,

Figure 6.77. SEM image of ‘NanObamas’ comprised of vertically-aligned carbon nanotubes. Image

provided courtesy of Prof. John Hart, Dept. o f Mechanical Engineering, Univ. of Michigan.

−

− − − − −

−

− − −

− − − −

+ + + + +

−

−−−−

−

−−

−

−−−−

−

−−

−

+++

− − −

−

quartz

repeating

the LBL

process

plasma

PSS

CNT growth

iron colliod

deposition

COOH

reducing

Figure 6.78. Schematic and cross-section SEM image of a layer-by-layer (LbL) approach to deposit iron

nanoclusters onto a surface to yield vertically aligned SWNTs. Reproduced with permission from Liu, J.;

Society.

6.3. Nanoscale Building Blocks and Applications 549

CoMoCAT

(ﬂuidized-bed CVD) and HiPCO

(high-pressure CO CVD), have

recently been developed for the commercial production of SWNTs (Figure 6.79).

Though the experimental setup of these methods are signiﬁcantly more complex

than standard hot-walled CVD, these techniques are still considered an extensi on of

Figure 6.79. Illustration of two methods used for the commercial production of SWNTs. Shown are

(a) the CoMoCat ﬂuidized bed method using CO as the precursor and a Co/Mo bimetallic catalyst,

[204]

and

(b) the HiPco “ﬂoating catalyst” process using the thermal decomposition of iron pentacarbonyl at

pressures of 1–10 atm.

[205]

550 6 Nanomaterials

CVD, as the precursor is decomposed on the surface of the catalyst. Most

noteworthy about these and o ther recent CVD methods is that they offer the ability

to ﬁne-tune resultant lengths, diameters, and even chiralities

[202]

by varying the

operating conditions. Think of the possibilities for applications, as it would not be

long until you can place an order such as: “10 g of pure (5,5) SWNTs with an

average diameter of 7 nm and length of 500 nm.” However, this degree of control

will only be possible once the exact growth mechanism is known, rather than current

efforts that invoke “blind” modiﬁcations of experimental variables.

[203]

Growth mechanisms of carbon nanotubes

Though CNT growth dates back to the early 1990s, details of the growth mechanism

are still unclear. The most commonly accepte d description is based on the vapor-

liquid-solid (VLS) model, which was ﬁrst proposed for the growth of semiconductor

whiskers (Figure 6.80).

[206]

This model assumes that the catalyst liqueﬁes , which

then acts as a preferential adsorption site for gaseous precursors. Subsequent growth

of the nanotubes/nanowires occurs by supersaturation of the catalyst droplet, and

precipitation at the liquid–solid interface.

[207]

However, the amount of carbon in a

supersaturated catalyst nanocluster will never be sufﬁcient to account for the length

or number of nanostructures observed in practice. As a result, additional precursor

atoms must be supplied continuously to the catalyst nanocluster in order for sus-

tained growth to occur.

VAPOR

111

SILICON

CRYSTAL

SILICON SUBSTRATE

Au-Si LIQUID

ALLOY

Figure 6.80. The original schematic used to describe vapor–liquid–solid (VLS) growth of semiconductor

nanowires. Reproduced with permission from Wagner, R. S.; Ellis, W. C. Appl. Phys. Lett. 1964, 4, 89.

6.3. Nanoscale Building Blocks and Applications 551

By examining the phase diagram for a binary system (precursor and catalyst

species), one can easily determine the best temperature range for VLS growth –

any value above the eutectic temperature, where the cat alyst remains a liquid

(Figure 6.81). As we saw earlier for gold nanoparticles, as the diameter decreases,

the melting point is signiﬁcantly decreased relative to the bulk solids. As a result,

“size-corrected eutectics” must be determined for the speciﬁc nanopart icle diameter

in order to determine experimental conditions for nanowire growth.

The VLS mechanism is generally sufﬁcient to model the growth of nanowires; in

addition, nanowire growth may occur in solution via solution-liquid-solid (SLS)

[208]

or via solid-liquid-solid

[209]

via simple annealing of substrate-immobilized nano-

clusters (Figure 6.82). However, the VLS route is som ewhat controversial in its

description of CNT growth, since there are many recent reports of CNTs being

Figure 6.81. Silicon nanowire growth from a gold nanocluster catalyst. Shown is (a) the phase diagram

for the Au/Si system, showing the eutectic temperature/composition; (b) SEM image; and (c) high-

resolution TEM image of the nanowires grown at a temperature of 450



C. The dark tip of the nanowire

is from the gold nanocluster. Reproduced with permission from Hu, J.; Odom, T. W.; Lieber, C. M. Acc.

552 6 Nanomaterials

grown at temperatures below the size-corrected eutectic,

[210]

or from non-melting

nanoclusters such as SiO

,Al

, etc.

[211]

Though small nanoparticles may exhibit

“ﬂuid-like” behavior by altering their surface geometry, empirical data has shown

that the catalyst particle may remain crystalline, rather than an amorphous liquid

phase, duri ng growth. This suggests that nucleation may be dominated by the

catalyst surface – not unlike most heterogeneous catalysis processes (e.g.,

Ziegler–Natta polymerization).

[212]

As with any complicated system, it is much easier to study growth mechanisms

using computational techniques. For CNT growth, it has been shown that small

graphitic islands form on the supersaturated catalyst surface. When the island cover s

half of the catalyst particle, it lifts off and forms the SWNT – with the same diameter

as the catalyst. Based on empirical data and theoretical predictions,

[213]

the catalytic

ability of the metal/alloys for SWNT growth follows the order: Ni/Mo > Ni/Cr >

Figure 6.82. Schematic illustrations of (a) VLS, (b) SLS: solution-liquid-solid, and (c) SLS: solid-liquid-

solid 1-D nanostructural growth.

6.3. Nanoscale Building Blocks and Applications 553

Ni/Co > Ni/Pt > Ni/Rh > Ni/Fe > Ni > Fe/Mo > Fe/Cr > Fe/Co > Fe/Pt > Fe/Rh

> Fe > Ni/Mo > Fe/Mo > Co/Mo > Co > Pt > Cu. Depending on whethe r the

mechanism proceeds via VLS or surface-addition mechanism, the above order may

be related to the ease of carbide formation and carbon diffusion through the

nanocluster interior, or the morphology/composition of the catalyst surface and

rate of carbon diffusion on the catalyst surface, respectively.

[214]

Figure 6.83

shows some recent mechanistic proposals, based on experimental and theoretical

data. The VLS-based proposals (Figure 6.83, top and middle) illustrate the following

basic steps:

(i) At initial growth stages, carbon dissolves in the (semi)-molten “liquid-like”

catalytic nanocluster .

[216]

Calculations indicate that there is a dynamic process

of carbon precipitation onto the catalyst surface and re-dissolution, until a

highly supersaturated catalyst is obtained.

(ii) Carbon precipitates on the surface of the highly supersaturated catalyst

nanoclusters, forming carbon strings/polygons. This causes a decrease in the

dissolved carbon concentration.

(iii) The carbon nuclei form graphitic islands on the surface of the catalyst, which

aggregate into larger graphitic clusters.

(iv) At low temperatures, the graphitic islands are not able to lift off the catalyst

surface, resulting in graphite-encapsulated metal nanoclusters.

[217]

(v) At relatively high temperatures (ca. 500–1,200



C), when the diameter of the

island becomes ca. 1/2 that of the catalyst, the graphitic nucleus lifts off the

catalyst surface to form the SWNT endcap. Subsequent graphitization and

growth propagation of the SWNT may occur through two routes (Figure 6.83 –

middle):

(v.1) “Root (base) Growth” (c–d): carbon atoms migrate through the catalyst

nanocluster and precipitate from the liquid-like catalyst to the open end

of the growing CNT. This route has recently been conﬁrmed for

MWNT growth via in situ TEM measurements, from Fe

C catalyst

nanoclusters (Figure 6.84 – top) .

[218]

Recent theoretical

[219]

and empir-

ical

[220]

studies have addressed the question of how growth occurs

at the atomic level, described by a ‘screw-dislocation-like’ (SDL)

mechanism (Figure 6.85).

[221]

(v.2) “Tip (folded) Gr owth” (e–g): carbon atoms precipitate directly onto the

catalyst surface, and are added to the graphitic endcap. This route has

not been demonstrated for CNTs, but has been shown to occur for

carbon nanoﬁber (CNF) growth using in situ TEM measurements

(Figure 6.84 – bottom).

[222]

In contrast, a surface-governed route is characterized by negligible carbon dissolu-

tion in the catalyst bulk. Since the growth is not propos ed to occur through

supersaturation/precipitation, any species that are chemisorbed onto the catalyst

surface will have a dramatic inﬂuence on SWNT growth. Figure 6.83 (bottom)

shows a proposed route for a surface-based SWNT growth mechanism, which

progresses through the continual addition of C

units onto the leading edge of the

554 6 Nanomaterials

Low Temperature

Intermediate

Temperature

High

Temperature

Supersaturated

Highly

Supersaturated

Unsaturated

(a)

(b)

(c)

(d)

(f)

(g)

(e)

Initial tube

Nucleation stage Growth stage

(A)

(B)

(C)

(D)

(E)

Figure 6.83. Comparison of a VLS-based mechanism (top and middle) and a surface-mediated route

[215]

(see text for details).

6.3. Nanoscale Building Blocks and Applications 555

Figure 6.84. Nucleation and growth of a MWNT from a nanoparticulate catalyst (NPC) on a substrate.

(a) Graphene layers are formed on a NPC and then a MWNT is suddenly expelled from the deformed

NPC. The recording time is shown in images. Copyright: Yoshida, H.; Takeda, S.; Uchiyama, T.; Kohno,

H.; Homma, Y. Nano Lett. 2008, 8, 2082. (a–d) ETEM image sequence showing a growing CNF in 3:1

at 1.3 mbar and 480



C. The video was recorded at 30 frames/s, and the time of the respective

stills is indicated. Drawings (lower row) indicate schematically the Ni catalyst deformation and C-Ni

interface. Hofmann, S.; Sharma, R.; Ducati, C.; Du, G.; Mattevi, C.; Cepek, C.; Cantoro, M.; Pisana, S.;

Parvez, A.; Cervantes-Sodi, F.; Ferrari, A. C.; Dunin-Borkowski, R.; Lizzit, S.; Petaccia, L.; Goldoni, A.;

Robertson, J. Nano Lett. 2007, 7, 602.

556 6 Nanomaterials

growing nanotube. As indicated in this scheme, the absence of an active catalyst will

result in the formati on of fullerenes rather than SWNTs. In this route, nanotube

growth is proposed to initiate from surface metal atoms adding to the edge of a

polyyne ring, which is subsequently able to add additional carbon atoms.

Whereas path C of Figure 6.83 (bottom) shows the successful addition of C

units

at the growth edge, path D shows the formation of a defect, which leads to tube

closure. It should be noted that this route is 1.3 eV more favorable than route C when

there is no metallic catalyst present, thus explaining the preference for closed-cage

fullerenes. Most likely, this is a consequence of pentagonal defe cts that are formed

when carbon atoms add to the open end of the growing SWNT. Hence, calculations

have shown that catalyst-assisted defect repair is a crucial step in the overall

mechanism (Figure 6.83 – bottom, route E). This is proposed to occur through a

“scooter mechanism,”

[223]

whereby the catalyst atoms diffuse along the open-end

surface of the SWNT, facilitating the formation of hexagons that prevent premature

nanotube closure.

Figure 6.85. Illustration of an axial screw dislocation in a CNT. An achiral zigzag (n, 0) tube (a) can be

viewed as a perfect crystal, and transformed into a chiral one by cutting, shifting by a Burgers vector

(red arrows in b–d), and resealing a tube-cylinder (b). The free-energy proﬁle shown in (e) illustrates

the difference in growth of a chiral or achiral CNT with an increasing number of added carbon

atoms. Reproduced with permission from Ding, F.; Harutyunyan, A. R.; Yakobson, B. I. PNAS 2009,

106, 2506.

6.3. Nanoscale Building Blocks and Applications 557