Fahlman B.D. Materials Chemistry

Подождите немного. Документ загружается.

Figure 6.60. Top: TEM micrographs and sketches of AB

superlattices (isostructural with intermetallic

phase NaZn

) of 11-nm g-Fe

and 6-nm PbSe NCs. Bottom: TEM micrographs and sketches of AB

superlattices (isostructural with intermetallic phase AlB

) of 11-nm g-Fe

and 6-nm PbSe NCs.

Reproduced with permission from Redi, F. X.; Cho, K. -S.; Murray, C. B.; O’Brien, S. Nature 2003,

528 6 Nanomaterials

The most popular method used to achieve 3-D arrays such as layered nanostructural

thin ﬁlms, is the layer-by-layer (LbL) self-assembly pioneered by Decher in the early

1990s.

[170]

This technique is based on the sequential adsorption of species with

complimentary functional groups (e.g., ionic charges) on a variety of substrates.

There are many device applications for LbL thin ﬁlms, such as self-cleaning surfaces,

surface deactivation of warfare agents, solar energy, drug delivery, and optoelectron-

ics. An interesting recent precedent for LbL nanostructural ﬁlm growth uses sequential

layering of cyclodextrin-stabilized Au nanoparticles and adamantyl-terminated

dendrimers (Figure 6.61).

[171]

The ﬁlm thickness was reported as 2 nm per bilayer,

allowing for strict control over the resultant ﬁlm thickness.

6.3.2. One-Dimensional Nanostructures

The second class of nanoscale building blocks, referred to as 1-D nanostructures, is

reserved for those materials that have nanoscale dimens ions that are equivalent

in all but one direction. Recall that a 0-D nanostructure is analogous to the period

following this sentence (length ¼ width); a 1-D nanostructure is analogous to the

number “1” (length > width).

Since we began the discussion of 0-D nanostructures with nomenclature, we will

follow suit in this section. Once again, it is easy to be confused by the common

synonymous use of the terms nanotube, nanoﬁber, nanowire, and nanorod. However,

if you think of the analogous bulk materials without the preﬁx “nano,” there should

be no ambiguity regarding the proper use of these descriptors (Figure 6.62). The

common thread among all of these structures is that their diameters must be within

the 1–100 nm range; typically, their lengths are within the micron (or larger) regime.

Figure 6.61. Illustration of LbL growth of cyclodextrin-stabilized Au nanoparticles (dark spheres) and

adamantyl-terminated PPI dendrimers. Reproduced with permission from Crespo-Biel, O.; Dordi, B.;

Society.

6.3. Nanoscale Building Blocks and Applications 529

Figure 6.62. Classiﬁcations of 1-D nanostructures. The top portion shows SEM images of various

nanostructures.

[172]

Whereas the morphology of nanowires, nanotubes, and nanoﬁbers look identical by

SEM, nanorods are notably different, with much shorter lengths and straight sidewalls. The bottom

portion shows high-resolution TEM images, which provide morphological details of the

nanostructures.

[173]

The top TEM image shows crystalline nanorods grown on the surface of amorphous

nanoﬁbers. The bottom two TEM images illustrate the difference between nanowires/nanotubes – the

latter contains a hollow core. It should be noted that crystalline nanorods may also be termed

“nanocrystals,” as their morphology resembles that of needle-like bulk crystallites.

530 6 Nanomaterials

A nanotube is a 1-D structure that contains a hollow core, whereas the other three

nanoarchitectures are solid throug hout. The term nanoﬁber should be reserved for

1-D nanostructures that are amorphous (and usually nonconductive) such as poly-

mers and other non-graphitized carbonaceous structures. By contrast, a nanowire

designates a structure that is crystalline, with either metallic or semiconductive

electrical properties.

A nanorod is typically a crystalline 1-D nanostructure, with an overall length

comparable to its width (i.e., both dimensions are < 100 nm). As their name implies,

another feature of nanorods is their rigid sidewall structures. However, since crystalline

nanorods exhibit the same overall shape as needle-like bulk crystals, the term “nano-

crystal” is probably more appropriate for these structures (or, more explicitly: “rod-like

nanocrystals”). Whereas nanowires, nanoﬁbers, and nanotubes exhibit an interwoven

array, nanorods are completely linear in morphology. As such, nanorods are capable of

stacking onto each other to yield interesting 2-D and 3-D arrays – not usually as easy to

perform with the “spaghetti-like” morphology of the other 1-D nanostructures.

Carbon nanotubes (CNTs)

Without questio n, the most widely studied 1-D nanomaterial is the carbon nanotube

(CNT). These structures were ﬁrst discovered by Iijima in 1991,

[174]

and may be

thought of as a graphitic sheet(s) of sp

hybridized carbon atoms (i.e., graphene

[175]

)

rolled into a tubular array. Based on the layers of graphene sheets that comprise the

CNT, the structures are designated as single-walled, double-walled, or multiw alled

nanotubes (SWNTs, DWNTs, or MWNTs, respectively – Figure 6.63). The dia-

meters of CNTs range from 1 nm (SWNTs) to >30 nm (MWNTs), with aspect ratios

(length:width) ranging from 100 to greater than 1  10

. Even though the diameters

of CNTs are orders of magnitude smaller than a human hair, their tensile strength is

ca. 20 times greater than steel – a property attributed to extremely strong sp

bonding between neighboring hexagonal units.

[176]

The electrical conductivity of SWNTs may vary from metallic to semiconducting,

depending on the way a graphene sheet is folded (Figure 6.64). In particular, the

diameter and helicity of a SWNT are uniquely characterized by the chirality vector

(or Hamada vector),

C, which connects crystallographically equivalent graphene

lattice sites. Vector indices designated by (n, m) are used to indicate the direction

and length of the chirality vector (Eq. 21). Equation 22 provides a formula to

calculate the diameter, D,ofa(n, m) nanotube, and Eq. 23 is the relation to

determine the chirality angle, f. When n ¼ 0, the SWNT is denoted as the zigzag

conformation; when n ¼ m, the SWNT is in its armchair form. For all values in

between these extremes, the nanotubes are designated simply as chiral .

C ¼n

þm

ð21Þ

D¼

ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

CC

ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

þnmþm

ð22Þ

6.3. Nanoscale Building Blocks and Applications 531

cos f ¼

2n þ m

ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

þ m

þ nm

ð23Þ

where n, m are the integers denoting the number of unit cell vectors along two

directions in the crystal structure of graphene,

;

are the graphene unit cell

vectors, and d

C–C

is the C—C bond length (1.41–1.44 A

For metallic SWNTs, the electrical conductance may exceed silver or copper by

three orders of magnitude. Calculations have shown that (n, 0) or zigzag SWNTs

exhibit metallic conductivity when n/3 is an integer, and semiconducting properties

for all other values of n. Similarly for chiral SWNTs, when (2n + m)/3 is an integer ,

the tubes are metallic (otherwise are semiconducting). Finally, armchair SWNTs,

with n ¼ m, exhibit metallic conductivity (Figure 6.65). Most importantly, elec-

tronic band structure calculations show that metallic and semiconducting CNTs are

dependent only on (n, m) – that is, slight structural variations result in dramatic

Figure 6.63. TEM images of MWNTs, formed from the folding (a) 5-, (b) 2- (i.e., DWNT), and

distance between adjacent planes in graphite. Reprinted from Dresselhaus, M. S.; Dresselhaus, G.;

Elsevier.

532 6 Nanomaterials

Figure 6.64. Illustration of the honeycomb 2-D graphene network, with possible unit cell vector indices

(n, m). The dotted lines indicate the chirality range of tubules, from y ¼ 0



(zigzag) to y ¼ 30



(armchair). For y values between 0 and 30



, the formed tubules are designated as chiral SWNTs. The

electrical conductivities (metallic or semiconducting) are also indicated for each chiral vector. The

number appearing below some of the vector indices are the number of distinct caps that may be joined

to the (n, m) SWNT. Also shown is an example of how a (5, 2) SWNT is formed. The vectors AB and

, which are perpendicular to the chiral vector (AA

) are superimposed by folding the graphene sheet.

Hence, the circumference of the SWNT becomes the distance between AB and A

axes. Reprinted from

Dresselhaus, M. S.; Dresselhaus, G.; Eklund, P. C. Science of Fullerenes and Carbon Nanotubes.

6.3. Nanoscale Building Blocks and Applications 533

changes in their electronic properties. For example, the bandgap of semiconducting

SWNTs may be ﬁne-tuned from ca. 10 meV to 1 eV – with no required addition of

dopants, unlike bulk Si counterparts (Eq. 24). If semiconducting nanotubes exhibit

the same chirality, the bandgap is inversely proportional to the diameter, D (e.g.,

(7, 0) > E

(10,0)). The density of states (DOS) for SWNTs exhibit sharp

maxima, which are known as van Hove singularities resulting from quantum

conﬁnement in directions perpendicular to the tube axis. Optical transitions are

possible between these maxima, which may be observed by Raman spectroscopy

(Figure 6.66).

[177]

Armchair nanotube

Zigzag nanotube

(5,5)

(7,0)

metal

Semi-

conductor

Density of States

−10.0 −5.0 0.0 5.0 10.0

Energy (eV)

Density of States

−10.0 −5.0 0.0 5.0 10.0

Ener

y (eV)

Figure 6.65. Comparison of the density of states, and resultant electronic properties of armchair and

zigzag SWNTs. Reproduced with permission from Charlier, J.-C. Acc. Chem. Res. 2002, 35, 1063.

534 6 Nanomaterials

3:13 eV ðd

cc

ð24Þ

The tunable electronic properties of CNTs are being explored for next-generation IC

architectures. As you may recall from Chapter 4, traditional Si-based microelec-

tronic devices will likely reach a fundamental limit within the next decade or so,

necessitating the active search for replacement materials. Accordingly, an area of

intense investigation is molecular electronics – in which the electronic device is

built from the placement of individual molecules.

[178]

Not surprisingly, the inter-

connects of these devices will likely be comprised of CNTs and other (semi)

conductive 1-D nanostructures such as nanowires.

Since CNTs have a high electrical conductivity and contain sharp tips, these

nanomaterials are the best-known ﬁeld-emitter

[179]

of any material to date. In

general, the smaller the radius of curvature of the tip, the more concentrated the

electric ﬁeld will be, which corresponds to increased ﬁe ld emission at low required

voltages. This property is currently being exploited for the design of ﬂat-panel

ﬁeld emission displays. Though plasma and LCD displays are hot ticket items

at electronics stores, they both possess inherent disadvantages. Plasma displays

are extremely heavy, consume a signiﬁcant amount of energy, and are prone to

Figure 6.66. Density of states in a semiconducting single-walled carbon nanotube. The sharp maxima

are van Hove singularities. The dominant optical absorption/emission transitions are indicated by arrows.

Reproduced with permission from Weisman, R. B.; Subramoney, S. Electrochem. Soc. Interface, Summer

2006, 42.

6.3. Nanoscale Building Blocks and Applications 535

“burn-in,” which permanently degrades the screen. On the other hand, LCD screens

are expensive to produce and often lack the response time required to view

fast-paced sporting events/movies without blur. In contrast to popular belief, the

new ‘LED TVs’ available at the high end of the HDTV market are not truly LEDs.

These displays are in fact LCDs that simply use LEDs instead of cold cathode

ﬂuorescent lamps as the backlight; in contrast, true LED TVs would not require a

backlight, since each LED pixel would be self-illuminating.

[180]

Hence, as far as

picture quality is concerned, traditional CRT displays are still among the ﬁnest

quality. The replacement of this technology with CNTs is a logical step in the

evolution of display panels. Rather than a single electron gun, CNT-based screens

will contain a separate nanotube electron gun for each indivi dual pixel in the

display – dramatically enhancing the resolution and clarity of the picture. Further,

in contrast to current large ﬂat-panel televisions, the overall weight of CNT-based

analogues will be signiﬁcantly lower, and they will consume far less power. This

concept has already been proven in prototypes

[181]

and is scheduled to reach the

commercial market wi thin <5 years.

Another electronic application for CNTs is for next-generation ﬁeld-effect

transistor (FET) designs. The “proof-of-concept” for CNTFETs was demonstrated

in the late 1990s, with a simple bridging of two noble metal electrodes with a

SWNT (Figure 6.67 – top). However, the electrical characteristics of this new

FET design were less than desirable, with high contact resistanc e (>1MO) and

low drive currents. Researchers at IBM have since modiﬁed the original design

wherein the semiconducting SWNTs are placed directly onto an oxidized Si

wafer, followed by the deposit ion of the source and drain (Co or Ti) electrodes

(Figure 6.67 – bottom). Through subsequent annealing, a stronger interaction is

afforded between the electrodes and CNT channel, which reduces the contact

resistance.

Though CNTFETs are in a relatively early stage of development, Avouris at IBM

provided a recent comparison of the output from a top-gate CNTFET (Figure 6.67 –

bottom) and existing Si-based FETs.

[182]

For CNTs with an average diameter of

1.4 nm, the ON current for a top-gated CNTFET is on the order of 2,100 mAmm

1

(drain voltage) ¼ V

(gate voltage)–V

(threshold voltage) ¼ 1.3 V. In com-

parison, the highest drive current in a p-CMOS under the same conditions is 650

mAmm

1

for a gate length of 50 nm. The transconductance (ratio of the output

current variation to the input voltage variation) of the CNTFET is 2,300 mSmm

1

;

the value for an analogous Si p-CMOS is 650 mSmm

1

. Hence, the values obtained

thus far show that CNTFETs outperform Si-based FETs. As improvements continue

to be made to the design of nanotube-based transistors, this technology should be

ready for the market just as the “glass ceiling” of Si CMOS devices is reached. It

should be noted that the extremely efﬁcient thermal conductivity of CNTs will also

be exploited for cooling applications for future computers – of increasing concern as

the chip density continues to soar.

In addition to the above tunable conductive properties, CNTs are the strongest and

stiffest materials known to date (Table 6.6). The hollow, closed morphology of

536 6 Nanomaterials

nanotubes results in rather intriguing deformation modes in response to a mechanical

stress (Figure 6.68). In particular, it has been suggested that Stone–Wales defects may

become mobile under stress, which results in a change in tube diameter/chirality.

[183]

This also causes a change in the electronic properties of the CNT, opening up

possibilities for sensor applications. However, for MWNTs, studies have shown that

only the outer graphitic shell is able to support stress,

[184]

and SWNT bundles/ropes

(Figure 6.69) exhibit smaller Young’s moduli, relative to isolated SWNTs, due to weak

intertube cohesion.

[185]

Hence, although an individual SWNT has an elastic modulus of

1.2 TPa, the value for bundles is ca. 100 GPa for diameters in the range of 15–20 nm.

To improve the load distribution throughout the CNT and improve its mechanical

strength, there are efforts to create crosslinks between individual shells of MWNTs,

and among SWNTs within ropes.

The desirable mechanical properties of CNTs have been exploited in recent years

for the structu ral reinforcement of polymers.

[186]

This is a natural extension of

−1.6 −0.8 0.0 0.8

−1.1 V

−0.9 V

−0.7 V

−0.5 V

−0.3 V

−0.1 V

−1.6 −1.2 −0.8 −0.4 0.0

−5.0

−4.0

−3.0

−2.0

−1.0

0.0

Id [μA]

[A]

[V]

− V

−10

−6

−10

−7

−10

−8

−10

−9

−10

−11

[V]

= −0.6 V

Figure 6.67. Atomic force microscope (AFM) topographical image of an original CNTFET (top). The

source/drain electrodes were fabricated on a SiO

thin ﬁlm grown on a Si wafer. The wafer itself served as

the gate electrode. Also shown (bottom) is an illustration of a top-gated CNTFET design, with an oxide

thickness of 15 nm and CNT length of 300 nm. The device electrical characteristics at room temperature is

also provided. The inset shows the transfer characteristic of the FET (output voltage of the device as a

function of the input voltage). Reproduced with permission from Avouris, P. Acc. Chem. Res. 2002, 35,

6.3. Nanoscale Building Blocks and Applications 537