Fahlman B.D. Materials Chemistry

Подождите немного. Документ загружается.

instance, SiO

nanoparticles of ca. 50–100 nm are obtained through the controlled

hydrolysis of a silicon alkoxide (e.g., tetraethyl orthosilicate, TEOS) in a mixture of

ethanol, aqueous ammonia (21%) and water. Particle sizes were controlled by the

addition of an entraining silane such as 3-aminopropyltriethoxysilane that was added

to TEOS prior to hydrolysis.

[151]

Iron oxide nanoparticles may be synthesized by

the dissolution of an iron salt in an acidic solution (Eqs. 12–14).

[152]

2 NaOH þ Hf(OHÞ

! Na

HfO

þ 3H

Oð11Þ

In addition to oxides, a number of other compositions may be synt hesized.

In general, any stoichiometry is “fair game” through the same reactions that one

does in bulk scale. For example, sulﬁdes through reaction of precursors with H

nitrides through NH

exposure, etc.

[153]

Reactions with sodium chalcogenides

represent a safer means to introduce chalcogens to the structure; for instance, the

formation of L-cysteine stabilized Mn

-doped ZnS photoluminescent phosphor

nanoparticles (Eq. 15). The reaction of Eq. 16 shows a clever way to generate

SnO

2−

SnO

2 Na

2(g)

Figure 6.51. Schematic of the growth of tin oxide nanoclusters at room temperature. The TEM images on

the right illustrate dendritic-stabilized nanoclusters using (a) PAMAM and (b) amine-terminated poly

(ethyleneimine) hyperbranched polymer hosts. The latter is a less efﬁcient entraining agent, as observed

by the greater degree of nanocluster agglomeration. Reproduced with permission from Juttukonda, V.;

Paddock, R. L.; Raymond, J. E.; Denomme, D.; Richardson, A. E.; Slusher, L. E.; Fahlman, B. D. J. Am.

518 6 Nanomaterials

S gas in situ, rather than having to handle a cylinder of the dangerous nerve gas.

As long as a suitable entraining agent is used, one is able to control the resultant

size/morphology of the nanostructures. With current advances in compound and

metallic nanoclusters, and the properties of each being ﬁne-tuned via quantum

conﬁnement effects, one can begin to imagine intriguing designs for future micro-

electronic devices.

Step 1:

[Fe(H

OÞ



3þ

$ [Fe(OH)(H

OÞ



2þ

þ H

(pH: 0  2Þð12Þ

Step 2:

2 [Fe(OH)(H

OÞ



2þ

$½ðH

OÞ

Fe(OHÞ

Fe(H

OÞ



4þ

þ 2H

O (pH: 2  3Þ

ð13Þ

Step 3:

½ðH

OÞ

Fe(OH)

Fe(H

OÞ



4þ

$ isopolyoxo - cations

$ Fe

 xH

O (pH : 3  5Þð14Þ

Overall:

ZnSO

+Na

S + (MnCl



cysteine

ZnS(:Mn

2þ

)+Na

+ (Cl



Þð15Þ

C  C

NH

Thiacetamide

þH

O ! H

S+CH

COO



+NH

ð16Þ

Mechanisms for the nucleation/growth/agglomeration of metal nanoclusters

In order to maintain control over the composition and morphology of a 0-D

nanostructure, it is essential that we understand the self-assembly mechanism for

these structures. Only within the last decade have we ﬁgured out ways to repeat-

ably control the morphology and composition of nanoparticles/nanoclusters. This

explains why the mechanistic details of nanocluster growth have not surfaced until

recently. The ﬁrst system to be investigated was the synthesis of metallic iridium

nanoclusters, formed through the hydrogenation of a polyoxoanion-supported Ir

complex.

[154]

Since Ir does not have an observable surface plasmon resonance

proﬁle, the rate of nanocluster growth was determined by following the hydrogena-

tion of alkenes over time. This is possible in this system since the initial complex is

not an active toward alkene hydrogenation; catalytic activity arises only from the

reduced metal (Ir

The overall four-step mechanism for nanocluster growth/agglomeration is shown

in Figure 6.52.

[155]

Although this pathway is based on metallic nanocluster studies,

other types of nanoclusters/nanoparticles would likely follow a similar route.

[156]

The ﬁrst step involves the slow, continuous nucleation of clusters that are much

6.3. Nanoscale Building Blocks and Applications 519

smaller than 1 nm. When an energetically favored critical nucleus size is reached

(ca. 15 atoms for Ir

; varies depending on the metal), additional atoms rapidly

aggregate to the surface – a process referred to as autocatalytic surface growth.

The term “autocatalytic” is used since the nanoclusters formed from the nucleation

step are also reactants for subsequent surface growth. Accordingly, the autocatalytic

step will proceed faster as the reaction progresses, effectively shutting down the

nucleation step – essential to achieve monodisperse nanocluster growth. That is, the

size of the growing nanocluster may be controlled by varying the relative values of

and k

, as well as the availabili ty of additional precurso r molecules, both of which

separate the nucleation and growth events in time. In this fashion, metal nanoclusters

have been designated as “living metal polymers,” analogous to organic living

polymers discussed in Chapter 5.

It is widely believed that Au nanoparticle growth by the Tu rkevich method occurs

via the LaMer model,

[157]

in which the concentration of metal atoms in solution

constantly increases (by decomposition of a precursor compound) until supersatura-

tion occurs. Atomic nucleation then occurs, followed by growth via the adsorption of

additional gold atoms. This may be obser ved experimentally by measuring a sharp

drop in the concentration of metal atoms in solution. Below a certain level of

supersaturation, no further nucleation events will take place, which is required to

effectively control the monodispersity of the resultant nanoparticles. Further growth

A+B

B+C 1.5C

2B C

B n Pt''

n-1

n+m

+ Pt''

+ Pt

bulk

Figure 6.52. Schematic of the four-step mechanism for transition metal (e.g., Pt) nanocluster formation.

Shown are (i) nucleation to a desired cluster size; (ii) autocatalytic growth onto the cluster surface;

(iii) diffusive agglomerative growth of two nanoclusters; and (iv) autocatalytic agglomeration into bulk

metal particulates. Reproduced with permission from Besson, C.; Finney, E. E.; Finke, R. G. J. Am. Chem.

520 6 Nanomaterials

will be accompanied by coarsening or Ostwald ripening, whereby larger partic ulates

grow at the expense of smaller ones. Though this may yield a high degree of

monodispersity, it may also yield bulk metal particles via a second autocatalytic

surface-growth process (step 4, Figure 6.52). A general prediction of the overall

mechanism is that lower concentrations and higher temperat ures are most conducive

to yield nanoclusters rather than bulk metal (i.e., k

>> k

, k

). Indeed, this

prediction is backed up by experimental data.

The sizes of resultant Au nanoparticles may be ﬁne-tuned by varying the experi-

mental conditions

[158]

; in fact, a recent study has even elucidate d a non-LaMer

growth mode via nanowire intermediates (Figure 6.53).

[159]

It should be noted that

in the Turkevich process, the citrate ion serves as a reducing agent by undergoing

oxidative decarboxylation to yield an acetone dicarboxylic acid intermediate. This

species is then proposed to chelate gold [AuCl

]

n

ions as they undergo stepwise

reduction from Au

! Au

[160]

Hence, it appears as though the reduction

of metal ions in solution occurs via the aggregation/nucleation of unreduced ions,

rather than the nucleation of zero-valent atoms that then coalesce into larger metallic

nanostructures.

[161]

Interestingly, it has been shown that the growth of nanoclusters proceeds through

the formation of “magic number” (or closed shell) clusters that exhibit unusual

electronic stability. It should be noted that once the diameter of the cluster is <ca.

10 nm, a signiﬁcant portion of the total atoms lie at its surface, which will result in

different physical/optical properties than exhibited by the bulk metal.

[162]

A fcc or

hcp transition metal will form stable clusters with icosohedral or truncated octahe-

dral (TO; analogous to C

) symmetry (Figure 6.54 ). The number of metal atoms

Figure 6.53. TEM micrographs of various stages of Au nanoparticle synthesis via the citrate reduction of

AuCl



, showing nanowire intermediate formation. The colors of the solutions are: (a) colorless, (b) dark

blue, (c) dark purple, (d) - (e) purple, and (f) ruby-red. Reproduced with permission from J. Phys. Chem.

6.3. Nanoscale Building Blocks and Applications 521

within an icosohedral cluster is given by the Mackay sequence of Eq. 17; i.e., 13, 55,

147, 309, 561, 923, 1,415, 2,057, 2,869, etc.:

þ5n

nþ1;ð17Þ

where n ¼ packing shell (n ¼ 1, 2, 3, 4,. ..). The number of surface atoms in the nth

shell is given by 10n

+2.

[163]

For instance, the ﬁrst stable shell consists of a central

atom surrounded by 12 surface atoms, corresponding to 92% of the total atoms

housed on the surface.

In contrast, the number of total and surface atoms within a TO array are given by

Eqs. 18 and 19, respectively. For instance, the ﬁrst stable cluster of 38 atoms consists

of 32 surface atoms surrounding a core of 6 atoms (84% of the total atoms located on

the surface). Subsequent shells yield stable clusters with 201, 586, 1289, 2406, 4033,

6266, etc. atoms. Further, the number of TO edge atoms is represented by Eq. 20,

corresponding to the most reactive atoms that exhibit the lowest number of nearest

neighbors. Hence, the ﬁrst shell of a TO structure would contain 24 edge atoms –

63% of the total number of atoms in the cluster. A regular TO cluster may be

Figure 6.54. Illustrations of icosahedral and truncated octahedral (TO) clusters. The numbers listed on

the TO faces indicate atomic coordination.

522 6 Nanomaterials

designated by its (n, m) indices with n – m ¼ 0(n, m > 1), where n is the number of

atoms on an edge adjoining (111) facets and m that between adjoining (111) and

(100) faces.

[164]

However, gold nanoclusters with N

> 20 have been shown to

exhibit the irregular TO

structure, with 0 < n – m  4, m > 1. Stable clusters

within the TO

set contain 79, 140, 225, 314, 459, ... metal atoms.

¼ 16ðn þ1Þ

 33ðnþ1Þ

þ24ðnþ1Þ6ð18Þ

¼ 30ðn þ1Þ

 60ðnþ1Þþ32ð19Þ

¼ 36ðn þ1Þ48ð20Þ

It should be noted that intermediate magic-number nanoclusters represent only local

minima in the potential energy surface, relative to the globa l minimum of a bulk

metal with the lowest possible surface area (Figure 6.55).

[165]

The high yield

of magic number nanoclusters is a consequence of kinetically controlled surface

growth. That is, once these favored intermediate structures are formed, they are less

reactive toward autocatalytic surface growth relative to non-magic number clusters.

Thus far, we have considered mechanisms to explain the relat ive sizes of 0-D

nanostructures, without consideration of their shapes. The observed shapes of

nanocrystalline facets may be controlled by either thermodynamics or kinetics.

For those reactions under thermodynamic control, growth will occur in such a way

to minimize the total interfacial free energy, g – i.e., the energy associated with the

nucleation/growth of a surface of a unit area. For a fcc structure, g {110} > g

{100} > g {111}; hence, a single crystal should prefer an octahedral structure to

maximize the numb er of (111) facets. However, as we have seen, a truncated

Figure 6.55. Comparative energy levels for “magic number” noble metal nanoclusters relative to bulk

metal. Also shown are the perfect fcc arrays of the nanoclusters, in which more than 75% of the atoms are

located on the surface. Adapted with permission from Finke, R. G. in Metal Nanoparticles: Synthesis,

Characterization, and Applications Feldheim, D. L.; Foss, C. A. eds., Dekker: New York, 2002.

6.3. Nanoscale Building Blocks and Applications 523

octahedron (TO) structure is often observed since a cube will have a lower g than an

octahedron of the same volume.

In contrast, to achieve kinetic control over the resultant nanostructure, one must

generally slow down the precursor decomposition/reduction while taking advantage

of Ostwald ripening. Once the atomic nucl eation and growth becomes slow, the

nanostructures will exhibit random hexagonal close-packing in addition to a variety

of stacking faults, leading to thermodynam ically-disfavored shapes (Figure 6.56).

Further, if twinned regions are present rather than single-crystal nuclei, one is able

Figure 6.56. Reaction pathways that give rise to varying shapes of fcc metal nanocrystals. First, a precursor

is reduced or decomposed to form the nuclei (small clusters). Once the nuclei have grown past a certain size,

they become seeds with a single-crystal, singly twinned, or multiply twinned structure. If stacking faults

are introduced, then plate-like seeds will be formed. The green, orange, and purple colors represent the {100},

{111}, and {110} facets, respectively. Twin planes are delineated in the drawing with red lines. The parameter

R is deﬁned as the ratio between the growth rates along the <100> and <111> directions. Reproduced

524 6 Nanomaterials

to guide the morphology into 1-D or 2-D arrays via selective oxidation of speciﬁc

crystal facets, or by the prefere ntial adsorption of entraining agents to

speciﬁc regions of the growing nanocrystal.

[166]

Self-assembly of 0-D nanostructures into arrays

Now that you understand how 0-D nanostructures are synthesized and stabilized, it

is worthwhile noting how these structures are aligned into more complex arrays.

As you recall, the “bottom-up” approach to materials design features the purposeful

placement of individual nanoarchitectures, in order to build speciﬁc functional

devices one unit at a time. Though one would have the greatest control over the

device properties through the growth/arrangement of individual atoms, current syn-

thetic methodology most easily yields nanoclusters/nanoparticles that consist of small

groups of atoms (N

atoms

~ 50+). Advanced lithographic techniques such as dip-pen

nanolithography (DPN), micro/nano-contact printing, focused-ion beam (FIB), and

deposition of nanoparticles within lithographically generated surface features have all

been used to pattern 0-D nanostructures into regular 2-D and 3-D arrays.

[167]

As we have seen, a stabilizing agent is used to prevent agglomeration of growing

0-D nanostructures. In fact, this component also provides an effective “handle”

to bind the nanostructure to a particular surface. Once appropriate reactive groups

(e.g., —OH, NH

) are placed on a surface through monolayer formation, the stabi-

lizing group surrounding the nanostructure spontaneously becomes chemisorbed

(Figure 6.57). If the nanostructures are encapsulated with stabilizing agents of

well-deﬁned sizes, then the spacing between adjacent nanostructures will also be

highly ordered and predictable (Figure 6.58). Though this most often results in a

2-D matrix of nanostructures, it is also possible to create alignment into 1-D chains.

For instance, Au nanostructures that are stabilized by a long-chain thiolated poly

(ethylene oxide) polymer form a linear array via interactions with sulfate/acid groups

of a polysaccharide compound (chondroitin sulfate c, IX) (Figure 6.59). In addition to

chemisorptive methods, light actuation (e.g., optoelectronic tweezers, OET) and elec-

trostatic forces generated from biased p-n junction-patterned substrates have also been

an effective means to align metallic nanoparticles.

[168]

6.3. Nanoscale Building Blocks and Applications 525

TiO

Ti(On-Bu)

Hydrolysis

Gold

Nanoparticles

n -Bu

-Bu

Figure 6.57. Illustration of the chemisorption of a nanostructure onto a hydroxylated TiO

surface. The

hydroxyl groups on the periphery of the nanostructure stabilizing agent provide the handle for surface

adsorption. Reproduced with permission from Liu, S.; Zhu, T.; Hu, R.; Liu, Z. Phys. Chem. Chem. Phys.

Figure 6.58. Schematic of the controlled spacing between individually stabilized gold nanoparticles by

the formation of fullerene inclusion complexes. Reproduced with permission from Yonezawa, T.;

526 6 Nanomaterials

An important precedent for the self-assembly of 0-D nanostructures into a 3-D

superlattice array was reported by O’Brien and coworkers.

[169]

In this work, oleic

acid stabilized g-Fe

nanoclusters were synthesized by the decomposition of Fe

(CO)

and subsequent oxidation using trimethylamine N-oxide within a high boiling

solvent (trioctylamine, b.p. 365



C). Depending on the relative size ratios and

concentrations of Fe

:PbSe QDs, AB

or AB

superlattice arrays were generated

(Figure 6.60) – isostructural to intermetallic compounds such as NaZn

and AlB

respectively. The superlattices were made by simply placing a substrate in a

suspension containing the two types of nanoparticles. Upon evaporation of the

solvent in a low-pressure chamber, the nanoparticles self-assembled into the ordered

arrays. Binary superlattices of this type that feature both magnetic and semiconduct-

ing QDs may have appl ications for data storage applications, whereas arrays of 2+

semiconductor QDs may be exploited for next-generation solar or thermoelectric

devices.

Figure 6.59. TEM images of a 1-D array of Au nanoparticles formed through the interaction of PEO and

polysaccharide chains. Reproduced with permission from Takagi, K.; Ishiwatari, T. Chem. Lett. 2002,

6.3. Nanoscale Building Blocks and Applications 527