Ultrathin Nanowires - A Materials Chemistry Perspective

Ultrathin Nanowires - A Materials Chemistry Perspective

(Parte 1 de 3)

Ultrathin Nanowires—A Materials Chemistry Perspective

By Ludovico Cademartiri, and Geoffrey A. Ozin*

1. Introduction: Why Ultrathin Nanowires?

Theterm‘‘ultrathin’’isunscientific,asitdoesnotexplicitlyspecifya size range. Ultrathin nanowires are considered to be less than 10nm in diameters in most of the literature; we will see, though, that an especially large interest is emerging in nanowires below 2–3nm. The attractive characteristics of ultrathin nanowires are summarized in Figure 1. One aspect correlated with decreased diameter is the increased surface area, which in turn, has immediate repercussions on sensors,[7] catalyst supports,[8] and environmental-remediation applications.[9,10]

Less obviously, the decrease in size of the nanowires leads to a certain degree, to an increased colloidal stability: the gravitational force is easier to counteract by applying colloidal-stabilization strategies involving coulombic repulsion or steric stabilization.

This important characteristic potentially allows the inorganic ultrathin nanowires to outperform carbon nanotubes (CNTs) in a variety of applications that involve solution processibility, such as microfluidics[1] and paint-on electronics; CNTs are known to be scarcely dispersible in solvents. Colloidal stability also allows the study of certain properties of nanowire synthesis (like absorption cross-sections) in a more comfortable way.[12,13]

Closely related with their potential colloidal stability is the possibility of designing the surface chemistry of the nanowires by ligand exchange; while this is also true for thicker nanowires, the ultrathin diameter amplifies the importance of this possibility. Bioconjugation techniques to employ nanowires for in vivo biodetection[14] or as regeneration scaffolds[15] can now be imagined, together with the tuning of their surface charge, to create layer-by-layer films of nanowires with nanometer-controlled thickness.[16]

Nevertheless, what really inspires a large number of scientists is the possibility of synthesizing, stabilizing, and isolating large numbers of ultrathin nanowires with the unprecedented structures predicted by theory.[17] Most of the work has been performed on metallic nanowires, from elements such as Al,[17] Pb,[17,18] Bi,[19] Si,[20–2] Rh,[23] Ag,[24] Cu,[25,26] and Au.[27,28] In the case of Al, Pb, Au, Cu, Rh, and Bi, wires below a certain size have been found to preferentially adopt a multiwalled architecture. The stability of the outer surface determines the stability and properties of the entire structure; this explains the theoretical findings that surface vacancies on such wires would migrate to the core,[25] that melting would first occur in the core,[28] or that the core might be entirely missing, leading to a tube architecture.[29] The outer surface of the wires is predicted to show, in certain ranges of diameter, a helical twist, which would give an unprecedented chirality to the nanowires,[17] similarly to CNTs. Preliminary experimental evidence for such helicity has been obtained in Au nanowires[30] and Pt nanotubes.[29]

While there are a few reports of single ultrathin metallic wires and tubes suspended between electrodes in ultrahigh vacuum conditions,[29,30] there is the necessity to access such structures in a much less demanding way. It is possible that solution-phase syntheses will soon allow obtention of that size regime, where new structures emerge. One concern is that the ligand strategies w.advmat.de

This paper is dedicated to Geoff Ozin on the occasion of his 65th birthday

Materials Chemistry Research Group Department of Chemistry University of Toronto 80 St. George Street, Toronto Ontario M5S 3H6 (Canada) E-mail: gozin@chem.utoronto.ca

[+] Present address: Dept. of Chemistry and Chemical Biology, Harvard University, 12 Oxford Street, Cambridge, MA 02138, USA.

DOI: 10.1002/adma.200801836

The recent years have seen an explosiveinterest in one-dimensional nanostructures,[1] as testifiedby the numberof citationsthis fieldhasaccrued;as customary, its blossomingwas enabledby chemicalbreakthroughsthat allowed the reproducible and affordable synthesis of such structures.[2,3] The limitations of those syntheses was in the diameter of the nanowiresthat it could produce (hardly<10nm), and in the use of expensive and low-yield techniques,such as chemical vapor deposition(CVD). This paper attempts to summarize the very recent chemical breakthroughs that have allowed the production of ultrathin nanowires, often in solution, and often in gram-scale quantities. By no means is this a comprehensive coverage of the field, which can in part be found in other excellent reviews,[1,2,4–6] but a selection of those contributions that we feel would most help put this emerging field in perspective. We will review the various synthetic strategies, their pros and cons, and we will give our best guesses as to the future directions of the field and what we can expect from it.

NEWS w.advmat.de that are generally used to maintain nanostructures colloidally stable might alter the very reason why such ‘‘weird’’ structures emerge: the delicate equilibrium between surface and core energies.

The properties of such ultrathin nanowires are of great interest, as they have been demonstrated to show quantum conductance and ballistic conduction.[31] Negative magnetoresistance,[32] localization,[31] low thermal conductivity,[3] ferromagnetism,[34] and quantum size effects[35] are other properties that are of great importance for a variety of applications. For example, the low thermal conductivity of one-dimensional nanostructures is expected to boost the efficiency of thermoelectric devices, as recently demonstrated in silicon nanowires.[36,37]

The last point of interest for ultrathin nanowires we would like to stress is the possibility of studying the fuzzy interface between nanowires and molecular entities. Questions that arise are whether we can observe polymer-like topological properties in inorganic nanowires, if we can couple molecular properties, like monodispersity, with the quantum size effects of nanoscale materials, and what the interface between these two worlds would look like.

2. Ultrathin-Nanowire Synthesis Methodologies

The synthetic methodologies that have been employed for the synthesis of ultrathin nanowires are all within the nanowire-synthesis framework described by Xia et al.[1] More specifically they mostly involve templating, ligand control, and oriented attachment. In Figure 2, we have listed the tools, the pros, and the cons of each of these strategies, which we will individually cover through some case histories.

Figure 2. Ultrathin-nanowire synthesis methodologies: the tools and pros and cons of the different synthetic strategies to obtain ultrathin nanowires are shown.

Figure 1. Potential of ultrathin nanowires: the various areas of interest behind the research on ultrathin nanowires are shown, together with possible areas of application.


2.1. Templating

The templating strategy involves the use of a hard or soft structure, called the template, as the directing agent of the growth of the nanowires. The isotropicity of crystal growth is constrained by the asymmetry of the template. As shown in Figure 2, the templates can be zeolites, mesoporous materials, nanocrystals, or micelles.

Romanov showed that molten bismuth could be infiltrated in the natural zeolite Mordenite to obtain ultrathin nanowires only two atoms across.[38] Hong et al. used an organic analog to a zeolite instead, obtained by the self-assembly of calix-4-hydroquinones.[39] The redox activity of the calixarene was used to reduce silver ions in situ, and the result was the formation of silver nanowires 0.4nm thick embedded in the matrix (Fig. 3a and b). In these cases, the wires were not extracted from the template.

In the case of mesoporous materials, like periodic mesoporous silicas (PMS), the pores are between 2 and 30nm in diameter, permitting an easier infiltration. Holmes and coworkers showed how silicon, germanium, copper, cobalt, and iron oxide nanowires could be created within pores by supercritical inclusion techniques, which take advantage of the extremely low viscosity and surface energy of supercritical fluids.[40–43] An example of the magnetite nanowires produced by this technique is shown in Figure 3c. Ko and Ryoo showed, instead, how wet impregnation with metal-salt solution allowed the formation of metallic nanowires within the pores of mesoporous materials, which could also be extracted from the template.[4] This technique has then been used to produce a variety of nanowires of different metals.[45,46] The reduction of the salts was performed by UV irradiation. Ultrathin polymer nanowires have been produced by in situ polymerization of phenolformaldehyde within the pores of PMS.[47] As shown in Figure 3d, those fibers had a strong tendency to aggregate in bundles, a very common feature of nanowires liberated from mesoporous materials, as they lack specific surface stabilization.

Templating can also be performed by nanocrystals through the vapor–liquid–solid (VLS) reaction protocol, in which a molten metallic nanocrystal templates the dissolution and precipitation of the precursors, and gives the onedimensional symmetry.[5] In Figure 3e, a 5nm Si nanowire growing along the (110) axis with the templating Au nanocrystal at its tip is shown.[48] A similar mechanism was used by Korgel and coworkers to form Si and GaAs nanowires in supercritical solvents (Fig. 3f and g).[49] In this case, the nanowires were not stuck to a surface, as in the case of VLS methods, but were instead dispersed in the fluid phase. By using Bi/Au alloy colloidal nanocrystals, Kuno’s group demonstrated the synthesis of veryhigh-quality ultrathin CdSe[50] and CdTe[51] nanowires in solution at relatively low temperatures.

Recently, a flurry of papers has appeared in the literature regarding Au ultrathin nanowires. In the first report, Xia’s group reported the reduction of gold chloride salts through oleylamine and silver nanocrystals.[52] The mechanism they propose is the formation of an [(oleylamine)AuCl]n inorganic polymer, which templates the formation of ultrathin nanowires with uniform thickness. The polymer is shown in Figure 3h, and the product in Figure 3i. Preliminary evidence of such polymer chains was demonstrated.

A few days after, Sun’s and Lieber’s group reported the electrical characterization of samples obtained in very similar conditions.[53] Au single-crystal nanowires of 9nm diameter were found to have a good conductivity and a high failure current density. This work demonstrates that ultrathin nanowires can circumvent the electromigration problems of nanowires, which leads to failure. These wires are thus expected to be very promising for nanoelectronic circuitry. In a report which followed by few days, Yang’s group obtained a similar

Figure 3. Templatingstrategies:a)HRTEMimageofAgnanowirearrays,formedinaself-assembled calix [4] hydroquinone matrix. [39] b) Proposed atomic structure for the Ag nanowires in a). [39] c)

HRTEM image of Fe3O4 nanowires produced in a mesoporous silica material by supercritical inclusion. [42] d)TEM image ofpoly(phenolformaldehyde) ultrathin nanowire bundlesformed after removalofthePMStemplate.[47]e)UltrathinSinanowiresgrownbyVLSmethod—thescalebaris 5nm. [48] f,g) TEM and HRTEM images of ultrathin Si nanowires grown in supercritical solvents by nanocrystal templating. [49]h) Drawing of an [(oleylamine)AuCl]n inorganic polymer, which templates the gold nanowires depicted in i). l) HRTEM image of gold nanowires grown via a micellar mechanism.[54]Reproduced withpermissionfrom [4]. Reproduced withpermissionfrom [39,49]. Copyright 2001, 2000 American Association for the Advancement of Science. Reproduced with permission from [42,48,54]. Copyright 2003, 2004, 2008 American Chemical Society. Reproduced with permission from [47]. Copyright 1998 Royal Society of Chemistry.

w.advmat.de product (shown in Fig. 3l) via a similar procedure involving gold chloride salts and oleylamine.[54] In that work, the formation of the wires is attributed to a micellar mechanism.

The micellar growth mechanism is based on the assumption that the micelles formed by the long-chain alkylamines would guide the growth of the nanowires in one dimension. The curvature of the micelle at the tip of the nanowires would lead to a very loose packing of the alkyl chains, which in turn would allow for easier diffusion of the reagents to the tips.[5] At the same time, the alkylamines can slowly reduce the metal salts. Some doubts remain on this proposed mechanism, based on the fact that oleylamine (sometimes with oleic acid impurity) is generally used as the ligand in these reactions. Oleylamine has a double bond in the middle of the chain, which usually prevents the formation of ordered monolayers stabilized by van der Waals interactions. The micelles produced by oleylamine were thus not expected to be strong, well ordered, and dense enough to completely prevent the growth of the nanowires in two directions. In Xia and coworkers’ interpretation of the Au-nanowire growth mechanism,[52] the micelle is not really the template. According to their results, the nanowires are formed by the reduction and coalescence of [(oleylamine)AuCl] inorganic polymers. In this case, the stability of the one-dimensional architecture during growth would be granted by the Au–Au bonds along the backbone of the polymer, instead of by an oleylamine micelle.

The micellar-templating mechanism was also proposed for the formation of FePt nanowires by Sun and coworkers,[5] for the growth of calcium phosphate[56] and barium sulfate[57] nanofibers by Mann and coworkers, and for the growth of ultrathin Te nanowires by Qian and co-workers.[58] The attractiveness of the micelle- and polymer-templating mechanisms is that they can marry the best aspects of a templating strategy with the advantages of ligand-control strategies, which are routinely used for the synthesis of monodisperse spherical nanocrystals. Together with monodispersity and the large know-how available, the possibility exists to obtain large amounts of colloidally stable material. In addition, one is not necessarily limited by anisotropic structures, as often happens for ligand-control strategies.

2.2. Ligand Control

Ligand-control strategies for the synthesis of nanowires was an offspring of the work that was being done on shape control of nanocrystals. The first demonstration of the controllable synthesis of CdSe nanorods demonstrated that using mixtures of ligands on anisotropic structures (like CdSe wurtzite) allowed for the growth of anisotropic structures.[59] The principle is simple: one ligand caps very strongly the sides of the nanostructure, thus strongly reducing its surface energy and therefore its growth rate, while the other ligand caps selectively and loosely the tips. This selective ligand shielding, coupled with the high chemical potential of the tips (due to their curvature and different lattice terminations), allows for the formation of nanowires with high aspect ratios.[60] In some cases, the presence of two ligands is not necessary, as the difference in surface energy of the different facets of the nanostructure are already high enough to favor one-dimensional growth.

Some exceptionally thin nanowires have been obtained by this methodology. Qian and coworkers demonstrated the growth of

Cu2S nanowires as thin as 1.7nm (Fig. 4a).[61] A mixture of decanethiol and oleic acid was used, and the morphology of the wires was found to be strongly dependent on the precursor ratio.

Similarly, the morphology of Bi2S3[62] and Sb2S3[63] nanowires and nanotubes was also dependent on the precursor ratio, as reported by our group and by Son and coworkers.[64]

Bi2S3 nanowires were formed via a heterogenous nanocrystal route,[65] by reacting a bismuth citrate slurry in oleylamine with a sulfur solution in oleylamine[62] (Fig. 4b). The diameter was found to be extremely monodisperse and independent of reaction

Figure 4. Ligand control strategies: a) HRTEM image of 2nm Cu2S nanowires produced by decomposition of single-source precursor Cu{S2CNEt2} in the presence of decanethiol and oleic acid. [61] b) TEM image of 1.6nm necklace Bi2S3 nanowires produced by the reaction of bismuth citrate with sulfur in oleylamine. [62] c,d) TEM image of 1.1nm 2.2nm samaria nanoribbons produced by high-temperature decomposition of samarium acetate in presence of oleylamine and decanoic acid. [6] e) TEM and HRTEM images of 1.2nm ceria nanowires produced by hightemperature decomposition of cerium nitrate in the presence of diphenyl ether, oleic acid, and oleylamine. [67] Reproduced with permission from [61,6]. Copyright 2005, 2006 American Chemical Society. Reproduced with permission from [62,67].

w.advmat.de time. The structure of the wires was compared to a necklace, as it was composed of nanocrystalline grains along the backbone of the wire. Optical spectroscopy and photoconductivity experiments demonstrated the presence of quantum confinement effects and of excitons at room temperature, never before seen in Bi2S3. The exact mechanism of growth is still largely unknown, but we postulated, on the basis of our experiments, that some specifically stable one-dimensional configuration of the atomic structure of

Bi2S3 must lie at about 1.6nm,[62] much like what happens for magic clusters. This is further confirmed by the ubiquitous

1.6nm size in nanostructures formed from the isostructural

Sb2S3.[64,65] This strong valley in the energy landscape would prevent the nanowire to grow sideways, and instead would foster growth along one direction only. Sb2S3 nanowires and nanoplatelets could be formed with a thickness of 1.7nm, by reacting antimony oleate with sulfur dissolved in oleylamine.[63] Similarly, nanotubes formedby therolling up ofa 1.7nm-thick ribbon could be formed by reacting SbCl3 with sulfur in oleylamine.[64] In this case, different from Bi2S3, the nanowires were single crystalline. Metal oxide nanowires were also produced in ultrathin diameters by Hyeon’s group. Sm2O3 nanowires, shown in Figure 4c and d, had a rectangular cross-section of 1 2 unit cells, corresponding to 1.1nm 2.2nm.[6] The nanowires were grown by decomposing hydrated samarium acetate in the presence of oleylamine and decanoic acid at high temperatures. The peculiarity of this reaction was its sensitivity to the specific fatty acids. The same reaction performed with oleic acid gave ultrathin platelets instead of wires.

By dissolving and decomposing cerium nitrate in diphenyl ether, oleylamine, and oleic acid, ceria nanowires with a diameter of 1.2nm were obtained.[67] Wires with different lengths were produced by using different amounts of oleic acid, but also in this case the thickness was found to remain constant.

The ligand-controlled reactions have now demonstrated the feasibility of producing ultrathin nanowires of a variety of materials, irrespective of anisotropicity of the lattice. The last two examples of samaria and ceria both have cubic structures. The use of more than one ligand is also unnecessary, as the cases of Sb2S3 and Bi2S3 demonstrate. What is somehow clear is that a more thorough understanding of the mechanisms of these reactions is needed. Most of these examples show a remarkable selectivity for some determined thicknesses, which is not consistent with traditional ligand-controlled growth models. It is also possible that some of these examples are unrecognized cases of micelle-templated growth.

2.3. Oriented Attachment

Oriented attachment is maybe the most fascinating growth strategy for ultrathin nanowires,[68–70] where single-crystalline nanowires are produced by oriented aggregation of faceted nanocrystals.[4,71] This methodology has some unique features, such as the constancy of the nanowire diameter during growth, and the similarity with polymerization reactions, in which monomers are added to the tip of the growing polymer.[72] The attachment is usually driven by the formation of a permanent or temporary electric (or magnetic) dipole in each nanocrystal. The dipoles then orient in chains and attach. The matching of the dipoles with specific facets of the structure allows for the final nanowire to often maintain single-crystallinity. The advantage of oriented attachment is that the diameter of the nanowire is largely determined by the diameter of the nanocrystals. The ease with which nanocrystals with small diameters (<4nm) can be prepared renders this technique promising in the production of ultrathin nanowires. Ultrathin CdSe nanowires were obtained by oriented attach- ment by reacting cadmium acetate (CdAc2) and selenourea in long-chain amines.[73] The diameter of the final wires could be tuned by changing the reaction temperature, or by using different long-chain amines. Longer-chain amines produced larger wires (Fig. 5a and b). In a following theoretical paper,[74] the same authors found that the formation of low-aspect-ratio quantum dots or rods followed by oriented attachment was the most energetically viable route to nanowires, confirming the experimental data. In the wurtzite structure of CdSe, the c axis can show facets entirely capped by cations or anions; this is a strong premise to the formation of dipoles in the structure during growth. ZnSe, another wurtzite solid, was also formed in the shape of ultrathin nanowires by oriented attachment, by reacting selenourea with cadmium acetate in the presence of molten octadecylamine (Fig. 5c).[75]

In all the examples mentioned above, very strong quantum confinement effects were observed, testifying to the crystallinity and small diameter of the wires. While it is true that oriented attachment is easier to obtain on structures possessing an anisotropic lattice, such as wurtzite, recent experiments have demonstrated that cubic structures can also undergo oriented attachment in spectacular ways.

(Parte 1 de 3)