You are on page 1of 12

www.afm-journal.

de

FULL PAPER

Controlled Synthesis of CdSe Nanowires by Solution LiquidSolid Method


By Zhen Li,* Ozgul Kurtulus, Nan Fu, Zhe Wang, Andreas Kornowski, Ullrich Pietsch, and Alf Mews*
reaction schemes to prepare nanostructures with distinct size and shape.[1] Different wet and nonwet chemical techniques have been developed to generate various nanostructures of different dimensionality.[2] Compared with nonwet chemical approaches, such as molecular beam epitaxy[3] and chemical vapor deposition,[4] wet chemical methods show advantages for tuning the size, shape, and properties of the nanostructures, especially in the preparation of very small nanostructures.[5] In addition, wet chemical methods are attractive in terms of cost and sustainability. One typical class of systems is semiconductor nanostructures, where CdSe is chosen as a prototype material. Hence in this paper, we will rstly summarize the controlled synthesis conditions to prepare CdSe quantum dots (QDs) and rods through advanced colloidal chemical methods, where the importance of reaction parameters, such as precursor concentrations, ligands, precursor molar ratios, and reaction temperatures, will be emphasized.[617] Since these parameters have been intensively studied in the past, they serve as a basis for the synthesis of extended CdSe nanowires by the more recently developed technique, namely the solutionliquidsolid (SLS) approach,[1820] where the 1D wire growth is catalyzed by a catalyst particle with a low melting point. In this method, the range of parameters is even broader in comparison to the unanalyzed growth of CdSe nanostructures as mentioned above, due to the additional presence of the catalyst. Hence we will summarize the different synthetic approaches from literature and compare it to our own results. In particular, we will present a detailed investigation on the synthesis, characterization, and optical properties of CdSe nanowires generated by the SLS method and nally we will provide an outlook of this research direction. A milestone in the preparation of CdSe nanostuructures by wet chemical methods was presented in 1993 where it was shown that high quality CdSe QDs with a narrow size distribution can be prepared by thermal decomposition of organometallic precursors in hot solvent.[6] Since then, several CdSe nanostructures such as dots, rods and tetrapods with tunable shapes and dimensions in the nanometer or even subnanometer range have been

Semiconductor nanowires prepared by wet chemical methods are a relatively new eld of 1D electronic systems, where the dimensions can be controlled by changing the reaction parameters using solution chemistry. Here, the solutionliquidsolid approach where the nanowire growth is governed by low-melting-point catalyst particles, such as Bi nanocrystals, is presented. In particular, the focus is on the preparation and characterization of CdSe nanowires, a material which serves a prototype structure for many kinds of low dimensional semiconductor systems. To investigate the inuence of different reaction parameters on the structural and optical properties of the nanowires, a comprehensive synthetic study is presented, and the results are compared with those reported in literature. How the interplay between different reaction parameters affects the diameter, length, crystal structure, and the optical properties of the resultant nanowires are demonstrated. The structural properties are mainly determined by competing reaction pathways, such as the growth of Bi nanocatalysts, the formation and catalytic growth of nanowires, and the formation and uncatalytic growth of quantum dots. Systematic variation of the reaction parameters (e.g., molecular precursors, concentration and concentration ratios, organic ligands, or reaction time, and temperature) enables control of the nanowire diameter from 6 to 33 nm, while their length can be adjusted between several tens of nanometers and tens of micrometers. The obtained CdSe nanowires exhibit an admixture of wurtzite (W) and zinc blende (ZB) structure, which is investigated by X-ray diffraction. The diameter-dependent band gaps of these nanowires can be varied between 650 and 700 nm while their uorescence intensities are mainly governed by the Cd/Se precursor ratio and the ligands used.

1. Introduction
The adjustment of the structural, optical, electronic, and magnetic properties of nanomaterials strongly relies on the development of
[*] Dr. Z. Li, Prof. Alf Mews, Dr. N. Fu, Z. Wang, A. Kornowski Department of Physical Chemistry University of Hamburg Grindelallee 117, 20146 Hamburg (Germany) E-mail: zhenli76@hotmail.com; mews@chemie.uni-hamburg.de Dr. O. Kurtulus, Prof. U. Pietsch Department of Physics University of Siegen Walter-Flex Str. 3, 57072 Siegen (Germany)

DOI: 10.1002/adfm.200900569 3650

2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Funct. Mater. 2009, 19, 36503661

www.afm-journal.de

reported.[6,7,9] It is interesting to note that only slight modications of reaction parameters, such as ligands, reaction temperature, or precursor concentration (ratios), can lead to the broad variety of sizes and shapes as mentioned above.[79,1315] For example, Peng et al. demonstrated that the size and size distribution of CdSe dots can be manipulated by the monomer concentration.[7] At high monomer concentrations, the smaller nanoparticles grow faster than larger ones, which results in the size distribution being focused. If the monomer concentration drops below a critical threshold, the smaller particles are depleted as larger ones grow (i.e., Ostwald ripening), and the size distribution becomes broader or is defocused. Further attempts to adjust the growth kinetics of the CdSe nanomaterial incidentally lead to the development of CdSe nanorods.[8,9] By using very high precursor concentrations and a dened admixture of alkylphosphonic acids and trioctylphosphine oxide (TOPO), CdSe rodsand later even more complex structures, such as arrows, teardrops or tetrapodshave been synthesized.[9] Peng et al. could also show that the shape evolution of CdSe nanorods goes through three stages upon decrease of the monomer concentration: i.e., 1D growth, 3D growth, and 1D/2D ripening.[11] On the other hand an extremely high monomer concentration could also lead to complex 3D nanostructures, but only after the formation of magic-sized nuclei.[14] For instance, initial monomer concentrations as large as [Cd] 433 mmol (g TOPO)1 resulted in tetrapod-shaped or branched CdSe nanocrystals while successive lowering of the concentration generated nanorods, rice-shaped nanorods, and nally spherical nanocrystals. In summary, it is possible to generate very complex CdSe nanostructures by precisely controlling reaction parameters, such as the precursors and their concentrations, the molar ratio of Cd and Se precursors, the molar ratio between alkylphosphonic acidand TOPO-ligands, the chain length of alkylphosphonic acids, and the reaction time and temperature.[14] Hence, the preparation and shape-evolution of CdSe nanorods is well investigated, but it is not possible to grow high-yield CdSe quantum wires with a length of more than a few micrometers by controlling the reaction kinetics.[21] However, the preparation of very long 1D nanostructures is of particular importance, for example, electronic devices.[22] Therefore an alternative method to grow very long nanowires with high yield has been established; this is the so-called SLS approach.[1820] Here low-melting-point (LMP) nanoparticles are used as catalysts,[19] and the reaction is performed at temperatures where the LMP nanoparticles are melted into nanodroplets. The nanowire precursors are then adsorbed on or dissolved into the nanodropets and grow into 1D nanowires until the precursors are consumed. Hence, in addition to the reaction parameters as mentioned above, the surface properties, solvating abilities, and reactivities of the nanocatalysts should also determine the formation of the nanowires.[19] The rst example of synthesizing CdSe nanowires by the SLS method was reported by the Buhro group in 2003.[23] In contrast to the recipes applied for the preparation of nanorods,[14] these nanowires were prepared from cadmium stearate and n-R3PSe (R butyl or octyl) in TOPO at 240300 8C using bimetallic Au@Bi (diameter, D 8.7 nm) or pure Bi (D 21.2 or 23.6 nm) nanoparticles as catalysts. In addition the initial Cd precursor concentrations ([Cd] 3.829.7 mmol (g TOPO)1) and the molar

ratios of Cd and Se precursor (Cd/Se 0.010.1) were much lower as compared to the values used in nanorod preparation ([Cd] 267 mmol (g TOPO)1; Cd/Se 2).[14] This indicates that the growth of CdSe nanowires can be already realized at low precursor concentrations and also at low reaction temperatures in the presence of Bi or Au@Bi nanocatalysts. By adjusting the reaction parameters, the authors could even prepare CdSe nanowires with diameters below the bulk exciton dimension, to compare the respective band gaps (DEg) with theoretical values.[23] Unfortunately, the authors did not present a systematic study of the growth conditions at that time. So the individual chemical parameters, which lead to the impressively narrow diameter distributions of their nanowires remain unclear. Following a similar strategy, the Kuno group prepared straight and branched CdSe nanowires from CdO and TOPSe (TOP: tri-nactylphosphine) in TOPO in the presence of octanoic acids at 330 8C using Au@Bi nanoparticles as catalysts.[24] The diameters of the nanocatalyst used (1.43.0 nm) were much smaller than those used by Buhro et al.(D 8.7, 21.2, and 23.6 nm).[23] Also they used much higher Cd precursor concentrations ([Cd] 64.7 mmol (g TOPO)1) and very different Cd to Se precursor ratios (Cd/Se 71.7).[23] However, their observed nanowire diameters were always much larger than the original catalyst diameter[24] while in the work of the Buhro group the diameter of the CdSe nanowires could be either smaller or bigger than the nanocatalyst dimensions, depending on the reaction conditions.[23] In this article, the authors empirically discussed synthetic considerations, such as reaction mixture concentration, reaction temperature, apparent Cd to Se ratio, catalyst Au@Bi size, catalyst volume, and TOP doping. They showed for example, that parameters such as low concentrations, high temperatures, high Cd/Se ratios, large catalyst sizes and lower TOP concentrations are benecial to the growth of straight nanowires, in contrast to branched nanowires However, the nanowire diameter and length distribution was not discussed. It should be noted that this SLS approach has also been applied to synthesize other semiconductor nanowires such as IV (Si, Ge),[25,26] IVVI (PbS, PbSe, PbTe),[27] IIVI (CdS, CdTe, ZnS, ZnSe, ZnTe)[2832] and IIIV (InP, InAs, GaAs).[18,3337] This method can even be used to prepare diode-nanowires such as CdSeCdS[38] and ZnSeZnTe.[39] In these reports, the nanowire diameters are larger than their corresponding exciton radius, and how to control the nanowire diameter and length remains unclear, especially below the exciton dimension region. Hence, the inuence of the reaction parameters on the preparation of nanowires and the details of the reaction mechanism are still not well understood. One of the main issues in this context is the formation and function of the different Bi nanocatalysts.[23,25,26,3032,35,36,3840] For example, pure Bi particles with diameters of approximately 20 nm were either prepared by thermal decomposition of Bi[N(SiMe3)2]3 in the presence of Na[N(SiMe3)2] and poly(1-hexadecene)0.67-co-(1-vinylpyrrolidinone)0.33,[23,41] or by reduction of Bi(III) 2-ethylhexanoate with NaBH4 in presence of TOP.[35] The coreshell Au@Bi nanoparticles[23,24,2729] were synthesized from Au seeds that were successively covered by Bi.[23,24,42] By controlling the amount of Bi precursor, the particle size can be adjusted between 1.5 and 8.7 nm. However, it is quite difcult to separate the coated from the noncoated nanoparticles. In our previous study, we developed a

Adv. Funct. Mater. 2009, 19, 36503661

2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3651

FULL PAPER

www.afm-journal.de

simple way to prepare pure Bi nanoparticles (D 3.3 0.5 nm) by the reduction of Bi[N(SiMe3)2]3 or BiCl3 with TOP at room temperature.[43] These nanocatalysts can also be in situ produced by mixing the Bi precursors and TOPSe solution (the Se concentration is 1 or 2 M) without additional TOP. We showed that the Bi nanocatalysts grow into large nanoparticles (4080 nm) upon heating.[43] Also we were able to show that the growth of the Bi nanoparticles could be slowed down in the presence of Cd and Se precursors upon formation of the CdSe nanowires.[43] In summary, the growth mechanism of CdSe nanowires by the SLS method is even more complex than the kinetic reaction control to form different CdSe nanostructures, due to the additional presence of the catalytic nanodroplets. In addition, the structural and optical properties[44] of CdSe nanowires generated from slightly different SLS conditions (e.g., different catalyst particles) can be different. Hence we performed a systematic variation of reaction parameters to evaluate their inuence on the CdSe nanowires nucleation and growth during the SLS process. We propose that at least three parallel competing reactions are taking place during the reaction; these include 1) the growth of nanocatalysts, 2) the formation and catalytic growth of nanowires, and 3) the formation and noncatalytic growth of QDs. These competitive processes determine the diameter, length, shape, and yield of the nanowires produced. We will show that distinct changes in reaction parameters allow guidance of the reaction pathway towards certain directions, and thus, allow adjustments of the structural and optical properties of the nanowires.

2. Results and Discussion


This part of the paper is organized as follows: rstly, we will describe the effect of different reaction parameters on the nanowire morphology. Here, we will rstly change only one of the reaction parameters as described in the experimental section and discuss the effect on the diameter and length of the nanowires on the basis of transmission electron microscopy (TEM) results. This is followed by an iterative variation of several reaction parameters to highlight their interactions. Secondly, we will show synchrotron X-ray results of an investigation of the crystallinity of the nanowires, and nally, we will focus on the optical properties.

because a longer reaction time should lead to longer nanowires. We performed experiments in which toluene was injected at different time intervals after the precursor injection. Due to the high reaction temperatures, the toluene immediately evaporated leading to a very fast temperature decrease, and hence, termination of nanowire growth (Table S1, code 13, Supporting Information (SI)).[45] For example, Figure 1ac shows TEM images of CdSe nanowires quenched with toluene at different times. The diameter and length distributions of the resultant CdSe nanowires are presented in Figure S1, SI. Relatively short rods (38.8 10.0 nm) were produced when toluene and the mixture of Bi nanocatalysts and TOPSe were co-injected into the Cd precursor solution (Fig. 1a). When the mixture of Bi nanocatalysts and TOPSe was injected rst, followed by toluene injection after about a second, pencil-shaped CdSe nanorods with a diameter of 16 nm and length of 150 nm were generated (Fig. 1b). If the growth was maintained for 15 s before quenching, the obtained nanowires exhibited a similar diameter, but an increased length of about 625 nm (Fig. 1c). Further prolonging the reaction time to 3 min resulted in partial precipitation, probably due to a 3D ripening process similar to the case of nanorods.[14] Therefore, we chose 1 min as the standard reaction time for most of the investigations. In comparison with other reports where a reaction time of 35 min was commonly used,[23] our short reaction time is assigned to the high reactivity of the small and pure Bi nanoparticles. As we have already shown in our previous report, these small (3 nm) particles grow very fast at high temperatures forming larger Bi nanodroplets. Most likely, the Bi particle growth happens simultaneously with the wire growth, which essentially explains the tip-shape of the nanowire.[43]

FULL PAPER

2.1. Nanowire Morphology The individual reaction parameters which were changed to adjust the diameter and length of the nanowires were: reaction time, precursor ratio, and nanowire and nanocatalyst ligands, nanocatalyst amounts, and reaction temperature. 2.1.1. Reaction Time Reaction time is the most evident reaction parameter affecting the nanowire length

Figure 1. TEM images of CdSe nanowires prepared using Bi@OA as nanocatalysts. Images (ac) show the effects of reaction time on CdSe nanowires obtained from the molar ratio of 7/1 (Cd/Se) after direct quenching through toluene injection: a) toluene and the mixture of Bi nanocatalysts and TOPSe were co-injected, b) the mixture of Bi nanocatalysts and TOPSe was injected rst, followed by toluene, c) the mixture of Bi nanocatalysts and TOPSe was injected rst, and toluene was injected after 15 s. Images (df) show CdSe nanowires prepared from different Cd/Se precursor ratios: d) 7/1, e) 1/1, and f) 1/7.

3652

2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Funct. Mater. 2009, 19, 36503661

www.afm-journal.de

2.1.2. Precursor Ratio The initial Cd and Se precursor ratio also affects nanowire morphology. This value has been found to be an important factor in inuencing the aspect ratio of CdSe nanorods during kinetic growth control[9,14] or the branching CdSe nanowires in SLS growth.[24] Figure 1df shows the TEM images of CdSe nanowires obtained from three different Cd/Se precursor ratios, i.e., 7/1, 1/1 and 1/7 (Table S1, code 46, SI). With increasing Se precursor ratio, the nanowires decrease in length from 1.5 mm to 663 and 334 nm with a respective slight decrease in diameter from 22 to 17 nm (Fig. S2, SI). A decrease in length upon an increase of the Se ratio has also been observed in the case of CdSe nanorods without additional catalyst particles. Peng et al. attributed this to the low concentration of remaining monomers because more Se precursor would lead to more magic-sized nuclei.[14] In this context, it is worth mentioning that we also observed the formation of a large quantity of noncatalytically grown nanocrystals, as presented in Figure 1f. On the other hand, for the catalytic growth of nanowires, different ratios have been reported for different nanocatalysts. For smaller Au@Bi nanocatalysts (D 1.43.0 nm), it was observed that under Cd-rich conditions a higher Cd/Se ratio (Cd/Se 7) yields straight wires, and a lower ratio (Cd/Se 1.7) generates branched nanowires.[24] For larger pure Bi or Au@Bi nanocatalysts (D 8.7, 21.2, and 23.6 nm), Se-rich conditions (Cd/Se < 1/10) were used, and the best Cd/Se ratio for producing uniform thin nanowires was around 1/30.[19,23] Obviously the nature of the different catalyst particles used is of major importance for wire growth. For example, the mean diameter of the pure Bi catalyst particles used in this study is similar to the Au@Bi nanocatalysts (D 1.43.0 nm) as used by the Kuno group. However, the diameter of the nanowires from the pure Bi particles is 22 and 28 nm as shown in Figure S2 (SI), which is larger than the reported diameters of Au@Bi nanocatalysts (515 nm).[24] Additionly, in our case, we obtained almost straight nanowires even at lower Cd/ Se ratios (1/1 and 1/7). These differences are attributed to the different reactivities of nanocatalysts. The Au@Bi nanocatalysts (D 1.43.0 nm) typically consist of Au cores approximately 1.5 nm in diameter covered with a Bi layer of 0.40.8 nm in thickness.[24,42] Such hybrid nanocatalysts may release the Bi more slowly, and the formation of large Bi particles is hampered, which also leads to the formation of thinner nanowires. Hence in our case, the pure Bi particles become 710 times larger in diameter after wire growth while the Au@Bi nanocatalysts used by Kuno et al. are around 10 nm, which is only about 45 times the initial diameter. 2.1.3. Nanowire and Nanocluster Ligands Different ligands covering the Bi catalyst or CdSe surface can also affect the nucleation and growth of the CdSe nanowires. In this context, it should be mentioned that the small bimetallic Au@Bi nanocatalysts (D 1.43.0 nm) which grow during the nanowire formation are covered with TOP and oleic acid ligands.[24,42] For those particles, it was reported that the resulting nanowire diameter is always larger than the original catalyst size.[24,42] On the other hand, if large bimetallic Au@Bi (D 8.7 nm) or large pure Bi nanocatalysts (D 21.2 or 23.6 nm) covered with a polymer ligand (poly(1-hexadecene)0.67-co-(1-vinylpyrrolidinone)0.33) were used,

the resultant nanowire diameter can be either larger or smaller than the original size of the nanocatalysts.[23] It might well be that the small particles fuse more easily due to thermodynamic effects, such as melting point depressions or differences in surface tension. Another reason might be that the polymer shell exhibits stronger protective effects on the Bi particles in comparison with the small organic molecules, which prevent the nanocatalyst from growing. Finally the ligands might also inuence the adsorption kinetics and reactivity of the Cd and Se precursors with the Bi nanocatalyst surface. To test the effects of ligands on the nanowire growth, we compared the SLS reactivity of amine-covered Bi particles with those covered with TOP only. Firstly, we investigated the inuence of different amines such as hexadecylamine (HDA) (Table S1, code 7, SI) instead of oleylamine (OA; Table S1, code 4, SI). Figure 1d and 2a shows that both amine ligands lead to nanowires which are approximately 20 nm in diameter. Only the length of the nanowires resulting from Bi@HDA (35 mm, Fig. S3, SI) is longer than that of Bi@OA (1.5 mm). Also, the use of different acids, such as octanoic acid and oleic acid, had only a minor inuence on the resulting nanowires (Table S1, code 8, SI). In both cases (Fig. 2a,b), the mean diameter was in a range of 2025 nm, and the length was between 3 and 5 mm (see also Fig. S3, SI). On the other hand, if Bi@TOP nanoparticles are used, the morphology of the resulting nanowires is totally different (Table S1, code 911, SI). For example Figure 2c,d shows the TEM images and diameter histograms of CdSe nanowires prepared from 7/1 and 1/1 Cd/Se precursor ratios. Compared with those using Bi@OA or Bi@HDA as nanocatalysts (Fig. 1d, 2a), the nanowires

FULL PAPER
3653

Figure 2. TEM images of CdSe nanowires prepared from Bi@HDA nanocatalysts using a) oleic acid and b) octanoic acid as CdO ligands. The molar ratio of Cd and Se precursor in these two cases is 7/1. Images (cd) show CdSe nanowires prepared from two different Cd/Se ratios, c) 7/1 and d) 1/ 1, using Bi@TOP as nanocatalysts. The insets are the corresponding histograms of diameter distributions.

Adv. Funct. Mater. 2009, 19, 36503661

2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.afm-journal.de

tion conditions with respect to the controlled growth of uniform obtained from Bi@TOP nanocatalysts (Fig. 2c) are noticeably and thin nanowires. thinner and shorter. Their mean diameter is only 14 nm, and their length is approximately 100 nm. It is interesting to note that for 2.1.4. Nanocatalyst Volume these Bi@TOP particles, the decrease of Cd/Se ratio from 7/1 to 1/ 1 results in an increase of the nanowire length from 100 nm to Since thin nanowires emerge from small nanocatalyst particles, a 1.7 mm (Fig. 2d). This is in contrast to the Bi@OA particles as suppression of the Bi nanocatalysts fusion might be accomplished reported above, where a higher Se ratio leads to shorter nanowires by using lower quantities of Bi nanoparticles. Hence different (Fig. 1e). Further decrease of this ratio to Cd/Se 1/7 produced volumes of Bi@HDA catalyst solutions ranging from 25 to 300 mL even thinner nanowires wires (7.5 1.7 nm, Fig. S4, SI), which (0.091.09 mmol) were used for the growth of CdSe nanowires however was more irregular in shape. (Table S1, code 1517, SI). For the smallest volume (25 mL, Obviously the ligands strongly affect the growth of nanowires 0.09 mmol), only a few nanowires with a majority of nanocrystals from Bi nanodroplets. In particular, an excess of HDA is were obtained. If the volume was increased to 50 mL, several CdSe detrimental to catalytic growth of CdSe nanowires from the Bi nanowires 18 nm in diameter and 3 mm in length were obtained nanocatalyst surface, and favors the noncatalytic growth of (Fig. S6a,d, SI). When 100 mL of nanocatalyst solution was used, a nanocrystals. This can either be attributed to a strong afnity of majority of nanowires with a diameter and length of 23 nm and 5 amines to the Bi nanodroplet surface or to the amine complexation 7 mm, respectively, was formed (Fig S6b,e, SI). In comparison, the abilities of the formed CdSe species. The different afnities of TOP nanowires which were prepared using 200 mL nanocatalyst and amines to the Bi catalyst can be explained in a very solution (Fig. 2a) were similar in diameter, but slightly shorter rst approximation from the viewpoint of electronegativities of (35 mm). In general however, a further increase of the Bi catalyst [46] the interacting Bi, P, and N atoms. For example since solution volume to 300 mL leads to an additional increase of the the electronegative difference between Bi (2.02) and N (3.04) is nanowire diameter (33 nm) and a decrease in the length (23 mm; much higher than that between Bi and P (2.19), the latter one Fig. S6c,f, SI). Again this not only shows that the Bi particles grow might be replaced more easily from the surface of the Bi upon fusion, but also that a decrease of the nanocatalyst particle. Another argument might be that the higher packing of concentration is not sufcient to grow thin nanowires, the linear amines versus the branched TOP on the surface because of the parallel nucleation of nanoparticles. Hence, of the Bi nanodroplets leads to reduced surface activity in we used a volume between 100 and 200 mL, which relates the case of amines. The complexation abilities of amines to a concentration 0.360.72 mmol Bi for further optimization with the CdSe species on the other hand has been observed studies. already during the growth of CdSe QDs because an addition of HDA to the TOPOTOP system resulted in a more rapid focusing of the size distribution during particle growth.[13] 2.1.5. Reaction Temperature Therefore it might well be that CdSe clusters are generated as The temperature dependence of the wire formation is important intermediate species during the Bi-catalyzed nanowire growth, because the reaction should only be catalyzed by molten Bi and that these CdSe clusters are stabilized by amines. particles. It has been reported however, that the melting point of Bi While the details of adsorption and reaction are certainly much more complicated, there seems to be a general competition between the growth of Bi nanodroplets, the formation and catalytic growth of nanowires, and also the formation and noncatalytic growth of nanocrystals. For example, if the HDA/Bi ratio was further increased to 20/1, 50/1, or 100/1 (Table S1, code 1214, SI), the average diameters of the nanowires slightly decreased, and more CdSe nanocrystals were formed (Fig. S5, SI). It can be seen that the use of Bi@TOP particles leads to thinner, but more irregular nanowires. On the other hand, the reaction can be more easily controlled upon addition of HDA ligands (see below). In addition, we found that the nanowires from Bi@HDA nanocatalysts show higher uorescence intensities (see nanowire optical properties, Section 2.3). Most likely, the HDA molecules not only cover the surface of the Figure 3. AFM images of CdSe nanowires prepared at different temperatures and Cd/Se ratios: nanocatalysts, but also attach to the surface of ac) nanowires obtained from Bi@HDA nanocatalysts with high HDA/Bi molar ratio (10/1) the nanowires and hence increase their prepared at 150, 180 and 250 8C, respectively. df) Nanowires obtained from Bi@HDA nanouorescence.[13] Therefore, we also used catalysts with low HDA/Bi molar ratio (5.5/1) prepared from Cd/Se molar ratios of 7/1, 1/1 and 1/ Bi@HDA nanocatalysts to optimize the reac- 2, respectively.

FULL PAPER
3654

2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Funct. Mater. 2009, 19, 36503661

www.afm-journal.de

nanoparticles depends on size, i.e., the bulk melting point of 272 8C is only reached for particles as large as 20 nm in diameter, while particles of 4 nm in diameter show a melting point of approx 150 8C.[47] This means that any nanocatalyst particle will be molten at the high reaction temperature of 330 8C, which has been used so far throughout this study. On the other hand, the small initial Bi particles with a diameter of approximately 3 nm would already melt at temperatures far below the bulk melting point. Hence we investigated the temperature-dependence of the growth of nanowires at 150, 180, and 250 8C. For these studies we used a low CdO concentration (22.6 mmol (g TOPO)1) and a Cd/Se precursor ratio of 1/1 (Table S1, code 1820, SI). The atomic force microscopy (AFM) images of the respective sample are shown in Figure 3ac. Most interestingly it can be seen from Figure 3a, that CdSe nanowires can already be prepared if the reaction temperature is only 150 8C. However, here we found that after a reaction time of 1 min, only QDs could be observed. On the other hand, if the reaction time was prolonged to 10 min, a few nanowires with an average diameter of 12 nm and a length of 0.8 mm were obtained from the same solution. Again this shows that QDs might be formed as intermediates and successively transform into nanowires, if there is a sufcient interaction with suitable nanocatalysts. Also it shows that the reactivity of the nanocatalyst is already given at temperatures far below the bulk eutectic temperature of Bi and CdSe (265 8C)[48] yet to a somewhat lower degree. These results could be explained by the fact that there is a dramatic decrease of the eutectic temperature upon decrease in nanoparticle size. Another explanation would be that solid Bi nanoparticles might also catalyze the CdSe nanowire growth, a model which has been recently proven in GeAu systems synthesized by an vaporliquidsolid (VLS) approach.[49] With the reaction temperature increasing from 150 to 180 8C (Fig. 3b), the resulting nanowires already increase from 12 to 22 nm in diameter and from 0.8 to 6.0 mm in length. This is not much different upon further increase of the reaction temperature from 180 to 250 8C (Fig. 3c) where the diameter is also in the range of 21 nm and the average wire length is 7.0 mm. In comparing our results to literature values, it should be mentioned that the synthesis of semiconductor nanowires from Bi or Au@Bi nanoparticles was always reported for temperatures above 240 8C.[19,2327,2931,3540,42] The lower reaction temperature reported in this study is attributed to the low melting temperature[47] and high reactivity of the small and pure Bi nanocatalysts. 2.1.6. Cd Precursor Reactivity The effects of Cd precursor reactivity on the formation of CdSe nanowires is investigated by using dimethyl cadmium (CdMe2) as an alternative precursor of high reactivity. In contrast to the use of CdO as precursor, the synthetic scheme had to be slightly modied such that a Bi/CdMe2 mixture was injected into a TOPO solution containing Se precursor (Table S1, code 2123, SI). Figure S7a (SI) shows the TEM image of CdSe nanowires prepared from CdMe2 using a Cd/Se precursor ratio of 1/2 and 100 mL of Bi@HDA nanocatalysts. The mean diameter of the resultant nanowires was 15 nm with a standard deviation of 4.4 nm. Upon further decrease of the Cd/Se precursor ratio to 1/10, the nanowire

diameter decreased to 7.1 3.0 nm (Fig. S7b, SI). However, the diameter distribution was broader compared to the wires prepared using a high Cd/Se ratio (7/1) and Bi@OA nanocatalysts (7.9 1.9 nm).[43] In all cases, the nanowires prepared from CdMe2 precursor were shorter than those synthesized from CdO precursor, most likely due to competitive growth of QDs, which becomes more signicant due to the high reactivity of CdMe2. However, a large fraction of the nanowires was in the same size range as the size of the Bi nanocatalysts ($25% of D < 5 nm, Fig. S7c, SI), which indicates that very thin nanowires can be prepared by using highly reactive precursor. So far we have investigated the effect of several reaction parameters, which have been changed individually. In summary, we could show that the reaction is very fast at high temperature, and short wires can only be prepared if the reaction is terminated within seconds. At temperatures far below the Bi melting point, the reaction becomes slow, and predominantly nanocrystals are formed. This competitive growth of nanocrystals happens also at low Bi volumes and high Cd precursor reactivities. The morphology of the nanowires on the other hand is mainly affected by the interplay between the Cd/Se precursor ratio and the ligands used. If only TOP is used as the Bi ligand, a low Cd/Se ratio results in thin and long nanowires, which become irregular in shape. This situation is reverse upon addition of amines, where a high Cd/Se ratio results in longer and thinner wires, which are more uniform in shape. 2.1.7. Iterative Changes All the above results show that the growth of Bi nanocatalysts and the growth of CdSe nanowires cannot be individually controlled by changing just one reaction parameter. Therefore we performed iterative changes of the reaction parameters as mentioned above to achieve the goal of preparing long, homogeneous CdSe nanowires with a uniform diameter below the bulk exciton dimension (DB 11.2 nm). The general strategy is to enhance the growth of CdSe nanowires and suppress the growth of the Bi catalyst as well as the formation of CdSe QDs. Since it is rather difcult to slow down the Bi particle growth, we combined several reaction parameters by which the growth of the nanowires can be accelerated. Since wire formation can be more easily controlled using CdO rather than CdMe2 as precursor, we optimized the reaction conditions using CdO by a stepwise change of the reaction parameters as mentioned above. In the rst step, we decreased the HDA/Bi precursor ratio from 10/1 to 5.5/1 and used a high Cd/Se precursor ratio of 7/1 to prepare nanowires at 330 8C (Table S1, code 24, SI). Figure 3d shows the AFM image of CdSe nanowires obtained. Compared with nanowires prepared from the same Cd/ Se ratio, but a high HDA/Bi ratio (10/1) as shown in Figure S6b (Table S1, code 16, SI), the diameter was reduced from 23 to 16 nm (Fig. S8, SI), while the length of the nanowires remained almost unchanged (6 mm). In the second step, we decreased the Cd/Se precursor ratio from 7/1 to 1/1 (Table S1, code 25, SI). This resulted in a further decrease of nanowire diameter from 16 to 8.5 nm (Fig. 3e and Fig. S8, SI), which is already below the bulk exciton dimension. In addition, some of the nanowires become very long and exceeded 10 mm in length, and became more exible. Further decrease of the Cd/Se

Adv. Funct. Mater. 2009, 19, 36503661

2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3655

FULL PAPER

www.afm-journal.de

FULL PAPER

The nanowire diameter (8.2 1.3 nm) is similar to these shown in Figure 4, but the length of the wires is much shorter, which is attributed to the difference of Bi precursors.

2.2. Nanowire Crystal Structure Before we discuss the crystal structure of the resulting nanowires, we will briey focus on the structure of the Bi catalyst particles after solidication. Previously we investigated the size and composition of the Bi nanoparticles before and after nanowire growth.[43] The results showed that within experimental error of the energydispersive X-ray spectroscopy (EDAX) method (<1%) the Bi nanoparticles only consist of bismuth despite of the presence of the Cd and Se precursor. Here, we want to emphasize that upon fast quenching of the reaction solution most of the Bi nanoparticles showed a typical coreshell structure, i.e., a crystalline core and an amorphous shell. The quenching can be either realized by injection of toluene, or upon removal of the heating mantle and cooling the reaction ask with an air-blower or an ice-water bath. Figure 5ac shows high-resolution TEM (HRTEM) images of several typical Bi nanocatalysts attached to nanowires. The amorphous shell thickness is in the range of 2.910.2 nm for Bi particles of 1040 nm in diameter, respectively. Most likely the nanodroplet surface solidies very quickly and forms an amorphous layer while the inner part of the catalyst can equilibrate to form a crystalline structure. For example, the contrast of the amorphous shell and the crystalline core is most obvious if the reaction is rapidly quenched upon toluene injection (see Fig. 5b and Fig. S11a, SI). In contrast, if the cooling rate is very slow (only several 8C min1), almost no Bi particles can be found at the end of the nanowires, as shown in Figure S11b (SI). This might be due to the crystallization of the entire Bi particle which leads to a large

Figure 4. TEM and AFM images of CdSe nanowires prepared using different reaction times and several temperatures using Bi@HDA nanocatalysts and a low HDA/Bi ratio (2.1/1). The temperature and reaction time of nanowires shown in (a) and (c) are 330 8C and 30 sec, respectively; while the temperature and reaction time in (b) and (d) are 250 8C and 1 min. The height scale bar in the AFM images is 10 nm.

ratio to 1/2 (Table S1, code 26) leads to even thinner (6.5 nm, Fig. S8, SI) and longer nanowires as shown in Figure 3f. Hence in the third step, we started again from a Cd/Se precursor ratio of 1/1 (Fig. 3e) and decreased the HDA/Bi ratio from 5.5/1 to 2.1/1 (Fig. 4a,c; Table S1, code 27, SI). The results show that even at a shorter reaction time of 30 s, the obtained nanowires are 8 mm in length and only 9.1 nm in diameter. This is somewhat larger than the nanowire diameter observed using a small Cd/Se ratio, but the obtained nanowires are more uniform in shape (Fig. S8c and S9a, SI). In the next step, we decreased the reaction temperature from 330 to 250 8C and kept the reaction time for 1 min (Table S1, code 28, SI). Figure 4b shows the typical TEM image of obtained nanowires, from which a mean diameter of 8.1 nm (1.4 nm) and a nanowire length of about 7.4 mm can be extracted. Figure S9 (SI) presents the low magnication TEM images of these two samples prepared at 330 and 250 8C. Further decrease of reaction temperature to 180 8C and an increase of reaction time to 2 min produced thicker and shorter wires again (Table S1, code 29, SI). Therefore these results suggest that the optimum temperature for growing thin uniform nanowires is 250 8C. As a nal check, we also synthesized CdSe nanowires using similar Bi@HDA nano- Figure 5. ac) HRTEM images of different sized coreshell Bi nanocatalysts attached to CdSe catalysts prepared from the BiCl3 precursor nanowires. The particle sizes are 13.8, 19.7 and 29.4 nm, respectively. df) HRTEM images of (Table S1, code 30, SI). Figure S10 (SI) shows CdSe nanowires synthesized from different Cd precursors and Bi catalysts: CdMe2 and Bi@OA the TEM image of CdSe nanowires obtained. (de); CdO and Bi@HDA (f).

3656

2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Funct. Mater. 2009, 19, 36503661

www.afm-journal.de

lattice mismatch between the rhombohedral Bi particle and the hexagonal (or cubic) CdSe nanowire. While more detailed experiments are needed to investigate the BiCdSe interface, the slow temperature decrease provides a simple and effective way to remove the Bi nanocatalysts from the CdSe nanowires.[37] The crystal structure of the CdSe nanowires was investigated by HRTEM and synchrotron X-ray diffraction (XRD). Firstly, we will focus on the HRTEM results of CdSe nanowires prepared from different precursors and using amines as Bi nanoparticle ligands. Figure 5de shows HRTEM images of CdSe nanowires obtained using CdMe2 as Cd precursor and oleylamine as Bi ligand. It can be seen that the lattice pattern was partially distorted along the wire axis accompanied by a variations of contrast (Fig. 5d) and/or thickness (Fig. 5e). The different thickness along the wire axis can be clearly seen in low magnication TEM images. If CdO was used as Cd precursor and HDA was used as the Bi nanocatalyst ligand, more uniform nanowires were obtained which exhibited coherent lattice fringes as shown in Figure 5f. Besides the detailed morphology, the HRTEM images also reveal that all the CdSe nanowires exhibit a mixture of zinc blende (ZB) and wurtzite (W) sections, as has been previously described.[24] Here we performed a rst attempt to estimate the ratio of ZB and Won the basis of powder XRD. Figure 6a,b presents the typical XRD patterns of four samples referring to the images shown in Figure 3ef, S8 (SI), 4, and S9 (SI), respectively; these data were obtained at beamlines at the DELTA and the European Synchrotron Radiation Facility (ESRF). The angular values and

Figure 6. Four typical XRD patterns of CdSe nanowires. a) Patterns obtained from DELTA beamline measurements using 15.5 keV radiation resulting from an image plate with a beam size of 0.7 mm 0.7 mm. b) Patterns obtained from ESRF measurements using an X-ray energy of 25 keV and a point detector with a beam size of 0.2 mm 0.2 mm.

background of the raw data were calibrated using a silicon powder as reference material. It can be seen that the measured peak positions t perfectly to the W structure of CdSe since the hexagonally close packed (hcp) Miller indices (hkl) are used to index the reections. However, the intensity ratios between the integrated reections differ considerably from a pure W phase, which is due to the admixture of ZB and W. Hence, since several W reections also exist for the ZB lattice, the intensity ratio of the pure phases can be used to estimate the W to ZB ratio within the nanowires, in principle. Therefore we calculated the CdSe structure factors for pure W and ZB structure, respectively, using the atomic scattering factors and DebyeWaller factors of Cd and Se from literature.[50,51] Assuming the kinematical scattering law where the Bragg intensity is proportional to the square of the structure factors, some peaks such as 100 or 103, for example, could be interpreted as resulting only from the W crystal structure (indexed by 100W and 103W), while some others could be interpreted to be due to a combination of W and ZB. For example, it can be shown that the 111ZB and 002W, 220ZB and 110W, 131ZB and 112W, 224W and 300W, and 044ZB and 220W reections coincide in peak position. Taking into account the respective multiplicity factors, the relative contributions of pure ZB and W phases within the nanowires can be estimated from the ratios of intensities of these reections.[52] It has to be mentioned however, that a reliable quantication of ZB and W phases in this way is possible only assuming ZB and W phases with perfect crystal structure, i.e., without intrinsic stacking faults. Although this is not the case (see HRTEM), we used this approximation to estimate the relative contribution of ZB fractions between different samples. In any case, we could show that all the samples crystallize predominantly in the W phase. Based on our rough estimation described above the contribution of the ZB phase for both samples, No. 1 and No. 2, should only be of the order of 3%. However, this value is about twice as high for sample No. 3 and No. 4, where our model results in a ZB contribution of 6 and 8%, respectively. Even though the absolute values for the W and ZB fraction can be very different, the analysis of the diffraction pattern clearly shows that the relative ZB contribution is higher for sample No. 3 and No 4. This could be explained by the reaction conditions, which are summarized in Table S1, code 2528, SI. Here it can be seen that the short reaction time used to prepare No. 3 (30 s) resulted in nanowires exhibiting a higher ZB ratio as compared to the longer reaction time of No. 1 (1 min). Also the lower reaction temperature of No. 4 (250 8C) leads to a higher ZB ratio as compared to a high temperature used for No. 1 (330 8C). Obviously a long reaction time as well as a high temperature leads to the thermodynamically more favored W phase. In contrast, a lower reaction temperature at the initial stage of the wire growth leads to a somewhat larger amount of the kinetically favored ZB phase. However, it should be mentioned again that this model provides only a rough estimate of the coherent ZB sections within the nanowires, and further investigations and theoretical modeling is needed to quantify the amount of Wand ZB fractions. Besides the intensity (ratios), the full width at half maximum (FWHM), Du, of the diffraction peaks can be analyzed to determine the coherence length of the diffracted X-rays. In principle these values can be used to determine the crystalline domain size d and

Adv. Funct. Mater. 2009, 19, 36503661

2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3657

FULL PAPER

www.afm-journal.de

FULL PAPER

Table 1. Diameters and lengths of CdSe nanowires measured by TEM, AFM, and XRD.
Wire sample TEM No. No. No. No. 1 2 3 4 8.5 2.6 6.5 2.5 9.1 2.2 8.1 1.4 Diameter [nm] AFM 7.3 2.9 6.4 2.5 6.8 1.8 6.7 1.2 XRD(100) 8.0 6.8 7.6 6.1 TEM >4 >6 >8 >6 Length [mm] AFM 5.7 2.3 7.1 1.7 8.5 2.9 7.4 1.6 XRD(002) 0.018 0.012 0.022 0.019 3 3 6 8 Estimated ZB ratio [%]

the strain e acting in different directions of the nanowires, by using Equation 1:[53]

Du l=d cos u " tan u

(1)

where l refers to the beam wavelength. In order to extract Du, the peak shape of a measured Bragg reection was t by a Gaussian program, and deconvoluted from the instrumental peak prole modelled by another Gaussian program. In our previous paper, we have shown that the strain cannot be resolved from our experiment so far because of the low number of high-angle reections, which are not additionally affected by stacking faults.[52] Therefore strain effects shall be neglected at this point. This drawback can be overcome in future considering modern achievements in X-ray powder analysis, as shown for example for metals by Scardi et al.[54] Here we used the 002W reection to determine the length of the coherent blocks along the nanowire axis, since the nanowires grow predominantly along this crystallographic direction. Likewise the perpendicular 100W reection is used to determine the (coherent) radial diameter of the nanowires. Table 1 compares the calculated diameters and lengths with those obtained from TEM and AFM measurements. It can be seen that the calculated diameters are in good agreement with TEM and AFM data, but the calculated lengths are much shorter (1020 nm) than those determined by TEM and AFM (several micrometers). This can be explained on the basis of the HRTEM results which showed that the coherence in length direction is relatively short due to a change of alternating Wand ZB sections. Both sections are separated by stacking faults. In addition, within a homogeneous W or ZB section additional stacking faults also could be a reason for the discontinuity of the coherence. For example, a stacking fault within the W phase shows up if the sequence of layer stacking is changed, such as ABABjCBCB.[55] In the ZB phase a single change in stacking sequence as ABCjAjCBA is not a stacking fault, but a twin. In this case, the formation of a stacking fault requires at least two sequential twin planes, such as ABCAjCjB. However, for XRD the size of a coherently scattering domain is limited by any change in the stacking sequence. In both cases a single domain consists of at least two or more complete stacking units such as ABAB for the W phase or ABCABC for the ZB phase.

Figure 7. a) UVvis absorption spectra of different sized CdSe nanowires. b) Lowest-energy excitonic peaks are plotted as a function of nanowire diameter. c) UVvis absorption and photoluminescence spectra of CdSe nanowires prepared from different molar ratios of Cd and Se precursor; the AFM and TEM images of these samples are shown in Fig. 3df and Fig. S8, SI. The excitation wavelength was 450 nm.

2.3. Nanowire Optical Properties The optical properties of CdSe nanowire solution are investigated by UVvis absorption and photoluminescence spectroscopy.[23]

Figure 7a shows the UVvis absorption spectra of different sized CdSe nanowires. The energetic position of the lowest-energy absorption band was plotted in Figure 7b. A blue-shift of the absorption edge from 702 to 657 nm can be clearly observed with a decrease of nanowire diameter from 33 to 6 nm, respectively. All

3658

2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Funct. Mater. 2009, 19, 36503661

www.afm-journal.de

the absorption energies are higher than that of the bulk bandgap of CdSe (712 nm, 1.74eV), which is marked by dashed line in Figure 7b.[24] This shows that a quantum connement effect can be observed even for nanowires with a diameter of 33 nm, which is almost 3 times the value of bulk Exciton Bohr diameter, which is 11.2 nm for CdSe. These results are in good agreement with those observed for CdSe nanorods[17] and nanowires[23] in previous reports. Besides the UVvis absorption, the photoluminescence of the nanowire samples was investigated. Firstly it should be noted that every sample showed a weak bandgap luminescence, where the quantum yield was always lower than 1%, as compared to standard dyes, which is in agreement with previous reports.[44] Hence we compared the relative change of the uorescence intensity upon different reaction conditions. For example, Figure 7c shows the absorption and photoluminescence spectra of CdSe nanowires synthesized from different Cd/Se precursor ratios using Bi@HDA as catalysts. Both UVvis and uorescence spectra were blue-shifted upon decrease of the Cd/Se precursor ratios, due to the decrease of nanowire diameter as shown in Figure 3df and S8, SI. The uorescence intensity reached a maximum when the Cd/Se ratio was 1/1 and decreased when using an excess of either precursor. This is in contrast to previous reports on CdSe QDs where it has been shown that the excess of one of the precursors leads to a high uorescence quantum yield, especially when excessive Se precursors were used.[15] This difference could possibly be attributed to the stacking faults that show signicant effects on nanowire optical properties,[44] but only slight effects on QDs.[6,13,15] To further investigate the change of the uorescence intensity in dependence of the reaction conditions, we measured the photoluminescence of CdSe nanowires synthesized from Bi@OA or Bi@TOP as nanocatalysts. Figure S12 (SI) presents the UVvis and photoluminescence spectra of CdSe nanowires, which were introduced in Figures 1de, 2cd, and S4 (SI), respectively. Here it can be seen that the nanowires prepared from Bi@TOP nanocatalysts show lower uorescence intensities as compared with those synthesized from Bi@OA and Bi@HDA nanocatalysts. This result indicates that alkylamines play an important role in improving the uorescence intensity of CdSe nanowires. Such effects of surface ligands on the uorescence intensity have been well-studied in CdSe QDs systems.[13,15] For example, the quantum yield (QY) of CdSe QDs prepared from the HDA TOPOTOP system can be as high as 4060%, while TOPOcovered QDs exhibit only 515% QY.[13] Most likely, the HDA ligands not only cover the surface of the Bi nanocatalyst, but also passivate the surface of the nanowires. However, as already mentioned above, a large excess of amines (OA and HDA) leads to the formation of irregular nanowires. These results show that the role of the surface ligands on the formation and uorescence intensity of nanowires is very complex. Especially the surface properties of the nanowires need to be further investigated to synthesize nanowires with high uorescence intensity.

FULL PAPER
Figure 8. Proposed scheme for the formation and growth of CdSe nanowires and CdSe nanocrystals. After injection at high temperatures, larger Bi nanodroplets are formed upon coagulation and react with the Cd and Se precursors or intermediately formed CdSe clusters to initiate the growth of CdSe nanowires. Depending on the reaction conditions, this will lead to extended CdSe nanowires or to the parallel growth of semiconductor nanocrystals.

3. Summary
In summary, we have presented a detailed investigation of the growth of semiconductor nanowires by systematic variation of

reaction parameters. The presence of Bi nanoparticles at the end of nanowires shows that CdSe nanowires were grown via a typical SLS mechanism.[19,20] Although this concept was proposed more than 10 years ago,[18] the growth mechanism is still unclear since it depends on several reaction parameters. In Figure 8, we summarize how the formation of nanowires happens in solution. When the mixture of Bi nanoprecursors, TOPSe and TOPO CdOalkyl acid (or other reactive species as mentioned in the Experimental section) are brought together at high temperature, Bi nanodroplets grow very rapidly by aggregation. At the same time, Cd and Se precursors or even intermediate CdSe clusters will dissolve into (or decompose on the surface) of these nanodroplets, and grow into nanowires. The parallel growth of Bi nanodroplets and growth of nanowires from those nanodroplets might explain the tip-shape at the end of the nanowire. In any case, both the reactivity and concentration of the Bi precursors as well as the reactivity and concentration of the Cd precursors determines the morphology, and partially also the crystal structure of the nal nanowires. At this point, we shall not repeat the different effects of concentration (ratios), reaction temperatures, and ligands used. However, the general strategy for a reproducible synthesis of semiconductor nanowires is control of the competition between the growth of Bi nanocatalysts, the formation and catalytic growth of CdSe nanowires, and the formation and noncatalytic growth of CdSe QDs. This general strategy can be extended to the SLS controlled synthesis of other semiconductor nanowires below their exciton dimensions. To this end, ultrathin (D < 5.0 nm) semiconductor nanowires which exhibit strong quantization effects as their counterpart QDs and rods will be one of the future directions. Another direction will be the synthesis of highly uorescent semiconductor nanowires, since the highest-uorescence quantum yield of nanowires reported is only 1%. One effective approach to improve uorescence is to passivate nanowire surfaces with a wide bandgap shell to result in coreshell nanowires.[56] In addition, integration of multicomponents into a single hetero-nanostructure represents another future direction.[3840,57] These homo- and hetero-nanostructures are excellent building blocks for future nanometer-sized technological devices.[22]

Adv. Funct. Mater. 2009, 19, 36503661

2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3659

www.afm-journal.de

FULL PAPER

4. Experimental
Materials: Cadmium oxide (CdO, 99.99%), Se powder (99%), HDA (98%), octanoic acid (99%), oleic acid (70%), OA (70%), and octyl ether (99%) were purchased from Aldrich. TOPO (98%) and TOP (90%) were received from Merck and Fluka, respectively. BiCl3 (99.99%) was obtained from Acros. Dimethyl cadmium (CdMe2) was used as received (Strem Chemicals). Bi[N(SiMe3)2]3 was synthesized as described elsewhere [43,58]. TOPSe stock solutions were prepared by dissolving Se into TOP under air free conditions. Preparation of Bi Nanocatalysts: Bi nanocatalysts can be prepared either from Bi[N(SiMe3)2]3 or BiCl3 using TOP as reducing agent at room temperature [43]. Typically, 15.0 mg Bi[N(SiMe3)2]3 was dissolved in 5 mL octyl ether, followed by a dropwise addition of 100 mL of TOP. The mixture was stirred for 30 min and the resultant Bi nanocatalysts were only stabilized with TOP (referred to as Bi@TOP). For comparison, the other two Bi nanocatalyst solutions, i.e., Bi@OA and Bi@HDA, were prepared by addition of OA or HDA as stabilizing ligands. It should be noted that these two nanocatalyst solutions also contain TOP, but are called Bi@OA and Bi@HDA only for convenience. All the Bi nanocatalyst solutions were stored in a glovebox (H2O < 0.1 ppm, O2 < 0.1 ppm, T < 5 8C) and used directly without further purication. Preparation of CdSe Nanowires: The preparations of the CdSe nanowires is in principle based on a standard TOPO/TOP SLS scheme as previously reported [24,43]. While the general procedure will be described in the following, a summary of reaction parameters and quantities of reagents used is given in Table S1 (SI) for each synthesis [45]. For all syntheses a mixture of CdO, TOPO, and oleic acid (or octanoic acid) was loaded into a 50-mL three-necked ask. This mixture was dried and degassed for 30 min at 100 8C under vacuum (1 mbar, 1 bar 103 kPa). Then the ask was backlled with Ar and the temperature was increased to the respective reaction temperature (150330 8C). For the syntheses operated at 150 and 180 8C, the temperature was rstly increased to 200 8C to completely dissolve CdO and then decreased to the corresponding temperature. Afterwards a mixture of Bi nanoparticles and TOPSe was injected. The solution was kept at the reaction temperature for a certain time before cooling down. When the temperature was decreased to 80 8C, 24 mL of toluene was added to the solution to prevent the TOPO from solidifying. Then an excess of butanol (or isopropanol) was added to precipitate the CdSe nanowires, which were separated by decantation after centrifugation at 4000 rpm for 10 min. The dark brown precipitate was washed with toluene/butanol several times and then redissolved in toluene or chloroform. It should be noted that butanol was used instead of short chain alcohols, only if the nanowires should be separated from nanocrystals, because the latter ones precipitate only upon addition of methanol or ethanol [24]. An alternative option to separate nanowires from nanocrystals was the use of high-speed centrifugation (18000 rpm, 10 min). Characterization: Low-resolution TEM images were obtained on a Hitachi-8100 electron microscope operating at an acceleration voltage of 200 kV. HRTEM images were collected on a Philips CM 300 UT microscope operating at an acceleration voltage of 200 kV. TEM samples were prepared by dropping a diluted toluene solution of CdSe nanowires onto carboncoated copper grids. The AFM samples were prepared by putting one droplet of a CdSe nanowire solution on a freshly cleaned Si/SiO2 substrate. The AFM images were recorded on a NanoScope III operated in tapping mode. The nanowire crystal structure was determined by XRD performed in BL9 at DELTA and ROBL beamlines at the ESRF with 15.5 and 25 keV radiation, respectively. The dried CdSe nanowires were ground and then loaded into glass capillaries with a diameter of 0.3 mm. The X-ray energy for ESRF measurements was 25 keV and a point detector with a beam size of 0.2 mm 0.2 mm was used. In DELTA measurements, the X-ray energy was 15.5 keV and an image plate with a beam size of 0.7 mm 0.7 mm was adopted. The distance between the image plate and the sample was 26.4 cm. UVvis spectra were measured with a Cary 50 UVvis spectrometer. The uorescence of nanowire solutions was investigated by a Cary Eclipse uorescence spectrometer. The excitation wavelength was 450 nm and both the emission and excitation slit width were 10 nm.

Acknowledgements
Z. Li gratefully acknowledges the Alexander von Humboldt Foundation for nancial support. Supporting Information is available online from Wiley InterScience or from the author. Received: April 2, 2009 Revised: July 25, 2009 Published online: October 1, 2009
[1] Y.-W. Jun, J.-S. Choi, J. Cheon, Angew. Chem. Int. Ed. 2006, 45, 3414. [2] Y. N. Xia, P. D. Yang, Y. G. Sun, Y. Y. Wu, B. Mayers, B. Gates, Y. D. Yin, F. Kim, Y. Q. Yan, Adv. Mater. 2003, 15, 353. [3] H. Sakaki, J. Cryst. Growth 2003 251, 9. [4] C. Ma, Z. L. Wang, Adv. Mater. 2005, 17, 2635. [5] Y. Yin, A. P. Alivisatos, Nature 2005, 437, 664. [6] C. B. Murray, D. J. Norris, M. G. Bawendi, J. Am. Chem. Soc. 1993, 115, 8706. [7] X. Peng, J. Wickham, A. P. Alivisatos, J. Am. Chem. Soc. 1998, 120, 5343. [8] X. G. Peng, L. Manna, W. D. Yang, J. Wickham, E. Scher, A. Kadavanich, A. P. Alivisatos, Nature 2000, 404, 59. [9] L. Manna, E. C. Scher, A. P. Alivisatos, J. Am. Chem. Soc. 2000, 122, 12700. [10] Z. A. Peng, X. G. Peng, J. Am. Chem. Soc. 2001, 123, 183. [11] Z. A. Peng, X. G. Peng, J. Am. Chem. Soc. 2001, 123, 1389. [12] L. H. Qu, Z. A. Peng, X. G. Peng, Nano Lett. 2001, 1, 333. [13] D. V. Talapin, A. L. Rogach, A. Kornowski, M. Haase, H. Weller, Nano Lett. 2001, 1, 207. [14] Z. A. Peng, X. G. Peng, J. Am. Chem. Soc. 2002, 124, 3343. [15] L. H. Qu, X. G. Peng, J. Am. Chem. Soc. 2002, 124, 2049. [16] X. G. Peng, Adv. Mater. 2003, 15, 459. [17] L. S. Li, J. T. Hu, W. D. Yang, A. P. Alivisatos, Nano Lett. 2001, 1, 349. [18] T. J. Trentler, K. M. Hickman, S. C. Goel, A. M. Viano, P. C. Gibbons, W. E. Buhro, Science 1995, 270, 1791. [19] F. D. Wang, A. G. Dong, J. W. Sun, R. Tang, H. Yu, W. E. Buhro, Inorg. Chem. 2006, 45, 7511. [20] M. Kuno, Phys. Chem. Chem. Phys. 2008, 10, 620. [21] N. Pradhan, H. F. Xu, X. G. Peng, Nano Lett. 2006, 6, 720. [22] W. Lu, C. M. Lieber, Nat. Mater. 2007, 6, 841. [23] H. Yu, J. B. Li, R. A. Loomis, P. C. Gibbons, L. W. Wang, W. E. Buhro, J. Am. Chem. Soc. 2003, 125, 16168. [24] J. W. Grebinski, K. L. Hull, J. Zhang, T. H. Kosel, M. Kuno, Chem. Mater. 2004, 16, 5260. [25] A. T. Heitsch, D. D. Fanfair, H. Y. Tuan, B. A. Korgel, J. Am. Chem. Soc. 2008, 130, 5436. [26] X. M. Lu, D. D. Fanfair, K. P. Johnston, B. A. Korgel, J. Am. Chem. Soc. 2005, 127, 15718. [27] K. L. Hull, J. W. Grebinski, T. H. Kosel, M. Kuno, Chem. Mater. 2005, 17, 4416. [28] J. Puthussery, A. D. Lan, T. H. Kosel, M. Kuno, ACS Nano 2008, 2, 357. [29] M. Kuno, O. Ahmad, V. Protasenko, D. Bacinello, T. H. Kosel, Chem. Mater. 2006, 18, 5722. [30] J. Sun, W. E. Buhro, Angew. Chem. Int. Ed. 2008, 47, 3215. [31] J. Sun, L. Wang, W. E. Buhro, J. Am. Chem. Soc. 2008, 130, 7997. [32] D. D. Fanfair, B. A. Korgel, Chem. Mater. 2007, 19, 4943. [33] H. Yu, W. E. Buhro, Adv. Mater. 2003, 15, 416. [34] H. Yu, J. B. Li, R. A. Loomis, L. W. Wang, W. E. Buhro, Nat. Mater. 2003, 2, 517. [35] D. D. Fanfair, B. A. Korgel, Cryst. Growth Des. 2005, 5, 1971. [36] F. Wang, H. Yu, J. Li, Q. Hang, D. Zemlyanov, P. C. Gibbons, L. W. Wang, D. B. Janes, W. E. Buhro, J. Am. Chem. Soc. 2007, 129, 14327. [37] F. D. Wang, W. E. Buhro, J. Am. Chem. Soc. 2007, 129, 14381. [38] L. Ouyang, K. N. Maher, C. L. Yu, J. McCarty, H. Park, J. Am. Chem. Soc. 2007, 129, 133.

3660

2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Funct. Mater. 2009, 19, 36503661

www.afm-journal.de

[39] A. G. Dong, F. D. Wang, T. L. Daulton, W. E. Buhro, Nano Lett. 2007, 7, 1308. [40] A. Dong, R. Tang, W. E. Buhro, J. Am. Chem. Soc. 2007, 129, 12254. [41] F. Wang, R. Tang, H. Yu, P. C. Gibbons, W. E. Buhro, Chem. Mater. 2008, 20, 3656. [42] J. W. Grebinski, K. L. Richter, J. Zhang, T. H. Kosel, M. Kuno, J. Phys. Chem. B 2004, 108, 9745. [43] Z. Li, A. Kornowski, A. Myalitsin, A. Mews, Small 2008, 4, 1698. [44] V. V. Protasenko, K. L. Hull, M. Kuno, Adv. Mater. 2005, 17, 2942. [45] See supporting information for details. [46] L. Pauling, J. Am. Chem. Soc. 1932, 54, 3570. [47] E. A. Olson, M. Y. Efremov, M. Zhang, Z. Zhang, L. H. Allen, J. Appl. Phys. 2005, 97, 034304. [48] P. Villars, A. Prince, H. Okamoto, Handbook of Ternary Alloy Phase Diagrams, Vol. 5, ASM International, OH, USA 1995. [49] S. Kodambaka, J. Tersoff, M. C. Reuter, F. M. Ross, Science 2007, 316, 729.

[50] U. Shmueli, in Reciprocal space, Vol. B, (Ed: U. Shmueli), Kluwer Academic Press, Dortrecht/Boston/London 2006. [51] H. X. Gao, L. M. Peng, Acta Crystallogr. A 1999, 55, 926. [52] O. Kurtulus, Z. Li, A. Mews, U. Pietsch, Phys. Status Solidi A 2009, 206, 1732. [53] B. E. Warren, X-Ray Diffraction, Dover Publications Inc., New York 1990, chapter 13.6. [54] P. Scardi, M. Leoni, R. Delhez, J. Appl. Cryst. 2004, 37, 381. [55] P. Caroff, K. A. Dick, J. Johansson, M. E. Messing, K. Deppert, L. Samuelson, Nat. Nanotechnol. 2008, 4, 50. [56] J. A. Goebl, R. W. Black, J. Puthussery, J. Giblin, T. H. Kosel, M. Kuno, J. Am. Chem. Soc. 2008, 130, 14822. [57] N. Fu, Z. Li, A. Myalitsin, M. Scolari, R. T. Weitz, M. Burghard, A. Mews, unpublished. [58] M. Vehkamaki, T. Hatanpaa, M. Ritala, M. Leskela, J. Mater. Chem. 2004, 14, 3191.

Adv. Funct. Mater. 2009, 19, 36503661

2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3661

FULL PAPER

You might also like