Synthesis of Hybrid CdS− Au Colloidal Nanostructures

Nov 10, 2006 - Three growth stages are observed when Au growth is performed under air: (1) Au nanocrystal formation at both nanorod tips, (2) growth o...
0 downloads 0 Views 420KB Size
J. Phys. Chem. B 2006, 110, 25421-25429

25421

Synthesis of Hybrid CdS-Au Colloidal Nanostructures† Aaron E. Saunders,‡,§ Inna Popov,§ and Uri Banin*,‡,§ Institute of Chemistry and the Center for Nanoscience and Nanotechnology, The Hebrew UniVersity of Jerusalem, Jerusalem 91904, Israel ReceiVed: August 29, 2006; In Final Form: September 26, 2006

We explore the growth mechanism of gold nanocrystals onto preformed cadmium sulfide nanorods to form hybrid metal nanocrystal/semiconductor nanorod colloids. By manipulating the growth conditions, it is possible to obtain nanostructures exhibiting Au nanocrystal growth at only one nanorod tip, at both tips, or at multiple locations along the nanorod surface. Under anaerobic conditions, Au growth occurs only at one tip of the nanorods, producing asymmetric structures. In contrast, the presence of oxygen and trace amounts of water during the reaction promotes etching of the nanorod surface, providing additional sites for metal deposition. Three growth stages are observed when Au growth is performed under air: (1) Au nanocrystal formation at both nanorod tips, (2) growth onto defect sites on the nanorod surface, and finally (3) a ripening process in which one nanocrystal tip grows at the expense of the other particles present on the nanorod. Analysis of the hybrid nanostructures by high-resolution TEM shows that there is no preferred orientation between the Au nanocrystal and the CdS nanorod, indicating that growth is nonepitaxial. The optical signatures of the nanocrystals and the nanorods (i.e., the surface plasmon and first exciton transition peaks, respectively) are spectrally distinct, allowing the different stages of the growth process to be easily monitored. The initial CdS nanorods exhibit band gap and trap state emission, both of which are quenched during Au growth.

Introduction One of the key goals of nanocrystal research is the development of experimental methods to selectively control the composition1-5 and shape6-12 of nanocrystals over a wide range of material combinations. Over just the past 5 years, significant advances have been made in our understanding of how to control nanoparticle morphology by manipulation of the reaction chemistry or growth kinetics during synthesis,6,13-17 by use of nanocrystal catalysts to promote anisotropic growth,18,19 or through the directed attachment of preformed nanocrystals.8,20,21 Work to develop methods to control nanocrystal composition extends back nearly to the beginning of nanocrystal research. The most common of these methods is the formation of core/ shell semiconductor heterostructures in which the properties of the core and shell semiconductor materials are chosen to engineer the electronic band structure, and hence the optical properties, of the resulting composite particle.22-25 Similarly, strategies to modify the optical or magnetic properties of semiconductor nanocrystals through impurity doping have also advanced significantly over this time frame.5,26-29 Within the past few years, both shape and compositional control have been applied to semiconductor nanoparticles, resulting in discrete domains of different semiconductor materials joined together to form anisotropic shapes.2,16,30 These ideas have also recently been extended beyond semiconductor heterostructure systems to synthesize “hybrid” nanocrystals, which combine disparate material systems (such as metal/magnet,31-35 metal/ semiconductor,1,33,36-40 or magnet/semiconductor33,39,41-43 systems) into one colloidal nanocrystal. †

Part of the special issue “Arthur J. Nozik Festschrift”. * To whom correspondence should be addressed. E-mail: uri.banin@ huji.ac.il. ‡ Institute of Chemistry. § Center for Nanoscience and Nanotechnology.

One of the key challenges in synthesizing multicomponent nanostructures is understanding how to form an interface between materials which may have very different crystallographic structures, lattice dimensions, thermal stability, and/ or chemical reactivity. Use of these new materials then requires identifying and understanding the emergence of new properties due to the combination of different material systems on the nanometer length scale. The optical properties of these nanostructures, for example, often exhibit interesting deviations from either their individual components or from a physical mixture of the two components. These optical effects include a shift in the plasmon resonance of noble metal nanocrystals when combined or coated with other materials31,33 or changes in the photoluminescence intensity of semiconductor nanocrystals which can be attributed to the overlap of the electronic structures of the different components.44 In 2004 we reported a straightforward, room-temperature method for growing gold nanocrystals onto the tips of CdSe nanocrystals, nanorods, and tetrapods.1 Since then, variations on this method have been successfully applied to grow gold nanocrystals onto many other nanoscale materials, such as CoPt3,32 PbS,37 and InAs,38 for example, or onto heterostructure semiconductor nanorods in which one portion of the heterostructures acts as a sacrificial template for subsequent Au growth.40 In this contribution we extend this method of heterogeneous gold nanocrystal growth to include CdS and examine the growth behavior, optical properties, and structural characteristics of the resulting hybrid nanoparticles. Importantly, our findings show that even within the general system of cadmium chalcogenide nanocrystals, significant differences exist in the gold growth mechanism and kinetics when moving between different materials. We demonstrate that gold growth onto CdS nanorods is sensitive to the presence of oxygen: without the presence of oxygen during the reaction, gold growth

10.1021/jp065594s CCC: $33.50 © 2006 American Chemical Society Published on Web 11/10/2006

25422 J. Phys. Chem. B, Vol. 110, No. 50, 2006 occurs at only one end of the nanorod. In contrast, with air in the system, additional nucleation sites are activated which results in Au growth at both CdS nanorod tips, followed by growth onto the nanorod surface, and eventually enhanced growth onto one side through an electrochemical ripening process.45 Because the gold plasmon does not overlap with the excitonic absorbance of the cadmium sulfide, it is possible to separately examine the effects of metal growth on the optical properties of the semiconductor nanorod. The properties of the CdS/Au system presented here may be attractive for a variety of future applications in photocatalysis or photovoltaics. Experimental Section Trioctylphosphine (TOP, 90%), trioctylphosphine oxide (TOPO, 90%),46 cadmium oxide (99.99+%), gold(III) chloride (99%), dodecylamine (DDA, 98%), and didodecyldimethylammonium bromide (DDAB, 98%) were purchased from Aldrich; sulfur was obtained from Merck; and n-tetradecylphosphonic acid (TDPA) was obtained from PolyCarbon Industries Inc. CdS Nanorod Synthesis. CdS nanorods with different aspect ratios were grown through the slow addition of S-TOP to a hot solution containing CdO, TDPA, and TOPO.16 Typically, 7 g of TOPO, 0.83 g (3 mmol) of TDPA, and 0.23 g (1.8 mmol) of CdO were degassed for 1 h at 70 °C and then heated to 320 °C under argon flow. During heating the CdO decomposes in the presence of the phosphonic acid to produce a Cd-TDPA complex, at which point the solution color changes from murky red to clear and colorless.47 Separately, 0.18 g (5.6 mmol) of elemental sulfur are dissolved in 16 mL of degassed TOP at room temperature. The S-TOP solution is injected into the CdTDPA/TOPO solution using a syringe pump, at a rate of 12 mL/h, keeping the growth temperature at 300 °C. After the nanorod mixture is cooled to room temperature, excess organic material could be removed from the solution by adding small amounts of methanol to the crude mixture to induce nanorod precipitation and then centrifuging. The nanorod precipitate could be redissolved in organic solvents for further cleaning or processing. Nanorod samples grown with this method exhibit narrow diameter distributions (approximately 4 nm ( 8%); the length of the sample varies with the growth time or the injection rate, and the average length could be tuned between 20 and 200 nm, with variation in the length of approximately 15-40%. The as-synthesized product also contains a small amount (approximately 5%) of branched structures. The impurity multipods could be removed from the cleaned nanorod solution by adding a few drops of methanol to induce flocculation of the longest nanorods. Centrifuging at 3000 rpm for 10 min resulted in a precipitate containing the nanorods, while most of the tetrapods were left in the supernatant and could be decanted off. The resulting samples contained a substantially lower amount of tetrapods (50 nm) were more likely to exhibit the largest gold tips; shorter rods also had one dominant tip, but it was typically smaller than that for the longest rods (d). Gold nanocrystal growth onto small aspect ratio nanorods, with average dimensions of 3.9 nm × 20 nm after 80 min (e). The time evolution of the gold nanocrystal size distribution for the samples in (b-d) is shown in (f).

As the smaller nanocrystals on the nanorods are now less stable compared with the larger ones, there is a strong thermodynamic driving force for ripening processes to occur, and the larger gold nanocrystals grow via the consumption of smaller ones.49-51 As the histogram for this last stage shows, the majority of the nanocrystal size distribution has shifted toward smaller sizes (reflecting nanocrystal dissolution), with the exception of a single particle on each nanorod that grows at the expense of the smaller Au nanocrystals present on the nanorod, resulting in a second population of particles that range in size from 5 to 15 nm (Figure 1d). Au Growth without Air. Despite the presence of surface defects on the nanorods, which provide low-energy sites for Au reduction and growth and inhibit the ripening process, we find that it is also possible to obtain asymmetric structures with Au nanocrystals grown at only one tip on the CdS nanorods. Figure 2 compares Au nanocrystal growth after 3 h under anaerobic conditions (Figure 2a) and in the presence of air (Figure 2b). Without air present in the reaction system, accomplished by bubbling argon gas through both the CdS nanorod solution and the metal growth solution for half an hour and conducting the growth under an argon atmosphere, Au growth occurs only at one nanorod tip (∼2 nm Au particles can be seen in Figure 2a) and defect growth is almost entirely suppressed. While most of the nanorods in Figure 2a show gold growth at only one tip, there are cases where no gold growth is observed, or where growth occurs at a defect site. Such behavior is likely due to variation from rod to rod in the degree of ligand passivation and the specific surface chemistry at the end of each rod, resulting in a statistical distribution of growth behavior as was seen in the growth of PbSe onto CdS nanorods.52 In contrast, under the same reaction conditions except with air present in the system, gold growth occurs via the same route as shown in Figure 1sat both nanorod tips as well as onto defect sides. The result of asymmetric growth under argon suggests that the nanorod tips are not equally reactive, and in fact that

the presence of air contributes toward making the second tip and sidewall sites available for growth. Optical Properties. The difference in growth behavior between air and argon is also strongly reflected in the absorbance spectrum of the hybrid nanostructures. Figure 3a shows absorption spectra from a gold growth reaction performed under an argon atmosphere, where the nanorod solution and the gold growth solution were both initially saturated with argon. Over the course of the reaction, spanning more than 5 h, the absorbance changes very little: there is a small shift of the absorption edge toward the red, and the first exciton peak and the fine structure show some reduction in intensity, indicative of nucleation and growth of small gold particles on the nanorod tips. Significantly, however, no strong gold plasmon is observed in the absorption spectra, indicating that the majority of gold particles have diameters below approximately 3 nm.53 In contrast, Figure 3b shows the evolution in the hybrid nanocrystal absorption of an aliquot of the reaction mixture which was removed from the argon atmosphere after 180 min (the dark blue curve in Figures 3a and 3b) and exposed to air. Over the same remaining time period as in Figure 3a, the spectra of the hybrid nanocrystals grown under air show the rapid appearance and enhancement of a peak below the CdS band gap, corresponding to a gold plasmon peak. After 1 h of exposure to air (240 min from the beginning of the reaction) a small peak at around 525 nm is evident, which increases in intensity and shifts slightly to 535 nm at the end of the reaction. The intense plasmon peak at the end of the growth stage indicates both that the gold nanocrystals are larger than 3 nm and that there are many of themsi.e., each nanorod contains multiple particles due to growth at both tips and onto sites on the nanorod surface. Concurrently, the CdS first exciton peak decreases in intensity and the peak position shifts to the blue by approximately 20 nm, possibly due to slight etching of the nanorod surface during gold growth or modification of the semiconductor energy levels due to overlap with the metallic states in the Au.

25424 J. Phys. Chem. B, Vol. 110, No. 50, 2006

Figure 2. Comparison of gold growth under argon (a) and under air (b) after 3 h. Without oxygen gold growth is slowed and only occurs at one end, producing small (typically e2 nm) nanocrystals. With oxygen, growth occurs at both tips followed by growth onto defect sites.

The photoluminescence (PL) spectrum of an unmodified sample of CdS nanorods is shown in Figure 3b (dotted line) and exhibits a sharp peak at 460 nm from band-edge recombination along with a broad low-energy peak due to trap-state emission arising from surface defect sites. Band-edge peaks from various CdS nanorod samples are extremely narrow, with typical full-widths at half-maximum (fwhm) between 17 and 20 nm (100-170 meV); the PL peak widths reflect the narrow diameter distribution and are comparable with reported results from monodisperse CdS quantum dots.54,55 The nucleation and growth of Au nanocrystals onto the nanorods completely quenches the PL (Figure 3c, solid line), most likely as the result of electron or energy transfer from an exciton formed in the nanorod to the metal tip.1,44 While many of the reported semiconductormetal hybrid nanostructures exhibit PL quenching,1,33,36 a few recent reports have suggested that this quenching may be related to the crystal structure, geometry, and material combination of the hybrid particles. In 2005, Klimov and co-workers reported the growth of a shell of a polycrystalline CdSe onto spherical Co nanocrystals; the hybrid nanocrystals exhibited PL, possibly due to inefficient electron or energy transfer through grain boundaries within the shell to the metallic core.43 Recent work by Lin et al. explored the optical properties of Au/CdS and FePt/ CdS core/shell nanocrystals.39 In contrast to the CdS/Au nanorods in our work, they found that the PL was enhanced for core/shell Au/CdS nanocrystals, which they attribute to the injection of electrons from the Au surface plasmon into the CdS shell. Lin et al. propose that this electron-transfer process becomes more efficient in core/shell structures, compared with our NDB structures, due to the extended contact between the

Saunders et al.

Figure 3. Evolution of the optical properties during gold nanocrystal growth. Absorbance spectra during gold growth in deaerated solvent under argon flow (a); the number by each spectrum refers to the time (minutes) after the gold growth solution was first added. After 180 min, part of the solution was withdrawn and exposed to air (b), after which the gold plasmon peak develops rapidly at 535 nm, concurrent with reduction of the intensity of the first exciton peak in the CdS absorption. For comparison, aliquots taken from each sample at the same time are shown in the same colors in (a) and (b). The photoluminescence (c) of the unmodified CdS nanorods (dotted line) is completely quenched upon gold growth (solid line) (the excitation wavelength was 440 nm).

plasmon at the Au surface and the CdS coating. Such enhancement may also be material-dependent as well, however, as Lin and co-workers did not observe similar results for core/shell FePt/CdS.39 Previous reports of hybrid nanocrystals have demonstrated that the optical properties of the colloids are not simply a linear combination of the properties of the individual components.31,33 To test whether this was also true for the CdS/Au structures, we compared the optical spectra of CdS nanorods, Au nanocrystals, CdS/Au hybrid nanostructures, and a physical mixture of free Au and CdS nanoparticles (Figure 4). The free CdS nanorods exhibit a strong peak in the absorbance at 451 nm corresponding to the first exciton transition (Figure 4a). The absorbance of the hybrid CdS/Au nanocrystals is shown in Figure 4b (and is the same as the final absorbance spectrum in Figure 3b); compared with the free CdS nanorods, the exciton peak has clearly shifted to higher energies and has also broadened slightly. For comparison, colloidal Au nanocrystals were synthesized as previously reported,56 resulting in a core metal diameter of approximately 6 nm which is coated with an alkylammonium ligand similar to the DDAB used in the hybrid synthesis; the plasmon peak for the sample of free Au nanoc-

Hybrid CdS-Au Colloidal Nanostructures

J. Phys. Chem. B, Vol. 110, No. 50, 2006 25425

Figure 5. Powder X-ray diffraction (XRD) spectra of a CdS nanorod sample before (a) and after (b) Au growth has occurred. The stick spectra represent the expected position and intensity of the most intense XRD reflections for, from top to bottom, bulk hexagonal (wurtzite) CdS, cubic Au, and cubic Au2S, respectively. The diffraction peaks in CdS/Au spectrum (b) can be indexed to a combination of CdS and Au peaks, but do not suggest the presence of significant amounts of Au2S. Figure 4. Comparison of the absorption spectra of (a) free CdS nanorods, (b) hybrid CdS/Au nanostructures such as those shown in Figure 1, (c) a physical mixture of free CdS nanorods and Au nanocrystals, and (d) free Au nanocrystals coated with tetraoctylammonium bromide. The CdS exciton peak and the Au plasmon peak in the hybrid nanostructures (b) are shifted with respect to the individual components and a physical mixture of the components.

rystals in toluene was located at 527 nm (Figure 4d). The plasmon peak of the hybrid CdS/Au particles is red-shifted by 8 nm compared with the pure Au nanocrystals, an effect which could be due to the overlap of the electronic states of the different components of the hybrid particles which modifies the surface plasmon resonance.31,44 Alternatively, the shift may reflect that the Au portion of the hybrid nanocrystals is partially covered with CdS, which possesses a higher index of refraction than toluene or the organic capping ligands. The presence of a material with a higher index of refraction is expected to shift the Au plasmon toward longer wavelengths and has been observed experimentally by varying the refractive index of the solvent57,58 as well as in hybrid nanocrystals containing Au domains.33 Finally, we mixed together various ratios of free Au nanocrystal and free CdS nanorods; the absorbance of one dilute mixture is shown in Figure 4c. Although the magnitudes of the Au plasmon peak and the CdS absorption peak are comparable to the hybrid nanocrystals, the peak locations of the physical mixture clearly do not match that of the hybrid nanocrystals. In fact, the absorbance spectrum of the physical mixture in Figure 4c can be well approximated by a linear sum of the absorbance of the free CdS and Au particles; however, such a simple linear combination is unable to reproduce the spectrum of the hybrid nanocrystals in Figure 4b. Structural Analysis. The CdS/Au hybrid nanocrystals were also analyzed using XRD and HRTEM to evaluate the structural characteristics. Figure 5a shows the powder XRD spectra from unmodified CdS nanorods which can be indexed to the hexagonal (wurtzite) structure. The reflection corresponding to the (002)CdS planes, which run perpendicular to the long axis of the nanorods, is much narrower than the other diffraction peaks due to the anisotropic nanoparticle shape. The (002)CdS peak is also enhanced in intensity relative to what is expected

for the bulk; this may again be due to the anisotropic shape of the nanorods or due to the preferential alignment of nanorods parallel to the substrate during deposition (i.e., the particle orientation is not truly random as in an “ideal” powder diffraction sample). After gold nanocrystals are grown onto the nanorods, additional peaks appear in the XRD spectrum (Figure 5b) corresponding to the powder diffraction pattern for gold. In addition to the observed Au nucleation and growth onto the nanorod surface, a secondary reaction could involve cation exchangesin which Au3+ ions are substituted for Cd2+ ions in the nanorodsas was recently observed by Alivisatos and coworkers for the case of Ag+, Pb2+, and Cu2+ ions when combined with CdS, CdSe, and CdTe nanocrystals.3 The XRD spectrum of the final CdS/Au product (Figure 5b), however, does not suggest the presence of appreciable amounts of Au2S, and we conclude that ion exchange is negligible under these experimental conditions. HRTEM images showing gold nanocrystals grown onto the ends of CdS nanorods are shown in Figure 6. The observed lattice spacings from the nanorods and the particles attached to the surface match the expected spacings for CdS and Au, respectively, confirming the XRD results. Such images were also analyzed to determine if the gold nanocrystals grow heteroepitaxially onto the ends of the gold nanorods, as has been observed for several other hybrid systems.31,33,42 From fast Fourier transforms (FFTs) of the TEM micrographs, the angle between the commonly observed (002) planes of the CdS nanorods and the (111) planes of the Au nanocrystals was determined. If epitaxial growth did occur for this system, we would expect to see a set of discrete, well-defined angles between these sets of lattice planes. For example, from geometrical considerations, if a (110)Au facet was joined to the (002)CdS facet at the end of a nanorod, we should measure only angles of 35.3° and 90° between the (111)Au and (002)CdS planes. Similarly, if the (111)Au and (002)CdS facets were joined, we would expect to measure angles of 0° and 70.5° between the planes. Instead, as can be observed from the histogram of observed angles in Figure 6c, a broad and nearly uniform range of angles is found from the HRTEM images. From the observed

25426 J. Phys. Chem. B, Vol. 110, No. 50, 2006

Saunders et al. that the metal particle be oriented in a certain direction, thus resulting in a lower occurrence of obserVation of these angles. Discussion

Figure 6. High-resolution TEM images of gold nanocrystals grown onto the tips of CdS nanorods (a, b). Measured distribution of angles between the (002) of the CdS nanorods planes and the (111) planes of the gold nanocrystal tips (c).

angles, we expect that low-index gold facets ((100), (110), (111), etc.) are joined to the ends of the CdS nanorods, but conclude that growth is nonepitaxial. Nonepitaxial growth was also observed for CdSe/Au hybrid nanocrystals,1 and in general is not surprising given the differences in crystal structure and lattice spacings between Au and the cadmium chalcogenides. Analysis of the lattice spacings for Au and CdS suggests that in general epitaxial growth could be possible for this system; e.g., the (200)Au and (110)CdS lattice planes have a lattice mismatch of only 1.3%. Growth of an Au nanocrystal onto the end of a CdS nanorod for this situation would require an interface between the (022)Au facet and the (002)CdS facet; the rectangular lattice of the (022)Au facet, however, would necessitate significant lattice distortion for epitaxial growth onto the hexagonal atomic arrangement of the (002)CdS facet. Additionally, while atoms in both the (111)Au and (002)CdS facets have hexagonal arrangements, the lattice mismatch is approximately 30%, which would again require significant distortion for epitaxial growth to occur. In contrast, semiconductor heterostructure systems often have a smaller lattice mismatch between components, and hence exhibit near epitaxial growth. As one example, CdS/HgS/CdS quantum well systems are “lattice matched” (that is, exhibit nearly 0% lattice mismatch) and thus show epitaxial growth.59 Other similar CdS-containing systems, such as ZnS/CdS/ZnS, have a more significant lattice mismatch of ∼7%, which for bulk systems would tend to inhibit good epitaxial growth.60 On the nanometer scale, however, the curvature of the quantum well systems may help to reduce lattice strain, allowing for epitaxial growth in systems which exhibit relatively large lattice mismatches, though we note that these systems typically have much smaller mismatches than the CdS/Au system. We note that, from the histogram, there does appear to be a preferred orientation of the gold nanocrystals that results in a small angle between the measured planes. This result may be somewhat misleading, however, as in these cases the gold nanocrystal may be grown onto the CdS nanorod in such a way that it may not require a specific orientation with respect to the electron beam in order for the (111)Au planes to be visible. Other growth directions for the Au nanocrystals may require

Nanorod Etching and Au Growth. Anisotropic Au growth under airless conditions may be due to the lack of chemical symmetry of the CdS nanorod tips: the unit cell of the wurtzite crystal structure is not centrosymmetric; hence, one end facet of the nanorod is sulfur-terminated while the other end facet is cadmium-terminated.61,62 Under airless conditions, the sulfurterminated end likely represents the lowest energy site for gold nucleation due to the high S-Au bond enthalpy. The high surface energy of the nucleated gold clusters promotes further growth at those sites to increase the gold cluster size and reduce the surface curvature. Under these conditions, Au growth onto the second tip, or onto defect sites on the nanorod surface, is suppressed. A second possibility is that the difference in surface chemistry at the nanorod ends leads to a difference in the ligand coverage.62 Hence, one nanorod apex may be less well passivated and thus much more reactive. We propose that in both cases further Au growth onto these sites in the presence of air is the result of a mild etching process which exposes additional favorable nucleation sites onto the nanorod surface (the growth process is summarized in Figure 7). The differences in surface passivation or chemistry of the terminal nanorod facets may also be responsible for the recent observations of Kudera et al., who studied the epitaxial growth of spherical PbSe nanocrystals onto the ends of CdS nanorods.52 In this work, it was found that the PbSe tended to grow at only one end of the CdS nanorods, and based on the specific chemistry of this system, they reason that the Cd-terminated facets should be the most reactive. Certainly, we would expect differences in the facet reactivity due to differences in chemistry between the noble metals compared with binary semiconductors; for the CdS/Au system, we believe that the etching process attacks the Cd-terminated facet at the end of the nanorod as well as defect sites, exposing sulfur-rich facets. The negative charge on these facets, from the dangling sulfur bonds, is attractive to the positively charged Au ions, and the very high Au-S bond enthalpy (a diatomic bond enthalpy of 418 kJ/mol)63 promotes the deposition and subsequent growth of Au at these sites. Additionally, Manna and co-workers have suggested that the chalcogenide ions may in fact be oxidized to allow Au reduction,40 which would also promote the site-selective growth observed here; in this case, the S2- ions in the nanorods could be oxidized to S0 during the reduction of Au3+. Taken together, these results suggest that by tuning of the reaction conditions or choice of material systems, it is feasible to take advantage of differences in surface termination of nanorods and other structures to generate anisotropic heterostructure and hybrid nanomaterials. We previously observed that a toluene solution of DDA and DDAB will attack the surface of CdSe nanorods, significantly etching both the tips and the sidewalls1 and amine-induced etching of cadmium chalcogenide nanocrystals has also been previously observed in aqueous dispersions.64-66 Addition of typical amounts of DDA and DDAB (without gold) to a dispersion of CdS nanorods did not suggest significant etching over a time period of a few hours: the exciton peak in the nanorod absorption spectrum did not shift to indicate rapid oxide formation or the removal of atomic layers. However, slow etching was observed through a shift of several nanometers in the absorbance peak over 24 h, after increasing the concentration of DDA and DDAB by a factor of 5. The etching rate is

Hybrid CdS-Au Colloidal Nanostructures

J. Phys. Chem. B, Vol. 110, No. 50, 2006 25427

Figure 7. Schematic of the growth process of Au nanocrystals onto CdS nanorods. Without oxygen, gold grows only at one tip (Figure 2a). In the presence of oxygen, the Cd-terminated end is etched, providing a second high-energy site for gold growth (Figures 1b and 1e), followed by side growth (Figure 2b). At long reaction times gold growth ceases and electrochemical ripening leads to the migration of gold from the smallest nanocrystals to the largest nanocrystal.

enhanced or activated when performed under air, and likely relies upon the presence of dissolved oxygen or trace amounts of water in the solvent.64-66 Etching and subsequent gold growth does not appear to be part of a photochemical process, however. Our experiments show that Au growth under air is the same whether the growth is carried out in the dark or if the solutions are exposed to typical laboratory levels of light. Previous experiments with the Au/CdSe system also show that photogenerated carriers in the nanorods do not play a role in this process,1 in contrast to what has been observed for metal growth onto ZnO nanocrystals under intense illumination.67,68 Although the etching of CdS nanorods at normal amine concentrations is slower than that for CdSe, the small amount that does occur appears to be enough to make the second tip and defect sites available for gold growth. Under air, the initial growth of nearly symmetric Au tips suggests that etching of the second tip occurs relatively quickly, compared with etching of the nanorod sides, such that nucleation at both tips occurs nearly simultaneously. Slow etching also occurs at defect sites on the nanorod surface, and eventually Au nucleation and growth occurs here as well. The appearance of gold on the sides of the nanorods provides a physical indication of the presence of defect sites, in addition to the optical signature that normally accompanies such surface states (e.g., the surface trap emission in the PL spectrum in Figure 3c). Au Nanocrystal Ripening. At very long reaction times the concentration of free gold monomer in solution decreases below a critical value and the system transitions from a nanocrystal growth state to a nanocrystal ripening state. This stage is closely related to Ostwald ripening in which the smallest (and hence the least thermodynamically stable) gold nanocrystals dissolve into solution and the mass is transferred to increase the size of the largest nanocrystal on the nanorod. The ripening process was earlier proposed to occur by an electrochemical method in which the oxidation of a gold atom from smaller nanocrystals is followed by the transfer of electrons through surface states along the nanorod to the larger gold nanocrystal, whereupon free gold atoms in solution could be reduced.45 For the previously examined case of Au growth onto CdSe nanorods, ripening of NDB structures leads to the formation of asymmetric structures in which one tip grows while the other tip is

consumed, due to the lack of surface defects. When surface defects were purposely introduced into this system, however, the ripening process was interrupted and gold was deposited onto the defect site.45 In the current CdS/Au system, the presence of surface defects also inhibits the ripening process and does not lead to the formation of a solitary large nanocrystal at each nanorod tip. Instead, while one tip on each nanorod does grow to be much larger than the others (Figure 1d, for example), the process is very slow and gold remains at defect sites even after several days. It appears that even though oxidation from small gold nanocrystals does release electrons into the nanorod, the electrons likely only travel to a nearby defect site where they can reduce free gold atoms in solution. Thus, the gold likely “hops” from defect site to defect site, before it finally reaches the largest nanocrystal. TEM analysis shows that the longest CdS nanorods typically have the largest (>10 nm) Au nanocrystals at the tips after the ripening stage. This observation is most likely due to the availability of larger numbers of gold nanocrystals along the surface of the long nanorods, compared with shorter nanorods, which can be oxidized to provide additional electrons and hence accelerate the growth of the largest gold tip. This observation is consistent with the proposed electrochemical ripening mechanism in which electrons are confined to a single nanorod and cannot be transferred between nanorods in solution.45 Comparison with CdSe/Au Nanodumbbells. We have observed two important differences between our current work with the CdS/Au system and our previous work with CdSe/ Au.1,45 As mentioned previously, the growth kinetics and mechanism are distinctly different between the two systems. Apparently because of differences in the material properties of CdS and CdSe, amine-assisted etching occurs rapidly without air for CdSe promoting Au growth at both nanorod tips. In contrast, etching is much slower without air for CdS, resulting in Au growth onto only one tip of the nanorods under anaerobic conditions. In air, the CdS etching process is enhanced, but still appears to be slower than that for CdSe, which results in a slower overall growth rate of Au. The slower reaction kinetics allow closer monitoring of the Au growth, as the reaction takes place over hours compared with a few minutes for CdSe. Such monitoring of the Au growth is also made easier because the

25428 J. Phys. Chem. B, Vol. 110, No. 50, 2006 absorption from the CdS is significantly blue-shifted with respect to the surface plasmon of the growing Au nanocrystals, which can be clearly seen in Figure 3b. For CdSe/Au, the plasmon and excitonic absorption overlap, making it difficult to quantify the appearance of the surface plasmon. The ability to shift the surface plasmon relative to the semiconductor band gap may prove to be optically interesting; the cadmium chalcogenide/ Au system would allow the surface plasmon to be tuned to lay below the band gap (CdS), at the band gap energy (CdSe), or above the band gap (CdTe). Conclusion We have demonstrated the growth of Au nanocrystals onto preformed CdS nanorods and compared the growth mechanism with our previous work on CdSe nanorods. Although the two material systems are relatively similar, we observe remarkably different Au growth behavior. For CdS, the growth of Au nanocrystals is strongly dependent on the absence or presence of air. When air is carefully excluded during the reaction, Au grows onto only one tip of the CdS nanorods, resulting in asymmetric metal/semiconductor hybrid nanocrystals. The presence of air promotes etching of the nanorod tips and sides due to the presence of amine and provides additional nucleation sites for Au deposition. In this case, symmetric CdS/Au NDBs are obtained, with gold growth occurring at both tips. At later reaction times, gold growth onto defect sites is observed, followed by a ripening mechanism that results in the transfer of gold from the smallest Au nanocrystals on a nanorod to the largest. Understanding these growth processes, and the differences between similar material systems, will provide a basis for the rational design of other hybrid nanomaterials and their inclusion into future technologies. Acknowledgment. The authors thank Taleb Mokari for helpful discussions. This work is supported in part by the EUFP6 under project SA-NANO and by the US-Israel Binational Science Foundation (BSF). A.E.S. thanks the Lady Davis Fellowship Trust for partial financial support through a postdoctoral fellowship. References and Notes (1) Mokari, T.; Rothenberg, E.; Popov, I.; Costi, R.; Banin, U. Science 2004, 304, 1787. (2) Milliron, D. J.; Hughes, S. M.; Cui, Y.; Manna, L.; Li, J.; Wang, L.-W.; Alivisatos, A. P. Nature 2004, 430, 190. (3) Son, D. H.; Hughes, S. M.; Yin, Y. D.; Alivisatos, A. P. Science 2004, 306, 1009. (4) Wang, X.; Zhuang, J.; Peng, Q.; Li, Y. D. Nature 2005, 437, 121. (5) Erwin, S. C.; Zu, L. J.; Haftel, M. I.; Efros, A. L.; Kennedy, T. A.; Norris, D. J. Nature 2005, 436, 91. (6) Peng, X.; Manna, U.; Yang, W.; Wickham, J.; Scher, E.; Kadavanich, A.; Alivisatos, A. P. Nature 2000, 404, 59. (7) Sun, Y. G.; Xia, Y. N. Science 2002, 298, 2176. (8) Tang, Z. Y.; Kotov, N. A.; Giersig, M. Science 2002, 297, 237. (9) Manna, L.; Milliron Delia, J.; Meisel, A.; Scher Erik, C.; Alivisatos, A. P. Nat. Mater. 2003, 2, 382. (10) Jin, R. C.; Cao, Y. C.; Hao, E. C.; Metraux, G. S.; Schatz, G. C.; Mirkin, C. A. Nature 2003, 425, 487. (11) Yin, Y.; Rioux, R. M.; Erdonmez, C. K.; Hughes, S.; Somorjai, G. A.; Alivisatos, A. P. Science 2004, 304, 711. (12) Dumestre, F.; Chaudret, B.; Amiens, C.; Renaud, P.; Fejes, P. Science 2004, 303, 821. (13) Manna, L.; Scher, E. C.; Alivisatos, A. P. J. Am. Chem. Soc. 2000, 122, 12700. (14) Peng, Z. A.; Peng, X. J. Am. Chem. Soc. 2002, 124, 3343. (15) Puntes, V. F.; Zanchet, D.; Erdonmez, C. K.; Alivisatos, A. P. J. Am. Chem. Soc. 2002, 124, 12874. (16) Shieh, F.; Saunders, A. E.; Korgel, B. A. J. Phys. Chem. B 2005, 109, 8538.

Saunders et al. (17) Cozzoli, P. D.; Snoeck, E.; Garcia, M. A.; Giannini, C.; Guagliardi, A.; Cervellino, A.; Gozzo, F.; Hernando, A.; Achterhold, K.; Ciobanu, N.; Parak, F. G.; Cingolani, R.; Manna, L. Nano Lett. 2006, 6, 1966. (18) Kan, S.; Mokari, T.; Rothenberg, E.; Banin, U. Nat. Mater. 2003, 2, 155. (19) Yong, K. T.; Sahoo, Y.; Choudhury, K. R.; Swihart, M. T.; Minter, J. R.; Prasad, P. N. Nano Lett. 2006, 6, 709. (20) Cho, K. S.; Talapin, D. V.; Gaschler, W.; Murray, C. B. J. Am. Chem. Soc. 2005, 127, 7140. (21) Lu, W. G.; Gao, P. X.; Bin Jian, W.; Wang, Z. L.; Fang, J. Y. J. Am. Chem. Soc. 2004, 126, 14816. (22) Hines, M. A.; Guyot-Sionnest, P. J. Phys. Chem. 1996, 100, 468. (23) Dabbousi, B. O.; RodriguezViejo, J.; Mikulec, F. V.; Heine, J. R.; Mattoussi, H.; Ober, R.; Jensen, K. F.; Bawendi, M. G. J. Phys. Chem. B 1997, 101, 9463. (24) Cao, Y. W.; Banin, U. J. Am. Chem. Soc. 2000, 122, 9692. (25) Kim, S.; Fisher, B.; Eisler, H.-J.; Bawendi, M. J. Am. Chem. Soc. 2003, 125, 11466. (26) Mikulec, F. V.; Kuno, M.; Bennati, M.; Hall, D. A.; Griffin, R. G.; Bawendi, M. G. J. Am. Chem. Soc. 2000, 122, 2532. (27) Norris, D. J.; Yao, N.; Charnock, F. T.; Kennedy, T. A. Nano Lett. 2001, 1, 3. (28) Stowell, C. A.; Wiacek, R. J.; Saunders, A. E.; Korgel, B. A. Nano Lett. 2003, 3, 1441. (29) Pradhan, N.; Goorskey, D.; Thessing, J.; Peng, X. G. J. Am. Chem. Soc. 2005, 127, 17586. (30) Talapin, D. V.; Koeppe, R.; Goetzinger, S.; Kornowski, A.; Lupton, J. M.; Rogach, A. L.; Benson, O.; Feldmann, J.; Weller, H. Nano Lett. 2003, 3, 1677. (31) Yu, H.; Chen, M.; Rice, P. M.; Wang, S. X.; White, R. L.; Sun, S. H. Nano Lett. 2005, 5, 379. (32) Pellegrino, T.; Fiore, A.; Carlino, E.; Giannini, C.; Cozzoli, P. D.; Ciccarella, G.; Respaud, M.; Palmirotta, L.; Cingolani, R.; Manna, L. J. Am. Chem. Soc. 2006, 128, 6690. (33) Shi, W. L.; Zeng, H.; Sahoo, Y.; Ohulchanskyy, T. Y.; Ding, Y.; Wang, Z. L.; Swihart, M.; Prasad, P. N. Nano Lett. 2006, 6, 875. (34) Zhang, L.; Dou, Y. H.; Gu, H. C. J. Colloid Interface Sci. 2006, 297, 660. (35) Wang, H.; Brandl, D. W.; Le, F.; Nordlander, P.; Halas, N. J. Nano Lett. 2006, 6, 827. (36) Pacholski, C.; Kornowski, A.; Weller, H. Angew. Chem., Int. Ed. 2004, 43, 4774. (37) Yang, J.; Elim, H. I.; Zhang, Q.; Lee, J. L.; Ji, W. J. Am. Chem. Soc. 2006, 128, 11921. (38) Mokari, T.; Aharoni, A.; Popov, I.; Banin, U. Angew. Chem., Int. Ed. 2006, in press. (39) Lin, H. Y.; Chen, Y. F.; Wu, J. G.; Wang, D. I.; Chen, C. C. Appl. Phys. Lett. 2006, 88, 161911. (40) Carbone, L.; Kudera, S.; Giannini, C.; Ciccarella, G.; Cingolania, R.; Cozzoli, P. D.; Manna, L. J. Mater. Chem. 2006, 16, 3952. (41) Gu, H. W.; Zheng, R. K.; Zhang, X. X.; Xu, B. J. Am. Chem. Soc. 2004, 126, 5664. (42) Kwon, K. W.; Shim, M. J. Am. Chem. Soc. 2005, 127, 10269. (43) Kim, H.; Achermann, M.; Balet, L. P.; Hollingsworth, J. A.; Klimov, V. I. J. Am. Chem. Soc. 2005, 127, 544. (44) Steiner, D.; Mokari, T.; Banin, U.; Millo, O. Phys. ReV. Lett. 2005, 95, 056805. (45) Mokari, T.; Sztrum, C. G.; Salant, A.; Rabani, E.; Banin, U. Nat. Mater. 2005, 4, 855. (46) Some variation in the nanorod synthesis was observed due to varying levels of impurities in different batches of TOPO; these variations could be eliminated by purifying the TOPO via vacuum distillation. (47) Peng, Z. A.; Peng, X. J. Am. Chem. Soc. 2001, 123, 183. (48) While methanol may act as a weak reducing agent for metals, we did not observe significant numbers of free gold particles which would indicate homogeneous Au nucleation upon methanol addition. It is possible that additional gold ions may be reduced onto existing particles, but the difference in Au nanocrystal size compared with samples washed with acetone, or not at all, appears to be