Organosulfur-Functionalized Au, Pd, and Au−Pd Nanoparticles on 1D

Aug 4, 2005 - fine structures (EXAFS)] at the Au L3-, Pd K-, S K-, and Si K-edges. It was also ..... Figure 6 shows the S K-edge XANES of the sample a...
5 downloads 0 Views 516KB Size
Download PDF
8502

Langmuir 2005, 21, 8502-8508

Organosulfur-Functionalized Au, Pd, and Au-Pd Nanoparticles on 1D Silicon Nanowire Substrates: Preparation and XAFS Studies Peng Zhang,† Xingtai Zhou, Yuanhong Tang,‡ and Tsun Kong Sham* Department of Chemistry, University of Western Ontario, London, Ontario, Canada N6A5B7 Received January 13, 2005. In Final Form: June 8, 2005 A hybrid preparative method was developed to prepare organosulfur-functionalized Au nanoparticles (NPs) on silicon nanowires (SiNWs) by reacting HAuCl4 with SiNW in the presence of thiol. A number of organosulfur moleculessdodecanethiol, hexanethiol, 1,6-hexanedithiol, and tioproninswere used to functionalize the Au surface. Size-selected NPs ranging from 1.6 to 7.5 nm were obtained by varying the S/Au ratio and the concentration of HAuCl4. This method was further extended to the preparation Pd and Pd-Au bimetallic NPs on SiNWs. The morphology of the metal nanostructures was examined by transmission electron microscopy (TEM) and high-resolution transmission electron microscopy (HRTEM). The local structure and bonding of the SiNW-supported metal nanostructures were studied using X-ray absorption fine structures (XAFS) [including both X-ray near-edge structures (XANES) and extended X-ray absorption fine structures (EXAFS)] at the Au L3-, Pd K-, S K-, and Si K-edges. It was also found that the annealing of the thiol-capped Au NPs up to 500 °C transforms the surface of the thiol-capped NPs to gold sulfide, as identified using Au L3- and S K-edge XANES. We also illustrate that this preparative approach can be used to form size-controllable Au NPs on carbon nanotubes.

Introduction The chemical preparation of nanomaterials usually involves the use of hard nanosubstrates or soft capping ligands that can be considered nanoreactors where metal and semiconductor nanocrystals are formed.1-10 The hard nanosubstrates refer to materials with rigid nanoscale geometry such as zeolites, porous membrane, nanotubes, and porous silicon,1-6 whereas soft capping ligands refer to molecules, such as surfactants, that can form size/shapechangeable nanoreactors.7-9 The hard nanosubstrate approach to the preparation of nanomaterials exhibits advantages such as simple procedure and short reaction time. Thus, it offers a simple route to immobilizing the nanoproducts onto functional substrates, which is usually required for practical applications. In comparison, the main advantage of using soft capping ligands is the flexibility in controlling the size and shape of the products.7-10 It is natural to ask whether we can combine the advantages of both approaches to prepare desired nanomaterials. The preparative approach described here is based on the recent findings that the reduction of metals in the presence of thiols can result in metal nanoparticles * Corresponding author. Phone: 519-6612111 ext. 86341. Email: [email protected]. † Present address: Department of Chemistry, McGill University, Montreal, Quebec, Canada. ‡ Present address: College of Materials Science and Engineering, Hunan University, Changsha, Hunan, P. R. China. (1) Martin, C. R. Science 1994, 266, 1961. (2) Sloan, J.; Kirkland, A. I.; Hutchison, J. L.; Green, M. L. H. Acc. Chem. Res. 2002, 35, 1054. (3) Zach, M. P.; Ng, K. H.; Penner, R. M. Science 2000, 290, 2120. (4) Deng, Z.; Mao, C. Nano Lett. 2003, 3, 1545. (5) Nakao, H.; Shiigi, H.; Yamamoto, Y.; Tokonami, S.; Nagaoka, T.; Sugiyama S.; Ohtani, T. Nano Lett. 2003, 3, 1391. (6) Coulthard, I.; Zhu, Y.-J.; Degan, S.; Sham, T. K. Can. J. Chem. 1999, 76, 1707. (7) Pileni, M. P. J. Phys. Chem. 1993, 103, 6961. (8) Peng, X.; Manna, L.; Yang, W.; Wickham, J.; Scher, E.; Kadavanich, A.; Alivisatos, A. P. Nature 2000, 404, 59. (9) Jana, N. R.; Gearheart, L.; Murphy, C. J. Chem. Commun. 2001, 617. (10) Sun, Y.; Xia, Y. Adv. Mater. 2002, 14, 833.

(NPs) of a few nanometers11-14 and that nanostructures such as porous silicon, silicon nanowires (SiNWs), germanium nanowires, and carbon nanotubes (CNTs) can serve as both hard nanosubstrates and reducing agents for the growth of metal nanocrystals.15-27 We will show herein that by simultaneously using soft capping ligands (organosulfur molecules) and hard nanosubstrates (SiNWs) in the preparation, Au NPs of 1.6-7.5 nm in size can be prepared in a simple, fast, and controlled manner. This method is then extended to the preparation of NPs such as Pd and Pd-Au on SiNWs and size-controllable Au NPs on CNTs. The morphology of the nanostructures is revealed by transmission electron microscopy (TEM). The structure and bonding properties of these metal nanostructures are investigated using synchrotron X-ray absorption fine structures (XAFS), including both X-ray near-edge structures (XANES) and extended X-ray ab(11) Brust, M.; Walker, M.; Bethell, D.; Schiffrin, D. J.; Whyman, R. J. Chem. Soc., Chem. Commun. 1994, 801. (12) Templeton, A. C.; Chen, S.; Gross, S. M.; Murray, R. W. Langmuir 1999, 15, 66. (13) Templeton, A. C.; Wuelfing, W. P.; Murray, R. W. Acc. Chem. Res. 2000, 33, 27. (14) Huang, T.; Murray, R. W. J. Phys. Chem. 2003, 107, 7434. (15) Zhang, P.; Kim, P. S.; Sham, T. K. Appl. Phys. Lett. 2003, 82, 1470. (16) Zhang, P.; Sham, T. K. Appl. Phys. Lett. 2003, 82, 1778. (17) Coulthard, I.; Jiang, D. T.; Lorimer, J. W.; Sham, T. K. Langmuir 1993, 9, 3443. (18) Sham, T. K.; Coulthard, I.; Lorimer, J. W.; Hiraya, A.; Watanabe, M. Chem. Mater. 1994, 6, 2085. (19) Sun, X. H.; Peng, H. Y.; Tang, Y. H.; Shi, W. S.; Wong, N. B.; Lee, C. S.; Lee, S. T.; Sham, T. K. J. Appl. Phys. 2001, 89, 6396. (20) Sun, X. H.; Li, C. P.; Wong, N. B.; Lee, C. S.; Lee, S. T.; Teo, B. K. Inorg. Chem. 2002, 41, 4331. (21) Mayer, B.; Jiang, X.; Sunderland, D.; Cattle, B.; Xia, Y. J. Am. Chem. Soc. 2003, 125, 13364. (22) Wen, X.; Yang, S. Nano Lett. 2002, 2, 451. (23) Li, Q.; Wang, C. Chem. Phys. Lett. 2003, 375, 525. (24) Choi, H. C.; Shim, M.; Bangsaruntip, S.; Dai, H. J. Am. Chem. Soc. 2002, 124, 9058. (25) Xue, B.; Chen, P.; Hong, Q.; Lin, J.; Tan, K. L. J. Mater. Chem. 2001, 11, 2378. (26) Banerjee S.; Wong, S. S. Nano. Lett. 2002, 2, 195. (27) Banerjee S.; Wong, S. S. J. Am. Chem. Soc. 2003, 125, 10342.

10.1021/la0501031 CCC: $30.25 © 2005 American Chemical Society Published on Web 08/04/2005

Au, Pd, & Au-Pd Nanoparticles on Silicon Nanowires

Langmuir, Vol. 21, No. 18, 2005 8503

Figure 1. Representative TEM images of (a) as-prepared SiNWs, (b) gold on SiNWs in the absence of thiol (Au-NW), and (c) gold on SiNWs in the presence of dodecanethiol (AuSRNW). Scale bars of a-c: 50 nm.

sorption fine structures (EXAFS), at the Au L3-, Pd K-, S K- and Si K-edges. EXAFS data were analyzed with the Winxas software package. Experimental Section The SiNWs were prepared by a slightly modified thermal evaporation method reported by Shi et al.28 As seen from the TEM image in Figure 1a, they are several micrometers long with diameters of ∼20-30 nm. The metal nanocrystals were prepared by the following procedure. First, stock solutions containing HAuCl4 and organosulfur molecules (alkanethiol in methanol and tiopronin in water, see Figure 3 for their molecular structures) were prepared with the desired Au/S ratios (from 1:6 to 6:1) and concentrations (1-0.01 mM). Next, SiNWs (5% hydrofluoric acidtreated for 2 min) or CNTs (20-30 nm in diameter and 1 µm in length, used as received from Alfa) were placed into the HAuCl4/ organosulfur solution for 5 min. After this, the Au/SiNWs or Au/CNTs samples were rinsed with ethanol or distilled water several times to remove the unreacted chemicals and were allowed to dry in air. Other metal NPs such as Pd and Pd-Au on SiNWs were prepared in an identical manner using Pd(NO3)2 and the mixture of Pd(NO3)2 and HAuCl4, respectively. The morphology of the products was characterized with TEM by either sandwiching the samples between a double-ring copper grid or casting the ultrasonicated samples onto a carbon-coated copper grid. The Au L3- and Pd K-edge XAFS measurements were conducted at the Pacific Northwest Consortium Collaborative Access Team (PNC-CAT) ID-20B beamline of the Advanced Photon Source (APS) at the Argonne National Laboratory, and the S K- and Si K-edge XAFS measurements were conducted at the Canadian Synchrotron Radiation Facility (CSRF) DCM beamline at the Synchrotron Radiation Center (SRC), University of Wisconsin-Madison.

Figure 2. (a) Au L3-edge XAFS raw data (inset XANES), (b) k-space EXAFS, (c) FT-EXAFS, and (d) S K-edge XANES of Au foil, AuNW, and AuSC12NW samples.

Results and Discussions 1. Gold on SiNWs. 1.1. Au on SiNWs with and without Thiols: A Comparison. Figure 1b,c shows the TEM images of gold reductively formed on HF-refreshed SiNW onedimensional (1D) substrate without and with thiol capping agents, respectively. It is immediately apparent from Figure 1b,c that, in the presence of dodecanethiol, the size of Au NPs is considerably smaller than that of those formed in the absence of thiols. The reductive formation of gold on a silicon surface has been well-documented.16,17-20 However, the size of gold NPs is usually on the order of 10-100 nm17-20 because of the fast growth of Au upon its reduction on the silicon surface. This is clearly seen in Figure 1b, in which the size of Au NPs (denoted AuNW) is in the range of 10-30 nm. Because of their

Figure 3. Representative TEM images of gold on SiNWs in the presence of (a) dodecanethiol, (b) hexanethiol, (c) hexanedithiol, and (d) tiopronin. The four samples were all prepared using HF-treated SiNWs and 1 mM HAuCl4, with a Au/S ratio of 6:1. Samples in a-c were prepared in methanol, and d was prepared in water. Scale bars of a-d: 50 nm. Molecular structures of the capping molecules are also illustrated in each figure.

(28) Shi, W. S.; Peng, H. Y.; Zheng, Y. F.; Wang, N.; Shang, N. G.; Pan, Z. W.; Lee, C. S.; Lee, S. T. Adv. Mater. 2000, 12, 1343.

large size, the Au NPs tend to separate from the 1D substrate.19 In the presence of thiol, however, further

8504

Langmuir, Vol. 21, No. 18, 2005

Zhang et al.

growth of Au particles is inhibited because of the wellknown gold-sulfur self-assembling interaction. As shown in Figure 1c, the sizes of the thiol-capped Au NPs (denoted AuSC12NW, for dodecanethiol) are mainly within the range of 2-3 nm. These results clearly indicate that the simultaneous use of thiol soft capping ligands and SiNWs hard nanosubstrates can significantly reduce the NP size relative to the NPs prepared using only SiNWs hard nanosubstrates. We then used Au L3-edge XAFS to further compare the structural properties of bulk Au (foil), AuNW, and AuSC12NW. The characteristic face-centered cubic (fcc) structure of the three Au samples is clearly observed in both Au L3-edge XANES and EXAFS (Figure 2a,b). As shown in Figure 2a and the inset, the characteristic three-peak pattern following the edge jump at ∼11.919 keV is an indication of the existence of an fcc structure.29,30 In the postedge region, the nanosize effect of the AuSC12NWs is evident in that the resonance peak is significantly broadened and attenuated relative to that of the foil and the non-thiol-capped Au, the postedge features of which are identical. This fact is clearly seen in the EXAFS in the k space in Figure 2b in which the EXAFS oscillations of AuSC12NW, the pattern of which is identical to those of Au foil and AuNW, are considerably broadened and attenuated. The nanosize effect observed in the Au L3edge XANES and EXAFS can be understood by considering the increase in the percentage of surface atoms, which results in a decrease in the number of neighboring Au atoms, on average, when the NP size decreases. In the Fourier transform (FT) of the EXAFS shown in Figure 2c, the nanosize effect in thiol-capped Au NPs can be seen from two aspects. One is the decrease in magnitude of the first-shell Au-Au peak due to the lower coordination number of Au, on average, in the NPs and to increasing disorder, which is in agreement with the observation in Figure 2a,b. The other is the emergence of the surface S-Au contribution near 2 Å, which is usually observable for very small thiol-capped metal NPs, for which the percentage of surface atoms increases markedly.29,30 The existence of Au-S bonding is also revealed in the S K-edge XANES in Figure 2d, in which the S K-edge whiteline (1s-3p transition) in the thiol-Au NPs is blue-shifted by ∼0.7 eV relative to the free thiol, and their postedge features are completely different because of the formation of the Au-S bond.15,29 1.2. Au/SiNWs with Various Capping Molecules. The spontaneous interaction between gold and organosulfur allows us to extend this method to a large variety of organosulfur molecules with a thiol functional group. This approach is of great practical importance because it provides chemical and size tunability in the selection of capping molecules. Thus, it opens up a simple pathway to prepare and immobilize surface-functionalized Au NPs on SiNWs and to introduce functional groups onto the 1D nanosubstrates. Four commercially available organosulfur moleculessdodecanethiol, hexanethiol, hexanedithiol, and tioproninswere chosen in this study. In Figure 3, we show the TEM images of the Au NPs of several nanometers formed on SiNWs with all four different capping molecules. Several trends are observed. First, when the capping molecule is changed from dodecanethiol to hexanethiol, the NP packing density is increased considerably, whereas the mean size of Au NPs is only slightly increased. The increase in NP size from dodecanethiol to hexanethiol is understandable because

the chain-length of hexanethiol is considerably shorter; thus, it is less effective to prevent the NP from further growth than dodecanethiol. The increase in NP packing density on SiNWs is useful in understanding the mechanism for the reaction. It implies that the formation of gold on SiNWs is a diffusion-controlled process if one keeps in mind that the mixture of HAuCl4 and thiol exists in the form of polymeric gold-thiolate in solution31 and that the polymeric species resulting from shorter-chain thiols move faster and reach the SiNWs surface much more easily. This observation indicates that it is possible to tailor both the NP size and the packing density on the 1D substrates by choosing proper organosulfur capping molecules. Second, we focus on hexane-dithiol, which can be considered a double-anchoring capping molecule. As might be expected, the TEM micrograph in Figure 3c clearly shows the existence of a considerable amount of complex nanostructures consisting of two or more linking Au NPs (two representative areas are marked by white circles in the figure). Finally, tiopronin, a water-soluble molecule is chosen to form Au NPs on SiNWs in aqueous solution. In Figure 3d, it can be seen that Au NPs of ∼5 nm with a broad size distribution are formed. The above observations clearly show that the molecular structure of the organosulfur molecule plays an important role in determining the size distribution of the NPs. It should be noted that although it is clear that the presence of thiol molecules controls the size of the Au NP, it is not very clear at present whether there is also a significant interaction between the SiNWs and different organosulfur molecules that also influence the growth of the NPs. For instance, the HF-refreshed SiNWs is terminated with Si-H bonds that are hydrophobic. Thus, the effect of the interactions between the SiNWs and various organosulfur molecules (hydrophilic or hydrophobic) on the size and packing density of the NPs cannot be ruled out. However, it is certain that the Au NPs themselves can be directly bound to the SiNWs because Au NPs without capping ligands are known to form on SiNWs6,20 and that the Authiol interaction is stronger than that of Si-thiol. On the basis of all the observations in Figure 3, we believe it is possible to use a large variety of HS-R molecules to functionalize the immobilized Au NPs on SiNWs and, more importantly, by choosing HS-R molecules with the proper molecular structures (or functional groups), one can considerably tune the size, packing density, and structure of the capped NPs. 1.3. Size Control of Tiopronin-Capped Au NPs on SiNWs. We then chose tiopronin as an example to further explore the possibility of tuning the NP size because tiopronin is bioactive and water soluble, and the NPs capped with it show intriguing physical properties.14 It has been welldocumented that the size of colloidal thiol-capped Au NPs can be controlled by varying such parameters as thiolAu ratio and solution concentration.11-13 Here we adapted a similar strategy to control the size of tiopronin-capped Au NPs on SiNWs. From Figure 4a-c, it can be seen that Au NPs with an average size of 1.6, 2.4, and 4.9 nm can be obtained by keeping a constant HAuCl4 concentration (1 mM) while varying the Au/S ratio from 1:6, 1:1, and 6:1. These results are noteworthy in that they show that this method can provide 1D-substrate-supported Au NPs with sizes comparable to those colloidal Au NPs prepared with the classic two-phase method.10-12 We then kept the Au/S ratio as 1:1 and decreased the concentration of HAuCl4 100-fold to 0.01 mM. It is shown in Figure 4d that Au NPs

(29) Zhang, P.; Sham, T. K. Phys. Rev. Lett. 2003, 90, 245502. (30) Zhang, P.; Sham, T. K. Appl. Phys. Lett. 2002, 81, 736.

(31) Bourg, M.; Badia, A.; Lennox, R. B. J. Phys. Chem. B 2000, 104, 6562.

Au, Pd, & Au-Pd Nanoparticles on Silicon Nanowires

Langmuir, Vol. 21, No. 18, 2005 8505

Figure 4. Representative TEM images of tiopronin-capped Au on SiNWs prepared with 1 mM HAuCl4 with a Au/S ratio of (a) 1:6, (b) 1:1, and (c) 6:1. (d) Tiopronin-capped Au on SiNWs using 0.01 mM HAuCl4 with a Au/S ratio of 1:1. Scale bars of a-d: 30 nm. (e) A typical HRTEM image of the sample shown in a. (f) Au L3-edge XANES of the gold foil and NPs presented in a-d.

of even larger size (∼7.5 nm) are formed. This is surprising because it has been reported that, in the absence of thiols, the size of metal NPs reductively formed on SiNWs and porous silicon decreases with decreasing metal concentration.17,18 The discrepancy observed here indicates that the concentration of the capping molecules also plays an important role in determining the NP size in this method. It has been reported32 that, in the case of a two-dimensional (2D) self-assembled monolayer (SAM) of thiol on Au, at lower thiol concentration, the formation of SAM is much slower. This observation helps one to understand why the NPs found in Figure 4d are larger relative to those found (32) Ulman, A. An Introduction to Ultrathin Organic Films, Academic Press: San Diego, 1991.

in Figure 4b because a slower capping effect can result in the continued growth of the Au NPs (i.e., capping competes with Au aggregation). The crystalline structure of the smallest Au NPs (1.6 nm on average) is confirmed by the high-resolution transmission electron microscopy (HRTEM) image shown in Figure 4e, in which a value of 0.23 nm, which is typical for the fcc gold d-d spacing, is found. In addition, the HRTEM image in Figure 4e reveals that the surface of SiNWs is partially covered with amorphous silicon oxide. The oxides are the product of the redox reaction between the gold compound (oxidant) and hydrogen-passivated silicon surface (reductant).19 The effect of metal reduction on the property of SiNWs will be further discussed below.

8506

Langmuir, Vol. 21, No. 18, 2005

Figure 5. Au L3-edge XANES of the 1.6-nm Au NPs on SiNWs annealed at various temperatures. (a) Spectra with baseline shifted vertically for calarity. (b) Spectra with edge jump normalized to unity showing the annealing temperature dependence of the whiteline.

In Figure 4f, the Au L3-edge XANES provides spectroscopic evidence of the existence of crystalline gold in all the tiopronin-capped NP samples. The broadening and attenuation of the postedge resonance in XANES when the NP size decreases, particularly in the region marked by the arrow in Figure 4f, is a clear indication that the gold absorber is surrounded by less neighboring atoms (i.e., a lower coordination number) and that there is an increase in the disorder and degradation of the long-range order in the NPs.15 1.4. The Effect of Annealing Studied By XANES. The SiNW-supported 1.6-nm Au NPs described above were annealed in an Ar environment for 2 h at temperatures of 80, 200, and 500 °C. The samples were then monitored with Au L3-edge XANES (shown in Figure 5). As shown in Figure 5a, the characteristic fcc XANES pattern observed in the preannealed sample (25 °C) was essentially maintained at all temperatures albeit with increasing broadening, but the whiteline intensity changes somewhat at 200 °C and changes more pronouncedly at 500 °C. When aligned at the edge threshold and normalized to the unit edge jump, the XANES spectra show an increased whiteline intensity on the order of 25-80 < 200 < 500 °C. It has been generally recognized that the whiteline (i.e., the first peak after the edge jump) of Au L3-edge XANES is mainly due to the 2p3/2 f 5d electronic transition. The area under the Au L3-edge whiteline is related to the unoccupied d states of Au character. A more intense whiteline relative to metallic gold in the bulk corresponds to either more unoccupied d states at the Au site or d-charge depletion relative to elemental Au. The Au L3-XANES data shown in Figure 5 clearly indicate that increasing the annealing temperature (to 200 and 500 °C) results in a more intense whiteline and, hence, more unoccupied d-states. To further investigate this behavior of AuSRNW upon annealing, we used S K-edge XANES to obtain elementspecific structural information from the sulfur perspective. Figure 6 shows the S K-edge XANES of the sample at 25 and 500 °C. It is evident from Figure 6 that upon heating the sample, the Au-S interaction, as revealed by the resonance at 2475 eV, becomes more intense in the XANES spectra. The Au-S peak in the S K-edge XANES in thiolcapped Au NPs is usually very weak,29 whereas that of metal sulfide is more intense. The results in Figure 6 imply that at high annealing temperatures, the surface of the thiol-capped Au NP was turned into sulfide-capped NP. This notion is in agreement with the results in Figure 5, which show that a stronger gold-sulfide interaction induces a more significant charge transfer from Au to S because sulfur has a greater electronegativity than

Zhang et al.

Figure 6. S K-edge XANES of the 1.6-nm Au NPs on SiNWs which are as-prepared and annealed at 500 °C.

Figure 7. Pd K-edge FT-EXAFS of Pd foil, PdNW, and PdSRNW samples.

gold.29,30 The core remains fcc Au, however, because the XANES feature beyond the whiteline remains essentially unchanged. 2. Pd NPs on SiNWs. In an attempt to extend this preparative method to other metals, tiopronin-capped Pd NPs on SiNWs were prepared. Like the Au NPs, the organosulfur-capped Pd NPs are much smaller than those prepared without thiol (TEM results not shown). The structure of the Pd samples was examined with Pd K-edge EXAFS, as depicted in Figure 7. It can be seen from the FT of the EXAFS that the amplitude of the first-shell Pd-Pd interatomic distance decreases considerably on the order of Pd foil > PdNWs > PdSRNW, which is in agreement with nanostructure morphology observed by TEM. Furthermore, in the capped Pd NPs, the first-shell interatomic distance no longer bears a fcc signature, and a new contribution in the shorter bond distance region emerges. It is plausible that this shorter bonding feature originates from the Pd-S or Pd-Si interaction. It is conceivable that PdSRNW is a small cluster capped by the thiol molecules, as opposed to being an NP with fcc structure. 3. Au-Pd Bimetal NPs on SiNWs. Using a similar method, we prepared Au-Pd bimetal NPs with and without tiopronin capping ligands. It was found that NPs of 1-3 nm were formed on the SiNWs substrates. The structure and bonding of the Pd-Au NPs together with a series of reference samples, including bulk Au, AuPd alloys (1:3 and 3:1 ratios), and SiNW-supported NPs of Au, AuSR, and Pd-Au (without capping molecules), were studied by Au L3-edge XANES and EXAFS. The Au L3edge XANES and EXAFS techniques have been established as powerful tools to study the structure and bonding in Au-Pd alloys.33,34 (33) Davis, R. J.; Boudart, M. J. Phys. Chem. 1994, 98, 5471.

Au, Pd, & Au-Pd Nanoparticles on Silicon Nanowires

Langmuir, Vol. 21, No. 18, 2005 8507

Figure 8. (a) Au L3-edge XANES and (b) FT-EXAFS of a series of gold and alloy samples.

Generally, two important features can be seen from the Au L3-edge XAFS upon Au-Pd alloy formation. First, the whiteline intensity of Au, which is closely related to the d-charge population, decreases noticeably upon alloy formation.33,34 The Au L3-edge whiteline arises from a dipole transition from the 2p3/2 level to the unoccupied densities of states with 5d3/2,5/2 character. In atomic Au, the electronic configuration is 5d106s1, which means that the 5d orbital is full. In the solid state, hybridization of the 5d state with the 6s and 6p states results in some unoccupied d-states above the Fermi level, the amount of which is ∼0.401 (d-holes) according to theoretical calculations.30 In a Au-Pd alloy, there are fewer like nearest Au neighbors for the Au atoms, which reduces the dominant Au-Au interactions.33,34 Thus, the electronic configuration of the alloy should be intermediate between the atomic Au (with no d-hole) and the bulk Au (with ∼0.4 d-hole). Thus, the Au 5d unoccupied states in the alloy should be less than those of the bulk Au, resulting in a decreased whiteline intensity.35 Second, the bond distance of PdAu alloys should be shorter than that of Au metal and longer than that of Pd metal (to a first approximation), with the extent of the change being linearly dependent on the concentration of Pd (Pd-Pdnearest 2.75 Å) and Au (AuAunearest 2.88 Å). In Figure 8, some of the above-noted features are observed in the XANES and in the FT of the EXAFS of the Au-Pd bulk alloy samples. The AuPdNW sample shows a considerably decreased whiteline intensity, whereas the bond length difference cannot be straightforwardly extracted from the FT. It should be noted that the FT values are used here qualitatively in a fingerprint manner; hence, the apparent two-peak feature does not automatically indicate the presence of two bond lengths in this case. This is because the phases of both the absorber and the backscatterer atoms and the amplitude of the backscatterer atom are not linear in k space (Å-1) but are modulated instead. This is further complicated by multiple scattering pathways and disorder in the alloy. Thus, the Au FT does not appear as a clean single peak [this is the case even in pure Au or more generally in high Z (atomic (34) Keifsnyder, S. N.; Lamb H. H. J. Phys. Chem. B 1999, 103, 321. (35) It should be noted that Au-Au interaction is the strongest in Au binary alloys. Au is the most electronegative metallic element. In general, Au tends to withdraw charge from the other metallic constituent in alloys via a charge compensation mechanism in which it loses 5d charge and gains 6s charge, resulting in a more intense L3-edge whiteline. The 6s charge gain often overcompensates the 5d charge loss, therefore the overall charge redistribution is in agreement with electronegativity arguments. Thus, reduced whiteline intensity is more an exception than a norm and only appears in Au-noble metal alloys. See Bzowski, A.; Yiu, Y. M.; Sham, T. K. Phys. Rev. B 1995, 51, 9979 for details.

Figure 9. Si K-edge XANES of plain SiNW, AuNW, and AuSRNW samples.

number) absorber and high Z scatterer situations]. The resemblance between the XANES and the FT for AuPdNW and those of bulk Pd-Au alloy clearly indicates alloy formation in the AuPdNW sample. Interestingly, when organosulfur capping molecules were used, the appearance of the XANES features in Figure 8a of the bimetal sample (AuPdSRNW) is considerably different from the alloy (Au3Pd and AuPdNW) and gold samples (foil, AuNW, and AuSRNW). It can be seen that the whiteline intensity of the AuPdSRNW sample is even larger than that of the Au foil, indicating a decrease in d-electron density at the Au site in the sample. This can be attributed to the capping effect of the more electronegative thiol molecules. Meanwhile, the postedge XANES features (Figure 8a) are greatly broadened and dampened relative to those of the other Au samples, which corresponds to a lower coordination number for the Au absorber and increased disorder, even though the characteristic three-peak feature for fcc structure is still present. It has been reported that thiolcapped Au NPs tend to lose d-electrons because of the charge transfer from Au to S, and this effect becomes more pronounced when the NP size decreases.29,30 The fact that the Au atoms in AuPdSRNW lost more d-charge (i.e., Au L3-whiteline intensity increased) relative to the bulk Au and AuSRNW indicates the existence of very small organosulfur-capped Au nanoclusters. This notion is supported by the observation of dampened postedge XANES features. This feature comes about when there is a significant static disorder (overlapping of EXAFS arising from surface atoms and atoms in the core having significantly different oscillatory behavior in their EXAFS). The very short first-shell bonding peak of AuPdSRNW may thus be a reflection of the dominant contribution of the surface Au-S bond, which is much shorter than the AuAu bond. These results lead to the notion that the AuPd bimetal NPs prepared in the presence of tiopronin do not have the same structure as those in bulk alloys. Instead, very small organosulfur-capped nanoclusters were formed. The structure of these clusters awaits further investigation (both experimental and theoretical), a situation reminiscent of the observation in PdSRNW discussed in section 2. 4. Effect of Metal Reduction on the SiNW Substrates. The effect of metal (Au and Pd) reduction on the SiNW substrates was monitored by Si K-edge XANES. As shown in Figure 9, the near-edge feature of the as-prepared SiNWs near 1840 eV arises from the resonance (1s to 3p character in the conduction band) in elemental silicon,

8508

Langmuir, Vol. 21, No. 18, 2005

Zhang et al.

size-tunable tiopronin-capped Au NPs on multiwalled carbon nanotubes (MWNT) (shown in Figure 10). In comparison, if no capping molecules were used, much bigger Au NPs, which tend to separate from the CNTs, were always formed. Thus, we have demonstrated the advantages of this hybrid preparative method; that is, that it not only allows for the functionalization of the NP surface, but it also provides a facile route to controlling the size of the 1D-substrate-supported metal NPs. Figure 10. Representative TEM images of tiopronin-capped Au on an MWNT using 1 mM HAuCl4 with Au/S ratios of (a) 6:1 and (b) 1:1.

whereas the peak near 1848 eV arises from silicon dioxide (SiO2). It can be seen that upon metal (both Au and Pd) reduction, the intensity of the Si resonance decreases considerably, whereas the contribution of SiO2 increases. Furthermore, in the presence of organosulfur capping molecules, the intensity of the Si peak becomes even lower than that of the Si peak generated without using capping molecules. It has been proposed that the reduction of metals by SiNWs involves the use of a Si-H bond as the reducing agent, and, consequently, the silicon is oxidized to silicon oxide. The observation in Figure 9 that, upon metal reduction, the intensity of the Si peak decreases, whereas the intensity of the Si-O peak increases, is in good agreement with this proposed mechanism. The fact that there is less contribution from the Si peak when organosulfur molecules are used indicates that more silicon sites were involved in the redox reaction. This is understandable because, in the presence of capping agents, the growth of metal nanocrystals was largely prohibited by the Au-S interaction, and thus more nuclei were created, covering a larger area of the surface, instead of forming bigger nanocrystals. 5. Size-Controlled Au NPs on CNT 1D Substrates. Finally, we want to illustrate that this hybrid preparative method can be readily applied to the preparation of sizecontrolled metal NPs on different 1D substrates. Theoretically, a number of 1D substrates with reductive activity17-25 should also be usable in this method. An interesting example is the CNT, for which the attachment of NPs is considered a route to add new functions.23-26 Using a completely analogous approach, we have prepared

Summary and Conclusion We have reported a hybrid preparative approach in which organosulfur ligands and SiNWs are simultaneously used to prepare SiNW-immobilized and surface-functionalized metal NPs. The preparative procedure is systematically studied by varying the type of organosulfur ligands, the metal compound reactants, and their concentrations. It is found that the size, packing density, and composition of the metal NPs can be controlled. This technique has a considerable advantage over Au NP formation in the absence of thiol. In the latter technique, Au NPs are often larger and nonuniform.6,15,16,20 The local structure and bonding of the SiNW-supported metal and bimetal NPs, including the Au NPs annealed at various temperatures up to 500 °C, have been revealed from the results of XANES and EXAFS at the Au L3-, Pd K-, S K-, and Si K-edges. The preparative method developed here is expected to be applicable to other 1D substrate systems. We also demonstrated here that a similar procedure was successfully applied to 1D CNT systems. Acknowledgment. The APS is funded by the U.S. DOE under contract No. W-31-109-Eng-38, and the SRC is funded by the U.S. NSF grant DMR-00-84402. The PNCCAT beamline is funded by the U.S. DOE and the NSERC of Canada, and the CSRF is funded by the NRC and NSERC of Canada. The research at UWO is supported by the NSERC, CRC, and CFI of Canada and the OIT of Ontario. We thank Mr. Y. Yao and Prof. S. T. Lee at the City University of Hong Kong for HRTEM measurements. We are also grateful to Drs. R. A. Gordon and S. M. Heald at the APS and Drs. Y. F. Hu, A. Jurgensen, and K. H. Tan at the CSRF for technical XAFS assistance. LA0501031