Porous, catalytically active palladium nanostructures by tuning nanoparticle interactions in an organic medium

Abstract

We present a simple template-free method for the synthesis of interconnected hierarchical porous palladium nanostructures by controlling the aggregation of nanoparticles in organic media. The interaction between the nanoparticles is tuned by varying the dielectric constant of the medium consistent with DLVO calculations. The reaction products range from discrete nanoparticles to compact porous clusters with large specific surface areas. The nanoclusters exhibit hierarchical porosity and are found to exhibit excellent activity towards the reduction of 4-nitrophenol into 4-aminophenol and hydrogen oxidation. The method opens up possibilities for synthesizing porous clusters of other functional inorganics in organic media.

---

Introduction

Controlling the morphology is critical for reaping maximal benefit from nanostructures. Porous structures with accessible internal porosity are critical for many applications including catalysis,^1,2 chemical sensing³ and storage materials.⁴ A variety of template based methods are available for the synthesis of such structures.⁵ For gas storage applications, nanostructures with high active surface areas are desirable.⁶ Three-dimensional porous structures with hierarchical porosity combine efficient transport with high surface area and are ideal for such applications⁵ where the physical dimension and design of the porous path are important for molecular transport. While mesopores provide hindered transport to the molecules, nanostructure with a hierarchical porous structure enables faster diffusion of molecules with the measured diffusion rates almost comparable to the rates in an open medium.⁶

Palladium is an excellent material for hydrogen storage applications. The dissolution and diffusion of hydrogen atoms in the region of the alpha phase of the H–Pd/palladium hydride system makes it a potentially interesting and useful subject of study. The chemical reactivity of palladium towards hydrogen and the selective absorption of hydrogen by Pd have led to the use of Pd as a chemical sensor for the detection of hydrogen and as a hydrogen storage material. Synthesis of different nanostructures of palladium in the form of nanoclusters,⁷ films,^3,8,9 islands,¹⁰ nanoballs,¹¹ mesowires^12–14 have been demonstrated. Here, we demonstrate a new method for the synthesis of a hierarchically porous structure of palladium by controlling the aggregation of ultrafine amine-capped Pd nanoparticles in organic media. Conventionally, pH is used as a controlling factor to induce aggregation of uncapped nanoparticles.¹⁵ However, the stability of the nanoparticles under different pH conditions imposes a limitation on this method. Here, we present a synthesis route in organic media by tuning the dielectric constant of the synthesis medium to control the aggregation of amine-capped Pd nanoparticles and show the formation of a hierarchical porous structure in this case. We show that the morphology of the aggregate can be controllably changed by using varying toluene–methanol mixtures for synthesis which is consistent with the DLVO calculations of interparticle interactions as a function of the dielectric constant of the medium. The hierarchical porous structures exhibit excellent catalytic activity for the reduction of p-nitrophenol and a high electrochemical activity for hydrogen absorption. The method is general and can serve as a convenient means for synthesizing highly active porous clusters of other metals and functional inorganics.

Experimental section

Materials

All the analytical grade chemicals were used as-received without any further purification. Potassium hexachloropalladate (K₂PdCl₆·4H₂O), oleylamine, oleic acid (Sigma-Aldrich) and toluene, methanol and sodium borohydride (S.D. Fine Ltd, India) were used for synthesis.

Synthesis

14 mg of potassium hexachloropalladate in 50 ml of toluene with 400 μl of oleylamine and 200 μl of oleic acid were refluxed at 120 °C for 11 h and allowed to cool at room temperature. The resultant solution is transparent orange yellow in colour and contains a Pd intermediate in a +2 oxidation state. Potassium hexachloropalladate is insoluble in toluene medium; the presence of the amine solubilises the Pd salt and leads to the formation of this intermediate phase. 46 mg of sodium borohydride was solubilized in 50 ml of methanol and slowly added in to 50 ml of the yellowish orange solution under vigorous stirring conditions at room temperature. The solution turns black due to the formation of Pd nanoclusters. The solution mixture was stirred for one hour and at the end of which the black coloured product settles down at the bottom of the reaction vessel. The product was centrifuged at 2000 rpm for 5 min and dried under vacuum and used for all subsequent characterization. The powder was dispersed in hexane and drop-cast on a carbon-coated Cu grid and dried for transmission electron microscopy investigations.

Characterization

The Pd nanoclusters were characterized by powder X-ray diffraction (XRD) and X-ray photoelectron spectroscopy (XPS). XPS data was collected in a ThermoScientific Multilab 2000 instrument and the binding energies are with respect to graphitic C_1s at 284.5 eV. The binding energy is accurate to within ±0.1 eV. Microstructural characterization was carried out using a field-emission scanning electron microscope (Sirion, FEI) operated at 5–20 kV and a field-emission transmission electron microscope (Tecnai F30, FEI) operated at 300 kV. BET surface area measurement was carried out using Quantachrome Autosorb automated gas sorption system. Prior to the measurement, the aggregates were heated at 100 °C for 5 h in vacuum. UV-visible spectra were recorded at room temperature with a Perkin-Elmer Lambda 35B double-beam UV–visible spectrophotometer to study the catalytic activity of the reduction of p-nitrophenol. Cyclic voltammetry and chronoamperometric studies were carried out using a potentiostat/galvanostat EG&G PARC model Versastat II or Solartron model 1286.

Electrode preparation

To study the hydrogen storage property of the nanoporous palladium, electrodes were prepared on amorphous carbon paper (Toray). A mixture of palladium clusters (80 wt %) and acetylene black (20 wt %) was subjected to grinding in a mortar. A few drops of Nafion suspension (Aldrich) was added to form a slurry which was coated on the carbon paper. Coating and drying steps were repeated to get the required loading level (∼2.8 mg cm⁻²) of the active material. Finally the catalyst-loaded carbon paper was dried at 80 °C for 12 h. For electrical contact, a copper wire was attached to the carbon paper with the help of conducting silver paste. A Sartorius balance (CP225D-OCE) with 0.01 mg sensitivity was used for weighing the electrodes.

Electrochemical measurements

A conventional cell with three electrode configuration was used for all experiments. A glass cell of 70 ml capacity with suitable ground–glass joints to introduce a working electrode, Pt foil auxiliary electrodes, and a saturated calomel electrode (SCE) as reference were used for electrochemical studies. All solutions were prepared using doubly distilled water. Potential values are reported against SCE.

Results and discussion

Structural and microstructural studies

Fig. 1a shows the XRD pattern of the black product obtained as a result of reduction by sodium borohydride that clearly confirms the formation of crystalline Pd. From the (111), (200), (220) and (311) peaks observed in the XRD pattern, the lattice constant is found to be 3.91 Å which is close to the value (3.89 Å) observed for bulk Pd with an FCC structure. An estimation of the particle size using the Scherrer formula indicates an average crystallite size of 4 nm. Fig. 1b is a secondary-electron image of the product illustrating the formation of a porous structure with particles of the order of 100 nm and pore sizes of the order of 500 nm. The higher magnification TEM bright field image (Fig. 1c) shows that the 100 nm clusters comprise aggregates of fine particles of the order of 4–5 nm with a much finer scale of porosity within each cluster. The particle size of these subunits is clearly revealed in the dark field TEM micrograph shown in Fig. 1d. From the combination of SEM and TEM images, it is thus confirmed that the clusters of palladium exhibit hierarchical porosity which is ideal for catalyst materials.⁶ The Pd 3d band in the X-ray photoelectron spectrum from the clusters confirms the metallic state of Pd.

Fig. 1 (a) XRD pattern from the nanoporous clusters of palladium. (b) SEM image of Pd clusters showing the interconnected porous structure. (c) Higher magnification shows both the microporous and nanoporous nature of palladium clusters. (d) Dark-field image showing the size of the individual crystallites in the aggregate.

The specific surface area and pore size distribution were measured by multiple-point BET method using the adsorption–desorption isotherms of nitrogen at 77 K. The lower part of the adsorption isotherm is used for measurement of specific surface area, where as desorption branch of the isotherm is used for pore size analysis. The specific surface area of the Pd nanocluster is found to be 24 m² g⁻¹ and the average pore diameter is 4 nm estimated from the Barrett–Joyner–Halenda (BJH) pore size distribution analysis, consistent with the microscopy investigations.

Formation mechanism of Pd nanoclusters

Along with the microstructural and spectroscopic studies, the formation mechanism of these porous nanoclusters in organic media is also investigated in detail. The formation of monodisperse aggregates has been observed in several systems and is well-documented in the literature.^15–25 The combination of short-range attraction and the long-range repulsion has been shown to result in the formation of equilibrium clusters during protein aggregation.²⁴ The formation of platinum nanoclusters in aqueous media has been reported¹⁵ wherein the reaction-limited aggregation of platinum nanoparticles is utilized by exploiting surface charge of the nanoparticles by varying the pH of the reaction medium. In this study, the interaction between Pd nanoparticles in organic media has been investigated by considering the electrostatic repulsive interactions as varied by changing the effective dielectric constant of the interaction medium. As the concentration of surfactant molecules is kept unchanged, the steric repulsive interaction is not considered for the calculation, i.e., the net interaction energy is normalized with respect to steric factor. Mixtures of different proportions of toluene and methanol (dielectric constants of 2.4 and 32.6 respectively) have been used to controllably vary the dielectric constant. This provides an excellent means to control the aggregation of nanoparticles in organic media to achieve porous clusters of Pd with high surface area without forming oxides. It is important to note that achieving porous morphology with a high surface area in the case of metals less noble compared to Au is difficult as they are prone to forming oxides, especially if synthesized in an aqueous medium.

In the modified form of DLVO theory, the attractive potential is represented by the van der Waals force between the particles in the medium. The repulsive force between the particles capped with organics arises due to the double layer overlap of surface charges, polarisability of the medium and steric forces and can be estimated semi-quantitatively by knowing the dielectric constant of the medium, the surface coverage of capping organics and the surface charge. The net interaction energy between two nanoparticles is the summation of the attractive and repulsive components and can be represented as follows.²⁶

V = van der Waals attractive + electrostatic repulsive

V = [−C_vdW/r⁶] + [q₁q₂ e^{−k(r − σ)}/4πεε_or (1 + kσ)]

where C_vdW is a constant that is related to the Hamaker constant and r is the separation distance between two particles, σ is the radius of the particle and the k is inverse double layer thickness. q₁ and q₂ are the charge of the nanoparticles. ε and ε_o are the dielectric constant of the medium and the permittivity of free space.

Using the above expression, the net interaction energy between the two Pd nanoparticles in media with different dielectric constant has been calculated. For simplification, it is assumed that the surface charge of the amine capped Pd particles equals one electron charge and the inverse double layer thickness is taken to be 1 nm which is a typical value at low concentrations. In Fig. 2a, the interaction energy between two Pd nanoparticles of 2 nm radius as a function of separation distance is plotted and it shows the net interaction energies with corresponding barriers at different dielectric constant. At very high dielectric constant, the repulsive component is weaker and hence there is no barrier between two nanoparticles and the net interaction energy is attractive. At low dielectric constant, the repulsive component increases and exhibits a barrier as is evident from the Fig. 2a.

Fig. 2 (a) Interaction energy between 2 nm radius Pd nanoparticles plotted as a function of separation distance for media with different dielectric constant. Panel (b) shows the schematic images of cluster formation mechanism with increase of dielectric constant. Panels (c) and (d) show the bright field images of the palladium nanostructures synthesized at dielectric constants of the binary mixture of the reaction solution of ε_m = 6.9 and ε_m = 24.3 respectively.

The above calculations indicate that it is possible to control the aggregation process by changing the dielectric constant of the medium. Although, there are several assumptions used here due to uncertainties in experimental values of surface charge and double layer thickness, the trends clearly show that the interaction between the particles can be varied over a large range by tuning the dielectric constant of the medium. The experimental realization of this controlled aggregation by using different proportions of methanol and toluene for the synthesis of Pd aggregates and stable Pd nanoparticles at room temperature is given in Fig. 2c and 2d.

All the palladium nanostructures prepared in media with different dielectric constants have been studied using transmission electron microscopy. The nanoparticles prepared at a very low dielectric constant (ε_m = 6.9) show separate nanoparticles of palladium (Fig. 2 c). An increase in the dielectric constant of the medium helps to overcome the repulsive barrier and particles start to aggregate, e.g. at ε_m = 9.9 the nanoparticles show slight aggregation but not completely to overcome the repulsive barrier. At a further increase in dielectric constant (ε_m = 24.3), the nanoporous clusters form where the net interaction energy is optimum for the nanoparticles to undergo multiple collision before aggregation, leading to the formation of reaction-limited aggregates (Fig. 2d). The schematic representation of the mechanism of porous cluster formation is given in Fig. 2b.

Catalytic activity of the porous clusters

The porous palladium nanoclusters are ideally suited for use as catalysts owing to their high surface-to-volume ratio. The reduction of 4-nitrophenol into 4-aminophenol, using excess sodium borohydride has been taken as a model system to explore the catalytic activity of porous palladium. The reaction solution is prepared by the addition of 1.7 ml of water, 0.3 ml of 2 mmol dm⁻³4-nitrophenol, 1 ml of aqueous 0.03 mol dm⁻³sodium borohydride and 0.5 mg of palladium nanoparticles. The addition of sodium borohydride is done at the end, after mixing all the other reagents. Fig. 3a shows the successive UV–visible spectra of the reduction of 4-nitrophenol by NaBH₄ in the presence of porous palladium nanoclusters. In the absence of any catalyst, the peak due to 4-nitrophenol at 400 nm remains unaltered as reported.^15,27–30 Addition and mixing of an aliquot of 400 μl of palladium cluster to the reaction mixture causes the fading, and ultimately bleaching, of the yellow colour of the 4-nitrophenol. The gradual reduction of the peak at 400 nm is observed with the new peak appearing at approximately 300 nm due to the formation of 4-aminophenol.

Fig. 3 (a) Successive UV–visible spectra of reduction of 4-nitrophenol by porous palladium clusters. The changes in absorbance at different wavelengths as measured at different times is indicated using different colors. The reaction conditions are as follows: [p-nitrophenol] = 2 mmol dm⁻³, [NaBH₄] = 0.03 mol dm⁻³ and [Pd nanoclusters] = 1.6 mmol dm⁻³. Panel (b) shows the apparent rate constant of reduction of p-nitrophenol for different concentrations of palladium nanoclusters. The reaction conditions are as follows: [p-nitrophenol] = 2 mmol dm⁻³, [NaBH₄] = 0.03 mol dm⁻³ and (I) [Pd nanoclusters] = 1.6 mmol dm⁻³ and (II) [Pd nanoclusters] = 3 mmol dm⁻³.

The concentration of the sodium borohydride is in a large excess such that its concentration remains constant during the reaction and thus the reaction can be considered to follow pseudo-first-order kinetics with respect to 4-nitrophenol. The reaction rate constant has been evaluated by plotting the log A versus time, where A stands for absorbance at any time t. A good linear correlation with time is obtained and value of the pseudo-first-order rate constant (k) or apparent rate constant (k_app) is calculated from the linear plot.

The apparent rate constant is assumed to be proportional to the surface area S of the metal nanoparticles catalyst present in the system

-dc_t/dt = k_appc_t = k₁Sc_t

where, c_t is the concentration of p-nitrophenol at time t and k₁ is the rate constant, S is the surface area of Pd nanoclusters per unit volume of the system. For the concentration of 1.6 mmol dm⁻³ of Pd nanoclusters, the specific surface area per unit volume has been calculated to be 6 m² l⁻¹ (calculated from BET surface area measurement) and the rate constant is found to be 1.33 × 10⁻⁴ s⁻¹m⁻² l. Similarly, for the concentration of 3.0 mmol dm⁻³ of Pd nanoclusters, the rate constant found to be 2.50 × 10⁻⁴ s⁻¹m⁻² l. The data clearly indicates that with an increase in surface area the catalytic activity increases, as expected for this kind of heterogeneous catalysis.³¹

The enhanced catalytic activity is attributed to the higher surface-to-volume ratio. The hierarchical porous structure influences the catalytic activity by providing the diffusion of 4-nitrophenol to the metal surface.³²Fig. 3b shows the apparent rate constant of the reduction of p-nitrophenol for different concentrations of palladium nanoclusters. The rate constants for the reaction using the nanoclusters synthesized by aggregation are higher compared to PPI dendrimer–Pd catalysts with similar concentrations synthesized by other means.²⁸ We attribute this increased activity to the higher active surface area in the present case due to the nanoporous nature of the catalyst.

Electrocatalytic activity

The electrocatalytic activity of nanoporous Pd clusters was studied using cyclic voltammetry. Cyclic voltammogram of Pd/C in 1 M H₂SO₄ at a sweep rate of 50 mV s⁻¹ is shown in Fig. 4a. The electrochemical behaviour of the nanoporous palladium is quite similar to the reported data in the literature.^33–35 Current peaks appearing between −0.2 and 0.0 V are due to adsorption and desorption of hydrogen atoms on Pd surface. The peak (peak I in Fig. 4a) appearing at −0.167 V can be attributed to hydrogen desorption from the bulk of the metal or from the bulk along with desorption from the surface.^35,36 The peak at more positive potential (peak II in Fig. 4a) is due to the oxidation of hydrogen adsorbed on the surface and that absorbed in the α-phase in the bulk of the metal. The peak marked III (Fig. 4a) at 0.43 V is considered due to the surface oxide formation and dissolution.³³ Adsorbed oxygen is reflected in the increase in oxidation current at about 0.40 V and its reduction by a cathodic current peak at about 0.45 V in the reverse sweep. At 0.37 V (peak IV), reduction of the surface oxides takes place.³³ The cathodic peaks V and VI are due to the hydrogen adsorption and absorption. These features of the voltammograms suggest that Pd nanoparticles are electrochemically active.

Fig. 4 Panel (a) shows the characteristic CV curve of porous palladium nanoclusters in 1.0 M H₂SO₄ at a sweep rate of 50 mV s⁻¹ and (b) shows the cyclic voltammogram of Pd/C in 0.1 M KNO₃ at a sweep rate of 20 mVs⁻¹ after keeping the electrode at −1.0 V for (1) 0 s, (2) 1800 s, (3) 3600 s and (4) 5400 s.

The Pd/C electrode is further considered for hydrogen oxidation studies. Pd metal has been receiving considerable attention in fuel cell technology due to low cost, greater availability compared to platinum and its ability to absorb large quantities of hydrogen.³⁷ The cyclic voltammogram of Pd/C before purging H₂ shows a typical capacitive behaviour with very low current densities. A dramatic change in the response was observed when aqueous KNO₃ solution was purged with H₂ where a large increase in oxidation current was observed. Fig. 4b shows the cyclic voltammogram of Pd/C electrode in aqueous 0.1 M KNO₃ solution after keeping the electrode at −1.0 V for a definite time. At −1.0 V, there is visible hydrogen evolution from the working electrode surface (Pd/C electrode). Hydrogen gets absorbed spontaneously on the surface of Pd nanoparticles which is reflected in an increase in the oxidation as well as reduction current during subsequent cycle. Post electrochemical investigation of the microstructure indicates some morphological transformation while the porous nature is still maintained showing that the porous catalyst is stable under electrochemical conditions.

Conclusions

A novel method of synthesis of hierarchical porous clusters of palladium of very high surface area has been described. The mechanism of the formation of the cluster aggregation has been explored and it has been found that by changing the dielectric constant of the medium it is possible to control the aggregation of the particles in organic media. We have demonstrated that by controlling the dielectric constant, a wide range of structures with different degrees of aggregation can be synthesized in a deterministic manner. The nanoporous clusters thus prepared exhibit high catalytic activity for the reduction of p-nitrophenol and also exhibit good hydrogen storage capacity as investigated by electrochemical methods. The method is general and can be adopted for synthesis of high surface area structures for other functional inorganics as well.

Acknowledgements

The authors would like to acknowledge financial support from the NSTI, Department of Science and Technology, Govt. of India. A. H. acknowledges CSIR for student fellowship. The Tecnai F30 TEM is a part of the National Microscopy Facility at the Indian Institute of Science. The XPS is a part of the Surface Science Facility at the Indian Institute of Science.

Notes and references

1. X. W. Teng, X. Y. Liang, S. Rahman and H. Yang, Adv. Mater., 2005, 17, 2237–2241.
2. A. Corma, Chem. Rev., 1997, 97, 2373–2420.
3. H. Lin, T. Gao, J. Fantini and M. J. Sailor, Langmuir, 2004, 20, 5104–5108.
4. H. Takagi, H. Hatori, Y. Soneda, N. Yoshizawa and Y. Yamada, Mater. Sci. Eng., B, 2004, 108, 143–147.
5. O. D. Velev and E. W. Kaler, Adv. Mater., 2000, 12, 531–534.
6. D. R. Rolison, Science, 2003, 299, 1698–1701.
7. S. Y. Huang, C. D. Huang, B. T. Chang and C. T. Yeh, J. Phys. Chem. B, 2006, 110, 21783–21787.
8. J. C. Hierso, R. Feurer and P. Kalck, Chem. Mater., 2000, 12, 390–399.
9. S. E. Nam and K. H. Lee, Ind. Eng. Chem. Res., 2005, 44, 100–105.
10. Y. Gimeno, A. HernandezCreus, S. Gonzalez, R. C. Salvarezza and A. J. Arvia, Chem. Mater., 2001, 13, 1857–1864.
11. G. Surendran, F.a. Ksar, L. Ramos, B. Keita, L. Nadjo, E. Prouzet, P. Beaunier, P. DieudonneÌ, F. Audonnet and H. Remita, J. Phys. Chem. C, 2008, 112, 10740–10744.
12. E. C. Walter, F. Favier and R. M. Penner, Anal. Chem., 2002, 74, 1546–1553.
13. A. Fukuoka, H. Arakia, Y. Sakamotoa, S. Inagakib, Y. Fukushimab and M. Ichikawa, Inorg. Chim. Acta, 2003, 350, 371–378.
14. F. Favier, E. C. Walter, M. P. Zach, T. Benter and R. M. Penner, Science, 2001, 293, 2227–2231.
15. B. Viswanath, S. Patra, N. Munichandraiah and N. Ravishankar, Langmuir, 2009, 25, 3115–3121.
16. A. K. Boal, F. Ilhan, J. E. DeRouchey, T. Thurn-Albrecht, T. P. Russell and V. M. Rotello, Nature, 2000, 404, 746–748.
17. G. H. Bogush and C. F. Zukoski, J. Colloid Interface Sci., 1991, 142, 19–34.
18. J. Groenwold and W. K. Kegel, J. Phys. Chem. B, 2001, 105, 11702–11709.
19. E. Matijevic, Chem. Mater., 1993, 5, 412–426.
20. E. Matijevic, Colloid J., 2007, 69, 29–38.
21. J. Park, V. Privman and E. Matijevic, J. Phys. Chem. B, 2001, 105, 11630–11635.
22. V. Privman, D. V. Goia, J. Park and E. Matijevic, J. Colloid Interface Sci., 1999, 213, 36–45.
23. W. Stober, A. Fink and E. Bohn, J. Colloid Interface Sci., 1968, 26, 62–69.
24. A. Stradner, H. Sedgwick, F. Cardinaux, W. C. K. Poon, S. U. Egelhaaf and P. Schurtenberger, Nature, 2004, 432, 492–495.
25. I. M. Yakutik, G. P. Shevchenko and S. K. Rakhmanov, Colloids Surf., A, 2004, 242, 175–179.
26. D. Leckband and J. Israelachvili, Q. Rev. Biophys., 2001, 34, 105–267.
27. K. Hayakawa, T. Yoshimura and K. Esumi, Langmuir, 2003, 19, 5517–5521.
28. K. Esumi, R. Isono and T. Yoshimura, Langmuir, 2004, 20, 237–243.
29. M. H. Rashid, R. R. Bhattacharjee, A. Kotal and T. K. Mandal, Langmuir, 2006, 22, 7141–7143.
30. A. Murugadoss and A. Chattopadhyay, J. Phys. Chem. C, 2008, 112, 11265–11271.
31. S. Panigrahi, S. Basu, S. Praharaj, S. Pande, S. Jana, A. Pal, S. K. Ghosh and T. Pal, J. Phys. Chem. C, 2007, 111, 4596–4605.
32. F. Balfanz and D. Gelbin, J. Eng. Phys. Thermophys., 1973, 25, 1298–1306.
33. M. Grden, M. Lukaszewski, G. Jerkiewicz and A. Czerwinski, Electrochim. Acta, 2008, 53, 7583–7598.
34. A. Sarkar, A. V. Murugan and A. Manthiram, J. Phys. Chem. C, 2008, 112, 12037–12043.
35. A. Czerwinski, R. Marassi and S. Zamponi, J. Electroanal. Chem., 1991, 316, 211–221.
36. A. Czerwinski, J. Electroanal. Chem., 1994, 379, 487–493.
37. C. R. K. Rao and D. C. Trivedi, Coord. Chem. Rev., 2005, 249, 613–631.

---

Footnote

† Electronic supplementary information (ESI) available: Additional figures of the Pd nanostructures. See DOI: 10.1039/c0nr00640h

---