Simplifying the creation of hollow metallic nanostructures: One-pot synthesis of hollow palladium/platinum single-crystalline nanocubes

Full text of DOI:10.1002/anie.200900199

Communications
DOI: 10.1002/anie.200900199
Nanostructures
Simplifying the Creation of Hollow Metallic Nanostructures: One-Pot
Synthesis of Hollow Palladium/Platinum Single-Crystalline
Nanocubes**
Xiaoqing Huang, Huihui Zhang, Changyou Guo, Zhiyou Zhou, and Nanfeng Zheng*
[
1–4]
Hollow nanoparticles
represent an emerging class of
alloyed nanoparticles have been prepared where both metal
precursors are present simultaneously. Together with the use
of iodide ions as the morphology controller, the selection of
acetylacetonates as the metal precursors is critical to the
formation of hollow nanocubes. With respect to the solid
palladium/platinum nanocubes of identical size, the hollow
nanocubes have an increased accessible surface area, leading
to improved electrocatalytic activity.
nanomaterials that exhibit unique catalytic, magnetic, elec-
tronic, and optical properties.
structures are mainly created by using sacrificial tem-
[5–8]
Currently, hollow nano-
[
9–15]
plates.
The templates necessary for the creation of void
space are either preformed or produced in situ. After the
growth of shell materials, the templates are removed by
thermal/chemical treatments to generate the void space.
Nevertheless, without any necessary post-treatment, it is also
possible to fabricate hollow nanostructures by depleting the
core templates during the shell growth. The outer diffusion of
the inner core is normally driven by oxidation, sulfidation,
phosphidation, galvanic replacement, or reduction between
the core materials and reactants in the reaction environ-
In a typical synthesis of palladium nanocubes, palladium
acetylacetonate ([Pd(acac) ]), polyvinylpyrrolidone (PVP),
2
and an aqueous solution of sodium iodide (NaI) were mixed
together with DMF (see the Experimental Section). The
resulting homogeneous solution was transferred to a teflon-
lined stainless-steel autoclave. The sealed vessel was then
heated at 1508C for 8 h. The black nanoparticles were
collected by centrifugation, and washed several times with
ethanol and acetone. The procedure for synthesis of palla-
dium/platinum was similar to that of the palladium nano-
[
3,4,16,17]
ment,
which is analogous to the Kirkendall diffusion
effect.
When hollow metal nanostructures are taken as the
[
18]
[
2,17,19,20]
example,
it should be noted that they are dominantly
prepared by galvanic replacement reactions that strictly
require perfect redox match between the oxidizable metal-
core template and oxidizing metal cations in the solution.
Furthermore, the preparation of the metal core and its redox
reaction with oxidizing metal species have to be carried out
sequentially by supplying the metal reagents at different
stages to prevent their co-reduction. Very recently, hollow
monometal nanocrystals were also prepared in situ by
reducing the corresponding metal oxide nanoparticles,
cubes, except that half of the [Pd(acac) ] was replaced by
2
platinum acetylacetonate ([Pt(acac) ]).
2
As illustrated in the transmission electron microscopy
(TEM) images, the prepared particles are cubic in shape and
substantially uniform in size, averaging (12.5 Æ 0.8) nm in
dimensions (Figure 1a). The single-crystalline nature of the
nanocubes is evident from the appearance of continuous
lattice fringes oriented in the same direction across each cube.
The spacing of the fringes is 0.194 nm (Supporting Informa-
tion, Figure S1), corresponding to the {200} interplanar
distance of face centered cubic (fcc) palladium.
[
17]
which was demonstrated only in the case of cobalt.
Herein we report a synthetic strategy that allows the one-
pot fabrication of uniform single-crystalline palladium nano-
cubes and hollow palladium/platinum alloyed nanocubes. To
the best of our knowledge, it is the first time that hollow
The iodide ions play an important role in the formation of
the monodispersed palladium nanocubes. Unlike many pre-
vious studies in which disodium tetrachloropalladate
[*] X. Q. Huang, H. H. Zhang, Dr. C. Y. Guo, Dr. Z. Y. Zhou,
Prof. N. F. Zheng
State Key Laboratory for Physical Chemistry of Solid Surfaces and
Department of Chemistry, College of Chemistry and Chemical
Engineering, Xiamen University
Xiamen 361005 (China)
Fax: (+86)592-218-3047
E-mail: nfzheng@xmu.edu.cn
Homepage: http://chem.xmu.edu.cn/person/nfzheng/index.html
[
**] Our research was financially supported by NNSFC (20871100,
0721001), the 973 projects (2009CB930703) from MSTC, the Key
Project (108077) from the Chinese Ministry of Education, RFDP
200803841010), and the Key Scientific Project of Fujian Province of
2
(
China (2009HZ10102).
Figure 1. TEM images of Pd nanoparticles: a) nanocubes prepared
with the use of I ion; b) multiply twinned nanoparticles prepared
without the use of any halide.
À
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200900199.
4
808
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 4808 –4812

Angewandte
Chemie
(
Na PdCl ) was applied in the synthesis, the use of [Pd(acac) ]
platinum. The high-quality single-crystalline nature of the as
2
4
2
as the palladium precursor in our synthesis provides the
opportunity to examine the role of halides without the
influence of chloride from the precursor. The reactions at the
same conditions, but without any halide yield semiregular
polyhedral nanoparticles with a size of about 50 nm (Fig-
ure 1b). These particles are characterized by their multiply
twinned structure (Supporting Information, Figure S1). Such
a structural feature is also presented in the particles prepared
from the systems where NaI is replaced by an equal amount of
NaF, NaCl, or NaBr (Supporting Information, Figure S2). The
dimensions of the twinned particles prepared using halides
prepared hollow cubes is also confirmed by the high-
resolution TEM analysis. The analysis presents lattice fringes
with an interplanar distance of 0.195 nm perpendicular to the
edge direction of each cube. The spacing of 0.195 nm is
between 0.194 and 0.196 nm, corresponding to the {200}
interplanar distance of fcc palladium and platinum, respec-
tively. Similar to the palladium nanocubes, the hollow
palladium/platinum alloy nanocubes are enclosed mainly by
{100} facets. While the HAADF-STEM (high-angle annular
dark-field scanning transmission electron microscopy) images
confirm the hollow feature of the palladium/platinum nano-
cubes, the EDX line scanning profile across single palladium/
platinum nanocubes shows a rather homogenous distribution
of both palladium and platinum in each nanocube.
À
À
À
are in the order of F > Cl > Br , however, they are all
smaller than those synthesized without halide.
Different halides result in the formation of different
palladium nanostructures, suggesting the importance of
halide in controlling the shape and size of noble metal
nanoparticles, which is consistent with many previous
Being able to fabricate hollow metal alloy nanocrystals in
one step is the most striking feature of the synthesis reported
herein. Considering that the standard reduction potentials
[
21–23]
0
II
II
reports.
Xia and co-workers have demonstrated that
(E ) for Pd /Pd are typically more negative than Pt /Pt pairs
when they are in the same coordination environment, the
formation of hollow palladium/platinum nanocrystals is not
expected from the simple mixture of palladium(II) and
platinum(II) with the same ligands coordinated. As expected
from the reduction potentials, platinum(II) should be reduced
before palladium(II). Therefore, the galvanic reaction
between platinum(0) and palladium(II) would not occur in
the system to create the hollow structure. To better under-
stand the formation mechanism of hollow nanocubes in our
system, we have monitored their formation process by
running the reactions under the same conditions but for
different lengths of time.
bromide ions but not other halides facilitate the formation of
palladium nanocubes in ethylene glycol. They have proposed
that the chemisorption of chloride or iodide on the surface of
the nanocrystals is either too weak or too strong for particles
[23]
to develop the cubic shape. However, the results from our
system suggest that the halide effect in the formation of metal
nanoparticles should be discussed in more depth by consid-
ering the different reaction media or precursors. For example,
iodide is better than bromide in the preparation of uniform
palladium nanocubes in the DMF system reported herein.
The more interesting result is that if half of the [Pd(acac)2]
is replaced by [Pt(acac) ] in the synthesis of palladium
2
nanocubes the reaction leads to the formation of uniform
hollow nanocubes (Figure 2). As measured by EDX (energy
dispersive X-ray spectroscopy), these hollow nanocubes are
essentially alloy particles containing 52% palladium and 48%
As shown by TEM and EDX analysis (Figure 3; Support-
ing Information Figure S3), only dense palladium nanocubes
are formed at the end of one hour. When the reaction time is
increased to two hours, edges of the cubes become partly
etched. A small amount of platinum is found in these etched
nanocubes. Together with the appearance of the hollow
feature, an increasing platinum content in the particles is
evidenced at the end of three hours of reaction time. By
further increasing the reaction time to four hours, the void
space in each cube becomes more distinct. The palladium/
platinum ratio in the sample after six hours is close to the ratio
supplied in the reaction. The time-domain composition
Figure 2. a,b) TEM images of Pd/Pt hollow nanocubes at low and high
magnification; c) representative HAADF-STEM image of hollow nano-
cubes; d) line-scanning profile across a hollow Pd/Pt nanocube is
indicated (left).
Figure 3. TEM images of nanoparticles collected at different reaction
times: a) 1 h, b) 2 h, c) 3 h, d) 4 h, e) 6 h, f) 8 h, and g) change in
mole % of Pt (DPt) during 8 h of reaction. All scale bars are 10 nm.
Angew. Chem. Int. Ed. 2009, 48, 4808 –4812
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 4809

Communications
change revealed by EDX analysis is confirmed by our
inductively coupled plasma mass spectroscopy measurements
ICP-MS). Beyond six hours, no significant change in either
the structure or composition is observed up to eight hours.
Therefore, it can be concluded that the formation of hollow
palladium/platinum particles relies upon the initial formation
of palladium nanocubes followed by its galvanic replacement
by platinum(II) species in the solution. This mechanism is
further confirmed by the result of our two-step synthesis in
which palladium nanocubes are first produced and then used
hollow nanocubes, the absence of iodide ions led to no
formation of hollow particles (Figure 4). Based on EDX
analysis, only trace amount of platinum was observed on the
obtained solid particles. Similarly, when the iodide ions were
(
as templates to react with [Pt(acac) ] separately in two steps.
2
This procedure also yields monodispersed hollow palladium/
platinum nanocubes (Supporting Information, Figure S4).
As discussed above, the one-pot temporal separation of
palladium nanocubes from their galvanic replacement with
platinum(II) species is the key to the formation of the hollow
palladium/platinum nanocubes. To achieve such a separation,
the selection of metal precursors, reducing agent, and also
reaction temperature is crucial. The simultaneous use of both
iodide ions and acetylacetonate precursors makes it possible
for the palladium(II) to get reduced before the platinum(II)
species. The reactions in the absence of either iodide or
acetylacetonates did not yield hollow cubes (Supporting
Information; Figure S5, S6). Solid palladium/platinum nano-
cubes were obtained when palladium chloride (PdCl ) and
2
platinum(II) chloride instead of acetylacetonates were used
as the metal precursors (Supporting Information, Figure S7).
While critical to the development of the cube morphology, the
Figure 4. TEM images of the Pd/Pt nanoparticles prepared a) with and
b) without the I ions; nanoparticles collected after 8 h of reaction at
c) 1808C and d) 1208C, respectively.
À
presence of iodide ions in the solution of [Pt(acac) ]/[Pd-
2
(
acac) ] also significantly alters the dominating forms of the
2
metal precursors and therefore the reduction kinetics of metal
precursors. When iodide ions were introduced into DMF
solutions of [Pd(acac) ] and [Pt(acac) ] respectively, the
À
À
À
substituted with other halides (i.e., Br , Cl , and F ), the
formation of hollow cubes was not observed either. The
yielded solid particles contained mainly palladium and a small
amount of platinum (Supporting Information, Figure S11).
This result might be ascribed to the strong coordination of
iodide ions to palladium(II) ions, which facilitates the
galvanic replacement between palladium nanocubes and
2
2
[
Pd(acac) ] solution became much darker and the [Pt(acac) ]
2 2
one remained almost unchanged (Supporting Infirmation,
2
À
2À
Figure S8). Since [PtI4] and [PdI4] are both colored much
darker than their acetylacetonate counterparts, this observa-
2
À
tion indicates the stability of [Pt(acac) ] over [PtI ] , and
PdI4] over [Pd(acac) ]. The additon of iodide ions to the
2
Pd(acac) ]/[Pt(acac) ] mixture results in [PdI ] and [Pt-
2
4
2
À
[
[
2
À
[Pt(acac) ] and therefore fast diffusion of palladium atoms
2
2
4
2
(
acac) ] becoming the dominating precursors. In DMF, we
for the creation of the hollow structure. This observation is
somewhat consistent with the findings in dealloying studies.
For example, Corcoran and co-workers reported that the
iodide ions enhance surface diffusion compared to other
2
2
À
found evidence that the reduction of [PdI4]
favorable than [Pt(acac) ] (Supporting Information, Figure S9
and S 10). Therefore, it is not surprising that the dense
palladium nanocubes formed before deposition of the plat-
inum is the prerequisite to create hollow structure.
Following the formation of the dense nanocubes, the
galvanic replacement between palladium cubes and the
platinum(II) is the main driving force for the palladium
atoms to diffuse outwards to create the palladium/platinum
hollow structure. In addition to facilitating the formation of
palladium nanocubes, an important role of the iodide ions in
the galvanic replacement process was also evidenced. To
identify the role of the iodide ion in the process of formation
of the hollow nanocubes, we prepared and purified palladium
nanocubes to remove the iodide ions. The purified palladium
is more
2
[
24]
halides during the dealloying of gold/silver alloys.
In addition to the coordinated use of iodide ions and the
acetylacetonates, the selection of the reducing agent and
temperature is important to distinguish the reduction of
palladium(II) and platinum(II) in two steps, so that their co-
reduction can be prevented. When a stronger reducing agent
(i.e., ascorbic acid) was added to the system, the reduction of
palladium(II) and platinum(II) occurred simultaneously,
leading to a phase separation (Supporting Information,
Figure S12). A similar situation happened when the reaction
temperature was elevated from 1508C to 1808C. The reac-
tions at 1808C led to formation of palladium nanocubes and
large platinum nanoparticles separately (Figure 4c; Support-
ing Information, Figure S13). In comparison, much longer
time, 48 hours, was required to obtain the hollow cubes at
nanocubes were then reacted with [Pt(acac) ] for eight hours
in a DMF solution of PVP containing different halides. While
the presence of iodide ions in this reaction produced uniform
2
4
810 www.angewandte.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 4808 –4812

Angewandte
Chemie
1
208C (Supporting Information, Figure S14). While four
Synthesis of palladium nanocubes: [Pd(acac)
160.0 mg), and an NaI solution (2 mL, 0.15 gmL ) were mixed
] (50.0 mg), PVP
2
À1
(
hours at 1508C was long enough to obtain the hollow cubes,
reactions at 1208C for four hours yielded only solid cubes of
palladium. At 1208C, the etching of the solid palladium cubes
by platinum(II) species was observed only when the reaction
was extended beyond eight hours (Figure 4d). This result
implies that the diffusion rate of palladium atoms caused by
the galvanic reactions is much slower at lower temperatures.
Therefore, there is a particular temperature window in which
the production of the hollow structure can be optimized.
As a galvanic reaction is involved during the formation of
the hollow structures, the void space inside each hollow
palladium/platinum nanocube is interconnected with its
external surface (Figure 2). The inner void space can there-
fore be considered as internal indentation of the nanocubes.
Such a structural feature increases the overall surface area
accessible by small molecules (e.g., carbon monoxide, formic
acid), which is expected to be beneficial to their catalytic
applications. To demonstrate the increased surface area
resulting from the hollow structure, carbon monoxide strip-
ping voltammetry studies were performed on both hollow and
solid palladium/platinum nanocubes. Both types of nanocubes
are of similar sizes (ca. 12 nm). The hollow nanocubes show a
current density that is twice that of the solid cubes (Support-
ing Information, Figure S15). This observation is consistent
with the increased accessible surface area caused by the
hollow structure discussed above. Such a surface-area incre-
ment in the hollow cubes might explain the enhanced
performance of the hollow nanocubes in the electrooxidation
of formic acid (Supporting Information, Figure S15).
In summary, a simplified synthetic strategy has been
developed to prepare single-crystalline hollow palladium/
platinum nanocubes with both palladium and platinum
precursors present at the same time. Together with the use
of iodide ions as the shape regulator, the use of acetylacet-
onate precursors is demonstrated to alter the reduction
kinetics of metal cations and therefore critical to the one-
pot formation of hollow palladium/platinum nanocubes.
Compared to the solid palladium/platinum nanocubes, the
hollow palladium/platinum nanocubes increase accessible
surface area and therefore improve electrocatalytic activity
in formic acid oxidation.
together in DMF (10 mL). The resulting homogeneous brown
solution was then treated under the same conditions as those used
in the preparation of multiply twinned nanoparticles.
Synthesis of hollow palladium/platinum nanocubes: [Pd(acac)2]
(
25.0 mg), [Pt(acac) ] (30.0 mg), PVP (160.0 mg), and an NaI solution
2
À1
(2 mL, 0.15 gmL ) were mixed together in DMF (10 mL). The
resulting solution was transferred to a teflon-lined stainless-steel
autoclave. The sealed vessel was then heated at 1508C for the desired
time before it was cooled to room temperature. The black nano-
particles were precipitated by acetone, separated using a centrifuge,
and further purified by an ethanol/acetone mixture.
Synthesis of solid palladium/platinum nanocubes: PdCl2
(13.3 mg), PtCl2 (20.0 mg), PVP (160.0 mg), and an NaI solution
À1
(2 mL, 0.15 gmL ) were mixed together in DMF (10 mL). The
resulting homogeneous brown solution was then sealed and heated at
1508C for 8 h to yield the solid nanocubes.
Electron microscopy characterization: The high-resolution trans-
mission electron microscopic (HRTEM) observations were per-
formed on a FEI TECNAI F30 microscope operated at 300 kV.
Electrochemical measurements: An ethanol dispersion of puri-
fied nanoparticles was deposited on a glassy carbon electrode to
obtain the working electrodes after evaporation of the solvent under
an IR lamp. A saturated calomel electrode (SCE) and a platinum foil
were used as the reference and counter electrode, respectively. For
the CO stripping voltammetry measurements, CO gas (99.99%) was
bubbled for 20 min through a 0.5m H SO solution in which the
2
4
electrode was immersed. The electrode was quickly moved to a fresh
solution and the CO stripping voltammetry was recorded at a sweep
À1
rate of 50 mVs . For the electrooxidation of formic acid, the cyclic
À1
voltammgrams were recorded at a sweep rate of 50 mVs in a
solution (25 mL) containing H SO (0.5m) and formic acid (0.25m).
2
4
Before the cyclic voltammetry measurements, two cycles of potential
À1
sweeps between À0.2 V and 1.2 V at a sweep rate of 250 mVs were
applied.
Received: January 13, 2009
Published online: May 14, 2009
Keywords: alloys · formic acid · nanoparticles · palladium ·
.
platinum
[
[
1] H. J. Fan, U. Gꢀsele, M. Zacharias, Small 2007, 3, 1660 – 1671.
2] X. W. Lou, L. A. Archer, Z. Yang, Adv. Mater. 2008, 20, 3987 –
4019.
[
[
3] Y. G. Sun, Y. N. Xia, Science 2002, 298, 2176 – 2179.
4] Y. D. Yin, R. M. Rioux, C. K. Erdonmez, S. Hughes, G. A.
Somorjai, A. P. Alivisatos, Science 2004, 304, 711 – 714.
Experimental Section
[5] S. W. Kim, M. Kim, W. Y. Lee, T. Hyeon, J. Am. Chem. Soc. 2002,
124, 7642 – 7643.
Chemicals: [Pd(acac) ] (99%), [Pt(acac) ] (99%), PdCl (99%), and
2
2
2
PtCl (99%) were purchased from Alfa Aesar, PVP (MW= 30000,
[6] H. P. Liang, H. M. Zhang, J. S. Hu, Y. G. Guo, L. J. Wan, C. L.
Bai, Angew. Chem. 2004, 116, 1566 – 1569; Angew. Chem. Int.
Ed. 2004, 43, 1540 – 1543.
[7] X. M. Lu, L. Au, J. McLellan, Z. Y. Li, M. Marquez, Y. N. Xia,
Nano Lett. 2007, 7, 1764 – 1769.
2
AR), NaX (X = F, Cl, Br, I) and DMF were purchased from
Sinopharm Chemical Reagent Co. Ltd. (Shanghai, China). The
water used in all experiments was ultrapure with a resistivity of
1
8.2 MWcm. All reagents were used as received without further
purification.
Synthesis of Pd multiply twinned nanoparticles: Pd(II)acetyl-
[8] Y. Vasquez, A. K. Sra, R. E. Schaak, J. Am. Chem. Soc. 2005,
127, 12504 – 12505.
acetonate ([Pd(acac) ], 50.0 mg), PVP (160.0 mg), and water (2 mL)
[9] P. M. Arnal, C. Weidenthaler, F. Schꢁth, Chem. Mater. 2006, 18,
2733 – 2739.
[10] I. D. Hosein, C. M. Liddell, Langmuir 2007, 23, 2892 – 2897.
[11] Y. X. Hu, J. P. Ge, Y. G. Sun, T. R. Zhang, Y. D. Yin, Nano Lett.
2007, 7, 1832 – 1836.
2
were mixed together in DMF (10 mL). The resulting homogeneous
yellow solution was transferred to a teflon-lined stainless-steel
autoclave. The sealed vessel was then heated at 1508C for 8 h
before it was cooled to room temperature. The black nanoparticles
were precipitated by acetone, separated using a centrifuge, and
further purified by an ethanol/acetone mixture.
[12] S. H. Im, U. Y. Jeong, Y. N. Xia, Nat. Mater. 2005, 4, 671 – 675.
[13] X. W. Lou, L. A. Archer, Adv. Mater. 2008, 20, 1853 – 1858.
Angew. Chem. Int. Ed. 2009, 48, 4808 –4812
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
4811

Communications
[
14] X. W. Lou, C. L. Yuan, L. A. Archer, Adv. Mater. 2007, 19, 3328 –
332.
[19] G. S. Mꢂtraux, Y. C. Cao, R. C. Jin, C. A. Mirkin, Nano Lett.
2003, 3, 519 – 522.
3
[
15] Z. Z. Yang, Z. W. Niu, Y. F. Lu, Z. B. Hu, C. C. Han, Angew.
Chem. 2003, 115, 1987 – 1989; Angew. Chem. Int. Ed. 2003, 42,
[20] Y. D. Yin, C. Erdonmez, S. Aloni, A. P. Alivisatos, J. Am. Chem.
Soc. 2006, 128, 12671 – 12673.
1
943 – 1945.
16] A. E. Henkes, Y. Vasquez, R. E. Schaak, J. Am. Chem. Soc. 2007,
29, 1896 – 1897.
[21] B. Wiley, Y. G. Sun, Y. Xia, Acc. Chem. Res. 2007, 40, 1067 – 1076.
[22] B. J. Wiley, Y. C. Chen, J. M. McLellan, Y. J. Xiong, Z. Y. Li, D.
Ginger, Y. N. Xia, Nano Lett. 2007, 7, 1032 – 1036.
[23] Y. J. Xiong, H. G. Cai, B. J. Wiley, J. G. Wang, M. J. Kim, Y. N.
Xia, J. Am. Chem. Soc. 2007, 129, 3665 – 3675.
[
[
1
17] K. M. Nam, J. H. Shim, H. Ki, S.-I. Choi, G. Lee, J. K. Jang, Y. Jo,
M.-H. Jung, H. Song, J. T. Park, Angew. Chem. 2008, 120, 9646 –
9
650; Angew. Chem. Int. Ed. 2008, 47, 9504 – 9508.
[24] A. Dursun, D. V. Pugh, S. G. Corcoran, J. Electrochem. Soc.
2003, 150, B355 – B360.
[
18] H. Jin Fan, M. Knez, R. Scholz, K. Nielsch, E. Pippel, D. Hesse,
M. Zacharias, U. Gꢀsele, Nat. Mater. 2006, 5, 627 – 631.
4
812
www.angewandte.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 4808 –4812