Synthesis and mechanistic study of palladium nanobars and nanorods

Article abstract of DOI:10.1021/ja0688023

This paper describes a simple and versatile method for growing highly anisotropic nanostructures of Pd, single-crystal nanobars bounded by {100} facets and single-crystal nanorods with their side surfaces enclosed by {100} and {110} facets. According to thermodynamic arguments, Pd atoms should nucleate and grow in a solution phase to form cuboctahedrons of spherical shape with their surfaces bounded by a mix of {111} and {100} facets. Anisotropic nanostructures can only form under kinetically controlled conditions, while the cubic symmetry is broken. In the present system, we found that one-dimensional growth could be induced and maintained through an interplay of the following processes: (i) speedy reduction of the precursor to ensure prompt addition of atoms to the seed; (ii) chemisorption of bromide on the seed to promote the formation of {100} and {110} facets; and (iii) localized oxidative etching on one specific face of the seed to initiate preferential growth on this face. Experimentally, the anisotropic growth can be achieved by varying the type and concentration of reducing agent, as well as by adjusting the reaction temperature. This methodology developed for Pd has also been extended to both Au and Pt. As expected for a kinetically controlled product, the anisotropic nanostructure evolved into the thermodynamically favored shape during an aging process.

Full text of DOI:10.1021/ja0688023

Published on Web 03/03/2007
Synthesis and Mechanistic Study of Palladium Nanobars and
Nanorods
Yujie Xiong,† Honggang Cai,† Benjamin J. Wiley,‡ Jinguo Wang,§ Moon J. Kim,§ and
Younan Xia*,†
Contribution from the Department of Chemistry, and Department of Chemical Engineering,
UniVersity of Washington, Seattle, Washington 98195, and Department of Electrical
Engineering, UniVersity of Texas at Dallas, Richardson, Texas 75083
Received December 8, 2006; E-mail: xia@chem.washington.edu
Abstract: This paper describes a simple and versatile method for growing highly anisotropic nanostructures
of Pd, single-crystal nanobars bounded by {100} facets and single-crystal nanorods with their side surfaces
enclosed by {100} and {110} facets. According to thermodynamic arguments, Pd atoms should nucleate
and grow in a solution phase to form cuboctahedrons of spherical shape with their surfaces bounded by
a mix of {111} and {100} facets. Anisotropic nanostructures can only form under kinetically controlled
conditions, while the cubic symmetry is broken. In the present system, we found that one-dimensional
growth could be induced and maintained through an interplay of the following processes: (i) speedy reduction
of the precursor to ensure prompt addition of atoms to the seed; (ii) chemisorption of bromide on the seed
to promote the formation of {100} and {110} facets; and (iii) localized oxidative etching on one specific
face of the seed to initiate preferential growth on this face. Experimentally, the anisotropic growth can be
achieved by varying the type and concentration of reducing agent, as well as by adjusting the reaction
temperature. This methodology developed for Pd has also been extended to both Au and Pt. As expected
for a kinetically controlled product, the anisotropic nanostructure evolved into the thermodynamically favored
shape during an aging process.
1. Introduction
nanoscale channels in alumina or polycarbonate membranes
have been most commonly used for this purpose.3 Other types
of templates include mesoporous silica,4 cylindrical micelles,5
and organic block copolymers,6 as well as the edges or grooves
on solid substrates.7 Although a template-directed synthesis can
be very simple and straightforward, it is limited in terms of the
quantity of nanostructures that can be produced in each run of
synthesis. It often yields polycrystalline nanostructures, which
are less valuable for both fundamental study and device
fabrication. In addition, the template needs to be removed in a
postsynthesis step so the metal nanostructures can be harvested
and put to use. As a result, it seems to be impractical to rely on
template-directed synthesis if one needs single-crystal, 1-D
nanostructures of Pd.
Single-crystal, one-dimensional (1-D) nanostructures of Pd
are attractive as interconnects for fabricating nanoscale electronic
devices. For example, Pd can form reliable and reproducible
ohmic contacts with carbon nanotubes (CNTs) because Pd has
a relatively high work function and can easily wet the carbon
surface. This capability allows one to elucidate the intrinsic
properties of CNTs and to maximize the performance of CNT-
based devices such as field-effect transistors (FETs).1 Another
important property of Pd is its exceptional sensitivity toward
hydrogen. To this end, polycrystalline, mesoscopic wires made
of Pd have been utilized for resistance-based detection of
hydrogen gas.2 However, polycrystalline wires containing gaps
between adjacent grains often shrink after initial exposure to
hydrogen and may cause random, irreversible changes to the
resistance of a sensing device. It should be possible to overcome
this problem by switching to single-crystalline Pd nanowires
with better controlled characteristics.
Solution-phase growth has received considerable interest for
its capability to produce single-crystal nanostructures with high
quality. However, it is not easy to grow 1-D nanostructures of
(3) (a) Martin, C. R. Science 1994, 266, 1961. (b) Martin, B. R.; Dermody, C.
J.; Reiss, B. D.; Fang, M.; Lyon, L. A.; Natan, M. J.; Mallouk, T. E. AdV.
Mater. 1999, 11, 1021.
(4) (a) Huang, M. H.; Choudrey, A.; Yang, P. Chem. Commun. 2000, 1063.
(b) Han, Y.; Kim, J.; Stucky, G. D. Chem. Mater. 2000, 12, 2068. (c) Lee,
K.-B.; Lee, S.-M.; Cheon, J. AdV. Mater. 2001, 294, 348.
(5) Kijima, T.; Yoshimura, T.; Uota, M.; Ikeda, T.; Fujikawa, D.; Mouri, S.;
Uoyama, S. Angew. Chem., Int. Ed. 2004, 43, 228.
(6) Cornelissen, J. J. L. M.; Van Heerbeek, R.; Kamer, P. C. J.; Reek, J. N.
H.; Sommerdijk, N. A. M.; Nolte, R. J. M. AdV. Mater. 2002, 14, 489.
(7) (a) Bishop, D. J.; Licini, J. C.; Dolan, G. J. Appl. Phys. Lett. 1985, 46,
1000. (b) Jones, E. T. T.; Chyan, O. M.; Wrighton, M. S. J. Am. Chem.
Soc. 1987, 109, 5526. (c) Penner, R. M. J. Phys. Chem. B. 2002, 106,
3339.
One of the simplest ways to generate 1-D nanostructures of
metals is to confine their growth within a template. The
† Department of Chemistry, University of Washington.
‡ Department of Chemical Engineering, University of Washington.
§ University of Texas at Dallas.
(1) (a) Javey, A.; Guo, J.; Wang, Q.; Lundstrom, M.; Dai, H. Nature 2003,
424, 654. (b) Mann, D.; Javey, A.; Kong, J.; Wang, Q.; Dai, H. Nano Lett.
2003, 3, 1541. (c) Zhu, W.; Kaxiras, E. Nano Lett. 2006, 6, 1415.
(2) Favier, F.; Walter, E. C.; Zach, M. P.; Benter, T.; Penner, R. M. Science
2001, 293, 2227.
9
10.1021/ja0688023 CCC: $37.00 © 2007 American Chemical Society
J. AM. CHEM. SOC. 2007, 129, 3665-3675
3665

A R T I C L E S
Xiong et al.
2. Experimental Section
Pd in a solution phase. As a face-centered cubic (fcc) metal, Pd
has no intrinsic driving force for the growth of anisotropic
structures when the seeds are surrounded by an isotropic
medium. As dictated by thermodynamics, Pd atoms are expected
to nucleate and grow into cuboctahedrons (with a nearly
spherical shape) enclosed by a mix of {111} and {100} facets
to minimize the total surface energy.8 This prediction has
recently been verified by experimental studies where 8-nm
cuboctahedrons were obtained as the major product when a Pd
precursor was reduced at a sufficiently fast rate to exclude any
kinetic effect.9 In general, an fcc metal can only be forced to
grow into anisotropic nanostructures through the kinetic control.
As demonstrated in our previous work, triangular and hexagonal
thin plates of Pd could be prepared by operating at an extremely
slow reduction rate to induce the formation of stacking faults
and thus break the cubic symmetry. For such thin plates, the
top and bottom faces account for >70% of the surface and are
terminated in {111} facets, while the side faces (<30% of the
surface) are enclosed by {100} and {111} facets.10 It is evident
that one needs to not only break the cubic symmetry but also
substantially increase the coverage of {100} and/or {110} facets
on the surface to generate nanostructures with 1-D morphol-
ogy.11
Chemicals and Materials. Ethylene glycol (EG, J. T. Baker, 9300-
01), sodium palladium(II) chloride (Na2PdCl4, Aldrich, 379808-1g),
potassium bromide (KBr, Fisher, P205-100g), and poly(vinyl pyrroli-
done) (PVP, MW ) 55 000, Aldrich, 856568-100g) were all used as
received without further purification.
Synthesis of Pd Nanorods with an Aspect Ratio of 8. In a typical
synthesis, 5 mL of EG was hosted in a 25-mL, three-neck flask
(equipped with a reflux condenser and a Teflon-coated magnetic stirring
bar) and heated in air under magnetic stirring at 100 °C. Meanwhile,
0.0486 g of Na2PdCl4 and 0.600 g of KBr were dissolved in 3 mL of
water, and 0.0916 g of PVP was dissolved in 3 mL of EG at room
2-
temperature. These two solutions (with the molar ratio of PdCl4 to
Br- and the repeating unit of PVP being 1:30:15) were then injected
simultaneously into the flask using a two-channel syringe pump (KDS-
200, Stoelting, Wood Dale, IL) at a rate of 45 mL per hour. The reaction
mixture was heated at 100 °C in air for 1 h before the product was
collected by centrifugation and washed with acetone once and with
ethanol three times to remove most of the EG and excess PVP. The
as-obtained samples were characterized by transmission electron
microscopy (TEM), high-resolution TEM, electron diffraction (ED),
energy-dispersive X-ray (EDX) analysis, powder X-ray diffraction
(PXRD), and X-ray photoelectron spectroscopy (XPS).
Synthesis of Pd Nanobars with Aspect Ratios of 2-4. The
procedure was similar to what was used for the Pd nanorods, except
that the PVP was dissolved in 3 mL of water instead of EG.
Synthesis of Pd Nanobars with Aspect Ratios of 1-1.2. The
procedure was similar to the one used for the Pd nanorods, except that
the solvent for all solutions was water instead of EG. In this case, the
hydroxyl end group of PVP served as the reducing agent.
Instrumentation. TEM images were captured using a Phillips 420
transmission electron microscope operated at 120 kV. High-resolution
TEM images and ED patterns were taken on a JEOL field-emission
transmission electron microscope (2100F) operated at 200 kV. Filtered
images were generated by inverse FFT with a Gatan Digital Micrograph
program. EDX analysis was performed on a FEI field-emission scanning
electron microscope (Sirion XL) operated at an accelerating voltage of
10 kV. Samples for TEM and EDX studies were prepared by drying a
drop of the aqueous suspension of particles on carbon-coated copper
grids (Ted Pella, Redding, CA) or silicon wafers under ambient
conditions. Before imaging, the TEM and SEM samples were placed
in a homemade, gravity-fed flow cell and washed for 1 h with deionized
water to remove excess PVP. Finally, the samples were dried and stored
in vacuum for TEM and EDX characterization. All powder XRD
patterns were recorded using a Philips 1820 diffractometer equipped
with a Cu KR radiation source (λ ) 1.54180 Å), and the samples were
deposited onto glass slides.
XPS measurements were performed using a Surface Science X-Probe
spectrometer equipped with an Al KR monochromatized X-ray source
(E ) 1486.6 eV) with a penetration depth of 6-10 nm. The instrument
was operated at a pressure of 5 × 10-9 Torr in the analysis chamber.
The X-ray spot had an elliptical shape with a short axis of 800 µm
when focused on the surface, and the analyzer had an angle of 55°
with respect to the surface normal of the sample. Photoelectrons were
collected with a pass energy of 150 eV for surveys and 50 eV for high-
resolution spectra. The binding energies were calculated relative to the
maximum intensity of the C 1s signal at 285.0 eV. Analysis of the
spectra was executed using the ESCA VB data reduction software
provided by Service Physics. A submonolayer of Pd nanobars was
deposited on small pieces of silicon (0.5 × 0.5 cm2 in size) for all
XPS samples.
Here, we demonstrate that the formation of {100} and {110}
facets can be greatly promoted by introducing bromide into the
reaction solution. The bromide is able to chemisorb onto the
surface of Pd seeds and alter the order of surface free energies
for different facets. By fine-tuning the experiment conditions,
we can selectively generate nanobars, structures with a square
cross-section and enclosed by {100} facets, and nanorods,
structures with an octagonal cross-section whose side surface
is bounded by a mix of {100} and {110} facets. Systematic
studies suggest that the anisotropic growth is most likely induced
by localized oxidative etching on one specific face of a seed,
which creates an active site for atomic addition (i.e., growth)
and thus facilitates the preferential growth on this face. When
the rate of atomic addition is sufficiently fast, the preferential
growth on this particular face leads to the formation of an
elongated nanostructure with a square cross-section. More
specifically, when the reduction rate is in the medium region,
the seeds take a cubic shape with slight truncation at the corners,
and the product mainly contains nanobars. As the reduction rate
becomes much faster, more seeds are formed in the nucleation
step. These cubic seeds take a smaller size but with more
significant truncation at the corners, and the final product is
dominated by nanorods thinner than the nanobars. Because the
nanobars and nanorods are kinetically controlled products, they
can be aged and transformed into cuboctahedrons, a shape more
favorable from the viewpoint of thermodynamics. This approach
developed for Pd can also be applied to both Au and Pt.
(8) (a) Wulff, G. Z. Kristallogr. 1901, 34, 449. (b) Marks, L. D. Rep. Prog.
Phys. 1994, 57, 603. (c) Pimpinelli, A.; Villain J. Physics of Crystal Growth;
Cambridge University Press: Cambridge, UK, 1998.
(9) Xiong, Y.; Chen, J.; Wiley, B.; Xia, Y.; Aloni, S.; Yin, Y. J. Am. Chem.
Soc. 2005, 127, 7332.
(10) Xiong, Y.; McLellan, J. M.; Chen, J.; Yin, Y.; Li, Z.-Y.; Xia, Y. J. Am.
Chem. Soc. 2005, 127, 17118.
(11) (a) Wang, Z. L.; Mohamed, M. B.; Link, S.; El-Sayed, M. A. Surf. Sci.
1999, 440, L809. (b) Wang, Z. L. J. Phys. Chem. B 2000, 104, 1153. (c)
Wang, Z. L.; Gao, R. P.; Nikoobakht, B.; El-Sayed, M. A. J. Phys. Chem.
B 2000, 104, 5417.
3. Results and Discussion
Correlation between Shape and Reduction Kinetics. We
first conducted the synthesis in solutions that contained KBr,
9
3666 J. AM. CHEM. SOC. VOL. 129, NO. 12, 2007

Synthesis of Palladium Nanobars and Nanorods
A R T I C L E S
Figure 1. TEM images of Pd nanostructures that were obtained when the reduction kinetics was controlled by adjusting the volume percent of ethylene
glycol (EG) in the solvent mixture: (A) 0%; (B) 9.1%; (C) 45.5%; and (D) 72.7%. The reaction temperature was 100 °C. The product shown in (E) was
obtained under the same condition as in (D), except for the use of di(ethylene glycol) (DEG) instead of EG. The product shown in (F) was obtained under
the same condition as in (D), except that the temperature was increased to 120 °C. All of these syntheses were carried out with a fixed molar ratio of KBr
to Na2PdCl4 at 30, in the presence of 75 mM PVP, and in an 11-mL mixture of EG (or DEG) and water.
PVP, and different amounts of EG in water, with the addition
of Na2PdCl4 as a precursor to Pd atoms. The key reaction
involved in such a synthesis is the co-reduction of Na2PdCl4
by both EG and PVP. We have recently demonstrated that the
ends of commercially available PVP are terminated in the
hydroxyl group, which can serve as a mild reducing agent for
the generation of metal nanoparticles.12 In contrast to the weak
reducing strength of PVP, EG has proven to be a more powerful
reducing agent for the reduction of a metal salt.9,10 Therefore,
in the current work, the reduction rate should increase as the
percent of EG in the solvent mixture is increased. Through this
kinetic control, the reaction can generate a number of well-
defined nanostructures, nanobars and nanorods with different
aspect ratios (RAs). Figure 1A-D shows TEM images of four
representative samples, illustrating the dependence of the shape
of Pd nanostructures on the concentration of EG. It is clear that
a faster reduction rate at a higher EG concentration induced
anisotropic growth for the nanocrystals. More specifically, slow,
medium, and fast reduction rates favored the formation of Pd
nanobars (width, ∼8 nm; RA, 1-1.2), nanobars (width, ∼6 nm;
RA, 2-4), and nanorods (diameter, ∼2 nm; RA, ∼8), respec-
tively.13 For nanobars and nanorods, the yield was typically
>95% in each run of synthesis (see Figure 1A and D).13 This
strong dependence implies that the shape of product can be
effectively altered by manipulating the reduction rate. This point
is further supported by the result of a synthesis where di(ethylene
glycol) (DEG) instead of EG was used as the reductant. As
compared to EG, DEG has a weaker reducing power14 and
therefore shows a slower reduction rate for the Pd precursor.
As a result, the reaction yielded short nanobars instead of
nanorods (see Figure 1E). In addition to the change of reducing
(13) A minimum of 100 particles were used to measure lengths and widths.
The values listed are averages; standard deviations in lengths are at most
20%, and standard deviations in widths are at most 10%. The yield in this
paper means the fraction of all observed particles that are of a certain shape.
(12) Xiong, Y.; Washio, I.; Chen, J.; Cai, H.; Li, Z.-Y.; Xia, Y. Langmuir 2006,
22, 8563.
9
J. AM. CHEM. SOC. VOL. 129, NO. 12, 2007 3667

A R T I C L E S
Xiong et al.
Figure 2. High-resolution TEM images of the Pd nanobars and nanorods shown in Figure 1A, B, and F: (A) a nanobar with an aspect ratio of ∼1 recorded
along [010]; (B) a nanobar with an aspect ratio of ∼2 recorded along [010]; (C) a nanorod recorded along [11h0]; and (D) a nanorod recorded along [010].
agent, the reduction kinetics can also be manipulated by
adjusting reaction temperature. To this end, we found that the
reduction could be accelerated to induce anisotropic growth by
increasing the temperature from 100 to 120 °C while other
parameters were kept the same as in Figure 1D. This increase
in temperature doubled the aspect ratio of Pd nanorods from 8
to 16 while the diameter remained at 2 nm (see Figure 1F).
The structures of nanobars and nanorods were characterized
by high-resolution TEM studies, as detailed in Figure 2 and
Figure S1. Figure 2A shows a high-resolution TEM image of a
nanobar with an aspect ratio of about 1. The fringes show a
period of 2.0 Å, which was consistent with the {200} lattice
spacing of fcc Pd. This image also displayed well-resolved,
continuous fringes with the same orientation, implying that the
Pd nanocube is a piece of single crystal bounded by {100}
facets. Similarly, the Pd nanobar with a higher aspect ratio is
also a single crystal enclosed by {100} facets, except that its
dimension along one direction is much larger than those along
the other two directions (see Figure 2B). All of them were
slightly truncated at the corners. Different from the nanobar,
the nanorod has an octagonal cross-section and its side surface
is bounded by a mix of both {100} and {110} facets. It can lie
on a flat surface against any one of these facets. Figure 2C shows
a high-resolution TEM image of a Pd nanorod recorded along
[11h0], suggesting that the nanorod is a single crystal without
any dislocation, stacking fault, or twin defects. This nanorod
grew along the [001] direction and was supported on the TEM
grid against one of its {110} facets. The ends of this nanorod
were enclosed by the (001) face and {111} facets. Figure 2D
(14) (a) Glycols; Curme, G. O., Jr., Johnston, F., Eds.; Reinhold Publishing
Co.: New York, 1952. (b) Fievet, F.; Lagier, J. P.; Figlarz, M. MRS Bull.
1989, 14, 29. (c) Bandwar, R. P.; Rao, C. P. Carbohydr. Res. 1995, 277,
197. (d) Yamaguchit, T.; Kitajima, K. J. Mater. Sci. 1998, 33, 653. (e)
Teranishi, T.; Hosoe, M.; Tanaka, T.; Miyake, M. J. Phys. Chem. B 1999,
103, 3818.
9
3668 J. AM. CHEM. SOC. VOL. 129, NO. 12, 2007

Synthesis of Palladium Nanobars and Nanorods
A R T I C L E S
shows a high-resolution TEM image of a nanorod that lies on
the TEM grid against the (010) face. The ends of this nanorod
were enclosed by the (001) face and {110} facets of relatively
small in area. These results are similar to the experimental
observations on single-crystal Au nanorods.11 We also observed
surface reconstruction for the less stable {110} facets, which
has a higher surface energy relative to the {100} and {111}
facets. In some regions, rows of atoms are missing along the
[11h0] axis, and the {110} surface was transformed into strips
of {111} facets. It is known that the {111} surface has a most
closely packed structure, and thus the lowest surface energy.
Through this surface reconstruction, the {110} facets were
stabilized on the Pd nanorods. A similar surface reconstruction
has also been observed for single-crystal Au nanorods.11c In
addition to the high-resolution TEM, the nanorods and nanobars
can also be easily distinguished by the sharpness of their ends,
with bars, in general, sharper than rods.
Oxidative etching has been extensively explored by our group
and others for the shape-controlled synthesis of metal nano-
structures. It can be used to slow a reduction process, selectively
remove multiply twinned seeds, generate hollow Pd nanostruc-
tures, or shorten single-crystal Au nanorods.9,10,17 Yet it can play
another role in the formation of highly anisotropic nanostruc-
tures. In the present synthesis, addition of bromide at a
sufficiently high concentration could cover the surface of a Pd
nanocrystal with bromide due to its strong binding to the Pd
surface.18 The chemisorbed bromide layer prevents further
addition of Pd atoms from solution to the nanocrystal surface,
so one has to activate the surface of this nanocrystal to continue
the growth. Oxidative etching, which is caused by oxygen (from
air) and chloride (from PdCl42-), could remove some of the
bromide from the surface and thus expose sites for addition of
Pd atoms. In the current case, the bromide layer would make it
difficult to perform oxidative etching over the entire surface.
However, with the assistance of water, oxidative etching was
able to occur locally on a specific face of a cubic nanocrystal.19
We have previously observed that corrosion of Pd nanocubes,
whose surface was protected by a high concentration of PVP,
occurs from only one side by a pitting process, even though all
sides are equivalent {100} facets.17b In the galvanic replacement
between Ag nanocubes and HAuCl4 in water, it was also
observed that etched pits were only formed at one of the six
{100} faces.20 Other groups have also reported that oxidative
etching can selectively take place on {100} tips of single-crystal
Au nanorods, although there are other {100} faces on the side
surface.17c This localized oxidative etching makes one face
become more active than others and thus provides favorable
sites for the addition of Pd atoms. When sufficient Pd atoms
are added to the etched sites, atomic addition will be faster than
the dissolution of atoms caused by etching. We believe that it
is the preferential growth at these active sites that breaks the
cubic symmetry of a Pd nanocrystal and leads to the formation
of Pd nanorods and nanobars.
The phase purity and high crystallinity of the Pd nanostruc-
tures are also supported by powder X-ray diffraction (XRD).
Figure S2 shows the typical PXRD pattern of an as-prepared
sample of Pd nanobars. In the XRD pattern, all of the peaks
can be indexed to fcc palladium (JCPDS card, 05-0681). No
characteristic peaks were observed for impurities such as PdBr2
and PdO. The ratio between the intensities of (111) and (200)
peaks is much lower than the value reported for the conventional
powder sample (1.33 versus 2.38), indicating that the diffraction
from {100} planes is enhanced for the sample of Pd nanobars.
It is likely that the nanobars preferentially lay on the substrate
against their flat {100} facets. As a result, there was a texturing
effect. The ED pattern (inset of Figure S2) taken from an
assembly of Pd nanobars also indicates that they were highly
crystalline. The concentric rings can be indexed to the diffraction
from {111}, {200}, {220}, and {311} planes of fcc palladium,
respectively.
Oxidative Etching and Its Role in the Anisotropic Growth
of Pd Nanostructures. What is the driving force for the
anisotropic growth of Pd nanobars and nanorods in the current
work? Two mechanisms have been proposed to account for the
anisotropic growth of a crystal in the solution phase when no
template is involved. In the first mechanism, the solid materials
are characterized by a highly anisotropic crystallographic
structure, and they naturally grow into nanorods and nanowires.
Typical examples include trigonal Se and Te, as well as wu¨rtzite
CdS and CdSe.15 The second mechanism involves twin or
stacking faults, which has been observed for both 5-fold twinned
nanorods and 2-fold twinned nanobeams of fcc metals.16 The
twin planes can generate re-entrant grooves, favorable sites for
the addition of atoms.16c The absence of an anisotropic crystal
structure or twin defects in both Pd nanobars and nanorods led
us to propose a third mechanism, in which localized oxidative
etching plays the pivotal role to induce anisotropic growth in a
short period of time.
This argument is supported by the experiments with different
degrees of etching. Figure 3A shows a TEM image of the
product obtained from a synthesis that was protected by
continuous Ar flow. The resultant nanorods exhibited a decrease
in aspect ratio as compared to the product prepared in air
because oxidative etching was partially blocked due to the
reduced level of air in the solution. However, because oxygen
binds strongly to Pd surface, it is impossible to completely block
oxidative etching in a Pd synthesis by simply bubbling an inert
gas through the reaction solution. Most recently, we demon-
strated that addition of citric acid was able to efficiently block
oxidative etching.21 Following this strategy, we added citric acid
to the current synthesis, and the product was found to contain
(17) (a) Xiong, Y.; Chen, J.; Wiley, B.; Xia, Y.; Yin, Y.; Li, Z.-Y. Nano Lett.
2005, 5, 1237. (b) Xiong, Y.; Wiley, B.; Chen, J.; Li, Z.-Y.; Yin, Y.; Xia,
Y. Angew. Chem., Int. Ed. 2005, 44, 7913. (c) Tsung, C.-K.; Kou, X.; Shi,
Q.; Zhang, J.; Yeung, M. H.; Wang, J.; Stucky, G. D. J. Am. Chem. Soc.
2006, 128, 5352.
(15) (a) Peng, X.; Manna, L.; Yang, W.; Wickham, J.; Scher, E.; Kadavanich,
A.; Alivisatos, A. P. Nature 2000, 404, 59. (b) Peng, X. Chem.-Eur. J.
2002, 8, 334. (c) Xia, Y.; Yang, P.; Sun, Y.; Wu, Y.; Mayers, B.; Gates,
B.; Yin, Y.; Kim, F.; Yan, H. AdV. Mater. 2003, 15, 353. (d) Peng, X.
AdV. Mater. 2003, 15, 459.
(16) (a) Germain, V.; Li, J.; Ingert, D.; Wang, Z. L.; Pileni, M. P. J. Phys.
Chem. B 2003, 107, 8717. (b) Sun, Y.; Mayers, B.; Herricks, T.; Xia, Y.
Nano Lett. 2003, 3, 955. (c) Lofton, C.; Sigmund, W. AdV. Funct. Mater.
2005, 15, 1197. (d) Wiley, B. J.; Wang, Z.; Wei, J.; Yin, Y.; Cobden, D.
H.; Xia, Y. Nano Lett. 2006, 6, 2273. (e) Elechiguerra, J. L.; Reyes-Gasga,
J.; Yacaman, M. J. J. Mater. Chem. 2006, 16, 3906.
(18) (a) Schimpf, J. A.; Abreu, J. B.; Soriaga, M. P. J. Electroanal. Chem. 1994,
364, 247. (b) Carrasquillo, A.; Jeng, J.-J., Jr.; Barriga, R. J.; Temesghen,
W. F.; Soriaga, M. P. Inorg. Chim. Acta 1997, 255, 249. (c) Lucas, C. A.;
Markovic´, N. M.; Ross, P. N. Phys. ReV. B 1997, 55, 7964. (d) Zou, S.;
Gao, X.; Weaver, M. J. Surf. Sci. 2000, 452, 44.
(19) (a) Newman, R. C.; Sieradzki, K. Science 1994, 263, 1708. (b) Scully, J.
C. The Fundamentals of Corrosion, 3rd ed.; Pergamon Press: Oxford, New
York, 1990; pp 1-57.
(20) Sun, Y.; Xia, Y. J. Am. Chem. Soc. 2004, 126, 3892.
(21) Xiong, Y.; McLellan, J. M.; Yin, Y.; Xia, Y. Angew. Chem., Int. Ed. 2007,
46, 790.
9
J. AM. CHEM. SOC. VOL. 129, NO. 12, 2007 3669

A R T I C L E S
Xiong et al.
Figure 3. TEM images of Pd nanostructures synthesized when oxidative etching was retarded (A) by bubbling with Ar gas and (B) in the presence of 0.13
M citric acid; or enhanced (C) in the presence of 0.20 M NaCl and (D) 1 M HCl. All of the syntheses were carried out under the same condition as in Figure
1D, except for the difference noted above.
nanocubes and short nanobars instead of nanorods (see Figure
3B). This observation indicates that the lack of oxidative etching
inhibited the anisotropic growth of nanocrystals. On the other
extreme, when oxidative etching was enhanced, etching would
take place over the entire surface of each particle rather than
locally on a specific face, in a way to eliminate the difference
between different faces. When NaCl was added to the reaction
system to enhance oxidative etching, we observed a decrease
in aspect ratio for the nanostructures as shown in Figure 3C.
The oxidative etching could be further enhanced by adding HCl
instead of NaCl.10 Figure 3D shows a typical TEM image of
the sample prepared with the use of HCl. The aspect ratio of
Pd nanostructures was further reduced to 1-3, as compared to
Figures 1D and 3C. From these results, it can be concluded
that localized oxidative etching is responsible for the anisotropic
growth of Pd nanobars and nanorods.
Adsorption of Bromide on the Surface of Pd Nanostruc-
tures. In the present synthesis, bromide played a critical role
in the formation of nanostructures enclosed by the {100} and
{110} side facets. According to the Wulff construction, the
thermodynamically favored shape of an fcc metal is cubocta-
hedron under the vacuum condition.8,9 In a solution-phase
process, impurities or capping agents can alter the surface free
energies via adsorption and thus induce new shapes.8c For
instance, the {100} facets of an fcc metal can be stabilized
through chemical interactions with the oxygen and/or nitrogen
atoms of the pyrrolidone units of PVP, leading to the formation
of truncated Pd nanocubes >25 nm in size.16b,17a However, a
polymer is too big to have a capping effect on smaller
nanocrystals. As a result, the PVP-stabilized Pd nanoparticles
<10 nm in size usually have a cubooctahedral shape.9 When
bromide was introduced into the reaction system, it is able to
chemisorb onto the surface of Pd seeds and alters the order of
surface free energies.18 This modification to surface energy also
plays an important role in facilitating the formation of aniso-
tropic Pd nanostructures such as nanobars and nanorods that
are enclosed by {100} and {110) side facets.
To decipher the role of bromide, we performed a series of
syntheses by altering the concentration of KBr in the synthesis,
and the results are summarized in Figure 4. As expected,
synthesis in the absence of bromide only produced cuboctahe-
drons. Addition of bromide gradually alters the shape of products
to nanobars and nanorods. The appearance of these new shapes
implies that the presence of bromide can promote the formation
of {100} and {110} facets. The concentration of bromide is
also critical to the control of product shape. At low concentration
of bromide, the product was a mixture of both nanobars and
nanorods (see Figure 4B). When the molar ratio of KBr to Na2-
PdCl4 was increased from 8 through 15 to 30, the major product
became nanorods (Figures 4B,C and 1D). Further increase of
bromide concentration led to a decrease for the aspect ratio, as
shown in Figure 4D. This dependence of morphology on the
concentration of bromide can be understood by the following
argument: when the concentration of bromide was increased,
the coverage of bromide on particle surface also increased. A
higher coverage would protect the nanocrystals better from
oxidative etching, and thus make the reaction faster, favoring
the formation of nanorods. Because oxidative etching is the key
to the formation of nanorods, lack of etching in the reaction
system would also restrain their anisotropic growth, resulting
in a decrease of the aspect ratio.
It is worth pointing out that the major function of bromide
in the current synthesis is its chemisorption on nanocrystal
surface. As bromide is less corrosive than chloride, it cannot
9
3670 J. AM. CHEM. SOC. VOL. 129, NO. 12, 2007

Synthesis of Palladium Nanobars and Nanorods
A R T I C L E S
Figure 4. TEM images of Pd nanostructures synthesized in the presence of KBr at different concentrations, demonstrating the role of bromide in promoting {100}
and {110} surfaces. The molar ratio of KBr to Na2PdCl4 was: (A) 0; (B) 8; (C) 15; and (D) 50. All of the syntheses were carried out under the same condition as in
Figure 1D, except the variation of KBr concentration.
Figure 5. (A,B) EDX spectra taken from the nanobars shown in Figure 1B: (A) as-synthesized and (B) after extensive washing with water. (C) High-resolution Pd
3d XPS core level spectrum taken from the nanobars. (D) High-resolution Br 3d XPS core level spectrum recorded from the nanobars. The nanobars were extensively
washed with water before XPS measurements.
make a major contribution to the oxidative etching. For example,
when Na2PdBr4 instead of Na2PdCl4 was used as the precursor,
the morphology exhibited a similar dependence on the concen-
tration of bromide, as shown in Figure S3. As compared to the
9
J. AM. CHEM. SOC. VOL. 129, NO. 12, 2007 3671

A R T I C L E S
Xiong et al.
case with Na2PdCl4, the samples prepared with Na2PdBr4
displayed lower aspect ratios. This difference can be attributed
to the fact that the rate of oxidative etching was reduced in the
absence of chloride (from the Na2PdCl4 precursor).
The presence of bromide on the surface of Pd nanostructures
was further confirmed by EDX analysis. Figure 5A and B shows
EDX spectra taken from Pd nanobars without washing and after
extensive washing with deionized water, respectively. The
presence of bromine signals after extensive washing implies that
there is a high concentration of bromide strongly bonded to the
particle surface. The binding energies of Pd and Br were
examined by XPS analysis. The depth of XPS penetration is
about 6-10 nm, which ensures the detection of all of the atoms
on or in each nanobar. Figure 5C shows the Pd 3d core level
spectrum, which includes two sets of 3d peaks. One set shows
Pd 3d5/2 and Pd 3d3/2 at 335.2 and 340.5 eV, respectively, which
are in accordance with the literature values for bulk Pd(0).22
The other set includes a Pd 3d5/2 peak at 337.5 eV and a Pd
3d3/2 peak at 342.6 eV, which are close to the reported values
for Pd(II)-type samples such as PdBr2.22 The ratio of Pd(0) to
Pd(II) is about 5.27. As shown by the XRD and ED patterns in
Figure S2, there should be no PdBr2, PdO, or other impurities
in the sample. A percentage of Pd(II) as high as 16% suggests
that Pd atoms on the surface donated electrons to Br for chemical
bonding. The Br 3d5/2 and Br 3d3/2 binding energies are
determined to be 68.7 and 69.8 eV, respectively (see Figure
5D), which were very close to the literature values for KBr (68.5
and 69.6 eV).22 As compared to the spectrum of KBr (Figure
S4), the Br 3d peaks in Figure 5D are much broader. The
broadening and small shift of the Br 3d peaks could be attributed
to the binding of Br to Pd surface.
The selection of an appropriate anion for chemisorption is
also instrumental in the growth of Pd nanocrystals with
anisotropic shapes. As reported in literature, the strength of
chemisorption or surface coordination to Pd surface increases
in the order of chloride < bromide < iodide.18b For this reason,
when KCl was used instead of KBr, the surface protection and
promotion of {100} and {110} surfaces were not remarkable,
and the product mainly consisted of cuboctahedrons (Figure 6A).
On the contrary, the chemisorption of iodide on the nanocrystal
surface is so strong that the particles are difficult to grow into
larger sizes with a well-defined shape (Figure 6B).
Controlling the Shape of Pd Nanostructures. Kinetic
control has been used to synthesize metal nanostructures with
a number of different shapes. For example, we and other groups
have prepared nanoplates of Ag, Au, and Pd by slowing the
reduction kinetics.10,12 The slow reduction can induce the
formation of stacking faults and thus break the cubic sym-
metry.16 However, over 70% of the surface of these nanostruc-
tures were covered by the {111} facets. In the current system,
a different mechanism seems to be involved, although it also
involves manipulation of reduction and growth kinetics. Based
on the results presented in the previous sections, it is not difficult
to recognize that three key factors are responsible for the
formation of Pd nanostructures enclosed by {100} and {110}
facets: bromide adsorption, oxidative etching, and reduction
kinetics. In the presence of surface adsorption and localized
Figure 6. TEM images of Pd nanoparticles synthesized by using KCl or
KI instead of KBr: (A) KCl and (B) KI. All of the syntheses were carried
out under the same condition as in Figure 1D, except for the substitution
of KBr by KCl or KI at the same molar concentration.
oxidative etching, the rates of nucleation and atomic addition
to the surface of seeds (i.e., growth) can be manipulated by
controlling the reduction kinetics to generate well-defined
nanostructures such as nanobars and nanorods with different
aspect ratios.
As shown in the schematic illustration (Figure 7A), the
adsorption of bromide can alter the order of surface free
energies, and slightly truncated nanocubes rather than cuboc-
tahedrons will become the most favorable shape. When the
reduction rate is in the low or medium regions, the seeds take
a cubic shape with slight truncation at the corners. Although
the fcc lattice of Pd is highly symmetric, localized oxidative
etching creates an active site on one of the six faces, for
example, the (100) face, for atomic addition and facilitates the
preferential growth along the a-axis in a short period. This
preference will break the symmetry of an fcc structure and lead
to the formation of nanobars. At a low rate of atomic addition,
the rate of growth along the a-axis is not much higher than that
of etching on the (100) face, so that the effect of preferential
growth along the a-axis is not distinct and thus the aspect ratios
of the resultant nanobars are very small (1-1.2). When atomic
addition is fast enough, preferential growth on the (100) face
becomes most favorable, leading to the formation of nanobars
with higher aspect ratios.
(22) (a) Kumar, G.; Blackburn, J. R.; Jones, M. M.; Albridge, R. G.; Moddeman,
W. E. Inorg. Chem. 1972, 11, 296. (b) Rochefort, A.; Bertolini, J. C.; Abon,
M.; Delichere, P. Surf. Sci. 1993, 294, 43. (c) Briggs, D.; Seah, M. P.
Practical Surface Analysis; John Wiley & Sons: New York, 1993; Vol. 1.
9
3672 J. AM. CHEM. SOC. VOL. 129, NO. 12, 2007

Synthesis of Palladium Nanobars and Nanorods
A R T I C L E S
Figure 7. Schematic illustration of the mechanisms responsible for the formation of nanobars and nanorods, as well as the morphological changes in an
aging process: (A) nucleation and formation of nanobars at a medium reduction rate; (B) nucleation and formation of nanorods at a fast reduction rate; (C)
decrease of aspect ratio for nanorods; (D) evolution of nanobars into nanorods; and (E) evolution of nanobars with an aspect ratio of ∼1 (i.e., nanocubes)
into cuboctahedrons.
As the reduction rate becomes much faster, the synthesis will
follow a different path. Under this condition, more seeds will
be formed in the nucleation stage. At the same concentration
of Pd precursor, an increase in the number of seeds results in
the production of nanocrystals with smaller sizes. On the other
hand, the fast nucleation makes the cubic seeds have more
significant truncation at the corners, with a shape more close
to cuboctahedron. As illustrated in Figure 7B, preferential
growth of a cubooctahedral seed along the a-axis will induce
{110} facets as well as {100} facets to be the side surface. The
appearance of {110} facets as the side faces will result in the
rod-shaped nanostructures with an octagonal cross-section. As
a result, the final product is dominated by nanorods thinner than
the nanobars. This mechanism can also explain why the shape
of the final product can be tailored by manipulating the reduction
kinetics.
Formation of Au Nanorods and Pt Nanobars. We have
also applied the same strategy to the synthesis of single-crystal
Au nanorods. Figure 8A shows a typical TEM image of Au
nanorods synthesized in the presence of KBr. The nanorods have
a diameter of ca. 2.5 nm and length up to 20 nm. The yield of
nanorods was around 40%, while the other 60% of the product
was microscale aggregates. Such Au nanorods have been
intensively studied by several groups, and here we note that
their structure is the same as that of the Pd nanorods with the
side faces being bounded by a mix of {100} and {110} facets.8,11
Murphy and co-workers have also noticed that the bromide in
the surfactant hexahecyltrimethylammonium bromide (CTAB)
might play an important role in the formation of Au nanorods.23
The bromide was thought to adsorb onto the side faces of Au
seeds and then confine the lateral growth of nanorods by
recruiting the cationic surfactant to form a bilayer. The long
surfactant tails were proposed as a major driving force for the
anisotropic growth. In addition, a small amount of Ag cations
must be added to induce the formation of single-crystal Au
nanorods in their work. Here, we found that the bromide alone
could also alter the surface energies and promote highly
anisotropic growth. This unique feature of bromide will certainly
attract the attention of the research community, and it is
anticipated that it will serve as a new knob for controlling the
shape of metal nanostructures. Note that the size of our Au
(23) (a) Gao, J.; Bender, C. M.; Murphy, C. J. Langmuir 2003, 19, 9065. (b)
Murphy, C. J.; Sau, T. K.; Gole, A. M.; Orendorff, C. J.; Gao, J.; Gou, L.;
Hunyadi, S. E.; Li, T. J. Phys. Chem. B 2005, 109, 13857. (c) Murphy, C.
J.; Gole, A. M.; Hunyadi, S. E.; Orendorff, C. J. Inorg. Chem. 2006, 45,
7544.
9
J. AM. CHEM. SOC. VOL. 129, NO. 12, 2007 3673

A R T I C L E S
Xiong et al.
Figure 8. TEM images of Au nanorods and Pt nanobars synthesized using a similar approach: (A) Au nanorods (with the inset showing a blow-up image),
and (B) Pt nanobars. (C,D) High-resolution TEM images of (C) a Pt nanobar recorded along [010] and (D) a Pt nanobar recorded along [11h0]. The syntheses
were performed with a molar ratio of KBr to HAuCl4 (or Na2PtCl6) at 30, in the presence of 75 mM PVP, and in an 11-mL mixture of EG (72.7%) and water
at 140 °C.
nanorods is much smaller as compared to that of the Au
nanorods made by the Murphy group (3 nm vs 15-30 nm in
thickness).
had been evolved into nanorods with both {110} and {100}
facets as side faces. The high-resolution TEM study also
revealed that the nanobars with an aspect ratio of ∼1 trans-
formed into cuboctahedrons, a shape more favored by thermo-
dynamics, as depicted in Figures 9D and S6. The short nanorods
formed through the aging process should be some intermediates
on the pathway to the thermodynamically favorable shape.
This method can also be extended to Pt. As shown in Figure
8B, the product consisted of nanobars with an aspect ratio of
2-3 and a yield above 90%. Figure 8C shows a high-resolution
TEM image of a Pt nanobar recorded along [010]. The fringes
show a period of 1.9 Å, which was consistent with the {200}
lattice spacing of fcc Pt. This image also displayed well-resolved,
continuous fringes with the same orientation, suggesting that
each Pt nanobar is a piece of single crystal enclosed by {100}
facets. Figure 8D shows a high-resolution TEM image of a Pt
nanobar with a different structure, which was recorded along
[11h0]. The side facets are {110}, and the growth direction is
also [001]. The ends of the nanobar are terminated by the (001)
face and {111} facets. These two types of 1-D structures are
different from the Pt nanorods and nanowires growing along
the [111] axis that were prepared using the iron-mediated polyol
process.24 This result also supports the argument that addition
of bromide favored the formation of {100} and {110} surfaces.
Stability of the Pd Nanostructures. As discussed above,
the formation of Pd nanobars and nanorods was governed by
kinetics. For this reason, they are not thermodynamically stable
due to their relatively high surface energies. To evaluate their
stability, the products were aged in the reaction solution for 2
weeks at room temperature. As shown in Figure 9A, the
diameter of nanorods increased while their length decreased
during the aging process. For the nanobars, the aging process
made them more severely truncated (see Figure 9B). High-
resolution TEM images (Figure 9C) indicate that the nanobars
These observations imply that the anisotropic nanostructures
formed through kinetic control were not stable shapes and they
would evolve into the thermodynamically favorable shape. This
aging process is supposed to be mainly governed by thermo-
dynamics, following a process illustrated in Figure 7C-E. The
loss of atoms from {111} facets and the addition of atoms to
{100} facets resulted in the enlargement of {111} facets and
the shrinkage of {100} facets, transforming Pd cubes into
cuboctahedrons (see Figure 7E). Multiple processes might be
involved in this shape evolution, such as oxidative etching,
desorption of bromide, and surface diffusion. However, mini-
mization of the surface energy of a particle should be the major
driving force for such a transformation.
It is worth pointing out that the Pd nanorods in Figure 1
should be formed via growth from small, near spherical seeds,
rather than through the evolution of nanobars in an aging
process. This point was supported by the evolution of shape,
which was monitored with TEM by taking samples at various
reaction times (Figure S7). The TEM observation suggests that
the Pd nanorods could be formed in the stage as early as t )
10 min. The smaller diameter of nanorods relative to nanobars
(2 vs 6 nm, Figure 1) also supports this argument because
nanorods evolved from nanobars during aging should have a
thicker diameter.
(24) (a) Chen, J.; Herricks, T.; Geissler, M.; Xia, Y. J. Am. Chem. Soc. 2004,
126, 10854. (b) Teng, X.; Yang, H. Nano Lett. 2005, 5, 885. (c) Song, H.;
Kim, F.; Connor, S.; Yang, P. J. Phys. Chem. B 2005, 109, 188.
9
3674 J. AM. CHEM. SOC. VOL. 129, NO. 12, 2007

Synthesis of Palladium Nanobars and Nanorods
A R T I C L E S
Figure 9. Electron microscopy characterization after the Pd nanorods and nanobars in Figure 1F and A had been aged in the reaction solution for 2 weeks
at room temperature: (A) TEM image of nanorods after aging; (B) TEM image of nanobars after aging; (C) high-resolution TEM image of a nanobar with
an aspect ratio of ∼2.5 after aging that was recorded along [11h0]; and (D) high-resolution TEM image of a nanobar with an aspect ratio of ∼1 after aging
that was recorded along [010]. The insets in panels (A) and (B) show two TEM images of the corresponding samples before the aging process, at the same
magnification as in (A) and (B).
4. Conclusion
synthesis. This work also suggests the role of bromide as a
capping agent to influence the surface energies of different facets
of an fcc metal and therefore induce the formation of new shapes
not favorable from the perspective of thermodynamics. Cer-
tainly, a more direct contribution of this work to the chemistry
community is the demonstration of a simple and versatile
method for growing highly anisotropic, single-crystal nano-
structures of noble metals, including Pd, Au, and Pt. Although
modification to the experimental procedures might be required,
it is expected that this generic method could be further extended
to grow longer Pd nanowires that will find immediate use in
the fabrication of electronic and sensing devices.
We have demonstrated a new strategy for growing highly
anisotropic nanostructures of Pd, single-crystal nanobars bounded
by {100} facets and single-crystal nanorods with their side
surfaces enclosed by {100} and {110} facets. According to both
the theoretical prediction and the experimental observation, Pd
atoms should nucleate and grow in a solution phase to form
the thermodynamically favored shape, cuboctahedrons with the
surface bounded by a mix of both {111} and {100} facets.9,10
In the current work, we combined the following three processes
to obtain elongated nanostructures: (i) speedy reduction of
precursor to ensure prompt addition of atoms to the seed; (ii)
chemisorption of bromide on the seed surface to alter the order
of free energies of different facets and promote the formation
of {100} and {110} facets; and (iii) localized oxidative etching
on one specific face of the seed to break the cubic symmetry.
The involvement of bromide absorption was confirmed by XPS
and EDX analyses on the products. These anisotropic nano-
structures formed via kinetic control were not stable, and they
tended to evolve into the thermodynamically favored shape
during aging.
This work demonstrates, for the first time, that cubic
symmetry of an fcc metal can not only be broken by forming
twin defects but also through localized oxidative etching. Note
that the localized oxidative etching is usually caused by a trace
amount of oxygen, chloride, and water, which could originate
from a minor change in experimental parameters such as the
source of chemical reagents and the choice of glassware. These
results emphasize again the importance and necessity to
precisely control the experimental parameters in a solution-phase
Acknowledgment. This work was supported in part by the
ACS (PRF-44353-AC10) and NSF (DMR-0451788), as well
as a fellowship from the David and Lucile Packard Foundation.
X.Y. is a Camille Dreyfus Teacher Scholar (2002-2007). J.W.
has been supported by a grant from CNMT-MOST (No.
M105KO010026-05K1501-02611). We are grateful to Dr. Lara
Gamble for her assistance with XPS analysis. This work used
the Nanotech User Facility (NTUF), a member of the National
Nanotechnology Infrastructure Network (NNIN) funded by the
NSF.
Supporting Information Available: Filtered high-resolu-
tion TEM images of the nanobars and nanorods of Pd and Pt;
PXRD and ED patterns of Pd nanobars; and TEM images of
Pd nanoparticles synthesized with Na2PdBr4 as the precursor.
This material is available free of charge via the Internet at
http://pubs.acs.org.
JA0688023
9
J. AM. CHEM. SOC. VOL. 129, NO. 12, 2007 3675