In situ study on the wet chemical synthesis of nanoscopic Pt colloids by reductive stabilization

Article abstract of DOI:10.1021/jp022615j

The synthesis of colloidal Pt particles obtained from the reaction between platinum acetylacetonate (Pt(acac)₂) and Al(CH₃)₃ was monitored, and the formation of the Pt colloid via reductive stabilization was established by a combination of techniques including NMR, anomalous small-angle X-ray scattering (ASAXS), X-ray absorption spectroscopy (XAS), density functional theory (DFT) computations, and transmission electron microscopy (TEM). For the first time, a defined organometallic compound was identified as the common intermediate in the formation of an organometallic Pt complex or a Pt colloid, depending on the type of ligands or stabilizers present. During the decomposition of this intermediate state, stable colloidal particles of just one size are observed the number of which increases with time.

Full text of DOI:10.1021/jp022615j

J. Phys. Chem. B 2003, 107, 7507-7515
7507
In Situ Study on the Wet Chemical Synthesis of Nanoscopic Pt Colloids by “Reductive
StabilizationӠ
‡
§
Klaus Angermund, Michael B u1 hl, Uwe Endruschat, Frank T. Mauschick, Reinhard M o1 rtel,
⊥
Richard Mynott,* Bernd Tesche, Norbert Wald o1 fner, and Helmut B o1 nnemann*
Max-Planck-Institut f u¨ r Kohlenforschung, D-45466 M u¨ lheim an der Ruhr, Germany
|
Gesa K o1 hl, Hartwig Modrow, and Josef Hormes
Physikalisches Institut, UniVersit a¨ t Bonn, Nussallee 12, D-53115 Bonn, Germany
Eckhard Dinjus and Franz Gassner#
Forschungszentrum Karlsruhe, Institut f u¨ r Technische Chemie, Postfach 3640, D-76351 Karlsruhe, Germany
Heinz-G u1 nter Haubold and Thomas Vad
Institut f u¨ r Festk o¨ rperforschung, Forschungszentrum J u¨ lich GmbH, D-52425 J u¨ lich, Germany
Martin Kaupp
Institut f u¨ r Anorganische Chemie, UniVersit a¨ t W u¨ rzburg, Am Hubland, D-97074 W u¨ rzburg, Germany
ReceiVed: December 18, 2002; In Final Form: May 8, 2003
The synthesis of colloidal Pt particles obtained from the reaction between platinum acetylacetonate
2 3 3
(Pt(acac) ) and Al(CH ) was monitored, and the formation of the Pt colloid via “reductive stabilization” was
established by a combination of techniques including NMR, anomalous small-angle X-ray scattering (ASAXS),
X-ray absorption spectroscopy (XAS), density functional theory (DFT) computations, and transmission electron
microscopy (TEM). For the first time, a defined organometallic compound was identified as the common
intermediate in the formation of an organometallic Pt complex or a Pt colloid, depending on the type of
ligands or stabilizers present. During the decomposition of this intermediate state, stable colloidal particles of
just one size are observed the number of which increases with time.
1. Introduction
Recently, we opened a novel, straightforward, and general
pathway to mono- and bimetallic nanoparticulates via the
reductive stabilization” of colloidal transition metals using
“
triorganoaluminum compounds, which act both as reducing
agents and as colloid stabilizers.1
Unreacted organoaluminum groups (e.g., Al-CH3 groups)
are still present in the colloidal stabilizer and may be substituted
subsequently by, for example, long-chain alcohols to tailor the
1
dispersion characteristics. Various bifunctional organic “spacer”
can collide in solution during the embryonic stage of the
nucleation to form an irreversible “seed” of stable metal nuclei.
Alternatively, the single metal atoms may be stabilized by
complexation with ligands such as olefins and phosphines to
molecules (e.g., R,ω-dialcohols) have been used to cross-link
aluminum-organic-stabilized platinum nanoparticles to form
2
three-dimensional networks. It is generally accepted that when
metal colloids are synthesized by wet chemical reductions, the
4
3
give organo-transition metal complexes.
metal salt is first reduced to give zerovalent metal atoms. These
5
In a preliminary communication, we reported that in the case
†
of Pt we were able to follow the formation of the colloid by a
combination of mass-spectrometry, NMR, anomalous small-
angle x-ray scattering (ASAXS), X-ray absorption spectroscopy
Part of the special issue “Arnim Henglein Festschrift”.
To whom correspondence should be addressed. E-mail addresses:
*
boennemann@mpi-muelheim.mpg.de, mynott@mpi-muelheim.mpg.de.
‡
Present address: W. C. Heraeus GmbH & Co. KG, Chemicals Division,
(XAS), molecular modeling, density function theory (DFT)
Heraeusstrasse 12-14, 63450 Hanau, Germany.
§
computations, and transmission electron microscopy (TEM)
(Figure 1). This publication documents these studies in detail.
Present address: Euro Engineering AG, Sedanstrasse 10, 89077 Ulm,
Germany.
⊥
Present address: MagForce Applications GmbH, Spandauer Damm
1
30, 14050 Berlin, Germany.
|
2. Experimental Section
Present address: Center for Advanced Microstructures and Devices,
Baton Rouge, LA 70806.
#
2.1. Synthesis of Colloidal Pt Nanoparticles. Platinum
acetylacetonate (1.97 g, 5 mmol) [Pt(acac)2] was dissolved under
Present address: Degussa Galvanotechnik GmbH, Klarenbergstr. 53-
7
9, 73525 Schwaebisch Gmuend, Germany.
1
0.1021/jp022615j CCC: $25.00 © 2003 American Chemical Society
Published on Web 07/09/2003

7508 J. Phys. Chem. B, Vol. 107, No. 30, 2003
Angermund et al.
analytical calculation of the harmonic vibrational frequencies
at that level. Nuclear magnetic shieldings σ have been evaluated
for the BP86/ECP1 geometries using a recent implementation
of the gauge-including atomic orbitals (GIAO)-DFT method,12
involving the functional combinations according to Becke
13
14
(
hybrid) and Lee, Yang, and Parr, (denoted B3LYP), together
with basis II′, that is, the same ECP and valence basis as in
ECP1 basis on Pt, and the recommended IGLO-basis II15 on
all other atoms except H, for which a double-ú basis was used
15
(
[2s] contraction). This particular combination of methods has
been employed successfully for chemical-shift computations of
1
6
transition metal complexes. Chemical shifts δ have been
13
calculated relative to TMS computed at the same level [σ( C)
181.0]. Unless otherwise noted, energies are reported at the
)
B3LYP/II′//BP86/ECP1 level, corrected for the BP86/ECP1
zero-point energies. These computations were performed using
1
7
the Gaussian 98 program package.
Third- and second-order spin-orbit corrections to the 13C
nuclear shieldings were computed separately by the combined
finite-perturbation/SOS-DFPT approach of ref 18 using modified
1
9
20
versions of the deMon-KS and deMon-NMR codes. The
Figure 1. Mechanism of the colloid formation.
underlying Kohn-Sham calculations with deMon-KS employed
2
1
the PP86 functional, a FINE angular integration grid with 64
argon atmosphere in dry toluene (200 mL). Trimethylaluminum
-
3
radial shells, and a finite-perturbation parameter λ ) 10 (test
calculations with other choices of λ gave essentially identical
results). The same quasi-relativistic ECP and valence basis set
on Pt were employed as described above, together with IGLO-
(1.44 g, 20 mmol) [Al(CH3)3] was dissolved in toluene (200
mL) and carefully added over 4 h at 333 K. After 24 h, when
the gas evolution had stopped, the solution was filtered, and all
volatile components were completely evaporated in vacuo. The
solid residue was colloidal platinum (2.4 g). Elemental analy-
sis: Pt 40%, Al 14%, C 31%, H 6%.
In Situ Experiments. Pt(acac)2 (60 mg, 0.15 mmol) was
dissolved in [D8]-toluene (5 mL) under argon atmosphere. A
solution of Al(CH3)3 (43.2 mg, 0.6 mmol) in [D8]-toluene (5
mL) was then added. The solutions were thoroughly mixed and
subsequently transferred to the respective instrument of char-
1
5
II all-electron basis sets for Al and C and a DZVD basis for
2
2
hydrogen. Auxiliary basis sets to fit density and exchange-
correlation potential were of the sizes 3,4 for Pt, 5,4 for Al, 5,2
for C, and 5,1 for H (n,m denotes n s-functions and m spd-
19
shells with shared exponents ).
18
The subsequent perturbation calculations with deMon-NMR
used the appropriate spin-orbit ECP for platinum10 combined
with all-electron atomic-mean field spin-orbit operators (AMFI
approximation) on all other atoms. A common gauge origin
was used on the respective carbon nucleus in question.
acterization.
2
2
3
.2. NMR. H and 13C NMR measurements were carried out
1
on a Bruker AMX-400 NMR spectrometer using an inverse 5
mm probe head. Some 13C NMR spectra were also recorded in
In addition, a topological analysis of the charge density (Bader
2
4
analysis) has been performed for the lowest minimum for 3
a separate experiment using a 10 mm BBX probe head. The
samples were prepared in deuteriotoluene. Chemical shifts were
determined relative to the solvent signals and converted to the
TMS scale [δH(CHD2) ≡ 2.08, δC(CD3) ≡ 20.4]. Spectrum
simulations were carried out using the computer programm
gNMR (Adept Scientific).
at the B3LYP/II′ level using the Morphy program.25
2.5. ASAXS. All measurements were performed with syn-
chrotron radiation from the storage ring DORIS at the small
2
6
angle scattering beamline JUSIFA (http://www.fz-juelich.de/
iff/personen/H.-G.Haubold). For the in situ small angle scattering
experiment, Pt-acetylacetonate, Pt(acac)2, was dissolved in
toluene, and a solution of trimethylaluminum, Al(CH3)3, in
toluene was then added under argon atmosphere. The reaction
mixture was transferred to a sealed capillary tube and quenched
to -78 °C. The samples were prepared at the MPI in M u¨ lheim
and brought to the JUSIFA beamline at the Hamburg Synchro-
tron Radiation Laboratory HASYLAB-DESY, where the mea-
surements were performed in situ after reaction times t of
between 0.8 and 63 h at room temperature.
2.3. XAS Measurements. Pt LIII-edge X-ray absorption near
edge spectroscopy (XANES) measurements were performed at
beamline BN3 at the ELectron Stretcher and Accelerator ring,
ELSA, in Bonn, operating in 2.7 GeV mode. The setup at this
6
beamline and type of the cell used for the measurements have
been described previously.7 Ge442 crystals were used for
monochromatization, allowing a step width of 0.7 eV between
the data points. The integration time per data point was 500
ms. The integration time per data point was 500 ms for XANES
and up to 1500 ms for extended X-ray absorption fine structure
2.6. Transmission Electron Microscopy. The samples were
investigated with a Hitachi HF 2000 transmission electron
microscope operating at 200 keV. The instrument was equipped
with a cold field emission source of W(310) and capable of
energy-dispersive X-ray (EDX) analysis. The objective lens was
a high resolution/analytical type ((15° specimen tilt) with a
spherical aberration coefficient of 1.2 mm, giving a point
resolution of 0.23 nm and an information limit of 0.16 nm. The
microscope was fitted with a Voyager 1000/S EDX system with
a high-angle takeoff geometry of 68°, 0.037 sr solid angle, and
an energy resolution for Mn of 140-143 eV. Spots as small as
(EXAFS) spectra.
2.4. DFT: Computational Details. Molecular structures have
been fully optimized in the given symmetry at the BP86/ECP1
level, that is, employing the exchange and correlation functionals
of Becke8 and Perdew,9 respectively, together with a fine
integration grid (75 radial shells with 302 angular points per
shell), relativistic MEFIT effective core potentials with the
corresponding valence basis sets for Pt1 (contraction schemes
0
[
6s5p3d]), and standard 6-31G* basis set11 for all other elements.
The nature of each stationary point was characterized by

Wet Chemical Synthesis of Nanoscopic Pt Colloids
J. Phys. Chem. B, Vol. 107, No. 30, 2003 7509
Figure 2. 1H NMR spectrum27 (400 MHz, panel a) of the reaction mixture after 5 h at 27°. The inset shows an expansion of the platinum methyl
2
195
1
4
195
1
2
195
1
4
195
1
signal of 3 at δ
H
) 0.51: (A) | J( Pt, H)| ) 35.7 Hz; (B) | J( Pt, H)| ) 5.7 Hz; (C) | J( Pt, H) + J( Pt, H)| (AA′X12X′12 spin system,
2
195
1
4 195 1
J( Pt, H) and J( Pt, H) have opposite signs). Panel b presents the analysis of the Pt-methyl multiplet using the coupling data from the isotopomer
with one Pt nucleus to analyze the AA′X X′′ spectrum. A value of 300 Hz was used for J( Pt, Pt) (see text).
n n
1
95
195
195
about 4 nm in diameter were analyzed in the present study. The
samples were supported on a 6 nm thick carbon film mounted
on a copper grid (400 meshes/inch).
establish whether a direct platinum-platinum bond is present.
However, to carry out such an analysis, it is necessary to find
the outermost lines of this multiplet. In this case, they are much
less intensive than the C satellites, and despite repeated
attempts, they could not be located. Therefore, while
1
3
3
. Results and Discussion
1
95
195
2
195
1
J( Pt, Pt) must be large compared with J( Pt, H) and
When Pt(acac)2 (1) and Al(CH3)3 (2) react in toluene, the
4
J(195Pt, H), an exact determination was not possible. The com-
puter simulations indicated that J( Pt, Pt) is at least 250 Hz.
The C NMR spectrum of the platinum-methyl groups is
made up, like the corresponding H NMR signal, of the
superimposed signals of isotopomers containing no Pt nuclei,
one Pt nucleus, and two Pt nuclei. For the same reasons
as for the H NMR spectrum, a full analysis of the AA′X
spectrum (i.e., of isotopomers with two Pt nuclei) was not
possible.
1
reaction mixture turns from bright yellow to black over the
1
95
195
1
13
course of about 24 h. H and C NMR, as well as Pt LIII-edge
XANES, and time-resolved ASAXS measurements were carried
out in situ to follow the initial stages of the reaction and to
identify the products.
1
3
1
1
95
1
95
195
1
1
95
1
Comparing the relative changes in intensity of the H signals
over the course of time assists in assigning the signals of the
1
reaction mixture. The only signal in the H NMR spectrum
1
showing the same variation as the platinum-methyl signal is a
singlet at -0.27 ppm within the group lines from aluminum-
methyl species. The integrated intensity of this resonance is one-
fourth of that of the Pt-methyl signal. Further evidence
supporting its assignment to the intermediate complex was the
observation of a cross-peak in a long-range COSY experiment
3
.1. NMR. In the first H NMR spectrum recorded ca. 30
min after the reagents had been mixed, a new signal was
observed at δH ) 0.51 (Figure 2). This is assigned to methyl
groups in an intermediate platinum complex on account of the
coupling with Pt ( Pt, I ) /2, 33.8% natural abundance).
After 60 min ca. 33% of the Pt(acac)2 had reacted. Over the
course of the next hours, this Pt-methyl signal became more
intensive and at the same time the number of signals from other
products increased.
Analysis of this platinum methyl resonance provides detailed
information about the structure of this intermediate complex.
The relative intensities of the lines in the multiplet are consistent
with this compound being a binuclear platinum complex. The
multiplet is made up of the superimposed signals of isotopomers
containing no Pt nuclei (43.8%), one Pt nucleus (44.8%),
and two 1 Pt nuclei (11.4%) (Figure 2b). The most informative
part is that due to the latter, which can be analyzed as the X
1
95
195
1
(optimized for small H,H couplings) between this signal and
the methyl signal at 0.51 ppm. Surprisingly, no coupling to
1
95
Pt was observed. These spectra are consistent with the
suggested structure 3.
The NMR study showed that to achieve a near quantitative
reaction of 1 the stoichiometric ratio of 1 to 2 should be greater
than 1:4. Also initially formed in the reaction is (CH3)2Al(acac)
1
(4). The H NMR spectrum indicates that after several hours a
195
195
number of additional products are formed by, for example,
reaction of excess 2 with 4. The time period over which NMR
can be measured is limited by the formation of metallic
particulates, hindering more extensive NMR studies.
95
region of an AA′XnX′n spin system (A ) 1 Pt, X ) H). The
complex must therefore be symmetrical, but the platinum nuclei
are not magnetically equivalent,28 that is, the methyl groups must
be terminal. Thus they cannot be located in bridges between
the platinum atoms. The form of the multiplet rules out any
exchange of the methyl groups between the platinum atoms in
the NMR time scale. No other platinum methyl signals are
observed.
95
1
The pure complex is rather unstable but can be kept in
solution for up to 2 days in the presence of a 5-fold excess of
Al(CH3)3. A sample of isolated 3 dissolved in [D8]-toluene
1
decomposed rapidly but not before a H NMR spectrum was
obtained that confirmed that just the two signals at 0.51 and
-0.27 ppm were present. The mass spectrum of 3 (M ) 594
g/mol) shows the characteristic mass of m/z 579. The observed
isotope distribution corresponds to the theoretical distribution
of two platinum atoms, confirming that these fragments originate
A full analysis of the AA′XnX′n spectrum was attempted
because it would allow J(1 Pt, Pt) to be determined and help
95
195

7510 J. Phys. Chem. B, Vol. 107, No. 30, 2003
Angermund et al.
Figure 3. Pt LIII edge XANES spectra of the ongoing reaction, which
leads to the intermediate state.
from a binuclear complex. The signal found at m/z 579 is due
to the cleavage of one methyl group. The byproduct 4 was also
detected by MS.
3
.2. XANES. The formation of the intermediate state was
also followed at room temperature by Pt LIII X-ray absorption
near edge spectroscopy (XANES) measurements. A liquid
sample cell similar to the one described in ref 7 was filled with
the educts of the synthesis in a glovebox, sealed with sticky
Kapton tape, and mounted in the sample chamber. A spectrum
was then recorded every 5 min for the next 24 h. The resulting
spectra are shown in Figure 3. Clearly, the intensity of the white
line, that is, the first intense absorption peak of the spectrum,
at about 11 568 eV decreases continuously, while at the same
time, an additional structure at approximately 11 576 eV gains
in importance. Whereas the first process can be attributed to a
reduction of oxygen coordination of the Pt absorber atoms,
which may be expected as a consequence of the decomposition
of the Pt(acac)2 molecule in the ongoing reaction, the appearance
of a second structure in the white line region can be assigned
to the presence of two different types of molecular orbitals into
which the electron can be excited. A less probable interpretation,
but one that cannot be excluded on the basis of the evidence
presented so far, is that two different new Pt compounds are
present in the solution.
Figure 4. Detailed comparison of the Pt LIII edge XANES spectra
taken during reaction and the optimal weighted addition of the basis
spectra as determined by MINUIT.
The XANES spectra change significantly during the first 80
min, but they remain more or less constant over the following
2
3 h. However, the reaction product is clearly different from
colloidal Pt particles. As illustrated in Figure 4, all of the spectra
that were measured while the reaction takes place can be
constructed using a basis set consisting of a spectrum of
Pt(acac)2 and the spectrum taken after 24 h. This indicates that
no further metastable intermediate reaction products involving
Pt are formed. The relative contributions of these two spectra
were estimated using a least-squares optimization based on the
routine “MINUIT”.29 The results, shown in Figure 5, indicate
a first-order reaction with a reaction constant k ) 0.03/min.
When it was attempted to fit the data assuming a reaction of
Figure 5. Relative contributions of product and educt material to the
combined spectra and least-squares fit assuming a reaction of first- (solid
line) and second-order (dotted line) reaction, respectively.
atoms at a distance of 244 ( 1 pm. Substitution of carbon by
oxygen gives a fit that is only slightly worse but is less
compatible with the XANES data. The data exclude the presence
of a direct Pt-Pt bond. While the radial distances are well
defined, the coordination numbers and Debye-Waller factors
are not given because they remained strongly correlated and
varied as a function of the k-weight.
The XANES spectrum of the final state, the colloid, is quite
different from that of the intermediate. EXAFS analyses were
performed on a dried sample to gain insight into its structure
(see Figure 6e,f and Table 3). As expected for a solid metal
nanoparticle, a clear dominance of Pt backscatterers at a distance
of 271 ( 1 pm is observed. This indicates that the Pt lattice
has contracted significantly. This is to be expected for a 1.2
nm particle, as a reconfiguration of the surface atoms, which
contribute significantly in a system of about 53 atoms, occurs.
Weak indications of the presence of a soft backscatterer such
as C at a distance of about 202 ( 2 pm are present, which are
related to the interaction of metal atoms at the surface of the
2
second order, the r factors were significantly worse.
To obtain further insight into the nature of the intermediate
state, EXAFS measurements were performed. As a first step,
Pt(acac)2 was analyzed as a reference to evaluate the degree of
detail, which could be extracted from the fit, and to determine
2
S0 . The results (see Figure 6a,b and Table 1) are in good
agreement with the XRD-based results given in ref 30, especially
bearing in mind that differences in bond length within a few
picometers cannot expected to be resolved by EXAFS. The
evaluation of the EXAFS-data of the intermediate state is
displayed in Figure 6c,d and Table 2. Carbon atoms are found
at a distance of 211 ( 1 pm from the absorbing Pt atom and Al

Wet Chemical Synthesis of Nanoscopic Pt Colloids
J. Phys. Chem. B, Vol. 107, No. 30, 2003 7511
Figure 7. Pt LIII edge XANES spectrum of the intermediate state
compared to the calculated spectrum for the input geometry determined
by the DFT calculations and for a system without bridging Al atoms.
SCHEME 1
the calculated structural input. For this calculation, the FEFF8
31
code was used, which uses a real space full multiple scattering
approach on a muffin-tin based, self-consistent potential. This
approach allows XANES spectra to be interpreted in great detail
[
see refs 32 and 33]. The result of this calculation is displayed
Figure 6. Modified Fourier transform (a,c,e) and weighted ø(k) (b,d,f)
functions of Pt(acac)
colloid (bottom).
in Figure 7. Good agreement with the experimental data is
reached after making corrections for a rather unphysical atomic
background. To evaluate the significance of this result, a
calculation was also performed for a structure in which Pt atoms
are coordinated only to methyl groups. The result proves that a
clear signature of a Pt-Al bond is visible in the experimental
data and that this bridging bond is necessary for the asymmetric
shape of the whiteline band.
2
(top), intermediate state (middle), and dried
TABLE 1: Fit Results for the EXAFS Analysis of Pt(acac)
2
2
Using S
0
) 0.95, a k-Weight of 3, the k-Range 2.1-14, and
the r-Range 1.2-2.7
path
degeneracy
E
0
[eV]
r [pm]
σ2
Pt-O-Pt
4
4
8
14.8(9)
11.7(3.1)
13.1(1.6)
200(1)
288(2)
303(5)
0.0017(4)
0.0020(10)
0.0032(20)
Pt-C-Pt
Crystallographic data for a diiridium complex (9a) having
bridging Al-alkyl units similar to 3 have recently been re-
Pt-C-O-Pt
3
4
ported. This is apparently the only literature reference to a
crystal structure of a binuclear transition metal complex with
such bridging monoalkyl-Al groups. The C2-symmetrical four-
membered ring formed by the two iridium atoms and the
bridging Al(C2H5) groups is planar with angles of 115.6°
TABLE 2: Fit Results for the EXAFS Analysis of the
2
Intermediate State Using S
0
) 0.95, a k-Weight of 3, the
k-Range 2.1-12, and the r-Range 1.3-2.5
path
E
0
[eV]
r [pm]
Pt-C
Pt-Al
10.8(1.4)
-1.1(1.9)
211(1)
244(1)
(Ir-Al-Ir) and 64.4° (Al-Ir-Al), as well as distances of 4.158
Å (Ir-Ir) and 2.456 Å (Ir-Al). While the two Ir atoms and the
ethyl ligand span a slightly distorted trigonal-planar coordination
polyeder at the Al atom, the phosphine substituent at the Ir atom
forms an angle of about 90° with the plane of the metalla cycle.
Further support for the unusual structure proposed for 3 is
provided by density functional calculations.
TABLE 3: Fit Results for the EXAFS Analysis of the Dried
2
Colloid Using S
0
) 0.95, a k-Weight of 3, the k-Range
2.1-11.0, and the r-Range 1.2-3.2
path
N
E
0
[eV]
r [pm]
σ2
Pt-Pt
Pt-C
5.0(5)
1.2(3)
0.0(9)
-3.6(3.2)
271(1)
202(2)
0.0104(7)
0.0096(31)
3.3. DFT Results. As detailed in the preceding sections,
NMR and EXAFS evidence suggested the presence of a Pt2Al2
core for 3. On the basis of the abovementioned precedence
for such an arrangement in the literature, namely, that for
(IrCp*P(CH ) ) (AlEt) (9a, Scheme 1), we first constructed a
colloid interacting with the surfactant. Relative coordination
numbers can be extracted from this analysis, yielding values of
about 5:1.2 for Pt and the soft backscatterer, respectively, which
can be related mainly to the small size of the particles, which
leads to a high surface contribution. A more detailed discussion
of other effects which might affect coordination numbers
determined by EXAFS is found in ref 48.
Further support for the structure of the intermediate state as
determined by the DFT calculations discussed below was gained
by an ab initio calculation of the Pt LIII XANES spectrum using
3
3 2
2
molecular model corresponding to conformation 3a depicted
in Scheme 2. The resulting Pt‚‚‚Pt distance was indicated to be
larger than 4 Å, in qualitative accord with the EXAFS data,
which precluded the presence of a bonding interaction between
the metal atoms.
To furnish additional support for this assumed structure, we
performed computations at a suitable level of density functional

7512 J. Phys. Chem. B, Vol. 107, No. 30, 2003
Angermund et al.
TABLE 4: Computed 13C Chemical Shifts for Pt(CH
a
SCHEME 2
3 4
)
Complexes
compound
nucleusb
B3LYP B3LYP+SOc exptd
Pt(CH3)4(TMEDA) (10) Cax
19.4
17.0
26.1
19.5
22.8
0.9
4.0
-9.3
17.6
12.0
14.8
-2.1
12.5
-8.6
Ceq
Pt2Al2(CH3)10 (3c)
Pt-Cterm
Pt-Cbr-Al
Pt-C avg
Al-Cterm
-0.1
-6.3
a
In ppm relative to TMS, basis II′ (including scalar relativistic
pseudopotentials on Pt). b ax ) axial; eq ) equatorial; term ) terminal;
c
br ) bridging; avg ) average of Pt-Cterm and Pt-Cbr-Al. Spin-
orbit correction evaluated at the PP86/II level. d Experimental data for
1
0 are taken from Kaim, W.; Klein, A.; Hasenzahl, S.; Stoll, H.; Z a´ li sˇ ,
S.; Fiedler, J. Organometallics 1998, 17, 237-247.
scale. Isomers 3c and 3d are thus plausible candidates for the
structure of 3 in solution.
To see whether the suggested structure 3c could be further
substantiated, we performed NMR chemical shift calculations.
This property is often quite sensitive to details of the molecular
structure. In particular, if the computed δ values deviate strongly
from the observed ones, this can be used to reject the structural
model on which the NMR calculation is based. In the present
case, special care must be taken to account for the heavy-atom
effect exerted by Pt on the chemical shifts of its neighboring
nuclei. Two theoretical approaches were applied, namely, first,
incorporation of scalar relativistic effects by means of a suitably
theory (DFT), denoted BP86/ECP1. It was first ensured that
the salient geometrical parameters for the Ir complex 9a can be
reproduced at that level. Model complex 9b was confirmed as
a minimum in Ci symmetry, and its optimized bond lengths
agreed within 4 pm with those observed for 9a.35 Thus, if 3
adopts such an arrangement, it should be well represented at
the chosen DFT level.
37
adjusted effective core potential on Pt at the B3LYP level and,
second, inclusion of spin-orbit (SO) coupling by an appropriate
18
perturbational treatment at the PP86 level.
13
The resulting δ( C) values are summarized for 3c in Table
When the Pt2Al2(CH3)10 complex in the assumed arrangement
4, together with those of Pt(CH ) (TMEDA) (10), a suitable
3
4
3a was subjected to a geometry optimization at the BP86/ECP1
reference molecule. For the latter, large SO effects are obtained,
level, isomer 3b (Ci symmetry) was obtained, in which the
Pt(CH3)4 moiety has rotated by approximately 45° along the
Pt‚‚‚Pt axis. This rotation allows the formerly axial CH3 groups
to interact with the adjacent electron-deficient Al centers. A
more thorough search on the potential energy surface afforded
two additional minima, 3c and 3d, in C2V and C2h symmetry,
respectively, both of which are also characterized by four
partially Pt-Al-bridging methyl groups (Scheme 2). Of these
minima, 3c is the most stable, but 3d is only 1-2 kcal/mol
higher in energy. In contrast, 3b is much less favorable, because
it is indicated to be at least 12 kcal/mol above 3c. Several
additional optimizations starting from different initial configura-
tions converged to any one of the isomers 3b-3d. In the lowest
minimum 3c, the optimized Pt‚‚‚Pt distance is 3.83 Å, consistent
with the estimated model structure and with the EXAFS result.
Also, the optimized Pt-Al and mean Pt-C distances, 2.440
and 2.123 Å,3 respectively, agree well with the EXAFS-derived
values of 2.44 and 2.11 Å, respectively (Table 2).
up to ca. 26 ppm for Cax (compare B3LYP and B3LYP+SO
13
entries), and the final, computed δ( C) values agree qualitatively
with experiment (maximum deviation 8.5 ppm). In contrast,
smaller SO effects are indicated for 3c, and larger deviations
from experiment remain, up to 14.9 ppm (Table 4). It is unclear
at this point whether this deviation originates in shortcomings
of the theoretical methods, for instance, the perturbational SO
treatment, in neglect of thermal and solvent effects (see below),
or in the presence of a different species in the NMR spectrum.
The calculated NMR data thus appear to be inconclusive.
In any event, the electronic structure of 3c is interesting.
Topological analysis of the B3LYP/II′ total electron density
affords bond paths between Pt and the four CH groups attached
3
to it and between Pt and its two neighboring Al atoms. Both
the Laplacian of the electron density and the electron localization
38
function (ELF) indicate increased charge density and electron
localization, respectively, between Pt and Al atoms, suggesting
the presence of polarized 2c2e bonds in the Pt Al core. Bond
6
2
2
39
In the NMR spectra of 3, all four CH3 groups bonded to Pt
are equivalent. In 3c and 3d, such equivalence can be brought
about by an exchange between bridged and terminal Pt-CH3
groups. In search of a corresponding transition state, a low-
lying second-order saddle point 3e (C2h symmetry, Scheme 2)
was located, which was found to be only 3.5 kcal/mol above
indices according to Wiberg’s definition (WBIs) are lower for
these Pt-Al bonds (0.38) than for the terminal Pt-CH3 or Al-
CH3 bonds (0.52-0.58). Despite a notable WBI between Al
and the bridging methyl carbons (0.24), no bond path is found
between these atoms. The remaining unoccupied d orbitals at
the Pt(IV) centers form the two lowest unoccupied MOs of 3c.
The presence of such low-lying acceptor orbitals would be
consistent with the chemistry of 3, for instance, with the rapid
degradation of the dinuclear complex upon addition of donor
ligands such as 1,5-cyclooctoadiene (COD). It is reasonable to
assume that the empty orbitals on Pt will facilitate the initial
addition of the diene, for example by formation of a monoden-
tate intermediate. Also, stabilization of 3c by solvation or by
interaction with other potential donors in solution is conceivable.
3
c, indicative of a highly fluxional character. We have not
located the true transition states, because the reduced symmetry
in combination with a flat potential energy surface would render
this search computationally quite expensive. Arguably, the actual
barriers on the potential-energy surface are even lower than the
3.5 kcal/mol required to reach 3e. This energy would thus suffice
to scramble all four methyl groups attached to Pt, in accordance
with the equivalence of these four groups on the NMR time

Wet Chemical Synthesis of Nanoscopic Pt Colloids
J. Phys. Chem. B, Vol. 107, No. 30, 2003 7513
Figure 8. Transmission electron micrograph (a) of Pt nanoparticles and (b) histogram of the particle size distribution.
SCHEME 3
To probe whether 3 could bind unsaturated donors such as
COD or the solvent toluene, we optimized two representative
model adducts, 3‚2C2H4, derived from 3c and 3d. Both
converged to the same C2 symmetric minimum 11 (Scheme 3).
The computed driving force for the formation of 11 is weak.
At the BP86/ECP1 level, the binding energy with respect to 3c
4
0
and two ethylene molecules is 11.3 kcal/mol (corrected for
basis-set superposition error), which is reduced to 1.2 kcal/mol
upon inclusion of the zero-point correction. In addition, ethylene
uptake is strongly disfavored entropically, so formation of stable
adducts such as 11 appears unlikely. Nonetheless, the availability
of two potential binding sites in 3 is confirmed, which could
be occupied by stronger donors or, possibly, by an aromatic
solvent present in large excess. Interestingly, on going from 3c
to 11, the computed δ( C) values are significantly shifted to
higher frequency. The mean value for the Pt-CH3 resonances
and that for the Al-CH3 signal decrease by ca. 6 and 9 ppm,
respectively (B3LYP/II′ level). Such a shielding would serve
to bring the B3LYP+SO values of 3c (Table 4) closer to
experiment. Unfortunately, the complexity of the system
precludes a rigorous computational treatment of structure and
properties of the bulk solution. It appears, however, that the
quadruply bridged structure 3c is not invalidated by the NMR
computations and that, taking all experimental and computa-
tional evidence together, it is a plausible structure for the
intermediate state in solution.
Figure 9. In situ ASAXS small angle scattering cross sections for
reaction times of 3.6 and 65.4 h at X-ray energies of E ) 11.46 keV
1
and E
2
3
) 11.43 keV in the preedge region of the L X-ray absorption
edge of Pt atoms and fitted difference cross sections of Pt nanoparticles
with mean radii 〈R〉 ) 5.8 Å and a monomodal lognormal particle size
distribution.39
To eliminate the unknown scattering contribution from the
organic molecules in the solution, we performed small angle
scattering measurements at an X-ray energy E1 ) 11.46 keV
and an energy E2 ) 11.54 keV in the preedge region of the L3
X-ray absorption edge of the Pt atoms. At this energy, the
squared atomic scattering amplitudes of the Pt atoms are reduced
13
4
2
by about 10% because of anomalous scattering, and conse-
quently, the scattering cross section of Pt structures becomes
smaller, whereas background scattering contributions remain
4
3-46
unaffected and can be subtracted out.
The ASAXS differ-
ence scattering cross section (E2 - E1) provides unbiased
information on the distribution of the Pt particle sizes even when
there are significant contributions from other nanostructures.43
Figure 9 gives the results after reaction times of 3.6 and 65.4
h at room temperature. From the fitted curves, one concludes
that platinum atoms nucleate into small particles with mean radii
〈R〉 ) 5.8 Å and a rather narrow monomodal log-normal particle
size distribution.
Figure 10 depicts the time dependence of the mean particle
radius 〈R〉 and of the mass fraction mparticle/mtotal of the Pt atoms
that have been transformed into particles as revealed by time-
resolved ASAXS. The amount of Pt in the particles, mparticle,
retrieved from the small angle scattering data, is measured in
quantitative absolute units; the total amount of Pt in the sample,
mtotal, was determined from the increase of the X-ray absorption
at the Pt-L3 X-ray absorption edge.
Confirmation that 5 is formed when the isolated complex 3
decomposes in solution in the presence of COD was obtained
from NMR and MS analyses.
When 3 decomposes, nondispersible globular Pt nanoparticles
of 1.4 nm diameter (6) are detected by TEM (Figure 8a). The
mean diameter is slightly greater than that in the (redispersible!)
Pt colloid as a result of agglomeration effects (see histogram,
Figure 8b).
3
.4. ASAXS. Time-resolved anomalous small-angle X-ray
scattering (ASAXS) was employed to monitor the formation
of colloidal Pt particles in solution in situ.41 The nucleation
process was studied for reaction times from 0.8 to 1000 h. The
strong increase in the intensity of small angle scattering by
almost an order of magnitude indicates the onset of cluster
formation in the solution.
During the experiment, the mass fraction of Pt atoms found
in stable particles of 1.2 nm in diameter increases. Remarkably,

7514 J. Phys. Chem. B, Vol. 107, No. 30, 2003
Angermund et al.
the absence of suitable ligandssfor nucleation processes on the
other, leading to the formation of nanoparticles (6) or metal
colloids (7). We find no evidence of stable configurations other
than 1.2 nm particles during the formation of nanoparticles or
metal colloids; in particular, no growth process is detected but
only an increase of the number of “completed” particles. Further,
the colloids 7 are suitable building blocks for Pt-nanoparticle
networks. This work is being extended to other metals to verify
this hypothesis.
Acknowledgment. We thank the Fonds der chemischen
Industrie and the Deutsche Forschungsgemeinschaft, Priority
program 1072, for financial support under Grant Nos. Bo 1135
and Ho 887 and for a Heisenberg fellowship for M.B., who
also thanks Prof. W. Thiel for continuous support.
References and Notes
(
(
1) B o¨ nnemann, H.; Richards, R. M. Eur. J. Inorg. Chem. 2001, 2460.
2) B o¨ nnemann, H.; Wald o¨ fner, N.; Haubold, H.-G.; Vad, T. Chem.
Mater. 2002, 14, 1115-1120. Vad, T.; Haubold, H.-G.; Wald o¨ fner N.;
B o¨ nnemann, H. J. Appl. Crystallogr. 2002, 35, 459-470.
(3) Leisner, T.; Rosche, C.; Wolf, S.; Granzer, F.; W o¨ ste, L. Surf. ReV.
Lett. 1996, 3, 1105.
4) ComprehensiVe Organometallic Chemistry; Stone, F. G. A., Wilkin-
son, G., Eds.; Pergamon: Oxford, U.K., 1982.
5) Angermund, K.; B u¨ hl, M.; Dinjus, E.; Endruschat, U.; Gassner,
(
(
F.; Haubold, H.-G.; Hormes, J.; K o¨ hl, G.; Mauschick, F. T.; Modrow, H.;
M o¨ rtel, R.; Mynott, R.; Tesche, B.; Vad, T.; Wald o¨ fner, N.; B o¨ nnemann,
H. Angew. Chem. 2002, 114, 4213-4216; Angew. Chem., Int. Ed. 2002,
4
1, 4041-4044.
Figure 10. Time dependence of the mean particle radius 〈R〉 (top)
and mass fraction mparticle/mtotal of Pt atoms transformed into particles
(
6) Bucher S.; Hormes, J.; Modrow, H.; Brinkmann, R.; Wald o¨ fner,
N.; B o¨ nnemann, H.; Beuermann, L.; Krischok, S.; Maus-Friedrichs, W.;
Kempter, V. Surf. Sci. 2002, 497, 321-332.
(bottom).
(7) Marcos, E. S.; Gil, M.; Martinez, J. M.; Muonz-Pa e´ z, A. M.;
Marcos, A. S. ReV. Sci. Instrum. 1994, 65, 2153-2154.
however, the mean size of the particles remains the same, within
(8) Becke, A. D. Phys. ReV. A 1988, 38, 3098-3100.
experimental error, and the width of their size distribution σ )
(
9) (a) Perdew, J. P. Phys. ReV. B 1986, 33, 8822-8824. (b) Perdew,
0
.17 ( 0.1 is also unchanged. A further sample showed that at
J. P. Phys. ReV. B 1986, 34, 7406.
the end of the reaction the entire Pt content is present, within
an experimental error of about 10%, in the form of particles.
The initial value of mparticle/mtotal of 0.206 arises from nucleation
that had occurred before the measurements were started.
Assuming an idealized icosahedral model, Pt particles of 1.2
nm size are composed of 53 atoms. This number is the second
in the row of the “magic numbers” of atoms, 13, 55, 147, ...,
which form energetically favorable icosahedra with complete
(10) Andrae, D.; H a¨ ussermann, U.; Dolg, M.; Stoll, H.; Preuss, H. Theor.
Chim. Acta 1990, 77, 123-141.
(
11) (a) Hehre, W. J.; Ditchfield, R.; Pople, J. A. J. Chem. Phys. 1972,
5
6, 2257-2261. (b) Hariharan, P. C.; Pople, J. A. Theor. Chim. Acta 1973,
28, 213-222.
(12) Cheeseman, J. R.; Trucks, G. W.; Keith, T. A.; Frisch, M. J. J.
Chem. Phys. 1996, 104, 5497-5509.
(13) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5642.
(14) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785-789.
15) Kutzelnigg, W.; Fleischer, U.; Schindler, M. NMR Basic Principles
and Progress; Springer-Verlag: Berlin, 1990; Vol. 23, pp 165-262.
(
47
outer shells of atoms. The experimental findings indicate that
when the particle size reaches 53 atoms to attain a closed outer
shell of metal atoms, the surrounding protective shell effectively
stops further nucleation. This is the basic requirement for the
synthesis of monodisperse particles.
(
16) This is true in particular for the δ(metal) values; for the ligand
shifts, other DFT variants would also be appropriate. See, for instance: (a)
B u¨ hl, M.; Kaupp, M.; Malkin, V. G.; Malkina, O. L. J. Comput. Chem.
1
999, 20, 91-105. For an application of this particular DFT combination
to ligand shifts, see: (b) B u¨ hl, M.; Sassmannshausen, J. J. Chem. Soc.,
Dalton Trans. 2001, 79-84.
The amount of platinum converted into particles, x ) (mparticle/
mtotal - 0.206)/(1 - 0.206), follows the exponential time
dependence x ) 1 - exp(-t/t0) shown as the solid line in Figure
(17) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.;
Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A.
D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi,
M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.;
Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick,
D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.;
Ortiz, J. V.; Baboul, A. G.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz,
P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-
Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe,
M.; Gill, P. M. W.; Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J.
L.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian 98, revision
A.9; Gaussian, Inc.: Pittsburgh, PA, 1998.
1
0. The rate of nucleation into particles dx/dt ≈ [1 - x(t)] is
linearly proportional to the number of precursor molecules in
the solution, [1 - x(t)]. The rate-controlling step for the
nucleation is the decomposition of the thermally unstable
binuclear precursor molecules and is not the subsequent
diffusion-controlled agglomeration process of the single zero-
valent Pt atoms into the particles. Fifty percent of the binuclear
platinum complex has decomposed at room temperature after a
(18) Vaara, J.; Malkina, O. L.; Stoll, H.; Malkin, V. G.; Kaupp, M. J.
Chem. Phys. 2001, 114, 61-71.
half time t0.5 ) t0 ln 2 of 52 h.
x
(19) Salahub, D. R.; Fournier, R.; Mlynarski, P.; Papai, I.; St-Amant,
A.; Ushio, J. deMon program. In Density Functional Methods in Chemistry;
Labanowski, J., Andzelm, J., Eds.; Springer: New York, 1991. St-Amant,
A.; Salahub, D. R. Chem. Phys. Lett 1990, 169, 387.
4
. Conclusions
On the basis of these experiments, we conclude that com-
(20) Malkin, V. G.; Malkina, O. L.; Eriksson, L. A.; Salahub, D. R. In
Modern Density Functional Theory: A Tool for Chemistry; Seminario, J.
M., Politzer, P., Eds.; Elsevier: Amsterdam, 1995; Vol. 2.
plexes of type 3 formed during the process of reductive
stabilization are common precursors for trapping reactions on
one hand, giving organometallic complexes such as 5, orsin
(21) Perdew, J. P. Phys. ReV. B 1986, 33, 8822. Perdew J. P.; Wang, Y.
Phys. ReV. B 1986, 33, 8800.

Wet Chemical Synthesis of Nanoscopic Pt Colloids
J. Phys. Chem. B, Vol. 107, No. 30, 2003 7515
(
22) Godbout, N.; Salahub, D. R.; Andzelm, J.; Wimmer, E. Can. J.
Chem. 1992, 70, 560.
23) Malkina, O. L.; Schimmelpfennig, B.; Kaupp, M.; Hess, B. A.;
Chandra, P.; Wahlgren, U.; Malkin, V. G. Chem. Phys. Lett. 1998, 296,
3-104.
24) (a) Bader, R. W. F. Atoms In Molecules. A Quantum Theory;
Clarendon Press: Oxford, U.K., 1990. (b) Bader, R. W. F. Chem. ReV.
(35) This degree of agreement is acceptable, given that a model complex
in the gas phase is compared to a bulkier system in the solid. For an
assessment of DFT-computed geometries and energies, see, for example:
Koch, W.; Holthausen, M. A Chemist’s Guide to Density Functional Theory;
Wiley-VCH: Weinheim, Germany, 2000.
(36) The Pt-C(bridge) and Pt-C(terminal) distances (computed values
2.153 and 2.093 Å, respectively) can probably not be resolved on the EXAFS
time scale.
(37) See, for instance: Kaupp, M.; Malkin, V. G.; Malkina, O. L.;
Salahub, D. R. Chem. Phys. Lett. 1995, 235, 382-388.
(38) Becke, A. D.; Edgecombe, K. J. Chem. Phys. 1990, 92, 5397-
5403.
(39) Wiberg, K. Tetrahedron 1968, 24, 1083-1096.
(40) Boys, S. F.; Bernardi, F. Mol. Phys. 1970, 19, 553-566.
(41) Haubold, H.-G.; Vad, T.; Wald o¨ fner, N.; B o¨ nnemann, H. J. Appl.
Cryst., in press.
(
9
(
1
991, 91, 893-928.
(
25) Popelier, P. L. A. Comput. Phys. Commun. 1996, 93, 212-240.
(26) Haubold H.-G.; Gruenhagen K.; Wagener, M.; Jungbluth, H.; Heer,
H.; Pfeil, A.; Rongen, H.; Brandenberg, G.; Moeller, R.; Matzerath, J.; Hiller,
P.; Halling, H. JUSIFAsA new user-dedicated ASAXS beamline for
materials science. ReV. Sci. Instrum. 1989, 60, 1943.
2
195
1
(
27) NMR data, 300 K, [D8]-toluene: δH ) 0.51 [24H, J( Pt, H) )
4 195 1
(
-
35.7 Hz, J( Pt, H) ) (5.7 Hz, Pt-CH3]; -0.27 [6H, Al-CH3]. δC )
1 195 13 3 195 13
0.1 [ J( Pt, C) ) (364 Hz, J( Pt, C) ) (16 Hz, Pt-CH3].
28) Abraham, R. J. Analysis of High-Resolution NMR Spectra; Elsevi-
(
(42) Cromer, D. T.; Libermann, D. A. Acta Crystallogr. 1981, A37, 267-
268.
er: Amsterdam, 1971.
29) James, F. Reference Manual Version 94.1; CERN program library
(
(43) Haubold, H.-G.; Gebhardt, R.; Buth, G.; Goerigk, G. In Resonant
Anomalous X-Ray Scattering; Materlik, G., Sparks, C. J., Fischer, K., Eds.;
Elsevier Science: Amsterdam, 1994; pp 295-304.
(44) Haubold, H.-G.; Wang, X. H. Nucl. Instrum. Methods Phys. Res.,
Sect. B 1995, 97, 50-54.
Long Write-up D506; European Organization for Nuclear Research
CERN: Geneva, 1998.
(
30) Katoh, M.; Miki, K.; Kai, Y.; Tanaka, N.; Kasai, N. Bull. Chem.
Soc. Jpn. 1981, 54, 611.
31) Ankudinov, A. L.; Ravel, B.; Rehr, J. J.; Conradson, S. D. Phys.
ReV. B 1998, 58, 7565.
32) Modrow, H.; Bucher, S.; Ankudinov, A.; Rehr, J. J. Phys. ReV. B
003, 035123.
33) Pantelouris, A.; Modrow, H.; Pantelouris, M.; Hormes, J.; Reinen,
D. Chem. Phys., in press.
34) Golden, J. T.; Peterson, T. H.; Holland, P. L.; Bergman, R. G.;
Andersen, R. A. J. Am. Chem. Soc 1998, 120, 223.
(
(45) Haubold, H.-G.; Wang, X. H.; Jungbluth, H.; Goerigk, G.; Schilling,
W. J. Mol. Struct. 1996, 383, 283-289.
(
(46) Haubold, H.-G.; Hiller, P.; Jungbluth, H.; Vad, T. Jpn. J. Appl.
Phys. 1999, 38 (Suppl. 38-1), 36-39.
2
(
(47) Echt, O.; Sattler, K.; Recknagel, E. Phys. ReV. Lett. 1981, 47,
1121.
(
(48) Modrow, H.; Bucher, S.; Hormes, J.; Brinkmann, R.; B o¨ nnemann,
H. J. Phys. Chem. B 2003, 107, 3684.