Structure and properties of the dendron-encapsulated polyoxometalate (C52H60NO12)12[(Mn (H2O))3(SbW9O33)2], a first generation dendrizyme

Article abstract of DOI:10.1021/ja017613b

Combining analytical and theoretical methods, we present a detailed study of a heteropolytungstate cluster encapsulated in a shell of dendritically branching surfactants, namely (C₅₂H₆₀NO₁₂)₁₂[(Mn- (H₂

Full text of DOI:10.1021/ja017613b

Published on Web 08/13/2002
Structure and Properties of the Dendron-Encapsulated
Polyoxometalate (C52H60NO12)12[(Mn(H2O))3(SbW9O33)2], a First
Generation Dendrizyme
Dirk Volkmer,*,† Bjo¨rn Bredenko¨tter,† Jo¨rg Tellenbro¨ker,† Paul Ko¨gerler,‡
Dirk G. Kurth,*,§ Pit Lehmann,§ Heimo Schnablegger, Dietmar Schwahn,|
Markus Piepenbrink,# and Bernt Krebs*,#
Contribution from the Department of Inorganic Chemistry 1, UniVersity of Bielefeld,
P.O. Box 100 131, D-33501 Bielefeld, Germany, Ames Laboratory and Department of Physics
and Astronomy, Iowa State UniVersity, Ames, Iowa 50011-3160, Max Planck Institute of
Colloids and Interfaces, D-14424 Potsdam, Germany, Institute for Physical Chemistry,
UniVersity of Hamburg, D-20146 Hamburg, Germany, Institut fu¨r Festko¨rperforschung (IFF),
Forschungszentrum Ju¨lich GmbH, D-52425 Ju¨lich, Germany, and Institute of Inorganic
Chemistry, UniVersity of Mu¨nster, D-48149 Mu¨nster, Germany
Received November 26, 2001. Revised Manuscript Received May 21, 2002
Abstract: Combining analytical and theoretical methods, we present a detailed study of a heteropolytung-
state cluster encapsulated in a shell of dendritically branching surfactants, namely (C52H60NO12)12[(Mn-
(H2O))3(SbW9O33)2], 3. This novel surfactant-encapsulated cluster (SEC) self-assembles spontaneously
from polyoxometalate-containing solutions treated with a stoichiometric amount of dendrons. Compound 3
exhibits a discrete supramolecular architecture in which a single polyoxometalate anion resides in a compact
shell of dendrons. Our approach attempts to combine the catalytic activity of polyoxometalates with the
steric properties of tailored dendritic surfactants into size-selective catalytic systems. The structural
characterization of the SEC is based on analytical ultracentrifugation (AUC) and small-angle neutron
scattering (SANS). The packing arrangement of dendrons at the cluster surface is gleaned from molecular
dynamics (MD) simulations, which suggests a highly porous shell structure due to the dynamic formation
of internal clefts and cavities. From analysis of the MD trajectory of 3, a theoretical neutron-scattering
function is derived that is in good agreement with experimental SANS data. Force field parameters used
in MD simulations are partially derived from a quantum mechanical geometry optimization of
[(Zn(H2O))3(SbW9O33)2]12-, 2b, at the density functional theory (DFT) level. DFT calculations are corroborated
by X-ray structure analysis of Na6K6[(Zn(H2O))3(SbW9O33)2]‚23H2O, which is isostructural with the catalytically
active Mn derivative 2a. The combined use of theoretical and analytical methods aims at rapidly prototyping
smart catalysts (“dendrizymes”), which are structurally related to naturally occurring metalloproteins.
Introduction
major reaction modes of POM-catalyzed oxidation processes
may be distinguished, depending on the chemical nature of the
Polyoxometalates (POMs) constitute a structurally diverse
class of discrete, negatively charged early transition metal oxide
clusters.1 Those POMs containing transition metal ions in their
highest possible oxidation states have found widespread use as
oxidation catalysts. According to Neumann and Hill2 several
oxidant, which often is composed of molecular oxygen, per-
oxygen compounds (e.g. H2O2, or tert-butylhydroperoxide), or
reactive transition metal-oxo species. In the latter case, “sand-
wich”-type transition metal substituted POMs have recently
attracted much attention, owing to their enhanced solvolytic
stability, their high turnover numbers in catalytic processes, and
their ability to catalyze alkane oxidation3 as well as alkene
epoxidation4 with high stereo- or regioselectivity.
Although many mechanistic details have yet to be elucidated,
a conceptual drawback of homogeneous catalytic processes
involving POMs is the lack of methods to direct the substrate
toward the catalytically active center. In contrast to many
* To whom correspondence should be addressed. D.V.: Phone, +49
(0)521 106 6142; fax, +49 (0)521 106 6003; e-mail, dirk.volkmer@uni-
bielefeld.de. D.G.K.: Phone, +49 (0)331 567 9211; fax, +49 (0)331 567
9202; e-mail, kurth@mpikg-golm.mpg.de. B.K.: Phone, +49 (0)251 83
33131; fax, +49 (0)251 83 38366; e-mail, krebs@uni-muenster.de.
† University of Bielefeld.
‡ Iowa State University.
§ Max Planck Institute of Colloids and Interfaces.
University of Hamburg.
| Institut fu¨r Festko¨rperforschung.
# University of Mu¨nster.
(3) (a) Neumann, R.; Khenkin, A. M. Inorg. Chem. 1995, 34, 5759. (b)
Neumann, R.; Dahan, M. Nature (London) 1997, 388, 353.
(4) (a) Khenkin, A. M.; Hill, C. L. MendeleeV Commun. 1993, 140. (b)
Neumann, R.; Gara, M. J. Am. Chem. Soc. 1994, 116, 5509. (c) Neumann,
R.; Gara, M. J. Am. Chem. Soc. 1995, 117, 5066. (d) Zhang, X.; Sasaki,
K.; Hill, C. L. J. Am. Chem. Soc. 1996, 118, 4809.
(1) (a) Pope, M. T.; Mu¨ller, A. Angew. Chem., Int. Ed. Engl. 1991, 30, 34. (b)
Pope, M. T., Mu¨ller, A., Eds.; Polyoxometalates: From platonic solids to
antiretroViral actiVity; Kluwer: Dordrecht, 1994.
(2) (a) Hill, C. L.; Prosser-McCarther, C. M. Coord. Chem. ReV. 1995, 143,
407. (b) Neumann, R. Progr. Inorg. Chem. 1998, 47, 317.
9
10.1021/ja017613b CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. 2002, 124, 10489-10496
10489

A R T I C L E S
Volkmer et al.
catalytic systems composed of organometallic or coordination
compounds, in which the access of substrates can be fine-tuned
by means of appropriate ligand design, likewise strategies have
not been successful in POM chemistry. Because of the chemical
inertness of the M(VI)-O bonds (M ) W or Mo), the covalent
attachment of organic ligands to the metal-oxygen framework
most often leads to product mixtures that display completely
rearranged cluster cores with concomitant loss of their original
catalytic activity.5
We recently described novel strategies to modify POMs while
retaining their structural integrity and their intrinsic properties.
One procedure employed rests on the exchange of the POM
countercations with suitable surfactants, leading to so-called
“surfactant-encapsulated clusters” (SECs).6 Surfactant encap-
sulation is a versatile, highly economic procedure to alter the
surface chemical properties of various POM compounds. The
anionic cluster is nested within a shell of surfactants, thus
leading to discrete, electrostatically neutral, hydrophobic core-
shell particles, the structures and properties of which we have
studied in great detail.7 An alternative strategy is based on the
exchange of the POM countercations with cationic macromol-
ecules, leading to a novel type of POM/polyelectrolyte hybrid
material.8
Figure 1. (a) Ball-and-stick and (b) polyhedral representation of the
heteropolytungstate 2b in the crystal structure of Na6K6[(Zn(H2O))3-
(SbW9O33)2]‚23H2O. (Counterions and water molecules are omitted for
clarity.)
Scheme 1
Here, we report on the structure and properties of (C52H60-
NO12)12[(Mn(H2O))3(SbW9O33)2], 3, which is composed of a
sandwich-type transition metal-substituted heteropolytungstate,
2a, that is encapsulated within a shell of novel, dendritically
branching amphiphiles, 1. Compound 2a was chosen because
of its excellent performance in the homogeneous catalytic
epoxidation of dienes.9
A central challenge in the characterization of complex soft
materials lacking crystalline order is to establish quantitative
structure-property relationships at a sufficiently accurate level.
Therefore, we apply a combination of small angle neutron
scattering and molecular dynamics simulations to reveal struc-
tural details of 3 at the nanoscopic scale.
presence of both components in the self-assembled SEC.
Although the characteristic UV-vis absorption band of 2a
(λmax ) 387 nm) remains unchanged upon encapsulation, some
of its characteristic IR-bands are slightly shifted due to
encapsulation. The UV-vis absorption bands of dendron 1 can
be found at 283 and 243 nm, whereas its characteristic
IR-frequencies occur at 3091, 3003, 2938, and 2839 cm-1 and
in the fingerprint region. These measurements confirm that 3
is composed of both the POM 2a and dendron 1. They also
demonstrate that the cluster anion remains intact and does not
decompose upon encapsulation. The hydrophobic nature of 3
is demonstrated by virtue of its solubility: the SEC easily
dissolves in organic solvents, such as CHCl3, toluene, or THF,
but is completely insoluble in water.
Results and Discussion
In the following, we will describe first the preparation and
characterization of SEC 3, then the isostructural POM 2b and
its crystal structure (Figure 1), the coordinates of which are used
in the following MD simulations, and finally, the characteriza-
tion of 3 by SANS.
Characterization of SEC 3. The dendron-encapsulated
cluster 3 was prepared according to previously published
procedures.7 The heteropolytungstate 2a was dissolved in water,
then the aqueous solution was treated with a trichloromethane
solution containing dendron 1‚Br (Scheme 1). This resulted in
spontaneous and complete transfer of 3 into the organic phase,
which is easily seen by the color change of the two phases.
Spectroscopic investigations (UV-vis and IR) showed the
Elemental analysis of 3 indicated a complete exchange of
alkali metal counterions. The POM anion is encapsulated in a
shell of ∼12 dendrons, therefore yielding a molar mass of
∼15778 g/mol for the supramolecular entity 3.10
The hydrodynamic radius of 3 was determined by analytical
ultracentrifugation (AUC). From sedimentation velocity experi-
ments, the sedimentation coefficient was calculated, which is
s ) 6.34 ( 2.9 s. The diameter of 3, as calculated from the
(10) The molar mass presented here corresponds to the idealized stoichiometric
ratio of 12:1 (dendron:POM). We are currently unable to quantify the exact
(minor) amount of “defect” SECs corresponding to a substoichiometric ratio
(11:1 or less). The 1H NMR spectrum of the dendron-encapsulated cluster
(5) Gouzerh, P.; Proust, A. Chem. ReV. 1998, 98, 77.
(6) Kurth, D. G.; Lehmann, P.; Volkmer, D.; Co¨lfen, H.; Mu¨ller, A.; Du Chesne,
A. Chem. Eur. J. 2000, 6, 385.
(7) (a) Volkmer, D.; Du Chesne, A.; Kurth, D. G.; Schnablegger, H.; Lehmann,
P.; Koop, M. J.; Mu¨ller, A. J. Am. Chem. Soc. 2000, 122, 1995. (b) Kurth,
D. G.; Lehmann, P.; Volkmer, D.; Mu¨ller, A.; Schwahn, D. J. Chem. Soc.,
Dalton Trans. 2000, 3989.
(8) (a) Caruso, F.; Kurth, D. G.; Volkmer, D.; Koop, M. J.; Mu¨ller, A. Langmuir
1998, 14, 3462. (b) Kurth, D. G.; Volkmer, D.; Ruttorf, M.; Richter, B.;
Mu¨ller, A. Chem. Mater. 2000, 12, 2829.
(9) Bo¨sing, M.; No¨h, A.; Loose, I.; Krebs, B. J. Am. Chem. Soc. 1998, 120,
7252.
(C52H60NO12)12[(Zn(H2O))3(SbW9O33)2] displays broad signals that are
significantly shifted against the (sharp) signals in the spectrum of free
dendron (1‚Br). As a result of the fact that signals of free 1‚Br are missing
in the 1H NMR spectrum of this SEC, we suggest a purity of at least 95%.
Photometric measurements at λmax ) 387 nm, furthermore, prove that the
(highly colorized) POM 2a is completely transferred from the aqueous
solution into the organic phase. We have, however, not yet been able to
establish the appropriate mass spectroscopic techniques that would have
shown a mass peak corresponding to nonfragmented SECs.
9
10490 J. AM. CHEM. SOC. VOL. 124, NO. 35, 2002

Dendron-Encapsulated Polyoxometalate
A R T I C L E S
Table 1. Crystal Structure Data of
Na6K6[(Zn(H2O))3(SbW9O33)2]‚23H2O
sedimentation coefficient, is D ) 2.7 ( 0.5 nm. The ap-
proximate dimensions of 2a are 1.5 × 1.1 × 1.1 nm, whereas
the length of a single dendron 1 amounts to 1.3 nm, assuming
a fully extended conformation. The diameter of SEC 3, therefore,
might be as large as 3.8 nm. The actual diameter, however, is
considerably smaller, owing to the propensity of the dendron
shell to adapt a dense packing. Since we could not observe
decomposition or aggregation products by AUC, the experi-
mentally determined hydrodynamic radius corresponds to single
(monodisperse) SECs. In summary, we conclude that SEC 3
consists of a single POM cluster residing within a shell of 12
dendrons.
Na6K [2b]‚23H O
6
2
formula
H52Na6K6O92Zn3Sb2W18
5645.84
0.24 × 0.4 × 0.32
colorless needles
hexagonal
P63/mmc (no. 194)
15.265(2)
24.461(5)
4936.3(14)
2
3.763
22.511
formula weight (M), g mol-1
crystal size, mm3
crystal color/habit
crystal system
space group
a, Å
c, Å
V, Å3
Z
F
calc, g cm-3
µ, mm-1
F(000)
4880
Crystal Structure of 2b. The open-shell electronic structure
of the three µ-oxo-bridged high-spin Mn(II) centers gives rise
to severe obstacles in calculating the electronic ground-state
properties of 2a. Since the primary goal here is to derive atomic
partial charges being used for MD simulations, we restricted
DFT quantum mechanics calculations on the Zn-substituted
derivative 2b, assuming (a) that the Zn-containing POM is
isostructural with 2a and (b) that the ground-state electron
density distributions of 2a and 2b, respectively, are (almost)
identical. To support this hypothesis, single crystals of 2b were
prepared, and the single-crystal X-ray structure of this novel
POM compound was solved.
T, K
213(2)
2θ range
9° < 2θ < 52°
reflections collected
independent reflections
GOF on F2
R (>2σ (I))
36 960
1832 [R(int) ) 0.1538]
1.227
R1a ) 0.0488
wR2 ) 0.1278
R1a ) 0.0512
wR2 ) 0.1294
3.06
R (all data)
(∆/F)max, e-/Å3
(∆/F)max, e-/Å3
-4.71
a R1 ) Σ||Fo| - |Fc||/Σ|Fo|. b wR2 ) |Σw(F0 - Fc )2/Σ(F0 )]1/2. w1 )
2
2
4
2
1/[σ2(Fo ) + (0.0620P)2 +160.86P].
Table 2. Experimental and Calculated Bond Lengths (Å) and
Angles (deg) of 2b in Na6K6[2b]‚23 H2O
A ball-and-stick as well as a polyhedral representation of the
Zn-substituted heteropolytungstate [(Zn(H2O))3(SbW9O33)2]12-
(2b), derived from X-ray structure analysis, is shown in Figure
1. It was found that compound 2b is isostructural with the Mn-
containing heteropolytungstate (2a) reported earlier.9 Compound
2b contains two identical R-B-[SbW9O33]9- fragments (≡{Sb-
W9}) of the R-Keggin structure, consisting of three corner-
sharing W3O13 groups, and is closely related to similar sandwich-
type POMs containing trivacant Keggin subunits.11 The
preparation procedure employed the lacunary {SbW9} fragment
as precursor, from which the sandwich-type heteropolytungstate
2b can be generated at pH 7.5. The X-ray structure analysis of
2b (for details, see Table 1) revealed a POM composed of two
{SbW9} subunits connected to each other by three zinc atoms
in a belt. The zinc centers are coordinated by four oxygen atoms
with an average Zn-OW-Zn bond length of 2.03 Å and one water
molecule (Zn-O(W1) bond length, 2.01 Å) to complete a
square-pyramidal coordination sphere. Positioned in the center
of each {SbW9} subunit, the SbIII heteroatom shows a distorted
ψ-tetrahedral coordination with a Sb-OW3-Sb bond length of
1.98 Å. Further selected bond lengths and angles are presented
in Table 2.
exptl
calc
W(1)-O(1), W(2)-O(5)
W(1)-O(2)
1.72(1)
1.92(1)
1.89(1)
2.40(1)
2.03(1)
1.79-1.80
1.97
W(1)-O(3)
1.92-1.93
W(1)-O(4)
W(2)-O(3)
2.40
2.08
W(2)-O(6), W(2)-O(7)
W(2)-O(4)
1.93(1)-1.96(1)
1.97-1.99
2.27(1)
2.32
W(2)-O(8)
1.79(1)
1.84
Sb(1)-O(4)
1.98(1)
2.05
Zn(1)-O(8)
2.03(1)
2.01-2.03
Zn(1)-O(W1)
O-W-Ocis
2.01(1)
2.38
73(1)-105(1)
159(1)-170(1)
91(1)
73-105
159-168
90
O-W-Otrans
O-Sb-O
O-Zn-Ocis
85(1)-103(1)
86-92
O-Zn-Otrans
154(1)
166
To model nonbonding electrostatic interactions between the
cationic dendrons 1 and the negatively charged surface of 2a,
DFT calculations were performed, and different computational
schemes were employed to derive partial atomic charges from
the resulting electron density distribution. The primary criterion
for the selected basis sets was to reproduce the experimental
geometry of the [(Zn(H2O))3(SbW9O33)2]12- anion 2b. The
geometry of the isolated anion 2b was optimized starting from
a C3h-symmetrical model structure. The symmetry elements are
sustained during optimization. The theoretically calculated
structure matches well with the experimental structure as derived
from single-crystal X-ray structure analysis (Table 2). The most
evident geometrical deviation concerned the Zn-O(W1) bond
length (calculated, 2.38; found, 2.01 Å); the values of all other
bond lengths and angles and the overall dimensions (intra-cluster
distances) of the anion differed less than 5%, mostly around
2%. Mulliken as well as Lo¨wdin atomic point charges for 2a
Molecular Dynamics Simulations. MD simulations were
performed in order to complement structural investigations on
dendron-encapsulated clusters by means of SANS and to get
further insights into the dynamic structure of the dendron shell.
Since the supramolecular ensemble consists of some 1600 atoms,
MD simulations at the ab initio level of theory are currently
not applicable. Thus, we followed a molecular mechanics
approach, using a modified Amber 94 force field12 to define
the intra- and intermolecular atomic interaction potentials.
(11) (a) Bo¨sing, M.; Loose, I.; Pohlmann, H.; Krebs, B. Chem. Eur. J. 1997, 3,
1232. (b) Mialane, P.; Marrot, J.; Rivie`re, E.; Nebout, J.; Herve´, G. Inorg.
Chem. 2001, 40, 44. (c) Loose, I.; Droste, E.; Bo¨sing, M.; Pohlmann, H.;
Dickman, M. H.; Rosu, C.; Pope, M. T.; Krebs, B. Inorg. Chem. 1999, 38,
2688.
(12) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K. M., Jr.;
Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.; Kollmann,
P. A. J. Am. Chem. Soc. 1995, 117, 5179.
9
J. AM. CHEM. SOC. VOL. 124, NO. 35, 2002 10491

A R T I C L E S
Volkmer et al.
Table 4. Scattering Lengthsa Used in SANS Investigations
Table 3. Calculated Partial Charges for the 2b Anion
atom
Mulliken partial charges
Lo¨wdin partial charges
element
b [10-13 cm]
c
Zn(1)
O(W1)
O(8)
Sb(1)
O(4)
W(1/2)
O(1/5)
O(2/3/4)
0.928
0.783
-0.568
-0.310 to -0.315
0.967
H
-3.739
6.671
6.6460
9.36
-0.566
D
-0.592
C
1.172
N
-0.775
-0.385
O
5.803
9.577
-3.73
5.57
1.383-1.435
-0.591 to -0.621
-0.607 to -0.683
0.250-0.361
-0.344 to -0.374
-0.304 to -0.372
Cl
Mn
Sb
W
4.86
a Neutron News 1992, 3, 29.
Table 5. Results from SANS
parameter
result
dΣ/dΩ
Rg
(Q ) 0) 0.016 cm-1
(1.44 ( 0.003)nm
0.0295 cm-1
6.2 × 1017 cm-3
1.74%
dΣ/dΩinc
N
ΦSANS
Φnominal
2.25%
ment, which should exclude access of medium to large-size
molecules to the cluster surface. The elemental analysis of 3,
on the other hand, shows that a certain number of water
molecules (≈15) reside within the SEC structure. It is not yet
clear, however, if these water molecules are distributed through-
out the dendron shell or if they are located close to the cluster
surface.
Small Angle Neutron Scattering of 3. To validate the
proposed structure model of 3, SANS investigations were carried
out. This method was particularly well suited to investigate such
a question, because the scattering contrast originates from the
deuterated solvent in which the SEC is dissolved. The scattering
pattern of 3 dissolved in a highly diluted CDCl3 solution is
depicted in Figure 6. Most of the scattering originates from
single SECs. The enhanced scattering observed at small Q-
values, however, indicates aggregation of a small fraction of
clusters.
The scattering function shown in Figure 6 (solid line) was
calculated by using the atomic coordinates as obtained from
the MD simulations and tabulated values (see Table 4) of the
atomic scattering lengths. Snapshots at 10 different evolution
times between 0 and 100 ps were used for the calculations, and
the resulting scattering functions were averaged thereafter. The
parameters obtained from scattering were collected in Table 5.
The size of the particles was derived from the radius of gyration
Rg ) 1.44 nm (eq 3) which, assuming a massive spherical
Figure 2. Numbering scheme of atoms in the asymmetric unit of the crystal
structure of Na6K6[2b]‚23H2O (orthographic projection).
were obtained from MO population analysis of POM 2b and
are listed in Table 3.
For MD simulations of SEC 3, the trajectory of dendrons 1
at the surface of 2a was calculated.
Figure 3 depicts a snapshot from a 100-ps MD simulation
run at T ) 300 K showing the typical supramolecular arrange-
ment of dendrons pointing with their charged headgroups toward
the cluster surface. Within the time limit of the MD simulation,
no dissociative processes occurred, which would lead to an
incompletely encapsulated cluster anion. To gain further insights
into the packing density of the dendron shell, surface models
of the solvent-accessible surface (SAS) were calculated for probe
radii ranging from 0.1 to 10.0 Å.13 The final solvent-accessible
surface area ( ASAS) is plotted versus probe radius in Figure 4.
x
The deviation from the straight line, representing the SAS of a
perfectly smooth sphere, becomes readily apparent for small
probe radii. Similar calculations have been reported on car-
boxylic acid terminated PAMAM dendrimers of generation
numbers 1-4,14 in which dendrimers of generation 3 (and 4,
respectively) reveal a greater internal than external surface area.
These data were interpreted with the appearance of internal voids
and clefts, indicating a porous dendrimer structure that should
be readily accessible to guest molecules of the appropriate size.
Our results suggest a porous shell structure in the SEC
comparable to those of high-generation PAMAM dendrimers.
However, in the case of the SEC, the porosity occurs already
with first generation dendrons!15 Cavities and channels were
visualized with surface contour models of 3 (Figure 5). The
average pore size of these computer-generated structure models
would let us assume a rather dense dendron packing arrange-
particle, would yield a particle diameter of D ) 3.7 nm
16
x
according to the relationship D ) 2 5/3Rg.
The SANS particle diameter is in good agreement with the
corresponding values of AUC measurements (D ) 2.67 ( 0.54
nm), or MD simulations (D ≈ 3.4 nm), respectively.
Summary and Outlook
In conclusion, we have presented a detailed picture of a novel
surfactant-encapsulated polyoxometalate cluster, namely the
dendron-encapsulated heteropolytungstate (C52H60NO12)12[(Mn-
(H2O))3(SbW9O33)2], 3. Spontaneous self-assembly of the het-
eropolytungstate and the dendritic surfactant produces 3 at
(13) Average SAS values were calculated from 10 different configurations of a
100-ps trajectory at 10-ps time steps.
(14) Newkome, J. R.; Moorefield, C. N.; Vo¨gtle, F. Dendritic Molecules; VCH-
Wiley: 1996; p 28.
(15) It should be noted here that the total number () 48) of terminal
3,5-(dimethoxy)benzyloxy moieties in the self-assembled dendron-
encapsulated cluster is comparable to a covalently assembled Fre´chet-type
dendrimer of generation 4-5 () 32 or 64 terminal groups).
(16) Guinier, A. X-ray Diffraction; W. H. Freeman and Company; San Francisco
1963; 329.
9
10492 J. AM. CHEM. SOC. VOL. 124, NO. 35, 2002

Dendron-Encapsulated Polyoxometalate
A R T I C L E S
Figure 3. Snapshot taken from a molecular dynamics simulation of the dendron-encapsulated cluster 3. Ball-and-stick model with color codes representing
Mulliken atomic partial charges. The gray envelope displays the solvent accessible surface (SAS) calculated for a probe radius of 1.4 Å. (The foremost part
of the SAS has been cut-away for clarity.)
By encapsulating a catalytically active POM within a shell
of cationic dendrons, we wished to combine the catalytic
properties of POMs with the steric properties of the dendrons.17
On the basis of size-exclusion principles, the dendron shell may
thereby control access of substrates to the active site of the POM
core and, thus, enhance the catalytic selectivity of the SEC. The
propensity of high-generation dendrimers to form internal voids
and cavities is a well-documented feature.18 Following this
concept, dendron-encapsulated POM clusters (“dendrizymes”19)
bear a certain resemblance to natural metalloproteins.20 In
addition to many other studies, which are devoted to the isolation
of active sites in dendrimer matrixes,21 SECs appear particularly
rewarding, considering the unsurpassed number of catalytically
active POMs.22 Future investigations will focus on the catalytic
Figure 4. Plot of SAS values of 3. Interpolated mean (SAS)1/2 values for
different probe radii ranging from 0.1 to 10 Å are depicted by the solid
line. The dashed line represents the SAS function (SAS)1/2 ) 2π1/2 (r + rp)
(17) Catalytically active POM/dendrimer hybrid materials have been reported
for a hard sphere, where r is the radius of the sphere and rp is the probe
recently which consist of an organic, dendritically branched core and POM
clusters covalently attached to the dendrimer periphery. Zeng, H.; Newkome,
G. R.; Hill, C. L. Angew. Chem., Int. Ed. 2000, 39, 1771.
radius.
(18) Baars, M. W. P. L.; Meijer, E. W. Top. Curr. Chem. 2000, 210, 131.
(19) Brunner et al. have coined the term “dendrizymes” for metal complexes
consisting of a dendritically branched (optically active) ligand and a strongly
binding chelate core. (a) Brunner, H.; Altman, S. Chem. Ber. 1994, 127,
2285. (b) Brunner, H. J. Organomet. Chem. 1995, 500, 39.
(20) The apparent link between natural enzymes and catalytically active
dendrimers has been pointed out by several authors. Selected reviews
referring to this concept include (a) Fre´chet, J. M. J. Science 1994, 263,
1710. (b) Smith, D. K.; Diederich, F. Chem. Eur. J. 1998, 4, 1353. (c)
Cardona, C. M.; Mendoza, S.; Kaifer, A. E. Chem. Soc. ReV. 2000, 29, 37.
(d) Gorman, C. B.; Smith, J. C. Acc. Chem. Res. 2001, 34, 60. (e) Twyman,
L. J.; King, A. S. H.; Martin, I. K. Chem. Soc. ReV. 2002, 31, 69.
(21) Hecht, S.; Fre´chet, J. M. J. Angew. Chem., Int. Ed. 2001, 40, 74.
almost quantitative yield. This result shows that surfactant-
encapsulation is a convenient and reliable approach to alter the
surface properties of POMs and is widely applicable to different
POM clusters and amphiphilic systems. To reveal details down
to the nanoscopic level, we employed a combination of
analytical (AUC, SANS, elemental analysis, UV-vis and IR
spectroscopy) and computational methods (DFT calculations,
MD simulations) in the structure analysis of SECs.
9
J. AM. CHEM. SOC. VOL. 124, NO. 35, 2002 10493

A R T I C L E S
Volkmer et al.
containing transition metal ions. Neglecting an atomistic
representation of the solvent represents another simplification
of the current structure model, which has been forced by the
expense of a (certainly more realistic) solvent box model
including some further 10 000 solvent molecules.24 As a result
of the lack of reliable experimental methods to adjust force field
parameters of complex self-assembling structures, such as 3,
we decided to restrict our computational efforts to the present
stage. However, at this level of analysis, our approach is readily
available to the bench chemist to develop and test prototypes
of artificial dendrizymes and related systems. The results of such
types of analysis may also become important in understanding
phenomena that occur on short-length scales, such as ion and
electron transport or structural relaxation processes in soft
materials.
Experimental Section
Preparative Procedures. Analytical TLC was performed on com-
mercial Macherey-Nagel plates POLYGRAM ALOX N/UV254. Col-
umns were packed with aluminum oxide (activated, neutral, 0.05-
0.20 mm). NMR spectra were recorded on a Bruker DRX 500 with
the solvent signal as standard. Mass spectra were obtained on an Esquire
3000 with electrospray ionization.
Figure 5. Ball-and-stick model of 3 displaying a SAS contour-plot projected
onto a cut plane running through the center of the SEC. (Cluster 2a appears
green in color). Color-coded SAS contour levels correspond to surface
probes with augmenting radii. (SAS probe radii ranging from 0.0 (red) to
1.4 Å (turquoise)). Blue regions that appear between the regions occupied
by the dendron shell indicate internal cavities and clefts.
IR spectra of Na6K6[(Zn(H2O))3(SbW9O33)2]‚23H2O (Na6K6[2a]‚
23H2O) were performed on a Perkin-Elmer 683 spectrometer in a range
4000-400 cm-1
.
Synthesis of Bis-{3,5-bis[3,5-(dimethoxy)benzyloxy]benzyl}-di-
methylammonium bromide (1‚Br). 3,5-Bis[3,5-(dimethoxy)benzyl-
oxy]benzyl bromide25 (2.00 g, 3.98 mmol) was dissolved in dry THF
(50 mL). To this solution was added sodium carbonate (106 mg, 1.00
mmol) and dimethylamine (500 µL, 1.00 mmol, 2 M solution in dry
THF). After stirring for 48 h at 20°C under argon, the insoluble
compounds were separated by filtration, and the solvent was removed
under vacuum. Upon column chromatography (MPLC, ALOX aceto-
nitrile f acetonitrile/methanol 10:1), a colorless oil was obtained in
1
94% yield (910 mg; 937 µmol). H NMR (CDCl3): δ 6.74 (d, 4, J )
1.9), 6.68 (t, 2, J ) 1.9), 6.55 (d, 8, J ) 2.2), 6.33 (t, 4, J ) 2.2), 5.00
(s, 8), 4.75 (s, 4), 3.75 (s, 24), 2.90 (s, 6); 13C NMR (CDCl3): δ 161.0,
159.9, 138.0, 129.0, 112.1, 105.3, 104.2, 99.9, 70.0, 68.0, 55.3, 48.7;
MS (ES): m/z ) 890 ([M - Br]+). Anal. Calcd for C52H60BrNO12
(970.94): C, 64.33; H, 6.23; N, 1.44. Found: C, 64.54; H, 6.23; N,
1.35.
Figure 6. SANS results and theoretical curve from MD simulations of 3.
Na11(NH4)[(Mn(H2O))3(SbW9O33)2]‚45H2O (Na11(NH4)[2a]‚45H2O).
The compound was synthesized as described previously.9
performance of these novel dendrizymes, for which we are
currently preparing a series of differently sized amphiphilic
dendrons.
Na6K6[(Zn(H2O))3(SbW9O33)2]‚23H2O (Na6K6[2b]‚23H2O). The
precursor compound Na9[SbW9O33]‚19.5H2O was synthesized as
described previously.11a Na9[SbW9O33]‚19.5H2O (6.0 g, 2.1 mmol) was
dissolved in water (30 mL), and ZnSO4‚7H2O (1.04 g, 3.6 mmol) was
added dropwise. The mixture was heated at 90 °C for 1 h. After the
reaction solution was cooled to room temperature, KCl solution
(12 mL, 1 M) was added. After 1 day, colorless crystals of Na6K6-
[(Zn(H2O))3(SbW9O33)2]‚23H2O (Na6K6[2b]‚23H2O) were obtained.
Yield: 4.4 g (0.78 mmol, 74%). Anal. Calcd. for H52Na6K6O92Zn3-
Sb2W18: Na, 2.44; K, 4.16; Zn, 3.47; Sb, 4.31; W, 58.3. Found: Na,
2.61; K, 4.08; Zn, 3.52; Sb, 4.25; W, 58.18. IR (KBr, cm-1): 3412
(vs) (ν(O-H)); 1615 (s) (δ(H-O-H)); 1115 (s); 948 (vs) (ν(W-Ot));
882 (vs) (ν(W-Oe)); 745 (vs) (ν(W-Oc)); 465 (s) (ν(Sb-O)) (t,
terminal; e, edge-sharing, c, corner-sharing).
Our results show that MD simulations are becoming an
indispensable tool to refine dynamic structure models of
supramolecular systems, such as 3, down to atomic resolution.
It should be pointed out that the computationally generated
structure models inherently bear the danger of misleading
interpretations. One critical point of the current MD simulations
concerns the use of DFT-derived Mulliken partial charges in
order to model electrostatic interactions. Although more power-
ful partition schemes have recently been developed to represent
the electrostatic properties of small organic molecules,23 these
methods still remain to be extended to include compounds
(C52H60NO12)12[(Mn(H2O))3(SbW9O33)2]‚15 H2O (3). Na11(NH4)-
[2a]‚45H2O (31.6 mg, 5.36‚10-6 mol) was dissolved in water (20 mL).
(22) (a) Hill, C. L.; Prosser-McCartha, C. M. Coord. Chem. ReV. 1995, 143,
407. (b) A series of 34 recent papers in a volume devoted to polyoxoanions
in catalysis: Hill, C. L. J. Mol. Catal. 1996, 114, 1-365. (c) Mizuno, N.;
Misono, M. J. Mol. Catal. 1994, 86, 319. (d) Okuhara, T.; Mizuno, N.;
Misono, M. AdV. Catal. 1996, 41, 113. (e) Kozhevnikov, I. V. Catal. ReV.-
Sci. Eng. 1995, 37, 311.
(24) Box models of such dimensions would require massive parallel computation
on a (very) huge computer cluster employing hand-optimized program
codes, which is clearly beyond the scope of the current investigations.
(25) Stewart, G. M.; Fox, M. A. J. Am. Chem. Soc. 1996, 118, 4354.
(23) Li, J.; Zhu, T.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. A 1998, 102,
1820.
9
10494 J. AM. CHEM. SOC. VOL. 124, NO. 35, 2002

Dendron-Encapsulated Polyoxometalate
A R T I C L E S
Twelve equiv of 1‚Br (6.43 × 10-5 mol ) 62.5 mg) dissolved in
trichloromethane (20 mL) were added, and the encapsulated cluster
was transferred into the organic phase by gentle shaking of the biphasic
solution. The organic phase was separated, evaporated, and dried for 3
days at 3 mbar (20 °C). Anal. Calcd. for (C52H60NO12)12[(Mn(H2O))3-
(SbW9O33)2]‚15H2O: C, 47.05; H, 4.59; N, 1.00; O, 22.72. Found: C,
47.45; H, 4.79; N, 1.06; O, 23.13. The molar weight of the SEC was
calculated according to the results of the elemental analysis to be 15 778
g/mol. Spectral data of SEC: UV-vis (CHCl3); λ [nm]: 387, 283,
243. IR (KBr, cm-1): 3442, 3091, 3003, 2959, 2938, 2839, 1600, 1460,
1431, 1376, 1347, 1323, 1300, 1205, 1156, 1055, 1019, 943, 869, 836,
775, 752, 713, 683, 514, 464, 438. The density of the SEC was
determined with a DMA 5000 density meter (solvent CHCl3) to be
F ) 1.82 g/cm3.
Analytical Ultracentrifugation. Analytical ultracentrifugation was
carried out with a Beckman Optima XL-I (Beckman Instruments, Palo
Alto, USA) analytical ultracentrifuge. The centrifuge was equipped with
an integrated scanning UV-vis absorption and Rayleigh interference
optical system. Sedimentation velocity experiments were performed
in toluene at a rotational speed of 60 000 rpm. The sedimentation
coefficient was determined to be s ) 6.34 ( 2.9 s. The calculated
diameter of the SEC is d ) 2.67 ( 0.54 nm.
From the movement of the sedimentation boundary, the sedimenta-
tion coefficient, s, was determined according to ln(rb/rm) ) sω2t, where
rb and rm are the radial positions of the boundary and the meniscus,
respectively, and ω is the radial velocity. A plot of ln(rb) versus time
gave a straight line of slope sω2. The diffusion coefficient, D, is
determined from the spreading of the sedimentation boundary. A plot
of time, t, versus [cp/(dc/dr)]2 gave a straight line of slope D/4π. Here,
cp is the concentration in the plateau region and dc/dr is the
concentration gradient at the boundary. The hydrodynamic particle
diameter, R, was determined according to d2 ) 18ηs/(Fp - Fs), were η
is the solvent viscosity and Fp and Fs are the densities of the particles
and solvent, respectively. The molecular weight was determined from
the Swedberg equation M ) sRT/D(1 - νjp), where νj is the experi-
mentally determined partial specific volume, and F, the solvent density.
The values of s and D were determined at different concentrations and
extrapolated to zero concentration.26
Crystal Structure Determination. Diffraction experiments were
performed on a STOE IPDS imaging plate system with Mo-KR
radiation (λ ) 0.71073 Å) and with corrections for absorption (using
DECAY/ABSCOR). The structures were solved by direct methods using
SHELXS 97 and refined with SHELXL 97 (on F2) by full-matrix least
squares. All metal atoms were refined anisotropically; the oxygen atoms
were refined isotropically. As usual in polyoxometalate chemistry, we
found little disorder in the range of counterions and water molecules.
Details of the crystal data collection, processing and structure analysis
and refinement are summarized in Table 1. Further information, such
as positional and thermal parameters, can be obtained from the cif-
files presented in the Supporting Information.
relevant bonding characteristics. In contrast, preliminary model inves-
tigations on smaller polyoxotungstate fragments (WO42-) showed that
extended basis sets (e.g., a triple-ú valence TZV(P) set, while operating
the aforementioned functionals) tend to overestimate the ionic W-O
bonding character, yielding unreasonably high Mulliken and Lo¨wdin
point charges. On the other hand, replacing the Perdew86 correlation
functional by more accurate but computationally more expensive
functionals, such as the Lee-Yang-Parr (LYP) correlation functional,
did not result in significant Mulliken and Lo¨wdin point charge
differences.31
Molecular Dynamics Simulations. MD simulations were performed
with the HyperChem program suite.32
Molecular models of dendron 1 were readily obtained from Amber94
force field calculations. With respect to the highly flexible nature of
the hydrophobic benzyl ether branches, the conformational space of 1
was scanned on the basis of the Metropolis search algorithm.33 The
conformational analysis yielded some 100 unique conformers, the
energetically most favorable of which was subsequently used in a single-
point DFT calculation, followed by population analysis.
Atomic coordinates and partial charges of 2a were obtained from a
DFT geometry optimization of a C3h-symmetric model structure, on
the basis of the crystallographic atomic positions of 2b. A customized
Amber94 force field was used to model atomic interaction potentials.
The atomic positions of 2a were fixed during simulation runs. An
energetically equilibrated starting configuration of 3 was obtained by
performing several high-temperature MD simulation runs (1200 K,
5-10 ps) that were interrupted by short cooling cycles (0 K, 1-2 ps).
The final MD simulation included a heating period (10 ps), allowing
the molecular ensemble to reach a final temperature of 300 K starting
from 0 K, followed by a 100 ps MD simulation run at constant (total)
energy. During simulation, the coordinate shifts of dendrons 1 were
recorded at 1-fs intervals.
Solvent Accessible Surface (SAS). SAS calculations were performed
with the program MSMS V2.53.34 SAS values were computed on the
basis of atomic coordinates of 3 that were extracted from a 100-ps
trajectory at constant (10-ps) time steps (T ) 300 K). Molecules were
represented as a set of overlapping spheres, each having the van der
Waals radius of its constituent atom. The SAS of a molecule is the
trajectory of the center of a spherical probe of radius rp (which
represents an imaginary solvent molecule) rolling over the van der
Waals surface of the molecule.35
Small-Angle Neutron Scattering (SANS). SANS experiments were
performed at the KWS1 diffractometer of the research center in Ju¨lich.36
The concentration of the sample was 41 mg/mL, which corresponds to
a volume fraction 0.0225. The measurements were performed with a
(31) For recent investigations on the effect of different methods, basis sets, and
functionals on the partial charges of a hypothetical WO2(OH)2 molecule,
see: Judd, D. A.; Nettles, J. H.; Nevins, N.; Snyder, J. H.; Liotta, D. C.;
Tang, J.; Ermolieff, J.; Schinazi, R. F.; Hill, C. L. J. Am. Chem. Soc. 2001,
123, 886. The results in this paper (see Table 1) also support our initial
finding that Hartree-Fock-type calculations overestimate the ionic bond
character of polymetalate species.
(32) Hyperchem Release 6.01; HyperCube Inc.: Gainesville, FL, 2000.
(33) Metropolis, N.; Rosenbluth, A. W.; Rosenbluth, M. N.; Teller, A. H.; Teller,
E. J. Chem. Phys. 1953, 21, 1087.
(34) Sanner, M. F.; Spehner, J.-C.; Olson, A. J. Biopolymers 1996, 38, 305.
(35) Tabulated van der Waals radii were taken from the website
at www.webelements.com. Values used in SAS calculations are as
follows: C, 1.70; H, 1.20; N, 1.55; O, 1.52 Å. For the other elements,
tabulated ionic radii were employed because of the lack of substantiate
van der Waals radii: Mn(II), high-spin, octahedral, 0.97 Å; Sb(III), 0.90
Å; W(VI), octahedral, 0.74 Å. The use of ionic radii is, strictly speaking,
incorrect in terms of the mathematical definition of the SAS; in practice,
however, the affected atoms are hidden in the POM framework and, thus,
do not contribute to the SAS. The most critical value is the (uncertain) van
der Waals radius of oxygen atoms belonging to the POM that was chosen
here to equal the well-established value for organic compounds. As a result
of the fact that the coordinates of the POM are fixed during MD simulation
runs, internal cavities originate from different packing arrangements of the
dendron shell, which consists of purely organic ligands. The (inadequate)
representation of the POM surface topology adds a constant factor to the
calculated SAS values, and the error thus introduced is negligible.
(36) A description of the SANS instrument is given in www.neutronscattering.de.
DFT Calculations. DFT investigations on the polyoxometalate
cluster (2b) and the amphiphilic dendrons (1) were performed using
the TURBOMOLE package27 employing the Becke88 exchange and
the VWN/Perdew86 and correlation functionals and split valence basis
sets incorporating polarization functions (SV(P)).28 Relativistic effective
core potentials (ECPs) were applied for Sb,29 and W.30 This chosen set
was found to accurately model the cluster anion’s geometry and the
(26) Ralston, G. Introduction to Analytical Ultracentrifugation; Beckman
Instruments: California, 1993.
(27) (a) Treutler, O.; Ahlrichs, R. J. Chem. Phys. 1995, 102, 346. (b) Eichhorn,
K.; Treutler, O.; O¨ hm, H.; Ha¨ser, M.; Ahlrichs, R. Theor. Chem. Acc. 1997,
97, 119.
(28) (a) Becke, A. D. Phys. ReV. A 1998, 38, 3098. (b) Perdew, J. P. Phys. ReV.
B 1986, 33, 8822.
(29) Bergner, A.; Dolg, M.; Ku¨chle, W.; Stoll, H.; Preuss, H. Mol. Phys. 1993,
80, 1431.
(30) Andrae, D.; Ha¨ussermann, U.; Dolg, M.; Stoll, H.; Preuss, H. Theor. Chim.
Acta 1990, 77, 123.
9
J. AM. CHEM. SOC. VOL. 124, NO. 35, 2002 10495

A R T I C L E S
Volkmer et al.
neutron wavelength of 0.7 nm, and the data were evaluated with a
standard procedure subtracting background and scattering from the
solvent. The data are given in absolute units as obtained from calibration
with a secondary standard. This allows the determination of the particle-
number density and the volume fraction from the scattering at Q ) 0
in eq 4. The total coherent scattering lengths of the molecule and of
the solvent molecules filling the same volume is given, respectively,
as ∑Nj)1 bi ) 2.867 × 10-10 cm and Nbhs ) 4.475 × 10-10 cm (Table 4).
The molar volume was evaluated from the molar weight M ) 15520
g/mol, and the corresponding density d ) 1.82 g/cm3, the molar weight,
and the density of the CDCl3 were MS ) 120.39 g/mol and dS ) 1.5004
g/cm3, respectively. From dΣ/dΩ(0) ) 0.016 cm-1, the number density
of n ) 6.2 × 1017 cm-3 and the volume fraction of 0.0174 was
calculated, which is approximately 77% of the nominal one.
of the molecules and of the solvent molecules filling the same volume.
The scattering length of the solvent is thus calculated by using the NS
tabulated scattering-length values bs,j of the solvent molecule of
molecular weight Ms and density ds and by using the molecular weight
Mc and the density dc of the cluster.
NS
MCdS
bhS )
bS,j
(5)
∑
MSdCNj)1
Measuring the scattering intensity in absolute units delivers the particle-
number density, n.
Equation 2 can be recast into a form that is computationally more
convenient:
N
sin(Qri)
dΣ
According to the classical scattering theory of J. J. Thomson and P.
Debye,37 the scattered intensity from molecular clusters consisting of
N atoms and having a particle-number density of n clusters/unit volume
is described by the macroscopic cross section
(Q) ) n p(ri)
(6)
∑
dΩ
Qri
i)1
where p(ri) is the so-called pair-distance distribution function (PDDF)
that represents the histogram of all existing distances r ) |brj - brk|
between atoms j and k inside the molecule. It is weighted by the
scattering lengths b(brj) and b(brk) at the positionsbrj andbrk, respectively.
N
dΣ
2
(QB) ) n| (bj - bhS) exp(iQBbrj)|
(1)
∑
dΩ
j)1
N
as a function of the momentum transfer Q ) |QB| ) (4π/λ) sin(Θ/2)
(λ, neutron wavelength; Θ, scattering angle). The bj is the coherent
scattering length of the jth atom positioned at brj, and bhS is the mean
atomic scattering length of the solvent (Table 4). Equation 1 can be
modified for the case of randomly oriented clusters.
p(r) )
[b(brj) - bhS][b(brj - br) - bhS]
(7)
∑
j)1
The
brackets indicate that the PDDF is the average over all
orientations ofbr. Because the number of atomic coordinates N is finite,
the average is automatically obtained when all N2 distances r are
collected into a histogram consisting of dr wide bins. This sampling
procedure is safe for our purposes as long as the bin width chosen is
smaller than the resolution limit dr < π/Qmax of the experiment. Qmax
corresponds to the largest accessible scattering angle.
2
N
sin(Qrj)
dΣ
(Q) ) n
(bj - bhS)
(2)
∑
(
)
dΩ
Qrj
j)1
The scattering of dendron-encapsulated clusters can be described by
the Guinier approximation according to
Acknowledgment. This project was funded by BMBF Grant
no. 03N8618A/B. D.V. thanks the Deutsche Forschungsge-
meinschaft for financial support. D.G.K. and P.L. thank the
Max-Planck-Society for financial support. M.P. and B.K.
gratefully acknowledge the support of the Fonds der Chemischen
Industrie. The authors are thankful to Helmuth Mo¨hwald for
valuable discussions.
dΣ
dΩ
dΣ
dΩ
2
(Q) =
(0) exp[-Rg Q2/3]
(3)
because their size is on the order of 4 nm, and the experimental Q
range is between 0.2 and 2.4 nm-1. Rg is the radius of gyration. The
scattering at Q ) 0 in eq 3 is
N
N
dΣ
Supporting Information Available: The Crystallographic
Information File (CIF) of Na6K6[(Zn(H2O))3 (SbWgO33)2]‚23H2O
(Na6K6[2b]‚23H2O). This material is available free of charge
via the Internet at http://pubs.acs.org.
(0) ) n| (bj - bhS)| ) n( bj - NbhS)2
(4)
2
∑
∑
dΩ
j)1
j)1
It is determined by the difference of the total coherent scattering lengths
(37) Debye, P. Ann. Phys. 1915, 46, 809.
JA017613B
9
10496 J. AM. CHEM. SOC. VOL. 124, NO. 35, 2002