Size-selective hydrogenation of olefins by Dendrimer-encapsulated palladium nanoparticles

Article abstract of DOI:10.1021/ja0105257

Nearly monodisperse (1.7 ± 0.2 nm) palladium nanoparticles were prepared within the interiors of three different generations of hydroxyl-terminated poly(amidoamine) (PAMAM) dendrimers. These dendrimer-encapsulated catalysts (DECs) were used to hydrogenate allyl alcohol and four α-substituted derivatives in a 4:1 methanol/water mixture. The results indicate that steric crowding on the dendrimer periphery, which increases with dendrimer generation, can act as an adjustable-mesh nanofilter. That is, by controlling the packing density on the dendrimer periphery, it is possible to control access of substrates to the encapsulated catalytic nanoparticle. In general, higher-generation DECs or larger substrates resulted in lower turnover frequencies (although some interesting exceptions were noted). Although the main products of the olefin hydrogenation reactions were the corresponding alkanes, ketones were also obtained when monosubstituted α-olefins were used as substrates. NMR spectroscopy was used to measure the size selectivity of DECs for the competitive hydrogenation of allyl alcohol and 3-methyl-1-penten-3-ol. The effect on catalytic rate as a function of nanoparticle size is also briefly discussed.

Full text of DOI:10.1021/ja0105257

6840
J. Am. Chem. Soc. 2001, 123, 6840-6846
Size-Selective Hydrogenation of Olefins by Dendrimer-Encapsulated
Palladium Nanoparticles
Yanhui Niu, Lee K. Yeung,† and Richard M. Crooks*,‡
Contribution from the Department of Chemistry, Texas A&M UniVersity, P.O. Box 30012,
College Station, Texas 77842-3012
ReceiVed February 26, 2001
Abstract: Nearly monodisperse (1.7 ( 0.2 nm) palladium nanoparticles were prepared within the interiors of
three different generations of hydroxyl-terminated poly(amidoamine) (PAMAM) dendrimers. These dendrimer-
encapsulated catalysts (DECs) were used to hydrogenate allyl alcohol and four R-substituted derivatives in a
4:1 methanol/water mixture. The results indicate that steric crowding on the dendrimer periphery, which increases
with dendrimer generation, can act as an adjustable-mesh nanofilter. That is, by controlling the packing density
on the dendrimer periphery, it is possible to control access of substrates to the encapsulated catalytic nanoparticle.
In general, higher-generation DECs or larger substrates resulted in lower turnover frequencies (although some
interesting exceptions were noted). Although the main products of the olefin hydrogenation reactions were the
corresponding alkanes, ketones were also obtained when monosubstituted R-olefins were used as substrates.
NMR spectroscopy was used to measure the size selectivity of DECs for the competitive hydrogenation of
allyl alcohol and 3-methyl-1-penten-3-ol. The effect on catalytic rate as a function of nanoparticle size is also
briefly discussed.
Introduction
the metal ions partition into the dendrimer where they are
strongly complexed by interior tertiary amine groups. Subse-
quent chemical reduction of the dendrimer/metal ion composite
yields a zerovalent metal nanoparticle that remains sterically
trapped within the dendrimer. Because the dendrimers are nearly
monodisperse in size and chemical composition they contain a
nearly monodisperse number of interior tertiary amine groups.
Because the maximum metal ion/dendrimer ratio is fixed for a
particular metal ion, it follows that the maximum number of
metal ions sorbed into the dendrimer is also nearly monodisperse
and that the resulting nanoparticles are, likewise, of similar size.
That is, the nanoparticle replica is a good representation of the
dendritic template from which it is prepared.
Dendrimers have been shown to be effective for enabling
chemical separations based on the size, shape, and charge of
the dendrimer or the species being separated, and they have
also been used as stationary phases and additives for various
liquid-phase separation methods such as capillary electrophore-
sis.9,10 Composite materials having dendrimers incorporated into
the structure have also been used to prepare selectively
permeable electrode coatings,11,12 xerogels,13 and organic mem-
branes for gas separations.14 All of these separation processes
are based on interactions between dendrimers and the much
smaller analyte molecules. There are two classes of interactions
between dendrimers and smaller molecules: those involving
physical interactions, such as size and shape, and those involving
electronic interactions, such as surface charge and polarity.
In this paper we show that dendrimer-encapsulated catalysts
(DECs), consisting of Pd nanoparticles contained within poly-
(amidoamine) (PAMAM) dendrimers,1 are highly selective
catalysts for hydrogenation reactions.2 Specifically, we have
examined the catalytic activity of a series of DECs toward
hydrogenation of allyl alcohol and four R-substituted derivatives
having different sizes and shapes. The interesting result is that
by controlling the steric crowding on the dendrimer periphery
it is possible to selectively control access of substrates to the
encapsulated metal particle. Because selectivity is induced by
the dendrimer, rather than by the intrinsically nonselective Pd
catalyst, this approach should be generally applicable to any
reaction that involves a catalyst that can be placed within the
interior of a dendrimer or related nanoporous materials.
We have previously shown that metal2-6 and semiconductor7
nanoparticles of nearly monodisperse size can be prepared within
dendritic hosts. These composite materials are easily prepared
by mixing together solutions containing dendrimers and metal
ions such as Pt2+, Pd2+, Au3+, Cu2+, Ni2+ 1,8
In favorable cases
.
* Author to whom correspondence should be addressed.
† Present address: The Dow Chemical Company, Freeport, TX 77541.
‡ E-mail: crooks@tamu.edu. Telephone: 979-845-5629. Fax: 979-845-
1399.
(1) Crooks, R. M.; Zhao, M.; Sun, L.; Chechik, V.; Yeung, L. K. Acc.
Chem. Res. 2001, 34, 181-190.
(2) Zhao, M.; Crooks, R. M. Angew. Chem., Int. Ed. 1999, 38, 364-
366.
(3) Zhao, M.; Sun, L.; Crooks, R. M. J. Am. Chem. Soc. 1998, 120,
4877.
(4) Zhao, M.; Crooks, R. M. AdV. Mater. 1999, 11, 217-220.
(5) Yeung, L. K.; Lee, C. T. J.; Johnston, K. P.; Crooks, R. M. Chem.
Commun. Manuscript submitted.
(6) Yeung, L. K.; Crooks, R. M. Nano Lett. 2001, 1, 14-17.
(7) Lemon, B. I.; Crooks, R. M. J. Am. Chem. Soc. 2000, 122, 12886-
12887.
(8) Grohn, F.; Bauer, B. J.; Akpalu, Y. A.; Jackson, C. L.; Amis, E. J.
Macromolecules 2000, 33, 6042-6050.
(9) Gray, A. L.; Hsu, J. T. J. Chromatogr., A 1998, 824, 119-124.
(10) Gao, H. Y.; Carlson, J.; Stalcup, A. M.; Heineman, W. R. J.
Chromatogr. Sci. 1998, 36, 146-154.
(11) Dermody, D. L.; Peez, R. F.; Bergbreiter, D. E.; Crooks, R. M.
Langmuir 1999, 15, 885-890.
(12) Liu, Y.; Zhao, M.; Bergbreiter, D. E.; Crooks, R. M. J. Am. Chem.
Soc. 1997, 119, 8720-8721.
(13) Kriesel, J. W.; Tilley, T. D. Chem. Mater. 1999, 11, 1190-1193.
(14) Kovvali, A. S.; Chen, H.; Sirkar, K. K. J. Am. Chem. Soc. 2000,
22, 7594-7595.
10.1021/ja0105257 CCC: $20.00 © 2001 American Chemical Society
Published on Web 06/20/2001

Correlation of DEC Structure to Catalytic Rate
J. Am. Chem. Soc., Vol. 123, No. 28, 2001 6841
Table 1. Hydrogenation Reaction Rates Using GnOH/Pd(0)40
While it not possible to fully separate these factors, Meijer et
al. demonstrated that dendrimers could selectively encapsulate
analyte molecules based on shape.15,16 Similarly, we have shown
that by controlling the pH of a solution dendrimers can act as
molecular-gates, admitting molecules on the basis of overall
charge.2,11,12
Catalysts for Structurally Related Allylic Alcoholsa
The selective nature of dendrimers can also be applied to
problems in catalysis. Specifically, we envision that dendrimers
encapsulating catalysts may act as selective gates that control
the access of small molecules to the embedded catalytic moiety.
In this regard, it is productive to think of the dendrimer as a
nanofilter having a mesh that can be controlled by chemically
altering the periphery. Indeed, others have previously recognized
that dendritic structures have the potential to function as selective
moieties in catalytic processes. For example, dendrons grafted
onto metalloporphyrins were shown to be selective for different-
shaped ligands17 as well as for epoxidation substrates.18,19
Dendrimers containing metal ion cores and dendritic ligands
have also been shown to affect both selectivity and reactivity
for Ru(II)-catalyzed asymmetric hydrogenations20 and Cu(II)-
catalyzed Diels-Alder reactions.21 These catalysts were com-
posed of a catalytically active species surrounded by a dendritic
shell designed to regulate both substrate selectivity and overall
catalytic reactivity.
There are two characteristics of DECs that make them
particularly attractive for catalysis. First, solubility is controlled
principally by the chemical composition of the dendrimer
periphery.1 Second, because the encapsulated particles are very
small (typically 1-3 nm),1 they have a high surface-area-to-
volume ratio, which is important for high efficiency.22,23 We
have taken advantage of these properties to show that DECs
are effective for the reduction of olefins in aqueous,2,4 organic,24
and fluorous25 solvents. Additionally, synthetically more useful
carbon-carbon bond-forming reactions have been catalyzed by
Pd(0) nanoparticles encapsulated in modified poly(propylene
imine) (PPI) dendrimers in both hydrocarbon/fluorocarbon6 and
supercritical CO2 solvent5 systems.
a Hydrogenation reactions were carried out at 25 ( 2 °C with 2 ×
10-4 M Pd(0) composite catalysts in MeOH-H2O (4:1 v/v) mixtures.
The turnover frequency (TOF) was calculated on the basis of H2 uptake
(mol of H2 per mol of Pd(0) per h). 1 Hydrogenation reactions were
carried out at 25 ( 2 °C with 2 × 10-4 M Pd(0) composite catalysts
in H2O only (no methanol). 2 Duplicate measurements were performed
to illustrate the level of run-to-run reproducibility.
indicate that the activity of DECs for a particular substrate can
be controlled by varying the generation of the dendrimer and
that for a particular DEC turnover frequencies (TOFs) differ
significantly depending on the size and shape of the substrate.
Experimental Section
Materials. Hydroxyl-terminated fourth-, sixth-, and eighth-generation
(G4OH, G6OH, and G8OH, respectively) poly(amidoamine) (PAMAM)
dendrimers having an ethylenediamine core were obtained as 10-25%
methanol solutions (Dendritech, Inc., Midland, MI). Prior to use, the
methanol was removed under vacuum at room temperature. The
unsaturated alcohols, methanol, deuterated solvents, and NaBH4 were
used as received from the Aldrich Chemical Co. (Milwaukee, WI). K2-
PdCl4 was purchased from Strem Chemicals, Inc. (Newburyport, MA)
and used without further purification. 18 MΩ‚cm Milli-Q water
(Millipore, Bedford, MA) was used to prepare aqueous solutions.
Cellulose dialysis sacks having a molecular weight cutoff of 12 000
were purchased from Sigma Diagnostics, Inc. (St. Louis, MO).
Instrumentation. 1H and 13C NMR spectra were recorded on a Unity
p300 spectrometer at 300 MHz (75 MHz for 13C). High-resolution
transmission electron micrographs (HRTEM) were obtained with a
JEOL-2010 transmission electron microscope having a point-to-point
resolution of 0.19 nm. Samples were prepared by placing a drop of
solution on a holey-carbon-coated Cu TEM grid and allowing the
solvent to evaporate in air. GC data were recorded on a HP 5890 GC
system, equipped with J&W DB-FFAP columns (30 m length, 0.32
mm i.d., 0.25 µm film, column head pressure 14 psi) and a flame
ionization detector. Helium was used as the carrier gas. The initial
temperature was set at 32 °C and increased at a rate of 10 °C/min.
Preparation of GnOH/Pd(0) Catalysts. A 5 × 10-3 M aqueous
solution of GnOH (n ) 4, 6, 8) PAMAM dendrimer was prepared from
dry GnOH. The GnOH/Pd(II) complex was prepared by adding 50 µL
of the GnOH solution into 9.75 mL of purified water, followed by 100
µL of a 0.1 M K2PdCl4 (aq) solution (40 eq) under vigorous stirring.
A light yellow solution was obtained immediately. After stirring for
30 min, 100 µL of a 1 M NaBH4 (aq) solution was added, which
changed the color of the solution to golden brown. The solution was
purified by overnight dialysis against water. Then 40 mL of methanol
was added to the solution to prepare the catalytic solution.
Although some data in our previous reports suggested high
levels of selectivity for either a particular substrate or a particular
product,1 no detailed study correlating dendrimer size and
structure to a catalytic process has thus far been reported.
Accordingly, this report is the first quantitative study correlating
DEC structure to catalytic rate. Specifically, the catalytic
efficiency of dendrimer encapsulated Pd nanoparticles toward
reduction of the double bond in differently shaped allylic
alcohols (Table 1, 1-5) is reported. Three generations of
PAMAM dendrimers were used as the template, while the size
of the catalytic nanoparticle was kept constant by controlling
the amount of Pd(II) salt used for the DEC synthesis. The results
(15) Jansen, J. F. G. A.; de Brabander-van den Berg, E. M. M.; Meijer,
E. W. Science 1994, 266, 1226-1229.
(16) Jansen, J. F. G. A.; de Brabander-van den Berg, E. M. M.; Meijer,
E. W. J. Am. Chem. Soc. 1995, 117, 4417-4418.
(17) Bhyrappa, P.; Vaijayanthimala, G.; Suslick, K. S. J. Am. Chem.
Soc. 1999, 121, 262-263.
(18) Bhyrappa, P.; Young, J. K.; Moore, J. S.; Suslick, K. S. J. Am.
Chem. Soc. 1996, 118, 5708-5711.
(19) Bhyrappa, P.; Young, J. K.; Moore, J. S.; Suslick, K. S. J. Mol.
Catal., A 1996, 113, 109-116.
(20) Fan, Q. H.; Chen, Y. M.; Chen, X. M.; Jiang, D. Z.; Xi, F.; Chan,
A. S. C. Chem. Commun. 2000, 9, 789-790.
(21) Chow, H. F.; Mak, C. C. J. Org. Chem. 1997, 62, 5116-5127.
(22) Lewis, L. N. Chem. ReV. 1993, 93, 2693-2730.
(23) Hirai, H. J. Macromol. Sci., Chem. 1979, A13, 633-649.
(24) Chechik, V.; Zhao, M.; Crooks, R. M. J. Am. Chem. Soc. 1999,
121, 4910-4911.
Hydrogenation Apparatus. Hydrogenation reactions were carried
out in a 150 mL round-bottomed Schlenk flask with an adapter
connected to the top of a buret filled with hydrogen gas. A three-way
stopcock was used with one arm attached to a hydrogen gas cylinder,
the second to the gas buret, and the third to the reaction flask. A dibutyl
phthalate-filled reservoir was connected to the bottom of the buret and
(25) Chechik, V.; Crooks, R. M. J. Am. Chem. Soc. 2000, 122, 1243-
1244.

6842 J. Am. Chem. Soc., Vol. 123, No. 28, 2001
Niu et al.
open to the atmosphere at the top. All of the hydrogenation reactions
were run at atmospheric pressure and room temperature (25 ( 2 °C).
Hydrogenation Reactions. 50 mL of the catalytic solution and a
magnetic stir bar were placed in a Schlenk flask. All of the joints of
the apparatus were sealed with silicone grease and checked for leaks
before adding the substrate. The system was purged with H2 for 15
min. To verify that H2 was not consumed in the absence of substrate,
the catalyst was stirred in solution over a known volume of H2. The
H2 volume did not change over a period of hours. Experiments were
carried out by adding 10 mmol of substrate by syringe under vigorous
stirring conditions. The rate of stirring was maintained high and constant
throughout all hydrogenation reactions. There was no evidence for
Pd(0) aggregation during or after hydrogenation of the substrates in
the MeOH-H2O solvent system, indicating that the catalysts are stable
on the time scale of the reactions.
Scheme 1
NMR Characterization of Hydrogenation Reaction Products. The
1
products of the hydrogenation reactions were identified by H NMR
by performing the catalytic reactions in deuterated solvents. When more
than one product was observed, 13C NMR spectra were also obtained.
The procedure was the same for all substrates and is illustrated using
allyl alcohol as an example. Ten microliters of a 5 × 10-4 M G4OH
aqueous solution was added to 170 µL of D2O in a reaction vial,
followed by 2 µL of 0.1 M K2PdCl4 (0.2 µmol) under vigorous stirring
for 30 min. Then 1 mg of NaBH4 (26 µmol) was added with stirring.
After 30 min, 682 µL of CD3OD was added (CD3OD/D2O ) 4) to aid
in solubilizing the substrate, and then 136 µL of allyl alcohol (2.0 mmol)
1
was added. A H NMR spectrum of this solution was obtained just
prior to the hydrogenation reaction. The hydrogenation reaction was
then performed in a reaction vial using the previously described
1
apparatus. After 24 h the reaction mixture was analyzed by H NMR
to identify the reaction product(s) (48 h after hydrogenation of substrates
4 and 5).
1
n-Propanol (1a): H NMR: δ 1.06 (t, CH3, 3H), 1.69 (m, CH2,
2H), 3.66 (t, CH2, 2H), 4.81 (b, OH, 1H).
2-Butanol (2a) and ethyl methyl ether (2b) mixture: H NMR:
1
δ 0.90 (t, CH3, 3H), 1.01 (t, CH3, 3H), 1.14 (d, CH3, 3H), 1.45 (m,
CH2, 2H), 2.16 (s, CH3, 3H), 2.53 (q, CH2, 2H), 3.66 (m, CH, 1H),
4.86 (b, OH, 1H); 13C NMR: δ 8.56, 10.56, 22.97, 29.76, 32.72, 37.64,
70.13, 214.70.
3-Pentanol (3a) and diethyl ether (3b) mixture: 1H NMR: δ 0.92
(t, CH3, 6H), 1.02 (t, CH3, 3H), 1.45 (m, CH2, 4H), 2.49 (q, CH2, 4H),
3.39 (m, CH, 1H), 4.63 (b, OH, 1H); 13C NMR: δ 8.44, 10.56, 30.36,
36.32, 75.36, 216.55.
1
2-Methyl-2-butanol (4a): H NMR: δ 0.90 (t, CH3, 3H), 1.17 (s,
CH3, 6H), 1.49 (q, CH2, 2H), 4.65 (b, OH, 1H).
dendrimer/metal-ion ratio.2-4 For the present study, we are
interested in probing the extent of substrate selectivity imparted
to the DEC by the surrounding dendrimer “nanofilter”. The
hypothesis is that dendrimer generation can be used to select
substrates based on their size, because higher generation
dendrimers have more crowded surfaces which limit access of
the substrate to the encapsulated nanoparticle (Scheme 1). To
test this hypothesis, we prepared dendrimer-encapsulated nano-
particles containing an average of 40 Pd atoms within the interior
of GnOH dendrimers (n ) 4, 6, 8, where n corresponds to
different PAMAM generations). This results in a homologous
series of DECs in which only the porosity of the dendrimer
and their diameter vary. The hydrogenation rates of chemically
similar, but structurally different, olefins were then measured
to test the aforementioned hypothesis. A decrease in turnover
frequency (TOF) is anticipated as the substrate size increases
(Scheme 1).
3-Methyl-3-pentanol (5a): 1H NMR: δ 0.88 (t, CH3, 6H), 1.10 (s,
CH3, 3H), 1.47 (q, CH2, 4H), 4.76 (b, OH, 1H).
NMR Study of G4OH Substrate Selectivity. Ten microliters of a
5 × 10-4 M aqueous solution of G4OH was added to 149 µL of D2O
in a reaction vial followed by 4 µL of 0.05 M K2PdCl4 (0.2 µmol)
under vigorous stirring for 30 min. One milligram of NaBH4 (26 µmol)
was then added with stirring. After 30 min, 650 µL CD3OD was added
(CD3OD/D2O ) 4) followed by 68 µL of allyl alcohol (1) (1.0 mmol)
and 120 µL of 3-methyl-1-penten-3-ol (5) (1.0 mmol). A 1H NMR
spectrum of this reaction mixture was obtained just prior to the treatment
with hydrogen. The hydrogenation reaction was then carried out using
the previously described apparatus. 1H NMR spectra were obtained after
2.5, 5.0, 20, and 42 h to determine the product distribution.
Molecular Modeling. Molecular models were created using the
Cerius2 (version 4.0) software package (Molecular Simulations, Inc.,
San Diego, CA). The free volume and dimensions of the substrates
(1-5) were measured on the basis of their optimized structures obtained
using MOPAC calculations.
Because allyl alcohol (1) is the simplest unsaturated alcohol
and was used as a substrate in a prior study,2 it was a natural
starting point for this work. The following structurally related
materials were also studied: 3-buten-2-ol (2), 1-penten-3-ol (3),
2-methyl-3-buten-2-ol (4), and 3-methyl-1-penten-3-ol (5) (the
structures of all five substrates are provided in Table 1).
Importantly, it has been shown previously that the intrinsic TOFs
Results and Discussion
We previously showed that hydroxyl-terminated PAMAM
dendrimers (GnOH) can be used as templates for the encapsula-
tion of Pd(0) and Pt(0) nanoparticles in aqueous solution.2,4 The
size of the entrapped metal nanoparticles can be controlled either
by using different dendrimer generations or manipulating the

Correlation of DEC Structure to Catalytic Rate
J. Am. Chem. Soc., Vol. 123, No. 28, 2001 6843
of C3-C6 unsaturated alcohols, which are structurally similar
to the allylic alcohols used here, differ only slightly.26 Accord-
ingly, it is appropriate to ascribe the differences in TOFs
discussed later to dendrimer-induced size selectivity rather than
to intrinsic properties of the substrates. Because the solubility
of substituted alcohols in H2O is not as great as in MeOH, a
MeOH-H2O (4:1, v/v) mixed solvent system was chosen to
maintain a homogeneous reaction mixture.
The composite catalysts used in this study are composed of
Pd(0) nanoclusters encapsulated within G4OH, G6OH, and
G8OH PAMAM dendrimers. In each case the DEC was
prepared using a Pd(II)-to-dendrimer ratio of 40. We expected,
and transmission electron microscopy confirmed, that following
reduction the average Pd(0) cluster size was independent of
dendrimer generation. Specifically, the encapsulated Pd(0)
nanoparticles had an average diameter and dispersity of 1.7 (
0.2 nm regardless of the dendrimer generation (see Supporting
Information).
between two O-H end groups (based on the van der Waals
surface of the outermost O atoms) in G4OH is 8.2 Å, while the
distance decreases to 5.4 Å for G6OH, and further to 3.2 Å for
G8OH. These data confirm that the “mesh size” of the dendrimer
“nanofilter” depends very sensitively on generation.
Molecular modeling was also used to calculate the free
volumes of the five substrates, which are as follows: (1) 60.7,
(2) 75.4, (3) 96.1, (4) 89.7, and (5) 103 Å3. It seems reasonable
that the substrates first penetrate into the dendrimer through
their hydroxyl groups, and therefore a good measure of substrate
size is the largest linear dimension perpendicular to the O-H
bond direction of the allylic alcohols. For substrates 1-5, these
values are 5.5, 7.0, 7.5, 7.0, and 8.0 Å, respectively. The results
of the molecular modeling parameters (substrate size and
dendrimer pore size) correlate well to the experimental data for
all three dendrimer generations (Table 1), except for substrate
4, which is smaller than 3 and has a smaller TOF. This suggests
that the model described by Scheme 1, which is based solely
on stationary steric considerations, is fairly good but does not
predict all of the experimental data correctly. Apparently there
are chemical and more complex structural aspects of the
substrate that must be taken into account to fully resolve this
issue, and therefore additional experiments are underway to
further refine the model proposed here.
It is interesting to compare hydrogenation rates for the same
substrate when different generation dendrimers encapsulate the
Pd(0) nanoparticle. Table 1 shows that there is a clear trend in
this regard: the lower generation DECs, with just two excep-
tions, result in the highest TOFs. In addition, for substrates 1-3
there is a much larger decrease in TOF when the generation
changes from G6 to G8 than for the change between G4 and
G6. These observations correlate well with the average distance
between terminal groups of the dendrimers (vide supra: G4OH,
8.2 Å; G6OH, 5.4 Å; G8OH, 3.2 Å). That is, when the
dendrimer generation increases to 8, the average distance
between terminal groups (3.2 Å) is significantly less than the
critical dimension of the substrates (range: 5.5-8.0 Å, vide
supra). Thus, because the dendrimer porosity can be controlled
by varying either the generation, as in this study, or the bulkiness
of the peripheral functional groups, it is possible to prepare
selective catalysts using as a first approximation the very simple
concept of steric crowding on the dendrimer surface.
We previously reported TOF values for 1 in water only (Table
1),2 and these rates are generally smaller than in the mixed
MeOH-H2O solvent used here. Unfortunately, it was not
possible to perform a thorough comparison of the reaction rates
of all five substrates in both solvent systems due to the limited
solubility of most of the substrates and products in pure water.
However, allyl alcohol (1) and its hydrogenation product are
both appreciably soluble in H2O. The TOFs of G4OH/Pd(0)40,
G6OH/Pd(0)40, and G8OH/Pd(0)40 for the hydrogenation of
allyl alcohol in water are 220, 200, and 130 mol H2(mol Pd)-1
h-1, respectively. In contrast, we found the maximum TOFs
for allyl alcohol hydrogenation in the mixed MeOH-H2O
solvent to be greater by about a factor of 2 for G4OH/Pd(0)40
and G6OH/Pd(0)40, but nearly the same for G8OH/Pd(0)40.
This result is consistent with previous findings that solvent plays
an important role in determining hydrogenation reaction rate at
colloidal nanoparticles.31,32 For example, the hydrogenation of
the ethyl ester of mandelic acid in ethanol is much faster than
The rate of hydrogenation for the five allylic alcohol
substrates (1-5) in the presence of the three catalysts was
determined by H2 uptake (Table 1). Turnover frequencies (TOFs,
mol of H2 per mol of Pd(0) per h) in the mixed solvent ranges
from 40 to 480 depending on the dendrimer generation and
substrate. Qualitatively, Table 1 indicates that for a particular
catalyst (dendrimer generation) the TOFs generally decrease as
the substrates become bulkier. For example, the maximum TOF
for the G4OH/Pd(0)40 catalyst for the hydrogenation of allyl
alcohol (1) was 480 mol H2(mol Pd)-1 h-1. This reaction rate
is comparable to that obtained using polymer-supported Pd(0)
catalysts in methanol (565 mol H2(mol Pd)-1 h-1 27
and
)
PAMAM encapsulated Pd(0) in fluorous-phase solvent (400 mol
H2(mol Pd)-1 h-1).25 A substrate having one R methyl group,
3-buten-2-ol (2), yielded a slightly lower TOF. When the methyl
group was changed to ethyl (3), the reaction rate decreased
further to 260 mol H2(mol Pd)-1 h-1. This trend continues when
two methyl groups are present at the R position (4) and when
both a methyl and an ethyl group are present at the R position
of the substrate (5). Overall, the TOF for the bulkiest substrate
(5) was nearly 5 times lower than the smallest (1). The same
trend was found for the G6OH/Pd(0)40 and G8OH/Pd(0)40
catalysts.
It is possible to correlate these experimental data to a
semiquantitative structural model of the dendrimer and in
particular to the average distance between functional groups on
the periphery of the dendrimer. That is, if we make the following
five assumptions, then the distance between adjacent terminal
groups for the different generation dendrimers can be calculated
using appropriate molecular models (see Experimental Sec-
tion): (1) the dendrimers are spheroidal in shape;28 (2) all of
the terminal groups are located at the surface of the den-
drimer;28,29 (3) the terminal C-O bonds are perpendicular to
the dendrimer surface with equal distances among adjacent
oxygen atoms; (4) the diameters of the G4OH, G6OH, and
G8OH dendrimers are 4.5, 6.7, and 9.7 nm, respectively;30 and
(5) the number of terminal hydroxyl groups for G4OH, G6OH,
and G8OH are 64, 256, and 1024, respectively. Using these
assumptions, the calculated average edge-to-edge distance
(26) Kacer, P.; Novak, P.; Cerveny, L. Collect. Czech. Chem. Commun.
2000, 65, 9-16.
(27) Selvaraj, P. C.; Mahadevan, V. J. Polym. Sci., Part A: Polym. Chem.
1997, 35, 105-122.
(28) Tomalia, D. A.; Naylor, A. M.; III, W. A. G. Angew. Chem., Int.
Ed. Engl. 1990, 29, 138-175.
(31) Augustine, R. L.; Warner, R. W.; Melnick, M. J. J. Org. Chem.
1984, 49, 4853-4856.
(32) Sulman, E.; Bodrova, Y.; Matveeva, V.; Semagina, N.; Cerveny,
L.; Kurtc, V.; Bronstein, L.; Platonova, O.; Valetsky, P. Appl. Catal., A
1999, 176, 75-81.
(29) Topp, A.; Bauer, B. J.; Klimash, J. W.; Spindler, R.; Tomalia, D.
A.; Amis, E. J. Macromolecules 1999, 32, 7226-7231.
(30) Information provided by Dendritech, Inc.; Midland; MI.

6844 J. Am. Chem. Soc., Vol. 123, No. 28, 2001
Niu et al.
Figure 1. 1H NMR spectra of reaction mixtures containing 5 × 10-6 M G4OH/Pd(0)40 and 2 M unsaturated 3-buten-2-ol (2) or 1-penten-3-ol (3)
before (A) and after (B) reaction with H2 in CD3OD-D2O (4:1 v/v). The data shown in parts C are the 13C NMR spectra of the reaction mixture
after hydrogenation.
in water.33 Other studies have shown that trace amounts of water
are very beneficial for increasing the hydrogenation rate in
ketonic solvent34 or methyl acetate.35 At this time we do not
fully understand why there are such large changes in TOF as a
function of solvent for nominally identical dendrimer-encap-
sulated catalysts, but it seems likely that this finding results
from changes in the surface chemistry of the encapsulated Pd
nanoparticles, structural changes of the dendrimer template
driven by differences in solvent polarity, or enhanced partition-
ing of the substrate into the dendrimer in the mixed solvent.36,37
The size of dendrimer-encapsulated nanoparticles can also
be used to tune the rate of chemical reactions. Specifically,
preliminary results indicate that increasing the size of a Pd
nanoparticle leads to a rate enhancement for the hydrogenation
reaction. For example, it is possible to hold the ratio of Pd(0)
to tertiary amines constant for all three generations of dendrim-
ers. That is, encapsulated Pd(0) clusters containing 40, 164, and
659 atoms can be prepared within G4OH, G6OH, and G8OH,
respectively. The allyl alcohol hydrogenation reaction rates
measured for the DECs containing large metal particles were
extremely fast compared with the data presented thus far: 1000
and 2500 mol H2(mol Pd)-1 h-1 TOF for G6OH/Pd(0)164 and
G8OH/Pd(0)659, respectively. Interestingly, these larger cata-
lytic particles are not very stable. There are many possible
explanations for these findings. For example, metal nanoparticles
in this size regime are known to exhibit size-dependent
electronic properties that can affect TOFs. Alternatively, mass-
transfer effects may be relevant: larger metal particles may be
more accessible to substrates than smaller particles when they
are encapsulated within the same generation dendrimer. Hy-
potheses such as these provide a means for refining the simple
model shown in Scheme 1.
(33) Ronchina, L.; Tonioloa, L.; Cavinato, G. Appl. Catal., A 1997, 165,
133-145.
The hydrogenation products were identified by NMR spec-
troscopy. The NMR spectra of reaction mixtures of 5 × 10-6
M G4OH/Pd(0)40 and 2.0 M unsaturated alcohols in CD3OD-
D2O (4:1, v/v) before and after treatment with H2 are shown in
Figure 1 and in the Supporting Information. The reaction
products for all five hydrogenation reactions are given in Scheme
2. Characterization of the reaction mixtures is simplified because
(34) Collier, P. J.; Iggo, J. A.; Whyman, R. J. Mol. Catal. A: Chem.
1999, 146, 149-157.
(35) Augustine, R. L.; Tanielyan, S. K.; Doyle, L. K. Tetrahedron:
Asymmetry 1993, 4, 1803-1827.
(36) Chai, M.; Niu, Y.; Youngs, W. J.; Rinaldi, P. L. J. Am. Chem. Soc.
2001, ASAP.
(37) Murat, M.; Grest, G. S. Macromolecules 1996, 29, 1278-1285.

Correlation of DEC Structure to Catalytic Rate
J. Am. Chem. Soc., Vol. 123, No. 28, 2001 6845
Figure 2. 1H NMR spectra of a reaction mixture containing 5 × 10-6 M G4OH/Pd(0)40, 1 M allyl alcohol (1), and 1 M 3-methyl-1-penten-3-ol
(5), before reaction with H2 and at various time increments after the hydrogenation reaction commenced (solvent: CD3OD-D2O, 4:1 v/v).
Scheme 2
appearance of the double bond peaks indicated that the reaction
went to completion. Substrates 1, 4, and 5 were cleanly
hydrogenated, yielding only the saturated alkyl alcohols without
isomerization or any other byproducts (see Supporting Informa-
tion).
Hydrogenation of substrates 2 and 3, both of which contain
an R hydrogen, generated not only the anticipated hydrogenation
products but also a small amount of byproducts (Scheme 2 and
Figure 1). In the case of 1-penten-3-ol (3), in addition to the
expected product (3-pentanol, 3a), which has peaks at 3.39 ppm
(m), 1.45 ppm (m) and 0.92 ppm (t) in the 1H NMR spectrum,
there are two extra sets of peaks at 2.49 ppm (q) and 1.02 ppm
(t) that correspond to 3-pentanone (3b). The characteristic
chemical shift at 216.55 ppm in the 13C NMR spectrum further
1
confirmed the presence of ketone. Integration of the H NMR
spectrum indicated that the ratio of the expected 3-pentanol (3a)
to the ketone byproduct is 4.7:1. Similarly, hydrogenation of
3-buten-2-ol (2) yielded the anticipated alcohol (2a) and
2-butanone (2b) as a byproduct in a 6.7:1 ratio. GC measure-
ments for both reaction mixtures were made to confirm the NMR
data. The formation of byproduct ketones from 2 and 3 is known
to involve a double bond migration resulting in the correspond-
ing enol isomers, which then slowly tautomerize to ketones.38
The important point is that reactions occurring within DECs do
not always yield the expected products. Once understood, this
finding is likely to be of great value in designing product-
selective DECs.
Having characterized the catalytic properties of the DECs in
single-substrate solvents, we endeavored to demonstrate their
the concentration of G4OH/Pd(0) catalyst is below the NMR
detection limit. Before the start of each reaction, the spectra of
the reaction mixtures showed the characteristic peaks of the
vinylic protons (5-6 ppm). After exposure to H2, the dis-
(38) Bergens, S. H.; Bosnich, B. J. Am. Chem. Soc. 1991, 113, 958-
967.

6846 J. Am. Chem. Soc., Vol. 123, No. 28, 2001
Niu et al.
selective nature by performing a competitive hydrogenation
reaction. Allyl alcohol (1) and 3-methyl-1-penten-3-ol (5) were
used for this study because the reaction does not yield
byproducts and because they have very different TOFs (Table
1). Because 1 is a more linear molecule than 5, we expected
that it would have an easier time navigating through the
Other issues that merit investigation include the effect of the
intradendrimer reaction medium on product yields. Specifically,
is it possible to develop a reasonable model to correlate the
interior structure of the dendrimer (which is undoubtedly a
strong function of solvent) to product distribution and subse-
quently design DECs that are both substrate- and product-
selective? Finally, we reported a few tantalizing initial results
that correlate TOF to nanoparticle size. Heterogeneous catalysis
studies have previously shown that either quantum electronic
or structural effects can have a dramatic influence on reaction
rates.39 DECs are ideally suited for studying such effects for
dispersed catalysts.
1
dendrimer branches than 5. H NMR spectra of the reaction
mixture of 1 and 5 (1:1 mol equiv) before and after treatment
with H2 were recorded as a function of time using G4OH/Pd-
(0)40 as the catalyst (Figure 2). After 2.5 h of hydrogenation,
50% of the allyl alcohol (1) was reduced to n-propanol (1a),
while only 1.6% 3-methyl-3-pentanol (5a) was generated. After
an additional 2.5 h, 100% of 1 is converted to 1a, but only
11% 5 was converted to 5a. After a total reaction time of 20 h,
a third of the 3-methyl-1-penten-3-ol was still unconverted to
product, but the conversion was finally complete after 42 h.
These results correlate well to the single-substrate data in Table
1, and they provide additional support for the model embodied
by Scheme 1.
Acknowledgment. We gratefully acknowledge the Office
of Naval Research and the Robert A. Welch Foundation for
supporting the research. We also thank Dr. Li Sun and Dr. Julio
Alvarez for valuable discussions, and Mr. Xu Li for performing
the GC measurements and helpful discussions regarding mo-
lecular diffusivity in catalysis. We also thank the Laboratory
for Molecular Simulation (LMS) at Texas A&M University for
providing software and computer time and Dr. Lisa M.
Thomson, Manager of the LMS, for her assistance with the
molecular modeling studies. Finally, we are particularly grateful
to one of the reviewers of this article for pointing out an
incorrect interpretation of two of the NMR spectra.
Summary and Conclusions
In this study we have described a general approach for
rendering intrinsically unselective catalysts selective. This
strategy involves encapsulation of the nonselective catalyst (the
Pd nanoparticle) within a selective nanoporous cage (the
dendrimer). As we have shown, these nanoreactors can distin-
guish between two substrates that differ only slightly in chemical
structure. On the basis of the experimental data and molecular
modeling results, we have proposed a model (Scheme 1) that
attributes selectivity principally to steric interactions between
the substrates and the functional groups on the dendrimer
periphery. Yet to be examined thoroughly are the effects of
different solvents and peripheral groups on reaction rates; those
issues are being addressed at the present time and will be
reported in due course. The important point, however, is that
by simply changing dendrimer generation reaction rates can be
varied by nearly a factor of 5 in favorable cases.
Supporting Information Available: TEM images and
particle size distributions of the GnOH/Pd(0)40 catalysts prior
to the hydrogenation reactions and 1H NMR spectra of mixtures
of 5 × 10-6 M G4OH/Pd(0)40 and three allylic alcohols (1, 4,
and 5) before and after exposure to H2 (PDF). This material is
available free of charge via the Internet at http://pubs.acs.org.
JA0105257
(39) Valden, M.; Lai, X.; Goodman, D. W. Science 1998, 281, 1647-
1650.