Noncovalent anchoring of homogeneous catalysts to silica supports with well-defined binding sites

Article abstract of DOI:10.1021/ja046956h

The efficient reversible functionalization of silica with catalytic sites using noncovalent interactions is described. We prepared silica materials with well-defined binding sites that selectively bind guest molecules that are equipped with the complement

Full text of DOI:10.1021/ja046956h

Published on Web 10/15/2004
Noncovalent Anchoring of Homogeneous Catalysts to Silica
Supports with Well-Defined Binding Sites
Ruifang Chen, Raymond P. J. Bronger, Paul C. J. Kamer,
Piet W. N. M. van Leeuwen, and Joost N. H. Reek*
Contribution from the Van’t Hoff Institute for Molecular Sciences, UniVersity of Amsterdam,
Nieuwe Achtergracht 166, 1018 WV Amsterdam, The Netherlands
Received May 24, 2004; E-mail: reek@science.uva.nl
Abstract: The efficient reversible functionalization of silica with catalytic sites using noncovalent interactions
is described. We prepared silica materials with well-defined binding sites that selectively bind guest molecules
that are equipped with the complementary binding motif, with the interaction between the two components
being based on either hydrogen bonds or metal-ligand interactions. Several phosphine ligands functionalized
with glycine-urea groups, required for hydrogen bond formation to the complementary host on the silica,
have been prepared. The resulting noncovalently immobilized complexes have been used as a ligand system
in the Pd-catalyzed allylic substitution and Rh-catalyzed hydroformylation of 1-octene. The supramolecular
interaction between the transition-metal catalyst and the binding site located at the support is sufficiently
strong to enable efficient catalyst recycling. In addition, the nature of the support facilitates the de- and
refunctionalization of support, allowing the recycling of both homogeneous catalysts and the functionalized
support. A rhodium catalyst based on a functionalized xantphos ligand was used in the hydroformylation
of 1-octene in 11 consecutive reactions without showing catalyst deterioration or metal leaching.
Introduction
the separation problem, does not exist, and especially, problems,
such as solubility of different components, catalyst instability,
The efficient separation and subsequent recycling of homo-
geneous transition-metal catalysts remains a topic that not only
constitutes scientific challenges but also is of commercial
relevance. The practical use of the ever-increasing number of
tailor-made transition-metal catalysts1 is indeed limited by the
cumbersome separation from the product phase, especially for
application in the fine-chemical industry.2 For bulk chemicals,
several simple unit operations are applied, such as distillation,
extraction, filtration, and two-phase catalysis.3 So far, several
strategies for advanced catalyst recycling have been explored,
including supported aqueous phase catalysis,4 fluorous phase
catalysis,5 and the use of ionic liquids6 and supercritical fluids.7
A widely studied approach to facilitate catalyst-product separa-
tion is the attachment of homogeneous catalysts to dendritic,8
polymeric organic, inorganic, or hybrid supports.9,10 Unfortu-
nately, the holy grail in this research area, a single solution to
and metal leaching, during the recycling procedure strongly limit
the utility of some of the concepts. For example, aqueous phase
catalysis is limited to substrates that are soluble in water.
Immobilization of inorganic materials, such as silica, has
advantages due to the physical strength and chemical inertness,
but the activity is generally lower than that of the homogeneous
system. A decrease in activity does not necessarily occur for
catalysts attached to dendrimers and (hyperbranched) soluble
polymeric supports, but here, the availability of suitable
membrane materials complicates the practical use.
In most supported catalysts reported so far, the catalyst has
been covalently linked to the support. An interesting alternative
(6) (a) Wasserscheid, P.; Welton, T. Ionic Liquids in Synthesis; Wiley-VCH:
Weinheim, Germany, 2003. (b) Welton, T. Chem. ReV. 1999, 99, 2071-
2083. (c) Wasserscheid, P.; Waffenschmidt, H.; Machnitzki, P.; Kottsieper,
K. W.; Stelzer, O. Organometallics 2000, 19, 3818-3823. (d) Bronger, R.
P. J.; Silva, S. M.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Chem.
Commun. 2002, 3044-3045. (e) Keim, W.; Vogt, D.; Waffenschmidt, H.;
Wasserscheid, P. J. Catal. 1999, 186, 481-484. (f) Wasserscheid, P.;
Waffenschmidt, H.; Machnitzki, P.; Kottsieper, K. W.; Stelzer, O. Chem.
Commun. 2001, 451-452.
(1) Herrmann, W. A.; Cornils, B. Angew. Chem., Int. Ed. Engl. 1997, 36, 1048-
1080.
(2) (a) Baker, R. T.; Tumas, W. Science 1999, 284, 1477-1479. (b) Cole-
Hamilton, D. J. Science 2003, 299, 1702-1706.
(3) Applied Homogeneous Catalysis with Organometallic Compounds; Cornils,
B., Herrmann, W. A., Eds.; VCH: Weinhein, Germany, 1996; Chapter 3.
(4) (a) Arhanchet, J. P.; Davis, M. E.; Merola, J. S.; Hanson, B. E. Nature
1989, 339, 454-455. (b) Davis, M. E. CHEMTECH 1992, 498-502. (c)
Sandee, A. J.; Slagt, V. F.; Reek, J. N. H.; Kamer, P. C. J.; van Leeuwen,
P. W. N. M. Chem. Commun. 1999, 17, 1633-1634.
(7) (a) Jessop, P. G.; Ikariya, T.; Noyori, R. Chem. ReV. 1999, 99, 475-494.
(b) Darr, J. A.; Poliakoff, M. Chem. ReV. 1999, 99, 495-542. (c) Leitner,
W. Acc. Chem. Res. 2002, 35, 746-756.
(8) (a) van Heerbeek, R.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.; Reek,
J. N. H. Chem. ReV. 2002, 102, 3717-3756. (b) Kreiter, R.; Kleij, A. W.;
Klein Gebbink, R. J. M.; van Koten, G. In Dendrimers IV: Metal
Coordination, Self-Assembly, Catalysis; Vo¨gtle, F., Schalley, C. A., Eds.;
Springer-Verlag: Berlin, 2001; Vol. 217, p 163. (c) Oosterom, G. E.; Reek,
J. N. H.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Angew. Chem., Int.
Ed. 2001, 40, 1828-1849. (d) Crooks, R. M.; Zhao, M.; Sun, L.; Chechik,
V.; Yueng, L. K. Acc. Chem. Res. 2001, 34, 181-190. (e) Reek, J. N. H.;
de Groot, D.; Oosterom, G. E.; Kamer, P. C. J.; van Leeuwen, P. W. N.
M. ReV. Mol. Biotechnol. 2002, 90, 159-181. (f) Astruc, D.; Chardac, F.
Chem. ReV. 2001, 101, 2991-3024.
(5) (a) Horva´th, I. T.; Ra´bai, J. Science 1994, 266, 72-75. (b) Horva´th, I. T.
Acc. Chem. Res. 1998, 31, 641-650. (c) Barthel-Rosa, L. P.; Gladysz, J.
A. Coord. Chem. ReV. 1999, 190-192, 587-605. (d) Rocaboy, C.;
Rutherford, D.; Bennett, B. L.; Gladysz, J. A. J. Phys. Org. Chem. 2000,
13, 596-603. (e) Richter, B.; Spek, A. L.; van Koten, G.; Deelman, B.-J.
J. Am. Chem. Soc. 2000, 122, 3945-3951. (f) de Wolf, E.; van Koten, G.;
Deelman, B.-J. Chem. Soc. ReV. 1999, 28, 37. (g) Fish, R. H. Chem.sEur.
J. 1999, 5, 1677-1680.
9
10.1021/ja046956h CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 14557-14566
14557

A R T I C L E S
Chen et al.
Scheme 1. General Concept of Supramolecular Anchoring of
approach is the noncovalent anchoring of the catalyst to the
support. Only a few examples have been reported so far, and
most examples involve the immobilization via ionic interactions.
Cationic transition-metal catalysts have been immobilized on
heteropolyacids11 and silica supports12 via ion pairing. This
appeared to be a viable approach for cationic rhodium catalysts
that are active in hydrogenation reactions, but the concept is
obviously limited to charged catalysts. In a similar approach,
Mecking and Schwab immobilized NaTPPTS to a polyelectro-
lyte providing a polymer-bound, soluble hydroformylation
catalyst that was recovered via ultrafiltration.14 Bianchini
introduced the so-called supported hydrogen-bonded catalysts
in which sulfonated phosphine-based ligands are tethered to a
silica surface via hydrogen bonds between the silanols and the
sulfonate groups.13 In this approach, the noncovalent interaction
is established between the support and the ligand, which makes
it applicable to a broader range of catalysts. Indeed, such
rhodium and ruthenium complexes have been used for (enan-
tioselective) hydrogenation13c and hydroformylation reactions.13a
Here, we report a new approach in which well-defined binding
sites, based on different binding motifs, are immobilized on
silica that can be noncovalently functionalized with catalysts
that have ancillary ligands with the complementary motif
(Scheme 1). This offers a high level of control and flexibility
since, in this approach, the noncovalent anchoring is independent
of the type of support and the transition metal. The reversible
nature of this type of anchoring allows controlled de- and
Catalysts Using Binding Sites that are Immobilized on Silica
Support (A), and the Two Binding Motifs Explored in this Paper (B)
refunctionalization of the support, which enables the simple
reuse of the support ultimately leading to a modular multipur-
pose system, and simplifies the variation of catalyst loading even
during catalysis. Furthermore, the noncovalent approach simpli-
fies the preparation of multicomponent catalysts by using
different orthogonal binding motifs in a modular fashion, which
can be interesting for tandem reactions. To ultimately reach this
goal, we have investigated the immobilization of two well-
defined binding sites on the support to which tailor-made
transition-metal catalysts, with a complementary binding motif,
have been assembled (Scheme 1). One binding motif is based
on hydrogen bonds and ionic interactions, an interaction that
was previously used to functionalize dendrimers,15 and the other
motif is based on metal-ligand interactions,16 and in principle,
these interactions are orthogonal.
(9) Extensive reviews on polymer-immobilized catalysts include: (a) Hartley,
F. R.; Vezey, P. N. AdV. Organomet. Chem. 1977, 15, 189-234. (b) Hartley,
F. R. Supported Metal Complexes. A New Generation of Catalysts;
Reidel: Dordrecht, The Netherlands, 1985. (c) Iwasawa, Y. In Tailored
Metal Catalysts; Ugo, R., James, B. R., Eds.; Reidel: Dordrecht, The
Netherlands, 1986. (d) Keim, W.; Driessen-Ho¨lscher, B. In Handbook of
Heterogeneous Catalysis; Ertl, G., Kno¨zinger, H., Weitkamp, J., Eds.;
Wiley-VCH: Weinhein, Germany, 1997; Vol. 1, p 231. (e) Lindner, E.;
Schneller, T.; Auer, F.; Mayer, H. A. Angew. Chem., Int. Ed. 1999, 38,
2154-2174. (f) Thomas, J. M. Angew. Chem., Int. Ed. 1999, 38, 3588-
3628. (g) Grubbs, R. H. CHEMTECH 1977, 512-518. (h) Leznoff, C. C.
Chem. Soc. ReV. 1974, 3, 65-85. (i) Thematic issue: J. Mol. Catal. A
2002, 177, 1-170. (j) Thematic issue: Chem ReV. 2002, 102, 3215-3892.
(k) de Miguel, Y. R.; Brule´, E.; Margue, R. G. J. Chem. Soc., Perkin Trans.
1 2001, 3085-3094. (l) Kobayashi, S.; Akiyama, R. Chem. Commun. 2003,
449-450. (m) Bra¨se, S.; Lauterwasser, F.; Ziegert, R. E. AdV. Synth. Catal.
2003, 345, 869-929. (n) Lu, Z.-L.; Lindner, E.; Mayer, H. A. Chem. ReV.
2002, 102, 3543-3578. (o) Wight, A. P.; Davis, M. E. Chem. ReV. 2002,
102, 3589-3614.
(10) Other interesting contributions: (a) Ajjou, A.; Alper, H. J. Am. Chem. Soc.
1998, 120, 1466-1468. (b) Sandee, A. J.; van der Veen, L. A.; Reek, J. N.
H.; Kamer, P. C. J.; Lutz, M.; Spek, A. L.; van Leeuwen, P. W. N. M.
Angew. Chem., Int. Ed. 1999, 38, 3231-3235. (c) Sandee, A. J.; Reek, J.
N. H.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. J. Am. Chem. Soc.
2001, 123, 8468-8476. (d) Buchmeiser, M. R.; Wurst, K. J. Am. Chem.
Soc. 1999, 121, 11101-11107. (e) Bergbreiter, D. E.; Liu, Y.-S.; Osburn,
P. L. J. Am. Chem. Soc. 1998, 120, 4250-4251. (f) Richmond, M. K.;
Scott, S. L.; Alper, H. J. Am. Chem. Soc. 2001, 123, 10521-10525. (g)
Bergbreiter, D. E. Catal. Today 1998, 42, 389-397. (h) Sandee, A. J.;
Petra, D. G. I.; Reek, J. N. H.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.
Chem.sEur. J. 2001, 7, 1202-1208. (i) Bergbreiter, D. E.; Osburn, P. L.;
Liu, Y.-S. J. Am. Chem. Soc. 1999, 121, 9531-9538. (j) Nozaki, K.; Itoi,
Y.; Shibahara, F.; Shirakawa, E.; Ohta, T.; Takaya, H.; Hiyama, T. J. Am.
Chem. Soc. 1998, 120, 4051-4052. (k) Farrell, M. O.; van Dyke, C. H.;
Boucher, L. J.; Metlin, S. J. J. Organomet. Chem. 1979, 172, 367-376. (l)
Bourque, S. C.; Alper, H. J. Am. Chem. Soc. 2000, 122, 956-957.
(11) Augustine, R.; Tanielyan, S.; Anderson, S.; Yang, H. Chem. Commun. 1999,
1257-1258.
Experimental Section
Materials. Amorphous silica powder (silica 60, 70-200 µm, surface
area ) 550 m2/g) was obtained from Fluka and dried under vacuum at
110 °C overnight before being used. 1-Octene was purified over neutral
alumina prior to its use. All other chemicals were purchased com-
mercially and used without further purification. Solvents were dried
prior to their use. Hexane, pentane, diethyl ether, THF, and toluene
were distilled from sodium. Dichloromethane and triethylamine were
distilled under nitrogen from calcium hydride/benzophenone. All
solutions and solvents not stated above were degassed under argon prior
to their use. All experiments were performed using standard Schlenk
techniques under an argon atmosphere.
A. Analytical Techniques. NMR spectra were recorded on a Varian
mercury 300 or Inova 500 spectrometer. 31P and 13C spectra were
1
1
measured H decoupled. TMS was used as a standard for H and 13C
NMR, and 85% H3PO4 in H2O was used for 31P NMR. Mass spectra
(15) (a) Baars, M. W. P. L.; Karlsson, A. J.; Sorokin, V.; De Waal, B. F. M.;
Meijer, E. W. Angew. Chem., Int. Ed. 2000, 39, 4262-4265. (b) Boas, U.;
Karlsson, A. J.; De Waal, B. F. M.; Meijer, E. W. J. Org. Chem. 2001, 66,
2136-2145. (c) De Groot, D.; De Waal, B. F. M.; Reek, J. N. H.;
Schenning, A. P. H. J.; Kamer, P. C. J.; Meijer, E. W.; van Leeuwen, P.
W. N. M. J. Am. Chem. Soc. 2001, 123, 8453-8458.
(16) (a) Slagt, V. F.; Reek, J. N. H.; Kamer, P. C. J.; van Leeuwen, P. W. N.
M. Angew. Chem., Int. Ed. 2001, 40, 4271-4274. (b) Slagt, V. F.; Kamer,
P. C. J.; van Leeuwen, P. W. N. M.; Reek, J. N. H. Angew. Chem., Int. Ed.
2003, 42, 5619-5623. (c) Slagt, V. F.; van Leeuwen, P. W. N. M.; Reek,
J. N. H. Chem. Commun. 2003, 2474-2475. (d) Slagt, V. F.; Kamer, P. C.
J.; van Leeuwen, P. W. N. M.; Reek, J. N. H. J. Am. Chem. Soc. 2004,
126, 1526-1536. (e) Slagt, V. F.; Roeder, M.; Kamer, P. C. J.; van
Leeuwen, P. W. N. M.; Reek, J. N. H. J. Am. Chem. Soc. 2004, 126, 4056-
4057.
(12) (a) Diaz-Requejo, M. M.; Belderrain, T. R.; Nicasio, M. C.; Perez, P. J.
Organometallics 2000, 19, 285-289. (b) de Rege, F. M.; Morita, D. K.;
Ott, K. C.; Tumas, W.; Broene, R. D. Chem. Commun. 2000, 1797-1798.
(c) Richmond, M. K.; Scott, S. L.; Alper, H. J. Am. Chem. Soc. 2001, 123,
10521-10525.
(13) (a) Bianchini, C.; Burnaby, D. G.; Evans, J.; Frediani, P.; Meli, A.;
Oberhauser, W.; Psaro, R.; Sordelli, L.; Vizza, F. J. Am. Chem. Soc. 1999,
121, 5961-5971. (b) Bianchini, C.; Dal Santo, V.; Meli, A.; Oberhauser,
W.; Psaro, R.; Vizza, F. Organometallics 2000, 19, 2433-2444. (c)
Bianchini, C.; Barbaro, P.; Dal Santo, V.; Gobetto, R.; Meli, A.; Oberhauser,
W.; Psaro, R.; Vizza, F. AdV. Synth. Catal. 2001, 343, 41-45.
(14) Schwab, E.; Mecking, S. Organometallics 2001, 20, 5504-5506.
9
14558 J. AM. CHEM. SOC. VOL. 126, NO. 44, 2004

Supramolecular Anchoring of Catalysts
A R T I C L E S
(FAB) were recorded on a JEOL JMS SX/SX102A instrument.
Elemental analysis was carried out by Kolbe, Microanalytische Labor-
atorium, Mu¨lheim. The amounts of palladium and rhodium were
determined with induced coupled argon plasma-atomic emission
spectrometry (ICP-AES). The ICP-AES measurements were determined
with a sequential Jarrell Ash upgraded (model 25) atomscan model
2400 ICP scanning monochromator and a Perkin-Elmer Optima 3000
XL instrument. Gas chromatography was performed on an Interscience
HR GC Mega 2 apparatus (split/splitless injector, J & W Scientific,
DB1 ) 30 m column, film thickness ) 3.0 mm, carrier gas ) 70 kPa
He, F. I. D. detector). FT-IR spectra were obtained on a Bio-Rad FTS-7
spectrophotometer. UV-vis spectroscopy experiments were performed
on an HP 8453 UV-visible system. Melting points were determined
on a Gallenkamp MFB-595 melting point apparatus in open capillaries
and were reported uncorrected.
(FAB+) m/z calcd for C50H53BrOP2 [M + H]+ 811.2. Found 811.2.
Anal. Calcd for C50H53BrOP2: C, 73.98; H, 6.58. Found: C, 73.77; H,
6.51.
Synthesis of 6. A quantity of 0.50 g (0.61 mmol) of 5 was dissolved
into 15 mL of THF. Under argon, the solution was transferred into an
autoclave (100 mL). After addition of 20 mL of liquid NH3, the mixture
in the autoclave was stirred and heated at 70 °C for 16 h. THF was
removed in vacuo, and the residue was dissolved in 20 mL of
dichloromethane and washed with 10 mL of water. The organic layer
was separated and dried over MgSO4. A white solid was obtained after
purification after flash column chromatography (eluent: 30% dichlo-
romethane in light petroleum ether), which yielded 0.41 g (89%). Mp
1
131-132 °C dec. H NMR (500 MHz, CDCl3): δ 7.24 (m, 24 H),
6.35 (br s, 2 H), 2.64 (t, J ) 6.50 Hz, 2 H), 2.43(m, 4 H), 1.67 (s, 6
H), 1.41 (m, 6 H), 1.26-1.21 (m, 8 H), 0.87 (t, J ) 6.5 Hz, 3 H).
31P{1H} NMR (121.4 MHz, CDCl3, versus H3PO4): δ -16.6. 13C NMR
(125.7 MHz, CDCl3): δ 150.8 (t, J ) 12.3 Hz), 137.6 (t, J ) 5.9 Hz),
137.1 (s), 136.8 (s), 133.8 (t, J ) 10.9 Hz), 131.8 (s), 129.6 (s), 129.4
(s), 128.0 (s), 126.1 (s), 125.0 (s), 42.0 (s), 35.2 (s), 34.4 (s), 33.4 (s),
31.8 (s), 31.5 (s), 31.2 (s), 31.0 (s), 28.5 (s), 26.1 (s), 22.5 (s), 14.0 (s).
FT-IR (KBr): 3453, 3053, 2961, 2854 cm-1. MS (FAB+) m/z calcd
for C50H55NOP2 [M + H]+ 748.4. Found 748.5. Anal. Calcd for C50H55-
NOP2: C, 80.29; H, 7.41; N, 1.87. Found: C, 80.38; H, 7.46; N, 1.84.
Synthesis of 7. To a solution of 0.25 g (0.33 mmol) of 6 in 10 mL
of dichloromethane was added 0.038 mL (0.35 mmol) of ethylisocy-
anatoacetate. After the mixture was stirred overnight at room temper-
ature, the solvent was evaporated. The product was recrystallized from
dichloromethane/pentane and obtained as a white solid, which yielded
Synthesis of 1. DAB-dendr-(NH2)4 (0.12 g, 0.38 mmol) was
dissolved in 5 mL of dichloromethane. 3-Triethoxysilylpropylisocyanate
(0.42 g, 1.70 mmol) was added, and the mixture was stirred for 3 h at
room temperature. The mixture was precipitated by addition to a stirred
diethyl ether solution (20 mL), and the pure white solid was collected
by filtration, which yielded 0.42 g (85%). Mp 91-92 °C dec. 1H NMR
(500 MHz, CDCl3): δ 5.87 (br, 2 H), 5.52 (br, 2 H), 3.83 (q, J ) 6.5
Hz, 8 H), 3.24 (br t, 8 H), 3.18 (br, 8 H), 2.40 (br, 12 H), 1.59-1.64
(m, 12 H), 1.24 (t, J ) 6.5 Hz, 12 H), 0.66 (br, 12 H). 13C NMR (75.5
MHz, CDCl3): δ 159.5 (s), 58.6 (s), 53.6 (s), 51.7 (s), 43.1 (s), 38.8
(s), 27.9 (s), 24.8 (s), 24.0 (s), 18.5 (s), 7.9 (s). FT-IR (KBr): 3338,
1629 cm-1. HRMS (FAB+) m/z calcd for C56H124N10O16Si4 [M + H]
1305.8352. Found 1305.8368. Anal. Calcd for C56H124N10O16Si4: C,
51.50; H, 9.57; N, 10.73. Found: C, 51.75; H, 9.77; N, 10.78.
1
0.25 g (85%). Mp 85-86 °C dec. H NMR (500 MHz, CDCl3): δ
7.29-7.18 (m, 24 H), 6.34 (s, 1 H), 6.33 (s, 1 H), 4.21 (q, J ) 7.5 Hz,
2 H), 3.98 (s, 2 H), 3.11 (t, J ) 7.0 Hz, 2 H), 2.41 (t, J ) 7.5 Hz, 4
H), 1.64 (s, 6 H), 1.43 (m, 6 H), 1.29 (t, J ) 7.0 Hz, 3 H), 1.25-1.20
(m, 8 H), 0.86 (t, J ) 6.5 Hz, 3H). 31P NMR (121.4 MHz, CDCl3,
versus H3PO4): δ -17.3. 13C NMR (125.7 MHz, CDCl3): δ 171.4
(s), 158.3 (s), 137.8 (m), 137.7 (s), 137.4 (s), 136.9 (s), 134.0 (m),
132.2 (m), 132.1 (m), 131.9 (s), 129.8 (s), 129.7 (s), 128.2 (br s), 126.4
(s), 61.5 (s), 42.4 (s), 40.7 (s), 35.5 (s), 34.7 (s), 32.0 (s), 31.8 (s), 31.4
(s), 30.1 (s), 28.8 (s), 26.4 (s), 22.7 (s), 14.3 (s), 14.2 (s). FT-IR
(KBr): 3364, 3051, 2930, 2856, 1751, 1637 cm-1. MS (FAB+) m/z
calcd for C55H62N2O4P2 [M + H]+ 877.4. Found 877.5. Anal. Calcd
for C55H62N2O4P2: C, 75.32; H, 7.13; N, 3.19. Found: C, 75.26; H,
7.06; N, 3.17.
Synthesis of 2. A quantity of 100 mg (0.13 mmol) of zinc(II) tetra-
4-aminophenylporphyrin was dissolved into 20 mL of dry DMF. After
132.8 mg (0.56 mmol) of 3-triethoxysilylpropylisocyanate was added,
the mixture was stirred at 80 °C for 5 h. The solvent and excess of the
reagent were removed under high vacuo to afford the pure product,
1
which yielded 200 mg (90%). H NMR (DMSO): δ 8.80 (s, 8 H),
8.01 (d, J ) 7.5 Hz, 8 H), 7.79 (d, J ) 8.4 Hz, 8 H), 6.40 (s, 4 H),
5.80 (s, 4 H), 3.77 (q, J ) 7.2 Hz, 24 H), 3.19 (t, J ) 7.2 Hz, 8 H),
1.58 (m, 8 H), 1.13 (t, J ) 6.9 Hz, 36 H), 0.63 (t, J ) 7.2 Hz, 8 H).
13C NMR (75.5 MHz, DMSO): δ 159.5 (s), 155.1 (s), 153.3 (s), 140.5
(s), 136.5 (s), 135.8 (s), 126.2 (s), 117.6 (s), 63.1 (s), 47.2 (s), 28.7 (s),
23.9 (s), 12.7 (s). FT-IR (KBr): 3360, 1635, 1618 cm-1. MS (FAB+)
m/z calcd for C84H116N12O16Si4Zn [M + H] 1728.7. Found 1728.7. Anal.
Calcd for C84H116N12O16Si4Zn: C, 58.40; H, 6.77; N, 9.73. Found: C,
58.51; H, 6.73; N, 9.82.
Synthesis of 8. A solution of 15.0 mg of NaOH (0.37 mmol) in 4
mL water was added to a solution of 0.27 g (0.31 mmol) of 7 in 5 mL
of THF. After the mixture was stirred overnight, the THF was
evaporated and the reaction mixture was neutralized by addition of 2
mL of 0.40 M aqueous HCl. The solvent was decanted, and the crude
product was washed with water. After recrystallization from chloroform,
a white powder was obtained, which yielded 0.13 g (49%). Mp 95-96
°C dec. 1H NMR (500 MHz, CDCl3): δ 7.28-7.17 (m, 24 H), 6.35 (s,
1 H), 6.33 (s, 1 H), 3.91 (s, 2 H), 3.46 (t, J ) 7.5 Hz, 2 H), 2.41 (t, J
) 7.5 Hz, 4 H), 1.65 (s, 6 H), 1.43 (m, 6 H), 1.29-1.20 (m, 8 H), 0.86
(t, J ) 6.5 Hz, 3 H). 31P NMR (121.4 MHz, DMSO, versus H3PO4):
δ -13.43. 13C NMR (75.5 MHz, CDCl3): δ 171.3 (s), 151.1 (t, J )
13.7 Hz), 137.8 (m), 137.7 (s), 137.4 (s), 136.8 (s), 134.1 (m), 132.3
(m), 132.0 (m), 131.9 (s), 129.8 (s), 129.7 (s), 128.2 (br s), 126.4 (s),
46.4 (s), 38.8 (s), 35.5 (s), 34.7 (s), 32.0 (s), 31.7 (s), 31.4 (s), 30.9 (s),
28.8 (s), 26.3 (s), 22.7 (s), 14.2 (s). FT-IR (KBr): 3386, 3051, 2927,
2855, 1773, 1654, 1565 cm-1. HRMS (FAB+) m/z calcd for
C53H58N2O4P2 [M - OH]+ 831.4. Found 831.4. Anal. Calcd for
C53H58N2O4P2: C, 74.98; H, 6.89; N, 3.30. Found: C, 74.62; H, 6.77;
N, 3.21.
Synthesis of 5. At -78 °C, 8.5 mL of n-butyllithium (1.64 M in
pentane, 13.90 mmol) was added dropwise to a stirred solution of 4,5-
dibromo-2-(5-bromopentyl)-7-hexyl-9,9-dimethylxanthene (4.0 g, 6.64
mmol) in 200 mL of Et2O. The resulting suspension was stirred for
another 2 h at -78 °C. Next, a solution of 2.55 mL of chlorodiphen-
ylphosphine (13.9 mmol) in 10 mL of ether was added, and the reaction
mixture was slowly warmed to room temperature and stirred overnight.
A quantity of 20 mL of 4 M hydrochloric acid was then added to quench
the reaction mixture. The water layer was removed, and the organic
layer was dried with MgSO4. The solvents were removed in vacuo,
and the residual white solid was washed with pentane and purified by
flash column chromatography (eluent: 10% dichloromethane in light
1
petroleum ether), which yielded 2 g (37%). Mp 140-141 °C dec. H
NMR (500 MHz, CDCl3): δ 7.19-7.22 (m, 24 H), 3.45 (t, J ) 6.9
Hz, 2 H), 3.32 (t, J ) 6.9 Hz, 2 H), 2.41 (t, J ) 7.5 Hz, 2 H), 1.76 (q,
J ) 7.2 Hz, 2 H), 1.62 (s, 2 H), 1.45-1.17 (m, 4 H), 0.84 (t, J ) 6.6
Hz, 3 H). 31P NMR (121.4 MHz, CDCl3, versus H3PO4): δ -17.2. 13
C
NMR (125.7 MHz, CDCl3): δ 150.2 (t, J ) 10.0 Hz), 137.5 (s), 136.8
(s), 134.1 (t, J ) 10.2 Hz), 132.2 (s), 132.1 (s), 129.9 (s), 129.6 (s),
128.5 (s), 128.4 (s), 128.3 (s), 126.6 (s), 45.1 (s), 35.5 (s), 35.2 (s),
33.9 (s), 32.7 (s), 32.0 (s), 31.8 (s), 31.4 (s), 30.5 (s), 28.8 (s), 27.6 (s),
22.7 (s), 14.2 (s). FT-IR (KBr): 2957, 2925, 2854, 1421 cm-1. MS
B. Synthesis of Silica-Immobilized Host Material: Si-I, Si-II, and
Si-III. Si-I. A quantity of 1.0 g of silica was dried at 180 °C under
reduced pressure for 2 h. To a suspension of the silica in 15 mL of
toluene was added 0.1 g (0.0767 mmol) of 1, and the resulting mixture
9
J. AM. CHEM. SOC. VOL. 126, NO. 44, 2004 14559

A R T I C L E S
Chen et al.
was refluxed for 5 h. The silica was isolated by filtration, washed with
toluene, and dried under vacuo. Analysis of the solution showed that
all of the host material 1 was attached to the silica. The subsequent
silica modification was performed by refluxing a mixture of 1 g of the
modified silica and 1 mL of dimethoxydimethylsilane in 15 mL of
toluene for 2 h. The resulting Si-I was washed with toluene, dried under
reduced pressure, and stored under argon.
Si-II. A quantity of 0.1 g of 1 (0.0767 mmol) was dissolved in 3
mL of THF. One milliliter of H2O and 1 mL of Si(OMe)4 were
subsequently added. Gelation took place within 1 h. After 36 h, the
gel was carefully dried under reduced pressure. The dried gel was made
into a powder and thoroughly washed with MeOH, THF, and Et2O.
The resulting silica was further modified by refluxing with 1 mL of
dimethoxydimethylsilane in 15 mL of toluene for 2 h. The resulting
Si-II was washed with toluene, dried under reduced pressure, and stored
under argon.
Si-III. To a suspension of the silica (1.55 g, predried at 180 °C
under vacuo for 2 h) in 15 mL of toluene was added 0.15 g (0.0868
mmol) of 2, and the resulting mixture was refluxed for 5 h. The silica
was washed with toluene and dried under vacuo. Subsequently, silica
modification was carried out by refluxing a mixture of 1 g of silica
and 1 mL of dimethoxydimethylsilane in 15 mL of toluene for 2 h.
The resulting Si-III was washed with toluene, dried under reduced
pressure, and stored under argon.
Determination of Binding Constants Using 1H NMR Titrations.
1H NMR titration experiments were performed to determine the
association constants (K1 and K2) related to the binding of the first and
the second guest ligand in the two binding pockets of 1, respectively.
(4.817 mg of 4 in 10 mL of dried CH2Cl2) was prepared. For each
data point, a different volume of stock solution (100-500 µl) was
transferred into a UV cuvette and then diluted into 5 mL of solution.
After different amounts of silica-I (1.0-7.0 mg) were added to this
solution, the mixture was shaken at room temperature for 2 h. The
concentration of guest 4 in free solution was determined by UV-vis
spectroscopy experiments using an HP 8453 UV-visible system.
Analysis of the data with a nonlinear regression program led to a value
of K ) 2 × 104 M-1 and a binding site of 17 × 10-5 (mol/g silica).
C. Catalytic Procedures. Allylic Amination Using Noncovalently
Immobilized Palladium Catalysts. At room temperature and under
argon, a mixture of 5 × 10-6 mol of [Pd(croty)Cl]2, 1 × 10-5 mol of
ligand 4, and 250 mg of the Si-I was stirred in 5 mL of dichloromethane
for 30 min. After the catalyst was washed with dichloromethane,
catalysis began with the addition of 1.0 mmol of crotyl acetate and 2.0
mmol of freshly distilled piperidine in 5 mL of dichloromethane. The
reaction mixture was monitored in time by quenching samples in DBA/
Et2O (DBA is dibenzylideneacetone) and analyzed by GC for the
conversion and product distribution. Subsequent catalytic runs were
performed after the products were removed and the noncovalently
anchored catalyst was washed with dichloromethane.
Allylic Alkylation Using Noncovalently Immobilized Palladium
Catalysts. A mixture of 5 × 10-6 mol of [Pd(allyl)Cl]2, 1 × 10-5 mol
of ligand 6, and 140 mg of Si-III in 5 mL of CH2Cl2 was stirred for 30
min. After the catalyst was washed with CH2Cl2, catalysis began with
the addition of 5 mL of CH2Cl2, 1.0 mmol of crotyl acetate, 2.0 mmol
of diethyl-2-methylmalonate, 2.0 mmol of BSA [N,O-bis(trimethylsilyl)-
acetamide], and a catalytic amount of KOAc (1.5 mg). The reaction
mixture was monitored by taking samples from the reaction mixture
that, after an aqueous workup, were analyzed by GC using decane as
the internal standard. Subsequent catalytic runs were performed after
the products were removed and the supported catalyst was washed with
CH2Cl2.
K1
H + G y\z HG
K2
HG + G y\z HG2
Hydroformylation of 1-Octene. A typical catalysis experiment
included the following: a stainless steel 50 mL autoclave, equipped
with a mechanical stirrer, a substrate vessel, and a cooling spiral. A
sample outlet was charged with 1 × 10-5 mol of Rh(acac)(CO)2, 1 ×
10-4 mol of ligand 4, 6, or 8, 1.2 g of Si-I or Si-III, and 20 mL of
toluene. The suspension was incubated for 1 h at 80 °C under 20 bar
CO/H2 (1:1). A mixture of 1 mL of 1-octene and 1 mL of decane was
added, and the CO/H2 pressure was brought to 50 bar. The mixture
was stirred for 20 h. The autoclave was cooled to 10 °C, and the
pressure was reduced to 1.8 bar. With this small overpressure, the liquid
was slowly removed from the catalyst with a 1.2-mm syringe. After
the catalyst was washed with 5 mL of toluene, 20 mL of toluene was
added and the pressure was brought to 20 bar. Finally, the mixture
was heated to 80 °C, and the second cycle was performed.
The 1H NMR titrations were carried out on an Inova 500 spectrometer
(500 MHz). For a typical experiment, stock solution A, containing 1.05
mM of 1 [14.692 mg of 1 in 10 mL of CD3Cl (dry and free of acid)],
and stock solution B, containing 2.90 mM of guest 3 and 1.05 mM of
1 (5.684 mg of 3 (or 4) in 5 mL of stock solution A), were prepared.
Twelve 5-mm NMR tubes were filled to a total volume of 0.6 mL by
the addition of different volumes of stock solutions A and B. All
titrations were performed at 298 K. Typically, the chemical shift of
the CH2 next to the amine of 1 and the urea protons was monitored,
and H NMR titration curves were analyzed with a fitting program
developed by Hunter et al., which details the equations given in the
paper.21
Determination of Adsorption Constants. The adsorption constants
were determined according to a procedure analogous to that described
previously.22 A stock solution of 0.97 mM of 4 in dichloromethane
1
Results and Discussion
(17) For silica-supported ruthenium and iron porphyrins, see: (a) Meunier, B.
Chem. ReV. 1992, 92, 1411-1456. (b) Battioni, P.; Cardin, E.; Louloudi,
M.; Schollhorn, B.; Spyroulias, G. A.; Mansuy, D.; Traylor, T. G. Chem.
Commun. 1996, 2037-2038. (c) Iamamoto, Y.; Ciuffi, K. J.; Sacco, H.
C.; Iwamoto, L. S.; Nascimento, O. R.; Prado, C. M. C. J. Mol. Catal. A
1997, 116, 405-420. (d) Ciuffi, K. J.; Sacco, H. C.; Valim, J. B.; Manso,
C. M. C. P.; Serra, O. A.; Nascimento, O. R.; Vidoto, E. A.; Iamamoto, Y.
J. Non-Cryst. Solids 1999, 247, 146-152. (e) Ciuffi, K. J.; Sacco, H. C.;
Biazzotto, J. C.; Vidoto, E. A.; Nascimento, O. R.; Leite, C. A. P.; Serra,
O. A.; Iamamoto, Y. J. Non-Cryst. Solids 2000, 273, 100-108. (f) Zhang,
R.; Yu, W. Y.; Wong, K. Y.; Che, C. M. J. Org. Chem. 2001, 66, 8145-
8153. (g) Zhang, J.; Liu, Y.; Che, C. Chem. Commun. 2002, 2906-2907.
(18) Bettelheim, A.; White, B. A.; Raybuck, S. A.; Murray, R. W. Inorg. Chem.
1987, 26, 1009-1017.
Synthesis of Binding Motifs and Guest Ligands. The
binding site previously used to anchor the catalyst to the
periphery of dendrimers comprises two urea groups and a basic
amine, and binding is based on four hydrogen bonds between
three urea groups and a proton transfer from the guest acid
function to the amine base of the dendrimer host. Since the
binding in this motif was compatible with the catalytic reaction
conditions and resulted in binding sufficiently strong to enable
catalyst recycling, we decided to immobilize a similar binding
motif on silica. For this purpose, we used the first generation
DBA dendrimer, which was reacted with 3-triethoxysilylpro-
pylisocyanate to give host 1 as a white powder in one step in
85% yield (Scheme 2).
(19) Bronger, R. P. J.; Kamer, P. C. J.; Reek, J. N. H.; van Leeuwen, P. W. N.
M. Organometallics 2003, 22, 5358-5369.
(20) (a) Deschler, U.; Kleinschmit, P.; Panster, P. Angew. Chem. 1986, 98, 237-
253. (b) Wieland, S.; Panster, P. Catal. Org. React. 1995, 62, 383-392.
(21) We have analyzed the titration curves with a fitting program developed by
Hunter et al.: Bisson, A. P.; Hunter, C. A.; Carlos, J. Chem.sEur. J. 1998,
4, 845-851.
(22) Jang, B. B.; Lee, K. P.; Min, D. H.; Suh, J. J. Am. Chem. Soc. 1998, 120,
12008-12016.
Porphyrin molecules are frequently used building blocks in
supramolecular chemistry, and we recently showed that as-
9
14560 J. AM. CHEM. SOC. VOL. 126, NO. 44, 2004

Supramolecular Anchoring of Catalysts
A R T I C L E S
Scheme 2. Synthesis of Host Molecule 1 in One Step from
Commercially Available Compounds
Scheme 4. Two Different Routes for Preparing Silica-Immobilized
Host Materials
From previous work in this area, we know that certain
bidentate phosphine ligands10b,10c lead to metal-ligand interac-
tions sufficiently strong to suppress metal leaching; so, we
decided to functionalize a xantphos analogue with the required
binding motifs. To this end, 4,5-dibromo-2-(5-bromopentyl)-
7-hexyl-9,9-dimethylxanthene19 was converted to the dilithio
derivative by treatment with n-BuLi, and subsequent reaction
with chlorodiphenylphosphine afforded the bromo derivative 5
(37% yield), which was further transferred into derivative 6 in
89% yield with liquid ammonia at elevated temperatures.
Reaction of 6 with the commercially available ethylisocyana-
toacetate yielded ester derivative 7 in 85% yield. After hydroly-
sis of the ester and recrystallization from chloroform, pure 8
was isolated in 49% yield as a white powder. Diphosphine guest
8 is thus prepared using four standard reactions (Scheme 3).
Immobilization of the Host Molecules of 1 and 2 on Silica.
Host molecule 1 was immobilized on silica by grafting it to
commercially available silica and by using the sol-gel tech-
nique20 (Scheme 4). Both methodologies are based on straight-
forward experimental procedures. Si-I was obtained after a
suspension of 1 (0.1 g) and silica (1.0 g) in toluene was refluxed
for 5 h, leading to quantitative grafting of the host material.
The sol-gel procedure involved the overnight stirring of a
solution of 1 and tetramethyl orthosilicate (TMOS) in THF/
H2O. The resulting gel was dried and crushed into free-flowing
silica, resulting in 1 being immobilized on a polysilicate support.
The surfaces of the silica materials were modified by reaction
with dimethoxydimethysilane in order to cap the free acidic
silanols. To this end, a suspension of either Si-I or Si-II was
refluxed for 2 h in toluene in the presence of dimethoxydi-
methylsilane.
semblies based on axial pyridine coordination to zinc porphyrins
are stable under catalytic conditions, as tested in various reac-
tions.16 In addition, this interaction is orthogonal to that of 1
since pyridines and amines do not bind to 1 and glycine-urea
functional groups have no significant interaction with 2. There-
fore, we decided to also investigate the use of such metal-lig-
and interactions as a means to noncovalently anchor transition-
metal catalysts. To this end, we immobilized zinc porphyrins
as binding sites on silica material, which could be used as non-
covalent support for phosphine-based transition-metal complexes
equipped with a nitrogen donor for the anchoring process.17 For
this purpose, we functionalized the para-tetraamino zinc(II)
porphyrin with trialkoxysilane groups using 3-triethoxysilyl-
propylisocyanate, giving building block 2 in 90% yield.18
Catalysts that can be noncovalently anchored to the binding
site of 1 are those functionalized with the complementary
binding motif (i.e., a glycine-urea group). We have prepared
phosphines 3 and 4 according to the procedures previously
reported.15c Note that guest phosphine 4 is functionalized with
an amide function instead of a urea group, and complexation
with the host material will be based on three hydrogen bonds.
In addition, the molecule contains an extra phenyl group to
improve solubility in apolar organic solvents.
Si-III was prepared by refluxing a suspension of Zn porphyrin
2 (0.15 g) and 1.0 g of commercially available silica in toluene
for 5 h. Again, the remaining free acidic silanols on the surfaces
of the silica were modified by reaction with dimethoxydi-
methysilane.
Binding of Guest Ligands to Silica Host Materials. The
binding of guest ligands 3 and 4 into the binding sites of 1 was
first studied in solution by NMR spectroscopy. Upon the
addition of 3, the protons of the urea groups of 1, as well as the
CH2 groups next to the amine, shifted downfield in a fashion
Scheme 3 . Synthesis of Glycine-Urea-Modified Xantphos Diphosphine 8 in Four Steps
9
J. AM. CHEM. SOC. VOL. 126, NO. 44, 2004 14561

A R T I C L E S
Chen et al.
Scheme 5. Palladium-Catalyzed Allylic Amination of Crotyl
similar to that previously observed, with the binding into the
dendrimers functionalized with the same binding sites.15 The
binding constants were determined by fitting the titration curves
and by assuming that K1 and K2 are not equal.21 A negative
cooperative effect was observed; for guest 3, we found that K2
(3 × 103 M-1) is an order of magnitude lower than K1 (8 ×
104 M-1), and for guest 4, a similar effect was observed (K1
and K2 are 3 × 104 and 3 × 102 M-1, respectively). This effect
is very likely due to the proximity of the two basic functions
of the host. As expected, the binding constant of 4 is lower
than that of 3 since only three hydrogen bonds are involved in
the binding process and the benzyl group might induce some
steric hindrance. In general, the binding of the first guest is
higher compared to that of the second one, and the exchange
of bound and free guests is fast on the NMR time scale under
the conditions applied (500 MHz, room temperature).
Acetate Using Piperidine as the Nucleophile
Scheme 6. Schematic Representation of the Control Experiment
that Proves Catalysis Takes Place at the Noncovalently
Functionalized Support
To compare the binding behavior of the guests in the silica-
immobilized binding sites with that of 1 in free solution, we
have measured the adsorption constant of 4 to Si-I.22 We
monitored the concentration of 4 in dichloromethane solution
using UV-vis spectroscopy in the presence of various amounts
of Si-I. From the analysis of the titration curve, we found that
17 × 10-5 (mol/g silica) binding sites for guest 4 are available,
with an adsorption constant of K ) 2 × 104 M-1. This is
consistent with the numbers from the experimental procedure,
which shows that we immobilized 9 × 10-5 mol of host 1 onto
the surface of 1 g of silica, which indicates that almost all of
the binding sites of 1 are accessible and able to bind two guests.
are situated in the narrow pores of the material. Therefore, we
decided to use Si-I for the subsequent catalytic reactions.
Compound 6 contains a primary amino group, which is a
functional group that can be used for noncovalent anchoring to
Zn porphyrin functionalized silica material. The binding constant
of 6 to Zn porphyrin 2 in toluene solution was determined via
titrations monitored by UV-vis spectroscopy and was found
to be in the expected range (Kass ) 2.2 × 104). The adsorption
constant of guest 6 to Si-III was determined in a way similar to
that described above and was found to be around 1.0 × 105.
The number of binding sites available (7.4 × 10-5 mol/g silica)
was, again, in line with that expected based on the experimental
procedure (8.7 × 10-5 mol of host 1 was grafted to the surface
of 1 g of silica), which indicates that nearly all binding sites
are accessible and able to bind to guest ligand 6.
We performed an additional experiment to determine the
binding efficiency with respect to the recycling procedures of
guest ligand 4 being anchored to Si-I or Si-II. A precise amount
of Si-I and 2 equiv of guest 4 were added to a dichloromethane
solution, and the suspension was stirred for 20 min. Subse-
quently, we filtered the solution and washed the silica with
dichloromethane (eight times) to simulate the recycling proce-
dure. The filtrate was concentrated to recover unbound guest
4. Next, the silica layer was washed with methanol several times,
and the filtrate was concentrated to afford guest 4, which
remained bound to Si-I during the previous “recycling proce-
dures”. We found that approximately 1 equiv of the guest was
bound sufficiently strong to the silica during the washing with
dichloromethane, which is in line with the difference in the
binding constant observed for the first and the second guest
molecule (K2 ) 300 M-1, which is insufficient). The guest
molecules that remained bound during the dichloromethane
washing procedure were recovered quantitatively after washing
with methanol, indicating that the binding is fully reversible
and controllable by choosing the proper solvent. When the ester
of guest 4, a ligand that does not significantly bind to the binding
site of host molecule 1, was subjected to the same experimental
procedure, we found that nothing remained on the silica during
the washing procedure with dichloromethane. In addition, guest
4 was found to have no interaction with the capped silica
material, indicating that the binding site is, indeed, essential in
these systems.
Catalysis. Allylic Amination Using a Palladium Catalyst
Immobilized via Hydrogen Bonds. We studied the Pd-
catalyzed allylic amination23,24 as a model reaction using crotyl
acetate and piperidine as the substrate molecules (Scheme 5)
and ligand 4 assembled to silica support Si-I. The ligand was
preassembled on the silica material, and when the slurry was
stirred in the presence of [Pd(crotyl)Cl]2, the silica material
turned yellow, indicating that the supported catalyst was formed
[(guest 4)2Pd(crotyl)(Cl)]-Si-I. After the slurry had been
washed with dichloromethane to remove the potentially unbound
catalyst, the reaction began by the addition of the substrates.
Samples were withdrawn from the reaction mixture and analyzed
by GC to determine the conversion and product distribution.
The first experiment was carried out to show that catalysis
takes place at the heterogeneous phase and not in the homo-
geneous phase (Scheme 6). To this end, a slurry reaction mixture
was stirred and monitored for 10 min, after which a part of the
homogeneous solution containing the substrate and the product
solvent was decanted. This mixture was transferred into another
Schlenk flask, and both the supernatant and the slurry residue
were stirred again under reaction conditions; the reaction
progress was further monitored. The results displayed in Figure
1A clearly show that the reaction continued only in the vessel
containing the heterogeneous phase. In a similar experiment
using the ester of ligand 4, a ligand that does not sufficiently
Similar experiments with Si-II showed that slightly less than
1 equiv remained on the silica during the washing process with
dichloromethane. Probably, part of the host molecules 1 become
inaccessible as a result of the sol-gel procedure20 because they
(23) Johannsen, M.; Jørgensen, K. A. Chem. ReV. 1998, 98, 1689-1708.
9
14562 J. AM. CHEM. SOC. VOL. 126, NO. 44, 2004

Supramolecular Anchoring of Catalysts
A R T I C L E S
Figure 1. Conversion in the allylic amination reaction using guest 4 supported on silica (A) as the ligand and the ester of 4 (B); see text and Scheme 6 for
explanation.
Table 1. Palladium-Catalyzed Allylic Amination Using the
Supported Ligand (4-Si-I) Studied in Four Consecutive Reactions
and Compared with the Homogeneous Systema
Scheme 7. Defunctionalization and Subsequent
Refunctionalization of Si-I
linear linear
time
conv
(%)
trans
(%)
cis
branched
(%)
entry
cycle
catalyst
(min)
(%)
1
2
3
4
5
6
HOM Pd(guest 4)2(crotyl)Cl
HOMb Pd(guest 4)2(crotyl)Cl
5
5
92.2 54.1 11.2
92.0 54.3 11.2
34.7
34.5
35.1
35.0
35.5
35.4
filtration step and could be recycled. In four consecutive
reactions, the selectivity did not change at all, while the activity
decreased slightly. Since there is no Pd leaching into the
homogeneous phase during the recycling process as determined
by ICP-AES (<0.1%), this decrease is attributed to the catalyst
decomposition, as is commonly observed for palladium catalysts.
1
2
3
4
(4-Si-I)2Pd(crotyl)Cl 30 90.5 53.2 11.7
(4-Si-I)2Pd(crotyl)Cl 30 85.6 54.5 10.5
(4-Si-I)2Pd(crotyl)Cl 30 84.9 53.4 11.1
(4-Si-I)2Pd(crotyl)Cl 30 72.8 53.7 10.9
a With 5 mL of CH2Cl2, [crotyl acetate] ) 0.2 M, [piperidine] ) 0.4 M,
ligand/Pd ) 2, [Pd] ) 0.002 M, room temperature. b Capped silica (250
mg) was added to the solution.
Hydroformylation Using Rhodium Catalysts Immobilized
via Hydrogen Bonds. To demonstrate the potential of the
concept of supramolecular anchoring, we decided to use the
same silica material as was used for the support for the rhodium-
catalyzed hydroformylation26 of 1-octene. To this end, the
palladium catalyst was removed from the Si-I surface by
washing it with methanol until the silica was completely white
and the eluent became colorless (Scheme 7). The absence of
palladium on the support was confirmed by ICP-AES analysis
of the silica material (<0.1%).
After the mixture was dried, the silica material was used as
a supramolecular support for the rhodium-catalyzed hydro-
formylation. Guest ligand 4, Rh(acac)(CO)2, 20 mL of toluene,
and the Si-I were added to a stainless steel 50-mL autoclave,
equipped with a mechanical stirrer and a substrate vessel. The
suspension was heated for 1 h at 80 °C under 20 bar CO/H2
(1:1) to allow catalyst formation before a mixture of 1 mL of
1-octene and 1 mL of decane (internal standard) was added,
and the CO/H2 pressure was raised to 50 bar. The mixture was
stirred for 20 h, after which the autoclave was cooled to 10 °C
and the pressure reduced to 1.8 bar. With this small overpressure,
the liquid was slowly removed from the supported catalyst via
a syringe while keeping the catalyst under a syngas atmosphere,
thereby suppressing catalyst decomposition. Subsequently, the
silica material was washed with 5 mL of toluene, making the
system ready for the next catalytic cycle. Using this procedure,
bind to the host molecule, the reactions in both vessels proceeded
with the same rate (Figure 1B). These experiments clearly show
that if the catalyst is noncovalently anchored to the binding sites
of the silica, the reaction indeed takes place at the heterogeneous
phase. In line with this, we observed that during the catalytic
experiment the silica remained yellow, whereas the organic
solvent was colorless. The homogeneous phase was analyzed
by atomic emission spectroscopy, confirming that this phase
indeed does not even contain traces of palladium (<0.1%).
We subsequently compared the catalytic properties of the
noncovalently anchored catalyst with the homogeneous catalysts,
and we studied the recycling properties of the systems (Table
1). In addition, a control experiment was carried out in which
the homogeneous reaction was performed in the presence of
capped SiO2 silica material (Table 1, entry 2). The yields and
the product distribution are the same as those of the homogen-
eous reaction (entry 1), indicating that the solid phase does not
interfere with the reaction. Moreover, the organic phase was
yellow, while the silica material remained white, clearly showing
that the capped silica does not adsorb the metal complex.
The noncovalently anchored catalyst was less active than the
homogeneous catalyst, as is commonly found for heterogenized
catalysts; 90% conversion was reached in 30 min (5 min for
the homogeneous reaction).25 The product distribution is the
same as that in the homogeneous phase, suggesting that the
catalytically active species is a similar palladium complex. The
noncovalently anchored catalyst was separated by a simple
(25) A control experiment showed that the reaction did not proceed in the absence
of the palladium catalyst.
(26) Rhodium Catalyzed Hydroformylation; van Leeuwen, P. W. N. M., Claver,
C., Eds.; Kluwer Academic Publishers: Dordrecht, The Netherlands, 2000.
(24) Trost, B. M.; van Vranken, D. L. Chem. ReV. 1996, 96, 395-422.
9
J. AM. CHEM. SOC. VOL. 126, NO. 44, 2004 14563

A R T I C L E S
Chen et al.
Table 2. Rhodium-Catalyzed Hydroformylation of 1-Octene Using
the Supported Ligand (4-Si-I) Studied in Eight Consecutive
Reactionsa
for 11 consecutive runs, no loss in either activity or selectivity
was observed and no rhodium was detected by ICP-AES
analysis of the organic phase (<0.5%). Compared to those of
the homogeneous system, the silica-anchored xantphos showed
a decrease in both activity and selectivity. Similar effects were
observed previously for covalently anchored Nixantphos on
silica (Rh-A).10c Interestingly, higher activity and selectivity
for the linear aldehyde are observed for the noncovalently
anchored ligand compared to those of the covalently anchored
Nixantphos, while in the homogeneous phase, these Nixantphos
and Xantphos ligands show similar activity and selectivity.28
The difference in spacer length might be the reason for this
slight improvement as the active site is located further from
the silica support. We previously reported the coexistence of
cationic and neutral rhodium species upon using covalently
anchored Nixantphos,10c which resulted in a hydrogenation side
reaction, which could be suppressed by adding a base or by
modifying the silica surface. The current system does not show
this behavior because the surface was modified in order to obtain
selective binding to the binding sites.
conv
(%)
linear
branched
isomers
(%)
cycle
aldehyde (%)
aldehyde (%)
l/b ratio
1
2
3
4
5
6
7
8
90.4
90.0
86.6
85.1
81.1
80.5
80.0
79.9
64.5
62.5
66.8
60.8
61.4
63.4
67.1
67.3
32.2
33.2
30.4
33.8
33.8
33.4
30.5
30.7
3.3
4.3
2.8
5.4
4.8
3.2
2.4
2.0
2.0
1.9
2.2
1.8
1.8
1.9
2.2
2.2
a With 1.2 g of Si-I, 0.01 mmol of Rh(acac)(CO)2, ligand/Rh ) 10, 1
mL of 1-octene, 1 mL of decane, and 20 mL of toluene; 80 °C, 50 bar of
CO/H2, 20 h.
Table 3. Rhodium-Catalyzed Hydroformylation of 1-Octene Using
the Supported Ligand (8-Si-I) Studied in 13 Consecutive
Reactions under Various Conditionsa
We decided to study the influence of syngas pressure on the
catalytic performance. A decrease in syngas pressure to 20 bar
resulted in a slightly higher turnover frequency up to a maximum
of 20.4 mol-1‚h-1 (aldehyde or rhodium). At this pressure, a
maximum in the l/b ratio was obtained, while the amount of
isomerization is not changed (Table 3, entry 9). These results
are in line with a partial negative order in p-CO and with a
zero order in p-H2, commonly found for Rh-diphosphine
catalyst systems.27 A further decrease of syngas pressure to 10
bar leads to both a decreased activity and a decreased regiose-
lectivity. The rate of isomerization is suddenly affected and
increases 3-fold to more than 30%. To test whether this was
due to catalyst decomposition, catalysis using the same condi-
tions of the first six catalyst experiments was performed (Table
3, entry 11). Similar results were obtained, indicating that no
decomposition had taken place.
At 100 °C, the activity of the catalyst was higher at the
expense of a small decrease in selectivity. Catalysis performed
at higher temperatures (120 °C) resulted in catalyst decomposi-
tion (Table 3, entry 13), as was clear from the large decrease in
selectivity (the high conversions of entries 12 and 13 do not
allow comparison of the TOF with other entries). In addition,
the product phase was yellow, and significant amounts of
rhodium in the product phase were detected by ICP-AES
analysis (1.6% of the total amount of rhodium was leached to
the organic phase).
linear
aldehyde
(%)
branched
aldehyde
(%)
conv
(%)
TOF
isomers
(%)
1
entry
cycle
(h-
)
l/b ratio
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
1
2
3
4
5
6
39.4
39.9
39.0
40.0
39.6
39.0
42.2
41.4
45.3
54.6
40.5
62.9
86.2
17.0
17.1
16.9
17.1
17.0
16.9
17.7
18.4
20.4
19.2
17.1
28.7
29.4
8
87.7
87.7
87.4
87.9
87.4
87.4
86.1
87.5
87.9
65.6
86.1
79.5
42.4
85.5
93.3
3.4
3.6
3.5
3.5
3.6
3.4
3.7
3.3
3.1
2.5
3.9
5.3
13.3
4.5
2.9
8.9
8.7
9.1
8.6
9.0
9.2
10.2
9.2
25.9
24.4
25.0
25.0
24.3
25.7
23.3
26.5
27.9
25.9
22.2
15.0
3.2
7b
8c
9d
9.0
10e
11f
12g
13h
31.9
10.0
15.2
44.3
9.9
Rh-Ai 24
Rh-Bi 19j
19
32
283
3.7
a With 1.2 g of Si-I, 0.01 mmol of Rh(acac)(CO)2, ligand/Rh ) 10, 1
mL of 1-octene, 1 mL of decane, and 20 mL of toluene; 80 °C, 50 bar of
CO/H2. b At 40 bar of CO/H2. c At 30 bar of CO/H2. d At 20 bar of CO/H2.
e At 10 bar of CO/H2. f At 50 bar of CO/H2. g At 100 °C. h At 120 °C. i Data
extracted from ref 10c. j After 2 h.
we performed eight consecutive reaction experiments. The
results are depicted in Table 2.
The chemo- and regioselectivities (linear/branched ratio (1/
b) ∼ 2) of the reaction are typical of bistriphenylphosphine-
based rhodium complexes.26 All reactions had 80-90% con-
version, and the activity of the catalyst was only modest (average
TOF ∼ 30 mol‚mol-1‚h-1). However, the results show clearly
that the catalyst can be recycled at least eight times with only
a slight loss in activity. This decrease in activity is a result of
small amounts of rhodium that leached from the ligand (Rh
leaching was determined by ICP-AES analysis) during the
recycling process, which has been observed previously for
systems based on covalently anchored monodentate phosphine
ligands applied in this reaction.10i We anticipated that the use
of bidentate ligands would suppress rhodium leaching and, in
addition, would afford systems that give rise to higher regiose-
lectivity. Indeed, when Xantphos-type ligand 8 was nonco-
valently anchored to Si-I, a rhodium catalyst that gave a high
regioselectivity (l/b ) 25, 87% linear aldehyde; see Table 3)
was obtained. Also, metal leaching was suppressed completely;
Allylic Alkylation Using Palladium Catalysts Immobilized
via Metal-Ligand Interactions. To investigate the use of
(27) (a) Lazzaroni, R.; Uccello-Barretta, G.; Benetti, M. Organometallics 1989,
8, 2323-2327. (b) Casey, C. P.; Petrovich, L. M. J. Am. Chem. Soc. 1995,
117, 6007-6014.
(28) van der Veen, L. A.; Keeven, P. H.; Schoemaker, G. C.; Reek, J. N. H.;
Kamer, P. C. J.; van Leeuwen, P. W. N. M.; Lutz, M.; Spek, A. L.
Organometallics 2000, 19, 872-883.
9
14564 J. AM. CHEM. SOC. VOL. 126, NO. 44, 2004

Supramolecular Anchoring of Catalysts
A R T I C L E S
Table 4. Palladium-Catalyzed Alkylation of Crotyl Acetate and Diethyl-2-methylmalonate Using the Supported Ligand (6-Si-III) Studied in
Three Consecutive Reactions and Compared with the Homogeneous Systema
cycle
catalyst
time (h)
conv (%)
linear trans (%)
linear cis (%)
branched (%)
(guest 6)2Pd(allyl)Cl
(guest 6)2Pd(allyl)Clb
(6-Si-III)2Pd(allyl)Cl
(6-Si-III)2Pd(allyl)Cl
(6-Si-III)2Pd(allyl)Cl
0.5
0.5
6
6
6
78.9
77.5
21.7
20.1
16.8
65.5
65.4
66.1
65.8
66.3
9.2
9.5
9.6
9.7
9.0
25.3
25.1
24.3
24.5
24.7
1
2
3
a With 5 mL of CH2Cl2, [Pd] ) 0.002 M, ligand/Pd ) 2, [crotyl acetate] ) 0.2 M, [diethyl-2-methylmalonate] ) 0.4 M, [BSA] ) 0.4 M, room temperature.
b In the presence of Zn tetraphenyl porphyrin 2.
Table 5. Rhodium-Catalyzed Hydroformylation of 1-Octene Using
metal-ligand interactions as a means for the noncovalent
the Supported Ligand (6-Si-III) Studied in Four Consecutive
Reactionsa
anchoring of catalysts, we studied ligand 6, attached to Si-III,
as the supported ligand for the palladium-catalyzed alkylation24
of crotyl acetate using the diethyl-2-methylmalonate anion as
the nucleophile. In a typical experiment, 5 µmol of [Pd(allyl)-
(Cl)]2, 20 µmol of ligand 6, and the proper amount of Si-III
were dissolved in 5 mL of dichloromethane, and the mixture
was stirred at room temperature for 15 min. Subsequently, the
silica material was washed to remove the nonbound catalyst,
and the reaction began by the addition of crotyl acetate, diethyl-
2-methylmalonate, BSA (N,O-bis(trimethylsilyl)acetamide), and
a catalytic amount of KOAc dissolved in 5 mL of dichlo-
romethane. The results of the catalytic reactions are shown in
Table 4. For comparison, we also performed the reaction with
the unsupported ligand in the absence and presence of zinc
porphyrin 2.25 This control experiment shows that in the
homogeneous phase, the assembly of the porphyrin does not
change the catalytic performance of the palladium catalyst; both
the selectivity and the activity are similar in the presence and
absence of 2. Comparing the noncovalently anchored catalyst
with the homogeneous analogue, we observed, as expected, a
small decrease in activity, but the selectivity remains the same
throughout all of the experiments. The catalyst was separated
by a simple filtration step and recycled. In three consecutive
reactions, the selectivity did not change at all, while the activity
decreased slightly, which again is attributed to catalyst decom-
position during the recycling process.
Similar to the experiments with the hydrogen-bonded system,
we decided to use the same silica material as the support for
the rhodium-catalyzed hydroformylation. Again, the support was
washed with methanol to remove all of the palladium complexes,
and we “uploaded” fresh ligand 6 to Si-III. The experimental
details are similar to those for the hydrogen-bonded system,
and the results are summarized in Table 5. Interestingly, the
activity of the supported catalyst is identical to that of the
hydrogen-bonded system, but the selectivity of the catalyst
system is slightly higher and more comparable to that of the
homogeneous analogue. Much to our surprise, we found that,
in contrast to the result with 8, the activity decreased upon
reusing the recycled catalyst. With the Xantene-based bidentate
diphosphine, we found previously10b,c that the metal-ligand
interaction is sufficiently strong to suppress rhodium leaching,
and the origin of the decrease must therefore be either leaching
of the whole complex or catalyst deactivation. Also, the results
conv
TOF
linear
branched
isomers
(%)
1
cycle
(%)
(h-
)
aldehyde (%)
aldehyde (%)
l/b ratio
1
2
3
4
69.1
59.6
53.4
44.3
16.8
14.2
13.7
10.7
94.2
93.9
93.2
93.8
2.2
2.4
2.3
2.5
3.6
3.7
4.5
3.7
43.2
40.0
40.5
36.7
a With 1.4 g of Si-III, 0.01 mmol of Rh(acac)(CO)2, ligand/Rh ) 10, 1
mL of 1-octene, 1 mL of decane, and 20 mL of toluene; 80 °C, 50 bar of
CO/H2, 20 h.
with ligand 8 show that the metal-ligand interaction is
sufficiently strong, and that the catalyst is very stable. The
binding of ligand 6 to the support is as equally strong as that of
ligand 8 to Si-I, and the equal activity observed for both systems
suggests that during the reaction, the ligand remains on the
support. We therefore reasoned that the product formed must
interfere with the binding process. From UV-vis titrations of
Zn(TPP) with butylamine performed in toluene in the presence
of nonanal (10%), the aldehyde produced during the hydro-
formylation reaction, we know that the product formed has no
effect on the binding strength (K ) 5 × 104 M-1 in the presence
and absence of aldehyde). Most likely, nonanal, formed during
the process, reacts with the primary amine of ligand guest 6
giving the corresponding imine, which results in a reduced
affinity for the zinc porphyrin.29 Indeed, in a control experiment.
we found that under hydroformylation conditions, 60% of
n-butylamine (1 equiv was added with respect to the ligand)
was converted into the corresponding imine,30 making this the
most plausible explanation for the drop in activity observed in
consecutive reactions.
Conclusions
We have introduced a new strategy for immobilizing transi-
tion-metal catalysts on silica supports that involves the use of
well-defined binding sites that are immobilized on silica
material. These binding sites can be reversibly functionalized
with homogeneous catalysts equipped with the complementary
(29) The binding constants of 1-butylamine and 1-butylimine with the Zn-
TPP complex are 5.4 × 104 and 1.9 × 103 M-1, respectively.
(30) A control hydroformylation experiment was carried out in the presence of
1 mL of 1-butylamine (and 1 mL of 1-octene, 1 × 10-5 mol Rh(acac)-
(CO), 1 × 10-4 mol Xantphos) at 80 °C and 50 bar of H2/CO for 20 h.
The reaction mixture was analyzed by GC-MS, showing the conversion
of 60% of the 1-butylamine into 1-butylnonylimine.
9
J. AM. CHEM. SOC. VOL. 126, NO. 44, 2004 14565

A R T I C L E S
Chen et al.
binding motif. The two binding motifs explored in the current
contribution are based on hydrogen bonds, in combination with
acid-base protonation, and on metal-ligand interactions. These
binding motifs are, in principle, orthogonal and can potentially
be used in one system. The host-guest binding behavior of the
silica-supported systems was found to correspond well with the
analogous systems in solution.
The supported transition-metal catalysts have been studied
in the palladium-catalyzed allylic substitution and the rhodium-
catalyzed hydroformylation. In general, the noncovalently
immobilized catalyst system shows an activity lower than that
of the homogeneous analogues, commonly observed for het-
erogenized catalysts, and importantly, the selectivity was
retained. Using the current noncovalent interactions, the catalyst
was bound to the support sufficiently strong to enable efficient
separation and reuse of the catalyst. Interestingly, the silica
material could be quantitatively defunctionalized by washing it
with methanol, and consequently, the support could be used in
different reactions (i.e., we first applied the systems in allylic
substitution and then in rhodium-catalyzed hydroformylation).
When noncovalently bound diphosphine 8 was used, the
rhodium catalyst was reused in 11 consecutive runs without any
noticeable deterioration of the catalyst. Employing the concept
of noncovalent anchoring clearly opens the way to modular
multipurpose systems since the catalyst and support can be
reused after separation. In this view, it is important to stress
that the catalysts used in this study can also be noncovalently
anchored to the dendritic materials previously reported.15c
Acknowledgment. We are grateful to the NRSC-C for
financial support, and L. Hoitinga is kindly acknowledged for
performing rhodium and palladium analyses.
JA046956H
9
14566 J. AM. CHEM. SOC. VOL. 126, NO. 44, 2004