Assembly of tunable supramolecular organometallic catalysts with cyclodextrins

Article abstract of DOI:10.1021/om1006215

Methylated-β-CDs were used to activate a catalyst by ligand solution trapping, to form second-sphere coordination by ligand complexation, and to increase the regioselectivity of biphasic and homogeneous hydroformylation reactions. Different methylation degrees of the β-CD allow assembly of hydrophobic or hydrophilic catalysts that can be employed in homogeneous or biphasic hydroformylation reactions.

Full text of DOI:10.1021/om1006215

3442 Organometallics 2010, 29, 3442–3449
DOI: 10.1021/om1006215
Assembly of Tunable Supramolecular Organometallic Catalysts
with Cyclodextrins
Loıc Leclercq† and Andreea R. Schmitzer*
ꢀ
ꢀ
ꢀ
Department of Chemistry, Universite de Montreal, C.P. 6128 Succursale Centre-ville, Montreal,
ꢀ
†
ꢀ
^
Quebec, H3C 3J7, Canada. Present address: UFR of Chemistry, Universite Lille 1, Bat. C6,
F-59655 Villeneuve d’Ascq Cedex, France.
Received June 28, 2010
Methylated-β-CDs were used to activate a catalyst by ligand solution trapping, to form second-
sphere coordination by ligand complexation, and to increase the regioselectivity of biphasic and
homogeneous hydroformylation reactions. Different methylation degrees of the β-CD allow
assembly of hydrophobic or hydrophilic catalysts that can be employed in homogeneous or biphasic
hydroformylation reactions.
Introduction
these equilibria simply by assembling host-guest complexes.
Therefore, Monflier et al. have reported the self-assembly of
a bidentate ligand formed by the complexation of an amino-
CD and a phosphane.6 The application of this catalyst was
limited to aqueous homogeneous or biphasic catalysis in
organic solvents. Considering the lack of flexibility of cur-
rent supramolecular ligands, we applied our experience with
CDs and their use in aqueous and organic solutions7 to
design water- or organic-soluble supramolecular catalytic
systems. Our approach relies on the self-assembly of di-
or trimethylated-β-CD (β-DIME or β-TRIME) and di(1-
adamantyl)benzylphosphine (DABP) (Scheme 2). The choice
of β-DIME or β-TRIME was dictated by their hydrophilic
and hydrophobic properties, respectively.8 The self-assembly
of supramolecular complexes between the DABP metal-
ligand and these CDs modifies the hydrophobic or hydrophilic
properties of the resulting supramolecular organometallic
catalyst. We present here the characterization and the applica-
tion of these complexes in homogeneous and biphasic catalytic
hydroformylation processes.
Over the last three decades, several supramolecular ap-
proaches have been developed in the field of organometallic
catalysis, where host-guest interactions have proved to be a
powerful tool to elaborate new strategies or to improve the
performances of various catalytic entities.1 The design of
supramolecular ligands is an elegant strategy to stabilize
organometallic complexes under catalytic conditions and
represents a potential route to produce a library of ligands.2
A straightforward way to assemble a supramolecular cata-
lytic entity is to generate second-sphere coordination by the
use of macrocycles such as cyclodextrins (CDs) (Scheme 1).3
The resulting supramolecular assembly can be used to
modify the selectivity of the initial catalyst. However, it is
difficult to obtain stable, flexible, and active catalytic sys-
tems by these means. The triphenylphosphine trisulfonated
sodium salt is known to readily form inclusion complexes
with randomly dimethylated-β-CD.4 The application of this
complex in hydroformylation reactions was previously re-
ported, but it resulted in a decrease of the regioselectivity
of the reaction.4 The formation of the inclusion complex
favors the formation of low-coordinated rhodium phosphine
species.5 In general it is difficult to avoid the modification of
Results and Discussion
Characterization of DABP/Methylated-β-CD Complexes.
i. Inclusion Complexes in Dichloromethane. First, DABP
complexation by methylated-β-CDs was investigated by
*To whom correspondence should be addressed. E-mail: ar.
schmitzer@umontreal.ca.
(1) Monflier, E.; Hapiot, F.; O’Hare, D. In Comprehensive Organo-
metallic Chemistry III, Vol. 12; Crabtree, R. H.; Mingos, D. M. P., Eds.;
Elsevier: Oxford, 2006; pp 781-834.
(2) (a) Wilkinson, M. J.; van Leeuwen, P. W. N. M.; Reek, J. N. H.
Org. Biomol. Chem. 2005, 3, 2371–2383. (b) Sandee, A. J.; Reek, J. N. H.
Dalton Trans. 2006, 3385–3391.
(6) (a) Machut, C.; Patrigeon, J.; Tilloy, S.; Bricout, H.; Hapiot, F.;
Monflier, E. Angew. Chem, Int. Ed. 2007, 46, 3040–3042. ; Angew. Chem.
2007, 119, 3100-3102. (b) Legrand, F. X.; Hapiot, F.; Tilloy, S.;
Guerriero, A.; Peruzzini, M.; Gonsalvi, L.; Monflier, E. Appl. Catal.
A: Gen. 2009, 362, 62–66. (c) Ferreria, M.; Bricout, H.; Sayede, A.; Hapiot,
F.; Tilloy, S.; Monflier, E. ChemSusChem 2008, 1, 631–636.
(3) (a) Reek, J. N. H. In Supramolecular Catalysis; Van Leeuwen, P. W.
N. M., Ed.; Wiley, VCH: Weinheim, 2008; pp 199-234. (b) Hapiot, F.;
Tilloy, S.; Monflier, E. Chem. Rev. 2006, 106, 767–781.
(4) (a) Monflier, E.; Fremy, G.; Castanet, Y.; Mortreux, A. Angew.
Chem., Int. Ed. Engl. 1995, 34, 2269–2271. ; Angew. Chem. 1995, 107,
2450-2452. (b) Dessoudeix, M.; Urrutigoıty, M.; Kalck, P. Eur. J.
Inorg. Chem. 2001, 7, 1797–1800. (c) Leclercq, L.; Sauthier, M.; Castanet,
Y.; Mortreux, A.; Bricout, H.; Monflier, E. Adv. Synth. Catal. 2005, 347, 55–
59.
(7) (a) Leclercq, L.; Lacour, M.; Sanon, S. H.; Schmitzer, A. R.
Chem.;Eur. J. 2009, 15, 6327–6331. (b) Leclercq, L.; Schmitzer, A. R. J.
Phys. Org. Chem. 2009, 22, 91–95. (c) Leclercq, L.; Schmitzer, A. R. J.
Phys. Chem. B 2008, 112, 11064–11070. (d) Leclercq, L.; Noujeim, N.;
Sanon, S. H.; Schmitzer, A. R. J. Phys. Chem. B 2008, 112, 14176–14184.
(e) Noujeim, N.; Leclercq, L.; Schmitzer, A. R. J. Org. Chem. 2008, 73,
3784–3790.
(8) (a) Szejtli, J. Cyclodextrin Technology; Kluwer: Dordrecht, 1988; p
€
51. (b) Immel, S.; Lichtenthaler, F. W. Starke 1996, 48, 225–232. (c)
(5) Monflier, E.; Bricout, H.; Hapiot, F.; Tilloy, S.; Aghmiz, A.;
Leclercq, L.; Bricout, H.; Tilloy, S.; Monflier, E. J. Colloid Interface Sci.
2007, 307, 481–487.
ꢀ
Masdeu-Bulto, A. M. Adv. Synth. Catal. 2004, 346, 425–431.
r
pubs.acs.org/Organometallics Published on Web 07/12/2010
2010 American Chemical Society

Article
Scheme 1. Concept of Supramolecular Catalyst Assembly
by Ligand (L)/Cyclodextrin (CD) Complexation
Organometallics, Vol. 29, No. 15, 2010 3443
Figure 1. H(e) and H(g) of DABP NMR titration profile obtained
at 25 °C for addition of (a) β-TRIME and (b) β-DIME in CD2Cl2
solution of DABP (5 mM).
Scheme 2. Methylated Cyclodextrins and Phosphorous Ligand
Used in This Work and Their Schematic Representation
NMR spectroscopy in deuterated dichloromethane. 1H
NMR titrations of the adamantyl protons showed that 2:1
complexes were formed (Figure 1). It is important to note
that no significant shifts were observed for the benzyl pro-
tons. The binding constants were estimated at 5900 ( 40 M-2
for the β-DIME and 10 100 ( 110 M-2 for the β-TRIME.
These values are weak compared to adamantyl inclusion in
aqueous solution.9
To gain more insight into the geometry of the host-guest
complexes, we performed a 2D NMR study (Figure 2).
ROESY experiments are well suited for this purpose;
although the COSY peaks were intense at the concentrations
used, it was possible to assign the NOE signals between the
˚
DABP and the CDs. The distance must be less than 5 A
to observe cross-peaks between protons. It is noteworthy
that no cross-peaks were detected between the phosphane
aromatic residue and the CDs, confirming the absence of
benzyl inclusion in all cases (see Supporting Information
(SI), Figure S1). For the adamantyl residues, the principal
cross-peaks were observed between the H(e), H(f), and H(g)
protons and the C(2) and/or the C(3) methoxy of the CD
(and C(2), C(3), and C(6) in the case of β-TRIME). This
indicates that the inclusion complexes in dichloromethane
are formed by the inclusion of the adamantyl residues of
DABP into the CDs hydrophobic cavity with insertion at the
wide rim. However, the adamantyl residues are not deeply
included into the CD cavity, probably due to the steric
hindrance of DABP. To confirm this assumption, we per-
formed 31P NMR experiments, and no significant changes
Figure 2. Partial ROESY NMR at 25 °C of (a) β-DIME and
DABP and (b) β-TRIME and DABP in CD2Cl2 (5 mM CD and
10 mM DABP).
were observed in the presence of the CD (see SI, Figure S2).
Thus the CD is positioned sufficiently far from the phos-
phorus atom.
In order to assess the energy content for the various inclu-
sion complexes formed in dichloromethane and to support the
geometries deduced from the ROESY NMR experiments,
semiempirical quantum calculations were undertaken.10 The
COSMO (conductor-like screening model) method11 was used
to approximate the effect of the dichloromethane solvent
(10) Stewart, J. J. P. Stewart Computational Chemistry, Version
7.213W, http://openmopac.net/.
€€
(11) Klamt, A.; Schuumann, G. J. Chem. Soc., Perkin Trans. 2 1993,
799–805.
(9) (a) Connors, K. A. Chem. Rev. 1997, 97, 1325–1358. (b) Rekharsky,
M. V.; Inoue, Y. Chem. Rev. 1998, 98, 1875–1918.

3444 Organometallics, Vol. 29, No. 15, 2010
Leclercq and Schmitzer
Figure 3. Geometry and enthalpy variation obtained by semiempirical calculations of the inclusion complexes of DABP and β-DIME
(top) and β-TRIME (bottom).
Figure 4. 1H NMR spectra in D2O at 25 °C of (a) β-DIME
(2 mM) and (b) residual aqueous phase of a DABP (1 mM) and
β-DIME (2 mM) in 1:1 CD2Cl2/D2O mixture.
Figure 5. Partial ROESY NMR at 25 °C of the residual
aqueous phase of a DABP (1 mM) and β-DIME (2 mM) in
1:1 CD2Cl2/D2O mixture.
surrounding the ligands and the complexes. Enthalpy changes
(ΔH) determine whether the complexation can take place
spontaneously.12 The formation of all inclusion complexes
complexation was fast on the NMR time scale, the observed
was exothermic (Figure 3). It is noteworthy that the predic-
chemical shifts were the average of the chemical shifts for all
ted geometries for the various complexes are in good agree-
the complexes present at equilibrium.
ment with the geometries deduced from the ROESY NMR
experiment.
ii. Inclusion Complexes in Water. To confirm the forma-
To gain more insights into the molecular interactions and
the geometry of the host-guest complex formed in water, we
performed a 2D ROESY NMR study. The ROESY spec-
trum of the residual aqueous phase in D2O gave intense
cross-peaks between the protons of adamantyl groups and
tion of the inclusion complex between DABP and β-DIME
in aqueous solution, we studied the residual aqueous phase
of a DABP (1 mM) and β-DIME (2 mM) mixture in CD2Cl2/
the internal H(2)-H(6) protons and the C(2) methoxy pro-
D2O (1:1). In the absence of β-DIME, no NMR signals were
tons of the β-DIME (Figure 5). These cross-peaks provide
obtained in the aqueous phase, confirming that DABP is
evidence of the formation of an inclusion complex between
completely insoluble in D2O. The formation of β-DIME/
the β-DIME and DABP in aqueous solution. Moreover,
the absence of cross-peaks between the hydrogens of the
DABP inclusion complexes was observed by 1H NMR,
where characteristic chemical shifts of the DABP complex
β-DIME and those of the benzyl residue (not shown) showed
were obtained (Figure 4). All chemical shifts of the β-DIME
that the inclusion of the DABP into the β-DIME occurs only
in the residual aqueous phase showed significant differences
by the adamantyl groups. It is noteworthy that similar
compared to those of the β-DIME alone. In particular, both
conclusions can be made with β-TRIME.
To estimate the association constants in aqueous solution,
internal H(3) and H(5) protons of the β-DIME underwent
we used the phase-solubility method.14 Phase-solubility
upfield shifts due to the inclusion of the adamantyl resi-
dues in the cavity. These results indicate the formation of
inclusion complexes between DABP and β-DIME.13 Since
analysis involves an examination of the effect of the CD on
the water solubility of DABP. Experimentally, the DABP
was added to several vials in a way to always be in excess. The
(12) Smith, M. B.; March, J. Mechanisms and Methods of Determin-
presence of solid DABP in these systems is necessary to keep
ing Them. In March’s Advanced Organic Chemistry, 6th ed.; John Wiley &
Sons: New York, 2007.
(13) Schneider, H. J.; Hacket, F.; Rudiger, V.; Ikeda, H. Chem. Rev.
1998, 98, 1755–1786.
€
(14) Higuchi, T.; Connors, K. A. Adv. Anal. Chem. Instrum. 1965, 4,
117–212.

Article
Organometallics, Vol. 29, No. 15, 2010 3445
Table 1. Effect of Methylated-β-CDs on DABP Partition
Coefficient (Porg/aqu) Estimated from UV Absorbance at 265 nma
b
entry
CD
CD/DABP
Porg/aqu
1
2
3
4
β-DIME
β-DIME
β-TRIME
β-TRIME
1
2
1
2
0.8
0.2
1.5
25.7
a Cary Bio 100 UV-visible spectrophotometer, [DABP] = 1 mM,
[CD] = 1 or 2 mM, 25 or 50 °C. b Calculated with eq 5.
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 þ 4K½CDꢁT - 8K½CDꢁT þ 1
½CD2Gꢁ ¼
ð3Þ
8K
[CD2G] can be calculated from eq 3 assuming a value of K.
For each [CD2G] value, the molar absorption coefficient, ε,
was calculated with eq 4.
Figure 6. Phase-solubility diagram of DABP/CD for ())
β-DIME and (() β-TRIME recorded at 265 nm and 25 °C.
The black line represents the calculated relative absorbance
obtained from eqs 3 and 4 (β-DIME: K = 7300 ( 100 M-1
,
ε = 21 790 ( 145 L mol-1 cm-1, R2 = 0.9999; β-TRIME: K =
4700 ( 50 M-1, ε = 43 980 ( 820 L mol-1 cm-1, R2 = 0.9998).
A ¼ εl½CD2Gꢁ
ð4Þ
where l is the cell length and A is the relative absorbance. This
algorithm provides estimation of the association constant, K,
as well as minimizes the dispersion of ε. The estimated values
were K = 7300 ( 100 M-1, ε = 21 790 ( 145 L mol-1 cm-1
for β-DIME and K = 4700 ( 50 M-1, ε = 43 980 ( 820 L
mol-1 cm-1 for β-TRIME. It is noteworthy that an increase
of the molar absorption coefficient ε was observed compar-
ing the values obtained for the β-DIME and β-TRIME.
Similar changes were observed when guest molecules were
dissolved in less polar solvents, suggesting that the guest is
incorporated in a more hydrophobic environment.16 There-
fore we can conclude that β-TRIME provides a more
hydrophobic environment to DABP than β-DIME.
The energy content for the inclusion complexes in water
was also studied by semiempirical quantum calculations,
similar to those undertaken in dichloromethane. The for-
mation of 1:1 and 1:2 inclusion complexes was exothermic,
-22 kcal/mol and -28 kcal/mol for the β-DIME and
-27 kcal/mol and -32 kcal/mol in the case of the β-TRIME.
As in dichloromethane, the 1:2 complexes are more stable
and favored in solution.
iii. Partition of the Complex between Dichloromethane and
Water. Complementary experiments were performed by UV
spectroscopy measurements. To a CH2Cl2 solution of DABP
ligand was added each of the two methylated-β-CDs (1 or
2 equiv). After water addition and vigorous stirring at 25 °C
(24 h), the UV spectra of the dichloromethane and the
aqueous phase were recorded. In the absence of CDs, no
absorbance was observed in the aqueous phase, suggesting
that the DABP was absent in water. The CD/DABP complex
partition coefficients (Porg/aqu), between the organic and the
aqueous phase, were calculated from eq 5.
the thermodynamic activity at a constant level. To the DABP
(G) was added a constant volume of water containing
progressively larger concentrations of the CD. The vials were
mixed at 25 °C until equilibrium was established (1 week).
The solid DABP was then removed, and the aqueous solu-
tion was studied by UV-visible spectroscopy. A phase-
solubility diagram was constructed by plotting the rela-
tive absorbance (obtained by multiplying the absorbance
of diluted solutions by the dilution factor) and the total
concentration of added CD, [CD]T (Figure 6).
Mass spectrometry was used to determine the stoichio-
metry of the DABP/CD complex. Electrospray ionization
(ESI) is used to “fish” loosely bonded supramolecular com-
plexes in solution and to transfer them to a mass spectro-
meter to investigate their assemblies. The positive ions are
formed in solution and then transferred by ESI directly to the
gas phase. ESI is characterized by the gentleness by which the
gaseous ions are formed, and loosely bonded supramolecules
can be observed, such as hydrogen-bonded amino-acid
assemblies.15 The ESI mass spectra, in the positive ion mode,
of DABP/CD solutions prepared by the phase-solubility
technique at 3 mM, give m/z peaks at 3083.5 corresponding
to (DABP/2 ꢀ β-DIME)þ and respectively at 3280.7 corre-
sponding to (DABP/2 ꢀ β-TRIME)þ. On the basis of this
results and the complete water insolubility of DABP in the
absence of the CD, the complexation equilibrium with CD is
dictated by a 2:1 equilibrium:
½CD2Gꢁ
2CD þ GuCD2G K ¼
ð1Þ
2
½CDꢁ
With respect to this model, the mass balance for CD provides
eq 2.
Abs
DABPorg
Porg=aqu
¼
ꢀ 100
ð5Þ
Abs
DABPaqu
½CDꢁT ¼ ½CDꢁ þ 2½CD2Gꢁ
ð2Þ
As shown in Table 1, the supramolecular complex was
prevalent in the aqueous phase for the β-DIME and in the
organic phase for the β-TRIME. However a clear influence
of the CD/DABP ratio is observed: (i) the 2:1 complex
The free CD concentration, [CD], can be calculated from eq 1
and substituted in the association constants to obtain eq 3.
(15) (a) Koch, K. J.; Gozzo, F. C.; Nanita, S. C.; Takats, Z.; Eberlin,
M. N.; Cooks, R. G. Angew. Chem., Int. Ed. 2002, 41, 1721–1724. (b)
Takats, Z.; Nanita, S. C.; Cooks, R. G. Angew. Chem., Int. Ed. 2003, 42,
3521–3523. (c) Cooks, R. G.; Zhang, D. X.; Koch, K. J.; Gozzo, F. C.;
Eberlin, M. N. Anal. Chem. 2001, 73, 3646–3655.
(16) (a) Bender, M. L.; Komiyama, M. Cyclodextrin Chemistry;
Springer-Verlag: Berlin, 1978. (b) Miyake, K; Irie, T; Arima, H.; Hirayama,
F.; Uekama, H.; Hirano, M.; Okamoto, Y. Int. J. Pharm. 1999, 179, 237–245.

3446 Organometallics, Vol. 29, No. 15, 2010
Leclercq and Schmitzer
Scheme 3. Rhodium Hydroformylation Reaction: (a) Homo-
geneous; (b) Biphasic Reaction of Hydrophilic Substrate (1);
(c) Biphasic Reaction of Hydrophobic Substrate (2)
Figure 7. Tolman angle for optimized structures (PM3/COS-
˚
MO, MOPAC2009) of (a) TPP (140 A), (b) DABP (180 A), and
˚
˚
(c) β-TRIME/DABP (230 A).
presence of β-DIME or β-TRIME are summarized in Table 2.
A conversion of <1% was obtained for 1 in the absence of
CD, whereas conversions of 42% and 33% were obtained
with 1 equiv of β-DIME and β-TRIME respectively. In the
same way, 2 gave also a low aldehyde conversion (<1%) in
the absence of CD and rather higher conversions with 1 equiv
of β-DIME or β-TRIME (9% and 25%, respectively). The
poor activity observed in the case of Rh(acac)(DABP)2 alone
is probably due to the steric hindrance of the DABP ligand,
where the metal site is not accessible to ensure a catalytic
process (see above). The increase of the reaction conversion
in the presence of CD is probably due to the solution
trapping of DABP by the CD, inducing the formation of a
more active species, i.e., a low-coordinated rhodium phos-
phine catalyst. From this point of view, the CD is a catalyst
promoter.5,18 Moreover, the hydrophobic or hydrophilic
character of the methylated-β-CD has an influence on the
conversion in a given solvent. For example, in the case of
aqueous hydroformylation, the use of β-TRIME decreased
conversion compared to β-DIME. We propose that a com-
plex is formed between the catalytic species and the CD via
the DABP ligand. The complex formed with the β-TRIME
is more hydrophobic than the complex formed with the
β-DIME, therefore the coordination of 1 is less favored in
the case of β-TRIME due to its partition in the organic phase
(Porg/aqu = 25.7). It is noteworthy that these observations
and explanations are inverted for organic homogeneous
hydroformylation (Table 2).
Table 2. Homogeneous Hydroformylation of Allyl Alcohol (1) in
Water and 1-Octene (2) in Dichloromethanea
entry
CD
CD/Rh substrate solv convb selectivityc l/bd
1
2
3
4
5
6
7
8
9
1
1
1
1
1
2
2
2
2
2
H2O
H2O
H2O
H2O
H2O
<1
42
76
33
23
β-DIME
β-DIME
β-TRIME
β-TRIME
1
2
1
2
91
95
67
60
3.3
4.3
3.6
4.5
CH2Cl2 <1
β-DIME
β-DIME
β-TRIME
1
2
1
2
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
9
15
25
59
89
85
93
89
3.9
4.1
3.7
4.4
10 β-TRIME
a Rh(acac)(DABP)2 (5.0 ꢀ 10-3 mmol), CD (5.0 ꢀ 10-3 or 10-2
mmol), olefin (1.5 mmol), solvent, 70 °C, CO/H2 (1:1): 45 atm, 12 h.
b Conversion. c Selectivity = (mol of aldehyde)/(mol of converted olefin) ꢀ
100. The formation of hemiacetals was not observed for 1. d Ratio of linear
to branched aldehydes.
formed with β-DIME is more hydrosoluble than the 1:1
complex, and (ii) the opposite is observed for β-TRIME.
Therefore, the solubility of the CD/DABP complex can be
controlled by the choice of CD and the CD/DABP ratio. In
other words, the hydrophobic or hydrophilic properties of
methylated-β-CDs can be extended to their supramolecular
complexes with DABP.
Catalytic Investigations. Methylated-β-CDs in combina-
tion with DABP are good candidates for application to
hydroformylation reaction for several reasons: (a) CD/
DABP complexes can be formed with adamantyl residues
more easily than with olefins,17 (b) the phosphorus atom
remains available for metal coordination, and (c) supramo-
lecular ligand solubility can be varied by the choice of CD.
Thus, flexible hydroformylation processes can be developed
depending on the polarity of the solvent and the substrate
(Scheme 3).
The formation of the complex between the CD and
catalytic species is also corroborated by the increase of the
regioselectivity. The increase of the linear/branched (l/b)
aldehyde ratio with the higher CD/Rh ratio is due to the
bulkiness around the rhodium coordination sphere. In other
words, steric hindrance in the supramolecular ligand around
the phosphorus atom leads to wider angles, enhancing the
steric effects and increasing the regioselectivity.19 Therefore,
First, the performances of CDs/Rh(acac)(DABP)2 cou-
ples were evaluated in the homogeneous hydroformylation
of allyl alcohol 1 (in water) and 1-octene 2 (in dichloro-
methane). The results obtained in the absence and in the
(18) Binkowski-Machut, C.; Canipelle, M.; Bricout, H.; Tilloy, S.;
Hapiot, F.; Monflier, E. Eur. J. Inorg. Chem. 2006, 1611–1619.
(19) (a) Leclercq, L.; Hapiot, F.; Tilloy, S.; Ramkisoensing, K.; Reek,
J. N. H.; van Leeuwen, P. W. N. M.; Monflier, E. Organometallics 2005,
24, 2070–2075. (b) Mingos, D. M. P. In Modern Coordination Chemistry:
The Legacy of Joseph Chatt; Leigh, G. J.; Winterton, N., Eds.; The Royal
Society of Chemistry: Cambridge, 2002; pp 69-78.
(17) Hapiot, F.; Leclercq, L.; Azaroual, N.; Fourmentin, S.; Tilloy,
S.; Monflier, E. Curr. Org. Synth. 2008, 15, 162–172.

Article
Organometallics, Vol. 29, No. 15, 2010 3447
Scheme 4. Schematic Representation of the Proposed Reaction Pathway of the Rhodium Hydroformylation Reaction with DABP Ligand
with or without CDs
Table 3. Biphasic Hydroformylation of Allyl Alcohol 1 in Di-
chloromethane and 1-Octene 2 in Watera
the major roles of CDs are trapping the DABP in solution
(CDs promote the catalytic species) and complexing the
DABP ligand (l/b increase for 2 equiv of CD). To evaluate
steric hindrance in the free ligand and in the supramolecular
complex, we compared the Tolman cone angles calculated
from semiempirical PM3 calculations (Figure 7).
entry
CD
substrate CD/Rh solvent convb selectivityc l/bd
1
2
3
4
5
6
2
2
2
1
1
1
H2O
H2O
H2O
CH2Cl2
CH2Cl2
CH2Cl2
<1
39
57
7
54
79
β-DIME
β-DIME
1
2
86
90
79
89
95
3.6
3.9
3.4
3.8
4.1
As a control experiment of the computational approach,
we determined the Tolman angle for the well-known triphe-
nylphosphine ligand (TPP). The obtained angle was 140°,
which is in good agreement with the literature value (145°).20
The obtained Tolman angles were 180° for DABP, 210° for
the β-DIME/DABP (1:1), and 230° for the β-TRIME/DABP
(1:1) complex. The increased cone angle can explain the l/b
ratio increase and the decrease of the conversion in the
hydroformylation reactions performed with these com-
plexes, compared to the classical TPP ligand.4c Specifically,
the steric hindrance related to the complexation of the DABP
phosphine by the CD favors the hydroformylation reaction
on the terminal alkenyl carbon. It is noteworthy that the
chemoselectivity is higher in the presence of the CD (near
90%, except for the β-TRIME with 1), probably due to an
inhibition of the β-H elimination reaction, again due to steric
hindrance in the supramolecular ligand.19 The homogeneous
catalysis results suggest that (i) the native Rh/DABP orga-
nometallic complex is inactive, (ii) the presence of 1 equiv of
CD activates the catalyst, and (iii) the presence of 2 equiv of
CD increases the regioselectivity. On the basis of these
observations, we propose the following schematic reaction
pathway (Scheme 4).20 In the presence of carbon monoxide
and hydrogen, the square-planar HRh(CO)(DABP)2 inter-
mediate is formed (a). The phosphine brings a large steric
hindrance in the complex; therefore the olefin cannot be
coordinated (b). In the presence of 1 equiv of CD, a phos-
phine is trapped in solution (c0); therefore the formed new
complex (c) allows olefin coordination and insertion (e). In
the presence of 2 equiv of CD, the organometallic species is
more congested, due to the steric hindrance of the CD/
DABP supramolecular ligand (d). The olefin coordination
occurs in this case, but the steric hindrance favors linear
insertion (f).
β-TRIME
β-TRIME
1
2
a Rh(acac)(DABP)2 (5.0 ꢀ 10-3 mmol), CD (5.0 ꢀ 10-3 or 10-2
mmol), olefin (1.5 mmol), H2O and CH2Cl2 (1 mL), 70 °C, CO/H2 (1:1):
45 atm, 12 h. b Conversion. c Selectivity = (mol of aldehyde)/(mol of
converted olefin) ꢀ 100. The formation of hemiacetals was not observed
for 1. d Ratio of linear to branched aldehydes.
Knowing that the interaction between DABP and
β-CD derivatives can be beneficially used in homogeneous
hydroformylation processes (catalyst activation, increase
of the regioselectivity, good catalytic activity, and high
chemoselectivity), further catalytic experiments were per-
formed with these CDs under biphasic hydroformylation
conditions (Table 3). The results described in Table 2 were
used to set the parameters of the biphasic experiments:
β-DIME and β-TRIME were used to maintain the cata-
lytic species in the aqueous or in the organic phase,
respectively. As can be seen in Table 3, the formation of
the CD/DABP inclusion complex was used to activate
the catalytic species (entries 1-3 and 4-6). The higher
CD/Rh ratio again affected the steric hindrance around
the phosphorus atom, forcing the hydroformylation re-
action to preferentially form the linear aldehyde. Finally,
all chemoselectivities were near 90% in these biphasic
reactions.
Conclusion
In conclusion, we have demonstrated that methylated-
β-CDs can be used in hydroformylation reactions as cata-
lytic promoters, as well as to increase the regioselectivity
of the reaction. Furthermore, the combination of these
CDs with a hydroformylation catalyst can also be used
to underpin versatile processes. Indeed, the judicious
choice of the methyl-β-CD allows obtaining a hydrophobic
or hydrophilic catalyst that can be used in aqueous or
organic homogeneous catalysis or in biphasic processes.
(20) van Leeuween, P. W. N. M.; Casey, C. P.; Whiteker, G. T. In
Rhodium Catalyzed Hydroformylation; van Leeuween, P. W. N. M.; Claver,
C., Eds.; Kluwer: Dordrecht, 2000; pp 64-66.

3448 Organometallics, Vol. 29, No. 15, 2010
Leclercq and Schmitzer
The versatility and the ease of the assembly of hydrophobic
or hydrophilic catalysts to obtain tunable catalytic pro-
cesses provide a valuable element for further developments
in organometallic catalysis. The extension of this system
to further organometallic catalysis is under study in our
group.
water batch. These solutions were decanted, and the absorbance
of dichloromethane (presented in the SI Figure S3, for example)
and in the aqueous layer was determined by measuring the
absorbance at 265 nm. UV spectroscopy was performed on
a Cary Bio 100 UV-visible spectrophotometer. In the absence
of CDs, no absorbance was observed for the aqueous
phase, showing the absence of DABP in water. Therefore,
the CD/DABP complex partition coefficients (Porg/aqu), bet-
ween the organic and the aqueous phase, have been calculated
from eq 5.
Experimental Section
Materials and General Information. Dicarbonylacetylaceto-
natorhodium(I) and organic compounds (2,6-dimethyl-β-cyclo-
dextrin (β-DIME), undecane, 4-allyl-1,2-dimethoxybenzene,
styrene) were purchased from Aldrich Chemicals in their highest
purity and used without further purification. 2,3,6-Trimethyl-
β-cyclodextrin (β-TRIME) was purchased from CycloLab
Cyclodextrin Research & Development Laboratory, Ltd. D2O
(99.95% isotopic purity) and CD2Cl2 were also purchased from
Aldrich Chemicals. Di(1-adamantyl)benzylphosphine (DABP)
was synthesized as reported by Beller et al.21 The purity of the
DABP was carefully controlled. In particular, 31P solution
NMR indicated that the product was a mixture of phosphine
(ca. 98%) and its oxide (ca. 2%). Carbon monoxide/hydrogen
mixtures (1:1) were used directly from cylinders (>99.9% pure;
PraxAir). Bidistilled deionized water was used in all experi-
ments. All reactor preparations are realized under a totally
isolated and controlled environment in a glovebag, except the
catalytic precursor (Rh(acac)(CO)2/DABP), which was per-
formed under nitrogen using standard Schlenk and vacuum line
techniques before being moved in a glovebag. All solvents and
liquid reagents were degassed by bubbling nitrogen for 15 min
before each use or by two freeze-pump-thaw cycles before use.
The stainless reactor was purchased from Parr Instrument
Company. The GC analyses were performed on a TRACE
GC Ultra (Thermo Scientific) coupled with a Windows XP
workstation.
Molecular Modeling. All calculations were performed on the
Windows Vista platform. The initial configurations were ob-
tained from UFF calculations under ArgusLab 4.0.1 software
(Mark Thompson and Planaria Sofware LLC).23 In order to
assess the energy content for various molecules designed and
minimized above, semiempirical quantum calculations were
undertaken using the PM3 method with COSMO solvation
parameters as implemented in MOPAC2009 (Stewart Compu-
tational Chemistry). The COSMO method (conductor-like
screening model) is useful for determining the stability of
various species in a solvent environment. Theoretical calcula-
tions were carried out at the restricted Hartree-Fock level
(RHF) using the PM3 semiempirical SCF-MO methods. For
dichloromethane simulation, a relative permittivity of 4.8 was
employed for dichloromethane and 74.8 for water, with up to
92 surface segments per atom for the COSMO model being
used to construct a solvent-accessible surface area based
on van der Waals radii. All structures were optimized to
a gradient inferior to 0.1 using the eigenvector following
method. Note that the MOPAC Cartesian coordinates are
generated with OpenBabel 2.2.0 graphical interface (Chris
Morley) from the geometries obtained with ArgusLab UFF,
and the keywords used are added in the file. After MOPAC
geometry optimization the out files are opened with Molda 6.5
and final pictures are obtained with WebLab ViewerLite 3.7
(Molecular Simulations Inc.). To measure the cone angle, after
geometry optimization by PM3 calculation, the classical defi-
nition of the cone angles is used.24 Indeed, Tolman’s original
proposal depended on the direct physical measurement
of cone angles from idealized space-filing CPK models.25
To make the calculations consistent with those reported by
Tolman, the metal-phosphorus length was initially set con-
1
NMR Measurements. H, 13C, 31P, and ROESY NMR ex-
periments were recorded on an Advance 300 Bruker, at 300.13,
75.49, and 121.49 MHz, respectively, with the samples not
spinning. 1H and 31P{1H} chemical shifts are given in ppm
relative to external references: sodium [D4]-3-(trimethylsilyl)-
propionate (98% atom D) in D2O for 1H NMR and H3PO4 in
D2O for 31P{1H} NMR. Chemical shifts are given in ppm (δ)
and measured relative to residual solvent. All NMR measure-
ments were carried out in pure deuterated water (pH = 6.5)
because it is known that electrolyte ions also form inclusion
complexes with CDs.22
˚
˚
stant at 2.28 A and a van der Waals radius of 1.00 A was used
for hydrogen.
General Biphasic Hydroformylation Experiments. All catalytic
reactions were performed under nitrogen using standard
Schlenk techniques. All solvents and liquid reagents were de-
gassed by bubbling nitrogen for 15 min before each use or by two
freeze-pump-thaw cycles before use. Rh(acac)(DABP)2 (5.0 ꢀ
10-3 mmol) and the required amount of methylated-β-CD were
dissolved in 1 mL of water. Olefin (1.5 mmol) and undecane
(0.2 mmol, GC internal standard) were dissolved in dichloro-
methane (1 mL). The two solutions were charged in a glovebag
into the 25 mL reactor, which was sealed and heated at 70 °C.
Magnetic stirring was started (1150 rpm), and the autoclave was
pressurized with 45 atm of CO/H2 (1:1) from a gas reservoir
connected to the reactor through a high-pressure regulator valve
enabling constant pressure in the reactor throughout the whole
reaction. After 12 h, the autoclave was cooled and depressurized
and the organic products were analyzed using a CG. The results
presented are the average of at least 3 runs under each set of
conditions.
ESI/HRMS. ESI mass spectra were recorded with a TSQ
Quantum Ultra (Thermo Scientific) mass spectrometer with
ꢀ
ꢀ
accurate mass options instrument (Universite de Montreal Mass
Spectrometry Facility).
Phase-Solubility Diagrams. Solubility measurements were
performed at 298 ( 0.5 K. For all CD solutions, DABP was
added until insolubility. These solutions were mixed for 24 h in a
temperature-controlled water bath. These solutions were fil-
tered, and the DABP solubility was determined by measuring
the absorbance of diluted solutions at 265 nm (Abs < 1).
Solubility was obtained by multiplying the absorbance of di-
luted solutions by the dilution factor. For solubility measure-
ments, we used a Cary Bio 100 UV-visible spectrophotometer.
Partition Coefficients between Organic and Aqueous Phases.
Partition coefficients studies were performed at 25 ( 0.5 °C. For
all experiments, dichloromethane solutions (4 mL) of DABP (1
mM) were prepared without or with CDs (1 equiv). After
vigorous stirring, 4 mL of water was added to each solution.
These solutions were mixed for 24 h in a temperature-controlled
(23) Thompson, M. A. ArgusLab 4.0.1; Planaria Software LLC, http://
www.arguslab.com.
(24) Mingos, D. M. P. In Modern Coordination Chemistry: The
Legacy of Joseph Chatt; Leigh, G. J.; Winterton, N., Eds.; The Royal
Society of Chemistry: Cambridge, 2002; pp 69-78.
(25) (a) Tolman, C. A. Chem. Rev. 1977, 77, 313–348. (b) Seidel, W. C.;
Tolman, C. A. Ann. N.Y. Acad. Sci. 1983, 415, 201–221.
(21) Beller, M.; Ehrentraut, W. H.; Ehrentraut, T. E.; Fuhrmann, C.;
Zapf, A. US Patent 7148176, 2006.
(22) Rohrbach, R. P.; Rodriguez, L. J.; Eyring, E. M.; Wojcik, J. F. J.
Phys. Chem. 1977, 81, 944–948.

Article
Organometallics, Vol. 29, No. 15, 2010 3449
Acknowledgment. We are grateful to the Natural
Sciences and Engineering Research Council of Canada,
We thank colleagues for careful reading and discussion of
the manuscript.
Supporting Information Available: ROESY NMR, 31P{1H}
NMR, and UV spectra of DABP ligand. This material is
available free of charge via the Internet at http://pubs.acs.org.
ꢀ ꢀ
the Fonds Quebecois de la Recherche sur la Nature
et les Technologies, the Canada Foundation for Innova-
ꢀ
tion, and Universite de Montreal for financial support.
ꢀ