Heterogeneously catalyzed asymmetric hydrogenation of C=C bonds directed by surface-tethered chiral modifiers

Article abstract of DOI:10.1021/ja906356g

Asymmetric hydrogenation of C=C bonds is of the highest importance in organic synthesis, and such reactions are currently carried out with organometallic homogeneous catalysts. Achieving heterogeneous metal-catalyzed hydrogenation, a highly desirable goal

Full text of DOI:10.1021/ja906356g

Published on Web 09/15/2009
Heterogeneously Catalyzed Asymmetric Hydrogenation of
CdC Bonds Directed by Surface-Tethered Chiral Modifiers
†
,‡
§
†
David J. Watson, R.J. Bennie Ram John Jesudason, Simon K. Beaumont,
†
§
,†
Georgios Kyriakou, Jonathan W. Burton, and Richard M. Lambert*
Department of Chemistry, UniVersity of Cambridge, Lensfield Road, Cambridge CB2 1EW,
United Kingdom, and Department of Chemistry, UniVersity of Oxford, Chemistry Research
Laboratory, 12 Mansfield Road, Oxford, OX1 3TA, United Kingdom
Received July 29, 2009; E-mail: rml1@cam.ac.uk
Abstract: Asymmetric hydrogenation of CdC bonds is of the highest importance in organic synthesis, and
such reactions are currently carried out with organometallic homogeneous catalysts. Achieving heteroge-
neous metal-catalyzed hydrogenation, a highly desirable goal, necessitates forcing the crucial enantiod-
ifferentiating step to take place at the metal surface. By synthesis and application of six chiral sulfide ligands
that anchor robustly to Pd nanoparticles and resist displacement, we have for the first time accomplished
heterogeneous enantioselective catalytic hydrogenation of isophorone. High resolution XPS data established
that ligand adsorption from solution occurred exclusively on the Pd nanoparticles and not on the carbon
support. All ligands contained a pyrrolidine nitrogen to enable their interaction with the isophorone substrate
while the sulfide functionality provided the required interaction with the Pd surface. Enantioselective turnover
numbers of up to ∼100 product molecules per ligand molecule were found with a very large variation in
asymmetric induction between ligands: observed enantiomeric excesses increased with increasing size of
the alkyl group in the sulfide. This likely reflects varying degrees of ligand dispersion on the surface: bulky
substituent groups hinder close approach of ligand molecules to each other, inhibiting close-packed island
formation, favoring dispersion as separate molecules, and leading to effective asymmetric induction.
Conversely, small substituents favor island formation leading to very low asymmetric induction. Enanti-
oselective reaction most likely involves initial formation of an enamine or iminium species, confirmed by
use of an analogous tertiary amine, which leads to racemic product. Ligand rigidity and resistance to self-
assembled monolayer formation are important attributes that should be designed into improved chiral
modifiers.
3
Introduction
in the case of ꢀ-ketoester hydrogenation (Ni) and details of
the corresponding reaction mechanisms are the subject of debate,
The major operational advantages offered by heterogeneous
over homogeneous catalysis are well-known and exploited
wherever possible. However, heterogeneously catalyzed enan-
tioselectiVe reactions are rarities, despite their huge potential
importance in the research laboratory and in the pharmaceutical,
fine chemicals, and advanced materials industries. Virtually all
published work in this field refers to the asymmetric catalytic
hydrogenation of CdO bonds in ketoesters carried out on the
one key aspect is very clear for both classes of reaction: the
crucial step that leads to enantiodifferentiation occurs at the
surface of the metal.
The situation with regard to asymmetric hydrogenation of
CdC bonds is very different. Moreover, unlike CdO asym-
metric hydrogenation, such processes are of the highest impor-
tance in organic synthesis and are often a critical step in an
4
,5
1
overall synthetic scheme (e.g., the synthesis of L-dopa).
surfaces of modified Pt or Ni surfaces. In these systems, a chiral
Currently, such reactions are carried out with organometallic
homogeneous catalysts, which depend on costly, usually phos-
phorus-based ligand systems. There are very few examples of
heterogeneously catalyzed asymmetric CdC hydrogenation. In
the late 1990s Nitta et al. and others claimed that cinchona
alkaloid-modified Pd surfaces may be used for the enantiose-
lective hydrogenation of CdC bonds in large aromatically
conjugated hydrogenation substrates, such as (E)-R-phenylcin-
agent is used to modify the otherwise achiral heterogeneous
catalyst, thus allowing enantioselective catalysis to occur.
Although the means by which such chiral templating of the
metal catalyst surface is accomplished in the case of R-ketoester
2
hydrogenation (Pt) appears to differ from that which operates
†
University of Cambridge.
Present address: Department of Chemistry, University of Reading,
‡
6
Whiteknights, Reading, United Kingdom RG6 6AD.
namic acid. However, large amounts of the chiral agent were
§
University of Oxford.
(
1) For a recent overview of heterogeneous enantioselective hydrogena-
tions including the Orito reaction see: Klabunovskii, E.; Smith, G. V.;
(3) De Vos, D. E.; Vankelcom, I. F. J.; Jacobs, P. A. Chiral Catalyst
Immobilization and Recycling; Wiley-VCH: Weinheim, 2000.
(4) Knowles, W. S. Angew. Chem., Int. Ed. 2002, 41, 1999–2007.
(5) Noyori, R. Angew. Chem., Int. Ed. 2002, 41, 2008–2022.
(6) Nitta, Y. Chem. Lett. 1999, 7, 635–636.
˜
Zsigmond, A. Heterogeneous EnantioselectiVe Hydrogenation - Theory
and Practice; Springer: Dordrecht, 2006; Vol. 31.
(2) Bartok, M.; Felfoldi, K.; Torok, B.; Bartok, T. Chem. Commun. 1998,
2
3, 2605–2606.
1
4584 9 J. AM. CHEM. SOC. 2009, 131, 14584–14589
10.1021/ja906356g CCC: $40.75  2009 American Chemical Society

CdC Bond Hydrogenation Directed by Chiral Modifiers
A R T I C L E S
Figure 1. Proline directed asymmetric hydrogenation of isophorone to (S)-
TMCH ((S)-3,3,5-trimethylcyclohexanone).
7
required, some of which underwent hydrogenation, so that the
behavior was not catalytic in the normal sense of the term.
Moreover, the effect was specific to this particular class of
substrates, and a recent review by Studer et al. indicates that
no significant progress has been made in extending this
Figure 2. Sulfide ligands used 1,2-bis((S)-pyrrolidin-2-ylmethyl)disulfane
(1), (S)-2-(methylthiomethyl)pyrrolidine (2), (S)-2-(iso-propylthiometh-
yl)pyrrolidine (3), (S)-2-(tert-butylthiomethyl)pyrrolidine (4), (S)-2-(ada-
mantan-1-ylthiomethyl)-pyrrolidine (5), and (S)-2-(phenylthiomethyl)pyr-
rolidine (6).
8
approach. Another system, the proline-directed asymmetric
hydrogenation of isophorone (Figure 1) and similar molecules,
typically carried out with a Pd catalyst, has been investigated
genation of isophorone. Above all else, the principal object of
this work is to demonstrate proof of concept.
9,10
by Tungler and co-workers who proposed that these reactions
proceed by the same general mechanism as that which operates
for asymmetric CdO hydrogenationsa reactive encounter
between the adsorbed organic substrate and the adsorbed chiral
agent (proline) in the presence of hydrogen adatoms.
Experimental Methods
Enantiomerically pure sulfide ligands 2, 4, and 6 were prepared
1
4
11
12
following known procedures (Figure 2); the remaining ligands
, 3, and 5 were prepared analogously. All of the ligands contain
However, we have shown and others have confirmed that
this interpretation is not correct. The metal surface merely carries
out a racemic hydrogenation of adsorbed isophorone, and the
observed enantiomeric excess (ee) in the product (trimethylcy-
clohexanone, TMCH) is merely due to subsequent kinetic
resolution that takes place in solution as a result of one
enantiomer of TMCH reacting with the chiral agent much faster
than the other. In other words, heterogeneous enantioselective
catalytic synthesis was not achieved. The mechanism we
1
the pyrrolidine motif which has had widespread use in enantiose-
lective organocatalysis; however, the carboxylic acid group is
15
replaced by a range of sulfide substituents. The pyrrolidine nitrogen
is intended to enable these ligands to interact with isophorone while
the sulfide functionality provides a strong interaction with the
16
surface, enabling the tethered molecules to behave as true hetero-
geneous chiral modifiers. A range of structurally diverse sulfide
substituents (Figure 2) was used so as to examine possible steric
and electronic effects on the effectiveness of the chiral ligand at
inducing asymmetry in the heterogeneous enantioselective hydro-
genation of isophorone. Molecules analogous to the disulfide ligand
1
1
proposed also explains why the maximum attainable yield of
enantiopure TMCH cannot exceed 50% and why the chiral agent
1
7,18
(which is necessarily consumed) has to be used in stoichiometric
1
undergo S-S scission upon adsorption on a range of metals,
19
amounts rather than in catalytic quantities.
including Pd. Ligand 1 therefore provides a means of depositing
the unsubstituted sulfide on the surface.
To achieve true heterogeneous enantioselective catalysis it
is necessary to force the crucial enantiodifferentiating step to
take place at the metal surface. In this connection, using single
Catalytic testing of the hydrogenation reactions was carried out
by the following procedure: Methanolic solutions of the enantio-
merically pure ligands were added to a reaction mixture of
isophorone (3 mL, Sigma-Aldrich) and 10% reduced Pd/C catalyst
1
3
crystal methods, we have shown that a critical obstacle to
heterogeneous catalysis in this system is the much faster
(
Alfa Aesar, 0.050 g) in methanol (40 mL, Fisher Scientific HPLC
5
(
(
∼×10 ) and stronger adsorption from solution of the reactant
grade), such that the molar ratio of ligand to isophorone was varied
up to a maximum of 1:500. These solutions were then hydrogenated
isophorone) compared to the chiral agent (proline). As a result,
by stirring in autoclaves under 15 bar of H
2
(after thorough flushing
the metal surface becomes saturated with isophorone to the
complete exclusion of proline so that only racemic hydrogena-
tion is possible at the metal surface. Clearly, overcoming this
impediment requires radically changing the surface chemistry
so as to tether the chiral agent to the metal surface sufficiently
strongly. Here we report a significant advance: the successful
attainment of this goal by purposeful synthesis of chiral ligands
that anchor robustly to the metal surface, resist displacement,
and direct the heterogeneous enantioselective catalytic hydro-
with H at the start of each reaction) to fixed time (168 h) or fixed
2
conversion (60%). Analysis of reaction products was carried out
by dilution of aliquots in 50/50 dichloromethane/methanol (Fisher
Scientific HPLC grade) and addition of a decane (Sigma-Aldrich)
internal standard. These were then analyzed by a gas chromatograph
(
Hewlett-Packard 5890 Series II) equipped with an R-cyclodextrin
capillary column (Chirasil-Dex CB, Varian, Inc.) for chiral separa-
(14) Cran, G. A.; Gibson, C. L.; Handa, S. Tetrahedron: Asymmetry 1995,
6
, 1553–1556.
(
15) For recent reviews of organocatalysis with catalysts bearing a
pyrrolidine motif see: (a) Erkkila, A.; Majander, I.; Pihko, P. M. Chem.
ReV. 2007, 107, 5416–5470. (b) Mukherjee, S.; Yang, J. W.; Hoffmann,
S.; List, B. Chem. ReV. 2007, 107, 5471–5569. (c) Gaunt, M. J.;
Johansson, C. C. C.; McNally, A.; Vo, N. T. Drug DiscoV. Today
2007, 12, 8–27.
(
(
7) Ferri, D.; Burgi, T.; Baiker, A. J. Catal. 2002, 210, 160–170.
8) Studer, M.; Blaser, H. U.; Exner, C. AdV. Synth. Catal. 2003, 345,
4
5–65.
(
9) Tungler, A.; Kajtar, M.; Mathe, T.; Toth, G.; Fogassy, E.; Petro, J.
Catal. Today 1989, 5, 159–171.
(
(
(
(
10) Tungler, A.; Mathe, T.; Petro, J.; Tarnai, T. J. Mol. Catal. 1990, 61,
(16) Jensen, S. C.; Baber, A. E.; Tierney, H. L.; Sykes, E. C. H. ACS Nano
2007, 1, 22–29.
(17) Biebuyck, H. A.; Bain, C. D.; Whitesides, G. M. Langmuir 2002, 10,
1825–1831.
(18) Driver, S. M.; Woodruff, D. P. Surf. Sci. 2001, 488, 207–218.
(19) Love, J. C.; Wolfe, D. B.; Haasch, R.; Chabinyc, M. L.; Paul, K. E.;
Whitesides, G. M.; Nuzzo, R. G. J. Am. Chem. Soc. 2003, 125, 2597–
2609.
2
59–267.
11) McIntosh, A. I.; Watson, D. J.; Burton, J. W.; Lambert, R. M. J. Am.
Chem. Soc. 2006, 128, 7329–7334.
12) Mhadgut, S. C.; Torok, M.; Esquibel, J.; Torok, B. J. Catal. 2006,
2
38, 441–448.
13) McIntosh, A. I.; Watson, D. J.; Lambert, R. M. Langmuir 2007, 23,
6
113–6118.
J. AM. CHEM. SOC. 9 VOL. 131, NO. 40, 2009 14585

A R T I C L E S
Watson et al.
Figure 3. Dependence of TMCH ee on initial ligand concentration. Each data point corresponds to running the reaction up to a conversion of ∼60%. (a)
, (S)-2-(tert-butylthiomethyl)pyrrolidine (4); O, (S)-2-(methylthiomethyl)pyrrolidine (2); [, (S)-2-(phenylthiomethyl)pyrrolidine (6); (b) 2, (S)-2-(adamantan-
-ylthiomethyl)-pyrrolidine (5); 3, (S)-2-(iso-propylthiomethyl)pyrrolidine (3); b, 1,2-bis((S)-pyrrolidin-2-ylmethyl)disulfane (1). A ligand concentration of
1
1
0
-
3
.47 mmol dm corresponds to 0.1 mol % ligand with respect to isophorone.
Table 1. Dependence of TMCH ee on Identity of Ligand for Fixed
tion. The dimethyl acetal of TMCH is formed in varying amounts
under the reaction conditions. In cases where the dimethyl acetal
was formed, a few crystals of pyridinium p-toluene sulfonate, one
drop of water, and one drop of acetone were added to a sample of
the crude reaction mixture to allow complete hydrolysis of the acetal
back to TMCH. The sample was then analyzed as before. The
enantiomeric excess of the TMCH was the same before and after
hydrolysis demonstrating that the dimethyl acetal was formed post
hydrogenation and not from the isophorone prior to hydrogenation.
High resolution X-ray photoelectron spectroscopy measurements
were recorded in NCESS Laboratory, Daresbury, U.K., using the
SCIENTA ESCA300 spectrometer and employing monochroma-
tized Al KR (hν ) 1486.6 eV) radiation. Thin films of (i) carbon
support (carbon black, Cabot), (ii) Pd reference sample (powder,
Johnson Matthey), and (iii) 10 wt % Pd/C catalyst (Alfa Aesar)
reduced in hydrogen at 150 °C for 12 h were obtained by pressing
the three powders on Al plates (Advent 99.9%) until no Al signal
could be detected by XPS. Adsorption of the (S)-2-(tert-butylthi-
omethyl)pyrrolidine chiral modifier was performed by dipping the
plates in a 0.47 mmol dm- solution of the chiral modifier (4) in
methanol for 10 min. The samples were dried briefly in air then
vacuum prior to acquisition of XP spectra.
-3
Ligand Concentration (0.23 mmol dm , corresponding to 0.05
mol % ligand with respect to isophorone) as in Figure 2
modifier ligand
% ee of TMCH obtained
(S)-2-(adamantan-1-ylthiomethyl)-pyrrolidine (5)
(S)-2-(tert-butylthiomethyl)pyrrolidine (4)
13.8
7.3
6.0
3.3
0.0
0.0
(
(
S)-2-(iso-propylthiomethyl)pyrrolidine (3)
S)-2-(methylthiomethyl)pyrrolidine (2)
1
,2-bis((S)-pyrrolidin-2-ylmethyl)disulfane (1)
(
S)-2-(phenylthiomethyl)pyrrolidine (6)
initial ligand concentration, consistent with strong Langmuir-
like adsorption of the chiral modifier from solution.
It is striking that the gradients of the lines in Figure 3 increase
systematically with increasing size of the alkyl group in the
thioether. Another way of illustrating this trend is shown in
Table 1 which lists the ee achieved at ∼60% for each modifier
-3
for the same initial modifier concentration of 0.23 mmol dm .
3
(
Typically, yields calculated from an internal standard were
found to be consistent with conversion to within a few percent
for all ligands, indicating that no significant decomposition of
the substrate occurred).
Results and Discussion
These results clearly indicate that heterogeneous enantiose-
lective hydrogenation did indeed occur in the presence of
adsorbed chiral modifier molecules. Only catalytic amounts of
our modifiers were used (typically 1:2000 modifier/isophorone
molar ratio) in contrast to the far greater stoichiometric amounts
of proline that were required to achieve an enantiomeric excess
Figure 3a,b show the measured ee as a function of modifier
concentration for the six different chiral ligands. A measure of
the uptake of the ligand by the catalyst is provided by specifying
the concentration of the ligand in the initial reaction solution:
exactly where on the Pd/C catalyst the ligand resides is revealed
by XPS data presented below. Here, it is important to note that
only Very small (i.e., catalytic) amounts of ligand were used, in
contrast with the necessarily large stoichiometric amounts of
proline that were consumed when the latter was used to achieve
kinetic resolution of the TMCH product in the hydrogenation
9
,11
in the product Via kinetic resolution.
Rationalizing the
observed dependence of ee on the degree of steric hindrance in
the vicinity of the stereogenic center on the pyrrolidine moiety
requires discussion of the mode of ligand adsorption. First, it
must be the case that our ligands adsorb intact and remain so
during reaction. Dissociative adsorption of the modifier by
cleavage of the C-S bond (see Figure 4) to yield two adsorbed
fragments would isolate the stereogenic carbon atom on the
pyrrolidine ring from the alkyl or aryl functionality. If this were
the case the ee achieved would be the same for all the sulfides
as all would produce the same chiral entity on the surface.
Clearly, this is not what is observed.
1
0
of isophorone.
Reactions were run until a yield of at least 60% was obtained
as monitored by hydrogen gas consumption during reaction
(
and subsequently by chiral gas chromatography). In every case,
increasing the ligand concentration decreased the reaction rate,
and this retardation effect was most pronounced with the largest
ligands. For this reason, the concentration range explored was
limited because the length of time required to achieve ∼60%
conversion was infeasible.
How then does the R functionality affect ee? A plausible
hypothesis is that the identity of R determines the spatial
distribution of the adsorbed chiral modifier on the surface of
the Pd catalyst. In general, adsorbates on metal surfaces may
be dispersed as an ordered or disordered array of individual
molecules, they may agglomerate into islands of close-packed
molecules separated by regions of bare surface, or the islands
may be in equilibrium with a “sea” of dispersed molecules.
Which of these three possibilities actually occurs depends on
Two features are apparent. First, there is a large variation in
the effectiveness of the ligands to induce asymmetry in the
reduction of isophorone: 1,2-bis((S)-pyrrolidin-2-ylmethyl)dis-
ulfane (1) and (S)-2-(phenylthiomethyl)pyrrolidine (6) are totally
ineffective, whereas the other four ligands are effective to
varying degrees. Second, for the effective ligands, the ee
achieved at ∼60% conversion varies essentially linearly with
1
4586 J. AM. CHEM. SOC. 9 VOL. 131, NO. 40, 2009

CdC Bond Hydrogenation Directed by Chiral Modifiers
A R T I C L E S
bare (i.e., racemic chemistry). Activity is therefore high, and
because only very few ligand molecules can interact with
adjacent reactant molecules, ee is very low, as indeed observed
2
1,22
(
Figure 3 and Table 1).
Initially, with increasing ligand
concentration, the islands grow, most of the surface remains
bare, activity remains high, and the reaction is essentially
racemic. At a certain point, the entire surface becomes covered
with close-packed ligand molecules, a self-assembled sulfide
monolayer that is impervious to reactant molecules, and activity
collapses. Further increase in ligand concentration in the solution
phase can have no effect on the saturated surface which therefore
remains catalytically inert.
Figure 4. Showing that dissociative adsorption of sulfide ligands would
preclude any dependence of ee on the R group.
Although the ee’s we report are modest compared to those
encountered in the well-established field of homogeneous
enantioselective catalysis, their significance is considerable. We
have demonstrated a hydrogenation reaction that is both truly
heterogeneous and genuinely catalytic and which leads to
enantioenriched product. Furthermore, we have demonstrated
that the effectiveness of the stereogenic center on the ligand in
inducing enantioselective catalysis depends in no small part on
the collective behavior of the adsorbed chiral molecules which
depend on the interplay between intermolecular and molecule-
surface interactions.
the interplay between molecule-surface and molecule-molecule
interactions. Consider now the consequences of these alternative
behaviors for enantioselective catalysis. Fully dispersed chiral
modifiers should be the most effective: each molecule can
interact with a neighboring adsorbed substrate molecule (iso-
phorone) allowing asymmetric induction in the subsequent
reaction with H adatoms. Close-packed islands of modifier
should be very ineffective; only molecules at the periphery can
interact with adjacent substrate molecules. And of course only
hydrogenation to give racemic product can occur on the “bare”
It is revealing to calculate the enantioselective turnover
23
number because this enables a distinction to be made between
(modifier-free) metal surface. Bulky R groups will hinder close
a catalytic process and a kinetic resolution by reductive
amination, that is, the mechanism of the proline mediated
reaction. For example, carrying out the reaction with 0.47 mmol
approach of modifier molecules to each other, thus inhibiting
island formation, favoring dispersion as separate molecules, and
resulting in effective asymmetric induction. Conversely, small
R groups should favor island formation leading to very low
asymmetric induction. Another factor that may contribute to
the observed increase in ee with bulkiness of the R group is
that a larger R may push the stereogenic center closer to the
surface constraining the reaction conformation during hydro-
genation and promoting enantioselective reaction. The results
-3
dm (corresponding to 0.1 mol % ligand with respect to
isophorone) of (S)-2-(tert-butyl-thiomethyl)-pyrrolidine modifier
(4) present in the initial reaction solution, hydrogenation gave
TMCH in 70% yield with an ee of 15%. This corresponds to
a enantioselective turnover number of g100 product mol-
ecules per ligand molecule. (Mere kinetic resolution would
of course correspond to an enantioselective turnover number
of exactly 1).
i
t
for R ) H, Me, Pr, Bu, and adamantyl, where bulkiness and
ee increase together, are in agreement with these explanations
To confirm that the observed enantiomeric excess did indeed
arise purely from the chirality of the modifier ligand, a control
experiment was performed with the opposite enantiomer of the
originally used chiral modifier. Thus a concentration of 0.47
(
Figure 3 and Table 1). Interestingly, R ) phenyl does not fit
this pattern: no detectable ee was observed in this case. It
appears that the presence of an aromatic substituent strongly
modifies adsorption behavior compared to the other cases,
possibly the consequence of π-π interactions which favor island
-
3
mmol dm (corresponding to 0.1 mol % ligand with respect
to isophorone) of (R)-2-(tert-butyl-thiomethyl)-pyrrolidine pro-
duced TMCH with an ee of -14.3% in very good agreement
with the result given above (Figure 3) where it was found that
the (S) form gave an ee of +14.8% at similar reactant
conversion.
2
0
formation and, hence, poor enantioselectivity.
Support for this explanation is provided by the data shown
in (Figure 5a,b) which shows the total conVersion of reactant
to TMCH product after a fixed time as a function of ligand
concentration for two different ligands. These results provide a
direct indication of the extent to which reactant molecules can
access the surface and undergo hydrogenation. For R ) tert-
butyl (Figure 5a) the catalytic activity decreases approximately
linearly with ligand concentration while at the same time the
ee rises (Figure 3 and Table 1). This is consistent with well-
dispersed ligand molecules which (i) create chiral adsorption
sites and (ii) necessarily decrease overall activity simply as a
result of denying surface sites to the reactants (hydrogen and
isophorone). The behavior for R ) phenyl (Figure 5b) is
dramatically different, with a pronounced step decrease in
Our current hypothesis regarding the mechanism of the
reaction is as follows (Figure 6). The ligand is tightly bound to
the surface of the palladium through sulfur. The ligand and
surface-bound isophorone react to give an iminium ion (or
enamine) with loss of water. The iminium ion or enamine
undergoes diastereoselective olefin hydrogenation to give a
second iminium ion/enamine which is hydrolyzed to give
product which desorbs from the surface.
(21) In our case the ligand does not cause rate enhancement, unlike the
case of Pt-catalyzed ketoester hydrogenation, which involves a different
mechanism. Such effects are specific to the reaction in question.
-3
activity at 0.2 mmol dm . This may be rationalized as follows.
Even at low concentration the ligand molecules form islands,
(
22) Walsh, P. J.; Koslowski, M. C. Fundamentals of Asymmetric Catalysis;
University Science Books: Sausalito, CA, 2009; Chapter 1, pp 15-
2
0
likely due to π-π interactions, leaving most of the surface
1
7.
(
23) We define enantioselective turn-over number as (product ee × product
yield)/(ligand loading × ligand ee); if this number is >1 then enan-
tioselective catalysis has occurred.
(
20) Jin, Q.; Rodriguez, J. A.; Li, C. Z.; Darici, Y.; Tao, N. J. Surf. Sci.
999, 425, 101–111.
1
J. AM. CHEM. SOC. 9 VOL. 131, NO. 40, 2009 14587

A R T I C L E S
Watson et al.
Figure 5. Dependence of activity (reactant conversion after a fixed time (168 h)) on ligand concentration. (a) (S)-2-(tert-butylthiomethyl)pyrrolidine (4), (b)
S)-2-(phenylthiomethyl)pyrrolidine (6).
(
2
5
sulfide. In every case, only very small enantiomeric excesses
were observed (<4% ee at g60% yield). This may be attributed
to the tethering sulfur atom now being in a much less rigid
relationship with the pyrrolidine allowing the two moieties to
rotate with respect to each other so that isophorone molecules
interacting with the pyrrolidine ring undergo hydrogenation from
both prochiral faces with approximately equal probability.
Finally, we address a point that is critically important to any
mechanistic interpretation that might be offered: where does
the adsorbed chiral ligand actually reside? On the palladium?
On the support? On both? Accordingly, we carried out high
resolution XPS measurements on three samples that had been
exposed to (S)-2-(tert-butyl-thiomethyl)-pyrrolidine (4) under
exactly the same conditions as those used for catalytic testing.
These were (i) pure Pd powder (ii) pure carbon support (iii)
the Pd/carbon catalyst itself. The key result is given in Figure
8a which shows the relevant S 2p XP spectra.
Figure 6. Scheme showing proposed mechanism via iminium ion/enamine
intermediate.
Figure 8a confirms extensive adsorption of (S)-2-(tert-butyl-
thiomethyl)-pyrrolidine (4) by palladium. It is noteworthy that,
when due allowance is made for photoionization cross sections,
the corresponding N 1s XP spectra (shown in the Supporting
Information) yield a S:N ratio of 1:1 in accord with the
stoichiometry of the chiral modifier. Figure 8b shows Pd 3d
spectra obtained from the Pd powder reference sample and from
the catalyst. It serves to illustrate the point that the metal content
of the catalyst is of course far lower than that of pure Pd powder,
which is why the S 2p spectrum of the catalyst is ∼10× less
intense than that of pure Pd powder (Figure 8a). The broadening
of the S 2p signal observed for the (dilute) Pd/C catalyst likely
reflects inelastic scattering of the sulfur photoelectrons (kinetic
energy 1320 eV) arising from Pd nanoparticles located within
the carbon support. Thus it is established that within experi-
mental error ligand uptake occurred exclusively on the Pd
nanoparticles, justifying the implicit assumption that underlies
the preceding discussion. (The arrow in Figure 8a indicates the
S 2p binding energy characteristic of sulfur adatoms on
Figure 7. Modified sulfide ligands used in additional experiments (S)-2-
(
tert-butylthiomethyl)-1-methylpyrrolidine (7) and ligands (8) containing
an additional methylene unit between the pyrrolidine ring and the sulfide.
To test the hypothesis that interaction of the ligand with the
substrate via the nitrogen atom on the pyrrolidine ring is
critically important, we synthesized and tested the effectiveness
of the analogous tertiary amine (S)-2-(tert-butylthiomethyl)-1-
methylpyrrolidine (7) to compare its performance with that of
the secondary amine (4) (S)-2-(tert-butyl-thiomethyl)-pyrrolidine
(
15% ee). Strikingly, use of the tertiary amine (7) as a ligand
in the catalytic asymmetric hydrogenation of isophorone resulted
in racemic product, a result which is in keeping with our model
20
palladium. This indicates that the species we observe contains
26,27
2
4
organically bound sulfur,
confirming our earlier conclusion
of enamine/iminium formation.
that adsorption does not destroy ligand integrity).
The effect of the distance between the secondary amine and
the sulfide in influencing enantioselectivity of the reduction of
TMCH was examined by synthesizing and then using a set of
ligands 8 (Figure 7) analogous to 1-6 but containing an
additional methylene unit between the pyrrolidine ring and the
Conclusions
Chiral sulfide ligands that anchor robustly to Pd nanoparticles
and resist displacement are effective in directing the truly
heterogeneous enantioselective catalytic hydrogenation of iso-
(
24) Another mechanistic possibility involves the delivery of the isophorone
to the metal surface via hydrogen bonding with the secondary amine
of the ligand. This would also account for the production of racemic
TMCH with the tertiary amine ligand 7, although it is less likely given
the previous observation of the condensation product with proline and
the lack of asymmetric induction in this reaction by other chiral
modifiers with a capacity to hydrogen bond.
(25) The ligands 8 were prepared from (S)-ꢀ-homoproline in an analogous
manner to the preparation of the ligands 1-6.
(26) Corthey, G.; Rubert, A. A.; Benitez, G. A.; Fonticelli, M. H.;
Salvarezza, R. C. J. Phys. Chem. C 2009, 113, 6735–6742.
(27) Turner, M.; Vaughan, O. P. H.; Kyriakou, G.; Watson, D. J.; Scherer,
L. J.; Davidson, G. J. E.; Sanders, J. K. M.; Lambert, R. M. J. Am.
Chem. Soc. 2009, 131, 1910–1914.
1
4588 J. AM. CHEM. SOC. 9 VOL. 131, NO. 40, 2009

CdC Bond Hydrogenation Directed by Chiral Modifiers
A R T I C L E S
Figure 8. (a) Sulfur 2p XP spectra showing uptake of (S)-2-(tert-butyl-thiomethyl)-pyrrolidine modifier (4) from methanol solution by (top to bottom)
carbon support, 10 wt % Pd/C catalyst, and pure palladium reference sample. Samples were left in contact with ligand solution for 10 min. (b) Palladium
3
d XP spectra for the fresh Pd/C catalyst and pure Pd reference.
phorone. Enantioselective turnover numbers of up to ∼100
product molecules per ligand molecule have been found. High
resolution XPS data show that ligand adsorption from solution
occurred exclusively on the Pd nanoparticles and not on the
carbon support. A large variation in asymmetric induction across
the series of ligands correlated with size of the alkyl group in
the sulfide. This is thought to reflect different degrees of ligand
dispersion on the surface, bulky substituent groups inhibiting
close-packed island formation, favoring dispersion, and hence
resulting in effective asymmetric induction. Control experiments
carried out with an analogous tertiary amine lead to a purely
racemic product, consistent with a mechanism that involves
initial formation of an enamine or iminium species. Use of a
set of ligands equivalent to the original set of six but containing
an extra methylene unit gave only very small enantiomeric
excesses. This suggests that placing the tethering sulfur atom
in a much less rigid relationship with the pyrrolidine allows
the two moieties to rotate with respect to each other so that
isophorone molecules interacting with the pyrrolidine ring
undergo hydrogenation from both prochiral faces with ap-
proximately equal probability. Ligand rigidity and resistance
to self-assembled monolayer formation are important attributes
that should be designed into improved chiral modifiers.
Acknowledgment. D.W. and R.J.B.R.J.J. acknowledge financial
support from the UK Engineering and Physical Sciences Research
Council. S.K.B. acknowledges financial support from Cambridge
University and Trinity Hall, Cambridge. J.W.B acknowledges the
Royal Society (London) for the award of a University Research
Fellowship. We thank Dr. Graham Beamson and Dr. Danny Law
for their assistance during the XPS experiments at Daresbury
Laboratory.
Supporting Information Available: XP spectra. This material
is available free of charge via the Internet at http://pubs.acs.org.
JA906356G
J. AM. CHEM. SOC. 9 VOL. 131, NO. 40, 2009 14589