Communications
À
flowing D2 show that the OH bands of the zeolite initially
present at 3630 and 3565 cmÀ1 readily disappeared—as new
bands grew in the range of 2550–2780 cmÀ1 (Figure SI-3),
classes of active sites are required to be close enough to each
other to allow a cooperative mechanism, consistent with our
observation that a physical mixture of the bare H-form zeolite
and the MgO-supported rhodium complex (5:1 mass ratio)
led to the formation of ethane as the only observed reaction
product, corresponding to the performance of the rhodium
complex anchored to the basic support. Thus, we rule out the
possibility that the dimerization occurs as a result of action of
the acidic zeolite sites, even when hydrogen may spill over
onto this support. However, we do not discard the possibility
that the reaction takes place by activation of one of the ethene
molecules on the same Al site where the rhodium complex is
bonded, as it is known that the Al sites can be partially
À
À
corresponding to the formation of Si OD and Al OD groups
resulting from HD exchange.[29] Upon introduction of C2H4 to
À
the IR cell, the Al OD bands at 2677 and approximately
2600 cmÀ1 declined drastically in intensity (Figure SI-3)—just
as the dimerization reaction started. Mass spectra of the gas-
phase products show that the butenes were characterized by
C3H5 and C4H8 fragments (m/z 41 and 56, respectively,
corresponding to the typical fragmentation of n-butenes) with
no deuterated dimers. This result demonstrates that C2H5
species, either as ethyl ligands formed on the rhodium after a
partial hydrogenation of the initially p-bonded ethene
ligands[31] or as carbenium ions on the acidic zeolite sites,[20]
are not reaction intermediates in the formation of n-butenes,
in contrast to the accepted mechanisms for dimerization of
alkenes on halide-containing rhodium complexes[12,13] and on
strong solid Brønsted acids.[18–24] (Scheme SI-1 illustrates the
reported reaction mechanisms for ethene dimerization on
these catalysts.) The results of the D2 exchange experiments
are thus consistent with the observation that ethene retained
removed from the zeolite framework under some conditions
[32]
À
to yield another type of acidic Al OH species, which we
have observed for our catalyst in the presence of H2 at 303 K
and 1 bar, leading to the appearance of a new band in the
OH stretching region at 3710 cmÀ1 that readily disappeared
upon contact with C2H4.
À
To account for the observed high selectivity to butenes
(Table 2), we infer that the interaction between two adsorbed
=
ethene molecules is faster than the hydrogenation of the C C
=
À
À
the C C bond upon interaction with the zeolite Al OH
groups.
bond. The data indicate that the formation of the C C bond,
which is the rate-determining step when the reaction is
catalyzed by rhodium halide complexes[12] and is quite slow on
acidic zeolites,[20,21] is facilitated by a cooperation between the
RhI species and the acidic sites of the support. In contrast, the
accumulation of relatively large amounts of hydrocarbons on
the surface of the catalyst during reaction (evidenced by
intense IR bands in the range 2960–2875 cmÀ1, Figure 1,
bottom), together with the observed deactivation of the
catalyst over time (Figure SI-6), suggests that product
desorption might be rate-determining (notwithstanding the
deactivation, more than 2000 turnovers of the catalyst were
achieved in 5 h on stream, demonstrating that the reaction is
catalytic). This observation, together with the fact that H2
accelerates the dimerization reaction (Table 2) without being
incorporated into the product butenes, suggests that the role
of hydrogen is to facilitate the regeneration of the active sites
by accelerating the desorption of the products.
Moreover, the mass spectra characterizing the product
observed with the zeolite-supported rhodium catalyst in the
presence of flowing C2H4 + D2 demonstrate the formation of
C2H5D and C2H4D2 in low concentrations, resulting from a
=
minor, parallel hydrogenation of the C C bond. Accordingly,
we infer that this hydrogenation, in contrast to the dimeriza-
tion process, proceeds through ethyl ligands on the rhodium,
as expected.[31]
On the basis of these results, we propose a mechanism
(Scheme 1) for the dimerization of ethene on the zeolite
incorporating the rhodium complex. In this scheme, an ethene
À
molecule weakly adsorbed on an Al OH site of the zeolite
(which is not reactive enough to couple with another ethene
from the gas phase as occurs with stronger Brønsted acids)
reacts with a second ethene molecule activated on a nearby
rhodium site (Figure SI-7). According to this model, the two
To understand the role of the H2 activation
in the dimerization process, as well as in the
=
competitive hydrogenation of the C C bond, we
compared the performance of the rhodium
complex catalyst supported on zeolite DAY
with that of a sample consisting of extremely
small rhodium clusters on the same support,
prepared by treatment of the former catalyst
with H2 at 303 K. EXAFS spectra taken at the
Rh K edge after 1.2 h in flowing H2 indicate a
À
Rh Rh coordination number of nearly
2
(Table SI-1), which corresponds to an average
cluster nuclearity of 3, and thus to the presence
of extremely small rhodium clusters together
with unconverted rhodium complexes.[30]
The catalytic performance of the zeolite
containing the rhodium clusters in flowing
mixtures of C2H4 + H2 at 303 K and 1 bar
indicates that the hydrogenation of the double
Scheme 1. Simplified reaction mechanism for the conversion of ethene on a zeolite-
supported rhodium complex catalyst in the presence of H2. The reaction intermediates
are proposed on the basis of IR and EXAFS data, results of H2/D2 exchange
experiments, and catalyst performance data.
5530
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 5528 –5531