ACS Catalysis
Research Article
hexanol molecule to the hydroxy group of the FOX ligand.
This hydrogen bond network could be another reason for the
equatorial displacement. Halide analogs of A, namely,
No rearrangement of the disubstituted olefin 5 was observed
during step (VI). This indicates that the cationic intermediate
formed after the nucleophilic attack of 2 is short-lived. In
similar reactions, the positive charge can migrate via a [1,3]-
hydride shift from a secondary carbon to a tertiary carbon.
Since the cationic intermediate is converted into an alkene by
proton elimination, Brønsted acid conditions extend the
lifetime of such intermediates and favor this rearrangement
process. However, the NMR data indicates that no [1,3]-
[
M(X) (FOX)] (M(X) = Mn(Br) , Fe(Cl) , Fe(Br) , or
2 2 2 2 2
15
Co(Br)2), did not catalyze the dehydration of 1.
The reaction mechanism depicted in Scheme 4 summarizes
the findings of the NMR, GC−MS, and crystallographic
Scheme 4. Proposed Mechanism of the Conversion of 1-
a
21
Phenylethanol (1) by the Catalyst [Fe(OTf) (FOX)] (A)
hydride shift occurred during the formation of 5. This
2
confirms that the reaction conditions are not Brønsted-acidic.
The complete mechanism shown in Scheme 4 was tentatively
formulated before but was not supported with the substantial
22
amount of evidence gained during this study.
Reaction Scale-Up. Since the formation rates of all
byproducts are increased at high concentrations of 1, we
decided to feed 1 slowly to a larger-scale reaction at a constant
addition rate (Scheme 5). 2 was continuously distilled from the
Scheme 5. Conversion of 1 to 2 via Reactive Distillation
reaction mixture. Such reactive distillation setups are common
for liquid-phase transformation processes in industrial
8
productions. The solvent, o-dichlorobenzene (ODCB), and
the starting material 1 have higher boiling points than the
product 2. The challenge is to distill 2 from the reaction
mixture before it reacts with the cationic intermediate to form
5
or before it polymerizes at the relatively high reaction
temperature. The high-temperature polymerization of 2 can
occur independently of a catalyst and involves free radicals.
23
a
(
I) Displacement of the equatorial triflate ligand of A by 1. (II)
The optimization of such distillation processes is complex
and involves sophisticated equipment and engineering skills.
Our experiment was aimed to demonstrate the potential for
the chemical optimization of this process. After only 1 h of
operation, common Brønsted-acidic catalysts, such as zeolites
Formation of a cationic intermediate. (III) Elimination of a proton
from a cationic intermediate yielding styrene (2). (IV) Recovery of A
via water elimination. (V) Nucleophilic attack of 1 at a cationic
intermediate resulting in formation of α-methylbenzyl ethers 3, 4, and
4
′. (VI) Formation of a styrene dimer (5) caused by nucleophilic
8
,24
attack of 2 at a cationic intermediate. (III), (V), and (VI) are
competing reaction pathways. Water formed in the dehydration
process may displace triflate ligands at iron centers in the catalytic
cycle.
and silica-aluminas, show signs of significant deactivation.
In contrast, A was proven to be very robust. The turnover
number of 310 is high for a homogeneous iron catalyst. Also,
the reaction rate did not notably decrease over the course of
the reaction, and even higher turnover numbers could have
been achieved. This is possible because the FOX ligand
inhibits the aggregation of the iron catalyst. After the
completion of the distillation, [Fe(H O) (FOX)][OTf] (C)
experiments. In step (I), the equatorial triflate ligand of A is
displaced by the alcohol 1. The activation of 1 by the Lewis
acid A leads to the formation of a cationic intermediate
2
2
2
was crystallized from the reaction mixture (Figure 4). In this
(
Scheme 4, step (II)). This benzylic carbocation can undergo
an elimination of a proton yielding styrene (2) (Scheme 4, step
III)). The E -type conversion of 1 to 2 competes with the
off-white compound, the two triflate ligands of A are displaced
by water molecules. The equatorial Fe−OH bond length
2
(
(2.06 Å) is significantly shorter than the axial Fe−OH bond
1
2
S 1-type transformation of 1 to 3, 4, 4′, or 5. Another
length (2.13 Å). The water did not induce oxidation or
oligomerization, and the catalytic activity was retained. It is
likely that water completely displaces the triflate ligands at A in
the catalytic cycle depicted in Scheme 4 as the water content in
the reaction mixtures increases during the dehydration of 1-
phenylethanol (1). Therefore, complex C could replace
complex A in Scheme 4.
N
molecule of 1 can attack the cation to form the α-methylbenzyl
ethers (AMEs) 3, 4, or 4′ (Scheme 4, step (V)). The side
reaction yielding the styrene dimer (5) likely proceeds via a
nucleophilic attack of a styrene molecule (2) at the cationic
intermediate (Scheme 4, step (VI)). While the formation of 5
was only observed at a temperature of at least 120 °C, 2, 3, 4,
and 4′ were already obtained at 100 °C. This is probably due
to the lower nucleophilicity of 2 than that of 1.
At high temperatures, A was found to react with air to form
the dinuclear iron(III) complex [μ-O{Fe(OTf)(FOX)} ]-
2
1
0888
ACS Catal. 2021, 11, 10885−10891