H. Iken et al. / Tetrahedron Letters 53 (2012) 3474–3477
3475
scribes the multigram preparation and characterization of ILs con-
stituted of alkylpyridinium cations (Fig. 1) substituted by various
alkyl radicals, and bromide anion; a brief discussion on the effect
of the cation size on the ILs physico-chemical properties is pre-
sented. Syntheses of the most representative IL were carried out
both in a stirred classical reactor and in a heat exchanger/micro-
structured reactor (HEx/MSR),9 designed in our laboratory. Com-
parison of mass balance results for various conditions was
performed in terms of conversion rate and selectivity, in order to
evaluate the interest to use HEx/MSR at production scale.
of temperature to 110 °C leads to a significant decrease in the
viscosity.
A strong influence of the temperature was observed in the case
of [OcDePy][Br] (assumed to be a non-Newtonian IL), in compari-
son with the [OcPy][Br] (assumed to be a Newtonian IL). Arrhenius
analysis allows to determine the viscosity activation energy:
37.9 kJ/mol for [OcPy][Br] (Newtonian IL) and 17.2 kJ/mol for [Oc-
DePy][Br] (non-Newtonian IL) and confirms the influence of the
temperature in the case of the Newtonian IL.
In order to compare the performance of stirred reactor and HEx/
MSR, the synthesis of octylpyridinium bromide was chosen as the
model reaction, because of the appropriate physical properties of
this IL (low viscosity and melting point, non-hygroscopic behav-
ior). Pyridine alkylation with 1-bromooctane was achieved under
equimolar conditions and mass balances were performed under
various operating conditions of temperature and residence time.
The efficiency of both reactors for syntheses at the multigram scale
The stainless steel heat exchanger/microstructured reactor
(
HEx/MSR) is composed of a plate (250 mm  120 mm) with rect-
angular (1 mm  1 mm) micro-channels (Fig. 2). The total volume
of the microstructured plate (including inlet and outlet chambers)
3
is 10.2 cm . The specific area of the MSR is S/V = 251/10.2 = 24.6 cm
À1
.
The reactants were introduced at rt into the reactor by a two
syringe pump, and distributed over the micro-channels by appro-
priate geometry of the inlet channel. The appropriate shape and
geometry of the inlet compartment allows a good mixing of re-
agents, so the solution was assumed to be homogeneous before
reaching the reaction area (microchannels). A similarly designed
collecting channel allows removing the reaction mixture.
We first embarked in the synthesis in batch reactor of a series of
alkylpyridinium salts (Fig. 1); although most of them were already
described in the literature,10 we needed to examine by ourselves a
wide range of physico-chemical properties (Table 1) in order to se-
lect the best example for comparing synthetic performances of
batch vs MSR. Experimental and analytical data are given in Sup-
plementary data. The most significant properties for our studies
are melting points and viscosity, which are directly relevant to IL
behavior, and are discussed below.
1
was compared in terms of conversion rate by H NMR.
In the stirred reactor, the operating procedure was as follows:
the flask was charged with an equimolar amount of pyridine and
1-bromooctane (0.10 ± 0.01 mol) under N atmosphere, and the
2
stirred solution was heated to a fixed temperature using a pre-
heated oil bath. When the reaction duration was reached, the flask
was cooled to room temperature and then the content was diluted
1
in chloroform-d to be analyzed by H NMR. The first experiment
was carried out at 60 °C for reaction durations ranging from 0.8
to 4 h. The results (Fig. 4(left)) show that the residual mole number
of pyridine linearly decreases against the reaction duration. This
means that the apparent reaction order of pyridine is zero, and
the estimated reaction rate is constant and equal to 2.7 Â 10
À5
À1 À1
mol L
s . During the reaction, the formation of IL involves
Melting points of synthesized ILs are lower than the boiling
point of water (100 °C) except for butylpyridinium bromide
the apparition of two distinct liquid phases: the lower phase com-
prises the IL and the upper one contains the residual reagents,
which remain practically in the same concentration throughout
the reaction because no mixing with the synthesized IL occurs. In
order to validate these results, a second reaction was carried out
at 70 °C. As expected, the decrease of residual mole number of pyr-
idine was more important than at 60 °C. Compared to 60 °C, the
conversion rate increases at least 2 times with a reaction rate of
(Fig. 3). The phase transition temperature of ILs (from solid to li-
quid phase) is governed by Van der Waals and electrostatic inter-
1
1
action forces. The melting points fall steadily with increasing
alkyl chain length from 4 carbons to reach a minimum value of
1
8 °C for 8 carbons; when the carbon number exceeds 8 the melt-
ing points increase (Fig. 3). At the shortest chain length, the melt-
ing point is mainly governed by ionic interactions that decrease
with the increase of the size of the cation: the melting point de-
creases when the alkyl chain length increases. However, at the lon-
gest chain length, Van der Waals interactions become prevalent,
increasing the melting point. ILs are generally viscous liquids,
and viscosity is mainly governed by Van der Waals interactions
as well as intra or extra molecular H bonding. Higher length alkyl
chain exalts these phenomena,12 and increases the viscosity due to
reduced rotation freedom of the IL. Viscosity measurements were
carried out versus shear rate for the synthesized ILs under two dif-
ferent experimental conditions: (1) at 5 °C above melting point for
each IL; (2) at 110 °C, a temperature where all ILs are liquid. The
viscosity increases with the number of carbon atoms in the alkyl
group (Fig. 3). Worthy of note is that the viscosity of ILs containing
a low carbon number alkyl group does not vary with increasing
shear rate, depicting a Newtonian behavior. On the opposite, for
À5
À1 -1
7.9 Â 10 mol L s , suggesting that the reaction is temperature
limited under these conditions. The increase of the working tem-
perature to 80 °C led to a significant increase in the IL rate produc-
À4
À1 À1
tion with a reaction rate of 1.5 Â 10 mol L
s . For higher
temperatures (Fig. 4(left), curve at 100 °C), the evolution of resid-
ual mole number of pyridine versus time was not linear, indicating
that the apparent reaction order of pyridine is not zero anymore.
Higher temperature promotes the solubilization of pyridine within
the IL, consequently the concentration of pyridine changes during
the increase of the volume of synthesised IL. Conversion of pyridine
reaches 90% after 4 h of reaction at this temperature. The estimated
1
1
values of the apparent rate constant (kapp = k[Py] [Alk] , alkylation
follows SN2 mechanism) allow access to the real rate constant of
the alkylation. Arrhenius analysis of k versus temperature led to
À1
the activation energy value of 85 kJ mol , a relatively high value.
As for the reaction in stirred reactor, the conversion rates of IL
1
[
DoDePy][Br] and [OcDePy][Br], increase of the shear rate causes
obtained with HEx/MSR were determined by H NMR, and results
the viscosity to decrease asymptotically, which indicates a shear
thinning behavior. Moreover, except for [OcPy][Br], the increase
are presented versus residence time in Figure 4(right). The increase
of temperature from 60 to 100 °C also involves an increase in the
relative conversion rate, confirming that the reaction is tempera-
ture limited. The obtained curves (npy=f(s)) at various tempera-
tures are not linear indicating an apparent reaction order close to
zero (order = 0.1). Nevertheless, in almost all experiments, conver-
sion rate were higher from HEx/MSR: for example at 1.3 h reaction
time np StirredR/np MSR = 1.04, 1.08, 1.12 for respectively 60, 70, 80 °C,
showing that the HEx/MSR was significantly more efficient than
the stirred one. For temperatures higher than 100 °C there are no
Br
N
n
n = 2, 4, 6, 10, 16
Figure 1. Structure of the alkylpyridinium salts.