2
M. Goldbach et al. / Tetrahedron Letters xxx (2015) xxx–xxx
Table 1
Mg
CO2
Settling velocities of spherical magnesium particles in THF and CPME
(powder)
Br
MgBr
CO2H
Particle
diameter (lm)
Settling velocity (cm/s)
in THF (0.89 g/mL)
Settling velocity (cm/s)
in CPME (0.785 g/mL)
THF
Scheme 1. The model Grignard reaction.
100
200
250
400
1000
0.010
0.038
0.059
0.15
0.010
0.043
0.067
0.17
the reactor column and used for the low-field spectroscopy
measurements. The permanent magnetic produced a magnetic
field of 1 T (42 MHz 1H Larmor frequency) with a homogeneity of
about 0.01 ppm across the 5 mm diameter sample tube, which
completely filled the solenoidal radio-frequency coil. The signal-
to-noise ratio of a single-scan spectrum for a water sample was
about 200,000. NMR spectra of the pure solution flowing through
the magnet were measured every 10–15 s depending on the case.
Reference spectra of bromobenzene (1 M) and phenylmagne-
sium bromide (1 M) are depicted in Figure 2A while Figure 2B
shows individual spectra of the Grignard solution aromatic region
of a typical 2 h run monitored on-line at temperatures between 4
and 80 °C at a constant flow rate of 5 mL/min. From these spectra
it was concluded that within 2 h the reaction to give the Grignard
reagent was complete at Tset temperatures of 40 and 80 °C, but not
at lower temperatures due to the absence of the bromobenzene
signal in the frequency range between 7.2 ppm and 7.5 ppm. The
small signal peak near 7.4 ppm was caused by the oxidation of
the Grignard reagent. The amount of the oxidation side-product
was much smaller for the reaction product generated under flow
conditions than in a commercial sample of PhMgBr (spectra not
shown). The evolution of the aromatic region in the NMR spectra
as a function of reaction time for an incomplete run is depicted
in Figure 2C. During the two-hour experiment at a fixed Tset of
25 °C (Tactual = 26–28 °C), different flow rates were used in order
to determine in real time the influence on conversion. Under the
chosen conditions it took approximately 30 min for the Grignard
reaction to stabilize to constant conversion at a given flow rate.
The reduced conversion to the Grignard reagent in this run from
76% to 39% upon increasing the flow rate from 2 to 6 mL/min can
also be observed in Figure 2D and Table 2, entries 6–8.
0.95
1.00
therefore focused on the use of a fluidized bed tube reactor with
spherical magnesium particles ranging in size from 100 to
1000 lm. In the vertically positioned tube reactor the magnesium
particles are kept as a fluidized bed by the upward solvent flow.
The settling velocities—the solvent flow velocity to prevent the
magnesium particles from settling—in tetrahydrofuran (THF) or
cyclopentyl methyl ether (CPME) differ with particle diameter, sol-
vent density, and flow rate (Table 1).
As a model Grignard reaction for the continuous flow setup
(Fig. 1) the preparation of phenyl magnesium bromide from bro-
mobenzene in THF or CPME was chosen, followed by quenching
with gaseous CO2 to give benzoic acid (Scheme 1).
A stainless steel vertical tube reactor with 75 mL volume and a
diameter of 1.1 cm together with a heating/cooling jacket was con-
nected to a 1.2 mL flow cell by 1/800 Teflon tubing. The temperature
of the reaction was controlled by a Lauda thermostat containing
Marlotherm SH medium. The model substrate was fed to the reactor
using a Gilson 307 HPLC pump. The nitrogen flushed tube reactor
was loaded with 10 g (0.41 mol) of magnesium powder (250 lm
particles for fluidized bed conditions). To activate the magnesium
particles, commercially available phenylmagnesium bromide
(PhMgBr) solution in THF (1 M, 50 mL) was added by syringe via
the top of the tube. Subsequently the column was sealed, connected
to the in-line flow NMR spectrometer, heated to the reaction tem-
perature (80 °C), and left for 30 min. A solution of bromobenzene
in dry THF (1 M) was then pumped through the reactor at
5 mL/min. A mild exothermic effect of +3 °C was visible. The reac-
tion mixture of the first 30 min was discarded before collecting
the reaction data, by leading a fraction of the solution through the
flow cell at 1 mL/min. Excess flow was collected under nitrogen
for GC comparison and further reaction of the Grignard with
gaseous CO2.
The conversions of the Grignard reaction in flow experiments at
different temperatures and flow rates are summarized in Table 2.
NMR conversion rates were obtained by a weighted fit of the aro-
matic region of the measured spectra using as base functions the
reference samples of bromobenzene (1 M) and Grignard reagent
(1 M) in THF (Fig. 2A). In the run shown in Table 2, complete con-
version to the Grignard reagent was observed at Tset of the reaction
For analysis of the continuous flow process,
a Magritek
SpinsolveTM NMR spectrometer was connected to the output of
Figure 1. Scheme (A) and photo (B) illustrating the experimental setup of the continuous flow reactor with the in-line bench-top NMR spectrometer (SpinsolveÒ by Magritek)
under a fume hood.