titration and was within specification. All proton NMR
spectra were recorded on a Varian 300 or 400 MHz
spectrometer with TMS as internal standard. Large-scale
2 3
mmol), K CO (13.27 g, 96 mmol), tetrabutylammonium
bromide (TBAB, Fluka, 0.26 g, 0.8 mmol), MIBK (40 mL),
and 1-bromododecane (Acros, 12.37 g, 50 mmol) were added
to a 100-mL three-necked flask. Subsequently, the reaction
mixture was heated to reflux and stirred for 2 h. The
conversion of the reaction was monitored by TLC (eluent
3
reaction was carried out in a 10-dm fully automated (semi)-
batch-wise operated reactor. This Belatec reactor is able to
perform under a variety of conditions: a temperature range
from -50 to 200 °C, solvent distillation, different agitator
types, a pressure range from 0.03 to 1.1 bar. The reactor is
controlled by a PLC, a special computer, which monitors
physical data (batch history), and secures optimal process
and safety conditions.
Determination of Particle Sizes of Potassium Carbon-
ate. Potassium carbonate was ground for 1 min and was
separated into fractions by sieving. Four different sieves were
used in the separation, which took 0.5 h: 315, 250, 125,
and 80 µm. SEM photomicrographs were taken from all of
the five fractions, which were used to determine the average
particle diameter. Each photo presented a representative
amount of particles of a fraction. The boundaries of the
fractions were checked by visual inspection of the SEM-
photos and turned out to be in good agreement with the
particle range shown. The average particle size was calcu-
lated by determining the projected area of the particles on
the SEM-photo and treating them like spheres.
1
hexane//ethyl acetate 24:1) and H NMR. Upon completion
the brown mixture was cooled below 100 °C, and water (40
mL) was added. The aqueous layer was separated and the
organic layer was washed with water (40 mL), diluted HCl
solution (40 mL 1.0 M), and water (40 mL) again. Solvent
evaporation from the resulting organic layer gave a yellow
oil (11.7 g), which crystallises at approximately 40 °C to
afford a light brown solid (2). Purification by column
2
chromatography (flash SiO ; eluent hexane//ethyl acetate 96:
1
4
) yielded a white powder (10.2 g): mp 43.2-43.8 °C. H
NMR (acetone-d
6
, 300 MHz): δ ) 7.27 (s, 2H, o-H), 4.11-
), 3.85 (s, 3H, OCH ), 1.90-1.70 (m,
), 1.55-1.20 (m, 54H, (CH ), 0.86 (t, 9H,
). Anal. Calcd for C44 (MW 689.12): C 76.69, H
1.70. Found: C 77.3, H 11.8.
Alternative Purification of Methyl 3,4,5-tri-dodecyl-
3
6
CH
1
.98 (m, 6H, OCH
H, OCH CH
2
3
2
2
2 8
)
3
80 5
H O
oxybenzoate (2). Crude yellow 2 (27.4 g) was heated above
its melting point and poured gently into 1 L of methanol
Procedure for Just-Suspended Stirrer-Rate Measure-
ments. The experiments were performed on three different
scales. Geometrically similar glass reactors were built having
(tech) under stirring. The initial yellow clusters were
dispersed, and a white precipitate was formed within 1 h.
Filtration yielded 97% pure (based on NMR analysis)
material, yield: 25.1 g ) 91%.
3
volumes of 1, 10, and 50 dm , respectively. The reactor
3
configurations are presented in Table 3. On 1-dm scale in
Master Recipe to Synthesise 3,4,5-Tri-dodecyloxyben-
each experiment 500 mL of MIBK and 166 g of K
2
CO
3
were
3
zoic Acid (3). The recipe on 10-dm scale has been described
3
used. A similar ratio was used on 10-dm scale, where on
5
previously.
3
3
5
2 3
0-dm scale 20 dm of MIBK and 6.005 kg of K CO were
Analysis of FBRM Data. The FBRM probe gives a chord
length distribution in graphical form. The maximum value
of the distribution, which is an indication of the particle size,
can easily been read from the data. We measured all fractions
used. In the synthesis of 2 at the onset of the reaction the
3
same ratio is used as in the 1- and 10-dm configurations,
but when bromododecane has been added to the mixture,
the solid/liquid ratio will decrease, and this has a positive
influence on the just-suspended stirrer rate. Ten experiments
2 3
of K CO with the FBRM probe to determine whether the
obtained distributions were comparable with sphere-like
particles. We have concluded that in our case the FBRM
data can be used to give an approximation of the real particle
size distribution.
3
were performed on 1- and 10-dm scale. In these experiments
the stirrer type and the particle size were varied (see five
fractions of potassium carbonate as given in Table 1). On
3
1
0-dm scale no experiments were performed with particle
sizes of 80-125 µm for practical reasons. Also the number
of baffles was varied during the experiments. No major
distinction in the final results was observed. Consequently,
only the two-baffled system was considered. Just the pitched-
blade impeller and the fractions with the smallest and largest
Acknowledgment
This research is supported by Professor E. W. Meijer
(Laboratory of Macromolecular and Organic Chemistry,
TUE, Netherlands) and DSM Fine Chemicals, Netherlands.
We gratefully acknowledge the assistance of Mr. J. Janssen
of Lasentec during experimentation with the FBRM and
PVM Lasentec techniques. We also thank Ing. J. J. H. M.
Lemmen of DSM Fine Chemicals, the Netherlands, for his
comments, based on an in-house validation study, and
suggestions regarding the Q-factor.
3
particle sizes were used in two experiments on 50-dm scale.
In all experiments the bottom of the reactor was illuminated
by a light bulb, and observations were made by visual
inspection. For this reason all observed values of the just-
suspended stirrer rate are the mean values of five to ten
observations starting alternatively from high or low stirrer
speed which then had been decreased and increased respec-
tively until the off-bottom condition was met.
Received for review January 16, 2002.
OP020007W
Procedure for the Synthesis of Methyl 3,4,5-tri-dodec-
yloxybenzoate (2). Methyl gallate (1) (Fluka, 2.95 g, 16
Vol. 6, No. 5, 2002 / Organic Process Research & Development
•
651