initiation. The exotherm of the reaction was then easily
controlled by adjusting the addition rate of the remaining
aryl bromide solution.
The Grignard reagent derived from 3 was found to be
quite stable. Stress tests in the lab demonstrated that it could
be stored under a nitrogen atmosphere overnight with only
minimal decomposition. This was an important consideration
since the process planned for our pilot plant would require
that the Grignard reagent be stored for that period of time.
Control of the Grignard-generating stage of the process
was accompanied with a shift in focus to the carboxylation
stage. Early experiments used heptane as the solvent to slurry
the dry ice. Under these conditions, the yield of carboxylic
acid was not acceptable. Examination of the mass balance
of the reaction mixture indicated that as much as 15% of
the Grignard reagent was converted back to starting dimethox-
ytoluene. Analysis of the reaction mixture for water content
and an investigation into the stability of the Grignard reagent
eventually revealed that the loss in yield occurred during
the aqueous workup. Addition of the Grignard reagent
mixture to heptane resulted in precipitation of the Grignard
reagent. This observation led to the conclusion that the
Grignard reagent precipitated during the quench into the dry
ice/heptane mixture. The precipitated Grignard reagent was
then not able to react with the carbon dioxide and remained
as an insoluble mass until water was added, when it dissolved
and was quenched by water.
Further refinement of the reaction conditions revealed that
preparing the dry ice slurry in tetrahydrofuran resulted in
less dimethoxytoluene. Subsequent experiments showed that
the product carboxylic acid would precipitate from the
reaction mixture upon addition of 3 N HCl. Dimethoxytolu-
ene was not entrained by the product and remained entirely
in the filtrate.
The process was scaled up to the 5-kg scale, and afforded
the desired product in 72% yield. The filtration was rather
slow, and fines passed though the cake and filter screen.
Experiments in the lab showed that the recovery of product
could be increased if the tetrahydrofuran was partially
removed prior to acidifying the reaction mixture. Dilute
sulfuric acid was substituted for hydrochloric acid to allow
stainless steel filters to be used if necessary.
Figure 2. Reaction profile for reduction of 2,3-dimethoxyben-
zaldehyde to 2,3-dimethoxytoluene.
aid such as diatomaceous earth or cellulose (Solka-Floc) and
removing the solvent at ambient pressure.
The next stage in developing the process was to couple
the reduction stage together with the bromination step. By
telescoping the reduction and bromination together, the
solvent requirement and unit operations would be reduced,
resulting in significant cost savings and better throughput.
This was easily accomplished, and as an added benefit, the
bromination was even more selective in ethyl acetate,
affording the desired regioisomer in a selectivity of 73:1. In
the pilot plant, 10 kg of dimethoxybenzaldehyde was reduced
to dimethoxytoluene as described above. The catalyst was
removed via an in-line cartridge filter and the filtrate added
directly to a reactor which contained DBDMH in ethyl
acetate. When the bromination was complete, the reaction
mixture was quenched with an aqueous sodium sulfite
solution. Dropping the lower aqueous layer out of the reactor
and switching the solvent to a methanol-water mixture
afforded the desired aryl bromide 3 in a yield of 96% from
dimethoxybenzaldehyde. The purity of the isolated product
was greater than 99% when measured using HPLC.
The aryl bromide also had desirable solid-state properties.
Batch sizes of 12-14 kg were easily filtered using a 13-gal
Nutsche filter within 5 min. Drying of the product was
accomplished by simply passing a stream of filtered nitrogen
through the cake overnight.11
With a route to the required aryl bromide secured,
attention was shifted to formation of the Grignard reagent
and to the quench with dry ice. Several excellent reports are
available regarding the safe generation and scale-up of
These changes were incorporated into the process, and
the chemistry was demonstrated on a 20-kg scale. Once
again, formation of the Grignard reagent was smooth when
methylmagnesium chloride was used catalytically. The
generated Grignard reagent was then quenched into a mixture
of tetrahydrofuran and dry ice. The dry ice slurry resulted
12,13
Grignard reagents.
On the basis of these reports, a process
that involved controlled addition of the aryl bromide to a
slurry of magnesium metal in tetrahydrofuran was developed.
Initial laboratory experiments were quite promising, with the
Grignard reaction being initiated smoothly at ambient tem-
perature. However, when aryl bromide more representative
of the pilot-plant batches was used, significant induction
times were observed. Addition of catalytic amounts of
methylmagnesium chloride (in tetrahydrofuran) to the mix-
ture of magnesium and aryl bromide led to more consistent
14
in an internal reactor temperature of -60 °C, which allowed
the reaction mixture to be quenched relatively quickly. Under
these conditions, the reaction temperature rose to ca. 0 °C.
Attempts to cool the Grignard reagent solution prior to the
quench resulted in precipitation of the Grignard reagent. One
interesting observation was the need to precool the reactor
used to prepare the dry ice slurry prior to charging dry ice.
The large surface area of the reactor and ambient temperature
(
(
(
11) Aryl bromide 3 melts at 55 °C; therefore, drying the product on the filter
was a desirable option.
12) am Ende, D. J.; Clifford, P. J.; DeAntonis, D. M.; SantaMaria, C.; Brenek,
S. J. Org. Process Res. DeV. 1999, 3, 319-329.
13) Leazer, J. L., Jr.; Cvetovich, R.; Tsay, F.-R.; Dolling, U.; Vickery, T.;
Bachert, D. J. Org. Chem. 2003, 68, 3695-3698.
(14) A reactor rated for low temperature was used to prepare the dry ice and
THF mixture.
626
•
Vol. 8, No. 4, 2004 / Organic Process Research & Development