Scale-up synthesis of the dopamine uptake inhibitor GBR-12909

Article abstract of DOI:10.1021/op020211j

1-[2-[Bis(4-fluorophenyl)methoxy]ethyl]-4-(3-phenylpropyl)piperazine (GBR-12909) is a dopamine uptake inhibitor. The development of a robust process for the preparation of this compound in kilogram quantities is described. The primary aims of the development work were to eliminate chromatographic purifications, to minimize the use of environmentally unacceptable reagents, and to improve the overall yield of the three-step convergent process. These objectives were met, with significant improvements obtained in the key coupling reaction of N-(3-phenylpropyl)piperazine dihydrochloride salt with 1-[bis(4-fluorophenyl)methoxy]-2-chloroethane, which was previously low-yielding and lacking in reproducibility.

Full text of DOI:10.1021/op020211j

Organic Process Research & Development 2002, 6, 621−627
Scale-Up Synthesis of the Dopamine Uptake Inhibitor GBR-12909
Michael D. Ironside,* Priyantha M. Sugathapala, Jerod Robertson, Mark C. P. Darey, and Jianzhong Zhang
Chemical DeVelopment Department, Albany Molecular Research Inc., 21 Corporate Circle, PO Box 15098,
Albany, New York 12212, U.S.A.
Abstract:
1-[2-[Bis(4-fluorophenyl)methoxy]ethyl]-4-(3-phenylpropyl)piper-
azine (GBR-12909) is a dopamine uptake inhibitor. The
development of a robust process for the preparation of this
compound in kilogram quantities is described. The primary
aims of the development work were to eliminate chromato-
graphic purifications, to minimize the use of environmentally
unacceptable reagents, and to improve the overall yield of the
three-step convergent process. These objectives were met, with
significant improvements obtained in the key coupling reaction
of N-(3-phenylpropyl)piperazine dihydrochloride salt with
Figure 1.
1-[bis(4-fluorophenyl)methoxy]-2-chloroethane, which was pre-
viously low-yielding and lacking in reproducibility.
diamine moieties can be introduced instead of piperazine,
and it has been demonstrated that the unsubstituted phenyl
ring can be replaced with a thiophene, furan, or pyridine
ring without compromising the affinity and selectivity in
many of these analogues.12,13
To support planned clinical trials, kilogram quantities of
GBR-12909 were required. The existing synthesis, supplied
by the National Institute on Drug Abuse (Scheme 1), which
had been used to deliver initial small quantities, was
evaluated critically and was deemed suitable for scale-up.
However, it was clear that a number of major process issues
needed to be resolved to allow successful manufacture of
kilogram batches.
The dihydrochloride salt formation of both intermediate
2 and GBR-12909 employed undesirable reagents and
produced difficult-to-filter solids, which required vacuum-
drying prior to proceeding. A further obstacle to scale-up
was the necessity for column chromatographic purification
of intermediate 4. Finally, the key synthetic step, namely
the coupling of intermediates 2 and 4, was low-yielding and
not reproducible.
Introduction
Cocaine, a powerful drug of abuse, is known to bind to
all three monoamine transporter systems in the brain which
mediate the neuronal uptake of dopamine (dopamine trans-
porter, DAT), serotonin (SERT), and norepinephrine (NET).1-3
The reinforcing effect of cocaine is believed to be initiated
by binding to, and thus causing inhibition of, DAT.4-6 As a
result, this protein has become one target for the development
of the medications for treating cocaine addiction. Develop-
ment targeting the DAT has resulted in the generation of
some very potent and selective molecules of diverse
structure.7-9 In 1980, Van der Zee and co-workers developed
the GBR series of compounds in which the tropine moiety
of benztropine was replaced by a substituted piperazine.9
Some well-known GBR compounds, GBR-12935 and its
bisfluorinated analogue GBR 12909 (vanoxerine), were
shown to have high potency for the DAT (Figure 1).10,11
Structure-activity relationship (SAR) studies in this class
of compounds have revealed that different cyclic and acyclic
Results and Discussion
* Corresponding author. E-mail: michaeli@albmolecular.com. Telephone:
(518) 464-0279. Fax: (518) 464-0289.
Preparation of N-(3-Phenylpropyl)piperazine Dihy-
drochloride Salt (2). In the original process, N-(3-phenyl-
propyl)piperazine free base was formed by reacting 1-bromo-
3-phenylpropane 1 with an excess of piperazine (10 equiv)
in refluxing acetonitrile for 2 h. The large excess of
piperazine was required to minimize competing polyalky-
lation side reactions. It was found that the reaction temper-
ature needed to be carefully maintained at 75-80 °C to
(1) Reith, M. E. A.; Meisler, B. E.; Sershen, H.; Lajtha, A. Biochem. Pharmacol.
1986, 35, 1123.
(2) Ritz, M. C.; Cone, E. J.; Kuhar, M. J. Life Sci. 1990, 46, 635.
(3) Javitch, J. A.; Blaustein, R. O.; Snyder, S. H. Mol. Pharmacol. 1984, 26,
35.
(4) Kuhar, M. J.; Ritz, M. C.; Boja, J. W. Trends Neurosci. 1991, 14, 299.
(5) Robinson, T.; Barridge, K. C. Brain. Res. ReV. 1993, 18, 249.
(6) Ritz, M. C.; Lamb, R. J.; Goldberg, R.; Kuhar, M. J. Science 1987, 237,
1219.
(7) Carroll, F. I.; Lewin, A. H.; Boja, J. W.; Kuhar, M. J. J. Med. Chem. 1992,
35, 969.
(8) Chaudieu, I.; Vignon, J.; Chicheportiche, M.; Kamenka, J.-M.; Trouiller,
G.; Chicheportiche, R. Pharmacol. Biochem. BehaV. 1989, 32, 699.
(9) Van der Zee, P.; Koger, H. S.; Gootjes, J.; Hespe, W. Eur. J. Med. Chem.
1980, 15, 363.
(10) Anderson, P. H. J. Neuro. Chem. 1987, 48, 1887.
(11) Anderson, P. H. Eur. J. Pharmacol. 1989, 166, 493.
(12) Matecka, D.; Lewis, D.; Rothman, R. B.; Dersch, C. M.; Wojnicki, F. H.
E.; Glowa, J. R.; DeVries, C.; Pert, A.; Rice, K. C. J. Med. Chem. 1997,
40, 705.
(13) Matecka, D.; Rothman, R. B.; Radesca, L.; De Costa, B.; Dersch, C. M.;
Partilla, J. S.; Pert, A.; Glowa, J. R.; Wojnicki, F. H.; Rice, K. C. J. Med.
Chem. 1996, 39, 4704.
10.1021/op020211j CCC: $22.00 © 2002 American Chemical Society
Published on Web 07/18/2002
Vol. 6, No. 5, 2002 / Organic Process Research & Development
•
621

Scheme 1
firmed by HPLC analysis of filtrates. No optimization of
solvent volumes to improve product recovery was attempted.
Nonetheless, sufficient material was in hand, and time
constraints did not permit a full study at that point.
Although the process had been demonstrated to scale-up
smoothly, efforts were made to further improve the yield
and throughput. Initial development focused on identification
of the major process impurities. HPLC analysis of the
reaction mixture showed a 3:1 ratio of desired product to a
single major impurity, subsequently identified as the dialky-
lated piperazine. Formation of this byproduct clearly had a
deleterious effect on the final yield but was otherwise not a
major issue since it was efficiently removed in the filtrate
during dihydrochloride salt formation. It was shown that
levels of this impurity could be reduced somewhat by
employing an extended addition time for the 1-bromo-3-
phenylpropane. As a result of this simple change, the isolated
yield could be increased to 72%.
Further development work centered on investigating
alternative solvents that could serve for both the coupling
reaction between 1 and piperazine and for the subsequent
dihydrochloride salt formation. This strategy was attractive
since it would eliminate a solvent exchange. It was discov-
ered that both methyl ethyl ketone and n-butyl acetate were
satisfactory solvents for both conversions. However, neither
approach was scaled up beyond a 100-g batch size since the
isolated yields obtained in the dihydrochloride salt formation
were lower than those observed with acetonitrile as the
coupling reaction solvent.
In an attempt to improve volume efficiency, alternative
means of removing the unreacted piperazine were investi-
gated. It was discovered that by cooling the reaction mixture
to 30 °C, a mobile slurry formed and the bulk of the excess
piperazine could be easily removed by filtration. HPLC
analysis confirmed that no product was lost in this filter cake.
This approach dramatically reduced the volume of water
necessary in the subsequent workup, thereby improving
volume efficiency.
Preparation of 1-[Bis(4-fluorophenyl)methoxy]-2-chlo-
roethane (4). In the original process, compound 4 was
prepared by heating 4,4′-difluorobenzhydrol in 2-chloroet-
hanol at 90 °C in the presence of molecular sieves and
p-toluenesulfonic acid. The reaction was quenched by adding
ethyl acetate, cooling to 0-5 °C, and then adding aqueous
sodium bicarbonate solution. The product was extracted into
ethyl acetate and concentrated under reduced pressure to
remove the solvent and residual 2-chloroethanol. The crude
product was purified by column chromatography to >85%
purity by HPLC analysis. A major issue with this procedure
was the formation of interfacial solids and emulsions upon
extraction with ethyl acetate; therefore the replacement of
this solvent prior to scale-up was essential. In addition, the
distillation to remove excess 2-chloroethanol was not trivial,
and subsequent chromatographic purification was tedious and
impractical at a large scale.
minimize the sublimation of piperazine, which was observed
at higher reaction temperatures. Upon complete consumption
of 1, the reaction was worked up by adding ethyl acetate
and water (21 vols) to the reaction mixture at 50 °C. Attempts
to improve throughput by reducing this large volume of water
in the workup were unsuccessful since lesser volumes failed
to completely dissolve the unreacted piperazine. The free
base was extracted into ethyl acetate, and the extracts were
then concentrated under vacuum to afford the product as an
oil. Dissolving the free base in 1,4-dioxane and then adding
4 M hydrochloric acid in 1,4-dioxane subsequently formed
the dihydrochloride salt 2, which was isolated by filtration,
although the brown-colored solid proved to have very poor
filtration characteristics and only 80-85% purity by HPLC
analysis. Crude 2 was subsequently purified by reslurrying
in heptane followed by vacuum-drying the heptane-wet filter
cake. Finally, a recrystallization from methanol-acetone and
subsequent vacuum-drying afforded dihydrochloride salt 2
of acceptable quality (>95% purity by HPLC) in 45% yield.
Initial development work focused on eliminating the
solvent 1,4-dioxane, a suspected carcinogen,14 from the
process and finding a safer, more cost-efficient alternative
to 4 M hydrochloric acid in 1,4-dioxane. It was demonstrated
that the dihydrochloride salt 2 could be efficiently formed
by dissolving the N-(3-phenylpropyl)piperazine free base in
ethanol and treating the resultant solution with 2.2 equiv of
concentrated hydrochloric acid at 20-30 °C. Addition of
acetone as an antisolvent and chilling the resultant suspension
to 0-5 °C afforded crude 2 as an easily filterable white solid
in >95% purity by HPLC analysis. The crude wet cake could
be further purified by recrystallization from methanol-
acetone to furnish 2 in 65% overall yield and >97% purity
by HPLC analysis.
The improved process was scaled up, and 1740 g of 2
was isolated (52% yield). The lower than anticipated yield
was due to product losses incurred during both the dihydro-
chloride salt formation and recrystallization steps, as con-
Development work revealed that heptane could be utilized
instead of ethyl acetate to simplify the extraction step. The
modified procedure involved diluting the reaction mixture
(14) Ninth Report on Carcinogens (PB2000-107509, 2000) III-122.
622
•
Vol. 6, No. 5, 2002 / Organic Process Research & Development

Scheme 2
with heptane, filtering to remove molecular sieves, and then
quenching with sodium bicarbonate at 30-40 °C. The
subsequent separations were clean, and the 2-chloroethanol,
which has low affinity for heptane, remained in the aqueous
layer. Removal of heptane under reduced pressure afforded
4 in >95% purity by HPLC analysis, which was suitable
for use in the following step without further purification.
This obviated any requirement for column chromatography,
which was previously a serious bottleneck. The improved
process was immediately scaled up to afford 2511 g of 4 in
93% yield.
Further development work aimed at eliminating 2-chlo-
roethanol as solvent was carried out on this reaction. A
procedure was developed in which compound 4 was syn-
thesized in almost quantitative yield by heating a solution
of 4,4′difluorobenzhydrol 3 and one equivalent of 2-chlo-
roethanol in toluene, with concentrated sulfuric acid as a
catalyst. This resulted in a 98% yield of the desired
compound and a facile workup since removal of excess
2-chloroethanol was not an issue. Heptane was also dem-
onstrated to be a suitable reaction solvent, which conferred
the added advantage that the batch did not need to be
concentrated to an oil, since residual heptane can be removed
azeotropically using methyl ethyl ketone (MEK), the reaction
solvent for the final step.
Preparation of 1-[2-[Bis(4-fluorophenyl)methoxy]ethyl]-
4-(3-phenylpropyl)piperazine (GBR-12909). The original
process was a one-pot coupling reaction between compounds
2 and 4 in refluxing acetonitrile, catalyzed by sodium iodide,
to afford GBR-12909 free base. Formation of GBR-12909
dihydrochloride salt was subsequently achieved by treatment
of a solution of the free base in 1,4-dioxane with 4 M
hydrochloric acid in 1,4-dioxane in 25-30% overall yield.
Consequently, this reaction also suffered from the same
drawbacks as the previous hydrochloride salt formation (vide
supra).
Investigative work showed that chloroether 4 reacted
extremely slowly in the absence of sodium iodide. It was
also demonstrated that the conversion of chloroether 4 to
the more reactive iodoether 5 was inefficient in acetonitrile,
thereby contributing to the slow reaction rate. Consequently,
it was envisioned that by cleanly preforming the iodo
derivative and then coupling with compound 2, a quicker
reaction should result. Acetone and acetonitrile proved to
be poor solvents for the conversion of chloroether 4 to its
corresponding iodoether 5 (Scheme 2), whereas MEK proved
to be an excellent solvent for this transformation.
With a procedure to cleanly generate the reactive iodo-
ether 5 in hand, the coupling reaction was carried out using
potassium carbonate, the base utilized in the original process.
Unfortunately, the coupling still proved to be extremely slow,
presumably due to the heterogeneous nature of the reaction
mixture. Typically, the reaction required at least 72 h at 80
°C to reach completion.
2.2 equiv of concentrated hydrochloric acid at 20-30 °C
and then cooling to 0-5 °C. The salt was conveniently
isolated as a white solid by filtration. This new procedure
eliminated 1,4-dioxane from the process and simplified
product isolation by eliminating a solvent-exchange step. The
crude product was further purified by recrystallization from
aqueous ethanol to afford the target molecule. The improved
process was scaled up to afford 1470 g of GBR-12909 in
45% yield. The yield was lower than anticipated, due to
product losses in the mother liquors incurred during the
hydrochloride salt formation and recrystallization steps, as
evidenced by HPLC analysis. No optimization of solvent
volumes to improve product recovery was attempted, how-
ever, since an improved process was already in hand (Scheme
2).
Further process work revealed that a significant improve-
ment in both reaction time and yield could be achieved
utilizing homogeneous reaction conditions. Thus, a solution
of the iodoether 5 in MEK was heated at 80 °C with a
solution of salt 2, dissolved in aqueous sodium carbonate.
Typically the reaction was complete within 18 h. Workup
was facile, involving a phase separation and subsequent back-
extractions of the aqueous phase with MEK to isolate the
desired free base. The resulting extracts could be treated
directly with concentrated hydrochloric acid to furnish the
crude dihydrochloride salt in 55-60% yield. Improved yields
and better reproducibility could however be obtained by
conducting the hydrochloride salt formation in ethanol as
solvent, which necessitated a solvent swap after reaction
completion. Using this procedure, crude GBR-12909 dihy-
drochloride was isolated in 70% yield on a 100-g scale.
Impurity Identification. HPLC analysis of the GBR-
12909 batch indicated the presence of a single impurity at
>0.1% (by area normalization). LC/MS analysis indicated
an [M + 1] value of 451.5 for the main peak, corresponding
to the molecular weight of GBR-12909 minus 2 mol of
hydrochloric acid. The impurity peak yielded an [M + 1]
value of 432.5, corresponding to the monofluoro analogue
of GBR-12909 minus 2 mol of hydrochloric acid. Presumably
this impurity emanated from the presence of 4-fluoroben-
After aqueous workup, the GBR-12909 free base was
converted to its dihydrochloride salt. As per intermediate 2,
it was discovered that crude dihydrochloride salt could be
formed more efficiently by treating the organic extracts with
Vol. 6, No. 5, 2002 / Organic Process Research & Development
•
623

Scheme 3
zhydrol 7 in the 4,4′-difluorobenzhydrol 3 starting material.
To confirm the structural assignment of this impurity, the
monofluoro analogue of GBR-12909 was prepared on a
small-scale following a similar route used for the synthesis
of GBR-12909 (Scheme 3). An HPLC spiking experiment
using this independently synthesized monofluoro GBR-12909
confirmed that it was indeed the impurity detected in the
batch of GBR-12909. Subsequently, a batch of monofluoro
GBR-12909 was prepared for evaluation in clinical trials
using the route outlined in Scheme 3. Many of the improve-
ments developed for the original GBR-12909 process were
incorporated. Gratifyingly, a marked improvement in yield
for the final coupling step (70%) was recorded on a kilogram
scale using an aqueous base.
as the solvent. Infrared spectra were obtained as KBr pellets
on a Perkin-Elmer Spectrum 1000 infrared spectrophotom-
eter. Mass spectra analyses were performed on a Shimadzu
QP-5000 GC/MS (CI mass spectrometry). Melting points
were obtained on a Mettler Toledo Star DSC. Elemental
analyses were obtained from Quantitative Technologies, Inc.,
Whitehouse Junction, NJ. Silica gel for flash chromatography
was purchased from E-M Scientific (mesh 230-400). Thin-
layer chromatography (TLC) was performed using 2.5 × 10
cm Analtech silica gel GF plates (25-µm thick). Visualization
of TLC plates was performed using ultraviolet light.
Initial Scale-Up of N-(3-Phenylpropyl)piperazine Di-
hydrochloride Salt (2). A solution of piperazine (9804 g,
113.80 mol) in acetonitrile (12 L) was heated to 75-80 °C.
To this solution was added 1-bromo-3-phenylpropane (1820
mL, 11.97 mol) dropwise over a period of 4 h. Upon
completion of the addition, no residual 1-bromo-3-phenyl-
propane was detected by HPLC analysis. The reaction was
quenched with water (50 L), followed by extraction into ethyl
acetate (20 L). The aqueous layer was separated and back-
extracted with ethyl acetate (2 × 10 L). The combined
organic extracts were washed with saturated aqueous sodium
chloride solution (15 L), dried over anhydrous sodium sulfate
(1 kg), filtered, and concentrated under reduced pressure at
40-45 °C to afford 2678 g of the free base as a thick oil.
The free base was dissolved in ethanol (7.2 L), treated with
concentrated hydrochloric acid (2.4 L) at 20-30 °C, and
stirred for 1 h. The batch was diluted with acetone (22 L) at
20-30 °C and the resultant slurry was stirred for 2 h prior
to cooling to 0-5 °C. The crude dihydrochloride salt was
filtered and found to be 95% pure by HPLC (AUC). The
filter cake was dissolved in refluxing methanol (9 L), and
the resultant solution was cooled to 50-55 °C and diluted
with acetone (20 L). The slurry was cooled to 0-5 °C, then
filtered, and the filter cake was washed with acetone (3 L),
then dried under vacuum at 40 °C to constant weight to afford
1739 g (52% yield) of compound 2 in 97% purity by HPLC
(AUC). 1H NMR (300 MHz, CD3OD) δ 7.30-7.16 (m, 5H),
3.69-3.51 (b, 5H), 3.30-3.21 (m, 3H), 2.79-2.69 (m, 3H),
and 2.20-2.08 ppm (m, 3H). HPLC method: Waters
Symmetry C18, 3.5 µm, 2.1 mm × 150 mm, 0.25 mL/min,
267 nm, 15% CH3CN:85% water to 90% CH3CN:10% water
Conclusions
In summary, the original synthetic strategy was employed
although several significant improvements were incorporated
to allow smooth scale-up and give higher yields for all steps.
The undesirable solvent 1,4-dioxane and the reagent 4 M
hydrochloric acid in 1,4-dioxane were replaced with ethanol
or MEK and concentrated hydrochloric acid, respectively,
in the hydrochloride salt formation reactions. The hydro-
chloride salts formed via the new process displayed improved
filtration characteristics and were of higher purity than those
produced via the original route. The introduction of heptane
as the extraction solvent in the ether formation step elimi-
nated a problematic phase separation and a chromatographic
purification step. Preformation of the iodoether 5 prior to
the last coupling step facilitated the formation of GBR-12909.
In the coupling step, MEK proved to be a superior solvent
than acetonitrile for the reaction, providing a cleaner reaction
profile by HPLC analysis. The use of an aqueous base in
the key final coupling step was key to achieving an improved
reaction rate and yield versus the original heterogeneous
conditions.
Experimental Section
Reagents and solvents were obtained from commercial
sources and used as received. Proton magnetic resonance
spectra were obtained on a Bruker AC-300 spectrometer at
300 MHz using either dimethyl sulfoxide-d6 or chloroform-d
624
•
Vol. 6, No. 5, 2002 / Organic Process Research & Development

over 30 min. MS m/z 205 [M + 1] (free base).
(100 mL) over a period of 45 min. The resulting light-yellow
solution was heated to 85 °C and stirred for 3 h until only
trace amounts of 4,4′-difluorobenzhydrol were detected by
TLC analysis in 80:20 heptane/ethyl acetate. The reaction
mixture was cooled to 15-25 °C, diluted with toluene (100
mL), and neutralized with potassium bicarbonate (powder)
to pH 6-7. The mixture was filtered, and the filtrate was
washed with water (3 × 300 mL) and concentrated under
reduced pressure at <40 °C to afford 102.3 g (98% yield)
of compound 4 as a yellow oil in 99% purity by HPLC
(AUC). 1H NMR (300 MHz, CDCl3) δ 7.32-7.21 (m, 4H),
7.05-6.97 (m, 4H), 5.36 (s, 1H), and 3.71-3.62 ppm (m,
4H). HPLC method: Waters Symmetry C8, 5 µm, 3.9 mm
× 150 mm, 1.0 mL/min, 267 nm, 30% CH3CN with 0.1%
TFA:70% water to 90% CH3CN with 0.1% TFA:10% water
over 30 min.
Final preparation of N-(3-Phenylpropyl)piperazine
Dihydrochloride Salt (2). To a solution of piperazine (500.0
g, 5.5 mol) in acetonitrile (1 L) at 75-80 °C was added
1-bromo-3-phenylpropane (85 mL, 0.56 mol) over a period
of 8 h. The mixture was then cooled to 15-25 °C and filtered
to remove piperazine. The filter cake was washed with
acetonitrile (250 mL), and the filtrate was concentrated under
vacuum at <40 °C to furnish a thick, mobile slurry. The
concentrate was diluted with ethyl acetate (500 mL) and
water (500 mL). The organic phase was separated, and the
aqueous layer was back-extracted with ethyl acetate (3 ×
200 mL). The combined organic extracts were washed with
saturated aqueous sodium chloride (250 mL) and then
concentrated under vacuum at <40 °C to afford the free base
as a thick oil. The concentrate was dissolved in ethanol (250
mL), treated with concentrated hydrochloric acid (110 mL),
at 20-30 °C, and stirred for 1 h. The batch was diluted with
acetone (700 mL) at 20-30 °C, and the resultant slurry was
stirred for 2 h. The batch was cooled to 0-5 °C and then
filtered; the filter cake was washed with acetone (150 mL)
and then dried under vacuum at 40 °C to constant weight to
afford 111.5 g (72% yield) of compound 2 in 98.4% purity
Initial Scale-Up of 1-[2-[Bis(4-fluorophenyl)methoxy]-
ethyl]-4-(3-phenylpropyl) Piperazine Dihydrochloride
(GBR-12909). A mixture of compound 4 (1948 g, 6.89 mol),
sodium iodide (2782 g, 18.5 mol), and MEK (17 L) was
heated at 80 °C for 24 h, at which point less than 5% residual
chloroether 4 was detected by HPLC analysis. The reaction
mixture was cooled to 25 °C and filtered through a Celite
pad to remove the inorganic salts. To the filtrate was added
potassium carbonate (5217 g) and salt 2 (1739 g, 6.27 mol).
The batch was heated at 80 °C for 68 h, at which point there
was <10% residual iodoether 5 by HPLC (AUC). The
reaction mixture was cooled to <35 °C, quenched with water
(3.5 L) and saturated aqueous sodium bicarbonate (7 L). The
organic layer was separated, and the aqueous layer was back-
extracted with MEK (3 × 3.5 L). The combined organic
extracts were washed with saturated aqueous sodium chloride
solution (5 L) and then treated with concentrated hydrochloric
acid (1.5 L) and stirred at 20-30 °C for 2 h. The reaction
mixture was then cooled to 0-5 °C and stirred for an
additional 4 h to afford a thick slurry. The crude GBR-12909
was isolated by filtration in reasonable yield (2800 g wet)
and purity 99.0% (AUC by HPLC analysis). The crude
product was purified by dissolution in 5% aqueous ethanol
(25 L) at 80 °C. The solution was cooled to 50-60 °C,
clarified by filtration (to remove extraneous matter), and then
diluted with acetone (34 L). The resultant slurry was cooled
to 0-5 °C and filtered. The filter cake was dried at 45 °C
until the residual ethanol and acetone levels were <0.5%
by GC analysis, to afford 1470 g (45% yield) of GBR-12909
dihydrochloride as a white solid with 99.8% purity by HPLC
1
by HPLC (AUC). H NMR (300 MHz, CD3OD) δ 7.30-
7.16 (m, 5H), 3.69-3.51 (b, 5H), 3.30-3.21 (m, 3H), 2.79-
2.69 (m, 3H), and 2.20-2.08 ppm (m, 3H). HPLC method:
Waters Symmetry C18, 3.5 µm, 2.1 mm × 150 mm, 0.25
mL/min, 267 nm, 15% CH3CN:85% water to 90% CH3CN:
10% water over 30 min.
MS m/z 205 [M + 1] (free base).
Initial Scale-Up of 1-[Bis(4-fluorophenyl)methoxy]-2-
chloroethane (4). A mixture of 4,4′-difluorobenzhydrol
(2100 g, 9.5 mol), 2-chloroethanol (19 L), p-toluenesulfonic
acid (1800 g), and molecular sieves (4 Å) (125 g) was heated
at 85 °C for 10 h. At this point no 4,4′-difluorobenzhydrol
was detected by TLC analysis in 80:20 heptane/ethyl acetate.
The reaction mixture was cooled to 25 °C, diluted with
heptane (12 L), and filtered through a Celite pad to remove
solids. The resulting filtrate was washed with saturated
aqueous sodium bicarbonate solution (3 L) and stirred for
30 min. The aqueous layer was separated and back-extracted
with heptane (3 L). The combined organic extracts were
washed with saturated aqueous sodium chloride solution (3
L), dried over sodium sulfate (500 g), filtered, and concen-
trated under reduced pressure to afford 2511 g (93% yield)
of ether 4 as a yellow oil in 97% purity by HPLC (AUC).
1H NMR (300 MHz, CDCl3) δ 7.32-7.21 (m, 4H), 7.05-
6.97 (m, 4H), 5.36 (s, 1H), and 3.71-3.62 ppm (m, 4H).
HPLC method: Waters Symmetry C8, 5 µm, 3.9 mm × 150
mm, 1.0 mL/min, 267 nm, 30% CH3CN with 0.1% TFA:
70% water to 90% CH3CN with 0.1% TFA:10% water over
30 min.
1
(AUC): mp ) 228-233 °C; H NMR (300 MHz, DMSO-
d6): δ 7.42-7.05 (m, 13H), 5.56 (s, 1H), 3.82-3.54 (m,
10H), 3.3-3.29 (m, 4H), 2.75-2.71 (m, 2H), and 2.11-
2.09 (m, 2H); 13C NMR (300 MHz, DMSO-d6): δ 163.4,
160.2, 140.9, 138.3, 138.2, 129.1, 129.0, 128.8, 128.6, 126.5,
115.7, 115.5, 81.9, 63.1, 55.6, 48.9, 48.3, 40.7, 40.4, 40.2,
39.9, 39.3, 39.0, 32.3, 25.1; IR (KBr) 3433, 2947, 2390,
1603, 1507, 1453, 1222, 1155, 1098, and 1014 cm-1. Anal.
Calcd for C28H34Cl2F2N2O: C, 64.24; H, 6.55; N, 5.35.
Found: C, 63.91; H, 6.47; N, 5.27. HPLC method: Waters
Symmetry C8, 5 µm, 3.9 mm × 150 mm, 1.0 mL/min, 267
nm, 30% CH3CN with 0.1% TFA:70% water to 90% CH3-
Final Preparation of 1-[Bis(4-fluorophenyl)methoxy]-
2-chloroethane (4). A mixture of 2-chloroethanol (37.0 mL,
555 mmol), toluene (60 mL), and sulfuric acid (6.5 mL, 131.0
mmol) was gently heated to 40 °C and treated with a solution
of 4,4′-difluorobenzhydrol (81.26 g, 369 mmol) in toluene
Vol. 6, No. 5, 2002 / Organic Process Research & Development
•
625

CN with 0.1% TFA:10% water over 30 min. MS m/z 451
[M + 1] (free base).
4.6 mm × 250 mm, 1.0 mL/min, 220 nm, 15% CH3CN with
0.05% TFA:85% water with 0.05% TFA to 100% CH3CN
with 0.1% TFA over 25 min. MS m/z 185 [M - OH].
1-[(4-Fluorophenyl)phenylmethoxy]-2-chloroethane (8).
A solution of 4-fluorobenzhydrol 7 (970 g, 4.80 mol) in
heptane (1810 mL) and dichloromethane (100 mL) at 40 °C
was added over 1 h in 270-mL portions to a solution of
2-chloroethanol (440 mL, 528.4 g, 6.56 mol, 1.37 equiv) and
concentrated sulphuric acid (73 mL, 1.3 mol, 0.27 equiv) in
heptane (800 mL), keeping the temperature at 35-40 °C.
After the addition was complete, the temperature was
increased to 80-85 °C, and the batch was stirred at this
temperature for 5 h. The reaction was complete as judged
by TLC in 80:20 heptane/ethyl acetate. The batch was cooled
to 15-25 °C, and washed with saturated sodium bicarbonate
solution (2 × 6 L) and purified water (3 × 6 L). The batch
was then concentrated under vacuum at 45 °C. The concen-
trate was dissolved in MEK (3 L) and the batch reconcen-
trated under vacuum to give 1-[(4-fluorophenyl)phenyl-
methoxy]-2-chloroethane (1254 g, 98.7%) as an oil in 95%
Final Preparation of 1-[2-[Bis(4-fluorophenyl)methoxy]-
ethyl]-4-(3-phenylpropyl) Piperazine Dihydrochloride
(GBR-12909). A mixture of compound 4 (93.6 g, 331.1
mmol), sodium iodide (71.0 g, 473.4 mmol), and MEK (800
mL) was heated at 80 °C for 24 h, at which point less than
5% residual chloroether was detected by HPLC analysis. The
reaction mixture was cooled to 25 °C and filtered through a
pad of Celite to remove the inorganic salts. The filtrate was
treated with a solution of sodium carbonate (53.3 g), salt 2
(82.0 g, 260.0 mmol), and water (400 mL). The reaction
mixture was heated at 75-80 °C for 24 h until complete by
HPLC (<10% iodoether 5) and then cooled to <35 °C. The
aqueous layer was separated and back-extracted with MEK
(3 × 400 mL). The combined organic extracts were washed
with saturated aqueous sodium chloride solution (400 mL)
and concentrated under vacuum at <40 °C to a volume of
100 mL. The concentrate was dissolved in ethanol (450 mL),
clarified by filtration (to remove extraneous matter) and
treated with concentrated hydrochloric acid (60 mL) at 15-
25 °C. The batch was stirred at 15-25 °C for 2 h, treated
with acetone (800 mL), cooled to 0-5 °C, and stirred for
another 2 h. The product was collected by filtration, washed
with chilled acetone (2 × 80 mL), and dried under vacuum
at 45 °C until the residual ethanol and acetone levels were
<0.5% by GC analysis. A total of 108.4 g (70% yield) of
GBR-12909 dihydrochloride was isolated as a white solid
in 99.2% purity by HPLC (AUC). A second crop [8.5 g
(5.5% yield)] of GBR-12909 dihydrochloride was obtained
upon vacuum concentration of the filtrate at <45 °C to a
volume of 40 mL, although of slightly lower purity at 97.8%
1
purity by HPLC (AUC). H NMR (300 MHz, CDCl3) δ
7.35-7.21 (m, 5H), 7.05-6.97 (m, 4H), 5.41 (s, 1H), and
3.74-3.62 ppm (m, 4H). HPLC method: Phenomenex
Synergi Hydro RP, 4 µm, 4.6 mm × 250 mm, 1.0 mL/min,
220 nm, 15% CH3CN with 0.05% TFA:85% water with
0.05% TFA to 100% CH3CN with 0.1% TFA over 25 min.
MS m/z 185 [M - OCH2CH2Cl].
1-[2-[(4-Fluorophenyl)phenylmethoxy]ethyl]-4-(3-phe-
nylpropyl)piperazine Dihydrochloride (Monofluoro GBR-
12909). 1-[(4-Fluorophenyl)phenylmethoxy]-2-chloroethane
8 (1227 g, 4.63 mol) was added to a mixture of MEK (10.5
L) and sodium iodide (1672 g, 11.15 mol) and heated at 80
°C for 24 h. After this time, the amount of residual
chloroether 8 was 8.3% by HPLC analysis. The batch was
cooled and filtered through a pad of Celite to remove the
inorganic salts. Sodium carbonate (679 g, 6.4 mol) was
dissolved in purified water (5.2 L), and N-(3-phenylpropyl)-
piperazine dihydrochloride salt 2 (1045 g, 3.77 mol) was
added carefully to this solution, avoiding excessive frothing.
Once the N-(3-phenylpropyl)piperazine dihydrochloride salt
had dissolved, the aqueous solution was added to the filtered
batch, and the mixture was heated to 75-80 °C. The batch
was stirred at this temperature for 22.5 h, after which time
the amount of residual iodoether 9 was 9.6% by HPLC. The
batch was cooled to below 40 °C, and the lower aqueous
phase was separated. The aqueous phase was extracted with
MEK (3 × 4 L), and the combined organic layers were
washed with saturated sodium chloride solution (3 L). The
batch was concentrated under vacuum at 45 °C and the
concentrate dissolved in ethanol (5 L). Concentrated hydro-
chloric acid (731 mL, 7.3 mol) was added over 1 h,
maintaining the batch temperature at 15-30 °C. The batch
was then stirred at 10-25 °C for 2 h prior to addition of
acetone (10.5 L), which induced crystallization of the
dihydrochloride salt. The resultant slurry was cooled to 0-5
°C, stirred for 16 h in this temperature range, and then
filtered. The filter cake was washed twice with chilled
acetone (2 L). The purity of the wet cake was 93.3% by
1
(AUC). H NMR (300 MHz, DMSO-d6): δ 7.42-7.05 (m,
13H), 5.56 (s, 1H), 3.82-3.54 (m, 10H), 3.3-3.29 (m, 4H),
2.75-2.71 (m, 2H), and 2.11-2.09 ppm (m, 2H); IR (KBr)
3433, 2947, 2390, 1603, 1507, 1453, 1222, 1155, 1098, and
1014 cm-1. Anal. Calcd for C28H34Cl2F2N2O: C, 64.24; H,
6.55; N, 5.35. Found: C, 63.91; H, 6.47; N, 5.27.
4-Fluorobenzhydrol (7). A solution of 4-fluorobenzophe-
none 6 (966 g, 4.83 mol) in 2-propanol (12 L) was treated
with sodium borohydride (70 g, 1.85 mol). The reaction
mixture was stirred for 16.5 h, whereupon TLC analysis in
80:20 heptane/ethyl acetate showed no starting material and
a single product. The reaction mixture was concentrated
under reduced pressure and the crude product partitioned
between water (12 L) and ethyl acetate (6 L). The layers
were separated, the aqueous layer was extracted with ethyl
acetate (5 L), and the combined organic layers were washed
with saturated aqueous sodium chloride (5 L), dried over
sodium sulfate (200 g), filtered, and concentrated under
reduced pressure to afford 4-fluorobenzhydrol 7 as a light
orange solid (982 g, 98% theory) of 99% purity by HPLC
(AUC). Analytical data was consistent with data published
in the literature.15 1H NMR (300 MHz, CDCl3) δ 7.36-7.23
(m, 5H), 7.08-6.97 (m, 4H), 5.83 (s, 1H), and 2.22 ppm (s,
1H). HPLC method: Phenomenex Synergi Hydro RP, 4 µm,
(15) Sell, M. S.; Hanson, M. V.; Rieke, R. D. Synth. Commun. 1994, 24(16),
2379.
626
•
Vol. 6, No. 5, 2002 / Organic Process Research & Development

HPLC (AUC). The wet cake was dissolved in a mixture of
ethanol (8 L) and purified water (1.4 L) at 80 °C. The hot
solution was clarified by filtration, cooled to 40-45 °C, and
stirred at this temperature for 30 min. Acetone (14 L) was
added over 1 h, and the batch was cooled to 0-5 °C in an
ice/ethanol bath and stirred for 30 min prior to filtration.
The filter cake was washed twice with prefiltered acetone
(5 L). The batch was dried under vacuum at 40 °C to afford
1-[2-[(4-fluorophenyl)phenylmethoxy]ethyl]-4-(3-phenylpro-
pyl)piperazine dihydrochloride (1334 g, 70%) as a white solid
Anal. Calcd for C28H35Cl2FN2O: C, 66.53; H, 6.98; N, 5.54.
Found: C, 66.76; H, 6.95; N, 5.39. Mp 226-229 °C. HPLC
method: Phenomenex Synergi Hydro RP, 4 µm, 4.6 mm ×
250 mm, 1.0 mL/min, 220 nm, 15% CH3CN with 0.05%
TFA:85% water with 0.05% TFA to 100% CH3CN with
0.1% TFA over 25 min. MS m/z 433 [M + 1] (free base).
Acknowledgment
We thank Thomas Gregory and Andrew Ignatz for
providing analytical support and Harold Meckler and Paul
Vogt for helpful discussions with regards to scale-up of this
chemistry. In addition we are indebted to Farahnaz Moham-
madi for her invaluable assistance in drafting this article.
This work was supported by NIDA Contract N01DA-8-8092.
1
in >99.8% purity by HPLC (AUC). H NMR (300 MHz,
DMSO-d6): δ 7.52-7.10 (m, 14H), 5.58 (s, 1H), 3.92-3.35
(m, 10H), 3.3-3.05 (m, 4H), 2.75-2.63 (m, 2H), and 2.11-
2.00 ppm (m, 2H); 13C NMR (300 MHz, DMSO-d6): δ
163.6, 160.5, 141.9, 140.9, 138.4, 129.2, 129.0, 128.9, 128.8,
128.6, 127.9, 127.0, 126.5, 115.7, 115.4, 82.7, 49.0, 40.7,
40.4, 40.2, 39.9, 39.3, 39.0, 32.3, 25.1; IR (KBr) 2971, 2358,
1598, 1504, 1453, 1373, 1331, 1217, 1098, and 1017 cm-1.
Received for review March 27, 2002.
OP020211J
Vol. 6, No. 5, 2002 / Organic Process Research & Development
•
627