Development of the route of manufacture of an oral H1-H 3 antagonist

Article abstract of DOI:10.1021/op1002598

A new route to an H₁-H₃ antagonist was developed to address scalability and environmental and cost of goods issues associated with the initial route.

Full text of DOI:10.1021/op1002598

Organic Process Research & Development 2011, 15, 112–122
Development of the Route of Manufacture of an Oral H1-H3 Antagonist
Guillaume Bret,‡ Sandra J. Harling,† Karim Herbal,† Nathalie Langlade,† Mike Loft,‡ Alan Negus,‡ Mahesh Sanganee,†
Steve Shanahan,‡ John B. Strachan,*,† Peter G. Turner,† and Matthew P. Whiting†
Synthetic Chemistry, GlaxoSmithKline Pharmaceuticals, Medicines Research Centre, Gunnels Wood Road, SteVenage SG1 2NY, U.K.,
and Synthetic Chemistry, GlaxoSmithKline Pharmaceuticals, Old Powder Mills, Leigh, nr Tonbridge TN11 9AN, U.K.
Abstract:
A new route to an H1-H3 antagonist was developed to address
scalability and environmental and cost of goods issues associated
with the initial route.
to degradation of the pTSA catalyst under the forcing conditions
of the process (either by conversion to pTSA anhydride or by
Introduction
desulphonation). Laboratory studies showed that 14 would
Compound 10 is an H1-H3 antagonist which is being
convert to 12 with a recharge of pTSA. This approach led to a
developed as an oral medicine for the treatment of allergic
90% reduction of the diamide level with formation of more
rhinitis. The initial route to this compound is shown in Sch-
product.
Parallel studies showed that the impure plant batch could
There were a number of issues associated with this synthesis
be reworked by selective crystallisation and filtration of 14. In
which precluded its use as the long-term route of manufacture.
this way the contaminant could be reduced to <2%. These
Use of p-iodophenol generates iodine-containing waste streams
changes were successfully implemented, and yields of at least
and associated high disposal costs. The use of a BOC protecting
84% were obtained for subsequent batches with <2% diamide
group for the piperidone raises issues of thermal stability and
impurity.
eme 1.
the generation of isobutylene as a byproduct. The cryogenic
The original method for reduction of compound 12 with
temperature and low yield for the Grignard reaction was
lithium aluminium hydride presented two main challenges: an
unsatisfactory and the resulting tetrahydropyridine 6 had limited
unattractive filtration of the alumina byproduct through a
stability. The use of expensive, thermally unstable TBTU as
Hyflow bed and a final distillation of 1-benzyl-3,3-dimethyl
the final coupling agent was also of concern. Neither the
piperidine (13) under high vacuum at 150-200 °C. The first
piperidine (3) nor naphthalene (7) fragments were commercially
issue was resolved by quenching the reaction mixture with 40%
available. Finally, the later stage intermediates were all non-
w/w aqueous sodium hydroxide and separating the layers at
crystalline, hindering the development of scalable isolations.
60 °C. Resolution of the second issue required significant capital
Indeed the drug substance (10) itself was poorly crystalline,
investment, and so the decision was made to telescope this stage
displaying a wide metastable zone width in potential crystal-
with the final debenzylation. This was reinforced by the fact
lisation solvents. We therefore initiated a programme of work
that the reaction mixture was reasonably free of impurities with
to improve the synthetic route in preparation for large-scale
the exception of low level formation of enamine 15 (presumably
manufacture.
from dehydration of the hemiaminal intermediate) which could
be hydrogenated in the next stage.
Results and Discussion
Preparation of 3,3-Dimethylpiperidine. 3,3-Dimethylpi-
peridine (3) was initially prepared by cyclisation of com-
mercially available dimethylglutaric acid (11) with benzylamine
followed by reduction with lithium aluminium hydride and then
debenzylation (Scheme 2).
The cyclisation was conducted at 100-kg scale in o-xylene
under Dean-Stark conditions in the presence of p-toluenesul-
The process was conducted in pilot plant on a 38-kg scale
phonic acid (pTSA) catalyst. Despite very good results in the
with an overall yield of 72%. All reactions proceeded smoothly
laboratory, initial scale-up led to unexpected formation of around
and were complete within two hours. Compound 13 was
10 mol % of diamide 14.
isolated as a THF solution after alkaline workup.
The exact reason for the elevated levels of 14 are unclear;
Debenzylation of 13 had previously been accomplished using
however, one could postulate that it could be directly related
1-chloroethyl chloroformate (ACE-Cl) in 60% yield. This
chemistry was considered undesirable for a pilot-plant campaign
due to the moderate yield and environmental issues associated
with handling the reagent and its byproducts, benzyl chloride
and acetaldehyde.
* Author to whom correspondence may be sent. E-mail: john.b.
strachan@gsk.com.
† GlaxoSmithKline Pharmaceuticals, Medicines Research Centre, Gunnels
Wood Road.
‡ GlaxoSmithKline Pharmaceuticals, Old Powder Mills.
112
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development
10.1021/op1002598  2011 American Chemical Society
Published on Web 11/30/2010

Scheme 1. Initial synthesis of compound 10
Scheme 2. Initial supply route to 3,3-dimethylpiperidine
aluminium hydride reduction was not stable under acidic
conditions. The hydrogenation was therefore carried out se-
quentially. First, the enamine was reduced (alkene reduction
was rapid), and second, acetic acid was charged and the
hydrogenation restarted to cleave the N-benzyl protecting group.
Distillation was chosen as the preferred purification method for
3 over isolation by salt formation and crystallisation. The latter
was investigated; however, the presence of basic impurities
prevented a robust method from being developed.
Although the above methodology successfully afforded 70
kg of 3, a more efficient approach was sought. Michael addition
of isobutyraldehyde (16) to acrylonitrile (17) produced 2,2-
dimethyl-4-cyanobutyraldehyde (18) in an unoptimised yield
of 45%. 18 was converted to compound 3 in an unoptimised,
one-pot nitrile reduction-reductive amination sequence (Scheme
3).2 This route was demonstrated on a 150-g scale in the
laboratory. The cost of goods is significantly cheaper with this
route, and with only two chemical steps, it also potentially offers
considerable savings in processing time.
Debenzylation of 13 by hydrogenation had previously been
reported on small scale in the literature, offering the prospect
of a clean and scalable transformation.1 Initial results showed
that the hydrogenation was stalling, probably due to catalyst
poisoning by the amine product. This issue was easily overcome
with the addition of acetic acid to the hydrogenation mixture.
However, the enamine impurity generated in the lithium
Phenol Alkylation and Grignard Reaction. p-Bromophe-
nol is a cheaper alternative to p-iodophenol which would avoid
the iodine waste issue and was expected to participate success-
fully in the desired chemistry. Sequential coupling with bro-
(1) See for example, Berardi, F.; Ferorelli, S.; Colabufo, N. A.; Leopoldo,
M.; Perrone, R.; Tortorella, V. Bioorg. Med. Chem. 2001, 9, 1325.
(2) Schreyer, R. C. J. Am. Chem. Soc. 1952, 74, 3194.
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
113

Scheme 3. Concise route to 3,3-dimethylpiperidine
1
Scheme 4. Addition of aryl lithium reagent to N-benzyl piperidin-4-one, conversions determined by H NMR analysis of the
crude reaction product
mochloropropane and 3,3-dimethylpiperidine yielded the desired
bromide. In the initial process, potassium carbonate was used
in two different solvents with an intermediate isolation. By using
aqueous sodium hydroxide and MIBK under phase transfer
conditions the two transformations could be effected without
an intermediate workup. In addition, the reaction time was
reduced from 24 to 5 h. The major byproduct was the diether
19 formed by over-reaction of bromophenol with bromochlo-
ropropane. No reaction conditions could be found which
eliminated its formation; however, it was easily removed by
acid-base extraction.
n-BuLi adds to the piperidone, giving rise to butyl piperidine
derivative 24. Low levels of the des-bromo compound 22 were
also observed, presumably arising from competitive deproto-
nation of the piperidone.
As tertiary alcohol 21 was a gum, we decided to telescope
the crude reaction mixture directly into the deprotection step
(Scheme 5). We found we were able to remove the N-benzyl
group and concomitantly reduce out the benzylic alcohol directly
by hydrogenolysis over palladium on carbon in the presence
of a strong acid. Specifically hydrogenolysis at 3 bar hydrogen
pressure using 0.2 weights of 50% wet 10% Pd-C (∼3 mol
% Pd) in 19:1 EtOH/water with 3 equiv of methanesulphonic
acid gave clean conversion to the desired product in 8 h at 65
°C. In the absence of added acid or in the presence of weaker
acids such as acetic only N-debenzylation was observed. Unlike
the tetrahydropyridine 6, this compound proved stable.
Unfortunately, neither the methanesulphonic acid salt
nor the free base was suitable for isolation. We were,
however, able to show that the benzoic acid salt could be
readily crystallised from a number of solvents with TBME
being shown to give a good cleanup of all the impurities.
One minor issue with this crystallisation was that the
product was obtained as a mixture of mono- and diben-
zoate salts, due to the lower solubility of the monoben-
zoate. This was not considered to be a major issue at this
stage, and the process was progressed to the plant.
The aryl bromide 20 was initially purified by acid base
extraction to remove residual dimer 19. The pure solution
in TBME was azeotropically dried before being transferred
into a solution of n-BuLi in THF and hexanes and then
treated with benzyl piperidone. After quenching and
aqueous workup, solvent exchange to ethanol for the
hydrogenolysis was achieved by distillation. Aqueous
workup under basic conditions enabled transfer of the free
base to a TBME phase. This was again azeotropically dried
We were encouraged by literature examples suggesting that
a high-yielding Grignard reaction could be achieved at non-
cryogenic temperatures using a benzyl protecting group in place
of a BOC group.3 This would also avoid the associated thermal
instability and isobutylene issues and provide a cheaper starting
material.
As the aryl bromide intermediate 20 was not reactive enough
to be metalated under Knochel conditions, we turned to
bromine-lithium exchange using n-BuLi in hexanes (Scheme
4). This was successful; however, a significant amount of the
butylated product 23 was obtained by reaction of the aryl lithium
species with the bromobutane byproduct. This could be mini-
mised by using two equivalents of n-BuLi and conducting the
reaction at -20 °C. Under these conditions the majority of the
bromobutane formed is consumed by reaction with the excess
n-BuLi. Use of a larger excess was counterproductive as residual
(3) See for example Sakamuri, S.; Enyedy, I.; Kozikowski, A.; Zaman,
W.; Johnson, K.; Wang, S. Bioorg. Med. Chem. Lett. 2001, 11, 495.
114
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development

Scheme 5. Telescoped alkylation-hydrogenolysis process
Table 1. Alkylation selectivities for different ArM species,
(instead of the desired addition), and this was confirmed
by quenching the Grignard reagent with R,R,R,R-d4
deuterated N-benzylpiperidone, where we observed a large
kinetic isotope effect for the deprotonation and also
significant levels of deuterium incorporation into 22.
The ArMgCl reagent derived from aryl iodide 4 by iodine
magnesium exchange using iPrMgCl had previously been
shown to give highly selective addition, with only a trace of
the deprotonation product 22 being observed. This implied that
the halide ion has an effect on the reactivity of these Grignard
reagents. Formation of the ArMgCl and ArMgBr species under
otherwise identical reaction conditions showed that this was the
case, with the ArMgBr reagent being significantly less selective
(Scheme 6). It is in fact known that the rate of aryl Grignard
addition to ketones can be highly dependent on the halide ion.6
Formation of the ArMgX reagent from the aryl iodide had
been previously ruled out by cost and waste issues, and direct
formation from the aryl chloride was not practical due to very
slow Mg insertion. We therefore looked to optimise the ArMgBr
reaction. It is well-known that solvent and concentration can
play a large role in the reactivity of Grignard reagents thought
to be due, in large part, to alteration of their aggregation states.7
In order to carry out Mg insertion efficiently, THF was required
as solvent. The effect of alternative solvents on the selectivity
was therefore investigated by dilution of a concentrated THF
solution of the Grignard reagent with a range of solvents prior
to quenching with the piperidone. Of those used, only toluene
gave a beneficial effect. Further experiments showed that
increasing both the toluene/THF ratio and the reaction concen-
tration further improved the selectivity for addition. An ad-
ditional increase was gained by initiating the Mg insertion using
iPrMgCl, instead of TMSCl,8 which served to remove any
residual water from the solvent and Mg surface prior to Grignard
formation. Using the information gained, the selectivity for
addition vs deprotonation could be increased from ∼3:1 to
>10:1.
1
determined by H NMR analysis of the crude reaction
products
T
21
22
23
entry
ArM
ArLi
Ar3MgLi
ArMgBr
solvent
(°C)
-20
0
20
(%)
(%)
(%)
1
2
3
THF
THF
THF
84
69
76
6
26
23
10
4
0
before benzoic acid was added, and the product was
isolated in 64% yield and >99% purity through a seeded
cooling crystallisation.
Although this process worked very well on scale, there
were several obvious areas which could potentially be
improved for future campaigns. First, >15% of the material
was lost in the initial alkylation as the butylated or des-
bromo byproducts 23 and 22, respectively. Second, the
process was quite lengthy, with several solvent and salt
swaps. We therefore decided to investigate both alternative
metalation and alkylation procedures to try and improve
the selectivity for the desired product, and also try to
streamline the hydrogenation and isolation procedure,
ideally removing the need for the salt and solvent swap.
Initially we looked into formation of less reactive aryl
metal species. The electron-rich aryl bromide 20 did not
undergo halogen-magnesium exchange (to give the
Grignard reagent) under Knochel-type conditions.4 How-
ever, the corresponding lithium triarylmagnesiate5 was
readily formed by addition of 0.66 equiv of n-BuLi to a
solution of the aryl bromide and 0.33 equiv of iPrMgCl.
This reagent exhibited much reduced reactivity towards
bromobutane, forming only 4% of 23, even in the absence
of excess base and at the elevated temperature of 0 °C
(Table 1). However, upon quenching with the piperidone,
a large amount of protonated impurity 22 was formed.
The bromomagnesium Grignard reagent itself was readily
formed by direct insertion of Mg in THF at 40 °C. In this
case formation of 23 was, of course, not a problem;
however, upon quenching with the piperidone a significant
amount of 22 was again observed. We assumed that this
must arise from competitive deprotonation of the ketone
We reinvestigated the use of alternative solvents for
the lithiation reaction, with the expectation that the ArLi
would be more stable in less polar solvents. The initial
(6) Holm, T. Acta Chem. Scand. B 1983, 37, 567–584.
(7) Silverman, G. S.; Rakita, P. E., Eds. Handbook of Grignard Reagents;
Marcel Dekker: New York, 1996.
(8) Ace, K. W.; Armitage, M. A.; Bellingham, R. K.; Blackler, P. D.;
Ennis, D. S.; Hussain, N.; Lathbury, D. C.; Morgan, D. O.; O’Connor,
N.; Oakes, G. H.; Passey, S. C.; Powling, L. C. Org. Process Res.
DeV. 2001, 5, 479–490.
(4) Krasovskiy, A.; Knochel, P. Angew. Chem. Int. Ed 2004, 43, 3333,
and references therein.
(5) Gallou, F.; Haenggi, R.; Hirt, H.; Marterer, W.; Schaefer, F.; Seeger-
Weibel, M. Tetrahedron Lett. 2008, 49, 5024–5027, and references
therein.
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
115

Scheme 6. Differing reactivities of ArMgCl and ArMgBr species
1
Table 2. Optimised alkylation selectivities determined by H
lithiation showed high solvent dependency. Addition of a
solution of 20 in 4 volumes of TBME to n-BuLi in 8
volumes of THF gave immediate Br-Li exchange even
at -50 °C. In contrast, addition of the bromide to n-BuLi
in TBME or toluene only gave slow lithiation at -20 °C
and required ∼1 h at 0 °C for complete reaction (Figure
1). The rate of formation of the butylated impurity (23)
was also monitored in these reactions. In THF rapid
reaction was observed with ∼50% formation of 23 over
1 h. In contrast, the reactions using TBME or toluene gave
essentially no butylation at -20 °C with only 0.5% being
formed even after 3 h at 0 °C (Figure 1 and Table 2).9
This vastly altered reactivity meant we were able to carry
out the lithiation in TBME at 0 °C using only a small excess
of n-BuLi without formation of significant levels of 23.
Disappointingly, upon addition of the piperidone we again
observed significant levels of competitive deprotonation to form
22. The product solution yield was, however, still slightly
increased compared to that from the BuLi/THF process, and
importantly the reaction appeared much more robust to time
and temperature variations. The straightforward and robust
nature of the process made this our favored option for further
development.
NMR analysis of the crude reaction products
T
21
22
23
20
entry
ArM
ArLi
ArLi
ArMgBr THF/toluene
solvent
THF
TBME
(°C) (%) (%) (%) (%)
1
2
3
-20 84
88
20 91
6
10
9
10
0
0
0
2
0
0
the catalyst, then dilution with TBME, the product could be
directly crystallised as the ditosylate salt in good purity (Scheme
7).
The improved procedures could be readily telescoped and
were demonstrated on laboratory scale, giving an unoptimised
67% isolated yield, only slightly higher than the initially scaled
process. However, it is expected this could be improved with
further development of the isolation, and importantly the new
process is significantly shorter and expected to be more robust
than that used initially.
Synthesis of Naphthalene Fragment. The naphthalene
fragment (7) was prepared by monocarboxylation of
dibromonaphthalene, followed by a Heck reaction with
benzyl acrylate (Scheme 8). We decided to continue with
the benzyl ester to avoid any problems in removing
inorganic byproducts from hydrolysis of simple alkyl
esters from the amphoteric drug substance in the down-
stream chemistry.
Isopropylmagnesium chloride and solid carbon dioxide
were used in the original method and this led to the
formation of high levels of desbromo 28, monoacid 29,
and diacid 30 impurities in the product (see below). Using
Turbo Grignard10 (1.5 equiv) gave a quicker reaction time,
Simplification of the hydrogenolysis/isolation process fo-
cused on finding a single acid that would both promote the
desired reduction and also enable direct crystallisation of the
product. The requirement of the reduction for a strong acid
limited our choices but fortunately p-toluenesulphonic acid
proved suitable. The reduction was slightly slower than using
methanesulphonic acid but crucially after filtration to remove
Figure 1. Rate of lithiation and rate of reaction of ArLi with BuBr.
116
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development

Scheme 7. Improved alkylation/hydrogenolysis telescope
carried out to identify a potential organic solvent for crystalli-
sation. Ethyl acetate gave the most promising results, and further
investigations showed that a 1.7:1 mixture of ethyl acetate/
heptane gave yield 74% on a 350-g scale.
For the Heck reaction, four solvents were screened (DMF,
NMP, toluene and Me-THF). Toluene gave the cleanest purity
profile and a complete reaction. Several catalysts (Pd(OAc)2,
PdCl2, Pd(PPh3)4 and Pd-C) and ligands (P(o-tolyl)3, PPh3 and
beta-alanine) were screened but only the combination of
Pd(OAc)2 and PPh3 gave complete reaction. It was demonstrated
that only 3 mol % of palladium acetate and 5 mol % of
triphenylphosphine were required to get complete reaction
within 6 h.
The original workup process involved several solid isola-
tions, loose charcoal treatments, and two crystallisations to
remove palladium residues. Hence, we sought to simplify the
workup and isolation. Once the reaction was complete, triethy-
lamine hydrobromide precipitates upon cooling to room tem-
perature. The reaction mixture was then filtered the cake washed
with toluene, and the liquors were extracted with water.
Hydrolysis of the product to the diacid was observed when
leaving it in contact with basic aqueous solutions for too long.
Other extractions were studied using an acidic aqueous solution
which led to the precipitation of triethylamine hydrochloride
as well as the product; basic extraction led to high levels of
debenzylation. Once the product was extracted into the aqueous
phase, an organic solvent was added to the solution and the
aqueous phase acidified to pH < 3 to extract the product into
the organic phase. Me-THF was chosen as the extractive solvent
as it could also be used for the crystallisation when combined
with an antisolvent such as heptane.
and use of dry carbon dioxide gas led to low levels of the
three described impurities, consequently giving a better
impurity profile in the product.
Final Coupling and Salt Formation. A screen of activating
agents was conducted to find a replacement for TBTU, and it
was found that thionyl chloride could be used to form the acid
chloride of 7 in either toluene or ethyl acetate. Reacting 25 with
the solution of the acid chloride, using either Hu¨nig’s base or
Schotten-Baumann conditions, gave the desired amide.
Schotten-Baumann conditions in ethyl acetate were chosen for
further development to avoid issues with removing excess
Hu¨nig’s base. On scale-up ethyl acetate was hydrolysed by the
aqueous sodium hydroxide; this issue was resolved by switching
the solvent to isopropyl acetate and changing the base to
aqueous potassium carbonate. The product was a gum, and a
salt screen revealed one isolable solid, the fumaric acid salt.
Hence, the fumarate salt 33 was isolated after a solvent swap
to isopropanol (Scheme 9). Two batches of this process were
run in the pilot plant, delivering 75 kg with an average yield of
83%.
Previously the CO2 sparging was carried out at 0 °C and
took up to 12 h on scale. Looking at the solubility of CO2 in
THF (Figure 2) suggested working at a lower temperature, and
the reaction was successfully repeated at -10 °C. At that
temperature the rate of formation of the desbromo impurity is
slower, allowing further control of its levels in solution and
consequently in the product.
Initially the product was crystallised from water by acidifica-
tion. However, in large-scale laboratory experiments the solid
showed a tendency to form a gum. A solubility study was
Scheme 8. Synthesis of the naphthalene fragment
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
117

Figure 2. Solubility of CO2 in THF.
Scheme 9. Proposed manufacturing route for compound 10
To convert 33 to 10 a reduction of the olefin and benzyl
acrylate followed by hydrochloride salt formation was required.
The API was poorly crystalline, had a wide metastable zone
and a high tendency to gum. The free base was a glassy foam,
and thus, the decision was taken to telescope the reduction and
salt formation and not include a final recrystallisation stage.
The reduction was achieved with 5% palladium on carbon, 3
bar hydrogen pressure at 50 °C in a range of solvents: TBME,
(9) For related observations: Boros, E. E.; Burova, S. A.; Erickson, G. A.;
Johns, B. A.; Koble, C. S.; Kurose, N.; Sharp, M. J.; Tabet, E. A.;
Thompson, J. B.; Toczko, M. A. Org. Process Res. DeV. 2007, 11,
899–902.
(10) Krasovskiy, A.; Straub, B.; Knochel, P. Angew. Chem., Int. Ed. 2006,
45, 159.
118
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development

IPA, IPAc, EtOAc, or MEK. The conditions were finalised once
a suitable salt formation had been developed.
equiv) added. The reaction mixture was heated to reflux for
10-12 h. The batch was cooled to 80-85 °C and the last
portion of p-toluenesulphonic acid monohydrate (2.3 kg, 0.02
equiv) added. The batch was then refluxed for 5-10 h. The
reaction was monitored by LC and mass of water evolved. The
reaction was judged complete by GC and then cooled to 15-25
°C. The mixture was transferred to a second workup reactor
and partitioned with 11%w/w aqueous potassium carbonate (300
L). After stirring for around 1 h at 15-25 °C, the layers were
allowed to separate, and the lower aqueous layer was discarded.
The remaining yellow o-xylene solution is repartitioned with 1
M HCl and stirred for 1 h at 15-25 °C before being allowed
to separate. The lower aqueous layer was again run out to waste.
The remaining o-xylene solution was then reduced in volume
and dried by distilling off o-xylene (approximately 100 L) under
reduced pressure to afford a yellow oil (140.9 kg total, 133.6
kg 12, 93% yield). 1H NMR (CDCl3, 400 MHz): δ 1.25 (s, 6
H), 1.78 (t, J ) 7 Hz, 2 H), 2.71 (t, J ) 7 Hz, 2 H), 4.93 (s, 2
H), 7.06-7.14 (m, 2 H), 7.20-7.32 (m, 3 H).
Preparation of 3,3-Dimethyl-1-(phenylmethyl)piperidine
(13). The diketopiperidine 12 solution in o-xylene (37.7 kg pure
12) was diluted with THF (360 L, 14 vol based on pure 12) at
20-25 °C under a nitrogen atmosphere. The yellow solution
obtained was heated to 60 °C ( 3 °C and a 1 M solution of
LiAlH4 in THF (295.2 kg, 2 equiv) added dropwise maintaining
the temperature below 65 °C. The resulting white slurry was
stirred for at least 2 h and analysed for completion of reaction
using HPLC. Once complete, 32% w/w aqueous NaOH (0.95
vol, 35.8 L, 47.5 kg, 1.26 equiv) was added very slowly
(hydrogen gas formation) followed by 40% w/w aqueous NaOH
(530 L, 14 vol). The biphasic mixture was then stirred for 1 h
at 60 °C ( 3 °C, and the layers separated leaving any emulsion
with the organic layer. 32% w/w aq NaOH (200 L, 5.3 vol)
was added and the biphasic mixture stirred at 60 °C ( 3 °C
for 30 min. Layers were allowed to separate, and the aqueous
layer was discarded. The THF layer was then concentrated to
4 vol at atmospheric pressure (155.5 kg total, 24.8 kg of 13,
75%). 1H NMR (CDCl3, 400 MHz) δ 0.92 (s, 6 H), 1.15-1.25
(m, 2 H), 1.54-1.61 (m, 2 H), 1.99 (s, 2 H), 2.26-2.35 (m, 2
H), 3.41 (s, 2 H), 7.00-7.35 (m, 5 H).
A solvent screen revealed that the final salt formation could
only be conducted in acetone, 2-methyl tetrahydrofuran, or
MEK. As the hydrogenation and salt formation were to be
telescoped, acetone was not considered a viable solvent due to
its miscibility with water. 2-Methyl tetrahydrofuran has the
potential to generate genotoxic impurities under the acidic
reaction conditions; thus, MEK was selected as the solvent. The
free base of 33 was formed with aqueous potassium carbonate
and the benzyl acrylate reduced with 5% palladium on carbon
(50% aqueous paste, 0.1 wt) under 2 bar hydrogen pressure at
50 °C.
The hydrochloride salt formation showed a large degree of
initial oiling. To avoid oiling, seeding was performed at lower
supersaturation by increasing the acid addition temperature to
70 °C and adding DMSO (0.75 vol) as a cosolvent. To form
the API, the solution was seeded followed by an MEK
antisolvent addition and cooled to crystallise 10. Three batches
of this process were run on plant to deliver 30 kg of API, with
an average yield of 79%.
Conclusions
The new route is shown in Scheme 9. Phase transfer catalysis
simplifiedandacceleratedthealkylationchemistry.Bromine-lithium
exchange eliminated iodide waste, and use of a benzyl protecting
group avoided cryogenic temperatures. Scalable isolations were
achieved by careful salt selection of the intermediates, and use
of the acid chloride under Schotten-Baumann conditions
simplified the downstream chemistry. In addition to addressing
all of the key sustainability and operability issues associated
with the initial process, the cost of goods was reduced by 80%.
Experimental Section
Reagents and solvents were purchased from commercial
1
suppliers and were used as received. H NMR spectra were
recorded at 400, 500, or 700 MHz. Chemical shifts are reported
in ppm on the δ scale with the residual solvent peaks or
tetramethylsilane as internal standard, respectively. Data are
presented as follows: chemical shift (ppm), multiplicity (s )
singlet, d ) doublet, t ) triplet, m ) multiplet), coupling
constant, J (Hz) and integration. HPLC-MS data were acquired
either on a Waters ZQ or LCT mass spectrometer, both fitted
with an electrospray (ESI) source and coupled to an Agilent
1100 HPLC. Mass spectra were obtained in positive ion mode.
Analyses were performed using a capillary voltage of 3.5 kV
and a cone voltage of 30 V.
Preparation of 3,3-Dimethyl-1-(phenylmethyl)-2,6-pip-
eridinedione (12). 2,2-Dimethylglutaric acid (100 kg, 1 equiv)
and p-toluenesulphonic acid monohydrate (2.3 kg, 0.02 equiv)
were placed in a reactor under an inert atmosphere. o-Xylene
(600 L) was charged and the resulting suspension heated to
reflux under Dean-Stark conditions. A solution of benzylamine
(67 kg, 1 equiv) in o-xylene (150 L) was added over 2-3 h,
maintaining reflux and azeotropic removal of water. Following
the addition, the reaction was maintained at reflux and water
continuously removed for 18 h. The batch was cooled to 80-85
°C and p-toluenesulphonic acid monohydrate (2.3 kg, 0.02
Preparation of 3,3-Dimethylpiperidine (3) Plant Process.
A solution of 13 in THF was filtered to remove NaOH
particulates prior to distillation under reduced pressure. Solvent
was removed to minimum stir volume. To the resulting oil was
added IPA (180 L) and the solvent again removed by distillation
under reduced pressure to minimum stir volume. To the
resulting oil was added IPA (143 L). THF content was checked
1
by H NMR and shown to be <5% w/w. To palladium on
carbon catalyst (3.58 kg, 0.1 wt) under an inert atmosphere was
added the IPA solution of 12. The stirred mixture was heated
to ∼55 °C and pressurised to 2.5 bar H2, and hydrogenation
continued for ∼3 h to consume the enamine 18. The vessel
was depressurised and purged as required and sample taken
(HPLC - to verify consumption of enamine). Acetic acid (∼1.1
mol equiv, 0.33 wt, 11.8 kg) was charged. The reaction mixture
was reheated to ∼55 °C and repressurised to 2.5 bar H2.
Hydrogenation was continued for 10 h before depressurising
and purging the vessel. A sample was taken (LC or GC to assess
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
119

chemical conversion). The reaction mixture was filtered and
the catalyst cake washed with IPA (53.7 L, 1.18 wt). The
resulting solution of 3 as its acetate salt was then concentrated
under reduced pressure at 70 °C and -0.7 bar. Water was added
(2 × 15 L), and further distillation removed the remaining
isopropanol at 80 °C and -0.7 bar pressure (IPA level <0.5%
M HCl (2 × 700 mL and then 1 × 200 mL). TBME (800 mL)
was added to the combined acid layers and heated to about 50
°C, and then 10 M NaOH (150 mL) was added. The aqueous
layer was separated, and then the TBME layer was heated to
reflux and 500 mL of solvent distilled out. IPA (1 L) was added
and 1 L of solvent distilled out of the vessel. A second portion
of IPA (500 mL) was added, and then the solution was
concentrated to about 300 mL by distillation. The solution was
then cooled to 0-5 °C and aged for 1 h. The solid was filtered
off, washed with cold IPA (200 mL), and dried under vacuum
at 40 °C to afford 20 as a white solid (142.5 g, 76% yield). 1H
NMR (CDCl3, 500 MHz) δ 0.92 (s, 6 H), 1.20-1.25 (m, 2 H),
1.54-1.61 (m, 2 H), 1.88-1.96 (m, 2 H), 2.01 (br s, 2 H),
2.25-2.33 (br s, 2 H), 2.38 (t, J ) 7 Hz, 2 H), 3.99 (t, J ) 6
Hz, 2 H), 6.78 (d, J ) 9 Hz, 2 H), 7.35 (d, J ) 9 Hz, 2 H);
MS: MH+ ) 327.
1
w/w by H NMR). The solution was basified with 16% w/w
sodium hydroxide solution (87 L, pH > 12) and extracted into
TBME (194 L). The layers were separated, and the TBME layer
was washed with 40% sodium hydroxide (25 L) and concen-
trated at atmospheric pressure. The product 3 was isolated by
vacuum distillation (boiling point 70-80 °C at -0.85 to -0.90
bar) to yield a colourless liquid (14.15 kg, 71% yield).1H NMR
(CDCl3, 400 MHz): δ 0.91 (s, 6 H), 1.32-1.35 (m, 2 H), 1.39
(br s, 1 H), 1.46-1.52 (m, 2 H), 2.47 (s, 2 H), 2.69-2.77 (m,
2 H).
Preparation of 4,4-Dimethyl-5-oxopentanenitrile (18). To
a solution of tetrabutylammonium hydroxide (38.9 g of a 40
wt % solution in methanol, 0.032 equiv, 0.06 mol) in TBME
(150 mL) was added a solution of isobutyraldehyde (150 g,
2.08 mol, 1.1equiv) and acrylonitrile (102 g, 1.89 mol, 1 equiv)
in TBME (150 mL) over 2 h at 50 °C. After completion of
reaction, acetic acid (4.7 mL, 81.7 mmol, 4.7 mol %) was added,
and the solvents were removed at 150 mbar/45 °C. The desired
product was then isolated and purified by distillation (20-10
mbar, 100-105 °C) to afford 106.85 g (45%) of a colourless
oil. 1H NMR (CDCl3, 400 MHz): δ 1.10 (s, 6 H), 1.84-1.88
(m, 2 H), 2.24-2.33 (m, 2 H), 9.41 (s, 1 H).
Preparation of 3,3-Dimethylpiperidine (3) by Alternative
Route. 18. (63.1 g, 0.505 mol) was charged to a 1 L
hydrogenator. Sponge Ni (21.45 g, Johnson Matthey type
A-5000) and absolute ethanol (630 mL) were added. Finally,
acetic acid (21 mL) was charged. The reaction was hydroge-
nated and stirred for 18 h at 1000 rpm at room temperature.
The catalyst was filtered through Celite and the filter bed was
washed with ethanol (250 mL). More acetic acid (1 equiv, 0.505
mol, 30.4 g) was added and the solution was concentrated at
70 °C/300 mbar until the distillation ceased. Water (63 mL, 1
vol) was added and the mixture was evaporated at 70 °C/200
mbar. The residue was diluted with TBME (250 mL). Water
(125 mL and 32% w/w aqueous NaOH (125 mL) were charged.
The layers were separated and the organics were washed with
40% w/w aqueous NaOH (125 mL). TBME was then evapo-
rated at 250 mbar/40 °C and the desired product 3 was purified
and isolated by distillation (62 °C/100 mbar) to yield 21.02 g
(37%) as a colourless oil.
Preparation of 3,3-Dimethyl-1-(3-{[4-(4-piperidinyl)-
phenyl]oxy}propyl)piperidine Benzoate (25a). Bromide 20
(34 kg, 104 mol) was dissolved in 0.5 M hydrochloric acid
(238 L, 119 mol) and washed with TBME (340 L). The aqueous
solution was basified with aqueous sodium hydroxide (10 M,
20 L) and extracted with TBME (306 L). The TBME solution
was washed with water (102 L) and then dried by Dean-Stark
azeotropic distillation to give a solution of 20 in TBME. THF
(272 L) at -20 °C was treated with 2.5 M n-butyl lithium (84.6
L, 212 mol); the TBME solution of 20 was then added over
3-6 h followed by N-benzylpiperidin-4-one (19.65 kg, 112
mol). The reaction mixture was warmed to 0 °C before being
quenched with 18.5% w/w ammonium chloride solution (170
kg). The reaction was warmed to 22 °C, the aqueous phase
separated; and the organic phase washed with water (170 L).
The organic solution was concentrated by atmospheric distil-
lation to 102 L, diluted with ethanol (306 L), and further
concentrated to 187 L. Water (9.5 L), 10% Pd-C/50% wet
(6.8 kg), and methanesulphonic acid (20.4 L, 314 mol) were
added, and the reaction was hydrogenated under 3 bar H2
pressure at 65 °C for 8 h. The reaction mixture was filtered to
remove the catalyst before being concentrated to 102 L by
atmospheric pressure distillation. Water (170 L) and TBME
(340 L) followed by 10 M aqueous sodium hydroxide (34 L)
were added, and the biphasic solution was filtered before the
aqueous layer was separated. The organic layer was further
washed with water (170 L) before being dried by Dean-Stark
azeotropic distillation to give a 340 L solution in TBME.
Benzoic acid (25.5 kg, 209 mol) was added and the mixture
seeded and cooled to room temperature. The slurry was aged
for 1 h before the product was collected by filtration, washing
with TBME (2 × 68 L). Drying under vacuum afforded 25a
as a white solid (35.4 kg, 65%). 1H NMR (MeOD, 400 MHz)
δ 0.98 (s, 6 H), 1.28-1.34 (m, 2 H), 1.65-1.75 (m, 2 H),
1.78-1.89 (m, 2 H), 1.97-2.04 (m, 2 H), 2.33 (br s, 2 H),
2.53-2.60 (br s, 2 H), 2.63-2.68 (t, J ) 8 Hz, 2 H), 2.76-2.84
(m, 1 H), 3.08 (td, 2 H), 3.45 (d, J ) 13 Hz, 2 H), 4.01 (t, J )
6 Hz, 2 H), 6.86 (d, J ) 8 Hz, 2 H), 7.15 (d, J ) 8 Hz, 2 H),
7.32-7.42 (m, 3 H), 7.94 (d, J ) 7 Hz, 2 H); MS: MH+ )
331.
Preparation of 1-{3-[(4-Bromophenyl)Oxy]Propyl}-3,3-
dimethylpiperidine (20). 4-Bromophenol (100 g, 0.58 mol)
and tetrabutyl ammonium bromide (18.6 g, 0.058 mol) were
charged to a vessel. MIBK (200 mL), water (127 mL), and 10
M NaOH solution (173 mL, 1.73 mol) were added followed
by bromochloropropane (60 mL, 0.61 mol). The mixture was
heated to 80 °C for 1 h and then cooled to 50 °C; 3,3-
dimethylpiperidine (89 mL, 0.58 mol) was then charged to the
vessel and the mixture heated at 80 °C for 4 h. The mixture
was cooled to about 50 °C, TBME (800 mL) was added, and
the aqueous layer was separated. The organic layer was washed
with water (200 mL). The organic layer was extracted with 0.5
120
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development

Preparation of 4-Bromo-1-naphthalenecarboxylic Acid
(27). To a solution of 1,4-dibromonaphthalene (25 kg, 87 mol)
at 0 °C in THF (175 L) was added a solution of iPrMgCl:LiCl
(1.35 M in THF, 100 L, 135 mol) under a nitrogen atmosphere.
The mixture was stirred at 0 °C for 30 min, warmed to 20 °C,
and stirred at that temperature until complete formation of the
Grignard intermediate (level of dibromonaphthalene <2% by
HPLC). The mixture was cooled to -10 °C and sparged with
CO2 gas for ∼1 h until there was no more consumption of the
Grignard intermediate (typical level of Grignard intermediate
in solution at the end of the reaction is 7%). The reaction
mixture was warmed to 20 °C and added to water (200 L), and
then THF was distilled off at atmospheric pressure (65 °C). At
the end of the distillation, the mixture was cooled to 45 °C;
ethyl acetate (325 L) was added followed by aqueous HCl (5
M, 75 L). The biphasic mixture was stirred at 45 °C for 20
min and allowed to settle. The aqueous phase was removed.
Water (175 L) was added, the biphasic mixture was stirred at
45 °C for 10 min, and the aqueous phase was removed. Heptane
(175 L) was added and the mixture concentrated down to 175
L at atmospheric pressure. The mixture was then cooled to 55
°C, and heptane (75 L) added. The mixture was then seeded
with an authentic sample of 27 (25 g) if required, cooled to 50
°C, and stirred for 1 h. It was then cooled to 20 °C over 2 h
and stirred at 20 °C for 1 h. It was finally cooled to 5 °C over
30 min and stirred at that temperature for 1 h, and then the
solid was filtered off and the cake washed with heptane (50 L)
and dried at 50 °C under vacuum to give a white solid (18.15
kg, 83%). 1H NMR (DMSO-d6, 400 MHz) δ 7.73-7.80 (m, 2
H), 7.98-8.05 (m, 2 H), 8.25-8.32 (m, 1 H), 8.89-8.97 (m,
1 H), 13.40 (s, 1 H); MS: MH+ ) 251-253.
Preparation of 4-{(1E)-3-Oxo-3-[(phenylmethyl)oxy]-1-
propen-1-yl}-1-naphthalenecarboxylic Acid (7). To a solution
of 27 (29 kg, 116 mol) at 20 °C in toluene (174 L) and
triethylamine (80 L, 574 mol) was added palladium acetate (0.78
kg, 3.5 mol), triphenylphosphine (1.5 kg, 5.7 mol), and benzyl
acrylate (26 L) under a nitrogen atmosphere. The mixture was
heated to 90 ( 3 °C and stirred for at least 11 h. The reaction
mixture was cooled to 70 °C and then sampled and analysed
by HPLC. The reaction was deemed complete when 25 was
<2%. After completion of the reaction the solution was cooled
to 20 ( 5 °C. The reaction mixture was filtered through a 5
µm cloth. The precipitate was washed with toluene (58 L), and
the washes were combined with the previous filtrate. HCl (1
M, 57 L) was added to the toluene solution, and the mixture
was stirred for at least 5 min and then allowed to settle. The
organic phase was removed. 2-Methyltetrahydrofuran (435 L)
was added to the aqueous phase followed by 1 M HCl (174 L)
over at least 5 min. The mixture was stirred for at least 5 min
and then allowed to settle. The pH of the aqueous phase was
controlled to <3, and the aqueous phase was removed. The
contents were concentrated to ∼145 L by atmospheric distil-
lation. The solution was cooled to 20 ( 3 °C over at least 3 h,
and n-heptane was (145 L) added over at least 2 h, maintaining
the contents at 20 ( 3 °C. The slurry was aged at 20 ( 3 °C
for at least 5 h before being filtered. The solid was then washed
with n-heptane (58 L) and dried in a vacuum oven at 40 °C to
constant probe temperature to give a white solid (32.1 kg, 84%).
1H NMR (DMSO-d6, 400 MHz) δ 5.30 (s, 2 H), 6.82 (d, J )
16 Hz, 1 H), 7.36 (t, J ) 7 Hz, 1 H), 7.42 (t, J ) 7 Hz, 2 H),
7.46-7.50 (m, 2 H), 7.67-7.74 (m, 2 H), 8.02 (d, J ) 8 Hz,
1 H), 8.11 (d, J ) 8 Hz, 1 H), 8.25-8.33 (m, 1 H), 8.51 (d, J
) 16 Hz, 1 H), 8.85-8.92 (m, 1 H), 13.34 (s, 1 H); MS: MH+
) 333.
Preparation of Phenylmethyl (2E)-3-(4-{[4-(4-{[3-(3,3-
dimethyl-1-piperidinyl)propyl]oxy}phenyl)-1-piperidinyl]-
carbonyl}-1-naphthalenyl)-2-propenoate fumarate (33). 7 (20
kg, 60 mol) was slurried in isopropyl acetate (200 L) and heated
to 79 ( 3 °C. Thionyl chloride (5.26 L, 72 mol) was added;
once a solution was obtained, the reaction was heated to reflux;
heating continued until there was less than 5% a/a 7 by HPLC
analysis (typically <1.0 h) and the solution cooled to 55 ( 3
°C. 25a (32 kg, 51 mol) was slurried in isopropyl acetate (160
L) and aqueous potassium carbonate (20% w/w, 160 L) added.
The biphasic solution was heated to 55 ( 3 °C. The biphasic
mixture was stirred for 10 min, and the layers were separated,
and the aqueous phase was discarded. Aqueous potassium
carbonate (20% w/w, 160 L) was added to the isopropyl acetate
layer, the biphasic mixture was stirred for 10 min at 55 ( 3
°C, and the layers were separated, the aqueous phase discarded.
Aqueous potassium carbonate (20% w/w, 160 L) was added to
the solution and the biphasic mixture heated to 55 ( 3 °C. The
acid chloride solution was added, maintaining the internal
temperature at 55 ( 5 °C; the biphasic mixture was heated until
there was less than 0.25% a/a 25 by HPLC analysis (typically
<0.25 h). The layers were separated, and the isopropyl acetate
layer was washed with water (160 L) at 55 ( 3 °C. The
isopropyl acetate layer was distilled to 320 L, isopropyl alcohol
(320 L) was added and the solution distilled to 320 L. Further
isopropyl alcohol (320 L) was added, and the solution was
distilled to 480 L and cooled to 70 ( 3 °C. Fumaric acid (7
kg, 60 mol) was added, and the mixture was stirred at 70 ( 3
°C to obtain a solution and cooled to 50 °C over 1 h. The
solution was aged for 4 h to form a slurry; the slurry was cooled
to 25 °C over 2 h and aged for 15 min. The slurry was filtered,
and the cake was washed with isopropyl acetate/isopropyl
alcohol (3× 40 L) and dried in a vacuum oven at 65 °C to
yield 33 as a white solid (36.6 kg, 81%). 1H NMR (500 MHz,
DMSO-d6) δ 0.91 (s, 6 H), 1.17-1.25 (m, 2 H), 1.51-1.60
(m, 2 H), 1.63-1.79 (m, 2 H), 1.84-1.92 (m, 4 H), 2.15 (br s,
2 H), 2.35-2.43 (m, 2 H), 2.70-2.79 (m, 1 H), 2.91-3.17 (m,
3 H), 3.26-3.30 (m, 1 H), 3.95-3.99 (m, 2 H), 4.79-4.85 (m,
2 H), 5.30 (s, 2 H), 6.61 (s, 2 H), 6.79-6.86 (m, 3 H),
7.13-7.20 (dd, J ) 21 Hz, 9 Hz, 2 H), 7.35-7.97 (m, 9 H),
8.02-8.04 (m, 1 H), 8.29 (d, J ) 5 Hz, 1 H), 8.51 (d, J ) 15
Hz, 1 H); MS: MH+ ) 645.
Preparation of 3-(4-{[4-(4-{[3-(3,3-Dimethyl-1-piperidinyl)-
propyl]oxy}phenyl)-1-piperidinyl]carbonyl}-1-naphthalenyl)-
propanoic Acid Hydrochloride (10). 33 (23.5 kg) was slurried
in butanone (235 L) and aqueous potassium carbonate (20%
w/w, 118 L) added. The biphasic solution was stirred at 25 (
3 °C for 15 min, the layers were separated, and the aqueous
phase was discarded. Aqueous potassium carbonate (20% w/w,
118 L) was added to the butanone layer, the biphasic mixture
was stirred for 15 min at 25 ( 3 °C, and the layers were
separated, the aqueous phase discarded. The butanone layer was
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
121

distilled to 118 L, further butanone (118 L) was added and the
solution distilled to 118 L. After cooling to room temperature
the solution was filtered before acetone (47 L) and 5% palladium
on carbon (50% aqueous paste, type 4530, 2.35 kg) were added
to the filtrate. The vessel was purged with nitrogen. An
atmosphere of 3 bar hydrogen was placed on the vessel and
the mixture stirred vigorously at 40 ( 3 °C until there was less
than 0.25% a/a 33 by HPLC analysis (typically 8 h). The
mixture was filtered through an R55S carbon CUNO and the
cake washed with butanone (70 L). The filtrate was distilled to
118 L, butanone (118 L) was added, and the solution was
distilled to 118 L. Further butanone (118 L) was added and the
solution distilled to 118 L and cooled to 70 ( 3 °C. Butanone
(47 L) and DMSO (17.6 L) were added; the solution was stirred
for 5 min, and concentrated hydrochloric acid (0.115 vol, 1.05
equiv) was added, maintaining the temperature at 70 ( 3 °C.
The mixture was stirred for 10 min and clarified, the filtrate
was heated to 70 ( 3 °C and seeded with 10 (0.12 kg slurried
in butanone 0.1 vol) and the slurry aged for 15 min. Butanone
(212 L) was added over 2 h, maintaining the temperature at 70
( 3 °C, cooled to 20 ( 3 °C over 3 h and aged for 4 h. The
slurry was filtered and the cake washed with butanone (2 × 94
L) and dried in a vacuum oven at 55 °C to yield 10 as a white
solid (14.7 kg, 80%). 1H NMR (700 MHz, DMSO-d6) δ 0.95
(s, 3 H), 1.14 (s, 3 H), 1.25 (ddd, 0.5 H), 1.32 (t, 1 H), 1.44 (d,
1 H), 1.49-1.61 (m, 1.5 H), 1.64 (ddd, 0.5 H), 1.68-1.77 (m,
1.5 H), 1.81-1.95 (m, 2 H), 2.13-2.18 (m, 2 H), 2.63-2.79
(m, 5 H), 2.91 (t, 0.5 H), 2.96 (t, 0.5 H), 3.02 (t, 0.5 H),
3.08-3.17 (m, 2.5 H), 3.19 (d, 1 H), 3.26 -3.41 (m), 3.97-4.04
(m, 2 H), 4.80 (d, 0.5 H), 4.84 (d, 0.5 H), 6.85-6.90 (m, 2 H),
7.17 (d, 1H), 7.21 (d, 1H), 7.33 (d, 0.5 H), 7.43 (m, 1 H), 7.48
(d, 0.5 H), 7.57-7.66 (m, 2 H), 7.74 (d, 0.5 H), 7.90-7.94 (m,
0.5 H), 8.12-8.17 (m, 1 H), 9.76 (br s, 1 H), 12.25 (br s, 1 H);
MS: MH+ ) 557.
Acknowledgment
We thank our colleagues at Stevenage and Tonbridge for
their contributions to this work: Andrea Childs, Ute Gerhard,
Ben Rigby, Marco Smith, Alex Stuart (analytical support),
Lindsey Wynne (material sourcing), Steve Gooding, Erica Vit
(reaction calorimetry), Alan Collier, Dave John, Elsa Vilminot
(process engineering), David Box, Frank Darabi, Charles Daka,
Charles Dexter, Marina Duffy (pilot plant), Julien Douillet, Mei
Lee (particle sciences), and Andy Searle (synthetic chemistry).
Received for review September 27, 2010.
OP1002598
122
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development