Investigation of an alternative route to ZD3638 and cost-benefit analysis comparison of raw materials with the previous route

Article abstract of DOI:10.1021/op0000935

A convergent synthesis to ZD3638 was proposed starting with alternative raw materials. A cost-benefit analysis for the new route was performed which demonstrated that significant savings in raw materials costs could be made over the previous linear sequence. The convergent route was then proved in principle by experimental work. LiTMP was shown to be a superior base to LDA in the modified lithiation reaction.

Full text of DOI:10.1021/op0000935

Organic Process Research & Development 2001, 5, 491−497
Investigation of an Alternative Route to ZD3638 and Cost-Benefit Analysis
Comparison of Raw Materials with the Previous Route
Jonathan D. Moseley,* William O. Moss, and Matthew J. Welham
AstraZeneca, Process Research and DeVelopment, AVlon Works, SeVern Road, Hallen, Bristol BS10 7ZE, U.K.
Abstract:
considerable efforts.4 Rather than attempt further develop-
ment of this process, we considered that a convergent
synthesis, desirable in its own right, might also solve this
issue. To further justify investigation on a potential new
route at this late stage, we conducted a cost-benefit analysis
of the raw materials costs. This clearly indicated that there
were considerable financial and material savings to be made
from any of our new route options, as discussed in detail
below.
A convergent synthesis to ZD3638 was proposed starting with
alternative raw materials. A cost-benefit analysis for the new
route was performed which demonstrated that significant
savings in raw materials costs could be made over the previous
linear sequence. The convergent route was then proved in
principle by experimental work. LiTMP was shown to be a
superior base to LDA in the modified lithiation reaction.
Chemistry Background. In earlier extensive work on the
key coupling step of 5 and 6, the process had lacked
robustness, varying in yield from 40 to 65%, and could not
be improved beyond 65% in the laboratory environment. The
reaction was less reliable on the plant scale, although after
much process development work, an overall yield of 63%
was reliably achieved for campaign 3 (25 kg). Further
problems concerned the output, operability, and a complex
work-up procedure. The manifestations of this moderate
yielding reaction were several; typically 10-15% unreacted
piperidone 5 could be recovered at the end of the reaction
(the use of higher equivalents of 6 and LDA did not help);
10-20% mass balance of the piperidone 5 input material
could not be accounted for (recovered, in the product, or in
the liquors); and there was some LC-MS evidence to suggest
that high-molecular weight aldol products were being formed.
All of this evidence strongly indicated that the highly basic
lithiation conditions were abstracting the enolisable protons
R to the ketone of 5, and either protecting it against further
reaction (until work-up) or promoting aldol condensations
with more 5 to give high-molecular weight, lipophilic
impurities. To test this hypothesis, the identical reaction
conditions were performed on nonenolisable benzaldehyde.
A strength-corrected overall yield of 86% of authenticated
alcohol5 10 was obtained without optimisation, indicating
that the reaction conditions were capable of delivering high
yields for well-behaved substrates. Attempts to soften the
Introduction
ZD3638 is an atypical antipsychotic agent for the treat-
ment of schizophrenia which was in development at Zeneca
Pharmaceuticals from 1993 to 1997. It was a follow-up to
AstraZeneca’s recently launched antipsychotic quetiapine
(AstraZeneca tradename “Seroquel”) and acts by a different
mechanism of action. It also had a potential further applica-
tion in the treatment of migraine. The project was transferred
within PR&D from Macclesfield to Avlon in January, 1997,
to continue development of processes for campaign 4.
The first three campaigns (0.7, 3.7, and 25 kg, respec-
tively) had relied on a linear sequence of 11 chemical steps
and a final recrystallisation, starting from 9-anthraldehyde
(1). For campaign 4 (170 kg demand), it was intended to
out-source methanoaldehyde (2) as the key raw material from
a commercial supplier using a six-step procedure.1 The
subsequent five steps (Scheme 1) would be essentially
unchanged from the previous campaigns. The reductive
alkylation of 2 with piperidone ketal 3 to give ketal 4 (now
not isolated), deprotection to piperidone 5, and the ethane
thiolate displacement to give sulphide 8 were all in good
shape. The two stages most in need of improvement were
the coupling of 2-fluoropyridine (6) with piperidone 5 to give
fluoropyridine 7, and the asymmetric Kagan oxidation2 of
sulphide 8 to give crude 9 (yields as shown in Scheme 1).
During the next 3-4 months, the yield achieved in the
asymmetric oxidation (with an acceptable ee) was signifi-
cantly improved to 55%.3 However, the organolithium-
mediated coupling of 6 to piperidone 5 had resisted
improvement over the three previous campaigns despite
(1) Sunagawa, M.; Sato, H.; Katsube, J.; Yamamoto, H. Chem. Pharm. Bull.
1979, 27, 1806. Zeneca modifications to this route on scale-up to be reported
in this journal in due course.
basicity of the nucleophile, for example by trans-metalation
(2) For reviews see: Zhao, S. H.; Samuel, O.; Kagan, H. B. Tetrahedron 1987,
43, 5135-5144. Kagan, H. B. Asymmetric Oxidation of Sulphides. In
Catalytic Asymmetric Synthesis; Ojima, I., Ed.; VCH: New York, 1993;
pp 203-226.
(4) Robinson, G. E. Zeneca internal 6-monthly reports (three), July, 1993, to
December, 1994.
(5) Gungor, T.; Marsais, F.; Queguiner, G. J. Organomet. Chem. 1981, 215,
(3) Moss, W. O. Zeneca internal 6-monthly report, January-July, 1997.
139.
10.1021/op0000935 CCC: $20.00 © 2001 American Chemical Society and The Royal Society of Chemistry
Published on Web 07/17/2001
Vol. 5, No. 5, 2001 / Organic Process Research & Development
•
491

Scheme 1
Scheme 2
with ClTi(OiPr)3,6 use of Lewis acids (ZnBr2, MgBr2), or
activation of the carbonyl oxygen with dried CeCl3, were
protected piperidones 11a-c. The urethanes were preferred
7
all unsuccessful, giving only trace amounts (<10%) of
product 7 in each case. Since much other work4 had been
carried out previously and had failed to improve the process
further, we decided a more radical strategy was now justified.
Convergent Route Strategy. A convergent route (Scheme
2) would make more efficient use of the expensive metha-
noaldehyde 2 by introducing it later in the synthesis
compared to the linear sequence (Scheme 1). However, for
this strategy to succeed, a readily available and cheap
alternative to piperidone ketal 3 would be required. As
replacements, we considered the CBz-, BOC-, and benzyl-
over the benzyl-protected compound 11c because the lower
basicity of the urethane nitrogen should reduce the potential
for enolisation. In addition, the benzyl piperidone 11c was
not favoured (although it was cheap) because of expected
problems with the removal of the benzyl group.8 The BOC
piperidone 11b was not readily available at the time on the
scale required, but fortunately the CBz piperidone 11a was
(6) Weidmann, B.; Widler, L.; Olivero, A. G.; Maycock, C. D.; Seebach, D.
HelV. Chim. Acta 1981, 64, 357.
(7) Dimitrov, V.; Kostova, K.; Genov, M. Tetrahedron Lett. 1996, 37, 6787
and references therein.
(8) The Research Department had investigated the use of this intermediate and
had indeed encountered problems with removal of the benzyl group later
in the synthesis.
492
•
Vol. 5, No. 5, 2001 / Organic Process Research & Development

Table 1. Projected raw material costs for old and new routes
old route
route A
route B
route C
cost/kg % cost cost/kg % cost cost/kg % cost cost/kg % cost
raw material
MW mol ratio cost/kga
pure
of route
pure
of route
pure
of route
pure
of route
methanoaldehyde (2)
piperidone ketal (3)
2-fluoropyridine (6)
(-)-diethyl tartrate
220
143
97
1.00
1.50
1.25
2.08
1.00
n/a
430
127
64
114
80
910
267
65
233
n/a
61.7
18.1
4.4
15.8
n/a
592
n/ab
61
233
179
1064
55.6
n/a
5.7
21.9
16.8
100
531
n/a
63
230
186
1011
52.5
n/a
6.3
22.8
18.4
100
294
n/a
61
254
180
789
37.3
n/a
7.7
32.2
22.8
100
206
CBz piperidone (11a) 233
ZD3638 pure (9) 459
n/a
1475
100
a Cost/kg in £’s. b n/a: not applicable.
both cheap and available on the desired scale.9
Table 2. Costs of proposed new routes relative to old route
For a convergent synthesis to be viable, we decided that
further criteria to be met were the following: yields
comparable to those previously achieved for equivalent
chemical transformations, no hazard or environmental issues,
and a seamless interface with the existing route. This would
make best use of the known late-stage chemistry so as not
to delay delivery of campaign 4 bulk drug and to minimise
the impact on quality and changes to the impurity profile.
However, the potential benefits were significant. There was
the strong possibility of making more efficient use of the
expensive raw materials, there was the potential for yield
and process improvements, and there was the possibility that
performing the asymmetric oxidation on an alternative
intermediate might give an improved resolution, either
directly or on crystallisation.
Raw Materials Costs Analysis. Methanoaldehyde 2 was
the most expensive of several raw materials at £430/kg, but
the contribution from the others was also significant (Table
1). In particular, piperidone ketal 3 and 2-fluoropyridine (6)
were also relatively expensive, as were some of the reagents
for the asymmetric oxidation (e.g., (-)-DET), since they
were required at stoichiometric levels. The high cost of the
raw materials was therefore an additional factor in searching
for a different route.
The predicted demand of ZD3638 was between 5 and 10
tonnes per annum at mature peak sales for the antipsychotic
indication. We calculated the raw materials costs alone, based
on the current campaign 4 processes and yields (Scheme 1).
The costs and equivalents of reagents are shown in Table 1,
the cost figures being correct at the time (early 1997). For
convenience, the costs of solvents and commodity chemicals
were assumed to be negligible.10 The three convergent routes
shown in Scheme 2 were also costed under the same criteria,
changing the raw materials and reagents where appropriate.
For comparative purposes, some assumptions had to be made.
The major assumption was that identical yields could be
achieved for the equivalent chemical transformations in the
proposed new routes. This seemed reasonable to us, given
the similarity of most of the intermediates. Furthermore, with
the chemical experience that had already been generated, we
route
cost/kg pure (£) cost of old route (%) saving (%)
old route
route A
route B
route C
1475
1064
1011
789
100.0
72.1
68.5
53.5
0.0
27.9
31.5
46.5
were confident that given time to develop the processes, we
could at least achieve comparable yields.
The figures in Table 1 also show the actual cost and the
percent contribution that each kilogram of raw material
makes per kilogram of pure 9 for each of the four routes.
From these figures, the total relative cost for each route was
derived, and the results proved to be significant, as shown
in Table 2. Any of the three proposed convergent routes
indicated that major financial savings could be made, with
savings of 28, 32 and 47% for Routes A, B, and C,
respectively (always assuming the linear route yields could
be reproduced). Any improvement in yield would gain yet
further savings. This did not take account of possible plant
efficiencies due to using a convergent synthesis, which might
also be substantial. Route C also involved the asymmetric
oxidation of the alternative CBz sulphide 13. This might
result in both an improved ee during the oxidation, and more
efficient improvement by crystallisation to enantiomerically
pure material, leading to a higher overall conversion for this
key step.
Unfortunately we had only a few months and limited
resources before we had to decide on the route if we were
to avoid a delay in the manufacture, and hence in the clinical
programme. This was largely due to the long lead time on
our key raw materials, 2 and 3 (or its substitute, 11a). We
dismissed Route B because although the displacement on
12 with ethane thiolate was expected to be facile, the assumed
high yield (89%) of this additional step over Route A meant
that the financial saving was marginal (only a further 4%).
In addition, we had in mind to investigate hydrogenation
methods on the resulting CBz sulphide 13 in the presence
of methanoaldehyde 2 to achieve an efficient reductive
alkylation, which we anticipated might prove troublesome
in the presence of the thioether functionality. We postponed
work on the most advantageous Route C, because our
experiences of the asymmetric oxidation on this and other
projects led us to believe that the optimal conditions for high
yield and ee in the conversion of 13 to 14 might not be
(9) CBz piperidone had been required for a previous compound (ZD7944) and
was potentially available from Ubichem on the scale required.
(10) The figures were put through Zeneca’s in-house “Procost” system, an Excel-
based software package, to generate the predicted raw material production
costs.
Vol. 5, No. 5, 2001 / Organic Process Research & Development
•
493

Scheme 3
quickly discovered. Therefore, we concentrated on Route A.
If we could prove the principle of the convergent synthesis
by Route A, this would buy extra time and resources which
could then be used to investigate Route C.
particularly to remove the chromatographic purification and
also to recycle the TMP, but the project was canceled before
much progress had been made in this area.
Results and Discussion
Ortho lithiation of 2-fluoropyridine (6) was achieved
under literature conditions5,11 by adding 1.30 equiv of 2.0
M LDA in THF/heptane/ethyl benzene to 1.25 equiv of 6 at
-78 °C for 1 h. CBz piperidone 11a was added dropwise to
the 2-fluoro-3-lithiopyridine anion, followed by quenching
at -20 °C with a simplified work-up which led to 32-40%
yields of CBz fluoropyridine 12 after chromatography. Some
CBz piperidone was recovered in up to 15%, giving a
corrected yield of 47% with recycled starting material. This
compared favourably with the original route under identical
conditions (i.e., 5 and 6 gave 7 in 65% without chromatog-
raphy) and became the baseline experiment against which
other preparations were then compared.
Alternative Bases. LDA.mono-THF (1.5 M in cyclohex-
ane) gave a yield similar to that of the baseline experiment
(36% after chromatography), indicating that the solvent
mixture was not critical. An experiment with LDA in hexane
was not attempted due to the known low solubility of the
lithiopyridine anion at -78 °C even in THF.11 Attempts to
use bases weaker than LDA (pKa 36)12 to reduce unwanted
enolisation were unsuccessful. For example, NaHMDS (pKa
30) gave no reaction even on warming to room temperature
(the Li and K salts were not investigated due to their similar
basicity). The less basic magnesium diisopropylamide13 also
failed to react at room temperature. Stronger bases such as
n-BuLi and s-BuLi (pKa ∼40) both gave only trace quantities
of the desired CBz fluoropyridine 12 as determined by HPLC
(<10%). It is well reported that these strong alkyllithiums
readily add to the pyridine ring even at low temperatures,14
and although such products were not identified here, they
had been seen in previous work.4 Addition of stoichiometric
TMEDA to an n-BuLi preparation to modify its reactivity
also gave no improvement.
Alternative Raw Materials. The standard reaction condi-
tions were also applied to the considerably cheaper 2-chlo-
ropyridine (16), although the literature indicated that the
deprotonation with LDA is less efficient than for 2-fluoro-
pyridine (6) which might necessitate the use of higher
equivalents of LDA/16.15 Furthermore the intermediate anion
is not as stable and is more prone to pyridyne formation.11,16
There was also concern over the ease of displacement of
the chloride group by sodium ethane thiolate in the subse-
quent stage, and whether the chloride would survive the
proposed hydrogenation conditions, for which there is good
precedent.17 The coupling reaction15,17 was not a success, and
no further work was attempted, although other substituted
pyridine substrates were briefly considered.18
Deprotection and Reductive Alkylation Stages. Re-
moval of the CBz group by hydrogenolysis was facile at
room temperature and occurred in under 3 h in near
quantitative yield to give the crude amino alcohol 17 (Scheme
3). The HPLC purity of the product was 95%, but this could
not be improved by normal phase chromatography due to
the amino alcohol’s high polarity. The structure was con-
firmed by NMR and MS and by subsequent synthesis to
authentic fluoropyridine 7 by two methods. The crude amino
alcohol 17 was subjected to the reductive alkylation condi-
tions of the existing ketal 4 process using BH3‚pyridine in
methanol with methanoaldehyde 2 and acetic acid, to give
the product 7 after chromotography in 73% overall yield from
CBz fluoropyridine 12. The structure of the product was
confirmed by comparison with authentic material generated
from the original route by HPLC tR’s and by NMR and MS
data. Since both the hydrogenation and the borane reduction
It was reasoned that a more hindered base might result
in greater nucleophilic addition by the anion on the carbonyl
rather than directly abstract the enolisable R protons itself.
Addition of n-BuLi to 2,2,6,6-tetramethylpiperidine (TMP,
15) generated the lithium tetramethylpiperidide (LiTMP)
anion, slightly more basic than LDA (pKa 37)12 but consider-
ably more hindered. When the standard reaction conditions
were employed, a reproducible yield of 65% of CBz
fluoropyridine 12 was readily achieved after chromatography.
Some modest process development work was then conducted
on these conditions to improve the yield and work-up,
(11) Gribble, G. W.; Saulnier, M. G. Tetrahedron Lett. 1980, 21, 4137.
(12) Encyclopaedia of Reagents for Organic Synthesis; Paquette, L. A., Ed.; John
Wiley and Sons: Chichester, 1995.
(13) Bradlee, M. J.; Helquist, P. Org. Synth. 1997, 74, 137.
(14) Marsais, F.; Granger, P.; Queguiner, G. J. Org. Chem. 1981, 46, 4494.
(15) Trecourt, F.; Marsais, F.; Gungor, T.; Queguiner, G. J. Chem. Soc., Perkin
Trans. 1 1990, 2409.
(16) Gribble, G. W.; Saulnier, M. G. Heterocycles 1993, 35, 151.
(17) Comins, D. L.; Myoung, Y. C. J. Org. Chem. 1990, 55, 292.
(18) For example, 3-alkoxypyridines have been successfully ortho-lithiated, but
the results were poor for a limited study on the comparable 2-substituted
series: Marsais. F.; Queguiner, G. Tetrahedron 1983, 39, 2009.
494
•
Vol. 5, No. 5, 2001 / Organic Process Research & Development

were conducted in methanol, it should have been possible
to telescope this sequence into one operation. The yield of
73% was very encouraging for a first small-scale reaction
and was not far from the 85% yield for the equivalent
transformation in the original route. Furthermore, the highly
crystalline nature of 7 should have obviated the need for
chromatography after only moderate process development
effort.
None of the proposed new reaction stages had a high
environmental impact factor or low output, and hazard issues
were unlikely to be a major concern, largely because much
of the chemistry was conserved and judged to be already
acceptable (e.g., the BH3‚pyridine complex had already been
used on the pilot plant at Macclesfield for campaign 3
delivery).
Effecting the deprotection, iminium ion formation, and
subsequent reduction all under reductive alkylation conditions
in a hydrogenation vessel was an obvious alternative ap-
proach for a telescoped reaction sequence. Two reactions
were tried under slightly differing conditions. CBz fluoro-
pyridine 12, methanoaldehyde 2, catalyst, and a few drops
of HCl were dissolved in methanol and hydrogenated under
3 bar hydrogen at room temperature for 22 h. After filtration
and flash chromatography authentic fluoropyridine 7 was
again produced, albeit in modest 43% yield.19
In the second case, the methanoaldehyde 2 was not added
until the CBz group had been quantitatively cleaved from
the precursor, as monitored by HPLC after 3 h. Methanoal-
dehyde 2 and acetic acid were then added to the crude amino
alcohol 17 and stirred under nitrogen for 2 h to allow
formation of the iminium ion, before the hydrogenation was
restarted for a further 22 h. This had the advantage of not
consuming any valuable methanoaldehyde before it could
react with 17 (by reduction to the unreactive methanoalcohol
18). The yield of fluoropyridine 7 after an identical work-
Conclusions
The strategy adopted for rapid investigation of concerted
Route A (Scheme 2) starting from the readily available CBz
piperidone 11a looked very promising. Initial studies on the
coupling reaction using LiTMP as the base equalled the 65%
yields obtained for the original route’s coupling reaction
using LDA. Subsequent transformations encompassing re-
moval of the CBz group and some form of reductive
alkylation showed that nearly comparable yields could be
obtained compared to the original linear route. With more
effort in this area, all of these stages should have been
amenable to further process improvements and elimination
of the chromatography. TMP was commercially available
on a useful bulk scale, and although expensive, the cost might
be expected to drop with usage. It was hoped that it might
be recycled by acid-base extraction during work-up, and
preliminary lab work suggested this would be possible.
In summary, the proposed new Route A has been proven
to work in principle on the basis of exploratory lab work.
The cost-benefit analysis demonstrated a clear financial
saving in raw materials costs, and proved to be a valid
exercise in itself in directing our limited resources at the
time. In the longer term we would have investigated the more
efficient Route C with the additional possibility of an
improved asymmetric oxidation of CBz sulphide 13 to CBz
sulphoxide 14 over that of 8 to 9.
up and chromatography was again 43%, but with 36%
recovered methanoaldehyde 2, giving an overall yield of 67%
allowing for unreacted starting material. Indeed, the recov-
ered methanoaldehyde indicated that more forcing conditions
could be used in the hydrogenation, and there is clearly scope
for process optimisation on this stage. The quality of the
product 7 generated by either of these new routes was very
good, even more so if the borane reduction was used instead
of the hydrogenation method. Once again, chromatography
should easily have been avoided with process improvements
and the new route was expected to be advantageous in the
quality aspect also.
Hazard and Environment. A DSC on a sample of CBz
fluoropyridine 12 showed it to have a melting point from
about 122 °C and to be free from self-heating to 286 °C.20
Although further tests would be needed for greater scale-
up, this indicated that the sample could be safely manipulated
up to 60 °C, the temperature of the hot extractions in the
original process and a reasonable temperature for drying.
Experimental Section
General Procedures. Melting points were determined
using a Griffin melting point apparatus (aluminium heating
block) and are uncorrected. Elemental analysis (C, H, and
N) were determined by the Research Department analytical
service, Mereside. 1H NMR spectra were recorded on a JEOL
GX 270 MHz spectrometer with chemical shifts given in
ppm relative to TMS at δ ) 0. Electron impact (EI) mass
spectra were determined on a Micromass Autospec. The
reaction mixtures and products were analysed by reverse
phase HPLC on a Hewlett-Packard 1050 according to the
following conditions: column, Waters Spherisorb S5ODS-
1, 250 mm × 4.6 mm i.d.; eluent, 1:1 acetonitrile:water with
0.1% v/v triethylamine; flow rate 1.5 mL/min.; wavelength
228 nm; injection volume 20 µL. Typical retention times
(tR’s) were: 2-fluoropyridine (6), 2.6 min; CBz piperidone
(11a), 3.0 min; benzaldehyde, 3.2 min; methanoaldehyde (2),
5.0 min.; toluene, 5.7 min. Other retention times were as
noted in the individual procedures. HPLC purities were
generally % w/w against a standard of known strength as
(19) There was some evidence that ZD3638 fluoropyridine sublimed under
reduced pressure (e.g., on concentration), and this may have led to
mechanical losses that would have been exaggerated on the small scale.
(20) Ramp rate 5 °C /min. Actual mp in the range 131-134 °C as determined
by slow ramp (<1 °C /min) on hot-stage melting point apparatus; see
Experimental Section.
1
determined by H NMR, except where noted a.n. (area %
normalised). Analytical TLC was carried out on com-
mercially prepared plates coated with 0.25 mm of self-
Vol. 5, No. 5, 2001 / Organic Process Research & Development
•
495

indicating Merck Kieselgel 60 F254 and visualised by UV
light at 254 nm. Preparative scale silica gel flash chroma-
tography was carried out by standard procedures using Merck
Kieselgel 60 (230-400 mesh).
15 min, which produced an orange suspension. The mixture
was stirred for 15 min, and then water (35 mL) was added
and stirred for a further 30 min. The reaction mixture was
then subjected to the hot extraction and distillation procedures
described above for 7. The residual toluene was removed
under reduced pressure to yield a dark red gum (14.25 g).
The crude gum was purified by silica gel flash chromatog-
raphy, eluting with 1:1 hexane:ethyl acetate to yield the title
compound 12 as a yellow solid (4.5 g, 32%) and recovered
11a (1.5 g, 15%), giving an overall yield of 37% allowing
for recycling of recovered 11a. Data for 12 as noted below,
except mp 133-134 °C.
Preparation of Fluoropyridine (7) using Standard
Conditions with LDA. 2-Fluoropyridine (6) (1.78 mL, 20.6
mmol) in THF (5.0 mL) was added over 10 min to a pre-
cooled (-78 °C) solution of LDA (2.0 M in THF/heptane/
ethyl benzene (10.7 mL, 21.5 mmol) and THF (12.5 mL)
under an inert atmosphere, maintaining the temperature
below -70 °C. The resulting orange solution was stirred for
1 h at below -70 °C. Piperidone (5) (5.0 g, 16.5 mmol)
dissolved in THF (20 mL), and toluene (2 mL) was added
dropwise to the solution of the 2-fluoro-3-lithio-pyridine
anion over 30 min, maintaining the temperature below -70
°C. The addition funnel was rinsed with additional THF (20
mL) and added over 10 min. The resulting solution was
stirred at below -70 °C for 1 h. A sample was taken after
this time and analysed by HPLC (or before, as appropriate).
The reaction mixture was warmed to -20 °C as evenly as
possible over 1.5 h and added to a flask containing a stirred
mixture of toluene (25 mL), water (17.5 mL), and glacial
acetic acid (2.2 mL) over 10 min. The reaction flask was
rinsed with further THF (20 mL), and this was also added.
The resulting solution was heated to 55 °C for 15 min with
stirring and then allowed to stand for 5 min. The lower
aqueous layer was tapped off, along with some dark brown
solid, and further water was added (17.5 mL). The wash and
extraction procedure was repeated. The THF was removed
by atmospheric distillation at 85 °C, and 64 mL of distillate
was collected. Further toluene (25 mL) was added to the
hot flask over 10 min, and the distillation then continued at
108 °C to remove 98 mL of toluene distillate. The resulting
brown slurry was cooled evenly at 20 °C/h to 15 °C. Further
toluene (4 mL) was added to mobilise the slurry, and the
buff coloured solid was collected by filtration, washed twice
with toluene (19 mL each), and pulled dry on the water pump
for 15 min. The solid was further dried in vacuo at 70 °C
over 3 days to yield the title compound 7 (4.19 g, 63% based
on 5). HPLC purity 98.3%, tR 9.5 min; mp 210-212 °C.
Other data for 7 as noted below.
Preparation of CBz Fluoropyridine (12) using LDA.
2-Fluoropyridine (6) (4.7 mL, 53 mmol) in THF (5 mL) was
added over 15 min to a pre-cooled (-78 °C) solution of LDA
(2.0 M in THF/heptane/ethyl benzene, 27.6 mL, 55.1 mmol)
and THF (50 mL) under an inert atmosphere, maintaining
the temperature below -70 °C. The resulting orange solution
was stirred for 1 h at below -70 °C. CBz piperidone (11a)
(10.0 g, 42.4 mmol) dissolved in THF (50 mL) was added
dropwise to the solution of the 2-fluoro-3-lithio-pyridine
anion over 35 min, maintaining the temperature below -70
°C. The addition funnel was rinsed with additional THF (5
mL) and added. The resulting solution was stirred at below
-70 °C for 1 h. A sample was taken after this time and
analysed by HPLC (or before, as appropriate). The reaction
mixture was warmed to -20 °C evenly over 1 h which
formed a brown slurry and was quenched by the addition of
a solution of toluene (50 mL) and acetic acid (5.6 mL) over
Preparation of CBz Fluoropyridine (12) using LiTMP.
TMP (9.4 mL, 55.2 mmol) was added to nBuLi (2.5 M in
hexanes, 22.9 mL, 57.3 mmol) in THF (50 mL) cooled to
-78 °C under an atmosphere of nitrogen, and stirred for 30
min. A solution of 2-fluoropyridine (6) (4.7 mL, 53.0 mmol)
in THF (10 mL) was added dropwise over 10 min keeping
the temperature below -70 °C, and then stirred for 1.25 h
at below -70 °C. A solution of CBz piperidone 11a (10.0
g, 42.4 mmol) dissolved in THF (50 mL) was added dropwise
over 30 min, maintaining the temperature below -70 °C.
The addition funnel was rinsed with THF (10 mL), and the
reaction mixture was stirred for a further 1 h. The resulting
orange solution was allowed to warm to -20 °C over 1.5 h
before being quenched by the addition of acetic acid (5.6
mL, 97.5 mmol) in toluene (50 mL). Water (35 mL) was
added dropwise over 5 min and the mixture stirred for 30
min before heating to 55 °C. The aqueous layer was separated
and the organic layer given a second water wash (35 mL).
The combined organic layer was concentrated under reduced
pressure to a yellow gum (15.4 g) which was purified by
silica gel flash chromatography, eluting with 1:1 hexane:
ethyl acetate to yield the title compound 12 as a white
crystalline solid (9.1 g, 65%). HPLC purity 98.0%, tR 3.9
min.; TLC Rf 0.25 (1:1 hexane:ethyl acetate, SiO2); mp 131-
1
133 °C; H NMR (270 MHz, CDC13) δ 1.78 (2H, d, J )
13.1 Hz), 2.20 (2H, bs), 2.22 (1H, d, J ) 4.0 Hz), 3.31 (2H,
bs), 4.13 (2H, bs), 5.16 (2H, s), 7.19-7.24 (1H, m), 7.31-
7.38 (5H, m), 7.90-7.97 (1H, m), 8.12-8.15 (1H, m); MS
(EI+) 330 (M+, 16%), 206 (M+ - PhCHO - H2O, 18), 91
+
(PhCH2 , 100). Anal. Calcd for C18H19FN2O3 requires C,
65.4; H, 5.8; N, 8.5. Found: C, 65.5; H, 5.8; N, 8.3%.
Preparation of Amino Alcohol (17). CBz Fluoropyridine
(12) (9.91 g, 30.0 mmol) and the catalyst (10% Pd on C,
0.99 g, 1.0 mol %) were suspended in methanol (180 mL)
and stirred for 3 h under an atmosphere of hydrogen (3 bar).
The reaction mixture was filtered through a pad of Celite,
washed copiously with methanol, and concentrated under
reduced pressure to yield the crude title compound 17 as an
off-white solid (6.07 g, ∼100%). HPLC purity 95% a.n., tR
16.6 min.; TLC Rf < 0.05 (10% CH3OH in CH2Cl2 with
1.0% NH4OH, SiO2); mp 188-191 °C; 1H NMR (270 MHz,
CDCl3) δ 1.76 (2Η, dd, J ) 13.9, 2.0 Hz), 2.23 (2H, dt, J
) 12.9, 4.9 Hz), 2.97-3.04 (2H, m), 3.14 (2H, dt, J ) 12.3,
1.5 Hz), 7.19-7.24 (1H, m), 7.94-8.01 (1H, m), 8.11-8.14
(1H, m); MS (EI+) 196 (M+, 35%), 178 (M - H2O, 100),
82 (21), 57 (62), 42 (78). Anal. Calcd for C10H13FN2O‚
496
•
Vol. 5, No. 5, 2001 / Organic Process Research & Development

0.2H2O requires C, 60.1; H, 6.7; N, 14.0. Found: C, 60.3;
H, 6.8; N, 13.5%.
mmol) suspended in methanol (40 mL) were added, and the
reaction mixture was stirred for 2 h at 20 °C under nitrogen.
The hydrogenation was then continued as above, for 22 h in
total before the addition of a solution of sodium bicarbonate
(1.09 g, 13.0 mmol) in water (20 mL). The reaction mixture
was filtered through a pad of Celite, washed copiously with
methanol, and concentrated under reduced pressure to a white
paste (6.0 g). This was purified by silica gel flash chroma-
tography eluting with 1:1 hexane:ethyl acetate to yield
recovered 2 (1.03 g, 36%, Rf 0.86) and the title compound
7 as a white solid (2.29 g, 43%, Rf 0.25), giving an overall
yield of 67% allowing for recovered 2. Data for 7:- HPLC
purity 98.8%, tR 10.5 min. Other data for 7 as noted above.
Preparation of Benzaldehyde Adduct 10. The method
described above for the preparation of 12 using LDA was
used on a 62.5 mmol scale of 2-fluoropyridine (6) to generate
the 2-fluoro-3-lithiopyridine anion. A solution of benzalde-
hyde (5.1 mL, 50.0 mmol) dissolved in THF (10 mL, 1.96
volumes) was added dropwise over 15 min, maintaining the
temperature below -70 °C, and stirred for a further 1.25 h.
The reaction mixture was quenched as for 12, and after a
simplified aqueous extractive work-up, the organic layer was
dried over MgSO4, filtered, and concentrated under reduced
pressure to yield the title compound 10 as an orange/brown
oil which solidified on standing (9.8 g, 96% crude, 86%
Preparation of Fluoropyridine (7) by Borane Reduc-
tion from 17. Crude amino alcohol (17) (1.64 g, 8.4 mmol)
and methanoaldehyde (2) (1.86 g, 8.4 mmol) were suspended
in methanol (25 mL) under an atmosphere of nitrogen. Acetic
acid (240 µL, 4.2 mmol) was added, and the resulting orange
solution was stirred at 20 °C for 1.5 h. Borane-pyridine
complex (8 M, 470 µL, 3.8 mmol) was added in toluene
(1.9 mL) over 4 min, with water bath cooling, and allowed
to stir at 20 °C for 22 h. A precipitate formed after 10 min,
and the orange colour disappeared. Aqueous HCl (2 N, 20.9
mL, 41.8 mmol) was added over 10 min after which a second
solid formed and then dissipated with stirring over the next
3 h at 20 °C. Aqueous NaOH (2 N, 30 mL, 60 mmol) was
added dropwise over 5 min to achieve a pH of 13-14. The
combined solid and aqueous suspension were extracted with
dichloromethane (4 × 50 mL). The combined organic
extracts were dried over MgSO4 and concentrated under
reduced pressure to yield a crude gum (3.8 g) which was
purified by silica gel flash chromatography eluting with 3:2
to 1:1 hexane:ethyl acetate to yield the title compound 7 as
an off-white solid after drying in vacuo at 60 °C (2.43 g,
73%). This could be further purified by trituration in 1:1
hexane:ethyl acetate to give a white solid, mainly improved
in colour rather than quality (2.08 g, 62%). HPLC tR 9.6
min.; TLC Rf 0.21 (3:2 hexane:ethyl acetate, SiO2); mp 209-
1
strength corrected by H NMR). HPLC purity 89% a.n., tR
1
3.1 min; mp (crude) 68-72 °C (lit.5 74 °C); H NMR (270
1
211 °C; H NMR (270 MHz, CDCl3) δ 1.78 (2H, dd, J )
MHz CDC13) δ 1.8-2.7 (1H, bs), 6.06 (1H, s), 7.18-7.24
(1H, m), 7.29-7.41 (5H, m, Ph), 7.97-8.04 (1H, m), 8.08-
8.11 (1H, m); MS (EI+) 203 (M+, 100%), 185 (M-H2O,
12), 124 (34), 105 (PhCO+, 28), 97 (55).
13.7, 2.4 Hz), 2.17-2.32 (3H, m) 2.62 (2H, d, J ) 1.5 Hz),
2.75 (2H, dt, J ) 11.8, 1.8 Hz), 2.90-2.96 (2H, m), 3.48
(2H, s), 4.27 (1H, s), 6.89-6.99 (4H, m), 7.15-7.28 (5H,
m), 7.88-7.95 (1H, m), 8.08-8.11 (1H, m); MS (EI+) 400
(M+, 50%), 206 (100), 195 (52). Anal. Calcd for C26H25-
FN2O requires C, 78.0; H, 6.3; N, 7.0. Found: C, 77.7; H,
6.4; N, 6.9%.
Preparation of Fluoropyridine (7) by Hydrogenation
Telescope from 12. CBz Fluoropyridine (12) (4.31 g, 13.0
mmol) and the catalyst (10% Pd on C, 0.43 g, 1.0 mol %)
were suspended in methanol (125 mL) and hydrogenated for
2 h under an atmosphere of hydrogen (3 bar). TLC (1:1
hexane:ethyl acetate) and HPLC indicated all starting mate-
rial had been consumed by this time. Methanoaldehyde (2)
(2.90 g, 13.1 mmol) and glacial acetic acid (372 µL, 6.5
Acknowledgment
We thank Anne Butlin and Graham Robinson (PR&D,
Macclesfield) for helpful discussions at the time of this work
and for comments on this manuscript; Paul Gillespie (Process
Hazards Section, Blackley) for the DSC data on 12; and
Phillip Hopes (PR&D, Avlon) for additional experimental
work.
Received for review August 29, 2000.
OP0000935
Vol. 5, No. 5, 2001 / Organic Process Research & Development
•
497