Application of a process friendly morpholine synthesis to (S,S)-Reboxetine

Article abstract of DOI:10.1016/j.tetlet.2010.07.106

We report our results on the construction of a morpholine ring system from the corresponding epoxide and amino alcohol. From this study, we were able to convert a previous four-step synthesis into a more efficient two-step process.

Full text of DOI:10.1016/j.tetlet.2010.07.106

Tetrahedron Letters 51 (2010) 5048–5051
Contents lists available at ScienceDirect
Tetrahedron Letters
journal homepage: www.elsevier.com/locate/tetlet
Application of a process friendly morpholine synthesis to (S,S)-Reboxetine
*
Georges Assaf , Gemma Cansell, Doug Critcher, Stuart Field, Stewart Hayes, Suju Mathew, Alan Pettman
Department of Chemical Research and Development, Pfizer Ltd, Ramsgate Road, Sandwich, Kent CT13 9NJ, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
We report our results on the construction of a morpholine ring system from the corresponding epoxide
and amino alcohol. From this study, we were able to convert a previous four-step synthesis into a more
efficient two-step process.
Received 2 June 2010
Revised 1 July 2010
Accepted 16 July 2010
Available online 22 July 2010
Ó 2010 Elsevier Ltd. All rights reserved.
The morpholine ring structure is an important heterocycle
present in many compounds of biological and pharmaceutical rel-
evance.1 (S,S)-Reboxetine free base 1 is an example of a morpho-
line-containing compound that displays strong activity against
many diseases including neuropathic pain.2 The initial manufac-
turing process involved a four-step synthesis from epoxide 2 and
was successfully applied to synthesize Reboxetine in its racemic3
and enantiopure4 forms. This strategy has also been used in the
synthesis of various analogues.5 The epoxide 2 was converted into
the amino alcohol 3 with ammonia followed by reaction with chlo-
roacetyl chloride to form the amide 4. This compound was con-
verted into the lactam 5 by reaction with a strong base and was
subsequently reduced to the morpholine 1 in the presence of a
suitable reducing agent (Scheme 1).
There is scope to optimize the published procedures in terms of
yield, stoichiometries, and concentrations. However, these ap-
proaches do not address the issues of some of the reagents and
by-products, or the throughput and overall step count. Our work
therefore focused on the alternative strategies to build the target
morpholine ring guided by the SELECT criteria.6
The first strategies examined utilized the amino alcohol inter-
mediate 3 as the starting point. We targeted removing 1–2 steps
by replacing chloroacetyl chloride by
a suitable two-carbon
synthon at the correct oxidation level, thus avoiding the lactam
reduction stage in the original process. A further chemical step
could be removed if the two bonds (N–C and O–C) could be formed
in one pot. A large set of conditions, including over 200 reactions,
was designed and run in our screening facilities with 10 different
reagents (symmetrical and unsymmetrical two-carbon units, see
Table 1) and using the most common combinations of solvents
and bases for alkylation reactions.
This well-established procedure has some drawbacks for a com-
mercial scale manufacturing process:
(1) The length of the synthesis (four steps) requires time in
expensive facilities increasing manufacturing costs and
waste produced.
(2) The chlorinated intermediates have a potential toxicity that
may also impact workers’ health and safety.
(3) The significant aluminium waste produced in the reduction
step.
The majority of the reactions did not give the desired product or
were lacking in selectivity giving multiple alkylation products.
Only the two cyclic substrates gave interpretable reaction profiles.
Ethylene carbonate reacted at the carbonyl centre to yield the
carbamate 6 that rapidly and quantitatively converted into the
oxazolidinone 7 in the presence of a base. The cyclic sulfate, on
the other hand, reacted at the desired carbon centre to yield a mix-
ture of mono-alkylated 8 and bis-alkylated 9 products (Scheme 2).
(4) The overall yield is moderate to low (30–40%).
EtO
O
EtO
O
EtO
O
EtO
EtO
O
ii)
i)
O
iii)
iv)
Ph
NH2
Ph
NH
Ph
NH
Ph
NH
Ph
O
OH
OH
O
O
O
O
2
3
4
Cl
5
1
Scheme 1. Reagents: (i) NH4OH; (ii) THF, chloroacetyl chloride, H2O, NaOH; (iii) KOtBu; (iv) RedAl or LAH.
* Corresponding author. Tel.: +44 1304644062.
E-mail addresses: georges.assaf@pfizer.com, georges.assaf@m4x.org (G. Assaf).
0040-4039/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2010.07.106

G. Assaf et al. / Tetrahedron Letters 51 (2010) 5048–5051
5049
Table 1
List of two-carbon units screened
OH
Br
Cl
OH
OH
O
O
S
I
Br
Cl
Br
Br
O
Cl
OTs
OMs
Cl
O
O
O
O
HO
Cl
EtO
EtO
O
O
O
O
O
base
3
Ph
NH
NH
O
Ph
O
OH
OH
6
O
O
7
EtO
O
EtO
EtO
O
O
O
O
S
O
O
NaOH
OSO3H
+
3
NH
1
+
Ph
N
Ph
N
Ph
OSO3H
OH
O
OH
9
OSO3H
OSO3H
8
10
Scheme 2. Reactions of amino alcohol 3 with cyclic substrates.
The resulting mixture of mono-alkylated 8 and bis-alkylated 9
compounds was treated with a base and the formation of the
expected morpholines 1 and 10 was observed. After optimization,
the selectivity7 was increased from 57% to 82% by running the
chemistry as a flow process: the separate reagent streams were
combined in a plug flow reactor before quenching. As the conver-
sion could not be increased above 70% without impacting on the
selectivity we looked at alternative strategies to achieve the
desired transformations.
diate 11 and would avoid bis-alkylated by-products. Using hetero-
geneous hydrogenation conditions we found that only platinum
could be used to facilitate this transformation as other metals in-
duced partial or complete loss of the 2-ethoxyphenol fragment.
The selectivity was also poor with the major component identified
as the fully reduced compound 12. The amount of desired product
11 could be increased to 70%, this however required the use of
toxic sodium cyanoborohydride as the reductant. On treating the
crude mixture with aqueous sodium hydroxide, we observed the
formation of the desired morpholine 1 as well as the terminal alco-
hol 13 (Scheme 3).
We believed that a reductive alkylation of the amino alcohol 3
with chloroacetaldehyde would provide the same type of interme-
EtO
O
EtO
O
O
+
H2, Pt/C
3
+
NH
Ph
NH
Ph
MeOH
Cl
OH
OH
Cl
11
12
> 60%
5-40%
EtO
O
EtO
O
NaBH3CN
O
+
NaOH
H2O / CH2Cl2
+
3
1
NH
Ph
CH2Cl2
NH
Ph
Cl
OH
OH
Cl
11
50-70%
Scheme 3. Reductive amination of 3 with chloroacetaldehyde.
OH
13
30%
25%

5050
G. Assaf et al. / Tetrahedron Letters 51 (2010) 5048–5051
EtO
O
EtO
O
EtO
O
OEt
O
EtO
3
HCl
+
1
NH
Ph
NH
NH
Ph
Ph
H2, Pt/C
MeOH
O
O
OH
EtO
OEt
OEt
OH
16
Scheme 4. An attempted route to Reboxetine via acetal intermediates.
14
15
allow for telescoping with the first step. Unfortunately, most of
the reactions did not yield the desired morpholine, but many
impurities instead. Some reaction conditions produced up to 10%
of the desired morpholine but they were not reproducible.
Table 2
Comparison of the bases used: base (5 equiv); 2-AEHS (5 equiv); solvent = methanol;
70 °C
Base
Producta 8 (%)
Dimera 17 (%)
Adducta 18 (%)
The introduction of additional steps could be avoided by revert-
ing to the epoxide intermediate 2 and opening it with a fragment at
the correct oxidation state with the leaving group already in place.
Clearly 2-aminoethanol hydrogen sulfate (2-AEHS) should facili-
tate this as it is precedented in the literature, and we had already
shown that the sulfate intermediate 8 can be converted into the
target molecule.1,10 2-AEHS exists as a zwitterion and needs to
be activated by deprotonation. We screened a range of bases suit-
able for this transformation (Table 2), targeting bases strong en-
ough to activate 2-AEHS (pKa = 9.25 in water) and with low
nucleophilicity to minimize the formation of by-products 18 due
to nucleophilic addition to the epoxide (Scheme 5).
Imidazole
DBU
13
92
40
21
0
85
69
1
0
6
15
8
0
5
87
2
45
71
0
10
2
99
Triethylamine
N-Methyl morpholine
2,6-Lutidine
Tetramethyl guanidine
Ethyl diisopropylamine
DABCO
29
0
a
% Area by HPLC, adducts were confirmed by LC–MS.
The use of toxic chloroacetaldehyde and cyanoborohydride
could be avoided by switching to the less toxic acetal 3. In this ap-
proach, reductive amination was followed by acid-catalyzed ring-
closure and silane reduction.
The bases screened confirmed the need for activation of 2-AEHS
by deprotonation (no reaction with 2,6-lutidine due to its basicity
being too weak) and demonstrated that even relatively non-nucle-
ophilic amines (triethylamine and N-methyl morpholine) opened
the epoxide to give the undesired adduct 18. The most suitable
base identified for this transformation was 1,8-diazobicyclo-
[5.4.0]undec-7-ene (DBU; pKa = 11.5) as this minimized the adduct
18 formation. We also repeated the literature conditions using so-
dium hydroxide,10 but with the expectation of high diol impurity
formation. The diol 13 and ether by-products were formed when
using alcohol/water systems, however, the major impurity was
now dimer 17. To overcome this issue, large excesses of 2-AEHS
and sodium hydroxide were needed (see Table 3). With DBU, high-
er selectivities could be achieved with lower equivalents of 2-AEHS
We were delighted to find that the first step worked well using
Pt/C as the catalyst (75% isolated yield) and treatment of the result-
ing acetal 14 with hydrochloric acid delivered the target interme-
diate as a 9:1 mixture of acetals 15 and hemiacetals 16 (Scheme 4).
The third transformation was well precedented in the literature,8
typically using triethylsilane with trifluoroacetic acid in dichloro-
methane. We used this in the design of a reaction screen combin-
ing a range of silanes and acids (Lewis and Brønsted). We also
screened metal-catalyzed reductions. The most attractive from a
process point of view was the use of a specific platinum-based
hydrogenation catalyst9 for reductive etherification as this may
EtO
EtO
EtO
O
EtO
O
NH+3
O
O
O
S
O
O
O
+
2-AEHS
Base
R1
+
NH
Ph
Ph
N
Ph
N+
R3
2
Ph
R2
OH
8
OH
OH
OH
OSO3H
OSO3H
18
17
Scheme 5. Epoxide opening with 2-aminoethanol hydrogen sulfate in the presence of a tertiary amine as a base (NR1R2R3).
Table 3
Comparison of NaOH and DBU for epoxide opening with different amounts of 2-AEHS and base
Base
2-AEHS (equiv)
Base (equiv)
Epoxidea 1 (%)
Producta 8 (%)
Dimera 17 (%)
Othersa,b (%)
NaOH
NaOH
DBU
5
10
3
5
10
3
3
2
0
0
84
90
92
96
9
5
7
3
4
3
1
1
DBU
5
5
a
% Area by HPLC, conversion measured after 3 h at 65 °C.
Others refer to the sum of all other impurities present. Solvent = water/methanol for NaOH and methanol only for DBU.
b

G. Assaf et al. / Tetrahedron Letters 51 (2010) 5048–5051
5051
EtO
O
no alcohol 3 and the epoxide 2. We have demonstrated that the
synthesis described in Scheme 1 can be replaced effectively by a
shorter and more efficient two-step process via epoxide opening
with 2-AEHS and base-mediated ring-closure using an atypical
solid NaOH/THF/EtOH system. The combined transformations
delivered (S,S)-Reboxetine in more than 60% overall yield through
a more efficient process both in an environmental and in an eco-
nomical respect.
concentrated
base >15 M
dilute base
8
1
NH
water / toluene
water / toluene
OH
OH
13
Acknowledgements
Scheme 6. Behaviour of zwitterion 8 in the presence of dilute and concentrated
base.
We wish to thank all the contributors to this work, especially
Trevor Newbury, John Deering, Laurence Harris, Adam Scott, Alex
Wilder, Catherine Dunne, Pieter de Koning, Mark Ridgill, Julian
Smith and Steve Challenger.
and it was therefore selected as the optimum base. The solvent sys-
tem was also optimized and resulted in a further decrease of the
stoichiometry of base and 2-AEHS. Most of the screened solvents
tried gave at least 70% (in situ) product but only the alcohol-based
solvent mixtures gave a yield greater than 80% while providing a
convenient work-up.
With the target intermediate in hand, the base-mediated ring-
closure was explored, by screening a range of bases and solvents.
KOH and NaOH were identified as optimum bases in toluene and
high concentrations were required to minimize the formation of
the diol impurity 13 (Scheme 6).
With other bases in anhydrous systems (typically KOtBu/THF)
many side products were observed and the desired morpholine 1
was formed in only 60–70% yield. Amongst the side products, 2-
ethoxyphenol and the amino alcohol 3 were observed. We found
that the use of solid sodium hydroxide in THF combined with an
additive in the solvent (water or alcohol at 1–5% levels) increased
dramatically the reaction rate and improved the yield from 65% to
90%.
References and notes
1. Wijtmans, R.; Vink, M. K. S.; Schoemaker, H. E.; van Delft, F. L.; Blauww, R. H.;
Rutjes, F. P. J. T. Synthesis 2004, 641–662.
2. Ellis, A. J.; Junor, R. W. J.; Stoker, M. J.; Whelan, L. J. PCT Int. Appl. WO
2010044016; Chem. Abstr. 2010, 152, 493509.
3. Henegar, K. E.; Ball, C. T.; Horvath, C. M.; Maisto, K. D.; Mancini, S. E. Org. Process
Res. Dev. 2007, 11, 346–353.
4. Henegar, K. E.; Cebula, M. Org. Process Res. Dev. 2007, 11, 354–358.
5. Ronald, M.; Tarazi, F. L.; Baldessarini, R. J.; Jarkas, N.; Goodman, M. M.; Seilby, J.
P. Bioorg. Med. Chem. Lett. 2007, 17, 533–537.
6. Butters, M.; Catterick, D.; Craig, A.; Curzons, A.; Dale, D.; Gilmore, A.; Green, S.
P.; Marziano, I.; Sherlock, J.-P.; White, W. Chem. Rev. 2006, 106, 3002–3027.
7. The selectivity in percent is defined as the ratio of 8/[9+8] Ã 100.
8. Srikrishna, A.; Viswajanani, R. Tetrahedron 1995, 51, 3339–3344; Olah, G. A.;
Yamato, T.; Iyer, P. S.; Prakash, J. K. S. J. Org. Chem. 1986, 51, 2826–2828; Benko,
Z.; Shinkle, S.; Van Heertum, J.; Jackson, J.; McQuiston, J.; Webster, J.; Turner, J.;
Weimer, M.; Paterson, E. Chimia 2003, 57, 720–724; Soderquist, J. A.; Kock, I.;
Estrella, M. E. Org. Process Res. Dev. 2006, 10, 1076–1079; Hansen, M. C.;
Verdaguer, X.; Buchwald, S. L. J. Org. Chem. 1998, 63, 2360–2361.
9. Gooßen, L. J.; Linder, C. Synlett 2006, 3489–3491.
The effect of water and ethanol was compared and we found
that the use of water required very tight control: while 1% water
proved beneficial, 3% or more retarded considerably the desired
reaction rate and favoured the formation of the amino diol 13. In
comparison, ethanol could be tolerated at higher levels (up to
20% volume in THF) without interfering with the reaction rate
and purity profile, typically more than 99% conversion within
3–4 h producing the morpholine 1 in 90–95% yield. Common alco-
hols were investigated and we found that methanol, isopropanol,
tert-butanol and tert-amyl alcohol gave equivalent results.
This process was successfully scaled up to kilogram-scale deliv-
ering (S,S)-Reboxetine succinate in high yield and chemical
purity.11
10. Brown, G. R.; Forster, G.; Foubister, A. J.; Stribling, D. J. Pharm. Pharmacol. 1990,
42, 797–799.
11. Typical reaction conditions: Epoxide opening of 2: 2-Aminoethanol hydrogen
sulfate (2.50 equiv) was slurried in toluene (2 vol) and EtOH (2 vol). DBU
(2.48 equiv) was added and the contents were heated to 65 °C for 1 h. Epoxide
(2, limiting reagent) in toluene (5 vol) was added dropwise over 1 h and the
mixture was stirred for a further 2 h at 70 °C. The reaction mixture was cooled
to room temperature and the reaction was quenched with 1.3 M NaOH
(3.0 equiv). The product crystallized from the aqueous layer by pH adjustment
(with 1.0 M HCl) and was isolated by filtration, washed with H2O (2 vol) and
reslurried in EtOH (6 vol) to give 8 in 74% yield. Ring closure to 1: To a slurry of
8 (limiting reagent) in THF (7 vol) and EtOH (0.21 vol) at room temperature
was added NaOH (3.0 equiv). The reaction mixture was heated to reflux for 3 h
and cooled to room temperature. H2O (5 vol) and cyclohexane (4 vol) were
added and the phases separated. The solvent was distilled and replaced with
EtOH (5 vol) and succinic acid (1 equiv) was added to crystallize the product as
the desired salt (82% yield).
In conclusion, we have explored a number of approaches to syn-
thesize the morpholine ring of (S,S)-Reboxetine from both the ami-