Optimisation and scale-up of microwave assisted cyanation

Article abstract of DOI:10.1016/j.tet.2005.11.095

A microwave enhanced palladium catalysed cyanation procedure was optimised for the final step of a production method for citalopram 2. The method was demonstrated on multigram batch scale for the synthesis of escitalopram (S)-2 and then in a stop-flow continuous process for citalopram.

Full text of DOI:10.1016/j.tet.2005.11.095

Tetrahedron 62 (2006) 4705–4708
Optimisation and scale-up of microwave assisted cyanation
*
Michael R. Pitts, Peter McCormack and John Whittall
StylaCats Ltd, The Heath Business & Technology Park, Runcorn, Cheshire, WA7 4QX, UK
Received 2 September 2005; revised 20 November 2005; accepted 20 November 2005
Available online 24 March 2006
Abstract—A microwave enhanced palladium catalysed cyanation procedure was optimised for the final step of a production method for cit-
alopram 2. The method was demonstrated on multigram batch scale for the synthesis of escitalopram (S)-2 and then in a stop-flow continuous
process for citalopram.
Ó 2006 Elsevier Ltd. All rights reserved.
1
. Introduction
various innovative methods for stabilising the catalyst. Cya-
nide ions poison homogeneous palladium catalysts and thus
the cyanide concentration in solution needs to be kept low.
The levels of cyanide can vary significantly with even small
Microwave-assisted organic synthesis is a rapidly expanding
area of research, aided, in recent years, because of the avail-
ability of commercial systems that offer safe and reproduc-
ible heating. The last decade has seen a growing number
of papers and reviews appearing in the literature on the
8
amounts of other additives. Most solutions to this problem
involve slow addition of soluble cyanide sources such as
9
10
TMSCN or acetone cyanohydrin, or the use of an insolu-
ble cyanide source with an additive to transmetallate the
cyanide. Examples of the latter utilised potassium cyanide
in solvents in which it is virtually insoluble are: in toluene
1
subject. Microwave heating seems to be particularly com-
patible with transition metal catalysed processes normally
associated with long reaction times (>4 h), bringing
these times down to minutes and reducing the levels of by-
1
1
12
with TMEDA, in THF with copper(I) iodide, or in aceto-
nitrile with catalytic tributyltin chloride, generate low levels
of Bu SnCN. Recently potassium hexaferricyanide has
2
products from (thermal) side-reactions.
1
3
3
1
4
The substituted benzonitrile motif is present in a significant
proportion of pharmaceuticals, agrochemicals and natural
products. The nitrile group also offers a useful functionality
for subsequent manipulations to important functional groups
such as acids, ketones, oximes, amines and also various het-
erocycles. The nitrile group can be used as a convenient
starting point for short-lived radiocarbon-labelled functional
been employed as a non-toxic source of cyanide. The slow
release of cyanide from the complex maintains low levels in
solution. This procedure was further refined to obviate the
3
y
1
5
needs for palladium by employing a copper catalyst.
Our interest in cyanation was for the improvement of the last
step in the synthesis of citalopram 2 from the parent bromide
1 (Scheme 1). The cyanation via Rosemund–von Braun
conditions took 24 h, and was low yielding after exhaustive,
time consuming wash cycles used to purify the material.
It was our belief that a transition metal catalysed method
could be employed, and enhanced with microwave-assisted
heating, to provide a highly selective transformation with
a good impurity profile. The work by Hallberg and Alter-
4
groups.
Although simple benzonitriles can be easily prepared by
ammoxidation, the most common, and direct method for
the introduction of the cyano group is via cyanation of the
parent aryl halide. Industrially this tends to be achieved by
the Rosemund–von Braun reaction, requiring stoichiometric
5
6
16
amounts of copper(I) cyanide. The alternative Sandmeyer
man had shown a precedent for using microwave heating to
increase the rate of cyanation of various bromides with the
reaction also uses copper(I) cyanide. The selectivity and
waste considerations involved in these reactions led to the
use of transition metal (usually palladium) catalysed cyana-
tions being examined by various industrial groups. The
relatively high catalyst loading requirements have prompted
1
7
previously utilised palladium–tetrakistriphenylphosphine.
However, Maligres and co-workers at Merck Process Re-
search had screened a range of phosphine ligands and found
7
0
1,1 -diphenylphosphinoferrocene (dppf) to be superior and
y
Keywords: Microwave assisted chemistry; Cyanation; Citalopram; Palla-
dium catalysis.
4 6
During the course of this work we had tested K [Fe(CN) ] (see Table 2,
entry 4) as a cyanide source—it seems the slow breakdown of the complex
is unsuitable for the rapid microwave timescale, giving only trace amounts
of product.
*
Corresponding author at present address: Reaxa Ltd, Hexagon Tower,
Blackley, Manchester, M9 8ZS, UK; e-mail: michael.pitts@reaxa.co.uk
0
040–4020/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tet.2005.11.095

4706
M. R. Pitts et al. / Tetrahedron 62 (2006) 4705–4708
more consistent and had demonstrated the cyanation on kilo-
scale. A room temperature palladium catalysed cyanation
Table 2
1
8
Entry
Cyanide source
Conversion (%)
has recently been reported employing tert-butylphosphine as
the ligand.
1
9
1
2
3
4
5
Acetone cyanohydrin
KCN
CuCN
0
1
60
1
F
F
KFe(CN)
TMSCN
6
O
'M', ligand
O
–
0
'
CN', solvent
microwaves
All reactions with 1 mol % Pd–dppf, 1 equiv cyanide source, 0.2 equiv
ꢀ
TMEDA, DMF, MW at 180 C for 300 s.
Br
CN
NMe2
NMe2
1
2
TMEDA as an additive provided a quantitative yield with
high purity. Changes in the solvent were briefly examined;
acetonitrile and THF both gave lower yields than in DMF.
Addition of TMEDA also allowed the temperature to be
dropped to 140 C with no loss in conversion or purity of
the crude material. At this lower temperature, discolouring
of the reaction mixture was eliminated.
Scheme 1.
All these methods utilised zinc cyanide in DMF. Zinc cya-
nide is very sparingly soluble in DMF, a factor that keeps
the concentration of cyanide ions at the required low level.
ꢀ
2
0
The Merck process provided the cleanest reaction with DMF
and 2–5% water. Other additives used to improve the zinc
cyanide/DMF/palladium system include zinc, zinc with
2
1
2
2
23
Further optimisation showed zinc cyanide could also be used
at 0.6 equiv, consistent with previous work suggesting both
bromine and zinc acetate. These observations formed
the starting point for our investigations.
1
8
cyano groups are transferred. Alternative cyanide sources
weretried, all giving lower yields than zinc cyanide, as shown
in Table 2. Interestingly the addition of only 10 mol % of
acetone cyanohydrin to the zinc cyanide reaction increased
cyanide ions in solution sufficiently to poison of the catalyst
and reduce the yield to 10%.
2
. Results and discussion
Initially we tested Pd (dba) and dppf (1:1) at 5 mol % in
2
DMF in the cyanation of 4-bromoacetophenone. The reac-
tion (1.0 mmol starting material, 2 mL solvent) was heated
3
ꢀ
with microwave irradiation to 120 C and after 5 min it
Altering the ligand had a significant effect on the yield. Tri-
phenylphosphine or tri-2-furylphosphine in place of dppf
gave no reaction. Bidentate ligands dppe and dpppe gave
trace amounts of product whereas DPE-phos gave complete
reaction. This prompted the test of Xantphos, which in con-
trast to dppf, allowed the use of 0.5 mol % Pd (1 mol %
showed complete conversion. It was determined to keep
the reaction time at a maximum of 5 min to assist eventual
transfer to a continuous flow system for scale-up. With no
catalyst present starting material was recovered quantita-
tively. Reduction of the level of palladium to 1 mol % was
not detrimental to the yield. Further reduction to 0.1 mol %
ꢀ
ligand) at 140 C. These conditions are the lowest reported
catalyst levels for microwave-enhanced cyanation to date
(Fig. 1).
ꢀ
gave incomplete conversion at 120 C, but was quantitative
ꢀ
catalysts under similar conditions resulted in little or no
conversion.
25
at 180 C. Repeating the reactions with a range of nickel
2
4
Lastly, examination of reaction time revealed the cyanation
to be complete after 120 s at the target temperature. The re-
action takes around 100 s to achieve the target temperature
(max. power 300 W) and around 1 min to cool to below
Surprisingly, subjecting the bromo-precursor 1 to cyanation
ꢀ
at 180 C with 0.1 or 1 mol % of palladium–dppf gave no
reaction. A range of additives were then investigated with
the results shown in Table 1.
ꢀ
40 C giving an overall time for the process of under 5 min.
Scale-up was achieved in the CEM Voyager SF microwave
2
6
Although water as an additive allowed good conversion,
detectable quantities of the amide, from hydrolysis of the
nitrile, and debrominated material were formed. Since these
impurities needed to be kept below 0.2%, exclusion of protic
solvents became important. Levels of the amide, bromo-
reduced material and unreacted 1 were measured by HPLC.
Full conversion was critical in achieving low impurity levels.
reactor. The Voyager is based on the same microwave cav-
ity as the Discover, but uses a larger (80 mL) glass vessel,
with a sealed head that has inlets for a temperature probe
and a tube for addition/removal of reaction mixtures. Valves
seal the vessel during operation through a loop containing
a pressure sensor. The addition/removal of solutions/slurries
requires calibration, but can then be automated to give a
stop-flow continuous process capability.
Table 1
The optimised conditions were exactly reproduced at 50 mL
(14 g of 1) batch scale. The same program could be used,
Entry
Additives
None
Conversion (%)
only by changing the increase in target temperature
ꢀ
1
2
3
4
5
0
97
90
15
(160 C), which was required for complete conversion (iso-
2
1% H O
5% Zn
5% CuI
20% TMEDA
lated yield 99%, >98.5% purity). The target temperature dif-
ference is presumably due to the reported discrepancy in
temperature measurement between the Discover and Voy-
100
2
7
ꢀ
2
, DMF, MWat 180 C
ager systems (infra-red vs fibre optic). With the larger ves-
sel, but same maximum power input (300 W), the overall
All reactions with 1 mol % Pd–dppf, 1 equiv Zn(CN)
for 300 s.

M. R. Pitts et al. / Tetrahedron 62 (2006) 4705–4708
4707
PPh2
PPh2
PPh2
PPh2
dppf =
dpppe =
PPh2
PPh2
dppe =
Fe
PPh2
PPh2
PPh2
PPh2
O
O
O
P(furyl)3 =
) P
Xantphos =
3
DPEphos =
Me Me
Figure 1.
heating step takes lightly longer. Target temperature is still
reached in less than 2 min, but cooling takes around 3 min
rather than one to reach a safe temperature. This equates
to around a 7 min heating step. When the addition/removal
automation time is added for stop-flow use, this gives a cycle
time of around 10 min per batch, with approximately 12 g of
starting material processed per batch.
3. Conclusions
We have developed a robust cyanation procedure for 1, using
the lowest catalyst loadings yet reported for microwave-
enhanced cyanation. The process cuts the reaction time from
24 h to 2 min with an excellent impurity profile. These con-
ditions have been demonstrated to be generally applicable
to cyanation for a range of substrates. The cyanation was
successfully run in a continuous stop-flow reactor.
Two batches of (S)-1 were subjected to the procedure yield-
ing over 20 g of escitalopram with no loss of enantiomeric
purity. Racemic 1 (56 g) was run under a continuous (stop-
flow) process in four cycles yielding 47 g of citalopram 2
4. Experimental
(
>98% purity) in a total run time of 40 min. A sample
from each cycle was analysed by HPLC, showing a very
high degree of consistency between batches. A longer run
provided 150 g in 11 cycles (under 2 h). Extrapolation of
this shows multikilogram quantities could be produced in
a useful timeframe (days).
Microwave reactions were carried out in a commercially
available monomode system (CEM Discover or Voyager).
The reactor has a variable power output from 0 to 300 W.
Small scale (<2 g) test reactions were carried out in the Dis-
cover with the accompanying 10 mL (5 mL working vol-
ume) tubes with septum tops. The Voyager batch reactions
were performed in a thick-walled glass vessel (capacity
The system developed for citalopram was tested on a series
of simple aryl halide substrates. Xantphos as ligand was
demonstrated to give more active catalysts for aryl chlorides
than dppf. Cyanation conditions were rapidly optimised in
the microwave for a small selection of aryl halides; all giving
very high yields in clean reactions (see Table 3).
80 mL, maximum working volume 50 mL). The vessel is
isolated from the computer controlled system for charging
the reaction contents by a valve. The pressure is controlled
and monitored by a load cell connected through this valve
with a 300 psi release valve for safety. The temperature of
the reaction mixture was monitored using a fibre-optic probe
inserted into the reaction vessel in a sapphire immersion
well. The contents of the vessel are stirred by means of a
rotating magnetic plate located below the floor of the micro-
wave cavity and a Teflon-coated magnetic stir bar in the ves-
sel. Temperature, pressure and power profiles are recorded
by the accompanying software.
Table 3
Entry Starting material
Product
Catalyst Conversion
mol %) (%)
(
O
O
Me
Me
Me
Ph
1
0.1
0.5
0.2
100
100
Br
CN
CN
CN
4.1. Preparation of escitalopram (S)-2, batch method
O
O
O
To a solution of bromo compound (S)-1 (>98% ee, 13.88 g,
36.7 mmol) in DMF (35 mL) in an 80 mL CEM Voyager
Me
2
0
0
Cl
microwave tube N,N,N ,N -tetramethylethylenediamine
1.1 mL, 7.3 mmol, 0.2 equiv), zinc cyanide (2.58 g,
2.0 mmol, 0.6 equiv), tris(dibenzylideneacetaone)dipalla-
(
2
O
Ph
3
100
93
dium(0) (84 mg, 92 mmol, 0.5 mol %) and Xantphos
(213 mg, 0.37 mmol, 1.0 mol %) were added successively.
The reaction tube was sealed and heated to 160 C under
Br
ꢀ
F3C
F3C
microwave irradiation with a 200 s hold time, and 300 W
maximum power input. After cooling under a stream of com-
pressed air, the reaction mixture was washed out (with
dichloromethane) through a pad a Celite with a thin layer
of silica in the middle.
4
1.0
0.5
Cl
CN
N
N
N
5
100
N
Br
CN
The combined filtrates were concentrated, and then dried ex-
tensively under high vacuum to give escitalopram as a yellow
All reactions with Pd–Xantphos, 0.6 equiv Zn(CN) , 0.2 equiv TMEDA,
2
DMF, MW at 180 C for 300 s.
ꢀ

4708
M. R. Pitts et al. / Tetrahedron 62 (2006) 4705–4708
1
gum (w 22 g). Structure confirmed by GC–MS and H NMR;
chiral purity measured by cHPLC (OD-H column, 95:5 hex-
A. J. Org. Chem. 2001, 66, 4165–4169; (e) Walla, P.; Kappe,
C. O. Chem. Commun. 2004, 564–565.
3. Sundermeier, M.; Zapf, A.; Beller, M. Eur. J. Inorg. Chem.
2003, 3513–3526; Ager, D. J.; Erickson, R. A.; Froen, D. E.;
Prakash, I.; Zhi, B. Org. Process Res. Dev. 2004, 8, 62–71;
U.S. Patent 6,699,993.
. Ponchant, M.; Hinnen, F.; Demphel, S.; Crouzel, C. Appl.
Radiat. Isot. 1997, 48, 755–762.
. L u¨ cke, B.; Narayana, K. V.; Martin, A.; J a¨ hnisch, K. Adv.
Synth. Catal. 2004, 346, 1407–1424.
. Ellis, G. P.; Romney-Alexander, T. M. Chem. Rev. 1987, 87,
779–794.
. Excellent review: Sundermeier, M.; Zapf, A.; Mutyala, S.;
Baumann, W.; Sans, J.; Weiss, S.; Beller, M. Chem.—Eur. J.
2003, 9, 1828–1836.
. Marcantonio, K. M.; Frey, L. F.; Liu, Y.; Chen, Y.; Strine, J.;
Phenix, B.; Wallace, D. J.; Chen, C. Org. Lett. 2004, 6,
ꢁ
1
ane with 0.4% diethylamine to isopropanol, 0.5 mL min
2
,
20 nm detection) shows >98% ee.
4.2. Preparation of citalopram 2, continuous stop-flow
method
4
5
6
7
The cyanation was also carried out in a continuous process
for racemic citalopram.
The CEM Voyager stop-flow reactor was fed from two stirred
vessels: (1) tris(dibenzylideneacetaone)dipalladium(0)
400 mg, 0.87 mmol, 0.6 mol %) and Xantphos (920 mg,
.59 mmol, 1.1 mol %) slurried in DMF (40 mL), and (2)
the bromo starting material 1 (56.0 g, 148 mmol), zinc cya-
(
1
8
9
nide (ground to 600 micron powder, 10.8 g, 92 mmol,
0
3
723–3725.
. Sundermeier, M.; Mutyala, S.; Zapf, A.; Spannenberg, A.;
.62 equiv) and TMEDA (4.8 mL, 32 mmol, 0.22 equiv)
slurried in DMF to a total volume of 160 mL. A program
cycle was calibrated to add 10 mL from vessel 1 and
Beller, M. J. Organomet. Chem. 2003, 684, 50–55.
0. Sundermeier, M.; Zapf, A.; Beller, M. Angew. Chem., Int. Ed.
2003, 42, 1661–1664.
11. Sundermeier, M.; Zapf, A.; Beller, M.; Sans, J. Tetrahedron
Lett. 2001, 42, 6707–6710.
2. Anderson, B. A.; Bell, E. C.; Ginah, F. O.; Harn, N. K.; Pagh, L.
M.; Wepsiec, J. P. J. Org. Chem. 1998, 63, 8224–8228.
3. Yang, C.; Williams, J. M. Org. Lett. 2004, 6, 2837–2840.
4. Schareina, T.; Zapf, A.; Beller, M. Chem. Commun. 2004,
1
40 mL from vessel 2, then heat with microwave irradiation
to 160 C (300 W max. power) for a 200 s hold time, cool
ꢀ
to 50 C then remove to a clean vessel. The program was
ꢀ
repeated continuously over four cycles with a total run time
of 40 min. The slurry generated was filtered through Celite
and analysed by GC–MS to show >98% conversion. Chemi-
cal purity was further determined by HPLC as >98% (Zor-
bax SB C18 column, 25 cmꢂ4.6 mm, MeCN/[NaH PO /
1
1
1
2
4
ꢁ
1
1
388–1389.
5. Schareina, T.; Zapf, A.; Beller, M. Tetrahedron Lett. 2005, 46,
585–2588.
OctSO H aqueous buffer pH 2.8] 32:68, 1.5 mL min
,
3
1
1
1
2
10 nm detection).
2
6. Alterman, M.; Hallberg, A. J. Org. Chem. 2000, 65, 7984–
7989.
7. Tschaen, D. M.; Desmond, R.; King, A. O.; Fortin, M. C.;
Pipik, B.; King, S.; Verhoeven, T. R. Synth. Commun. 1994,
24, 887–890.
Acknowledgements
We thank Ajit Bhobe and Nandu Chodankar of Sekhsaria for
important suggestions for parts of this work and supply of 1.
1
1
2
8. Maligres, P. E.; Waters, M. S.; Fleitz, F.; Askin, D. Tetrahedron
Lett. 1999, 40, 8193–8195.
9. Stazi, F.; Palmisano, G.; Turconi, M.; Santagostino, M. Tetra-
hedron Lett. 2005, 46, 1815–1818.
References and notes
ꢀ
ꢁ4
ꢁ1
0. Zinc cyanidesolubility in DMF (at 80 C) is 1.8ꢂ10 g mL
.
1
. (a) Microwave Assisted Organic Synthesis; Lidstr o¨ m, P.,
Tierney, J. P., Eds.; Blackwell: Oxford, 2004; (b) Microwaves
in Organic Synthesis; Loupy, A., Ed.; Wiley-VCH: Weinheim,
002; (c) Microwave Synthesis: Chemistry at the Speed of
Light; Hayes, B. L., Ed.; CEM: Matthews, NC, 2002; (d)
Kappe, C. O. Angew Chem., Int. Ed. 2004, 43, 6250–6284;
e) Lidstr o¨ m, P.; Tierney, J. P.; Wathey, B.; Westman, J. Tetra-
hedron 2001, 57, 9225–9283; (f) Caddick, S. Tetrahedron 1995,
1, 10403–10432.
21. Jin, F.; Confalone, P. N. Tetrahedron Lett. 2000, 41, 3271–3273.
22. Hatsuda, M.; Seki, M. Tetrahedron Lett. 2005, 46, 1849–1853.
23. Chidambaram, R. Tetrahedron Lett. 2004, 45, 1441–1444.
24. For cyanation with nickel catalysts: Sakakibara, Y.; Ido, Y.;
Sasaki, K.; Sakai, M.; Uchino, N. Bull. Chem. Soc. Jpn.
1993, 66, 2776–2778; Sakakibara, Y.; Okuda, F.; Shimobaya-
shi, A.; Kirino, K.; Sakai, M.; Uchino, N.; Takagi, K. Bull.
Chem. Soc. Jpn. 1988, 61, 1985–1990; Arvela, R. K.; Lead-
beater, N. E. J. Org. Chem. 2003, 68, 9122–9125.
2
(
5
2
. (a) Leadbeater, N. E. Chem. Commun. 2005, 2881–2902; (b)
Stadler, A.; Yousefi, B. H.; Dallinger, D.; Walla, P.; Van der-
Eycken, E.; Kaval, N.; Kappe, C. O. Org. Process Res. Dev.
25. Zhang, A.; Neumeyer, J. L. Org. Lett. 2003, 5, 201–203.
26. For recent report using the Voyager: Arvela, R. K.; Leadbeater,
N. E.; Collins, M. J. Tetrahedron 2005, 61, 9349–9355.
27. Leadbeater, N. E.; Pillsbury, S. J.; Shanahan, E.; Williams,
V. A. Tetrahedron 2005, 61, 3565–3585.
2003, 7, 707–716; (c) Appukkuttan, P.; Van der Eycken, E.;
Dehaen, W. Synlett 2003, 1204–1206; (d) Erd e´ lyi, M.; Gogoll,